In collaboration with

Disclaimer

The results of the GMPPA-CDG drug repurposing project that we are sharing in the spirit of open science below are novel preclinical research findings and therefore they do not constitute the practice of medicine. Please consult a physician or clinical care team if considering off-label use of any approved drug or compassionate use of any experimental drug. The same caution applies to nutraceuticals, supplements and “generally recognized as safe” compounds.

For most rare genetic diseases, cure-focused research depends entirely on entrepreneurial families seared by a bolt of genetic lightning. The reverberations thunder into reality when the first life-shattering signs of disease manifest. Often at birth but sometimes lurking for years before the shockwaves suddenly make impact on the fateful day a confirmed diagnosis is delivered in a genetic test report.

A cruel stroke of the genetic pen converts the book of life into its opposite. Well-made plans go out the window. Dear readers of Cure Odysseys are by now all too familiar with the unrelenting tragedy of rare genetic diseases that unfolds mercilessly every day across the globe. And so they also know all too well that tragedy fuels triumph.

Not all parents are in a position to buck the advice of credentialed experts and bushwhack through the wild lands of basic science, translational research, and clinical trials. But some parents can’t help themselves. Through grit, self-instruction and mentorship, founder parents reveal their inner scientist, community builder, and drug developer — oh and toss fundraiser in the mix, too. Traits they might not have manifested in their professional lives before their child’s diagnosis. Or never even seen in themselves before.

No sane parent signs up for the rare disease life. They are wickedly thrust into the arena, like a righteous gladiator. The collective force of will of a community of affected families organized by its natural leaders makes rare disease medicines possible. Spinal muscular atrophy (SMA), Angelman syndrome, cystic fibrosis are just three top-of-mind scientific and commercial success stories in rare.

But a well-versed critic is fair to point out that those disease are common to medium rare. What about the long tail of rare genetic diseases affecting a few dozen to a few hundred patients? It’s lonely out there.

The only consolation of a catastrophic genetic lightning strike is that it mobilizes exceptional founder parents to put their children’s rare genetic disease on the map, and then devise a plan to wipe it from the face of the Earth.

CDGs — Congenital Disorders of Glycosylation — are a natural experiment in what happens when families are nucleated into a community by one or more founder parents. From the PIGN Cure Collective and Cure DHDDS on the larger side to Behind the Rainbow (PIGW-CDG) and Cure SRD5A3-CDG on the smaller side, we’ve seen these community nucleation events across CDG subtypes over and over again, and across the rare disease landscape for that matter.

We obviously wouldn’t be talking about GMPPA-CDG if there weren’t a founder parent whose child is personally affected. That parent is Diana Roetting and her daughter Rosie, who is living with the particularly uncommon CDG subtype called GMPPA-CDG. Diana knows of just six GMPPA-CDG patients total around the world, one of the rarest of the rare genetic diseases.

This story is about how one family is creating a scalable cure blueprint for a disease so tiny you can count the patient population on two hands.

Subscribe now

When we last checked in on the GMMPA-CDG cure oddyssey two years ago, we had evaluated and then ruled out a GMPPA-CDG worm model. As a hedge, we pursued a fly model in parallel. We were handsomely rewarded by a strong lethality phenotype, meaning any rescuer compound would have to overcome the high bar of lethality and propel affected embryos to adulthood.

CDGs are a family of inherited metabolic diseases. Nearly 200 genes encode a protein that serves as a node in the cellular sugar supply chain. Sugars are imported as raw materials, processed into precursors and advanced parts, assembled into manufactured goods, and exported as finished products. A robust sugar supply chain keeps the cell humming.

The GMPPA enzyme is specifically involved in generating the precursors of the glycan chains that get snapped onto proteins at specific amino acid attachment points, e.g., asparagine residues. The enzyme converts mannose-1-phosphate to GDP-mannose, a so called activated nucleoside sugar. Imagine Lego bricks that can be stacked into towers.

GMPPA has a dancer partner called GMPPB, so the pathomechanism is more intricate. Loss-of-function mutations in the GMPPA gene leads to overactive GMPPB protein activity. In other words, the loss of GMPPA function leads to a gain of GMMPB function, ultimately resulting in excess glycosylation, specifically hypermannosylation. Normally in CDG, the disease driver is insufficient glycosylation, or hypoglycosylation. But in the case of GMPPA-CDG, too little of something but also too much of something — in this case GDP-mannose precursor — can be a bad thing according to the still inscrutable wiles of the wire diagram of metabolism.

From a cellular biomanufacturing point of view, glycan chains have extremely tight specs. An extra mannose here, an unexpected branch point there, and soon enough the entire proteome is riddled with defective or suboptimal glycans. Some proteins and pathways are more vulnerable to glycosylation imbalance than others, a recurring theme across inherited metabolic diseases. This heterogeneity bubbles up as a spectrum of symptoms and symptom severity in patients living with GMPPA-CDG.

Two years ago, a fly screen of over 1,000 known drugs in Professor Clement Chow’s lab at the University of Utah surfaced three top hits that rescue lethality of GMPPA deficient flies, and, remarkably, share a hidden pharmacophore: repirinast, finasteride and bicalutamide.

Repirinast is a mast cell inhibitor and antihistamine developed for the Japanese market in the 1970s and 1980s, but has since been discontinued for commercial reasons. The known on-target of repirinast did not immediately present a connection to GMMPA or glycosylation. So what do these three compounds have in common? The breakthrough insight was realizing that finasteride targets SRD5A2, an enzyme closely related to SRD5A3, which was first described as a CDG disease-driver gene in 2010.

With the proper lens to refract the data, a clear mechanistic hypothesis emerges from the three hit structures in isolation, denoting a conserved structure-activity relationship, or what we refer to as the SRD5A3 active site pharmacophore, which is depicted below in cyan. Independent of chemical structure, the known shared pharmacology of the steroidal antiandrogen finasteride and the nonsteroidal antiandrogen bicalutamide further corroborates the prediction: repirinast is a cryptic SRD5A3 inhibitor.

We’ve revealed pharmacophores like this in almost every drug repurposing screen — now numbering over a dozen examples — regardless of which flavor of patient avatar is used to conduct the screen: yeast, worm, fibroblast, or in this case, fly. We’ve also seen pharmacophores in drug repurposing hit lists from Unravel Bioscience, which skips the patient avatar step and samples gene expression in patient cells from a nasal swab.

The SRD5A3 pharmacophore is observed in both the closed conformation, as with repirinast and finasteride, as well as in the open conformation of bicalutamide. That’s another pattern we’ve seen before: cyclic versus linear versions of the same pharmacophore. Coincidentally, it just so happens that we observed that same pattern in the worm-powered screen for SRD5A3-CDG, where the mevalonate pharmacophore (in purple) reveals itself, or a fragment thereof, in both linear and cyclic forms in the top hits from the screen.

Does the link between GMPPA-CDG and SRD5A3-CDG go deeper? You bet it does.

Based on the totality of the data collected thus far, we hypothesize that SRD5A3 inhibition is a rescue mechanism that compensates for the effects of GMPPA deficiency. Rebalancing glycosylation in the context of GMPPA deficiency means somehow compensating for or relieving the surfeit of GDP-mannose flooding into the sugar supply chain upstream of SRD5A3. Rebalancing metabolism is another recurrent theme: if one dial is throttled up, another is turned down.

We believe the fly results below are telling us that the cell copes with GMPPA deficiency by reducing the availability of dolichol phosphate mannose (Dol-P-Man), the universal mannose donor for multiple steps in the N-linked glycosylation pathway. Less Dol-P-Man is available, which in the absence of a way to reduce GDP-mannose levels directly appears to be the safest and most effective way for the cell to deal with excess GDP-mannose it cannot for some reason easily and safely shunt or metabolize away.

In the absence of molecular target data, the GMPPA-CDG fly model provides the most compelling evidence that repirinast is a cryptic SRD5A3 inhibitor. Knockdown of SRD5A3 alone is not lethal, but knockdown of GMPPA alone is. However, knockdown of both GMPPA and SRD5A3 results in near complete rescue of lethality. Fly genetics confirms the pharmacology and a “two wrongs make a right” mechanism.

We’ve seen the “two wrongs make a right” rescue mechanism before in CDG across multiple model systems, even in humans. In yeast, for example, where genetic inhibition of PGM1 rescues PMM2 deficiency. In flies, DPM1 is as a disease modifier of DPAGT1, just like in the GMPPA/SRD5A3 case. And in humans, ALG6 mutations modify PMM2-CDG disease severity.

The call is coming from inside the house!

If you think about it, where else could it be coming from? Like any small-world network, there are intimate and redundant connections between nodes in cellular metabolism. What’s not obvious is the disease modifier pairs among all CDG genes. Efforts should be made to map the CDG disease modifier network. What we can say for certain now is that the evidence for a linkage between GMPPA and SRD5A3 is based both on fly genetics and first-principles pharmacology.

Ironically, we found a different pharmacophore (mevalonate) enriched in SRD5A3 rescuers, but the rebalancing principle is the same. In SRD5A3 deficiency, upstream metabolic flux is reduced at the source by inhibiting mevalonate production by the enzyme HMG CoA. The amount of polyprenol flowing to SRD5A3 is reduced to match its lower catalytic capacity. Two wrongs make a right, genetically and pharmacologically, seems to be the order of the day.

Share

You might be asking yourself: why not start a single-patient IND study for repirinast just like we did years ago for epalrestat, another Japanese drug repurposed for a CDG?

Ideally to buttress the fly data we’d like data from Rosie’s cells showing improvement in glycosylation upon treatment with repirinast. That would connect the critical dots between flies and human cells; and between SRD5A3 inhibition and rebalanced glycosylation at the proteome level. Those experiments were completed and analysis did not reveal a clear rescue signal. We’ve seen this movie before though. Fibroblasts are not neurons, and neurons are not brains. A fly does happen to combine all all three bit to some extent.

However, there was a practical blocker to a repirinast spIND study: drug manufacturing. Unlike in the case of epalrestat, which is still marketed in Japan and available by doctor’s order, repirinast was commercially discontinued years ago. This is the fate of many drugs around the world: supplanted by next-generation successors with improved efficacy or simply on-patent derivatives that have greater revenue potential. Bloom for a Cure has both an ODD and RPDD granted for repirinast and GMPPA-CDG. But manufacturing costs are in the hundreds of thousands of dollars. Not a trivial commitment to make without having more confidence that the drug will actually work in a human.

As a middle ground, Diana decided to pursue a spIND for finasteride, which is available by prescription in the US. The fly data tell us that finasteride is targeting SRD5A3, close cousin to its intended target SRD5A2. After getting buy-in from a pediatric endocrinologist and in consultation with a team that includes a local physician so as to minimize travel for site visits, a finasteride single-patient observational study has officially begun for GMPPA-CDG pioneer Rosie under the direction of Dr Eva Morava-Kozicz at Mount Sinai in New York City.

Rosie took her first dose of finasteride on Sunday July 27, 2025.

A protocol was devised that assesses safety and a suite of exploratory outcome measures. We’re focused on tests that could objectively read out efficacy. One extremely practical outcome measure happens to be in play for GMPPA. Namely, alacrima, or lack of tear production. Similar to NGLY1 deficiency, alacrima is a core symptom of GMPPA deficiency. It’s a rare symptom in CDGs overall, and how deep the linkage between GMPPA and NGLY1 goes is still the bleeding edge of science.

Harkening back to Matt Might showing his dearly departed son Bertrand’s first tears after a N-of-1 drug repurposing trial, alacrima could act as a rapid indicator of target engagement and therapeutic effect. It’s reminiscent of keratosis pilaris responding rapidly to ibuprofen treatment in MAN1B1-CDG. Swallowing would be the next area of improvement, as Rosie takes in a good chunk of her daily calories through liquid or puree. If she is able to swallow food, that would be a quantitative biomarker with real-world positive impact day to day. Cognitive and motor improvements would be the other areas to look for potential treatment effect. We don’t know which symptoms or disease processes will respond to treatment.

This is why we N-of-1.

Share Perlara’s Cure Odysseys

Prepared for Laura Avery and the FBXO31 Foundation

Authors:

• Mathuravani Thevandavakkam, PhD, Cure Guide and Director of Drug Repurposing, Perlara PBC

• Uche Medoh, PhD. Cure Guide, Perlara PBC & Principal Investigator at the Arc Institute

• Ethan O. Perlstein, PhD. CEO of Perlara PBC & Maggie’s Pearl, LLC & Curetopia

Executive Summary

The goal of the FBXO31 Cure Roadmap is to accelerate the identification of a treatment modality for patients with FBXO31 D344N mutation whose parent-led foundation commissioned the development of this Roadmap. This work is expected to enhance our understanding of FBXO31 function and its impairments leading to new treatment options for FBXO31-related neurological disorders and potentially even for neurodevelopmental disorders associated with cerebral palsy.

Our key recommendations for the next 12-24 months are:

• Initiate a drug repurposing screen on patient iPSC-derived neural progenitor cells to identify clinically actionable repurposable hits. This effort driven by Perlara is expected to yield repurposable hits that can be translated into so-called “1-to-N” trials (coordinated, parallel single-patient studies expanded to broader patient communities).

• Build a drug repurposing consortium that would bring together a broader coalition of clinicians and researchers supported by the FBXO31 Foundation. The goal of this consortium is twofold: (a) enable the sharing of patient information and biosamples, and (b) delineate the role of repurposed drugs in alleviating FBXO31 pathology at a cellular level. It’s expected that these studies would be invaluable for accelerated drug development.

• Additionally explore alternative drug development opportunities for FBXO31 D344N variant, including ASO (antisense oligonucleotides), PROTAC (proteolysis-targeting chimeras), and other gene therapy approaches.

• Increase awareness of FBXO31-related neurological disorder, particularly via centers for neurological diseases or specializing in cerebral palsy. It’s expected that awareness leading to more sequencing efforts could further enhance the patient pool with FBXO31 variants.

Introduction

Disease Overview: Basic science

FBXO31 and Cell cycle: The cell cycle is a precisely timed series of events fundamental to normal cellular function and organismal health¹. Progression through the different phases of the cell cycle is tightly regulated by specific checkpoints and controlled by regulatory proteins including cyclins and cyclin-dependent-kinases (CDKs)². Aberrations in cell cycle mechanisms lead to an array of diseases, from cancers to neurodegenerative disorders.

The ubiquitin-proteasome system (UPS) plays a critical role in regulating the cell cycle by targeting specific proteins for degradation. The UPS consists of three subunits: E1 (ubiquitin-activating), E2 (ubiquitin-conjugating) and E3 (ubiquitin ligase), that activate ubiquitin and subsequently transfer it to the protein destined for degradation³. Specifically, UPS regulates transitions between the cell cycle phases via cyclin degradation, consequently tightly regulating the levels of cyclins and associated cyclin dependent kinases². This system ensures that cellular function and genomic integrity are maintained³.

FBXO31 (F-box only protein 31) was discovered as a tumor suppressor factor in 2005⁴. The F-box proteins (FBPs) are a diverse group of proteins characterized by the presence of a conserved F-box motif of ~40 amino acids⁵. This motif allows the protein to interact with SKP1 (S-phase kinase-associated protein 1) and constitutes the SCF (SKP1-Cullin-F-box protein) E3 ubiquitin ligase complex. The SCF complex is instrumental in targeting proteins for ubiquitin-mediated degradation, thereby regulating various cellular processes such as cellular proliferation, cell cycle progression, and apoptosis. To date, there are over 600 E3 ligases that have been identified, of which the SCF or Cullin-RING E3 ligase (CRL) family is the largest, with CRL1 being the most characterized^5–7.

FBXO31, belonging to the FBXO subfamily of F proteins, is expressed across tissue types with particularly high expression in the brain and is highly expressed in the G2→G1 phase of the cell cycle⁴. It has a well-defined role in DNA damage response and tumorigenesis. Acting as a tumor suppressor (alterations causing loss-of-function effects)⁸, FBXO31 has been implicated in breast cancer^4,9, melanoma^10,11, hepatocellular carcinoma¹², gastric cancer^13,14, and glioma¹⁵. Conversely, it functions as an oncogene (alterations causing gain-of-function effects) in esophageal squamous cell carcinoma¹⁶ and lung cancer¹⁷.

FBXO31's role in the DNA damage response includes halting cell cycle progression to allow for damage assessment and repair, particularly through pathways involving Ataxia telangiectasia mutated kinase (ATM)¹⁸, DNA-dependent protein kinase (DNA-PK), and Ataxia telangiectasia and Rad3-related protein (ATR) that maintain genomic stability by detecting DNA double-strand breaks by homologous recombination and non-homologous end joining, and in response to replication stress and single-strand breaks respectively¹⁹. In response to genotoxic stress, FBXO31 facilitates cyclin D1 degradation²⁰.

Substrate specificity and neuromodulation of FBXO31: Cyclin D1, a substrate of SCF^FBXO31, plays a pivotal role in initiating and advancing the cell cycle through the G1 phase in response to external stimuli^20,21. Considered a proto-oncogene²², changes in the ubiquitination process and the subsequent stability of cyclin D1 have been identified as cancer biomarkers ²³, indicative of the progression of the disease.

An unconventional function for cyclin D1 in neuronal stem cells is its effect on cell fate influencing whether cells proliferate or differentiate by regulating the G1 phase²⁴. In pathological contexts, nuclear cyclin D1 can prompt apoptosis in post-mitotic neurons²⁵. Interestingly, cyclin D1's presence in the cytoplasm of differentiating neurons and neuroblastoma cells suggests roles in preventing apoptosis or promoting cell cycle withdrawal²⁶. Additionally, cytoplasmic cyclin D1 in the cytoplasm supports neuritogenesis in response to nerve growth factor (NGF) and contributes to neuronal plasticity during hippocampal development^27,28. More recently, cyclin D1 was identified to modulate the activity of Gamma-aminobutyric acid (GABA), a crucial neurotransmitter in the brain, and thereby neuronal inhibition²⁹. Deficits in GABA receptor function are implicated in various neurological disorders, from anxiety to epilepsy, with a significant role in synaptic inhibition and neurodevelopment in the hippocampus and cortex^30,31.

An emerging role for FBXO31 points towards it being a regulator of neuronal morphogenesis and polarity³². These two distinct processes create the information flow necessary for brain functions such as memory, learning, and emotion. In addition to Parkinson’s disease protein Parkin I, FBXO31 is one of few E3 ubiquitin ligases that localizes to the centrosome, the microtubule organizing center, that regulates neuronal polarity, spindle formation motility and adhesion crucial to cell division³².Malfunctioning centrosomes potentiate cortical defects during development (the region of the brain associated with sensory, and motor development)^33,34. The ligase activity of FBXO31 drives neuronal morphogenesis and is attributed to axon and dendritic growth, and axon specification. FBXO31 also targets centrosome polarity protein Par6c for degradation³⁵. Reduction in Par6 degradation by another E3 ligase (Smurf1) drives the formation of a Par3/Par6/aPKC complex^36,37 and thereby axon specification and polarity. Conversely, FBXO31 drives the proteasomal degradation of Par6c (the most abundant Par6 protein) regulating axonal growth and neuronal polarity³².

These examples underscore the significance of FBXO31 in maintaining cellular function, supporting normal development, and potentially influencing the pathogenesis of neurodegenerative diseases. Other key FBXO31 substrates include MCM7, MDM2, CDT1, FOXM1, Slug, Histone H2AX, TRF1, FAT10 perturbations of which can impact neurogenesis, neuronal survival, synaptic plasticity, and in general brain development³⁸.

It has been proposed that the conserved binding pocket of FBXO31 enables the selective recognition of C-terminal peptides with an amide group (CTAPs), thereby serving as a distinct marker for proteins targeted for degradation by FBXO31, including hormones and neuropeptides³⁹. Mutations affecting the binding domain of FBXO31 disrupt this specificity, leading to the degradation of newly identified proteins lacking amidation, while proteins originally marked with CTAPs accumulate within the cell. This toxic gain-of-function mutation in FBXO31 affects the function of downstream substrates but also results in the accumulation of protein degradation products, thereby disrupting the cell cycle. Dysregulation of this process in tissues where FBXO31 is highly expressed such as the brain ⁴ and skeletal muscles⁴⁰, can significantly impact the development of neuro and skeleton-muscular disorders.

Disease Overview: Clinical Features of FBXO31-related cerebral palsy

To date, there have been six reported cases of a neurodegenerative disorder related to cerebral palsy caused by mutations in FBXO31⁴¹. All affected patients have a de novo D334N (p. Asp334Asn) mutation. This mutation is located in the cyclin D1 binding region of FBXO31 and alters the electrostatic charge in the binding pocket, leading to reduced levels of cyclin D1 in these patients compared to controls⁴². Conflicting studies suggest that FBXO31 amelioration increases levels of cyclin D1³⁹. Modulation via CTAP and switching in electrostatic charge in the binding domain likely together alter cyclin D1 binding and induce substrate-related toxicity.

The characteristic effect of mutation is a form of brain injury categorized as cerebral palsy typified by poor coordination, stiff muscles, weak muscles, and tremors. Infants, heterozygous for mutation, are born full-term and are not diagnosed with abnormalities prenatally. Hypotonia, low muscle tone, is identified in neonates followed by a diagnosis of cerebral palsy. MRI scans reveal a reduction in white matter distribution and content in certain patients. Gross motor milestones such as rolling over, sitting, crawling or walking are delayed in infants. Speech impairments (primarily receptive language) and cognitive deficits including attention deficit hyperactivity disorder are present. Central and peripheral motor pathways are affected leading to spasticity and dystonia. Children present with global developmental delay and varying levels of intellectual disability. However, underlying motor or neurodevelopmental issues do not worsen over time. Literature reports one patient with intestinal malrotation in whom bowel resection was required (see Figure 1 on clinical presentation below)⁴³ (includes observations by FBXO31 Foundation).

Recessive mutations in FBXO31 offer insights into a potential model for understanding the cerebral palsy phenotype associated with the D334N mutation. A 5-base pair deletion in FBXO31 results in a frameshift mutation that introduces a premature stop codon p.(Cys283asnfs*81). This mutation is characterized by dysmorphia and intellectual disability but does not manifest as cerebral palsy⁴⁴. The truncation of the protein leads to its degradation via nonsense-mediated decay resulting in reduced levels of FBXO31. Consequently, this reduction prevents any potential toxic gain-of-function effects. Although decreased FBXO31 levels likely disrupt normal neuronal development, compensatory mechanisms mitigate the extent of damage^44,45. This model posits that any intervention that corrects the gain-of-function toxicity will result in a significant correction of the pathophysiology. As such, there are many therapeutic approaches with the potential to positively impact the progression of FBXO31-related cerebral palsy (see Therapeutic Readiness).

Disease Models: Cellular Models

A crucial step in developing cures is establishing disease models to investigate disease mechanisms and test potential therapies. Reliable disease models provide essential evidence of safety and efficacy before clinical trials. The development of models here is essential for understanding how aberrant FBXO31 activity or lack thereof influences proteostasis and neurological dysfunction. Cellular models, the first to be developed, are relatively simple to generate and handle in vitro. These models are vital for large-scale therapeutic screenings and for exploring the activity and mechanisms of new treatments.

Developing a model system for FBXO31 is complicated by several factors:

• Substrate Interaction: FBXO31, like other F-box proteins, has multiple binding partners, making it difficult to delineate their interactions and downstream effects. Functional redundancy among F-box proteins⁴⁶ can also obscure the specific effects of FBXO31 mutations.

• Cell-Type Specificity: FBXO31's function varies across different cell types and tissues, making it challenging to develop models that accurately reflect its role in specific contexts, such as neuronal cells versus other cell types; developmental stages (during neuronal development vs post development).

• Model Organism Limitations: Common model organisms may not fully replicate human FBXO31-related disease phenotypes due to differences in gene expression, protein function, and developmental processes, limiting the model's relevance.

This often requires the establishment of different types of cellular models to answer questions iterating FBOX31 dysfunction and therapeutics. Thanks to efforts in academic discovery science, however, several options exist to overcome these challenges.

Patient iPSCs are currently in development and will serve as the foundational research asset in drug repurposing screens, and beyond.

Disease Models: Whole-Organism Models

At present, there is a lack of published whole-organism models for FBXO31. Owing to their genetic similarity to humans, mouse models are indispensable for studying the pathological mechanism of human diseases, as well as for identifying new treatments. The FBXO31 mouse and Zebrafish model are currently under development and/or validation.

Share

Subscribe now

Therapeutic Readiness:

Several therapeutic approaches have potential applicability to address the understood mechanisms of FBXO31-related cerebral palsy, offering potential benefits to patients. These modalities vary significantly in development strategy, precedence, timeline, and likelihood of success. In this section, we delineate these modalities within the context of an overarching treatment strategy for patients affected by FBXO31-related cerebral palsy.

The first approach involves drug screening, which Perlara is currently undertaking in collaboration with industry partners (refer to Biotechnology Readiness). The second modality, antisense oligonucleotide therapy, will be clinically developed by a biotechnology company and would be accessible through compassionate use pathways. Additionally, we discuss two emerging modalities that are currently in development. While these approaches have longer development timelines, they represent promising avenues for expanding treatment options over the next decade.

Modalities: Drug Repurposing Screen

The fastest route in developing a clinical treatment modality that could significantly help patients with FBXO31-related cerebral palsy is to screen drugs that are already approved for activities that might be beneficial therapeutically. Since we don't fully understand the specific reasons why the disease happens or how drugs work in these cases, unbiased drug repurposing screens can often give us valuable and promising results.

Lastly, a key challenge for this approach is the question of generalizability. While it is understood that FBXO31-related cerebral palsy shares common underlying molecular mechanisms, it is not guaranteed that the most promising hits with one patient’s cells will also apply to other patients due to the small patient pool. Validating the generalizability of a screening hit will require access to a broader array of patient-derived cell lines (see Biobanking).

Modalities: RNase H-Dependent ASO

A Gapmer ASO is a specific type of antisense oligonucleotide (ASO) designed to induce the degradation of target RNA through the recruitment of RNase H, an enzyme that specifically degrades the RNA strand of an RNA/DNA hybrid. This allele-specific ASO specifically targets the mutated RNA and forms a complementary RNA/DNA hybrid with the target RNA, which is recognized and cleaved by RNase H. This degradation of the target RNA reduces the levels of the corresponding mRNA (specifically the mutant FBXO31), leading to decreased production of the target protein. Gapmer ASOs are FDA-approved, Mipomersen and Inotersen are examples of FDA-approved ASO gapmers, used for treating familial hypercholesterolemia and hereditary transthyretin amyloidosis, respectively.

The FBXO31 D334N mutation is amenable to this intervention on two counts: (1) FBXO31 is haplosufficient, meaning one functional copy is enough to maintain normal function, and (2) this method specifically eliminates the toxic gain-of-function of the mutant allele. Recessive mutations in FBXO31 support this, as the severity of the phenotype is significantly reduced when the truncated FBXO31 is targeted for non-mediated decay, leaving only one functional FBXO31 copy.

Modalities: PROTACs

PROTACs (Proteolysis Targeting Chimeras) are a novel class of therapeutic molecules designed to target specific proteins for degradation within cells and are part of a larger class of molecules known as heterobifunctional molecules. These molecules contain two binding domains and function by recruiting their respective targets into proximity with each other They leverage the ubiquitin-proteasome system to achieve selective degradation of pathogenic proteins, offering a potential approach to treat diseases caused by aberrant protein function. They typically consist of two parts: one that binds to the target protein and another that recruits an E3 ubiquitin ligase⁷, which tags the protein for degradation by the cell's proteasome machinery. PROTACs have been described as “targeted protein degraders”.

In gain-of-function mutations where the FBXO31 D334N ubiquitinates and degrades target protein excessively (evinced by low cyclin D1 levels), PROTAC could be used to selectively degrade the mutant E3 ligase reducing toxic activity. Conversely, PROTACS can also be designed to recruit other functional E3 ligases to degrade substrates the mutant FBXO31 can no longer process.

The caveat with this approach is that it requires a clear biomarker to phenotype connection.

Modalities: Symptomatic Treatments

Common symptomatic treatments include pharmacological medications to manage symptoms of various neuromotor and psychiatric symptoms. Patients currently work with their physicians to develop the best management route. It’s expected that drug repurposing hits from the screen will broaden the scope of immediately accessible drugs that could improve the quality of life.

Deep Brain Stimulation (DBS) is an invasive treatment option aimed at managing symptoms. It involves using a neurostimulator to produce electrical pulses that regulate abnormal neural activity in precise brain regions through implanted electrodes. DBS effectively alters neural circuit activity by modifying neuronal firing patterns and potentially influencing neurotransmitter release⁴⁸. This therapy has demonstrated efficacy in treating movement disorders such as Parkinson’s disease and dystonia, as well as certain psychiatric conditions like treatment-resistant depression and obsessive-compulsive disorder⁴⁹.

Share

Share Perlara’s Cure Odysseys

Clinical Research and Development Readiness:

Developing new therapeutics, particularly in rare diseases, requires hand-in-glove coordination between clinicians and researchers, especially at the earliest stages. Having reviewed the scientific landscape, laboratory model systems, and promising technologies that define the space of possible therapeutic candidates, here we explore the readiness of the clinical healthcare system to support the development of these candidates as well as their evaluation in human patients - the final, costliest, and riskiest stage of the drug development process.

Clinical Research: Patient Identification and Community Engagement

Engagement of the patient community in the drug development process is essential. The recruitment of an adequate number of patients for clinical trials remains a significant bottleneck in the path to FDA approval, often making the clinical development phase the most time-consuming aspect of the entire drug development pipeline. Consequently, early initiatives that diagnose, identify, and register FBXO31 patients can yield substantial benefits, both in terms of cost-effectively accelerating drug development and improving health outcomes for the patient community.

Patients with FBXO31-related cerebral palsy are identified primarily through genetic testing, particularly in individuals presenting with atypical cerebral palsy features, such as non-spastic motor defects, cognitive deficits, and behavioral issues. These features, while characteristic, often require a multidisciplinary diagnostic approach, involving neurologists, developmental pediatricians, and psychiatrists, to prompt referral for genetic screening. Despite these diagnostic avenues, many patients face a prolonged diagnostic journey.

The challenge of accurately diagnosing FBXO31-related disorders is compounded by the need for broader awareness among clinicians regarding the condition and its genetic basis. To this end, whole exome sequencing is expected to emerge as a powerful diagnostic tool in the identification of FBXO31 mutations among cerebral palsy patients. WES allows for the comprehensive analysis of all coding regions in the genome, making it particularly useful in uncovering rare genetic variants like those associated with FBXO31⁵⁰. This approach is especially valuable when standard diagnostic methods fail to provide a clear diagnosis. This is also expected to uncover more patients with FBXO31-related cerebral palsy who may be mistaken for other neurodevelopmental disorders, leading to delays in appropriate identification. Establishing a patient registry could facilitate better tracking and management of these cases.

Community engagement is critical for advancing the understanding and treatment of FBXO31-related disorders. The FBXO31 Foundation would lead efforts in engaging patients and connecting them with resources, a coordinated effort involving awareness campaigns, support groups, and partnerships with advocacy organizations that are essential. These initiatives are expected to aid in patient identification and ensure that patients remain informed and engaged in research and clinical trials, ultimately accelerating the development of effective therapies.

Clinical Research: Natural History

A detailed understanding of the natural history of a disease is fundamentally important to a successful drug development campaign, particularly in the case of a rare disease. There are two major reasons for this: first, the natural progression of a disease offers clues as to how a drug that corrects it might work; second, a robust natural history study has significant regulatory implications. As part of a modernization of its handling of the regulation of new medicines for rare diseases, FDA now accepts natural history as a control group on a case-by-case basis. Much as a strong patient network can accelerate a clinical development timeline by facilitating enrollment, a natural history control group accelerates timelines by reducing the total number of patients required to produce a signal of efficacy that FDA requires for a drug to be marketed. This also has the advantage of reducing trial costs and addresses ethical concerns associated with withholding a potentially effective drug from a patient, often a child, with severe disease to recapitulate a natural history that we already understand (see Regulatory Landscape and Strategy).

Unfortunately, dedicated longitudinal natural history studies for FBXO31-related conditions are currently unavailable. The absence of such studies presents a significant gap in our understanding of FBXO31-related cerebral palsy and its progression. Natural history studies are essential for identifying clear prognostic indicators, such as the age of onset and genetic markers, and for establishing key clinical indicators of disease severity. The insights gained from such studies could play a critical role in the development of inclusion criteria and clinical endpoints for future therapeutic interventions.

Currently, natural history studies for FBXO31 are also complicated by the small patient pool. To this end, whole exome sequencing of cerebral palsy patients is expected to increase the FBXO31 patient pool. A larger pool enhances the statistical power of these studies and improves the identification of crucial surrogate endpoints, such as biomarkers. These endpoints are essential for monitoring disease progression, assessing therapy efficacy, and have significant implications for accelerating the drug development process through regulatory strategies.

Future investigations should focus on applying modern laboratory techniques, such as single-cell genetic analysis, proteomics, and metabolomics, to observe longitudinal changes associated with disease severity or progression in FBXO31-related conditions. Moreover, while large-scale genome-wide association studies may not yet be feasible due to the limited number of FBXO31 patients, exploring the role of nuclear genetic background in these conditions could yield valuable insights and potential biomarkers for disease progression.

Clinical Research: Biobanking

Reliable patient samples are essential for advancing research and therapeutic development for FBXO31-related disorders. The foundation of patient sample collection, cataloging, and distribution lies in the biobank or biorepository. Currently, there is no dedicated biobank for FBXO31, highlighting a critical gap in resources for researchers and drug developers in this space.

To support the creation of effective cellular models for research and drug screening, the most valuable samples for biobanking would include blood, skin, and muscle biopsies. While blood samples are easier to collect, skin and muscle biopsies offer cells that are more representative of the tissues most affected by FBXO31-related cerebral palsy. Fibroblasts from skin biopsies and myocytes from muscle biopsies, although more challenging to obtain and culture, are vital for creating robust and physiologically relevant models. These models are crucial for accurately replicating the disease environment in vitro, thereby enhancing the potential for developing effective therapies.

In addition to supporting research, biobanks play a key role in natural history studies by providing the high-quality, longitudinal samples needed to track disease progression. As more patients with FBXO31-related disorders are identified through methods such as whole exome sequencing, biobanking will become increasingly important for collecting and preserving these samples. This will not only aid in the discovery of biomarkers and therapeutic targets but will also strengthen the overall drug development pipeline by offering a centralized, accessible resource for researchers working to combat FBXO31-related conditions.

Cognizant of the above, the FBXO31 Foundation is leading efforts toward establishing an easily accessible biobank for FBXO31 and making any patient-derived cellular model(s) accessible to the wider research community that would significantly bolster efforts to understand and treat this rare disorder.

Clinical Development: Regulatory Landscape and Strategy

The FDA serves a critical function in the healthcare system by acting as the gatekeeper to the pharmaceutical market. Any new therapeutic must demonstrate both safety and efficacy in human clinical trials to gain FDA approval. Additionally, the FDA is responsible for the ethical evaluation and approval of these clinical trials, balancing the risks and potential benefits to patients. As such, a well-crafted regulatory strategy is essential for successful drug development.

Historically, the FDA has maintained a degree of detachment from drug developers to preserve its neutrality. However, the increasing need for innovative regulatory approaches, particularly for rare diseases, has led to the FDA becoming more involved early in the development process, often acting as a partner and resource. The FDA now offers several regulatory designations to aid the development of drugs in areas that are typically underserved by the biopharmaceutical industry. For therapeutic candidates targeting FBXO31-related cerebral palsy, several of these designations are likely applicable, each with significant implications for the development process.

Clinical Development: Biomarkers and Candidate Surrogate Endpoints

As we explore the clinical development pathway for FBXO31-related therapies, identifying and validating biomarkers remains a critical step in establishing robust surrogate endpoints. Biomarkers serve as key indicators that can provide early insights into disease progression, therapeutic efficacy, and overall patient outcomes. In the context of cerebral palsy, several biomarkers have emerged as potential candidates for further investigation, however, there are no approved biomarkers specific to FBXO31-related cerebral palsy.

This is a significant bottleneck to clinical readiness for an FBXO31 therapeutic. FDA will require significant evidence that any surrogate biomarker serving as a primary endpoint in a clinical trial is highly representative of the desired clinical outcome. As such, research toward the identification and validation of potential surrogate biomarker candidates and their specific relevance to FBXO31-related cerebral palsy disease progression should be prioritized and supported.

Neuroimaging biomarkers, including findings from MRI and Cranial ultrasound, provide critical information on structural and metabolic changes in the brain. These imaging modalities could serve as early indicators of disease severity and treatment response.

Existing research into biomarkers for cerebral palsy yields some interesting starting points. Biochemical biomarkers, such as metabolites in cerebrospinal fluid (CSF) or blood-based inflammatory markers, will offer another layer of insight into the underlying pathophysiology of FBXO31-related cerebral palsy⁵¹. At present, no biochemical biomarker consistently identifies patients with cerebral palsy, this could be attributed to clinical heterogeneity of the disease. Given the specific aberration in cyclin D1 function and levels in FBXO31 D334N, cyclin D1 could be developed as a potential biomarker specific to FBXO31-related cerebral palsy.

Additionally, abnormalities in neurotransmitter systems, such as altered levels of dopamine, serotonin, and GABA, have been associated with motor and cognitive impairments in FBXO31-related cerebral palsy. These neurotransmitters could serve as biomarkers to monitor the effectiveness of therapeutic interventions. Lastly, neuroinflammatory biomarkers such as elevated levels of cytokines such as IL-6, IL-8 could be instrumental in monitoring neuroinflammation and metabolic disturbances associated with the disorder⁵¹.

These biomarkers can be evaluated using either a targeted ELISA panel for protein quantification or a qPCR approach for assessing gene expression levels.

Functional biomarkers, assessed through tools like the Gross Motor Function Measure (GMFM) and neurodevelopmental scales, provide a direct link to patient outcomes, capturing the impact of disease on motor and cognitive functions⁴³. These assessments could be crucial in defining meaningful clinical endpoints, particularly in the context of long-term therapeutic efficacy.

Lastly, imaging-based functional biomarkers, such as those derived from functional MRI (fMRI) and electroencephalography (EEG), have the potential to reveal alterations in brain activity and connectivity patterns that correlate with clinical symptoms. As the development of FBXO31-targeted therapies progresses, the identification and validation of these biomarkers will be pivotal in establishing surrogate endpoints that can accelerate the drug approval process.

While challenges remain, particularly in the variability and validation of these biomarkers, their integration into the clinical development strategy for FBXO31-related cerebral palsy could significantly enhance our ability to track disease progression, evaluate therapeutic interventions, and ultimately improve patient outcomes.

Clinical Development: Clinical Endpoints

In nearly all cases, the primary relevant outcome to FDA (and, generally, to patients) is whether the drug produces some kind of clinical benefit that is meaningful. Even if surrogate endpoints are used to approve a drug, ongoing investigation to demonstrate such clinical benefit will continue, and if there is no evidence of clinical benefit the drug can be removed from the marketplace. As such, even though defining strong surrogate endpoints is key at this stage, it is important to consider clinical endpoints as well.

Clinical endpoints are generally more challenging, expensive, and time-consuming to measure than biomarker endpoints. Clinical endpoints are more subjective, measured over longer periods, and capture more inherent biological variability than biomarker endpoints do. Thus, reaching a clinical endpoint is a high bar that most therapeutic candidates do not achieve - but to offer a patient anything less than this high standard would be unacceptable.

Several clinical endpoints have been defined for cerebral palsy in general and their use can be cited in clinical trials for FBXO31-related cerebral palsy. These endpoints in both children and adults focus on evaluating the impact of the condition on functional abilities, quality of life, and overall health⁴³. For children, key endpoints include assessments of motor function through tools like the Gross Motor Function Measure (GMFM) and fine motor skill evaluations, functional independence using the Functional Independence Measure (FIM) and Pediatric Evaluation of Disability Inventory (PEDI), and cognitive and communication skills through relevant developmental tests. Quality of life is measured using instruments like the Pediatric Quality of Life Inventory (PedsQL), and muscle tone and spasticity are evaluated with the Ashworth Scale and Modified Tardieu Scale. In adults, similar endpoints are utilized, with additional focus on mobility and movement through tests like the Timed Up and Go (TUG) and Six-Minute Walk Test (6MWT), pain assessment with the Visual Analog Scale (VAS), and quality of life measures adapted for adult populations. Both age groups also benefit from assessments of psychosocial factors and secondary conditions to provide a comprehensive understanding of the disease’s impact and guide effective management strategies. Given the clinical heterogeneity of cerebral palsy, the above endpoints will be suitably refined to define standards for FBXO31-related cerebral palsy.

Biotechnology Ecosystem Readiness

Having explored the scientific landscape and considerations for the clinical development of therapeutic candidates, we now turn our attention to the biotechnology ecosystem crucial for advancing therapies for FBXO31-related cerebral palsy. This ecosystem encompasses academic institutions, industrial biotechnology, and both public and private funding sources. The successful development of new therapies for FBXO31 nearly always requires the engagement of multiple players within this ecosystem.

In the course of preparing this Roadmap, we’ve identified potential academic partners with knowledge and expertise that will enable accelerating the discovery of potential treatments and cures. At present, patient-derived neuronal cell lines are under generation and an in silico RNA sequencing drug repurposing screening approach employing multiple patients is being considered to identify potential drug repurposing candidates and lead-like molecules for FBXO31 therapeutics.

Share

References:

1. Vermeulen, K., Van Bockstaele, D. R. & Berneman, Z. N. The cell cycle: a review of regulation, deregulation and therapeutic targets in cancer. Cell Prolif 36, 131–149 (2003).

2. Johnson, D. G. & Walker, C. L. CYCLINS AND CELL CYCLE CHECKPOINTS. Annu. Rev. Pharmacol. Toxicol. 39, 295–312 (1999).

3. Hegde, A. N. Ubiquitin-proteasome-mediated local protein degradation and synaptic plasticity. Progress in Neurobiology 73, 311–357 (2004).

4. Kumar, R. et al. FBXO31 is the chromosome 16q24.3 senescence gene, a candidate breast tumor suppressor, and a component of an SCF complex. Cancer Res 65, 11304–11313 (2005).

5. Skowyra, D., Craig, K. L., Tyers, M., Elledge, S. J. & Harper, J. W. F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitin-ligase complex. Cell 91, 209–219 (1997).

6. Nguyen, H. C., Wang, W. & Xiong, Y. Cullin-RING E3 Ubiquitin Ligases: Bridges to Destruction. Subcell Biochem 83, 323–347 (2017).

7. Liu, Y. et al. Expanding PROTACtable genome universe of E3 ligases. Nat Commun 14, 6509 (2023).

8. Tan, Y., Liu, D., Gong, J., Liu, J. & Huo, J. The role of F-box only protein 31 in cancer. Oncol Lett 15, 4047–4052 (2018).

9. Song, Q. et al. Gene expression analysis on a photodiode array-based bioluminescence analyzer by using sensitivity-improved SRPP. Analyst 135, 1315–1319 (2010).

10. Wajapeyee, N., Serra, R. W., Zhu, X., Mahalingam, M. & Green, M. R. Oncogenic BRAF induces senescence and apoptosis through pathways mediated by the secreted protein IGFBP7. Cell 132, 363–374 (2008).

11. Bartkova, J. et al. Oncogene-induced senescence is part of the tumorigenesis barrier imposed by DNA damage checkpoints. Nature 444, 633–637 (2006).

12. Huang, H.-L., Zheng, W.-L., Zhao, R., Zhang, B. & Ma, W.-L. FBXO31 is down-regulated and may function as a tumor suppressor in hepatocellular carcinoma. Oncol Rep 24, 715–720 (2010).

13. Zhang, X. et al. F-box protein FBXO31 is down-regulated in gastric cancer and negatively regulated by miR-17 and miR-20a. Oncotarget 5, 6178–6190 (2014).

14. Malonia, S. K., Dutta, P., Santra, M. K. & Green, M. R. F-box protein FBXO31 directs degradation of MDM2 to facilitate p53-mediated growth arrest following genotoxic stress. Proc Natl Acad Sci U S A 112, 8632–8637 (2015).

15. Feng, Y., Liu, M., Xie, P., Dong, R. & Hao, Z. FBXO31 suppresses lipogenesis and tumor progression in glioma by promoting ubiquitination and degradation of CD147. Prostaglandins Other Lipid Mediat 163, 106667 (2022).

16. Kogo, R., Mimori, K., Tanaka, F., Komune, S. & Mori, M. FBXO31 determines poor prognosis in esophageal squamous cell carcinoma. Int J Oncol 39, 155–159 (2011).

17. Huang, H.-L. et al. FBXO31 promotes cell proliferation, metastasis and invasion in lung cancer. Am J Cancer Res 5, 1814–1822 (2015).

18. Choppara, S., Malonia, S. K., Sankaran, G., Green, M. R. & Santra, M. K. Degradation of FBXO31 by APC/C is regulated by AKT- and ATM-mediated phosphorylation. Proc Natl Acad Sci U S A 115, 998–1003 (2018).

19. Shiloh, Y. FBXO31: a new player in the ever-expanding DNA damage response orchestra. Sci Signal 2, pe73 (2009).

20. Santra, M. K., Wajapeyee, N. & Green, M. R. F-Box Protein FBXO31 Mediates Cyclin D1 Degradation to Induce G1 Arrest Following DNA Damage. Nature 459, 722–725 (2009).

21. Li, Y. et al. Structural basis of the phosphorylation-independent recognition of cyclin D1 by the SCFFBXO31 ubiquitin ligase. Proceedings of the National Academy of Sciences 115, 319–324 (2018).

22. Bates, S. & Peters, G. Cyclin D1 as a cellular proto-oncogene. Semin Cancer Biol 6, 73–82 (1995).

23. Chen, S. & Li, L. Degradation strategy of cyclin D1 in cancer cells and the potential clinical application. Front Oncol 12, 949688 (2022).

24. Lange, C., Huttner, W. B. & Calegari, F. Cdk4/cyclinD1 overexpression in neural stem cells shortens G1, delays neurogenesis, and promotes the generation and expansion of basal progenitors. Cell Stem Cell 5, 320–331 (2009).

25. Freeman, R. S., Estus, S. & Johnson, E. M. Analysis of cell cycle-related gene expression in postmitotic neurons: selective induction of Cyclin D1 during programmed cell death. Neuron 12, 343–355 (1994).

26. Sumrejkanchanakij, P., Eto, K. & Ikeda, M.-A. Cytoplasmic sequestration of cyclin D1 associated with cell cycle withdrawal of neuroblastoma cells. Biochem Biophys Res Commun 340, 302–308 (2006).

27. Tamaru, T., Trigun, S. K., Okada, M. & Nakagawa, H. Identification of cells expressing a D type G1 cyclin in matured brain: implication for its role in neuronal function. Neurosci Lett 153, 169–172 (1993).

28. Tamaru, T., Okada, M. & Nakagawa, H. Differential expression of D type cyclins during neuronal maturation. Neurosci Lett 168, 229–232 (1994).

29. Pedraza, N. et al. Cyclin D1—Cdk4 regulates neuronal activity through phosphorylation of GABAA receptors. Cell Mol Life Sci 80, 280 (2023).

30. Jiménez-Balado, J. & Eich, T. S. GABAergic dysfunction, neural network hyperactivity and memory impairments in human aging and Alzheimer’s disease. Seminars in Cell & Developmental Biology 116, 146–159 (2021).

31. Schmitz, T. W., Correia, M. M., Ferreira, C. S., Prescot, A. P. & Anderson, M. C. Hippocampal GABA enables inhibitory control over unwanted thoughts. Nat Commun 8, 1311 (2017).

32. Vadhvani, M., Schwedhelm-Domeyer, N., Mukherjee, C. & Stegmüller, J. The Centrosomal E3 Ubiquitin Ligase FBXO31-SCF Regulates Neuronal Morphogenesis and Migration. PLoS One 8, e57530 (2013).

33. Yang, J., Hu, X., Ma, J. & Shi, S.-H. Centrosome regulation and function in mammalian cortical neurogenesis. Curr Opin Neurobiol 69, 256–266 (2021).

34. Subramanian, L., Calcagnotto, M. E. & Paredes, M. F. Cortical Malformations: Lessons in Human Brain Development. Front Cell Neurosci 13, 576 (2019).

35. Jiang, T., McKinley, R. F. A., McGill, M. A., Angers, S. & Harris, T. J. C. A Par-1-Par-3-Centrosome Cell Polarity Pathway and Its Tuning for Isotropic Cell Adhesion. Curr Biol 25, 2701–2708 (2015).

36. Cheng, P., Lu, H., Shelly, M., Gao, H. & Poo, M. Phosphorylation of E3 ligase Smurf1 switches its substrate preference in support of axon development. Neuron 69, 231–243 (2011).

37. Chen, J. & Zhang, M. The Par3/Par6/aPKC complex and epithelial cell polarity. Exp Cell Res 319, 1357–1364 (2013).

38. Johansson, P. et al. SCF-FBXO31 E3 Ligase Targets DNA Replication Factor Cdt1 for Proteolysis in the G2 Phase of Cell Cycle to Prevent Re-replication*. Journal of Biological Chemistry 289, 18514–18525 (2014).

39. C-terminal amides mark proteins for degradation via SCF/FBXO31 | bioRxiv. https://www.biorxiv.org/content/10.1101/2023.06.29.547030v1.

40. Blondelle, J., Biju, A. & Lange, S. The Role of Cullin-RING Ligases in Striated Muscle Development, Function, and Disease. Int J Mol Sci 21, 7936 (2020).

41. Dzinovic, I. et al. Variant recurrence confirms the existence of a FBXO31‐related spastic‐dystonic cerebral palsy syndrome. Ann Clin Transl Neurol 8, 951–955 (2021).

42. Jin, S. C. et al. Mutations disrupting neuritogenesis genes confer risk for cerebral palsy. Nat Genet 52, 1046–1056 (2020).

43. Hastings, S. et al. A pilot study combining noninvasive spinal neuromodulation and activity-based neurorehabilitation therapy in children with cerebral palsy. Nat Commun 13, 5660 (2022).

44. Mir, A. et al. Truncation of the E3 ubiquitin ligase component FBXO31 causes non-syndromic autosomal recessive intellectual disability in a Pakistani family. Hum Genet 133, 975–984 (2014).

45. Moudi, M. et al. Novel variants underlying autosomal recessive neurodevelopmental disorders with intellectual disability in Iranian consanguineous families. J Clin Lab Anal 36, e24241 (2022).

46. Hermand, D. F-box proteins: more than baits for the SCF? Cell Div 1, 30 (2006).

47. Dawood, S. N. et al. Newly Discovered Cutting-Edge Triple Combination Cystic Fibrosis Therapy: A Systematic Review. Cureus 14, e29359 (2022).

48. Benabid, A. L. Deep brain stimulation for Parkinson’s disease. Current Opinion in Neurobiology 13, 696–706 (2003).

49. Larson, P. S. Deep brain stimulation for movement disorders. Neurotherapeutics 11, 465–474 (2014).

50. Lewis, S. A. et al. Insights From Genetic Studies of Cerebral Palsy. Front Neurol 11, 625428 (2020).

51. Alpay Savasan, Z., Kim, S. K., Oh, K. J. & Graham, S. F. Chapter Four - Advances in cerebral palsy biomarkers. in Advances in Clinical Chemistry (ed. Makowski, G. S.) vol. 100 139–169 (Elsevier, 2021).

In collaboration with

Disclaimer

The results of the PIGN-CDG drug repurposing project that we are sharing in the spirit of open science below are novel preclinical research findings and therefore they do not constitute the practice of medicine. Please consult a physician or clinical care team if considering off-label use of any approved drug or compassionate use of any experimental drug. The same caution applies to nutraceuticals, supplements and “generally recognized as safe” compounds.

The PIGN-CDG Cure Collective gelled two years ago under the leadership of a pioneer family from New Jersey. Susie Lam and Hiran Patel, two British ex-pats residing in the outskirts of New York City, were mobilized by their daughter Alice’s diagnosis. CDG CARE, the umbrella foundation representing the 190+ Congenital Disorders of Glycosylation and recent awardee of the prestigious Rare As One capacity-building grant from the Chan Zuckerberg Initiative, served as a fiscal sponsor for the PIGN-CDG community, enabling highly motivated families to rally for research.

Longtime readers of Cure Odysseys surely recognize the abbreviation CDG by now.

Seasoned by years in finance — one of their secret superpowers — Susie and Hiran efficiently organized a community-driven, “by us, for us” fundraising campaign focused on drug repurposing using yeast avatars. A tried-and-true playbook successfully deployed by Perlara in partnership with a dozen CDG communities. The most recent example is DHDDS.

As CDGs go, PIGN-CDG is more prevalent than some of the rarest subtypes, but still qualifies as ultra-rare. A one-in-a-million genetic lightning strike with the destructive force to forever alter a family, and at the same time, the generative spark for a cure collective. It shouldn’t be up to rare disease parents to make medicines for their sick kids. The harsh reality is that no one else is stepping up to the plate with dollars and devotion for rare genetic diseases that nobody has heard of except the people living with them, day in and day out.

But first a quick glycosylation fresher. PIGN is one of 24 enzymes employed in a buzzing factory that assembles GPI anchors on both sides of the endoplasmic reticulum membrane. You can think of a GPI anchor as the foundation on top of which a prefab house, or in this case a protein, is mounted. The PIGN enzyme is located smack dab in the middle of the assembly line.

You can get caught up to speed on the last project update from a year ago. Right after we completed a test-flight screen of the ~2,000-compound Pharmakon library, a best-of compilation of approved generic drugs and over-the-counter supplements you can order online. Before we dig into the details, let’s take a moment to compare the results of the PIGN Pharmakon screen to five other Pharmakon screens of yeast avatars of inherited metabolic diseases, four of which are fellow CDGs: PIGA, PIGS, PGAP3 and ALG11.

The plot below shows that the average yeast growth spans two logs, meaning some yeast avatars are inherently sicker than others. The SURF1 haploid and diploid screens serve as controls. The average yeast growth of the SURF1 diploid is 2-fold higher than the haploid, as expected given that diploid yeast cells are twice the size of haploid yeast cells.

The three-way comparison between PIGS, PIGA and PIGN is informative. The PIGN and PIGS Pharmakon screen both yielded relatively few rescuers. The PIGS Pharmakon screen identified more sensitizers than the PIGN Pharmakon screen. Overall, there are relatively few rescuers and sensitizers for PIGN. The PIGN yeast avatar is the sickest of the six yeast avatars we screened against the Pharmakon library.

Across the five CDG Pharmakon screens, there are relatively few rescuers compared to the SURF1 Pharmakon screens, indicating that it’s not the severity of the growth defect per se that results in fewer rescuers. The PGAP3 Pharmakon screen is an exception because the growth defect of the PGAP3 yeast avatar is leaky.

Why is the ability to rescue the PIGN yeast avatar low? It could mean that there are no robust compensatory mechanisms available to overcome a state of PIGN deficiency. Or the compensatory mechanisms just have a low ceiling. PIGN appears to be an exception among the five CDGs screened in terms of vulnerability to sensitizers, where PIGA takes the cake with ALG11 in hot pursuit, followed by PIGS. Interestingly, PIGN deficiency is difficult to rescue in yeast but at the same time also difficult to make worse.

Now let’s turn our attention to the PIGN Pharmakon screen results. Remarkably, a clear pharmacological class emerged as the winner of the PIGN Pharmakon screen. Half of the top 20 hits are azole antifungals, including the top four rescuers. Where did we see this same rescuer pattern before? Look no further than PIGS, the last stop on the GPI anchor biosynthesis assembly line.

A direct comparison between the PIGN Pharmakon screen and the PIGS Pharmakon screen shows that the PIGN and PIGS yeast avatars behave similarly, i.e,. they share a large fraction of rescuers and sensitizers. Does that mean perturbations to the second half of the GPI anchor biosynthesis pathway — the stops between PIGN and PIGS — result in a shared pathophysiological cascade that is somehow alleviated by a common rescue mechanism?

Notably, the atypical antipsychotic ziprasidone rescues the PIGN yeast avatar but not the PIGS yeast avatar. Antipsychotics are noteworthy in the context of drug repurposing for CDGs. Going back to Perlara’s first drug repurposing screen for a CDG, we found that aripiprazole was the top rescuer for NGLY1 deficiency. In the MAN1B1-CDG fly screen performed in collaboration with Prof Clement Chow’s lab at the University of Utah, and in the SRD5A3-CDG worm screen, we also saw antipsychotics come up as hits. Curiously, each CDG prefers its own antipsychotic for reasons that remain a polypharmacological mystery.

As luck would have it, we noticed the most clinically actionable hit shared between PIGN and PIGS is an over-the-counter supplement: ascorbyl palmitate, otherwise known by its street name: fat-soluble vitamin C. Ascorbyl palmitate turned out to be the star of the show, but we’ll get to that in a few moments.

Next we compared the PIGN Pharmakon screen to the PIGA Pharmakon screen. PIGA is an interesting comparison because it is the bookend to PIGS, kicking off the GPI anchor biosynthesis assembly line. We don’t see any common rescuers in either direction, except maybe, you guessed it, ascorbyl palmitate (albeit a very weak PIGA rescuer). None of the azoles were common rescuers; in fact a handful of azoles are PIGA sensitizers! The PIGN-selective rescuer ziprasidone does nothing for the PIGA yeast avatar.

The Pharmakon screening campaign taught us that the ability to compare different GPI anchor deficiencies reveals common disease modifying pathways and potential “platform in a pill” drugs, i.e., one medicine that treat a bundle of related rare genetic diseases.

Share

Subscribe now

At the outset of the project, we’d planned to screen the 8,400-compound TargetMol library after a successful Pharmakon pilot screen. Rewind the clock to the summer of 2023, when we generated the results summarized below.

The 8,400 TargetMol library compounds are distributed across 27 384-well plates. The Z score distributions for each of the 27 plates are individually displayed. The positive control (magenta points) is a wildtype (healthy) yeast strain. The negative control (cyan points) is the PIGN yeast avatar treated with placebo (compound-free solvent). Each of the orange points is a compound from the TargetMol library.

The PIGN TargetMol rescuers are weak to moderate, as we saw in the Pharmakon screen. The dearth of sensitizers was also observed in the expanded TargetMol screen. 13 rescuers from the combined TargetMol and Pharmakon screening campaigns were evaluated in dose-response validation experiments a year ago. Three finalists entered the winner’s circle.

In first place is voriconazole (posaconazole showed similar results). The azoles are the top pharmacological class overall and, reassuringly, the top hit from the Pharmakon screen is an azole. At the top dose tested at the 48-hour time point, voriconazole increased yeast growth by a whopping 300%. Voriconazole is approved for pediatric use but requires a prescription for off-label use.

So far everything checks out. In second place is ziprasidone. The kinetics of rescue by ziprasidone are slower than the kinetics of rescue by voriconazole. Could that mean ziprasidone targets a different mechanism of action, or acts further downstream? At the top dose tested at the 48-hour time point, ziprasidone increased yeast growth by 50%. Not too shabby. Ziprasidone (the brand name is Geodon) is also approved for use in children as young as 10 years old.

The third-place finisher is ascorbyl palmitate. Some of you reading this may already be taking ascorbyl palmitate as part of a daily supplement stack. The maximal growth rescue achieved by ascorbyl palmitate is 25%. Doesn’t seem that impressive, especially compared to the magnitude of rescue by voriconazole. But that could be due the stability and bioavailability of ascorbyl palmitate versus voriconazole in yeast media. Or differences in drug uptake into yeast cells among several other possibilities.

We don’t know how the therapeutic effect size in yeast correlates to the therapeutic effect size in humans. There’s really only one way to find out. So the PIGN-CDG Cure Collective advanced ascorbyl palmitate to what we call a 1-to-N trial, with Alice as the first PIGN-CDG ascorbyl palmitate pioneer. Ten other PIGN-CDG kiddos have followed her since.

In parallel, we are testing the top three PIGN yeast hits for rescue of GPI protein expression in PIGN-CDG patient fibroblasts. The PIGN-CDG Cure Collective organized fibroblast collections in the first phase of the project, so we have fibroblasts for almost every PIGN-CDG kiddo, including the ones who are taking ascorbyl palmitate as we speak, some of them for over a year now. I won’t keep you in suspense any longer.

Subscribe now

In consultation with Dr Andrew Edmondson at CHOP, Alice began her N-of-1 trial 14 months ago. IRB and IND approvals are not necessary since anyone with an internet connection and $40 can order ascorbyl palmitate online. What about all the other variables? Bryan Johnson, arguably the N-of-1 messiah, receives the same legitimate criticism.

Parents have their kids on multi-component cocktails, not to mention physical and occupational therapy. Can we really assign therapeutic efficacy to a single intervention? Kids naturally have growth spurts even in a neurodevelopmental disease. How do we know that the ascorbyl palmitate is responsible for the benefits that are described below?

It’s not the job of a 1-to-N trial to answer those questions. The main purpose is to see if parents, the best and most sensitive detectors of change in their children, the humans who know them best, notice any changes one way or another. Ideally a validated clinical outcome measure already exists that objectively captures those changes. Another purpose is dose finding, since the optimal therapeutic dose will likely not be the same for all kiddos, especially for a naturally occurring essential metabolite like vitamin C.

There are two easy ways to mitigate parental bias. First, a group of pioneer patients should synchronize the intervention. Starting with a larger N allows for an estimation of the treatment response rate. We learned from the epalrestat PMM2-CDG experience not to over-index (or for that matter under-index) expectations based on the first pioneer patient’s experience. Second, keep everyone except the core caretakers — ideally just the parents — blinded about the n-of-1 trial. If a physical therapist or teacher who sees the child regularly corroborate what the parents are seeing at home, then the changes are not just wishful thinking.

On top of that we can add quantitative hard measurements. Seizures are common in PIGN-CDG. We didn’t know if seizures or ability to walk would respond to ascorbyl palmitate. Turns out they do!

Here’s a PIGN-CDG pioneer kiddo whom we’ll refer to as Patient 20. This is a before video, which shows that she is unable to take independent steps without active support and physical scaffolding.

Here’s Patient 20 after 9 months on ascorbyl palmitate. The video speaks for itself.

Share

Some call it parental bias. I call it parental insights. Here are PIGN-CDG Cure Collective parents in their owns words describing the improvements they observed in their kiddos while on ascorbyl palmitate — and where available before vs after video capture evidence.

Patient 9:

I have been giving AP to my son for about 5, 6 months. Before starting I stopped all the supplements he was taking, waited about a month for him to clear everything, and did the lab research they recommended on the Drug repurposing project. Vitamin C levels were below normal when I started giving him AP. He was pretty worn out after a month without any supplements. At first it didn't seem to make any difference with or without AP, but I kept giving it. He began to recover after more than a month of taking 500 mg daily. So far, I have only been giving AP at the same dose, and after repeating the tests, the vitamin C levels had gone up. His behavior also changes, his body becomes more stable, he wants to walk and is constantly trying. He started to stand up on the furniture, he understands more, he doesn't get sick often (at least not once during the summer). I can't be sure that this is all due to AP, but I am sharing personal observations. We will be doing new lab research soon.

Patient 11:

Ever since we started ascorbyl palmitate, she didn’t have any seizures. So far, she is on keto and Lamotrigin, but it seemed like something changed a lot. She had an EEG and it looked a lot better than before. It really helps her, not in behavior matters or in changing her moods, but we still think that having no seizures is the greatest thing of all. The doctor said we have to wait a year before we can wean her off the medication or the ketogenic diet. It’s been more than half a year since her last big seizure.

Patient 17:

In our opinion he has made a lot of progress....certainly partly due to the ascorbyl palmitate and partly to a growth spurt, but in our opinion the association of progress is partly due to the vitamin. A before video:

He understands better, if we ask him simple things he understands more quickly, he copies our gestures a lot, he seems to have more muscle strength. Thursday 10 days ago he took his first 2 steps alone and now he walks short distances just giving him one hand. He has started to make some more noises like mamamamama papapapa which I hope will develop into dad and mum soon, or so it seems.

He points much more precisely, he recognizes us in photos and if you ask him where mummy is he points to me. He has added to the yes and no that he used to do with his head, the enough with sign language. Even with occupational therapy he seems better at movements and more precise. If someone shows him a movement he tries to copy, hit a ladle on the pot, build with duplo lego, use a screwdriver. An after video:

When I take him to sleep he does bye bye with hand to say goodnight. He has more sense of attachment towards me, if I wake up in the morning he cries because I go away (to the bathroom). From February to June he ate very little, he had lost weight. Since he took the vitamin his appetite returned and he regained the lost weight. Initially I thought it was related to the pancreas as per Alexa but luckily the enzyme is there, he only has a low pancreatic amylase value. Thinking about it now his appetite has improved a lot since he took ascorbyl palmitate, and in fact he has also added new foods that he didn't eat before like bread and pizza, pears and grapes, which he didn't even touch before.

The thing where he got worse is falling asleep, he has much more energy but he can't fall asleep and he never stays still for more than 5 minutes. Someone told me to give him the vitamin in the morning, since we give it in the morning he gets a little better in the evening but it happens that he is awake until 11pm and at 5am he already wants to get up, then cuddling him he falls asleep again for 1-2 hours. In the afternoon after lunch he is tired but only I can put him to sleep or sometime with other only in the stroller he collapses. He was already a light sleeper before and he struggled before but his hours of sleep have decreased.

Last year we had a polygraphy, he had a lot of sleep apnea.... in my opinion these have improved a little. Maybe it was because it was summer but the wheezing in the lungs has decreased and he also drools less (this could be because we lowered the depakine in May). He has become much more stubborn, if he gets his mind set on something he wants to do it, if he decides he wants to stand in front of the door he throws himself down making himself almost heavy so that we can't lift him and he has started to throw tantrums.

Recent update following an EEG:

No paroxysmal anomalies were observed as before, but he has not reached a stable sleep. For me it is the effect of the ascorbyl palmitate. The neurologist was impressed how he is much more active and cognitively alert. Now we too will remain for the next 6 months with unchanged drugs, in May we will decide whether to start reducing or removing the depakine. So in my opinion it is all related.

Patient 18:

Since starting it I’ve found that she recovers quicker after a seizure (she could eat dinner that night, usually she is in bed for 1 -2 days after seizures). She has had less illnesses (although this is compared to the flu season we went through a while ago). She has better appetite She has been making more sounds! (She is usually very quiet) and this a HUGE thing! Nowhere near talking but you get what I mean. Otherwise gross motor and fine motor is similar not much change. To be honest I don’t know if it’s coincidental or related to the medicine.

Community medicines are created by a community of personally affected families. Imagine a world where community governance and community ownership allows even the smallest rare disease communities to fund research, from concept to cure. That world is coming faster than you think thanks to the efforts and success of rare families like the PIGN-CDG Cure Collective. Onward!

Share Perlara’s Cure Odysseys

In collaboration with

Disclaimer

The results of the FAM177A1 drug repurposing project that we are sharing in the spirit of open science below are novel preclinical research findings and therefore they do not constitute the practice of medicine. Please consult a physician or clinical care team if considering off-label use of any approved drug or compassionate use of any experimental drug. The same caution applies to nutraceuticals, supplements and “generally recognized as safe” compounds.

One day a mighty AGI oracle will effortlessly divine drug repurposing recommendations for the long tail of 10,000+ rare diseases. We are already seeing glimmers of that promised land on the still faint horizon. So what are rare disease families with sick kids supposed to do today?

Especially in cases of rare diseases that are caused by a gene whose function has not yet been fully elucidated at the protein level. Zero-shot learning on data from related genes or symptoms will only take us so far when the long tail of biology is terra incognita and full of uncharted surprises. The same symptom can arise via multiple distinct mechanisms. No knowledge base, no knowledge graphs. No appropriate training data, no ready-made AI solution.

Take the example of the protein encoded by the FAM177A1 gene, where two loss-of-function mutations lead to a complex neurodevelopmental and progressive neurodegenerative disease. When we started working with Jill and Doug Hawkins, founders of the FAM177A1 Research Fund, and parents to two affected children Charlotte and Cooper, the FAM177A1 knowledge base was sparse — and that’s being generous.

A mere seven papers about FAM177A1 had been published in January 2023 when we set out on the drug repurposing leg of the multi-modality FAM177A1 cure odyssey. Only one of those seven papers involved a gene-specific investigation of FAM177A1 function. The rest were casting a wide genome-wide association study net and hauled in FAM177A1 along with other candidate disease-causing genes. Not to mention that the FAM177A1 protein has unstructured domains whose functions are anyone’s guess.

Over the past 18 months, three new attention-worthy FAM177A1-focused papers have emerged, filling key gaps in the FAM177A1 knowledge base. The first paper is a formal description of FAM177A1 deficiency as a rare genetic disease and includes disease modeling in patient fibroblasts and a zebrafish model (Kohler et al., 2024).

The second paper is a cell-biology-oriented study that independently confirmed FAM177A1 protein localization to the Golgi and also showed that FAM177A1 appears to be involved in lipid trafficking based on its association with a known Golgi lipid trafficker, VPS13B (Ugur et al., 2024).

Intriguingly, a third paper centered on a population-wide association study found that higher expression of FAM177A1 protein is associated with protection from a progressive autoimmune disease called primary biliary cholangitis, where lipid trafficking specifically runs amok in the liver (Yang et al., 2023).

In order to keep filling in the fragmentary albeit slowly expanding FAM177A1 knowledge base, we applied some Organic Intelligence, or OI. At Perlara we love all model organisms equally. Choosing between them can be like picking a favorite child. The evolutionary history of the FAM177A1 gene made the selection process easy. Yeast and worms don’t have orthologs (versions) of the FAM177A1 gene, but flies do! So we connected Jill with Professor Clement Chow’s lab to discuss the possibility of running a drug repurposing screen in a fly model of FAM177A1 deficiency.

Two previous fly-powered drug repurposing screens, MAN1B1-CDG and GMPPA-CDG, led to the discovery of at least one clinically actionable hit. Ibuprofen in the case of MAN1B1-CDG, which was trialed in three parallel family-directed N-of-1 observational studies done in close consultation with physicians. In both of those instances, the fly version of the human disease gene had not be studied before. For our purposes, that doesn’t matter. All we need for an unbiased drug repurposing screen to work is a robust rescuable phenotype.

The project with Chow lab got rolling about a month after our introductory call. Initially, the Chow lab generated a series of RNAi knockdown models of FAM177A1 deficiency in flies. A robust “rough eye” phenotype was observed right off the bat for MAN1B1-CDG, so that’s where we looked first.

By April 2023, Chow lab had tried three different FAM177A1 RNAi lines, each line with a different eye driver, and incubated flies at two temperatures. Alas, they found no effect on the fly eye. They even tried incubation at 30˚C or “heat shock” to stress the flies and force a rough eye phenotype, but no dice. So they pivoted to whole-animal survival as a readout.

Case in point: in the GMPPA-CDG screen, we successfully rescued a strong lethality phenotype. Chow lab tested if lethality could be induced by knocking down FAM177A1 ubiquitously, i.e., in all cells of the fly, or only in restricted tissues or cell types such as neuron, glia, muscle or wing.

By June 2023, five months into an unexpectedly protracted model qualification process, we were getting nervous that we might have hit a dead-end. Then one day we were relieved to receive this email from Prof Chow:

“This was the most difficult pre-screen analyses we’ve had to date. But, the good news is, we can do a screen. The most robust, reproducible phenotype is a male-based lethality when we eliminate FAM117A1 expression in the muscles. 100% of all males die before adulthood – it’s very reproducible. There is lethality in the females as well, but it’s not 100% and varies a bit from trial to trial (not ideal). So we can run the screen based on whether we see any living males, which will make it robust.”

It was finally go time.

Share

A representative live wild-type adult fly and two representative dead FAM177A1 knockdown flies are shown in the image below. The drug repurposing screening campaign took flight in the summer of 2023.

A screen of a ~1,400-compound tried-and-true drug repurposing library was completed and results were delivered in January 2024, a year after we made first contact. We screened that same library on MAN1B1 knockdown flies. We also screened that library on GMPPA knockdown flies, where rescue of lethality, i.e., whole-animal survival, was the chosen robust rescuable phenotype. However, instead of whole-body knockdown, which wasn’t as penetrant, muscle-specific knockdown of FAM117A1 in male flies did the trick for some reason.

A prodrug form of riluzole called troriluzole is being developed by Biohaven for several therapeutic areas including spinocerebellar ataxia (SCA) with a positive Phase III trial completed recently completed. SCA shares clinical symptoms with FAM177A1 deficiency.  Troriluzole was designed to improve both bioavailability and safety, allowing a more consistent exposure to the parent compound riluzole.

Using a cutoff of 45% survival, we identified 33 hit compounds, or rescuers of lethality, which are summarized in the table below. Strikingly, half of the hits have pharmacology associated with action on acetylcholine signaling and metabolism. Other notably actionable compounds in the top 20 hits include allopurinol, loperamide and ranitidine, all three being safe and well-tolerated generic drugs. We always keep our eyes peeled for over-the-counter nutraceuticals and supplements. Pyridoxine brought up the rear with 46% rescue of lethality.

A PCA plot for FAM177A1 screen was done to understand how the hits bucket into mechanistic classes. The 33 hits with > 45% rescue of lethality in the primary screen are denoted in red. Almost all are on a vector in PCA space pointing concentrated in the upper right quadrant.

The upper right quadrant corresponds to male and female survival, the outcome we desire. Compounds that lie on undesirable vectors are eliminated from consideration. For example, compounds populating the upper left quadrant are doing the opposite of what we want: precocious lethality at the pupal stage or reduced fecundity.

From the 33 primary screen hits we down-selected to 11 compounds for dose-response hit validation experiments. This post-screening phase began in March 2024 and was wrapped up by May. Three doses were tested: 1 µM (low), 5 µM (medium) and 25 µM (high). Asterisks indicate statistically significant (p < 0.05) treatment effect.

Allopurinol and loperamide rescued at all three doses. Riluzole rescues at the 1 µM and 5 µM doses, but the rescue effect is lost at the highest dose, indicating dose-limiting toxicity at 25 µM. Pyridoxine is less potent with rescue only at the 5 µM dose, but dose-limiting toxicity at 25 µM. Ranitidine rescues at 1 µM and 25 µM, but strangely not at 5 µM. The other compounds either had either no effect (atracurium and clomipramine) or increased lethality at one or more doses (phentolamine, cyproheptadine, ethosuximide, linezolid).

In total, 5/11 primary screen hits validated, including the screen first-place winner riluzole. A 50% hit validation rate is consistent with our historical average.

Why did some rescuer compounds from the primary screen end up making the FAM177A1-knockdown flies worse in the hit validation experiments? Are they false positives, which pop up in any screen? There was only one statistically significant dose that caused increased lethality. Phentolamine and ethosuximide increased lethality at the highest dose (25 µM), which could be attributable to off-target toxicity.

We consider these academic questions; worthwhile to consider but not on the critical path to a clinical candidate. We had in hand the data package to convince us that riluzole was that clinical candidate, but we wanted as airtight a case as possible in flies.

Share

Subscribe now

Three months ago, we received a final report summarizing a battery of phenotypic rescue experiments that were performed focusing on riluzole in the close-out stage of the drug repurposing project with the Chow lab. Male and female flies were assessed in parallel. Significant p-values are displayed.

The original robust rescuable phenotype was muscle-specific-knockdown lethality. In this final round of validation studies, riluzole partially rescued muscle-knockdown lethality at 1 µM and 5 µM in both males and females. In absolute terms, riluzole doubles survival in both males and females in the muscle knockdown regime. There was no effect at 25 µM. These results are reassuring because they corroborate the results of the primary screen and the first round of hit validation. In other words, this is the third set of experiments that demonstrate a statistically significant rescue effect by riluzole.

Riluzole strongly rescues whole-animal knockdown lethality at 1 µM in males but has no effect on whole-animal knockdown lethality in females. In absolute terms, riluzole triples survival in the whole-body knockdown paradigm. What pharmacologists would call dose-limiting toxicity appears to preclude the rescue effect at 5 µM or 25 µM in either sex. It’s also possible that dose-limiting aversion is to blame, meaning the drug simply tastes bad to the flies.

FAM177A1 knockdown in glia causes seizures in a percentage of animals. Seizures are prevented at 1 µM in females, but oddly not in males. There was no effect at the two higher doses (5 µM and 25 µM) in either sex. We have no idea what is behind the male-specific and female-specific sex effects of FAM177A1 knockdown and rescuability. We also don’t know whether those sexually dimorphic phenotypes are a fly phenomenon or evolutionarily conserved in humans.

Seeing is believing, whether it’s a fly or a kiddo. This first video is of glial FAM177A1 knockdown in females. A handful of flies scurry up the wall immediately after shaking and inversion of the tube. But at least five animals are seizing and immobile on the bottom of the tube, eventually recovering and clambering up the wall but after a delay.

Now observe glial FAM177A1 knockdown female flies treated with 1 µM riluzole. None of the flies seized. Within 5 second of shaking and inversion the tube, all animals in the vial were scurrying up the wall or were already at the top.

The totality of the fly data consisted of reproducible and robust rescue of lethality in two sexes and in two knockdown regimes albeit with a dose-limiting ceiling. Riluzole has a maximum tolerated dose in people, too. The optimal dose that rescues lethality is the same dose (1 µM) that rescues a whole-animal brain-dependent phenotype like seizures as well as a functional outcome measure like wall climbing, which is the fly version of the 6-minute walk test. More accurately, the 6-second crawl test.

Share Perlara’s Cure Odysseys

We had our eye on riluzole soon after the commencement of screening. Riluzole emerged as a 95% rescuer early in the drug repurposing screening campaign, but we waiting anxiously and patiently for the entire screen to be completed so we could contextualize riluzole against the backdrop of other hits. Through happenstance, we were able to be introduced to the CEO and CSO of Biohaven and presented the fly data package to them. They were incredibly receptive to the project and have been supportive ever since.

In the first six months of this year, Jill and Doug steadily built up conviction about giving riluzole to Charlotte. But they had to convince a clinician to become a champion too. They carefully diligenced what the starting dose and dosing schedule should be. The conditions finally came together over the summer, helped by the fact that their daughter Charlotte turned 18 and so is technically an adult from a pharmacokinetics and dosing perspective.

Here’s a before (baseline) video of Charlotte.

Here’s Charlotte after a week on the lowest dose of riluzole. Jill’s narration says it all.

Other improvements were noted in what turned out to be an initial honeymoon period. Unfortunately, a few weeks later Charlotte backslid to baseline. So far the riluzole journey has progressed in fits and starts. Immediately after a dose, Charlotte does great. But then the effect fades. After further consultation with her physician, a dose-escalation plan was implemented. Just last week, Charlotte reached the target dose of 200 mg/day. We excited and anxious to see how the next few months go.

Because science never sleeps, we’re collaborating with Unravel Biosciences and utilizing their rareSHIFT drug repurposing services platform. Charlotte will have a nasal swab RNAseq time course over several 48-hour intervals including during her dose escalation on riluzole. Meaning the levels of mRNA extracted from the epithelial cells in her nose will be measured resulting in a genome-wide gene expression profile snapshot. We’re excited for the potential of patient RNAseq as a N-of-1 pharmacodynamic (PD) marker. We might also consider running the SomaScan serum proteomics panel again to make a before riluzole vs after riluzole assessment.

This personalized drug repurposing and biomarker validation study opens the door to expanding riluzole (ideally ultimately troriluzole) to the other four FAM177A1 kiddos. Next on deck for riluzole is Jill’s and Doug’s youngest son and Charlotte’s younger brother, Cooper. But first we need to collect more observational data from Charlotte’s N-of-1, especially as she’s now ramped up to the maximum daily dose.

Given uncertainty about the FAM177A1 mouse model that so far has not yielded obviously actionable phenotypes, the FAM177A1 Research Fund is also hopping on the frog train to take advantage of a frog model of FAM177A1 deficiency. We have now flies, fish and frogs; all housed separately, of course. And fibroblasts and iPSCs at our disposal if necessary. The FAM177A1 disease model base is populating the FAM177A1 knowledge base, which is informing smart drug repurposing and preparing the ground for the modalities that come next.

After all, drug repurposing is a gateway drug: a springboard to targeted and curative therapies that build on the foundational successes of symptomatic treatments.

Share

In collaboration with

Disclaimer

The results of the DHDDS drug repurposing project that we are sharing in the spirit of open science below are novel preclinical research findings and therefore they do not constitute the practice of medicine. Please consult a physician or clinical care team if considering off-label use of any approved drug or compassionate use of any experimental drug. The same caution applies to nutraceuticals, supplements and “generally recognized as safe” compounds.

Just one spark from a bolt of genetic lightning can ignite an inherited metabolic disease that wreaks havoc across metabolism. Down to the unit of the cell and bubbling up as symptoms at the level of tissues and organs. Take the case of the ultra-rare genetic disease DHDDS deficiency.

The DHDDS gene gets its acronym from the enzyme it encodes: dehydrodolichyl diphosphate synthase. It’s responsible for the conversion of the one-to-many metabolite farnesyl to polyprenol, which is the precursor of dolichol. Dolichol serves as the foundation on top of which a glycan chain is elaborated by an assembly line of worker bee enzymes, one sugar brick at a time, as summarized in the diagram below.

No dolichol, no glycoproteins, yes disease. But how exactly?

In the simplest cases of inherited metabolic diseases (IMDs) where just one self-contained subnetwork of the grid is knocked out by mutation, dietary supplementation or restriction of a single metabolite can be curative if implemented in infancy. Think biotin supplementation for biotinidase deficiency, and phenylalanine restriction for phenylketonuria (PKU). Rephrased in the parlance of biochemists, dietary supplementation becomes substrate replacement therapy, and dietary restriction transforms to substrate reduction therapy.

Longtime readers will remember that substrate reduction therapy constitutes the therapeutic basis for treating SRD5A3-CDG with atorvastatin. SRD5A3 happens to be the enzyme two stops down the line from DHDDS. We hypothesize that lowering the flux of the upstream metabolite mevalonate prevents toxic overproduction of polyprenol, which competes head-to-head with dolichol leading to immature glycan formation and hypo-glycosylation. Polyprenol is one of many downstream products generated from mevalonate, whose biosynthesis by the enzyme HMG-CoA is inhibited by statins.

However, the critical difference between SRD5A3 and DHDDS is that human life is incompatible with complete loss of DHDDS, while SRD5A3 has a close cousin called SRD5A2 that partially backfills the loss of SRD5A3. In other words, there must always be some residual DHDDS enzymatic activity in every cell, even if only a modicum of normal levels. Instead of slowing down the tempo, what if the remedy in cases like DHDDS deficiency is to make the enzymes positioned after DHDDS in the assembly line work harder, i.e., substrate enhancement therapy?

A single hobbled enzyme like DHDDS can be a flashpoint for multidirectional disruptions that sweep unevenly across the grid of cellular metabolism, surging flows at some chokepoints while tripping circuit breakers at others; simultaneously causing there to be too little or too much of key metabolites; and, therefore, demanding multi-pronged therapeutic strategies that restore balance to the diseased metabolic network as a whole.

For most IMDs, including DHDDS, the toxic presence of unwanted and normally transient metabolites must certainly be factored into the pathophysiology equation, but so too must the toxic absence of essential metabolites. Especially central hub metabolites that are rate-limiting precursors for numerous downstream metabolic pathways. In the case of DHDDS, that means coping with rock-bottom levels of dolichol.

The simplicity and beauty of unbiased drug repurposing screens of yeast patient avatars mean that we don’t need to know a priori which enzymes act as the key circuit breakers or backup generators when a given node in metabolism is compromised. We don’t need to know a priori whether we need to increase metabolic flux or decrease metabolic flux at those critical nodes in the grid. Fortunately, yeast are here to give us answers.

AI certainly couldn’t know the answer because there is no training data on under-studied or new-to-medicine ultra-rare diseases, a state of affairs describing the majority of the 1,500 inherited metabolic diseases. Per Sam Altman’s recent AGI dawn forecast, for at least the next 1,000 days AI has to play nicely with OI — Organic Intelligence, which is a cute way of saying we need to learn all we can from predictive disease models like yeast, worms and flies. Gradually, reinforcement learning with patient avatar feedback will turn shaky in silico hallucinations into undeniable in vivo validations. But we have to create the knowledge base first!

Share

Cure DHDDS is an Anglo-American alliance between the Dixon family in the UK and the Manville-Gourley family in US. The latter happen to be the founding members of the band Portugal The Man. Proving once again that genetic lightning can strike anyone. Anytime. Anyplace.

When it does it can reveal a hidden superpower in parents who propel themselves into the multi-faceted role of caretaker, advocate, fundraiser, scientist, and, in the fullness of time, biotech founder and CEO. No one else will care as much about a rare disease than those who are personally affected by it. Instead of running away from this hard truth, let’s embrace it and unleash the uncapped potential of patients as the platform.

Over the past two years, Perlara and pioneer IMD families have ironed out a three-phase workflow for yeast-powered drug repurposing screens where hits from the screen can be rapidly translated to kiddos based on the weight of the evidence from a yeast screen combined with a first-principles approach to the practice of medicine. Let’s not forget that we also need objective, quantitative, and data-rich functional outcome measures to replace outdated, pen-and-paper, one-shot assessments like the ataxia rating scale ICARS. I’ll return to this key point at the end of this post.

In the first phase of drug repurposing workflow, growth defects of yeast avatars are verified and shown to be statistically separable from wildtype yeast. In the case of DHDDS, we were fortunate to be able to leverage previously published and experimentally vetted DHDDS yeast avatars.

In the second phase, we screen the 8,400-compound TargetMol library. It has served us and families well. In all eight yeast-powered drug repurposing screens we’ve performed to date, we’ve found at least one clinically actionable hit, almost always an over-the-counter nutraceutical/supplement, starting back in 2018-2019 with PMM2-CDG and the discovery of alpha-hydroxy-cinnamic acid (aka cinnamon), which in turn led to epalrestat.

Examples of parent-led studies we’ve disclosed so far on our Substack include antioxidants for PGAP3, steroidal saponins for AFG3L2, disulfiram for ADSL, and ascorbyl palmitate for PIGS-CDG. For PIGN-CDG, PIGW-CDG, and ALG11-CDG, we have active 1-to-N studies underway with compounds whose identities we will disclose in due time.

In the third phase, we generate a hit list and carry out hit validation experiments such as dose responses, and if necessary to build conviction — as proved to be the case for DHDDS — mechanistic studies that include evaluating a therapeutic hypothesis like substrate enhancement.

It took 12 months to go from project launch to the first N-of-1 parent-led observational study that began over the summer. Some projects move faster and multiple steps can be skipped, for example hit validation and mechanistic studies. We took more time because we made a novel discovery along the way, as I’ll describe. The DHDDS drug repurposing project timeline is shown below.

Subscribe now

We had a little help from some friends in the Grabinska lab at Yale. They created the yeast DHDDS avatars in the first place. I will focus on one DHDDS variant, R205Q. As you can see in the original publication from 2022, the R205Q avatar is extremely sick — it barely grows — and its DHDDS protein barely has detectable enzymatic activity (referred to as cis-PT). We advanced the DHDDS R205Q yeast avatar to a drug repurposing screen. The severe presentation of R205Q creates a reproducible night-and-day growth difference with the wildtype.

We previously shared high-level results just under a year ago.

The plot below shows the distributions of growth values from the DHDDS TargetMol screen. The positive control (aka wildtype) are the magenta points, soaring above the rest. The negative control (aka R205Q avatar plus placebo) are the cyan points way down below. Each of the orange points is one of the 8,400 tested compounds.

I highlight in yellow two metabolic cofactors, NADH and NADPH, as being in the top 10 rescuers of the DHDDS R205Q avatar. Notably, these cofactors are required for dolichol biosynthesis. In hit validation dose-response experiments, NADPH repeated but NADH did not. (Solubility and stability issues in yeast media are the likely culprits).

NADH and NADPH have never come up as a pair of rescuers in any other TargetMol screen we’ve done. We have observed UTP and CTP as rescuers in a few other CDG screens, but never alongside NADH or NADPH. When looking more closely at the dolichol production pathway and the list of top rescuer compounds, we were struck by the enrichment for enzyme cofactors that stimulate activity of the enzymes that are immediately downstream of DHDDS.

It suggested to us that the scarcity of polyprenol forced the enzymes downstream of DHDDS to work extra hard to utilize whatever polyprenol eekes out of the hobbled DHDDS protein. CTP just so happens to be a cofactor for dolichol kinase, the enzyme one step after SRD5A3 that phosphorylates dolichol and primes it to receive sugar building blocks. In the dose-response experiment above, we see impressive rescue of the DHDDS R205Q yeast avatar by CTP. (The high luminescence values suggests that some of the rescue effect is an artifact).

It had been thought that the SRD5A3 enzyme converts polyprenol directly to dolichol. But a recent paper smashed that model to bits, assigning the conversion of polyprenal to dolichal to SRD5A3, and implicating another enzyme called DHRSX as the missing link that bridges polyprenol to dolichol via a pair of side reactions that involve polyprenal and dolichal intermediates.

Remarkably consistent with our DHDDS TargetMol hit list, the newly recognized enzyme DHRSX utilizes both NAD+ and NADPH as cofactors. The former in the reaction that converts polyprenol to polyprenal, and the latter in the reaction that converts dolichal to dolichol. In fact, DHRSX works upstream and downstream of SRD5A3, which itself is a NADPH-dependent enzyme.

Altogether, these data lead us to hypothesize that steady-state NAD+ levels in DHDDS-deficient cells are not sufficient to drive conversion of polyprenol to dolichol. Therefore, the cell needs to boost NAD+ levels in order to drive flux through the dolichol production pathway. This ensures that every last molecule of polyprenol is immediately sopped up by the DHRSX and SRD5A3 tandem bucket brigade, which have to work overtime to compensate for diminished DHDDS activity.

Much to our delight, we observed robust growth rescue of the DHDDS R205Q yeast avatar by NMN, niacin and NMMH (in that order). NR and NAM didn’t rescue but that could be a bioavailability issue in yeast, or yeast vs human NAD+ biosynthesis and salvage pathway evolutionary divergence. The NAD+ precursors we tested in yeast are all available as over-the-counter supplements. NMN is particular is being pursued for multiple disease indications, and is often bound up in conversations (and claims) about healthspan extension and longevity.

Share Perlara’s Cure Odysseys

The proof is in the pudding, or in the iPhone as it were. A major learning from our first N-of-1 studies is that the index (pioneer) patient, i.e., the first N-of-1, may not be representative of the median patient with the same disease. In fact we should even expect an “over-response” that may be amplified by environmental factors like an optimized diet, access to supplemental physical and occupational therapies, and a family’s level of hope and optimism for the future.

We believe the way to mitigate over-indexing on the outcome of a single N-of-1 is to start multiple N-of-1’s in parallel. If the same improvements are observed in three kiddos, it becomes harder to explain away the results as a fluke or an exaggeration of the true responder rate. So Mel’s two children, Rosie and Tom, and Zoe’s daughter Frances, are the three DHDDS NMN pioneers. Over the summer, Rosie’s NMN N-of-1 started first, followed by Frances’s NMN N-of-1 a few weeks later. Rosie’s brother Tom started his NMN N-of-1 several weeks ago.

Working with Dr Eva Morava, the pioneer DHDDS families are exploring the sensitivity and robustness of at-home video capture as a universal functional outcome measure that is completely disease-agnostic. One of the assessments is the “walk the line” test.

Here’s Rosie’s baseline video before she started taking NMN:

And here’s Rosie after 11 weeks on NMN. Do you notice a difference?

In two weeks, the DHDDS NMN pioneers have an appointment with Dr Morava at Mt Sinai in New York City to review the totality of the data and make a concrete plan for next steps.

Let’s conclude by taking a moment to acknowledge the predictive power of yeast avatars and the ocean-parting powers of pioneer families battling a rare disease. That potent combo allowed us to springboard rapidly and responsibly from a compound hit list to a data-driven intervention in the real world. Onward!

Share

Share Perlara’s Cure Odysseys

In collaboration with

Disclaimer

The results of the MEPAN drug repurposing project that we are sharing in the spirit of open science below are novel preclinical research findings and therefore they do not constitute the practice of medicine. Please consult a physician or clinical care team if considering off-label use of any approved drug or compassionate use of any experimental drug. The same caution applies to nutraceuticals, supplements and “generally recognized as safe” compounds.

According to prevailing TradBio tastes, drug repurposing is the unsexiest therapeutic modality and rare diseases are the least desirable — let’s face it, money-making — patient populations. Now try getting Pharma jazzed about drug repurposing for rare diseases. Good luck with that. We all know why rare diseases don’t attract Pharma’s gaze: they’re just numerically too small in an era when global blockbusters for chronic diseases of aging like Ozempic, and its entourage of fellow GLP-1 agonists, have made a comeback. Statins 2.0, you could say.

The double irony is that Pharma doesn’t seem to be aware of its own history. Both statins and GLP-1 agonists were originally developed for other indications, and in the case of statins a smaller indication, before expanding to their current mega blockbuster reach. Statins were originally approved in 1987 for patients living with refractory high LDL levels caused by familial hypercholesterolemia. In 1998, the first statin was approved for primary prevention of coronary heart disease. In 2024, 1 in 4 people in America takes a statin.

GLP-1 agonists were originally approved in 2005 for treatment-resistant type 2 diabetes. Nine years later, the label was expanded for the management of obesity in 2014. Today, it is estimated that 1 in 8 Americans is on a GLP-1 agonist. The industry and media are heralding GLP-1 agonists as the first longevity drugs.

But it’s not just economics that retard the potential of drug repurposing. The principal non-economic reason why drug repurposing is overlooked is the impression that the best outcome patients can hope for is a symptomatic treatment that compensates for disease sequelae without snuffing out the fire at its source. In some cases, that’s true. Yet the same can be lamented of recently approved drugs, especially in cancer and Alzheimer’s, that are disease modifying but not disease ending.

In the positive telling, drug repurposing is about getting a runner on base. First base will do. The expectation is not a walk-off grand slam. Those high hopes are usually exclusively reserved for targeted therapies that fix the genetic lesion in the DNA or restore function to the lesioned protein.

The unexpected results of the MEPAN drug repurposing project demonstrate that drug repurposing is underestimated at our collective peril. Like in almost every other drug repurposing project we’ve run or helped manage — 26+ different rare disease programs over the last 5 years — screens powered by patient avatars or patient fibroblast led to the discovery of at least one compelling clinical candidate, and usually several actionable contenders that share a common mechanism of action.

Just our yeast-powered screens alone have a 90% conversion rate from screen to pioneer, parent-led n-of-1 studies. So any argument that drug repurposing is a desperate fishing expedition can be handily dismantled by the evidence amassed by Perlara alone, let alone countless other drug repurposing success stories made possible by others. Yet one is within their rights to steelman the other side and pose the following pointed questions:

• When treating older children where the disease is entrenched and some damage is irreversible, can we expect a repurposed drug — either as a monotherapy or multiple compounds stacked in combination — to provide meaningful benefit?

• Can a repurposed drug be expected to target the root cause of a disease it wasn’t originally developed for?

• Will a repurposed drug just activate compensatory bypass mechanisms and therefore quickly hit an efficacy ceiling?

• What are the chances that polypharmacological lightning strikes and drug repurposing becomes a gateway to novel drug discovery?

We all agree that if a treatment is started early enough, it can snuff out pathophysiology in the womb and prevent the disease from even manifesting. Take the example of a fetus with cystic fibrosis (CF) that was washed in Vertex’s Trikafta in utero and was born without classic CF symptoms. Several similar case studies have been reported.

Vertex spent years and billions of dollars developing their fleet of targeted CF small molecule designer drugs, starting with Kalydeco (ivacaftor), the inaugural pharmacological chaperone to be approved for CF after the idea was first proposed in the early 2000s. Shown below is a crystal structure of CFTR bound to second-generation and third-generation CFTF correctors, the culmination of thousands of optimization cycles.

Every protein implicated in a rare disease should be able to receive the CFTR treatment. But it can’t take $10,000,000,000+ and over a decade per program. Times have changed since the idea for a pharmacological chaperone was born two decades ago and when GPUs were just powering graphics cards for video games.

Fast-evolving and rapidly improving GenAI will accelerate the development of designer small molecule chaperones, small molecule correctors, small molecule potentiators — small molecule “you-name-its” depending on how it salves the broken protein. What we found by equal part luck and persistance is that a drug repurposing hit could be the seed for GenAI-enabled designer drug discovery.

Share

The pantheon of CF designer drugs proves that in theory we can scale the development of designer pharmacological chaperones and more expansively designer small molecules that bind to a mutant protein and counteract or reverse the effects of a pathogenic mutation on protein function. Self-proclaimed AI-first drug discovery companies like In Silico Medicines and Excistencia have taken AI-designed drug candidates into the clinic but not yet across the finish line.

Chemical space is certainly large enough that a molecule exists for every mutant protein. Vertex resorted to brute force unbiased phenotypic screening, which don’t get me wrong is the tried and true methodology. Drug repurposing can take advantage of unbiased phenotypic screening too, but instead of a library of novel lead-like compounds, the starting point is a library of repurposable compounds. Using yeast avatars of MEPAN, we did just that. Turns out the seed phrase for MEPAN is echinocandin.

Echinocandins are natural products that humanity turned into antifungals. What jumps off the page is that they are cyclic peptides: larger than conventional small molecules, but still dwarfed by biologics like antibodies. As shown in the figure above, an echinocandin molecule is comprised of six amino acids (each rendered in a different color). The R groups can be derivatized as well.

Peptides are miniature proteins. They can fold. They adopt conformational states. The on-target for echinocandins is a glycosyltransferase that is essential for the assembly of the yeast cell wall. As previously described here, the top hits from the yeast-powered MEPAN drug repurposing screen are caspofungin, micafungin and anidulafungin. We were at first struck by the fact that antifungals of wildtype yeast were acting as rescuers of MECR-deficient yeast. Wouldn’t the antifungal effect of echinocandins swamp out any other secondary pharmacology or providential off-targets?

We in fact see that with caspofungin, the first-generation echinocandin and also the dirtiest with the simplified acyl tail, has an upside down U-shaped dose response. The antifungal effect does eventually swamp out the MECR rescue effect, but interestingly in the second-gen micafungin and third-gen anidulafungin, that high dose tox is gone.

Critically, only MECR missense mutants that still create a folded protein are rescued by echinocandins. There was no rescue observed in the MECR knockout, demonstrating that at least a quasi-stable full-length MECR protein is required for the echinocandin rescue effect. The killer experiment will be to measure the binding of echinocandins to different variants of the MECR protein. For now, the most parsimonious explanation is that the echinocandins bind to the MECR protein and are acting as unoptimized MECR pharmacological chaperones.

How could that even be possible? Let’s take a close look at the echinocandin structures and in particular the modifications between them as the chemical series was optimized for antifungal activity and superior pharmacokinetic properties over time. Rezafungin differs from anidulafungin by the addition of a single choline group at the R2 position.

The ultimate test is to apply the echinocandins in a setting where the on-target antifungal target is off the table, i.e., the yeast target of echinocandins does not exist in human cells. So in the absence of a fungal infection, enchinocandins should be inert in the human body. Therefore we tested the echinocandin series on MEPAN patient fibroblasts and focused on mitochondrial function as a readout using Seahorse assays.

Based on OCR (oxygen consumption rate) and ATP production rate, anidulafungin rose to the top of the pile, followed by micafungin in not-too-distant second. With micafungin, the uptick in ATP production rate at the lower doses disappears at the highest dose. Caspofungin was toxic from the get-go, reducing ATP consumption rate at all three doses tested, and it had no effect on OCR.

Rezafungin to our mild surprise did not have any effect on OCR but did seem to induce a bump in ATP production rate. The same rank order of rescue we observe in the MECR yeast avatars is recapitulated in MEPAN patient fibroblasts. Rezafungin and anidulafungin differ by a single substitution that presumably makes rezafungin an even better antifungal but a slightly worse MECR chaperone. Even with just four members of a chemical series, a structure-activity relationship between the echinocandins and pharamcological rescue of MECR is starting to emerge.

Share

We have evidence from yeast. We have evidence from fibroblasts. But no data from an organism establishing rescue of a non-cellular phenotype. In collaboration with Prof. Hugo Bellen’s lab, micafungin and anidulafungin were evaluated on a fly model of MEPAN that they generated. As shown in the figure below, anidulafungin rescues a wall climbing behavioral phenotype in the fly, but micafungin does not.

Altogether, we think the data package for a single-patient IND with anidulafungin is set. Working with the MEPAN Foundation, we’re in discussions with clinicians to see who will step up to the plate.

Looking further into the future, the results of this project open up the possibility of using GenAI and medicinal chemistry to synthesize of a library of anidulafungin derivatives in search of a near neighbor with optimized MECR pharmacological chaperone activity. We’ve already had preliminary discussions with Prof. Scott Lokey who specializes in cyclic peptides. Echinocandins are 6-mer cyclic peptides, so there are 64 million (20^6) possible total combinations to explore.

GenAI can perform an in silico screen to narrow that vast chemical space to 100-1,000 cyclic peptides that can be empirically assessed using reinforcement learning from yeast, fibroblast and fly feedback. We can also try to dial in oral bioavailability so that the we don’t need IV administration, which is how the echinocandins are currently administered to patients.

We still have to cross the threshold of the clinic but after several years of collaboration and research, we’re almost there. Onward!

Share Perlara’s Cure Odysseys

In collaboration with

Rory Belle started showing symptoms very early in life, which concerned her parents. And her mother, Rachel Heilmann, whose background was in clinical pharmacology, knew what she wanted for her child. But knowing and getting are two different things, and Rory Belle lost her life to a genetic variant in the gene NARS1, barely nine months after her diagnosis. 

Like so many parents dealing with rare diseases, the internet was the first place Rachel went, which eventually brought Rachel in contact with others dealing with the same issue — in this case, another mother, Teri Declercq, whose son Grady was diagnosed when he was three years old, following many rounds of testing and re-testing. After months of reaching out to researchers who looked to have some connection with NARS1, Rachel finally heard back from a group at University College London, Henry Houldlen and his post-doc Stephanie Efthymiou, who had collected data on around 10% of NARS1 cases. After Stephanie and Rachel spoke, they decided to start a FaceBook page. “That’s where our story begins,” says Rachel.

What is NARS1?

The NARS1 gene encodes the asparagine aminoacyl tRNA synthetase gene, one of twenty such enzymes that operate in the cytoplasm to charge tRNAs with their cognate amino acids. This being an essential step in protein synthesis, these enzymes (ARSs) are found ubiquitously and highly conserved in nature. Mutations in a few of them have been reported to lead to neurological problems — developmental and intellectual delays, movement difficulties, seizures, being the most common — which vary in severity. Notably, NARS1 stands out among the ARSs with its neurodevelopmental delay and multi-system presentation.

The NARS1 gene codes for a 548 amino acid protein, located in humans on chromosome 18. It has three domains: a catalytic domain (CAT), which charges the tRNA with its cognate amino acid, an anticodon binding domain (ABD) which positions the tRNA at its appropriate place along the mRNA, and a unique domain (UNE-N). At least, two forms of NARS1 disorders exist, toxic gain of function variants which are believed to be de novo mutations that cause progressive peripheral neuropathy, and recessive partial loss of function mutations, which are in most cases heterozygote (bi-allelic) mutations that affect brain development in particular.

What researchers have discovered about the mutations so far are that a number map to two mutational hotspots (arginine 534 and arginine 545) located within the last 40 amino acids of the protein. Mutations close to the beginning of the protein lead to severe phenotypes, and other mutations are scattered throughout the gene, as shown below:  

Where NARS1 disorder differs from other tRNA synthetase mutations is that there is a clear CNS phenotype including severe neurodevelopmental delay in some cases, but also a peripheral neuropathy, which has shown up in all the kids so far, according to Rachel. This puts NARS1 in a similar camp as Charcot-Marie Tooth disease, a common neurological disease that causes peripheral neuropathy.

One life forever changed and another never to be forgotten

Rory Belle’s life and death have changed the course of her mother’s life, even the way she talks about her life. Rachel’s marks time by how long after Rory Belle died some event occurred. Not content with the work she had been doing before Rory Belle came into this world — work that had no connection to her life and her loss — Rachel quit her job in pharma and eventually found herself working with several non-profits, all of which are tackling rare diseases in different combinations. And in 2022, Rachel and Teri founded the Rory Belle Foundation.

In the beginning, most of the positions were held by family members, but the group has grown and diversified, as more families with different backgrounds and expertise become part of the NARS1 family. In the year and a half since its foundation, they have been figuring out the lay of the land — who you need to talk to, where you need to go, when you get money, what you need to do with it, the order of operations. Just trying to sort it all out. 

Share

Each ultra-rare disease family is like a firefly. Fragile and floating through the forest, solitary and in search of solidarity. Invisible to everyone else except fellow firefly families, off in the distance. One teams up with a friend to make two. More and more come together. A speck becomes a spark. Ten thousand sparks shine like a sun.

The metaphor extends beyond families to communities. Ultra-rare disease groups are also fireflies self-organizing in the night quietly, unrelentingly, undeterred by obstacles or darkness. Then the light ignites, illuminating corridors that lead to treatments and cures.

Drug repurposing is the first door that needs to be opened. Over the past two years, Perlara working alongside pioneer families has amassed laboratory and real-world data proving that yeast avatars are therapeutically relevant and predictive disease models for patients with inherited metabolic diseases.

On paper, NARS1 is a near 10/10 fit for disease modeling and drug repurposing using yeast avatars. NARS1 is required for an essential metabolic process that operates at the molecular level nearly identically in cells of the human body and in yeast cells: the assembly of proteins, stitched together one amino acid at a time. If proteins were constructed out of Legos, NARS1 would be one of 20 essential bricks.

Not surprisingly, deleting DED81, the yeast version of human NARS1 gene, is lethal to yeast. The killer experiment, as they say in the lab, demonstrates that the human NARS1 gene complements (can replace) the yeast NARS1 gene while NARS1 constructs containing known pathogenic variants do not complement.

In other words, yeast cells cannot grow if the DED81 gene is deleted, but giving back a healthy copy of the human NARS1 gene restores the ability of DED81-deficient yeast cells to grow. The effect requires a functional NARS1 protein because non-functional copies of NARS1 fail to rescue growth.

We kindly received the ded81-5001 temperature-sensitive mutant yeast strain from the Boone lab. As expected, initial growth characterization experiments showed that the ded81-5001 mutant exhibits exquisite heat sensitivity at 38˚C. In other words, the ded81-5001 mutant is unable to grow at high temperature. Our drug repurposing goal was to find compounds that, like Lazarus, would bring yeast cells on the brink of death back to life.

Next on the yeast drug repurposing workflow checklist: seeding density experiments. The initial growth characterization experiment is a quick-and-dirty spot check that ensures the mutant strain is sick. We always test multiple incubation timepoints, e.gs., 24 hours and 48 hours. We also test several temperatures when working with heat-sensitive mutants. Sometimes we’ve gotten tripped up at this step. In the case of NARS1, the ded81-5001 temperature-sensitive mutant yeast strain performed to expectations.

The seeding density experiments locks in the most critical variables of yeast cell density and the incubation time for the lumiscence-based growth assay. The final pre-screening step is the Z’ optimization experiment. Our target Z’ value is any number greater than 0.5, which in this case we achieved at the lowest concentration of yeast cells (OD 0.0025) and after the shortest incubation time post-cell lysis (10 minutes).

All checks were a go — so we progressed to the TargetMol screen.

Subscribe now

Below is the Z-score distribution of the 8,400-compound TargetMol library screen.

Each of the TargeMol compounds is an orange point. The gray points in the upper segment of the y-axis are the ATP and ADP positive controls, or benchmarks that establish the assay ceiling. The four gray points in the lower segment of the y-axis are artifacts of one of the ATP wells that creates a cross-shaped halo.

The cyan points (negative control) are the ded81-5001 temperature-sensitive mutant yeast strain treated with DMSO, which can be considered a placebo. The negative control wells all performed as expected. The magenta points (positive control) are the wildtype yeast strain. The positive controls on one of the assay plates underperformed. Those points can be disregarded.

Reminiscent of PIGN, PIGA, ALG11, ADSL and DHDDS, the rescue ceiling is low. Meaning the rescuers allow the heat-shocked and extremely sick ded81-5001 mutant strain to grow to numbers that are several fold greater than the negative control (placebo-treated) ded81-5001 mutant strain, but that growth benefit translates to only a small percentage of the total growth achieved by the wildtype strain.

By contrast, the screens we conducted on SURF1, FARS2, AFG3L2, MT-ATP6, PIGW, PGAP3 resulted in rescuers that rescued mutant strain growth to levels approximating, equating, or even exceeding maximal wildtype growth. Why are mutants that are equally sick more or less rescuable? It’s not clear to us today what factors determine the altitude of the rescue ceiling. But we can ask a simpler question: how does NARS1 compare to other inherited metabolic diseases in yeast?

The obvious first comparison is between NARS1 and FARS2, a fellow ARS gene. NARS1 is an asparagine (abbreviated N) tRNA synthetase, and FARS2 is a phenylalanine (abbreviated F) tRNA synthetase. Two different Lego bricks. Adding another wrinkle of complexity, the mitochondria has its own posse of ARS genes, which usually get the number 2 designation, e.g., FARS2.

Even though NARS1 encodes a cytoplasmic tRNA synthethase and FARS2 encodes a mitochondrial tRNA synthetase, we naively could have assumed that the two ARS diseases share more in common than not. Much to our surprise, NARS1 and FARS2 only share a few rescuers — ATP and ADP are positive controls and are expected to set the assay ceiling. Strikingly, the compounds boxed in orange above rescue FARS2 mutant yeast but make the NARS1 mutant yeast more sick! This was an important clue.

Across the set of 16 different yeast screens completed to date, we’ve noticed a peculiar and provocative pattern. Compounds that rescue mitochondrial mutants make congenital disorder of glycosylation (CDG) mutants more sick. Curiously, we don’t observe the converse, i.e., compounds that rescue CDG mutants and make mitochondrial mutants more sick. That got us thinking how NARS1 would stack up against a CDG.

Surprise number two! NARS1 and ALG11-CDG have a lot in common. The magenta box contains shared rescuers and the cyan box contains shared sensitizers. Those boxes are almost empty in the NARS1 vs FARS2 pairwise comparison. The sensitizers in the cyan box are sensitizers in almost every other TargetMol screen of a CDG mutant.

So little is known about NARS1 that nothing should come as a surprise. Yet how do we connect the dots between NARS1 and congenital disorders of glycosylation? Maybe the answer is staring us in the face. Recall that the N in NARS1 stands for asparagine.

Guess where most of the sugars that decorate proteins are attached? The N in N-linked glycosylation, the most common type of protein glycosylation, also stands for asparagine. What if NARS1 is actually more accurately described pathophysiologically as a type of Congenital Disorder of Glycosylation rather than a type of Charcot-Marie Tooth?

Where there’s smoke, there’s fire. NARS1 and most CDGs feature infant-to-child onset of neurodevelopmental delay and intellectual disability with variable seizures and behavioral disturbances, followed by progressive peripheral neuropathy as kiddos age. Remember that NARS1 stands out among the ARSs clan. It’s the first (and most diagnosed) ARS that primarily presents early in life with multi-organ and neurodevelopmental disorder instead of a later onsetting progressive peripheral neuropathy.

A molecular explanation neatly presents itself. NARS1 deficiency would lead a decrease in the incorporation rate of asparagine residues into nascent proteins, meaning newly minted proteins would have fewer attachment points for glycans (sugar chains). Such a model makes a clear prediction: proteins in a NARS1-deficient cells are under-glycosylated, or hypoglycosylated to use the technical term. A concomitant testable prediction is that asparagine levels will be altered in NARS1 patients.

We presented this results to Rachel and Teri from the Rory Belle Foundation this week and we’re discussing how to proceed with hit validation. Excitingly, several of the top rescuers are clinically actionable nutraceuticals/supplements. Onward for NARS1!

Share

Share Perlara’s Cure Odysseys

In collaboration with

Disclaimer

The results of the PIGW-CDG drug repurposing project that we are sharing in the spirit of open science below are novel preclinical research findings and therefore they do not constitute the practice of medicine. Please consult a physician or clinical care team if considering off-label use of any approved drug or compassionate use of any experimental drug. The same caution applies to nutraceuticals, supplements and “generally recognized as safe” compounds.

A scattering of PIGW-CDG patients is known to medicine today, and young Hannes is one of them. His parents, Kelsey and Brian, had to take matters into their own hands because no TradBio white knight was coming to their rescue. That meant sticking their necks out — and stretching their wallets thin — to pay for research that had never been done before. It’s what pioneer families do. We’ve seen it over and over again at Perlara, settling any question about repeatability.

One of the clutch learnings of the last two years of offering yeast-powered “drug repurposing as a service” is that inherited metabolic diseases, especially CDGs, are where yeast patient avatars shine the brightest. To date, we have zig-zagged across the GPI (glycosylphosphatidylinositol) anchor biosynthesis landscape, hopping from enzyme deficiency to enzyme deficiency purely based on which pioneer families show up on our doorstep.

Whether by chance or providence, we started with biosynthetic bookends: the GPI-GnT complex member PIGA, at the start of the supply chain; and the GPI-TA complex member PIGS, which attaches a protein to a GPI anchor that later goes onto post-production editing in the secretory pathway. Following the lead of pioneer families, we also sampled the middle of the GPI supply chain: the PIGN enzyme, which plops the first phosphoethanolamine moiety onto the nascent GPI anchor.

The PIGW enzyme is sandwiched between PIGA and PIGN, where the GPI anchor assembly flips from the cytoplasmic face of the ER to the lumen of the ER. This step in the pathway depends on fatty acid availability because PIGW is an inositol acyltransferase. Its job is to attach a specific type of fatty acid to the inositol ring of the GPI anchor.

The resulting three-tailed GPI anchor intermediate exists while the GPI anchor is built up and the protein is appended. After the protein and GPI anchor are covalently linked, the lipid tails undergo a multi-step removal and remodeling process. PGAP1, the alter ego of PIGW, is an inositol deacylase. It clips the acyl tail that was added by PIGW.

Why does nature attach a third acyl chain to the growing GPI anchor only to later shear it off? Our intuition around the intersection of lipid and sugar supply chains is defined by our experience with PGAP3-CDG, the first instance where we went straight from yeast drug repurposing hits to a parent-driven n-of-1 study in the fall of 2022.

The antioxidant PGAP3-CDG rescuers like alpha-lipoic acid and folic acid that are standard components of mitochondrial cocktails were a clue. In parallel, it just so happened that disease modeling experiments on patient-derived iPSC derived astrocytes were being conducted in Dr Kat Meyer’s lab at Nationwide Children’s Hospital. Remarkably, Lucy’s astrocytes and the PGAP3-deficient yeast avatar were both experiencing a lipid peroxidation crisis aka ferroptosis.

PGAP3 is a phospholipase that removes an acyl chain while PIGW is an acyltransferase that adds an acyl chain. Even though these enzymes catalyze opposing reactions, maybe they both can cause a lipid traffic jam? In the same way a revolving door can get stuck in either the clockwise or the counter-clockwise rotation.

One more item to note about PGAP3-CDG. In the case of PGAP3-CDG, Lucy’s parents are both medical doctors. Another pair of M.D. parents whose son Jake has MAN1B1-CDG were also drug repurposing first movers and perhaps willing to take on more risk than a doctor not personally affected, let alone pioneer parents who are not doctors. Now imagine that one parent was trained as an architect and the other as a software engineer, neither with scientific or medical credentials. Can they still drive clinical research forward?

When a disease is as tiny as PIGW-CDG, the answer is a resounding yes.

Subscribe now

In the previous installment of the PIGW-CDG Cure Odyssey, we were about to perform the TargetMol library screen. The first attempt last August yielded a noisy (unusable) dataset, so we had to repeat the screen. The second bite at the apple a month later proved successful.

Small molecules don’t have personalities in the conventional sense but they are nonetheless colorful. Here’s a representative plate from the TargetMol library. Note that part of the reason we use a luminescence-based readout is so as not to have issues with false positives due to overlapping emission spectra, which affect fluorescence-based and absorbance-based measures of yeast cell growth.

Let’s start the TargetMol screen analysis with the Z-score distribution. For ease of viewing, the y-axis is segmented. The positive control is a wild-type (healthy) yeast strain. The negative control is the temperature-sensitive PIGW yeast avatar.

There are 35 presumptive rescuers with a Z-score greater than 1. The grey circle is ADP, which is a positive control that tells us the screen was performant and had a dynamic range comparble to previous TargetMol screens. That takes us down to 34 presumptive rescuers. One of those is what we refer to as a jackpot: a runaway screening hit that exceeds the positive controls. Like elsewhere in life, it turned out to be too good to be true. We’ll return to Compound 1 and Compound 2 in a moment.

First, let’s zoom in closer on the presumptive rescuers and presumptive sensitizers:

There’s an enrichment of presumptive rescuers in the Z-score range 0.5 to 1 even though the distributions are overlapping. The PIGW temperature-sensitive yeast avatar is on the leakier side, meaning sometimes the mutant grew better than expected. Something about the intrinsic biology makes the growth defect not fully penetrant. Interestingly, the PGAP3-deficient yeast avatar was also leaky.

On the other hand, there’s an obvious and substantial enrichment of presumptive sensitizers with a Z-score lower than -0.5. We never observed any of the negative control wells with a Z-score below -0.5. We observed this skewed ratio of sensitizers to rescuers in PIGA-CDG, and to lesser degrees in ALG11-CDG and PMM2-CDG. However, PIGN-CDG and PIGS-CDG screens did not yield an excess of sensitizers. If anything, the sensitizer-to-rescuer ratio is skewed in the opposite direction. Based on available data, there doesn’t seem to be any rhyme or reason to which way that ratio skews. But as we add more nodes to the map, we’re confident that an explanation will emerge.

In the meantime, we can start to compare and contrast different CDG TargetMol screens. For example, here’s a Z-score scatterplot with PIGW-CDG results on the x-axis and PMM2-CDG result on the y-axis. The only common rescuer is the positive control ADP, which serves as a sanity check that the screens worked. On the other hand, there is a trend toward convergent sensitizers.

Okay, remember presumptive rescuer Compound 1 and presumptive rescuer Compound 2? We focused on these two presumptive PIGW-CDG rescuers because they are both over-the-counter nutraceutical supplements available for purchase online from multiple reputable manufacturers. We first performed 8-point dose response experiments on Compound 1 and Compound 2 individually and measured yeast cell growth after 24 hours and 48 hours of compound treatment.

Compound 1 rescues, albeit with high variance, at the highest two doses at 24 hours, but not at 48 hours. On the other hand, Compound 2 exhibits robust rescue at both 24 hours and 48 hours, and also appears to be more potent than Compound 1.

Next, we performed a Compound 1 + Compound 2 combination experiment. In contrast to ALG11-CDG which has a ceiling to the rescue effect, the absolute rescue effect of the combination treatment increases from 24 hours to 48 hours. The magnitude of the rescue is quite large, suggesting that the primary cellular pathophysiology is being rescued.

Share

Buoyed by those encouraging hit validation experiments using the PIGW-CDG yeast avatar, Kelsey and Brian put the presumptive rescuer hits to the ultimate test: their son Hannes. Using video capture by iPhone as a digital outcome measure, Kelsey performed baseline assessments of gross motor function, to use clinical trial parlance. Here’s a video taken in the lead-up to administration of Compound 1. Hannes is unable to stand on his own and required physical support to maintain an upright posture.

Here’s another baseline video of Hannes a few days before Compound 1 treatment began last fall.

Kelsey supplemented objective videos with real-world accounts of Hannes’ reaction to treatment with Compound 1. Despite being a parent-led n-of-1 study, Kelsey and Brian enforced pseudo-blinding by not telling anyone else in Hannes’ orbit about what was going on. That means other caretakers and family members in Hannes’ life were unaware that he was on any intervention. So, if those blinded individuals independently observed and reported similar changes in Hannes, it would soften concerns that Kelsey and Brian are biased observers.

And I can definitely fill you in on Hannes' progress recently, but our overarching opinion is that for the first couple weeks we didn't see much difference and then in the last 2 weeks we feel like there was a change. He seems to be a lot more actively aware rather than passively aware. The best way to describe it is that it seems like he's processing more quickly and purposefully (like if there's a toy that you put down that he wants, he makes the decision almost instantaneously to pivot and move to it). He's choosing to move to where his brothers are instead of just being excited when they come to him, and he's more purposeful in his movements (like trying to pick up everything on a table in front of him rather than just what's immediately in front of him). Interestingly, I think we're noticing more cognitive changes than physical, but that might just be because he's been on a more consistent physical development path (definitely behind typical development, but continuously developing) whereas the cognitive piece has been harder to see consistent development or jumps in development until now.

Three weeks later, Kelsey emailed this dispatch from the n-of-1 frontlines:

Otherwise, we're continuing to see awareness way beyond anything he was doing before we started the supplement (and I think we may be seeing some improvement in his physical strength now that's faster than what we'd seen before too). For example, I feel like he's playing with his brothers differently which is coming through in terms of his awareness/decision making/social back and forth and the speed he's able to physically do things to engage more with them (for example, his 4 year old brother crawled away from him to try to get him to chase, and he immediately crawled after him, which I've never seen him do before).

Here’s the video receipt:

After several months on Compound 1 alone, Kelsey and Brian added Compound 2 to the stack. Hannes continued to show gains and improvements, as demonstrated by this video of Hannes on a balancing board. Look out, Tony Hawk! You’ve got competition now.

Can we ascribe with 100% certainty Hannes’ progress to Compound 1 and Compound 2 combination treatment? No. We don’t know how Hannes would have progressed over these past 6-9 months if he hadn’t been on any interventions. The only way to find out is to go from N-of-1 to 1-to-N. If a second and third PIGW-CDG pioneer respond similarly to the Compound 1 + Compound 2 combination, the case in favor of a treatment effect grows stronger.

Moves are afoot. Stay tuned for the next installment of the PIGW-CDG Cure Odyssey.

Share Perlara’s Cure Odysseys

In collaboration with

Disclaimer

The results of the ECHS1 drug repurposing project that we are sharing in the spirit of open science below are novel preclinical research findings and therefore they do not constitute the practice of medicine. Please consult a physician or clinical care team if considering off-label use of any approved drug or compassionate use of any experimental drug. The same caution applies to nutraceuticals, supplements and “generally recognized as safe” compounds.

Three years ago, we embarked on a drug repurposing expedition on behalf of a family affected by ECHS1 deficiency, or ECHS1D, an extremely rare autosomal recessive mitochondrial disease (Leigh-like Syndrome) that typically affects children at birth or infancy. Fly and fibroblast patient avatars were indispensable compasses as we navigated an uncharted landscape, converging on a common disease-modifying mechanism.

ECHS1, which stands for Enoyl-CoA Hydratase, Short Chain 1, encodes an evolutionarily conserved enzyme that localizes to the mitochondrial matrix and is involved in branched chain amino acid (e.g., valine) and mitochondrial fatty acid beta-oxidation pathways that are essential for life (Peters et al., 2014; Ferdinandusse et al., 2015).

Common clinical manifestations include encephalopathy, developmental delay, movement disorders, and cardiomyopathy. To date, there have been less than 100 people — mostly children — identified with ECHS1D around the world (Muntean et al., 2022). Like all ultra-rare diseases, ECHS1D suffers from near-zero awareness and visibility, and little reward incentive for investigation by pharmaceutical companies or academic researchers.

Time and time again, pioneer families fill the vacuum. We made disease modeling forays in worms and flies, scientific firsts. We ultimately opted to focus resources on ECHS1D patient fibroblasts as the primary workhorse for drug repurposing screens, which were performed by Charles River Laboratories. ~10,000 compounds were screened across the Prestwick, Selleck and SPECS libraries. Two metabolic challenge assays were deployed in 384-well plates, at two concentrations (1µM and 10µM), in duplicate. A HTS tour d’force.

Fibroblast hits were funneled into a fly model of ECHS1D for in vivo hit validation studies conducted by Dr. Clement Chow’s lab at the University of Utah. As will be recounted below, those distributed and persistent efforts led to the novel and unexpected discovery of p38 MAPK inhibitors as drug repurposing candidates for ECHS1D. There are currently no approved medicines for ECHS1D.

Subscribe now

Perlara’s drug repurposing playbook for inherited metabolic diseases like ECHS1D calls for a portfolio approach and multi-model authentication, i.e., discover compound in model 1, validate compound in model 2. Within a few months of the project launching in the spring of 2021, we booted up three parallel work streams for worms, flies, and fibroblasts.

Evolution dealt a favorable hand for ECHS1D disease modeling, at least on paper. Ancient versions aka orthologs of the ECHS1 gene are encoded in the genomes of animals, including the invertebrates C. elegans (worms) and Drosophila melanogaster (flies):

The worm actually encodes a second ortholog of ECHS1 called ECH-7 in addition to the bona fide ortholog ECH-6. The fly version of the ECHS1 gene has still not been christened in a formal publication, hence the nondescript alphanumeric identifier CG6543. ECHS1D patient-derived fibroblasts are the third model in the mix, in part as a hedge in case the worm and fly models of ECHS1D hit a wall.

Said ECHS1 patient is compound heterozygous for two different missense variants: N59S and T124I. N59 is evolutionarily conserved in all animals. While T124 is a more variable position, the hydrophobic amino acid isoleucine is not observed at this site in the animal kingdom. The N59S and T124I variants are computationally predicted to be damaging – N59S potentially more so – but the exact loss-of-function defect caused by each mutation has not yet been elucidated.

InVivo Biosystems (IVB), experts in CRISPR-edited animals for biomedical discovery, was commissioned to create ECHS1 knockout worm avatars — an ech-6 single knockout mutant and an ech-7 single knockout mutant — as well as a worm avatar expressing the N59S variant in the ECH-6 gene, highlighted in green below.

According to Wormbase, RNAi-mediated knockdown of ECH-6 produces a variety of phenotypes from slow larval growth to defects in foraging behavior. However, a ech-6 mutant expressing patient-derived variants had not been generated or studied to date by any lab. 

IVB conclusively showed that a homozygous ech-6 knockout worm avatar exhibits 100% larval lethality, while the homozygous ech-7 knockout worm avatar is viable but developmentally delayed. We reasoned that the fully penetrant larval lethal phenotype of the ech-6 knockout would be difficult if not impossible for a compound to rescue in the context of a drug repurposing screen. That was our first dead-end.

Disappointingly and somewhat curiously, the ech-6 N59S worm avatar did not exhibit lethality or statistically significant developmental delay (blue-filled circles). The sticking point with the worm model of ECHS1D was the fact that expression of the human ECHS1 gene in the ech-6 knockout worm avatar did not rescue lethality. What’s more, the ech-6 N59S worm avatar behaved unexpectedly upon exposure to excess dietary valine, meaning indistinguishable from the wildtype worm strain (N2).

People living with ECHS1D may benefit from valine-restricted diets (Ganetzky et al., 2016), so we hypothesized that the ech-6 N59S worm avatar might also be intolerant of high valine supplementation. Alas, we observed no change in valine sensitivity, as shown in the above figure. In fact, the ech-6 N59S variant engenders resistance to high valine exposure when expressed in the ech-7 knockout genetic background (purple-shaded bars).

We made one last-ditch effort to salvage the ECHS1D worm modeling project and identify a phenotype amenable to rescue in a high-throughput drug repurposing screen. We treated the ech-6 N59S worm avatar with excess propionate, a short-chain fatty acid that is normally metabolized by ECHS1. As shown below, 120mM propionate supplementation did not affect the growth of the ech-6 N59S worm avatar one bit.

All told, the ECHS1D worm modeling exercise consumed six months. No disrespect to worms, which can be perfectly suitable patient avatars when there’s a biological fit. However, the 100% lethality of the ech-6 knockout worm avatar, the risk of moving forward with the ech-7 knockout given the unexplained behavior of the ech-6 N59S, ech-7 knockout double mutant, and the failure to find sensitizing environmental conditions made a high-throughput primary drug screen with worms too risky. We swiped left on worms and focused our attention on flies and fibroblasts.

Share

Soon after the InVivo Biosystems project was in motion, we commissioned a biotech startup called Vivan Therapeutics to create an ECHS1 RNAi knockdown model in flies. The attempt with flies proved successful: ECHS1 knockdown is larval lethal, consistent with the larval lethality of the ECHS1/ech-6 knockout worm.

In contrast to the worm, knockdown of ECHS1 expression is tunable by temperature in flies using a nifty array of genetic tricks.

At the standard growth temperature (18˚C/64.4˚F), the GAL80 protein sops up the GAL4 protein, inhibiting its function. When flies are reared at a balmy temperature (29˚C/84˚F) the GAL80 protein wilts from the heat, freeing the GAL4 protein to activate expression of a gene-silencing cassette that specifically shuts off ECHS1 expression in all fly tissues. The activity of GAL80, and by extension the level of ECHS1 knockdown, is inversely proportional to the temperature.

Results of a temperature scan experiment to determine the optimal level of lethality are summarized in the table below.

Notice that at 28˚C, there were surviving larvae or pupae. The Goldilocks temperature is wedged between 26˚C and 25˚C. Importantly, the timing of lethality was shown to be during larval stages, not at the embryonic stage. If lethality emerges too early in development it would not be possible to rescue in a drug repurposing screen because larvae have to be viable enough to feed and ingest the test compound dissolved in the fly food.

We faced a fork in the road: fibroblast or fly. As with the worm model of ECHS1D, we were concerned at that time that larval lethality might be too difficult to overcome in a drug repurposing screen. So we held the ECHS1 fly model in reserve for secondary hit validation studies and proceeded with a primary screen in fibroblasts.

Share

In the Fall of 2021, CRL was expanding the ECHS1D patient fibroblast line alongside a healthy age-matched and sex-matched control fibroblast line. Based on the successful execution of a previous drug repurposing project conducted at CRL using a different mitochondrial disease patient-derived fibroblast line, we were reasonably confident that ECHS1 patient fibroblasts could serve as the primary drug screening disease model.

We challenged ECHS1 patient fibroblasts with growth in two different stress conditions, or metabolic challenges: galactose supplemented media and propionate supplemented media.

We knew from a previous drug repurposing project that mitochondrial disease patient-derived fibroblasts grow normally in media supplemented with glucose as the sole carbon source, but they grow poorly in media supplemented with galactose as the sole carbon source. Galactose forces mitochondrial disease patient-derived fibroblasts to utilize their mitochondria to fuel cell growth and division.

Thinking about other ways to metabolically challenge cells, we hypothesized that ECHS1D patient-derived fibroblasts would exhibit hypersensitivity to excess exogenous propionate, which increases the levels of toxic ECHS1 substrates. In other words, ECHS1 patient fibroblasts “fail to thrive” in media supplemented with either galactose or propionate.

CRL performed assay optimization experiments in galactose and propionate challenge media separately because we decided to pursue a two-track drug repurposing screen: a galactose challenge assay and a propionate challenge assay. We reasoned that compounds that rescue both ECHS1 patient fibroblasts in both challenge paradigms would be more likely to have therapeutic potential because they might alleviate both the bioenergetic defect observed in ECHS1 patients (galactose challenge) as well as the buildup of toxic metabolites upstream of ECHS1 (propionate challenge) observed in patient fibroblasts and in patient biosamples such as urine or blood.

In the beginning of 2022 — nine months after we initially set sail — CRL performed dual pilot screens using a small collection of ~2,500 bioactive compounds (the Prestwick and Selleck libraries), followed by a more exhaustive screen of the ~7,500-compound SPECS (aka Broad Repurposing Hub) library. In both instances, test compounds were screened at 1μM and 10μM so that we could stratify hits by potency.

Over 100 hit compounds from the galactose challenge screen and the propionate challenge screen were reordered, solubilized powder stocks as fresh solutions, and then tested in dose-response assays in both challenge paradigms. (We never subjected ECHS1 patient fibroblasts to simultaneous galactose and propionate challenges).

Hit validation studies confirmed p38 MAPK inhibitors, e.g., losmapimod, as the top cluster. p38 MAPK inhibitors were among the only class of primary screening hits that validated in both the galactose challenge condition and the propionate challenge condition. Here’s the potency determination results for losmapimod in galactose challenge.

Share

Next, we evaluated the p38 MAPK inhibitors and several other mechanistically unrelated hit compounds in a fly ECHS1 RNAi knockdown model. It was one thing to observe modest enhancement of growth of fibroblasts by rescue hits in response to two physiologically relevant metabolic stresses. It’s literally another beast entirely to observe rescue effects of a test compound in a whole animal.

The Chow lab created an ECHS1 RNAi knockdown line for hit validation experiments. Just as Vivan Therapeutics observed with their temperature-sensitive ECHS1 RNAi knockdown line, the Chow lab showed that ECHS1 knockdown larvae are developmentally delayed and drastically smaller in size compared to the control.

A summary of the effect of four p38 MAPK inhibitors — losmapimod, talmapimod, ralimetinib and neflamapimod — on larval size is shown below. The top row labeled ECHS1 RNAi is the negative control: we expect no growth in the absence of intervention. Statistically significant rescue of the larval size defect was observed for 25µM losmapimod and for 25µM and 50µM talmapimod.

Remarkably, all of the p38 MAPK inhibitors rescued not only larval developmental delay, i.e., drug-treated larvae were bigger than untreated control larvae, but they also rescued the “sinking” phenotype. Normally, wildtype larvae sink to the bottom of a well if floated on top of a well filled with liquid. ECHS1 RNAi knockdown larvae float instead of sinking but ECHS1 RNAi knockdown fly larvae treated with p38 MAPK inhibitors drop like anchors, though not to the degree of the wildtype fly larvae.

Together, these fly hit validation studies demonstrate that p38 MAPK inhibitors rescue ECHS1 deficiency in vitro in patient-derived fibroblasts and in vivo in ECHS1 RNAi knockdown fly larvae. Going from fibroblast to flies led to this discovery. Would we have arrived at the same place if the directionality was flipped?

Meaning, if we had run a fly screen and a fibroblast screen at the same time, would the two models have converged on the same answer? Turns out the answer is, yes! A group of ECHS1 families crowdfunded a fly screen at Vivan Therapeutics after we completed the fibroblast screening campaign. Lo and behold, Vivan identified losmapimod as the top rescue hit of fly larval lethality.

There is compelling support for the use of p38 MAPK inhibitors in mitochondrial diseases like ECHS1 deficiency. An emerging body of literature from Professor Simon Johnson’s lab (Stokes et al., 2022), Professor Peter McGuire’s lab (Kapnick et al., 2018) and others, suggests that Leigh and Leigh-Like Syndrome mitochondrial diseases have an immune cell component to pathophysiology. p38 MAPK inhibitors were originally developed for autoimmune diseases with the immune cells as the primary target tissue. It seems drug repurposing works in not so mysterious ways.

Losmapimod is currently in clinical development by Fulcrum Therapeutics for adults living with another rare disease called facioscapulohumeral muscular dystrophy (FSHD). Plans are underway for expansion into the pediatric FSHD population. In the meantime, we’re planning mouse efficacy studies to determine the range of effective and well-tolerated doses of p38 MAPK inhibitors in a mammal. While we wish the path to the clinic were more straightforward, the science has to take its course. Onward with our search for a cure to ECHS1D!

Share Perlara’s Cure Odysseys

In collaboration with

Steadfast readers of Cure Odysseys will recognize ALG11-CDG as one of the 12 congenital disorders of glycosylation (CDG) drug repurposing programs in Perlara’s direct-to-patient, family-driven pipeline which we affectionately dub the Anti-Portfolio of Pharma.

NGLY1 deficiency is Perlara’s first CDG program. Aripiprazole (Abilify) was the last hit standing after a three-species — worm, fly, human cell — drug repurposing campaign. Abilify has so far been trialed in three NGLY1 kiddos, including one patient with a durable positive response to treatment. Epalrestat for the treatment of PMM2-CDG is the most advanced CDG program to emerge from Perlara’s pipeline. What began in 2017 as a PerlQuest collaboration is now a standalone drug development company called Maggie’s Pearl, established in August 2020. Maggie’s Pearl is the commercial sponsor of a fully enrolled pivotal Phase 3 trial currently in the open label stage.

Eight of Perlara’s 12 CDG programs moved from screening hits in a model organism to testing in humans. The 1-to-N medicine model we’ve pioneered is on the precipice of scaling. That scale is made possible by an unexpected observation that we didn’t predict at the outset (though in retrospect it’s kind of obvious). In 6 out of 12 CDG programs, at least one clinically actionable nutraceutical was among the top 10 hits, if not the top hit.

By “clinically actionable” we mean available without a prescription — over-the-counter — which could be a generic drug like the NSAID ibuprofen for MAN1B1-CDG or a nutritional supplement like alpha-lipoic acid for PGAP3-CDG. The catchall term “nutraceuticals” runs the gamut from bioavailable forms of naturally occurring metabolites (or metabolic precursors) to bioavailable forms of natural products like plant-based polyphenols.

In the case of ALG11-CDG, we believe we’ve identified what fits the definition of a novel therapeutic: a broad term which in this case refers to a specific and nonobvious combination of two nutraceuticals that could not be predicted by any AI model — at least for now. In fact, it’s the collective “Organic Intelligence” of billions upon billions of yeast cells that narrowed down a library of 10,000 compounds to just two compounds that demonstrate a reproducible rescue effect.

Subscribe now

A year ago we summarized our disease modeling experience working with an initial ALG11-CDG pioneer family. You can get caught up to speed here. We ended the last installment with a cliffhanger: we were about to screen the Pharmakon library. Before we get to the results, a quick refresher on ALG11.

ALG11 is located in a part of the N-glycosylation supply chain that we haven’t yet explored as much as the GPI anchor biosynthesis and dolichol biosynthesis branches. In fact it’s the only ALG program we’ve worked on to date!

ALG11 forms a mannosyltransferase complex with partner proteins ALG1 and ALG2. The nascent glycan chain receives its first mannose additions at this step in the supply chain. We’ve worked on mannose metabolism itself in the form of PMM2-CDG and GMPPA-CDG, which fall under the umbrella of nucleoside sugar synthesis. Our intuition is that the metabolic rebalancing mechanisms at work in other CDGs will also manifest in ALG11-CDG. The question is, which specific ones?

We might expect based on our experience with PGAP3-CDG that a compensatory enzyme can moonlight in the stead of the hobbled enzyme. In this case another phospholipase, maybe one that doesn’t normally prefer PGAP3’s substrate but can be coaxed to substitute for PGAP3 under the right conditions.

The solution may be in the substrate, not the enzyme. If mannosyltransferase enzymatic rates are decreased, does mannose metabolism need to be decreased in lockstep in an attempt by the cell to maintain homeostasis, like we saw in the case of HMG-CoA inhibitors (statins) rescuing disease models of SRD5A3-CDG, where polyprenols accumulate to toxic levels?

One can even imagine a third class of rescue mechanism that is a counterintuitive inverted mirror of a substrate reduction model. A substrate surge approach, if you will, whereby flux of an upstream rate-limiting metabolite is increased to compensate for the reduction of a downstream rate-limiting metabolite. Perhaps that surge is an orthogonal attempt by the cell to restore homeostasis by juicing a moonlighting enzyme or bypass synthesis pathway.

The reality is likely complicated. All three rebalancing mechanisms may contribute to rescue in ALG11-CDG. The only way to find out is to run some screens.

Share Perlara’s Cure Odysseys

Recall that the growth of the ALG11 knockout yeast avatar could be finicky. In the Pharmakon screen, the ALG11 knockout yeast avatar grew better than expected, so the delta between the negative control and the positive control was smaller than the deltas in other Pharmakon screens. Despite the higher baseline growth of the ALG11 knockout yeast avatar, there was room for a few rescuers to separate from the pack. A notably large contingent of sensitizers was observed.

The ratio of rescuers to sensitizers is markedly skewed toward sensitizers. There are only a handful of compounds that make the ALG11 knockout yeast avatar grow better, but many more compounds that make the ALG11 knockout yeast avatar fare worse. In genetics speak, complete loss of ALG11 is difficult to suppress but relatively easy to enhance. When interpreting these findings, keep in mind that patients living with ALG11-CDG have some residual ALG11 enzymatic activity versus none in the ALG11 knockout yeast avatar.

We had similar issues wrangling the PIGA-CDG yeast avatar, whose growth defect was extraordinarily sensitive to temperature. Off by less than one degree centigrade, the growth defect is lost. A comparison to other Pharmakon screens reveals that ALG11-CDG and PIGA-CDG have similarly shaped distributions.

Both show a pronounced skew toward sensitizers relative to the number of rescuers, PIGA-CDG much more so than ALG11-CDG. By contrast, the ratio of rescuers to sensitizers is balanced in the cases of PIGN-CDG and PGAP3-CDG. By contrast, the mitochondrial disease gene SURF1 has an excess of rescuers relative to the number of sensitizers.

Share

Unlike PMM2 and the PIG genes, the ALG11 gene is not essential for yeast growth at standard growth conditions. Lethality is only revealed under high-temperature stress. The ALG11 knockout yeast avatar — the upper-left plate — cannot grow at 37˚C.

We encountered road bumps working with ALG11 yeast avatars in the past. So in seeding density experiments we knocked out the ALG11 gene in two different yeast strains (W303 and BY4741). We also tested growth in two different media. We had to get the conditions just right, otherwise the bake would be off.

Z-score distributions of the 8,400-compound TargetMol library screen are shown below. Reassuringly, a nearly identical distribution of hits was observed in both the Pharmakon and TargetMol screens, which were completed at different times and locations. We also observed the familiar skewed ratio of sensitizers to rescuers.

There was a runaway jackpot hit that grew better than wildtype yeast, the positive control. We were immediately suspicious, as we’ve seen exaggerated screen winners like this before. However, three moderate rescuers, and about a dozen modest rescuers, caught our eye.

We ordered 12 hits for hit validation and potency determination studies. The runaway was indeed too good to be true. It showed absolutely no effect in a dose response assay.

The clear winner with a robust rescue effect is a potential novel therapeutic (Compound 2), which happens to be the overall second-ranking hit in the TargetMol screen once the false positives are removed from consideration. The only other hit that validated has a much more modest rescue effect (Compound 1), which is a plant-based nutraceutical available as an over-the-counter direct-to-consumer product.

Next steps are to pursue mechanistic studies in ALG11-CDG patient fibroblasts for both top hits, and to start a n-of-1 observational study with the plant-based OTC nutraceutical.

Share Perlara’s Cure Odysseys

In collaboration with

There are only three families in the United States who are living with the cruel reality of a genetically confirmed diagnosis of FAM177A1 (FAM) deficiency. There are without question more FAM families out there, stuck in the invisible purgatory of an interminable diagnostic odyssey. Or yet to be born.

If Jill Hawkins, the founder of the FAM177A1 Research Fund, could install giant neon signs emblazoned with her foundation’s website on every street corner in America, she’d turn up more patients to be sure — but not that many more. No matter how you slice it, FAM deficiency is a tiny rare disease. Outside of a handful of personally devastated families, who even knows it exists?

Overcoming the cold start problem is by no means a unique challenge for families affected by FAM deficiency. All rare disease families must ascend the vertical face of a cliff at the outset of a cure odyssey: convincing (cajoling, really) scientists and clinicians and companies to join their cause instead of competing opportunities, while simultaneously asking (begging, actually) a community of friends, relatives and coworkers to step in as funders of last resort for risky, ground-breaking research projects.

In the absence of basic understanding of the disease — when all you have is a blank slate and blank stares — what are cure-focused, parent-led foundations like FAM177A1 Research Fund supposed to do? As previous n-of-1 medical pioneers have proved, the answers lie within. Literally, coursing through the veins of FAM patients and their families. The ultimate disease expert is the disease itself. The most learned scholar is the parent herself.

Share

As previewed in our last FAM dispatch, Jill in collaboration with COMBINEDBrain was able to organize a multi-state mobile phlebotomy roadshow last Fall where blood samples were collected from three FAM-affected families, two affected siblings per family. Healthy controls (parents and unaffected siblings) also volunteered samples. In total, 15 samples — 6 FAM affected plus 9 healthy controls — were shipped to SomaLogic for processing on their SomaScan platform, which precisely measures the levels of thousands of serum proteins.

The samples were processed over the winter. Last month, SomaLogic delivered the data package. Perlara Cure Guide Dr. Arun Ramani performed a preliminary analysis. As a sanity check, Arun examined the levels of the FAM177A1 protein itself. FAM patients are by definition homozygous for loss-of-function mutations, including large deletions which render a nonfunctional truncated protein or no protein at all due to nonsense-mediated decay.

So the expectation was that FAM177A1 protein levels would be significantly reduced in FAM patients compared to controls. This is indeed the case, as shown above. The level of FAM177A1 protein in the sera of six FAM patients is at the lower limit of detection compared robust expression in all control samples. The false discovery rate (FDR) corrected p-value is 0.00031.

Including FAM177A1, there are 12 proteins whose mean expression levels are different in FAM patients relative to healthy controls:

FAM177A1 is the only protein whose expression is lower in FAM patients versus healthy controls. In the 11 other cases of differentially expressed proteins, expression is higher in FAM patients versus healthy controls. However, none of those 11 hits reached statistical significance given the small sample sizes. The eleven proteins are:

• Tumor necrosis factor receptor superfamily member 21 (TNFRSF21)

• Myelin-associated glycoprotein (MAG)

• Netrin receptor UNC5D (UNC5D)

• Matrilin-2 (MATN2)

• Tyrosine-protein kinase transmembrane receptor ROR2 (ROR2)

• Brorin (VWC2)

• Ephrin type-B receptor 2 (EPHB2)

• Neurexin-3-beta (NRXN3)

• Ephrin type-B receptor 4 (EPHB4)

• Serine/threonine-protein kinase DCLK1 (DCLK1)

• Fibromodulin (FMOD)

If any of the proteins on the above hit list ring a bell, please reach out. Perlara Cure Guide Helen Hernandez did some initial detective work keeping in mind that we’re still dealing with trends toward significance and a small sample size of six FAM patients ranging in age from 4 to 18.

The hit list implicates vascular development, inflammation and apoptosis. RNAseq data is publicly available from FAM patient-derived PBMCs and fibroblasts, as well as a FAM deficiency zebrafish model. We’re conducting a preliminary meta-analysis. In parallel, we await hit validation data from the FAM fly repurposing screen. We’re working as diligently as we can to find convergences across datasets and models.

Share Perlara’s Cure Odysseys

In collaboration with

When genetic lightning strikes

Megan Nolan knew early on that something was wrong with her baby, Domenico.  When he was just a few months old, she noticed some disturbing behaviors: not making eye contact, atypical limb movements, and hours-long crying spells that could at times go all night.

During a 14-month diagnostic odyssey spanning two continents, Megan and her husband consulted a dozen doctors, none of whom provided any answers or suggested genetic testing.

When Domenico was 14 months old.  Megan and Kamal, who at that point felt like they were “grasping at straws”, moved forward with double eye surgery for Domenico’s strabismus, thinking that it might be the cause of some of his movement and visual problems.  

After the surgery, perhaps coincidentally, perhaps not, Domenico had his first seizure. This, combined with his medical history, brought a neurologist at CHOP to recommend testing with a panel of genes associated with epilepsy and seizures caused by a variety of causes. Two months later, they had a diagnosis — FOXG1 syndrome (FS), a rare genetic disorder affecting a master regulator of neurological development.

Finally, they had the answer they had been seeking, but it still left them with a severely disabled child and no idea how to manage an increasing number of symptoms or what the future holds for their child and their family.

FUNCTIONS OF FOXG1

What is FOXG1?

FOXG1 stands for fork-head box G1, aka brain factor 1, a member of the fork-head box family of transcription factors.  Transcription factors, considered master regulators of gene expression since they control the expression of groups of genes. are highly conserved proteins, which affect a number of different organ systems.

In the case of FOXG1, the target organ is the brain. The name, fork-head comes from the first FOX gene to be discovered, which happened to be in fruit flies Drosophila melanogaster); mutations in the fly version of FOX causes gnarled fork-like bristles, whence the name fork-head comes.

FOXG1 switches on early in fetal brain development and is essential for the proper formation of a functional brain cortex, interacting with a large number signaling pathways and processes.

It’s no wonder then that FS patients display serious brain abnormalities, such as microcephaly and disruptions of the corpus callosum, which connects the right and left sides of the brain. The children exhibit numerous symptoms, which vary in severity, including intellectual disability, dyskinesis of movement, absence of verbal and language acquisition, irritability as well as some autistic features.

The first heterozygous de novo mutation in FOXG1 was identified in a 7-year-old child in 2005, who had microcephaly, complete disruption of the corpus callosum, and severe cognitive deficits.  As genetic testing has become more commonplace, the numbers of FOXG1 patients has been steadily growing, and may be close to a thousand now.

The FOXG1 gene has 4,890 base pairs, with a single exon (uninterrupted coding region) and multiple binding domains.

Most FS cases are caused by de novo mutations, although there are a few cases in which the parents appear to be mosaics, which means only some of their cells (presumably germ cells, as the trait was passed on to their children) have the mutation. 

FOXG1 mutations, of which there are more than 100 (as of January 2023), come in all varieties — missense, frameshift, deletion and even duplications. All result in a state of FOXG1 haploinsufficiency, or the inability of the body to cope with only 50% FOXG1 gene expression.

Mapping studies have shown that mutations can occur anywhere along the gene and in all binding domains, but there are several hotspots, mainly in areas where runs of single bases occur--- c.256dupC and c.460dupG, which have seven repeated cytosines (CCCCCCC) and guanines (GGGGGGG) respectively.  

These kinds of sequences are prone to replication errors. Studies connecting genotype to phenotype are inconclusive, but the severity of symptoms decreases as you move from the N-terminus to the C-terminus of the protein.

DISTRIBUTION OF MUTATIONS ALONG THE FOXG1 GENE

The awesome power of yeast genetics

FOXG1 Syndrome (FS) has two things going for it. First, FOXG1 is well-studied as rare disease genes go — over 600 publications on PubMed. Second, FS sits at the upper end of the ultra-rare disease distribution, meaning multiple groups have funded a portfolio of research projects over the years, and new families are unfortunately struck by genetic lightning each year. Although yeast models of FS have been studied with support from the FOXG1 Research Foundation, these results are not published and we are not aware of any ongoing or completed FS drug repurposing projects using yeast.

Yeast models with gene defects often exhibit growth defects. Slow growth can be
rescued in a pathway-agnostic high-throughput drug repurposing screen. The yeast gene HCM1 is the ortholog (equivalent) of human FOXG1. HCM1 performs the function of many FOX family members in humans. This is a fairly typical situation — humans have about five times as many genes as yeast, and many of these new additions arose through gene duplication events over the course of evolutionary history.

Understanding that one can talk themselves out of almost any experiment that hasn’t been tried before, we set out to rescue growth defects of HCM1 mutant yeast. In our experience, in cases where yeast have a single ancestral gene that diverged into an entire family of genes in humans, screening using yeast lacking that ancestral gene has produced biologically-relevant results. This fact, combined with the fact that yeast screens are expeditious and cost-effective, led us to recommend this yeast-driven approach.

We conducted growth assays including both haploid and diploid FOXG1 (yeast HCM1) deletion strains, along with their respective controls, under two temperature conditions: 30°C (ambient) and 37°C (heat shock). We observed a significant growth impairment in the diploid HCM1 knockout strain at 30°C.

After several failed attempts at further optimization, we pivoted away from using high temperature as a sensitizer to using a pharmacological sensitizer. Based on FOXG1 biology in yeast and human cells, we reasoned that HMG1 mutant yeast would be sensitized by the microtubule stabilizer benomyl.

Here’s the initial seeding density experiment using benomyl sensitization to reveal a growth defect.

Next, our yeast team completed a dose-finding Z’ optimization experiment at BADASS Labs to determine which concentration of benomyl to advance to the screening stage. That Goldilocks dose turned out to be 15µg/mL.

Here’s the Z’ optimization experiment performed at HTSF using the 15µg/mL dose.

The TargetMol screen was just completed today, auspiciously on the eve of Rare Disease Day 2024. More details to come soon.

Rare Uncommon Disease Day

In thinking about all the other families — potentially millions of them — who find themselves in the same situation, as she did, Megan decided to use the skills that she had developed in media and advertising, to fill in these gaps of knowledge and to create support systems, systems that were non-existent when Domenico was diagnosed.

Thus beget The Children's Rare Disorders Fund (CRDF), an ultra-lean non-profit biotech, founded by Megan and her husband, Kamal Sandu, in July, 2022, just three months after their son was diagnosed. The CRDF’s goal is to support research into therapies for rare diseases. While inspired by Megan and Kamal’s son, the foundation is exploring a variety of modalities that could be applied to any or all underserved rare disease. 

Among the modalities is, of course, drug repurposing. Perlara’s work on FOXG1 syndrome is the first project to receive funding from CRDF, which came into existence in July 2022, just three months after Domenico’s diagnosis.

More recently, Megan launched a journal, Rare Parenting, a treasure trove of information for families in need of guidance and information, what Megan calls "service journalism." There you can find articles that provide step-by-step guidance on what parents of undiagnosed children can do, what kind of symptoms might be a sign of neurological disorders, financial planning to mention a few.

Subscribe now

Share Perlara’s Cure Odysseys

In collaboration with

Perlara rebooted as a remote first, fully distributed biotech company in Q4 2021. The first pioneer family to engage us as scientific consultants — Cure Guides — asked us to develop a personalized cure roadmap for their child living with an ultra-rare form of Ehlers-Danlos Syndrome (EDS) caused by deficiency of the enzyme PLOD1, which is required for proper collagen production by the body.

In that roadmap, we proposed a cell-based drug repurposing project as our main objective. We thought the project would take at most two years to complete. We enter the third year of this cure odyssey guided by the same north star, and lugging a suitcase full of accumulated scientific and operational learnings. Most importantly, we have a line of sight on what appears to be a robust assay that would finally enable us to consummate a high-throughput drug repurposing screen.

There’s still no guarantee we will pull it off, but we have confidence in our plan. Perlara Cure Guides L Handy and Dr Shiri Zakin have been working with Dr Yoshi Ishikawa at UCSF over the past year to develop a cell-based assay for drug repurposing for PLOD1-kEDS. Last summer, L completed proof-of-concept of experiments that green lit advancing to where we are today.

Rewinding the clock back to 2022, the first major roadblock was not being able to visualize a statistically discernible defect in collagen fiber structure inside patient fibroblasts cultured in the lab. Those cell imaging studies are summarized here. Last year, we pivoted to a biochemical strategy: measuring levels of extracellular collagen secreted by PLOD1-deficient cells. PLOD1 adds hydroxyl groups to lysines in the collagen protein, encouraging cross-linking between collagen strands that can be detected by antibodies in a conventional ELISA assay.

First, however, we had to establish the specificity and fidelity of our PLOD1-kEDS research toolkit: PLOD1-deficient mouse and human cell lines; antibodies specific to different types of collagen. The launch phase of our collaboration with Yoshi is described here for the collagen aficionados in the audience.

Preliminary results were encouraging but we wanted to run additional controls to make sure we were interpreting the results correctly. Yoshi tested two different antibodies specific for lysyl hydroxylase 1 (LH1), the protein encoded by the PLOD1 gene. No LH1 protein was detected by immunoblot in two different PLOD1-kEDS patient fibroblast lines obtained from the nonprofit biobank Coriell: a line derived from an affected infant and an older child’s line.

Next, Yoshi biochemically surveyed the maturation of collagen fibers by immunoblotting with a panel of antibodies. Type 1 and Type 5 collagen are significantly reduced in the two PLOD1-kEDS patient lines compared to the two control lines. On the other hand, Type III collagen appears to be spared or modestly decreased in the two PLOD1-kEDS patient lines.

Subscribe now

The two LH1-specific antibodies tested both target an epitope close to the C-terminus of the protein. We would expect to see the same result using an LH1-specific antibody that binds to a stretch of amino acids near the N-terminus (opposite end) of the protein. Indeed, the N-terminal-specific anti-LH1 antibody failed to detect any LH1 protein.

This cure odysseys has taken unexpected twists and turns, much like an errant collagen fiber. The story made a sharp and sudden 90-degree turn when Yoshi measured LH1 protein levels by immunoblot in the PLOD1-kEDS family trio. LH1 levels are reduced in mother (“M”), father (“F”) and affected child (“P”), but notice how the paternal and maternal banding pattern is different.

The original genetic testing report concluded that the affected child is homozygous for a nonsense variant in the fifth exon of the PLOD1 gene. But Mom and Dad do not carry the same PLOD1 genotype according to LH1 biochemistry!

Let this serve as a lesson in edge cases, which are more common than we realize. Clinical geneticists often will insist on having a DNA sequencing test result and a corroborating protein activity test result that mutually agree. A possible genetic explanation for the disagreement we observe is that the father actually carries a large deletion that is invisible to old-school PCR-based genetic tests.

The large deletion hypothesis was suggested in a follow-up genetic report generated by the Collagen Diagnostic Laboratory (CDL). The report acknowledges that there could be a multi-exon deletion that would not be detected by standard gene sequencing.

We’re sorting out the exact genotype of the PLOD1 locus to confirm the presence of a paternal multi-exon deletion variant.

Share

After ruling out the feasibility of a LH1 enzymatic activity assay, we narrowed on the final option: a biochemical assay to measure extracellular collagen secreted by PLOD1-kEDS patient fibroblast lines grown in a multi-well-plate format amenable to high-throughput screening. This schematic summarizes our proposed workflow.

Before using the precious patient samples, we optimized the hydroxylysine assay using purified collagen from mouse fibroblasts. Right off the bat, we ran into assay development challenges. We quickly learned that the first hydroxylysine antibody we chose was incompatible with an ELISA use case. We found a new antibody that binds to human, mouse and bovine Type I collagen. Problem solved. L ran the first successful proof-of-concept experiment right before the holidays.

The binding of non-hydroxylated collagen purified from the LH1 knockout mouse cell is reduced by 50% compared to the control collagen purified from a wildtype mouse cell line. We finally found an assay that we could deploy in a drug repurposing screen. We also validated a functional cell-based assay for differentiation between hydroxylated and non-hydroxylated collagen that anyone of the EDS space can run with beyond our immediate focus on drug repurposing.

Yoshi, L and Shiri have come up with a scope of work that would take us through the rest of assay development. To minimize unnecessary depletion of patient fibroblasts, Yoshi is creating a LH1 knockout human cell line for the final stages of assay development and possibly for use in a drug repurposing screen itself if we elect to save patient fibroblasts exclusively for hit validation studies. More to share soon!

Share Perlara’s Cure Odysseys

There’s a concept in investing called the anti-portfolio, the ones that got away. An anti-portfolio is a counterfactual fund comprised of all the passed-on investments that later became fabled unicorns — much to the chagrin of the investor who talked themselves out of the deal or simply didn’t grok the founder’s vision. Perhaps the most recognized example of a public anti-portfolio is courtesy of Bessemer Venture Partners.

The concept of anti-portfolio is not much discussed in TradBio (traditional biotech) circles. Perhaps because the 90%+ failure rate of clinical trials has become so normalized. 2023 is being heralded as a bumper crop year with 55 FDA drug approvals. While that’s a healthy uptick from the 37 approvals in 2022, the number of approvals has wobbled around the 50 mark for the past seven years.

55 new drug approvals is not an annual high-water mark. 2018 boasted 57 approvals. In 1996, there were 53 approvals. However, famed drug hunter Dr John Lamattina, who oversaw global R&D for years at Pfizer, noted in a January 2015 Forbes commentary that 1996 was a fluke year in which FDA processed a backlog of approvals thanks to staff hires at the agency made possible by commercial sponsor user fees that had been enacted by Congress. As a sign that the rate of drug approvals is trending up despite a momentary COVID dip in 2022, Dr Lamattina’s piece was about how the year 2014 and its 41 new drug approvals spanning multiple modalities (not just small molecules, mind you) offered a glimpse into what felt like an imminent acceleration. 2015 went on to bag 45 new drug approvals.

Was the Pharma rocket ship about to take off? The very next year in 2016 there were only 22 approvals. Looking back over a 30-year window, the number of FDA drug approvals per year has ranged between 17 to 57. Of course many breathtakingly efficacious medicines have been approved in the modern era, including the first CRISPR medicine approved for sickle cell anemia a few weeks ago. There’s no question that the quality and diversity of new medicines is increasing unrelentingly, and dramatically.

But the output — the total number of drugs approved by FDA in a year — has stayed basically flat, though admittedly the last half-decade is averaging more drug approvals than the preceding five years. Medicine is not accelerating fast enough. The industry has come to accept process failure and capital inefficiency. $1 billion. 10 years. 99 busts. One success. Not good enough.

Inspired by the trajectory of SpaceX and the success rate of rocket launches over time, what will need to happen to 10X annual new drug approvals, from 50 to 500? What will it take for the 90% clinical trial failure rate to flip to a 90% clinical trial success rate? Remember SpaceX circa 2009 when every launch countdown concluded either in a scrubbed mission or a viral video of a half-exploded rocket?

Check out how SpaceX is doing today:

Medicine feels like it’s stuck where SpaceX was in 2009. Blowup after blowup. If your first reaction is to pin the blame on FDA or over-regulation, wrong answers. Or that Janssen and his contemporary drug gatherers picked the pharmaceutical tree clean of all low hanging fruit. Also, wrong answer.

The industry is not learning fast enough from its mistakes; not spending each dollar like it could be its last; not making ruthlessly dispassionate decisions at every step of the process based on hard data; and with the exception of COVID, not executing fast enough like our lives, and the lives of those we love, depend on it. After guiding pioneer families on their cure odysseys for almost a decade, we think we’ve found one path toward biomedical acceleration, or b/acc for short.

Subscribe now

Anyone who has operated in the rare disease space over the last decade is all too familiar with the whiplash business cycle. 10 years ago, rare diseases were the talk of the town, the belles of the ball. Big Pharma companies launched rare disease divisions. The rare FOMO was real. At the peak of the 2020-2021 biotech bubble, a new preclinical platform company bragging about dozens of programs, even ultra-rare diseases with no actual viable commercial paths, was going public weekly.

Today, rare diseases are more like the unwanted stepchild from a previous marriage. Rare disease families are left to ask their friends, colleagues and neighbors: hey, brother can you spare a dime. Most of those biotech companies that IPO’d in the boom years are either dead, dying or have focused on one or two lead programs at the expense of their once mighty preclinical pipelines. Companies large and small are still shedding orphan assets — and shattering hopes.

In response to that extinction event — the industry wide de-prioritization of ultra-rare disease programs — Perlara coevolved alongside pioneer families into the first direct-to-consumer — more accurately, direct-to-family — biotech company. Genetic lightning keeps striking. Pioneer families emerge from the fog. Organically, just following the lead of those families, disease indications came into focus until they coalesced into a pipeline.

Instead of a modality or asset chasing a disease as is the norm in TradBio, Perlara empowers families to chase after their cures, to borrow drug repurposing champion Dr David Fajgenbaum’s expression. Perlara’s key insight was that rare disease communities are fragile and fleeting on their own but when linked together they form a battering ram of precision medicine that can break into the clinic faster and with high probability of success.

Bootstrapping the company since 2021 led to revenue-positive business development. Instead of spending time and burning runway trying to secure TradBio aka B2B partnerships, Perlara pivoted hard to a B2C “drug repurposing as a service” for pioneer families. We had to reimagine a distributed work flow involving fractional team members, month-to-month subscription access to a turnkey lab, as well as a hybrid yeast team with remote-first project managers and on-site staff scientists.

Perlara 1.0 (2014-2018) was centralized and operated by full-time employees. Perlara 2.0 (2019-present) is unbundled, a network of fractional consultants. We popped up a lab and focused internal resources on yeast. We outsourced worm, fly and fibroblast to academic partner labs, CROs or platform startups, and offered a consulting layer of project management, data analysis and collaboration choreography on top. We had finally found product-market fit.

Perlara did $190,000 ARR in 2021. In 2022, we did $1,000,000 ARR, an increase of 5X. We’re still waiting on the audited numbers from the last month of the year, but 2023 revenues will be ~$1,350,000, which is 35% year-over-year growth. We haven’t paid a cent in marketing or sales. Pioneer families find us online via word of mouth, social media, earned media and Substack.

As a result of being totally modality agnostic, we’ve been able to build a business working on any overlooked disease, even the tiniest ultra-rare diseases where there are only 6 known patients in the United States, such as FAM177A1 deficiency. We are truly scale-free, or 1-to-N as we like to call it. As a result of focusing on (for some reason still) contrarian but more predictive disease models and the rare diseases to which they’re most well suited, we have been able to create commercial optionality and upside potential on a substrate of overlooked assets, even nutraceuticals and decades-old generics, by creating proprietary non-obvious combination therapies.

As a result of liberating ourselves from the shackles of a B2B2C business model that otherwise dominates the current system, we can partner with overlooked founders, parents of rare disease kiddos who are expected to make way for the professionals at some point. The result is the Anti-Portfolio of Pharma.

Share

We tripled down on overlooked diseases, overlooked assets, and overlooked founders. Rivaling some of the biggest companies and most ambitious platforms in the industry, we’ve assembled what we affectionately call the Anti-Portfolio of Pharma: 22 programs, all made possible by pioneer families taking control of their medical destiny. Nine of the 22 programs are clinical stage, meaning trialed in at least one pioneer kiddo. Several n-of-1 studies have progressed to the n-of-few stage.

We’ve successfully run yeast-powered screens, worm-powered screens, fly-powered screens and cell-powered screens. Remarkably, a nutraceutical was the top hit — or in the top 3 — in half of all programs. The accessibility and safety profile of nutraceuticals only lowers the barrier to parent-led, bottom-up, single-patient observational studies, as we described last month for a n-of-3 study of ibuprofen for MAN1B1-CDG.

The bet is already paying off with Maggie’s Pearl, though we still have to get across the NDA finish line. In less than five years and for less than $5 million dollars, we have taken our flagship program for PMM2-CDG from a yeast hit to a fully enrolled pivotal Phase III study (NCT04925960). We believe there are many more baby biotechs like Maggie’s Pearl yet to be born.

Our Anti-Portfolio is clearly enriched in inherited metabolic diseases. Congenital disorders of glycosylation, or CDGs, will be familiar to longtime subscribers of Cure Odysseys. Mitochondrial disease known collectively as Leigh or Leigh-like Syndrome are the next biggest group. To be clear we are truly disease agnostic, trekking to the far-flung corners of biology led by the unstoppable directive of pioneer families.

For example, a potassium channelopathy caused be a de novo variant in KCNC1. We’ve been cure guiding Stephanie Telesca and the KCNC1 Foundation for almost two years. We found KCNC1 drug repurposing candidates including a nutraceutical that we will take into the clinic this year. Our DTC approach serendipitously reveals diseases and genes that no one’s ever heard of because they’re so new to medicine. Even then we can ramp from 0 to 1. For example, the aforementioned FAM177A1 deficiency. The top hit from a just concluding fly-powered drug repurposing screen is a repositionable drug that is already in clinical trial for another indication. We’re in active discussions with the commercial sponsor about launching two single-patient INDs.

In the last year, we’ve focused on the potential of yeast itself as a drug repurposing platform. We’ve completed 12 yeast-powered screens, averaging one screen per month in 2023. We have translated directly from yeast to human in nine programs, leveraging the fact that 50% of hits are over-the-counter nutraceuticals or over-the-counter generics.

Despite the progress we’ve made, we’re still not moving fast enough. Instead of a reactive and bespoke approach where we wait for genetic lightning to strike families with the means and determination to pursue drug repurposing on their own one gene at a time, in 2024 we will shift to a proactive and standardized approach that takes advantage of pooling many yeast disease models and performing a single growth assessment of compound treatment on all yeast disease models at the same time.

Based on Perlara’s cumulative experiences over the past decade, it is time to industrialize yeast as a scalable and cost-effective drug repurposing platform for inherited metabolic diseases, which comprise approximately 15% of the 10,000+ rare diseases. Just focusing on a subset of inherited metabolic diseases, there’s an opportunity to ~20X our current pipeline from 12 yeast-powered programs to ~250 yeast-powered programs. And we’re building with others a distributed (90% at-home), sensor-driven, device-powered clinical trial infrastructure to keep up with the pace of clinical candidates flowing down the funnel.

Metabolism is so well conserved between yeast and humans that yeast have been used to delineate interconnectivity between different metabolic pathways, identify the roles of various metabolites in the pathway, and study perturbations in the context of diseased cellular states. While the use of yeast as disease models or the automation involved in pooled drug screens is not especially novel, Perlara’s innovation is the discovery that yeast are sufficiently faithful models of inherited metabolic diseases to be used as patient avatars and the application of these models at scale. The answer has been right under our noses the entire time.

The 22 programs comprising Perlara’s B2C/DTC drug repurposing pipeline are just the beginning of the latent biomedical acceleration that’s waiting to be unlocked. The 1-to-N clinical trial operational playbooks developed in the effort to scale drug repurposing will be generalizable to any intervention and all modalities, unlocking second-order exponentials.

Here's to a pipeline that we hope will launch a thousand cures!

Share

Share Perlara’s Cure Odysseys

In collaboration with

Two families, two continents, two medical systems — one, the United Kingdom’s presumably universal National Health Service, the other, the United States’s anything but universal network of public and private payers. Nonetheless the answer the two families, desperate to find what was causing their children’s symptoms, got was the same:

We have nothing, sorry. Pharma doesn’t care about your disease so you’re on your own. We may have more information in 5-10 years. In the meantime, find funding for research. Good luck with that.

And so they did, like so many pioneer families are compelled to do. Cure DHDDS Foundation was set up earlier this year in the UK by Mel and Charles Dixon, whose two children — 14 year old Tom and 8 year old Rosie, pictured below — were found to have variants in the gene, dehydrodolichyl diphosphate synthase, abbreviated DHDDS, which affects protein glycosylation. The Dixons were profiled by ABC News over the summer, making them a rare breed of family that receives earned media.

Joining the Dixons from the US are John Gurley and Zoe Manville aka the band Portugal. The Man, no strangers to the spotlight. Their 12-year-old daughter Frances also has a heterozygous variant — in her case, p.Arg205Gln — in one copy of her DHDDS gene.

To date, only 70 people in the world have been diagnosed with DHDDS deficiency. Internet searches and cold calling scientists, clinicians, anyone who had published anything on DHDDS, brought the two families together and ultimately to Perlara.

These two families have been on a parallel journey, which will be eerily familiar to anyone who has dealt with an ultra-rare disease. The Dixon’s then four year old son, who had developed a tremor, received various diagnoses, from ataxia to autism, before whole genome sequencing identified his heterozygous pathogenic DHDDS variant affecting a different arginine residue, p.Arg37His. (An earlier more selective genetic panel had come back negative).

Frances began exhibiting symptoms during lockdown, when she was nine, myoclonic seizures that cause brief muscle movement and fluttering, that Zoe and John might not even have noticed had they all not been home together 24/7. After months of waiting just to get appointments because of the pandemic, they too received the devastating news that Frances has a rare genetic disease, about which little is known and for which there is no therapy. Both families have been juggling various over-the-counter regimens to alleviate some of the more pernicious symptoms, with no idea on the likely course of the disease, and what the future holds for these children.

DHDDS joins a large list of genes that affect glycosylation. There are well over a hundred such inherited metabolic diseases, collectively known as congenital disorders of glycosylation (CDGs) that have a wide range of health problems from strokes and seizures to development and intellectual delays. DHDDS is not Perlara’s first rodeo dealing with this complicated and essential set of pathways; Perlara is shepherding a dozen CDGs families through drug repurposing protocols, a number of which have advanced to various stages of 1-to-N clinical testing, including a fully enrolled pivotal Phase 3 study with 39 participants.

Tom, Rosie and Frances on the surface appear more mild compared to some in the group with the worst manifestations of mutations in DHDDS, which can include ataxia, seizures, severe cognitive impairment, and even premature death. According to her mom, Frances has had seizures and ataxia, though “they are not intractable yet.” Zoe clarified in an email recently: “I suppose they were spared the severe things early on that some endure.”

Being new on the scene — the first DHDDS variant was described in the literature in 2011 — not much is known about the mechanism or how individual variants produce particular symptoms and phenotypes.  No obvious connection between genotype and phenotype has yet to emerge. But it’s early days.

Subscribe now

What is known about DHDDS is that it encodes the catalytic subunit of the enzyme cis-prenyltransferase (cis-PTase) which performs the rate limiting step in the biosynthesis of dolichol, a glycosyl carrier lipid (depicted as the yellow rectangle below) required for the biosynthesis of several classes of glycoproteins and glycolipids.

The clinical presentation caused by DHDDS variants are not your typical CDGs, however. For one thing, under or hypo-glycosylated proteins, a hallmark of CDGs, have not been detected in the serum of people with DHDDS variants. Rather, changes in lipid profiles have been seen, which puts DHDDS somewhere between a CDG and a lysosomal storage disease, another class of inherited metabolic diseases with rather severe neurological consequences.

With now 70 known cases of DHDDS variants worldwide, 24 different variants have been reported, almost all de novo mutations. In just a couple of instances, one parent must be a mosaic, which means that only some of their cells have the mutation, among them necessarily germ cells, since the same variant showed up in more than one of their children.

There are three common variants and a bunch more found in only one or two patients; in a detailed study of 25 patients, 19 of them had one of these three common variants — p.Arg211Gln, p.Arg37His or p.Arg37Cys. This same study mapped out which symptoms have been observed with different variants, and some trends are beginning to emerge: one of the more common variants, for example, was associated with a more severe phenotype. Zooming in on Arg37His, Galosi et al., 2022 describes the scene:

In addition, the side chain of Arg37 is stabilized by a network of salt bridges with the β-phosphate group of FPP, Glu89 on DHDDS and the main chain carboxylate group of Lys293 on the C-terminus of NgBR. Introducing a histidine residue at this position would abolish binding interactions with the β-phosphate group of FPP substrate as well as with the C terminus of NgBR.

Although the FPP (farnesyl pyrophosphate) substrate is not visible in this 3-D video rendering by AlphaFold, you can see the salt bridge connecting arginine 37 and glutamate 89. (Arginine 37 starts out behind the play button).

Shifting camera angles for the close-up shot of arginine 205, we return to Galosi et al., 2022 for the intramolecular play-by-play:

Similarly, the side chains of Arg205 and Arg211 form salt bridges with the pyrophosphate of IPP, and replacement of either residue with Gln would eliminate binding interactions with IPP. Although substituting Ser213 with Asn can still maintain hydrogen bonding with the β-phosphate group of IPP, introducing a bulky side chain at this region would result in steric clashes with nearby residues, hence leading to structural destabilization.

Again, although the IPP (isopentenyl pyrophosphate) substrate is not visible in this 3-D video rendering by AlphaFold, you can clearly see that arginine 205 forms a nexus of critical salt bridges without which the enzyme’s active site doesn’t work. The guanidino group of arginine 205 forms a blue trident that starts out pointing to the upper left corner of the screen.

Share

There are several hopeful signs that answers for these and other families with DHDDS children may be forthcoming. The gene is highly conserved evolutionarily, so animal models could provide helpful clues on the mechanisms involved in the pathology and potential targets for therapies. Already mouse models exist that providentially contain these two families’ variants. There is even a zebrafish model for a particular DHDDS variant that uniquely affects the retina. Human fibroblast lines derived from patient samples have been established and more recently several research groups are started on making induced pluripotent stem cells (iPSCs) from the fibroblasts of these children. These cells, which can be differentiated into any cell type with the right kind of coaxing, will provide a research tool for studying the actual cell types affected in this syndrome.

An international team of researchers has been studying the enzyme in a yeast (Saccharomyces cerevisiae) model, in which the natural homolog of cis-PTase (ddol-PP synthase) has been replaced with a number of human variants. Working with these yeast avatars, they found that the enzyme complex formed with the human variants has much less activity both in vitro and in vivo, in some cases, restricting or inhibiting growth to a standstill.

The collection of yeast strains with human DHDDS variants was generously shared with Perlara at the end of last summer. Shout out in particular to Dr Kariona Granbińksa at Yale University. As shown in the figure from her 2022 paper, both R37H and R205Q variants have nearly undetectable cis-PTase activity. The question we all wanted to answer next: could we find compounds that bring these DHDDS yeast avatars back from the brink?

After multiple rounds of seeding density experiments with multiple DHDDS yeast avatars, it turned out that the R205Q variant behaved the best in our hands. Consistent with having < 1% of normal DHDDS enzymatic activity — kissing the lower limit of detection — the R205Q yeast avatar is really sick. That did, however, create a situation where the difference between the mutant yeast and the healthy yeast is huge, so there we’d have no trouble discerning even a modicum of growth rescue.

Z’ optimization experiments were performed twice, once at BAL and again at HTSF, just to be sure. Is the R205Q yeast avatar too sick to rescue? If the accumulation of farnesyl pyrophosphate is toxic to cells, it will take a metabolic miracle. If on the other hand the cells are merely in a quiescent state due to one or more metabolite imbalances, then a small molecule rescuer has a fighting chance to increase growth above the rock-bottom baseline.

We completed the 8,400-compound TargetMol screen the week after Thanksgiving. Much to our excitement, the screen revealed several dozen weak-to-moderate rescuers. None of the rescuers fully rescued the R205Q yeast avatar to wildtype growth levels but that was an impossibly heavy lift. Many of the top rescuers are dietary supplements aka nutraceuticals — ripe for combo therapies to boot.

In the meantime, families have been flocking to the Cure DHDDS Facebook group. Just in the time since Zoe and Mel joined forces, the group has grown from a mere handful to 85 members as of this writing.

Zoe and John are using their platform as world renown rock musicians, to bring attention to their story, which they hasten to add, is the story of millions of other families with children suffering from rare genetic diseases. Frances, a natural performer, has been a part of some of these events.

A recent appearance of John and Zoe on NBC’s Today Show ended with the audio of Frances’s sweet voice, singing her favorite line from one of her parents’ song, “Everything is going to be alright.” Let’s hope so, Frances.

Share Perlara’s Cure Odysseys

In collaboration with

Disclaimer

The results of the MAN1B1-CDG drug repurposing project that we are sharing in the spirit of open science below are novel preclinical research findings and therefore they do not constitute the practice of medicine.

Please consult your physician or clinical care team if you’re considering off-label use of any approved drug.

The same caution applies to nutraceuticals, supplements and “generally recognized as safe” compounds.

When a bolt of genetic lightning strikes a family and the smoke clears, a superpower is revealed. When genetic lightning struck their family, Claire and Matt mobilized on behalf of their son Jake. Claire’s and Matt’s shared superpower is a being a medical doctor by profession. As such, they’re in an ideal position to translate drug repurposing discoveries from patient avatars in the lab into 1-to-N medicine.

Physicians, heal thy child.

Holly Carmichael, cofounder of precious Maggie and Maggie’s Pearl, introduced me to Claire and Matt in early 2021. Over a series of Zoom calls, we devised a research plan with a fixed budget and contingencies. Little did we know that we’d have to invoke plan B several times. Claire and Matt found themselves in a succession of morale-draining research ruts lasting months, as often happens when improvising on the unexplored frontier of small-batch science.

Not all disease modeling expeditions are gently sloping straight lines. Complexity sneaks out of nowhere. We hit a wall with worms, and then got stuck in a cul-de-sac with fibroblasts. Flies were the last-ditch pivot a year into the project. Turned out more like flies to the rescue.

Professor Clement Chow’s lab at the University of Utah completed a fly-powered drug repurposing screen in summer 2022, and then followed up with hit validation studies in fall 2022. That’s where the MAN1B1-CDG Cure Odyssey left off and where we pick up the thread — or bar, as it were.

The Chow lab reduced the expression of the fly version of the MAN1B1 gene only in the cells comprising the fly eye. In cases where loss of a gene is not toxic, the fly eye develops normally. But if loss of a specific enzyme causes cytotoxicity in the fly eye, it results in what Drosophilists call a “rough eye” phenotype.

On the left, a healthy photogenic adult ready to explore the fruit bowls of the world. On the right, Rocky Balboa after a few rounds with Apollo Creed. 1,500 compounds that make up the Prestwick library were screened in the Chow lab, by hand, one vial at a time, over four months, in search of clinically actionable compounds that turn shriveled rough eyes into vibrant rubies.

Small-batch science paving the way for small-batch medicine.

The Chow lab’s MAN1B1-CDG fly screen yielded 54 rescuers with a Z score greater than 2. At the same time, the MAN1B1-CDG fly screen revealed 56 sensitizers with a Z score less than to -2. (By contrast, recall that the GMPPA-CDG fly screen, albeit a rescue of whole-animal lethality versus tissue-specific cytotoxicity, yielded only six rescuers). With so many rescuers from which to choose, we had to prioritize compounds by real-world clinical data, known pharmacology, and commercial availability.

In the course of poring over chemical structures, we noticed a peculiar pattern across the MAN1B1-CDG fly hits. There are repeated examples of structurally related pairs of compounds where one member of the couple is a rescuer but its partner is a sensitizer. Typically, structural analogs have the same effect in a screen, i.e., they’re either both rescuers or both sensitizers.

However, in the case of MAN1B1-CDG, the pattern of a pair of structural analogs that have opposite effects is evident across a dozen pharmacological classes that include: steroids, catecholamine action, antibiotics, local anesthetics, alpha adrenergics, beta adrenergics, diuretics, and sulfonylurea anti-diabetics. The dose makes the poison? Can there be too much potency, pharmacologically speaking?

As a first example, consider a tale of two glucocorticoids. Fluticasone is a MAN1B1 rescuer (Z score = 2.31) but its doppelgänger clobetasol is a MAN1B1 sensitizer (Z score = -2.45).

Across 20 drug repurposing programs spun up by Perlara and pioneer families since 2021, closely related hit compounds with the same mechanism of action usually have the same effect in a drug repurposing screen. We’ve only occasionally seen the contrarian pattern where chemical cousins have opposite effects, most of these instances are congenital disorders of glycosylation like MAN1B1-CDG.

Here’s a second duo of divergent doppelgängers. The atypical antipsychotic risperidone is a MAN1B1 rescuer (Z score = 2.55), but its pal pirenperone is a MAN1B1 sensitizer (Z score = -2.24).

A third illustration is a pair of penicillin derivatives. Pivampicillin is a MAN1B1 rescuer (Z score = 4.09). A few cosmetic adjustments leads to cyclacillin, a MAN1B1 sensitizer (Z score = -2.41).

Not to belabor the point, but here’s a fourth example involving a pair of beta adrenergic antagonists. You know the drill: atenolol is the MAN1B1-CDG rescuer, bisoprolol is the MAN1B1-CDG sensitizer.

It gets really interesting when we examine the largest pharmacological class of MAN1B1-CDG rescuers. None other than some of the most widely used medicines on the planet: nonsteroidal anti-inflammatory drugs, or NSAIDs.

NSAIDs target cyclooxygenase 1 (COX-1) and cyclooxygenase 2 (COX-2) enzymes, which convert the lipid arachidonic acid into drivers of inflammation called prostaglandins. COX-1 is constitutively active and has a long protein half-life, whereas COX-2 is inducible and has a short protein half-life. The alias of COX-1 and COX-2 is prostaglandin synthase (PGSH). The fly genome encodes a single cyclooxygenase isoform called COX.

9/54 (16%) MAN1B1-CDG fly rescuers are NSAIDs. Two MAN1B1-CDG fly sensitizers are also NSAIDs, indicating a dose-limiting toxicity that will be discussed in greater detail. The fly screen also revealed five rescuers with anti-inflammatory properties, and two prostaglandin-like rescuers, which act as competitive inhibitors of COX-1 and COX-2.

It was a completely unexpected result. A reminder why we do unbiased drug repurposing using patient avatars in the first place. The most exciting part was the clinical actionability. A bottle of ibuprofen is sitting in everyone’s medicine cabinet. No need for an IRB or spIND. Although chronic long-term use is not an option, ibuprofen is demonstrably one of the safest medicines on the planet. Also on the Essential Medicines List published by the World Health Organization.

Here are the chemical structures of the nine NSAID rescuers.

Phenybutazone and feprazone are chemical twins separated at birth, the pattern we expect to see. Ibuprofen, like most of the NSAIDs shown above, is a nonselective COX inhibitor. Phenacetin approval was revoked by FDA. Fortunately, we could swap it out for acetaminophen.

Three COX-2 selective NSAIDs were included in the hit validation phase: celecoxib (Celebrex), rofecoxib, and the structurally distinct lumiracoxib, which was actually a sensitizer in the primary screen.

16 compounds, mostly NSAIDs, were evaluated in rough eye rescue dose response experiments performed in the Chow lab. Two compounds were not soluble enough in fly food to work with, so 14 hits were actually validated, as summarized in the multi-panel figure below.

Subscribe now

With the sole exception of 4-aminosalicylic acid, the 13 other compounds all had at least one dose that resulted in a statistically significant rescue of the rough eye phenotype (red asterisks). Tolmetin produced the cleanest dose response, but no generics company currently manufacturers a US supply. Ibuprofen has a smaller effect size relative to other NSAIDs but importantly lacks any dose-limiting toxicity observed at 25 µM celecoxib.

Overall, 7 of the 14 compounds exhibit statistically significant rescue at all three doses tested. Interestingly, all three COX-2 selective NSAIDs exhibit dose-limiting toxicity. For celecoxib and lumiracoxib, the peak rescue occurs at 5 µM. For rofecoxib, the peak rescue occurs at 1 µM.

Is the fly telling us that we want to inhibit COX-1 and COX-2 equally, or just weakly? Or is it okay to hit COX-2 selectively using a potent inhibitor but we need intermittent versus continuous dosing to avoid on-target toxicity? At this time, all we can do is make educated guesses based on our data and published data.

NSAIDs target the active site of the enzymes COX-1 and COX-2, both of which are cyclooxygenases that feed on a metabolic substrate called arachidonic acid.

As recounted in Pharma Chameleon, a shape-shifting pharmacophore has emerged from nearly every Perlara drug repurposing hit list. NSAIDs are echoes of arachidonic acid and its high-energy transition states that form in the active site of COX-1 and COX-2. The arachidonic acid pharmacophore coil ups like a cobra into the twin benzene rings of phenylbutazone and celecoxib. Rosuvastatin is a curious hybrid of a NSAID and a statin, structurally speaking.

Can we connect the dots between MAN1B1 and NSAIDs? A PubMed search reveals tantalizing clues. A 2006 paper (Mbonye et al., 2006) directly tied MAN1B1 and COX-2 together via a process called endoplasmic reticulum associated degradation (ERAD), pinpointing glycosylation of asparagine 594 as indispensable for the degradation of the COX-2 enzyme. MAN1B1 initiates a glycan trimming process that sends COX-2 on its way to the cellular trash bin. Meaning once the inflammatory trigger subsides, COX-2 protein is meant to be rapidly cleared from the cell.

Another paper (Singh et al., 2012) showed that decreasing MAN1B1 protein levels led to an increase in COX-2 protein levels. Perhaps loss of MAN1B1 disturbs glycosylation of COX-2 and its naturally fast turnover? So instead of MAN1B1 being an indiscriminate housekeeping enzyme that clips of mannose resides willy nilly from any glycoprotein, MAN1B1 specifically regulates the glycosylation of particular client proteins, in this case COX-2.

A similarly counterintuitive pathomechanism has emerged for another CDG: NGLY1 deficiency. NGLY1, which encodes an N-glycanase, ostensibly has cellular housekeeping functions that affect the deglycosylation of all glycoproteins. But all published evidence from disease models like worms and flies and human cells paints the opposite picture. NGLY1 regulates the activity of a handful of critically important client proteins like NRF1, BMP and NKCC1 using deglycosylation of specific asparagines as a form of post-translational regulation.

If something analogous is happening between MAN1B1 and COX-2, what would happen if a kiddo with MAN1B1-CDG was treated with a NSAID?

Share

Innocent until proven guilty. That’s the guiding principle of our legal system. Safe until proven dangerous. That should be the comparable guiding principle of our medical system. Especially small-batch medicine where the standard of care is nothing because there is no standard of care for 95% of the 10,000+ rare diseases.

Of all the NSAIDs, ibuprofen seemed the most reasonable to try. Remember the old McDonald’s tag line: Billions served. Ibuprofen has been taken by over billion people at this point in its 62-year history. That safety record speaks for itself. But what about more convincing evidence that would predict treatment effect on the autism symptoms of MAN1B1-CDG?

One experiment we could only do in the lab versus a human was to knock down COX in flies in which MAN1B1 is also knocked down to see if loss of COX enzymatic activity rescues the rough eye phenotype. Two genetic wrongs make a phenotypic right. In other words, is genetically knocking down COX in flies equivalent to pharmacological treatment of flies with ibuprofen?

The answer is: yes! This result suggests the rescue effect of ibuprofen is caused by inhibition of COX. In the fly, we infer that loss of the MAN1B1 enzyme somehow leads to an increase in COX enzyme activity, which can be pharmacologically relieved by a NSAID like ibuprofen, or genetically relieved by not having COX in the first place.

The primary screen and the hit validation studies we just presented all use the fly rough eye model. One could reasonably argue for evidence of rescue of a neurological phenotype in flies. Turns out when MAN1B1 is knocked down in the fly brain versus the fly eye, there’s a strong “bang sensitivity” phenotype as shown in the two videos below.

Bang sensitivity is when you literally bang a vial full of flies on the benchtop, thrusting them all downwards. Normal flies hastily start climbing up the walls of vial. MAN1B1-brain-knockdown flies don’t recover the bang and remain dazed or have seizures at the bottom of the vial.

Here’s what happens at 1 µM ibuprofen, which is below the effective dose and comparable to the no-drug control. This video (and the next one) start right after the bang. A few flies attempt to scale the sides but fall down immediately. None of the flies make it off the bottom.

By contrast, here’s what happens at 50 µM ibuprofen, which is a bioactive, non-toxic dose. You’ll observe that about half of the flies are climbers.

Clearly, ibuprofen has a rescue effect on a brain-based phenotype, not just rescue of eye cell proliferation and survival in what could be argued is an artifactual physiological setting. But would ibuprofen have any positive effects on a child living with MAN1B1-CDG?

Subscribe now

Based on the totality of evidence summarized above, Claire and Matt decided to move forward with ibuprofen. A pioneer observational study with Jake a little over one year ago on November 15, 2022. Claire and Matt consulted with CDG expert Dr Eva Morava a few weeks later about how long to dose for and what additional in vitro tests could be done on MAN1B1-CDG patient fibroblasts to test the NSAID hypothesis.

Most of the data are diary entries and other observational notes, even text messages from group chats. Some clinical trial traditionalists will dismiss any bottom-up, parent-led effort as riddled with bias or unreliable data. Those who are more charitable about the concept of small-batch medicine might still reasonably insist on knowing what outcomes will be measured to demonstrate treatment effect. This is why they’re called observational studies. If rare disease parents are really good at one thing it’s observing their kids.

To wit, an unexpected skin biomarker revealed itself in the course of Jake’s pioneer observational study. Keratosis pilaris (KP), a relatively common skin condition colloquially referred to as chicken skin, is a common symptom in MAN1B1-CDG. Within two weeks of taking ibuprofen, Jake’s KP resolved. It’s hard to tell because of the swelling and redness caused by a bug bite, but the bumps characteristic of KP have all but vanished in the pictures below.

But then the KP started creeping back by Week 10, as shown below. It had fully returned to the pre-ibuprofen baseline by Week 14. Claire noted in the summary spreadsheet: “We noticed the KP had slightly returned after decreasing from tid to bid dosing.”

What about cognitive or neurological changes? With respect to sleep, Claire observed:

Waking 30 minutes earlier as soon as we started, waking at 5:15. This fluctuated when he got influenza but now he's back to waking at 5:15. We thought the bid dosing improved his sleep but I think that may have been more related to the influenza.

Other parent observations from Claire and Matt:

• more interested in engaging with people, even people he is meeting for the first time (grabs their arm to get their attention and smiles at them)

• communicating more: pointing to things, now consistently saying 'no', trying to say 'off' and 'on'

• now able to write all his lower case letters independently (previously required support at the wrist to guide his writing)

• more relaxed: willing to try new foods, go to new places (ski lessons), be at a crowded party without issue

In terms of weight, MAN1B1-CDG is unlike other CDGs. Weight gain is an issue, whereas in PMM2-CDG and most other CDGs the kiddos knock back several PediaSures per day but are still underweight. Claire and Matt noticed that Jake’s weight did not change during his two months on ibuprofen. After four weeks off ibuprofen, Jake gained 2.5 pounds.

Share

Small-batch medicine is another way of saying 1-to-N medicine. In this case, from 1 to 3. Two more North American MAN1B1-CDG kiddos — Remi in the US and the Olivia in Canada — started their ibuprofen observational studies three and four weeks after Jake started his study, respectively. No clinical trial coordinator. No clinical trial overhead. In retrospect, a geographically distributed, staggered start, open label, Phase 1/2 study design. Lean startup meets precision medicine.

Olivia’s response to ibuprofen is similar to Jake’s treatment response along multiple axes. First, her KP resolved in the first week of treatment. However, by the third week it was back in full force if not a little worse by the fourth week. What was causing this rebound effect, evident in 2/2 kiddos?

For Olivia, we have four weeks of data. Like Jake, Olivia’s weight stayed constant during ibuprofen treatment. In fact, she lost half a pound while on ibuprofen. However, she gained 15 pounds four weeks after going off ibuprofen.

In terms of any neurological, cognitive or behavioral improvements, the second week on ibuprofen Olivia’s parents noted that she was not as “whiny for food" and “it seems that she is maturing a lot.” Caretakers and teachers noted improvements during the fourth week of ibuprofen treatment. The same early riser lark effect seen with Jake was noted with Olivia in the third week on ibuprofen.

At roughly the same time that Olivia started on ibuprofen, Remi did too, becoming the third pioneer MAN1B1-CDG kiddo to try ibuprofen. Remi appears to be the hyper-responder among the trio of MAN1B1-CDG pioneers. She is also the least affected of the three kiddos. The convergence on the same changes, many of which can be objectively called improvements, across three out of three pioneer MAN1B1 kiddos is noteworthy.

First, Remi’s KP improved in Week 1 but was returning to baseline by Week 2, and then remained unchanged for the rest of the 6-week observational study. Second, Remi’s weight held basically flat over the 6-week treatment period, but then she gained 2.5 pounds four weeks after stopping ibuprofen.

Third, here’s a compilation of text and video updates from Remi’s mom which capture the communications and cognitive gains. Note that Remy was not doing any adding before ibuprofen:

Remi’s elementary school teacher offered the following real-world evidence:

It was agreed that after 6 weeks on drug, and then the kiddos would go 4-6 weeks off drug to see if treatment effects were lost. Ibuprofen is not prescribed for chronic use. When all three kiddos went off drug, they eventually regressed and lost some or all of their gains, an indication that the drug was responsible for the effect.

Were the gains seen in all three kiddos real or just coincidental spontaneous developmental spurts that are sometimes seen in the natural history of CDG? Or should this be viewed from the lens of variable treatment response where one out of the three kiddos exhibited remarkable communication and cognitive gains on ibuprofen?

This past spring, Dr Morava’s lab generously tested COX-1 enzymatic activity and COX-2 enzymatic activity in four MAN1B1 fibroblast lines. Frustratingly, the results are not consistent with the results of the MAN1B1-CDG fly models.

What happens when disease models disagree? Which one overrules? If we were testing COX-1 and COX-2 activity in neurons or glia instead of fibroblasts, would we see concordance with the fly data package? We are left with more questions than answers at this juncture.

Share

The convergent flickers of efficacy in the ibuprofen n-of-3 parent-led observational study leave us wanting more but the risk of rebound and regression gives us pause. MAN1B1 turns out not to be an easy gene. Loss of MAN1B1 can be shrugged off by most cells but some cells, in particular neurons and/or glia are particularly sensitive. If any researcher reading this wants to help figure out why, by all means join the MAN1B1-CDG cause.

With NSAIDs on pause, what about rosuvastatin (Z score = 2.75) or guaicol (Z score = 4.68), which is actually chemically similar to vanilla? Both rosuvastatin and guaicol validated in the 3-point dose response study alongside ibuprofen. It is worth reconsidering those but also keep in mind every intervention is a several month commitment if it’s to be done right and produce meaningful data. Risperidone showed a nice dose response in the hit validation study. The concern though is the weight gain that is a known side effect of antipsychotics like risperidone. MAN1B1-CDG patients already struggle with hyperphagia and weight gain.

In parallel, a new dawn rises. More recently diagnosed MAN1B1-CDG families are taking up the drug repurposing and disease modeling torches where Claire and Matt have left off. Cure odysseys are ultramarathons. One family can’t be expected to go the distance by themselves. 1,500 compounds were screened in the first drug repurposing campaign. The goal now is to screen at least 4,000 compounds in a fly model, ideally using the brain-specific MAN1B1 knockdown model and screening for compounds that rescue a neurological phenotype.

Onward for MAN1B1-CDG in 2024! Jake, Olivia, Remi and other kiddos like them are counting on it.

Share Perlara’s Cure Odysseys

In collaboration with

Last year Cure Mito Foundation made a bet on Perlara and yeast as a drug repurposing platform. The bet has paid off preclinically, so far. For the first time, a rational mitochondrial cocktail is in sight if we can demonstrate therapeutic effect and elucidate rescue mechanisms in human cells.

Yet it will take more than just a silver bullet monotherapy to defeat a disease as pernicious and complex as Leigh Syndrome. Combination therapies will one day be the norm for inborn errors of metabolism, and likely most genetic diseases.

In analogy to Vertex’s Trikafta, a transformative novel triple combination therapy for cystic fibrosis, we envision combining three different (in this case repurposed) molecules for SURF1 deficiency, with each compound contributing an additive — or possibly synergistic — rescue mechanism. That’s a very ambitious goal still on the horizon, but we’re rocketing forward. We have no choice.

A year ago we completed our first SURF1 drug repurposing screen using the 2,250-compound Pharmakon library at the Small Molecule Discovery Center at UCSF. Six months ago, we completed a second SURF1 screen using the much larger 13,500-compound ReFRAME library, which had been made available to researchers by Scripps Research. Results of the Pharmakon and ReFRAME screens are summarized here.

In the spirit of completeness, we embarked on a third SURF1 screen using the 8,400-compound TargetMol library, which is now our first-line option for yeast-powered drug repurposing. Here’s a Z score summary plot of the SURF1 TargetMol dataset. The TargetMol data are in line with the Pharmakon and ReFRAME results.

A head-to-head comparison of the Z score distributions across the three different drug repurposing libraries totaling 24,150 compounds is shown below. The shape of the Z score distributions is similar in all three screens. The Pharmakon screen had the lowest experimental variance and therefore much higher Z scores. Not too surprising given that there are only seven 384-well plates in the entire Pharmakon library, which also meant the screen was performed in a single run. In contrast, there are 52 plates in the ReFRAME library and 27 plates in the TargetMol library. Those libraries were screened in batches at different times.

Each library yielded dozens of rescuers. Notably, each screen resulted in a unique top rescuer. Consistent across all the libraries, there are only a handful number of sensitizers, especially relative to the number of rescuers. We don’t know what the excess of rescuers relative to sensitizers means other than that there may be multiple SURF1 rescue (or bypass) mechanisms available to activate pharmacologically.

Share

After applying filters like pediatric approval status, oral bioavailability, and blood-brain barrier penetrance, we down-selected 12 yeast hits from the three libraries and reordered fresh powder stocks for mitochondrial stress tests in SURF1 patient fibroblasts. Linsey Stiles is the scientist at the UCLA Mitochondria and Metabolism core facility who performed the experiments summarized below.

The Seahorse instrument measures oxygen consumption rate (OCR) and also distinguishes between ATP production by glycolysis versus mitochondrial respiration (aka oxidative phosphorylation/OxPhos). As shown in the schematic below, cells are exposed to a series of stress tests that target the start, middle and end of the electron transport chain.

We tested a SURF1 patient fibroblast line (GM28010) and an age-matched control fibroblast line (GM00038). First, we established baselines measurements under standard growth condition — glucose as the carbon source — in the absence of compound treatment.

As shown in the figure below, ATP production from mitochondrial respiration (oxphos) is reduced by ~67% in the SURF1-deficient patient fibroblast line. There’s a compensatory increase in ATP production by glycolysis. What we’re looking for is a compound that increases ATP production by mitochondrial respiration, returning the disease profile back to normal.

Share

Technically, four out of 12 yeast hits increased ATP production by mitochondrial respiration in a representative SURF1 patient fibroblast line. The fourth yeast hit increased basal respiration and ATP-linked respiration in galactose-cultured cells; however, this compound showed some toxicity in glucose-cultured cells with a dose-dependent decrease in all respiratory parameters measured.

Therefore, we considered three out of 12 yeast hits as validated. That translates to a 25% bioactivity conversion rate from a yeast cell to a human cell. Hereafter the three yeast hits will be referred to as Compounds 1-3.

Treatment with Compound 1 for 48 hours resulted in 12-15% boosts in ATP production by mitochondrial respiration and maximal respiration in both glucose-cultured and galactose-cultured cells. There is also concomitant decrease in ATP production through glycolysis.

Treatment with Compound 2 for 48 hours did not result increase maximal respiration in either glucose or galactose conditions. However, there was an increase in ATP production through both mitochondrial respiration and glycolysis.

Treatment with Compound 3 for 48 hours did not result in changes to maximal respiration in either glucose or galactose conditions, but there was an increase in ATP production through mitochondrial respiration at the 10 µM concentration.

Let’s look closer at maximal respiration in response to treatment with Compound 1 in galactose media versus glucose media.

Seven of the twelve yeast hits appear not to be active in fibroblasts.

Could it be because they’re yeast-specific hits, i.e., target a mechanism that is not evolutionarily conserved? Or do we need to treat with compound longer, or in different SURF1 patient fibroblast lines from genetically unrelated individuals? Or perhaps fibroblasts are the wrong cell type to study and we need to be interrogating neurons or muscle or other high-energy-demand cells. We’re pursuing the neuron option with Prof Alessandro Prigione’s lab in Germany.

Interestingly, two of the twelve yeast hits selectively affected ATP production by glycolysis. One of the hits decreases the fraction of ATP produced by glycolysis, while the other hit increases the fraction of ATP produced by glycolysis.

Altogether, half of the 12 SURF1 yeast hits are bioactive in SURF1 patient fibroblasts, a remarkable testament to the predictive power of Perlara’s yeast-powered drug repurposing platform. For any remaining skeptics who doubt the therapeutic relevance of yeast, remember that statistic: 50% conversion rate.

Next steps are to repeat the Seahorse assay on the 12 yeast hits in a genetically unrelated SURF1 patient line, and test combinations of Compounds 1-3. Onward!

Share Perlara’s Cure Odysseys

In collaboration with

Disclaimer

The results of the AFG3L2 drug repurposing project that we are sharing in the spirit of open science below are novel preclinical research findings and therefore they do not constitute the practice of medicine. Please consult your physician or clinical care team if you’re considering off-label use of any approved drug. The same caution applies to nutraceuticals, supplements and “generally recognized as safe” compounds.

Genetic lightning can strike any family, anytime, anywhere. Sometimes it reveals parental superpowers. When I say superpower I mean a skill or expertise or personality trait that served the parents particularly well in their professional lives, and now this alpha has found a purpose like no other job or mission before. Once unleashed, kiddo-inspired superpowers spark into existence 1-to-N odysseys that begin with a single affected child and then spread across patient communities by word of mouth and social media like a chain reaction of hope.

Like every family we work with at Perlara, Rita and Daniel are compelled to lead even if that means going from zero to one. They launched Miracles for Mighty Milo, a cure-focused research foundation, on behalf of their son Milo, who suffers from Spastic Ataxia Type 5 (SPAX5). SPAX5 is caused by inheriting two pathogenic variants in the evolutionarily conserved gene AFG3L2.

The AFG3L2 gene encodes a protease, or a protein that chews up other proteins. Specifically, it’s a mitochondrially localized protease that is responsible for recycling unfinished or damaged proteins that reside in the mitochondrial membrane. For that reason, autosomal recessive AFG3L2 can be considered an ultra-rare mitochondrial disease. As we’ll show in the results of our yeast-powered drug repurposing screen for AFG3L2, the mitochondrial connection is quite salient.

Over the summer, we performed a 8,400-compound TargetMol screen at the HTSF in Berkeley, California. Here are the summary results. Per usual for our mitochondrial yeast avatar workflow, the positive control (magenta circles) is a wildtype yeast strain grown in 2% lactate media, and the negative control (cyan circles) is the mutant AFG3L2 knockout yeast strain grown in 2% lactate media.

The orange circles represent the mutant AFG3L2 knockout yeast strain grown in 2% lactate and treated with one of the 8,400 test compounds.

We observed exceptionally tight distributions of the positive and negative control well values, i.e., the magenta and cyan circles are all clustered around the mean (black bars) with very little spread. 26 compounds are discernible rescuers. Although none of the rescuers achieved wildtype growth levels, that’s not unusual in our experience across over many yeast drug repurposing projects. There are at least a dozen sensitizers depending on where we draw the cutoff.

Here’s the AFG3L2 TargetMol Z score summary figure with compound identities listed. The dark blue shaded rectangle contains the strongest rescuers. The light blue shaded rectangle contains the moderate to weak rescuers. Sensitizers are highlighted in the red shaded rectangle.

The enlarged orange circles and bolded names correspond to the sprawling family of Chinese Traditional Medicines called steroidal saponins. Intriguingly, steroidal saponins, quaternary amine containing lipids, and sphingolipid analogs are also rescuers of FARS2, another ultra-rare mitochondrial disease.

Just as we’ve seen convergence of rescue mechanisms in yeast between different congenital disorders of glycosylation (CDGs), the same pattern of convergence is emerging for mitochondrial diseases.

Share

Three out of the top six AFG3L2 rescuers are saponins. Eight out of the 26 AFG3L2 rescuers are saponins. With the exception of FARS2, we have not previously seen such a strong enrichment of saponins as rescuers. The chemical structures of six out of the eight saponins are displayed below. Note that unlike in the case of FARS2, there appears to be a strict requirement for a disaccharide or trisaccharide attachment. Many of these Chinese Traditional Medicines (CTM) are available as over-the-counter extracts and in a variety of formulations. Batch-to-batch effects and consistency of manufacturing process are issues for therapeutic natural products like CTM.

Next we compared the AFG3L2 and FARS2 TargetMol Z score datasets so we could evaluate the degree of convergence. It’s immediately apparent that the upper right quadrant is populated with shared rescuers. However, the lower left quadrant is empty, meaning there are no shared sensitizers.

Here’s the same Z score plot with hit compound IDs added. The quaternary amine lipid octenidine is the top common rescuer. RP-54745, an immunomodulating preclinical tool compound, is another common rescuer. Diclazuril and to a lesser extent muscone are examples of FARS2-specific rescuers.

There are only a handful FARS2 sensitizers, while there are at least a dozen AFG3L2 sensitizers highlighted by the red shaded rectangle. Let’s zoom in closer on the AFG3L2 sensitizers. We set a cutoff of Z scores below -0.35, the dashed line in the figure below. The top 12 AFG3L2 sensitizers are listed below by name.

Remarkably, almost all AFG3L2 sensitizers are also SURF1 sensitizers! For example, the iron containing porphyrins shown here.

Two AFG3L2-specific sensitizers are particularly noteworthy because of their mechanism of action.

Genipen is a protein cross-linker, a chemical agent that randomly staples together proteins that happen to be close by each other. When AFG3L2 is missing, mitochondria have diminished capacity to deal with protein stress, also known as proteotoxicity. Treating AFG3L2-deficient yeast with genipen is adding fuel to the fire of mitochondrial protein misfolding and respiratory complex mis-assembly.

MKT077 is a HSP70 inhibitor. HSP70s are a family of heat shock proteins whose expression is increased during heat stress or heavy metal stress. Just as a protein cross-linker would make proteotoxicity matters worse, inhibiting HSP70 would reduce cellular capacity to deal with mitochondrial protein stress.

Finally, a third group of chemically and pharmacologically related AFG3L2 sensitizers are polyphenols — specifically phenylpropanoids — which for reasons that we don’t yet understand are rescuers in CDG yeast drug repurposing screens.

Piceatannol and oxyresveratol are both stilbenoids. As we’ve seen in previous screens, there’s a hidden pharmacophore, in this case a stilbenoid scaffold that is accented in purple in the structure of hypericin, the second most potent AFG3L2 sensitizer. Hypericin is actually a criss-crossed double stilbenoid in a locked conformation.

Over the summer, Rita and Daniel tried out a few CTM in a series of n-of-1 observational studies with Milo. There was no playbook to recommend dosing levels or dosing frequency. Flickers of efficacy were observed, but then seizures led them to pause CTM. Seizures are part of the natural history of SPAX5.

What are the saponins, quaternary amines and sphingolipid analogs doing to rescue AFG3L2 deficiency? Is it the same mechanism at work underlying rescue of FARS2 deficiency too? Our hypothesis is that changing the ratio of positively charged lipids to negatively charged lipids leads to increased cellular capacity to handle proteotoxic stress. Perhaps the saponins themselves are acting directly on mitochondrial membrane proteins as miniature heat shock proteins?

The mitochondrial convergence we’re seeing is striking because the monogenic driver genes affect different aspects of mitochondrial physiology. FARS2 is involved in mitochondrial protein production, AFG3L2 is involved in mitochondrial protein degradation, and SURF1 is involved in Complex IV assembly. It’s fascinating to speculate about shared rescue mechanisms that would impact a wide range of mitochondrial diseases.

However, more funding is required for hit validation studies in Milo’s fibroblasts. Consider donating to Miracles for Mighty Milo to build on these exciting preliminary results which may have implications beyond kiddos like Milo and AFG3L2-related mitochondrial disease like SPAX5.

Share Perlara’s Cure Odysseys

In collaboration with

Disclaimer

The results of the FARS2 drug repurposing project that we are sharing in the spirit of open science below are novel preclinical research findings and therefore they do not constitute the practice of medicine. Please consult your physician or clinical care team if you’re considering off-label use of any approved drug. The same caution applies to nutraceuticals, supplements and “generally recognized as safe” compounds.

It took the discerning eye of a grandmother to notice that something was not quite right with Alexander, when at an early age, he had an unusual way of standing in his crib. Rather than standing on his soles, Alexander stood on the top of his feet. And as he grew, his walking led to tripping and skipping. As so often happens with ultra-rare diseases, it wasn’t until some years later that the family received a diagnosis for then 9-year old Alexander. Genetic analysis revealed that Alexander has a rare autosomal recessive disorder, FARS2 deficiency, a defect affecting mitochondrial function.

Dozen of disorders exist that impair mitochondrial function, which can result in various maladies from vision or hearing loss to muscular and cognitive deficiencies. FARS2 deficiency is one of a subset of mitochondrial disorders involving nuclear-coded mitochondrial tRNA synthetase proteins (mt-aaRS). These proteins play an essential role in protein synthesis, as they catalyze the attachment via aminoacylation of amino acids onto their cognate tRNAs.

This is a schematic of an aminoacyl-tRNA with a yeast tRNAPhe and generic amino acid. An arrow indicates where an ester bond is formed between a 3' terminal adenosine on the tRNA and the amino acid carboxyl group.

First described in 2012, FARS2 deficiency is among rarest of mt-aaRS deficiency disorders; as of this writing, only 41 variants/cases of the FARS2 gene, which codes for the phenylalanine tRNA synthetase gene, have been reported, putting it in the ultra-rare category. Mt-aaRS variants all do pretty much the same thing—disrupt the proper translation of proteins in the mitochondria, which in turn affects energy production in the form of ATP in cells throughout the body. However, the brain and the musculature appear to be the greediest organ systems, as these disorders most often result progressive motor and cognitive disabilities. Two forms of FARS2 disorders have been described: early onset epileptic mitochondrial encephalopathy, an aggressive form, appearing in the very young, and a late form characterized by spastic paraplegia. The early onset form is more prevalent—occurring in about two thirds of the cases — and has a poorer prognosis, often accompanied with developmental delays and various brain abnormalities.

Mt-aaRs disorders are inherited in an autosomal recessive fashion. Most patients are compound heterozygotes, which simply means they have different mutations in the two copies of the FARS2 gene, one originating from each parent. Although the actual amino acid changes caused by the mutations are known, the disease mechanism is not well understood. For one thing, it’s unclear why certain variants give rise to the more aggressive form.

Subscribe now

By performing trio exome analysis of Alexander’s family (mother, father, child), researchers discovered two unique mutations in Alexander’s FARS2 genes; a paternally inherited nonsense mutation (Q216X) and a maternally inherited missense mutation (P136H). The Q216X gives rise to a truncated protein near the midpoint of the protein, thus removing both the catalytic site and anti-codon binding domain while the P136H variant results in a nonconservative amino acid replacement in what is otherwise a conserved region of the protein—- which suggests functionality might be affected, since Nature chose to conserve that part of the protein.

This schematic view of the location of the FARS2 variants reported to date shows that they are scattered rather uniformly throughout the protein. Those found either in epileptic or spastic patients are underlined, and those found only in epileptic patients are indicated in black, while the variants associated only with spastic paraplegia are indicated in blue.

As FARS2 syndrome goes, Alexander had what would be considered a mild case, lacking cognitive decline or debilitating epileptic encephalopathy, seen in the more severe cases. Looking for some explanation for this, the Cedars-Sinai group expressed the proteins and measured their activity in vitro, and found some residual activity with the P136H protein. (The Q216X proved impossible to purify, but given that it lacks two essential parts of the protein, it would be surprising if it were to have any activity).

And now for some good news. As a group, these proteins are conserved evolutionarily, which means that it might be possible to crack the code on the connection between genotype and phenotype is some simple model systems. In Drosophila, for example, Chinese researchers found that knocking out FARS2 two entirely led to developmental delays and seizures. Drilling down a little more precisely, they showed that one human variant linked to early onset epilepsy in people (pG309S) when expressed in flies, led to seizures, while another associated with spastic paraplegia (pD142Y) affected locomotion.

Share

FARS2 is also conserved in Saccharomyces cerevisiae, aka our old friend budding yeast. The yeast version of FARS2 is called MFS1. Dusting off our established mitochondrial gene knockout yeast avatar playbook — see SURF1 and AFG3L2 examples — we could hit the ground running using off-the-shelf MFS1-knockout yeast mutants.

We performed a 96-well-plate growth assessment in 2% lactate media of an allelic series at two temperatures (30˚C vs 37˚C) and two timepoints (24 h vs 48 h). The allelic series consists of a wildtype homozygous diploid control (MSF1/MSF1), a heterozygous knockout diploid mutant (MSF1/∆msf1), and a homozygous knockout diploid mutant (∆msf1/∆msf1). As expected, the homozygous diploid knockout mutant exhibited a severe growth defect under all conditions tested.

We advanced to the next gate: a seeding density experiment in 384-well plates with the homozygous diploid knockout mutant at a 24-hour timepoint. Again, the results were as expected, with a large delta between the wildtype strain (black circles) and the FARS2 patient avatar (magenta squares).

The last step before progressing to the 8,400-compound TargetMol drug repurposing library is a Z’ optimization experiment that simulates the conditions of the screen. Think of it as the final dress rehearsal before opening night.

Everyone hit their mark, as demonstrated by the stark night-and-day absolute difference between the growth of the wildtype strain and the growth of the FARS2 patient avatar. Notably, the well-to-well variance is low, resulting in Z’ values close to the gold standard value 0.5.

Below is a summary plot of the 8,400-compound TargetMol screen. As per usual, the orange circles represent individual test compounds and the FARS2 knockout yeast avatar. The negative control is the FARS2 knockout yeast avatar treated with DMSO (cyan circles), the solvent in which all test compounds are dissolved. The positive control is the wildtype yeast treated with DMSO (magenta circles). While there was some slippage of wildtype yeast that grew sub-optimally, the FARS2 knockout yeast avatar grew consistently poorly, as expected.

Share

The identities of the rescuers are listed below. We created two buckets: strong-to-moderate rescuers (boxed in dark blue) and moderate-to-weak rescuers (boxed in light blue). Several mechanistically and structurally related classes of rescuers are evident even to the chemically untrained eye. They include three types of fatty acids, and a sprawling family of plant-based Chinese Traditional Medicines called steroidal saponins.

Let’s focus our attention on the top two hits that have the largest effect size: perifosine and miltefosine.

Both compounds belong to a pharmacological class called lysophosphatidylcholines, or LPCs. Technically, they are classified as alkylphosholipids. LPCs are a modified form of the naturally occurring phospholipid called phosphatidylcholine (PC) where one of its two fatty acids tails is lopped off. LPCs mimic PC inside the cells, inhibiting the enzymes that produce PC and gumming up the protein transporters that normally control the flow of lipid tides from one organelle to another.

Since 2014, miltefosine (trade name Imavido) is a FDA approved oral drug for use as a 28-day treatment regiment for leishmaniasis, a parasitic infection. It was originally developed in the 1970s and 1980s as an anti-cancer agent. It was approved in India as an antileishmaniasis therapy in 2002. Miltefosine sits on the WHO’s List of Essential Medicines. For the avoidance of doubt, it is not FDA approved for chronic use in any disease indication.

Perofosine and miltefosine are not the only lysophosphatidylcholines that scored as FARS2 rescuers in the yeast screen. All of the compounds whose structures are displayed below are united by the presence of a positively charged choline or terminal quaternary amine, but there is tolerance for a range of acyl chain lengths. What’s more, the phosphate group is dispensable for activity. Interestingly, two versions of phosphatidylcholine are also rescuers, albeit weak ones. The unifying structure-activity relationship appears to be phosphatidylcholine mimicry.

The screen also yielded a pair of sphingolipid analogs, amiselimod and fingolimod. Phospholipids and sphingolipids both have nitrogen-containing head groups like choline or ethanolamine. Why are positively charged (cationic) lipids protective in FARS2-deficient yeast? We’ll return to this question momentarily.

Curiously, in addition to the cationic fatty acids shown above, the FARS2 yeast screen also yielded a group of structurally related negatively charged (anionic) fatty acids. Elaidic is the trans isomer of oleic acid, which is famously found in olive oil. Glycerol alpha-monomyristate is a commonly used food additive. Agaric acid is a natural product found in fungi but is not generally recognized as safe (GRAS). Phorbol 12-myristate 13-acetate (PMA) is a laboratory tool compound historically used to stimulate the protein kinase C (PKC) pathway. Agaric acid and PMA shared a diacylglycerol backbone.

Finally, there are two groups of steroidal saponins, which are Chinese Traditional Medicines derived from plants. The first group shares an oleanane scaffold. Intriguingly, the aglycosidic Hederagenenin appears alongside its monosaccharide (Calenduloside E), disaccharide (Momordin Ic and Ginsenoside Z-R1) and trisaccharide (Hederacolchiside A1 and Raddeanin A) versions. There doesn’t seem to be a preference for the number of sugars attached to the steroidal core.

The second group of saponins belong to the sarsasapogenin family. As is the case for the first group of saponins, 1-3 sugar attachments are seemingly equally tolerated: the monosaccharide (Timosaponin AI), disaccharide (Timosaponin AIII) and trisaccharide (Dioscin and Pennogenin 3-O-beta-chacotrioside) versions all rescue.

At this juncture, we can only speculate about mechanisms of action. It is a noteworthy coincidence that 5 out of the top 10 rescuers are choline-bearing lipids, and 3 out of the top 12 rescuers are steroidal saponins. A 2010 paper in on the effects of miltefosine (and other LPCs) and lipid metabolism in a well studied hepatic carcinoma cell line (HepG2) found that miltefosine changes the balance between choline-bearing lipids and cholesterol. Specifically, miltefosine decreases the levels of phosphatidylcholine and sphingomyelin, which also has choline head group, while at the same time increasing the levels of cholesterol.

Why might shifting the balance of lipids matter to the mitochondria? Membrane lipids with positively charged head groups interact with other membrane lipids with negatively charged head groups. More cholesterol in mitochondrial membranes might somehow increase the cell’s capacity to deal with misfolded mitochondrial membrane proteins that build up when there’s not enough tRNA-Phe to fuel protein production at full speed and high fidelity.

Next steps on the preclinical research front: test the top 10-15 yeast hits in Alexander’s fibroblasts (and potentially myoblasts) using the Seahorse instrument followed by mitochondrial lipidomics and mitochondrial protein stability studies.

Of course, for the families, treatment for FARS2 can’t come soon enough. Although Alexander has not experienced some of the more serious symptoms — with the infantile form, children die before the age of two — his disease is not without consequence.  Now 16, Alexander lives normally, his mother describes him running up and down subway steps on a recent trip to Manhattan.  Yet, she also reports that is walking has worsened lately — he is more apt to lose his balance and he is finding it hard to lift his legs or step sideways. And people are noticing in a new way. Before he had a little bit of a weird walk and people might or might not notice. But now people are seeing him as disabled, according to his mother — offering golf carts, pointing out elevators, something that has not happened to Alexander so far, who has refused to identify as handicapped. His family as well has not applied for services or handicapped parking.  Whether access to intervention with these therapies will come in time to allow him to retain his identity — “that’s our hope.”

Share Perlara’s Cure Odysseys

In collaboration with

Let’s start by centering this story around our pioneer patient and her family. Eliana has a de novo pathogenic variant, V434L, in her KCNC1 gene, which encodes a potassium channel called Kv3.1 that is predominantly expressed in inhibitory GABAergic interneurons and cerebellar neurons. A genetic lightning strike to the KCNC1 gene can present as myoclonic epilepsy and ataxia due to potassium channel (MEAK), developmental epileptic encephalopathy (DEE), or intellectual disability with hypotonia but no seizures. It all comes down to where lightning strikes.

A cryo-EM model of Kv3.1 is the featured image above. Four Kv3.1 subunits intertwine to form a pore through which the purple potassium ion flows. Imagine that you are facing down the barrel of the channel. Highlighted in red, V434 resides near the center of the channel and at the interface between helices. This interface is key for channel function, as these dynamic helices need to slide past each other for the channel to switch between closed and open states. Could V434 have a role in maintaining Kv3.1 in the open-like state? You might think leucine and valine are not that different as amino acids go, what’s an extra methylene group among friends? It’s the difference between typical neurodevelopment and a neurodevelopmental disorder.

Eliana lives with her parents and her new baby sister in Canada. Phenotypically, she does not present as a typical DEE. However, she has significant hypotonia, cortical visual impairment, vertical nystagmus, and global delays. You could say genetic lightning struck again because Eliana’s mom Stephanie happens to be a genetic counselor by profession. Stephanie and her husband Chris sprung into action after Eliana’s diagnosis and founded the KCNC1 Foundation to drive research forward. Today, the Foundation is aware of 36 patients affected by a total of 14 variants: ~25% of patients share the same variant, A421V; 12.5% have MEAK caused by the R320H variant, which is thought to be an under-representation; a handful of patients have V432M; the rest of the variants affect 1-3 patients each.

Stephanie engaged Perlara for “cure guidance” a year and a half ago. Our core recommendation was to focus on drug repurposing as the fastest, cheapest and safest route to a clinically actionable intervention. Dr Whitney Dolan has served as what we call a generalist Cure Guide, providing comprehensive project management and continuity over time. Whitney is paired up with specialist Cure Guide Dr Van Duesterberg who got a PhD in biophysics studying ion channel function. We’ve found that this Cure Guide generalist-specialist duo model works well operationally across dozens of projects regardless of disease gene or therapeutic modality.

When we started working together 18 months ago, we weren’t certain whether Eliana’s V434L variant is loss-of-function or gain-of-function. Normally that’s a showstopper — but not in The Great British Baking Show sense. That thorny issue got sorted out earlier in the year by an academic group at the same KCNC1 Foundation elected to move forward with an unbiased V434L-focused drug repurposing screen in collaboration with Metrion Biosciences, a CRO that specializes in high-throughput ion channel screens.

7 months ago when we posted the last project update, the results of the Clatot et al study — the first to describe the V434L variant as having a gain-of-function effect on channel activity — had just been by reproduced by the experienced team of electrophysiologists at Metrion. The structural insights gleaned from the cryo-EM model were right: the mutant V434L Kv3.1 channel appears to open too easily and then have a hard time closing once open. Paradoxically, even though the Kv3.1 channel is more active, the neurons that express it become less active.

However in science, no result is sacred until it’s convincingly replicated and properly controlled. It was critical for Metrion to independently reproduce the published academic findings, which were performed on HEK293 cells that were transiently transfected with a plasmid expressing wildtype KCNC1 or one of three pathogenic variants: M430I, V432M, and Eliana’s V434L.

We realized early in the project planning stages that doing a drug repurposing screen in neurons differentiated from Eliana’s iPSCs would drive up costs substantially, as in tripling the budget, not to mention doubling (at least) the timeline. Clatot et al validated the minimum viable channelopathy model: a generic cell line expressing the mutant V434L Kv3.1 channel. One of the hoops that Metrion needed to jump through in the assay optimization stage was assessing if transient transfection would be sufficient or whether they would need to stably transfect cells with V434L KCNC1. They also needed to confirm which exact channel activity assay to deploy in a high-throughput screen. All doable based on a time-tested ion channel screening playbook.

The major caveat of the minimum viable channelopathy model is that a strictly electrophysiological readout will not capture intracellular trafficking defects or protein stability defects that result from the V434L gain-of-function variant. That said, it also doesn’t take much sleuthing on PubMed to find multiple examples of drugs that hit potassium channels like Kv3.1 as an off-target. Polypharmacology is no stranger to ion channels. Which brings us back full circle: centering the research process on Eliana’s V434L variant is the only way to make a data-driven decision since we can’t assume even neighboring KCNC1 variants have the same effects on Kv3.1 channel function.

In this section, we get in the weeds of high-throughput screening for the benefit of families and foundations looking to apply the minimum viable disease model playbook to their channelopathy. Metrion started out with transient transfection of CHO cells with plasmids expressing wildtype KCNC1 or Eliana’s gain-of-function variant V434L. As shown in the figure below, the transient transfections of CHO cells using plasmid encoding human KCNC1-WT and human KCNC1-V434L worked equally well, allowing progression to the biophysical characterization of both KCNC1-WT and KCNC1-V434L currents. Moreover, as an important negative control, non-transfected cells were shown not to exhibit electrophysiological activity confirming that the currents recorded from transfected cells were generated by human KCNC1-WT and KCNC1-V434L channels. So far, so good.

The FLIPR assay uses a fluorescent dye to measure ion flow through the Kv3.1 channel, thereby converting an electrophysiological readout into a fluorescence readout. The dye is sensitive to changes in intracellular thallium ions (Tl⁺), which are used as a surrogate for potassium ions (K⁺). The rate and quantity of Tl⁺ ions flowing into the cell can be regulated by opening and closing the Kv3.1 channels in the plasma membrane. An increase in fluorescence measured on the FLIPR directly correlates with an increase in Tl⁺ flux into the cell.

The V434L gain-of-function variant is activated at lower concentrations of K⁺. In a series of benchmarking experiments, Metrion characterized WT versus V434L Kv3.1 channels in response to two known Kv3.1 channel modulators: AUT1 and 4-aminopyridine (4-AP).

The application of 4-AP 3 mM almost completely inhibited the KCNC1 WT mediated currents, while the same 4-AP concentration only blocked the KCNC1 V434L current by 40%. The application of Aut1 increased the current amplitude when tested on the KCNC1 WT channel in a dose-dependent manner, while the application of Aut1 at 30 µM strongly blocked the current mediated by the V434L Kv3.1 channel. Altogether, these results indicate that the V434L Kv3.1 channel shows a different pharmacological profile when compared to the WT channel.

Assay performance of the transiently transfected polycloncal cells (meaning the cells were not derived from a single clone) was not ideal. We half-expected this outcome. So Metrion isolated monoclonal lines expressing either WT or V434L channels. As shown below, clone 2G6 showed the most reproducible behavior across 120 replicate wells.

Further QC and assay performance was done by Metrion prior to the primary screen of the Broad Repurposing Hub library. The spiked plate assessment is where a dilution series in triplicate is randomly distributed across the wells of a plate and then a dose-response curve is computationally reconstructed. Two known Kv3.1 channel inhibitors were assessed: AUT1 and the antidepressant fluoxetine (aka Prozac).

Well-fit dose response curves were successfully generated from the randomly distributed data points. AUT1 and fluoxetine are virtually indistiguishable and equipotent Kv3.1 channel inhibitors.

The IC50 of fluoxetine on V434L Kv3.1 channels is 18.5µM when the two replicates are averaged, modestly exceeding the plasma concentration achieved in clinical dosing in the on-label antidepressant context. That’s not what any pharmacologist would call potent inhibition. So the antidepressant pharmacology/mechanism of action of fluoxetine will certainly be in play. Perhaps neurogenesis is an on-target side effect?

Intriguingly, a recent paper described a nearby valine residue V425 in KCNC1 that is mutated to methionine (V425M) and results in a modest gain-of-function and electrophysiological profile somewhat similar to V434L, with the key and perhaps critical difference being that V425M Kv3.1 channels behave more normally when co-expressed with wildtype Kv3.1 channels.

Remarkably, a child with V425M has been taking fluoxetine as part of a 3+ year single-patient observational study in Italy. Seizures have been completely controlled and the child has experienced neurodevelopmental gains and psychomotor improvements. Will fluoxetine emerge as a top inhibitor of V434L? What other compounds will be revealed by an unbiased screen of the Broad Repurposing Hub library?

Share

While much of Europe was sunning itself on a Mediterranean beach in August, Metrion received compound plates from the Broad Institute, with plans to begin screening the last week of August. Compounds were screened at 10 µM in duplicate. Below is an example the FLIPR assay output in relative fluorescence units. The green traces indicate 0% Kv3.1 channel inhibition (aka the negative control) while the blue traces indicate 100% Kv3.1 channel inhibition (aka the positive control).

The goal of this precision drug repurposing campaign is to identify V434L Kv3.1 channel inhibitors that behave like the red trace 7, i.e., full inhibition.

For high-throughput screening aficionados, the signal-to-background (S:B) ratios and the Z-prime values (RZ’) of each replicate pair of plates comprising the entire Broad Repurposing Hub library are plotted below. Well-performing screens have a S:B ratio greater than 3, and RZ’ greater than 0.5. Metrion easily cleared that hurdle.

The Broad Repurposing Hub library fills 23 384-well plates. One way to visualize the concordance between the two replicates is by using plate heatmaps where the intensity of channel inhibition or activation by each test compound in a well corresponds to the color intensity of red and blue, respectively. Test compounds that have no effect on V434L Kv3.1 channel function are rendered in white.

Here’s the distribution of test compounds plotted as a function of percent inhibition. Several dozen test compound inhibit V434L Kv3.1 channel function as well or more as the positive control. Reassuringly, hit compounds are randomly distributed across the library and not clustered in a handful of plates.

Now let’s merge the test compounds from the two replicates into a single plot while subtracting out the control wells. When the threshold for inhibition is set at 90%, there are 34 test compounds that hit the mark in both replicates. That translates to a hit rate of 0.5%, the sweet spot for a robust unbiased phenotypic screen.

What happened with fluoxetine? Turns out fluoxetine and its primary active metabolite norfluoxetine are both present in the Broad Repurposing Hub library. Based on the results from the assay optimization stage, we expected at least 50% inhibition by fluoxetine at the 10 µM dose. However, what we observed was an underperformance for whatever reason. The average inhibition by fluoxetine was 7% (4.7% in replicate 1 and 9% in replicate 2). The average inhibition by norfluoxetine was -6.4% (technically slight channel activation) in replicate 1 and 3.4% in replicate 2, meaning no difference from control.

While there’s still justification for trying fluoxetine in a V434L single-patient study based on the successful V425M single-patient study, the V434L-focused drug repurposing screen produced one more surprise. There’s a safe, over-the-counter nutraceutical (marked by the yellow arrow on the plot below) that caused 112% inhibition of the mutant V434L Kv3.1. channel in both replicates. Now that the primary screen is done, the next and final stage of the screening campaign is hit validation and potency determination studies. We are also considering a test of selectivity, i.e., do the top hits inhibit WT channels as potently as V434L channels?

We’re excited to work with KCNC1 Foundation and a clinician to boot up a single-patient study for Eliana based on the above results. All eyes had been on fluoxetine until the Broad Repurposing Hub library screen was finished last month. We can only test one drug at a time in pioneer patients. And of course it’s possible that combo therapy will be superior to a monotherapy.

Onward for Eliana and KCNC1 kiddos like her!

Share Perlara’s Cure Odysseys