CHDI FOUNDATIONCHDI

Accelerating Therapeutic Development for Huntington's Disease

2025 Conference

CHDI’s 20th Annual HD Therapeutics Conference took place February 24 – 27, 2025 in Palm Springs, California. This unique conference series focuses on translational drug discovery and development for Huntington’s disease and draws participants from the biotech and pharmaceutical sectors as well as academia. The conference is intended as a forum where all participants can share ideas, learn about new disciplines, network with colleagues, and build new collaborative partnerships. We are indebted to all of the conference speakers, and especially grateful to those who are able to make their presentations available here for a wider audience.

3D genome architecture across the life span and in Huntington’s disease

Longzhi Tan, PhD, Stanford University

How do different cells in our nervous system develop different functions, and degenerate differently in aging and Huntington’s disease, despite having (approximately) the same genome? An emerging mechanism is 3D genome architecture: the folding of our 2-meter-long genome into each 10-micron cell nucleus. This architecture brings together genes and distant regulatory elements to orchestrate gene transcription and has been implicated in many neurodevelopmental and degenerative diseases including Huntington’s. However, genome architecture is extremely difficult to measure.

We developed a series of DNA sequencing-based technologies (Dip-C, LiMCA, Pop-C, vDip-C, and scMicro-C) and solved the first 3D structure of the human genome in a single cell. Applying these methods across the body, we made a surprising discovery that a cell’s functional type and developmental/aging state can be inferred from its genome structure alone, providing a new structural basis for development and aging. For example, in highly specialized sensory neurons, we discovered unique genome structures that functionally drive development. In the brain, we determined the dynamics of three facets of our genome—linear sequence, gene transcription, and 3D structure—after birth. We obtained the true spectrum of somatic mutations in the normal human brain and discovered a major transformation of both transcriptome and 3D genome in the first month of life in mice. Most notably, in the aging human and mouse brains, we discovered evolutionarily conserved life-long changes in 3D genome architecture, including the formation of ultra-long-range intra-chromosomal and specific inter-chromosomal contacts in cerebellar granule cells across the life span.

More recently, we discovered drastic, cell type–specific, genome-wide DNA mis-wiring in a knock-in mouse model of Huntington’s, Q140. In collaboration with William Yang’s lab at UCLA, we created the world’s first single-cell 3D genome atlas of Huntington’s mice, by sequencing 4,108 striatal cells across 16 mice. Q140 led to global changes in genome architecture, which can be nearly completely prevented by knocking out a top Huntington’s-implicated DNA mismatch-repair gene, Msh3. Our work provides a first look into the “black box” of 3D genome regulation in Huntington’s and offers new therapeutic opportunities for precisely rewiring DNA in 3D to revert neurodegeneration.

Top

Targeting arbitrary sequences of intrinsically disordered proteins

Kejia Wu, PhD, University of Washington

A general approach to design proteins that bind tightly and specifically to intrinsically disordered regions (IDRs) of proteins and flexible peptides would have wide application in biological research, therapeutics, and diagnosis. However, the lack of defined structures and the high variability in sequence and conformational preferences has complicated such efforts. We sought to develop two main computational methods combining biophysical principles with deep learning tools to readily generate de novo binders for any disordered sequence. Following distinct design principles, the two approaches showed robust performance at targeting disordered regions 1) with secondary structural preference and 2) without, in 50+ broadly diverse cases tested. A typical affinity range of <1-100nM was routinely achieved with dozens of designs tested. A subset of all by all binding experiments showed most designed binders were highly specific toward their intended targets. Co-crystal and NMR structures indicated designs were consistent with the models. Some designs were tested functional in multiple downstream applications. Our approaches thus could provide a general solution to the intrinsically disordered protein and peptide recognition problem. Utilizing the power of rational design, we now are targeting length-dependent poly-GLN patch from mutant HTT protein, with an idea of developing mutant HTT specific protein degraders.

Top

Targeting HTT1a: The pathogenic consequence of somatic CAG-repeat expansion

Gillian Bates, PhD, University College London

The mutation that causes Huntington’s disease (HD) is a CAG repeat in the HTT gene that encodes a polyglutamine tract in the HTT protein. The CAG repeat is unstable, expanding in certain cell types with age, such that a small proportion of cells contain expansions of hundreds of CAGs. It is now accepted that it is the rate of somatic CAG repeat expansion in brain that drives the age of onset and rate of disease progression. Expanded CAG repeats have direct consequences on the processing of the HTT pre-mRNA causing cryptic polyA activation in intron 1 to generate the HTT1a transcript that encodes the aggregation prone and highly pathogenic HTT1a (exon 1 HTT) protein.

We have used the allelic series of knock-in mice to model the molecular consequences of repeat expansion between 50 and 190 CAGs. We found that as the CAG repeat expanded to ~190, the level of the HTT1a protein increased whilst full length mutant HTT decreased to ~10% of wild type HTT levels. The HTT1a protein shuttles between the nucleus and cytoplasm, with soluble HTT1a being expelled from the nucleus because of a very potent nuclear export sequence. HTT1a is retained in the nucleus, because it forms high molecular weight aggregation complexes that cause transcriptional dysregulation.

It has yet to be resolved as to whether HTT lowering approaches should target full-length HTT, HTT1a or both isoforms. In collaboration with the Khvorova lab, we identified siRNAs that specifically target Htt1a (486 and 634) and evaluated these, alongside a potent siRNA that is specific for full-length Htt (10150). Wild-type and zQ175 mice were treated with 10150 or 634/486 alongside treatment controls at 2 months of age with treatment to 6 or 10 months, or at 6 months with treatment to 10 months. The siRNA potency and durability were most effective in the hippocampus. Targeting Htt1a was more effective at delaying HTT aggregation and transcriptional dysregulation than targeting full-length Htt, despite the greater potency of 10150. These data support HTT-lowering strategies that are designed to target the HTT1a transcript, either alone, or in combination with full-length HTT.

This work was supported by the CHDI Foundation.

Top

Nuclei with huntingtin aggregates have long somatic CAG-repeat expansions (>150 CAGs), phase C/D transcriptional pathology, and N-terminal HTT encoded by exon 1

Won-Seok Lee, PhD, Harvard Medical School & Broad Institute of MIT and Harvard

An important direction in Huntington’s disease research is to identify the toxicity mechanism of somatically expanded mutant Huntingtin. Early studies identified widespread nuclear aggregate formation in mouse models with very long CAG repeats, but the sparseness of such aggregates in brain tissue from human HD patients subsequently raised doubts about whether the aggregates were an essential feature of HD pathology in humans.

We recently discovered that the cell-autonomous component of transcriptional pathology in HD neurons (that is instructed directly by a neuron’s own CAG-repeat expansion) is present only in sparse individual cells with very long somatic expansions (>150 CAGs), and that individual neurons acquire this pathology asynchronously (and then are lost) rather than simultaneously. Since our model thus predicts that the key pathological features will in fact be sparse, it elevates the potential pathological significance of nuclear aggregates.

To determine whether HTT aggregation associates, at a single-cell level, with long somatic CAG-repeat expansions and gene-expression disturbances, we developed an approach for purifying nuclei with visible HTT aggregates from the brain tissues of persons with HD. Using this approach, we discovered that intranuclear aggregates are found in precisely the subset of HD-vulnerable neurons that have very long somatic CAG-repeat expansions (>150 CAGs) and exhibit transcriptional dysregulation of the form that we have recently described as “Phase C” and “Phase D” transcriptional changes. We found that these nuclei with aggregates also exhibited pervasive epigenetic dysregulation that may underlie their transcriptional pathology.

Biochemical analysis and immunofluorescence staining suggest that only N-terminal fragments of HTT, including the Exon1 HTT protein, form intranuclear aggregates. We also found evidence that Exon1 HTT may be generated at least in part by inefficient splicing in the nuclei with aggregates.
Our data suggest that intranuclear aggregates (i) can serve as a cytological marker of neurotoxicity in HD, (ii) may mediate the neuronal toxicity that results from long somatic CAG-repeat expansions, and (iii) strongly support our model in which long (>150) CAG-repeat expansion provokes pathological changes in the nucleus.

Top

Non-clinical testing of SPK-10001 for patients with Huntington’s disease

Liz Ramsburg, PhD, Spark Therapeutics, Inc.

Surgically delivered AAV-vectored gene therapies enable delivery of HTT-targeting miRNA to the basal ganglia and connected brain regions without significant spread to the blood or peripheral organs. We have developed a vectorized miRNA (SPK-10001) using a proprietary capsid (Spark100) which, when injected directly into the basal ganglia, induces durable dose-dependent reduction of total HTT for at least 12 months in non-human primates. SPK-10001 was well tolerated in a pivotal GLP-compliant safety study in non-human primates, with no adverse events up to the highest dose tested. The miRNA cargo of SPK-10001 binds downstream of exon 1, which we found to minimize changes in the abundance of non-target mRNA. The selected miRNA was processed with high fidelity and did not perturb endogenous miRNA processing or abundance. Consistent with other studies, the miRNA encoded by SPK-10001 induced consistent reduction in mHTT in rodent and sheep models of HD and exerted neuroprotective changes at relatively low doses without targeting the CAG repeat tract directly. In parallel, we assessed the abundance of mRNA arising from alternate splicing of exon 1 in HD patient brain samples across disease stages and found low expression of the HTT1a isoform compared to the full-length transcript, which correlated with disease severity, assessed by Vonsattel grading. These data suggest that SPK-10001 has the potential to benefit individuals with Huntington’s Disease and provide rationale for further development of SPK-10001.

Top

Single-molecule studies of the proteins and protein-DNA complexes responsible for mismatch repair and trinucleotide repeat expansion

Dorothy A Erie, MS, PhD, University of North Carolina

The DNA mismatch repair (MMR) system corrects DNA synthesis errors that occur during replication and is also involved in several other DNA transactions. MMR is initiated by the highly conserved MutS and MutL homologs. They are both dimers and contain DNA binding and ATPase activities that are essential for MMR in vivo. MMR is initiated by MutS homologs binding to a mismatch. Subsequently, ATP induces a conformational change in MutS homologs, which promotes their interaction with MutL homologs that preferentially nick the daughter strand. This MutL-induced nicking signals the initiation of excision and resynthesis of the newly synthesized DNA strand containing the incorrect nucleotide. Two MutS homologs with different but overlapping specificities are responsible for MMR in eukaryotes. MutS (MSH2-MSH6) preferentially recognizes base-base mismatches and small insertion deletion loops (IDLs), and MutS (MSH2-MSH3) preferentially recognizes larger IDLs. Interestingly, MMR itself can be mutagenic, with MutS promoting trinucleotide-repeat (TNR) expansions in genes associated with neurodegenerative diseases. Like MMR, TNR expansions also require the recruitment of MutL homologs by MutS. Notably, in vitro, MutL (MLH1-PMS2), which is the primary homolog in MMR, promotes both expansions and deletions of TNR insertions; whereas, MutL (MLH1-MLH3) preferentially promotes TNR expansions. Furthermore, mutation of the endonuclease site in MutL eliminates CAG expansion in Huntington’s disease mice, suggesting that MutL along with MutS are the major drivers of TNR expansion. Structural and biochemical studies indicate that the conformational dynamics and assembly states of the proteins and protein-DNA complexes are central to the regulation of MMR. To develop effective treatments for the associated cancers and TNR diseases, it will be essential to understand the molecular mechanisms that underlie these different processes. We use a combination of atomic force microscopy (AFM) and single molecule fluorescence to characterize the stoichiometries and the conformational and dynamic properties of MutS and MutL homologs and their assembly on DNA containing a mismatch or TNR hairpins. I will discuss conformational properties of the individual proteins, especially MutLα and MutL, which are highly dynamic, and I will present a molecular mechanism of the initiation of MMR and discuss how it may relate to TNR expansion.

Top

Multiple approaches to targeting MutSβ function in triplet nucleotide repeat expansion in HD

Brinda Prasad, PhD, CHDI

MutSβ, a key initiator of the DNA mismatch repair (MMR) pathway, is a heterodimeric ATPase composed of MSH2 and MSH3 subunits. MSH3 was identified in people with Huntington’s disease (HD) as a human genetic modifier of several clinical landmarks and of mHTT somatic instability in blood. Based on target assessment, knowledge available from genetic variants and phenotypic effects, mode of action, position in the MMR pathway, and therapeutic index profile, we have prioritized identification of small molecule modulators of MSH3/MutSβ with a goal of halting somatic expansion in HD.

We have pursued three approaches to modulate MSH3/MutSβ: (i) inhibiting MSH3’s ATPase activity, (ii) preventing protein interactions with MutL complex(es) needed for DNA resection and repair; and (iii) disrupting MutSβ recognition of DNA. We have identified compounds that are: (i) specific and potent orthosteric inhibitors of MutSβ ATPase activity which are active in cell-based assays; (ii) macrocyclic peptide inhibitors of the MutSβ and MutL protein interaction that target the MLH1 S2 site and inhibit mismatch repair in a nuclear extracts; and (iii) covalent binders of MSH3 that alter MutSβ interaction with a (CAG)1 extrahelical loop containing DNA and are active in cell-based assays. We are using a suite of biochemical, biophysical, cellular and animal model assays, combined with structural biology, to evaluate structure/function relationships of the MutSβ enzymatic cycle, mode of action of bound ligands, and genotype:phenotype relationships of genetic variants of MSH3/MutSβ to support hit identification and lead development. We will present our studies supporting advancement of MutSβ modulation as a candidate therapy for HD and, potentially, additional triplet repeat diseases.

Top

Using genome-editing tools to reveal novel aspects of DNA mismatch repair

Presentation not made available by presenter

Britt Adamson, PhD, Princeton University

Efforts to repurpose CRISPR-Cas systems have produced a suite of genome-editing tools, including programmable nucleases, base editors, and prime editors. These tools have greatly enabled the study of genomes and gene function, and their advancement to therapeutic development has demonstrated promise for addressing a host of unmet medical needs. Our understanding of how endogenous cellular processes influence the activity of these tools, however, lags behind their application and, due to the rapid pace of technology development, behind efforts to build new approaches. Our work focuses on identifying cellular determinants of genome-editing tools to better understand how they work. In pursuit of this goal, we consider that to introduce sequence changes, genome-editing tools must damage DNA. Development of genome-editing tools therefore requires a basic understanding of the cellular systems that sense and respond to DNA damage and presents new opportunities to study those basic cellular systems. Our results provide key insights into how genome-editing tools interact with the cellular environment, suggest general strategies for improvement of these important technologies, and reveal new aspects of cellular DNA damage response.

Top

Distinct mismatch repair complex genes set neuronal CAG-expansion rate and drive selective pathogenesis in HD mice

X William Yang, MD, PhD, University of California, Los Angeles

Genome-Wide Association Studies (GWAS) of Huntington’s disease (HD) reveal mismatch repair (MMR) genes as modifiers of age-at-onset. However, the mechanistic connections between MMR genes and selective neuronal pathogenesis in HD remains poorly understood. Here we crossed knockout (KO) alleles for 9 HD GWAS modifiers/MMR genes to the Q140 murine Huntingtin (mHtt) knock-in mouse model of HD. We found 4 KO mice strongly (Msh3 and Pms1) or moderately (Msh2 and Mlh1) rescue a triad of adult-onset, striatal medium-spiny-neuron (MSN)-selective phenotypes: somatic mHtt CAG-repeat expansion, transcriptionopathy, and mHtt protein nuclear aggregation. Comparatively, Q140 cortex also exhibits a much later-onset and progressive pathogenic triad that is Msh3-dependent. Msh3-deficiency reverses mHtt-induced open chromatin dysregulation in striatal neurons, which corresponds to distinct murine chromatin states and transcription factor binding sites. Remarkably, Q140/Msh3-KO lacks any visible PHP1+ or EM48+ mHtt aggregates in the striatum and cortex up to 20-month age. Moreover, Msh3-deficiency prevents striatal synaptic marker loss, astrogliosis, and locomotor impairment in HD mice, while Q140/Pms1-KO also has substantial reduction of mHtt aggregation load. Surprisingly in purified MSN nuclei, mHtt exhibits a fast rate of modal-CAG repeat expansion at a rate of +8.8 repeats/month (R2=0.98), which is slowed or stopped by Msh3 gene-dosage reduction. We computationally identify an MSN somatic DNA repeat expansion threshold for mHtt protein nuclear (about 150 CAG) and neuropil (about 192 CAG) aggregation. We also identify 479 genes that are significantly associated with somatically-expanded Modal-CAG length but not with age in Q140 MSNs, which could be mined for PET or fluid biomarker development. Together, our study reveals Msh3 and Pms1 (and likely their encoded MutSb-MutLb complex) are drivers of fast somatic mHtt CAG-expansion rate in HD-vulnerable neurons, which in turn elicit repeat-length/threshold-dependent, selective, and progressive pathogenesis in vivo. Moreover, our study provides a robust genetic model platform for dissecting the in vivo disease mechanisms and testing candidate therapeutics that could benefit HD and other repeat expansion disorders (REDs).

Top

RNAi-based modulation of MMR pathway and HTT expression as a path towards disease-modifying therapy for HD

Anastasia Khvorova, PhD, University of Massachusetts, Chan Medical School

Somatic repeat expansion (SRE), misregulated splicing/polyadenylation leading to HTT1a isoform expression, and mutant huntingtin (mHTT/HTT1a) production are critical processes driving Huntington’s disease (HD) pathogenesis.

Using divalent siRNAs—compounds enabling durable gene modulation in the CNS—we evaluated the role of mismatch repair (MMR) pathway genes in somatic repeat expansion (SRE) within the Q111 HD model. While results align with genetic studies, partial gene modulation revealed not only key factors but also their relative contributions to SRE. Potent protein silencing (~90%) of MSH2, MSH3, and MLH1 produced varied effects on SRE, with MSH3 and PMS1 emerging as optimal therapeutic targets based on safety, efficacy, and dose-response relationships.

We found that MutLγ (MLH3) and MutLβ (PMS1) subunits are required for expansion, while MutLα (PMS2) modulation increased expansion rates. Three models—proximity, multisubunit, and criss-cross—are proposed to explain these findings.

In the Q111 (120 CAG) model, robust SRE leads to mHTT aggregation and neurodegenerative transcriptomic changes by 12 months. A yearlong silencing of MSH3 results in blocking expansion reduced mHTT aggregation and reversed neurodegenerative signatures. While a yearlong HTT lowering alone had minimal effects, combining expansion blockade with HTT lowering synergistically abolished aggregates and enhanced transcriptomic rescue.

This study highlights somatic expansion blockade as a potential disease-modifying strategy for HD, with MSH3/HTT co-modulation offering further therapeutic promise. These findings provide a foundation for advancing RNAi-based clinical approaches targeting MSH3 and HTT.

Top

Non-MMR genetic modifiers: Targets or path to uncover events after CAG expansion?

Seung Kwak, PhD, CHDI

Exciting developments are underway within the HD R&D community, with multiple industry partners actively seeking to intervene the CAG expansion process in somatic cells. The focus on DNA mis-match repair (MMR) enzymes and the phenomenon of somatic trinucleotide expansion has been ushered by the initial identification of >7 human genetic modifiers of HD that coalesce in the MMR pathway, and further supported by observations in HD post-mortem brains that most affected cells in cortex and caudate harbor the greatest repeat expansion. While the notion that CAG length must expand beyond the inherited allele size for pathogenesis is now generally accepted, the question of what happens after CAG repeat expands remains unanswered. Hints for insight into toxicity mechanisms may reside on the 4 other modifier genes (RRM2B, CCDC82, TCERG1, and MED15) which, at first blush do not appear to participate in MMR. Among the possible avenues that exist to pursue this question, I will briefly describe our planned path at CHDI and call to the research community to help join the effort to answer this important question.

Top

Genetic modification of Huntington’s disease beyond DNA repair

Marcy MacDonald, PhD, Massachusetts General Hospital & Harvard Medical School

Genome-wide Association Studies (GWAS) of the timing of Huntington’s disease (HD) clinical manifestations revealed that HD pathogenesis involves sequential components, with somatic expansion of the HTT CAG repeat preceding damage to vulnerable neurons. These GWAS have identified genetic modifiers that implicate a mismatch-repair-related process in influencing the somatic CAG repeat expansion that drives HD pathogenesis. However, the GWAS strategy has also identified an almost equal number of modifier loci harboring genes not directly implicated in DNA repair. These non-mismatch repair modifiers do not converge on a single molecular process, offering multiple potential entrées to elucidate other components of pathogenesis that could provide targets for therapeutic intervention. The most intriguing of the non-mismatch repair modifiers is specific variation adjacent to the CAG repeat that could help to resolve the role of HTT in pathogenesis beyond somatic expansion.

Top

Insights into the causes and consequences of DNA repeat expansions from 700,000 biobank participants

Margaux Hujoel, PhD, Brigham and Women’s Hospital & Harvard Medical School

Expansions and contractions of tandem DNA repeats are a source of genetic variation in human populations and in human tissues: some expanded repeats cause inherited disorders, and some are also somatically unstable. We analyzed DNA sequence data, derived from the blood cells of >700,000 participants in UK Biobank and the All of Us Research Program, and developed new computational approaches to recognize, measure and learn from DNA-repeat instability at 15 highly polymorphic CAG-repeat loci. We found that expansion and contraction rates varied widely across these 15 loci, even for alleles of the same length; repeats at different loci also exhibited widely variable relative propensities to mutate in the germline versus the blood. The high somatic instability of TCF4 repeats enabled a genome-wide association analysis that identified seven loci at which inherited variants modulate TCF4 repeat instability in blood cells. Three of the implicated loci contained genes (MSH3, FAN1, and PMS2) that also modulate Huntington’s disease age-at-onset as well as somatic instability of the HTT repeat in blood; however, the specific genetic variants and their effects (instability-increasing or -decreasing) appeared to be tissue-specific and repeat-specific, suggesting that somatic mutation in different tissues—or of different repeats in the same tissue—proceeds independently and under the control of substantially different genetic variation. Additional modifier loci included DNA damage response genes ATAD5 and GADD45A. Analyzing DNA repeat expansions together with clinical data showed that inherited repeats in the 5’ UTR of the glutaminase (GLS) gene are associated with stage 5 chronic kidney disease (OR=14.0 [5.7–34.3]) and liver diseases (OR=3.0 [1.5–5.9]). These and other results point to the dynamics of DNA repeats in human populations and across the human lifespan.

Top

Defining ALS mechanisms at single-cell resolution

Aaron D Gitler, PhD, Stanford University

Amyotrophic lateral sclerosis (ALS) is a devastating human neurodegenerative disease caused by a selective loss of motor neurons from the brain and spinal cord. Intriguingly, in ALS, some motor neurons are vulnerable and others are resistant. We used single-cell transcriptomics and epigenomics to explore the heterogeneity of the spinal cord and to discover the changes that occur during ALS. We identified differentially expressed genes associated with degeneration as well as resilience. Our studies have revealed unexpected diversity in the autonomic nervous system, gradients of fast-and slow-firing motor neuron types within motor pools. These findings provide insight into mechanisms of degeneration, molecular underpinnings of selective vulnerability, and may suggest novel therapeutic strategies. We have also been extending studies on the adult human spinal cord, revealing similar logic as in mouse motor neurons.

We also use single cell approaches to define the RNA processing changes associated with ALS pathology. A hallmark pathological feature of ALS is the depletion of RNA-binding protein TDP-43 from the nucleus of neurons in the brain and spinal cord. We discovered that TDP-43 represses a cryptic exon splicing event in the ALS risk gene UNC13A. Loss of TDP-43 from the nucleus causes inclusion of a cryptic exon in UNC13A mRNA and reduced UNC13A protein expression. Our data provide a direct functional link between one of the strongest genetic risk factors for ALS UNC13A genetic variant and loss of TDP-43 function. Beyond cryptic splicing, we have also discovered loss of TDP-43 in ALS leads to widespread alternative polyadenylation changes, impacting expression of disease-relevant genes and providing evidence that alternative polyadenylation is a new facet of TDP-43 pathology.

Finally, we use simple model systems like yeast to screen for modifiers of ALS disease proteins TDP-43 and FUS and identified the polyglutamine protein ataxin 2 as a potent modifier. We then identified mutations in the ataxin 2 gene as genetic risk factor for ALS and have pursued studies of ataxin 2 as a therapeutic target. The approaches and insights we are learning for ALS will hopefully be applicable to other neurodegenerative diseases, including HD.

Top

HOX transcription factors toxicity in Huntington’s disease, a mechanism induced by mHTT in vulnerable neurons that then becomes self-sustained

Julien Mamet, PhD, Core Biotherapeutics, Inc.

Transcriptomic studies of vulnerable neurons in Huntington’s disease (HD) show transcriptional dysregulation is an early and central mechanisms of the disease, supporting the fact that mHTT triggers toxicity by altering the operations of key transcription factors (TFs). Since TFs operate as networks to control gene expression, we built the largest library of single-cell TF networks using data of >1 million cells from >11,500 patients across numerous diseases including HD. Analysis of these networks led to developing the Core Coefficient algorithm that measures the functional importance of each TF in a network, enabling the agnostic decoding of TF dysregulations driving diseases. For HD, we built networks of striatal and cortical cells from post-mortem data and transgenic mice. The Core Coefficient analyses of post-mortem medium spiny neurons (MSNs) networks from the caudate nucleus and the putamen of the striatum show that in HD, the HOX family of TFs take over MSN networks and become functionally dominant. HOX dominance is also observed in vulnerable pyramidal neurons of the motor cortex. Additional TFs are involved in the dysregulation process with discrete gain of functionality measured in HD networks depending on cell type, location or disease grade, including FOXD1, ONECUT1, SREBF2 and TBX5. HOXs lack functional importance in normal adult neurons, but in HD they systematically control the networks of vulnerable neurons and execute a neurodegeneration program by inhibiting thousands of genes required for the structural integrity of neuron compartments, including membranes, cell bodies, organelles, axons, dendrites, and synapses, and for their proper function, including excitability, conduction, cell body size regulation, calcium signaling or communication. Analysis of zQ175DN mice networks (snRNAseq source: GSE270728, CHDI study) shows that HOXs dominate in heterozygous animals, recapitulating humans. In these animals, lowering mHTT initially inhibits HOX dominance but the effect is only temporary and HOXs regain dominance despite the sustained inhibition of mHTT. This shows HOX toxicity is self-sustained once induced by mHTT and supports the need to inhibit HOXs independent of mHTT levels to stop neurodegeneration. Since HOXs lack functionality in normal adult neurons, blocking HOXs is anticipated to be therapeutic and safe in the treatment of HD.

Top

Rethinking Huntington: Untangling toxicity, aggregation, and repeat expansion

Don Cleveland, PhD, University of California, San Diego

Despite absence of consensus on the mechanism(s) of toxicity in Huntington’s disease (HD) from CAG-repeat expansions in the huntingtin gene, it initially seemed highly likely that diminishing synthesis of mutant huntingtin would directly target that toxicity if delivered to the key affected cells. Using antisense DNA oligonucleotides (ASOs) to catalyze rapid degradation of huntingtin mRNA, intraventricular infusion was demonstrated in mice and non-human primates to reduce huntingtin expression within cell targets thought to be central to HD pathogenesis. In HD model mice, this not only delayed onset of symptoms when administered early, but mediated sustained partial reversal of disease phenotype that persisted much longer than the mRNA knockdown. Disappointingly, when taken to human trial, however, there was no clinical benefit despite reduction in mutant huntingtin, perhaps because therapy was initiated too late, because the ASO reduced both mutant and wild type huntingtin, or – maybe – because primary toxicity was not from mutant huntingtin and we need to rethink what is pathogenic, including ongoing somatic repeat expansion!
To test this, we have now developed a spatial transcriptomics method (multi-modal MERFISH) that enables imaging (within tens of thousands of individual cells in an intact tissue section) of RNAs encoded by thousands of genes, immunolabeling for tens of proteins including huntingtin protein aggregates, and measuring CAG repeat number in each HTT RNA. Using repeated in situ hybridizations in intact human tissue sections, we have measured the expression levels of 1128 genes and quantified the extent of CAG repeat expansion using single-molecule FISH probes targeting CAG repeats in HTT RNAs. Excitatory neurons are found to be the cells with the highest CAG repeat expansion and those neurons are dispersed rather than clustered. Those excitatory neurons (both in normal and HD brain) also express the highest levels of HTT RNAs, consistent with somatic repeat expansion driven by HTT transcription. Furthermore, within the same cells, intranuclear inclusions of huntingtin increase as the CAG repeat expands (albeit inclusions do not require large somatic expansion). Taken together, these single-cell, multi-modal measurements of HTT RNAs, CAG expansion, and huntingtin protein inclusion define molecular events occurring in HD.

Top

Selective excision of pathogenic CAG repeats in HTT using CRISPR and novel delivery strategies

Ross Wilson, PhD, Innovative Genomics Institute at UC Berkeley

Allele-selective inactivation of the pathogenic HTT repeat expansion has emerged as a promising strategy for treating or preventing Huntington’s Disease (HD). The stringent selectivity of CRISPR enzymes makes genome editing an appealing candidate for an allele-selective intervention in HD. Such an approach could underlie a therapeutic intervention that disables the pathogenic HTT allele while leaving the healthy (wild type) copy of HTT intact. To this end, we have assembled an interdisciplinary team of genome editors, neurosurgeons, and macromolecular delivery experts to carry out pre-clinical research assessing the potential of an allele-selective, CRISPR-mediated intervention in HD.

We have established a high-efficiency genome editing strategy that will target mutant HTT and is not anticipated to impact wild type HTT. We are advancing novel delivery technologies in parallel, comparing viral and non-viral approaches for CRISPR delivery. Our viral approach features a novel capsid with exceptional affinity for neurons and an innovative strategy for encoding CRISPR components. Our non-viral approach relies on carrier-free CRISPR enzyme delivery with tethered peptides enhancing cell entry and neuron-specific targeting. These delivery and editing strategies are being evaluated in small animal models, and have been used to extend the life of HD model mice.

Our program also features off-target assessment, and our studies will soon progress into large animal models. Overall, we are optimistic about the possibility of a CRISPR-based therapy for HD.

Top

Precision genome editing of CAG repeats reduces somatic repeat expansions in Huntington’s disease models

Zaneta Matuszek, PhD, Broad Institute of MIT and Harvard

Trinucleotide repeat (TNR) diseases are neurological disorders affecting approximately 1 in 3,000 individuals worldwide. These diseases arise from genomic expansions of trinucleotide repeats, which become unstable in somatic cells in a length-dependent manner. The most common pathogenic triplet sequence, CAG•CTG, is implicated in nearly one third of all known pathogenic TNR loci, including the HTT gene responsible for Huntington’s disease (HD). Currently, no approved interventions effectively halt progression of this disease.

We explored genome-editing strategies targeting pathogenic trinucleotide repeats to reduce the repetitiveness of these repeats in patient-derived cells and animal models. Using cytosine base editors, we introduced G•C-to-A•T interruptions at CAG repeats, mimicking naturally occurring, stable, non-pathogenic alleles found in the general population. We demonstrated efficient base editing at repeat expansions in vitro, with proclivity for editing longer repeat alleles in patient-derived cells. In vivo, AAV9-mediated delivery of optimized base editors in the Htt.Q111 mouse model of HD resulted in efficient repeat interruption in transduced tissues, and significantly reduced somatic repeat expansions in the central nervous system.

Comprehensive evaluations of unintended targets of the CAG repeat base editing strategy, using CIRCLE-seq and whole-genome sequencing, revealed efficient repeat interruption at pathogenic alleles alongside a wide range of alternative on-target and off-target loci, highlighting the need for further cell-type specific and longitudinal safety assessments.

Our findings demonstrate that introducing interruptions in trinucleotide repeats of pathogenic length mitigates a key disease-associated feature of repeat expansions in HD models.

Top

Application of antiviral receptor TRIM21 to protein aggregation and expansion diseases

William McEwan, PhD, University of Cambridge

Protein aggregates represent challenging substrates for targeted protein degradation (TPD) techniques; they are poorly processed by the ubiquitin-proteasome system and it is difficult to specify selectivity for aggregate over monomers, whose role may be required for cellular function. Thus, despite the promise of PROTACs and other degrader modalities, there remain few validated methods for degrading aggregates relevant to neurodegenerative disease. In recent years, we have deployed the cytosolic antibody receptor TRIM21 to target cytosolic aggregated proteins, under the rationale that its ability to rapidly degrade virus particles would make its activity suitable for protein aggregates. Mechanistic studies demonstrate a requirement for clustering in order to stimulate TRIM21 E3 ligase activity, further suggesting substrate assembly-state could drive its activity. We have exploited these properties to develop antibody-based and genetically encoded degrader approaches. We have demonstrated that aggregates of Tau can be successfully marked for TPD using this approach. Cells expressing degraders are protected from seeded aggregation, whereas cells bearing pre-existing aggregates can be returned to an aggregate-free state. In animal models of Tau pathology, AAV delivery of degraders can protect against protein aggregation and extend symptom-free period of life. Moreover, the requirement for clustering means that aggregation and genetic expansion are properties that may be selectively targeted, sparing monomeric or normal-length counterparts. The results offer new methods that could be therapeutically exploited for the depletion of cytosolic protein aggregates.

Top

Integrating iPSC-based models, genomics, and machine learning to uncover neurological disease mechanisms and therapeutic targets

Ajamete Kaykas, PhD, insitro

We have developed an integrated platform for modeling neurological diseases, combining iPSC-based systems, functional genomics, and large-scale automation. This platform is powered by machine learning, enabling deeper exploration of disease mechanisms and accelerating both target identification and therapeutic discovery. In this presentation, I will share examples of our advancements, focusing on foundational model development and applications in Tuberous Sclerosis (TSC) and Amyotrophic Lateral Sclerosis (ALS). Key highlights will include the use of cutting-edge techniques such as optical pooled screening and Perturb-seq to identify targets and uncover novel biological insights. I will also delve into the design of robust data pipelines and production-ready machine learning models, illustrating their critical role in tackling the complexities of neurological diseases and discussing their potential application to Huntington’s disease.

Top

Discovery of RNA-targeted small-molecule therapeutics

Presentation not made available by presenter

Kathleen McGinness, PhD, Arrakis Therapeutics

RNA is upstream of nearly all biology and is both a unique and robust target for therapeutic intervention across many human diseases. Most current therapeutic agents that bind directly to RNA are either antibiotics blocking bacterial ribosome function or oligonucleotides with their attendant pharmaceutical limitations. We have built a platform to discover drug-like small molecules that penetrate cell membranes, bind to specific structures in RNA, and modulate RNA biology. This presentation will highlight some of the unique challenges of drugging RNA and offer some solutions to those challenges.

We have applied our platform to therapeutic discovery programs across many targets and disease areas. Among these, repeat-expansion diseases are uniquely attractive targets for RNA-targeted small molecules as RNA is at the root of the pathology in many of these diseases. Myotonic dystrophy type 1 (DM1) is caused by a trinucleotide (CUG) repeat expansion in the gene encoding myotonic dystrophy protein kinase (DMPK), resulting in the formation of nuclear aggregates of the repeat-containing RNA that bind and sequester splicing factors such as muscleblind-like splicing regulator 1 (MBNL1). Sequestration of critical splicing factors leads to global splicing abnormalities and widespread pathology. We have identified small molecules that selectively bind to the CUG repeat RNA and release bound MBNL1, both in biochemical assays and in patient-derived cells. We have further demonstrated that these compounds correct splicing defects in both cellular and animal models and completely reverse myotonia in the HSALR mouse model of DM1. This program establishes the viability of small molecules as therapeutics for repeat-expansion diseases and provides a foundation for exploration of RNA-targeted small molecule intervention in other diseases of this type.

Top

Deep visual proteomics: A spatial medicine approach to illuminate neurodegenerative diseases

Andreas Mund, PhD, University of Copenhagen & OmicVision Biosciences

Spatial proteomics is transforming our understanding of complex diseases by revealing cellular functions within their native tissue context. Deep Visual Proteomics (DVP), recognized as 2024 ‘Method of the Year’ by Nature Methods is the state-of-the-art workflow in spatial proteomics. We developed DVP by combining high-resolution digital pathology, AI-driven image analysis, automated laser microdissection, and ultra-sensitive mass spectrometry. This approach enables cell-type-resolved or single-cell proteome mapping directly in situ, preserving critical spatial information lost in traditional bulk or transcriptomic analyses. DVP directly probes cellular function by quantifying thousands of proteins with spatial precision. DVP’s impact has been demonstrated across a host of different diseases. In cancer biology, DVP revealed critical spatial insights into the tumor microenvironment. In a “cold” colorectal tumor, we identified a spatial barrier of immunosuppressive macrophages, which may impede T cell infiltration and offer a potential explanation for immunotherapy resistance. In a “hot” tonsil cancer, we observed significant proteomic heterogeneity among cancer cells influenced by their proximity to cytotoxic T cells and uncovered metabolic adaptations of T cells in hypoxic regions, offering new avenues to understand and modulate immunotherapy responses. Most importantly, DVP guided the successful development of a therapeutic strategy for toxic epidermal necrolysis, a life-threatening skin condition. By precisely identifying the role of the JAK-STAT pathway in specific immune cell populations, DVP enabled targeted treatment with JAK inhibitors, leading to patient full recovery in 8(8) patients to date. As a proof of concept for neurodegenerative applications, we studied morphologically distinct cell classes in their spatial context within basal ganglia FFPE samples. Using our state of the art DVP workflow, we quantified >6000 proteins per cell class, revealing phenotype-specific proteomes from archived human brain tissue. Given the spatial and cellular complexity of neurodegenerative diseases like Huntington’s (HD), DVP offers a powerful approach to advance understanding. By applying DVP to HD brain, we aim to map cell-specific protein alterations driving pathology, delineate spatial mechanisms of disease progression, and identify therapeutic targets. This is the promise of spatial medicine: a path to accelerate biomarker discovery and develop targeted therapies based on human data, for this debilitating disease.

Top

HD biomarker discovery in the HDClarity cohort: Using proteomics and machine learning to predict HD-ISS stages

Jim Rosinski, PhD, CHDI

The discovery of new biomarkers for HD is a constant focus at CHDI. Whether they be for pharmacodynamic effect, stratification, progression monitoring or toward other goals, there are significant gaps in the HD biomarker space. CHDI has undertaken a task to leverage the HDClarity cohort samples along with multiomic profiling to aid the community’s efforts to discover and validate biomarkers of HD. In this work, we will detail the extensive analytical efforts that CHDI has undertaken to mine proteomics data acquired on the Somalogic and Olink platforms from around six hundred serum and CSF samples obtained in the HDClarity study. The HDClarity samples proved to be the highest quality with nearly all providing reliable results and absolute minimum levels of contamination from other fluids. Statistical analysis of the data focusing on proteins that change between stages of disease uncovered significant differences in protein levels in the CSF and serum of participants. Going one step further and applying machine learning techniques, ranging from traditional decision trees through modern gradient boosting machines, reveals a strong capacity for this proteomics data to predict HD stage, but also a very high level of baseline structure to the dataset even with randomized stages. Looking forward, this promising start with the first six hundred participants in the HDClarity cohort can now be validated using the over four hundred participants that have joined in the intervening years, giving us a truly independent validation cohort.

Top

cfDNA methylation as a potential liquid biopsy marker for HD

Leslie M Thompson, PhD, University of California, Irvine

Alterations in promoter DNA methylation play a crucial role in gene regulation. Specifically, promoter hypermethylation often acts as a surrogate for gene silencing, while hypomethylation leads to gene activation. DNA methylation also exhibits several distinctive characteristics: it can be easily detected in clinical specimens such as blood, which is relatively non-invasive, remains stable in samples over time, and is specific to certain cell types. These characteristics of DNA methylation, coupled with its strong correlation with gene expression, make it an ideal candidate for biomarker development or human fluid based “liquid biopsy” for use in early detection and prognosis of HD. As a proof-of-concept study, we isolated cfDNA from 1 mL plasma from 10 control, 10 late pre-manifest and 10 manifest HD (CAG range from 42 to 52) patients and followed with targeted enzymatic methyl-sequencing (EM-seq) of ~4 million CpG sites. Circulating cfDNA primarily consists of nucleosome-associated fragments that largely retain the chromatin structure information from the originating cell. Due to the endonuclease degradation of cfDNA, nucleosome regions are better protected against enzymatic degradation, whereas open chromatin regions are more susceptible to endonuclease activity. Since EM-seq preserves the integrity of cfDNA, such as its fragment pattern as compared to bisulfite conversion, we observed a peak around 166 bp in the cfDNA fragment length distribution. This corresponds to the length of DNA associated with a nucleosome and a linker histone. We were able to identify 8436 CpGs that show significant differences between symptomatic/manifest HD and control, 6006 CpGs between pre-manifest and control and 9197 between HD premanifest and manifest patient cfDNA. These groups can be separated when using the top 1000 differential methylated CpGs (ranked by adjusted-P value). Patients with HD also showed a preferential pattern of CC dinucleotides 4-mer end motifs (CCCA, CCCT, CCTT and CCTA), which is associated with elevated activity of Dnase1l3 gene. Furthermore, we also identified a set of CpG sites whose methylation status significantly correlated with total motor score. Our results reveal the potential of cfDNA-derived methylation as a liquid biomarker and monitor of disease progression. Validation will be carried out on expanded cohorts in future studies.

Top

Multimodal neuroimaging in Huntington’s disease: Preliminary findings from the iMarkHD study

Manuela Moretto, PhD, University of Padova & King’s College London

This presentation highlights preliminary results from the iMarkHD study, which aims to identify novel in-vivo neuroimaging biomarkers for Huntington’s disease (HD) progression and symptoms. The study employs a multimodal neuroimaging approach, combining four molecular positron emission tomography (PET) tracers—[11C]IMA107 (phosphodiesterase-10A activity marker), [11C]MePPEP (cannabinoid 1 receptor marker), [11C]MDL100907 (serotonin 2A receptor marker), and [11C]MK-8278 (histamine 3 receptor marker)— with advanced magnetic resonance imaging (MRI) techniques to assess brain volume, diffusivity, cerebral blood flow, iron deposition, and functional connectivity.

The study aims to recruit 108 participants, including 72 people with HD (PwHD) and 36 healthy controls (HC). Participants undergo 10 study visits, including screening, clinical, and MRI assessments, as well as PET imaging at baseline, year 1, and year 2.
This presentation focuses on dynamic PET imaging results and highlights from multimodal MRI in premanifest PwHD (N~15, HD-ISS stage 0/1) compared to age- and sex-matched HC (N~20).
We analyzed non-displaceable binding potential (BPND) for [11C]IMA107 and [11C]MDL100907 and the volume of distribution (VT) for [11C]MK8278 and [11C]MePPEP in predefined brain regions. Group differences in tracer binding values were assessed using ANOVA tests, adjusting for covariates like age, sex, body weight, and regional volumes.

Preliminary results showed no demographic differences between groups. In PwHD, phosphodiesterase-10A binding was significantly reduced in the striatum (caudate, putamen, nucleus accumbens), pallidum (globus pallidus), and substantia nigra. Serotonin 2A receptor binding increased significantly in the amygdala in PwHD. Histamine 3 receptor binding was significantly reduced in the striatum (caudate, putamen, nucleus accumbens), and in the pallidum (globus pallidus), consistent with post-mortem evidence of histaminergic disruption in PwHD. Cannabinoid 1 receptor binding showed a decreasing trend in the caudate, nucleus accumbens, cerebellum, and insular cortex, with a significant reduction in the substantia nigra in PwHD compared to HC. In line with the literature, we also observed a significant volumetric decrease in the caudate, putamen, and pallidum in PwHD compared to HC.

These preliminary results revealed widespread and early changes across several receptors and PDE10A in PwHD. While some of these changes are in line with previous human and animal studies of the disease, we also reported novel findings on the widespread decrease in H3R, only supported by post-mortem studies showing disruption of the histaminergic system in PwHD.

Top

Allele-selective lowering in Huntington’s disease: From bench to bedside

Peter McColgan, MD, PhD, F. Hoffmann-La Roche Ltd

Heterozygous single nucleotide polymorphisms (SNPs) in phase with the CAG repeat on the mutant Huntingtin gene allow for allele-selective suppression of the pathogenic HTT transcript. Preferential silencing of the mutant HTT allele, preserving wild type HTT expression, may minimize potential loss-of-function effects and yield greater therapeutic benefit than non-selective suppression of both HTT copies. In this presentation we will share preclinical data from our allele-selective huntingtin-lowering program. Using a fully humanized HD mouse model, we demonstrated sustained allele-specific reduction of mutant huntingtin while preserving the wild-type mRNA and protein. In order to understand the geographic prevalence of heterozygotes for our target SNP, we have leveraged genotyping data from the previous tominersen Phase III study GENERATION HD1. To accurately identify individuals who could benefit from our approach, we have developed a novel SNP phasing assay capable of identifying distant intronic and exonic SNPs in phase with the expanded CAG mutation, paving the way for identifying multiple SNPs in parallel. Finally we will share details of our ongoing epidemiology study where we are using this assay to identify the geographic prevalence of individuals with the target SNP in phase with the expanded allele.

Top

Fluid biomarkers in HD: A confluence of goals

Hilary Wilkinson, PhD, CHDI

Fluid biomarkers play a critical role in improving the design and supporting critical decision making in clinical trials for HD therapeutics. These biomarkers can guide patient selection, assess drug pharmacodynamics, and detect early changes before clinical symptoms become apparent. However, there remains a significant unmet need for robust and translatable biomarkers, especially for early HD stages (HD-ISS 0–2). I will provide illustrative examples of how we are studying high-value, translational biomarkers to support HD clinical trials, focusing on discovery, characterization, and application in preclinical and clinical stages. CHDI is establishing a formal biobank infrastructure to further support research and I will discuss the ongoing challenge of obtaining, curating, and sharing high quality data and biosamples for HD community research. This framework highlights CHDI’s comprehensive approach to biomarker discovery and development.

Top