Targeting MSH3 To Slow Huntington's Disease Progression

A conversation between P. Peter Ghoroghchian, MD, Ph.D., CEO, Latus Bio, and Morgan Kohler

Huntington’s disease (HD) is an inherited neurodegenerative disorder affecting more than 100,000 patients in major gene therapy markets. Extensively validated by human data, MSH3 is the key pathogenic driver of the somatic repeat expansion that results in HD onset and progression.

Latus Bio published a study in Nature Communications that integrates a predictive computational model of HD, built on human data, with validating preclinical data in support of LTS-201, the company’s investigational AAV gene therapy designed to reduce MSH3 expression in striatal and cortical neurons that are vulnerable to degeneration in HD. These findings support lowering MSH3 to slow somatic CAG-repeat expansion. The model further predicts prospective benefits from LTS-201, which will be tested in a forthcoming clinical trial. P. Peter Ghoroghchian, MD, Ph.D., CEO of Latus Bio, details more about these findings.

What was the key scientific evidence that led Latus Bio to prioritize MSH3 as a therapeutic target for Huntington’s disease, and how does reducing MSH3 address disease progression differently from directly targeting the huntingtin (HTT) gene or protein?

The strongest evidence comes from human studies, including genome-wide association studies, transcriptional analyses from postmortem brains, and investigations of cultured neurons generated from HD patient-derived induced pluripotent stem cells. While HD is inherited as an expansion of CAG repeats in the HTT gene (usually >36 repeats), neurodegeneration does not actually begin at these inherited lengths. Over an HD patient's lifetime, specific neurons in the deep brain (striatum) and in the cerebral cortex continuously expand this repeat tract. Importantly, these studies have shown that such somatic expansion must surpass a strict threshold to trigger severe, pathological gene expression changes and subsequent neuronal cell death. These research studies strongly validate therapeutic strategies aimed at halting or slowing DNA repeat expansions, including targeting the mismatch repair enzyme MSH3 that is the central driver of the process, before cells can hit the fatal CAG expansion limit.

Unlike other clinical-stage, experimental HD therapeutics that aim to directly lower HTT levels, our approach focuses on reducing MSH3 expression to slow or prevent the accumulation of additional repeats in vulnerable neurons, potentially preserving their function and delaying disease progression. Importantly, this approach does not require altering the expression levels of the HTT gene itself. Our computational modeling of human disease and preclinical findings suggest that MSH3 reduction may significantly suppress somatic instability, providing strong mechanistic rationale for LTS-201 as a potential disease-modifying therapy for both symptomatic and premanifest patients.

Your computational model links the degree of MSH3 knockdown and the percentage of medium spiny neurons (MSNs) targeted to predicted clinical benefit. How did you validate the model against human natural-history data, and which assumptions most strongly influence its predictions?

Our computational model integrates human genetics, human somatic CAG-repeat expansion rates, and longitudinal natural-history data from patients in the Enroll-HD database. It simulates repeat expansion within the population of medium spiny neurons over a patient's lifetime and connects percentages of healthy neurons and their survival to established Huntington's disease milestones, including biomarker changes, motor symptom onset, progression through disease stages, and changes in the composite Unified Huntington's Disease Rating Scale (cUHDRS).

We assessed the model's consistency through repeated simulations and sensitivity analyses across inherited CAG-repeat lengths, ages at intervention, and treatment parameters. The most influential assumptions include the rates of somatic expansion, the CAG-repeat thresholds associated with neuronal dysfunction, and the relationships between percentages of targeted MSNs, MSH3 reduction levels, and suppression of somatic instability.

The model is grounded in human biological and clinical data sets, but its predictions of treatment benefit remain prospective. In particular, the relationship between preservation of healthy neurons and clinical outcomes will require validation in patients. The model provides a quantitative framework to guide therapeutic development rather than a definitive prediction of individual patient outcomes.

The model suggests that deeper MSH3 knockdown in a smaller proportion of neurons can be more beneficial than more modest knockdown across a larger population. What biological mechanisms explain this finding, and how is it informing your target product profile for LTS-201?

One of the most important insights from our modeling is that the depth of MSH3 suppression within individual neurons may matter more than simply maximizing the number of neurons reached. Somatic CAG-repeat expansion is a progressive process, and neurons can accumulate increasingly long repeats that eventually compromise their function and survival.

Modest MSH3 reduction may slow this process without sufficiently preventing neurons from reaching pathogenic repeat lengths. In contrast, deeper suppression could substantially reduce the rate of expansion, potentially allowing treated neurons to remain functional for much longer.

This finding has important implications for LTS-201. Our goal is to achieve robust, durable MSH3 knockdown in a therapeutically meaningful proportion of vulnerable neurons. Our model identified approximately 50% MSH3 knockdown in 50% of MSNs as a potentially meaningful benchmark across a broad patient population. Encouragingly, our nonhuman primate studies demonstrated 48% to 94% MSH3 mRNA reduction in >90% of striatal MSNs, supporting the potential to surpass the levels of target engagement predicted to provide clinical benefit.

LTS-201 uses an engineered AAV-DB-3 capsid to deliver an artificial microRNA targeting MSH3. What were the most important considerations in selecting the microRNA sequence and scaffold, and how did you evaluate its potency, processing accuracy, and potential off-target effects?

Our selection process focused on three priorities: potent MSH3 suppression, efficient microRNA processing, and minimizing unintended effects on other genes. We used computational sequence-design tools to identify candidates targeting human, nonhuman primate, and murine MSH3, followed by experimental screening of their activity and specificity in cultured cell lines (human, monkey, mouse).

Two candidates, different MSH3 targeting sequences, demonstrated promising activity. We evaluated both within two microRNA (miRNA) scaffolds and found that one scaffold produced stronger in vivo MSH3 knockdown in mouse striatal tissue. The overall MSH3 miRNA candidates also demonstrated high processing accuracy and favorable guide-to-passenger strand ratios in both mouse brains and in cell lines (murine and human), helping reduce the potential for unintended RNA interference.

We further assessed off-target activity using transcriptomic analyses in Huntington's disease patient-derived neurons. The selected MSH3 miRNA candidate demonstrated a limited off-target transcriptional signature, with no significant changes in DNA mismatch repair or cancer-associated signaling genes in the reported analyses. These findings supported its advancement into our nonhuman primate and disease-model studies.

In nonhuman primates, LTS-201 achieved 48% to 94% MSH3 mRNA knockdown in striatal MSNs, depending on dose. How do you interpret the relationship between vector dose, target engagement, and protein reduction, particularly given the differences observed between bulk-tissue measurements and cell-specific analyses?

Our nonhuman primate studies demonstrated a clear dose-dependent relationship between LTS-201 administration and MSH3 mRNA reduction within medium spiny neurons. At low, intermediate, and high vector doses, we observed approximately 48%, 65%, and 94% reduction, respectively, in cytoplasmic MSH3 mRNA in >90% of analyzed striatal MSNs.

The distinction between bulk-tissue and cell-specific measurements is particularly important. Bulk-tissue analyses measure the activity of multiple cell populations, including non-neuronal cell types that are not transduced by our neuronal-specific capsid variant. As a result, bulk measurements can be misleading with respect to the degree of target suppression achieved in the cell types that are most relevant to Huntington's disease (MSNs and cortical layer 5/6 neurons).

We, therefore, used cell-specific RNAscope analysis to quantify MSH3 reduction directly within MSNs. We also observed MSH3 protein reduction in bulk tissue, consistent with downstream target engagement in the study. The magnitude of bulk protein reduction, while not uniformly dose dependent, surpasses the measured reduction in mRNA at each analyzed dose level.

Together, these results reinforce the importance of measuring target engagement in the relevant neuronal population rather than relying exclusively on whole-tissue averages. Longer-duration studies will help further characterize the relationship between cell-specific mRNA suppression, protein reduction, and sustained pharmacological activity.

The study found dose-dependent reductions in somatic instability in HdhQ111 knock-in mice, even though Msh3 mRNA knockdown appeared to plateau at higher doses. What do these results reveal about the relationship between MSH3 mRNA, MSH3 protein levels, and CAG-repeat expansion? Which biomarker do you consider most informative for assessing pharmacological activity?

These findings highlight why it is important to look beyond bulk mRNA measurements alone when evaluating therapeutic activity. Although bulk-tissue Msh3 mRNA reduction appeared to plateau at higher doses, bulk Msh3 protein levels continued to decline, accompanied by progressively greater reductions in somatic CAG-repeat expansion.

At the highest dose tested in HdhQ111 knock-in mice, we observed approximately 40% bulk Msh3 protein reduction and a corresponding 46% reduction in the somatic instability index. This suggests that bulk MSH3 protein levels may provide a more informative measure of functional target suppression than bulk mRNA measurements.

Ultimately, we believe the most compelling evidence of pharmacological activity is the reduction in somatic instability itself, because that is the pathogenic process that LTS-201 is designed to address. Our approach is, therefore, to evaluate multiple complementary measures, including cell-specific MSH3 mRNA suppression, protein reduction, and CAG-repeat instability, to establish a comprehensive understanding of target engagement and biological activity.

MSH3 is involved in DNA mismatch repair, raising questions about the consequences of sustained target suppression. What evidence would you need to establish that long-term MSH3 reduction can slow pathogenic CAG-repeat expansion without compromising essential DNA repair functions or increasing genomic instability elsewhere?

Safety is a central consideration in developing a therapy designed to provide sustained MSH3 suppression. Although MSH3 participates in DNA mismatch repair, human genetic evidence provides important reassurance that reduced MSH3 expression can be achieved without added risks for carcinogenicity. In subjects who are born with complete (biallelic) loss of MSH3, there are, however, higher rates of recessive adenomas in the colon; this highlights the potential need for tissue (brain)- and, ideally, target cell (neuronal)-specificity for therapeutic strategies aiming to reduce MSH3.

Our preclinical studies suggest that MSH3 reduction by LTS-201 was largely confined to target neurons and not quantifiably detected in most tissues outside the CNS, providing encouraging initial findings that could support tolerability. In Huntington's disease patient-derived neurons, MSH3 suppression did not produce significant changes in the expression of other key mismatch repair genes or cancer-associated signaling pathways in the reported analyses. However, these findings do not establish long-term genomic safety.

We recognize that additional studies are necessary to characterize the consequences of sustained MSH3 reduction, including potential effects on DNA repair activity, genomic stability, and neuronal health over time. Our objective is to establish a therapeutic window that provides meaningful suppression of pathogenic CAG expansion while maintaining an acceptable long-term safety profile.

The paper reports localized inflammation in the globus pallidus at the highest dose in nonhuman primates, although the study found no increase in glial fibrillary acidic protein (GFAP) expression and no clear evidence of broader toxicity in the reported assessments. How are you distinguishing potential effects of the AAV capsid, the delivery procedure, and MSH3 knockdown itself as you evaluate the therapy's safety profile?

In our nonhuman primate study, we observed a localized increase in a marker of microglial activation in the globus pallidus (injection site) at the highest examined dose of LTS-201. Importantly, we did not observe increased GFAP expression, evidence of increased cerebrospinal fluid cellularity, nor abnormalities in the reported blood chemistry assessments.

Our priority in the current study was to identify a dose that achieved robust target engagement in vulnerable neurons while minimizing unnecessary vector exposure and potential inflammatory responses. Localized inflammatory responses in tissue can occur following intraparenchymal AAV administration, particularly near the infusion site. It is important, however, to not attribute these findings to any single factor without further investigation. Distinguishing effects associated with the surgical procedure, vector delivery, and MSH3 suppression requires controlled studies and longer-term assessments. Subsequent evaluations of LTS-201 have considered dose dependence, regional distribution, inflammatory markers, and tissue pathology, alongside appropriate vector and procedural controls; we look forward to sharing a comprehensive update on these studies soon. While findings from preclinical investigations are encouraging, longer-term clinical assessments are essential to establishing the safety profile of LTS-201.

The model predicts that earlier intervention could provide greater benefit, but it also suggests that treatment later in the disease course may remain meaningful. How are you using these predictions to define the therapeutic window for LTS-201, and what preclinical experiments could help establish whether treatment can preserve neuronal function after substantial CAG-repeat expansion has already occurred?

HD progression results from the continued loss of healthy neurons over time. Our model predicted the putative benefits from treatments that reduce the rate of somatic expansion, predicting delays in the time points by which populations of healthy neurons pass critical CAG-repeat length thresholds, whereby they become pathogenic and eventually die. The model showed that irrespective of inherited genetics, earlier intervention has the greatest potential to preserve populations of healthy neurons over time. Simulations also suggested, however, that meaningful benefits remain possible throughout the disease course, particularly in preserving the survival of remaining neurons and delaying model-predicted thresholds that are matched to disease stage progression. This is an important distinction because Huntington's disease progresses heterogeneously; not all neurons reach advanced stages of pathology simultaneously. Even in patients with established symptoms, a large proportion of vulnerable neurons may remain capable of benefiting from reduced somatic expansion.

To further define the therapeutic window, studies that initiate treatment at different stages of disease progression will be particularly informative. They should evaluate not only CAG-repeat dynamics but also neuronal survival and functional outcomes. Our goal is to understand how effectively LTS-201 can preserve neuronal function across different stages of Huntington's disease, helping inform future clinical development and patient selection.

What are the most important experiments needed to strengthen the translational case for LTS-201 before clinical development? In particular, how do you plan to establish that reductions in somatic instability translate into durable preservation of neuronal function and disease-relevant outcomes, rather than changes in CAG-repeat measurements alone?

Our published findings established an important proof of mechanism: LTS-201 achieved robust MSH3 suppression in the neuronal populations most vulnerable in Huntington's disease, and reducing MSH3 led to dose-dependent suppression of somatic CAG-repeat expansion in a disease model of HD.

The next critical step is strengthening the connection between these molecular findings and meaningful preservation of neuronal function. This will require longer-term studies, evaluating the durability of LTS-201 to sustain MSH3 knockdown, and correlate this activity to sustained reductions in somatic instability. By combining human genetic evidence, predictive computational modeling, and rigorous preclinical evaluation, we aim to build a strong translational foundation for LTS-201 as a potential disease-modifying therapy. Prospectively, in clinical studies, we believe it is particularly important to assess whether limiting CAG expansion preserves neuronal survival and functional measures relevant to Huntington's disease progression. Additional studies examining therapeutic activity across different disease stages will also help define the optimal timing of intervention.

About The Expert

P. Peter Ghoroghchian, Ph.D., is a physician-scientist, bioengineer, and serial biotechnology entrepreneur. He is the CEO of Latus Bio, a company developing scalable gene therapies. He was previously the CEO/cofounder of Ceptur Therapeutics, head of therapeutic development at Repertoire Immune Medicines, CTO of Poseida Therapeutics, and CSO/cofounder of Vindico NanoBioTechnology. Earlier in his career, he was an oncologist at the Dana-Farber Cancer Institute (DFCI) and led a research lab at MIT. He earned his BS from MIT, his MD/Ph.D. from Penn, and completed training in internal medicine at the Brigham and Women’s Hospital and in hematology/oncology at DFCI and MGH.