← Writing

Mechanisms of Genetic Anticipation in Huntington's Disease

Huntington's diseasegeneticsDNA repair

In 1872, a 22-year-old physician named George Huntington described to the Meigs and Mason Academy of Medicine in Ohio a hereditary condition he had observed. This condition moved through families on Long Island whom he had known since childhood, and whom his father and grandfather had treated before him. Across three generations, the Huntingtons had documented something George Huntington himself skimmed over: once this disease appeared in a family, it passed to roughly half the children and never skipped a generation. Behind this clear Mendelian inheritance pattern, a further pattern was lurking — in many of these families, the disease arrived earlier, and more severely, with each generation. It took more than a century, and the discovery of an expanding CAG repeat in the HTT gene, to recognize this as genetic anticipation, where the mutation grows as it is inherited [15]. It took even longer to answer the question this essay addresses: why should a DNA sequence expand at all, and consistently in a biased direction?

Any candidate mechanism has to satisfy several empirical observations. First, the mechanism is directionally error-prone: it is not rigid, since contractions do occur, especially in maternal transmissions, but the expected number of expansions outnumbers contractions. So the mechanism can't simply be damaging to the tract — that would change both directions equally — nor can it be one that only ever invokes expansion; this will matter later when discussing how mechanisms can flip contractions into expansions. Second, the rate of expansion is not linear: longer alleles expand more, and more often, so the mechanism isn't a constant copying error but something about the tract itself becoming a better substrate as it lengthens. Third, the pathological threshold in Huntington's is more discrete than continuous. Below 27 repeats, alleles are stable across generations. From 27–35, they rarely cause symptoms but can expand in future generations. Above 36–40, the allele is pathogenic. So the mechanism must explain not just why longer is worse, but why a silent length can turn into a deadly one. Fourth, expansions are far more frequent and far more severe when inherited from the father — repeat counts above 60 are almost never inherited through mothers. This demands some sex-specific explanation, perhaps related to differences between spermatogenesis and oogenesis.

This essay argues that genetic anticipation in HD is best explained as a repair-driven event. The cell's own mismatch- and excision-repair machinery is biased toward expansion and concentrated in the male germline [4, 9]. Replication, local sequence context, and modifier genes all play a role in the outcome, but the causal mechanism is repair.

Intrinsic Instability of CAG/CTG

Before asking which cellular pathways drive expansion, we need to explain why the HTT repeat is an unusually reactive DNA substrate in the first place. CAG tracts can fold back on themselves into self-complementary hairpin structures rather than the normal duplex form. The complementary strand, made of CTG repeats, also forms fold-back structures; mismatched A–A and T–T pairs create the imperfect geometry that produces the hairpin-like conformation. Repetitive sequences additionally allow slipped-strand DNA formation, where complementary strands anneal out of register and leave out looped repeat units — and if a secondary structure forms on the newly synthesized strand, that favors expansion, while structure on the template strand favors contraction (more on this later) [12, 14]. This hints at why length matters: longer repeat tracts increase the likelihood of hairpin and slipped-strand formation, and the pathological thresholds exist because short CAG tracts don't fold into stable secondary structures, whereas as the tract lengthens it crosses a thermodynamic threshold and begins forming stable hairpins and slipped-strands far more readily. There is also some newer work on structures like R-loops and G-quadruplex-related conformations forming from repeats, though I couldn't find evidence that these relate specifically to HTT [9].

Replication-Based Mechanisms

Once CAG/CTG tracts were recognized to form these structures, the most intuitive explanation was that instability occurs during replication fork progression, where a looped-out nascent strand could be copied into a permanent expansion — and the earliest mechanistic models of Huntington's Disease were replication-based, with two main theories.

First, during replication, repeat tracts can misalign when the nascent and template strands briefly dissociate and reanneal out of register, creating a stable hairpin that is incorporated into the new strand, increasing its length and its repeat count. This is particularly likely on lagging strands, since discontinuous replication repeatedly exposes single-stranded repeat DNA.

Second, Okazaki fragments are normally joined by the enzyme FEN1, which cuts off the flap at the end of the last fragment — the displaced 5′ end of the downstream fragment — to allow smooth ligation. FEN1 recognizes a single-stranded substrate, but when that strand is made of CAG repeats, the dangling strand can fold back into a hairpin that FEN1 no longer recognizes. The flap is not removed, the extra repeat units are not trimmed away, and the tract grows.

Both models, however, suffer critical limitations relative to the four criteria above. The first model is directionally symmetric — out-of-register annealing can happen on either the template or the nascent strand — so it doesn't explain why repeat counts generally go up in Huntington's patients. The second model fails in the opposite way: if the problem is that FEN1 can't cut flaps, there is no explanation for the contractions seen in some patients. It is entirely expansion-leaning and doesn't account for the observed symmetry. Both models also share a fatal, somatic limitation: Huntington's expansion occurs in post-mitotic, non-replicating tissues like neurons, where there isn't an active replication fork at all [9, 14]. Replication models simply cannot explain this.

Repair-Based Mechanisms

The field consequently shifted toward DNA repair pathways, an interpretation heavily supported by human and mouse data [4, 9]. DNA mismatch repair (MMR) is the most strongly supported pathway linking HD to repeat instability [4, 6]. Normally, MMR is designed to remove errors; in Huntington's disease, however, it counterintuitively drives CAG-repeat expansion. MMR runs in two stages: recognition and processing. Recognition is carried out by the MutS complexes — MSH2 pairs with MSH6 to make MutSα, and MSH2 pairs with MSH3 to make MutSβ. MutSα recognizes single base-base mismatches and very small loops, while MutSβ recognizes larger insertion/deletion loops (large bulges). Once MutS has bound, it recruits MutL complexes, which partner with PMS2, PMS1, and MLH3 to form MutLα, β, and γ respectively; MutLα carries the endonuclease activity that cuts the mismatched stretch so a polymerase can resynthesize the correct base pairs. In Huntington's disease, an expanded CAG tract can extrude into a slipped-strand or hairpin structure (see above), and MutSβ binds this loop as if it were a normal mismatch. This recruits MutL, which is where the process breaks down: because the tract is repetitive and symmetric, MutL appears to lack reliable strand discrimination and cannot identify the loop as the "error" — though the precise mechanism of this failure remains an area of active, uncertain research. DNA polymerase then copies across the hairpin as if it were the template strand, and ligation seals the longer product. In the next round of replication, the extra repeats become part of the template, and the tract expands [14]. Consistent with this, FAN1 — a gene associated with stabilizing the repeat and restraining expansion by limiting MSH2 recruitment to MSH3 — is associated with later onset of Huntington's in GWAS studies, suggesting repair is indeed causal [6, 12, 14]. Studies have also shown that MSH3 suppression stalls CAG-repeat expansion, and that MSH2 is required for somatic instability of the CAG repeat [1, 11].

A complementary, less well-supported model involves changes to Base Excision Repair (BER), the pathway that fixes small, everyday chemical damage to individual bases, most often caused by oxidative damage that accumulates with age. Here, an oxidative lesion forms in or near the CAG tract, is recognized by the enzyme OGG1, and is incised; BER then fills the gap by resynthesizing DNA. The same problem can occur — if synthesis proceeds through a repeat tract, the displaced strand can fold into a loop or hairpin, and the polymerase creates a looped-out structure that seals the extra repeats in.

There is also evidence that expansion and contraction are at least partly separable processes rather than mirror images of the same pathway [14]. Paternal transmissions in an MSH2-null background showed contractions replacing the usual expansions — relating back to the first criterion described in the introduction. Removing MSH2 evidently replaces expansions with contractions, so the two processes are, in some way, dissociated [14]. This is also consistent with why paternal influence is more dominant: instability exists in both sexes, but the expansion-specific machinery appears to operate more strongly in the male germline, though exactly why remains unsettled.

More recent, less-researched proposals center on transcription rather than replication, since HTT is actively transcribed. Here, RNA can remain hybridized to the template strand, displacing the non-template strand to form an R-loop and offering another opportunity for hairpin formation — again creating abnormal DNA substrates that attract repair processes [9].

Overall, the most persuasive current view is not that any single repair pathway is responsible, but that several different damage-response pathways can end up stabilizing, rather than removing, extra repeats, lengthening the CAG tract in the process.

Parent and Cis-Modifiers

Can either mechanism explain the difference between expanded alleles passed through the mother versus the father? The most extreme expansions and earliest cases are almost always inherited through the father. The replication mechanism has an intuitive explanation here: sperm derive from a germline that divides constantly throughout a man's life, so there is an ever-growing number of mitotic divisions, each individually increasing the chance of instability, whereas eggs retain the same genome from early development. If instability increased purely with division count, sperm should accumulate far more of it, and older fathers should transmit longer expansions more often — but this is not observed in mouse studies. Researchers also grew a Huntington's-model monkey's stem cells into spermatogenic cells and saw the repeat grow progressively in those specific cells, localizing the expansion to this particular germline lineage — though this at most only weakly supports repair over replication, since the rates of expansion observed were continuous, suggesting they weren't part of a discrete replication event (though this remains quite suggestive).

It is also worth noting that the glutamine-coding tract in HTT is not a pure CAG repeat — it is (CAG)ₙCAACAG, with a CAA interruption near the 3′ end [5, 15]. CAA still codes for glutamine, so the change is invisible at the protein level, but it matters a great deal for structure at the DNA level. Since CAA is not complementary to CAG, it forms the hairpin loop less readily, breaking the instability [3, 10]. There is direct evidence that CAG repeat length, not polyglutamine length, determines the timing of Huntington's Disease onset [7, 13, 15], suggesting that the purity of the repeat — not just its length — controls instability [2, 8]. Pure tracts form better hairpins, which matters for both the repair and replication models.

Genetics

The strongest independent support for repair-based models comes from genome-wide association study (GWAS) results, which strongly implicate genes responsible for DNA maintenance and repair as modifiers of Huntington's onset and progression [6]. By 2025, the GWAS landscape included seven DNA maintenance genes (FAN1, MLH1, MSH3, PMS1, PMS2, and MLH3), most of whose functions are described above [6, 15]. Equally telling is what does not appear in the GWAS: replication factors such as polymerases and fork components are not the dominant modifiers [6]. There are nonetheless real weaknesses to this evidence. These studies measure modifiers of onset and progression, which reflect somatic expansion in neurons across a lifetime rather than the germline expansion responsible for anticipation itself [4, 13]. The genetics therefore identify the relevant machinery — repair, not replication — with confidence, but extending that conclusion to the germline relies on the assumption that the same factors operate in both compartments [4, 6].

Conclusion

Returning to the four criteria from the introduction, the repair-driven account fits them more completely than its rivals [4, 9]. The directional bias toward expansion, and its reversal to contraction in an MSH2-null background, points to a mechanism that actively processes the lesion in a biased way while still occasionally allowing the opposite outcome [14]. The non-linear dependence on length follows from the thermodynamics of hairpin and slipped-strand formation: longer, purer tracts form stable secondary structures that repair machinery stabilizes rather than, as it should, removes [14, 15]. And the paternal bias, while still debated, sits more comfortably with germline-concentrated repair activity than with simple replication. Replication-based mechanisms remain less persuasive given that somatic expansion still occurs in post-mitotic neurons, and given that replication factors do not show up in GWAS landscapes [6].

None of this amounts to a closed case. The GWAS evidence concerns modifiers of somatic onset rather than germline expansion directly [4]. The precise reason MutL fails to discriminate the loop as an error is unsettled, as is the source of the male germline's bias toward heightened expansion.

When George Huntington closed his 1872 lecture, he had described the condition so precisely that William Osler would later call it the finest delineation of a disease "in so few words." That anticipation was invisible to him not because of any failure of observation, but because it was caused by a molecular logic of an expanding repeat and a counterproductive repair process that would not be uncovered for another 121 years — and is still not fully understood today.

References

  1. Aldous, S. G., Smith, E. J., Landles, C., Osborne, G. F., et al. (2024). A CAG repeat threshold for therapeutics targeting somatic instability in Huntington's disease. Brain, 147(5), 1784–1798.
  2. Bao, Y.-F., et al. (2023). Loss of CAA interruption and intergenerational CAG instability in Chinese patients with Huntington's disease. Journal of Molecular Medicine, 101(7), 837–847.
  3. Choi, D. E., Shin, J. W., Zeng, S., Hong, E. P., Jang, J.-H., Loupe, J. M., Wheeler, V. C., & Lee, J.-M. (2024). Base editing strategies to convert CAG to CAA diminish the disease-causing mutation in Huntington's disease. eLife, 13, e89782.
  4. Donaldson, J., Flower, M., Wright, G. E. B., et al. (2025). Huntington disease: somatic expansion, pathobiology and therapeutics. Nature Reviews Neurology.
  5. Findlay Black, H., Wright, G. E. B., Collins, J. A., Caron, N. S., Kovalenko, M., Helmold, V., Anderson, T., & Hayden, M. R. (2024). Ascertainment of uninterrupted CAG repeat length and disease-modifying variants in fragment-based genetic testing for Huntington disease. Genetics in Medicine Open, 2, 101896.
  6. Genetic Modifiers of Huntington's Disease (GeM-HD) Consortium. (2025). Genetic modifiers of somatic expansion and clinical phenotypes in Huntington's disease highlight shared and tissue-specific effects. Nature Genetics.
  7. Gu, X., et al. (2022). Uninterrupted CAG repeat drives striatum-selective transcriptionopathy and nuclear pathogenesis in human Huntingtin BAC mice. Neuron, 110(7), 1173–1192.
  8. Heinzmann, A., et al. (2025). Sequence variants in small CAG repeat expansions of the HTT gene and disease onset and progression in Huntington disease. Neurology.
  9. Malik, I., Kelley, C. P., Wang, E. T., & Todd, P. K. (2021). Molecular mechanisms underlying nucleotide repeat expansion disorders. Nature Reviews Molecular Cell Biology, 22(9), 589–607.
  10. Matuszek, Z., Brown, B. L., Yang, B., et al. (2025). Base editing of trinucleotide repeats that cause Huntington's disease and Friedreich's ataxia reduces somatic repeat expansions in patient cells and in mice. Nature Genetics.
  11. Mouro Pinto, R., Arning, L., Giordano, J. V., et al. (2025). In vivo CRISPR-Cas9 genome editing in mice identifies genetic modifiers of somatic CAG repeat instability in Huntington's disease. Nature Genetics.
  12. Phadte, A. S., Bhatia, M., Ebenezer, J., Csonka, D., et al. (2023). FAN1 removes triplet repeat extrusions via a PCNA- and RFC-dependent mechanism. Proceedings of the National Academy of Sciences, 120(33), e2302538120.
  13. Scahill, R. I., Farag, M., Murphy, M. J., et al. (2025). Somatic CAG repeat expansion in blood associates with biomarkers of neurodegeneration in Huntington's disease decades before clinical motor diagnosis. Nature Medicine.
  14. Senoussi, I., et al. (2025). Mechanism of trinucleotide repeat expansion by MutSβ-MutLγ and contraction by FAN1. Nature Communications.
  15. Wright, G. E. B., Black, H. D., Collins, J. A., Gall-Duncan, T., Caron, N. S., Pearson, C. E., & Hayden, M. R. (2020). Interrupting sequence variants and age of onset in Huntington's disease: clinical implications and emerging therapies. The Lancet Neurology, 19(11), 930–939.