FAN1 controls mismatch repair complex assembly via MLH1 retention to stabilize CAG repeat expansion in Huntington's disease

Abstract

CAG repeat expansion in the HTT gene drives Huntington's disease (HD) pathogenesis and is modulated by DNA damage repair pathways. In this context, the interaction between FAN1, a DNA-structure-specific nuclease, and MLH1, member of the DNA mismatch repair pathway (MMR), is not defined. Here, we identify a highly conserved SPYF motif at the N terminus of FAN1 that binds to MLH1. Our data support a model where FAN1 has two distinct functions to stabilize CAG repeats. On one hand, it binds MLH1 to restrict its recruitment by MSH3, thus inhibiting the assembly of a functional MMR complex that would otherwise promote CAG repeat expansion. On the other hand, it promotes accurate repair via its nuclease activity. These data highlight a potential avenue for HD therapeutics in attenuating somatic expansion.

Keywords: CAG instability; DNA repair; FAN1; FAN1 nuclease activity; GWAS; Huntington’s disease; MLH1; MSH3; mismatch repair; repeat expansion.

Graphical abstract

Figure 1

The FAN1 N-terminal region (p.73-349) mediates its interaction with MLH1 and its effect on CAG stabilization activity (A) CoIP extracts from human HD iPSCs showing FAN1 interacts with MutLα components MLH1 and PMS2. Note that MSH3 is absent from the anti-FAN1 IP fraction (n = 3 biological replicates). (B) CoIP extracts from human HD lymphoblasts confirming FAN1 interacts with MLH1 (n = 3 biological replicates). (C) Pull-down assays using GFP-Trap beads in U2OS cells showing FAN1 interacts with MutL components, but not MutS components or PCNA (proliferating cell nuclear antigen). FAN1^−/− cells act as a negative control, demonstrating specificity of the pull-down (n = 4 biological replicates). (D) CoIP of cortical extracts from mouse zQ175 at 6 months of age confirming FAN1 interacts with MLH1. Observations were also confirmed in R6/2 HD mice at 12 weeks of age (zQ175, n = 3 biological replicates; R6/2, n = 2 biological replicates). (E) Crosslinks identified between FAN1, MLH1, and PMS2 in unstimulated HEK293T cells and HD lymphoblasts. Grey parts on the proteins are structurally unsolved (no PDB structure available). Turquoise line, interprotein; purple line, intraprotein; green line, crosslinks close to the SPYF motif. See also Figure S1A and Table S1. (F) Schematic illustrating FAN1 constructs cloned into U2OS system. Locations of UBZ null (C44A/C47A) and nuclease null (D960A) mutations are also outlined. UBZ, ubiquitin-binding zinc-finger domain; SAP, SAF-A/B, Acinus and PIAS domain; TPR, tetratricopeptide repeat domain; VRR_NUC, virus-type replication-repair nuclease domain. (G) Pull-down using GFP-Trap beads in U2OS cells expressing GFP-FAN1 deletion constructs. FAN1^Δ73–349 (highlighted in bold) did not interact with MLH1. Note that inactivation of UBZ or VRR_NUC domains (C44A/C47A and D960A mutants, respectively) does not affect FAN1-MLH1 interaction (n = 3 biological replicates). (H) CoIP extracts using αMLH1 antibody in U2OS cells showing N-terminal FAN1^1–349 is sufficient to interact with MLH1 (n = 4 biological replicates). (I) MMC viability curves in U2OS cells expressing FAN1 variants (mean ± SD) showing lower viability when FAN1 lacks an intact nuclease domain (n = 5–8 biological replicates; n = 3 technical replicates). See also Figure S1E. (J) CAG expansion rates in U2OS cells expressing truncated FAN1 constructs, including mutations within key functional domains (UBZ, C44A/C47A; VRR_NUC, D960A). Note that only FAN1^Δ73–349 shows a higher expansion rate than cells expressing FAN1^FL but does not equate to FAN1^−/− (mean ± SEM, n = 2–5 biological replicates, n = 3–6 technical replicates, F(5,97) = 40.8, p < 0.001 by one-way ANOVA with false discovery rate [FDR] correction of 5%). ^∗∗∗p < 0.001; ns, non-significant. (K–M) Fragment analysis traces illustrating expansion of the exogenous HTT 118 CAG repeat in U2OS cells expressing FAN1 constructs over 6 weeks in culture (K) with time courses plotted (L and M). Note that cells expressing FAN1^Δ73–349 (individual data points shown) expand at a rate between that of FAN1^FL or FAN1^1–349 (individual data points shown) and FAN1^−/− cells (mean ± SD, 95% confidence interval [CI] in shaded areas, n = 2–5 biological replicates, n = 3–6 technical replicates).

Figure 2

A conserved SPYF motif in FAN1 is required for MLH1 binding (A–C) CoIP extracts using GFP-Trap beads in U2OS cells expressing truncated FAN1 constructs (A and B) with quantification showing progressively longer FAN1 N-terminal fragments bind more MLH1 (C). Note residues 120–140 are essential for MLH1 binding (mean ± SEM, n = 4–5 biological replicates, F(5,22) = 88.31, p < 0.001 by one-way ANOVA with FDR correction of 5%). ^∗p < 0.05; ^∗∗∗p < 0.001; ns, non-significant. (D) Conservation analysis schematic showing SPYF motif is heavily conserved within common model species (residues with >80% consensus shown in yellow). (E) Schematic illustrating FAN1 constructs with mutations at conserved SPYF residues that were cloned into the U2OS system. Nuclease null mutation (D960A) is also outlined. UBZ, ubiquitin-binding zinc-finger domain; SAP, SAF-A/B, Acinus and PIAS domain; TPR, tetratricopeptide repeat domain; VRR_NUC, virus-type replication-repair nuclease domain. (F) MMC viability curves in U2OS cells expressing FAN1 SPYF mutants (mean ± SD). Note viability is only reduced in FAN1^−/− line (n = 6–8 biological replicates, n = 3 technical replicates) (see also Figure S1G). (G and H) Input and GFP-Trap pull-down fractions from U2OS cell extracts expressing FAN1 SPYF mutants (G) with quantification (H) showing reduced MLH1-binding with mutation of SPYF motif relative to FL construct. Q123A is displayed as a control, having a mutation outside the conserved motif (mean ± SEM, n = 5 biological replicates; F(4,17) = 744.6, p < 0.001 by one-way ANOVA with FDR correction of 5%). ^∗∗∗p < 0.001.

Figure 3

FAN1 SPYF motif and nuclease activity stabilize the HTT CAG repeat (A) CAG expansion rates in U2OS cells expressing FAN1 constructs with mutations at conserved SPYF motif. Note that mutation of this domain results in hastened expansion of the HTT CAG repeat. Q123A is displayed as a control, having a mutation outside the conserved motif. (mean ± SEM, n = 2–5 biological replicates, n = 3–6 technical replicates, F(5,83) = 28.64, p < 0.001 by one-way ANOVA with FDR correction of 5%). ^∗∗p < 0.01, ^∗∗∗p < 0.001, ns = non-significant. (B) CAG expansion rates in U2OS cells expressing truncated N-terminal constructs of FAN1, showing residues 120–140 contribute significantly to HTT CAG repeat stability. (mean ± SEM, n = 2–5 biological replicates, n = 3–6 technical replicates, F(5,86) = 22.38, p < 0.001 by one-way ANOVA with FDR correction of 5%). ^∗p < 0.05, ^∗∗∗p < 0.001, ns = non-significant. (C) Input and GFP-Trap pull-down fractions from U2OS cell extracts expressing FAN1^FL and FAN1^F129A/D960A showing reduced MLH1-binding with mutation of SPYF motif relative to FL. Note equivalent FAN1^FL and FAN1^F129A/D960A expression (n = 2 biological replicates). (D) MMC viability curves in U2OS cells expressing FAN1^F129A and FAN1^F129A/D960A mutants (mean ± SD, n = 6–7 biological replicates, n = 3 technical replicates). Note resistance to MMC toxicity is only maintained in the F129A line. See also Figure S1H. (E–G) Fragment analysis traces illustrating expansion of the exogenous HTT 118 CAG repeat in U2OS cells expressing FAN1^F129A or FAN1^F129A/D960A mutants over 6 weeks in culture with time courses plotted (F; mean ± SD, 95% CI in shaded areas) and quantified (G). Cells expressing FAN1^F129A/D960A show equivalent expansion as FAN1^−/− cells (mean ± SEM, n = 2–5 biological replicates, n = 3–6 technical replicates, F(3,72) = 39.27, p < 0.001 by one-way ANOVA with FDR correction of 5%). ^∗∗p < 0.01; ^∗∗∗p < 0.001; ns, non-significant.

Figure 4

FAN1 regulates mismatch repair (MMR) activity through MLH1 binding (A and B) Western blots showing MMR protein expression in U2OS MLH1 (A) and MSH3 (B) knockout lines (n = 3 biological replicates). (C) CAG expansion rates in FAN1^−/−, MLH1^−/−, and MSH3^−/− U2OS cell lines. Note that knockout of MSH3 or MLH1 ablates CAG repeat expansion (mean ± SEM, n = 2–5 biological replicates, n = 3–6 technical replicates, F(2,72) = 272.5, p < 0.001 by one-way ANOVA with FDR correction of 5%). ^∗∗∗p < 0.001; ns, non-significant. (D) 6TG viability curves in U2OS cells expressing FAN1 constructs and MLH1, showing cells with an intact FAN1 SPYF motif have enhanced resistance to 6TG, indicating reduced MMR activity. MLH1^−/− cells serve as a control (mean ± SD, n = 5 biological replicates, n = 3 technical replicates). See also Figure S2A. (E) CoIP of MLH1 and binding partners from FAN1^−/− and FAN1^FL cells. Note FAN1 expression reduces MSH3 levels in MLH1 IP fractions but does not affect PMS2 (n = 4 biological replicates). See also Figure S2F. (F) CoIP of MLH1 and binding partners from 125 CAG HD MSNs with shRNA-mediated FAN1 knockdown. Untreated cells and an empty shRNA vector were used as controls. Note that FAN1 knockdown increases MSH3 levels in MLH1 IP fractions (n = 3 biological replicates). See also Figures S2G and S2H. (G) CoIP of myc-tagged FAN1 from HEK293T cells expressing strep-tagged MLH1 variants and endogenous MLH1. Note endogenous MLH1 and strep-tagged MLH1^FL bind to FAN1, whereas MLH1^E669A does not (n = 2 biological replicates). (H) FAN1 peptide competition assay in HeLa cell nuclear extracts showing FAN1 wild-type (wt) 60-mer peptides (15 μM) reducing MLH1-MSH3 interactions, whereas FAN1 mutant (mut.) peptides do not (n = 3 biological replicates). (I) ChIP extracts from FAN1^FL and FAN1^−/− U2OS cells, immunoprecipitated with αMLH1 antibodies and DNA amplified with primers targeting HTT CAG repeat region. Note the decreased levels of both long (exogenous HTT) and short (endogenous HTT) amplicons in FL ChIP fractions (n = 3 biological replicates). (J) Fragment analysis traces from U2OS FAN1^−/− extracts show the presence of the CAG repeat from the endogenous HTT allele (20 CAG units) and the longer exogenous repeat (118 CAG) from the exon 1 construct in both input and ChIP fractions. The lack of signal in the control IP (-Ab) shows the specificity of the procedure (n = 3 biological replicates, n = 3 technical replicates). (K) Quantification of DNA levels in ChIP fractions from FL and FAN1^−/− U2OS cells. Primer pairs proximal to the CAG repeat (P1 and P2) and toward the 3′ end of HTT (HTT2) were used (mean ± SEM, n = 3 biological replicates, n = 3 technical replicates; P1: F(2,6) = 20.76, p = 0.002; P2: F(2,6) = 17.84, p = 0.003; HTT2: F(2,6) = 23.56, p = 0.001 by one-way ANOVA with FDR correction of 5%). ^∗p < 0.05; ^∗∗∗p < 0.001.