Antagonistic roles of canonical and Alternative-RPA in disease-associated tandem CAG repeat instability

Abstract

Expansions of repeat DNA tracts cause >70 diseases, and ongoing expansions in brains exacerbate disease. During expansion mutations, single-stranded DNAs (ssDNAs) form slipped-DNAs. We find the ssDNA-binding complexes canonical replication protein A (RPA1, RPA2, and RPA3) and Alternative-RPA (RPA1, RPA3, and primate-specific RPA4) are upregulated in Huntington disease and spinocerebellar ataxia type 1 (SCA1) patient brains. Protein interactomes of RPA and Alt-RPA reveal unique and shared partners, including modifiers of CAG instability and disease presentation. RPA enhances in vitro melting, FAN1 excision, and repair of slipped-CAGs and protects against CAG expansions in human cells. RPA overexpression in SCA1 mouse brains ablates expansions, coincident with decreased ATXN1 aggregation, reduced brain DNA damage, improved neuron morphology, and rescued motor phenotypes. In contrast, Alt-RPA inhibits melting, FAN1 excision, and repair of slipped-CAGs and promotes CAG expansions. These findings suggest a functional interplay between the two RPAs where Alt-RPA may antagonistically offset RPA's suppression of disease-associated repeat expansions, which may extend to other DNA processes.

Keywords: Alternative replication protein A (Alt-RPA); BioID protein interactome; DNA repair (FAN1, MSH2, MSH3, MSH6, HTT); Huntington's disease (HD); RPA1, RPA2, RPA3, RPA4; Replication protein A (RPA); Spinocerebellar ataxia type 1 (SCA1); slipped-DNA; tandem repeat expansions; trinucleotide CAG repeat expansions.

Figure 1:. RPA1, 2, 3 &4 are upregulated in patient brain tissues, cell lines, and mouse brain tissues.

A) canonical RPA (RPA1, 2, 3) versus Alt-RPA (RPA1, 4, 3). B) RPA2 and RPA4 have homologous DNA-binding domains (DBD), winged helix domains, and a less similar N-terminal region. ddPCR data of all RPA subunits from C) striatum of HD patients (n=7 individuals/group/tissue, 3 replicates/person) and D) cerebellum of HD and SCA1 patients. (n=7 HD & unaffected individuals/group/tissue, and 3 SCA1 patients/group/tissue, 3 replicates/person). Colored dots indicate striatal neuropathological grade (HD 1–4). E) Representative western blots for RPA2 and RPA4 expression levels in HD and SCA1 patient striatum and cerebellum relative to unaffected control tissues. Actin loading control. F-I) densitometric quantification of RPA2 and RPA4 levels versus actin in striatum, cerebellum, cultured fibroblasts, and zQ175 HD mouse striatum. Unpaired student t-test comparing means. Data are represented as median ± Tukey whisker extent (box plots) or mean ± SD (dot plots).

Figure 2:. RPA and Alt-RPA competitively modulate slip-out repair.

A) Nick-in-flank substrates are structurally stable; whereas nick-in-repeat substrates permit multiple heterogenous structures. Nicks created by cutting and annealing strands of (CAG)30 and (CAG)50 DNAs (nicks-in-flank, EcoRI, and nicks-in-repeat, BsmI, grey, white, and black triangles). B-D) Repair reactions with 25 μg of HeLa or RPA-/Alt-RPA-depleted (RPA2 siRNA) HeLa extracts, dNTPs, and purified (6 μg) RPA or Alt-RPA. Reaction products were purified, repeat-containing fragments released (EcoRI/HindIII), electrophoretically resolved on native 4% PAGE, and visualized by Southern. Structural intermediates (SI; green triangles) and fully-duplexed repair products (black arrows). Densitometric quantification of 3–6 replicates (normalized for background). Graphed SI levels (green bars) and correct repair products (grey, blue, and orange bars). E-H) Repair reactions with indicated slip-out or G-T mismatch, cell extract, and purified protein(s), as in B-D. Where indicated reactions included 0.6 μg RPA (lane 4) or Alt-RPA (lane 4). Lanes 5–7 shows processing by RPA-deficient extract supplemented by 0.6 μg RPA or Alt-RPA at ratios of 1:0, 1:2, 1:6, or 1:10 Alt-RPA or RPA. Lane 8 shows substrate processed by RPA-deficient extract supplemented by only 6 μg RPA or Alt-RPA. I-J). Repair reactions were performed as above, included 0.6 μg of human RPA, bacterial bSSB or yeast scRPA. Quantitation of E-J as for B-D. Data are the mean ± SD.

Figure 3:. Alt-RPA has altered binding, poorly melts slipped-DNAs, and inhibits FAN1 cleavage.

A) γ-³²P-radiolabelled DNA substrates were incubated with 600 ng (lanes 2, 5), 1200 ng (lanes 3, 6), or 2400 ng (lanes 4, 7) of RPA or Alt-RPA, resolved on 4% native acrylamide. Free-DNA (black triangles); protein-DNA complexes (blue arrowheads (RPA) or orange arrowhead (Alt-RPA)). B) Time-course of melting of fluorescently labelled slipped-CAG oligonucleotides by RPA and/or Alt-RPA. Time needed to reach 0.475 FRET (i.e. half the DNA being melted) quantified rates of melting by each complex based on molar ratio of RPA:Alt-RPA (bar graph; n=3 replicates). Statistics: unpaired t-test comparing means. C) FAN1 endo-nuclease activity is enhanced by RPA and inhibited by Alt-RPA. Purified FAN1, RPA € and/or Alt-RPA (AR) were incubated with FAM-labelled oligonucleotides (mimics slipped-CAG; schematic at left) (- = no FAN1 added and 200 nM of RPA or Alt-RPA, + = 50 nM FAN1; triangles: 25 nM, 50 nM, 100 nM, or 200 nM RPA or Alt-RPA; straight line = 50 nM RPA or Alt-RPA) and resolved on a 4% denaturing gel. Schematic (right) indicates migration positions for labeled CAG strand and FAN1 endo-nucleolytic cleavage positions (red arrows). Nuclease activity was quantified densitometrically (cleavage products intensity/full-length substrate intensity, n=3 replicates. *=p<0.05, ***=p<0.001). Data are the mean ± SD.

Figure 4:. RPA and Alt-RPA BioID interactomes.

A) Shared and unique BioID hits for RPA and Alt-RPA subunits from HEK293T cells. B) Dot plot outlining subunit associated proteins (functionally categorized). Fill-color shows log2 fold enrichment versus BirA-only control (non-specific interactions): protein enrichment versus whole-dataset minimum (dark blue) and maximum enrichment (dark red) value. Dot size indicates relative abundance of interaction versus other subunits. Dot outline color shows significance (black=p<0.01, blue=p<0.05, and light blue, p>0.05). C) Co-IP of RPA (RPA2) or Alt-RPA (RPA4) from unaffected and HD patient-derived fibroblasts, or D) Co-Ips of purified RPA or Alt-RPA complexes with purified MutSα or MutSβ complexes using FLAG-tagged MSH2. In C and D co-Ips visualised by Western, n=3–6 replicates; representative images.

Figure 5:. Upregulation of RPA inhibits while Alt-RPA enhances somatic repeat expansions.

A) Small-pool PCR (spPCR) quantification of RPA (left panel) or Alt-RPA (right panel) overexpression versus transfection control (middle panel) in serum-starved HT1080 (CAG)850 cells grown10-days. (Light gray line=starting repeat length, gray bars=expansions (right) and contractions (left). Statistical analysis, with indicated comparisons, were by χ-square test^,. B) Representative fragment length analysis scans of GFP- and Rpa1-overexpressing 56-week-old SCA1 mouse cerebellum and striatum. (Gray bar=inherited repeat length, red brackets=ongoing expansions). C) Average instability and expansion/contraction indices for all GFP- and Rpa1-overexpressing SCA1 mouse cerebellum and striatum ***=p<0.001. Data are represented as mean ± SD.

Figure 6:. Canonical RPA upregulation in SCA1 mouse striatum reduces neuronal DNA damage and mutant Ataxin-1 aggregation.

A-B) Representative IF images of γ-H2AX and 53BP1 in DARPP32-posititive striatal medium spiny neurons (MSNs) and C) quantification (n=3 mice/group, 3 replicates with ≥30 neurons/replicate, ***=p<0.001). D) Representative images of Ataxin-1 and ubiquitin co-staining within MSNs and E) quantification (n=3 mice/group, 3 replicates with ≥50 neurons/replicate. **=p<0.01). Data are mean ± SD.

Figure 7:. Working model of Alt-RPA↔RPA antagonistic interactions in somatic expansions.

(Top) Canonical RPA enhances correct repair by rapidly melting slip-outs, enhancing FAN1-mediated slip-out excision reducing CAG expansions and diminishing downstream events. (Bottom) Alt-RPA inhibits repair by differential binding, inefficiently melting slip-outs, inhibiting FAN1-mediated excision, leading to expansions and exacerbating downstream events. For non-disease state, the top pathway may predominate whereas for diseased state (dysregulated RPA/Alt-RPA levels) the bottom path may predominate. Differential associations of RPA/Alt-RPA with DNA repair proteins known to modulate repeat expansions also contribute to instability. RPA preferentially interacts with MutSα (MSH2-MSH6) and MutLβ (MLH1-PMS1). Alt-RPA preferentially interacts with MutSβ (MSH2-MSH3) and XPG. Downstream events shown in this study include brain DNA damage and neuronal polyglutamine-aggregates (Figure 6A–B and S7), motor phenotypes, neuron morphology, dysregulated transcriptome, spliceosome, and cell cycle.