Three-dimensional chromatin interactions remain stable upon CAG/CTG repeat expansion

Expanded CAG/CTG repeats underlie thirteen neurological disorders, including myotonic dystrophy (DM1) and Huntington’s disease (HD). Upon expansion, CAG/CTG repeat loci acquire heterochromatic characteristics. This observation raises the hypothesis that repeat expansion provokes changes to higher order chromatin folding and thereby affects both gene expression in cis and the genetic instability of the repeat tract. Here we tested this hypothesis directly by performing 4C sequencing at the DMPK and HTT loci from DM1 and HD patient-derived cells. Surprisingly, chromatin contacts remain unchanged upon repeat expansion at both loci. This was true for loci with different DNA methylation levels and CTCF binding. Repeat sizes ranging from 15 to 1,700 displayed strikingly similar chromatin interaction profiles. Our findings argue that extensive changes in heterochromatic properties are not enough to alter chromatin folding at expanded CAG/CTG repeat loci. Moreover, the ectopic insertion of an expanded repeat tract did not change three-dimensional chromatin contacts. We conclude that expanded CAG/CTG repeats have little to no effect on chromatin conformation.


Introduction
the vicinity of the expanded repeat (12,23,31). Importantly, daSTRs are found predominantly at 69 TAD and sub-TAD boundaries, suggesting more generally that daSTR expansions may disrupt 70 TADs (31). This may then contribute to gene silencing in cis and to the high levels of instability 71 found at these sequences, ultimately altering disease progression (31). 72 The critical unknowns in this model are whether changes in higher-order chromatin structure are 73 confined to CGG and GAA repeats or if this is general to daSTRs and whether changes in

93
Chromatin conformation is stable upon repeat expansion at the HTT locus 94 To assess whether chromatin contacts change upon CAG repeat expansion at the HTT locus, we 95 used a series of HD patient-derived LCLs (GM02164, GM03620, and GM14044, referred to as 96 HD-A, HD-B, HD-C respectively) as well as two lines from unaffected individuals (GM04604 and 97 GM02180, UN-A and UN-B respectively). Their family relationships and their repeat sizes are 98 found in Fig. 1A, S1A, and Table S1.
To determine chromatin conformation, we used 4C-seq (36, 37) because it maximizes resolution 100 at the loci of interest. This allows a high sensitivity to small changes in conformation that may be 101 missed by other 3C-based methods. For the HTT locus, we used 2 viewpoints on chromosome 4 102 within the gene body -1 kb (HTT_d1) and 85 kb (HTT_d85) downstream of the CAG repeats. To 103 control for a potential effect of the pathology on genome-wide chromatin conformation, we also 104 used a viewpoint located near the ACTA1 gene on chromosome 1. We obtained three replicates 105 of each 4C viewpoint and compared the DNA interaction profiles between the unaffected and HD 106 cell lines (Fig. 1B). Replicates from the same cell lines show good correlation in fragments with 107 more than 20 mapped reads ( Fig. S2A-C). 108 We then identified fragments and regions that interact with the 4C viewpoint at frequencies higher 109 than expected (significant interactions, see methods). We also looked for regions that show 110 significant differences in interaction frequencies between patient-derived and unaffected cells 111 (differential interactions, see methods). To do so, we used two 4C-seq data analysis packages: 112 FourCSeq (38) and 4C-ker (39). We found that the interaction profiles were similar within a 2 Mb 113 region around the viewpoint (Fig. 1B, Fig. S3). Notably, the chromatin conformation remained 114 unaltered in HD-A, an HD patient cell line that has two expanded alleles (44/56 CAGs), as well 115 as in HD-B (18/70 CAGs) and HD-C (19/750 CAGs). The ACTA1 viewpoint also produced 116 indistinguishable interaction profiles between unaffected and HD patient LCLs (Fig. 1C). We 117 identified few small regions displaying differential interactions, but they were mainly outside of 118 regions with significant interactions (Fig. 1BC). In addition, most regions of differential interactions 119 were not exclusive to HD patient cells, as they were also found in comparisons between the two 120 unaffected cell lines. This suggests that the minor changes in interaction frequencies are due to 121 factors other than the presence of the expanded repeat tract in HTT, e.g. the different genetic 122 backgrounds. Taken together, these results show that a CAG repeat expansion at the HTT locus 123 does not cause significant alterations in chromatin conformation.

126
Chromatin conformation is stable upon repeat expansion at the DMPK locus 127 Repeat expansion at the HTT locus is not known to be associated with significant changes in 128 histone marks and chromatin accessibility, which might cause changes in chromatin 129 conformation. To determine whether our findings applied in a case where chromatin structure is 130 altered around expanded CAG/CTG repeats, we studied four viewpoints in the DMPK gene. We The top blue bar represents the HTT gene and the triangles represent the location of both HTT viewpoints. The interaction profiles for the HTT_d85 viewpoint (85 kb downstream of the CAG repeat) can be found in Fig. S3. (C) 4Cseq chromatin interaction profiles (average of triplicate smoothed and normalized counts) from the ACTA1 viewpoint (central purple triangle). For panels B and C, high-interacting regions were called using 4C-ker and significant interactions were called using FourCSeq. Regions of differential interactions are marked with black bars below each 4C-seq track and labeled as "diff. int.". The top blue bar represents the ACTA1 gene. used two DM1 patient-derived LCLs (GM06077 and GM04648, DM1-A and DM1-B respectively).

132
The DM1-A cell line harbored one expanded DMPK allele with 1,700 CTGs and the DM1-B cell 133 line, 1,000 CTGs ( Fig. 2A and S1B). We found that DM1-A cells had increased CpG methylation 134 levels at two CTCF binding sites flanking the repeats (Fig. S4B), with a concomitant loss of CTCF 135 binding at these sites (Fig. S4C). In DM1-B cells we observed normal methylation levels at both 136 CTCF binding sites and slightly reduced CTCF binding ( Fig S4B). Thus, DM1-A cells displayed  (Table S1). 139 We performed 4C-seq on the unaffected and DM1 cell lines using four different 4C viewpoints at 140 distinct distances away from the DMPK CTG repeats ( Fig. 2A). Replicates from the same cell   and TAIL PCR, we mapped the insertion site to the p-arm of chromosome 12, 1.2 Mb from the 182 telomere ( Fig 3A). We performed 4C-seq in both cell lines with a viewpoint located 1 kb upstream 183 of the CAG repeats. The chromatin interaction profiles of this ectopic CAG locus were also very 184 similar between the 15 and 270 CAG repeat cells, with few regions of differential interactions 185 overlapping with those displaying significant interactions using 4Cker. By contrast, FourCSeq did 186 not detect any changes that were significant (Fig. 3B). We concluded that CAG repeats at an 187 ectopic site causes few changes to chromatin conformation.

188
Some studies suggest that transcription may help define topological domain boundaries (44-47).

189
To determine whether transcription through expanded CAG repeats could lead to changes in  Here we showed that chromatin interactions remain stable at expanded CAG/CTG repeat loci.

200
This was true for two disease loci and one transgene and in two different cell types. Our findings 201 are also supported by allele-specific analysis of 4C-seq interactions. Furthermore, CpG 202 methylation and CTCF binding at two sites flanking the CTG repeats of DMPK did not impact 203 chromosome conformation. This is especially relevant because CTCF is a key architectural 204 protein involved in the demarcation of TAD and sub-TAD boundaries (48). Additionally, when we 205 inserted a hemizygous transgene with CAG repeats, we found that expanded CAG repeats did 206 not induce a significant re-organization of the chromatin contacts established at this exogenous 207 CAG/CTG repeat locus. These results further show that an expanded CAG/CTG repeat tract is 208 not sufficient to change chromatin folding. 209 One possibility is that expanded CAG/CTG repeats lead to changes in chromatin conformation in 210 specific cell types that are especially vulnerable in DM1 or HD, for example cardiomyocytes or    Repeat length determination and small-pool PCR 260 Genomic DNA was isolated from each LCL using the NucleoSpin Tissue kit (Macherey-Nagel).

261
PCR products containing the CTG repeats from DMPK were amplified with primers oVIN-1252 262 and oVIN-1251 (Table S3). PCR products with the CAG repeats from HTT were produced with 263 primers oVIN-1333 and oVIN-1334 (Table S3) Table S3.   Table S3.  Table S2. For plotting the data, fragment counts were 323 normalized (reads per million) and smoothed with a running mean (window size = 5 fragments).

324
The smoothed and normalized fragment counts were averaged among replicates of the same 4C

Competing interests
The authors declare no competing interests. All sequencing data underlying this study will be shortly deposited in the SRA of NCBI. All data 543 presented here are available from the corresponding author.