The D4Z4 macrosatellite repeat acts as a CTCF and A-type lamins-dependent insulator in facio-scapulo-humeral dystrophy

Abstract

Both genetic and epigenetic alterations contribute to Facio-Scapulo-Humeral Dystrophy (FSHD), which is linked to the shortening of the array of D4Z4 repeats at the 4q35 locus. The consequence of this rearrangement remains enigmatic, but deletion of this 3.3-kb macrosatellite element might affect the expression of the FSHD-associated gene(s) through position effect mechanisms. We investigated this hypothesis by creating a large collection of constructs carrying 1 to >11 D4Z4 repeats integrated into the human genome, either at random sites or proximal to a telomere, mimicking thereby the organization of the 4q35 locus. We show that D4Z4 acts as an insulator that interferes with enhancer-promoter communication and protects transgenes from position effect. This last property depends on both CTCF and A-type Lamins. We further demonstrate that both anti-silencing activity of D4Z4 and CTCF binding are lost upon multimerization of the repeat in cells from FSHD patients compared to control myoblasts from healthy individuals, suggesting that FSHD corresponds to a gain-of-function of CTCF at the residual D4Z4 repeats. We propose that contraction of the D4Z4 array contributes to FSHD physio-pathology by acting as a CTCF-dependent insulator in patients.

Figure 1. A single D4Z4 acts as a boundary that interferes with position effect and enhancer-promoter communication.

A. The different constructs carry a hygromycin resistance gene fused to the herpes simplex virus type 1 thymidine kinase suicide gene (HyTK, white box) and an eGFP reporter gene (speckled box), each driven by a CMV promoter (pr). In the T construct, a telomere seed (grey triangles) is added downstream of the eGFP reporter gene in order to create a de novo telomere after random integration followed by a telomeric fragmentation . A single D4Z4 repeat (black box) is cloned downstream of the eGFP gene in pCMV construct (C1X) or between eGFP and the telomere seed in T construct (T1X). We further compared D4Z4 with the canonical chicken 5′ HS4 boundary by cloning this latest sequence into the vectors used for de novo telomere seeding (5′HS4-T) or for random integration (5′HS4). Each constructs were linearized and transfected into the human cervical carcinoma cells (C33A). The level of eGFP was measured by flow cytometry (FACS) for an extended period of time in the presence or absence of Hygromycin B (Figure S1A). Histograms show the average percentage of eGFP positive cells from day 18 to day 29 of three independent transfections ±S.D. shown by error bars, when eGFP expression reaches a plateau (Figure S1B). The integrity of each construct was verified in stable populations of cells (Figure S1D). B. In order to evaluate the enhancer blocking activity of D4Z4, we used the test previously described . The K562 human erythroleukemia cell line was stably transfected with the constructs shown on the left. Each construct carries the neomycin resistance gene driven by the human A β-globin promoter (γ-Neo) flanked with the mouse 5′HS2 enhancer (E). Most constructs contain the 5′HS4 insulator upstream of the promoter in order to block from the influence of regulatory elements at the site of integration. For each assay, colony number was normalized to the un-insulated control (pNI). Data are the average of three independent transfections. The mean values with S.D. are plotted. As controls, the following constructs, kindly provided by Dr. G. Felsenfeld, were used: pNI, no insert; pJC3-4, 2.3 kb of λ DNA; pJC5-4, chicken β-globin 1.2 kb 5′HS4 insulator .

Figure 2. Mapping of the regulatory fragments within D4Z4.

A. Schematic representation of the D4Z4 element from position 1 to 3303 given relative to the two flanking KpnI sites (K) (to scale). The different regions within D4Z4 are indicated: LSau repeat (position 1–340), Region A (position 869–1071), hhspm3 (position 1313–1780), DUX4 ORF (position 1792–3063). The different restriction sites used for the cloning of D4Z4 subfragments are indicated (B: BamHI; Bl: BlpI; F: FseI; E: EheI). B. Different fragments obtained after digestion of D4Z4 were cloned downstream of the eGFP reporter (“C” constructs) or between the reporter gene and the telomeric seed (“T” constructs). Linearized plasmids were transfected into C33A cells and the percentage of eGFP positive cells was monitored by flow cytometry for an extended period of time. The histogram represents the mean value of the percentage of eGFP positive cells from day 18 to day 29 when eGFP expression reaches a plateau±S.D shown by error bars. Fragments DB2-3 (position 1 to 382), DB1-2 (position 814 to 1381) and DF (position 1549 to 3303) do not abrogate TPE or CPE while fragment DB1 (position 1 to 1381), DB1-3 (position 382 to 814) and DE (deleted of a distal 623 bp fragment from position 2269 to 2892) protect from CPE and TPE. Asterisks denote statistically significant values relative to control vectors (pCMV or T) (Student's t test). * p<0.001; **p<0.005; *** p<0.05.

Figure 3. CTCF and A-type Lamins bind to D4Z4 in vivo.

We searched in silico for CTCF binding sites across the 3.3 kb D4Z4 sequence (Genbank accession number AF117653) using the consensus binding site at the chicken β-globin locus ,. Two sites were identified at the 5′ end of D4Z4 and the binding was investigated by ChIP using antibodies to CTCF. We also investigated the involvement of A-type Lamins using specific antibodies. Enrichment of the immunoprecipitated DNA fraction with antibodies compared to input DNA was determined after real-time Q-PCR amplification (y-axis) for different primer pairs. Values were normalized to the Histone H4 internal standard. Each bar is the average of at least three independent experiments with the S.D. shown by error bars. “eGFP” amplifies the eGFP sequence. The position of the primers within D4Z4 is indicated (sets 1–4). Using high-throughput analysis, numerous CTCF binding sites were recently identified and many of these sites also correspond to Cohesins enrichment . We then asked if Cohesins/CTCF complex also contains A-Type Lamins and amplified DNA immunoprecipitated with Lamins A/C antibodies with primers corresponding to chromosome 6 (Chr 6). We observed a strong enrichment for CTCF but not Lamins at this site suggesting that CTCF/Cohesins and CTCF/Lamins bind distinct sites. A sequence on chromosome 20 (Chr 20) was reported as a site for CTCF only and does not bind A-type Lamins. Chr 7 primers are CTCF-negative control. Asterisks denote statistically significant values (** p<0.001; *p<0.005; Student's t test).

Figure 4. Depletion in CTCF or A-type Lamins abrogates D4Z4 anti-silencing activity.

The involvement of CTCF and A-type lamins in the insulating activity of D4Z4 was studied by knocking-down their expression. A. Different populations of cells transfected with randomly integrated constructs were transfected with siRNA against CTCF (CTCF) or negative control siRNA (mock) and the level of eGFP was monitored by FACS. The mean value±S.D of three independent experiments are presented. B. Telomeric constructs harboring protection against TPE (T1XDB1-3, T1XDE) and control (T, T-5′ HS4) were transiently transfected with pools of siRNA against CTCF. The expression of the eGFP reporter gene was analyzed by FACS. C. D. The different constructs allowing the protection against CPE (C1X, C1XDB1-3, C1XDE, pCMV-5′ HS4) (panel C) or protection against TPE (T1XDB1-3, T1XDE) (panel D) were transiently transfected with siRNA against Lamins A/C and the expression of the eGFP reporter gene was measured by FACS. Asterisks denote statistically significant values relative to control siRNA (Student's t test). **p<0.01; *p<0.001.

Figure 5. The multimerization of D4Z4 abrogates CTCF binding and insulation activity.

A. The expression of eGFP was measured by FACS on populations of cells transfected with telomeric constructs carrying 0, 1, 4, 8 or 12 copies of D4Z4 downstream of the telomeric seed (T, T1X, T4X, T8X, T12X) or internal constructs (pCMV, C1X, C4X, C8X) containing respectively 0, 1, 4 or 8 copies of D4Z4. The integrity of each construct was verified in stable populations of cells after either random integration or telomeric fragmentation (Figure S1D). Histograms show the average percentage of eGFP positive cells from day 18 to day 29±S.D. shown by error bars, when eGFP expression reaches a plateau. In the different constructs containing D4Z4 inserted at random sites (C4X, C8X), the level of eGFP is proportionally decreased when the number of repeats is increased suggesting that the repeated element loses its anti-CPE activity upon multimerization. On the opposite, eGFP level is slightly increased at telomeres (see main text). B. The binding of CTCF was investigated by ChIP on the different populations of cells carrying different number of D4Z4 element downstream of the eGFP reporter gene. Input DNA and DNA fraction immunoprecipitated with antibodies to CTCF were amplified by a real-time Q-PCR method (x-axis) using primers encompassing the 5′ CTCF site. The y-axis shows the fold enrichment of CTCF in the bound fraction versus input chromatin. Each data point is the average of at least three independent experiments with the S.D. shown by error bars. C. ChIP analysis of CTCF binding in two different control (CT1 and CT2, >11 D4Z4 repeats) and three different myoblasts from FSHD patients (FSHD1, 5 repeats; FSHD 2, 6 repeats; FSHD 3, 7 repeats).

Figure 6. Model explaining the role of the D4Z4 insulator and its implication in the epigenetic alteration of FSHD.

In normal cells, the multimerization of D4Z4 compromises CTCF binding and the boundary activity is counteracted (upper panel). In this conformation, the D4Z4 array might repress gene expression either at the 4q35 locus or at a long distance from the array. In patients, D4Z4 acts as an insulator that protects the expression of different loci from repressive structures such as the 4q terminus or other subtelomeric surrounding sequences. This boundary activity depends upon CTCF and Lamins A/C (lower panel). The exclusion of CTCF from multiple repeats and the presence of a silencer element within D4Z4 might suggest that the D4Z4 array behaves as a silencer. However, the presence of up to 12 copies of the repeat does not repress the expression of the neighboring eGFP reporter in our experimental settings where the D4Z4 array directly flanks the telomere and argues against the hypothesis that multiple D4Z4 repress in cis the expression of genes. An alternative explanation is that multiple D4Z4 cooperates with other elements of the 4q region to form a silencer, as suggested by the link between D4Z4 array contraction and a particular allele of 4q35 in patients .