Identification of conserved sequence and Ctcf sites

Multipipmaker (Schwartz et al. 2000) was used to align sequences corresponding to the following: Mus musculus chrX: 100,565,208–100,688,561; Homo sapiens chrX: 72,871,740–73,007,631 [University of California at Santa Cruz (UCSC) genome browser database coordinates] (Karolchik et al. 2003); and GenBank accession nos. 13445265 (Microtus arvalis), 13445266 (Microtus rossiaemeridionalis), 13445263 (Microtus transcaspicus), 1575005 (Equus caballus), 1575009 (Oryctolagus cuniculus), and 21425595 (Bos taurus). To identify potential Ctcf-binding sites, the Blat function (Kent 2002) was used with the UCSC genome browser database (Karolchik et al. 2003) to find sequences from various species corresponding to the 2.3-kb ClaI–BamHI fragment (bp 126,980–129,252 of GenBank accession no. AJ421479) from the 3′ end of M. musculus Xist, and sequences from Rattus norvegicus (chrX: 914,45,420–91,447,696; UCSC genome browser database coordinates for this and subsequent species), Oryctolagus cuniculus (scaffold_214795:36610:38552), Equus caballus (chrUn: 222,426,453–222,428,947), Canus familiaris (chrX: 60,374,176–60,376,292), Pan troglodytes (chrX: 73,151,334–73,153,817), Felis catus (scaffold_203485:24-1902), and Macaca mulatta (chrX: 72,942,832–72,945,325). These sequences, along with sequence from H. sapiens Xic (bp 3974–6554 of GenBank accession no. AL353804), were searched with the Ctcf-binding site matrix (Kim et al. 2007) using the program rVista (Loots et al. 2002) (parameters: 0.95 Core Matrix Similarity, 0.75 Matrix Similarity) to locate potential binding sites within RS14.

Electrophoretic mobility shift assay

Complementary single-stranded oligonucleotides were ordered from Integrated DNA Technologies (RS14d) or from the Massachusetts General Hospital DNA Core Facility. The complementary oligonucleotides were as follows: H19, 5′-CAGTTGTGGGGTTTATACGCGGGAGTTGCCGCGTGGTGGCAGCAAAATCGATTGCGCCAAACCTAAAGAG-3′ (MS1) (Hark et al. 2000); H19-reverse, 5′-CTCTTTAGGTTTGGCGCAATCGATTTTGCTGCCACCACGCGGCAACTCCCGCGTATAAACCCCACAACTG-3′; RS14c, 5′-TGTTATACCCGTGTAGGCCAGCAGAGGGTGTCGGATCCCAAGGAAA-3′; RS14c-reverse, 5′-TTTCCTTGGGATCCGACACCCTCTGCTGGCCTACACGGGTATAACA-3′; RS14c-mut, 5′-TGTTATACCCGTGTAGGCCAGCAAAAAATGTCGGATCCCAAGGAAA-3′; RS14c-mut-reverse, 5′-TTTCCTTGGGATCCGACATTTTTTGCTGGCCTACACGGGTATAACA-3′; RS14d, 5′-CTGTGCTTTGACCCTTTCCATTTTTCTTTCACTCACAGAGGGTGGGACAGGAGGCCGAGTGAAGGAAAGGGTCCAGCCTG-3′; and RS14d-reverse, 5-CAGGCTGGACCCTTTCCTTCACTCGGCCTCCTGTCCCACCCTCTGTGAGTGAAAGAAAAATGGAAAGGGTCAAAGCACAG. The complementary single-stranded oligonucleotides were suspended together in duplex buffer from Integrated DNA Technologies (RS14d) or PCR buffer, heated to 90° for 2 min, and cooled to room temperature to form double-stranded oligonucleotides. Double-stranded oligonucleotides were labeled using phosphonucleotide kinase and γ³²P-ATP. For gel-shift analyses, full-length purified recombinant Ctcf protein (Novus Biologicals catalog no. H00010664-P01, amino acids 1–728) and 0.2 pmol ³²P-labeled oligonucleotides were incubated at room temperature for 20 min in 20 mm HEPES (pH 7.5), 50 mm KCl, 5 mm MgCl₂, 3.3 μm ZnSO₄, 1 mm dithiothreitol, 0.3 mg/ml BSA, 0.5 μg poly(dI:dC), 5% glycerol, and 0.5% triton X-100. Cold competitors were added at 1000× molar excess. Shifts were resolved in 5% acrylamide/0.5× TBE gels at 4° which were fixed in 50:10:40 methanol:acetic acid:water and dried.

Chromatin immunoprecipitations

Chromatin immunoprecipitation (ChIP) analysis was carried out as described (Lee et al. 2006) on each of day 0 and day 4 male and female ES cells, and the results were averaged for three independent biological replicates. α-Ctcf antibodies (sc-15914) and IgG polyclonal serum (sc-2028) were obtained from Santa Cruz Biotechnology. Quantitative PCR was performed using an iCycler iQ real-time PCR detection system (Bio-Rad) by amplifying with primer pairs specific to RS14c (p142/p139, 87-bp product), RS14d (p140/p141, 102-bp product), control site A (p134/135, 87-bp product, which is 1.2 kb from RS14c), and control site B [p136/137, 86-bp product, which is 0.8 kb from RS14d (1.0 kb from RS14c)]. Primer sequences were the following: p134, ACTCATCAGTGGATACGTTTGAGTGCC C; p135, CTCCAAGAAGCCCATGAAACCTTACC; p136, GCAAATCCTTGTAGAACAGGCGAACC; p137, CCCACAGTTTGCCTCAGAGGTTGAATAC; p139, TAGGTTTGGGTGTTATACCCGTGTAGGC; p14, CTTAGATTCCAGATAGACAGGCTGGACC; p141, AATGCCTGTGCTTTGACCCTTTCC; and p142, GGCTACACACAAGATGGCGTCTGTAACTTG.

Enhancer-blocking assay

The enhancer-blocking assay was carried out essentially as described (Chung et al. 1993; Bell et al. 1999). To produce the RS14 forward and reverse plasmids, a 2.3-kb ClaI–BamHI fragment [bp 126,980–129,252 of GenBank no. AJ421479 (Chureau et al. 2002)] at the 3′ end of Xist was ligated, using AscI linkers, into the AscI site of the no-insulator plasmid in the forward and reverse directions. Constructs were linearized by SalI digestion, phenol–chloroform extracted, and ethanol precipitated. Control constructs with β-globin insulators or λDNA are as described (Chung et al. 1993). A transfection efficiency control vector, pTK–hygromycin (Chao et al. 2002), was linearized with BglI and prepared in the same way. K562 cells were grown in Iscove’s Modified Dulbecco’s Medium with L-glutamine, 10% fetal bovine serum, 1% penicillin–streptomycin (Gibco 15140-122), and 1.5 g/liter sodium bicarbonate. K562 cells (10⁷) were electroporated with 3 pmol of each construct and 23 μg of pTK–hygromycin and were recovered for 2 days in K562 media. Electroporated cells were split into two pools and were plated in soft agar composed of K562 media with 0.42% agar, 1 μg/ml fungizone, 10 μg/ml ketoconazole, and either 1 mg/ml of neomycin (G418) or 200 μg/ml hygromycin. G418 and hygromycin-resistant colonies were counted 3–4 weeks after plating, and sample pairs with <50 hygromycin colonies were discarded. The ratios of G418 to hygromycin-resistant colonies were normalized to the value obtained with the no-insulator construct. P-values were calculated using the unpaired two-tailed Student’s t-test in pairwise comparisons.

Targeted mutagenesis

The PGKneobpAlox2DTA vector (Soriano 1997) was used to construct the targeting vector for RS14. Fragments corresponding to bases 119,800–126,978 and 129,252–1,322,274 of GenBank no. AJ421479 (Chureau et al. 2002) were ligated into the SacII site and SalI–HindIII sites, respectively, to create a targeting vector with a 7.2-kb homologous arm, followed by a PGK-driven neomycin resistance expression cassette flanked by loxP sites, a 3-kb homologous arm, and a diphtheria toxin (PGK-DTAbpA; negative selection) expression cassette in a pBluescript II KS+ backbone. PvuI was used to linearize the construct, and 40 μg was electroporated into ~10⁷ J1 and EL16 cells. Colonies were selected with G418, and homologous recombination events were screened by Southern blot using SpeI and probe 1, a probe external to the targeting construct (bp 118,989–119,739 of GenBank no. AJ421479). Single insertion of the vector and correct targeting of the 129 allele was confirmed with Southern blot of a BtgI digest using probe 2, which is located internal to the targeting construct (bp 129,826–130,568 of GenBank no. AJ421479). Targeted clones were transfected via electroporation with pMC-CRE to delete Neo via Cre-mediated recombination, and deletion was confirmed by Southern blot using BlpI and probe 1. As is the case with all female ES cell lines, the targeted female clones have a heterogeneous mixture of diploid and tetraploid cells.

Fluorescent in situ hybridization and antibody stains

RNA/DNA fluorescent in situ hybridization (FISH) was carried out as previously described (Lee and Lu 1999; Zhang et al. 2007). Wild-type male (J1) (Li et al. 1993) and female (EL16) (Lee and Lu 1999) ES cells and the RS14 mutant male (B7Δ2.3) and female (A9Δ2.3) ES cells generated in this study were grown and treated under identical conditions. Embryoid body differentiation was induced by LIF withdrawal, and cells were trypsinized and harvested at days 0, 4, and 10. Cells were cytospun and fixed onto superfrost plus slides (Fisherbrand). Detection of Xist RNA was carried out using a pSx9 probe labeled by a nick translation kit (Roche) using Cy3-dUTP (Amersham). Images of the cells were taken with 0.5-μm z-sections using Volocity software (Improvision) and a Nikon Eclipse 90i microscope. DNA FISH with FITC-labeled X-paint (Cambio) was subsequently used to identify X chromosomes. Signals corresponding to overlapping Xist RNA coating of an X chromosome were counted using the extended focus view of Volocity. A minimum of 271 cells was characterized at each time point. For co-labeling of the Xic (FITC) and Neo cassette (Cy3), DNA FISH was performed with nick-translated probes generated from pSx9 and a 2.3-kb BtsI–XcmI fragment from the RS14 deletion plasmid. Prior to DNA FISH slide denaturation, samples were treated with both RNAseA and RNAseH. Images of the cells were taken with 0.2-μm z-sections.

Immunofluorescence was performed as previously described (Zhang et al. 2007). Undifferentiated ES cells and day 8 cells from embryoid bodies (EBs) were fixed and incubated with rabbit anti-Nanog antibody (Novus) at 1:750 dilution and visualized with goat anti-rabbit-alexa555 (Invitrogen) at 1:1000 dilution. For all FISH and antibody imaging, nuclei were stained with DAPI in Vectashield mounting medium.

Quantitative RT-PCR

RNA was isolated from embryoid bodies derived from wild-type male (J1) and female (EL16.7) cells and RS14 mutant male (B7Δ2.3) and female (A9Δ2.3) cells using Trizol (Invitrogen). First-strand synthesis was performed with Superscript III RT (Invitrogen) using gene-specific primers to Xist (BD127, 5′-TAAACAGCCAGAGTCTGATGTAACGG-3′), Tsix (NS19) (Stavropoulos et al. 2001), and tubulin α1A (YJ2, 5′-CTTGCCAGCTCCTGTCTCAC-3′). Real-time reactions were performed in a CFX96 Real-Time PCR System using iQ SYBR Green Supermix (Bio-Rad). Xist was amplified with primers NS66 (Stavropoulos et al. 2001) and BD127. Tsix was amplified with NS18 and NS19 (Stavropoulos et al. 2001). Tubulin α1A was amplified with primers YJ1 (5′-CTCGCCTCCGCCATCCACCC-3′) and YJ2. Expression of Xist and Tsix was calculated relative to tubulin α1A normalized to primer pair efficiency.

Allele-specific RT-PCR

Allele-specific RT-PCR was performed as described (Stavropoulos et al. 2001). Briefly, total RNA was isolated from retinoic acid differentiated cells, and 2 μg was reverse transcribed with random hexamer primer. Primers NS33 (5′-CAGAGTAGCGAGGACTTGAAGAG-3′) and NS66 (5′-GCTGGTTCGTCTATCTTGTGGG-3′) and 32 cycles of PCRwere used to amplify polymorphic Mus musculus castaneus and 129 fragments which were digested with ScrfI. These fragments were fractionated on an agarose gel, blotted, and probed with the ³²P-end-labeled probe NS67 (5′-CCAGAGTCTGATGTAACGGAGG-3′). Relative allelic amounts were quantitated by phorphorimage using the software ImageQuant. P-values were calculated using the unpaired two-tailed Student’s t-test in pairwise comparison.

Quantitation of cell death

Day 0 cells were trypsinized and 6 × 10⁶ cells were differentiated with 100 nM retinoic acid. On day 1, cells were plated on gelatinized plates (0.2%) with 200 nM retinoic acid, and for all subsequent days were differentiated with 13 nM of all trans-retinoic acid. At each time point, the level of cell death was measured in triplicate using the Multitox-Fluor Multiplex Cytotoxicity Assay (Promega).

RS14 binds Ctcf protein and displays enhancer-blocking activity

Given the presence of a boundary-like activity at the 3′ end of Xist, we asked if RS14 might bind the only known mammalian chromatin insulator, Ctcf, especially in view of the fact that Ctcf is implicated in formation of an ACH at the H19/Igf2 imprinted locus (Murrell et al. 2004; Kurukuti et al. 2006; Ling et al. 2006) and at the β-globin locus (Splinter et al. 2004). A search with the 20-bp Ctcf-binding site matrix (Kim et al. 2007) revealed two potential elements within RS14 that are conserved among multiple eutherian mammals. One is located in RS14d and shows conservation among nine eutherian species (Figure 2, A and B). We also found a cluster of closely spaced potential Ctcf motifs in rodents downstream of RS14d in a 110-bp domain that we designate RS14c, the best conserved of which is shown in Figure 2B.

Open in a separate window

Figure 2

RS14 binds Ctcf in vitro and in vivo. (A) The RS14 region contains two putative Ctcf-binding sites within a ClaI–BamHI restriction fragment that harbors RS14d and RS14c (locations shown by blue circles). Control sites for Ctcf ChIP are designated “A” and “B.” (B) Multispecies alignment of the consensus Ctcf motif (Kim et al. 2007) at RS14d and RS14c. The RS14d M. musculus sequence corresponds to chrX: bp 100,656,561–100,656,608 (UCSC Genome Browser); RS14c M. musculus sequence corresponds to chrX: bp 100,655,589–100,655,636. Control sites A and B lack obvious Ctcf motifs and are 1.2 kb from RS14c and 0.8 kb from RS14d, respectively. The BamHI site next to RS14c is underlined in red. Asterisks mark mutated bases in RS14c-mut. Arrow marks the highly conserved C nucleotide of the Ctcf-binding consensus. (C) EMSA shows that RS14c binds Ctcf protein and depends upon four core G nucleotides for binding. Radioactively labeled probe RS14c was shifted with purified Ctcf protein with or without unlabeled competitor DNA as indicated. Arrow, specific protein–DNA complex. (D) Ctcf and control IgG ChIPs at RS14 and flanking positions in day 0 and day 4 male and female ES cells. Statistical significance (P < 0.05) is determined by a paired two-tailed Student’s t-test comparing Ctcf ChIP to the corresponding IgG ChIP samples. Statistically significant localizations are marked by an asterisk.

To determine if the sites could bind Ctcf in vitro, we performed electrophoretic mobility shift assays (EMSA) using purified recombinant Ctcf protein and labeled 46- and 80-bp oligonucleotides corresponding to RS14c and RS14d, respectively. Although RS14d contains a conserved Ctcf motif, the DNA did not shift Ctcf protein under our conditions (data not shown). A more careful examination of the RS14d motif indicated that the highly conserved C-nucleotide in the sixth position of the Ctcf consensus is a T nucleotide instead in RS14d (Figure 2B, arrow). This difference may explain the inability to bind Ctcf protein. By contrast, murine RS14c carries the C nucleotide and consistently shifted Ctcf protein (Figure 2C). The protein–DNA complex was eliminated when four conserved G nucleotides within the core motif were mutated (RS14c-mut) (Figure 2B, asterisks). Furthermore, the protein–DNA complex was competed away by addition of unlabeled self-DNA or H19, which contains a known Ctcf-binding site (Bell and Felsenfeld 2000; Hark et al. 2000; Kanduri et al. 2002), but was still present with the addition of unlabeled RS14c-mut (Figure 2C). Combined, these data indicate that Ctcf specifically binds RS14c in vitro.

To determine if RS14 is occupied by Ctcf in vivo, we carried out ChIPs using α-Ctcf antibodies in male and female ES cells before (day 0) and during XCI (day 4) and compared the results to control IgG ChIP (Figure 2D). Consistent with EMSA, we observed significant enrichment of Ctcf at RS14c over background (IgG) in both male and female ES cells and at both day 0 and day 4, indicating that Ctcf strongly occupies RS14c in vivo. At RS14d, the occupancy was much lower than at RS14c, although the enrichment was statistically significant in day 4 male and female ES cells when compared to the results of IgG ChIP. Analysis of two published ChIP-seq data sets for ES cells showed that RS14c but not RS14d was detected above the significance cutoffs in both data sets (Chen et al. 2008; Nitzsche et al. 2011).

We then performed a reciprocal bioinformatics analysis and asked whether the RS14c motif appeared more frequently in known Ctcf-binding sites, as defined by the published ChIP-seq data sets for ES cells. We generated sequence motifs based on alignments of RS14d and RS14c sequences in various species and used them as queries to search the published binding sites with the FIMO method (Grant et al. 2011). In total, the RS14c motif matched 438 sequences in the Chen et al. (2008) data set and 10,008 sequences in the Nitzsche et al. (2011) data set with a significance cutoff of q-value <0.05 (Figure 2E; Supporting Information, Table S1; Table S2). By contrast, the RS14d motif did not identify any significant matches in either data set [the best q-values were 0.345 and 0.277 for the Chen et al. (2008) and Nitzsche et al. (2011) data sets, respectively].

Taken together, the in vivo, in vitro, and in silico analyses argue strongly that RS14c is a bona fide binding site for Ctcf. Regarding RS14d, we consider one of three possibilities: (1) that RS14d is also directly bound by Ctcf in vivo, albeit at a much reduced level that could be identified by ChIP-qPCR (Figure 2D) but escaped detection by ChIP-seq (Figure 2E); (2) that RS14d indirectly interacts with Ctcf via other motifs and proteins; or (3) that RS14d’s proximity to RS14c resulted in its co-immunoprecipitation with the RS14c–Ctcf complex. Given EMSA results and the base difference at the conserved “C” in RS14d, one of the latter two explanations may be more likely. Likewise, the control flanking sites A and B, which do not have any obvious Ctcf motifs, also showed very slight enrichment for Ctcf binding in day 4 female ES cells, perhaps due to similar reasons. We conclude that Ctcf strongly occupies RS14c in vivo.

We next investigated if RS14 might possess enhancer blocking activity (Figure 3A). In this assay, a test sequence is inserted between the β-globin locus control region and a neomycin-resistance reporter (Neo^R) to determine the degree to which the test element can decrease the number of Neo^R colonies in a K562 background [a flanking insulator (INS) protects against position effects] (Chung et al. 1993; Bell et al. 1999). Whereas a test sequence containing either λDNA or no insulator yielded plentiful colonies, two copies of the established β-globin insulator led to a more than fourfold reduction in colony number, as expected (Figure 3B). A 2.3-kb test fragment containing RS14 likewise resulted in a significant reduction of Neo^R colonies when compared to λ or no insulator controls (P < 0.005), consistent with RS14 acting as a chromatin insulator. RS14 could block enhancer–promoter interactions in both forward and reverse orientations, as is the case with insulators in other contexts (Ohlsson et al. 2001; West et al. 2002). Taken together with the EMSA and ChIP experiments discussed above, these data demonstrate that RS14 binds Ctcf and exhibits classic enhancer blocking activity.

Open in a separate window

Figure 3

RS14 contains enhancer-blocking activity. (A) The schematic shows the RS14 test region and the vector construct (below) used in the enhancer-blocking assay. INS, β-globin insulator. NEO, neomycin resistance gene. LCR, locus control region (enhancer). (B) (Left) The enhancer-blocking assay using the constructs. (Right) Graph showing the relative number of neomycin-resistant (NEO^R) colonies, normalized to hygromycin-resistant colonies (the transfection efficiency control) and presented relative to the number of colonies from the “no insulator” construct (standardized to 1.0). The average and standard errors are shown for 10–11 experiments. A 2.3-kb RS14 fragment produced a statistically significant reduction in Neo^R colony number in either orientation when compared to λ or no insulator controls (P < 0.005), as determined by unpaired two-tailed Student’s t-tests in pairwise comparisons.

Targeted deletion of RS14

To investigate the function of RS14 during XCI in vivo, we generated a deletion of a 2.3-kb sequence corresponding to RS14 (Figure 4A). One concern that arises when creating Xist deletions in this context is the need to preserve the silencing function of Xist. Two previous studies indicated that the 2.3-kb deletion would not affect Xist’s silencing function. In one study, Xist cDNA transgenes lacking the 3′ end of Xist and RS14 retained the RNA’s ability to bind the X, spread in cis, and establish XCI (Wutz et al. 2002). In a second study, a 65-kb deletion of the Xic (Δ65; see Figure 1A), which also removed the 2.3-kb sequence in addition to Tsix and Xite, resulted in a fully functional Xist RNA that constitutively silenced the X in cis (Clerc and Avner 1998; Morey et al. 2004). Here, we matched the 5′ border of Δ2.3 to Δ65 to preserve Xist RNA function. Among the ~2500 XY and 1700 XX clones picked, two correctly targeted clones (Neo⁺) of each sex were identified (XY, B7Δ2.3 and D7Δ2.3; XX, A9Δ2.3 and B5Δ2.3) and verified in Southern blot analysis using 5′ and 3′ external probes (Figure 4, B and C). Two Neo⁻ clones were subsequently derived for each (Figure 4D). A BtgI polymorphism (B′) present between the M. musculus (129) and M. castaneus (cas) Xs confirmed that the deletion was targeted to the 129 chromosome in each case.

Open in a separate window

Figure 4

Targeted deletion of RS14. (A) Targeting scheme for the 2.3-kb RS14 deletion. The expanded region depicts the targeted area of the Xic. Blue rectangles, RS14. Black rectangles, Xist exons. WT, wild type. S, SpeI. L, BlpI. B, BtgI. B′, BtgI restriction site found only on 129 allele. HI, BamHI. C, ClaI. M, MnlI. Black triangles, LoxP sites. Neo, neomycin-resistance gene. DTA, diphtheria toxin A. Δ2.3, RS14 knockout allele. ΔNeo, neomycin-resistance reporter excised with transient Cre expression. (B–D) Southern blot analysis of new RS14 alleles. Knockout alleles are indicated by “Δ2.3” or “ΔNeo” labels of the expected bands hybridizing to the relevant probes. (B) Wild-type male ES (J1), wild-type female ES (EL16), knockout female A9Δ2.3 and B5Δ2.3, and knockout male B7Δ2.3 and D7Δ2.3. SpeI-digested genomic DNA was hybridized with external probe 1 from A. (C) Same lines as in B. BtgI-digested genomic DNA was hybridized with external probe 2 from A. (D) ΔNeo lines produced from Cre-mediated excision of the Neo^R cassette. BlpI-digested genomic DNA was hybridized with probe 1 from A.

We are grateful to all lab members for discussion and support. R.J.S. is supported by the Medical Scientist Training Program at Harvard Medical School. S.F.P. is supported by a Deutche Forschungsgemeinschaft research fellowship. This work is funded by a grant from the National Institutes of Health (RO1-GM58839). J.T.L. is an Investigator of the Howard Hughes Medical Institute.