UBC Faculty Research and Publications

XIST-induced silencing of flanking genes is achieved by additive action of repeat a monomers in human… Minks, Jakub; Baldry, Sarah E; Yang, Christine; Cotton, Allison M; Brown, Carolyn J Aug 1, 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-13072_2013_Article_294.pdf [ 279.44kB ]
JSON: 52383-1.0223296.json
JSON-LD: 52383-1.0223296-ld.json
RDF/XML (Pretty): 52383-1.0223296-rdf.xml
RDF/JSON: 52383-1.0223296-rdf.json
Turtle: 52383-1.0223296-turtle.txt
N-Triples: 52383-1.0223296-rdf-ntriples.txt
Original Record: 52383-1.0223296-source.json
Full Text

Full Text

RESEARCH Open AccessXIST-induced silencing of flanking genes isachieved by additive action of repeat amonomers in human somatic cellsJakub Minks, Sarah EL Baldry, Christine Yang, Allison M Cotton and Carolyn J Brown*AbstractBackground: The establishment of facultative heterochromatin by X-chromosome inactivation requires the longnon-coding RNA XIST/Xist. However, the molecular mechanism by which the RNA achieves chromosome-widegene silencing remains unknown. Mouse Xist has been shown to have redundant domains for cis-localization, andrequires a series of well-conserved tandem ‘A’ repeats for silencing. We previously described a human inducibleXIST transgene that is capable of cis-localization and suppressing a downstream reporter gene in somatic cells, andhave now leveraged these cells to dissect the sequences critical for XIST-dependent gene silencing in humans.Results: We demonstrated that expression of the inducible full-length XIST cDNA was able to suppress expressionof two nearby reporter genes as well as endogenous genes up to 3 MB from the integration site. An inducibleconstruct containing the repeat A region of XIST alone could silence the flanking reporter genes but not the moredistal endogenous genes. Reporter gene silencing could also be accomplished by a synthetic construct consistingof nine copies of a consensus repeat A sequence, consistent with previous studies in mice. Progressively shorterconstructs showed a linear relationship between the repeat number and the silencing capacity of the RNA.Constructs containing only two repeat A units were still able to partially silence the reporter genes and could thusbe used for site-directed mutagenesis to demonstrate that sequences within the two palindromic cores of therepeat are essential for silencing, and that it is likely the first palindrome sequence folds to form a hairpin,consistent with compensatory mutations observed in eutherian sequences.Conclusions: Silencing of adjacent reporter genes can be effected by as little as 94 bp of XIST, including two‘monomers’ of the A repeat. This region includes a pair of essential palindromic sequences that are evolutionarilywell-conserved and the first of these is likely to form an intra-repeat hairpin structure. Additional sequences arerequired for the spread of silencing to endogenous genes on the chromosome.Keywords: X-chromosome inactivation, XIST, Long non-coding RNA, Eutherian dosage compensation, GenesilencingBackgroundTo ensure dosage compensation of X-linked genes be-tween males and females, eutherian females silence oneX chromosome [1]. The minimal region required for X-chromosome inactivation contains the non-coding (nc)RNA gene XIST, which is expressed solely from the in-active X chromosome [2]. Experiments in mice haveshown that Xist is both required and sufficient forinactivation; however, the mechanism by which theXIST/Xist RNA causes chromosome-wide gene silencingremains elusive (reviewed in [3]). XIST localizes in cis tothe chromatin of the inactive X chromosome [4],suggesting a potential role in targeting silencing com-plexes to the chromosome. The alternatively spliced andpolyadenylated RNA is over 15 kb long in all eutheriaexamined. Overall the gene is only weakly conservedamong mammals, but its regions of repetitive sequencescalled repeat A to F show better conservation [4,5]. Add-itionally, exon 4 of XIST/Xist is well-conserved, and* Correspondence: carolyn.brown@ubc.caDepartment of Medical Genetics, Molecular Epigenetics Group, University ofBritish Columbia, Life Sciences Institute, 2350 Health Sciences Mall,Vancouver, BC V6T 1Z3, Canada© 2013 Minks et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.Minks et al. Epigenetics & Chromatin 2013, 6:23http://www.epigeneticsandchromatin.com/content/6/1/23shows homology with the protein-coding Lnx3 gene,from which the Xist gene may have evolved by theaddition of sequences from transposable elements [6,7].Intriguingly, in marsupials, Lnx3 remains protein-codingand Rsx3 encodes an RNA that is similar to XIST in thatthe long non-coding, repeat-rich RNA is transcribedfrom and associates with the inactive X chromosome [8].While there is no sequence conservation between Rsx3and XIST, both are able to silence in cis, and show re-gions of putative stem-loop structure, supporting theidea that these long ncRNAs may be serving as adaptermolecules containing different protein-recognition mo-tifs to recruit components of the gene-silencing machin-ery to the inactive X chromosome.As X-chromosome inactivation is a developmentalprocess, most studies of Xist function have been under-taken in mice, where embryonic stem (ES) cells or em-bryos can be analyzed during the inactivation process.Human ES cells have demonstrated considerable epigen-etic instability (for example, [9]) and studies of humanembryos are necessarily restricted ([10,11]). However,the potential differences in the inactivation process be-tween mice and humans, suggested by both differencesin the regulation of the XIST gene and the number ofgenes escaping inactivation (reviewed in [12]), led us todevelop an inducible model to study human XIST action[13]. Induced expression of XIST in the immortalHT1080 fibrosarcoma cell line is able to induce somefeatures of an inactive X, including XIST localization, si-lencing of a co-integrated reporter gene, depletion of re-petitive (CoT1) RNA, and the acquisition of someheterochromatic histone modifications associated withthe inactive X.Previous studies in mice targeted a panel of truncatedinducible Xist transgenes to the single X chromosome in amale ES cell and demonstrated that redundant sequenceswere involved in localization of the mouse Xist RNA tothe chromosome [14], with a construct containing onlyapproximately 3 kb of Xist cDNA, including the well-conserved A repeat region, able to localize to and repressthe single X chromosome. Furthermore, chromosomalsilencing was fully compromised when the 5’ regionencompassing repeat A was deleted [14], but concatamersof a synthetic version of these repeats were able to replacethe A repeat region. A near normal complement of 7.5 re-peats or an increase to 12 repeats fully recapitulated silen-cing, while 5.5 repeats showed less silencing and 4repeats were only minimally active [14]. Therefore, inmice the A repeats are necessary for silencing, but add-itional redundant domains of Xist are involved inlocalization to the chromosome and the presence ofdifferent domains supports models that the RNA isserving as an adaptor to bring different epigenetic silen-cing proteins to the inactive X.A number of chromatin remodeling proteins associatewith the inactive X chromosome resulting in the acquisi-tion of many histone modifications characteristic of het-erochromatin (reviewed in [3]). The binding of many ofthese proteins is Xist-dependent; and it has been shownthat the A repeat region interacts in vitro and in vivowith components of PRC2 [15-17]. Surprisingly, how-ever, a silencing defective Xist RNA that lacks the repeatA region is still able to recruit PRC2, PRC1, SAF-A,ASH2L and macroH2A1 to the inactive X chromosomein ES cells (reviewed in [3]). In contrast, a similar dele-tion in transgenic mice failed to produce Xist RNA,suggesting an important regulatory role for the repeat Aregion [18]. Furthermore, interaction with the transcrip-tional repressor YY1 [19] has been shown to occur atthe mouse C repeat region and while a direct interactionwith the A repeat region has been reported for the spli-cing factor ASF/SF2, this has been proposed to have arole in enabling proper processing of the Xist RNA fa-cilitating choice of the future inactive X chromosome[20]. Therefore, despite the growing body of literatureon XIST/Xist-interacting partners and identification of acritical role for the A repeat region, understanding howXIST/Xist expression leads to gene silencing remainselusive. Contributing to the challenge is the large size ofthe XIST RNA, and that the monitoring of silencing atdistal sites requires both silencing and spread of theRNA along the chromosome.The palindromic nature of the repeat A core se-quences suggests their involvement in forming a distinctsecondary RNA structure, and several alternative butmutually exclusive structures have been suggested. Thefirst model proposed that each of the two palindromesforms a hairpin and thus the repeat A region of XISTRNA folds into a two-hairpin 8- or 9-mer [14]. Thisstructure was supported by the abrogation of silencingactivity in a construct with two base alterations thatwould disrupt the putative first hairpin. However, anin vitro analysis of repeat A structure by fluorescenceresonance energy transfer, as well as sensitivity to RN-ases that specifically digest single- or double-strandedRNA regions, proposed an alternative structure. The firstpalindrome was suggested to engage in pairing betweentwo separate monomers, rather than within repeat Amonomers, and the model proposed that the second pal-indrome did not form a defined structure [16]. Recently,a third option, supported by nuclear magnetic resonanceanalyses of repeat A monomer and dimer structures,suggested that under in vitro conditions, the first palin-drome forms a hairpin, while the second palindrome en-gages in pairing between repeat A units [21,22].Our previously reported inducible transgenic system inthe immortal fibrosarcoma line HT1080 provides a tract-able system to study the RNA sequences involved in geneMinks et al. Epigenetics & Chromatin 2013, 6:23 Page 2 of 10http://www.epigeneticsandchromatin.com/content/6/1/23repression by XIST [13]. Here, we focus on refining theminimal XIST sequence necessary for cis-regulated silencing,independent of the developmental signals that establishmono-allelic XIST expression in females. We demonstratesilencing of reporter genes by expression of less than 100 bpof XIST containing two consensus repeat A monomers.Results and discussionRepeat A is sufficient for XIST-dependent reporter genesilencingWe have previously shown that an inducible transgenicXIST is capable of silencing an Enhanced Green FluorescentProtein gene (EGFP) reporter in human somatic cells, whilea construct lacking the repeat A region failed to silence theEGFP gene [13]. Similarly, inducible mouse constructs havebeen shown to require the repeat A region for silencing ofthe X chromosome in mouse ES cells [14]. The RNA in-duced from the full-length XIST cDNA construct localizesto the autosome upon which it has integrated [13]; however,the EGFP reporter construct is located only 7.7 kb 3’ ofXIST in HT1080 male fibrosarcoma cells (see Figure 1A)and, thus, may not require the localization domains of XISTfor silencing. Therefore, to test whether the repeat A is suf-ficient for proximal gene silencing, we induced expression~3Mb50 kbtelomerecentromere~20Mb ~10Mb ~3Mb ~2MbA BC0. p2 p3 p4cDNArelative to genomic DNA replicate 1 replicate d1 d2 d3 d4 d5full length 5'AEGFPexpression normalized to no DOX2 kbXISTp1 p2 p3 p4p5vectordel 5’A5’Afull lengthD EExpression relative to ACTB[normalized to no DOX]Δ (no DOX-DOX) / no DOX-20%-10%0%10%20%30%40%50%60%70%SKIL TTC14 BCL6 LPP OPA1full length 5'A del 5'A vector0. EGFP CLDN16 IL1RAPfull length 5'A del 5'A vector*****Figure 1 The repeat A region of XIST is necessary and sufficient for silencing of flanking reporter genes. (A) Approximate location ofgenes analyzed on chromosome 3 relative to the schematic of full-length XISTcDNA construct showing regions included in shorter XISTconstructs and location of qRT-PCR primer pairs p1 to p4 and p5 (vector primer pair used to amplify all XIST constructs). (B) Enhanced GreenFluorescent Protein gene (EGFP) expression following one to five days (d1 to d5) induction of full-length XIST or 5’A, measured by flow cytometryand shown relative to d0. (C) qRT-PCR analysis of expression within full length XIST transgene (p2) and upstream (p1) and downstream (p3, p4) ofXIST sequence. Genomic DNA was used to normalize for amplification efficiency. Location of qPCR amplicon positions is shown in Figure 1A.(D) Expression of the reporter genes (Hygromycin gene (Hyg) and EGFP) and endogenous genes CLDN16 and IL1RAP following five days oftransgene induction measured by qRT-PCR, relative to expression in uninduced cells (d0) and normalized to ACTB expression. Transgeneconstructs were full XIST, 5’A only, full XIST lacking the 5’A region or vector with no XIST as indicated. Error bars indicate ± 1 S.D. of four to sixbiological replicates. Significance (P-value <0.05) was calculated using a Mann–Whitney test comparing each transgene construct with the vectoralone construct. (E) Allele-specific silencing of flanking endogenous genes following five days of transgene induction. The percent change inallelic ratio upon DOX induction relative to the ratio without DOX was measured by pyrosequencing for expressed polymorphisms in five genesup to 20 Mb from the integration site (see A). Transgene constructs were full XIST, 5’A only, full XIST lacking the 5’A region or vector with no XISTas indicated. Two technical replicates of three biological replicates were averaged for each datapoint.Minks et al. Epigenetics & Chromatin 2013, 6:23 Page 3 of 10http://www.epigeneticsandchromatin.com/content/6/1/23of a construct containing only repeat A sequence (5’A) andmeasured EGFP expression by flow cytometry (Figure 1B).The extent and dynamics of EGFP silencing by repeat Amimicked that of the full XIST construct over five days fol-lowing induction of the construct’s expression by doxycyc-line (DOX), suggesting that the ability of XIST to silencethe proximal EGFP reporter gene is attributable to the re-peat A region.To confirm that silencing results from an XIST RNA-related, sequence-specific effect rather than transcrip-tional interference, we demonstrated that transcriptionof the DOX-induced XIST transgenes ceased before thereporter construct. While some transcripts were presentdownstream of the polyadenylation site, transcriptionwas completely absent at a site approximately 2 kb 5’ ofthe EGFP promoter (Figure 1C). Our conclusion that si-lencing is not due to transcriptional interference is fur-ther supported by XIST-dependent attenuation of theexpression of the hygromycin resistance gene (Hyg) lo-cated upstream of XIST and absence of gene silencingwith vector lacking XIST sequences (Figure 1D).Endogenous gene silencing induced by full-length XISTIn order to explore whether XIST RNA is able to inducesilencing of the endogenous genes flanking the integra-tion site, we identified the FRT integration site intowhich subsequent single-copy integrations were directed.DNA-FISH using the XIST cDNA identified the full-lengthtransgene as integrated onto the der(11)t(3;11) of 46,XY,del(1)(p21),i(3)(p10),i(3)(q10),der(4)t(1;4)(p21;p16),der(5)t(5;5)(p15;?),der(11)t(3;11)(q11;q25) cells. We used inverse PCRfrom primers in the pFRT/lacZeo plasmid to identify the3q FRT integration site as just downstream of the CLDN1gene (Figure 1A). Low expression levels of the CLDN1,TMEM207 and LEPREL1 genes prevented a reliable ana-lysis of these adjacent genes by qRT-PCR. Using qRT-PCRfollowing induction of full-length XIST, we observed sig-nificant silencing of CLDN16, a gene located approxi-mately 100 kb downstream of XIST (Figure 1D). Neitherthe construct consisting only of repeat A, nor the con-struct containing a deletion of repeat A showed significantsilencing of CLDN16 upon induction, although there wasa non-significant reduction with the repeat A-containingconstruct. IL1RAP, which is located a further 120 kbdownstream (that is, 220 kb from XIST) did not showsignificant XIST-induced silencing, although there wasa non-significant drop in expression. The decrease inCLDN16 transcription is consistent with the almostcomplete silencing of the cis-located allele; however, at-tempts to confirm silencing of the XIST-associated al-lele by FISH failed, presumably due to the relatively lowexpression levels of CLDN16. In order to examinewhether one allele of endogenous genes was being si-lenced, we identified more distal genes that containedan expressed polymorphism, and thus provided an op-portunity to probe allelic silencing. At the DNA levelthese genes show an allelic ratio of approximately 66%,consistent with the presence of a single allele on theder(11)t(3;11) and the alternate allele in two copies onthe isochromosome 3q. Upon DOX treatment therewas a significant decrease in relative expression of thesingle allele for BCL6, LPP and OPA1 (Figure 1D),shown as a change upon DOX induction, relative to theexpression in cells without DOX treatment, as therecan be variations in allelic expression levels. Similar tothe q-PCR results with CLDN16, constructs containingXIST lacking repeat A, or no XIST (vector sequencesonly) showed no change in allelic ratio upon DOX in-duction; however, in these cell lines the DNA ratioshowed an equivalent allelic DNA ratio, reflectingkaryotypic instability of the HT1080 line. There was asignificant allelic silencing of BCL6 with the constructcontaining only repeat A; however, the reduced silen-cing seen with this construct suggests that additionalsequences are required for the spread of the XIST-induced silencing effect beyond the immediate XISTdomain.As repeat A binds the polycomb group 2 proteins thatare responsible for trimethylation of H3K27, we askedwhether there would be a differential ability of full-length versus repeat A alone to recruit H3K27me3.However, we did not observe any H3K27me3 enrich-ment by ChIP at the EGFP, Hyg or the CLDN16 pro-moters (Additional file 1: Figure S1). H3K27me3 is amark of the inactive X, and has been shown to beenriched at the promoters of inactivated genes [23];however, given that the silencing we have observed inthis system is reversible ([13] and data not shown), it isperhaps not surprising that this heritable mark of silentchromatin is not recruited. Similarly, we had previouslyshown that there was no recruitment of DNA methyla-tion in this reversible system [13]. A similar inducibletransgene in mice had identified a developmental windowduring which inactivation could occur [24], yet we observeinduction of silencing in our somatic cell model; possiblyreflecting a more epigenetically dynamic state to thesecancer-derived cells, or differences in the genes being ex-amined, as we observed variability between genes in theirability to be silenced. By recapitulating XIST-induced genesilencing, but not requiring sequences involved in thespread of XIST, the A repeat construct exposes the mostbasal aspects of XIST silencing function. To identify theminimal functional unit for silencing, we further dissectedthe repeat A sequences.Repeat A monomers contribute additively to silencingIn order to further characterize the link between repeatA sequence and its silencing ability, we generated anMinks et al. Epigenetics & Chromatin 2013, 6:23 Page 4 of 10http://www.epigeneticsandchromatin.com/content/6/1/23artificial repeat A construct that tested the potential im-pact of sequence variations in the individual monomers,which are particularly prevalent in the T-rich linker re-gions. This artificial repeat A consisted of a nine-foldrepetition of a 46 bp consensus monomer sequence, andcontained restriction enzyme sites in the T-rich stretchesto allow for the creation of constructs with reducednumbers of repeats (Figure 2A). Flow cytometry and q-PCR showed that the artificial repeat A silenced EGFPto the same extent as full-length XIST or human repeatA constructs. Since variability within the individual re-peats and spacer regions did not contribute to silencing,we were then able to test the silencing ability of con-structs with fewer repeats. Transgenes harboring two tosix repeat A monomers were functional, with an ap-proximately linear relationship between the number ofrepeats and their silencing ability (Figure 2B). Silencinginduced by the repeat A 2-mer gradually increased be-tween day 2 and approximately day 8; however, longerinduction of the repeat A 2-mer did not promote furtherEGFP down-regulation (Figure 2C).These observations provide strong evidence that thesilencing of an adjacent EGFP reporter is achievedthrough an additive effect of repeat A monomers, witheven a 2-mer repeat A inducing partial EGFP silencing.The number of repeat A units was previously reportedto correlate with the ability of Xist to induce silencing indifferentiating mouse ES cells [14]. Also, in agreementwith a previous report on mouse Xist [14], artificial re-peat A retains full silencing potential when compared tohuman repeat A, suggesting that neither sequence vari-ation within the CG-rich core nor the varying length ofthe T-rich spacers in individual repeat A monomers isessential for XIST function. The remarkable ability of aconstruct with only two repeats to silence EGFP in a re-producible and statistically significant fashion providedus with a well-defined template for further dissection ofthe relationship between repeat A sequence and its si-lencing ability.The core repeat A sequence consists of two palin-dromes; the first potentially allowing for perfect C-Gpairing linked by ‘ATCG’ and the second involving C-Gpairing as well as a G-U pair linked by ‘ATAC’ with theT-rich stretches serving as spacers [14]. While alterna-tive structures have since been proposed, for simplicity,we refer to the four components of CG-rich consensuscore as stem 1 (S1), loop 1 (L1), stem 2 (S2) and loop 2(L2). We initially created a variant of the 2-mer repeat Ain each of these elements in order to probe their role incis-silencing of EGFP (Figure 3A). Mutations of L1, S2and L2 completely ablated the transgenes’ ability to si-lence EGFP, as measured by flow cytometry of two rep-resentative clones for each mutation, compared to acanonical repeat A 2-mer (Figure 3B). Analysis by qRT-PCR showed the same trends and allowed examinationof the Hyg gene (Figure 3C); however, flow cytometry af-fords considerably greater sensitivity as 30,000 eventswere combined into each datapoint. Mutation of S1resulted in partial abolition of EGFP silencing. Thus, themost conserved regions of XIST both among the individ-ual repeats in human (Figure 2A) and among differentspecies (Additional file 2: Figure S2), the CG-rich palin-dromes and their intervening ‘ATCG’ and ‘ATAC’ se-quences, are critical for XIST function. All of the0. 5'A9-mer6-mer5-mer4-mer3-mer2-merEGFPexpression[relative to no DOX]Cttttctatattttgcccatcggggctgcggatacctggttttattattttttctttgcccaacggggccgtggatacctgccttttaattcttttttattcgcccatcggggccgcggatacctgctttttatttttttttccttagcccatcggggtatcggatacctgctgattcccttcccctctgaacccccaacactctggcccatcggggtgacggatatctgctttttaaaaattttctttttttggcccatcggggcttcggatacctgctttttttttttttattttccttgcccatcggggcctcggatacctgctttaatttttgtttttctgcccatcggggccgcggatacctgctttgatttttttttttcatcgcccatcggtgctttttatggatgaaaaa[ttttattttctttgcccatcggggccgcggatacctgcttttataa]A9xHinDIII BamHIXhoIEcoRV NotIAflII0. d4 d6 d8 d10 d12 d14 d16EGFPexpression [relative to no DOX]Days of repeat A 2-mer inductionBFigure 2 Repeat A monomers additively contribute tosilencing. (A) Human repeat A sequence consists of 8.5 copies of awell-conserved CG-rich core and T-rich spacer sequences.Palindromic sequences hypothesized to form a secondary structureare underlined. Artificial repeat A was constructed as a 9-merrepetition of consensus monomer sequence and restriction enzymesites were introduced to allow for the creation of shorter constructs.(B) Enhanced Green Fluorescent Protein gene (EGFP) expressionfollowing five days of transgene induction as measured by qRT-PCR,relative to d0 and normalized to changes in expression caused byinduction of the vector alone and to ACTB expression for twobiological replicates. (C) EGFP expression was measured by flowcytometry every 2 days for 16 days following induction of repeat A2-mer. Data are normalized to EGFP expression in cells that were notinduced with DOX.Minks et al. Epigenetics & Chromatin 2013, 6:23 Page 5 of 10http://www.epigeneticsandchromatin.com/content/6/1/23previously proposed structures predict the existence ofan ‘ATCG’ loop and mutation to ‘TTTT’ in our systemcompletely abolished human repeat A function. Simi-larly, mutation to ‘TAGC’ in mice has been shown topartially abolish Xist function [14], suggesting that thesequence of the tetraloop, and not just its presence, iscritical for XIST/Xist function.The palindromic nature of the repeat A core se-quences strongly suggests their involvement in forminga distinct secondary RNA structure. Several alternativebut mutually exclusive structures were previously pro-posed in which the CG-rich palindrome encompassingthe ‘ATCG’ tetraloop (‘stem 1’) may either form a hairpinwith pairing within each repeat A monomer [14,21,22]or pairing between two separate monomers [16]. Theability of the repeat A 2-mer to reproducibly inducegene silencing allowed us to use mfold, an RNA struc-ture prediction algorithm [25], to design repeat A mu-tants that would compare the silencing effectivenesswhen inter- or intra-repeat pairing was enforced. Wefound that modeling mutated constructs of larger than2-mer repeat A structures was highly unreliable as mul-tiple structures of similar minimum free energies (dG)were predicted. We designed a quartet of mutations inthe 2-mer repeat A that were predicted to enforcepairing either within (A1, A2) or between (B1, B2) eachmonomer (Figure 4A and Additional file 3: Figure S3).The repeat A 2-mer mutations were constructed so thata single prominent structure was predicted to fold witheither higher (A1, B1), or lower (A2, B2) dG, comparedto the unmodified repeat A 2-mer.Measured by flow cytometry, the mutants predicted toenforce pairing within each monomer functioned betterthan those enforcing the interaction between the mono-mers; although none of the four mutants silenced EGFPas efficiently as the canonical repeat A 2-mer (Figure 4B),suggesting that there may be more complex structuresinvolved. While the differences in EGFP expression wererelatively subtle due to a limited silencing effect of therepeat A 2-mer, they were highly statistically significantand equivalent results were obtained for two single-cellclones of independent integrations and a total of sevenbiological replicates. More representative structurescontaining greater than two repeat units were not testedas they could not be predicted to reliably form only asingle thermodynamically favored structure. However,given the number of eutherian genome sequences thathave now been assembled, we turned instead to acharacterization of the full repeat A sequences that areavailable in genome databases.Survey of repeat A mutations shows strong preferencefor stem 1 and mild preference for stem 2 formationTaking advantage of the increasing number of sequencedmammalian genomes, we generated an alignment of re-peat A sequences from 27 mammalian species (Additionalfile 2: Figure S2A). Repeat A consists of 24 bp-long CG-rich core sequences separated by approximately 20 to 50bp-long T-rich spacers. The CG-rich core is formed bytwo palindromes, each of which is broken by four bp-longsequences. As expected, repeat A was well conserved, inparticular within the CG-rich core sequences (Additionalfile 2: Figure S2B). Interestingly, 22/27 mammalian XISTsequences contained either eight or nine monomers of re-peat A, and at least one of the remaining five was incom-plete across the region, supporting the need for eightmonomers to achieve full XIST functionality.Of the defined stem-loop structures, loop 1 showedthe highest frequency of deviation from the canonical‘ATCG’ sequence, with approximately 10% (20/202) ofB0. 5'A2-mer2-mer  S12-mer  S22-mer  L12-mer  L2Reporter expression DOX d5/d0EGFP Hyg0. 5'A2-mer2-mer  S12-mer  S22-mer  L12-mer  L2EGFPexpression DOX d5/d02-mer:  ttttattttctttgcccatcggggccgcggatacctgcttttataa2-mer L1:  -----------------TTTT-------------------------2-mer S1:  ---------------G------A-----------------------2-mer L2:  ------------------------------TTTT------------2-mer S2:  -----------------------------C------A---------2xAp = 0.0003n. s.p = 0.02CFigure 3 Mutation of the core repeat A sequences abrogatesits silencing ability. (A) Sequence of the canonical repeat Amonomer and four mutant constructs that were created to targetthe hypothesized repeat A hairpins. Underlined sequencescorrespond to stem 1 and stem 2. Dashes indicate no sequencechange. (B) Mean Enhanced Green Fluorescent Protein gene (EGFP)expression following five days of transgene induction, measured byflow cytometry, relative to d0 (two-tailed paired t-test). Error barsindicate ± 1 S.D. of two single-cell clones. (C) EGFP and hygromycin(Hyg) gene expression following five days of transgene induction,measured by qRT-PCR, relative to d0 and normalized to ACTBexpression for two independent single-cell clones.Minks et al. Epigenetics & Chromatin 2013, 6:23 Page 6 of 10http://www.epigeneticsandchromatin.com/content/6/1/23repeat A units harboring an ‘AACG’ tetraloop instead(Additional file 2: Figure S2). To ask whether there wasan evolutionary preference for reciprocal mutations thatsupported the formation of an intra- or interloop config-uration we examined deviations from the canonical stemsequences across the bona fide monomers of the 27mammals (Figure 5 and Additional file 4: Figure S4).Despite the strong conservation there were 50 stem 1changes, allowing us to determine whether fully comple-mentary double-stranded sequences could form due toexisting reciprocal mutations either within the sameunit, or in another unit of the same species. Of the 50stem 1 mutations we analyzed, 24 could not be linkedwith a reciprocal mutation; 12 of the remaining 26 mu-tations were accompanied by a reciprocal mutation ex-clusively within the same unit; and a further 10 couldpair either within the same unit, or with another unit(Figure 5A). These findings strongly argue in favor ofthe predicted stem-loop 1 formation. Survey of stem 2mutations uncovered 46 deviating repeat A units, 28 ofwhich could not pair with any reciprocal mutation(Figure 5B). Of the remaining 18 mutations, 8 could ex-clusively form a stem-loop by pairing within each unit,with a further 3 allowing for pairing either within a unit orwith other units (Figure 5B). While the propensity of thestem 2 region to harbor reciprocal mutations retainingA1...5....10...15...20...25...30...35...40...45...50...55...60...65...70...75...80...85...90...952-mer   : ttttattttcttt.gcccatcggggc.cgcggatacctgcttttataattttattttcttt.gcccatcggggc.cgcggatacctgcttttataa2-mer B1: ------------A-C---------C--------------------------------------G---------G----------------------2-mer B2: -------------G-G-G--------C----------------------------------G--------C-C-C---------------------  A12-mer  A22-mer  B12-mer  B2EGFPexpressionDOX d5/d0p < 2*10-7p < 10-81...5....10...15...20...25...30...35...40...45...50...55...60...65...70...75...80...85...90...952-mer   : ttttattttcttt.gcccatcggggc.cgcggatacctgcttttataattttattttcttt.gcccatcggggc.cgcggatacctgcttttataa2-mer A1: ------------A-C----------G--------------------------------------G------C------------------------2-mer A2: -------------G-G-G----C-C-C----------------------------------G------------C---------------------BFigure 4 Stem-loop 1 structure is required to maintain repeat A silencing ability. Silencing ability of 2-mer repeat A construct is retainedwhen forced to form stem-loop 1 structure, but abrogated when pairing between the monomers is enforced. (A) Sequence of the canonicalrepeat A 2-mer and four mutant constructs that either enforce formation of stem-loop 1 (A1, A2) or an alternative folding (B1, B2) of repeat Asequences, as indicated by schematics. Dashes indicate no change in sequence. (B) Mean Enhanced Green Fluorescent Protein gene (EGFP) expressionfollowing five days of transgene induction, measured by flow cytometry, relative to d0 (two-tailed paired t-test). Error bars indicate ± 1 S.D. of twoindependent single-cell clones and a total of seven biological replicates.ATotal mutations in stem 15020How many possible pairsby reciprocal mutations?112 3Withinvs. between1616Total bases in stem 1 probed15 1110 1+Within + between vs. both between24+BTotal mutations in stem 246≥ 20How many possible pairsby reciprocal mutations?18 4Withinvs. between1083Total bases in stem 2 probed12 63 3+Within + between vs. both between28+Figure 5 Compensatory changes in putative stems of repeat A hairpin sequences of 27 mammals. (A) All bona fide repeat A coresequences that deviated in sequence from the canonical stem 1 sequence were categorized by their potential to form a base pair with areciprocally-mutated base within the same repeat A unit or within another unit. (B) As in (A), but stem 2 is analyzed.Minks et al. Epigenetics & Chromatin 2013, 6:23 Page 7 of 10http://www.epigeneticsandchromatin.com/content/6/1/23stem-loop 2 formation is less striking than for stem 1, it isstill remarkably high, arguing either that stem 2 indeedforms a stem-loop by pairing within each unit, or that re-peat A structure involves a combination of both modes ofpairing.Several secondary structures of repeat A have beenproposed based on analysis of repeat A mutants [14],NMR data, [21,22] and RNase footprinting and fluores-cence resonance energy transfer data [16]. The first pal-indrome was suggested to form either a hairpin bypairing within each monomer [14,21,22] or, alternatively,to pair between monomers [16]. Both our targeted mu-tations in an artificial 2-mer repeat A construct, as wellas our assessment of evolutionary sequence conserva-tion, support the intra-repeat pairing model, consistentwith outcomes observed in mice [14], that the first pal-indrome indeed forms a stem to expose the ‘ATCG’tetraloop. The mutations we introduced to the secondpalindrome also resulted in a complete loss of silencingby XIST (Figure 3), supporting the importance of thesesequences; however, these mutations did not directly ad-dress secondary structure formation. While the secondpalindrome was proposed to pair within each monomerto form the second stem-loop [14], recent studies sug-gest that the secondary structure may rather involvepairing between individual repeat A monomers [21,22]or with the T-rich spacers [16]. Our assessment of evolu-tionary sequence conservation provides evidence in favorof second stem-loop formation, though the frequency ofcompensatory mutations is less striking than observedfor stem-loop 1.ConclusionsWe utilized a single-copy FRT integration site to gener-ate DOX-inducible XISTcDNA integrations allowing thedelineation of the repeat A monomers as the minimalfunctional unit that additively contributes to gene silen-cing. The ability of only two copies of repeat A to repro-ducibly silence a flanking EGFP reporter gene allowedfor further dissection of repeat A sequence to elucidatethe relationship between repeat A structure and func-tion. Disruption of either the putative stems or loops ofthe repeat A abrogated silencing, and mutations of thefirst palindrome to enforce pairing within a repeat, orbetween the first and second repeat, supported modelsthat the first palindrome forms a hairpin. An evolution-ary analysis of sequence changes within the palindromesallowed assessment of intra- versus inter-repeat pairingin full-length XIST sequences. Again, the model of intra-repeat pairing was favored. The intricate set of eventsthat ultimately lead to X-chromosome inactivation infemale mammals remains a vanguard to mammalianepigenetic research. By focusing only on the ability to si-lence a proximal reporter we have reduced thecomplexity of deciphering the critical roles of XIST. Wedemonstrate that a mere 94 bp-long sequence of repeatA can silence flanking reporter genes, but not more dis-tal endogenous genes that are silenced by induction ofthe full-length XIST RNA. Further data on the relation-ship of repeat A sequence and function will provide afoundation for elucidating the yet unclear connectionbetween the sequence of long non-coding RNAs, likeXIST/Xist, and their ability to silence chromatin.MethodsConstruct generationThe artificial repeat A construct and its shorter derivativesand mutants were synthesized by GeneArt (now Life Tech-nologies Inc, Burlington, ON, Canada) and cloned into thepcDNA5/FRT/TO plasmid (Life Technologies Inc.) usingstandard techniques. HT1080 cells were transfected as de-scribed previously [13]. Mfold server version 2.3 was usedto predict secondary RNA structures (http://mfold.rna.al-bany.edu) of new constructs.Cell cultureClones harboring single-copy FLP-mediated integrations ofXIST constructs into HT1080 fibrosarcoma cell lines weregenerated and cultured as described previously [13]. TheXIST transgenes were induced by doxycycline (1 μg/mL)and cell culture medium was changed every 24 hours.Identification of the transgene integration siteInverse PCR utilizing primers complementary to a se-quence within the integrated pEGFP-N1 plasmid (LifeTechnologies) was used to identify the precise integra-tion site of the XIST - EGFP transgene in the HT1080 2-3-0.5 + 3#4 cell line. PCR primer sequences are listed inAdditional file 5: Supplementary methods.qRT-PCRRNA was isolated from frozen cell pellets by TRIZOL(Life Technologies Inc.) and treated with DNase I(Roche Diagnostics, Laval, QC Canada) according to themanufacturers’ recommendations. Following phenol-chloroform extraction, RNA concentration was assessedby spectrophotometry and 0.5 to 2.5 μg of RNA wasreverse-transcribed by M-MLV reverse transcriptase(Invitrogen). Fermentas HS Taq (Thermo Scientific,Waltham, MA USA) and EvaGreen (Biotium Inc., Hay-ward CA USA) were used in quantitative PCR under thefollowing conditions: 5 minutes 95°C, 40x (15 sec. 95°C,30 sec. 60°C, 60 sec. 72°C). PCR primer sequences arelisted in Additional file 5: Supplementary methods.Flow cytometryHT1080 cell pellets were washed with PBS andresuspended in 0.5 mL of PBS supplemented with 10%Minks et al. Epigenetics & Chromatin 2013, 6:23 Page 8 of 10http://www.epigeneticsandchromatin.com/content/6/1/23FCS. A total of 30,000 events were recorded using anLSRII flow cytometer (BD Biosciences, Mississauga, ON,Canada). Mean fluorescence intensity of EGFP wasassessed by using a combination of 488 nm laser excita-tion and a 530/30 nm bandpass filter.Allelic discrimination by pyrosequencingA total of 2 μL of cDNA was added to a standard 25-μLpyrosequencing reaction containing 1 × PCR buffer(QIAGEN, Valencia, CA, USA), 0.2 mM dNTPs, 0.625unit Hot Start Taq DNA polymerase (QIAGEN), 0.25μM forward primer and 0.25 μM reverse primer. PCRconditions were: 95o for 15 minutes, 35 cycles of 94° for30 sec, 56.3° or 58.3° for 30 sec (see Supplementarytable), 72° for 30 sec, and finally 72° for 10 minutes.Template preparation for pyrosequencing was doneaccording to the manufacturer’s protocol, using 10 to 15μl of PCR products.Analysis of repeat A core in mammalsRepeat A sequences in a panel of mammalian species wereidentified using a combination of BLAST, BLATand in silicoPCR searches of mammalian genomes available throughNCBI (http://blast.ncbi.nlm.nih.gov) and ENSEMBL (http://www.ensembl.org/Multi/blastview) databases and the UCSCgenome browser (http://genome.ucsc.edu). A table listingthe accession numbers or genomic locations of repeat A se-quences is provided in Additional file 5: Supplementarymethods. Sequences were aligned using clustalw2 (http://www.ebi.ac.uk/Tools/msa/clustalw2) and screened to ex-clude all non-bona fide repeat A CG-rich core sequencesfrom further analyses. CG-rich core sequences thatcontained bases deviating from the canonical sequence ofeither stem 1 or stem 2 were identified. Finally, we testedwhether such a mutation was reciprocated by a mutationwithin the same repeat A unit, or in all other repeats of thatspecies.Additional filesAdditional file 1: Figure S1. ChIP for H3K27me3 at silenced promoters.ChIP for H3K27me3 at the EGFP, CLDN16 (2 locations, P1 and P2) and Hyg(2 locations, P1 and P2) promoters that are shown to be silenced byDOX-induced expression of XIST. H3 shows pull-down for all promoters,while IgG shows limited pull-down. MYT1, a silenced gene, is a positivecontrol for H3K27me3 recruitment, and the active APRT gene is used as anegative control.Additional file 2: Figure S2. Alignment of repeat A sequences in 27mammals. (A) Sequence alignment of repeat A region in 27 mammalianspecies. Black circles mark sequences that were not considered bona fiderepeat A units and were thus excluded from further analyses. (B) Sequenceconservation of 202 core repeat A units among 27 mammalian species.Lines on the X axis depict (from top to bottom) the position of bases,percent of units that deviate from canonical sequence, the canonicalsequence and arrows corresponding to bases forming the hypothesizedstem 1 and stem 2.Additional file 3: Figure S3. In silico prediction of repeat A mutantstructure. Structures and free energies of 2-mer repeat A and mutantscreated to enforce pairing within each monomer (A1, A2) or betweenthe two monomers (B1, B2) predicted by mfold. Bases diverging from thecanonical repeat A sequence are capitalized and highlighted.Additional file 4: Figure S4. Analysis of repeat A sequences in 27mammals. (A) Illustration of the approach used to analyze repeat Asequence alignment data in Figure 5A, B and Additional file 3: Figure S3B,C. (B) Analysis of reciprocal mutations in the stem 1 of individual repeatA units. The table depicts the number of occurrences when mutation ina repeat A unit would allow pairing due to the existence of a reciprocalmutation within the same unit (highlighted), in a different unit or whenno reciprocal mutation exists in the species’ repeat A (listed in the lastcolumn). (C) As in B), but stem 2 is analyzed.Additional file 5: Supplementary methods. List of accession numbersor sequence coordinates of repeat A sequences used in sequenceanalyses and table of primer sequences and ChIP methods.AbbreviationsChIP: Chromatin immunoprecipitation; DOX: Doxycycline; EGFP: EnhancedGreen Fluorescent Protein gene; ES: Embryonic stem; FCS: Fetal calf serum;Hyg: Hygromycin gene; ncRNA: Non-coding RNA; PBS: Phosphate-bufferedsaline.Competing interestsThe authors declare no financial or non-financial competing interests.Authors’ contributionsJM designed the study, carried out the cloning and molecular geneticstudies, and drafted the manuscript. SELB generated the cell lines describedand reviewed the manuscript. CY and AMC contributed to the moleculargenetic studies and data analysis, and revised the manuscript. CJBparticipated in the study design and co-ordination, data interpretation andmanuscript preparation. All authors read and approved the final manuscript.AcknowledgementsThis work was funded by a CIHR operating grant (MOP-13690) to CJB.Received: 6 March 2013 Accepted: 18 June 2013Published: 1 August 2013References1. Lyon MF: Gene action in the X-chromosome of the mouse (Mus musculusL.). Nature 1961, 190:372–373.2. Brown CJ, Ballabio A, Rupert JL, Lafreniere RG, Grompe M, Tonlorenzi R,Willard HF: A gene from the region of the human X inactivation centre isexpressed exclusively from the inactive X chromosome. Nature 1991,349:38–44.3. Wutz A: Gene silencing in X-chromosome inactivation: advances inunderstanding facultative heterochromatin formation. Nat Rev Genet2011, 12:542–553.4. Brown CJ, Hendrich BD, Rupert JL, Lafreniere RG, Xing Y, Lawrence J, WillardHF: The human XIST gene: analysis of a 17 kb inactive X-specific RNAthat contains conserved repeats and is highly localized within thenucleus. Cell 1992, 71:527–542.5. Brockdorff N, Ashworth A, Kay GF, McCabe VM, Norris DP, Cooper PJ, SwiftS, Rastan S: The product of the mouse Xist gene is a 15 kb inactive X-specific transcript containing no conserved ORF and located in thenucleus. Cell 1992, 71:515–526.6. Duret L, Chureau C, Samain S, Weissenbach J, Avner P: The Xist RNA geneevolved in eutherians by pseudogenization of a protein-coding gene.Science 2006, 312:1653–1655.7. Elisaphenko EA, Kolesnikov NN, Shevchenko AI, Rogozin IB, Nesterova TB,Brockdorff N, Zakian SM: A dual origin of the Xist gene from a protein-coding gene and a set of transposable elements. PLoS One 2008, 3:e2521.8. Grant J, Mahadevaiah SK, Khil P, Sangrithi MN, Royo H, Duckworth J,McCarrey JR, VandeBerg JL, Renfree MB, Taylor W, Elgar G, Camerini-OteroRD, Gilchrist MJ, Turner JM: Rsx is a metatherian RNA with Xist-likeproperties in X-chromosome inactivation. Nature 2012, 487:254–258.Minks et al. Epigenetics & Chromatin 2013, 6:23 Page 9 of 10http://www.epigeneticsandchromatin.com/content/6/1/239. Hoffman LM, Hall L, Batten JL, Young H, Pardasani D, Baetge EE, Lawrence J,Carpenter MK: X-inactivation status varies in human embryonic stem celllines. Stem Cells 2005, 23:1468–1478.10. van den Berg IM, Laven JS, Stevens M, Jonkers I, Galjaard RJ, Gribnau J, vanDoorninck JH: X chromosome inactivation is initiated in humanpreimplantation embryos. Am J Hum Genet 2009, 84:771–779.11. Okamoto I, Patrat C, Thepot D, Peynot N, Fauque P, Daniel N, DiabangouayaP, Wolf JP, Renard JP, Duranthon V, Heard E: Eutherian mammals usediverse strategies to initiate X-chromosome inactivation duringdevelopment. Nature 2011, 472:370–374.12. Yang C, Chapman AG, Kelsey AD, Minks J, Cotton AM, Brown CJ: X-chromosome inactivation: molecular mechanisms from the humanperspective. Hum Genet 2011, 130:175–185.13. Chow JC, Hall LL, Baldry SE, Thorogood NP, Lawrence JB, Brown CJ:Inducible XIST-dependent X-chromosome inactivation in human somaticcells is reversible. Proc Natl Acad Sci U S A 2007, 104:10104–10109.14. Wutz A, Rasmussen TP, Jaenisch R: Chromosomal silencing andlocalization are mediated by different domains of Xist RNA. Nat Genet2002, 30:167–174.15. Kanhere A, Viiri K, Araújo CC, Rasaiyaah J, Bouwman RD, Whyte WA, PereiraCF, Brookes E, Walker K, Bell GW, Pombo A, Fisher AG, Young RA, Jenner RG:Short RNAs are transcribed from repressed polycomb target genes andinteract with polycomb repressive complex-2. Mol Cell 2010, 38:675–688.16. Maenner S, Blaud M, Fouillen L, Savoye A, Marchand V, Dubois A, Sanglier-Cianférani S, Van Dorsselaer A, Clerc P, Avner P, Visvikis A, Branlant C: 2-Dstructure of the A region of Xist RNA and its implication for PRC2association. PLoS Biol 2010, 8:e1000276.17. Zhao J, Sun BK, Erwin JA, Song JJ, Lee JT: Polycomb proteins targeted by ashort repeat RNA to the mouse X chromosome. Science 2008,322:750–756.18. Hoki Y, Kimura N, Kanbayashi M, Amakawa Y, Ohhata T, Sasaki H, Sado T:A proximal conserved repeat in the Xist gene is essential as a genomicelement for X-inactivation in mouse. Development 2009, 136:139–146.19. Jeon Y, Lee JT: YY1 tethers Xist RNA to the inactive X nucleation center.Cell 2011, 146:119–133.20. Royce-Tolland ME, Andersen AA, Koyfman HR, Talbot DJ, Wutz A, Tonks ID,Kay GF, Panning B: The A-repeat links ASF/SF2-dependent Xist RNAprocessing with random choice during X inactivation. Nat Struct Mol Biol2010, 17:948–954.21. Duszczyk MM, Zanier K, Sattler M: A NMR strategy to unambiguouslydistinguish nucleic acid hairpin and duplex conformations applied to aXist RNA A-repeat. Nucleic Acids Res 2008, 36:7068–7077.22. Duszczyk MM, Wutz A, Rybin V, Sattler M: The Xist RNA A-repeat comprisesa novel AUCG tetraloop fold and a platform for multimerization. RNA2011, 17:1973–1982.23. Calabrese JM, Sun W, Song L, Mugford JW, Williams L, Yee D, Starmer J,Mieczkowski P, Crawford GE, Magnuson T: Site-specific silencing ofregulatory elements as a mechanism of X inactivation. Cell 2012,151:951–963.24. Wutz A, Jaenisch R: A shift from reversible to irreversible X inactivation istriggered during ES cell differentiation. Mol Cell 2000, 5:695–705.25. Zuker M: Mfold web server for nucleic acid folding and hybridizationprediction. Nucleic Acids Res 2003, 31:3406–3415.doi:10.1186/1756-8935-6-23Cite this article as: Minks et al.: XIST-induced silencing of flanking genesis achieved by additive action of repeat a monomers in human somaticcells. Epigenetics & Chromatin 2013 6:23.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitMinks et al. Epigenetics & Chromatin 2013, 6:23 Page 10 of 10http://www.epigeneticsandchromatin.com/content/6/1/23


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items