3734 Research Article

IntroductionLarge-scale genomic studies have estimated that ~50% of themammalian genome is transcribed into RNA on one or both strands,whereas all of the predicted protein-coding part of transcriptstogether constitute only ~2–5% of the genome (Cheng et al., 2005;Mattick and Makunin, 2006; van Bakel et al., 2010). This suggeststhat a large proportion of genome-encoded transcripts consist ofnon-protein coding RNAs (ncRNAs) (Mattick, 2009; Mattick andMakunin, 2006; Prasanth and Spector, 2007). Such ncRNAs includehousekeeping ncRNAs (examples include ribosomal RNA, transferRNA, small nuclear and nucleolar RNAs) as well as regulatoryncRNAs, whose expression is modulated in a cell-specific ordevelopmental stage-specific manner (Prasanth and Spector, 2007).

Regulatory ncRNAs are generally categorized into small [18–200 nucleotides (nts)] and long (200 nts to 100 kb) ncRNAs(Prasanth and Spector, 2007). MicroRNAs, an example of smallncRNAs, serve regulatory roles in several crucial biologicalprocesses such as cellular proliferation, differentiation anddevelopmental timing (Mattick, 2005). Genome-wide transcriptomeanalyses have now shown that a major fraction of the regulatoryncRNAs in mammals consists of long ncRNAs (lncRNAs) (Dingeret al., 2009; Mercer et al., 2009; Wilusz et al., 2009). LncRNAswere initially thought to be an artifact of chromatin remodeling or‘transcriptional noise’ in the cell (Struhl, 2007), but recent datahave indicated that a large number of mammalian lncRNAs arehighly conserved and are involved in diverse cellular functions(Dinger et al., 2009; Guttman et al., 2009; Khalil et al., 2009).Some of the well-characterized mechanisms involving lncRNAsinclude dosage compensation and genome imprinting, programmed

whole-genome rearrangements in ciliate differentiation, humanbrain evolution, T-cell receptor recombination, maintenance oftelomeres, stress-response and cellular differentiation (Amaral etal., 2008; Wilusz et al., 2009).

Nuclear-retained RNAs (nrRNAs) constitute a subclass ofeukaryotic regulatory RNAs that are predominantly enriched in thecell nucleus and are suggested to play structural roles or act asriboregulators (Pederson, 2009; Prasanth and Spector, 2007). Well-characterized examples of mammalian nrRNAs include Xist andTsix that are involved in dosage compensation (Brockdorff et al.,1992; Lee et al., 1999), Air and Kcnq1ot1 RNAs involved ingenome imprinting (Braidotti et al., 2004; Pandey et al., 2008) andMEN e/b (Neat1) involved in the structural organization ofparaspeckles (Chen and Carmichael, 2009; Clemson et al., 2009;Sasaki et al., 2009; Sunwoo et al., 2009). In Schizosaccharomycespombe, repeat-containing nrRNAs are involved in heterochromatinorganization (Grewal and Elgin, 2007). Recent studies inmammalian cells also indicated crucial roles for repeat-containingnrRNAs in chromatin-specific functions including heterochromatinorganization, neocentromere formation and X-chromosomeinactivation (Chow et al., 2010; Chueh et al., 2009; Maison et al.,2002; Wong et al., 2007). Interestingly, a significant number ofnovel nrRNAs were recently identified in mammalian cells thatshow tissue- or cell-type specific expression (Mercer et al., 2008;Sone et al., 2007). Understanding how novel nrRNAs regulategene expression will provide insights into how a higher degree ofcellular complexity is achieved in eukaryotes.

In the present study, we have identified a novel species ofnrRNAs in mammalian cells: GAA-repeat-containing RNAs (GRC-

Polypurine-repeat-containing RNAs: a novel class oflong non-coding RNA in mammalian cellsRuiping Zheng1, Zhen Shen1, Vidisha Tripathi1, Zhenyu Xuan2, Susan M. Freier3, C. Frank Bennett3, Supriya G. Prasanth1 and Kannanganattu V. Prasanth1,*1Department of Cell and Developmental Biology, University of Illinois at Urbana-Champaign, Chemical and Life Sciences Laboratory,601 South Goodwin Avenue, Urbana, IL 61801, USA2Department of Molecular and Cell Biology, University of Texas at Dallas, 800 West Campbell Road, Richardson, TX 75080, USA3ISIS Pharmaceuticals, 1896 Rutherford Road, Carlsbad, CA 92008, USA*Author for correspondence (kumarp@life.illinois.edu)

Accepted 18 July 2010Journal of Cell Science 123, 3734-3744 © 2010. Published by The Company of Biologists Ltddoi:10.1242/jcs.070466

SummaryIn higher eukaryotic cells, long non-protein-coding RNAs (lncRNAs) have been implicated in a wide array of cellular functions. Cell-or tissue-specific expression of lncRNA genes encoded in the mammalian genome is thought to contribute to the complex genenetworks needed to regulate cellular function. Here, we have identified a novel species of polypurine triplet repeat-rich lncRNAs,designated as GAA repeat-containing RNAs (GRC-RNAs), that localize to numerous punctate foci in the mammalian interphase nuclei.GRC-RNAs consist of a heterogeneous population of RNAs, ranging in size from ~1.5 kb to ~4 kb and localize to subnuclear domains,several of which associate with GAA.TTC-repeat-containing genomic regions. GRC-RNAs are components of the nuclear matrix andinteract with various nuclear matrix-associated proteins. In mitotic cells, GRC-RNAs form distinct cytoplasmic foci and, in telophaseand G1 cells, localize to the midbody, a structure involved in accurate cell division. Differentiation of tissue culture cells leads to adecrease in the number of GRC-RNA nuclear foci, albeit with an increase in size as compared with proliferating cells. Conversely, thenumber of GRC-RNA foci increases during cellular transformation. We propose that nuclear GRC-RNAs represent a novel family ofmammalian lncRNAs that might play crucial roles in the cell nucleus.

Key words: Non-coding RNAs, Nuclear domains, Nuclear matrix, Triplex repeats, Midbody

Jour

nal o

f Cel

l Sci

ence

RNAs) that are distributed throughout the nucleus in amicropunctate pattern. GRC-RNAs primarily consist of polypurinerepeats and are heterogeneous in size, ranging from 1.5–4 kb.GRC-RNAs associate with the nuclear matrix and interact withseveral bona fide nuclear matrix proteins. Our studies furtherindicate that the nuclear GRC-RNA foci is altered in cells thateither are highly proliferating or are cell-cycle-arrested owing toserum deprivation. We propose that GRC-RNAs are novel membersof the nrRNA family of RNAs that interact with components of thenuclear matrix and play crucial structural roles in the maintenanceof the nuclear matrix and/or regulate gene expression.

ResultsGRC-RNAs are members of the nrRNA family of lncRNAsRNA in situ hybridization studies have previously cataloged ~850lncRNAs that are expressed in specific cell types or in specificsubcellular compartments within the adult mouse brain (Lein et al.,2007; Mercer et al., 2008). To gain further insight into the cellularfunction of these novel ncRNAs, we analyzed their distribution invarious mammalian cell lines by RNA-FISH (fluorescent in situhybridization). One of these ncRNAs (EST clone AK082015; ~2.2kb) showed a punctate distribution in the nucleus of mouse NIH-3T3 cells (Fig. 1A). In most of the cells, apart from numerousnuclear foci, 2–5 distinct brightly labeled GRC-RNA-positive fociwere also observed in the cytoplasm (see arrows in Fig. 1A;supplementary material Fig. S1B,C). To identify the minimal regionof the cDNA probe that hybridized to the nuclear foci-associated

3735GRC-RNAs and nuclear organization

RNA, we subcloned AK082015 cDNA into multiple smaller regionsand used them individually as probes in RNA-FISH analysis(supplementary material Fig. S1). RNA-FISH using probescontaining the fragment-2 region (170 bp) displayed similarmicropunctate nuclear staining, whereas the flanking regions ofthe cDNA failed to show any specific staining (supplementarymaterial Fig. S1B,C). These results indicate that the AK082015ncRNA by itself does not localize to these nuclear foci but thefragment-2 in AK082015 cDNA hybridizes to a class of RNA(s).Sequencing of the fragment-2 probe revealed that it contains acharacteristic GAA.TTC tandem triplet repeat sequence (20continuous GAA.TTC repeats; supplementary material Fig. S1A).A recent study using a fluorescently labeled GAA.TTC probe or(GAA)20 oligonucleotide probe and performing FISH in non-denatured human cells demonstrated the presence of similarsubnuclear foci (Ohno et al., 2002). Based on their results, theauthors claimed that the GAA-repeat probe hybridized to the TTC-repeat DNA single strand that was displaced owing to the formationof GAA-GAA-TTC triple-helix DNA structure (Ohno et al., 2002).Our studies using RNA-FISH in mouse NIH-3T3 cells usingfluorescent-labeled single-stranded DNA oligonucleotide probesconsisting of (GAA)15, (TTC)15, (CCT)15 or (CT)20 revealed thatonly the (TTC)15 probe displayed the micropunctate nuclear signal(Fig. 1B), indicating the presence of GAA-repeat-containing nuclearRNA. Based on our results (described below), we named thisnovel species of RNA as GAA-repeat-containing RNA (GRC-RNA). Our results are strikingly in contrast to the previous study

Fig. 1. GRC-RNA is a member of the nrRNA family oflncRNAs. (A)RNA-FISH using an AK082015 probe(green) in NIH-3T3 cells revealed the punctate nuclearlocalization of GRC-RNA. In addition to nuclear foci, GRC-RNA also localized to a few cytoplasmic foci (arrows).Scale bar: 10mm. (B) RNA-FISH using FITC-labeled(GAA)15 (a,b), (TTC)15 (c,d), (CCT)15 (e,f) and (CT)20 (g,h)single-stranded oligonucleotide probes in NIH-3T3 cellsrevealed that only the (TTC)15 probe hybridized to nuclear-enriched GRC-RNA (c,d). DNA was counterstained withDAPI (blue). Scale bar: 10mm. (C)Northern blot using a32P-labeled (TTC)15 probe from total RNA revealed thatGRC-RNAs represent heterogeneous transcripts. Actin RNAwas used as a loading control. (D)Northern blot using a32P-labeled (TTC)15 probe in nuclear and cytoplasmic RNAfractions from EpH4 cells revealed nuclear enrichment ofGRC-RNA. MALAT1 and lysine tRNA were used asmarkers for nuclear and cytoplasmic RNA fractionation,respectively.

Jour

nal o

f Cel

l Sci

ence

that claimed that nuclear foci are apparent only when hybridizedwith the (GAA)20 probe (Ohno et al., 2002).

Northern hybridization using a 32P-labeled (TTC)15

oligonucleotide probe revealed that GRC-RNAs wereheterogeneous in length (Fig. 1C). However, the high-molecular-weight (~8 kb) RNA was present only in the mouse cell line.Northern hybridization on nuclear and cytoplasmic RNA frommouse mammary cells using the (TTC)15 oligonucleotide probereconfirmed the RNA-FISH observations that GRC-RNAsconstitute a heterogeneous population of RNAs that arepredominantly enriched in the nuclear fraction (Fig. 1D).Surprisingly, the (TTC)15 probe detected the ~8 kb RNA only inthe total RNA fraction and failed to detect it in the fractionatedRNA samples from multiple cell lines (Fig. 1D; Fig. 2C).

GRC-RNA foci are sensitive to RNase but not DNase ItreatmentTo further confirm the identity of the RNA-FISH signal observedwith the fluorescently labeled (TTC)15 probe, we treated NIH-3T3 cells with RNase A prior to RNA-FISH. RNase Apretreatment failed to eliminate the nuclear-foci-like distributionof GRC-RNA (Fig. 2A), whereas the localization of Neat1lncRNA at paraspeckles was clearly sensitive to RNase Atreatment. Moreover, double-strand RNA-specific ribonucleases,RNase VI (supplementary material Fig. S2a-c) and RNase III

3736 Journal of Cell Science 123 (21)

(supplementary material Fig. S2d–f) failed to remove GRC-RNAfoci when they were incubated with the cells prior to RNA-FISH.However, pretreatment of cells with Riboshredder, a cocktail ofmultiple RNases (Epicenter Inc.) or a single-strand-specificribonuclease, RNase I, completely abolished GRC-RNA foci (Fig.2Bd–i). Systematic analysis with various RNase treatmentsshowed that RNase H treatment eliminated the GRC-RNA focionly when the RNase H treatment was performed after the RNA-FISH (but not prior to RNA-FISH) using the oligonucleotide(TTC)15 probe (supplementary material Fig. S2g–l). Because the(TTC)15 probe consists of single-strand oligonucleotide DNA,the sensitivity of GRC-RNA foci to RNase H treatment afterhybridization demonstrates the presence of single-stranded RNAin these subnuclear foci. Finally, transfection of NIH-3T3 cellswith a modified antisense DNA oligonucleotide that specificallytargets GAA-repeat-containing RNA resulted in the completedisappearance of GRC-RNA foci, further confirming the presenceof RNA in these subnuclear foci (supplementary material Fig.S3). RNase A preferentially cleaves the phosphodiester bonds at3�end of pyrimidines in single-strand RNA, whereas RNase Icleaves the phosphodiester bonds at 3� end of both purines andpyrimidines (D’Alessio and Riordan, 1997; Meador et al., 1990).Sensitivity of GRC-RNA to RNase I and not RNase A indicatedthat GRC-RNA is predominantly enriched in homopurinenucleotide repeats. Furthermore, DNase I treatment of cells prior

Fig. 2. GRC-RNA nuclear foci are sensitive to RNase butnot DNase I treatment. (A)Co-RNA-FISH in NIH-3T3cells for GRC-RNA (green; a,d) and Neat1 RNA (red; b,e)revealed that, unlike Neat1 RNA (e), GRC-RNA foci wereinsensitive to RNase A treatment (d). Merged images areshown in c and f, with chromatin stained with DAPI in blue.Scale bar: 10mm. (B)Co-RNA-FISH in NIH-3T3 cells forGRC-RNA (green; a,d,g,j) and Neat1 RNA (red; b,e,h,k)revealed that both GRC-RNA and Neat1 RNA weredegraded by Riboshredder (d–f) and RNase I (g–i)treatments, but not by DNase I (j–l) treatment. Chromatinwas stained with DAPI in blue. Merged images are shownin c, f, i and l. Scale bar: 10mm. (C)Northern blot using a32P-labeled (TTC)15 probe from nuclear RNA of NIH-3T3cells revealed that RNase A treatment resulted in thecleavage of GRC-RNA into smaller fragments. Longerexposure of the RNase-A-treated RNA blot (RNase A*)revealed the presence of RNase-A-resistant GRC-RNA.(D)Control and RNase-A-treated nuclear RNA were run onan acrylamide gel and hybridized with a 32P-labeled(TTC)15 probe.

Jour

nal o

f Cel

l Sci

ence

to RNA-FISH failed to remove the GRC-RNA nuclear signal(Fig. 2Bj-l), indicating that GRC-RNA foci do not compriseDNA.

To understand the nature of homopurine stretches in GRC-RNA,northern hybridization using a (TTC)15 probe was conducted inuntreated and RNase-A-treated nuclear RNA (Fig. 2C). Theuntreated nuclear extracts from NIH-3T3 cells displayed severalbands of GRC-RNA ranging between 1.5–4 kb. The northernprofile in NIH-3T3 cells looked different from that of mousemammary cells (EpH4; Fig. 1D) and could be attributed to the cell-type-specific changes in the size of GRC-RNAs. RNase A treatmentresulted in the disappearence of most of the bands but a smear of<1 kb in size was observed in the gel. Interestingly, when the blotwas exposed for a longer duration (48 hours; Fig. 2C, RNase A*),specific higher-molecular-weight bands (~3 kb) were also observedin the RNAse-A-treated extracts. To understand the nature of thelow-molecular-weight RNAs (<1 kb) produced upon RNase Atreatment, untreated and RNase-A-treated total nuclear RNA wasresolved in an acrylamide gel and hybridized using the (TTC)15

probe (Fig. 2D). Northern hybridization using a (TTC)15 probe inRNAs isolated from untreated cells failed to detect any specificsignal below 1 kb (Fig. 2D, Control). However, RNase-A-treatedsamples produced several distinct bands of GRC-RNA ranging insize from 0.1–1 kb. These results indicate that the long GRC-RNAs comprise several stretches of homopurine repeats (100 bpto 1 kb) that are interspersed with pyrimidine nucleotides that aresensitive to RNase-A-mediated cleavage. Strikingly, the RNase-A-cleaved homopurine stretches of GRC-RNA continue to localize tothe subnuclear foci (Fig. 2A). This result indicates that either thesubnuclear-foci-enriched GRC-RNAs primarily consist ofhomopurine strands of ~1 kb or association with specific subnuclearstructure(s) maintains the distribution of GRC-RNA in thesesubnuclear domains.

GRC-RNAs associate with GAA.TTC-repeat DNA elementsin the chromatinOur initial attempts to clone the mouse GRC-RNAs remainunsuccessful, probably owing to the large number of repeats presentin the transcripts. Genome-sequencing analyses in mammals haveindicated the presence of a large number of extended GAA.TTC-repeat regions (more than 20 GAA tandem repeats) in several partsof the genome (Clark et al., 2006). We were interested to examinewhether long GAA-repeats show any preference in specific regionsof the mouse genome. We have conducted a bioinformatics-basedsearch in order to estimate the relative positions of GAA-repeatregions in the mouse genome (Fig. 3A). Our analysis indicated thepresence of ~1142 polypurine-rich regions with GAA-repeats inthe mouse genome (at least 7 continuous GAA-repeats and thewhole polypurine repeat length is ~300 bp). Further, a significantlylarge proportion of GAA-repeat regions were preferentially locatedat the 5�or 3�end of the annotated genes, whereas genic regionsseemed to contain fewer GAA-repeats (Fig. 3A). Our results implythat GAA-repeat regions are not randomly distributed in the genomebut are located in close proximity to the regulatory elements in alarge number of genes. Next, we examined whether the GRC-RNAs associate with any of the GAA.TTC-repeat DNA regions inthe mouse genome. We performed RNA-FISH using a digoxigenin-labeled (TTC)15 probe in non-denatured NIH-3T3 cells followedby DNA-FISH using a fluorescently labeled (GAA)15 probe post-DNA denaturation (Fig. 3B; supplementary material Fig. S4). DualRNA-DNA-FISH results revealed that a large number of the GRC-

3737GRC-RNAs and nuclear organization

RNA foci (~80% of GRC-RNA-positive foci; n50 nuclei)colocalized with the (GAA)15 DNA FISH signal, indicating thatseveral of the GRC-RNA foci associate with GAA.TTC-repeatDNA elements (arrows in Fig. 3B; supplementary material Fig.S4) and could be transcribed from these regions. The remainingfraction of the GRC-RNA foci did not associate with GAA.TTC-repeat DNA (~20%; circles in Fig. 3B; supplementary materialFig. S4). Conversely, ~30% of the GAA.TTC DNA foci did notcontain GRC-RNA (arrowheads in Fig. 3B; supplementary materialFig. S4), indicating that these DNA elements might representtranscriptionally inactive regions.

To determine the identity of the GRC-RNA foci that did notassociate with GAA.TTC DNA elements, we conducted a series ofdual-labeling fluorescence microscopic analyses utilizing GRC-RNA RNA-FISH with markers for known subnuclear domains. Noassociation of GRC-RNA with either centromeres (supplementarymaterial Fig. S5a–d) or telomeres (supplementary material Fig.S5e–h) was evident. Moreover, dual RNA-FISH andimmunostaining studies revealed that GRC-RNA foci did notoverlap with any of the known subnuclear domains includingparaspeckles (supplementary material Fig. S5i–l), nuclear speckles(supplementary material Fig. S5m–p), Cajal bodies (supplementarymaterial Fig. S5q–t) and PML bodies (supplementary material Fig.S5u–x), indicating that some of the GRC-RNA foci represent novelsubnuclear domains.

To determine whether GRC-RNA foci coincide with activelytranscribing regions, we performed in vivo short-pulse Br-UTPincorporation assays (5 minutes; Fig. 3C). Surprisingly, the majorityof the GRC-RNA-containing nuclear foci did not colocalize withBr-UTP-labeled nascent transcripts, indicating that the GRC-RNAfoci do not represent highly active transcription sites or are sitesthat are active only at specific stages of the cell cycle. Based onthe DNA-RNA FISH results, we are tempted to speculate that theabsence of Br-UTP incorporation in the GRC-RNA foci that areassociated with GAA.TTC DNA elements could be because GRC-RNAs transcribed from these regions constitute a stable populationof RNA that associate with transcription sites after beingsynthesized. In order to study the cellular turnover of GRC-RNA,RNA-FISH using a (TTC)15 probe was conducted in NIH-3T3 andwild-type mouse embryonic fibroblast (WT-MEF) cells that wereincubated with various RNA polymerase II inhibitors: a-amanitin,5,6-dichloro-1-beta-D-ribobenzimidazole (DRB) and actinomycinD. RNA-FISH revealed that the GRC-RNA foci number within thenuclei remained unaltered after RNA polymerase II transcriptioninhibition (Fig. 3D). However, treatment of WT-MEFs with a-amanitin for more than 14 hours resulted in the decrease of GRC-RNA foci, implying that GRC-RNAs might constitute a stablepopulation of RNA polymerase II transcripts (supplementarymaterial Fig. S6).

GRC-RNAs localize to the midbody in mitosisMany of the subnuclear domains present in interphase nuclei alsoform analogous structures in the cytoplasm during mitosis (Chen etal., 2008; Leung et al., 2004; Prasanth et al., 2003). We examinedthe fate of GRC-RNA foci in mitotic NIH-3T3 cells by RNA-FISHanalysis. The nuclear GRC-RNA foci were disorganized duringprophase soon after the nuclear envelope breakdown (data notshown). GRC-RNA was concentrated in distinct bodies (20–40/cell)in cytoplasm of metaphase cells (Fig. 4Aa,b) and these structureswere more prominent in anaphase cells (Fig. 4Ac,d). The GRC-RNA foci were mostly excluded from the chromosomes and were

Jour

nal o

f Cel

l Sci

ence

distributed in the cytoplasm. Interestingly, in early telophase cells,most of the GRC-RNA foci were congregated in the mid-zone area,a cytoplasmic region located between the two newly formingdaughter nuclei (Fig. 4Ae,f). A similar localization of Xist nrRNA atthe midzone area in telophase cells has been reported previously(Clemson et al., 1996). Finally, in late-telophase cells, a fraction ofthe GRC-RNA was localized in the cytokinetic furrow or midbody(arrow in Fig. 4Ag,h). GRC-RNA continued to associate with themidbody in early G1 cells (Fig. 4Ba–f). The midbody forms acytoplasmic bridge between dividing daughter cells and is ultimatelycleaved to complete cytokinesis (Eggert et al., 2006). The midbodyis characterized by a specific set of proteins and comprises thickbundles of actin and tubulin (Eggert et al., 2006). It is speculated thatcomponents present at the midbody play vital roles in the successfuldivision of daughter cells (Eggert et al., 2006). Interestingly, themidbody-localized GRC-RNA displayed similar characteristics toGRC-RNA that is localized at the subnuclear domains in interphasenuclei. For example, both were resistant to DNase I (Fig. 4Bc,d) and

3738 Journal of Cell Science 123 (21)

RNase A (Fig. 4Be,f) treatments, and sensitive to RNase I treatment(Fig. 4Bg,h). Transfection of an antisense DNA oligonucleotide thatspecifically targets GAA-repeat-containing RNA resulted in completedisappearance of GRC-RNA in the midbody (Fig. 4Bi,j). OthernrRNAs that are known to form cytoplasmic foci in mitotic cells[Neat1 (supplementary material Fig. S7); MALAT1 and CTN-RNA(data not shown)] did not associate with the midbody, indicating thatmidbody distribution is specific to GRC-RNA. A recent study, byperforming RNA-FISH analyses, identified ~10 RNA species (outof ~3000 RNAs tested) to be localized in the midzone area inDrosophila embryos, further emphasizing the importance of RNAlocalization to the midbody (Lecuyer et al., 2007).

GRC-RNAs associate with the nuclear matrix and interactwith nuclear matrix proteinsEarlier studies have reported the association of nrRNAs with thenuclear matrix, a biochemically defined nuclear skeletal non-chromatin structure that is suggested to be involved in crucial

Fig. 3. GAA-repeats are preferentially enriched inthe 5� and 3� regulatory regions of genes. (A)GAA-repeats are preferentially enriched in the 5� and 3�

regulatory regions of genes. Based on the Refseq genedefinition from the UCSC genome browser, the mousegenome is divided into four parts: the gene region (fromstart to end), 50 kb upstream of the gene 5� end, 50 kbdownstream of the gene 3� end, and others called ‘far-intergenic regions’. The distribution of 1142 GAA-enriched regions (with at least 7 continuous GAAs andat least 300 bp of A/G stretch) in each of these four partsis shown by the gray bar. The significance of enrichmentof GAA-repeats in each part is estimated based on thedistribution of randomly selected genomic DNA regions(10 times) with same amount in each chromosome andsame length distribution as the GAA fragments.(***P<0.001). The P-values are: 50 kb up-gene0.00007; in gene0.00002; and 50 kb down-gene0.0006. (B)GRC-RNA foci associate withGAA.TTC-repeat DNA elements: RNA-FISH using a(TTC)15 (red; a,c) probe followed by DNA-FISH using a(GAA)15 probe (green; b,c) revealed association ofGRC-RNA with GAA.TTC-repeat DNA elements in thegenome. Arrows show a few of the representative GRC-RNA foci that colocalize with GAA.TTC-repeats; opencircles represent the GRC-RNA foci that do notassociate with GAA.TTC-repeats; arrowheads representa few of the GAA.TTC-repeat DNA regions that do notassociate with GRC-RNA. (C)GRC-RNA foci do notrepresent highly active transcription sites: RNA-FISHusing a (TTC)15 (red; a,c) probe coupled with in vivoBr-UTP incorporation assays (5 minutes, Br-UTP pulse;green; b,c) revealed weak to no significant overlapbetween GRC-RNA foci and actively transcribing sites(c). (D)GRC-RNA foci contains stable nuclear RNAs:RNA-FISH for GRC-RNA in control (a,e) and RNApolymerase II inhibitor-treated NIH-3T3 cells[(actinomycin D for 1 hour; 5mg/ml; b,f), (DRB for 3hours; 32mg/ml: c,g), (a-amanitin for 6 hours; 50mg/ml:d,h)] showed a similar number of nuclear foci.Chromatin was stained with DAPI (blue). Scale bars:10mm.

Jour

nal o

f Cel

l Sci

ence

nuclear functions including DNA replication, transcription, pre-mRNA processing and mRNP transport (Clemson et al., 1996;Nickerson, 2001; Sone et al., 2007). GRC-RNA nuclear foci wereresistant to detergent and nuclease extraction (data not shown), acharacteristic feature of nuclear-matrix-associated molecules(Nickerson, 2001). RNA-FISH using a GRC-RNA probe clearlydemonstrated the presence of GRC-RNA foci in high-salt-extracted,

3739GRC-RNAs and nuclear organization

DNase-I-treated nuclear matrix preparations (Fig. 5A). We speculatethat the interaction with specific nuclear matrix proteins stabilizesthe GRC-RNA foci in the nucleus. Alternatively, GRC-RNA couldbe involved in the structural organization of the nuclear matrix. Toidentify the factors that interact with GAA-rich GRC-RNA,(GAA)15-, (UUC)15- and scrambled (scr)-RNA oligonucleotideswere covalently coupled to agarose beads and were used in parallelfor RNA affinity purification of proteins from NIH-3T3 nuclearextract. Several protein bands were consistently enriched in the(GAA)15-RNA oligo affinity-purified samples but were also presentat lower concentrations in the (UUC)15-RNA oligo-purified fraction(Fig. 5B). Because the GRC-RNAs in the nuclear extract willhybridize to (UUC)15-RNA oligos, we speculated that a loweramount of the GRC-RNA-interacting proteins can also potentiallybe present in the (UUC)15-RNA oligo-purified fraction. Therefore,bands that showed enrichment in the (GAA)15-RNA oligo affinity-

Fig. 4. GRC-RNAs form cytoplasmic foci in mitotic cells and localize tothe midbody during cell division. (A)Co-RNA-FISH for GRC-RNA (red)and immunostaining for a-tubulin (green) in mitotic NIH-3T3 cells[(metaphase; a,b), (anaphase; c,d), (early telophase; e,f), (late telophase; g,h)]revealed the presence of GRC-RNA in cytoplasmic foci. The arrows in g and hshow the presence of GRC-RNA in the midbody. (B)Co-RNA FISH for GRC-RNA (red) and immunostaining for a-tubulin (green) was conducted in earlyG1 cells that were treated with scrambled DNA oligonucleotide (control; a,b),DNase I (c,d), RNase A (e,f), RNase I (g,h) or GAA-antisense DNAoligonucleotides (i,j). GRC-RNA showed midbody localization (arrows) incontrol, DNase-I- and RNase-A-treated cells. However, RNase I treatment(g,h) and GAA-antisense DNA oligonucleotide transfection (i,j) resulted in theloss of GRC-RNA signal. The chromatin is stained with DAPI (blue). Scalebars: 10mm.

Fig. 5. GRC-RNA is associated with the nuclear matrix. (A)GRC-RNA-FISH in nuclear matrix preparations from transformed MEFs revealed thepresence of GRC-RNA nuclear foci (a,b). The chromatin was stained withDAPI (blue). The absence of DAPI staining in b is due to DNase I treatment,which degrades most of the nuclear DNA. (B)Isolation of the GRC-RNAbinding protein complex. RNA affinity chromatography was performed asdescribed in the Materials and Methods, using affinity matrices containingdifferent RNA sequences [(GAA)15, (UUC)15 and scrambled (scr) RNAoligonucleotides]. For identification, specific bands that showed enrichmentwere excised, digested with trypsin and subjected to mass spectrometry. a-drepresent some of the bands that were excised and analyzed by massspectrometry (a: hnRNP U; b: nucleolin; c: hnRNP I; d: hnRNP A1, A2/B1,hnRNP C1). (C)Western blot analysis of GAA- and UUC-RNA affinity-purified samples using antibodies against proteins that were identified by massspectrometry. (D)Depletion of La/SSB in NIH-3T3 cells does not influencethe nuclear distribution of GRC-RNA. Immunoblot analysis using La-antibodyin cell extracts revealed >75% depletion of La protein in La-siRNA-treatedcell extracts. GRC-RNA FISH (red) in control (a) and La/SSB-siRNA-treated(b) NIH-3T3 cells showed similar distributions of GRC-RNA. The chromatinis stained with DAPI. Scale bars: 10mm.

Jour

nal o

f Cel

l Sci

ence

purified fraction but were also present in the (UUC)15-purifiedfraction at lower amounts (Fig. 5B) were excised from silver-stained gels and subjected to mass spectrometry analysis. Theproteins that displayed higher affinity to GAA-repeat RNAsincluded hnRNP U/SAF-A (Fig. 5B, band a), Nucleolin (C23; Fig.5B, band b), hnRNPs A1 (Fig. 5B, band d), A2/B1 (Fig. 5B, bandd), C1, R and Q. The GRC-RNA also showed a strong associationwith the nuclear-enriched La/SSB antigen. The polypyrimidine-tract-binding protein PTB/hnRNP I (Fig. 5B, band c), which isknown to specifically interact with RNA containing polypyrimidinerich regions, was enriched specifically in the (UUC)15-purifiedfraction, indicating the specificity of the RNA affinity-purificationprocedure. To validate the mass spectrometry results, we carriedout immunoblot analysis of the affinity-purified samples usingantibodies against the proteins that were identified by massspectrometry. The immunoblot assays demonstrated the enrichmentof hnRNP U/SAF-A, hnRNP R, Q, hnRNP A1, hnRNP C1,Nucleolin (C23) and La/SSB antigen in the (GAA)15 RNA oligo-purified fraction compared with (UUC)15 or scr-RNA oligo-purifiedfractions (Fig. 5C). Most of the GAA-repeat RNA-interactingproteins identified in the present study contain RNA-interactingdomains and were previously shown to be integral components ofthe nuclear matrix (Nickerson, 2001).

3740 Journal of Cell Science 123 (21)

The interaction between GAA-repeat RNA and La/SSB is ofparticular interest owing to the suggested role of La in the nuclearretention of cellular and viral RNAs (Brenet et al., 2005; Fok etal., 2006; Intine et al., 2002). In order to understand the role of La-protein in the nuclear retention of GRC-RNA, we examined thedistribution of GRC-RNA in La-depleted cells. RNA-FISH againstGRC-RNA revealed comparable levels of nuclear GRC-RNA inthe control and La-siRNA-treated cells, indicating that La/SSB isnot involved in the nuclear distribution of GRC-RNA (Fig. 5D).

GRC-RNA nuclear foci number is altered during celldifferentiation and proliferationTo determine whether the distribution of GRC-RNA foci aresensitive to changes in gene expression, RNA-FISH wasconducted in NIH-3T3 cells that were serum-starved for 3 days.Serum deprivation of cells is known to alter the expression ofseveral genes that are involved in cell proliferation anddifferentiation (Mandl et al., 2007; Muise-Helmericks et al.,1998). Serum starvation typically causes cell cycle arrest (G0arrest or quiescence) eventually leading to senescence, apoptosisor cellular differentiation (Alexandre et al., 2000; Cooper, 2003;Simm et al., 1997). Interestingly, upon serum starvation, thenumber of GRC-RNA foci within the nuclei was dramatically

Fig. 6. GRC-RNA nuclear foci number is alteredupon serum starvation and during cellulartransformation. (A)GRC-RNA-FISH (green) inuntreated (a,c) and serum-starved (3 days: b,d) NIH-3T3 cells reveal a reduced number of GRC-RNAfoci in serum-deprived cells. Scale bar: 10mm.(B)GRC-RNA-FISH in myoblasts (a,c) andmyotubes (b,d; differentiated by incubating themyoblasts in serum-free media) revealed a reducednumber, but prominent, GRC-RNA foci inmyotubes. Note that the myogenesis marker, myosinheavy chain (MHC) specifically stains differentiatedmyotubes (d) and not proliferating myoblasts (c).Chromatin was stained with DAPI. Scale bar:10mm. (C)GRC-RNA-FISH (red) in WT-MEF(a,c), transformed MEF (b,d), WI38 (e,g), WI38VA13 (f,h), WT-EpH4 (i,k), empty vector (EV)-transfected EpH4 (m,o), transformed constitutivelyactive MEKDD (j-l) or Erbb2 (n,p) EpH4 cell linesrevealed increased numbers of GRC-RNA foci inthe transformed highly proliferating cell lines(SV-40 T-antigen-transformed MEF, WI38 VA13human fibroblast, EpH4-MEKDD and EpH4-Erbb2) compared with their wild-type counterparts(WT-MEF, WI38 primary fibroblast and EpH4 andEpH4-EV mammary cells). DNA wascounterstained with DAPI. Scale bar: 10mm.

Jour

nal o

f Cel

l Sci

ence

decreased (5–8 instead of 50–60) but the existing nuclear focibecame larger and more prominent (Fig. 6A). We have alsoobserved a concomitant increase in the homogenous pool ofGRC-RNA in the serum-starved cell nuclei (compare Fig. 6Abwith Aa).

Similarly, upon serum starvation, C2C12 murine myoblast cellsexit the cell cycle; such a cell cycle arrest is essential for theirdifferentiation into multinucleated myotubes (Yaffe and Saxel,1977). C2C12 cells are studied extensively to understand themolecular events associated with skeletal myogenesis (Yaffe andSaxel, 1977). RNA-FISH using a (TTC)15 probe coupled withimmunofluorescence staining against a myogenesis marker, myosinheavy chain (MHC; expressed specifically in myotubes and inmyoblasts that are cell-cycle-arrested and are committed towardsforming myotubes), revealed that the GRC-RNA formed fewer butlarger and brighter foci in the MHC-expressing cell-cycle-arresteddifferentiating myoblasts and multinucleated myotubes comparedwith proliferating myoblasts (Fig. 6B). The reduced number ofGRC-RNA foci in serum-starved non-dividing NIH-3T3 fibroblastsand C2C12 myotubes suggests that the specific distribution ofGRC-RNA in the form of intranuclear foci might be linked to cellproliferation status. To test this hypothesis, we have compared theGRC-RNA foci distribution in various diploid and transformedcell line pairs. In WT-MEFs, GRC-RNA formed ~10 foci pernucleus (Fig. 6Ca,c). However, SV40 T-antigen-transformed MEFscontained 60–80 GRC-RNA foci (Fig. 6Cb,d). Similarly, thetransformed human fibroblast WI38 VA13 cells showed anincreased number of GRC-RNA foci (~60) relative to its primarywild-type counterpart (WT-WI38; 10–20 foci/nucleus; Fig. 6Ce–h).We have analyzed the GRC-RNA foci status in non-tumorigenic[WT-EpH4 and EpH4-EV (empty vector)] versus tumorigenicmouse mammary epithelial cell line pairs (due to a stableoverexpression of the constitutively active form of the MEK,MEK-DD, EpH4-MekDD or Erbb2 oncogene or EpH4-Erbb2)(Pinkas and Leder, 2002). Consistent with the results in mouse andhuman fibroblasts, the tumorigenic MEK or Erbb2-transformedEpH4 cells displayed greater numbers of GRC-RNA foci than theirwild-type counterparts (Fig. 6Ci–p).

DiscussionRecent studies have indicated that a large number of lncRNAspresent in eukaryotic cells play vital roles in regulating geneexpression (Mercer et al., 2009; Wilusz et al., 2009). The increasingfunctional diversity among lncRNAs observed in the mammaliangenome suggests that they contribute to the complex networksneeded to regulate cell function and could be the ultimate answerto the ‘genome paradox’ (Mercer et al., 2009; Wilusz et al., 2009).

We have identified GRC-RNA, a GAA-repeat-rich lncRNA thatis localized at numerous puncate nuclear foci in both primary andtransformed human and mouse cell lines. GRC-RNAs consist of aheterogeneous population of nuclear RNA. It remains to bedemonstrated whether these transcripts are produced by a singlegene or are encoded by multiple genes in the genome. Genomicanalyses have identified several genomic regions, both in mouseand human, that contain long polypurine-rich sequence elements(including GAA-repeat regions) preferentially located at thepromoter, 5�end or 3�end of protein-coding genes (Goni et al.,2004; Goni et al., 2006; Wu et al., 2007). Interestingly, several ofthese GAA-repeat regions in the mouse genome display higherhistone-3 lysine-36 methylation modifications (H3K36methylation), a histone mark that is associated with transcriptionally

3741GRC-RNAs and nuclear organization

active regions (data not shown). The colocalization of several ofthe GRC-RNA foci with GAA.TTC-DNA-repeat elements supportsthe notion that multiple transcribing regions could be responsiblefor the increased number of GRC-RNA nuclear foci.

Our studies have revealed that GRC-RNA is part of the nuclearscaffold and interacts with several nuclear-matrix-associatedproteins. NrRNAs are known to be an integral part of the nuclearmatrix (examples include Xist RNA and Gomafu RNA) and aresuggested to be involved in maintaining the higher-order chromatinarchitecture in cells (Clemson et al., 1996; Nickerson, 2001;Nickerson et al., 1995; Rodriguez-Campos and Azorin, 2007; Soneet al., 2007). The components that provide structural integrity tothe nuclear matrix remain enigmatic. Evidence indicates that thenuclear matrix comprises a highly ordered network ofribonucleoprotein (RNP) complexes, and in vitro studies haveindicated that proteins like hnRNP A2 and B1 play crucial roles inthe assembly of ordered filament-like nuclear structures (Nickerson,2001; Tan et al., 2000). Because GRC-RNA interacts with hnRNPA2 and B1 and other nuclear matrix proteins, it is tempting tospeculate a possible structural role for GRC-RNA in the formationof the core filament structure underlying nuclear matrixorganization. HnRNP U/SAF-A, another GRC-RNA-interactingprotein, is a bona fide component of the nuclear matrix and is alsothe only nuclear matrix protein shown to specifically interact withscaffold-attachment-regions-DNA (SAR-DNA) (Fackelmayer etal., 1994; Mattern et al., 1996). SAF-A is a multifunctional proteinthat binds to both SAR-DNA and RNA and is suggested to regulategene expression (Eggert et al., 2001; Eggert et al., 1997; Kim andNikodem, 1999). A previous study indicated the role for SAF-A asa plausible candidate for an anchor point of the Xist nrRNA toinactive X-chromosomes (Helbig and Fackelmayer, 2003). SAF-Ainteraction with GRC-RNA could be crucial for the properorganization of GRC-RNA into subnuclear foci.

At present, we do not know the functional significance ofGRC-RNA nuclear foci. The insensitivity of GRC-RNA nuclearfoci to RNase A treatment indicates that GRC-RNAs contain longstretches of polypurine ribonucleotides (GAA-repeats). To ourknowledge, our results on GRC-RNA are the first report showingthe existence of long GAA-repeat-rich RNA in normal cell lines.In vitro studies have indicated that long polypurine RNAs,especially GAA-repeat-containing RNAs, can partially hybridizewith their template DNA strand (dTTC) to form RNA:DNAhybrids (Grabczyk et al., 2007; Grabczyk and Usdin, 2000). Suchhybrids are favored when the non-template strand (dGAA) isengaged as the third strand to form a triple-helical structure (GAA-GAA-TTC) by hybridizing with a DNA duplex via non-Watson-Crick hydrogen bonds (Hoogsteen base pairing and reverseHoogsteen base pairing) (Grabczyk et al., 2007; Grabczyk andUsdin, 2000). Furthermore, it has been demonstrated thatRNA:DNA hybrids and triple-helix DNA structure formed onsuch templates arrest RNA polymerses, resulting in transcriptioninhibition (Grabczyk et al., 2007; Grabczyk and Usdin, 2000;Ohshima et al., 1998; Wells, 2008). In situations described above,GAA-repeat RNA coexists with the triplex DNA structure.Abnormal amplification of GAA-tandem repeats in the transcribingregion of the frataxin (FXN) gene resulting in its transcriptionalsilencing owing to the formation of an RNA:DNA hybrid andtriple-helix DNA structure has been implicated in case ofFriedreich’s ataxia (Bidichandani et al., 1998; Grabczyk et al.,2007; Grabczyk and Usdin, 2000; Ohshima et al., 1998; Sakamotoet al., 2001). Previous studies by Ohno et al. demonstrated the

Jour

nal o

f Cel

l Sci

ence

3742 Journal of Cell Science 123 (21)

association between GAA.TTC foci and triple-helix structures(Ohno et al., 2002). However, this study claimed that theGAA.TTC nuclear foci primarily consisted of GAA-GAA-TTCtriple-helix DNA and the single-stranded template (TTC)n DNA.By contrast, we have provided several lines of evidence confirmingthe existence of polypurine-repeat-rich GRC-RNA in these novelsubnuclear domains. These include: (1) sensitivity of GRC-RNAto RNase I, a single-strand-specific ribonuclease that cleavesphosphodiester bonds at the 3� end of all the ribonucleotides, andto Riboshredder, a cocktail of multiple ribonucleases; (2) absenceof GRC-RNA nuclear foci in cells that are transfected withantisense DNA oligonucleotides complementary only to GAA-repeat (GAA-antisense), but not TTC-repeat, sequences (GAA-sense); and (3) differentiation- or proliferation-specific changes inthe distribution of GRC-RNA foci in cell lines. Further,insensitivity of GRC-RNA foci to DNase I treatment ruled out therole of DNA in the formation of these nuclear foci. Taken together,we demonstrate the presence of GRC-RNAs in these subnuclearfoci that could coexist with triple-helix DNA structure in thesedomains. Based on our data, we speculate that the GRC-RNAtranscribed from several of the GAA.TTC-repeat elements tend toassociate with those genomic regions and facilitate the formationor stabilization of triple-helical structures, exerting a regulatoryeffect on transcription of those genes that are located in closeproximity to the repeat elements. In such a scenario, the GRC-RNAs might behave as riboregulators and modulate the expressionof several genes in a developmental stage- or cell-cycle-dependentmanner. Its upregulation in cancerous cells and redistribution afterserum starvation, as well as during myogenesis, suggestphysiological and developmental relevance. Future studies onhomopurine-rich RNAs like GRC-RNA will provide insights intothe role of such unique structures in regulating gene expression.

Materials and MethodsCell culture and treatmentNIH-3T3, WT-MEFs, U2OS, HeLa, C2C12, WT and mutant EpH4, WI38 andWI38-VA13 cells were grown in Dulbecco’s Modified Eagle Medium (DMEM)containing high glucose (Invitrogen, Carlsbad, CA) supplemented with penicillin-streptomycin and 10% fetal bovine serum or calf serum (NIH-3T3; Hyclone, Logan,UT). For the serum-starvation experiment, cells were cultured in DMEMsupplemented with 0.1% calf serum for 3 days. Cells were seeded onto acid-washedcoverslips for immunofluorescence or RNA-FISH experiments (Prasanth et al.,2005). For transient transfection experiments, plasmid DNA (2 mg) was transfectedinto cells using lipofectamine 2000 (Invitrogen) according to the manufacturer’sinstructions.

To inhibit RNA polymerase II transcription, cells were incubated with a-amanitin(50 mg/ml, Sigma-Aldrich, St Louis, MO), DRB (32 mg/ml, Sigma-Aldrich) oractinomycin D (5 mg/ml) for indicated time-points. Nascent transcription sites weredetected using Br-UTP incorporation assays using a previously published protocol(Prasanth et al., 2003).

Nuclease treatmentCells were fixed in 2–4% formaldehyde for 15 minutes at room temperature,permeabilized using 0.5% Triton X-100 for 10 minutes at 4°C and were treated withdifferent nucleases for 1 hour in a moist chamber at 37°C: RNase A (1 mg/ml inCSK buffer, Sigma-Aldrich); RNase I (200 U/ml in 1� RNase I reaction buffer,Promega, Madison, WI); Riboshredder (50 U/ml, Epicenter Biotechnologies,Madison, WI); RNase H (100 U/ml, Epicenter Biotechnologies); RNase V1 (EpicenterBiotechnologies); RNase III (100 U/ml, Epicenter Biotechnologies); and DNase I(200 U/ml, Roche, Madison, WI). After treatment, cells were washed with PBS andprocessed for RNA-FISH.

Cellular fractionation and northern blotNuclear and cytoplasmic fractionation and RNA isolation from cell lines and northernblot hybridization using random-labeled a-32P dCTP-labeled probes were performedaccording to previously published procedures (Prasanth et al., 2005; Topisirovic etal., 2003). Northern hybridization using g-32P ATP-labeled DNA oligonucleotideprobes (TTC)15 was performed as per the manufacturer’s instructions (Ambion,Austin, TX).

Nuclear matrix preparationCells were extracted according to the protocol described previously with modifications(Sone et al., 2007). Cells grown on poly-L-lysine-coated coverslips werepermeabilized in cytoskeleton (CSK) buffer (100 mM NaCl, 300 mM sucrose, 10mM PIPES, pH 6.8, 3 mM MgCl2) with 0.5% Triton X-100 on ice for 10 minutesand then extracted in extraction buffer (250 mM ammonium sulfate, 300 mMsucrose, 10 mM PIPES, pH 6.8, 3 mM MgCl2, 0.5% Triton X-100) on ice for 5minutes. After two washes in digestion buffer (50 mM NaCl, 300 mM sucrose, 10mM PIPES, pH 6.8, 3 mM MgCl2), cells were treated with 500 U/ml DNase I(Roche) for 1 hour at 37°C. Cells were then fixed in 4% formaldehyde at roomtemperature for 15 minutes and were processed for RNA-FISH.

RNA-FISH, dual RNA-DNA-FISH and immunofluorescence microscopyFor RNA-FISH, cells were either fixed in 4% formaldehyde in PBS (pH 7.4) for 15minutes at room temperature and then permeabilized with 0.5% Triton X-100 in PBSon ice for 10 minutes, or pre-extracted in CSK buffer plus 0.5% Triton X-100 for 5minutes on ice and then fixed using 4% formaldehyde. Cells were washed threetimes in PBS and rinsed once in 2� SSC prior to hybridization. Hybridization wascarried out in a moist chamber at 37°C for 12–16 hours as previously described(Prasanth et al., 2005). Texas Red or FITC-labeled oligonucleotide probes [(TTC)15,(GAA)15, (CCT)15 and (CT)20] were synthesized by Sigma-Genosys, St Louis, MOand nick-translated probe was made as per the manufacturer’s instructions (AbbottLaboratories, Abbott Park, IL).

Dual RNA-DNA-FISH was performed as previously described (Xing et al., 1995).In brief, RNA-FISH using digoxigenin (DIG)-labeled (TTC)15 (Sigma Genosys)probe was performed in CSK plus 0.5% Triton X-100 pre-extracted NIH-3T3 cellsand the DIG label was detected using anti-DIG-rhodamine antibody (1:200; Roche).Cells were refixed in 4% formaldehyde, DNA was denatured using heat denaturation(4 minutes at 72–75°C in 70% formamide plus 2� SSC) and hybridized using FITC-labeled (GAA)15 oligonucleotide probe.

For RNA-FISH coupled with immunofluorescence staining, cells were fixed andpermeabilized as described above, and then blocked in PBS containing 1% RNase-free fetal bovine serum (Invitrogen). Cells were incubated with primary antibodiesfor 1 hour at room temperature in a moist chamber: anti-centromere antibody (AnaC;human IgG, 1:10, Sigma-Aldrich); anti-coilin (rabbit IgG, 1:100, Santa CruzBiotechnology, CA), anti-SF2/ASF mAb103 (mouse IgG, 1:100, a kind gift from DrA. Krainer, CSHL, NY); and anti-BrdU [mouse IgG, 1:100, Sigma-Aldrich]. Cellswere then washed in PBS containing 1% RNase-free fetal bovine serum, andsecondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA)were added for 45 minutes at room temperature: goat anti-mouse (GAM) IgG TexasRed (1:1000); GAM-FITC (1:500); donkey anti-rabbit IgG Texas Red (1:1000);DAR-FITC (1:500); and goat anti-human IgG-FITC (1:500). After secondaryantibodies, cells were washed three times in PBS, fixed for 5 minutes in 4%formaldehyde and processed for RNA-FISH.

Cells were examined using an AxioImager.Z1 fluorescence microscope (CarlZeiss Inc., Thornwood, NY) equipped with Chroma filters (Chroma Technology,Bellows Falls, VT) using a 63�/1.4 objective (PlanApochromat, Carl Zeiss, Inc.).Images were acquired by a Hamamatsu ORCA-ER camera and processed usingAxioVision Rel 4.7 software (Carl Zeiss, Inc.). Alternatively, cells were imagedusing a Deltavision microscope (Applied Precision, Issaquah, WA) with a 60�/1.42objective (PlanApochromat, Olympus, Canter Valley, PA) and images weredeconvolved using SoftWorx software.

Antisense oligonucleotide and siRNA treatmentSense and antisense DNA oligonucleotides targeting GAA-repeats (GAA-antisenseoligo, TTCTTCTTCTTCTTCTTCTTC; GAA-sense oligo, GAAGAAGAAGAAGA -AGAAGAA) were modified with six 2�-o-methoxyethyl nucleotides on the 5� and3� ends and nine consecutive oligodeoxynucleotides to support RNase H activity andwere used to deplete mouse GRC-RNA (ISIS Pharmaceuticals, Carlsbad, CA). Theoligonucleotides were modified with phosphorothioate internucleosidic linkages formaximum stability. The oligonucleotides (final concentration: 100 nM) weretransfected to cells using Lipofectamine RNAiMAX Reagent as per themanufacturer’s instructions (Invitrogen). Depletion of mouse La protein wasperformed by using pooled double-stranded siRNAs (sc-40916; Santa CruzBiotechnology, CA) as previously described (Prasanth et al., 2004).

RNA affinity chromatographyRNA oligonucleotides [(GAA)15, (UUC)15 and scrambled (scr; GAUGACCG -UUCAGCAACCUUCUACGCAUACGCCAUCUCGACGGGAUCCAGACGGA)Dharmacon Inc., Lafayette, CO] were covalently coupled to agarose beads (Sigma-Aldrich) using a method previously described (Caputi et al., 1999). To each aliquotof resin, 250 ml of NIH-3T3 nuclear extract was loaded in a total volume of 600 mland was incubated at 30°C for 1 hour. Unbound proteins were removed by washingwith 1 ml Dignam buffer D (20 mM HEPES, pH 7.9, 20% glycerol, 100 mM KCl,0.2 mM EDTA, 1 mM DTT, 0.5 mM PMSF) (Dignam et al., 1983) containing 0.2 MNaCl and 0.1 mg/ml yeast tRNA at 4°C followed by two washes with buffer Dcontaining 0.04 mg/ml yeast tRNA, and two washes with buffer D alone. Specifically-bound proteins were released by boiling in Laemmli buffer for 5 minutes and wereresolved by SDS-PAGE and silver staining (SilverQuest Kit, Invitrogen). Specific

Jour

nal o

f Cel

l Sci

ence

3743GRC-RNAs and nuclear organization

protein bands were excised from the silver-stained gels and were used for massspectrometric analysis.

Western blot analysisProtein samples were resolved by 10% SDS-PAGE, transferred onto nitrocellulosemembrane and used for western blot (WB) analysis. The primary antibodies included:anti-mouse hnRNP U/SAF-A antibody (3G6; 1:2000) (Kiledjian and Dreyfuss,1992); anti-mouse hnRNP A1 (4B10; 1:5000; Dr G. Dreyfuss, University ofPennsylvania, PA); anti-mouse hnRNP C1/C2 (4F4; 1:1000) (Choi and Dreyfuss,1984); anti-mouse hnRNP Q & R (18E4; 1:2000; Sigma-Aldrich); anti-mouse hnRNPL (4D11; 1:1000; Sigma-Aldrich); anti-rabbit nucleolin (C23; 1:10; NovusBiologicals, Littleton, CO); anti-rabbit La (1:1000) (Park et al., 2006) and anti-rabbitPTB (1:20; Dr S. Huang, Northwestern University, Chicago, IL). HRP-labeledsecondary antibodies (Jackson ImmunoResearch) followed by ECL assays (PierceInc., Rockford, IL) were used for detecting the specific signal.

We thank Dr P. M. Yau for mass spectrometry analysis. We thankDrs M. Bellini, G. Carmichael, J. Chen, Z. Deng, G. Dreyfuss, M.Hastings, S. Huang, A. Krainer, P. Lieberman, R. Maraia, S. Nakagawaand Z. Zhang for sharing reagents, protocols and for helpful discussions.We thank Drs S. Ceman, J. Chen, P. Leder, C. Mizzen and Y. Ge forproviding cell lines. We thank A. Zhanar for the technical help. Wealso thank Drs P. Bubulya, A. Lal, P. Newmark and Prasanth laboratorymembers for helpful discussions and for critically reading themanuscript. This work is supported by grants UIUC-ICR-MCB toK.V.P. and S.G.P. and NSF0843604 to S.G.P.

Supplementary material available online athttp://jcs.biologists.org/cgi/content/full/123/21/3734/DC1

ReferencesAlexandre, S., Rast, C., Nguyen-Ba, G. and Vasseur, P. (2000). Detection of apoptosis

induced by topoisomerase inhibitors and serum deprivation in syrian hamster embryocells. Exp. Cell Res. 255, 30-39.

Amaral, P. P., Dinger, M. E., Mercer, T. R. and Mattick, J. S. (2008). The eukaryoticgenome as an RNA machine. Science 319, 1787-1789.

Bidichandani, S. I., Ashizawa, T. and Patel, P. I. (1998). The GAA triplet-repeatexpansion in Friedreich ataxia interferes with transcription and may be associated withan unusual DNA structure. Am. J. Hum. Genet. 62, 111-121.

Braidotti, G., Baubec, T., Pauler, F., Seidl, C., Smrzka, O., Stricker, S., Yotova, I. andBarlow, D. P. (2004). The Air noncoding RNA: an imprinted cis-silencing transcript.Cold Spring Harbor Symp. Quant. Biol. 69, 55-66.

Brenet, F., Dussault, N., Borch, J., Ferracci, G., Delfino, C., Roepstorff, P., Miquelis,R. and Ouafik, L. (2005). Mammalian peptidylglycine alpha-amidating monooxygenasemRNA expression can be modulated by the La autoantigen. Mol. Cell. Biol. 25, 7505-7521.

Brockdorff, N., Ashworth, A., Kay, G. F., McCabe, V. M., Norris, D. P., Cooper, P. J.,Swift, S. and Rastan, S. (1992). The product of the mouse Xist gene is a 15 kb inactiveX-specific transcript containing no conserved ORF and located in the nucleus. Cell 71,515-526.

Caputi, M., Mayeda, A., Krainer, A. R. and Zahler, A. M. (1999). hnRNP A/B proteinsare required for inhibition of HIV-1 pre-mRNA splicing. EMBO J. 18, 4060-4067.

Chen, L. L. and Carmichael, G. G. (2009). Altered nuclear retention of mRNAscontaining inverted repeats in human embryonic stem cells: functional role of a nuclearnoncoding RNA. Mol. Cell 35, 467-478.

Chen, Y. C., Kappel, C., Beaudouin, J., Eils, R. and Spector, D. L. (2008). Live celldynamics of promyelocytic leukemia nuclear bodies upon entry into and exit frommitosis. Mol. Biol. Cell 19, 3147-3162.

Cheng, J., Kapranov, P., Drenkow, J., Dike, S., Brubaker, S., Patel, S., Long, J., Stern,D., Tammana, H., Helt, G. et al. (2005). Transcriptional maps of 10 humanchromosomes at 5-nucleotide resolution. Science 308, 1149-1154.

Choi, Y. D. and Dreyfuss, G. (1984). Isolation of the heterogeneous nuclear RNA-ribonucleoprotein complex (hnRNP): a unique supramolecular assembly. Proc. Natl.Acad. Sci. USA 81, 7471-7475.

Chow, J. C., Ciaudo, C., Fazzari, M. J., Mise, N., Servant, N., Glass, J. L., Attreed,M., Avner, P., Wutz, A., Barillot, E. et al. (2010). LINE-1 activity in facultativeheterochromatin formation during X chromosome inactivation. Cell 141, 956-969.

Chueh, A. C., Northrop, E. L., Brettingham-Moore, K. H., Choo, K. H. andWong, L. H. (2009). LINE retrotransposon RNA is an essential structural andfunctional epigenetic component of a core neocentromeric chromatin. PLoS Genet. 5,e1000354.

Clark, R. M., Bhaskar, S. S., Miyahara, M., Dalgliesh, G. L. and Bidichandani, S. I.(2006). Expansion of GAA trinucleotide repeats in mammals. Genomics 87, 57-67.

Clemson, C. M., McNeil, J. A., Willard, H. F. and Lawrence, J. B. (1996). XIST RNApaints the inactive X chromosome at interphase: evidence for a novel RNA involved innuclear/chromosome structure. J. Cell Biol. 132, 259-275.

Clemson, C. M., Hutchinson, J. N., Sara, S. A., Ensminger, A. W., Fox, A. H.,Chess, A. and Lawrence, J. B. (2009). An architectural role for a nuclear noncodingRNA: NEAT1 RNA is essential for the structure of paraspeckles. Mol. Cell 33, 717-726.

Cooper, S. (2003). Reappraisal of serum starvation, the restriction point, G0, and G1 phasearrest points. FASEB J. 17, 333-340.

D’Alessio, G. and Riordan, J. F. (1997). Ribonucleases: Structures and Functions. SanDiego: Academic Press.

Dignam, J. D., Lebovitz, R. M. and Roeder, R. G. (1983). Accurate transcriptioninitiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei.Nucleic Acids Res. 11, 1475-1489.

Dinger, M. E., Pang, K. C., Mercer, T. R., Crowe, M. L., Grimmond, S. M. andMattick, J. S. (2009). NRED: a database of long noncoding RNA expression. NucleicAcids Res. 37, D122-D126.

Eggert, H., Schulz, M., Fackelmayer, F. O., Renkawitz, R. and Eggert, M. (2001).Effects of the heterogeneous nuclear ribonucleoprotein U (hnRNP U/SAF-A) onglucocorticoid-dependent transcription in vivo. J. Steroid Biochem. Mol. Biol. 78, 59-65.

Eggert, M., Michel, J., Schneider, S., Bornfleth, H., Baniahmad, A., Fackelmayer, F.O., Schmidt, S. and Renkawitz, R. (1997). The glucocorticoid receptor is associatedwith the RNA-binding nuclear matrix protein hnRNP U. J. Biol. Chem. 272, 28471-28478.

Eggert, U. S., Mitchison, T. J. and Field, C. M. (2006). Animal cytokinesis: from partslist to mechanisms. Annu. Rev. Biochem. 75, 543-566.

Fackelmayer, F. O., Dahm, K., Renz, A., Ramsperger, U. and Richter, A. (1994).Nucleic-acid-binding properties of hnRNP-U/SAF-A, a nuclear-matrix protein whichbinds DNA and RNA in vivo and in vitro. Eur. J. Biochem. 221, 749-757.

Fok, V., Friend, K. and Steitz, J. A. (2006). Epstein-Barr virus noncoding RNAs areconfined to the nucleus, whereas their partner, the human La protein, undergoesnucleocytoplasmic shuttling. J. Cell Biol. 173, 319-325.

Goni, J. R., de la Cruz, X. and Orozco, M. (2004). Triplex-forming oligonucleotidetarget sequences in the human genome. Nucleic Acids Res. 32, 354-360.

Goni, J. R., Vaquerizas, J. M., Dopazo, J. and Orozco, M. (2006). Exploring the reasonsfor the large density of triplex-forming oligonucleotide target sequences in the humanregulatory regions. BMC Genomics 7, 63.

Grabczyk, E. and Usdin, K. (2000). The GAA*TTC triplet repeat expanded in Friedreich’sataxia impedes transcription elongation by T7 RNA polymerase in a length and supercoildependent manner. Nucleic Acids Res. 28, 2815-2822.

Grabczyk, E., Mancuso, M. and Sammarco, M. C. (2007). A persistent RNA.DNAhybrid formed by transcription of the Friedreich ataxia triplet repeat in live bacteria, andby T7 RNAP in vitro. Nucleic Acids Res. 35, 5351-5359.

Grewal, S. I. and Elgin, S. C. (2007). Transcription and RNA interference in the formationof heterochromatin. Nature 447, 399-406.

Guttman, M., Amit, I., Garber, M., French, C., Lin, M. F., Feldser, D., Huarte, M.,Zuk, O., Carey, B. W., Cassady, J. P. et al. (2009). Chromatin signature reveals overa thousand highly conserved large non-coding RNAs in mammals. Nature 458, 223-227.

Helbig, R. and Fackelmayer, F. O. (2003). Scaffold attachment factor A (SAF-A) isconcentrated in inactive X chromosome territories through its RGG domain.Chromosoma 112, 173-182.

Intine, R. V., Dundr, M., Misteli, T. and Maraia, R. J. (2002). Aberrant nucleartrafficking of La protein leads to disordered processing of associated precursor tRNAs.Mol. Cell 9, 1113-1123.

Khalil, A. M., Guttman, M., Huarte, M., Garber, M., Raj, A., Rivea Morales, D.,Thomas, K., Presser, A., Bernstein, B. E., van Oudenaarden, A. et al. (2009). Manyhuman large intergenic noncoding RNAs associate with chromatin-modifying complexesand affect gene expression. Proc. Natl. Acad. Sci. USA 106, 11667-11672.

Kiledjian, M. and Dreyfuss, G. (1992). Primary structure and binding activity of thehnRNP U protein: binding RNA through RGG box. EMBO J. 11, 2655-2664.

Kim, M. K. and Nikodem, V. M. (1999). hnRNP U inhibits carboxy-terminal domainphosphorylation by TFIIH and represses RNA polymerase II elongation. Mol. Cell.Biol. 19, 6833-6844.

Lecuyer, E., Yoshida, H., Parthasarathy, N., Alm, C., Babak, T., Cerovina, T., Hughes,T. R., Tomancak, P. and Krause, H. M. (2007). Global analysis of mRNA localizationreveals a prominent role in organizing cellular architecture and function. Cell 131, 174-187.

Lee, J. T., Davidow, L. S. and Warshawsky, D. (1999). Tsix, a gene antisense to Xist atthe X-inactivation centre. Nat. Genet. 21, 400-404.

Lein, E. S., Hawrylycz, M. J., Ao, N., Ayres, M., Bensinger, A., Bernard, A., Boe, A.F., Boguski, M. S., Brockway, K. S., Byrnes, E. J. et al. (2007). Genome-wide atlasof gene expression in the adult mouse brain. Nature 445, 168-176.

Leung, A. K., Gerlich, D., Miller, G., Lyon, C., Lam, Y. W., Lleres, D., Daigle, N.,Zomerdijk, J., Ellenberg, J. and Lamond, A. I. (2004). Quantitative kinetic analysisof nucleolar breakdown and reassembly during mitosis in live human cells. J. Cell Biol.166, 787-800.

Maison, C., Bailly, D., Peters, A. H., Quivy, J. P., Roche, D., Taddei, A., Lachner, M.,Jenuwein, T. and Almouzni, G. (2002). Higher-order structure in pericentricheterochromatin involves a distinct pattern of histone modification and an RNAcomponent. Nat. Genet. 30, 329-334.

Mandl, A., Sarkes, D., Carricaburu, V., Jung, V. and Rameh, L. (2007). Serumwithdrawal-induced accumulation of phosphoinositide 3-kinase lipids in differentiating3T3-L6 myoblasts: distinct roles for Ship2 and PTEN. Mol. Cell. Biol. 27, 8098-8112.

Mattern, K. A., Humbel, B. M., Muijsers, A. O., de Jong, L. and van Driel, R. (1996).hnRNP proteins and B23 are the major proteins of the internal nuclear matrix of HeLaS3 cells. J. Cell Biochem. 62, 275-289.

Mattick, J. S. (2005). The functional genomics of noncoding RNA. Science 309, 1527-1528.

Jour

nal o

f Cel

l Sci

ence

3744 Journal of Cell Science 123 (21)

Mattick, J. S. (2009). The genetic signatures of noncoding RNAs. PLoS Genet. 5,e1000459.

Mattick, J. S. and Makunin, I. V. (2006). Non-coding RNA. Hum. Mol. Genet. 15 SpecNo 1, R17-R29.

Meador, J., 3rd, Cannon, B., Cannistraro, V. J. and Kennell, D. (1990). Purificationand characterization of Escherichia coli RNase I. Comparisons with RNase M. Eur. J.Biochem. 187, 549-553.

Mercer, T. R., Dinger, M. E., Sunkin, S. M., Mehler, M. F. and Mattick, J. S. (2008).Specific expression of long noncoding RNAs in the mouse brain. Proc. Natl. Acad. Sci.USA 105, 716-721.

Mercer, T. R., Dinger, M. E. and Mattick, J. S. (2009). Long non-coding RNAs: insightsinto functions. Nat. Rev. Genet. 10, 155-159.

Muise-Helmericks, R. C., Grimes, H. L., Bellacosa, A., Malstrom, S. E., Tsichlis, P. N.and Rosen, N. (1998). Cyclin D expression is controlled post-transcriptionally via aphosphatidylinositol 3-kinase/Akt-dependent pathway. J. Biol. Chem. 273, 29864-29872.

Nickerson, J. (2001). Experimental observations of a nuclear matrix. J. Cell Sci. 114, 463-474.

Nickerson, J. A., Blencowe, B. J. and Penman, S. (1995). The architectural organizationof nuclear metabolism. Int. Rev. Cytol. 162A, 67-123.

Ohno, M., Fukagawa, T., Lee, J. S. and Ikemura, T. (2002). Triplex-forming DNAs inthe human interphase nucleus visualized in situ by polypurine/polypyrimidine DNAprobes and antitriplex antibodies. Chromosoma 111, 201-213.

Ohshima, K., Montermini, L., Wells, R. D. and Pandolfo, M. (1998). Inhibitory effectsof expanded GAA.TTC triplet repeats from intron I of the Friedreich ataxia gene ontranscription and replication in vivo. J. Biol. Chem. 273, 14588-14595.

Pandey, R. R., Mondal, T., Mohammad, F., Enroth, S., Redrup, L., Komorowski, J.,Nagano, T., Mancini-Dinardo, D. and Kanduri, C. (2008). Kcnq1ot1 antisensenoncoding RNA mediates lineage-specific transcriptional silencing through chromatin-level regulation. Mol. Cell 32, 232-246.

Park, J. M., Kohn, M. J., Bruinsma, M. W., Vech, C., Intine, R. V., Fuhrmann, S.,Grinberg, A., Mukherjee, I., Love, P. E., Ko, M. S. et al. (2006). The multifunctionalRNA-binding protein La is required for mouse development and for the establishmentof embryonic stem cells. Mol. Cell. Biol. 26, 1445-1451.

Pederson, T. (2009). The discovery of eukaryotic genome design and its forgottencorollary-the postulate of gene regulation by nuclear RNA. FASEB J. 23, 2019-2021.

Pinkas, J. and Leder, P. (2002). MEK1 signaling mediates transformation and metastasisof EpH4 mammary epithelial cells independent of an epithelial to mesenchymaltransition. Cancer Res. 62, 4781-4790.

Prasanth, K. V. and Spector, D. L. (2007). Eukaryotic regulatory RNAs: an answer tothe ‘genome complexity’ conundrum. Genes Dev. 21, 11-42.

Prasanth, K. V., Sacco-Bubulya, P. A., Prasanth, S. G. and Spector, D. L. (2003).Sequential entry of components of the gene expression machinery into daughter nuclei.Mol. Biol. Cell 14, 1043-1057.

Prasanth, K. V., Prasanth, S. G., Xuan, Z., Hearn, S., Freier, S. M., Bennett, C. F.,Zhang, M. Q. and Spector, D. L. (2005). Regulating gene expression through RNAnuclear retention. Cell 123, 249-263.

Prasanth, S. G., Prasanth, K. V., Siddiqui, K., Spector, D. L. and Stillman, B. (2004).Human Orc2 localizes to centrosomes, centromeres and heterochromatin duringchromosome inheritance. EMBO J. 23, 2651-2663.

Rodriguez-Campos, A. and Azorin, F. (2007). RNA is an integral component of chromatinthat contributes to its structural organization. PLoS ONE 2, e1182.

Sakamoto, N., Ohshima, K., Montermini, L., Pandolfo, M. and Wells, R. D. (2001).Sticky DNA, a self-associated complex formed at long GAA*TTC repeats in intron 1of the frataxin gene, inhibits transcription. J. Biol. Chem. 276, 27171-27177.

Sasaki, Y. T., Ideue, T., Sano, M., Mituyama, T. and Hirose, T. (2009). MENepsilon/betanoncoding RNAs are essential for structural integrity of nuclear paraspeckles. Proc.Natl. Acad. Sci. USA 106, 2525-2530.

Simm, A., Bertsch, G., Frank, H., Zimmermann, U. and Hoppe, J. (1997). Cell deathof AKR-2B fibroblasts after serum removal: a process between apoptosis and necrosis.J. Cell Sci. 110, 819-828.

Sone, M., Hayashi, T., Tarui, H., Agata, K., Takeichi, M. and Nakagawa, S. (2007).The mRNA-like noncoding RNA Gomafu constitutes a novel nuclear domain in asubset of neurons. J. Cell Sci. 120, 2498-2506.

Struhl, K. (2007). Transcriptional noise and the fidelity of initiation by RNA polymeraseII. Nat. Struct. Mol. Biol. 14, 103-105.

Sunwoo, H., Dinger, M. E., Wilusz, J. E., Amaral, P. P., Mattick, J. S. and Spector, D.L. (2009). MEN varepsilon/beta nuclear-retained non-coding RNAs are up-regulatedupon muscle differentiation and are essential components of paraspeckles. Genome Res.19, 347-359.

Tan, J. H., Wooley, J. C. and LeStourgeon, W. M. (2000). Nuclear matrix-like filamentsand fibrogranular complexes form through the rearrangement of specific nuclearribonucleoproteins. Mol. Biol. Cell 11, 1547-1554.

Topisirovic, I., Culjkovic, B., Cohen, N., Perez, J. M., Skrabanek, L. and Borden, K.L. (2003). The proline-rich homeodomain protein, PRH, is a tissue-specific inhibitor ofeIF4E-dependent cyclin D1 mRNA transport and growth. EMBO J. 22, 689-703.

van Bakel, H., Nislow, C., Blencowe, B. J. and Hughes, T. R. (2010). Most “darkmatter” transcripts are associated with known genes. PLoS Biol. 8, e1000371.

Wells, R. D. (2008). DNA triplexes and Friedreich ataxia. FASEB J. 22, 1625-1634.Wilusz, J. E., Sunwoo, H. and Spector, D. L. (2009). Long noncoding RNAs: functional

surprises from the RNA world. Genes Dev. 23, 1494-1504.Wong, L. H., Brettingham-Moore, K. H., Chan, L., Quach, J. M., Anderson, M. A.,

Northrop, E. L., Hannan, R., Saffery, R., Shaw, M. L., Williams, E. et al. (2007).Centromere RNA is a key component for the assembly of nucleoproteins at the nucleolusand centromere. Genome Res. 17, 1146-1160.

Wu, Q., Gaddis, S. S., MacLeod, M. C., Walborg, E. F., Thames, H. D., DiGiovanni,J. and Vasquez, K. M. (2007). High-affinity triplex-forming oligonucleotide targetsequences in mammalian genomes. Mol. Carcinog. 46, 15-23.

Xing, Y., Johnson, C. V., Moen, P. T., Jr, McNeil, J. A. and Lawrence, J. (1995).Nonrandom gene organization: structural arrangements of specific pre-mRNAtranscription and splicing with SC-35 domains. J. Cell Biol. 131, 1635-1647.

Yaffe, D. and Saxel, O. (1977). Serial passaging and differentiation of myogenic cellsisolated from dystrophic mouse muscle. Nature 270, 725-727.

Jour

nal o

f Cel

l Sci

ence