Abstract
The mechanisms by which epigenetic modifications are established in gene regulatory regions of active genes remain poorly understood. The data presented show that the establishment and recycling of a major epigenetic mark, the acetylated form of the replacement histone H2A.Z, is regulated by cell cycle-specific long noncoding RNAs encoded in regions adjacent to the promoters of active genes. These transcripts, termed SPEARs (S Phase EArly RNAs), are induced in early S phase: their expression precedes that of the downstream genes on which they exert their regulatory action. SPEARs drive the modification and deposition of the acetylated form of histone H2A.Z by bringing together the replacement histone and the histone acetyl transferase TIP60. This widespread bimodal pathway constitutes a novel RNA-mediated mechanism for the establishment of epigenetic marks and cell-specific epigenetic profiles, thereby providing a unifying explanation for the accuracy and persistence of epigenetic marks on chromatin.
A wealth of epigenomics data have recently provided meaningful biological and clinical information but little attention has been paid to the mechanisms for the establishment (‘writing’) and the maintenance of specific epigenetic patterns. Epigenetic aberrations are a frequent feature of deleterious conditions such as cancers and other genetic diseases1. In particular, subsets of histone variants and their post-translational modifications have emerged as prominent players among other epigenetic marks. The histone H2A.Z (the major H2A variant), for instance, is associated with gene activation2–5 by means of the balance between the acetylated vs. the non-acetylated forms at the promoters of genes6–8. Acetylation of H2A.Z is catalyzed by the histone acetyltransferase TIP-60 subsequent to its inclusion within nucleosomes9 but it remains unexplained how the balance between the acetylated and unmodified forms is inherited and preserved.
The concepts underlying epigenetic inheritance have recently undergone significant change. A number of studies demonstrated that epigenetic marks can cross generation borderlines10–12 but the mechanistic aspects of this intergenerational transmission of information, the so-called ‘epigenetic memory’, remained unclear. In particular, how is the balance between the acetylated and unmodified forms inherited and preserved. Here, we show that cell type-specific RNAs are conduits of the accurate recycling of the epigenetic mark and thereby provide transgenerational design marks for regulation of the transcriptional activity of the respective genomic loci.
Widespread transcriptional activity across the mammalian genome results in the production of many functional noncoding RNAs (ncRNAs)13–18 that play critical roles in multiple biological processes18–29. Such RNAs are essential components of the chromatin architecture30–32 that is shaped by epigenetic marks such as DNA methylation, histone modifications, nucleosome positioning, and the incorporation of histone variants into nucleosomes. Despite their well-documented role, few of these RNAs have been fully characterized and the majority still lack a functional classification.
The overall effect of these ncRNAs is to regulate cell-type specific gene expression at both the transcriptional and post-transcriptional level, including: (i) chromatin remodeling by recruiting complexes that lead to epigenetic changes (e.g., polycomb repressive complex (PRC2)–mediated transcriptional regulation33; (ii) transcriptional interference by directly interacting with ubiquitous or tissue specific transcription factors (e.g., dissociation of the pre-initiation complex through sequestration of the general transcription factor IIB34 or Sp1 in Myotonic Dystrophy cases24; or (iii) splicing interference by affecting the distribution of serine/arginine splicing factors35.
A targeted and global genomic approach is used here to demonstrate that the deposition and modification of the active versus inactive H2A.Z mark is mediated by locally induced cell-cycle specific RNAs, termed SPEARs (S Phase EArly RNAs).
RESULTS
Expression of newly minted promoter RNAs correlates with that of neighboring genes
Since the majority of active genes undergo replication in early S phase, which is followed by formation of the appropriate active chromatin36, we reasoned that transcripts generated in early S phase might control expression of the respective mRNAs by playing a role in reassembly of the chromatin. Furthermore, we hypothesized that the epigenetic balance between acetylated and unmodified forms of H2A.Z could be maintained by locally induced ncRNAs.
To establish the global occurrence of such newly minted RNAs, nascent S-phase transcripts were captured and sequenced (nasRNA-Seq). Synchronized human HL-60 cells were labeled with the ribonucleotide homolog 5-ethynyl uridine (EU) for one hour upon release into S phase. The collected RNAs were then biotinylated by click chemistry, isolated on streptavidin beads and deep-sequenced to produce nasRNA-Seq libraries37,38 (Extended Data Fig. 1a and Methods for details). Analysis of such libraries demonstrated a strong correlation of expression levels between nascent transcripts arising from gene promoter regions (“nascent promoter RNAs”) and those of their linked genes (Fig. 1a). In particular, such correlations are more prominent for nascent transcripts arising from the promoter regions of highly expressed genes (Fig. 1a). Among the early expressing genes marked by the presence of early S-phase promoter RNAs we identified c-MYC (Extended Data Fig. 1b,c), the oncogene most frequently altered in cancers39, myeloid master regulator transcription factor PU.1 (ref. 40) and proto-oncogene MYB (ref. 41) (uninterrupted sequence segments are in Supplementary Data #2); a list of all gene loci expressing early promoter RNAs is in Supplementary Data #1. The c-MYC early S-phase promoter transcripts were shown to be represented by ~13 copies in the nucleus of HL-60 cells (Extended Data Fig. 1d) and were mapped by primer extension and 5’, 3’-RACE (Extended Data Fig. 1e,f).
A fraction of early S-phase promoter RNAs interacts with H2A.Z, acH2A.Z and TIP60
Given the close proximity of promoter RNAs to TSS and the growing body of literature showing that highly expressed genes are marked by the acetylated form of the variant histone H2A.Z (acH2A.Z) (ref. 5), the possibilities of a link between early S-phase promoter RNAs and H2A.Z/acH2A.Z was investigated by: (i) Ribonucleoprotein (RNP) pull-down experiments followed by mass spectrometry; and (ii) RNA immunoprecipitations followed by RNA sequencing (RIP-Seq).
To design the RNA probes for RNP pull-down, the sequences of three early S-phase promoter RNAs, identified by nasRNA-Seq (Supplementary Data #1) and derived from gene loci c-MYC, PU.1, and MYB, were verified by primer extension and 5’, 3’ RACE (Figures S1E and S1F). The approximately 500 nt uninterrupted sequence segments (Supplementary Data #2) were cloned under a T7 RNA polymerase promoter to enable expression of biotinylated sense and antisense early promoter RNAs probes for RNP pull-down experiments (Extended Data Fig. 2a). Recovered RNPs interacting with both the antisense and sense early S-phase promoter RNAs probes and a negative control (D-Biotin) were analyzed by mass spectrometry. Among peptides pulled down by the early promoter RNAs probes were several corresponding to H2A.FZ/ H2A.FV (Supplementary Data #3). It was not possible to distinguish the very homologous forms of H2A.FZ vs. H2A.FV, and no peptides containing acetylation sites were detected, due to the relatively low fragment coverage. Other peptides pulled down by the early promoter RNA probes corresponded to histones H2A, H2B, H3, H4 and H1. No peptides corresponding to H2A/H2A.Z or other histones were detected by the negative control probe (Supplementary Data #3).
It is possible that early S-phase promoter RNAs exert their function through direct interaction with H2A.Z and the histone acetyl-transferase (HAT) TIP60, the enzyme reported to acetylate H2A.Z9,42. To directly test the interaction of early S-phase promoter RNAs with H2A.Z, acH2A.Z, and TIP60, RIP-Seq experiments were performed using antibodies to H2A.Z (i.e. recognizing all forms of the histone), to acetylated H2A.Z (acH2A.Z), and to TIP60 (Extended Data Fig. 2b). A significant overlap of RIP-Seq peaks was observed from all three immunoprecipitations and considerable co-localization of the three data sets was detected in the regions around the mRNA TSSs. In particular, a significant enrichment was observed of both sense and antisense transcripts in the 2 kb window upstream of TSSs (total number of early S-phase nascent promoter RNAs: 4689; 2,411 and 2,278 transcripts on the sense and antisense strands, respectively; Fig. 1c,d). At this stage, all early promoter RNAs which did not have detectable expression above background in the nasRNA-Seq library were discarded, which restricted the list of early S-phase nascent promoter RNAs to 3,560 (1,810 and 1,750 transcripts on the sense and antisense strands, respectively) (Fig. 1e). To assess the functional impact of the interaction of the early S-phase nascent promoter RNAs with TIP60, H2A.Z, and acH2A.Z, the identified 3560 early S-phase nascent promoter RNAs were aligned with the acH2A.Z ChIP-Seq library. To ensure consistency and stringency, only gene loci with a detectable acH2A.Z ChIP signal in the 2 kb upstream window of the TSS were accepted (Fig. 1f). This identified 2363 promoter RNAs (1209 sense and 1154 antisense transcripts) which are expressed in the first hour of S phase and which overlap with the signals of all four data sets: the three RIP-Seq libraries (antibodies to H2A.Z, TIP60, and acH2A.Z) and the acH2A.Z ChIP-Seq library. The co-location of acH2A.Z ChIP-Seq signals in the same genomic region as the signals from the three independent RIP-Seq experiments suggests a functional role of the early S-phase nascent promoter RNAs in positioning the acetylated H2A.Z marks. We termed these 2,363 early S-phase nascent promoter RNAs SPEARs (S Phase EArly RNAs). Comparison of the H2A.Z and acH2A.Z ChIP-Seq data sets implies that it is the acetylated form of H2A.Z that is largely bound to the SPEARs (Fig. 1g). Taken together, the histone immuno-precipitations and the RIP experiments indicate: i) a triple interaction of SPEARs with H2A.Z, acH2A.Z, and with TIP60 in the 1 kb upstream of the TSS (Extended Data Fig. 2c); and ii) a correlation of SPEARs expression level with that of the linked genes.
In summary, identification of 2,363 SPEARs linked to expressed genes, direct correlation of SPEARs expression levels with those of the corresponding genes, and, importantly, with the occupancy of acH2A.Z within the respective loci (Fig. 1f-h), implies a global involvement of SPEARs in cooperation with TIP60 and H2A.Z/acH2A.Z in establishing an active expression mode at the corresponding genes. This warranted further examination.
SPEARs carry common binding motifs
To further test the SPEARs-H2A.Z/TIP60 relationship, we performed motif discovery analysis on the SPEARs (see Methods for details). Among the twenty-five identified motif candidates, three motifs (##3, 5, and 9; Supplementary Data #4) exhibited the most significant enrichment among the SPEARs-regulated gene loci. Importantly, no similar motifs were found within the largest classes of nuclear RNAs, i.e. transcripts arising from SINEs and LINEs, nor from ribosomal RNA genes (see Methods for details). Among the top three candidates, motif 9 (corresponding to the RNA oligonucleotide RM9A; Fig. 2a and Extended Data Fig. 3a) was ranked as the strongest motif enriched in the c-MYC SPEARs sequence. This motif was enriched in acH2A.Z and TIP60 RIP-Seq overlapping peaks (13%), located near the peak centers (Fig. 2b).
RNA electrophoresis mobility shift assays (REMSAs) using the RM9A motif demonstrated that it is able to form RNPs with H2A.Z and with TIP60 in vitro (Fig. 2c, lane 3; and Fig.2d, lane 4). A shift in migration was also observed by incubation of RM9A with synthetic peptides corresponding to the N-terminal sequences of H2A.Z, both unmodified and acetylated at lysine 7 (K7) (Fig. 2c, right panel). Acetylation of the H2A.Z tail peptide did not abrogate the binding (Fig. 2c, lanes12, 13 – acetylated peptide, and lanes 8, 9 – not acetylated peptide), indicating that simple charge-charge interactions are not responsible for SPEARs-H2A.Z or SPEARs-acH2A.Z complexes. Interestingly, the presence of a predicted stem-loop-like structure (RNAfold43,44; Fig. 2e) seems to be required for RM9A binding to H2A.Z and TIP60 (Fig. 2c,d), but unlike the previously identified DiRs (DNMT1-interacting RNAs)45, the secondary structures are not the only requirement for RNP formation. Indeed, mutation of RM9A so as to prevent the predicted stem-loop structure reduced but did not totally abrogate the formation of the RNA-TIP60 complexes (Fig. 2d, lanes 4-5). Importantly, unrelated oligonucleotides (UR2 and UR3) lacking the common binding motif did not form comparable RNP complexes with TIP60/H2A.Z (Extended Data Fig. 3), indicating the involvement of both primary and secondary structure elements in the recognition of TIP60/H2A.Z by RM9A.
RNAs might be important for binding of remodeling complexes to chromatin, as found for the polycomb repressive complex 2 (PRC2) and Tip60–p40013,46–48. To test if SPEARs facilitate binding of TIP60 to the chromatin, we compared TIP60 binding to the RNA (single-stranded RM9A) and DNA (double-stranded DM9A) having the same primary sequence. The REMSA/EMSA gels indicate very strong TIP60 binding to the ssRM9A and weaker binding to the dsDM9A (Fig. 2d, lane 4 vs. lane 6).
In conclusion, a specific interaction is seen between the RM9A region of the c-MYC SPEAR and the histone acetyltransferase TIP60 and histone H2A.Z, in particular its modified form acH2A.Z.
SPEARs are involved in H2A.Z acetylation and exchange
To directly test if SPEARs are required for the deposition of acH2A.Z at the TSS of their corresponding gene, transcription was pharmacologically modified and followed by chromatin immunoprecipitation/sequencing (ChIP-PCR and ChIP-Seq) using H2A.Z and acH2A.Z antibodies (outlined on Extended Data Fig. 4a). Two transcription inhibitors, Actinomycin D (ActD; a RNA Polymerase I, II, and III inhibitor) and 5,6-Dichlorobenzimidazole 1-β-D-ribofuranoside (DRB; a RNA Polymerase II Inhibitor), were used at concentrations sufficient to interfere with the transcripts produced by both RNA Polymerases II and III. The rationale behind the use of ActD and DRB is: (i) to globally assess the effect of changes in SPEARs expression levels over a short time frame; (ii) to compare effects resulting from the two different pathways triggered by ActD and DRB; and (iii) to take advantage of the reversibility of DRB treatment. Cells were synchronized with a double thymidine block and released into S phase with ActD/DRB added (Extended Data Fig. 4a), treated with the drugs for two hours, an interval during which the overall levels of H2A.Z and TIP60 proteins were not affected (Extended Data Fig. 4b), while the global RNA levels of both the coding genes and the SPEARs were significantly altered (Fig. 3a). Relative changes were then investigated in the distributions of unmodified H2A.Z and acH2A.Z in the vicinity of the TSS of the SPEARs-linked genes, in response to DRB- and ActD-induced changes in SPEARs expression levels. The enrichments of ChIP-Seq signals for H2A.Z and acH2A.Z in the region adjacent to the TSSs of the 2363 SPEARs-linked genes following drug treatment were compared with their occupancy in mock-treated (DMSO) cells. We observed substantial reductions in acH2A.Z enrichment at the TSS of samples treated with either DRB or ActD (Fig. 3b, right panels), in contrast to only modest changes in the occupancy levels of unmodified H2A.Z (Fig. 3b, left panels). In cells treated with DRB, 26% of identified SPEARs loci showed a drop in acH2A.Z occupancy, while only 6% of such loci showed a similar trend in the H2A.Z signal (Fig. 3b, upper panels). ActD-treated cells showed a more pronounced effect, with 42% of SPEARs loci showing reduced acH2A.Z signal as compared with 4% of such loci showing decreased occupancy of H2A.Z (Fig. 3b, bottom panels). Thus, Fig. 3b demonstrates that loci with suppressed expression of SPEARs showed diminished enrichment in acH2A.Z, implying that SPEARs are involved in the precise placement of this epigenetic mark. The observation of a 10-fold greater number of acH2A.Z-affected TSS in drug-treated samples as compared to H2A.Z, indicates the involvement of SPEARs in the maintenance of normal levels of H2A.Z acetylation at TSS, i.e. to the generation of the active epigenetic mark acH2A.Z. The upper panel of Fig. 3c depicts snapshots of the c-MYC locus for which the SPEARs are repressed by ActD and DRB treatment (results of qRT PCR are shown in Extended Data Fig. 4c, left panel), resulting in a decreased intensity of acH2A.Z peaks. In detail, for the c-MYC the ratios of control (DMSO) to drug treated acH2A.Z peak intensities were [DMSO/DRB/ActD]c-MYC = 1/0.45/0.39. These ChIP-Seq data were confirmed by quantitative ChIP-qPCR for several amplicons covering the c-MYC locus (Extended Data Fig. 4d). By contrast, loci escaping the down-regulating effects of the drugs on their SPEARs (the “DRB paradox”49) exhibit an acH2A.Z intensity ratio in simple correlation with the change in SPEARs levels. For example, the MYB locus gives rise to SPEARs positively affected by ActD and DRB (Extended Data Fig. 4c, right panel), and exhibits an increase in the intensity of the acH2A.Z peaks, with no significant changes in the intensity of the unmodified H2A.Z peaks (Fig. 3c, bottom panel). For MYB, the ratios of control (DMSO) to drug treated acH2A.Z peak intensities are [DMSO/DRB/ActD]MYB =1/1.52/1.60, which correlates with the increases in its SPEARs levels (Extended Data Fig. 4c, right panel). Thus, loci at which expression of SPEARs are positively affected by ActD/DRB illustrate how an increase in SPEARs levels leads to a corresponding rise in the intensity of the acH2A.Z peaks.
Nascent chromatin immunoprecipitation (nasChIP-PCR50–52) is particularly useful for tracking histone deposition and was therefore used with the antibodies to H2A.Z, acH2A.Z and TIP60. Cells were synchronized with a double thymidine block, released into S phase and treated with DRB for 2 hours. The medium was supplemented with the EdU DNA analog to enable collection of nascent chromatin DNA (Extended Data Fig. 4a). It was observed that when transcription was inhibited with DRB, the c-MYC locus showed diminished enrichment in H2A.Z, acH2A.Z and TIP60 (Extended Data Fig. 4e, three upper panels). This demonstrates a link between suppressed c-MYC SPEARs expression and reduced levels of the replacement histone H2A.Z on the nascent chromatin and an even greater reduction of the acetylated acH2A.Z at the TSS of the c-MYC locus. Similar results were obtained for the PU.1 gene locus (Extended Data Fig. 4e, three bottom panels).
Taken together, the negative effects of ActD/DRB – a global “loss-of-function”, exemplified by c-MYC, – and their positive effects, a global “gain-of-function”, exemplified by MYB –monitored by ChIP-Seq and nasChIP-PCR experiments, define the role of SPEARs transcripts in deposition of the replacement histone H2A.Z. The more robust drop in the acetylation level of H2A.Z as compared to the more modest decrease in the unmodified form, as well as loss of TIP60 at the same sites close to the TSSs of the genes, all point to the role of SPEARs transcription in mediating the acetylation of H2A.Z.
In complementary gene-specific experiments, we then tested whether RNAi-mediated downregulation and dCas9-VP64-mediated upregulation of specific SPEARs followed by ChIP-Seq and ChIP-qPCR analyses leads to a corresponding reduction or increase of H2A.Z and acH2A.Z levels at the TSS of targeted loci and in reduced or increased expression of the corresponding gene. Knockdown of the c-MYC SPEARs (~75%; Fig. 3d, left panel) was associated with a significant decrease of acH2A.Z levels at the TSS of the c-MYC gene (Fig. 3d, middle panel), followed by the corresponding decrease of c-MYC mRNA expression of similar magnitude (~70%; Fig. 3d, right panel). The results of the RNAi/ChIP knockdown experiments demonstrate that reduced expression of the c-MYC SPEARs is linked to lower levels of H2A.Z acetylation at the c-MYC TSS, which accords with the initial hypothesis of direct causality between the expression of SPEARs and the deposition of the activating histone mark, leading to transcriptional activation of the adjacent gene.
Additionally, we performed a “gain-of-function” test using the dCas9-VP64 gene activation system (outlined on Extended Data Fig. 4f). Increasing levels of c-MYC SPEARs expression (Fig. 3e, left panel) were associated with a significant increase of acH2A.Z levels at the TSS of the c-MYC gene (Fig. 3e, middle panel) accompanied by a ~2-fold increase in c-MYC gene expression (Fig. 3e, right panel).
Overall, downregulation of SPEARs using independent pharmacological and RNAi-induced methods leads to a substantial reduction in the activating epigenetic mark acH2A.Z. Loss of the acetylated form of H2A.Z from chromatin in the vicinity of active TSS implies a reduction of its activity, i.e. reduced expression of the corresponding gene (Fig. 3d, right panel). On the other hand, upregulation of SPEARs leads to a substantial induction of the activating epigenetic mark acH2A.Z and, consequently, to elevated expression of the corresponding gene (Fig. 3e, right panel).
SPEARs regulate the expression of the corresponding mRNA via TIP60 recruitment and acH2A.Z deposition
To demonstrate that the SPEARs-mediated regulation of their corresponding mRNA is executed through the TIP60/acH2A.Z pathway, the effects of TIP60 inhibition on the c-MYC locus were tested. Two TIP60/HAT inhibitors, TH183453 and MG-14954, were used in HL-60 cells to reduce the total level of H2A.Z acetylation (Extended Data Fig. 5a). This led to significant downregulation of mRNAs, in contrast to almost unchanged levels of SPEARs transcripts (Fig. 4a). To examine the direct regulation of c-MYC expression by TIP60 and acH2A.Z interacting with the SPEARs, we monitored the regulating effect of nascent SPEARs on the level of nascent mRNA expression in the presence or absence of the two TIP60 inhibitors, taking advantage of the reversibility of the DRB transcriptional inhibitor (ref. 55. The rationale was to analyze only transcripts re-appearing after release from the transcriptional block whilst the HAT is inhibited. The short time of inhibitor treatment should rule out secondary effects of HAT inhibition.
Cells were synchronized with a double thymidine block and released into S Phase with DMSO/DRB added. After 2 hours (during which H2A.Z levels remained unaffected, but SPEARs and mRNA levels drop (see Extended Data Fig. 5a,b), cells were washed and then incubated for a further 2 hours with or without the TIP60 inhibitors. This interval is sufficient for the levels of both the c-MYC mRNAs and its SPEARs to be restored to the levels when the TIP60 inhibitors are absent (results shown in Extended Data Fig. 5b). The medium was supplemented with the EU RNA and EdU DNA analogs to enable collection of nascent RNAs and newly formed chromatin that had escaped the drug-induced inhibition of transcription or acetylation. After 2 hours with the TIP60 inhibitors present (MG-149 at 200 μM and TH1834 at 500 μM), during which time the levels of acH2A.Z dropped (see Extended Data Fig. 5a), cells were crosslinked and subjected to nasChIP-PCR and nascent RNA expression analyses (nas-qRT-PCR) (outlined on Fig. 4b). Collected RNAs were biotinylated by click chemistry, isolated on streptavidin beads and analyzed by nas-qRT-PCR (see Methods for details and Extended Data Fig. 1a, bottom panel). The results in Fig. 4c show that the correlation between expression of the SPEARs and the mRNA no longer holds when TIP60 activity is inhibited, i.e. the restored levels of the SPEARs are incapable of rescuing the expression of c-MYC mRNA to the level defined by the reversal of DRB repression. These data suggest that the function of SPEARs in regulating c-MYC mRNA expression is impeded by inhibition of the HAT activity of TIP60.
To further check that SPEARs realize their c-MYC regulatory function directly through the TIP60/acH2A.Z pathway, we assessed the differences in chromatin occupancy of H2A.Z, acH2A.Z and TIP60 after inhibition of transcription (“DRB”), reversal of inhibition (“DRB REV”), and reversal of inhibition in the presence of the two TIP60 inhibitors, using nasChIP-qPCR. In particular, it was important to check whether the drop in c-MYC expression after treating the cells with DRB and TIP60 inhibitors is correlated with diminished enrichment in TIP60, acH2A.Z, or H2A.Z. Nascent DNA was isolated from the chromatin immuno-precipitated with antibodies to H2A.Z, acH2A.Z, or TIP60, biotinylated by click chemistry, isolated on streptavidin beads (Methods) and finally analyzed by qPCR at amplicons corresponding to maximum enrichment within the c-MYC locus (see Fig. 3e). By analyzing the nascent RNAs and chromatin, we examined only the immediate changes in c-MYC expression and TIP60, acH2A.Z, and H2A.Z occupancy imposed by the inhibition of transcription and/or the reduced acetylation of histones. The middle and right panels of Fig. 4d demonstrate that the c-MYC locus with suppressed expression of its SPEARs shows diminished enrichment of both acH2A.Z and TIP60, and the reversal of the transcription inhibition led to the restoration of both enrichments. This indicates that the SPEARs are involved in the recruitment of TIP60 and in the proper placement of the acetylation mark. When the two HAT inhibitors are present, the reversal of transcriptional inhibition can only restore the enrichment levels of unmodified H2A.Z and of TIP60 (Fig. 4d, left and right panels) but not the levels of acH2A.Z (Fig. 4d, middle panel). In contrast, no enrichments of H2A.Z, acH2A.Z, or TIP60 were detected within the Gene Desert locus56 (Extended Data Fig. 5c). Given that local levels (i.e. in the vicinity of the c-MYC locus TSS; Fig. 4d, left panel) as well as overall levels of H2A.Z were unchanged throughout the experiment (Extended Data Fig. 5a), these results establish the leading role of the TIP60/acH2A.Z pathway in c-MYC SPEARs-mediated regulation of its corresponding gene, c-MYC. Similar results were obtained for the PU.1 locus (Extended Data Fig. 5d,e) and with another HAT inhibitor – NU 9056 (which at high concentration inhibits acetylation of H2A.Z) (Extended Data Fig. 5f-h). As with HAT inhibitors MG149 and TH1834 (Fig. 4d, middle panel), we observed loss of promoter occupancy by acH2A.Z after treatment with the HAT inhibitor NU 9056 in cells pre-treated with DRB (“DRB/NU 9056”) (Extended Data Fig. 5h).
Collectively our results suggest that the TIP60/acH2A.Z pathway is a general mechanism for SPEARs-mediated regulation of the corresponding adjacent coding genes.
SPEARs regulate recycling of the epigenetic mark - acH2A.Z
To demonstrate the role of SPEARs in the recycling of the acH2A.Z mark, we performed the experiments outlined in Fig. 5a. Cells were cultured for six days in medium supplemented with “heavy” 15N213C2-Lysine and 13C6-D-glucose and then synchronized with a double thymidine block. After washing out “heavy” lysine and glucose, cells were released into S phase in the presence of HAT inhibitor or DMSO (mock treatment). The rationale of the experiment was to distinguish histones deposited in the chromatin before entering S phase from those deposited after this point. Thus, histones containing “heavy” lysines and their acetylated forms carrying “heavy” acetyl groups are “old”, whilst “new” histones and their acetylated forms generated during early S phase should contain “light” lysines and “light” acetyl groups. The medium was supplemented with the EdU DNA analogs to enable collection of newly formed (nascent) chromatin. The HAT inhibitor NU 9056 was added at the start of S-phase and the efficiency of inhibition was demonstrated by HAT assays on the collected cells (Extended Data Fig. 6a) showing efficient suppression of H2A.Z acetylation. This experiment enabled investigation of whether “old” histones with “old” modifications are preserved and transmitted through the generational border. At 4 hours after release from synchrony, cells were collected and subjected to gene expression and Western Blot analyses, HAT assay, ChIP-PCR, nasChIP-PCR, and mass spectrometry. As with TH1834 and MG148 inhibition (Fig. 5a,b), treatment with NU 9056 resulted in a drop in mRNA but not in SPEARs expression (Fig. 5b) and levels of H2A.Z remained unaffected but acH2A.Z levels dropped (Fig. 5c).
To verify that the c-MYC locus indeed undergoes replication and chromatin re-assembly in the early part of S phase, nasChIP-PCR was performed (with antibodies to acH2A.Z) and a significant enrichment of acH2A.Z-occupied nascent c-MYC locus chromatin was observed (Fig. 5d).
The tests of experimental conditions being satisfactory (Extended Data Fig. 6a and Supplementary Data #5), ChIP-PCR analyses of acH2A.Z occupancy at the target promoters was performed. Strikingly, analysis of promoter occupancy by acH2A.Z in HL-60 cells after even longer treatment in NU 9056 showed levels of acH2A.Z to be diminished by only one half in comparison to the mock treatment (Fig. 5e, right panel). Two points need emphasis: (i) when acetylation of all newly synthesized histones is strongly inhibited, persistent promoter occupancy by the acetylated forms of H2A.Z can occur only through the recycling of the previously acetylated (“old”) histones; and (ii) this recycling process is interrupted by the inhibition of transcription, particularly with the abrogation of SPEARs expression (see Fig. 4 and 5).
To unequivocally demonstrate that the “recycling” of acetylated forms of histones indeed represents “old” histones, we performed mass spectrometry analyses of the immunoprecipitates from cells subjected to HAT inhibitor for 4 hours” (Fig. 5a), in which all “new” acetylation was suppressed. The samples were digested with chymotrypsin and analyzed on an Orbitrap Elite mass spectrometer. The Proteome Discoverer 2.4 search engine analyzed the raw data. Searches were performed against the H2A.Z protein sequence and common contaminants and all results filtered to 1 % FDR (false discover rate). The analysis focused on the N-terminal H2A.Z peptide Ac-AGGKAGKDSGKAKTK (Extended Data Fig. 6b) with variable acetylation of lysines K4, K7, and K11 and showed a high level of “heavy” lysines and the presence of heavy acetyl groups in this N-terminal peptide. Analysis of isotopic distributions showed that among all N-terminal peptides carrying two acetyl groups, the content of heavy lysines was ~70% and of heavy acetyl groups was ~20%. In peptides carrying three acetyl groups the content of heavy lysines was ~55% and of heavy acetyl groups ~30%. Extended Data Fig. 6c shows examples of the peptides detected. Fig. 5f demonstrates an increase of heavy acetyl groups in the “HATs inhibitor 4 hours” sample (Left panel; ~70%) as compared to their natural distribution (Right panel; ~38%). These analyses show that some old H2A.Z histones carrying heavy acetyl groups are preserved in the early stages of S phase and that recycling of old acH2A.Z is affected by brief suppression of transcription in the first hours of S phase.
Discussion
The roles of ncRNAs in regulating deposition and recycling of acH2A.Z at gene promoters has been investigated. Transcription of SPEARs that control the acetylation of nearby H2A.Z could be regarded as a mechanism whereby information encoded in the genome becomes expressed as ‘epigenetic’ information. Locally induced SPEARs bind to the replacement histone H2A.Z and to a nuclear factor, the histone acetyl transferase TIP60, leading to deposition/acetylation of H2A.Z, a process that represents recycling of the mark across generations. SPEARs transcripts guide deposition/acetylation of this replacement histone through RNP formation with TIP60/H2AZ. The function of these transcripts depends on a combination of their expression levels, primary and secondary structures, and physical co-compartmentalization of RNAs, their parental loci, and their protein partners. SPEARs databases can provide a valuable resource for studying the dynamics of RNP assembly and their role in the formation of active/inactive chromatin. This mechanistic model is depicted in Figure 6.
The demonstrated link between SPEARs and the activated form of the replacement histone H2A.Z raises the possibility that other long ncRNAs (lncRNAs) play a role in the recruitment of chromatin modifiers that deposit activating or repressive epigenetic marks. Although more detail is still needed, such lncRNAs may act not only as signaling molecules but also as the scaffolds around which chromatin writer/eraser complexes are assembled. This notion is supported by an observation in Drosophila, in which the deposition of the replacement histone H3.3 occurs within transcriptionally active loci and is not replication-coupled57.
In summary, the discovery of SPEARs reveals a novel mechanism for the biosynthesis and maintenance of a major epigenetic mark. Taken together with the study of identified DiRs45, the present investigations demonstrate that epigenetics and genetics are intrinsically linked, and such lncRNAs modulate distinct epigenetic marks by using site-specific targeting molecules. Indeed, epigenetic modulating agents are already employed in the clinic58,59, but assessment of their function is difficult, their efficacy being compromised by lack of specificity. This research provides a new tool to overcome these drawbacks by using the specificity of RNA molecules to drive deposition and selective modification of histones. This study indicates the need for further investigations uncovering how epigenetic information is encoded in the genome and regulated under normal conditions and in disease states.
METHODS
Detailed methods are provided in the online version of this paper and include the following:
- KEY RESOURCES
- CONTACT FOR REAGENT AND RESOURCE SHARING
- EXPERIMENTAL MODEL
Mammalian Cell Culture
- METHOD DETAILS
RNA isolation;
qRT-PCR;
Nascent RNA/DNA capture;
Primer extension and 5’/3’ RACE;
Double Thymidine block (early S-phase block);
DRB and Actinomycin D treatments;
Down-regulation of c-MYC SPEARs;
Ribonucleoprotein (RNP) fractionation;
Western Blotting Analysis;
Tandem Mass Spectrometry (LC-MS/MS);
Nuclear Chromatin (ChIP) and RNA immunoprecipitation (nRIP);
RNA electrophoretic gel mobility shift assays (REMSAs).
- QUANTIFICATION AND STATISTICAL ANALYSIS
ChIP- Sequencing Analyses;
Comparison of the enrichment of H2A.Z and acH2A.Z ChIP-Seq signal surrounding gene TSS loci (+/- 2kb);
RIP-Sequencing Analyses;
RNA-Sequencing Analyses;
Motif Discovery;
LC-MS/MS Analyses.
- DATA AND SOFTWARE AVAILABILITY
Sequencing Data are available on the gene omnibus database under the accession ID number: GSE165526; enter token: slehymyczncvfqd. List of all data sets is in Supplementary Data #6.
Mass Spectrometry data will be deposited to the ProteomeXChange Consortium via PRIDE upon acceptance of the manuscript. Private partial submission with MASSIVE: https://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=38a5eae58dd247f5b3c42f2be181e692; Password: Alex_Acetyl
Author Contributions
AKE, ADR, CCR and DGT conceived and designed the study and wrote the manuscript; AKE, SU, HZ, ADR, LY, DT, SU, EM, AW, BQT, RC, MRS, HZ, YZ performed experiments; MAB performed bioinformatics and statistical analyses, and analyzed RNA-Seq and RIP-Seq data; VEA, RTM, MAB and TB analyzed ChIP-Seq data, EM and VP performed the RNA motif discovery, BB and DK analyzed Mass Spectrometry datasets.
Acknowledgements
This work was supported by the award of NCI R35 CA197697, R01DK103858, W81XWH-15-1-0161, and P01HL131477-01A1 to DGT; and R50 CA211304 to AKE; ADR was supported by NCI R00 CA188595, W81XWH-20-1-0518, the Italian Association for Cancer Research (AIRC) and the Giovanni Armenise-Harvard Foundation; AW was supported by the Belgian American Educational Foundation (BAEF) and by Wallonie-Bruxelles International (WBI); K01CA222707 was awarded to BQT; TB was supported by the National Research Foundation and the Singapore Ministry of Education under its Centres of Excellence initiative. This work was also supported by the Singapore Ministry of Health’s National Medical Research Council under its Singapore Translational Research (STaR) Investigator Award; by the National Research Foundation Singapore and the Singapore Ministry of Education under its Research Centres of Excellence initiative; and by the Singapore Ministry of Education Academic Research Fund Tier 3, grant number MOE2014-T3-1-006. We thank Bruno Amati for thorough discussion and insightful comments on the manuscript and for providing the TIP60 antibodies; likewise Pier Paolo Pandolfi for critical reading of the manuscript and all the members of the Tenen Laboratory for the helpful suggestions.