Dynamic bookmarking of primary response genes by p300 and RNA polymerase II complexes

p300 and pol II Coassemble at Rapidly Expressed Genes.

Prior studies from our group showed that p300 associates rapidly with the promoters of several genes in Jurkat human T cells following mitogen stimulation with phorbol ester and ionomycin (P/I) (5, 6). We expanded this observation using ChIP-chip strategies and qChIP to profile genome-wide changes in promoter occupancy of p300 and pol II 45 min following mitogen stimulation (Fig. 1 A and B). p300 and pol II ChIP DNA from resting and P/I-treated cells was amplified, labeled, and hybridized to a proximal promoter array representing 37,364 promoters spanning −1,000 to +500 bp relative to the transcription start site (TSS). Promoters showing significant occupancy (−log₁₀ P ≥ 6.8) by either p300 or pol II were identified. As shown in a graphic representation of these binding data (Fig. 1A), p300 undergoes a dramatic redistribution across the genome following mitogen stimulation. By hierarchical clustering, this redistribution correlates with dynamic changes in the patterns of both pol II and p300 promoter co-occupancy (Fig. 1B).

Five major classes of promoter co-occupancy [supporting information (SI) Table S1A] were identified: (i) genes that show significant p300 occupancy in resting cells (p300-enriched, total = 2,188); (ii) genes that show p300 occupancy that increases in significance following mitogen induction (p300 inducibly enriched, total = 1,837); (iii) genes that show significant pol II occupancy in resting cells (pol II-enriched, total = 3,031); (iv) genes that show pol II occupancy that increases in significance following mitogen induction (pol II inducibly enriched, total = 2,142); and (v) genes that show both p300 and pol II occupancy that increases in significance following mitogen stimulation (p300 and pol II inducibly enriched, total = 1,057).

To define a gene set that had the strongest correlation between p300/pol II assembly and gene expression, we determined how many genes (744 annotated of the total 1,057) also showed rapidly increased RNA expression. Gene expression analysis of Jurkat T cells following mitogen (P/I) stimulation identified 151 annotated genes that increase significantly (P ≤ 0.05) following 1 h of mitogen stimulation (Fig. 1C and Fig. S1). Twenty-seven of these genes are among the 744 annotated genes that rapidly coassembled p300 and pol II (Fig. 1C, Table S1B, Fig. S1 and S2). Interestingly, many of these rapidly expressed genes are classic primary response or “immediate early” genes, including FOS, JUN, and EGR2, suggesting a prominent role for p300 in rapid genetic responses (Fig. 1D, Table S1B, Fig. S1 and S2). Members of this class of 27 rapidly induced p300 targets were therefore selected for further study.

Bookmarking of Immediate Early Genes by p300/pol II Complexes.

A common feature of many primary response genes, including FOS, is their rapid yet transient expression or “spike” kinetics following mitogen induction (7, 8). This is readily demonstrated in the 0–8 h RNA expression profiles of several p300 and pol II targeted genes following mitogen stimulation (Fig. 2A). Prior observation of the kinetics of mitogen-stimulated p300 assembly at rapidly induced genes using semiquantitative PCR showed dynamic changes as the promoter evolved over 15-min intervals within 1 h following mitogen activation, where p300 appears to leave the template shortly after transcriptional stimulation in some cases (6). To assess these dynamic changes over a longer time frame and at higher resolution, similar but more sensitive measurements were done using ChIP followed by quantitative real-time PCR. FOS was chosen as the prototype gene for study, and primers were designed to detect both mature (spliced) and nascent (unspliced) FOS RNA transcript profiles by qRT-PCR. For comparison, factor occupancy was profiled at locations upstream, near the promoter, and through the body of the gene (Fig. 2B). This allowed us to profile the in vivo coupling of active transcription with specific factor assembly events at the FOS locus. Unexpectedly, although ChIP analysis showed rapid recruitment of pol II to the promoter regions (−0.2 kB) within 30 min following mitogen stimulation, this assembly persisted for up to 8 h. In contrast, pol II transcriptional elongation, as reflected by its appearance at the 3′ end of the gene (+3.0 kB), was very transient, with a kinetic profile similar to that of mature and nascent FOS transcript production (Fig. 2 A and B and Fig. S3A). A kinetic overlay of factor assembly and transcription at the FOS locus (Fig. 2C) illustrates that although pol II assembly and elongation occur rapidly, this phase of transcription attenuates quickly within 2 h, leaving a large population of cells with polymerase accumulated at the proximal promoter in a potentiated or bookmarked state, possibly “primed” for response to subsequent signaling events (Fig. 2 B and C).

One major role of p300 is to increase the DNA accessibility for pol II and other factors by histone acetylation (3). To correlate changes in FOS expression with histone acetylation, the time- and location-dependent changes in the appearance of acetylated histone H3 (K9, K14) and histone H4 (K5, K8, K12, and K16) were profiled at the FOS gene. Although both histone H3 and histone H4 acetylation were rapidly induced at the proximal promoter following mitogen stimulation, these levels remained high for 8 h (Fig. 2 D and E). In a similar manner to pol II (Fig. 2 B and C), we found that the population of resting cells containing proximally accumulated p300 increases dramatically and persists for several hours following stimulation (Fig. S3B and Fig. 2F). Similar bookmarking by p300 and pol II is observed at the proximal promoters of EGR2 and NFKBIA (Fig. S4 A and B).

p300 Is Required for Immediate Early Gene Bookmarking by pol II and Transcriptional Induction.

Several in vitro protein interaction studies have implicated a role for p300 in the recruitment of pol II and the basal machinery to form the preinitiation complex (PIC) at target genes (3). To define the requirement for p300 in the placement and recruitment of a proximally positioned PIC during immediate early gene expression, we used different gene depletion strategies (Fig. 3 A–D and Fig. S4C). As shown in Fig. 3A, depletion of either p300 or CBP by RNAi results in loss of basal FOS gene expression in resting cells. Moreover, clonal cell lines in which endogenous p300 expression has been deleted (9) show significantly reduced immediate early gene expression (Fig. 3B) that is reduced further on CBP depletion by RNAi (Fig. S4C). Finally, cells depleted of p300 show both a dramatic reduction in the level of promoter proximal polymerase and transcriptional responsiveness of the immediate early genes FOS, EGR2, and NFKBIA against a broad array of extracellular stimuli known to activate RAS/MAP kinase and protein kinase C pathways (10) (Fig. 3 C and D and Fig. S4D). These agents include anisomycin, okadaic acid, Trichostatin A (TSA), EGF, insulin-like growth factor 1 (IGF-1), cycloheximide, and sodium butyrate (Fig. 3D).

Dynamically Bookmarked Pol II Can Be “Restarted” by Sequential Restimulation.

The use of the term dynamic bookmarking implies a transiently “poised” or “potentiated” state that could facilitate a secondary response to subsequent stimuli that, alone, may have little effect in the absence of preconditioning. To test this idea, P/I-stimulated Jurkat T cells were treated, once the initial spike of transcription had subsided, at 2 h with TSA, a histone deacetylase inhibitor that produces minimal stimulation when added to cells that have not been preconditioned by prior mitogen stimulation with P/I (Figs. 3D and 4B). As shown in Fig. 4, the transcriptionally arrested polymerase can be reinduced to elongate (Fig. 4A) and transcription can be returned to maximal mitogen-induced levels (Fig. 4 B and C) by sequential addition of TSA at 2 h after a mitogen stimulus, whereas TSA added alone to untreated cells has an undetectable influence on transcription (Fig. 4B, open symbols).

Similar to the profile of FOS expression, there is a rapid spike of transcription initiation-associated pol II C-terminal domain (CTD) Ser-5 phosphorylation (11, 12) that peaks at the FOS proximal promoter within 1 h after stimulation (Fig. 4D). The transient nature of this peak indicates that the polymerase, accumulated at the proximal promoter following elongation arrest, is relatively hypophosphorylated. As expected, this modification of pol II is transiently reinduced after the addition of TSA (Fig. 4E). Consistent with this observation, the proximal promoter recruitment profile of the TFIIH complex (containing the CDK7 kinase responsible for CTD Ser-5 phosphorylation) correlates well with the profile of CTD Ser-5 phosphorylation (Fig. S5 A and B).

The transient nature of stimulated and restimulated transcription implies that the assembly and activity of factors that control elongation should be equally transient. To interrogate this possibility, we focused on elongation factors that have been previously implicated in the control of immediate early gene expression. One factor is ELL, which had been initially identified as a factor that associates with pol II to increase its elongation efficiency (13, 14). Earlier studies showed that ELL is required for FOS-dependent transformation, and recent studies indicate that ELL may be recruited to stalled polymerases at heat shock genes in Drosophila (15, 16). Another factor is the positive transcriptional elongation factor (P-TEFb), which is composed of cyclin T1 or T2 and CDK9 active subunits and is responsible for the Ser-2 phosphorylation of the polymerase CTD that promotes elongation and other subsequent steps in the transcription cycle (17). Yet another factor is the bromodomain-containing protein (BRD4), which has been shown to associate with P-TEFb and possibly assist in targeting it to active genes (18, 19). ChIP analysis of all 3 factors shows that they are rapidly recruited to the FOS locus following mitogen stimulation. ELL shows higher enrichment at the 3′ end, which is consistent with its association with the elongating polymerase. BRD4 shows significantly higher association with the 5′ end of the gene, suggesting that it does not travel with the polymerase. P-TEFb shows rapid recruitment and significant enrichment at the 3′ end of FOS, indicating that it travels with the elongating polymerase. Notably, all 3 factors show rapid return to the FOS loci following the addition of TSA (Fig. 5). Consistent with the dynamic recruitment of P-TEFb, Ser-2 phosphorylation of the pol II CTD also shows transient 3′ enrichment and return following TSA treatment (Fig. 5). Interestingly, this striking reactivation of transcription does not appear to be attributable directly to HDAC inhibition, because multiple other agents, distinct from HDAC inhibitors, including anisomycin and okadaic acid, also reactivate transcription to levels that are significantly higher in mitogen-preconditioned cells compared with cells treated in the absence of prior stimulation (Fig. S5C).

Dynamic Bookmarking in Primary Human T Cells and Memory T-Cell Subsets.

The concept of dynamic bookmarking raises the possibility that this process may play a more general role in biological systems, where preconditioning by environment stimuli is an essential component of the physiological response. The process of T-cell activation is a robust model system for the study of potential mechanisms of transcriptional memory, and examples of preconditioned transcriptional status have been reported in prior studies (20, 21). To pursue this hypothesis, we assessed whether dynamic bookmarking could be observed in primary T cells (Fig. 6 A and B). P/I stimulation of primary human CD3 cells induces FOS expression with spike kinetics nearly identical to those observed in Jurkat T cells (Fig. 6A). In contrast, profiles of the pol II distribution at the FOS locus show persistent accumulation of pol II at the proximal promoter, which is consistent with dynamic bookmarking (Fig. 6B). This memory-like activity of preconditioned cells has been compared by analogy with true immune memory within the T-lymphocyte compartment (20). True T-cell memory is a process in which naive T cells develop into populations of memory cells capable of more rapid “recall” or enhanced response (22) and is an example of how preconditioning events play a physiological role. Several recent studies aimed at defining genetic signatures specific for memory T-cell function have identified multiple different primary and secondary response genes, including classic immediate early genes like FOS, that show higher expression in resting memory compared with unstimulated naive T cells (23–25). A reasonable speculation would be that prior conditioning by antigen exposure may result in bookmarking of genes like FOS by pol II. As predicted, resting populations of both human donor memory CD4 and CD8 T cells show higher levels of both pol II and p300 at the FOS proximal promoter than their naive controls (Fig. 6 C and D), and basal FOS is significantly higher in memory compared with naive CD4 and CD8 T cells (Fig. 6E). These observations clearly suggest a role for dynamic bookmarking in selective genomic “priming” for specialized responses.

Cell Culture and ChIP.

Complete methods for cell culture propagation; cell sorting; and transfection of Jurkat, HEK293, and HCT116 cells are provided in SI Materials and Methods. Cells were stimulated with 1 μM ionomycin (Calbiochem) and 50 ng/mL PMA (Sigma-Aldrich) at indicated time points as previously described (5). Cells were then fixed in 1% formaldehyde for 5 min and subjected to ChIP analysis as previously described (6). The complete protocol is provided in SI Materials and Methods. Primary human T cells were obtained by apheresis from the National Institutes of Health (NIH) Clinical Center Division of Transfusion Medicine and expanded by 3 cycles of phytohemagglutinin (PHA)/IL-2 stimulation before resting for 24 to 48 h before use. Human CD4 and CD8 T cells were sorted from healthy donors according to their memory differentiation under sterile conditions using a FACS Aria system (Becton Dickinson) in a BSL-3 facility (complete methods are described in SI Materials and Methods).

p300 and pol II Genome-Wide Promoter Location Analysis.

To identify p300/pol II binding sites, Jurkat T cells were treated with P/I for 0 or 45 min before cross-linking with 1% formaldehyde and were subjected to ChIP. ChIP-chip, DNA purification, and ligation mediated PCR (LMPCR) were performed as previously described (42) using antibodies against p300 and the CTD of pol II. ChIP-chip samples were labeled and hybridized to NimbleGen HG17 proximal promoter microarrays (Roche). Details of the procedure and data analysis are provided in SI Materials and Methods.

Gene Expression Analysis.

Gene expression analysis was performed using U133 plus 2.0 Genechips (Affymetrix) and is described in SI Materials and Methods.

Real-Time PCR.

PCR reactions were performed using SYBR Green Mix (Invitrogen) and 7300 Thermal Cycler (Applied Biosystems) instructions and were analyzed for ChIP and RT-PCR according to the manufacturer's instructions. Primer sequences for ChIP, qRT-PCR for mRNA, and qRT-PCR for nascent RNA (amplicon-containing intronic sequences) are provided in SI Text.