Report

Reciprocal Binding of PARP-1 and Histone H1 at Promoters Specifies Transcriptional Outcomes

See allHide authors and affiliations

Science  08 Feb 2008:
Vol. 319, Issue 5864, pp. 819-821
DOI: 10.1126/science.1149250

Abstract

Nucleosome-binding proteins act to modulate the promoter chromatin architecture and transcription of target genes. We used genomic and gene-specific approaches to show that two such factors, histone H1 and poly(ADP-ribose) polymerase-1 (PARP-1), exhibit a reciprocal pattern of chromatin binding at many RNA polymerase II–transcribed promoters. PARP-1 was enriched and H1 was depleted at these promoters. This pattern of binding was associated with actively transcribed genes. Furthermore, we showed that PARP-1 acts to exclude H1 from a subset of PARP-1–stimulated promoters, suggesting a functional interplay between PARP-1 and H1 at the level of nucleosome binding. Thus, although H1 and PARP-1 have similar nucleosome-binding properties and effects on chromatin structure in vitro, they have distinct roles in determining gene expression outcomes in vivo.

Gene expression outcomes are determined, in part, by the composition of promoter chromatin, including the posttranslational modification state of nucleosomal histones (1), the incorporation of histone variants (2), and the presence of nucleosome-binding proteins (3). Linker histone H1 and poly(ADP-ribose) polymerase-1 (PARP-1) are examples of nucleosome-binding proteins that modulate the chromatin architecture and transcription of target genes (4, 5). H1 and PARP-1 bind to overlapping sites on nucleosomes at or near the dyad axis where the DNA exits the nucleosome (6, 7). Unlike H1, PARP-1 has an intrinsic nicotinamide adenine dinucleotide (NAD+)–dependent enzymatic activity that regulates its association with chromatin (7). Previous work from our laboratory has shown that H1 and PARP-1 bind in a competitive and mutually exclusive manner to nucleosomes in vitro and localize to distinct nucleosomal fractions in vivo (7), suggesting distinct roles for these factors in the regulation of gene expression. However, little is known about how H1 and PARP-1 are distributed across the mammalian genome and how they interact to regulate global patterns of gene expression in vivo.

To determine the patterns of H1 and PARP-1 localization across selected regions of the human genome, we performed chromatin immunoprecipitation (ChIP) in MCF-7 breast cancer cells using antibodies specific to PARP-1 and H1 (7, 8), coupled with hybridization of the enriched genomic DNA to custom microarrays (i.e., ChIP-chip) (9). Each array represented 57 Mb of genomic DNA, including all 44 of the ENCODE regions (10), as well as an additional 1117 promoter regions selected from genes regulated by enzymes in the nuclear NAD+ signaling pathway (5) [approximately –25 to +5 kb relative to the transcription start site (TSS)]. The raw ChIP-chip signal to input ratios were processed (11) and aligned to the TSSs for all 1517 RNA polymerase II (Pol II)–transcribed promoters on the array (i.e., ENCODE + selected). We observed an enrichment of PARP-1 and a depletion of H1 in the region surrounding the TSSs (Fig. 1A and fig. S1). Significant peaks of PARP-1 and troughs of H1 [P < 0.01, Wilcoxon signed-rank test (12, 13)] were clustered around the TSSs, but were also found in upstream and intergenic regions (Fig. 1, B and C, and figs. S2 and S3). This pattern of PARP-1 and H1 localization was also revealed by averaging the ChIP-chip data over the 30-kb tiled region for all promoters on the array or in a 20-kb region centered around significant PARP-1 peaks or H1 troughs (P < 0.01, Wilcoxon signed-rank test) (fig. S4). Collectively, our ChIP-chip data identify a reciprocal relation for chromatin binding by PARP-1 and H1 across the genome.

Fig. 1.

Distinct patterns of genomic localization for H1 and PARP-1. (A)Heat maps of H1 and PARP-1 ChIP-chip data shown from –10 to +5 kb relative to the TSS. The data are limited to 758 promoters with the highest PARP-1 ChIP-chip signals and are ordered based on the intensity of the PARP-1 signal at the promoter. See fig. S1 for the full data set. (B) Histograms showing the number of statistically significant peaks and troughs of PARP-1 and H1 across the entire 30-kb tiled region for the 1517 promoters on the ChIP-chip array. (C) Diagram of statistically significant peaks and troughs of PARP-1 and H1 across an ENCODE region from chromosome 1 (Chr. 1). Annotated Reference Sequence (RefSeq) genes are represented by arrows indicating the length of the gene and direction of transcription. Green arrows: expressed in MCF-7 cells as determined by expression microarrays; gray arrows: ambiguous or no expression information available. Asterisks indicate genomic locations with a PARP-1 peak and a H1 trough.

Although eukaryotic promoters generally show reduced nucleosome occupancy (14, 15), this was not an important determinant for the reciprocal pattern of PARP-1 and H1 binding. For example, whereas PARP-1 peaks and H1 troughs are strongly correlated at promoters (Spearman rank correlation: –0.495, P = 3.7 × 10–94), they show little correlation with the presence of H3 (Fig. 2A; see also SOM Text). In addition, the pattern of PARP-1 and H1 binding at promoters (e.g., low versus high PARP-1/H1 ratios) is independent of the pattern of H3 occupancy at promoters (Fig. 2B). Finally, the reciprocal pattern of PARP-1 and H1 binding is observed in intergenic regions where H3 is not depleted (fig. S4B). Despite the reduced H3 occupancy at promoters, well-positioned nucleosomes are present at PARP-1–bound promoters that likely serve as targets for the binding of PARP-1 (fig. S5).

Fig. 2.

A high PARP-1:H1 ratio specifies actively transcribed promoters. (A) Correlation analyses of PARP-1, H1, and H3 occupancy as determined by ChIP-chip (at the –250 bp-centered window) with gene expression (Expr.) as determined by microarrays. (B) Averaging analysis of the log2 enrichment ratios from H1 and PARP-1 ChIP-chip for unambiguously expressed (top) or unambiguously unexpressed genes (bottom). (C) Top: Averaging analysis of the log2 enrichment ratios from H1 and PARP-1 ChIP-chip for genes (i) having both a PARP-1 peak and an H1 trough within 1.5 kb of the TSS (left) or (ii) unambiguously lacking both a PARP-1 peak and an H1 trough within 1.5 kb of the TSS (right). Bottom: Percentage of expressed and unexpressed genes in each category. P values are from a Chi-squared test and indicate significant differences relative to the total gene set (n = 878; percent expressed = 71.1).

In a previous study (7), we concluded that PARP-1 may act to repress Pol II transcription based on the observations that (i) PARP-1 represses in vitro transcription by Pol II with chromatin templates in the absence of NAD+ and (ii) PARP-1 does not colocalize with active Pol II (Ser5-P) on Drosophila polytene chromosomes. Our current ChIP-chip results suggest that the latter observation may simply be a consequence of the localization of PARP-1 and active RNA Pol II to distinct regions of a gene (i.e., upstream versus downstream of the TSS; see SOM text). To explore the relations between PARP-1, H1, and gene expression in more detail and under physiological NAD+ concentrations, we coupled our ChIP-chip analyses with gene expression microarray analyses for MCF-7 cells grown under the same conditions. PARP-1 peaks showed a significant positive correlation with gene expression (Spearman rank correlation, P = 7.1 × 10–49), whereas H1 showed a significant negative correlation with gene expression (Spearman rank correlation, P = 7.85 × 10–39) (Fig. 2A). In addition, PARP-1 was enriched and H1 was depleted near the TSSs of expressed genes relative to unexpressed genes (Fig. 2B) (16). We then grouped all genes containing both a significant PARP-1 peak and a significant H1 trough (P < 0.01, Wilcoxon signed-rank test) and compared them to a group that lacked both a PARP-1 peak and an H1 trough (17). More than 90% of the genes containing both a PARP-1 peak and an H1 trough at the promoter were expressed, whereas less than 45% of the genes lacking both a PARP-1 peak and an H1 trough at the promoter were expressed (Fig. 2C). This correlation was also observed when looking broadly across ENCODE regions enriched in expressed or unexpressed genes (Fig. 1C and fig. S2; see asterisks). Together, these results indicate that the pattern of PARP-1 and H1 promoter localization is indicative of gene expression outcomes.

Finally, to explore further the functional relations between PARP-1, H1, and gene expression, we identified subsets of PARP-1–bound genes either down-regulated or up-regulated in MCF-7 cells by stable short hairpin RNA (shRNA)-mediated knockdown of PARP-1 (Fig. 3A) (18). For each gene, we assayed (i) promoter binding by PARP-1 and H1 using ChIP-qPCR (quantitative polymerase chain reaction) and (ii) expression by reverse transcription (RT)-qPCR, with or without PARP-1 knockdown. The subset of genes positively regulated by PARP-1 (i.e., genes whose expression decreased upon PARP-1 knockdown) showed a three- to fivefold increase in H1 binding at the promoter in response to PARP-1 knockdown without changes in H3 occupancy (Fig. 3B and figs. S6 and S7). These results provide a functional link between the chromatin binding and gene-regulatory actions of PARP-1 and H1 at this subset of target promoters. Specifically, they suggest that PARP-1 acts to exclude H1 from these promoters and that upon PARP-1 knockdown, H1 is able to rebind and inhibit transcription. In contrast, the subset of genes negatively regulated or not regulated by PARP-1 (i.e., genes whose expression decreased or was unchanged upon PARP-1 knockdown) showed little or no change in H1 binding at the promoter in response to PARP-1 knockdown (Fig. 3C and fig. S8). These genes, some of which show a reciprocal pattern of PARP-1 and H1 localization at their promoters (Fig. 3C and fig. S8), may be subject to other PARP-1–related transcriptional regulatory mechanisms (5, 19)or indirect regulatory effects.

Fig. 3.

PARP-1 excludes H1 from PARP-1-regulated promoters. (A) Western blot showing the shRNA-mediated depletion of PARP-1 in MCF-7 cells versus control luciferase (Luc) knockdown cells. (B and C) Gene-specific analysis of PARP-1, H1, and H3 promoter binding by ChIP-qPCR and mRNA expression by RT-qPCR in MCF-7 cells with or without PARP-1 knockdown. Expression data are standardized to β-actin transcripts. Bars represent the mean + SEM, n ≥ 3.

Collectively, our data reveal the genomic localization patterns of H1 and PARP-1, highlighting the reciprocal relation for their binding at promoters and other genomic locations. In addition, our results provide a functional link between chromatin binding by PARP-1 and H1 at a subset of target promoters and the corresponding gene expression outcomes. Finally, our results suggest that PARP-1 acts to exclude H1 from a subset of PARP-1–regulated promoters in vivo. Our data fit well with and extend the results of previous biochemical and cell-based assays showing a role for PARP-1 in the transcription-related regulation of chromatin structure (7, 20, 21) and functional interplay between H1 and PARP-1 (7, 22, 23). Further, our results show that although H1 and PARP-1 have similar nucleosome-binding properties and effects on chromatin structure in vitro (7, 20), they have distinct roles in regulating gene expression outcomes in vivo. Future studies will examine the determinants that direct the specific pattern of H1 and PARP-1 binding at promoters, including the role of PARP-1's NAD+-dependent enzymatic activity.

Supporting Online Material

www.sciencemag.org/cgi/content/full/319/5864/819/DC1

Materials and Methods

SOM Text

Figs. S1 to S8

References

References and Notes

View Abstract

Navigate This Article