Transcription initiation patterns indicate divergent strategies for gene regulation at the chromatin level

Abstract

The application of deep sequencing to map 5' capped transcripts has confirmed the existence of at least two distinct promoter classes in metazoans: "focused" promoters with transcription start sites (TSSs) that occur in a narrowly defined genomic span and "dispersed" promoters with TSSs that are spread over a larger window. Previous studies have explored the presence of genomic features, such as CpG islands and sequence motifs, in these promoter classes, but virtually no studies have directly investigated the relationship with chromatin features. Here, we show that promoter classes are significantly differentiated by nucleosome organization and chromatin structure. Dispersed promoters display higher associations with well-positioned nucleosomes downstream of the TSS and a more clearly defined nucleosome free region upstream, while focused promoters have a less organized nucleosome structure, yet higher presence of RNA polymerase II. These differences extend to histone variants (H2A.Z) and marks (H3K4 methylation), as well as insulator binding (such as CTCF), independent of the expression levels of affected genes. Notably, differences are conserved across mammals and flies, and they provide for a clearer separation of promoter architectures than the presence and absence of CpG islands or the occurrence of stalled RNA polymerase. Computational models support the stronger contribution of chromatin features to the definition of dispersed promoters compared to focused start sites. Our results show that promoter classes defined from 5' capped transcripts not only reflect differences in the initiation process at the core promoter but also are indicative of divergent transcriptional programs established within gene-proximal nucleosome organization.

Figure 1. Promoter Classes Reflect Distinct Profiles of Nucleosome Organization.

Profiles are based on promoters classified as Narrow Peak (NP), Broad with Peak (BP), and Weak Peak (WP), and show the region of –1 kb to +1 kb around the designated TSS. RI refers to average levels at random intergenic sites, which is used as a baseline. (A) Increased H2A.Z levels (p<10E-36), (B) increased bulk levels, and consistent spacing were observed for human BP and WP promoters compared to NP. DNase hypersensitive sites revealed a more accessible nucleosome-free region at BP and WP but not at NP promoters (C), yet pol II levels were higher at NP promoters (D).

Figure 2. Heatmap of Nucleosome Occupancy within Individual Promoters.

Raw H2A.Z nucleosome occupancy values for each human promoter were partitioned into the three classes. The lower panel shows the average occupancy profile across all three classes. Within each class, promoters were arranged by location of their maximum occupancy value in the range of the −1 to +1 nucleosome (−400:+250 with respect to the TSS; the diagonal pattern is thus implied by this ordering and not the data). WP and BP promoters clearly reflected the periodic H2A.Z nucleosomes flanking the NFR, especially downstream of the TSS. Between promoters, the strongest enrichments were often observed at different nucleosomes, likely due to the sparse nucleosome occupancy data.

Figure 3. The Presence of a CpG Island Alone Does Not Imply Distinct Chromatin Architecture.

When stratifying promoters according to the presence of CpG islands as defined by Takai and Jones , no deviation in the nucleosome organization of the promoter classes is observed; WP and BP promoters maintain a higher association to H2A.Z than NP promoters. This pattern is consistent in alternative definitions of CpG islands (cf. Figure S2).

Figure 4. Computational Models Using Chromatin Features Show Different Accuracy for Promoter Classes.

Classification accuracy of two epigenetic models (i.e., using chromatin features) was evaluated on test sets for each promoter class (evaluated with auROC and auPRC). Values of 1 indicate perfect classification; auROC values close to 0.5 and auPRC values close to 0 reflect random results. At the bottom, relative weights of chromatin profile features included in each model are depicted.

Figure 5. Computational Models Support the Stronger Contribution of Chromatin Features to the Definition of Dispersed TSSs.

Changes in accuracy (auPRC) when using sequence models and combined (sequence and epigenetic feature) models are given, relative to the baseline performance in Figure 4. Below, the relative contribution of sequence and epigenetic features in the combined models is shown.

Figure 6. Distinct Nucleosome Organization Is Conserved in Insects.

(A) Fruit fly H2A.Z profiles show that BP and WP promoters had increased H2A.Z levels (p<10E-07). Nucleosomes in BP and WP promoters had a more precise spacing, with an average separation of 170 bp and deviations of up to 10 bp, compared to a mean distance of 183 bp between H2A.Z peaks at NP promoters, with deviations of up to 33 bp. (B) Differences between promoter classes were less pronounced in the available lower-resolution Drosophila bulk nucleosome data, with a slight shift compared to H2A.Z as originally reported . (C) Average bulk nucleosome occupancy profiles were computed by an in silico model, which assigned the predicted probability that a nucleosome was present at any given location . An average occupancy score of .5 indicated no preference for nucleosome presence or absence, as reflected in the scores at random intergenic locations. A clear separation of NP, BP, and WP profiles was observed, and the NFR for NP promoters was clearly much less pronounced; all predicted profiles were significantly different from each other (p<10E-09). (D) NP promoters had noticeably higher levels of pol II binding than BP and WP promoters (12–16 hr embryos). (E,F) Stalled NP, BP, and WP promoters in Drosophila mixed stages embryos (0–16 hr) maintained the same associations to H2A.Z and bulk nucleosomes as observed for the set of all actively transcribed 0–12 hr promoters.

Figure 7. Insulator Classes Are Characteristic of Promoter Classes.

(A) Human CTCF had higher occurrences in BP and WP promoters (p<10E-11). (B,C,D) Two classes of fruit fly insulators were compared to promoters classes on embryonic data from 0–12 hr. Class I insulators (including dCTCF, CP190, and BEAF32) supported the same pattern of increased BP and WP levels as observed for human CTCF. (E,F) Class II insulators had equal occurrence across promoter classes, with Su(Hw) not being bound to proximal promoter regions. (G,H) ChIP-chip profiles of the chromatin-remodeling transcription factor GAF, as well as presence of GAGA binding sites in the genome, showed a clear enrichment at NP promoters (p<10E-02).

Figure 8. Promoter Classes Are Indicative of Divergent Strategies for Transcription Initiation.

The aggregation of differences in transcription factor binding sites, nucleosome organization, histone variants and chromatin marks as well as insulator elements paint a picture of divergent strategies for transcription initiation in metazoans. (A) NP promoters are marked by a ‘fuzzy’ nucleosome organization (noted by alternative bulk -1 nucleosome locations in the figure) yet precise positioning of transcription initiation, which is reflected in the presence of location specific core promoter motifs that interact with a canonical TBP-containing basal complex , . NP promoters show higher levels of pol II bound around the TSS, possibly due to an enriched presence of stalled polymerase. They are also associated with specific chromatin remodelers in fly, namely GAF. (C) Initiation events in WP promoters spread over a larger genomic span, reflected in the presence of motifs with lower positional enrichment that have been linked to remodeled basal complexes containing TRF2 in fly . They exhibit a well-defined NFR and well-positioned H2A.Z nucleosomes as well as associated histone marks such as H3K4 tri-methylation. WP promoters in fly contain an enrichment of Class I insulators (CTCF, CP190, BEAF32). (B) BP promoters have a combination of features from both transcriptional programs. While chromatin organization is conserved, some of the known core promoter sequence elements depicted appear to be fly specific (Motif 1, DRE, Motif 6, Motif 7, MTE) , , . Pol II and insulator proteins are depicted at the maximum binding locations; sizes of the transcriptional components are not drawn to scale.