Inter-chromosomal transcription hubs shape the 3D genome architecture of African trypanosomes

Abstract

The eukaryotic nucleus exhibits a highly organized 3D genome architecture, with RNA transcription and processing confined to specific nuclear structures. While intra-chromosomal interactions, such as promoter-enhancer dynamics, are well-studied, the role of inter-chromosomal interactions remains poorly understood. Investigating these interactions in mammalian cells is challenging due to large genome sizes and the need for deep sequencing. Additionally, transcription-dependent 3D topologies in mixed cell populations further complicate analyses. To address these challenges, we used high-resolution DNA-DNA contact mapping (Micro-C) in Trypanosoma brucei, a parasite with continuous RNA polymerase II (RNAPII) transcription and polycistronic transcription units (PTUs). With approximately 300 transcription start sites (TSSs), this genome organization simplifies data interpretation. To minimize scaffolding artifacts, we also generated a highly contiguous phased genome assembly using ultra-long sequencing reads. Our Micro-C analysis revealed an intricate 3D genome organization. While the T. brucei genome displays features resembling chromosome territories, its chromosomes are arranged around polymerase-specific transcription hubs. RNAPI-transcribed genes cluster, as expected from their localization to the nucleolus. However, we also found that RNAPII TSSs form distinct inter-chromosomal transcription hubs with other RNAPII TSSs. These findings highlight the evolutionary significance of inter-chromosomal transcription hubs and provide new insights into genome organization in T. brucei.

Fig. 1. A highly contiguous T. brucei genome assembly was generated with ultra-long nanopore reads.

a An overview of the features and regions on the megabase chromosomes of the Lister 427 T. brucei assembly that have been modified. Allele ‘A’ and ‘B’ are shown on the top and bottom of each chromosome, respectively. Modified regions (gaps or collapsed regions, fully or partially corrected) are marked in blue. Regions with open gaps or collapsed repeats that could not be corrected and remained unmodified are marked in red. b Sizes of closed gaps (top) and change in size of corrected repeat regions (bottom). c Illustration of our strategy to confirm the correctness of the modifications made to the genome. Modified regions are considered corrected if there are at least 2 VPRs with less than 5 kb difference between the distance where the VPRs map on the assembly and their distance based on ONT read. d Correctness evaluation of modified regions in the new genome assembly (version 12).

Fig. 2. Micro-C reveals interactions among RNAPII transcription start sites.

a IC-normalized and ploidy-corrected heat maps of Hi-C (bottom left) and Micro-C (top right) of megabase chromosomes 1-11 (haplotype A) at 50 kb resolution (left) with each chromosome displayed with the core region in white, subtelomeric regions in light gray and centromeres displayed as black lines; part of chromosome 10 of the same heat maps is displayed at 5 kb resolution (right), with polycistronic transcription units of chromosome 10 illustrated as black arrows (top). b Aggregate analysis of Micro-C data around pairs of intra-chromosomal TSSs (n = 2342 TSS-TSS pairs); control loci are obtained from the same pairs of loci shifted downstream by 20 kb. c Aggregate analysis of Micro-C data around pairs of inter-chromosomal TSSs (n = 20,008 TSS-TSS pairs); control loci are obtained from the same pairs of loci shifted downstream by 20 kb. d Aggregate peak analysis of Hi-C and Micro-C data around chromatin loops (i.e., pairs of bins with high interaction frequency) identified in the Micro-C matrix using Mustache (with a p-value threshold of 0.02 and a sparsity threshold of 0.85; n = 367).

Fig. 3. Transcription start sites localize to distinct kilobase-scale compartments.

a Examples of correlation matrices for chromosomes 4, 5 and 11 (haplotype A) and corresponding second eigenvector calculated from the Micro-C data analyzed with the distiller pipeline at 10 kb resolution (using FAN-C). Each chromosome is displayed under the eigenvector plot with the core region in grey and subtelomeric regions in dark red; centromeres are represented as black dots. b Examples of domain boundaries identified manually with insulation scores calculated from the Micro-C data analyzed with the distiller pipeline at 2.5, 5 and 10 kb resolution using FAN-C (top); mappability of the region, TAD-like domains identified with HiCExplorer, no visibility regions (i.e., regions where signal was removed by IC normalization, for details see Methods), tRNA genes, SCC1 ChIP-seq fold enrichment and PTUs are displayed (bottom).

Fig. 4. Cohesin is depleted from sites engaging in TSS-TSS interactions.

a Top panel: schematics of polycistronic transcription unit types (head-to-head and head-to-tail) with divergent and single transcription start sites (dTSS and sTSS, in green) and single and convergent transcription termination sites (sTTS and cTTS, in red). Bottom panels: for each of the sites, aggregate Micro-C signal (observed/expected), metaplot of SCC1 ChIP-seq fold enrichment, RNAPII ChIP-seq fold enrichment and H2A.Z MNase-ChIP-seq enrichment.

Fig. 5. Cohesin levels gradually increase along PTUs.

a Metaplot of SCC1 ChIP- seq fold-enrichment (yellow) and of RNAPII ChIP- seq fold enrichment (purple) along polycistronic transcription units (n = 208). b Aggregated plot of interaction frequencies calculated from Micro-C data between pairs of PTUs (n = 208). PTUs were scaled to the same length and extended by 10% up- and downstream. c UpSet plot showing the chromatin loops (i.e., pairs of bins with high interaction frequency) identified in the Micro-C matrix at 5 kb resolution (n = 367), where both loci of a pair overlap with annotated TSSs (n = 298), H2A.Z MNase-ChIP-seq peaks (n = 592; see Methods), annotated TTSs (n = 309), SCC1 ChIP- seq peaks (n = 381; see Methods), and/or annotated tRNA genes (n = 137).

Fig. 6. rRNA and tRNA gene loci form polymerase class-specific clusters.

a Top: Virtual 4C interaction profiles based on Micro-C data (10 kb resolution) showing the interaction frequencies of rDNA regions (defined as groups of adjacent rDNA loci separated by less than 10 kb) used as viewpoints with the Tb427 genome version 12, haplotype A. Significant interactions as calculated using a two-sample KS-test were marked below each track (pink: p < 0.05; red: p < 0.01). Bottom: rDNA regions are highlighted in green; red boxes represent centromeres; each chromosome is displayed with the core region in white and subtelomeric regions in light gray. b Top: Virtual 4C interaction profiles based on Micro-C data (5 kb resolution) showing the interaction frequencies of tRNA regions (defined as groups of adjacent tRNA loci separated by less than 10 kb) used as viewpoint with the Tb427 genome version 12, haplotype A. Significant interactions as calculated using a two-sample KS-test were marked below each track (pink: p < 0.05; red: p < 0.01). Bottom: tRNA regions are highlighted in yellow; red boxes represent centromeres; each chromosome is displayed with the core region in white and subtelomeric regions in light gray.

Fig. 7. T. brucei genome form widespread inter-chromosomal interactions with centromere clustering.

a Top: Virtual 4C interaction profiles based on Micro-C data (10 kb resolution) showing the interaction frequencies of a subset of centromeres used as viewpoints (black arrowheads and green background) with the Tb427 genome version 12, haplotype B. Significant interactions as calculated using a two-sample KS-test were marked below each track (pink: p < 0.05; red: p < 0.01). Bottom: Each chromosome is displayed with the core region in white and subtelomeric regions in light gray. Centromeres are marked by red rectangles; rDNA loci are marked in green. b Fluorescence in situ hybridization (FISH)-based quantification of CIR147 repeats (located on chromosomes 4, 5 and 8) clusters (N = 1; analyzed nuclei: 171). Cells displaying no signal were excluded. c Model of T. brucei genome organization. T. brucei chromosomes make frequent centromere-centromere interactions, interactions between RNAPII TSSs, the active VSG gene and the splice leader locus, rRNA genes and some tRNA genes (not displayed).