Search

Nucleosomes are involved in DNA compaction and transcriptional regulation. Yet it is unclear whether histone modification marks are primary or secondary to transcription and whether they interact to form a histone code. We investigated the relationship between transcription and four histone modifications (H4ac, H3ac, H3K4me2/3) using ChIP-chip and expression microarray readouts from two murine cell lines, one in two differentiation stages. We found that their association with transcript levels strongly depends on the combination of histone modifications. H3K4me2 coincides with elevated expression levels only in combination with acetylation, while H3ac positive association is diminished by co-occurring modifications. During differentiation, upregulated transcripts frequently gain H4ac, while most modification conversions are uncorrelated with expression changes. Our results suggest histone modifications form a code, as their combinatorial composition is associated with distinct readouts. Histones may primarily function as signaling marks for specific effectors rather than being a sufficient driving force for or a consequence of transcription.

In studies of their transcriptional activity, genomes have shown a high order of organization. We assessed the question of how genomically neighboring genes are transcriptionally coupled across tissues and what could be the driving force behind their coupling. We focused our analysis on the transcriptome information for 13 tissues of Mus musculus and 79 tissues of Homo sapiens. The analysis of coexpression patterns of genomically adjacent genes across tissues revealed 2619 and 1275 clusters of highly coexpressed genes, respectively. Most of these clusters consist of pairs and triplets of genes. They span a limited genomic length and are phylogenetically conserved between human and mouse. These clusters consist mainly of nonparalogous genes and show a decreased functional and similar regulatory relationship to one another compared to general genomic neighbors. We hypothesize that these clusters trace back to large-scale, qualitative, persistent reorganizations of the transcriptome, while transcription factor regulation is likely to handle fine-tuning of transcription on shorter time scales. Our data point to so far uncharacterized cis-acting units and reject cofunctionality as a driving force.