Using ChIA-PET with antibodies recognizing the H3K4me2 signal, which marks active enhancers in the genome, we identified 6 520 long-distance chromatin interactions in human primary CD4+ T cells. In particular, we identified 2 067 enhancers that interacted with 1 619 specific promoters. We found that 57% of the tested enhancer elements showed enhancer activity on the heat shock reporter promoter, and these enhancers globally increased the expression of their target genes in CD4+ T cells but not in HEK293 cells, consistent with our observation that selected enhancers exhibited T cell-specific interactions with their target promoters. Our results indicate that interacting promoters are co-expressed. Our data also revealed multiple levels of chromatin organization comprising subdomains and domains rich of “local” interactions.
Previous genome-wide studies predicted thousands of potential enhancers based on sequence features, various epigenetic signatures, and regulatory factors 10, 11
. Correlation analysis of histone modifications on the predicted enhancers with expression of genes in proximity indicated that a subset of enhancers may regulate the expression of their nearby target genes 12
. However, an enhancer can regulate target genes over long distance, sometimes skipping intervening genes or even from different chromosomes. In these cases, it is difficult to predict targets for enhancers based on proximity. Since enhancers must interact with target promoters to activate their expression, direct interaction assays provide a powerful tool to identify target genes for enhancers. In this study, we were able to assign 2 067 potential enhancers to 1 619 specific target promoters in human primary CD4+
T cells using direct interaction assays. Our data indicated that the genes that interact with the identified enhancers exhibited higher expression than the control group of genes, suggesting that the enhancer regions have enhancer activity as indicated by the reporter assays. However, we found that 23% of these enhancer elements were bound by CTCF, an insulator-binding protein. Therefore, the CTCF-bound regions may not function directly to activate transcription of target genes, but instead may act to bring other enhancers in close proximity to target promoters, as suggested by a recent study 23
. Nonetheless, our data suggested that CTCF-bound enhancers and p300-bound enhancers collectively control the expression of genes involved in T cell differentiation and function.
Subnuclear organization critically influences gene expression. Transcription of active genes is believed to take place at transcription factories that are foci of hyperphosphorylated RNA polymerase II and are associated with active transcription 25, 26
, suggesting that the co-localized genes may be co-regulated. However, there is only sparse evidence on co-localization of genes and their co-regulation in the genome 27
, although relocation of genes in the nucleus is observed upon repression or activation 28, 29
. Recently, Schoenfelder et al.30
presented an elegant data showing that active globin genes associate with hundreds of other active genes located both on the same and different chromosomes, highlighting the coordinated regulation of genes by co-localization. Consistently, the INS
gene promoter physically interacts with the SYT8
gene, located > 300 kb away, which is critical for the regulated expression of the SYT8
gene depending on insulin signals 31
. In our study, we comprehensively identified promoter-promoter interactions and showed that interacting promoter pairs are co-expressed. Furthermore, the genes interacting with the same enhancer were significantly co-expressed. Importantly, the co-expression of the interacting genes appeared to be tissue specific, as the expression of linked genes, which were identified from CD4+
T cells, was only significantly correlated in CD4+
T cells and not in HEK293 cells, thus providing genome-wide support for co-localization and co-expression of functionally important genes.
Interphase chromosomes adopt preferred conformations 32
, which bring different functional elements into proximity by compartmentalization of the nucleus 33, 34, 35, 36
. However, our knowledge about the folding of chromatin beyond the 30-nm fiber is very limited. A recent unbiased mapping of long-range chromatin interactions confirmed the presence of chromosome territories and the segregation of open and closed chromatin compartments 37
. The genome-wide interaction data support the fractal globule model of chromatin organization through multiple rounds of crumpling an unentangled chromatin fiber into a series of small globules in a beads-on-a-string configuration. Our data support a model of multiple levels of chromatin organization () such that chromosomes contain active domains rich in relatively local interactions, and these domains are connected by super long-distance interactions. These super long-distance interactions may be responsible for bringing different chromatin domains to spatial proximity required for the reciprocal translocation observed in various human diseases. What is the force that drives the chromatin organization in the nucleus? One of the best-studied factors of chromatin organization is CTCF, a chromatin insulator-binding protein. CTCF is associated with tens of thousands of sites on chromatin 7, 20, 21
and mediates the functional interaction of several well-studied genomic loci 38, 39
. A recent study has unveiled the genome-wide chromatin interactions mediated by CTCF 23
. One critical partner for the function of CTCF in forming long-distance chromatin loop is cohesin. CTCF and cohesin are broadly co-localized on mammalian chromosomes 40, 41, 42
. However, our data indicated that only a minor fraction of the EP interactions was associated with CTCF binding, suggesting that other factors may be involved. Indeed, we found that motifs for several transcription factor families such as ETS
, and GATA
, which represent key factors for T cell development and function, were identified in the enhancer regions. Future studies are needed to address the mechanisms that regulate the long-distance EP and promoter-promoter interactions.