Transgenerational Epigenetic Inheritance Mediated by a Core Promoter Element

Division Of Basic Sciences - Nci

Investigators

Abstract

Transcriptional control is achieved through interactions of transcription factors with core promoter elements, such as the CCAAT box, which is located about 60-70 bp upstream of transcription start sites in most genes. The CCAT box serves as a docking site for specific transcription factors like NF-Y and has been highly conserved from E.coli to humans. Although its regulatory role as a promoter element has been well established, structural roles for CCAAT boxes - either as boundary elements or in DNA looping - have not been described. We have shown previously that transgenic mice with a genomic DNA fragment of the major histocompability complex (MHC) class I gene, PD1, that encodes a cell surface glycoprotein which mediates immune surveillance, displays normal tissue patterns of expression and cytokine responses. To assess the roles of core promoter elements in the regulation of expression of MHC class I genes, a series of transgenic mice with mutations either in the CCAAT box, the TATA box, Sp1 binding site or Inr were generated. Surprisingly, all of the mutant promoters were capable of supporting expression of the class I transgene (described in Barbash et al, 2013). Of 32 independently derived transgenic founder mice with CCAAT element mutations, only 6 failed to express the PD1 transgene; the remaining 26 founder mice all expressed the transgene at levels equal to or greater than the wild type. To investigate the basis for the expression of the CCAAT mutant promoter, we selected eight independently derived founders for breeding to establish lines of CCAAT (CCAATm) mutant transgenes; pups were monitored for expression of the PD1 protein by FACS analysis of their peripheral blood lymphocytes (PBL). Of the 8 founders, five of which expressed the transgene, all gave rise to off-spring that also expressed the transgene. Surprisingly, all 8 lines also gave rise to off-spring that had the transgene but did not express PD1 on their PBL. In subsequent generations of inter se breeding, CCAATm mice that expressed the PD1 transgene gave rise to offspring that were genotype positive for the transgene but did not express it. Conversely, breeding of transgenic non-expressers mice inter se generated offspring that did express the transgene. This pattern of variegated expression continued in each line for up to 4 generations after which the phenotype was stable and remained stable even after 9 generations. Transgenerational variegation of transgene expression was not observed in any lines of mice with either the wild type promoter or any of the other core promoter mutants. The observed variegated expression among generations was not restricted to PBL: CCAATm transgenic mice that expressed the PD1 protein on their PBL also expressed RNA and protein in their tissues (i.e. spleen, kidney, brain). The CCAATm transgenic mice that did not express PD1 protein on their PBL, also did not express either PD1 protein or RNA in these tissues. However, the pattern of expression among expresser CCAATm transgenes differed from the wild type, with aberrantly high expression in kidney and brain, indicating that the CCAAT box also regulates tissue specific transcription. Expresser CCAATm transgenes, like the wild type promoter transgene and endogenous MHC class I genes, responded to gamma-interferon treatment with enhanced class I expression. In contrast, gamma-interferon treatment did not induce de novo expression of PD1 in the non-expresser CCAATm transgenic mice, although it did enhance endogenous H2-Kb expression. Thus, in the absence of a CCAAT box, global expression of the transgene is variegated and stochastic from generation to generation. To validate these findings, we generated a completely independent series of CCAATm transgenic mice, which again displayed variegated expression across generations. These reproducible findings suggested that the wild type CCAAT box normally functions to maintain epigenetic inheritance of expression. They also lead to the question of what distinguishes the CCAAT mutant expressers from the non-expressers. The CCAAT element is known to bind either NF-Y or C/EBP, depending on the context. We found that NF-Y, but not C/EBP, binds to the CCAAT element in vivo. Mutation of the CCAAT box abrogates binding in both the CCAATm expressers and non-expressers. Importantly, depletion of NF-Y phenocopies the CCAATm in resulting in increased expression in fibroblasts. We next asked whether mutation of the CCAAT element affects the association of Pol II with the transgene. Whereas Pol II was bound across both the wild type and CCAATm expresser transgenes, its binding to the non-expresser transgene was weak and limited to a region upstream of the CCAATm box. binding of cognate transcription factors. We next asked what effect mutation of the CCAAT element has on chromatin organization or structure, relative to the wild type and whether this differs between the expressers and non-expressers by comparing nucleosome occupancy and histone modifications. As expected, the wild type class I transgene displayed a nucleosome-depleted region around the core promoter and TSS. In contrast, nucleosome occupancy across the promoter was much higher in both the CCAAT mutant expresser and non-expresser mice. Analysis of the histone modifications across the transgenes gave similar results: chromatin across the transgene of both the WT and CCAATm expresser mice was acetylated, whereas the non-expresser transgene was not. Interestingly, none of the transgenes had repressive chromatin methylation marks. These findings suggest that NF-Y/CCAAT function as a barrier element to maintain a transcriptionally open chromatin structure. To further examine chromatin structure, the association of CTCF and cohesin with the transgenes was determined. CTCF functions as a barrier element and also anchors loops created by cohesin. Here, for the first time, we observed differences between the CCAATm non-expresser, compared with the expresser or WT. CTCF binds to both the CCAATm expresser and WT, but not the non-expresser, upstream of the CCAAT box around -150bp. In contrast, CTCF binds to the non-expresser transgene, but not the others, around -600bp; its binding is also enriched at exon 5. We did not observe any clear correlation between expression and cohesin binding. Since CTCF is known to mediate looping, we asked whether loops exist within the class I transgene. Interestingly, we observed marked changes in DNA looping amongst the WT, CCAATm expresser and non-expresser. Whereas all three lines have a predominant loop between -425 bp and exon 3, this loop is enhanced in the CCAATm expresser. Additionally, two new loops are formed in the CCAATm expresser that are not present in the other two lines, one of which has an anchor at exon 1, near the CCAAT box. These findings suggest that the CCAAT box regulates loop formation, presumably through the binding of cognate factors. To our knowledge, this is the first example of a single mutation resulting in the establishment of trans-generational epigenetic inheritance. These studies are being prepared for publication.