Deciphering CTCF code in mammalian host and viral epigenomes

CTCF是一种高度保守的DNA结合蛋白，是染色质结构的全局组织者。CTCF参与转录激活和抑制的调控、基因印迹、细胞增殖和凋亡的控制、染色质区隔化、x染色体失活、防止三核苷酸重复扩增以及其他染色质驻留过程。我们花了20多年的时间来研究CTCF，以说服其他人CTCF的多功能性确实是基于高度保守的“多价11 ZF DBD”结合多种不同DNA序列的能力，以及通过组合使用DNA连接和蛋白质连接ZF与伴侣蛋白相互作用的内在能力。去年，在果蝇Su（Hw）因子中，另一个poly-ZF DBD阵列显示了类似的多价性。随着下一代测序技术的出现，CTCF结合位点已经在果蝇、小鼠和人类基因组中被鉴定出来。数以千计的非同源CTS序列被发现与参与远程染色质相互作用的基因组区域相关，包括增强子、启动子和基因间边界元件，这反映了CTCF的多种功能。然而，对于任何给定CTS的特定DNA序列如何与同一位点的特定CTCF功能相关，仍然是模糊的。今年，我们在了解通过dna接触手指的不同组合形成的不同CTCF/ dna复合物的多种功能方面取得了额外的进展。通过在全基因组范围内同时绘制CTCF和BORIS占用图谱，我们发现了两类CTCF结合区域，它们在DNA序列中是预先编程的和进化保守的。我们发现，70%的CTCF结合区域包含一个单一的CTCF结合位点，即“1xCTSes”，而ChIP-seq检测到的其他30%的CTCF结合区域实际上包含两个CTCF结合位点，即二元“2xCTSes”。相邻CTSes在二元2xcts区域内的占用限制了2个相邻CTCF蛋白在正常体细胞中形成同型二聚体，或在CTCF上共表达BORIS的生殖细胞和癌细胞中在同一DNA位点组装CTCF+ BORIS共结合的异源二聚体。最近对2xcts区域的突破性发现（通过标准的CTCF特异性ChIP-Seq无法解析）使我们第一次能够解决长期存在的问题，即CTCF如何在同一细胞核中作为真正的转录因子，同时在假定的绝缘子/边界位点保持大量存在，而这些位点没有转录活性的指示。事实上，在任何给定的细胞类型中，只有20%的CTCF结合区域位于启动子区域，而其余的ctse与转录起始位点无关。这种不同的功能作用的决定因素的明显候选将是DNA序列本身和/或在这两种类型的位点染色质的不同身份。在我们的研究中，我们提出了全基因组证据，证明两种类型的CTCF靶点的DNA序列在结构上是不同的。两类CTCF结合位点之间的结构差异与其功能差异有关：2xctse优先位于Pol II共同结合的h4k27acs标记的启动子和增强子上，并且在减数分裂后的精子中发现相同的2xCTS元件与正常的CTCF-BORIS异源二聚体相关，其中BORIS标记了成熟的人类和小鼠精子中单倍体外基因组中保留修饰组蛋白的未来无蛋白DNA区。与此形成鲜明对比的是，基因间和内含子基因组区域包含一个或多个基于1xcts的CTCF峰，具有命名的5‘- ccc (C/t)CT(a/g)-3’基序，该基序经常受到影响三维组织的疾病相关SNP的打击，这些SNP影响着粘性C端和来自远端CTCF/DNA复合物的无DNA zf亚群之间的基本自相互作用，参与由黏结蛋白保留稳定的位点特异性二/多聚合。在基因负担的人类受试者中发现的与CTCF +/-单倍体缺陷的显著联系，可能为临床导向的CTCF研究开辟了一条新的途径，CTCF研究与异常组蛋白/ dna甲基化有关，包括h3k27ac标记的pol2结合启动子增强子对中含有2xCTS元素的CTCF结合ChIP-Seq峰，这些元素能够改变基因表达，与我们之前与Chris Kemp和Galina合作分析的CTCF +/-小鼠中发现的方式相同来自西雅图弗雷德·哈钦森癌症中心的Filippova。因此，CTCF缺失引起的疾病可能有一个共同的潜在病理机制，不同于同一研究小组首次报道的经典TSG CTCF功能的完全丧失。因此，类似的病理相关机制似乎是CTCF剂量不足引起的人类和小鼠遗传疾病的基础，不包括额外的ZnF突变，即使在具有16q22/CTCF LOH的肿瘤中，会导致CTCF的完全丧失，而不是DNA-CTCF相互作用的部分丧失，而DNA-CTCF相互作用是由多价11 ZnF CTCF DBD中有活力的单个a.a.取代的体内选择引起的，这种选择首先在CTCF中被描述（1996年），后来（2002年）发现，在CTCF衍生的“BORIS”（“印迹状态规则的兄弟”的首字母缩写）中进行了概述。接下来，对二进制2xCTS代码的最初发现和进一步研究开始挑战当前文献中普遍存在的误解，即所有CTCF位点彼此等效，单个CTCF分子结合在单个CTS序列上，尽管具有不同基因组坐标的CTS元件可能在单个或双CTCF基序上包含一个或两个相邻的DNase I足迹，而无需可靠的基于基序的预测所需的任何同源性。增强子/启动子相关的2xcts元件的功能和染色质结构特征不同于由CTCF单体结合的含有1xcts的区域，这些区域主要位于小鼠和人类染色体上的内含子区和长基因间区。我们认为这两类以前被忽视的CTCF结合区可能在调节多种染色质基础现象中具有不同的作用，并可能影响我们对癌细胞和正常生殖细胞中遗传表观遗传调控的理解。例如，精子核小体的非随机保留被发现选择性地放置在无蛋白蛋白位点上，这是由含有相同的2xcts的CTCF元件的nt环境预先决定的，这些CTCF元件通常与CTCF和BORIS 11 ZF同源物共同结合，这些同源物通常在减数分裂后的圆形精子中共同表达。此外，CTCF和内聚蛋白复合物被广泛认为是在所有哺乳动物细胞中建立和维持三维基因组结构的关键角色。这些蛋白质不仅在科学界广为人知，而且最近还进入了大众媒体，包括三月号的《科学美国人》，网址是https://www.scientificamerican.com/article/untangling-the-formation-of-dna-loops。综上所述，我们的研究结果提供了染色质动力学的全局视图，并为研究不同人类细胞系中基因表达的远程控制提供了资源，并解释了为什么从众多转录因子，只有CTCF被认为是一种普遍的、可能不可逆转的表观遗传标记，存在于所有细胞类型的功能不同区域，以便在ENCODE联盟（www.factorbook.org/human/chipseq/tf/CTCF）编目的数十万个CTCF相关DNA序列中协调修饰组蛋白和甲基化DNA的非随机定位。

CTCF, a highly conserved DNA binding protein, serves as a global organizer of chromatin architecture. CTCF is involved in regulation of transcriptional activation and repression, gene imprinting, control of cell proliferation and apoptosis, chromatin compartmentali-zation, X-chromosome inactivation, prevention of tri-nucleotide-repeat's expansions, and other chromatin-resident processes. It took us over 20 years of CTCF studies to persuade others that the multi-functionality of CTCF is indeed based on the ability of a highly-conserved 'multivalent 11 ZF DBD to bind a wide range of diverse DNA sequences, as well as on its intrinsic capacity to interact with a partner-proteins through the combinatorial usage of DNA-contating and protein-contacting ZFs. Last year, a similar multivalency was shown for another poly-ZF DBD array in the Drosophila Su(Hw) factor. With the advent of next generation sequencing techniques, CTCF binding sites have been identified across fly, mouse, and human genomes. Reflecting the multitude of CTCF functions, many thousands of non-homologous CTS sequences were found to be associated with genomic regions engaged in long-range chromatin interactions, including enhancers, promoters, and inter-genic boundary elements. It remained obscure, however, as to how a particular DNA sequence of any given CTS is related to specific CTCF functions at the same site. This year, we have made additional advances in the direction of understanding multiple functionality of distinct CTCF/DNA-complexes formed via different combinations of DNA-contacting fingers. By mapping simultaneous CTCF & BORIS occupancy genome-wide, we uncovered two classes of CTCF binding regions that are pre-programmed and evolutionary conserved in DNA sequence. We found that 70% of CTCF bound regions enclose a single CTCF binding site, aka "1xCTSes" while other 30% of CTCF-binding regions detected by ChIP-seq as single peaks are, in fact, shown to contain the dual CTCF binding sites, aka binary "2xCTSes". Occupancy of adjacent CTSes within binary 2xCTS-regions constrains 2 adjacent CTCF proteins to form homodimers in normal somatic cells, or to assemble heterodimers of CTCF+ BORIS co-bound at the same DNA spot in germ and cancer cells co-expressing BORIS on top of CTCF. The recent breakthrough discovery of 2xCTS-regions (unresolvable by a standard CTCF-specific ChIP-Seq) enabled us, for the first time, to address the long-standing question as to how CTCF can serve in the context of the same nucleus as a bona fide transcription factor, while maintaining a substantial presence at putative insulator/boundary sites that bear no indications of transcriptional activity. Indeed, only 20% of all CTCF binding regions are located in promoter regions in any given cell type, while the remaining CTSes are not associated with transcriptional start sites. The obvious candidates for the determinants of such distinct functional roles would be DNA sequences themselves and/or differential identity of chromatin at these two types of sites. In our study we presented genome-wide evidence that DNA sequences underlying the two types of CTCF target sites are structurally different. The structural difference between two classes of CTCF binding sites is connected to their functional differences: 2xCTSes are preferentially located at H4K27ac-marked promoters and enhancers co-bound by Pol II, and the same 2xCTS elements are found to be associated with normal CTCF-BORIS-heterodimers in post-meiotic spermatids wherein BORIS marks the future protamine-free DNA zones that retain modified histones along haploid epi-genome in mature human and mouse spermatozoa. In a stark contrast, intergenic and intronic genomic regions harboring one or more 1xCTS-based CTCF peaks with the name-giving 5'-CCC(C/t)CT(a/g)-3' motif which is often hit by a disease-associated SNP affecting three-dimensional organization imprinted upon essential self-interactions among sticky C-termini and DNA-free ZF-subsets from distal CTCF/DNA complexes engaged into site-specific di-/multi-merization stabilized by cohesin retention. A remarkable link with CTCF +/- haplo-insufficiency found in genetically burdened human subjects might open up a novel avenenue in a clinically-oriented CTCF studies associated with aberrant histone/DNA-methylation encompassing CTCF-bound ChIP-Seq peaks with 2xCTS elements in H3K27ac-marked Pol2-bound promoter-enhancer pairs capable of altering gene expression in the same way that we had previously found to act in context of Ctcf+/- mice analyzed in collaboration with Chris Kemp and Galina Filippova from the Fred Hutchinson Cancer Center in Seattle. Hence, it is possible that there is a common underlying patho-mechanism for the disorders caused by CTCF deletions distinct from complete loss of classic TSG CTCF functions reported for the first time by the same team. Therefore, similar pathology-associated mechanisms seem to underlie both human and mouse genetic disorders caused by insufficient CTCF dosage exclusive of additional ZnF mutations which, even in tumors with 16q22/CTCF LOH, would cause a complete CTCF loss leading to death rather than a partial loss of DNA-CTCF interactions caused by in vivo selection of viable single a.a. substitutions within the multivalent 11 ZnF CTCF DBD that were characterized first in CTCF (1996) and found later on (2002) to be recapitulated in the CTCF-derived paralog named "BORIS" (an acronym for "Brother Of the Regular of Imprinted States"). Next, the original discovery and further studies of the binary 2xCTS code begun to challenge a widespread misconception in the current literature claiming that all CTCF sites are equivalent to each other, with a single CTCF molecule bound at a single CTS sequence in spite of the fact that CTS elements with different genomic coordinates may contain either one or two adjacent DNase I footprints over single or dual CTCF motifs without any homologies necessary for reliable motif-based predictions. The functional and chromatin structural features of enhancer/promoter-associated 2xCTS-elements are distinct from those of 1xCTS-containing regions bound by CTCF monomers mostly within intronic and long inter-genic zones along chromosomal mouse and human DNA. We suggest that these 2 previously overlooked classes of CTCF binding regions may have different roles in regulating diverse chromatin-based phenomena, and may impact our understanding of heritable epigenetic regulation in cancer cells and normal germ cells. For instance, non-random retention of sperm nucleosomes placed selectively into protamine-free loci was found to be pre-determined by the nt context of the same 2xCTS-containing CTCF elements that are normally co-bound by both CTCF & BORIS 11 ZF paralogs, which are normally co-expressed in post-meiotic round spermatids. Moreover, CTCF and the cohesin complex are widely recognized as key players in the establishment and maintenance of 3D genome architecture in all mammalian cells. These proteins are not just well known in the scientific community but have recently entered the popular press including March issue of Scientific American at https://www.scientificamerican.com/article/untangling-the-formation-of-dna-loops. Taken together, our results provide a global view of chromatin dynamics and a resource for studying long-range control of gene expression in distinct human cell lineages, as well as explain why from a multitude of Transcription Factors, only CTCF has been recognized as a universal and possibly irreversable epigenetic mark present in all cell types at functionally distinct regions in order to orchestrate non-random positioning of modified histones and methylated DNA at hundreds of thousands of CTCF-associated DNA sequences catalogued by the ENCODE Consortium at www.factorbook.org/human/chipseq/tf/CTCF.