p53-induced Regulation of Transcription in the Chromatin Context

为了阐明p53-DNA复合物在溶液中的DNA轨迹，我们正在使用碘-125放射性探针（与I。Panyeland和R. Neumann，临床中心，NIH）。这种方法是基于分析的DNA链断裂所产生的电子发射放射性同位素，碘-125，掺入胞嘧啶的C5位置的衰变。DNA链断裂越弱，从放射性同位素到切割位点的距离越大。放射探测的主要优点是它适用于非常大的蛋白质-DNA复合物。特别地，该方法允许直接比较与p53核心结构域和野生型蛋白结合的DNA的构象，后者仍然超出常规方法如晶体学和NMR的范围。我们的研究结果表明，在与野生型p53的四聚体复合物中，共有的20-bp DNA片段（YYYRRR）的中心区域被弯曲成小沟（即，与我们的模型一致，并与p53四聚体结合到核小体）。DNA轨迹的详细可视化需要更多的放射探测数据。在不久的将来，我们预计将获得这样的数据的几个DNA序列，包括那些p53 RE激活细胞周期阻滞和凋亡基因（CCA和载脂蛋白基因）。比较p53位点在CCA基因和Apo基因中的分布，发现CCA位点位于靶基因转录起始位点（TSS）的2-3 kb处，而Apo位点大多聚集在TSS的1 kb处。注意，p53位点的这种分布是违反直觉的，因为p53与远端CCA位点的结合和相应CCA基因的诱导似乎比p53与近端Apo位点的结合和Apo基因的激活更有效。我们进一步表明，CCA位点的侧翼序列，具有中等或低GC含量（35- 55%GC），揭示了富含AT和富含GC的簇的强周期性，类似于在核小体DNA序列中观察到的，表明稳定定位的核小体可能在这里形成。这些核小体的预测旋转定位意味着p53 RE以有利于p53识别的弯曲构象暴露。换句话说，CCA位点附近的可弯曲DNA元件以这样的方式组织，即核小体DNA被预先形成用于p53四聚体结合。例如，p21 5-反应元件，体内最有效的p53 RE，与TSS分开2.5 kb，并且以与在结晶核小体中观察到的相同的有利构象弯曲。我们认为，暴露的p21和其他CCA网站加速的过程中p53结合在体内。反过来，p53将辅激活因子如p300/CBP和/或染色质重塑因子募集到启动子，从而促进染色质的打开并增加转录水平。相比之下，Apo位点位于极富含GC的区域（高达75- 80%GC）。这些序列的典型特征是多定位和相对容易的核小体重组，以及低H1水平。我们推测，这种动态环境干扰了p53寻找其同源结合位点，使其不那么有效。因此，两组p53反应元件的核小体组织的差异似乎是影响p53-DNA结合强度和p53靶基因诱导动力学的关键因素。我们的模型不同于早期的概念连接CCA和载脂蛋白基因的选择性激活的结合亲和力的RE的p53。相反，我们强调p53诱导的肿瘤抑制途径（细胞凋亡与细胞周期阻滞）的选择和相应的p53结合位点在染色质中的结构组织之间的直接相关性。我们增加了新的维度，现有的范例，相对定位和染色质环境的p53 REs。我们的计划不仅解释了上述情况下，但也提供了一个新的见解，数百个基因的激活的细胞机制的p53。我们特别注意了p53结合位点的全基因组分布及其与各种转座因子，特别是Alu重复序列的关系。首先，我们分析了迄今为止鉴定的160个功能性p53 RE，发现其中24个发生在重复序列中。这些重复序列相关的RE中有一半以上位于Alu元件中;它们位于内部RNA聚合酶III启动子的A盒和B盒附近。有趣的是，几种p53 RE与凋亡基因相关（例如，CASP-10）。此外，使用位置权重矩阵的方法，我们发现大约400，000个潜在的p53位点的Alu元件全基因组。这些位点位于与功能性p53 RE相同的Alu重复区，因此可以根据它们与框A或B的邻近程度分为两组。先前对Alu元件进行的核小体作图实验表明，来自这两组的p53位点具有不同的染色质环境，这对p53-DNA结合至关重要。最后，我们将p53位点与相应的Alu共有序列进行了比较，得出的结论是，这两组位点可能通过不同的机制进化，一组是由CG-to-CA：TG突变产生的;另一组显然预先存在于几个Alu亚家族的祖细胞中，例如AluSp和AluSx。值得注意的是，这种评估适用于功能性p53 RE和推定的BS。我们的观察可能是重要的理解p53调控网络在全基因组水平的演变。

To elucidate DNA trajectory in the p53-DNA complex in solution, we are using Iodine-125 radioprobing (in collaboration with I. Panyutin and R. Neumann, Clinical Center, NIH). This method is based on analysis of the DNA strand breaks produced by the decay of an electron-emitting radioisotope, Iodine-125, incorporated in the C5 position of cytosine. The weaker the DNA strand break the larger the distance from the radioisotope to the cleavage site. The major advantage of radioprobing is its applicability for very large protein-DNA complexes. In particular, this method allows direct comparison of the conformations of DNA bound to the p53 core domain and to the wild type protein, the latter still being beyond the scope of conventional methods such as crystallography and NMR. Our results indicate that in a tetrameric complex with wt p53, the central region of the consensus 20-bp DNA fragment (YYYRRR) is bent into the minor groove (that is, consistent with our model and with the p53 tetramer binding to nucleosome). The detailed visualization of the DNA trajectory requires more radioprobing data. In the near future, we anticipate obtaining such data for several DNA sequences, including those of the p53 REs activating cell cycle arrest and apoptotic genes (CCA and Apo-genes). Comparing distribution of the p53 sites in the CCA- and Apo-genes, we found that the CCA-sites are located 2-3 kb away from the transcription start sites (TSS) of the target genes, whereas most of the Apo-sites are clustered within 1 kb from TSS. Note that such a distribution of the p53 sites is counter-intuitive because the p53 binding to a distal CCA-site and induction of the corresponding CCA-gene appears to be more efficient than the p53 binding to a close Apo-site and activation of the Apo-gene. We further showed that the flanking sequences of the CCA-sites, with moderate or low GC content (35-55 % GC), reveal strong periodicity of the AT-rich and the GC-rich clusters, similar to that observed in the nucleosomal DNA sequences, suggesting that stable positioned nucleosomes are likely to form here. The predicted rotational positioning of these nucleosomes implies that the p53 REs are exposed in the bent conformation favorable for the p53 recognition. To put it differently, the bendable DNA elements in the vicinity of the CCA-sites are organized in such a way that the nucleosomal DNA is preformed for the p53 tetramer binding. For example, the p21 5-response element, the most effective p53 RE in vivo, is separated from TSS by 2.5 kb, and is bent in the same favorable conformation as observed in the crystallized nucleosomes. We suggest that exposure of the p21 and other CCA-sites accelerates the process of p53 binding in vivo. p53, in turn, recruits co-activators such as p300/CBP and/or chromatin remodeling factors to the promoters, thereby facilitating opening of chromatin and increasing the level of transcription. By contrast, the Apo-sites are located in extremely GC-rich regions (up to 75-80 % GC). Such sequences are typically characterized by multiple positioning and relatively easy reorganization of nucleosomes, as well as low H1 level. We hypothesize that this dynamic environment interferes with the p53 search for its cognate binding site and makes it less effective. Thus, the difference in nucleosomal organization of the two sets of p53 response elements appears to be a key factor affecting the strength of p53-DNA binding and kinetics of induction of the p53 target genes. Our model differs from the earlier concept connecting the selective activation of the CCA- and Apo-genes to the binding affinities of their REs to p53. Instead, we emphasize a direct correlation between the selection of p53-induced tumor suppression pathway (apoptosis versus cell cycle arrest) and structural organization of the corresponding p53-binding sites in chromatin. We add new dimensions to the existing paradigm, the relative positioning and chromatin environment of the p53 REs. Our scheme not only explains the above cases but also provides a new insight into the cellular mechanisms of activation of hundreds of genes by p53. We paid special attention to genome-wide distribution of putative p53 binding sites and their relationship to various transposable elements, in particular Alu repeats. First, we analyzed 160 functional p53 REs identified so far and found that 24 of them occur in repeats. More than half of these repeat-associated REs reside in Alu elements; they are located in the vicinity of Boxes A and B of the internal RNA polymerase III promoter. Interestingly, several Alu-residing p53 REs are associated with apoptotic genes (e.g., CASP-10). In addition, using a position weight matrix approach, we found approximately 400,000 potential p53 sites in Alu elements genome-wide. These sites are located in the same regions of Alu repeats as the functional p53 REs and thus can be divided into two groups depending on their vicinity to the Boxes A or B. The nucleosome-mapping experiments made on Alu elements earlier, suggest that the p53 sites from these two groups have different chromatin environments which is critical for the p53-DNA binding. Finally, we compared the p53 sites with the corresponding Alu consensus sequences and concluded that the two groups of sites probably evolved through different mechanisms one group was generated by CG-to-CA:TG mutations; the other group apparently pre-existed in the progenitors of several Alu subfamilies, such as AluSp and AluSx. Remarkably, this assessments holds both for the functional p53 REs and for putative BSs. Our observations may be important for understanding the evolution of the p53 regulatory network at the genome-wide level.