DNA Folding in Chromatin and Interaction with Transcription Factors

在2013-2014财政年度，我们进一步努力阐明指导核小体旋转和翻译定位的DNA序列模式。特别是，我们开发了一种新的DNA穿线算法，可以准确预测核小体在体外和体内的精确定位（与F. Cui, Rochester institute of Technology， NY）。我们还对许多双链DNA片段进行了全原子能量最小化，这些片段经历了类似于在结晶核小体中观察到的构象转变。这种结合的方法使我们朝着理解基因组DNA中编码的核小体代码迈出了重要的一步。核小体中DNA的折叠伴随着相邻碱基对的横向位移，这通常被忽略。然而，我们发现，被称为滑动的剪切变形在DNA折叠中扮演的角色比之前想象的要重要得多。首先，在DNA局部各向异性弯曲处观察到的横向滑动变形定义了其在染色质中的超螺旋轨迹。其次，核小体上DNA变形的计算成本是序列特异性的：在最佳位置的序列中，最容易变形的碱基对步骤（CA:TG和TA）发生在大的正滑动和负滚动位点（DNA强烈弯曲或扭成小槽的地方）。在这里，我们结合了“真实”DNA的所有自由度，从而超越了传统模型的限制，忽略了碱基对的横向滑动位移。请注意，我们的结果与体外序列选择（SELEX）实验非常一致。对不同gc含量序列的核小体定位的成功预测显示了我们基于DNA变形能计算的结构分析的潜在优势。最近，我们开发了一种新的计算方法来计算高阶染色质组织的拓扑多态性，即所谓的30nm原纤维。(这很重要，因为与30纳米纤维的结构有关的问题仍然没有明确的答案。最可能的是，最初提出的螺线管型纤维可能只形成于L= 50bp或更长的核小体间连接体。当连接体L小于或等于30bp时（与酵母和人类神经元中的染色质相对应），双起点核小体纤维就形成了。我们分析了连接体L=13至37 bp的双起始染色质纤维。通过优化纤维能量与超螺旋参数，我们发现纤维中有两种类型的拓扑转变：一种是由连接子DNA扭曲的突变引起的，另一种是由连接子的交叉引起的。(第一次转变的特征是DNA连接数的变化，δ （Lk) = 1，第二次转变的特征是δ (Lk) = 2。）据我们所知，这种染色质纤维的拓扑多态性在之前发表的计算中没有报道。重要的是，连接剂L = 10n和10n+ 5bp的光纤的最优配置在拓扑结构上是不同的。我们的结果与实验观察一致，如60-70度的倾角（核小体盘和纤维轴之间的角度），螺旋上升，直径和纤维的左旋性。此外，我们还做出了几个可验证的预测，其中L = 10n和10n+5 bp纤维中存在不同程度的DNA超卷曲，两种纤维的刚度不同，以及酵母基因组不同部分的转录水平与局部NRL之间存在相关性。为了验证这些预测，我们计划进行两种类型的实验。第一项将是测量含有“601”核小体阵列的质粒的连接数差异，这些核小体被连接体L = 20和25bp分开（与宾夕法尼亚州立大学的S. Grigoryev合作）。我们还将分析酵母转录激活时核小体定位的重排。使用上述MNase/exoIII酶切的组合将是绘制高分辨率核小体的必要条件。

During the fiscal year 2013-2014, we extended our efforts to elucidate the DNA sequence patterns guiding rotational and translational positioning of nucleosomes. In particular, we developed a novel DNA threading algorithm correctly predicting positioning of nucleosomes precisely mapped both in vitro and in vivo (in collaboration with F. Cui, Rochester Inst. of Technology, NY). We also ran all-atom energy minimization of numerous double-stranded DNA fragments undergoing conformational transitions similar to those observed in crystallized nucleosomes. This combined approach allowed us to make an important step forward, toward understanding the nucleosome code encripted in genomic DNA. The folding of DNA in nucleosomes is accompanied by the lateral displacements of adjacent base pairs, which are usually ignored. We have found, however, that the shear deformation, called Slide, plays a much more important role in DNA folding than was previously imagined. First, the lateral Slide deformations observed at sites of local anisotropic bending of DNA define its superhelical trajectory in chromatin. Second, the computed cost of deforming DNA on the nucleosome is sequence-specific: in optimally positioned sequences the most easily deformed base-pair steps (CA:TG and TA) occur at the sites of large positive Slide and negative Roll (where the DNA strongly bends, or kinks, into the minor groove). Here, we incorporate all the degrees of freedom of 'real' DNA, thereby going beyond the limits of the conventional model ignoring the lateral Slide displacements of base pairs. Note that our results are in remarkable agreement with the in vitro sequence selection (SELEX) experiments. The successful prediction of nucleosome positioning for sequences of various GC-content demonstrates the potential advantage of our structural analysis, based on calculations of the DNA deformation energy. Recently, we developed a novel computational approach for calculation of topological polymorphism of the higher-order chromatin organization, the so-called 30-nm fibril. (This is important because the questions related to the structure of the 30-nm fibril still remain unanswered unambiguously. Most likely, the solenoid-type fibers proposed initially may be formed only for the inter-nucleosomal linkers L=50 bp or longer. When the linker L is 30 bp or shorter, which corresponds to chromatin in yeast and in human neurons, the two-start nucleosome fibers are formed.) We analyzed the two-start chromatin fibers for linkers L=13 to 37 bp. By optimizing the fiber energy with respect to the superhelical parameters we found two types of topological transition in fibers: one caused by an abrupt change in the linker DNA twisting, and another caused by over-crossing of the linkers. (The first transition is characterized by change in the DNA linking number, delta(Lk) = 1, and the second one by delta(Lk) = 2.) To the best of our knowledge, this topological polymorphism of the chromatin fibers was not reported in the computations published earlier. Importantly, the optimal configurations of the fibers with linkers L = 10n and 10n+5 bp are topologically different. Our results are consistent with experimental observations, such as the inclination 60-70 degrees (the angle between the nucleosomal disks and the fiber axis), helical rise, diameter and left-handedness of the fibers. In addition, we make several testable predictions, among them existence of different degree of DNA supercoiling in the fibers with L = 10n and 10n+5 bp, different stiffness of the two types of fibers, and a correlation between the local NRL and the level of transcription in different parts of the yeast genome. To test these predictions, we are planning two types of experiments. The first will be measuring the linking number difference in the plasmids containing arrays of '601' nucleosomes separated by linkers L = 20 and 25 bp (in collaboration with S. Grigoryev, Penn State). We will also analyze rearrangement of nucleosome positioning upon activation of transcription in yeast. Using the combined MNase/exoIII digestion described above will be essential for mapping nucleosomes with the highest possible resolution.