Modeling Structural and Functional Mechanisms of TRP Channels

我们刚刚开始这个项目，所以进展有限。我们正在与谢尔盖·苏哈里夫和青空实验室合作，从酵母中开发TRPy1机械敏感通道的结构模型。这个项目是与这些小组在微生物机械敏感通道建模方面非常富有成效的合作的继续；即，我们与他们合作开发大型机械敏感通道MSCL的门控机制模型。我们的MSCL模型的主要功能已经得到了许多小组的验证，目前已被接受。这个项目对我们来说也很自然，因为我们已经对同源电压门控通道的结构和功能机制进行了20多年的建模。我们已经将TRPy1(靶蛋白)序列与已知晶体结构的钾通道的序列进行了初步比对。这些晶体结构被用作同源建模的模板结构。电压门控通道由两个跨膜区组成；一个位于中心的成孔区由S5-P-S6段形成，另一个更外围的电压敏感区由S1-S4段形成。TRPy1在开放和闭合构象中的造孔结构域(S5-P-S6段)已经建立了初步模型，我们很快也应该有电压敏感结构域(S1-S4)的模型。然而，靶序列和模板序列太远，以至于比对的准确性值得怀疑，而且包括离子选择性区域在内的一些部分差异太大，以至于K+通道结构不能用作模板。我们正在使用过去被证明成功的替代方法来模拟这些地区。我们计划同时使用计算和实验方法来测试和改进初步模型。计算方法将涉及对嵌入在类脂膜中的模型结构进行广泛的分子动力学模拟，膜的两侧和中央孔中都有水和离子。将使用苏哈雷夫实验室的安德烈·阿尼什金开发的一种新的计算方法。此方法使用可以在模拟过程中打开和关闭的软对称约束。模拟最初将在没有对称约束的情况下进行，以更好地允许结构向较低的能量状态移动，然后将使用对称约束来恢复模型的四重对称性。这些模拟将主要基于不同的路线，用许多可选的起始模型来执行，以试图确定更好的模型。Kung实验室已经在TRPy1上进行了大量的突变实验，并发现了几个显著改变通道门控特性的突变。我们将检查我们的初步模型，以确定它们是否与这些数据一致。接下来，还将进行额外的诱变实验，以测试各种模型。这些实验可能包括交叉连接实验，以确定预测在特定状态下相互作用的残基对是否确实在这些状态下相互作用。它们还将涉及模型预测的残基，这些残基在确定通道的离子选择性方面至关重要。在开发蛋白质的结构模型时，最困难的部分通常是开发一个家族成员的第一个模型。之后，通常具有相似主干结构的其他相关成员可以更容易地使用相对标准的同源建模方法来开发。我们选择TRPY1首先进行建模，因为我们的实验合作者可以测试初步模型。一旦开发和测试了令人满意的TRPy1通道模型，我们将把这些模型扩展到生物医学上更感兴趣的人类Trp通道。

We have just started this project, so the progress is limited. We are collaborating with Sergei Sukharevs and Ching Kungs laboratories to develop structural models of the TRPy1 mechanosensitive channel from yeast. This project is a continuation of a very productive collaboration with these groups on modeling microbial mechanosensitive channels; i.e., we worked with them to develop models of the gating-mechanism of the large mechanosensitive channel MscL. The principal features of our MscL models have been validated by numerous groups, and are currently accepted. The project is also natural for us since we have been modeling the structures and functional mechanisms of homologous voltage-gated channels for over two decades. We have developed tentative alignments of the TRPy1 (target protein) sequences with those of potassium channels that have known crystal structures. These crystal structures are used as template structures for homology modeling. Voltage-gated channels are composed of two transmembrane domains; a centrally located pore-forming domain formed by S5-P-S6 segments and a more peripheral voltage-sensing domain formed by S1-S4 segments. Preliminary models have been constructed of the pore-forming domains (S5-P-S6 segments) of TRPy1 in open and closed conformations, and we should soon have models of the voltage-sensing domain (S1-S4) as well. However, the target and template sequences are so distant that the accuracy of the alignment is questionable, and some portions, including the ion selective region, is so different that K+ channel structures cannot be used as a template. We are using alternative methods that have proven successful in the past to model these regions. We plan to use both computational and experimental approaches to test and improve the preliminary models. The computational methods will involve extensive molecular dynamic simulations of the model structures embedded in a lipid membrane with water and ions on each side of the membrane and in the central pore. A new computational method developed by Andriy Anishkin of the Sukharev lab will be used. This method uses soft symmetry restraints that can be turned on and off during the simulations. The simulations will be performed initially without symmetry restraints to better allow the structure to move toward lower energy states, and then the symmetry restraints will be used to restore the four-fold symmetry of the model. These simulations will be performed with a number of alternative starting models based primarily on different alignments to try to identify the better models. The Kung lab has already performed numerous mutagenesis experiments on TRPy1, and has identified several mutations that dramatically alter gating properties of the channels. Our preliminary models will be examined to determine whether they are consistent with these data. Next, additional mutagenesis experiments will be performed to test various models. These experiments will likely include cross-linking experiments to determine whether residue pairs that are predicted to interact in specific states do in fact interact in those states. They will also involve residues that are predicted by the models to be crucial in determining the ion selectivity of the channel. In developing structural models of proteins, the most difficult part is typically developing the first model of a member of a family. After that, other related members that normally have similar backbone structures can be developed more easily using relatively standard homology modeling methods. We have selected TRPy1 to model first because our experimental collaborators can test the preliminary models. Once satisfactory models of the TRPy1 channel have been developed and tested, we will extend the models to human TRP channels that are of more interest biomedically.