MECHANISMS OF GATING AND SELECTIVITY OF AQUAPORINS

这个子项目是利用由NIH/NCRR资助的中心拨款提供的资源的许多研究子项目之一。子项目和调查员(PI)可能从另一个NIH来源获得了主要资金，因此可能会出现在其他CRISE条目中。列出的机构是针对中心的，而不一定是针对调查员的机构。水通道蛋白(AQP)(网址：http://www.ks.uiuc.edu/Research/aquaporins/))是一类膜通道家族，专门负责通过生物膜的快速水传导[42，43]。它们广泛地分布在各种形式的生命中。水通道蛋白通过调节细胞膜的透水性，在活细胞的水分动态平衡中发挥重要作用。用NAMD[44]对不同AQP的膜嵌入、完全水化模型的MD模拟揭示了这些通道选择功能的新机制。要在脂类和水的自然环境中正确描述这些蛋白质，需要100,000个原子或更多的系统大小，以及数十纳秒量级的模拟时间。我们研究了水通道蛋白的水孔和中心孔的形成机制[45，46]。与A.YOOL(U·Arizona)密切合作，首次提出cGMP诱导水通道蛋白的离子电导[47，48]，研究了cGMP结合引起的蛋白质构象变化。基于我们的模拟，提出了一种门控机制：其中一个细胞质环(环D)在控制中央孔对水和水合离子的可访问性方面起着关键作用[46]。在模拟过程中，cGMP的自由扩散使这个柔性环的构象受到高度扰动，主要是通过核苷酸与环的富含精氨酸的区域的强烈相互作用。四个单体的D环从中心孔收回，不仅在物理上疏通了中心孔的入口，而且还导致了一环孔衬里的构象变化，疏水残基形成了一扇门，堵塞了水的通道。携带这些疏水残基的螺旋立即连接到环D，很可能环D和这个螺旋的构象耦合是检测中心孔内cGMP结合的分子机制[46]。我们的合作者对一个双突变物种进行的实验测量成功地证实了精氨酸参与了门控机制，其中两个精氨酸已经被敲除[46]。通过与R.Neutze(瑞典查尔默斯理工学院)的密切合作，研究了植物水通道蛋白中磷酸化介导的水孔门控的机制。通道闭合状态的X射线结构被用来研究由磷酸化引发的构象变化。在闭合(未磷酸化)形式中，环D通过与蛋白质N端的氢键保持在关闭位置，植物水通道蛋白的环D比哺乳动物的环D长4-5个残基。用可视化程序VMD[49]对模拟结果的检查表明，一旦磷酸化，N末端和环D之间的连接就会断开，后者可以自由地经历大的构象变化。这些变化又通过两个互补的机制导致水孔的打开：1)环D从通道的细胞质口移位，以及2)疏水残基(Leu197)从孔[45]中收回。除了研究AQP的门控机制外，我们还解决了这些水通道的选择性问题。对来自同一物种的两个细菌AQP进行了比较研究[50]，即E.coliGlpF和AqpZ，它们在结构上高度同源，但功能不同。虽然AqpZ是纯水通道[51，52]，但GlpF也传导甘油。在早期的研究[53]中，我们计算了与甘油通过GlpF渗透有关的PMF。从分子动力学模拟中计算准确的自由能分布是一个重要的挑战性问题。该资源与几个理论小组合作，如Chris Chipot(法国)和Ioan Kosztin(美国密苏里州)，以克服与此类计算相关的技术困难。虽然AqpZ不是甘油通道，但在SMD模拟中实现了甘油通过AqpZ的人工诱导通道，由此我们构建了甘油沿孔传导的PMF，并确定了在正常条件下使AqpZ对甘油不渗透的障碍[50]。结果表明，AqpZ和GlpF的主要结构差异不仅存在于AqpZ的选择性过滤区域，而且还存在于整个通道中。此外，AqpZ和GlpF底物选择性的差异可能不仅仅是因为那些直接排列在通道上的残基[50]。要将水通道转换为甘油通道，人们可能必须识别不直接与外来物相互作用但仍然通过移动和倾斜形成孔的螺旋来控制通道直径的遥远残基[50]。

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. Aquaporins (AQPs) (URL: http://www.ks.uiuc.edu/Research/aquaporins/) are a family of membrane channels specializing in rapid water conduction across biological membranes [42, 43]. They are widely distributed in all forms of life. Through modulating water permeability of cellular membranes AQPs play an important role in water homeostasis of living cells. MD simulations of membrane-embedded, fully hydrated models of various AQPs with NAMD [44] have revealed novel mechansims for selective function of these channels. A proper description of these proteins in their natural environment of lipid and water requires system sizes of 100,000 atoms or more, and simulation times on the order of tens of nanoseconds. We have investigated the mechanisms that gate the water pores and the central pore of AQPs [45,46]. In close collaboration with A. Yool (U. Arizona) who was the first to propose cGMP-induced ion conductivity of AQPs [47,48], protein conformational changes induced by binding of cGMP were studied. Based on our simulations, a gating mechanism has been proposed: one of the cytoplasmic loops (loop D) plays a critical role in controlling the accessibility of the central pore to water and hydrated ions [46]. The conformation of this flexible loop is highly perturbed by freely diffusing cGMP during the simulations, mainly through strong interactions of the nucleotide with an arginine-rich region of the loop. Retraction of the D loops of the four monomers away from the central pore not only physically unblocks the entrance of the central pore, but also results in conformational changes of a ring of pore-lining, hydrophobic residues that form a gate and block the access of water. The helix bearing these hydrophobic residues is immediately connected to loop D, and it is very likely that the conformational coupling of loop D and this helix is the molecular mechanism of detecting cGMP binding inside the central pore [46]. The involvement of arginines in the gating mechanism was successfully verified by experimental measurements performed by our collaborator on a double mutant species in which two of the arginines had been knocked out [46]. The mechanism of phosphorylation mediated gating of water pores in a plant AQP was investigated through a close collaboration [45] with the group of R. Neutze (Chalmers U. Tech., Sweden). The x-ray structure of the channel in its closed form was used to study conformational changes triggered by phosphorylation. In the closed (unphosphorylated) form, loop D, which is 4-5 residues longer in plant AQPs vs. mammalian ones, is held in its closing position through hydrogen bonds with the N-terminus of the protein. Examination of simulation results using the visualization program VMD [49] shows that upon phosphorylation, the connection between the N-terminus and loop D breaks, and that the latter is free to undergo large conformational changes. These changes, in turn, result in the opening of the water pores through two complementary mechanisms: 1) displacement of loop D from the cytoplasmic mouth of the channel, and, 2) retraction of a hydrophobic residue (Leu197) from the pore [45]. Besides the investigation of gating mechanisms of AQPs, we also tackled the problem of selectivity in these water channels. A comparative study [50] was performed on two bacterial AQPs from the same species, i.e., E. coli GlpF and AqpZ, which are structurally highly homologous, but functionally different. While AqpZ is a pure water channel [51,52], GlpF also conducts glycerol. In an earlier study [53], we calculated the PMF associated with permeation of glycerol through GlpF. Calculation of accurate free energy profiles from MD simulations presents an important challenging problem. The Resource collaborates with several theoretical groups, such as Chris Chipot (France) and Ioan Kosztin (U. Missouri) to overcome technical difficulties associated with such calculations. Although AqpZ is not a glycerol channel, an artificially induced passage of glycerol through AqpZ was achieved in SMD simulations, from which we constructed the PMF of glycerol conduction along the pore,and identified the barriers that make AqpZ impermeable to glycerol under normal conditions [50]. The results indicate that a high barrier against glycerol permeation arises not only in the selectivity filter region of AqpZ, which is believed to account for the main structural difference between AqpZ and GlpF, but along the entire channel. Furthermore, the difference in substrate selectivities of AqpZ and GlpF may not be simply due to only those residues that directly line the channel [50]. To convert a water channel into a glycerol channel one may have to identify remote residues that do not directly interact with the permeant, but nevertheless control the channel diameter through shifting and tilting helices forming the pore [50].