袁小兵
论文题目：Cdc42和RhoA可介导细胞外因子对轴突的吸引及排斥作用
作者简介：袁小兵，男，1973年03月出生，1995年09月师从于中科院上海生命科学研究院
蒲慕明教授，于2001年10月获博士学位。
摘
要
神经系统是一个由数以千亿计的神经元相互联系组成的复杂网络。动物的各种基本生理活动如呼吸、心跳、分泌、感觉、反射、运动、乃至高等心理活动如情感、学习、记忆、思维等都有赖于各条设计精确的神经回路的参与及调控。这些复杂的回路由动物的遗传信息编码，在发育过程中逐渐建立起来，并在后天的使用中进一步得到优化。神经导向的研究领域有两方面的问题值得深入研究：1、有哪些细胞外因子对轴突生长具有导向作用，这些因子在什么发育时间分布在脑内什么部位，对哪一类神经纤维起导向作用？ 2、各种导向因子引导神经纤维生长和转向的细胞内信号机制是什么？研究神经导向的机制不仅能够揭示神经发育的奥秘，更能够对神经损伤再生的研究提供理论借鉴，为开发和设计有利于神经再生的药物提供新思路。
目前已经发现大量对神经轴突生长具有导向作用的分子，这些分子可以分为两大类：一类分子固着在细胞膜表面或细胞外基质中，影响局部的神经纤维生长，这类因子包括ephrin, MAG (myelin-associated glycoprotein), Nogo等；另一类是分泌性分子，能扩散一定的距离并形成浓度梯度起作用，如netrin, slit, semaphorin 家族的大多数成员, 及各种神经营养因子等。神经轴突的前端有生长锥（growth cone）的结构起到对环境信号的探测作用。神经生长锥表面存在各种导向因子的受体，它们特异地识别环境中各种因子，并向细胞内传递吸引（attractive）或是排斥(repulsive) 的信号，从而引导神经纤维沿特定路线生长。无论是怎样的方向信号，最终都要通过调节生长锥内的细胞骨架的重组来改变神经的生长方向。这些生长锥中的细胞骨架成分主要包括微管（microtubule）, 微丝（microfilament 或actin filament）和肌球蛋白（myosin）等。微管通常只延伸到生长锥的基部，对神经纤维起稳定作用。肌动蛋白(actin) 呈束状（f-actin）分布在丝状伪足(filopodia)中，而呈网状分布在片状伪足(lamelipodia) 中。肌球蛋白通常与肌动蛋白结合成束形成张力纤维丝（stress fiber）。 张力纤维一端终止于局部粘联复合物（focal adhesion complex）, 另一端连接到微管束。这些骨架成分影响轴突导向的观点已被普遍接受，但各种细胞骨架成分究竟如何重组从而造成轴突转向目前尚无定论。一种假说认为，当轴突生长锥探测到方向性的信号时，生长锥内的 actin成分发生不对称聚合，filopodia及lamelipodia向一侧优先延伸，进而轴突内微管也朝同侧延伸，使生长方向进一步稳定下来。肌球蛋白在轴突导向中的作用被认为是通过其收缩引起轴突生长锥的回缩（retraction）及塌陷（collapse）。
Rho 家族的小分子鸟苷酸三磷酸酶（Small GTPases of Rho family, Rho GTPases）是对细胞骨架的调节作用起到关键作用的分子。许多已知对细胞骨架有调节作用的蛋白如cofilin, profilin, gelsolin等都是Rho GTPase的下游效应物。在该家族中，目前了解最清楚的有Cdc42, Rac1 和RhoA三种。Cdc42和Rac促进细胞内肌动蛋白聚合, 分别有利于丝状伪足和片状伪足的形成; 而RhoA可能具有与Cdc42 和Rac相反的效果, 通常造成细胞边缘和突起的收缩及胞体变圆。通过调节细胞骨架, Rho GTPases影响细胞分裂，细胞迁移，神经生长，细胞分泌，胞内囊泡运输等各种过程。在体（in vivo）和离体(in vitro)的许多研究表明，Rho GTPases 的突变影响神经细胞的形态、神经纤维的生长及其投射。 这提示Rho GTPases可能参与轴突的导向过程，但其机制尚有待研究。
在本实验中，我们使用由蒲慕明先生创立的轴突生长锥转向分析方法研究了Rho GTPases在两种脑内分泌的因子, 脑源性神经营养因子（Brain-derived neurotrophic factor, BDNF），溶血磷脂酸（Lysophosphatidic Acid, LPA）, 及一种直接影响细胞内钙离子浓度的药物ryanodine, 对爪蟾脊髓运动神经元轴突导向过程中所起作用。BDNF是一种在脑内广泛表达的神经营养因子，具有促进神经细胞生长、存活、再生、迁移的作用。离体和在体的研究都表明，BDNF对神经轴突可能是一种化学吸引性因子（chemoattractant），它通过受体Tyrosin Recptor Kinase B(TrkB)起作用。从BDNF能促进神经细胞生长的特性来看， Cdc42 或Rac 有可能在它的信号通路下游中起作用。 Ryanodine是一种最初从植物中提取出来的化合物。它被证明具有调节胞内钙库释放的作用。在低浓度时，ryanodine结合在细胞内钙通道上并打开内钙通道，在高浓度时，ryanodine使内钙通道失活。在细胞外使用ryanodine并形成浓度梯度，则在低浓度时产生吸引作用而在高浓度时产生排斥作用。LPA已被证明是一种磷脂类营养因子。在脑内, LPA由分裂后神经元分泌，影响神经前体细胞的形态和增殖。LPA 受体是一类七次跨膜的G蛋白耦联受体Ga12/13 或Gq。该受体的激活促使细胞内RhoA活性增加，促进细胞内张力纤维及局部粘连的形成。 在神经细胞中，已知LPA激活RhoA导致神经突起生长的抑制及生长锥塌陷。
用生长锥转向分析方法我们观察到 BDNF和LPA能分别对轴突生长锥产生吸引和排斥作用。ryanodine可以模拟BDNF的吸引作用。为进一步研究Rho GTPases在轴突转向中的作用，我们将Cdc42及RhoA的显性失活突变基因(dominant negative, DN)及组成性激活突变基因（constitutively active, CA）注射到爪蟾胚胎2-4细胞期的一个细胞中，使它们在胚胎神经元中表达，并观察这些突变基因对BDNF，LPA，ryonodine诱导的轴突转向的不同影响，从而推测Rho GTPases 是否直接参与并介导轴突导向。结果发现，表达DN-Cdc42或CA-Cdc42 能够阻断BDNF 及低浓度ryanodine的浓度梯度对轴突生长锥的化学吸引作用。LPA 的浓度梯度造成的生长锥排斥反应在神经元内表达DN-RhoA后被阻断，表达CA-Cdc42则使之转化为吸引反应。在PKA受抑制的情况下，BDNF能够诱导轴突排斥反应，该排斥反应在表达 DN-Cdc42 或 DN-RhoA 后均被逆转为吸引反应。对于高浓度ryanodine的浓度梯度所造成的排斥反应，DN-Cdc42, CA-Cdc42和DN-RhoA均能将其阻断。 这些结果初步证明Cdc42可以介导BDNF和低浓度ryanodine的吸引性信号，而RhoA能够介导LPA及高浓度ryanodine的排斥反应; 并且，Cdc42和RhoA两条信号通路之间存在相互crosstalk, 这种crosstalk影响到生长锥排斥反应。进而，细胞整体孵育在RhoA下游激酶ROCK的抑制剂Y-27632中则失去对LPA的排斥反应。而当给予生长锥一个Y-27632的浓度梯度时，生长锥呈显著的吸引性反应。这提示神经生长锥内RhoA活性的不对称性对引起生长锥转向是充分的。可能的机制是，当生长锥内一侧的RhoA活性较另一侧高时，丝状伪足在RhoA活性低的一侧受到的抑制较弱，延伸更快，因此生长锥朝向该侧转向。虽然我们目前还没有一个直接操纵细胞内Cdc42或其下游分子活性的方法，但从DN-Cdc42和CA-Cdc42均阻断到BDNF 和ryanodine的吸引作用来看，我们推测，Cdc42的活性梯度很可能也能够决定生长锥转动方向，即丝状伪足在Cdc42活性高的一侧优先延伸，生长锥朝向该侧转动。进一步用药理学方法发现，两种能够降低myosin活性的药物， Y-27632和ML-7都能够阻断排斥性生长锥转向，对吸引性转向反应没有影响。这说明myosin产生的回缩力在排斥性神经导向中起关键作用。由于myosin的活性能够受到Cdc42和RhoA的调节，因此，两种GTPases 能够通过对myosin的汇聚作用相互crosstalk。我们的实验结果还强烈支持“丝状伪足不对称的生长产生轴突生长锥转向”的观点。未来一个重要的研究方向是揭示细胞外及细胞内的信号是如何调节Rho GTPases 并告诉生长锥究竟采用哪一条信号通路，从而产生轴突生长锥不同反应的。
我们对神经导向的细胞和分子机制的认识较十年前已经有了长足的进展。仍有许多未解之谜有待揭开。特别希望再次指出的是，发育过程中的神经生长及导向过程与神经损伤后的再生及功能性神经回路重建过程有许多相似之处。克服成年人体神经损伤后的再生障碍是医学领域的世纪难题。每一天都有许许多多的人不幸在各种事故中遭受神经损伤，造成肢体瘫痪或神经功能紊乱。神经生长和导向机制的研究将提高我们对神经损伤后修复、再生机制的认识，帮助我们找到克服成年人中枢神经损伤后神经再生障碍的新方法。
Abstract
The nervous system is a super complicated network composed of hundreds of billions of neurons connecting with each other. Nearly all animal behaviors, from the general physiological processes including the breath, heartbeat, gland secretion, sense, reflex, motion, to the high level brain activities, such as learning and memory, feeling, consciousness, are controlled or adjusted by various precisely wired neural circuits, which are designed by the animal genes, formed gradually during development, and further refined during the postnatal usage. Two main questions deserve the detailed study in the axon guidance research: First, what factors that exist in proper special and temporal pattern guide axons from different origins? Second, what are the intracellular signaling mechanisms for the axon guidance triggered by these factors? The study of axon guidance will not only lead to the discovery of the mystery of nerve development, but will also shed new light on the study of nerve regeration after injury, and lend new strategy for the design and development of new drugs for the therapy of nerve injuries.
Many axon guidance molecules have been identified and they can be categorized into two groups: one group of molecules are normally membrane proteins or are proteins that bind to the extracellular matrix. These proteins, including ephrins, MAG, Nogo, can influence the local axon extension. The other group of guidance cues are secreted molecules and can form a long range concentration gradient to be detected by growth cones, including netrin, slit, some semaphorin family members, as well as neurotrophins. The growing tip of an axon has the tiny motile structure, named growth cone, which explores the extracellular environment. Receptors to the guidance factors exist on the growth cone surface, on binding the specific guidance factors in the surrounding tissue. They transduce attractive or repulsive signal cascades into the cell and lead the nerve growth to specific direction. No matter through which processes, the directional signaling should adjust the rearrangement of cytoskeleton in the growth cone to change the direction of axon extension. The three main cytoskeleton elements in the nerve growth cone are microtubule, actin microfilament, as well as actomyosin. Microtubules normally extend to the basal region of the growth cone and stabilize the axons. Actin filaments can form long and thin bundles in the filopodia, and can also exist as the actin meshwork in the lamelipodia. Myosin is often coupled to f-actin bundle to form the stress fiber, which connect the focal adhesion complex in one end and the microtubule bundle in the other end. It is widely accepted that cytoskeleton is involved in axon guidance processes, while it remains unknown as to how these cytoskeleton elements are rearranged to turn the nerve growth cone. One simplified hypothesis is that, when the growth cone detected the directional signal, actin will polymerize asymmetrically in the growth cone, so that the filopodia and lamelipodia will extend preferentially to one side, microtubule will further follow and stabilize this growing direction. The growth cone has thus been turned. The function of myosin was often considered to generate contractility and to cause the growth cone retraction or collapse.
Small GTPases of Rho family play critical roles the cytoskeleton regulation. Many molecules that are known to regulate the cytoskeleton, such as cofilin, profiling, gelsolin and Arp2/3, are the downstream effectors of Rho GTPases. The three main family members that are well studied are Cdc42, Rac1, and RhoA. Cdc42 and Rac1 have similar effects on cytoskeleton. They both promote the actin polymerization in the cell and profit the formation of filopodia and lamelipodia respectively. RhoA’s effect may be opposite to Cdc42 and Rac, which increases the contraction of cell periphery or cell processes, and cause the cell rounding. Through their regulation on cytoskeleton, Rho GTPases are involved in many cellular processes, including cell division, cell migration, nerve growth, cell secretion, and vesicle transport. Many in vitro and in vivo studies have shown that Rho GTPase mutations change the neuronal morphology, the nerve growth, as well as the axon projection, suggesting a role played by Rho GTPases in axon pathfinding, while how these happen awaits further clarification.
In the present study, we used the growth cone turning assay developed by Mu-ming Poo to examine the role of Rho family GTPases in mediating axon guidance in Xenopus spinal neurons by brain-derived neurotrophic factor (BDNF), and Lysophosphatidic acid (LPA), which are secreted factors existing in the developing brain, as well as ryanodine, a drug that regulates the internal calcium store.
BDNF is a neurotrophic factor widely expressed in the brain, which promotes nerve growth, cell survival, axon regeneration, and neuronal migration. In vivo and in vitro studies have shown that BDNF is a candidate chemoattractant to nerve growth cones which works through its specific receptor Trk B. Since BDNF promotes nerve growth, we guess that Cdc42 and Rac may function in its signaling processes. Ryanodine is a naturally occurring alkaloid which regulates the calcium release from the internal calcium store. At low concentrations, ryanodine binds to and opens the internal calcium channel, while at high concentrations, it blocks the intracellular calcium channels. When a gradient of ryanodine is applied to the growth cone, it triggers attraction at low concentraton and repulsion at high concentration. Lysophosphatidic acid (LPA), a bioactive lipid known to induce proliferative and/or morphological effects on neurons, has been proposed to be an extracellular signal involved in early neurogenesis. It acts through specific G protein-coupled, seven-transmembrane domain receptors Ga12/13 or Gq, and increases intracellular RhoA activity. In neuronal cells, activation of RhoA by LPA results in axon growth inhibition and growth cone collapse.
Using the growth cone turning assay, we found that gradients of BDNF and LPA can induce attraction and repulsion to growth cones respectively. Low concentration ryanodine can mimic BDNF to induce growth cone attraction and high dose ryanodine gradient is repulsive to the growth cone. To further study the role of Rho GTPases in growth cone turning, GTPase mutants，either dominant negative (DN) or constitutive active (CA), were expressed in Xenopus spinal neurons by injection of the cDNA of the GFP-fusion protein into one of the blastomeres of Xenopus embryos at 2- or 4-cell stage. The effects of these mutants on the growth cone turning triggered by BDNF, LPA and ryanodine were observed to determine whether Rho GTPases are directly involved in axon guidance. We found that, expression of either DN-Cdc42 or CA-Cdc42 in cultured Xenopus spinal neurons abolished the chemoattractive turning of their growth cones in a gradient of BDNF, and low concentration ryanodine, whereas chemorepulsive turning induced by a gradient of LPA was abolished by the expression of DN-RhoA and converted to chemoattraction by the expression of CA-Cdc42. Conversely, repulsive turning induced by BDNF in the presence of an inhibitor of cAMP pathway was also converted to attraction in DN-Cdc42 or DN-RhoA expressing neurons. High dose ryanodine gradient was rather ineffective in triggering any turning response in neurons expressing DN-Cdc42, CA-Cdc42 or DN-RhoA. These data suggest that Cdc42 can mediate attractive signal triggered by a gradient of BDNF or low concentration ryanodine, whereas RhoA can mediate repulsion of LPA, and that there is crosstalk between Cdc42 and RhoA pathways, which only influence the growth cone repulsion. Furthermore, the uniform presence of Y-27632, the specific inhibitor of ROCK, which is the direct RhoA downstream kinase, blocked the LPA-induced repulsion, and a gradient of Y-27632 by itself is capable of inducing attractive turning，suggesting that asymmetric RhoA or ROCK activity in the growth cone is sufficient to trigger growth cone turning toward the lower RhoA activity side. The possible mechanism is that, when there is an asymmetry of RhoA activity in the growth cone, the filopodia at the lower RhoA activity side will extend faster then the other side because of lower inhibition or retraction, and hence the growth cone will turn towards this side. Although we have not a tool to directly modulate the cellular Cdc42/Rac activity at present, nevertheless, from the abolish of BDNF-triggered attraction by either DN-Cdc42 and CA-Cdc42, we propose that an intracellular Cdc42 activity gradient may also be sufficient to determine the turning direction, say, filopodia extend preferentially at the higher Cdc42 activity side, and the growth cone turn towards this direction in the hence. Further pharmaceutical studies showed that, Y-27632 and ML-7, two drugs that are capable of inhibiting myosin acticity, can block chemorepulsion triggered by gradient of either LPA or high dose ryanodine or BDNF (in the presence of PKA inhibitor), although they have no influence on chemoattraction. This implies a key role played by myosin contractility during chemorepulsion. Since the myosin activity can be up-regulated by both Cdc42 and RhoA, these two Rho GTPases can thus crosstalk through their convergent effect on myosin. Our data also gave strong support to the hypothesis that asymmetric filopodial elongation determines the growth cone turning. One important future direction is to discover how the extracellular and intracellular signals regulate different Rho GTPases, dictating the growth cone to choose which pathway, and resulting in what kind of turning.
Our knowledge about the molecular and cellular mechanisms for axon pathfinding has been greatly enlarged during the past ten years. However, many mysteries remain to be disclosed. One point that deserves special attention is that, many common events may happen both in axon growth and guidance during brain development, and in the regeneration of the functional nerve circuits after nerve injury. It is one of the most difficult problems for centuries in clinical medicine to overcome the regeneration defects after nerve injury. Every day, many unfortunate persons suffer nerve injury from various accidents, resulting in paralysis of the bodies or dysfunction of the nervous system. Studies in axon guidance mechanism will boost the discovery of new methods in the therapy of nervous system injuries.
