The discovery of tau by the Kirschner lab was based on its ability to associate with microtubules and to promote microtubule assembly (Weingarten et al., 1975). After the primary sequence of tau and MAP2 were reported and functional studies performed, it became clear that both proteins contained a similar microtubule binding domain in the carboxy terminal portion of the protein (Butner & Kirschner, 1991; Ennulat et al., 1989; Himmler et al., 1989; Lee et al., 1988; Lewis et al., 1988). However, most of the remaining tau sequence was distinct from that of MAP2. Given the importance of tau and its novel role in neurodegenerative disease, a natural question was the functional significance of its amino terminal domain. This motivated our lab to seek out new functions for tau that did not require or involve microtubule association.

Our lab found that in response to nerve growth factor stimulation, tau potentiated AP-1 transcription factor activation (Leugers et al., 2013; Leugers & Lee, 2010). We also determined that this effect, at least in part, was mediated by the ability of tau to potentiate the activation of MAPK. Our findings were made using PC12-derived cell lines. In addition, we ascertained that phosphomimetic mutations in tau that compromised microtubule binding (S262D/S356D) did not affect its effects on MAPK activation. Moreover, T231D, a mutation that reduced microtubule association, significantly increased the ability of tau to potentiate MAPK activation beyond the extent exhibited by wild type tau. These data suggested that independent of its microtubule binding function, tau was capable of affecting signal transduction during the early response to NGF; a similar effect was found with EGF (Leugers & Lee, 2010).

A relationship between tau and AP-1 had previously been suggested by the microarray analysis of tau knockout mice where it was reported that the genes with the highest levels of alteration were FosB and c-fos (Supplemental Data in, Oyama et al., 2004). FosB and c-fos are part of transcription factor AP-1 and have been implicated as regulators of cell proliferation and differentiation. The finding that tau knockout mice, at 8 weeks old, expressed higher levels of FosB and c-fos relative to wild type might seem to contradict our finding that a PC12-derived cell line with tau depletion had a lower level of NGF-induced AP-1 activation relative to control. However, establishing a tau depleted cell line would not face the same pressures as when establishing a tau knockout mouse where mouse viability and ability to breed would be critical. We speculate that the undifferentiated tau depleted cell line did not need to up-regulate fosB/c-fos to proliferate, but once NGF was added, defects in MAPK and AP-1 activation were evident. An important question is by what mechanism did tau promote activation of AP-1 and MAPK?

When one considers tau's ability to influence signal transduction, there are a number of possibly relevant tau interactors that have been identified. We reported an interaction between tau and src-family non-receptor tyrosine kinases that was mediated by the SH3 domain of Src or Fyn and the proline-rich region of tau just upstream of the microtubule binding domain (Lee et al., 1998). Identification of the specific prolines involved in the interaction has been reported (Lau et al., 2016; Reynolds et al., 2008). However, the likelihood of Fyn or Src being involved in tau's effect on MAPK is low since both the Fyn and Src SH3-tau interactions were reduced by at least 8-fold when the T231D/S235D mutation was tested using surface plasmon resonance (Bhaskar et al., 2005). It is likely that the T231D mutation was solely responsible since the testing of S235E later showed no effect on the ability of tau to bind Fyn SH3 (Reynolds et al., 2008). On the other hand, we have identified an interaction between tau and protein tyrosine phosphatase SHP2 and have found that the association, as determined by proximity ligation assays, was significantly increased in PC12-like cells responding to NGF; complexes between tau and activated SHP2 were also detected (Kim et al., 2019). Moreover, in PC12-derived cells responding to NGF, phospho-T231-tau-SHP2 complexes were found, consistent with the finding that a significant increase in phospho-T231-tau occurred after 5 min of NGF stimulation. In comparing T231D against the T231A construct, significantly more T231D-SHP2 complexes occurred relative to T231A-SHP2 complexes, suggesting that phosphorylation at T231 increased SHP2 interaction. Tau-SHP2 complexes were also found in tau transfected COS cells, where EGF stimulation increased tau-activated SHP2 complexes and localized them to membrane ruffles (Kim et al., 2019). Given that SHP2 participates in the early NGF and EGF responses of PC12 cells and is required for MAPK activation and neurite outgrowth (D'Alessio et al., 2007; Okada et al., 1996; Shi et al., 2000; Wright et al., 1997), the possibility that the tau-SHP2 interaction has a role in MAPK activation needs to be investigated.

Another tau interactor that may have a role in AP-1 activation is phosphatidylinositol 3-kinase (PI3K). Tau has been shown to interact with the SH3 domain of the p85 subunit of PI3K (Reynolds et al., 2008) and while the function of the interaction remains unclear, PI3K is activated by NGF in PC12 cells (Ashcroft et al., 1999) and the ability of PI3K to activate AP-1 has been reported (Fleegal & Sumners, 2003). Interestingly, using prostate cancer cells, co-immunoprecipitates of tau and PI3K have been found (Souter & Lee, 2009) and such complexes might also exist in PC12 cells. The role of tau in non-neuronal cancer cells, where cell cycle regulation has been altered, may have relevance toward a role for tau in EGF-induced proliferation. In addition, when one considers growth factor signaling mechanisms, upstream of PI3K, Ras, and SHP2 is the grb2-SOS complex (reviewed in Belov & Mohammadi, 2012; Mendoza et al., 2011; Schlessinger, 1994; Tari & Lopez-Berestein, 2001) and, strikingly, tau also interacts with one of the SH3 domains of grb2 (Reynolds et al., 2008). Grb2 is an adapter protein that is critical for many signaling pathways and mice that are grb2 deficient are embryonic lethal (Cheng et al., 1998). It would be vital to determine the functional significance of its interaction with tau.

An additional tau interactor to consider is phospholipase Cγ1 (PLCγ1). Tau antibodies co-immunoprecipitated PLCγ (Jenkins & Johnson, 1998) and in vitro, tau interacted with the SH3 domain of PLCγ1 (Reynolds et al., 2008). Since the activation of PLCγ1 occurs following NGF treatment of PC12 cells and is essential for MAPK activation (Rong et al., 2003; Stephens et al., 1994), the tau-PLCγ1 interaction might have a role in MAPK activation. Lastly, additional tau interactors with known roles in signal transduction, such as Abl and 14-3-3, have been identified (Agarwal-Mawal et al., 2003; Derkinderen et al., 2005; Hashiguchi et al., 2000). Abl and 14-3-3 both participate in the NGF response of PC12 cells, so there is no shortage of mechanisms to consider.

Given the number of signal transduction related tau interactors that have been reported, much remains to be done. In particular, besides assuring that the associations occur in cells, as has been demonstrated for Fyn, SHP2, PI3K, Syk, and PLCγ (Jenkins & Johnson, 1998; Kim et al., 2019; Lebouvier et al., 2008; Lee et al., 1998; Souter & Lee, 2009), it would be important to determine if the association affects the function of either the interactor or tau. Tau has been established as a substrate for tyrosine kinases (Fyn, Src, Syk, Lck, and Abl; (Derkinderen et al., 2005; Lebouvier et al., 2008; Lee et al., 1998; Lee et al., 2004; Scales et al., 2011), unpublished data) which strongly suggests that tau is engaged in signal transduction pathways. Tyrosine phosphorylated tau is also a substrate for SHP2 (Kim et al., 2019). In regards to the interactors, tau together with arachidonic acid, has been shown to activate PLCγ1 (Hwang et al., 1996) and tau has also been shown to increase the in vitro activity of Fyn as well as the activity of Src in transfected cells (Sharma et al., 2007). An additional issue that also needs more investigation is the effect of tau phosphorylation on interactions. For instance, for PI3K or PLCγ1, while the interactions between tau and their SH3 domains were decreased by phosphomimetic mutations in tau, the tau construct tested contained several phosphomimetic mutations to more precisely identify the critical site(s) (Reynolds et al., 2008); therefore, more tests are needed.

In envisioning that tau has a role in signal transduction, it is relevant to mention that the association of the amino terminus of tau with the plasma membrane of neuronal cells has been reported (Brandt et al., 1995), suggesting the association of tau with a subdomain of the plasma membrane, namely lipid rafts. Lipid rafts contain receptors that initiate signal transduction in response to hormones and growth factors and lipid rafts are thought to be critical toward a cell's response to extracellular signals (reviewed in Paratcha & Ibanez, 2002; Simons & Toomre, 2000; Tsui-Pierchala et al., 2002). Moreover, tau has been found in lipid rafts, where its presence in lipid rafts correlated with cellular responses (Hernandez et al., 2009; Kawarabayashi et al., 2004; Klein et al., 2002). An investigation of tau's role in signal transduction might involve its localization in lipid rafts.

Lastly, the ability of tau to interact with a number of tyrosine kinases and other proteins involved in signal transduction is likely to stem from its proline-rich domain, given that SH3 domains interact with PXXP motifs and tau contains seven PXXP motifs. With current advances in protein structure determination, the structure of tau in filaments and the model of tau-microtubule interactions have been reported using cryo-electron microscopy (Fitzpatrick et al., 2017; Kellogg et al., 2018; Zhang et al., 2019). However, the reported structures focused on the microtubule binding domain and did not comment on the structure of the proline-rich domain. It is not clear if investigators have ever crystallized the amino terminal domain (amino acids 1–241), given the importance of the microtubule binding domain in neurodegenerative diseases. The tau-microtubule binding model of Kellogg et al has shown that the microtubule binding repeat region of tau stretches along the surface of the microtubule (Kellogg et al., 2018), with no indication that the amino terminal domain folds back toward the microtubule binding domain. However, this does not rule out the possibility that when tau is detached from microtubules, an interactor might be capable of interacting with both the proline-rich domain and the basic microtubule binding domain of tau. In addition, when the Thr231 residue on tau is phosphorylated, it can occur as either cis-pT231 or trans-pT231 (Nakamura et al., 2012), indicating that tau structure at that part of the proline-rich domain closest to the microtubule binding domain can occur in two conformations, with non-pT231 tau contributing a third conformation. It will certainly be interesting to determine how tau structure is altered by phosphorylation of T231 and how interactions with proteins such as Fyn and SHP2 are affected by changes in the structure of the proline-rich domain caused by phosphorylation.

As we reflect on 50 years of tau research, while many questions have been answered, many more have been raised. The presence of abnormal tau in Alzheimer's disease, frontotemporal dementia, epilepsy, traumatic brain injury, and other neurodegenerative diseases has led to many investigations into the properties of abnormal tau, meaning tau isolated from diseased brains, abnormally phosphorylated tau, or tau bearing FTDP-17 mutations. However, the function of normal tau, beyond its ability to promote microtubule assembly, remains to be fully established. Here we have focused on just one possible avenue for investigation, with the thought that identifying new mechanisms through which tau acts would serve as natural starting points for disease related studies. During the next 50 years, we expect tau research to elucidate new functions for tau that will lead to new disease therapies.

Kirschner 实验室发现 tau 是基于其与微管结合并促进微管组装的能力（Weingarten 等，  1975）。在报道了 tau 和 MAP2 的一级序列并进行了功能研究后，很明显这两种蛋白质在蛋白质的羧基末端部分都含有相似的微管结合域（Butner & Kirschner，  1991；Ennulat 等，  1989；Himmler）等人，  1989；Lee 等人，  1988；Lewis 等人，  1988）。然而，大部分剩余的 tau 序列与 MAP2 的序列不同。鉴于 tau 蛋白的重要性及其在神经退行性疾病中的新作用，一个自然的问题是其氨基末端结构域的功能意义。这促使我们的实验室寻找不需要或不涉及微管关联的 tau 蛋白新功能。

我们的实验室发现，在对神经生长因子刺激的反应中，tau 蛋白增强了 AP-1 转录因子的激活（Leugers 等，  2013；Leugers & Lee，  2010）。我们还确定，这种效应至少部分是由 tau 增强 MAPK 激活的能力介导的。我们的研究结果是使用 PC12 衍生的细胞系得出的。此外，我们还确定 tau 蛋白中影响微管结合 (S262D/S356D) 的拟磷突变不会影响其对 MAPK 激活的影响。此外，T231D（一种减少微管关联的突变）显着增强了 tau 增强 MAPK 激活的能力，超出了野生型 tau 所表现出的程度。这些数据表明，独立于其微管结合功能，tau 能够在 NGF 的早期反应过程中影响信号转导。EGF 也发现了类似的效果（Leugers & Lee，  2010）。

tau 和 AP-1 之间的关系之前已通过 tau 敲除小鼠的微阵列分析得出，据报道，改变水平最高的基因是 FosB 和 c-fos（补充数据，Oyama 等人，2004 年 ））。FosB 和 c-fos 是转录因子 AP-1 的一部分，被认为是细胞增殖和分化的调节因子。与野生型相比，tau 基因敲除小鼠在 8 周大时表达了更高水平的 FosB 和 c-fos，这一发现似乎与我们的发现相矛盾，即 tau 缺失的 PC12 衍生细胞系具有较低水平的 NGF 诱导的 AP -1 相对于控制的激活。然而，建立tau蛋白耗尽的细胞系不会面临与建立tau蛋白敲除小鼠相同的压力，其中小鼠的活力和繁殖能力至关重要。我们推测未分化的 tau 耗尽细胞系不需要上调 fosB/c-fos 即可增殖，但一旦添加 NGF，MAPK 和 AP-1 激活的缺陷就很明显。一个重要的问题是tau蛋白通过什么机制促进AP-1和MAPK的激活？

当人们考虑 tau 影响信号转导的能力时，已经发现了许多可能相关的 tau 相互作用因子。我们报道了 tau 和 src 家族非受体酪氨酸激酶之间的相互作用，该相互作用由 Src 或 Fyn 的 SH3 结构域以及微管结合域上游的 tau 富含脯氨酸区域介导（Lee 等，1998  ）。已报道了参与相互作用的特定脯氨酸的鉴定（Lau 等人，  2016；Reynolds 等人，  2008）。然而，Fyn 或 Src 参与 tau 对 MAPK 影响的可能性很低，因为当使用表面等离子共振测试 T231D/S235D 突变时，Fyn 和 Src SH3-tau 相互作用均减少至少 8 倍（Bhaskar 等）等，  2005）。T231D 突变很可能是唯一的原因，因为后来 S235E 的测试显示对 tau 结合 Fyn SH3 的能力没有影响（Reynolds 等，  2008）。另一方面，我们已经确定了 tau 和蛋白酪氨酸磷酸酶 SHP2 之间的相互作用，并发现通过邻近连接测定确定，这种关联在响应 NGF 的 PC12 样细胞中显着增加；还检测到了 tau 和激活的 SHP2 之间的复合物 (Kim et al.,  2019 )。此外，在对 NGF 做出反应的 PC12 衍生细胞中，发现了磷酸化 T231-tau-SHP2 复合物，这与 NGF 刺激 5 分钟后磷酸化 T231-tau 显着增加的发现一致。在比较 T231D 与 T231A 构建体时，相对于 T231A-SHP2 复合物，出现明显更多的 T231D-SHP2 复合物，表明 T231 处的磷酸化增加了 SHP2 相互作用。在 tau 转染的 COS 细胞中也发现了 tau-SHP2 复合物，其中 EGF 刺激增加了 tau 激活的 SHP2 复合物，并将其定位于膜皱褶处 (Kim et al.,  2019 )。鉴于 SHP2 参与 PC12 细胞的早期 NGF 和 EGF 反应，并且是 MAPK 激活和神经突生长所必需的（D'Alessio 等人，  2007；Okada 等人，  1996；Shi 等人，  2000；Wright 等人） .,  1997 )，tau-SHP2 相互作用在 MAPK 激活中发挥作用的可能性需要研究。

另一种可能在 AP-1 激活中发挥作用的 tau 相互作用蛋白是磷脂酰肌醇 3-激酶 (PI3K)。Tau 已被证明与 PI3K p85 亚基的 SH3 结构域相互作用（Reynolds 等，  2008），虽然相互作用的功能仍不清楚，但 PI3K 在 PC12 细胞中被 NGF 激活（Ashcroft 等，  1999）并且已经报道了PI3K激活AP-1的能力(Fleegal & Sumners,  2003 )。有趣的是，使用前列腺癌细胞，发现了 tau 和 PI3K 的免疫共沉淀物（Souter & Lee，  2009），并且此类复合物也可能存在于 PC12 细胞中。tau 在细胞周期调节发生改变的非神经元癌细胞中的作用可能与 tau 在 EGF 诱导的增殖中的作用相关。此外，当考虑生长因子信号传导机制时，PI3K、Ras 和 SHP2 的上游是 grb2-SOS 复合体（Belov & Mohammadi,  2012综述；Mendoza 等人，  2011；Schlessinger,  1994；Tari & Lopez-Berestein） ,  2001）并且引人注目的是，tau 还与 grb2 的 SH3 结构域之一相互作用（Reynolds 等，  2008）。Grb2 是一种接头蛋白，对许多信号传导途径至关重要，而 grb2 缺陷的小鼠会导致胚胎死亡（Cheng 等人，  1998）。确定其与 tau 相互作用的功能意义至关重要。

另一个需要考虑的 tau 相互作用因子是磷脂酶 Cγ1 (PLCγ1)。Tau 抗体共免疫沉淀 PLCγ（Jenkins & Johnson，  1998），并且在体外，tau 与 PLCγ1 的 SH3 结构域相互作用（Reynolds 等，  2008）。由于 PLCγ1 的激活发生在 NGF 处理 PC12 细胞后，并且对于 MAPK 激活至关重要（Rong 等，  2003；Stephens 等，  1994），因此 tau-PLCγ1 相互作用可能在 MAPK 激活中发挥作用。最后，已鉴定出在信号转导中具有已知作用的其他 tau 相互作用子，例如 Abl 和 14-3-3（Agarwal-Mawal 等，  2003；Derkinderen 等，  2005；Hashiguchi 等，  2000）。Abl和14-3-3都参与PC12细胞的NGF反应，因此不乏需要考虑的机制。

鉴于已报道的信号转导相关 tau 相互作用因子的数量，仍有许多工作要做。特别是，除了确保关联发生在细胞中之外，正如 Fyn、SHP2、PI3K、Syk 和 PLCγ 所证明的那样（Jenkins & Johnson,  1998；Kim 等人，  2019；Lebouvier 等人，  2008；Lee 等人） al.,  1998 ; Souter & Lee,  2009），确定这种关联是否影响相互作用子或 tau 的功能非常重要。Tau 已被确定为酪氨酸激酶的底物（Fyn、Src、Syk、Lck 和 Abl；（Derkinderen 等人，  2005；Lebouvier 等人，  2008；Lee 等人，  1998；Lee 等人，  2004 ） ；Scales 等，  2011），未发表的数据）强烈表明 tau 参与信号转导途径。酪氨酸磷酸化 tau 蛋白也是 SHP2 的底物 (Kim et al.,  2019 )。关于相互作用，tau 与花生四烯酸一起，已被证明可以激活 PLCγ1（Hwang 等人，  1996），并且 tau 还被证明可以增加 Fyn 的体外活性以及转染细胞中 Src 的活性（夏尔马等人，  2007）。另一个需要更多研究的问题是 tau 磷酸化对相互作用的影响。例如，对于 PI3K 或 PLCγ1，虽然 tau 及其 SH3 结构域之间的相互作用因 tau 中的拟磷突变而减少，但测试的 tau 构建体包含多个拟磷突变，以更精确地识别关键位点 (Reynolds et al.,  2008)）；因此，需要更多的测试。

在设想 tau 在信号转导中发挥作用时，值得一提的是，tau 的氨基末端与神经元细胞质膜的关联已被报道（Brandt 等人，  1995），这表明 tau 与质膜的一个子域，即脂筏。脂筏含有响应激素和生长因子而启动信号转导的受体，脂筏被认为对于细胞对细胞外信号的反应至关重要（Paratcha & Ibanez,  2002综述；Simons & Toomre,  2000；Tsui-Pierchala 等人） .,  2002 )。此外，tau蛋白已在脂筏中发现，其在脂筏中的存在与细胞反应相关(Hernandez等，  2009；Kawarabayashi等，  2004；Klein等，  2002 )。对 tau 在信号转导中作用的研究可能涉及其在脂筏中的定位。

最后，考虑到 SH3 结构域与 PXXP 基序相互作用，并且 tau 包含七个 PXXP 基序，tau 与许多酪氨酸激酶和其他参与信号转导的蛋白质相互作用的能力可能源于其富含脯氨酸的结构域。随着目前蛋白质结构测定的进展，利用冷冻电子显微镜报道了丝状 tau 蛋白的结构以及 tau 蛋白与微管相互作用的模型（Fitzpatrick 等人，  2017；Kellogg 等人，  2018；Zhang 等人，  2019）。然而，报道的结构主要集中在微管结合域，而没有评论富含脯氨酸的域的结构。鉴于微管结合域在神经退行性疾病中的重要性，尚不清楚研究人员是否曾经结晶过氨基末端域（氨基酸 1-241）。Kellogg等人的tau-微管结合模型表明，tau的微管结合重复区域沿着微管表面延伸（Kellogg等人，  2018），没有迹象表明氨基末端结构域向微管结合方向折叠回去领域。然而，这并不排除当 tau 与微管分离时，相互作用子可能能够与 tau 富含脯氨酸的结构域和基本微管结合结构域相互作用。此外，当 tau 上的 Thr231 残基被磷酸化时，它可以以顺式 pT231 或反式 pT231 的形式出现（Nakamura et al.,  2012），表明富含脯氨酸的结构域中最接近微管的部分的 tau 结构结合域可以以两种构象出现，非 pT231 tau 提供第三种构象。确定 T231 磷酸化如何改变 tau 结构，以及磷酸化引起的富含脯氨酸结构域的结构变化如何影响与 Fyn 和 SHP2 等蛋白质的相互作用，肯定会很有趣。

当我们回顾 50 年来的 tau 研究时，虽然许多问题得到了解答，但更多的问题也被提出。阿尔茨海默病、额颞叶痴呆、癫痫、创伤性脑损伤和其他神经退行性疾病中存在异常 tau，这引发了对异常 tau 特性的许多研究，即从患病大脑中分离出的 tau、异常磷酸化的 tau 或带有 FTDP 的 tau 17个突变。然而，除了促进微管组装的能力之外，正常 tau 蛋白的功能仍有待完全确定。在这里，我们只关注一种可能的研究途径，认为确定 tau 作用的新机制将作为疾病相关研究的自然起点。在接下来的 50 年里，我们预计 tau 研究将阐明 tau 的新功能，从而带来新的疾病疗法。