Selective suppression of the α isoform of p38 MAPK rescues late-stage tau pathology

Alzheimer’s disease (AD), a progressive neurodegenerative condition, is diagnosed in 60–80% of total dementia cases and presents with memory loss and other cognitive impairments. AD is the sixth leading cause of death in the United States. Neurofibrillary tangle (NFT) pathology, one of the major pathological hallmarks of AD and related tauopathies, occurs when microtubule-associated protein tau (MAPT or tau) undergoes hyperphosphorylation and aggregates as NFTs. Recent studies suggest that NFT pathology closely correlates with neurodegeneration and cognitive decline [1, 2].

Over the years, studies have suggested that one of the major functions of tau is to bind and stabilize microtubules. However, immune-mediated or genetic-mediated depletion of tau did not seem to affect microtubule stability in vitro and in vivo, suggesting the possibility of non-microtubule binding functions of tau [3, 4]. Indeed, our previous studies have suggested that the PXXP motif of tau can interact with SH3 domains of the Src family kinases (such as Fyn and Src) [5‐7], which in turn affect actin remodeling in growth cones [7]. These tau–fyn [8] and tau–actin [9] (also reviewed in [10]) interactions were also demonstrated in earlier studies. The non-microtubule binding functions of tau are equally important because a recent study demonstrated that tau has a very short interaction time with microtubules in physiological conditions [11]. Based on these studies, the current consensus in the field is that tau can contribute to the disease process either via loss of function (e.g., failure to bind to microtubules) or gain of toxic function (e.g., by affecting various types of non-microtubule binding functions). In further support of this, our previous studies have demonstrated that phosphorylation or disease-related point mutations in tau affect tau–Fyn interactions and may contribute to pathogenesis [12, 13]. Therefore, phosphorylation of tau by various Ser/Thr or Tyr kinases may directly impact on both microtubule-binding and gain of toxic functions of tau, which may contribute to neurodegeneration in AD [14, 15] (reviewed in [16, 17]). Most of these tau phosphorylation sites are considered to be targeted by more than one kinase [18]. In fact, there are currently more than 20 serine/threonine kinases reported to phosphorylate tau in vitro: glycogen synthase kinase 3 beta (GSK-3β), cyclin-dependent kinase 5 (Cdk5), microtubule affinity regulating kinase 1–4 (MARK1–MAPK4), extracellular signal-related kinase (ERK), p38 mitogen-associated protein kinase alpha (p38α MAPK), and c-Jun N-terminal kinase (JNK) are the best examples [19‐21]. However, their direct involvement in AD remains elusive and those kinases amenable to therapeutic intervention are not yet evident.

We have demonstrated recently that activated microglia induced tau hyperphosphorylation via activation of the p38α MAPK signaling pathway in the hTau mouse model of tauopathy [22, 23]. Notably, we observed elevated levels of active (phosphorylated) p38α MAPK, but not GSK3β or regulators (p35/p25) of Cdk5, correlating with increased tau phosphorylation in the hippocampus of hTau mice deficient for the chemokine receptor CX3CR1 [22]. This is consistent with a previous in vitro study in which neocortical neurons pretreated with the popular-mixed kinase inhibitor SB203580 and then treated with interleukin-1β (IL-1β) showed significantly reduced tau phosphorylation and restored synaptophysin levels [24]. Signaling through p38 MAPK, especially the p38α isoform, has also been discovered recently to be a critical component for cytokine production in activated microglia [25, 26]. The p38α-mediated cytokine overproduction in microglia has been linked to inflammation-induced neurotoxicity in a variety of mouse models [22, 27‐33]. These studies suggest that targeting the p38α MAPK signaling pathway should be tested as a potential strategy to suppress microglial innate immune responses and prevent inflammation-mediated tau pathology.

The p38α MAPK has been a drug target in many animal models of disease, including peripheral inflammatory disorders (e.g., asthma and rheumatoid arthritis) and central nervous system (CNS) disorders. A variety of small-molecule p38α MAPK inhibitors have been tested in preclinical animal models of CNS diseases. These inhibitors vary in their target specificity, which raises concerns about linkage of target modulation to phenotype—a key aspect of drug discovery target validation, as documented previously [34]. For example, first-generation p38α MAPK inhibitors such as MW01-2-069A-SRM were used in a mouse model of AD-relevant pathophysiology that involves intracerebroventricular (ICV) infusion of oligomeric Aβ_1–42 [35]. Other studies used commercially available mixed kinase inhibitors such as SB203580, SB239063, and FR167653 in models of amyotrophic lateral sclerosis [32], contusion model of spinal cord injury [33], controlled cortical impact model of traumatic brain injury [36], cerebral ischemia [37], neuropathic pain in a rat model of lumbar disc herniation [38], cerebral ischemia in a rat model of stroke [39], and Parkinson’s disease in MPTP-treated mice [40]. While kinase inhibition resulted in significant neuroprotection in a majority of these models, it failed to restore function in some models (e.g., spinal cord injury). However, none of the previous studies carefully characterized or investigated p38α MAPK inhibitors in vivo for their independent effects on amyloid and tau pathologies, which are each relevant to AD and related non-AD tauopathies.

In the current study, we have utilized MW01-10-181SRM (or MW181), a highly selective p38α MAPK inhibitor [26], for its effects on p38α MAPK activity and tau phosphorylation in vitro and in 20-month-old hTau mice with advanced stages of disease. We report here that MW181 suppresses p38α MAPK activation, reduces tau phosphorylation, and decreases proinflammatory cytokine production. We also show that MW181 may prevent cognitive impairment in the aged hTau mice, possibly attributed to increasing synaptophysin levels. The data with MW181 were compared in several key experiments with results obtained using the mixed kinase inhibitor SB239063. In most experiments the results were congruent, allowing normalization of our results to those in the literature. The results with MW181, however, were remarkable in their further validation of p38α MAPK as a potential therapeutic target for tauopathies.

There are three major findings in this study; MW181 can reduce inflammation-induced p38α MAPK activation and tau phosphorylation in primary neurons in vitro; MW181 can block tau phosphorylation and aggregation in a 20-month-old hTau mouse model of tauopathy with an advanced stage of tau pathology; and evidence consistent with a p38 MAPK-mediated pathway playing an important role in tau pathology was provided by the resistance of MK2^–/– mice to inflammation-induced tau hyperphosphorylation. These results identify p38α MAPK as a druggable target that can block inflammation-induced tau pathology and improve cognitive function in tauopathies.

Mitogen-activated protein kinase (p38α MAPK is one of the members of this large family of kinases) is associated with various cellular processes including cell division [53], immunity [54], differentiation, oncogenesis [55], neurogenesis, and apoptosis [56] as well as learning and memory [57] (reviewed in [58, 59]). In AD/tauopathies, MAPKs play various roles at different stages of disease [60]. p38α MAPK can play an important role both in amyloid-mediated and tau-mediated toxicities. First, p38α MAPK has been shown to mediate Aβ-induced expression of various proinflammatory cytokines/chemokines from microglial cells that causes neuroinflammation [25, 26, 29, 35]. Targeted inactivation of p38α MAPK with small molecule inhibitors prevents the production of proinflammatory cytokines, reduces neurodegeneration, and improves neuronal and cognitive functions in rodent models of amyloid pathology [25, 26, 29, 35]. Second, in neurons, p38α MAPK has been shown to act as an important kinase in inducing IL-1β-mediated hyperphosphorylation of tau [24, 61]. Our previous study suggested that enhancement of microglia-specific neuroinflammation via genetic deficiency of Cx3cr1 resulted in accelerated tau pathology and cognitive impairment in an hTau mouse model of tauopathy [22]. Furthermore, hyperphosphorylation of tau in this model was mediated by upregulation of p38α the MAPK pathway, but not GSK3β or CDK5 pathways [22]. In a recent study, we demonstrated that adoptive transfer of reactive microglia from hTau-Cx3cr1 ^–/– mice led to induction of tau hyperphosphorylation and aggregation within the brains of recipient mice, which was also mediated via the IL-1β–p38α MAPK signaling pathway [23].

To our knowledge, this report provides the most conclusive demonstration that selective inhibition of p38α MAPK alone can significantly reduce Sarkosyl-insoluble NFTs, increase synaptophysin, and reduce the expression of proinflammatory cytokines in a pure tauopathy model with advanced stages of disease. Notably, original reports on hTau mice suggested formation of mature tangles and neurodegeneration by 20 months of age [45, 62, 63]. However, a recent study by our group has suggested that deficiency of microglial Cx3cr1 can enhance neuroinflammation and lead to the formation of Sarkosyl-insoluble tau in hTau mice as early as 2 months of age [23]. The potency of MW181 to reduce NFTs and rescue cognitive function suggests the therapeutic potential of selectively targeting p38α MAPK to attenuate tau pathology and cognitive impairment. While 1 mg/kg b.w. of MW181 was the only tested dose in the present study, different dosing and durations of treatment are key considerations during medical chemistry optimization of an in-vivo molecule for pharmacological safety. Just as our tauopathy studies described here were being completed, a recent drug development report [25] described a MW181 analog optimized for such key drug development compliances. While the anti-NFT effects of MW181 (via p38α MAPK inhibition) on 20-month-old hTau mice observed here are very intriguing, our study has some limitations. First, as far as we know, this is the first study targeting p38α MAPK in a 20-month-old mouse model of tauopathy, which is equivalent to approximately 60 years of age in humans, and demonstrating beneficial effects. Therefore, it should be validated in a separate study by a different group. Second, we did not perform parallel comparisons with nontransgenic mice administered with MW181, which might be important to determine whether MW181 effects are specific for transgenic mice or whether they have an overall affect in a normal system. Finally, in view of the normal physiological functions of p38 MAPK within the CNS, it is important to note that MW181 did not completely abolish the levels of activated forms of p38α MAPK. This further supports the possibility of modulatory effects of MW181 in selectively ablating the excessively activated forms of p38α MAPK.

Another important consideration related to the selectivity of MW181 for the α isoform of p38 MAPK is that MW181 is a representative member of efficacious and CNS-penetrant small molecule p38α inhibitors that large-scale kinome screens show are selective for the p38 MAPK family, lack crossover to major G-protein-coupled receptor (GPCR) agonist or antagonist classes based on functional screens, exhibit low toxicity at high doses, and are efficacious in an AD amyloid-relevant and progressive brain injury models [25, 26]. The selectivity of MW181 action is further indicated by the failure to bring about its CNS-relevant pharmacodynamic effects in p38α MAPK inhibitor-resistant mice [26, 28]. Further, the high-resolution crystal structure of MW181 and other members of this class of highly selective, in-vivo efficacious p38 MAPK inhibitor probes reveals that the structural basis of this selectivity is their common chemical scaffold that exploits specific three-dimensional features found only in p38 MAPK [25, 26]. Taken in its entirety, an extensive body of evidence indicates that the pharmacodynamic effects of this class of novel, selective inhibitors of CNS p38α MAPK are via modulation of endogenous p38α MAPK. In our previous study, the X-ray diffraction data revealed that the MW181 presence in the active site of the p38α MAPK renders the kinase to become inactive [25, 26]. This is an important consideration because the MW181-mediated inhibition may be entirely independent of the phosphorylation status of p38α MAPK on T180/Y182 sites. In other words, even if the p38α MAPK show unchanged T180/Y182 phosphorylation, binding of MW181 at the active sites makes the kinase inactive. Therefore, it is important to assess the activation status of p38α MAPK via assessing the phosphorylation of its downstream targets such as ATF2 and MK2. Indeed, this is the reason why we observed no alterations in the levels of pT180/pY182 in p38α MAPK in vivo but we did in vitro (Figs. 1 and 3). To some extent, the levels of pT180/pY182 in p38α MAPK could also be attributed to the extent of autophosphorylation by itself [64]. Therefore, we confirmed MW181-mediated p38α MAPK inhibition by assessing pATF2 both in vitro and in vivo. Optimized p38 MAPK inhibitors in this novel class of small molecules, such as MW150 [25], will also address the issue of drug candidate specificity and safety. Regardless, the results shown here added increasing support to the hypothesis that p38α MAPK is the key tau kinase under inflammatory conditions. While our previous studies suggested that MW181 may possibly target the β isoform of p38 MAPK and Nemo-like kinase (NLK) MAPK at relatively higher Ki values (322 nM and 199 nM, respectively) compared with p38α MAPK (184 nM) [26], it is the α-isoform selective inhibition of p38 MAPK that reduces tau phosphorylation and cytokine production. This conclusion is based on earlier studies, which suggested that deletion of the β isoform of p38 MAPK did not affect LPS-induced cytokine production or induce inflammation, suggesting that p38α MAPK is the primary mediator of cytokine release and inflammation [65] and that NLK is not an inducer, but a suppressor, of NF-κB activity [66], which is the primary mediator of inflammation. Taken together, these results suggest that the effects observed for MW181 in the current studies are exclusively due to targeted inhibition of the α isoform of p38 MAPK.

In the current study, MK2-deficient mice demonstrate a strong resistance to LPS-induced tau hyperphosphorylation. This suggests that suppression of IL-1β and related proinflammatory cytokine secretion is inhibited in MK2^–/– mice, which are otherwise induced by MK2-mediated transcription [67]. If this scenario was to be true, then one would argue that p38α MAPK should still be able to phosphorylate tau in the absence of MK2. Because IL-1β is one of the key proinflammatory cytokines important for LPS-induced p38α MAPK activation and the deficiency of MK2 leads to ablation of IL-1β production, then this suggests there is lack of a sufficient trigger for the activation of p38α MAPK following LPS injections (due to reduced expression of IL-1β and other proinflammatory cytokines). Indeed, our results show a significant reduction in the total p38 MAPK levels in MK2-deficient mice, suggesting insufficient levels of p38α MAPK in MK2^–/– mice. An alternative argument would be that MK2 itself might serve as a potential tau kinase. While this possibility needs further careful characterization, it is important to note that the MK2 activation via phosphorylation (at T222, S272, and T334) by p38 MAPK happens within the nucleus. Because of this, there is a very unlikely chance for the activated (and possibly p38 MAPK bound) MK2 to physically interact with tau, unless it translocates to the cytoplasm after activation. Therefore, based on current evidence, it is conceivable that reduced LPS-induced tau phosphorylation in MK2 deficient mice is due to the feed-forward abrogation of the MK2 > IL-1β > IL-1R > p38α MAPK > tau axis.

The results reported in this article provide a clear validation of p38α MAPK as a target whose selective inhibition can result in improved outcomes in tauopathy models. The recent description [25] of a drug candidate that retains all of the positive attributes of MW181, but is further optimized for bioavailability and in-vivo safety, provides the means to extend the rationale presented here for tauopathies toward clinical proof of concept.