GSK3 in Alzheimer’s Disease: Mind the Isoforms

Abstract

Understanding the molecular signaling pathways that go awry in Alzheimer’s disease (AD) would provide insights into developing novel therapies for this devastating neurodegenerative disease. Previous work has established that hyperactive glycogen synthase kinase-3 (GSK3) is linked to both “sporadic” and “genetic” forms of AD, suggesting a crucial role of GSK3 in AD pathogenesis. Therefore, inhibition of GSK3 activity has been intensely investigated as a potential therapeutic intervention for AD. GSK3 exists in two isoforms: GSK3α and GSK3β. Markedly, recent studies indicate specific contributions of each of the α and β isoforms of GSK3 to AD pathogenesis, suggesting a role of both isoforms in the disease. Here I review recent relevant work investigating isoform-specific roles of GSK3 in AD pathophysiology, highlighting the emerging role of GSK3α, which has been largely overlooked in favor of the more extensive studies of GSK3β.

INTRODUCTION

To date, there is no cure or disease-modifying therapy for Alzheimer’s disease (AD), the most common form of dementia in elderly and one of the leading causes of death across all ages [1]. Given that the incidence of AD has been escalating considerably with population aging, and may become a global public health crisis if it continues unchecked [2], it is clearly urgent to develop novel mechanism-based therapeutics. Among the signaling molecules being explored as potential drug targets for the treatment of AD, the serine/threonine kinase glycogen synthase kinase-3 (GSK3) has recently drawn attention. Originally identified as a regulator of glycogen metabolism [3], GSK3 is now established as a critical component involved in many cellular processes, and GSK3 dysfunction has been implicated in major diseases [4, 5]. Previous work has established that hyperactive GSK3 activity is linked to both “sporadic” and “genetic” forms of AD. Therefore, inhibition of GSK3 is being studied as a novel therapeutic approach for AD [6-8]. Of note, there are two isoforms of GSK3: GSK3α and GSK3β. The majority of AD-related studies on GSK3 have focused on GSK3β, though there have been reports suggesting a distinct role for GSK3α in synaptic plasticity and amyloid-β (Aβ) biology [9, 10]. In this review, I will focus on recent studies investigating isoform-specific effects on AD pathophysiology, highlighting the emerging role of GSK3α, which has been largely overlooked compared to GSK3β. Readers who are also interested in aspects related to general GSK3 signaling and its role in AD should consult other recent reviews [7, 11].

ISOFORMS OF GSK3

The two isoforms of GSK3 are generated from different genes (GSK3α and GSK3β), while a splice variant of GSK3β, GSK3β2, has also been identified. GSK3α and GSK3β share 85% overall sequence homology, and are virtually identical in their kinase catalytic regions, while differing substantially in their N- and C-terminals. Therefore, it would be unlikely to discriminate the effects of two isoforms based on experiments of small molecular inhibitors [12, 13]. Furthermore, despite the similarity in their kinase domains, previous studies indicate that the GSK3 isoenzymes may diverge in tissue-specific distribution, cellular localization, and functions [5,12]. For instance, genetic studies demonstrated that mice totally lacking GSK3β die before birth, while GSK3α global knockout mice survive and appear normal [14-16]. As for distribution pattern in tissue, both isoforms are enriched in brain tissue, with GSK3α being especially abundant in the hippocampus, cerebral cortex, striatum, and cerebellum, while GSK3β is expressed in nearly all brain regions, albeit with significant differences in mRNA levels in subareas of human brain [13, 17-20]. Further, it has been recently shown in brain that GSK3 isoform-specific genetic deletion produces distinct changes in substrate phosphorylation, indicating that each enzyme has a different range of substrates [21].

GSK3 SIGNALING AND ITS EMERGING ROLE IN SYNAPTIC PLASTICITY, LEARNING, AND MEMORY

GSK3 is unusual in that the kinase displays high activity under resting/basal conditions, and its constitutive activity is reduced upon N-terminal serine phosphorylation—Ser 9 in GSK3β and Ser 21 in GSK3α—by upstream kinases including Akt, protein kinase A (PKA), protein kinase C (PKC), p90RSK1, and p38MAPK. In opposition to serine phosphorylation, GSK3 can be stimulated by tyrosine phosphorylation—Tyr 279 in GSK3α and Tyr 216 in GSK3β—but whether this occurs via autophosphorylation or other tyrosine kinases is under debate [5, 13, 22]. In addition, GSK3 is regulated independently of phosphorylation by Wnt signaling through complex formation [5].

Notably, multiple lines of evidence from recent studies indicate that GSK3 plays a pivotal role in normal synaptic plasticity, learning, and memory. For instance, activation of GSK3 has been associated with inhibition of long-term potentiation (LTP) [23], a major form of synaptic plasticity that is established as a cellular model for learning and memory [24]. Along the same line, overexpression of GSK3 results in spatial learning deficits [25]. Furthermore, it has been shown that suppression of GSK3 activity facilitates LTP induction, likely through mechanisms involving tonic repression of mammalian target of rapamycin (mTORC1) signaling pathway [9].

GSK3 IN AD: DOES ISOFORM MATTER?

A large body of evidence has indicated that activity of both GSK3 isoforms is increased in the brains of postmortem AD samples as well as transgenic mouse model of AD [26,27], which provides basis for proposing GSK3 inhibition as a therapeutic avenue for AD-related synaptic dysfunction and memory defects [6, 7, 11]. Indeed, it was reported that hippocampal LTP failure, either induced by Aβ or displayed in an AD transgenic mouse model (Tg2576), could be rescued by applying structurally distinct GSK3 antagonists [8, 28]. In addition to in vitro studies, effects of GSK3 inhibitors on AD have also been tested in vivo. A recent study examined the in vivo effects of lithium, an inhibitor of GSK3, on AD transgenic mice and observed that although lithium was able to improve cognitive function at early stages, such ability was lost in aged AD transgenic mice [29].

For the aforementioned reasons, none of the available small molecule inhibitors is able to suppress GSK3 activity in an isoform-specific manner. Therefore, genetic approaches have been utilized to investigate whether there exist isoform-dependent effects of GSK3 on AD pathogenesis. Here I review in detail two recently published articles, perhaps representing the most comprehensive studies on this topic, which drew seemingly contradictory conclusions [30, 31].

Hurtado et al. [30] developed and optimized adeno-associated virus (AAV) constructs expressing a short hairpin RNA (shRNA) that specifically targeted either GSK3α or GSK3β genes to reduce their protein expression. They then injected the constructs (shRNA-α or -β) intraventricularly into newborn AD model mice and examined their brain pathology at 11 months of age. Interestingly, the knockdown of GSK3α, but not GSK3β, expression lowered the brain amyloid-β (Aβ) load in AD model mice. In contrast, the knockdown of either GSK3α or GSK3β led to reduction of tau hyperphosphorylation and neurofibrillary tangle formation. Going a step further, Hurtado et al. [30] utilized the CaMKIIα-Cre and loxP system to generate a conditional mutant mouse line in which GSK3α, but not GSK3β, gene expression was reduced in specific areas (cortex, limbic system, etc.) later in development in AD model mice. Consistent with their findings with the viral injection approach, the brain Aβ load in AD model mice was significantly decreased. Importantly, by performing behavioral tests, they also observed that memory defects in aged AD model mice were improved by conditionally knocking down GSK3α. In short, the studies by Hurtado et al. [30] imply that GSK3α, but not GSK3β, plays a key role in AD pathophysiology. Moreover, these findings were in agreement with an earlier work in a cell culture system demonstrating that Aβ production is specifically dependent on the α but not β isoform of GSK3, presumably reflecting the regulation of γ-secretase-mediated AβPP cleavage [10]. One limitation of the studies by Hurtado et al. [30] though is that the knocking down of either GSK3α or GSK3β was done in newborn AD model mice, and it was unclear whether suppression of one isoform resulted in the compensatory upregulation of the activity of the other isoform. Also it will be intriguing to determine whether intervention to reduce GSK3 activity at later stages in AD model mice, i.e., either before or even after development of brain plaques or tangles, can produce similar results. In other words, the work by Hurtado et al. [30] convincingly demonstrates “prevention” but not “reversal” of AD pathogenesis.

In contrast, most recently Ly et al. [31] reported that specific suppression of the activity of GSK3β, but not GSK3α, reduced Aβ production via the mechanisms associated with β-site AβPP-cleaving enzyme 1 (BACE1) [31]. In support of their conclusion, first they transfected human neuroblastoma cells with isoform-specific siRNA to knock down either GSK3α or GSK3β, and found that BACE1 gene expression as well as the related AβPP processing were reduced only by GSK3β siRNA. They further demonstrated, via a molecular approach, that the effects of GSK3β on BACE1 transcription were mediated through NF-αB signaling. Moving into in vivo experiments, they treated AD model mice with the GSK3 inhibitor ARA and observed decreased levels of brain Aβ plaques, in correlation with reduced BACE protein level and AβPP processing. Finally, they also revealed that AD-associated memory deficits were alleviated by GSK3 inhibitor treatment. However, because of the nonselectivity of small molecular inhibitors on GSK3α and GSK3β, it remains elusive which isozyme really accounts for the behavioral effects.

Moreover, while in both studies the authors tried to link the GSK3 effects on AD with AβPP processing, it needs to be pointed out the ability of GSK3 (either α or β) to control Aβ creation through γ-secretase or BACE1 has been challenged by in vivo data showing that neither isoform affected the process of Aβ generation, based on biochemical analysis [32]. However, it should be noted that the aforementioned in vivo study [32] did not report the effects of GSK3 isoforms on the brain pathology associated with Aβ deposits, nor did they perform any behavioral experiment to assess the learning and memory change. Furthermore, it is also possible that the rescuing effects of GSK3 isoforms on AD-associated memory defects are independent of AβPP processing (e.g., through affecting signaling pathways that are critical for neural plasticity and memory, or tau phosphorylation, etc.), even though a correlating relationship may exist.

CONCLUSION

In summary, both Hurtado et al. [30] and Ly et al. [31], along with previous studies, presented compelling evidence to demonstrate in vivo the association between GSK3 signaling and AD pathogenesis. Importantly, their work suggests that a specific GSK3 isoform, either α or β, plays a distinct role in regulating AD pathophysiology. An important translational implication based on these findings is that the specific targeting of only one isoform of GSK3, which presumably would reduce off-target effects, might be able to alleviate AD pathogenesis, as well as abnormalities associated with other neurological diseases. Thus, future studies designed to identify isoform-specific approaches for GSK3 inhibition could result in effective therapeutics for AD with fewer side effects.