1. GSK-3—A Story of Two Isozymes

Glycogen synthase kinase-3 (GSK-3) is a highly conserved protein serine/threonine kinase that plays a central role in a wide variety of cellular processes concerned with coordinating catabolic and anabolic pathways and regulating cellular fate and cell growth. GSK-3 targeted phosphorylation typically inhibits the activity of the substrate, leading to attenuation of the signaling pathway. GSK-3 thus functions as a suppressor of hormone/growth factor-induced signaling cascades. For example, GSK-3 inhibits insulin signaling through the phosphorylation of glycogen synthase and the insulin receptor substrates, IRS-1/IRS-2, where the former leads to inhibition of glycogen synthesis, and the latter inhibits insulin receptor tyrosine kinase activity ^[1][2][3][4][1,2,3,4]. GSK-3 also inhibits the canonical Wnt signaling pathway through phosphorylation of β-catenin, which de-stabilizes the protein, leading to subsequent degradation in the proteasome ^[5][6][5,6]. In addition, GSK-3 phosphorylates a variety of transcription factors including Nuclear Factor of Activated T-Cells, NFAT [7], heat shock factor-1 [8], cAMP response element binding protein, CREB, [9], and nuclear factor-kappa B, NF-κB [10], to inhibit gene expression. The unique properties of GSK-3 may explain its involvement in such a wide variety of biological processes: Unlike most of protein kinases, GSK-3 is active under “basal” conditions and is inhibited when cells are stimulated. The substrate recognition is also unusual as it typically requires pre-phosphorylation of the substrate by another “priming kinase” ^[1][11][1,11]. This unique feature adds additional levels of regulation because the ability of GSK to phosphorylate a substrate is conditionally dependent upon the activation of the priming kinase, and that may be controlled by various factors including cell type and cellular context. It should be noted, however, that unprimed substrates had been reported, such as β-catenin, or presenilin-1, as demonstrated by using a GSK-3 mutant that cannot interact with primed substrates ^[12][13][14][12,13,14]. Finally, the versatility of GSK-3 also relies on its broad range of substrate, including a predicted number over 500 substrates and about 100 “physiological substrates” that are related to diverse cellular functions ^[15][16][15,16]. Another important feature of GSK-3 is the existence of two isozymes, GSK-3α and GSK-3β coded by two different genes [17], and a spliced variant of GSK-3β (GSK-3β2) containing a 13 amino acid insert has been described [18]. The GSK-3β2 variant is enriched in neurons and shows lower in vitro activity as compared to GSK-3β [19]. GSK-3 isozymes exhibit both similar and distinct functions. In some cases, the isozymes fulfil non-redundant physiological functions, but in others, there is a possibility of compensation. The GSK-3 isozymes share 97% identity in their catalytic domains, but there are significant differences at the N–and C-terminal domains [17]. Notably, GSK-3α has been largely overlooked in favor of studies with GSK-3β, although roles for GSK-3α in cellular regulation and diseases pathogenesis have recently been described. From the evolutionary perspective, the α and β isozymes split from a common precursor approximately at the time of emergence of vertebrates, and both genes are highly conserved in fish, amphibians, reptiles, and mammals [20]. An interesting finding is that the α gene is missing in birds. Although the initial findings were based on the available draft genome of three species, namely, chickens, domestic turkeys, and zebra-finches [20], searching the updated genomic data confirms the general selective loss of GSK-3α in the avian species (results from our laboratory).

The question of whether or not each GSK-3 isozyme possesses distinct functions has been addressed in many intense studies. One possible cause of the differences between the isozymes could stem from their distinct distribution in the brain, where GSK3α is especially abundant in the hippocampus, cerebral cortex, striatum, and cerebellum, while GSK3β is expressed in nearly all brain regions [21]. Another option could be that the differences are due to their distinct phosphorylation pattern of substrates. Thus, specific deletion of each of the GSK3 isozymes in the brain produced a distinct substrate phosphorylation pattern [19]. For example, phosphorylation of Collapsin response mediated proteins, CRMP2 and CRMP4 at phosphorylation sites Thr 509, Thr 514 and Ser 518 was not detectable in cortex lacking GSK3β but was normal in cortex lacking GSK-3α, and phosphorylation of tau at Thr 231, Thr 235, and Se 396 was predominantly catalyzed by GSK-3β [19], although there may also be redundant activity of the GSK-3 isozymes for other substrates such as β-catenin [22].

In the following section, we summarize the results obtained by genetic manipulations of the GSK-3 isozymes, focusing on phenotypes and processes related to the brain and the nervous system. A significant difference between the isozymes is clearly observed in embryonic development: while loss of GSK-3β is lethal, due to liver degeneration and impaired heart development ^[23][24][25][23,24,25], GSK-3α null mice are viable [26]. However, it is apparently more complex to distinguish the roles played by the individual GSK-3 isozymes in adult neurons. The brains of GSK-3α null mice are smaller, and the mice exhibit more aggressive behavior, reduced exploratory activity, and reduced social interaction than normal controls [27]. The GSK-3α null mice also have a shortened lifespan that is associated with age-related pathology related to cardiac dysfunction, early onset of sarcopenia, and cellular senescence [28]. Selective loss of GSK-3α in neurons has also been shown to alter neuronal architecture and behavior activity [29]. With respect to pathological conditions, knock-down of GSK3α, but not GSK3β, ameliorated amyloid plaque loads and memory deficits in an Alzheimer’s disease (AD) mouse model [30]. In contrast, manipulation of GSK-3β expression resulted in alterations in neuronal structure, mood behavior, and cognitive functions. Selective loss of GSK-3β in the forebrain pyramidal neurons produced anxiolytic (reduced anxiety) and pro-social effects [31], and loss of GSK-3β but not GSK-3α in GABAergic neurons, reversed gamma oscillation deficits and cognitive dysfunction in an NMDA hypofunction model related to schizophrenia [32]. Another “behavior” study reported that GSK-3β heterozygous mice exhibit reduced exploratory and anxiety behavior ^[33][34][33,34]. The impact of GSK-3β on neuronal structure was further demonstrated in cortical and hippocampal neurons where selective deletion of GSK-3β reduced dendritic spine stability and attenuated excitatory synaptic transmission [35]. Finally, overexpression of GSK3β reduced brain size in transgenic mice [36].

Conditional deletion of both GSK-3 isozymes further highlighted the prominent role of GSK-3 in regulating brain architecture and behavior skills. Genetic elimination of both GSK-3 isozymes by shRNA reduced axon growth, while localized inhibition of both isozymes at the distal axon resulted in axon elongation [37]. Conditional deletion of GSK-3α and GSK-3β in astrocytes resulted in a larger brain with an increased number of astrocytes [38]. These animals showed aberrant anxiety and altered social behavior [38]. Specific deletion of GSK-3 isozymes in new born cortical neurons, disrupted dendritic orientation and radial migration (moving neurons to a different brain layer) in all areas of the cortex and hippocampus [39]. Finally, deletion of both GSK-3 isozymes in neuronal progenitors resulted in a massive proliferation of cells and prevented progenitor differentiation [40].

The observation that birds lack GSK-3α provides an opportunity to distinguish the specific roles of GSK-3β. Inhibition of brain GSK-3β in a zebra finch model altered singing behavior and reduced neurogenesis in certain regions of the ventricular zone [41]. The results suggested that GSK-3α may be the major tau kinase in the adult brain, as levels of phosphorylated tau (at GSK-3 phosphorylation site) in the bird’s brain were largely reduced as compared to that of found in the mouse brain, a phenomenon that was also recapitulated in the brain of GSK-3α KO mice [20]. As high levels of tau phosphorylation was found in the bird’s embryo, it was further suggested that GSK-3β may be the dominant tau kinase during embryonic development [20]. Interestingly, overexpression of GSK-3β resulted in increased tau phosphorylation in the adult mouse brain ^[36][42][36,42]. Thus, it seems that GSK-3α may be the preferred tau kinase in adult; nevertheless, GSK-3β may become a “more dominant” tau kinase in pathological conditions ^[36][42][36,42].

An interesting alternative model for the study of isozyme function is the phosphorylation-resistant GSK-3α/β knock-in mouse [43]. In these mice, GSK-3 could not be inhibited (via serine phosphorylation) by an upstream kinase. The results confirmed a dominant role for GSK-3β (but not GSK-3α) in regulating muscle glycogen synthase [43], as well as in vivo tau phosphorylation by GSK-3 [36]. These mice showed hyperactivity and mania, which recapitulated symptoms of schizophrenia and manic phase in bipolar disorder [44]; in another study, they showed impairment of neuronal precursor cell proliferation [45]. The recent development of isozyme selective GSK-3 inhibitors also provides an opportunity to distinguish differences in function between the two GSK-3 isozymes. The use of BRD0705 to selectively inhibit GSK-3α (IC₅₀ 0.066 μM vs. 0.5 μM of α or β isoform respectively [46]), revealed that inhibition of GSK-3α corrects excessive protein synthesis and ameliorates the susceptibility to audiogenic seizures in Fragile X syndrome (FXS) mice [47]. Conversely, inhibition of GSK-3β by the selective inhibitor, BRD3731 (IC₅₀ 0.015 μM vs. 0.215 μM of β or α isoform respectively [46]), reversed gamma oscillation and cognitive dysfunction in a mouse model of schizophrenia [32].

2. GSK-3 in Neurodegeneration

GSK-3 is indeed a crucial player in the nervous system, and a significant factor that contributes to disease pathogenesis. Earlier studies revealed lithium salt, a drug approved for treating psychiatric disorders, as a GSK-3 inhibitor ^[48][49][48,49]. This finding implicated GSK-3 as a central regulator of mood behavior and psychiatric disorders, a notion that has since been supported by numerous studies. The current paradigm suggests that hyperactivity of GSK-3 is a causative factor in progressive neurodegenerative and psychiatric conditions, while inhibition of GSK-3 may be therapeutic. Indeed, hyperactive GSK-3 was found in the AD brain, and overexpression of GSK-3 in vivo induced AD pathology, cognitive deficits, and gliosis in a number of AD mice models ^{[36][50][51][52][53][54][55]}[36,50,51,52,53,54,55]. Additional studies have reported that alterations in GSK-3 activity (e.g., either excessive activation, or inhibition) influence emotion, mood behavior, sociability skills, and schizophrenia-like behavior ^{[31][33][44][56][57][58][59][60]}[31,33,44,56,57,58,59,60]. As a corollary, a reduction in GSK-3 activity reverses the severity of a number of diseases. For example, conditional deletion of GSK-3 in the brain of AD transgenic mouse models (mice expressing APP mutant, tau mutant, or double transgene expressing APP/PS1 mutants), was reported to reduce β-amyloid loads and levels of tau phosphorylation, and to decrease the formation of neurofibrillary tangles ^[30][61][30,61]. Likewise, treatment with GSK-3 inhibitors has been shown to improve disease symptoms in animal models of AD, Parkinson’s disease (PD), amyotrophic lateral sclerosis (ALS), Fragile X syndrome (FXS). Yet, no efficacy was achieved in phase 2 clinical trial for progressive supranuclear palsy (PSP) with the GSK-3 inhibitor tideglusib. Detailed descriptions of these studies have been published elsewhere ^{[62][63][64][65][66][67]}[62,63,64,65,66,67].

It is evident from the accumulated data that GSK-3 plays a prominent role in regulating structural and metabolic processes both in developing and adult neurons. In this review, we describe the role of GSK-3 in regulating cytoskeleton organization, the mammalian target of rapamycin (mTOR) pathway, and in mitochondria, all of which are components that link GSK-3 to neurodegeneration (see Figure 1). In addition, we provide an update of the field of GSK-3 inhibitors.

Figure 1. GSK-3 regulatory pathways in neurons. GSK-3 regulates microtubule (MT) stability and dynamics. Phosphorylation of MT binding proteins (MAPs) by GSK-3 reduces their binding to MT, and GSK-3 phosphorylation of kinesin 1 impairs anterograde and retrograde transport. GSK-3 activation of mTORC1 inhibits autophagic and lysosomal activity. GSK3 regulates mitochondrial energy metabolism and mitochondria-mediated cell death. GSK-3 destabilizes peroxisome proliferator-activated receptor γ, PGC1α, and inhibits its transcriptional activity, phosphorylation of dynamin-related protein1, DRP1, by GSK3 enhances mitochondria fission, and phosphorylation of Voltage-dependent anion-selective channel 1, VADC1, and bcl-2 associated proteins, Bax, by GSK-3 enhances their induced-apoptotic activity. Finally, GSK-3 impairs mitochondria and ER communication by disrupting proteins associated with the microdomain, mitochondria-associated membranes, MAM.