SEARCH

SEARCH BY CITATION

Keywords:

  • Alzheimer's disease;
  • bipolar disorder;
  • phosphorylation;
  • schizophrenia;
  • signal transduction;
  • Tau kinase

Abstract

  1. Top of page
  2. Abstract
  3. GSK3 in AD
  4. Tau hyperphosphorylation in AD
  5. GSK3, neuronal death and plasticity
  6. GSK3 and insulin resistance: implications for AD
  7. GSK3 and other CNS disorders
  8. GSK3 inhibitors
  9. Pros and cons
  10. Conclusion and summary
  11. References

Glycogen synthase kinase3 (GSK3) is emerging as a prominent drug target in the CNS. The most exciting of the possibilities of GSK3 lies within the treatment of Alzheimer's disease (AD) where abnormal increases in GSK3 levels and activity have been associated with neuronal death, paired helical filament tau formation and neurite retraction as well as a decline in cognitive performance. Abnormal activity of GSK3 is also implicated in stroke. Lithium, a widely used drug for affective disorders, inhibits GSK3 at therapeutically relevant concentrations. Thus while the rationale remains testable, pharmaceutical companies are investing in finding a selective inhibitor of GSK3. In the present review, we summarize the properties of GSK3, and discuss the potential for such a therapy in AD, and other CNS disorders.

Abbreviations used

β-amyloid

AD

Alzheimer's disease

CSF

cerebrospinal fluid

GSK3

glycogen synthase kinase3

MT

microtubule

NFT

neurofibrillary tangle

NIDDM

non-insulin-dependent diabetes mellitus

Glycogen synthase kinase 3 (GSK3) was initially discovered as a kinase involved in the regulation of glucose metabolism and later as a participant in wnt/wingless signaling. In mammals, two closely related isoforms GSK3α and GSK3β are present (Woodgett 1991). The GSK3β isoform is highly expressed in neural tissue where its expression is regulated during development.

Insulin increases the activity of protein kinase Akt/PKB that inhibits GSK3 by phosphorylating the Ser9 residue of GSK3β or Ser21 residue on GSK3α (Frame et al. 2001). Conversely, GSK3 activity is ∼200-fold higher upon phosphorylation on its Tyr216 residue (Hughes et al. 1993). GSK3 is also regulated by protein phosphatase 2A. GSK3 has multiple substrates including the tau protein implicated in AD and can phosphorylate both prephosphorylated and unmodified substrates. The main differences between GSK3α and GSK3β isoforms are found in the N- and C-terminal regions. Within the ATP pocket of GSK3 where most drugs bind to and compete with ATP, there appears to be only a single amino acid difference (Glu196 in GSK3α, Asp133 in GSK3β) making it difficult to identify an inhibitor that can distinguish the two isoforms.

GSK3 in AD

  1. Top of page
  2. Abstract
  3. GSK3 in AD
  4. Tau hyperphosphorylation in AD
  5. GSK3, neuronal death and plasticity
  6. GSK3 and insulin resistance: implications for AD
  7. GSK3 and other CNS disorders
  8. GSK3 inhibitors
  9. Pros and cons
  10. Conclusion and summary
  11. References

The postmortem diagnosis of AD rests on the presence of extracellular plaques of β-amyloid (Aβ), and intracellular neurofibrillary tangles (NFTs) consisting of hyperphosphorylated tau protein. Increased levels of total GSK3 have not been consistently observed in AD brain, however, active GSK3 localizes to pretangle neurones, dystrophic neurites and NFTs in AD brain (Pei et al. 1997). Neurons undergoing granulovascular degeneration also contain active GSK3β (Leroy et al. 2002). A spatial and temporal pattern of increased active GSK3β expression coinciding with the progression of NFT and neurodegeneration has been observed. These studies provide evidence that the active form of GSK3β is increased in AD brain. Although a small increase in Aβ was observed when the β isoform of GSK3 was inhibited, non-isoform selective inhibitors such as lithium and kenpaullones as well as small interfering RNA against the α isoform of GSK3 reduce (Aβ) production in cells and in an animal model of amyloidosis (Phiel et al. 2003). If this finding is consistent, GSK3 will gain significant importance as a drug target for AD since it will have the potential to interfere with both amyloidosis as well as neurofibrillary pathology.

Tau hyperphosphorylation in AD

  1. Top of page
  2. Abstract
  3. GSK3 in AD
  4. Tau hyperphosphorylation in AD
  5. GSK3, neuronal death and plasticity
  6. GSK3 and insulin resistance: implications for AD
  7. GSK3 and other CNS disorders
  8. GSK3 inhibitors
  9. Pros and cons
  10. Conclusion and summary
  11. References

Tau, a microtubule (MT) associated protein is abundantly expressed in the brain and binds to tubulin through three or four repeat sequences in its C-terminal region to promote MT assembly. Abnormally phosphorylated tau forms filaments in vitro and in cells (Pérez et al. 2002a). In AD brains it accumulates in the neuronal perikarya and processes as paired helical filaments (PHFs), the chief constituent of NFTs. Hyperphosphorylation of tau is thought to result in the destabilization of MTs giving rise to a ‘pretangle’ stage (Braak et al. 1994) and subsequently loss of dendritic MT and synapses, cytoskeletal degeneration, and eventually cell death (Kosik 1992). Although multiple protein kinases can phosphorylate tau, GSK3 and cdk5 were the key kinases purified from MT from AD brain (Ishiguro et al. 1993). GSK3 phosphorylates the majority of sites on tau that are abnormally phosphorylated in AD brain (Hanger et al. 1998; example Fig. 1).

image

Figure 1. Model showing docking of Ser396–404 residues of Tau (arrow) within the substrate binding groove of the GSK3β protein. ATP is show within the ATP binding pocket. (a) Low power and (b) high magnification of the ATP pocket and substrate-binding groove.

Download figure to PowerPoint

Phosphoepitope-specific antibodies can detect some of the phosphorylation sites on PHF-tau. The AT-100 epitope (phospho-Thr212/Ser214) has been linked specifically to PHF-tau from AD brain and this antibody does not appear cross react with fetal tau protein. The AT-100 epitope requires a sequential phosphorylation by PKA and subsequent phosphorylation by GSK3 is a requirement (Zheng-Fischhofer et al. 1998).

Interestingly, several animal models that exhibit persistent tau phosphorylation also display neuritic dystrophy. For example, mice lacking either Reelin, disabled1, VLDLR2 or ApoER2 which exhibit persistent tau phosphorylation have dystrophic neurites and cytoskeletal abnormalities (Sheldon et al. 1997; Hiesberger et al. 1999). Fibrillar Aβ induces the clustering of the integrin receptors and is thought to mediate neuritic dystrophy via interactions with focal adhesion proteins (Grace and Busciglio 2003). Interestingly, active GSK3 colocalizes with focal adhesion proteins (Bhat et al. 2000). GSK3 also phosphorylates and inhibits kinesin-mediated motility leading to disruption of fast axonal transport, a mechanism hypothesized in the pathogenesis of AD (Morfini et al. 2002). Thus it is conceivable that persistent tau phosphorylation by GSK3 could result in neuritic dystrophy.

A conditional GSK3 overexpressing transgenic mouse exhibits persistent tau hyperphosphorylation, pretangle-like somatodendritic localization of tau, neuronal death in hippocampus (Lucas et al. 2001) and cognitive deficits. In Drosophila, overexpression of tau and the GSK3 homologue Shaggy, exacerbates neurodegeneration induced by tau overexpression alone, leading to neurofibrillary pathology (Jackson et al. 2002). There is also evidence that GSK3 reduces axonopathy in mice overexpressing human 4R tau by hyperphosphorylating tau and decreasing the ability of tau to interact with MTs (Spittaels et al. 2000). Specific GSK3 inhibitors could prevent the consequences of the aberrant tau phosphorylation. As animal models do not recapitulate all the features of AD, coupling the in vivo pharmacokinetic profile of kinase inhibitors with a pharmacodynamic read-out such as tau phosphorylation would provide critical information for testing inhibitor profiles in vivo. Additionally, use of biomarkers related to proof-of-mechanism such as measuring cerebrospinal fluid (CSF) phosphorylated tau might provide estimates of dose selection in a clinical setting.

GSK3, neuronal death and plasticity

  1. Top of page
  2. Abstract
  3. GSK3 in AD
  4. Tau hyperphosphorylation in AD
  5. GSK3, neuronal death and plasticity
  6. GSK3 and insulin resistance: implications for AD
  7. GSK3 and other CNS disorders
  8. GSK3 inhibitors
  9. Pros and cons
  10. Conclusion and summary
  11. References

Whether neuronal death in AD occurs by apoptotic mechanisms is still a debated issue. However, the existence of activated or cleaved proteins that result from apoptotic pathways have been observed in postmortem AD brains. Evidence suggests that GSK3 is responsive to a variety of apoptotic stimuli, e.g. heat shock, glutamate, H2O2. Exposure of hippocampal neurons to Aβ increases GSK3β activity (Takashima et al. 1996) and decreases acetylcholine (ACh) levels by inhibiting pyruvate dehydrogenase (PDH) activity. GSK3 mediated inhibition of PDH decreases the conversion of pyruvate to acetyl Co-A, a precursor for the synthesis of ACh. Studies in neuronal cell lines (Bhat et al. 2000) and in cortical neurons (Hetman et al. 2000) confirm that expression of active GSK3 is sufficient to induce neuronal death. Although the precise mechanisms by which GSK3β facilitates neuronal death remain unclear, GSK3 is upstream of the pro-apoptotic protease caspase-3 (Bhat et al. 2002). As active GSK3 triggers events that participate in cell death, part of AD pathology could result from an increase in GSK3 activity.

The transcription factor CREB plays a key role in long-term memory, apoptosis and synaptic plasticity. The phosphorylation of CREB by GSK3 attenuates CREB DNA binding (Bullock and Habener 1998) and overexpression of GSK3 in neuroblastoma cells inhibits its transcriptional activity (Grimes and Jope 2001). GSK3 also phosphorylates the transcription factor NFATc4 leading to a decrease in NFAT-regulated genes such as IP3R, which is required for neuronal plasticity (Beals et al. 1997). Thus negative regulation of CREB and NFATc4 by GSK3 could result in impaired synaptic plasticity, and cellular survival. These hypotheses will need to be further validated in animal models.

GSK3 and insulin resistance: implications for AD

  1. Top of page
  2. Abstract
  3. GSK3 in AD
  4. Tau hyperphosphorylation in AD
  5. GSK3, neuronal death and plasticity
  6. GSK3 and insulin resistance: implications for AD
  7. GSK3 and other CNS disorders
  8. GSK3 inhibitors
  9. Pros and cons
  10. Conclusion and summary
  11. References

GSK3, a regulator of glucose is an attractive target for non-insulin-dependent diabetes mellitus (NIDDM). In biopsies, GSK3 activity inversely correlates with glycogen synthase activity implicating GSK3 as a contributor to insulin resistance (Nikoulina et al. 2000). Insulin receptors are expressed in the brain regions that support the formation of memory. Some NIDDM patients afflicted with verbal and visual memory decline independent of clinically diagnosed dementia (Ewing et al. 1998). Recent evidence suggests that the insulin resistance known to underlie NIDDM may also contribute to the clinical symptoms of AD. As a consequence of commonality between these two diseases, therapeutic inhibition of GSK3 is indicated for both diseases.

GSK3 and other CNS disorders

  1. Top of page
  2. Abstract
  3. GSK3 in AD
  4. Tau hyperphosphorylation in AD
  5. GSK3, neuronal death and plasticity
  6. GSK3 and insulin resistance: implications for AD
  7. GSK3 and other CNS disorders
  8. GSK3 inhibitors
  9. Pros and cons
  10. Conclusion and summary
  11. References

Tau mutations that result in Frontotemporal dementia consist of NFT pathology. Transgenic mouse models with these mutations (P301L, V337M, R406W) exhibit NFT like pathology and increased GSK3 and cdk5 activity. It is unclear whether the tau mutations drive the pathology independent of kinases; the kinases might be acting downstream where phosphorylation facilitates the process of making the tau filaments insoluble.

Following occlusion of the middle cerebral artery in mice which results in ischaemia, an increase in active GSK3 was found to coincide with cell death which could be attenuated by lithium (Bhat et al. 2000), indicating that GSK3 inhibition may be beneficial in stroke. Lithium is widely used in the treatment of bipolar disorders and directly inhibits GSK3 (Ki 1–2 mm; Klein and Melton 1996). The mechanism of action by which lithium exerts its therapeutic effects is unclear but efficacy is observed after several days of treatment raising the possibility that GSK3 inhibition alters gene regulation and alters neuronal plasticity. Chronic lithium treatment results in increased protein levels of the GSK3 substrate, β-catenin, in rat brain (Gould et al. 2003). If lithium's therapeutic actions in bipolar disorder result from the inhibition of GSK3, then GSK3 would be an important target for bipolar disorder.

Lithium also regulates circadian rhythm cycles via inhibition of GSK3. Deregulation of circadian rhythms is observed in both bipolar disorder and unipolar depression. In Drosophila, activation of GSK3 or SHAGGY advances the entry of the TIMELESS protein into the nucleus (Martinek et al. 2001); a translation in mammals would be an advancement of circadian rhythms. It remains to be determined whether GSK3 inhibition would be of value in disorders of circadian rhythm.

GSK3 inhibitors

  1. Top of page
  2. Abstract
  3. GSK3 in AD
  4. Tau hyperphosphorylation in AD
  5. GSK3, neuronal death and plasticity
  6. GSK3 and insulin resistance: implications for AD
  7. GSK3 and other CNS disorders
  8. GSK3 inhibitors
  9. Pros and cons
  10. Conclusion and summary
  11. References

Several GSK3 inhibitors have been described and most of the observed effects are in vitro and cellular studies. Among these inhibitors are paullones (Leost et al. 2000), indurubin (Leclerc et al. 2001), anilinomaleimides (Smith et al. 2001), thiadiazolidinones (Martinez et al. 2002a), and other small molecules (Cross et al. 2001). However, these inhibitors do not appear to be selective against the closely related kinases cdk2 and cdk5.

Indirubins inhibit GSK3 by competing with ATP but are not specific for GSK3 as they also inhibit cyclin dependent kinases (cdk). Glaxo SmithKline's maleimide derivatives, SB-415286 and SB-216763 also inhibit GSK3 activity by competing for the ATP binding site and are neuroprotective but may not be specific against CDKs. The thiadiazolidinones are ATP-non-competitive GSK inhibitors although these results will need confirmation. Most recently, a potent and specific inhibitor of GSK3 has been reported (AR-A014418) by AstraZeneca which is selective against cdk2, cdk5 and 25 other kinases tested (Bhat et al. 2003).

Pros and cons

  1. Top of page
  2. Abstract
  3. GSK3 in AD
  4. Tau hyperphosphorylation in AD
  5. GSK3, neuronal death and plasticity
  6. GSK3 and insulin resistance: implications for AD
  7. GSK3 and other CNS disorders
  8. GSK3 inhibitors
  9. Pros and cons
  10. Conclusion and summary
  11. References

The current therapeutics for AD provides marginal benefit at attenuating cognitive deficits by inhibiting the enzyme acetylcholinesterase and increasing the levels of the neurotransmitter ACh. Unfortunately, this type of therapy does not stop the progressive neuritic dystrophy and damage, and over time these therapies become ineffective. There exists an urgent medical need to develop agents that delay or reverse the progression of AD. The majority of recent efforts have focused on preventing amyloidosis, an early event in AD pathogenesis. While attractive, there is no clinical evidence that it will prevent neurodegeneration and NFT development.

Concerns relating to inhibition of kinases relate to degree of effect of inhibiting a specific kinase. Complete inhibition of GSK3 is not desirable; and could lead to potential side effects; a small inhibition to reset the stoichiometry of tau phosphorylation may be sufficient. It is also unclear whether a brief or sustained inhibition of GSK3 is necessary to attenuate tau phosphorylation for an extended period of time.

As GSK3 phosphorylates glycogen synthase, an enzyme that converts glucose to glycogen, there is the theoretical possibility that GSK3 inhibition may lower glucose levels. Under normal conditions, GSK3 inhibition may not be sufficient to cause hypoglycaemia. Unlike insulin, a GSK3 inhibitor would not affect glucose transport. In contrast, in NIDDM, GSK3 phosphorylates and inactivates insulin receptor substrate thereby making cells resistant to insulin's action and under these conditions of elevated glucose, a GSK3 inhibitor would be expected to decrease glucose levels.

The Wnt pathway has been implicated in some cancers and one of the proteins affected by this pathway is β-catenin. In some tumours, stabilized β-catenin has been reported while in other cases a decrease in β-catenin levels has been observed (Osborne 2002). GSK3 phosphorylates β-catenin and targets it for subsequent degradation. Although GSK3 inhibition would be expected to stabilize β-catenin levels it is unclear whether this will lead to tumours. GSK3 is not a proto-oncogene and several other kinases can phosphorylate β-catenin. Furthermore, proteins such as Siah-1 and LRP/arrow can contribute to β-catenin degradation independent of phosphorylation (Liu et al. 2001). These studies indicate a redundancy in mechanisms leading to β-catenin degradation. To address this theoretical concern, the effects of GSK3 inhibition by lithium were tested in the APC multiple intestinal neoplasia mouse model, a robust model of tumorigenesis. Chronic lithium treatment in this model did not increase the number of tumours indicating that GSK3 inhibitors may not exacerbate intestinal polyp formation and may pose a low risk for the tumour development (Gould et al. 2003). Other studies suggest that stabilization of β-catenin in the CNS does not result in brain tumours (Kratz et al. 2002). Interestingly lithium, which has been prescribed for more than 50 years, increases survival rates of patients with adenocarcinomas (Johnson et al. 2001) and has not been reported to increase the prevalence of cancer.

From a drug discovery standpoint, any potential side effects via inhibition of GSK3 must be linked to the age of the disease population and the length of time it takes for the effect to be observed in humans. For AD, diagnosis is made at ∼65 years of age. Most AD patients deteriorate substantially within 4–5 years and death is common ∼8–10 years after diagnosis. GSK3 may phosphorylate some substrates more efficiently than others. Tau is a more accessible substrate than those, which are complexed to large proteins like APC, and a possibility of disconnection exists between efficacy and safety liabilities. The majority of marketed drugs exhibit side effects at high doses and a balance between the doses that provide clinical benefit and the dose having side-effects needs to be attained. Biological markers for efficacy and safety may allow for appropriate dose settings in the clinic and may overcome potential liabilities.

Conclusion and summary

  1. Top of page
  2. Abstract
  3. GSK3 in AD
  4. Tau hyperphosphorylation in AD
  5. GSK3, neuronal death and plasticity
  6. GSK3 and insulin resistance: implications for AD
  7. GSK3 and other CNS disorders
  8. GSK3 inhibitors
  9. Pros and cons
  10. Conclusion and summary
  11. References

Elevations of GSK3 levels and activity in the ageing brain may result from several of the features associated with familial and sporadic AD. GSK3 appears to be associated with a multitude of adverse events linked to MT dynamics, neuritic dystrophy, PHF-tau, plasticity, cognitive deficits, neurodegeneration and potentially amyloid production, pointing to a key role in AD. GSK3's preclinical link to NIDDM, stroke and affective disorders broaden its therapeutic potential. Given the significant role of GSK3 in a variety of effects linked to mechanisms related to AD and other CNS disorders, GSK3 inhibition using small molecule inhibitors is a testable hypothesis in the clinic. For other recent reviews on GSK3 refer to Eldar-Finkelman (2002); Kaytor and Orr (2002); Martinez et al. (2002b); and Doble and Woodgett (2003).

References

  1. Top of page
  2. Abstract
  3. GSK3 in AD
  4. Tau hyperphosphorylation in AD
  5. GSK3, neuronal death and plasticity
  6. GSK3 and insulin resistance: implications for AD
  7. GSK3 and other CNS disorders
  8. GSK3 inhibitors
  9. Pros and cons
  10. Conclusion and summary
  11. References