Journal of Neuroimmune Pharmacology

, Volume 2, Issue 1, pp 93–96

Glycogen Synthase Kinase 3 Beta (GSK-3β) as a Therapeutic Target in NeuroAIDS


  • Stephen Dewhurst
    • Department of Microbiology and ImmunologyUniversity of Rochester Medical Center
  • Sanjay B. Maggirwar
    • Department of Microbiology and ImmunologyUniversity of Rochester Medical Center
  • Giovanni Schifitto
    • Department of NeurologyUniversity of Rochester Medical Center
  • Howard E. Gendelman
    • Center for Neurovirology and Neurodegenerative DisordersUniversity of Nebraska Medical Center
    • Department of Pharmacology and Experimental NeuroscienceUniversity of Nebraska Medical Center
    • Center for Aging and Developmental BiologyUniversity of Rochester Medical Center
    • Department of NeurologyUniversity of Rochester Medical Center
    • Department of Microbiology and ImmunologyUniversity of Rochester Medical Center
    • Department of PediatricsUniversity of Rochester Medical Center
    • Center for Aging and Developmental Biology
Invited Review

DOI: 10.1007/s11481-006-9051-1

Cite this article as:
Dewhurst, S., Maggirwar, S.B., Schifitto, G. et al. Jrnl Neuroimmune Pharm (2007) 2: 93. doi:10.1007/s11481-006-9051-1


Highly active antiretroviral therapy (HAART) has made a significant impact on the lives of people living with HIV-1 infection. The incidence of neurologic disease associated with HIV-1 infection of the CNS plummeted between 1996–2000, but unfortunately the number of people currently HIV-1 infected (i.e., prevalence) with associated cognitive impairment has been steadily rising. While the reasons for this may be multifactorial, the implication is clear: there is a pressing need for adjunctive therapy directed at reversing or preventing damage to vulnerable pathways in the central nervous system (CNS) from HIV-1 infection. Using a team of preclinical and clinical investigators, we have focused our efforts on defining how proinflammatory mediators and secretory neurotoxins from HIV-1 disrupt signaling of the survival-regulating enzyme, glycogen synthase kinase 3 beta (GSK-3β). In a series of studies initiated using in vitro, then in vivo models of HIV-1-associated dementia (HAD), we have demonstrated the ability of the mood stabilizing and anticonvulsant drug, sodium valproate (VPA), that inhibits GSK-3β activity and other downstream mediators, to reverse HIV-1-induced damage to synaptic pathways in the CNS. Based on these results, we successfully performed pharmacokinetic and safety and tolerability trials with VPA in a cohort of HIV-1-infected patients with neurologic disease. VPA was well tolerated in this population and secondary measures of brain metabolism, as evidenced by an increase in N-acetyl aspartate/creatine (NAA/Cr), further suggested that VPA may improve gray matter integrity in brain regions damaged by HIV-1. These findings highlight the therapeutic potential of GSK-3β blockade.


glycogen synthase kinase 3 betahistone deacetylase type 3human immunodeficiency virus type 1HIV-1 associated dementianeuroprotection

Between 1996–2000, HIV-1 positive patients treated with HAART have had a significantly reduced incidence of HIV-1-associated neurologic disease. This observation reinforces the notion that active viral replication is a major contributing factor in the pathogenesis of this disorder. What is also striking is that HAART and other antiretroviral therapies (e.g., AZT monotherapy) can affect the course of already active neurologic disease. That is to say, HAART can ameliorate or even reverse the course of HAD in some HIV-infected persons. In light of these findings, we hypothesized that the underlying etiologies of HIV-1-associated dementia (HAD) are due, at least in part, to reversible metabolic dysfunction of neuronal pathways vulnerable to candidate HIV-1 neurotoxins. As neurotoxic molecules continue to be produced by immune-activated mononuclear phagocytes (MPs), metabolically compromised neurons sustain irreversible insults, undergo programmed cell death, and neurologic deficits become permanent.

Because neurologic disease does not correlate with the relative numbers of apoptotic neurons, but rather the relative burden of MPs in vulnerable brain areas (Glass et al. 1995), the question arises whether the etiology for HAD is due to, in part, a reversible metabolic encephalopathy. This supposition is based on neuropathologic data obtained from patients in the pre-HAART era, but is buttressed by reports of adult and pediatric patients with HIV-1 and neurologic disease that experienced substantial improvement in both neuropsychologic and neuroradiologic indices (Gendelman et al. 1998; Tepper et al. 1998). Further support for antiviral amelioration of neurologic disease associated with lentiviral infection can be garnered from studies by Fox et al. (2000) using SIV-infected rhesus macaques, where the authors have demonstrated that treatment with PMPA ameliorates neuropsychological deficits and normalizes sensory-evoked potentials (SEPs), a measure of the integrity of brainstem polysynaptic pathways in response to an event-related stimulus (such as light, sound, or recognition of a singular event in a stimulation pattern), but does not reverse motor deficits. Cessation of therapy leads to a return of neuropsychological deficits and abnormal SEPs. Kolson and Gonzalez-Scarano (2000) have speculated that microglia may serve as reservoirs for HIV-1 infection because they have a long life span. These brain-resident MPs, along with infiltrating MPs, are also believed to be the major source of candidate HIV-1 neurotoxins.

Based on these clinical, neuropathologic, and electrophysiologic observations, we recognized that several key targets in vulnerable neurons in the central nervous system (CNS) might be amenable to small molecule pharmacologic therapies to reverse neurotoxicity from secretory neurotoxins released from HIV-1-infected MPs. Based on data from our in vitro studies, we focused on the enzyme glycogen synthase kinase 3 beta (GSK-3β), which is a component of the Wnt signaling pathway with a rapidly growing number of identified roles in the nervous system, including regulation of neuronal migration and neurodevelopment, cell death, and synaptic structure and function.

Our laboratories have previously investigated relationships in signaling pathways between the HIV-1 regulatory protein Tat and its downstream mediator platelet-activating factor (PAF) in MP, with the ultimate goal of understanding how these impacted on normal neuronal signaling and cell fate. We were intrigued that both Tat and PAF up-regulated GSK-3β activity in cerebellar granule neurons (CGNs), with a resultant biologic effect of apoptosis (Maggirwar et al. 1999; Tong et al. 2001). Additionally, PAF also mediates neuronal migration in newly explanted CGNs by inducing GSK-3β activation (Tong et al. 2001), suggesting a developmental role for this enzyme in neuronal path finding. We and others have also observed that inhibitors of GSK-3β activation, including VPA and lithium, have neuroprotective efficacy against HIV-1 gp120-induced neurotoxicity in cultures of cortical and hippocampal neurons (Everall et al. 2002; Dou et al. 2003; Dou et al. 2005).

Other studies implicate GSK-3β dysfunction in psychiatric and neurodegenerative disorders (Martinez et al. 2002), and link GSK-3β with control of redox status [reviewed in (Chong et al. 2005)]. Roles for GSK-3β in the nervous system, including neuronal migration, axonal remodeling, synaptogenesis, and synaptic plasticity have been described (Packard et al. 2003). GSK-3β and downstream mediators like β-catenin regulate changes in structural and synaptic proteins that in turn result in a wide variety of biologic effects intimately involved with synaptic transmission. These effects include destabilization of microtubule associated proteins (MAPs), which in turn permits axonal remodeling (Hall et al. 2002); recruitment of synaptic vesicle proteins such as synapsin I and increased presynaptic activity (Hall et al. 2002); and effects that can be inhibited by VPA at concentrations in the therapeutic range for its use as a mood-stabilizing agent. Additionally, GSK-3β can modulate increased recruitment of postsynaptic proteins to the dendritic arbor, including S-SCAM and PSD-95, presumably to help modulate synaptic throughput (Murase et al. 2002; Nishimura et al. 2002).

GSK-3β’s primary substrate, β-catenin, may serve as an intermediary for many of GSK-3β’s effects on synaptic function, as it has been shown that depolarization drives β-catenin into dendritic spines, where β-catenin directly increases size and intensity of synapsin I clusters, and frequency of mEPSCs (Murase et al. 2002). Thus, inhibition of GSK-3β activity results in increased levels of β-catenin (phosphorylation of β-catenin by GSK-3β, its primary kinase, leads to increased degradation via the ubiquitin pathway), which in turn may modulate increased synaptic activity and plasticity.

To further validate the relevance of GSK-3β activation in mediating neuronal damage and death, we used conditioned media from HIV-1 infected MP and demonstrated that the neurotoxicity of this media depended on the presence of soluble Tat and dephosphorylation of ser9 on GSK-3β (a key site for regulation of enzyme activity (Sui et al. 2006). Equally important to understanding how GSK-3β inhibitors like VPA and lithium protect neurons against HIV-1 neurotoxins is the observation that their biologic effects are mediated in part through activation of NFκB (Sanchez et al. 2003; Rao et al. 2004). In particular, we have demonstrated that either recombinant Tat or conditioned media from HIV-1 infected MP can lead to phosphorylation of ser468 of the RelA subunit of NFκB, which is a GSK-3β-sensitive site (Sui et al. 2006). This promotes the association of RelA with histone deacetylase type 3 (HDAC3) corepressor molecules and effectively blocks the ability of RelA to transactivate prosurvival target genes that can influence neuronal cell fate (Sui et al. 2006). VPA opposes this through its effects on both GSK-3β and HDAC activity. Indeed, VPA has been shown to function as an inhibitor of the catalytic activity of class I HDACs, as well as an inducer of the proteasomal degradation of HDAC2 (Gottlicher et al. 2001; Phiel et al. 2001; Kramer et al. 2003). Key interactions of VPA that may contribute to neuroprotection in HAD are depicted in cartoon form in Fig. 1 below:
Fig. 1.

Cartoon of VPA-mediated neuroprotection against secretory neurotoxins from HIV-1-infected MP

Despite having a reasonable understanding of how VPA might reduce or prevent damage from secretory HIV-1 neurotoxins in our in vitro models, it was necessary to advance testing of VPA as a neuroprotective agent to an in vivo model. While primate models of SIV infection are the gold standard of viral neuropathogenesis, we felt that rodent models of HIV-1-associated neurologic disease might allow us to evaluate multiple genetic, biochemical, electrophysiologic, and spectroscopic parameters for testing single or multiple pharmacologic agents that may have value as adjunctive therapies. Thus, we used a severe combined immunodeficient (SCID) murine model of HIV-1 encephalitis (HIVE) to test whether VPA could reverse or prevent neuronal damage in vivo. VPA treatment was able to significantly reduce GSK-3β activation in vivo, as reflected by increased levels of β-catenin and a reduction in GSK-3β dependent phosphorylation of the neurofilament protein, tau (Dou et al. 2003). Additionally, VPA was able to preserve synaptic architecture, without increasing gliosis, a pathologic hallmark of HIVE or enhancing viral replication in HIV-1-infected MP present in SCID mice brains (Dou et al. 2003). It should be noted that the SCID mouse model of HIVE does not allow us to study the effects of peripheral HIV-1 infection of lymphocytes because the infection is confined to MPs stereotactically introduced into brain parenchyma, but these findings are reassuring and mirror recent clinical findings which have shown that VPA’s potential effects on latent HIV-1 infection do not significantly alter the overall level of virus production during an active infection (Ances et al. 2006) despite the potential for its HDAC inhibitory activity to reactivate chromosomally integrated, latent virus genomes (Lehrman et al. 2005).

Given the encouraging results in our in vivo model, we felt that VPA could be advanced to phase I trials. We initially performed a phase 1a pharmacokinetic trial during which we administered a low dose of VPA (250 mg P.O. BID) with efavirenz (EFV), a nonnucleoside reverse transcriptase inhibitor (NNRTI) and lopinavir/ritonavir (LPV), a protease inhibitor (PI), taken as part of the HAART regimen in our patient cohort. Our results demonstrated that there were no significant interactions between VPA and either EFV or LPV in terms of plasma concentrations for these HAART agents; conversely, EFV and LPV did not significantly alter VPA trough concentrations (DiCenzo et al. 2004). Using the same dosing regimen of VPA, we subsequently initiated a 10-week safety and tolerability trial (1b) in our HIV-1-infected patient cohort (16 patients with and six without cognitive impairment). VPA was safe and well tolerated without a significant effect on plasma HIV RNA concentration, and patients experienced an overall trend toward improved neuropsychologic performance compared to the placebo control group (Schifitto et al. 2006). It is interesting to note that despite the short period of time for this study, patients who received VPA had a significant increase in the N-acetyl aspartate/creatine (NAA/Cr) ratio in their frontal white matter (FWM) (Schifitto et al. 2006). In contrast, HAART alone only partially ameliorates HIV-1-induced decreases in NAA/Cr in the FWM that occur during HIV-1 infection (Meyerhoff et al. 1993; Chang et al. 1999; Stankoff et al. 2001). These findings are encouraging and are supported by studies from Letendre et al. (2006), who have shown that low-dose oral lithium improved neuropsychological performance in persons with HIV-1-associated neurological impairment, most likely as a result of effects on neuronal GSK-3ß. In aggregate, these results underscore the need for adjunctive therapies to help patients living with HIV-1 maintain their neurologic function. This is likely to be especially important for the aging population coping with HIV-1 infection and susceptible to age-related neurodegenerative processes such as Alzheimer’s disease.


The investigative team for this work encompassed three institutions: S. Dewhurst, S. Maggirwar, D. Peterson, G. Schifitto, R. J. Zhong, and H.A. Gelbard of the University of Rochester Medical Center, R. DiCenzo and G. Morse at the State University of Buffalo, and H. Dou, M. Boska, and H. Gendelman of the University of Nebraska Medical Center. We would also like to acknowledge the generous support of the Geoffrey Waasdorp Pediatric Neurology Fund, University of Rochester Medical Center. This work was supported by NIH grants PO1 MH64570, R01 MH56838, and RO1 NS054578.

Copyright information

© Springer Science+Business Media, Inc. 2006