HIV induces synaptic hyperexcitation via cGMP-dependent protein kinase II activation in the FIV infection model

Over half of individuals infected with human immunodeficiency virus (HIV) suffer from HIV-associated neurocognitive disorders (HANDs), yet the molecular mechanisms leading to neuronal dysfunction are poorly understood. Feline immunodeficiency virus (FIV) naturally infects cats and shares its structure, cell tropism, and pathology with HIV, including wide-ranging neurological deficits. We employ FIV as a model to elucidate the molecular pathways underlying HIV-induced neuronal dysfunction, in particular, synaptic alteration. Among HIV-induced neuron-damaging products, HIV envelope glycoprotein gp120 triggers elevation of intracellular Ca2+ activity in neurons, stimulating various pathways to damage synaptic functions. We quantify neuronal Ca2+ activity using intracellular Ca2+ imaging in cultured hippocampal neurons and confirm that FIV envelope glycoprotein gp95 also elevates neuronal Ca2+ activity. In addition, we reveal that gp95 interacts with the chemokine receptor, CXCR4, and facilitates the release of intracellular Ca2+ by the activation of the endoplasmic reticulum (ER)-associated Ca2+ channels, inositol triphosphate receptors (IP3Rs), and synaptic NMDA receptors (NMDARs), similar to HIV gp120. This suggests that HIV gp120 and FIV gp95 share a core pathological process in neurons. Significantly, gp95’s stimulation of NMDARs activates cGMP-dependent protein kinase II (cGKII) through the activation of the neuronal nitric oxide synthase (nNOS)-cGMP pathway, which increases Ca2+ release from the ER and promotes surface expression of AMPA receptors, leading to an increase in synaptic activity. Moreover, we culture feline hippocampal neurons and confirm that gp95-induced neuronal Ca2+ overactivation is mediated by CXCR4 and cGKII. Finally, cGKII activation is also required for HIV gp120-induced Ca2+ hyperactivation. These results thus provide a novel neurobiological mechanism of cGKII-mediated synaptic hyperexcitation in HAND.


Introduction
Human immunodeficiency virus (HIV)-associated neurocognitive disorders (HANDs) occur in as many as 50% of individuals infected with HIV, including patients receiving combination antiretroviral therapy (cART) [1]. HANDs range from mild neurological disorder (MND) and asymptomatic neurocognitive impairment (ANI) to severe and disabling dementia, and confer an increased risk of early mortality [1,2]. As cART enables individuals infected with HIV to survive to older ages, the prevalence of HAND continues to increase [1], and thus treatments targeting HIV's pathological processes in the brain are greatly needed. Previous knowledge of HAND neuropathogenesis is dependent on studies that have been predominantly carried out in the pre-ART era [3]. In fact, the majority of basic research on HAND has been focused on evaluating neuronal damage in the context of active viral replication and outcomes related to encephalitis and neuronal death [3]. Despite suffering from HAND, in patients with cART, the classical features of HIV encephalitis and/or brain atrophy often are absent [4]. In fact, the severity of HAND is strongly associated with the loss of synaptic markers in patients on cART [5]. However, the molecular mechanisms underlying HAND-associated synaptic alteration remain largely unclear [6].
One of the major limitations in searching for HAND cures has been the lack of an animal model that recapitulates all of the features of HIV infection in humans [7]. Thus, new animal models to examine how chronicity and aging affect HIV-induced neuropathology are an important current and future need [8]. Previous work has heavily relied on rodent models for the study of HIV pathology [9]. However, results obtained in rodent models are often not easily translated to treatment of humans, given that rodents are not naturally susceptible to HIV infection and do not reflect the in vivo nature of infection [10]. Although nonhuman primates infected with simian immunodeficiency virus (SIV) or genetic chimeras of SIV and HIV have a number of important advantages over small-animal models, they have obvious disadvantages, including considerable genetic variation, that greatly complicate studies using small numbers of animals and high maintenance costs [7]. Moreover, SIV is unable to cause acquired immune deficiency syndrome (AIDS) in its natural host [11,12]. In contrast, feline immunodeficiency virus (FIV) infection in domestic cats represents an animal model of immunodeficiency and shares similarities in pathogenesis with that of HIV in humans [11][12][13]. Certain strains of FIV can infect the central nervous system (CNS), leading to neurological symptoms similar to those observed in some individuals infected with HIV [13,14]. Importantly, FIV is a naturally occurring virus inducing both AIDS and neurological complications in animal models [15]. Furthermore, the combination of HIV antiretroviral drugs on naturally infected FIV cats in the late phase of the asymptomatic state of the disease significantly reduces viral load, indicating a similar pathogenesis of these viruses [16]. Therefore, FIV infection of cats is an attractive model to study the chronic neuropathogenesis of HAND. Little is known, however, about neuronal mechanisms underpinning overlapping neuropathology between FIV and HIV. Both HIV and FIV are tropic for lymphocytes and monocytes, utilizing CD4 (HIV) and CD134 (FIV) primary receptors together with the alpha chemokine receptor CXCR4 as a coreceptor to infect cells [17][18][19]. Even though lentiviral infection in the brain produces cortical and subcortical neuronal loss [20,21], HIV and FIV do not directly infect neurons but instead use a noninfectious interaction between the viral envelope and the neuronal surface [22,23]. Among HIV-induced neuron-damaging products, HIV envelope glycoprotein gp120 is one of the most prominent viral antigens found in the lysates of HIV-infected cells [24]. HIV gp120 indirectly and/or directly interacts with neurons, which enhances excitatory synaptic receptor activity, resulting in synaptic damages, but the mechanisms are not currently understood [25][26][27]. In neurons, the gp120 interaction with CXCR4 enhances Ca 2+ -regulating systems through NMDA receptors (NMDARs) in the synaptic membrane and inositol trisphosphate receptors (IP3Rs) in the endoplasmic reticulum (ER), resulting in apoptosis [28][29][30][31][32][33][34]. In addition, Ca 2+ fluxes through NMDARs promoting the production of nitric oxide (NO) by neuronal nitric oxide synthase (nNOS), which is tethered by the scaffolding protein postsynaptic density 95 (PSD95) [35][36][37][38][39]. NO subsequently exerts its effects by activating cGMP-dependent protein kinase II (cGKII) through the production of cGMP [40]. Notably, the NMDAR-nNOS-cGK pathway has been implicated in HIV-induced neurotoxicity [41,42]. However, the exact cellular role of the pathway on synaptic dysfunction in HAND has not been determined.
We have shown previously that cGKII can phosphorylate serine 1756 in neuronal IP3Rs and increase ER Ca 2+ release [43]. cGKII also phosphorylates the AMPA receptor (AMPAR) subunit GluA1, which triggers its synaptic trafficking, a critical step for inducing synaptic plasticity [43,44]. This suggests that cGKII activation is critical for HAND-associated synaptic dysfunction. Here, we demonstrate that FIV envelope glycoprotein, gp95, binds to CXCR4 on the neuronal plasma membrane and utilizes the same pathway as HIV gp120 to significantly increase intracellular Ca 2+ activity and synaptic activity in neurons. Thus, our results indicate that FIV serves as a model for HAND-associated synaptic hyperexcitation. Most notably, our study reveals the inclusion of cGKII in both FIV gp95 and HIV gp120-induced Ca 2+ hyperactivity, suggesting that cGKII inhibition may be a novel therapeutic target for HAND.

Gp95 increases surface expression of the AMPAR GluA1 subunit via cGKII activation
cGKII mediates phosphorylation of serine 845 of GluA1 (pGluA1), important for activitydependent trafficking of GluA1-containing AMPARs, and increases the level of extrasynaptic receptors [43,44]. Moreover, cGKII-mediated GluA1 phosphorylation is critical for hippocampal long-term potentiation (LTP) and learning and memory [43,44]. As gp95 was sufficient to induce cGKII activation (Fig 3), we hypothesized that gp95-induced cGKII activation increased GluA1 phosphorylation, which led to enhanced AMPAR-mediated synaptic activity. To test this idea, we first biochemically measured GluA1 phosphorylation levels when gp95 was applied. Mouse cultured cortical neurons were treated with 1-nM gp95 for 1 hour, and synaptosomes were isolated as described previously [47] (S3 Fig). We found that gp95 treatment was sufficient to increase GluA1 phosphorylation, while total GluA1 and GluA2 levels were not affected ( Fig 4A). To confirm whether such an increase was dependent on cGKII, we treated neurons with gp95 and 1-μM RP for 1 hour and measured pGluA1 levels. We revealed that inhibition of cGKII activity abolished the gp95 effects, while RP by itself was unable to affect AMPAR synaptic expression ( Fig 4A). To further confirm the role of cGKII in the elevation of GluA1 phosphorylation, we used cGKII KO neurons and found that gp95 treatment had no effect on GluA1 phosphorylation in KO neurons ( Fig 4B). Given that GluA1 phosphorylation promotes AMPAR surface expression, we next measured surface GluA1 levels by biotinylation after 1-nM gp95 was applied for 1 hour. We found that gp95 treatment increased surface GluA1 levels, which was blocked by pharmacological and genetic inhibition of cGKII (Fig 4C and 4D). Notably, cGKII inhibition by itself had no effect on GluA1 surface expression ( Fig 4C). Furthermore, gp95 treatment was unable to alter GluA2 and NMDAR subunit NR1 surface expression (S4 Fig). This suggests that cGKII is required for gp95-induced GluA1 up-regulation. To further confirm whether such an increase in surface expression of AMPARs elevates AMPAR-mediated synaptic transmission, we measured miniature excitatory postsynaptic currents (mEPSCs) in DIV14-17 cultured mouse hippocampal neurons (Fig 4E). We found that acute treatment of 1-nM gp95 significantly increased both average mEPSC amplitude (CTRL, 11.45 ± 1.07 pA; gp95, 21.07 ± 5.06, p = 0.04) and frequency (CTRL, 5.35 ± 0.37 Hz; gp95, 23.17 ± 3.29, p < 0.0001) (Fig 4E). This suggests that gp95-induced activation of cGKII increases surface expression of AMPARs, contributing to enhanced synaptic transmission.

Discussion
Although synaptic dysfunction, not neuronal death, is strongly associated with HAND [5], the molecular mechanisms underlying HAND-associated synaptic impairment remain largely unclear [6]. Previous studies document that FIV envelope proteins also elevate neuronal Ca 2+ and induce cell death in neurons [22,62,63]. However, cellular mechanisms of such FIV envelope protein-induced neurotoxic effects are unknown. We reveal that FIV envelope glycoprotein gp95 binds to CXCR4 on the neuronal plasma membrane and subsequently elevates intracellular Ca 2+ through mobilizing ER Ca 2+ via the stimulation of IP3Rs, as well as NMDARs, the same pathway of HIV gp120-induced Ca 2+ overactivation [18,49,50] (Fig 8).
Most notably, our study identifies that gp95-stimulated NMDARs activate the nNOS-cGMP-cGKII pathway, which subsequently phosphorylates IP3Rs and AMPAR subunit GluA1, leading to the elevation of surface GluA1 expression and AMPAR-mediated synaptic activity, a cellular basis of synaptic dysfunction in HAND (Fig 8). Moreover, we show that cGKII activation is required for Ca 2+ hyperactivity caused by HIV gp120 (Fig 7A and 7B), suggesting that cGKII activation plays crucial roles in synaptic dysfunction in both HIV and FIV models and there is a conserved cellular pathophysiology from mice and cats to humans. Although treatment of a lower dose of CNQX or DL-APV was unable to inhibit basal Ca 2+ activity, lower doses in combination completely inhibited Ca 2+ activity in both control and gp95-treated neurons (S5 Fig), suggesting that inhibition of both receptors induces additive effects on Ca 2+ activity. Given that there is NMDAR-independent Ca 2+ influx via L-type voltage-gated Ca 2+ channels [64], a lower dose of DL-APV alone is unable to block neuronal Ca 2+ activity completely. In fact, we found that 10-μM nifedipine, an antagonist of L-type voltage-gated Ca 2+ channels, was sufficient to abolish GCaMP5 activity in both control and gp95treated cells (S6B Fig). In addition, NMDARs significantly contribute to signaling at rest in the absence of AMPAR activity [65], although Ca 2+ permeability through NMDARs at negative membrane potentials is restricted because of their blockade by extracellular Mg 2+ ions [66,67]. Taken together, although AMPAR-mediated dendritic depolarization is required for removal of Mg 2+ ions for NMDAR activity, both receptors can also contribute to neuronal Ca 2+ activity in parallel.
https://doi.org/10.1371/journal.pbio.2005315.g008 leading to microglial activation during neurodegenerative inflammation [69]. Importantly, gp120 elevates synaptic receptor activity by enhancing the release of pro-inflammatory cytokines from activated microglia [70,71]. Among those cytokines, tumor necrosis factor alpha (TNFα) induces a rapid increase in mEPSC amplitude and frequency [72][73][74], as seen in gp95-treated neurons (Fig 4E). Although we used Neurobasal Medium designed for significantly less proliferation of glia [75], we were unable to completely remove microglia in our culture. This thus suggests that microglial activation by gp120 and gp95 can promote TNFα release, resulting in elevation of mEPSC frequency and amplitude.
Both HIV gp120 and FIV gp95 interact with CXCR4 to produce IP3, which can induce IP3R-mediated Ca 2+ efflux from the ER (Fig 2). However, we found that TTX completely abolished Ca 2+ activity (S6 Fig). This suggests that IP3 production through the interaction between gp120/gp95 and CXCR4 is not sufficient to increase neuronal Ca 2+ activity in the absence of neuronal activity. Furthermore, gp95-induced stimulation of AMPARs and NMDARs is insufficient to elevate Ca 2+ activity in neuronal cell bodies in the absence of IP3R-mediated ER Ca 2+ release (Fig 8), as indicated by decreased Ca 2+ activity via the inhibition of ER Ca 2+ channels (Fig 2). In our model, we thus propose that neuronal activity-driven stimulation of extracellular Ca 2+ influx through L-type voltage-gated Ca 2+ channels and NMDARs and Ca 2+ efflux from the ER coordinate and contribute to somatic Ca 2+ activity (Fig 8). Further work is necessary to differentiate the roles of these Ca 2+ channels on HIV-induced neuronal hyperexcitability.
The chemokine receptors CCR5 and CXCR4 are co-receptors together with CD4 for HIV entry into target cells [17]. Macrophage-tropic HIV viruses use CCR5 as a co-receptor [76][77][78][79][80], whereas T-cell line-tropic viruses use CXCR4 [81,82]. Given that most of the HIVinfected cells in the brain are macrophages and microglia, it is thought that CCR5 strains of HIV are the predominant viral species in the brain [83,84]. However, once HIV infection is established, dual tropic and CXCR4-preferring viruses slowly evolve from macrophage-tropic HIV viruses as an indication of progression to AIDS and HIV-associated dementia [17,[85][86][87][88]. Moreover, CXCR4-and dual-tropic strains of HIV have been isolated from the brains of infected individuals [86]. Therefore, CXCR4-tropic strains of HIV also play critical roles in the pathogenesis of HAND. Several studies have shown that the HIV gp120 binding to both CCR5 and CXCR4, even without CD4, contributes to neuronal injury and death both in vitro and in vivo [29-31, [89][90][91][92][93]. Interestingly, a CCR5 antagonist prevents gp120 neurotoxicity [94,95], and natural CCR5 ligands confer protection upon neurons against gp120 toxicity [61]. Conversely, HIV-induced apoptosis can be prevented by AMD3100, a CXCR4 antagonist, both in vitro and in vivo [96][97][98]. This suggests that CXCR4-mediated signaling can trigger HIVinduced neurotoxicity while CCR5 either protects or disrupts the CNS, depending on the context, ligand characteristics, and resultant signaling pathway. Surprisingly, CCR5-tropic gp120 (JRFL) also requires cGKII activation to induce Ca 2+ hyperexcitation (Fig 7B). It has been shown that chemokines and their receptors coordinate the signaling at the immunological synapses. In fact, during T-cell activation, CXCR4 and CCR5 chemokine receptors are recruited into the immunological synapse by antigen-presenting cell-derived chemokines [99]. In addition, the co-stimulatory properties of CCR5 and CXCR4 depend on their ability to form heterodimers [100]. Thus, gp120 (JRFL)-induced stimulation to CCR5 can interact with CXCR4, resulting in cGKII activation. Notably, the natural ligand of CXCR4, SDF-1, is also sufficient to induce cGKII-dependent Ca 2+ overactivation (Fig 7C). Taken together, CCR5-CXCR4 stimulation is sufficient to induce hyperexcitation in neurons. While both CXCR4 and CCR5 are important in the neuropathogenesis of HIV, it is clear that further study of the downstream pathways of CCR5 and CXCR4 activation in neurons will widen the understanding of HIVinduced neuronal toxicity. Given that FIV also targets primary CD4 T cells but uses CD134 instead of CD4 as a primary receptor and uses its sole co-receptor CXCR4 for efficient infection of target cells, similarly to T cell-tropic strains of HIV [18,19,49], FIV infection of cats is an ideal in vivo model to investigate CXCR4-mediated neuropathology in chronic HIV infection.
Our work extends beyond understanding of molecular mechanisms underlying HIVinduced neuronal dysfunction. One of the challenges that the HAND research community has faced in the treatment of this disorder is the lack of a viable target [1]. By identifying cGKII as the downstream effector of the gp95/120-induced synaptic hyperexcitation, our study completes the pathway and identifies cGKII as a new therapeutic target for limiting gp95/120induced synaptic dysfunction. Moreover, we reveal that CCR5-tropic gp120-induced Ca 2+ overactivation is also dependent on cGKII (Fig 7B). This thus suggests that cGKII activation is important for CCR5 and CXCR4-dependent neuropathology in HAND. Inhibition of cGKII may be superior as a therapeutic target to other forms of ER Ca 2+ release control, as its inhibition will limit the NMDAR-induced and IP3R phosphorylation-dependent Ca 2+ increase specifically, which are likely to be elevated under hyperexcitable conditions, while leaving basal functions unchanged. Thus, use of cGKII inhibition as a means for neuroprotection in individuals infected with HIV is a novel and innovative approach to this therapeutically challenging disease pathway.

Ethics statement
Colorado State University's Institutional Animal Care and Use Committee reviewed and approved the animal care and protocol (16-6779A).

Mouse and feline neuron culture
Mouse hippocampal and cortical neuron cultures were prepared as described previously [43,46,47]. Neurons were isolated from embryonic day 17-18 or postnatal day 0.5 C57Bl6 or cGKII KO mouse embryonic brain tissues. For feline hippocampal neurons, embryos were obtained by cesarean section at approximately 35-40 days gestation from specific pathogenfree cats. Hippocampi were isolated from embryos and digested with 10 U/mL papain (Worthington Biochemical Corp., NJ). Mouse cortical neurons were plated on polylysinecoated 15-cm dishes (25 million cells per dish) and 6-well dishes (500,000 cells per well) for biochemical experiments. Mouse and feline hippocampal cells were prepared in glass-bottom dishes (500,000 cells in the glass bottom) for Ca 2+ imaging. Mouse hippocampal neurons were also plated on 12-mm coverslips for electrophysiology (200,000 cells per coverslip). Cells were grown in Neurobasal Medium with B27 and 0.5 mM Glutamax and penicillin/streptomycin (Life Technologies).

GCaMP5 Ca 2+ imaging
GCaMP5 Ca 2+ imaging was carried out by a modification of the previously reported method [43,46,47]. DIV4 Neurons were transfected with pCMV-GCaMP5 (a gift from Douglas Kin and Loren Looger, Addgene plasmid #31788) [105] by using Lipofectamine 2000 (Life Technologies) according to the manufacturer's protocol. The transfection efficiency is about 5%, and obvious cellular toxicity has not been observed. Neurons were grown in Neurobasal Medium without phenol red supplemented with B27 and 0.5-mM Glutamax and penicillin/ streptomycin (Life Technologies) for 8-12 days after transfection and during the imaging. Glass-bottom dishes were mounted on a temperature-controlled stage on Olympus IX73 and maintained at 37˚C and 5% CO 2 using a Tokai Hit heating stage and digital temperature and humidity controller. The imaging was captured for periods of 50 milliseconds using a 60× oilimmersion objective. A total of 100 images was obtained with 1-second interval, and Ca 2+ activity in the cell body (excluding dendrites) was analyzed using the Olympus CellSens software. F min was determined as the minimum fluorescence value during the imaging. Total Ca 2+ activity was obtained by combining 100 values of ΔF/F min = (F t − F min )/F min in each image, and values of ΔF/F min < 0.2 were rejected due to bleaching. Twenty to thirty neurons were used for imaging in one experiment, and one individual neuron was assayed in one imaging.

Statistics
Statistical comparisons were analyzed with the GraphPad Prism6 software. Unpaired twotailed Student t tests were used in single comparisons. For multiple comparisons, we used oneway ANOVA followed by Fisher's Least Significant Difference (LSD) test to determine statistical significance. Results were represented as mean ± SEM, and p < 0.05 was considered statistically significant. (TIF) S1 Data. Contains raw numerical values that underlie the summary data displayed in the following figure panels: Figs 1A-1C, 2, 3A-3D, 4A-4E, 5, 6, 7A-7C, S2, S4A, S4B