Osteopontin and Integrin Mediated Modulation of Post-Synapses in HIV Envelope Glycoprotein Exposed Hippocampal Neurons

The advent of Human Immunodeficiency Virus (HIV) antiretrovirals have reduced the severity of HIV related neurological comorbidities but they nevertheless remain prevalent. Synaptic degeneration due to the action of several viral factors released from infected brain myeloid and glia cells and inflammatory cytokines has been attributed to the manifestation of a range of cognitive and behavioral deficits. The contributions of specific pro-inflammatory factors and their interplay with viral factors in the setting of treatment and persistence are incompletely understood. Exposure of neurons to chemokine receptor-4(CXCR4)-tropic HIV-1 envelope glycoprotein (Env) can lead to post-synaptic degradation of dendritic spines. The contribution of members of the extracellular matrix (ECM) and specifically, of perineuronal nets (PNN) toward synaptic degeneration, is not fully known, even though these structures are found to be disrupted in post-mortem HIV-infected brains. Osteopontin (Opn, gene name SPP1), a cytokine-like protein, is found in abundance in the HIV-infected brain. In this study, we investigated the role of Opn and its ECM integrin receptors, β1- and β3 integrin in modifying neuronal synaptic sculpting. We found that in hippocampal neurons incubated with HIV-1 Env protein and recombinant Opn, post-synaptic-95 (PSD-95) puncta were significantly increased and distributed to dendritic spines when compared to Env-only treated neurons. This effect was mediated through β3 integrin, as silencing of this receptor abrogated the increase in post-synaptic spines. Silencing of β1 integrin, however, did not block the increase of post-synaptic spines in hippocampal cultures treated with Opn. However, a decrease in the PNN to βIII-tubulin ratio was found, indicating an increased capacity to support spine growth. From these results, we conclude that one of the mechanisms by which Opn counters the damaging impact of the HIV Env protein on hippocampal post-synaptic plasticity is through complex interactions between Opn and components of the ECM which activate downstream protective signaling pathways that help maintain the potential for effective post-synaptic plasticity.


Introduction
Neurocognitive problems due to human immunodeficiency virus type 1 (HIV-1) infection are prevalent and comprise a spectrum of impairments that cannot be fully controlled by suppressive antiretrovirals. The virus enters the brain early and is less receptive to current therapy [1]. The advent of antiretrovirals has reduced the severity of neuronal degeneration in HIV infection, but it still affects 15-55% of the HIV+ population [1]. Understanding the detailed molecular mechanisms that instigate neurological impairments due to HIV infection is required to develop therapeutic approaches targeting this comorbidity.
that Opn and β1 integrin functioning in this context are not co-dependent. Likely due to their tight association with the ECM, silencing β1 integrins resulted in decreased expression of perineuronal nets (PNN) and increased growth of PSD-95 post-synaptic structures. Treatment with Opn alone also led to a decrease in the density of PNNs. In contrast, silencing β3 integrins did not decrease the number post-synapses in the presence of both Opn and HIV-1 Env, but interestingly it did in Opn-only treated neurons. This finding suggests the possibility of convergent signaling between β3 integrin and Env receptor pathways. Surprisingly, in HIV-1 Env only treated neurons, silencing β3 integrins also decreased the number of post-synapses suggesting activation of protective compensatory signaling that ultimately serves to block further damage.

Primary Hippocampal Neuron Culture
Dissociated primary hippocampal neurons were cultured from rat hippocampi digested with 1 mg/mL papain from E18 embryos (Neuromics) and depending on the downstream experiment, 20-40,000 cells were cultured on poly-D-lysine (PDL) coated coverslips (Neuvitro) for microscopy. The cells were supplemented with Neurobasal plus media containing Glutamax and B27+. CultureOne supplement was used to limit glial growth (Gibco), which according to manufacturer's instructions, eliminates any glial growth when supplemented at DIV 0. Cells were grown at 37 • C and in 5% CO 2 for 7 or 14 days in vitro.

SiRNA Silencing and Treatments
Receptor expression was modulated using anti-integrin silencing performed with the SiGENOME SMARTpool siRNA directed against rat β1 and β3 integrins using Dharmafect 1 transfection reagent on DIV 7 neurons. GFP-SiRNA were utilized to determine the ideal concentration of the transfection reagent and confirm successful transfection. Green fluorescent protein (GFP) positive puncta were observed in the entire well along with a low-level of cell death as observed by neurite blebbing.
DIV 12 neurons were subjected to HIV-1 Env (400 pM) and rat recombinant Opn (R&D systems, 100 ng/mL) for 48 h. Cells were harvested on DIV 14 for synaptic analysis and subjected to 4% paraformaldehyde fixation and subsequent antibody staining for immunofluorescence.

Image Analysis
Confocal and apotome images (at least 12 per treatment) were quantified for the number of dendritic spines per neurons and the immunoreactivity of the markers used. All image analysis and quantification was done using custom codes on Matlab R2017a (MathWorks) available upon request.

Experimental Groups
The first portion of this study had four non-transfected groups. Vehicle + Opn-only received neurobasal plus media on DIV 12. Vehicle + Opn+ received neurobasal plus media with 100 ng/mL rat recombinant Opn. Env + Opn-were treated with neurobasal plus containing 400 pM of HIV-1 Env. Env + Opn+ neurons were co-treated with 100 ng/mL rat recombinant Opn and 400 pM of HIV-1 Env in neurobasal plus media.
Silencing studies were performed next with scrambled siRNA as controls, β1ITG siRNA to silence β1 integrins and β3ITG to silence β3 integrins. Silencing were performed on vehicle treated, Env + Opn+, Opn only, and Env only treated neurons. Scrambled controls were used for comparisons analysis with both β1 and β3 integrins.

Statistics
All graphs are represented as mean ± standard deviation (SD). Statistics were done using Graphpad Prism 7. Two-way analysis of variance (ANOVA) with Tukey post hoc analysis were performed on multiple comparisons. Column factor (Degrees of freedom DF = 1) for Figure 1 was Env treated or not and row factor (DF = 1) was Opn treated or not. Column factor for Figures 2-5 (DF = 1) where SiRNA transfection and row factors (DF = 3) were Opn only, Env only, Env + Opn, and vehicle treatments (exogenous treatments). Sampling (n) was determined by the number of times the experiments were repeated, the number of observations is the number of images acquired (individual point on the interleaved scatters), the number of neurons were the number of neurons analyzed, and the alpha value for significance was set to be 0.05.

HIV-1 Env Mediates Post-Synaptic Remodeling by Decreasing the Number of Dendritic Spines, but Co-Treatment with Opn Reverses This Damage
The density and architectural integrity of post-synaptic dendritic spines has been implicated as molecular correlates of long-term potentiation. The number of dendritic spines reflect the status of neuronal plasticity as they protrude out of the dendrites, serving as inputs for pre-synaptic boutons extending from axons [42]. Thus, a higher number of dendritic spines indicates an increased number of contacts are being formed with the pre-synapse, and reflect the degree of synaptic strength. Therefore, we evaluated the number of dendritic spines formed by the post-synaptic scaffolding protein PSD-95 in the presence of Opn and HIV-1 Env. Two-way ANOVA analysis revealed a significant difference in post-synapses between vehicle control ( Figure 1, black circles) and exposure of neurons to HIV-1 Env ( Figure 1, red circles) (Column factor p = 0.0023) ( Figure 1D,E). Exogenous application of HIV-1 Env only (Env+) induced a decrease in the number of dendritic spines per neuron when compared to vehicle-treated neurons (p = 0.0029) ( Figure 1E). Three-dimensional image stack reconstruction, analyses also showed PSD-95 concentration preferentially on dendrites as opposed to localization to spines in neuronal cultures exposed to HIV-1 Env ( Figure 1A,B). In contrast, on vehicle treated neurons PSD-95 post-synapses were predominantly distributed in spines. Co-treatment of neurons with HIV-1 Env and Opn (Env + Opn+) showed reversal of these damages as there were no differences between the vehicle+Env-or Opn only treated neurons ( Figure 1A-D, 1E: left pair under OPN+). Co-treatment with Opn also distributed PSD-95 to the spines ( Figure 1D).

Opn Acts Independently of Extracellular Matrix (ECM) Component, β1 Integrin, to Regulate Hippocampal Post-Synapses in the Presence of HIV-1 Env
To understand the mechanism utilized by Opn to achieve the reversal of post-synaptic damage inflicted by HIV-Env, one of its receptors, β1 integrin, was silenced (siRNA-KO). Two-way ANOVA revealed an interaction between silencing of β1 integrin and the exogenous treatments (Interaction p = 0.0003). This finding therefore, suggested that changes in spine density were contingent on both β1 integrin silencing and treatments with HIV-1 Env and Opn. Spine density was also found to be dependent on HIV-1 Env and Opn treatments (Row factor p = 0.0003). In this regard, there were no differences between β1 integrin silencing on vehicle-treated and scrambled transfected control neurons suggesting that this integrin does not regulate post-synaptic spine density during physiological conditions in DIV 14 neurons (Figure 2A,B,I:Vehicle circle pairs). Interestingly, β1 integrin silencing increased the number of post-synapses in neurons co-treated with Env + Opn+ (p = 0.0093) when compared to scrambled transfected ( Figure 2C,D,I:Env + Opn circle pairs). This increase was also significant when compared with vehicle-treated neurons transfected with reduced β1 integrin expression (p = 0.0026) ( Figure 2B,D,I:Vehicle, red circles vs Env + Opn, red circles). There were no differences between scrambled siRNA transfected, Opn-only treated neurons and β1 integrin siRNA-KO neurons ( Figure 2I: Opn (circle pairs). These finding suggests that Opn's ability to modulate spine PSD-95 post-synapses in hippocampal neurons is not dependent on β1 integrin. Additionally, no differences were found between scrambled siRNA transfected Env-only treated neurons and β1 integrin silenced neurons ( Figure 2G,H,I: Env circle pairs).    (I) shows grouped interleaved scatter graph for the number of synapses per neuron due to the different treatments. The mean ± SD are shown with the following number of sampling for each groups: n = 5 and 117 neurons for vehicle treated scrambled transfected, n = 2 and 42 neurons for vehicle treated β1 integrin SiRNA transfected, n = 7 and 120 neurons for Env + Opn treated and scrambled transfected, n = 2 and 13 neurons for Env + Opn treated β1 integrin siRNA transfected, n = 2 and 66 neurons for Opn treated scrambled transfected, n = 3 and 28 neurons for Opn treated β1 integrin siRNA transfected, n = 2 and 16 neurons for Env only treated and scrambled transfected and n = 2 and 13 neurons for Env only treated β1 integrin siRNA transfected neurons. Two-way ANOVA with Tukey's post-test was performed to reveal statistical difference between groups (* p < 0.01 and ** p < 0.001).

Perineuronal Net (PNN) Expression Is Decreased in β1 Integrin Silenced Neurons Co-Treated with HIV-1 Env and Opn, and in Neurons Treated Only with Opn Compared to Vehicle Controls
Due to β1 integrin's tight association with PNNs, we next wanted to understand how silencing of this integrin impacted the expression of PNNs, another key member of the ECM. We first quantified total PNN density on hippocampal neurons to determine whether the experimental treatments alone altered its expression. PNN expression was normalized to βIII-tubulin levels (PNN/β3-tubulin) and two-way ANOVA revealed a significant interaction between exogenous treatments and siRNA transfections (Interaction p = 0.006) ( Figure 3G). No differences were found between vehicle-treated neurons either transfected with scrambled siRNA, or β1 integrin silencing confirming that β1 integrins do not regulate the expression of PNN/βIII-tubulin during physiological conditions in DIV 14 neurons (Figure 3A,B,G:Vehicle circle pairs). However, in the co-presence of HIV-1 Env and Opn there was a significant decrease in the PNN/βIII-tubulin expression when compared to HIV-1 Env and Opn co-treated and scrambled siRNA transfected neurons (p = 0.0320) (Figure 3C,D,G: Env + Opn circle pairs). A decrease in PNN/βIII-tubulin expression in Opn-only treated neurons was observed when compared to scrambled siRNA transfected HIV-1 Env and Opn co-treated neurons (p = 0.0049) (Figure 3C,E,G: Env + Opn, black circles vs Opn, red circles), and to vehicle-treated neurons transfected with β1 integrin silencing (p = 0.0070) (Figure 3B,E,G: Vehicle, red circles vs Opn, red circles). The decrease in the PNN/βIII-tubulin ratio was not β1 integrin dependent since silencing this receptor in Opn-only treated neurons did not change the expression of PNN/βIII-tubulin when compared to the scrambled transfected Opn-only treated neurons ( Figure 3E-G: Opn, black vs red circles). Collectively, these results suggest that Opn treatment can override the damage signal induced by HIV Env to reduce the PNN/βIII-tubulin ratio, and does so more effectively when β1 integrin expression is blocked ( Figure 3G: Env + Opn, circle pairs).

β3 Integrins Are Required for Opn Induced Upregulation of Post-Synaptic Dendritic Spines
Next, β3 integrins were silenced to determine whether they were involved in the Opn mediated regulation of post-synapse phenotype. A significant interaction between siRNA transfections and the exogenous treatments was detected (Interaction p = 0.0010) ( Figure 4I). Changes to spine density were also dependent on siRNA transfections (Column factor p = 0.0030). There were no significant differences between scrambled siRNA transfected and β3 integrin silencing siRNA control treated neurons, suggesting that these integrins do not regulate post-synaptic spines during physiological conditions ( Figure 4A,B,I: Vehicle, circle pairs). There was however a decrease in the number of spines in Opn-only treated β3 integrin silenced neurons when compared to Opn-only treated scrambled siRNA transfected neurons (p = 0.0232) ( Figure 4E,F,I: Opn, circle pairs). This finding suggests that Opn interacts with β3 integrins to regulate the density of spines in hippocampal neurons. Surprisingly, there was a significant decrease in the number of spines in HIV-1 Env treated neurons when β3 integrins were silenced compared to vehicle treated scrambled transfected neurons (p = 0.0063) ( Figure 4I, Env, circle pairs), Env + Opn treated, β3 integrin silenced neurons (p = 0.0253) ( Figure 4I, Env + Opn, red circles), and Opn-only treated scrambled transfected neurons (p < 0.0001) ( Figure 4D

Opn Acts Independently of Extracellular Matrix (ECM) Component, β1 Integrin, to Regulate Hippocampal Post-Synapses in the Presence of HIV-1 Env
To understand the mechanism utilized by Opn to achieve the reversal of post-synaptic damage inflicted by HIV-Env, one of its receptors, β1 integrin, was silenced (siRNA-KO). Two-way ANOVA revealed an interaction between silencing of β1 integrin and the exogenous treatments (Interaction p = 0.0003). This finding therefore, suggested that changes in spine density were contingent on both β1 integrin silencing and treatments with HIV-1 Env and Opn. Spine density was also found

In HIV-1 Env and Opn Treated Neurons, PNN Expression Is not Regulated by β3 Integrins
Next, we tested whether β3 integrins regulate the expression of PNNs in the presence HIV-1 Env and Opn. There was a significant interaction between siRNA transfections and exogenous treatments as revealed by two-way ANOVA (p = 0.0045) ( Figure 5I). PNN expression was also highly dependent on the exogenous treatments (Row factor p = 0.0356). There were no differences between vehicle scrambled transfected neurons or their β3 integrin silenced counterparts suggesting that this receptor is not involved in regulating the PNN/βtubulin expression ( Figure 5I: all groups and their circle pairs). There was a decrease in Opn only treated scrambled transfected neurons when compared to Env and Opn co-treated and scrambled transfected (p = 0.0071) and Env only treated and scrambled transfected (p = 0.0466) ( Figure 5I: Env, black circles).

Opn Acts Independently of Extracellular Matrix (ECM) Component, β1 Integrin, to Regulate Hippocampal Post-Synapses in the Presence of HIV-1 Env
To understand the mechanism utilized by Opn to achieve the reversal of post-synaptic damage inflicted by HIV-Env, one of its receptors, β1 integrin, was silenced (siRNA-KO). Two-way ANOVA revealed an interaction between silencing of β1 integrin and the exogenous treatments (Interaction p = 0.0003). This finding therefore, suggested that changes in spine density were contingent on both β1 integrin silencing and treatments with HIV-1 Env and Opn. Spine density was also found to be dependent on HIV-1 Env and Opn treatments (Row factor p = 0.0003). In this regard, there were no differences between β1 integrin silencing on vehicle-treated and scrambled transfected control neurons suggesting that this integrin does not regulate post-synaptic spine density during physiological conditions in DIV 14 neurons (Figure 2A,B,I:Vehicle circle pairs). Interestingly, β1 integrin Two-way ANOVA with Tukey's post-test was performed to reveal statistical difference between groups (* p < 0.01 and ** p < 0.001).

Discussion
The molecular and electrical properties of neuronal circuitry are governed by synapses. Indeed, synaptic integrity is crucial to maintain healthy communication between neurons and any interruption will trigger aberrant activity. Synaptic activity is governed by many complex factors contributed by neurons and glia, and any changes in the neuronal microenvironment will have a positive or negative impact [43]. Indeed, neuronal dysfunction during HIV infection in the form of morphological and functional changes that lead to a disruption the neuronal circuitry manifests as neurocognitive problems [44]. In this regard, high levels of HIV-1 Env (5 nM) have been shown to induce cytoskeletal changes in neurons [45]. Gp120 transgenic mice also exhibit decreased dendrites and synapses in addition to reactive gliosis in the brain [8,46]. The neuropathological findings were reflected in altered behavior on learning and memory tasks [46,47]. Although antiretroviral therapy has greatly extended the normal lifespan of people living with HIV infection, neurocognitive impairment in the setting of aging and inflammation remains an urgent problem [48,49]. Therefore, it is imperative to evaluate the changes that take place in the CNS microenvironment during HIV infection and to assess the collective contributions from viral particles and the ensuing inflammatory signaling.
In this report, we studied the role of Opn, a secreted protein elevated during neuroinflammation in the HIV infected brain, in modulating hippocampal synaptic architecture [13,15]. Opn is a known biomarker for various pathologies both in the CNS and periphery with circulating serum levels found to be >100 ng/mL [20,21,37,38]. Opn's function in the brain has been regarded as both protective and detrimental, depending on the cellular location, pathology, and post-translational modification [16][17][18]. Opn expression has been previously shown to be elevated in developing neurons, implicating its role in the differentiation and maturation of neurons in the developing brain [50,51]. It is multifunctional, having pro-angiogenic, pro-survival, chemotactic, as well as adhesive properties. Opn can be post-translationally modified by phosphorylation, glycosylation, and by proteolytic cleavage by thrombin or matrix metalloproteinases [29,52,53]. Opn as a secreted protein can induce an outside-in stimulant for integrins by directly binding to them [54,55].
Due to its diverse features, its role in several neurological disorders including neuroHIV remains under-explored. Hence, we decided to investigate the role of Opn at serum level concentration of 100 ng/mL on neurons to understand its effect at the molecular level in the presence of HIV-1 Env. In contrast to other reports [45], we utilized a low concentration of HIV Env (400 pM) to also match the serum levels found in patients (100-800 pM) [56]. Additionally, we chose hippocampal neurons in our experiments due to their key function in the brain in learning and memory. Moreover, the hippocampus is one of the key regions in the brain that is negatively impacted by HIV [46]. Synaptic degradation in the hippocampus is apparent in HIV infection and the pathways are not completely understood [46,57]. The degradation is reportedly a combinatory effect by soluble factors released by the virus and circulating pro-inflammatory cytokines [7].
We showed here that elevated levels of extracellular Opn in the presence of HIV-1 Env can counteract the detrimental effect of HIV-1 Env on post-synaptic PSD-95 in primary rat hippocampal neurons. The presence of HIV-1 Env reduced the expression of PSD-95 and decreased its translocation to the spines thereby reducing synapse density. A decline in the number of post-synaptic spines indicates a decrease in the number of contacts being formed with the pre-synaptic membrane and hence, would result in the weakening of excitatory connections [10]. Adding Opn as a co-treatment recovered their number comparable to vehicle treated neurons, and distributed them to the spines according to our experiments.
We targeted β1 and β3 integrins to determine whether Opn acts on these receptors to exert a protective effect in the presence of HIV-1 Env. In that regard, if Opn utilizes any of these integrins then the number of dendritic spines would be expected to decrease with a reduction of integrin expression. However, silencing β1 integrins did not reduce, but rather increased the number of spines in the presence of Opn and HIV-1 Env. Therefore, we believe these integrins are involved in spine regulation in the co-presence of Opn and HIV-1 Env. β1 integrins have been previously shown to be highly involved in neuronal plasticity and can increase or decrease excitability depending on the ligand [25,26]. The proteolytic enzyme MMP9 is usually released by neurons due to increased activity [24]. This enzyme can then digest the ECM including PNNs and release soluble cell adhesion molecules (CAMs) [24,58]. One of these CAMs is the intercellular adhesion molecule-5 (iCAM), which was shown to stimulate β1 integrins [58][59][60]. This activation of integrins promoted dendritic spine enlargement via cofilin inactivation and F-actin polymerization in hippocampal slices [25,35]. On the other hand, in the presence of the astrocytic derived extracellular matrix protein thrombospondin, β1 integrins depolymerized F-actin and lead to endocytosis of synaptic receptors in cultured spinal cord neurons thereby reducing neuronal excitability [26]. From our results, we noted a decrease in the PNN/β3-tubulin expression in β1 integrin silenced neurons co-treated with Opn and HIV-1 Env. Degradation of PNNs have been suggested to promote dendritic spine formation [24] which could explain why we observed an increase in spine density in β1 integrin silenced neurons co-treated with Opn and HIV-1 Env.
We also observed similar decrease in PNN/βIII-tubulin expression and an increase in spine density in Opn only treated neurons. Opn has been shown to increase the expression of MMP9 via β3 integrins in Neuro2A cells, human chondrosarcoma cells, cardiac, skeletal muscles, and endometrial cells [22,[61][62][63]. It is possible that they also increase MMP9 in this context and promote spine density by decreasing PNN expression and that needs to be shown in future studies. The increase in spine density was reversed by silencing β3 integrins suggesting these to be targeted by Opn to modulate dendritic spines. However, β3 integrins did not modulate the expression of PNNs as we did not see any difference in PNN levels when they were silenced. β3 integrins have been previously found to regulate synaptic maturation by enhancing the probability of neurotransmitter release and increasing post-synaptic density in developing hippocampal neurons [64]. β3 integrins have also been shown to mediate hippocampal dendritic spine remodeling via actin reorganization through calcium dependent mechanisms [33]. Therefore, β3 integrins in the presence of Opn are believed to positively enhance the number and density of post-synaptic structures without regulating the expression of PNNs. In the absence of these integrins, HIV-1 Env-only treated neurons exhibited even lower spine density suggesting the dysregulation prompted by HIV-1 Env is exacerbated by knocking the expression of β3 integrins down. This alludes to an intrinsic function of β3 integrins in the presence of HIV-1 Env and adds to the evidence that integrins themselves may have roles in mediating synaptic plasticity in the presence of HIV-1 Env. However, we did not observe this protective effect of β3 integrins in the co-presence of Opn and HIV-1 Env, suggesting a complex cross-talk between the three. Future studies involving the detailed interaction between HIV-1 Env and β3 integrins are therefore warranted. Further investigations into the intracellular signals regulated by β3 integrins might also shed light on whether there is a feedback loop due to the presence of both Opn and HIV-1 Env.
In conclusion, our work demonstrates post-synaptic modulation of hippocampal neurons by HIV-1 Env, Opn and integrins and predicts some of the mechanism by which the synaptic integrity can be impaired during HIV infection. We show here that pathological levels HIV-1 Env can reduce the expression of PSD-95 and reduce the number of dendritic spines. This would result in a reduced number of contacts being formed with the post-synaptic machinery and thus suggests weaker synaptic transmission. The ECM acts to counteract this damage by allowing the neurons to be more plastic. The ECM is very complex and we have only studied a few of the components here but offers an exciting avenue for further exploration. This includes but is not limited to the role of MMP9s and the dynamics of PNNs in the presence of HIV-1 proteins. Future studies using specific PNN markers are therefore warranted to understand the complexity that arises in the presence of HIV-1 Env. Their composition changes, if any, might give a clue as to how synaptic function is dysregulated.
In addition, we have shown here how secreted phosphoprotein Opn exerts a protective effect on synapses. However, there are some limitations as our studies were performed in in vitro cultures and do not completely capture the complexity of the brain. Therefore, future in vivo or ex vivo studies with Opn knockout mice and gene transfected organoid cultures are warranted to confirm whether our in vitro findings are reflected as molecular changes in the brain, and associated with the expected behavioral changes perhaps recapitulating aspects of deficits in fact observed during neuroHIV.