The Role of Shed PrP^c in the Neuropathogenesis of HIV Infection

Thirty-seven million people are living with HIV worldwide (1). Even with the success of combined antiretrovirals (cARTs), ∼40–70% of HIV-infected people develop a spectrum of cognitive and motor abnormalities that are collectively termed HIV-associated neurocognitive disorders (HANDs) (2–4). Because HIV-infected people are living longer, the prevalence of cognitive impairment is increasing and low-level chronic neuroinflammation and neuronal damage remain a significant public health issue (5).

HIV enters the CNS early after peripheral infection, predominantly through HIV-infected monocytes that cross the blood–brain barrier (BBB) (3, 6–8). After entering the CNS, these infected cells differentiate into perivascular macrophages and release virus that can infect macrophages, microglia, and to a lesser extent, astrocytes, as well as activate resident CNS cells (9–11). Activated and infected cells elaborate neuroinflammatory factors including CCL2/MCP-1 and TNF-α (12–14). CCL2, a potent monocyte chemoattractant, is elevated in the CNS and cerebrospinal fluid of individuals with HIV and remains elevated even with successful cART (15, 16). This can result in ongoing monocyte transmigration, reseeding of the CNS with virus, and chronic low-level inflammation and neuronal damage (17). TNF-α is produced predominantly by astrocytes and microglia, and can induce the production of cytokines and neurotoxic host factors, as well as activate HIV replication and cause the production of viral proteins (18–20). In addition, in neurons, this cytokine has been shown to cause apoptosis (21). Thus, TNF-α plays a key role in the progression of HIV CNS disease.

The cellular isoform of the human prion protein (PrP^c) is the nonpathological isoform of the prion protein and is an adhesion molecule that is constitutively expressed by CNS cells including astrocytes, brain microvascular endothelial cells, neurons, as well as infiltrating leukocytes (22–25). PrP^c is expressed on the cell membrane, where it is GPI anchored and can be cleaved and shed to the extracellular environment (26–28). In physiological conditions PrP^c has been shown to be involved in leukocyte activation, monocyte transmigration, cell adhesion, macrophage phagocytosis, and glutamate uptake in murine astrocytes isolated from PrP^c-deficient mice (29–32). However, its role in HIV neuropathogenesis is unknown.

Previously, our group demonstrated that HIV-seropositive people with HAND have higher amounts of shed PrP^c (sPrP^c) in their cerebrospinal fluid as compared with seropositive people with no impairment (33). In addition, in a pigtail SIV macaque model, sPrP^c levels correlated with encephalitis (33). In these studies, cerebrospinal fluid sPrP^c levels also correlated with cerebrospinal fluid CCL2, suggesting that this chemokine plays a role in the shedding of PrP^c (33). Other groups have shown that a recombinant protein of human PrP^c fused with the Fc portion of human IgG1 can activate signaling pathways including NF-κB, PI3K, and ERK in monocyte/macrophages (34). We therefore hypothesized that neuroinflammation during HIV infection of the CNS results in increased PrP^c shedding, and that sPrP^c participates in the development of HAND.

In this study, we demonstrate that CCL2 and TNF-α increase PrP^c shedding from astrocytes. We examined astrocytes because they are the most abundant CNS cell type and have key functions in the brain including maintenance of neuronal function and survival (35, 36). To study the mechanism of this increased shedding, we examined the effect of these cytokines on the disintegrin metalloprotease ADAM10, a protease that had been shown to cleave PrP^c in neurons. ADAM10 is a membrane-associated protein that can be found in two forms, active and inactive (37). The inactive form of ADAM10 has a propeptide covering its active site, preventing it from binding to its substrate. The active form of ADAM10, which is still cell membrane associated, has the proprotein cleaved and the metalloprotease domain freed to cleave proteins in close proximity to it in the plasma membrane, including N-cadherin, Notch, fractalkine, and amyloid precursor protein (38–42). We found that CCL2 and TNF-α increased PrP^c shedding by inducing increased expression of the active form of ADAM10. We also showed that a specific ADAM10 inhibitor (GI254023X) decreased this CCL2- and TNF-α–mediated PrP^c shedding to below baseline, demonstrating that this protease cleaves PrP^c in astrocytes.

In addition, we determined that sPrP^c contributes to neuroinflammation by mediating the production of CCL2, CXCL12, and IL-8 from astrocytes. We showed that supernatants of astrocytes treated with full-length human recombinant PrP^c (rPrP^c) that has the native conformation of PrP^c, but not rPrP^c by itself, cause increased chemotaxis of both uninfected and HIV-infected human primary monocytes above baseline. This suggests that factors produced from astrocytes in response to this protein are mediating chemotaxis and contribute to monocyte recruitment across the BBB and progression of HIV CNS pathogenesis. We also examined the role of rPrP^c in another important astrocytic function, glutamate uptake. rPrP^c decreased uptake of this metabolite in astrocytes, and this could lead to neurotoxicity and neuronal loss. Thus, we have shown that sPrP^c is involved in the processes of HIV neuropathogenesis and contributes to inflammation and neuronal damage.

Primary human astrocytes were obtained as part of a research protocol at the Albert Einstein College of Medicine and were cultured as previously described (43). To obtain pure astrocytes, we plated mixed cultures of neurons and astrocytes at a density of 4.5 × 10⁷ in DMEM supplemented with 10% heat-inactivated FBS, 1% penicillin/streptomycin, 1% nonessential amino acids (all from Thermo Fisher Scientific, Waltham, MA), and 2% HEPES (USB, Cleveland, OH). After 1 wk in culture, cells were dissociated and replated to eliminate microglia and neurons. After this, all cells are GFAP⁺.

Primary human astrocytes were cultured in 60-mm plates at a starting density of 300,000 cells for 4–5 d before treatments. Human recombinant CCL2 (PeproTech, Rocky Hill, NJ) was added to astrocytes in serum-free media to a final concentration of 200 ng/ml. Human recombinant TNF-α (PeproTech) was added to cells in serum-free astrocyte media at a concentration of 10 ng/ml. Control cultures were treated with ddH₂O, the CCL2 and TNF-α diluent. To examine whether CCL2 and TNF-α have synergistic or additive effects, we added CCL2 (200 ng/ml) and TNF-α (10 ng/ml) concomitantly to astrocyte cultures for some experiments. For experiments with the ADAM10 inhibitor (GI254023X; courtesy of Dr. Schmidt Technische Universität Darmstadt, Darmstadt, Germany) (44), astrocytes were treated with GI254023X (5 μM) concomitantly with CCL2 (200 ng/ml) or TNF-α (10 ng/ml).

Astrocytes were cultured in DMEM with 10% FBS on 96-well tissue culture plates at 10,000 cells per well for 2–3 d. Full-length human rPrP^c (Jena Bioscience, Jena, Germany) was added to fresh serum-free DMEM media to a final concentration of 10 μg/ml. Media were aspirated from wells, and 50 μl of rPrP^c-supplemented media was added to each well. Negative control cells received serum-free DMEM with an equal concentration of the diluent, NaAc. Positive control cells were treated with 10 ng/ml IL-1β to verify the astrocyte response to stimulation. After 24 h, media were collected from each well and analyzed by ELISA.

Leukopaks were obtained from the New York Blood Center, and PBMCs were isolated using Ficoll-Paque (GE Healthcare, Uppsala, Sweden). Monocytes were isolated from PBMCs using EasySep Human CD14 Positive Selection Kit according to the manufacturer’s instructions (STEMCELL Technologies, Vancouver, BC, Canada). CD14⁺ monocytes were cultured nonadherently in Teflon-coated flasks at a density of 2 × 10⁶ cells/ml in RPMI 1640 media (Thermo Fisher Scientific) supplemented with 10% heat-inactivated human serum type AB and 5% heat-inactivated FBS (Lonza, Walkersville, MD), 1% HEPES, 1% penicillin/streptomycin, and 10 ng/ml M-CSF (PeproTech) as previously described (45). This culture system yields 70% mature CD14⁺ CD16⁺ monocytes (45).

After 3 d of nonadherent culture, monocytes were infected with the R5 tropic isolate HIV-1_ADA (National Institutes of Health Repository, Germantown, MD) as previously reported (33). In brief, cells were pelleted and resuspended at a density of 2 × 10⁶ cells/ml in monocyte media. Monocytes were then infected with 20 ng/ml HIV-1_ADA. After 24 h, cells were pelleted, washed, and fresh media were added. On the third day postinfection, supernatants were collected and viral replication was assessed by HIV-p24 AlphaLISA according to manufacturer’s instructions (Advanced BioScience Laboratories, Kensington, MD).

To assay for cytokines released from astrocytes, we treated cells with full-length human rPrP^c for 24 h, and collected and centrifuged media at 3000 rpm for 5 min at 4°C to remove cellular debris, and assayed for CCL2/MCP-1, CXCL12/SDF-1, and IL-8 using DuoSet ELISA Kits according to the manufacturer’s instructions (all from R&D Systems, Minneapolis, MN). Limits of detection of these kits are 9.38 pg/ml for CCL2 and 31.25 pg/ml for both CXCL12 and IL-8.

To determine whether CCL2 and TNF-α induced PrP^c shedding, media were collected from both control and treatment plates and cell debris was removed by centrifuging at 3000 rpm 4°C for 5 min. After centrifugation, supernatants were concentrated to 250 μl using an Amicon Ultra-15, 10-kDa centrifugal filter device (Millipore, Bellerica, MA). Equal volumes of supernatants were loaded onto 4–20% polyacrylamide gels, separated by electrophoresis, and transferred to nitrocellulose membranes (GE Healthcare Life Sciences). Membranes were blocked for 2 h at room temperature with 5% nonfat dry milk and 3% BSA in TBST. Blots were probed with Ab against the extracellular portion of PrP^c (SAF 32 clone, 1:200 dilution; Cayman Chemical, Ann Arbor, MI) overnight at 4°C, washed with TBST, and probed with anti-mouse IgG-HRP secondary Ab (1:2000 dilution; Cell Signaling, Boston, MA) for 1 h at room temperature. Signal was detected using Western Lightning Plus-ECL (Perkin Elmer, Waltham, MA). To assay for active ADAM10, we treated astrocytes with CCL2 or TNF-α, and cells were lysed with cell lysis buffer (Cell Signaling) supplemented with protease inhibitor mixture (Roche, Indianapolis, MN). After removing debris by centrifugation, we determined total cellular protein concentration by Bio-Rad protein assay (Bio-Rad, Hercules, CA). Cell lysates were loaded onto 4–20% polyacrylamide gels, separated by electrophoresis, and transferred to nitrocellulose membranes. Membranes were blocked for 2 h at room temperature with 5% nonfat dry milk and 3% BSA in TBST, and blots were probed with Ab against ADAM10 (1:200 dilution; Santa Cruz, Dallas, TX) overnight at 4°C, washed with TBST, and probed with anti-rabbit IgG-HRP secondary Ab (1:2000 dilution; Cell signaling) for 1 h at room temperature. The ADAM10 Ab detects both the pro and active forms of ADAM10. Signal was detected using Western Lightning Plus-ECL. Blots were stripped using Restore Plus Western blot Stripping Buffer (Thermo Scientific), and reprobed with Ab against GAPDH (1:500 dilution; Cell Signaling), for 1 h at room temperature, washed with TBST, and probed with anti-mouse IgG-HRP secondary Ab (1:2000 dilution; Cell Signaling) for 1 h at room temperature. Data were quantified by densitometry using UN-SCAN-IT software (Silk Scientific).

Chemotaxis assays were performed with a Neuro Probe 48-well Micro Chemotaxis Chamber (Neuro Probe, Gaithersburg, MD) as previously described (46). Astrocytes were treated with rPrP^c (10 μg/ml) or its diluent NaAc. Supernatants were collected, spun, and placed in wells in the bottom chamber of the apparatus. A polycarbonate filter containing 5-μm pores was placed between the bottom and top chambers. Uninfected or HIV-infected mature monocytes (2.5 × 10⁵) in RPMI 1640 with 2% FBS were added to the wells in the top chamber and allowed to migrate for 1 h at 37°C. The membrane was then removed and the cells that had migrated through the membrane and bound to its underside were fixed and stained with the Diff-Quik Stain Set (Siemens, Munich, Germany). Chemotaxis was quantified by densitometry using UN-SCAN-IT software.

For glutamate uptake studies, astrocytes were cultured at 3 × 10⁴ cells per well on 24-well plates until they were confluent (2–3 d). Media for cells were changed to glutamine-free DMEM (Thermo Fisher Scientific) the day before the experiment. Astrocytes were treated with rPrP^c for 10 min after which [³H]l-glutamate (0.5 μCi/ml) (Moravek Biochemicals, Brea, CA) and 50 μM unlabeled glutamate (Sigma-Aldrich, Allentown, PA) were added to cells with rPrP^c still present. Cells were incubated for an additional 1 h. Glutamate uptake was stopped by addition of ice-cold solution of Kreps-HEPES buffer (140 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl2, 15 mM HEPES, and 1.2 mM MgSO4). After three consecutive washes, cells were lysed with NCS tissue solubilizing agent (MP Biomedicals, Boston, MA), and lysate was collected for scintillation counting. cpm were determined for 3 min.

Astrocytes are the most abundant cell type in the CNS with many specialized functions, and they play a prominent role in HIV CNS disease (10, 35, 47). They express and shed PrP^c at baseline levels in vitro at a concentration of ∼128 ng/ml per 10⁵ cells (33, 48). Our laboratory HIV-positive individuals with cognitive impairment have increased sPrP^c in their cerebrospinal fluid that correlates with cerebrospinal fluid CCL2, suggesting a dysregulation in the shedding of this protein by CCL2 (33). TNF-α is released by activated microglia and astrocytes during HIV infection of the CNS and has been shown to be increased in the cerebrospinal fluid of HIV-infected people with dementia (49). To determine the effect of CCL2 and TNF-α on PrP^c shedding, we treated astrocytes with CCL2 (200 ng/ml) and TNF-α (10 ng/ml), and supernatants were collected and analyzed for sPrP^c by Western blot. After treatments, cell viability of cultures was evaluated by trypan blue staining, and cultures were found to be free of dead or dying cells. CCL2 and TNF-α increased PrP^c shedding (37 kDa) after 24 h of treatment (Fig. 1). These experiments were performed four independent times with similar findings in which CCL2 increased PrP^c shedding 1.7-fold above baseline and TNF-α increased PrP^c shedding 2.0-fold above baseline (quantification shown in Figs. 2 and 3). We also determined whether concomitant treatment with CCL2 and TNF-α resulted in a further increase in sPrP^c. In these experiments we found that CCL2 increased PrP^c shedding by 2.06-fold above baseline and TNF-α increased shedding by 1.933-fold. Concomitant treatment of cells with CCL2 and TNF-α resulted in a 2.27-fold increase, n = 3 (data not shown). These results were not statistically different from data obtained with individual cytokine treatments and show that the effect of CCL2 and TNF-α is neither synergic nor additive. These results suggest that astrocytes are a major source of increased sPrP^c in the cerebrospinal fluid of HIV-positive individuals, and that CCL2 and TNF-α may play an important role in CNS PrP^c shedding.

In neurons, it was shown that the cell-associated active disintegrin metalloprotease (ADAM10) cleaves PrP^c from the cell membrane, but PrP^c shedding from human astrocytes had not been studied extensively (50). To examine a mechanism by which CCL2 and TNF-α increase PrP^c shedding, we studied whether these cytokines activate ADAM10 in astrocytes. Because both the active and inactive forms of ADAM10 are cell-associated protease, astrocytes were grown to confluency and treated with CCL2 or TNF-α, and cell lysate was prepared and analyzed by Western blotting. Results demonstrate that astrocytes express both the pro and active forms of ADAM10, and that CCL2 and TNF-α increase the active form. Densitometry was performed in which active ADAM10 was compared with a loading control, GAPDH, and these values were compared between control and CCL2-treated cells and control and TNF-α–treated cells. Quantification was performed at the time point with the maximal change. Results are reported as fold change relative to control. In four independent experiments, CCL2 increased active ADAM10, at either 2 or 4 h by 1.5-fold (Fig. 2B), and TNF-α increased ADAM10 at later time points, 6, 8, or 12 h by 1.5-fold (Fig. 2C).

We found that CCL2 and TNF-α increased PrP^c shedding and also induced increased expression of the active form of ADAM10, suggesting a mechanism for the increased PrP^c shedding caused by these factors. Therefore, we used a specific ADAM10 inhibitor (GI254023X), a small-molecule inhibitor that has been shown to inhibit ADAM10 100-fold more effectively than it affects ADAM17, to examine whether it will block this cytokine-mediated increased shedding (44). Astrocytes were treated with GI254023X concomitantly with CCL2 and TNF-α. Western blot analyses of supernatants from treated cells demonstrate that astrocytes shed PrP^c at baseline levels, and that CCL2 and TNF-α increase this shedding (Fig. 3A). Densitometric quantification of four independent experiments shows that CCL2 increased PrP^c shedding at 24 h by 1.7-fold, and the ADAM10 inhibitor brought this down to below baseline (Fig. 3B). Similarly, TNF-α increased PrP^c shedding at 24 h by 2-fold, and this increase was again brought below baseline by the inhibitor (Fig. 3C). These results demonstrate that ADAM10 cleaves PrP^c in astrocytes, and that the increased shedding mediated by CCL2 and TNF-α is by active ADAM10.

To examine the effect of sPrP^c on cytokine production, and therefore its contribution to CNS damage, we treated human primary astrocytes with rPrP^c (10 μg/ml). This concentration is within the range for 24 h of what is found in the cerebrospinal fluid of HIV-seropositive individuals with cognitive impairment (33). In a different study, this concentration of PrP^c was also found to induce the production of inflammatory factors and activate NF-κB and ERK signaling pathways in human macrophages (34). Soluble inflammatory mediators, CCL2, CXCL12, and IL-8, were quantified by ELISA. CCL2 and CXCL12 are chemokines that are important in recruitment of monocytes across the BBB, which is the first step in viral seeding of the CNS and subsequent establishment of HANDs. They are also increased in the cerebrospinal fluid of HIV-infected individuals even on cART (51). In addition, IL-8 increases monocyte adhesion to endothelial cells, contributing to the migration of monocytes across the BBB (52). At baseline conditions, astrocytes cultured in 96-well plates secreted 21,876 pg/ml CCL2, which increased to 42,653 pg/ml (1.9-fold) in response to rPrP^c treatment. CXCL12 secretion increased from 343 to 549 pg/ml (1.6-fold) and IL-8 secretion increased from 39 to 60 pg/ml (1.5-fold) in response to rPrP^c (Fig. 4). This suggests that sPrP^c alters astrocyte function and may play an important role in HIV-1 neuropathogenesis by increasing inflammatory factors within the CNS.

Astrocytes release cytokines in response to rPrP^c treatment. To test the effect of these inflammatory factors on monocyte recruitment, we treated astrocytes with 10 μg/ml rPrP^c or its diluent NaAc for 24 h. Supernatants from both control and rPrP^c-treated astrocytes were collected and applied to the lower part of a chemotaxis chamber. A total of 250,000 uninfected or HIV-1_ADA–infected primary human monocytes from our culture system (see 2Materials and Methods) in RPMI 1640 supplemented with 2% FBS were added to the top wells and allowed to chemotax for 1 h at 37°C across a polycarbonate filter with 5-μm pores that separated the upper and lower chambers. The membrane was removed, and cells that had migrated through the membrane and bound to its underside were fixed and stained with the Diff-Quik Stain Set. Chemotaxis was quantified by densitometry. Both uninfected and HIV-infected monocytes exhibited baseline chemotaxis to supernatants collected from diluent-treated astrocyte because of the low level production of chemokines by astrocytes under baseline conditions. Supernatants from astrocytes treated with rPrP^c mediated significant increases in chemotaxis of both uninfected and HIV-infected monocytes above baseline by 1.3- and 1.4-fold, respectively (Fig. 5). To determine whether rPrP^c is chemotactic by itself, media supplemented with 10 μg/ml rPrP^c was added to the lower part of the chemotaxis chamber and monocytes to the top chamber and allowed to chemotax for 1 h. Media supplemented with rPrP^c did not cause chemotaxis of either uninfected or infected monocytes by itself (results not shown), indicating that supernatants from rPrP^c-treated astrocytes contain factors produced in response to the rPrP^c treatment that mediate monocyte chemotaxis. These results indicate that sPrP^c may contribute to monocyte recruitment, viral seeding of the CNS, and inflammation in HIV neuropathogenesis.

Astrocytes are essential for maintaining neuronal health by regulating extracellular concentrations of ions and metabolites, including glutamate (53). Insufficient glutamate uptake by astrocytes leads to neurotoxicity and death (54). In mouse astrocytes, cell surface PrP^c was shown to facilitate glutamate uptake by increasing the binding of glutamate to its transporters (32). In different studies, surface PrP^c was shown to bind to itself, forming homotypic interactions (29, 55). Therefore, we hypothesized that increased sPrP^c will bind to cell-associated PrP^c on astrocytes, inhibiting the ability of surface PrP^c to facilitate glutamate uptake. To test this hypothesis, we grew human astrocytes in 24-well plates (30,000 cells per well) and placed them in serum-free and glutamine-free media 24 h before the study. Human astrocytes were pretreated with rPrP^c (10 μg/ml) for 10 min; then 0.5 μCi/ml [³H]l-glutamate and 50 μM unlabeled glutamate were added for 1 h. [³H]glutamate uptake was quantified and compared between control and rPrP^c-treated groups. We found that glutamate uptake was significantly decreased in astrocytes treated with rPrP^c by 0.4-fold (Fig. 6). These results indicate that increased sPrP^c in the CNS may lead to neurotoxicity and neuronal loss by impairing glutamate uptake by astrocytes.

PrP^c is a GPI-anchored protein found in lipid rafts of the cell membrane and is expressed on cells of the CNS. Our laboratory previously showed that sPrP^c is highly increased in the cerebrospinal fluid of neurocognitively impaired HIV-1–infected individuals, but not in HIV-infected individuals with normal cognitive function. In this study, we show that the cytokines CCL2 and TNF-α increase the shedding of PrP^c from astrocytes by inducing the activation of the metalloproteinase ADAM10. Although ADAM10 had previously been reported to be the sheddase for PrP^c for some cell types including murine neuronal cells, immortalized neuronal cells, and transfected HEK cells, to our knowledge, this is the first study to show that ADAM10 is the protease that cleaves PrP^c from human astrocytes. We showed that CCL2 and TNF-α increase expression of active ADAM10. In HEK293 cells, proprotein convertases (PC)7 and furin were shown to cleave the ADAM10 propeptide, resulting in increased expression of active ADAM10 (56).

We hypothesized that CCL2 and TNF-α increase active ADAM10 by activating PC7 and furin. Therefore, we examined whether these cytokines activate ADAM10 through a PC7- and furin-mediated mechanism. These convertases have been shown to be regulated through Notch1 signaling (57). The Notch protein has extracellular, transmembrane (TM), and intracellular domains (58). During activation, the Notch protein extracellular domain is cleaved from the TM domain and Notch intracellular domain (NICD) (58). After that cleavage, the TM domain is separated from NICD by γ-secretase, allowing the NICD to translocate to the nucleus, where it activates various transcription factors including Hey1 and Hes1 (59–62). TNF-α has been shown to activate Notch signaling and increase the gene expression of transcription factors Hey and Hes1 in fibroblasts, nucleus pulposus cells, and melanoma cell lines (57, 62–64). We studied whether CCL2 and TNF-α activated ADAM0 by increasing activation of PC7 and furin through Notch1-mediated signaling by measuring the gene expression of Hes1 and Hey1. We found that these cytokines decreased both Hes1 and Hey1 at various time points (data not shown). This suggests that TNF-α and CCL2 do not activate ADAM10 through PC7 and furin. A possible mechanism by which CCL2 and TNF-α are regulating ADAM10 activation may be through other PCs. Seven PCs have been identified: furin, PC1 (also known as PC1/3), PC2, PACE4, PC4, PC5/PC6, and PC7. Studies have shown that, in addition to furin, proprotein convertases, including PC1, PC2, PACE-4, and PC5/PC6, result in cleavage and activation of ADAM17, which is a close relative of ADAM10 (65). Therefore, further studies will address whether CCL2 and TNF-α result in activation of ADAM10 by affecting these PCs.

Another mechanism by which ADAM10 can be processed is through ADAM9. Studies have shown that ADAM9 is involved in the ectodomain shedding of ADAM10. After it is released from the ectodomain, some studies demonstrate that the activity of ADAM10 is diminished, whereas others show that it is actually increased, possibly because it is not restricted to the cell membrane and to cell membrane proteins surrounding it (66, 67). However, because the type of activation that is induced in astrocytes by CCL2 and TNF-α is maturation of ADAM10, it is unlikely that ADAM9 is involved in this process.

Although sPrP^c is found in human cerebrospinal fluid and serum, the specific functions of this protein have not been addressed extensively in the context of HIV CNS pathogenesis (33, 68). In this study, we examined the effect of sPrP^c on astrocyte cytokine production and glutamate uptake, functions dysregulated during HIV infection of the CNS. We found sPrP^c is an important mediator of the astrocyte inflammatory response by increasing CCL2, CXC12, and IL-8. Our studies suggest that PrP^c shedding in the CNS is self-amplifying, in that rPrP^c treatment of astrocytes increases the production of CCL2, and CCL2 increases the shedding of PrP^c from astrocytes. This is consistent with our previous finding that increased cerebrospinal fluid CCL2 and sPrP^c levels are correlative in HIV⁺ individuals with cognitive impairment (33). Both cell-associated PrP^c and sPrP^c are neurotrophins and have been associated with neuroprotection from toxic insult and maintenance of synaptic plasticity (69). Therefore, increased PrP^c shedding may be an initial protective response to neuronal injury in HIV-seropositive people with neurocognitive impairment. However, although increased sPrP^c might be initially protective, it is likely that it will contribute to CNS injury by inducing production of cytokines that mediate monocyte recruitment and viral seeding of the CNS, and by causing decreased glutamate uptake.

Some mechanisms of the effects of rPrP^c on monocyte/macrophages have been examined previously. In one study, rPrP^c bound to the surface of monocytes/macrophages and activated ERK and NF-κB pathways, resulting in TNF-α, IL-1β, and IL-6 production (34). In that study monocytes were treated with rPrP^c that had the Fc portion of the human IgG1 as a fusion protein, and it was shown by flow cytometry that rPrP^c bound to the surface of monocytes. In another study, recombinant murine PrP^c treatment of monocyte/macrophages was shown to activate the Src-like kinases, as well as Syk and Pyk2 (70). Although we and other groups have shown that rPrP^c induces the release of cytokines from cells, a possible binding partner for sPrP^c has not been identified. In neurons, PrP^c has been shown to bind to neural cell adhesion molecule and to the laminin receptor precursor protein (70, 71). Another possible interaction, and what we hypothesize is taking place in our study, is a homotypic interaction between soluble PrP^c and membrane PrP^c.

Astrocytes are essential in maintaining CNS homeostasis and regulating extracellular metabolite concentration including glutamate (72). Accumulation of high concentrations of glutamate results in neurotoxicity and neuronal degeneration. Because astrocytes are the principle means of clearing glutamate from the neuronal environment, factors that might inhibit or facilitate this glutamate uptake need to be characterized. Astrocytes from PrP^c knockout mice were shown to exhibit decreased glutamate uptake (32). Kinetic experiments from this study indicated that the difference in glutamate uptake in PrP^c-deficient mice is the result of glutamate affinity to its transporter, and that cell-associated PrP^c facilitates uptake of this neurotransmitter (32). Furthermore, treatment of wild type mice with the cellular prion protein–derived peptide PrP106-126 inhibited glutamate uptake. This effect was specific for wild type astrocytes, because astrocytes from PrP^c knockout mice were not affected by this treatment, suggesting that PrP106-126 interacts with cell-associated PrP^c and blocks its function. However, the use of PrP106-125 peptide as a model of cellular prion protein is controversial because this peptide forms toxic aggregates displaying properties of prion protein scrapie (73, 74). In our study, we found that pretreatment of astrocytes with full-length human rPrP^c caused decreased glutamate uptake. We propose that increased sPrP^c binds to cell-associated PrP^c on astrocytes, limiting the cell-associated protein’s ability to facilitate glutamate uptake. Binding studies to address this hypothesis will be performed in future experiments.

PrP^c in other CNS pathologies including Alzheimer disease (AD) has been studied (75–78). CNS PrP^c is increased in people with AD and regionally localized with amyloid-β (Aβ) plaques (79, 80). Cell-associated PrP^c is hypothesized to play a role in the progression of AD by acting as a receptor of Aβ–oligomers through which oligomers mediate their neurotoxic effect (76). Alternatively, the GPI anchor free rPrP^c was shown to prevent fibrillization and toxic oligomer formation, potentially by binding and sequestering Aβ oligomers and preventing them from plaque formation (81). Similar to our study in which cell-associated PrP^c and sPrP^c have different roles in HIV CNS pathogenesis, these findings suggest that cell surface anchored PrP^c may be injurious and sPrP^c may be protective during AD neuropathogenesis. In contrast with these studies, our findings suggest that sPrP^c plays an inflammatory and neurotoxic role during HIV CNS infection. Levels of sPrP^c in the cerebrospinal fluid have been examined in other CNS diseases including Lewy body dementia, Parkinson’s disease, multiple sclerosis, and cerebral ischemia (82). However, the role of sPrP^c in the development or progression of these diseases has not yet been examined.

Our data suggest that neuroinflammation during HIV CNS infection results in increased shedding of PrP^c from astrocytes by inducing active ADAM10. This increased sPrP^c, in addition to contributing to increased monocyte recruitment into the CNS by increasing the production of cytokines from astrocytes, may lead to neurotoxicity and neuronal loss by inhibiting glutamate uptake in astrocytes. Therefore, targeting PrP^c shedding may be a therapeutic approach to limit HIV neuropathogenesis.

In addition to cleaving PrP^c and contributing to neuroinflammation during HIV infection of the CNS, ADAM10 has been reported to be involved in dysregulated shedding that is associated with cardiovascular diseases, neurodegeneration, and cancer (83–85). Therefore, ADAM10 has increasingly become a target for therapy. Until recently, clinical trials inhibiting ADAM10 used broad-spectrum metalloprotease inhibitors and have failed because of toxicity resulting from nonselectivity and low bioavailability (86). However, the next generation of long-acting and highly selective inhibitors are now being developed (87). These might serve as appropriate therapy to prevent the shedding of PrP^c and reduce neuroinflammation caused by HIV infection of the CNS.

We thank Dr. Boris Schmidt at Technische Universität Darmstadt, Germany, for providing the ADAM10 inhibitor, GI254023X. We also thank Dr. Tina Calderon, Dr. Loreto-Torres Carvallo, Dr. Peter J. Gaskill, Dr. Dionna Williams, Matias Jaureguiberry, Lillie Lopez, Mike Veenstra, and Courtney Veilleux for valuable contributions to this project.

The authors have no financial conflicts of interest.