Neuroinflammation-Induced Interactions between Protease-Activated Receptor 1 and Proprotein Convertases in HIV-Associated Neurocognitive Disorder

Quantitative reverse transcription-PCR (RT-PCR) analysis.

Using TRIzol (Invitrogen) and an RNeasy purification kit (Qiagen), total RNA was prepared from frontal cortical samples from HIV-uninfected patients (group A; n = 10), HIV-infected patients without neurocognitive impairment (group B; n = 10), HIV-infected patients with neurocognitive impairment (group C; n = 10), and HIV-infected patients with neurocognitive impairment and encephalitis (group D; n = 10). HeLa cells cultured in serum-free medium were treated with a cytokine cocktail composed of tumor necrosis factor alpha (TNF-α; 50 ng/ml), gamma interferon (IFN-γ; 50 ng/ml), and interleukin-1β (IL-1β; 10 ng/ml) for 24 h before they were collected for RNA extraction in TRIzol. The cDNAs were synthesized using Superscript II reverse transcriptase (Invitrogen). Semiquantitative real-time PCR was performed using SYBR green (IQ SYBR Supermix; Bio-Rad) detection and the ΔC_T (where C_T is threshold cycle) method (31). Threshold cycle values for the gene of interest were normalized to those of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (for brain samples) and the TATA binding protein (for cell lysates) and are represented as the mean relative fold change compared to control sample values. Excel software was used for calculating standard errors of the means (SEM) and Student's t test probabilities.

Immunohistochemistry.

Human brain sections (8 μm) were deparaffinized and hydrated using decreasing concentrations of ethanol. Sections were boiled in 0.01 M citrate buffer, pH 6.0, for 10 min for PAR1 and glial fibrillary acidic protein (GFAP) immunostaining. Endogenous peroxidases were blocked by incubating sections in 0.3% hydrogen peroxide for 20 min. To prevent nonspecific binding, sections were preincubated with 10% normal goat serum–0.5% Triton X-100 for 1 h at room temperature. Rabbit anti-human PAR1 antiserum was used in accordance with previously described work (22). Mouse anti-GFAP antibodies were purchased from Dako (Copenhagen, Denmark). Primary antibodies were diluted in phosphate-buffered saline (PBS)–serum (PAR1, 1:500; GFAP, 1:1,000) and incubated overnight at room temperature, followed by washing. All washes were conducted for 15 min with 0.01 M PBS, pH 7.4, and antibodies were diluted in PBS containing 10% normal goat serum. Immunolabeling with primary antibodies was detected with biotinylated goat anti-rabbit or biotinylated goat anti-mouse antibodies (Vector Laboratories) and then with avidin-biotin-peroxidase complexes (Vector Laboratories) for 1 h at room temperature for each step. Immunoreactivity was detected using 3,3′-diaminobenzidine tetrachloride (brown) and/or 5-bromo-4-chloroindolylphosphate (blue) (32). All human brain specimens were collected with consent (19).

In situ hybridization.

Ten-micrometer-thick cryosections were prepared from brains of 3-month-old mice, fixed in 4% formaldehyde, and hybridized as previously described (33) with mouse sense (negative control) and antisense cRNA probes. The latter probes corresponded to the mouse PAR1, furin, PC5, PACE4, or PC7 coding region for aa 1 to 420, 1 to 793, 80 to 348, 1 to 214, 1 to 213, respectively, and were synthesized using ³⁵S-UTP (PerkinElmer).

In vitro assays.

Enzymatic in vitro activities of the purified soluble furin, PC5/6, PACE4, and PC7 (a gift of Robert Day, Université de Sherbrooke) were measured at 37°C in 100 μl of buffer (25 mM Tris-HCl, 1 mM CaCl₂, pH 7) in the presence of a 100 μM concentration of the fluorogenic substrate pyroglutamic acid-RTKR-7-amido-4-methylcoumarin (Pyr-RTKR-MCA; Peptide International). The release of free 7-amino-4-methyl-coumarin (AMC) was detected with a Spectra Max Gemini EM microplate spectrofluorimeter (Molecular Devices) (excitation, 370 nm; emission, 460 nm; emission cutoff, 435 nm). One unit of enzymatic activity was defined as 1 pmol of AMC released from 100 μM Pyr-RTKR-MCA/min at 37°C. The 19-mer synthetic peptide mimicking the tethered ligand and putative PC cleavage site (in boldface) of human PAR1, NATLDPR₄₁SFLLR₄₆↓NPNDKYE (hPAR1_35–53), was obtained from GenScript. To assess the in vitro cleavage of hPAR1_35–53 by PCs, the synthetic peptide was incubated at 37°C in vitro (100 μM peptide in 100 μl of buffer consisting of 25 mM Tris-morpholineethanesulfonic acid [MES], 1 mM CaCl₂, pH 7) with 90 nM active purified PCs (18 h) or with 2 μg/ml trypsin (5 min and 1 h). Trypsin cleaves at Arg₄₁↓ (preferentially) and at Arg₄₆↓ and was used as a reference for the reverse-phase high-pressure liquid chromatography (RP-HPLC) profile of the resulting peptide fragments; a 5-min digestion results in cleavage at Arg₄₁↓ and a 1-h digestion leads to cleavage at both Arg₄₁↓ and Arg₄₆↓. The digestion products were separated by RP-HPLC on a Varian C₁₈ column (particle size, 5 μm; pore size, 100 Å; 4.6 by 250 mm) using a 5% to 50% acetonitrile (plus 0.1% trifluoroacetic acid [TFA]) gradient over 45 min at a flow rate of 1 ml/min. Peptide bonds were detected at 214 nm, and Tyr-containing peptides were detected at 280 nm. By comparing the RP-HPLC profiles of trypsin- and PC-digested hPAR1_35–53, it was determined that PACE4 cleaves the 19-mer at the predicted site R₄₁SFLLR₄₆↓. The percent cleavage was calculated as the ratio of the normalized peak area (peak area/number of peptide bonds) of the C-terminal fragment NPNDKYE to that of the intact 19-mer peptide (at time zero).

Plasmids and reagents.

The cDNA encoding human PAR1 (hPAR1; pcDNA3.1 vector containing mCherry at the N terminus and enhanced yellow fluorescent protein [eYFP] at the C terminus) was used as reported previously (34). Site-directed mutagenesis was used to generate the human PAR1 mutants R46A and R41A and the double mutant R41A R46A. All PC constructions (human furin, mouse PC5A, mouse PC5B, human PACE4, and human PC7) were cloned in pcDNA3 or pIRES2-EGFP (where IRES is internal ribosome entry site and EGFP is enhanced green fluorescent protein) vectors, as previously described (35). Plasmid expressors encoding hemagglutinin (HA)-tagged Vpr as well as HxBc2 Env were previously described (10, 36). The plasmids coding for mouse growth differentiating factor 11 (GDF11) (37) and human transferrin receptor 1 (TfR1) (5) were used as reported previously.

Cell culture and cDNA transfections.

HEK293 (human embryonic kidney-derived epithelial) and COS-1 (monkey kidney fibroblast) cells were grown in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS) (Invitrogen) and were maintained at 37°C under 5% CO₂. Using JetPrime (PolyPlus) for HEK293 and Lipofectamine 2000 (Invitrogen) for COS-1 (unless otherwise indicated), 80 to 90% confluent cells were transiently transfected with pcDNA3 recombinants of dually tagged human PAR1 with full-length human furin, mouse PC5A or PC5B, human PACE4, human PC7, or an empty pcDNA3 plasmid. Twenty-four hours after transfection, the cells were washed and incubated in serum-free medium for an additional 20 h before medium collection and cell lysis.

Immunofluorescence.

HEK293 cells were plated on polylysine-coated coverslips and transfected. Twenty-four hours posttransfection, fresh serum-free medium with 1 unit/ml of thrombin (Calbiochem) and/or 5 μM E64 (a cysteine proteinase inhibitor; Calbiochem) was added to the cells and incubated overnight. Our previous work (34) indicated that many cells, including HEK293 cells, secrete PAR-regulating proteinases that can cleave the N terminus of PARs, including PAR1. Thus, E64 was found to preserve the N terminus so that it can in principle be targeted by other noncysteine proteinase enzymes. In experiments using PC-specific inhibitors, 25 μM cell-permeable decanoyl-RVKR-chloromethylketone (cmk) (RVKR-cmk; Bachem Bioscience) and 10 μM nonpermeable hexa-d-arginine (D6R; EMD Chemicals) were added at this point with fresh serum-free medium (38). After 20 h, the cells were washed three times with PBS and then fixed in 3.7% formaldehyde with 0.01 mM sucrose for 15 min at room temperature. Following a final three washes with PBS, the coverslips were mounted with ProLong Gold antifade reagent with 4′,6′-diamidino-2-phenylindole (DAPI; Invitrogen). Immunofluorescence analyses were performed with a Zeiss LSM-710 confocal microscope. When antibodies were used, the cells were washed three times with PBS and permeabilized with 0.25% Triton X-100 (or with PBS for nonpermeabilized cells) for 7 min following fixation. Blocking was performed subsequent to an additional three washes with PBS using 1% bovine serum albumin for 30 min. The cells were incubated overnight at 4°C with anti-golgin-97 monoclonal antibody (MAb) (Santa Cruz), V5 MAb (Invitrogen) for human furin, and monoclonal HA (in-house) primary antibodies for human PACS1 at 1:250 dilutions in 1% bovine serum albumin. Alexa Fluor 647-conjugated secondary antibodies (Invitrogen), diluted 1:500 in 1% bovine serum albumin, were incubated for 1 h at room temperature after three washes with PBS.

Western blot analyses.

Proteins were extracted from cells in 50 mM Tris-HCl, pH 8, 150 mM NaCl, 0.1% SDS, 1% Nonidet P40, and 0.25% Na deoxycholate (1× radioimmunoprecipitation assay [RIPA] buffer) in the presence of a cocktail of protease inhibitors (Roche). Protein concentrations were estimated by a Bradford assay. After SDS-PAGE and blotting onto polyvinylidene difluoride (PVDF) membranes, the latter were incubated with Chessie 8 anti-gp41 (1:10) (39), monoclonal HA (1:1,000; American Type Culture Collection), monoclonal V5 (1:5,000; Invitrogen), monoclonal Flag M2 (1:5,000; Stratagene), and rabbit polyclonal anti-β-actin (1:1,000; Sigma-Aldrich) antibodies and finally with appropriate secondary antibodies. Immunoreactive species were detected using enzymatic chemiluminescence (Amersham Bioscience) and quantified using ImageJ, version 1.41, software (NIH).

Calcium mobilization assay.

HEK293 cells from a T75 flask were detached in enzyme-free, EDTA-containing, cell dissociation buffer and resuspended in serum-containing growth medium, and approximately 100,000 cells/well were seeded into cell culture-treated clear-bottom 24-well CellBIND plates (Corning) coated with polylysine. The next day, cells were transfected with 0.5 μg of the PACE4 and PC5A cDNAs per well with a FuGene transfection reagent according to the manufacturer's protocol. At 24 h posttransfection the medium was replaced with serum-containing growth medium. The next day the culture medium was replaced with isotonic Hanks buffered salt solution (HBSS) and fluo-4-acetoxymethyl ester (AM) no-wash calcium indicator dye (Invitrogen), followed by incubation at 37°C for 30 min. Intracellular fluorescence (excitation at 494 nm and emission at 516 nm) was monitored for 2 min after the addition of 20 nM thrombin, using n Envision 2104 Multilabel Reader (PerkinElmer Life Sciences). The relative fluorescence measured per second was plotted, and the average values from each experiment for each condition were analyzed utilizing a nonlinear regression equation (GraphPad Prism). Peak height was calculated from the highest peak point to baseline of the average nonlinear regression lines.

PLA.

HEK293 cells seeded into 24-well plates were transfected with cDNAs with a FuGene transfection reagent according to the manufacturer's protocol and 24 h later reseeded into a μ-Slide eight-well ibiTreat microscopy chamber (Research Products/Ibidi). Cells were incubated with rabbit anti-PAR1 (Santa Cruz) and mouse anti-V5 (Invitrogen) primary antibodies overnight at 4°C. Duolink in situ (Olink Bioscience) proximity ligation assay (PLA) probes, anti-rabbit antibody minus sense and anti-mouse antibody plus sense, with a far-red detection kit were utilized to detect the interaction of mCherry-hPAR1-eYFP with V5-tagged furin or PC7 according to the Duolink in situ protocol. The number of dots (indicating interactions of PLA probes) per cell were manually counted utilizing ImageJ software (NIH).

PAR1 inhibition of gp160 processing and viral infectivity.

For protein expression assays, 1 × 10⁶ HEK293T cells were seeded in 100-mm plates and transfected using the calcium phosphate procedure with 5 μg of HIV-1 NL4-3 proviral DNA (NIH AIDS Reagent Program, NIH) (40) and either 1 μg of control vector EGFP-C1 (Clontech) or combinations of control vector and wild-type (WT) mCherry-hPAR1-eYFP or mutant mCherry-hPAR1 R46A-eYFP constructs. Cells were harvested after 2 days of culture and lysed in buffer (25 mM Tris, pH 7.4, 1% Triton X-100, 150 mM NaCl) containing protease inhibitors (Roche). Cell supernatants were briefly centrifuged and filtered on 0.45-μm-pore-size membranes, and virus pellets were obtained following ultracentrifugation on 20% (wt/vol) sucrose cushions for 2 h at 130,000 × g in a Beckman ultracentrifuge. Virus pellets were then resuspended in 100 μl of PBS and aliquoted. A total of 20 μg of protein in cell lysates was added to 5× Laemmli buffer, and the proteins were separated by 12.5% SDS-PAGE prior to transfer on nitrocellulose membranes (Bio-Rad). Membranes were blocked with 5% milk in Tris-buffered saline (TBS)–0.1% Tween 20, washed in TBS, and incubated with the appropriate antibodies overnight at 4°C: mouse anti-gp160 (Chessie 13-39.1, 1:10) (39), mouse anti-gp41 (Chessie 8, 1:10) (39), and mouse anti-gp120 (IIIB 902, 1:10) (41, 42) (all obtained from the NIH AIDS Reagent Program); mouse anti-p24 (31-90-25, 1:3,000; ME Biotech Services); rabbit anti-GFP (1:500; Life Technologies); and rabbit anti-GAPDH (1:1,000; Biolegend). Following three washes in TBS–0.1% Tween 20, blots were incubated with anti-mouse or anti-rabbit horseradish peroxidase (HRP)-labeled antibodies (1:5,000), washed, and processed for enhanced chemiluminescence assay (Pierce). Films were scanned, and protein signals were quantified using ImageJ (NIH) software. For infection assays, the TZM-bl reporter cell line (obtained from the NIH AIDS Reagent Program) (43, 44) was used. These HeLa-based modified cells harbor a copy of the firefly luciferase gene under the control of the HIV-2 long terminal repeat (LTR) and express CD4 and the HIV coreceptors CXCR4 and CCR5 (43). Following HIV-1 p24 quantitation by enzyme-linked immunosorbent assay (ELISA; XpressBio Life Science), virus obtained from the previously described HEK293T cells cotransfected with 0.1 μg of WT or mutant R46A PAR1 was used in the assay. Briefly, either 1 ng, 0.2 ng, or 0.04 ng of virus was added to 30,000 TZM-bl cells/well in 24-well plates. Following 2 days of culture, cells were lysed, and luciferase activity was measured using a Promega luciferase assay kit (Promega) on a GloMax luminometer. Infections using the most diluted virus (0.04 ng) were nonsaturating and were used to determine viral infectivity.

Ethics approvals.

The decedents used in this work were from the National NeuroAIDS Tissue Consortium (NNTC) (19). The use of autopsied brain tissues is part of ongoing research (Pro00002291) approved by the University of Alberta Human Research Ethics Board (Biomedical). Written informed consent documents were signed for all samples collected. The protocols for obtaining postmortem brain samples comply with all federal and institutional guidelines, with special respect for the confidentiality of the donor's identity, and were approved by the ethics committee of the University of Texas Medical Branch, Galveston.

Statistical analysis.

Quantifications are defined as means ± SEM and expressed as percentages of the control value, which was considered to be 100%. The statistical significance was evaluated by Student's t test. Probability values of <0.05 were considered significant. Data were analyzed and graphics were generated with the Prism program (GraphPad Software).

Furin, PC5, PACE4, and PC7 expression levels are upregulated along with PAR1 in the brains of HAND patients with neuroinflammation.

The frontal cortex is a particularly vulnerable region of the brain in HAND (11). Thus, to test our hypothesis that PCs are elevated in brains of HIV-infected subjects, we measured PC-expression in the frontal cortex from HIV-negative and HIV-positive (HIV⁻ and HIV⁺, respectively) individuals with neurocognitive impairment and encephalitis (n = 10 for each group). Brain transcript analyses revealed that furin, PC5, PACE4, and PC7 mRNA levels were all markedly increased (3- to 8-fold) in HIV⁺ brains of patients with neurocognitive impairment and encephalitis (HAND with encephalitis, or HIVE; group D), suggesting a role for PCs in neuroinflammation caused by HIV infection (Fig. 1A). Similar increases were observed in the expression levels of a prototypic inflammatory factor, IL-1β (10-fold), as well as those of PAR1 (6-fold), an inflammation-associated receptor that has previously been reported to be upregulated in HIVE (22). This G protein-coupled receptor (GPCR) has previously been linked to neuroinflammation and neurodegeneration (20–22). In contrast, the mRNA levels of PCs, IL-1β, and PAR1 did not change in HIV⁻ brains (group A), HIV⁺ brains without neurocognitive impairment (group B), and HIV⁺ brains with neurocognitive impairment (HAND; group C) (Fig. 1A). Notably, the mRNA levels of inflammation-modulated caspase-4 and IL-33 from the same cohort were not affected (Fig. 1A), affirming that the increases observed were specific to HIV-induced neuroinflammation and not to a general inflammatory response.

Additionally, an increased density of hypertrophied astrocytes (Fig. 2B) and microglia (Fig. 2D), which are widely assumed to be indicative of neuroinflammation, was evident in HIVE brain in contrast to observations in HIV⁻ patient brain sections (Fig. 2A and C). PAR1 expression was markedly upregulated (Fig. 2F) in HIVE compared to levels in HIV⁻ brain sections (Fig. 2E). This finding was limited to astrocytes and not apparent in neurons or microglia (data not shown).

The deleterious effect of inflammatory cytokines in promoting immune response and inflammation via various mechanisms, including the generation of nitrogen intermediates, has been reported in cell lines such as pancreatic β cells, epithelial cells, and HeLa cells. Many of these studies show synergetic effects by utilizing a cocktail of cytokines instead of a single cytokine (45, 46). We adapted this method of using a cytokine cocktail that simulates inflammation in HeLa cells, which exhibited a robust response and expressed all four PCs (Fig. 1B). Thus, upon a 24-h exposure of HeLa cells to a mixture of TNF-α, IL-1β, and IFN-γ, the mRNA levels of furin, PC5, PACE4, and PC7 were upregulated by these inflammatory cytokines (2- to 4-fold) (Fig. 1B), independent of HIV infection. This was similarly observed only in HIV⁺ brains displaying encephalitis (Fig. 1A).

PAR1 and the PCs are coincidently expressed in mouse brain.

Upon analysis of the sequence of the N-terminal tethered ligand of PAR1 (NPR₄₁SFLLR₄₆↓]NP; possible cleavage sites in boldface), we hypothesized that PCs may cleave at Arg₄₆↓ instead of its activating thrombin cleavage site Arg₄₁↓, potentially leading to anti-inflammatory effects (23, 26). For this cleavage to occur, we hypothesized that PAR1 and enzymatically active PCs must be coexpressed in the same tissue and meet along the secretory pathway. To support our hypothesis, we first examined PAR1 and PC expression profiles in adult mouse brain to determine if the two proteins might be coexpressed in similar regions (Fig. 3). PAR1 is highly expressed in the olfactory lobe, hippocampus, dentate gyrus, and cerebellum. Furin and PC7 are also highly expressed in the olfactory lobe, hippocampus, cerebellum, and frontal cortex region. PC5 is rich in the cerebral cortex and dentate gyrus, while PACE4 is particularly abundant in the cerebellum. Notably, the frontal cortex and the hippocampus are major sites of motor and/or cognitive functions, which are selectively damaged in HAND (11). Although indicative of coexpression, this does not prove that the PAR1 and PC proteins are colocalized intracellularly. However, as presented later, we provide evidence for the protein cellular colocalization of PAR1 with membrane-bound PCs.

The soluble PC5A and PACE4 cleave PAR1 at Arg₄₆↓.

To examine PC cleavage of PAR1 directly, a synthetic peptide (19-mer) mimicking the N-terminal tethered ligand and putative PC cleavage site of human PAR1 was incubated with recombinant soluble furin, PC5A, or PACE4 or soluble PC7 (using equal enzymatic activities) to assess in vitro cleavage by these PCs, as we previously published (38). Digestion with trypsin, which cleaves the 19-mer at Arg₄₁↓ (preferentially) and Arg₄₆↓, was used as a reference for the resulting peptide fragments. By comparing the RP-HPLC profiles of the trypsin- and PC-digested peptide corresponding to human aa 35 to 53 of PAR1, it was determined that in vitro only PACE4 inefficiently (10% in 19 h) cleaves human PAR1 at the expected Arg₄₆↓, separating most residues of the tethered ligand (SFLLR₄₆↓N₄₇) (Fig. 4).

To evaluate whether in cells such PC cleavage would be more efficient, we used a human PAR1 construct containing mCherry at its N terminus and eYFP at its C terminus (pcDNA3/mCherry-hPAR1-eYFP) (Fig. 5). Our published work evaluating the signaling properties and receptor dynamics of this dually tagged PAR1 construct indicates that the epitope-tagged receptor is fully functional, in keeping with the properties of the wild-type receptor (34). Accordingly, we coexpressed the mCherry-hPAR1-eYFP dually tagged PAR1 with and without furin, PC5A, PC5B, PACE4, or PC7 in HEK293 cells. We monitored the results of the coexpression of the PCs with the fluorescently tagged PAR1 utilizing confocal microscopy. The N-terminal ligand of PAR1 was intact when expressed alone as both N-terminal red and C-terminal yellow fluorescence signals were detected at the cell surface (Fig. 6, top panel). The cell surface expression of the PAR1 construct was confirmed by its colocalization with the low density lipoprotein receptor (LDLR), which is expressed and functions at the plasma membrane (Fig. 7) (47). However, LDLR does not necessarily colocalize with PAR1 in the endosomal-like structures near the cell surface (Fig. 7). Upon either short-term (e.g., 3 min) (34) or long-term (overnight) exposure of cells to thrombin (1 unit/ml), PAR1 lost its N-terminal red fluorescence and only a punctate C-terminal yellow fluorescence remained at the cell surface, confirming the thrombin cleavage and release of the N-terminal mCherry tag (Fig. 6, top row, Cherry-PAR1), as previously reported (23, 34). In addition, anti-thrombin III, which is a potent thrombin inhibitor, blocked the thrombin cleavage at Arg₄₁ so that the N-terminal mCherry tag was retained (Fig. 6, middle row). Interestingly, coexpression of PAR1 with soluble PC5A or PACE4 removed the N-terminal red tag while preserving the C-terminal yellow tag, which was more evenly spread around the cell surface than that of thrombin-cleaved PAR1 (Fig. 6). While cleavage by thrombin at Arg₄₁↓ does cause PAR1 to internalize (34) and results in a punctate pattern (Fig. 6), some enzymes, e.g., neutrophil elastase, can cleave PAR1 at a noncanonical site without causing receptor internalization (34). Our imaging data thus indicate that cleavage of PAR1 by PC5A and PACE4 results in a PAR1 form that loses its N terminus but does not internalize, resulting in a uniform distribution at the cell surface (Fig. 6). The PC cleavage site was confirmed at Arg₄₆↓ since expression of these soluble PCs with a mutant PAR1 (R46A; P1 Ala) construct prevented the loss of the N-terminal mCherry fluorescent tag (Fig. 5 and 8). As the motif R₄₁XXXXR₄₆↓ in PAR1 is a general basic amino acid-specific PC motif (2), we also confirmed that the mutant R41A (P6 Ala) was not processed by either PC5A or PACE4 (data not shown). These data also demonstrated that PC5A and PACE4 cleavage of PAR1 does not occur within the mCherry domain since cleavage is prevented in the R46A mutant and since the sequence of mCherry does not contain a PC recognition motif. We determined that PAR1 cleavage occurs at the cell surface by demonstrating that it was completely blocked by two PC inhibitors (38): cell-permeable decanoyl-RVKR-cmk and nonpermeable D6R (Fig. 9). Finally, cleavage at Arg₄₆↓ occurs only with soluble PCs as the membrane-bound PC5B and PC7 did not have any apparent effect on PAR1 (Fig. 10).

PACE4 and PC5A reduce calcium mobilization in response to thrombin stimulation.

PAR1 activation by thrombin signals via G proteins and leads to Gα_q-mediated calcium release, which ultimately results in inflammation (21). Thrombin cleavage of PAR1 has been previously demonstrated to induce calcium mobilization (34). In order to evaluate the inhibitory effects of PACE4 and PC5A on thrombin-induced calcium mobilization, we transfected HEK293 cells, which endogenously express PAR1, with cDNAs coding for PACE4 or PC5A and then monitored their effect on thrombin-induced calcium mobilization (Fig. 11). The data showed that PACE4 and, less so, PC5A significantly reduced by ∼2.5-fold the amount of calcium mobilization in response to thrombin stimulation. This result agrees with the hypothesis that PACE4 and PC5A would cleave PAR1 at Arg₄₆, thereby disarming PAR1 for thrombin activation and attenuating the calcium mobilization induced by thrombin cleavage of PAR1 at Arg₄₁. Notably, since PCs require both P1 Arg₄₆ and P6 Arg₄₁ for substrate recognition, the reduced thrombin-induced calcium signaling in cells overexpressing PACE4 or PC5A means that the PCs must have cleaved PAR1 at the cell surface before thrombin, which would have otherwise removed the P6 Arg₄₁ and prevented PC cleavage. However, we cannot exclude the possibility that proteins in addition to PAR1 in HEK293 cells are also sensitive to PACE4 and PC5A so as to diminish thrombin-induced calcium mobilization via a concurrent non-PAR1 mechanism.

Furin causes PAR1 retention in the TGN.

Unlike the membrane-bound PC5B and PC7 (Fig. 10), furin relocalizes PAR1 to the TGN (Fig. 12, colocalized with the golgin-97 marker). Additionally, furin does not seem to cleave PAR1 as both red and yellow fluorescence were detected in the TGN. In contrast, overexpression of the PAR1 R46A mutant did not result in PAR1 relocalization to the TGN, indicating that this effect requires furin to bind to the N-terminal domain of PAR1 in the Arg₄₆ region (Fig. 12). Furthermore, the mutant R41A (P6 Ala) or the double mutant R41A R46A (P6 and P1 Ala) does not affect the subcellular localization of PAR1 in the presence of furin (data not shown). These data argue for the binding of PAR1 at the active site of furin, which likely recognizes the R₄₁XXXXR₄₆ sequence (2) of PAR1 but does not cleave the receptor.

We next sought to define the mechanism behind this relocalization of PAR1 to the TGN in the presence of furin and to determine if the phospho-furin acidic cluster sorting protein 1 (PACS1) regulates this process. PACS1 has been reported to bind the serine-phosphorylated acidic domain of the cytosolic tail (CT) of furin (EECPS₇₇₃DS₇₇₅EEDE; serines are underlined), thereby redirecting the localization of furin to the TGN (48). Therefore, it was possible that via PACS1 binding to the CT of furin, the furin-PAR1 complex would be redirected to the TGN. Furthermore, since it was reported that the CT of PAR1 is phosphorylated because it contains an acidic serine-rich sequence, ESSDPSSYNSSGQLMAS₄₀₆ (49), we speculated that, like furin, PAR1 could also be redirected to the TGN in the presence of excess PACS1. In support of this hypothesis, a relocalization of PAR1 to the TGN was also observed upon coexpression of PAR1 with PACS1 (Fig. 13).

PAR1 reduces cleavage activity of the membrane-bound furin and PC5B, partially inhibits PC7 activity, and colocalizes with furin and PC7.

HIV glycoprotein gp160 is cleaved by furin to produce gp120 and gp41 (Fig. 14A and B), which are critical components for viral entry and infection of host cells (28–30). In COS-1 cells, cleavage of gp160 by endogenous convertases or by overexpressed furin (38) into gp120 was markedly reduced (40 to 50%) in the presence of PAR1 (Fig. 14A). PAR1 R46A does not inhibit the gp160 processing by furin (Fig. 14B), indicating that Arg₄₆ plays a critical role in the inhibitory effect of PAR1, likely through Arg₄₆ binding to the catalytic domain of furin.

The HIV accessory protein Vpr, which has neurotoxic properties, is cleaved and inactivated by soluble PC5A and PACE4 (10) (Fig. 14C). However, and in accordance with the ability of these convertases to process PAR1 (Fig. 6), this cleavage is not inhibited by PAR1 (Fig. 14C).

It was previously shown that PC5A and PC5B can similarly process the proform of GDF11 (proGDF11) into GDF11 (37). As evidence for a differential effect of PAR1 on the two PC5 isoforms, we demonstrate that the PC5A processing of proGDF11 is not affected, while cleavage by PC5B is blocked by PAR1 (Fig. 15A). This result reveals that PAR1 is an inhibitor of PC5B but not PC5A.

In HEK293 cells the PC7-specific cleavage of human TfR1 (5) is only partially (∼25 to 30%) inhibited by PAR1 (Fig. 15B), likely due to the fact that the steady-state localization of activated PC7 is in punctate-like structures just below the cell surface (6). This subcellular location may not be readily accessible to PAR1. We conclude that PAR1 inhibits the membrane-bound furin, PC5B, and partially PC7 but is cleaved by soluble PC5A and PACE4.

In order to monitor the colocalization of the membrane-bound PCs and PAR1 on a more cellular level, we employed a Duolink in situ proximity ligation assay (PLA). The PLA detects the physical proximity of two proteins by utilizing primary antibodies from two different species, such as rabbit and mouse, against each protein of interest. The secondary PLA probes (anti-rabbit antibody plus sense and anti-mouse antibody minus sense) interact with their respective primary antibodies, and if within close proximity this interaction will lead to the production of PCR-amplified distinct fluorescent dots that can be quantitated by fluorescence microscopy. Since PC5B is expressed only in the small intestine and adrenal cortex (50) while furin and PC7 are abundant in most tissues, including brain (2), we sought to analyze the colocalization of PAR1 with furin and PC7. The data clearly show that both enzymes interact with mCherry-hPAR1-eYFP when cotransfected into HEK293 cells, in contrast to results with cells cotransfected with mCherry-hPAR1-eYFP and an empty pIRES vector (Fig. 16). Thus, based on the distribution of the number of dots/cell, we estimate that on average cells ini which PAR1 colocalizes with furin, PC7, or an empty vector show 35, 23, or <2 dots/cell, respectively. This result affirms the close intracellular proximity of PAR1 with the membrane-bound furin and PC7. This interaction very likely explains the inhibitory effect of PAR1 on the proteolytic activities of furin and PC7 and the cellular relocalization of PAR1 by furin.

PAR1 inhibits the processing of HIV-1 gp160 and significantly reduces viral infectivity.

Having demonstrated that PAR1 affects the processing of gp160 into gp120 and gp41 (Fig. 14A and B), we next set out to determine if this inhibition of gp160 processing had a direct impact on virus infectivity. Accordingly, HEK293T cells were cotransfected with the NL4-3 HIV-1 proviral plasmid (40) and different amounts of cDNAs coding for either mCherry-hPAR1-eYFP or its R46A mutant. Following 2 days of cell culture, the cell lysates were analyzed by Western blotting using a panel of antibodies against p24, gp160/gp120, gp41, and GAPDH, as well as GFP as a measure of mCherry-hPAR1-eYFP protein. As previously observed in COS-1 cells (Fig. 14A and B), gp160 processing was also inhibited by overexpression of PAR1 in HEK293T cells but not by that of its R46A mutant (Fig. 17A and B).

In order to further quantify the impact of PAR1 overexpression on HIV-1, we measured the infectivity of viral particles produced from cotransfected HEK293T cells with 0.1 μg of cDNAs coding for PAR1 or PAR1 R46A using the TZM-bl reporter cell line (43). Three viral loads (1, 0.2, and 0.04 ng of p24) were tested, and the 0.04-ng virus load was found to be nonsaturating in the assay parameters (data not shown). Under these viral load conditions, we observed a significant (P = 3 × 10⁻¹³) 3-fold decrease in infectivity of viruses obtained from HEK293T cells overexpressing PAR1 compared to infectivity of the R46A mutant (Fig. 17C). Thus, PAR1 overexpression inhibits the infectious potential of released HIV-1 viral particles. These data represent the first evidence that PAR1 can reduce HIV-1 viral infectivity, likely due to its inhibition of viral glycoprotein processing by furin.