Cocaine-induced release of CXCL10 from pericytes regulates monocyte transmigration into the CNS

Cocaine use and its consequences continue to be a global epidemic. According to the World Drug Report, the number of cocaine users worldwide increased to 18.8 million in 2014 (United Nations Office on Drugs and Crime, 2016). Cocaine-related visits to hospital emergency departments remain a significant health care burden and accounted for ∼40.3% of all illegal drug-related emergency department visits in 2011 (Substance Abuse and Mental Health Services Administration, Center for Behavioral Health Statistics and Quality, 2013). Case reports indicate that cocaine use is often associated with seizures, cognitive impairment, depression, and an increased risk of stroke, all of which are major contributors to emergency department visits (Mendoza et al., 1992; Majlesi et al., 2010; Bodmer et al., 2014). Although the mechanisms underlying cocaine-associated central nervous system (CNS) disorders remain largely unknown, a variety of studies have implicated proinflammatory central immune signaling comprising both neuroexcitatory and neurotoxic effects as crucial factors in cocaine exposure/abuse (Rhoney, 2010; Fox et al., 2012; Li, 2016; Liao et al., 2016).

Monocytes are a subset of circulating white blood cells that can migrate across the blood–brain barrier (BBB) in pathological conditions and are implicated in the progression of many CNS neurodegenerative diseases, such as Alzheimer’s disease, multiple sclerosis, Parkinson’s disease, and human immunodeficiency virus (HIV)–associated neurocognitive disorders (Filion et al., 2003; Yao et al., 2010; Napuri et al., 2013; Grozdanov et al., 2014; Thériault et al., 2015). Intriguingly, cocaine has not only been found to enhance HIV-1 infectivity of monocyte-derived dendritic cells and macrophages (Dhillon et al., 2007), but also been shown to facilitate monocyte trafficking across the BBB, leading, in turn, to enhanced HIV disease progression and increased neuropathology (Yao et al., 2010, 2011b; Napuri et al., 2013; Dahal et al., 2015; Dash et al., 2015). The mechanisms by which cocaine elicits these responses, however, remain poorly understood.

Interstitial migration of monocytes is a dynamic, multistep process guided primarily by the local chemokine gradients that are formed by factors such as the C-C motif chemokine ligand 2 (CCL2), C-X3-C motif chemokine ligand 1, and C-X-C motif chemokine 10 (CXCL10; Taub et al., 1993; Yao et al., 2010; Pirvulescu et al., 2014). While CCL2 and C-X3-C motif chemokine ligand 1 have been widely linked with monocyte transmigration (Park et al., 2001; Butoi et al., 2011; Pirvulescu et al., 2014), studies of the role of CXCL10 in monocyte transmigration are limited. CXCL10, a proinflammatory chemokine produced by a variety of cell types including glia, dendritic cells, leukocytes, and endothelial cells (Taub et al., 1993; Vargas-Inchaustegui et al., 2010; Ioannidis et al., 2016), belongs to the CXCR3 (CD183) signaling family, including CXCL9/MIG (monokine-induced by γ-IFN), CXCL10/IP-10 (interferon inducible 10-kD protein) and CXCL11/I-TAC (inducible T cell-a chemoattractant). Elevated levels of CXCL10 have been associated with a variety of CNS diseases and viral infections such as tick-borne encephalitis, neuroborreliosis, Alzheimer’s disease, multiple sclerosis, and HIV-associated neurocognitive disorders (Lepej et al., 2007; Zajkowska et al., 2011; Mehla et al., 2012; Simmons et al., 2013; Krauthausen et al., 2015). One study demonstrated increased plasma levels of CXCL10 in HIV-infected cocaine abusers compared with nonusers, thereby underscoring its role as a biomarker (Kamat et al., 2012). CXCL10 is a critical chemokine that is dramatically up-regulated in HIV-associated neuropathogenesis (Lane et al., 2003; Cinque et al., 2005) and other neurodegenerative diseases (Sørensen et al., 2001; Corrêa et al., 2011). In this study, we sought to inquire if cells in proximity to the BBB played a role in contributing to the increased CXCL10.

Pericytes are fundamental components of the microvascular vessel wall and play a vital role in the development and regulation of the BBB and vascular function; however, their role in neuroinflammation (Hall et al., 2014) has not been explored. The aim of the present study was to identify the role of pericytes in cocaine-mediated monocyte transmigration, and the molecular mechanisms by which cocaine induces secretion of CXCL10 from human brain pericytes. Understanding the regulation of CXCL10 expression by cocaine could provide insights into the development of therapeutic targets for cocaine-mediated neuroinflammation.

In this study, we sought to assess the frontal cortices of postmortem brain tissues from no-cocaine controls and cocaine abusers for the presence of CD68⁺ (macrophage marker)/TMEM119⁻ (resident microglial marker) cells in proximity of Desmin⁺ pericytes. We found an increased number of CD68⁺/TMEM119⁻ cells accumulated around Desmin⁺ pericytes in brains of cocaine abusers compared with no-cocaine controls (Fig. 1 A). There was an increased number of CD68⁺/TMEM119⁻ cells in proximity to Desmin⁺ pericytes (Fig. 1 B). To understand the role of CXCL10 in pericyte–monocyte cross talk, we next sought to examine the expression of CXCL10 in the microvessels (MIVs) isolated from the frontal cortices of postmortem brain tissues from no-cocaine controls as well as cocaine abusers. MIVs were costained with anti-CXCL10 and the pericyte markers PDGF receptor-β (PDGFR-β), NG2, Desmin, or TBX18 as well as the endothelial cell marker CD31. There was increased CXCL10 expression in the pericyte marker PDGFR-β–positive cells in the MIVs isolated from cocaine abusers compared with no-cocaine controls (Fig. 1 C). Quantification of CXCL10 fluorescent intensity in PDGFR-β⁺ cells is shown in Fig. 1 D. Increased expression of CXCL10 was also observed in NG2⁺, Desmin⁺, and TBX18⁺ pericytes in MIVs of cocaine abusers, compared with no-cocaine controls (Fig. 1, E–J).

Based on the findings that in the brains of cocaine abusers, increased expression of CXCL10 was closely associated with enhanced monocyte transmigration, we next sought to determine the effect of cocaine exposure on the expression of CXCL10 in HBVPs. We first assessed dose and time course of CXCL10 expression following exposure of HBVPs to cocaine. For the dose curve, HBVPs were treated with varying concentrations (10⁻⁸, 10⁻⁷, 10⁻⁶, 10⁻⁵, and 10⁻⁴ M) of cocaine for 1 h, followed by assessment of CXCL10 mRNA by real-time PCR. In HBVPs, cocaine up-regulated expression of CXCL10 in a dose-dependent manner with a maximal response at 10⁻⁵ M (10 µM; 2.9-fold, P = 0.0049; Fig. 2 A). This concentration of cocaine was therefore used for all further experiments; physiologically relevant concentrations of cocaine found in the plasma of human cocaine addicts range from 0.4–13 µM (Stephens et al., 2004). All the cocaine concentrations tested failed to exert any toxicity on HBVPs, as determined by a cell viability assay (Fig. S1 C).

For the time-course study, HBVPs were exposed to cocaine (10 µM) for various time periods (1–24 h) followed by assessment of CXCL10 mRNA by real-time PCR. In HBVPs exposed to 10 µM cocaine, there was maximal induction of CXCL10 expression at 1 h (2.6-fold, P < 0.0001) with a drop in expression thereafter that persisted up to 3 h (Fig. 2 B). To validate the specificity of cocaine-mediated induction of CXCL10 in pericytes, we also assessed the effect of cocaine in another unrelated cell line—HEK293 cells. HEK293 cells were exposed to cocaine (10 µM) for various time periods (1–24 h) and assessed for the expression of CXCL10 mRNA by real-time PCR. As shown in Fig. S2 A and unlike the pericytes, cocaine exposure failed to up-regulate CXCL10 mRNA expression in HEK293 cells.

We next sought to determine whether cocaine-mediated induction of CXCL10 was regulated transcriptionally. For this we used six CXCL10 promoter–luciferase constructs reported previously by Dr. David Proud (University of Calgary, Calgary, Alberta, Canada; Koetzler et al., 2009)—the 972-bp full-length CXCL10 promoter (IFN-γ–inducible protein of pGL4), the 376-bp truncated CXCL10 promoter, and the CXCL10 promoter with point mutations in the activator protein-1 (AP-1), NF-κB1 (κB1), NF-κB2 (κB2), and IFN-stimulated response element (ISRE) recognition elements in the 972-bp full-length promoter. Herein, HBVPs were cotransfected with each of the indicated promoter plasmids described above, as well as with a Renilla luciferase–expressing plasmid. Following transfection, HBVPs were treated with cocaine, and promoter activity was assessed by monitoring relative luciferase activities using the Dual-Luciferase reporter 1000 assay system. Cocaine exposure resulted in significant activation of both full-length and truncated promoter luciferase constructs, with similar induction of luciferase activity by both the constructs (Fig. 2 C; 1.33-fold, P = 0.0446; 1.37-fold, P = 0.0408). Based on this result, it was inferred that the key promoter elements crucial for cocaine-mediated induction of CXCL10 transcription were contained within the 376-bp promoter, including AP-1, κB1, κB2, and ISRE recognition elements. To dissect the potential transcription factor binding sites engaged by cocaine on the CXCL10 promoter, the next step was to assess the ability of cocaine to induce/abrogate CXCL10 transcription in cells transfected with constructs containing various point mutations (AP-1, κB1, κB2, and ISRE recognition elements) in the full-length CXCL10 promoter. Mutations in the AP-1 site had no significant effect on the ability of cocaine to induce promoter activation (Fig. 2 C); however, mutations in the κB1, κB2, and ISRE sites all led to significant abrogation of cocaine-induced luciferase activity, thereby underscoring the role of these transcription factor binding sites in cocaine-mediated induction of CXCL10.

Since CXCL10 mRNA expression following cocaine exposure increased 2.9-fold in HBVPs, whereas the transcriptional activation only increased 1.33-fold, we hypothesized that the induction of CXCL10 mRNA by cocaine could be mediated, in part, by an increase in CXCL10 mRNA stability. To assess this, HBVPs were treated with actinomycin D (Act D; 10 µg/ml) in the presence or absence of cocaine for various time periods (30–180 min) to block the de novo mRNA synthesis, followed by assessment of CXCL10 mRNA levels by quantitative PCR. Cocaine exposure resulted in an increased half-life of CXCL10 mRNA (t_1/2 = 120 min) compared with the control group (t_1/2 = 77 min; Fig. 2 D).

To confirm whether increased CXCL10 mRNA levels manifested as increased protein translation, HBVPs were treated with 10 µM cocaine for 24 h followed by collection of supernatants that were assessed for a panel of cytokines/chemokines using the Luminex platform. As shown in the heatmap (Fig. 2 E), protein expression of CXCL10, VEGF, and CCL2 was significantly induced by cocaine in HBVPs compared with the unexposed controls (CXCL10: 5.19-fold, P = 0.000985; VEGF: 1.63-fold, P = 0.0289; CCL2: 1.46-fold, P = 0.0430); moreover, CXCL10 was the most prominently expressed chemokine among all 23 detectable chemokines. Out of a panel of 29 cytokines/chemokines, we were able to detect 23 cytokines/chemokines in our system. To validate our Luminex assay findings, HBVPs were treated with cocaine at varying concentrations (10⁻⁸, 10⁻⁷, 10⁻⁶, 10⁻⁵, and 10⁻⁴ M) for 24 h, followed by collection of supernatants that were assessed for expression of secreted CXCL10 using the ELISA assay. Similar to our Luminex results, cocaine induced CXCL10 release from HBVPs in a concentration-dependent manner with maximal response at 10⁻⁵ M (10 µM; 1,151 pg/ml, P < 0.0001) as determined by ELISA (Fig. 2 F). Taken together, these findings suggest that in HBVPs, cocaine mediated induction and release of CXCL10 by increasing the transcriptional activity, as well as mRNA stability of CXCL10 mRNA.

Since σ-1R is known to bind to a plethora of psychotropic drugs including cocaine (Hayashi and Su, 2003), we next sought to determine whether σ-1R was involved in cocaine-mediated induction of CXCL10 release from HBVPs. The first step was to examine the endogenous expression of σ-1R in HBVPs by RT-PCR. Basal expression level of σ-1R in HBVPs was lower compared with human brain microvascular endothelial cells (HBMECs; Fig. 3 A). Next, we sought to determine whether activation of σ-1R by cocaine involved translocation of these receptors from the ER lipid droplets to the plasma membrane. For this, we first isolated lipid rafts from HBVPs by sucrose gradient centrifugation and examined each fraction for the expression of σ-1R and GM1 by Western blot (Fig. 3 B). σ-1Rs were primarily present in fractions 8–10, while GM1 gangliosides were primarily present in fractions 4 and 5. Next, HBVPs were exposed to cocaine (10 µM) for 20 min, followed by lipid raft isolation and assessment of σ-1R and GM1 in fractions 4–6 by Western blot. Following cocaine exposure, there was a significant increase in σ-1R expression in the lipid raft fractions from HBVPs compared with controls (Fig. 3; expression of σ-1R in all the 10 fractions is shown in Fig. S2 B).

These findings were also validated by immunostaining of HBVP cells transfected with the σ-1R RFP plasmid. HBVPs were transfected with the σ-1R RFP plasmid for 24 h, exposed to cocaine for 15 min, followed by GM1 staining with Alexa-conjugated cholera toxin subunit B (CT-B), and assessed by fluorescence microscopy. In untreated control HBVPs, σ-1R (red) was mainly expressed in the cytoplasm and failed to colocalize with the lipid raft microdomains (Fig. 3 C, green). After cocaine exposure, however, there was colocalization of σ-1R within the lipid raft domains as evidenced by merging of green and red staining. This observation was quantified using Pearson’s correlation coefficient (Fig. 3 D).

Having determined that cocaine activated σ-1R via its translocation into the lipid domains of plasma membrane, we next sought to examine the role of σ-1R in cocaine-mediated induction of CXCL10 in HBVPs. Cocaine-mediated induction of CXCL10 expression was significantly attenuated in HBVPs pretreated with the σ-1R antagonist BD1047 (10 µM; Fig. 3 E). To further validate the involvement of σ-1R in cocaine-induced regulation of CXCL10 expression, we used a genetic knockdown approach by transfecting HBVPs with σ-1R siRNA (si–σ-1R). Transfection of HBVPs with si–σ-1R resulted in efficient knockdown of σ-1R expression (Fig. 3 F). In cells transfected with si–σ-1R, cocaine failed to up-regulate the expression of CXCL10 compared with cells transfected with scrambled siRNA. These findings thus underscore the role of σ-1R in cocaine-mediated induction of CXCL10 in HBVPs.

We next sought to examine whether cocaine-mediated induction of CXCL10 involved activation of c-Src. Exposure of HBVPs to cocaine resulted in a time-dependent increase in phosphorylation of c-Src, with activation as early as 5 min following exposure (Fig. 4 A). The specificity of Src signaling pathway was subsequently assessed using a pharmacological approach. Pretreatment of cells with the Src tyrosine kinase inhibitor PP2 resulted in abrogation of cocaine-induced phosphorylation of c-Src (Fig. 4 B). As expected, pretreatment of cells with the inactive analogue PP3 failed to abrogate cocaine-induced phosphorylation of c-Src. To examine the functional role of c-Src in cocaine-mediated induction of CXCL10 expression, HBVPs were pretreated with PP2 or PP3 for 1 h, followed by exposure of cells to cocaine for 24 h, followed by assessment of CXCL10 secretion by ELISA. Pretreatment of cells with PP2, not with PP3, significantly inhibited cocaine-mediated induction of CXCL10 secretion (Fig. 4 C).

Since σ-1R translocated to the lipid raft domain in the membrane following exposure to cocaine, we sought to determine the presence of any protein–protein interactions between σ-1R and c-Src kinase. First, a series of coimmunoprecipitation assays was performed using lysates from cocaine-treated/nontreated HBVPs. In HBVPs, the amount of σ-1R in the c-Src–immunoprecipitated protein complex was increased following cocaine exposure, while the amount of total c-Src remained unchanged (Fig. 5 A). These findings allude to a possible interaction between σ-1R and c-Src kinase following cocaine exposure. For additional validation of this interaction, σ-1R was immunoprecipitated from HBVP lysates and assessed for expression of c-Src by Western blot. Similar to the c-Src immunoprecipitated complex, the amount of c-Src in the σ-1R–immunoprecipitated protein complex was increased following cocaine exposure, while the amount of total σ-1R remained unchanged.

The protein–protein interaction between σ-1R and c-Src was further validated by confocal microscopy. HEK293 cells were cotransfected with c-Src–GFP and σ-1R–RFP and assessed for colocalization of c-Src and σ-1R by confocal microscopy. As expected, c-Src and σ-1R were found in the cytoplasm of HEK293 cells with strong colocalization (yellow) of green and red fluorescence observed within the cells, demonstrating an interaction of c-Src with σ-1R (Fig. 5 B).

Since p-Src is the active form of c-Src (Rutledge et al., 2014), we next examined the interaction of p-Src with σ-1R in HBVPs. HBVPs were treated with cocaine (10 µM) for 15 min followed by assessment of p-Src (green color) and σ-1R (red color) by double immunostaining. The intensity of p-Src fluorescence and the colocalization of p-Src and σ-1R were both dramatically induced following cocaine exposure in HBVPs compared with control cells (Fig. 5 C). This finding was quantified using Pearson’s correlation coefficient and shown in Fig. 5 D.

We next wanted to investigate whether activation of σ-1R was critical for cocaine-mediated phosphorylation of c-Src. Cocaine-mediated phosphorylation of c-Src was significantly attenuated in HBVPs pretreated with the σ-1R antagonist BD1047 (10 µM; Fig. 5 E). Similarly, transfection of HBVPs with σ-1R siRNA further validated the involvement of σ-1R in cocaine-induced phosphorylation of c-Src. In cells transfected with σ-1R siRNA, cocaine failed to mediate phosphorylation of c-Src, unlike the cells transfected with scrambled siRNA (Fig. 5 F). These findings thus demonstrate that cocaine-mediated activation of σ-1R is upstream of phosphorylation of c-Src.

Based on the premise that the activation of σ-1R and c-Src plays a role in cocaine-mediated induction of CXCL10, and the fact that Src kinases are important for PDGFR-β signaling transduction, the next step was to elucidate whether cocaine-mediated CXCL10 expression involved PDGFR-β signaling. For this, HBVPs were exposed to cocaine for various time periods (between 5 min and 3 h) followed by assessment of the cell lysates for phosphorylation of PDGFR-β by immunoprecipitation assay using the anti-phosphotyrosine antibody (clone 4G10). In HBVPs, cocaine exposure resulted in phosphorylation of PDGFR-β as early as 5 min (Fig. 6 A). We next sought to elucidate whether PDGFR-β activation played a role in cocaine-mediated induction of CXCL10 expression. To address this, HBVPs were pretreated with a tyrosine kinase inhibitor, STI571 (10 µM), followed by exposure of cells to cocaine and assessment of CXCL10 release by ELISA. Pretreatment of cells with STI571 resulted in significant amelioration of cocaine-mediated release of CXCL10 from HBVPs (Fig. 6 B). Further validation of the involvement of PDGFR-β in cocaine-mediated CXCL10 release from HBVPs was performed in cells transfected with the siRNA for PDGFR-β, followed by assessment of CXCL10 release by ELISA assay. Transfection of HBVPs with PDGFR-β siRNA (si–PDGFR-β) resulted in efficient knockdown of PDGFR-β expression (Fig. 6 C). In these cells, cocaine failed to up-regulate CXCL10 release from HBVPs compared with HBVPs transfected with scrambled siRNA (si-Con), wherein, as expected, cocaine exposure up-regulated the expression of CXCL10.

The next step was to elucidate the role of c-Src in cocaine-mediated phosphorylation of PDGFR-β in HBVPs. Briefly, before cocaine exposure, HBVPs were pretreated with the tyrosine kinase inhibitor STI571 (10 µM) followed by assessment of phosphorylation of c-Src by Western blot. In cells pretreated with STI571, cocaine was able to induce phosphorylation of c-Src, demonstrating thereby that c-Src was upstream of PDGFR-β signaling (Fig. 6 D). Reciprocally, HBVPs were also pretreated with the Src tyrosine kinase inhibitor PP2 or its inactive analogue PP3, followed by cocaine exposure and assessment of phosphorylation of PDGFR-β. In cells pretreated with PP2 but not PP3, cocaine-mediated phosphorylation of PDGFR-β was abrogated, thereby suggesting that c-Src lies upstream of PDGFR-β phosphorylation (Fig. 6 E).

For further validation of the role of NF-κB, we next sought to examine the cocaine-mediated translocation of NF-κB p65 into the nucleus in HBVPs. Briefly, HBVPs were exposed to cocaine for various periods of time (5–180 min) and assessed for translocation of the p65 subunit of NF-κB into the nucleus. Exposure of HBVPs to cocaine resulted in a time-dependent increase in translocation of the NF-κB p65 subunit in the nucleus with a concomitant decrease in its expression in the cytosol (Fig. 7, A and B). Additional confirmation of these findings by immunostaining also revealed enhanced translocation of NF-κB into the nucleus in HBVPs at 15 min after cocaine (10 µM) exposure. Images were captured by fluorescence microscopy, using a (63×/1.4) objective lens. Cocaine exposure resulted in a significant increase in the intensity of NF-κB fluorescence in the nucleus with a concomitant decrease in the cytoplasm (Fig. 7 C). Quantification of NF-κB nuclear translocation is shown in Fig. 7 D. We next sought to examine whether NF-κB was involved in secretion of CXCL10. For this, HBVPs were pretreated with SC514, a selective IκB kinase-2 (IKK-2) inhibitor, to specifically suppress the transcription of NF-κB, followed by exposure to cocaine and assessment of CXCL10 release by ELISA. Pretreatment of cells with SC514 resulted in significant amelioration of cocaine-mediated CXCL10 release from HBVPs (Fig. 7 E).

To assess if there was a link that could tie together the activation of σ-1R–c-Src–PDGFR-β signaling with NF-κB, HBVPs were transfected with either nontarget siRNA control or si–σ-1R followed by exposure of transfected cells to cocaine for 15 min and assessment of NF-κB expression in the nuclear fraction. As expected, in HBVPs transfected with si–σ-1R, cocaine exposure failed to increase the nuclear translocation of NF-κB compared with cells transfected with the nontarget siRNA control, wherein cocaine exposure mediated increased translocation of NF-κB (Fig. 7 F). To examine the role of c-Src in cocaine-mediated nuclear translocation of NF-κB, HBVPs were pretreated with either the Src tyrosine kinase inhibitor PP2 or its inactive analogue PP3, followed by cocaine exposure. Pretreatment with PP2, but not with PP3, resulted in abrogation of cocaine-mediated nuclear translocation of NF-κB (Fig. 7 G). Having demonstrated the σ-1R and c-Src were involved in cocaine-mediated nuclear translocation of NF-κB, we next sought to examine the role of PDGFR-β in cocaine-induced NF-κB nuclear translocation. HBVPs were transfected with either nontarget siRNA control or si–PDGFR-β, followed by exposure of transfected cells to cocaine for 15 min, and assessed for expression of NF-κB in the nuclear fractions. Cocaine mediated an increase in translocation of NF-κB in cells transfected with the nontarget siRNA control but not in cells transfected with si–PDGFR-β (Fig. 7 H). These findings thus underpin the involvement of σ-1R–c-Src–PDGFR-β signaling in cocaine-mediated activation of NF-κB in HBVPs.

The next step was to examine a link between NF-κB activation and the upstream signaling pathways in cocaine-mediated up-regulation of CXCL10 expression in HBVPs. We thus first sought to assess whether pharmacological inhibition of NF-κB would affect cocaine-mediated translocation of σ-1R. For this, HBVPs were pretreated with SC514 for 1 h, then exposed to cocaine for an additional 15 min and assessed for translocation of σ-1R. In cells pretreated with SC514, cocaine was still able to induce translocation of σ-1R to the lipid rafts (Fig. S3, A and B). Next, we sought to examine whether inhibition of NF-κB could affect cocaine-mediated phosphorylation of c-Src. Briefly, HBVPs were pretreated with SC514 (10 µM) followed by assessing phosphorylation of c-Src by Western blots. In the presence of SC514, cocaine continued to mediate phosphorylation of c-Src (Fig. S3 C). We next assessed cocaine-mediated phosphorylation of PDGFR-β in the presence of SC514. HBVPs were pretreated with SC514, followed by cocaine exposure, and assessed for phosphorylation of PDGFR-β. In the presence of SC514, cocaine-mediated phosphorylation of PDGFR-β continued to persist (Fig. S3 D).

We next sought to determine whether cocaine-mediated up-regulation of released CXCL10 in HBVPs could induce monocyte transmigration. We used an in vitro BBB model to assess monocyte transmigration (Fig. 8 A). Briefly, HBMECs (4 × 10⁴/well) were seeded onto a Transwell (Corning) until confluence was reached. Subsequently, green fluorescence cell tracker–labeled monocytes were added to the upper side of the Transwell, and pericyte-conditioned medium (PCM) from either the control or cocaine-exposed HBVPs was added to the lower chamber. Control PCM (without the cells) was incubated with 10 µM cocaine for 24 h and used as a negative control (CM-Cocaine) to eliminate any direct effect of cocaine on monocyte transmigration. Varying concentrations (25–100%) of conditioned medium incubated with cocaine were also added to the lower chamber of the Transwell and compared with conditioned medium collected from cocaine-treated and -untreated cultured pericytes. A dramatic increase in monocyte transmigration was observed in the presence of cocaine-treated PCM (PCM-cocaine) at all the concentrations tested compared with the PCM-Con (Fig. 8 B). Furthermore, we also sought to examine the transendothelial electrical resistance (TEER) and FITC Dextran-4 permeability of HBMECs following exposure to PCM-cocaine/PCM-Con/CM-cocaine. PCM-cocaine and PCM-Con at 50% concentration failed to exert any significant effect on TEER as well as permeability (Fig. S4, A and B). It was only at 100% concertation that PCM-cocaine showed significantly decreased TEER and increased permeability in the in vitro BBB model (Fig. S4, A and B).

Based on our findings that cocaine dose dependently increased CXCL10 expression in the culture medium of HBVPs (Fig. 2 G), we next sought to examine whether cocaine could also dose-dependently enhance monocyte transmigration across the BBB model. To address this, PCM was collected from HBVPs exposed to varying concentrations of cocaine (1, 10, and 100 µM) and assessed for its effect on monocyte transmigration across an in vitro BBB model. Control pericyte medium (without the cells) was incubated with 10 µM cocaine for 24 h and used as a negative control to eliminate any direct effect of cocaine on monocyte transmigration. Although control medium incubated with cocaine (10 µM) had the ability to induce a low level of monocyte transmigration, a more dramatic increase in monocyte transmigration was observed in the presence of PCM-cocaine, and this effect was dependent on the concentration of cocaine, with the maximal effect observed in PCM-cocaine (10 µM; Fig. 8 C). These findings thus demonstrate that PCM-cocaine enhanced monocyte transmigration compared with PCM-Con.

We next sought to examine whether CXCL10 released from HBVPs following cocaine exposure was a critical chemokine for enhanced monocyte transmigration. To answer this, PCM-Con and PCM-cocaine were incubated with either neutralizing anti-CXCL10 antibody or normal IgG antibody (2.0 µg/ml), followed by assessment of monocyte transmigration to these respective PCMs. PCM-cocaine incubated with neutralizing CXCL10 antibody failed to induce monocyte transmigration compared with PCM-cocaine incubated with the control IgG antibody, which resulted in enhanced monocyte transmigration (Fig. 8 D). To further confirm the effect of CXCL10 on monocyte transmigration, monocytes were pretreated with the CXCL10 receptor (CXCR3) antagonist AMG487 (1 µM) for 1 h and assessed as described before for transmigration across the BBB to either control or PCM-cocaine. Pretreatment of monocytes with AMG487 resulted in significant amelioration of PCM-cocaine to mediate monocyte transmigration. Control monocytes, not treated with the antagonist, exhibited increased migration to PCM-cocaine, as expected (Fig. 8 E).

Further validation of the involvement of σ-1R in PCM-cocaine–mediated transmigration of monocytes was performed in HBVPs transfected with the siRNA for σ-1R. Briefly, HBVPs were transfected with either σ-1R or nonsense (control) siRNA for 24 h followed by exposure of cells to cocaine for another 24 h, after which the PCM was collected and added to the bottom chamber of the Transwell. PCM-cocaine collected from HBVPs transfected with si–σ-1R failed to augment monocyte transmigration (Fig. 8 F). On the other hand, as expected, PCM-cocaine collected from HBVPs transfected with the nontarget siRNA control mediated an increase in monocyte transmigration. Next, we sought to examine the involvement of PDGFR-β in PCM-cocaine–mediated transmigration of monocytes by transfecting HBVPs with either control or si–PDGFR-β. HBVPs were transfected with either the PDGFR-β or nonsense (control) siRNA for 24 h followed by exposure of cells to cocaine for another 24 h, after which the PCM was collected and added to the bottom chamber of the Transwell for transmigration assays. PCM-cocaine collected from HBVPs transfected with si–PDGFR-β failed to augment monocyte transmigration, while, as expected, PCM-cocaine collected from HBVPs transfected with the nontarget siRNA control mediated an increase in monocyte transmigration (Fig. 8 G).

To further validate cocaine-mediated up-regulation of CXCL10 expression in pericytes in vivo, C57BL/6N mice (male, n = 4) were injected with cocaine (20 mg/kg) or saline i.p. for 7 d, followed by assessment of the expression of CXCL10 in brain capillaries isolated from cocaine- versus saline-treated mice. MIVs were coimmunostained for pericyte marker (NG2, PDGFR-β, Desmin, or TBX18; red color) and CXCL10 (green color) and assessed by fluorescence microscopy. There was increased expression of CXCL10 in NG2⁺ cells in MIVs isolated from cocaine-treated versus saline-treated mice (Fig. 9, A and B). Similar findings were observed in PDGFR-β⁺, Desmin⁺ or TBX18⁺ cells in the MIVs (Fig. S5).

Based on the premise that pericyte coverage and vascular permeability are both important indicators of MIV functioning (Proebstl et al., 2012), we next sought to assess pericyte coverage and vascular permeability in MIVs from cocaine-treated mice. Brain sections from cocaine or saline-treated mice were costained with the pericyte markers (PDGFR-β or Desmin; red) and an endothelial cell marker (CD31; green). There was significantly decreased pericyte coverage in MIVs of mice administered cocaine compared with the saline group (Fig. S6). Brain sections from mice administered saline or cocaine were costained for fibrinogen, as well as the pericyte marker (Desmin) and endothelial cell marker (CD31), and quantified for perivascular fibrinogen leak. There was significantly increased fibrinogen leak in the brains of mice administered cocaine versus saline (Fig. S4 C).

Since CXCL10 expression was increased in cocaine-treated mice, we next wanted to validate the role of CXCL10 in monocyte transmigration in vivo. For this, C57BL/6N mice (male, n = 4) were administered either saline or CXCL10 (200 µg/ml, 4 µl) stereotactically at the coordinates +1.34 mm behind the bregma, +1.25 mm lateral from the sagittal midline at the depth of −4.0 mm to skull surface, followed 24 h later by tail vein injection of mouse bone marrow–derived monocytes (BMMs) isolated from CX3CR1–GFP mice. After an additional 24 h, animals were euthanized and brain sections stained with anti-GFP and anti-F4/80 antibodies. The cortex, hippocampus, thalamus, and striatal regions were assessed for distribution of F4/80⁺ and GFP⁺/F4/80⁺ cells. Increased numbers of both F4/80⁺ and GFP⁺ and F4/80⁺ cells were found in the hippocampi of mice administrated CXCL10 compared with the saline group (Fig. 9, C and D); similar findings were observed in the thalamus, cortex, and striatum of cocaine-treated mice (Fig. S7).

To validate the role of the monocyte CXCL10 receptor, CXCR3, in mediating monocyte transmigration, mice were injected with either saline or cocaine (20 mg/kg) for 7 d followed by tail vein injection of mouse BMMs isolated from CX3CR1-GFP mice pretreated with or without the CXCR3 antagonist AMG487. 24 h after the cell infusion, animals were euthanized, and brain sections were stained with anti-GFP, anti-F4/80, and anti-PDGFR-β antibodies. F4/80⁺ as well as double-positive GFP⁺/F4/80⁺ cells in the cortex, hippocampus, thalamus, and striatum were quantified. In the presence of cocaine there were increased numbers of F4/80⁺ as well as double-positive GFP⁺/F4/80⁺ cells in all the brain regions examined compared with the saline group (Fig. 9, E–L). In animals infused with BMMs isolated from CX3CR1-GFP mice that were pretreated with AMG487, cocaine failed to induce double-positive GFP⁺/F4/80⁺ cells, but not F4/80⁺ cells, in all the brain regions examined.

Cocaine abuse is closely linked with HIV infection (Dahal et al., 2015; Dash et al., 2015). Many studies have demonstrated an elevated risk of HIV acquisition and an enhanced progression to neurological HIV in cocaine abusers (Anthony et al., 1991; Chiasson et al., 1991; Doherty et al., 2000). Both cell culture and animal studies demonstrate that cocaine not only facilitates HIV replication in various cell types (Roth et al., 2002; Steele et al., 2003) but also results in glial cell activation, neurotoxicity, and BBB breakdown (Fiala et al., 1998; Gekker et al., 2006; Kousik et al., 2012). Cocaine has a detrimental effect on most cell types of the CNS, including the endothelium (Brailoiu et al., 2016), neurons (Mills et al., 2017), microglia (Liao et al., 2016), and astrocytes (Yang et al., 2010), likely underpinning the link between cocaine and HIV-associated neurocognitive impairment. Our findings add yet another novel role of pericytes in cocaine-mediated increased monocyte transmigration across the BBB. Here, we demonstrate a novel molecular mechanism underlying cocaine-mediated induction of CXCL10 expression in HBVPs, involving sequential activation and translocation of σ-1R, with subsequent interaction with c-Src, which in turn resulted in phosphorylation of PDGFR-β and nuclear translocation of NF-κB, leading ultimately to enhanced expression of CXCL10 (Fig. 10).

Pericytes are mural cells that are essential components of the microvascular vessels in direct contact with the endothelial cells (Sweeney et al., 2016). The integral role of pericytes in the development and maintenance of the cerebrovascular unit has been well recognized (Bergers and Song, 2005; Bell et al., 2010; Armulik et al., 2011), including regulation of the cerebral blood flow and neurovascular coupling (Kisler et al., 2017). Additionally, it has been reported that reduced coverage of pericytes on the endothelium can lead to MIV dysfunction and leakage (Sagare et al., 2013; Villaseñor et al., 2017). Whether these cells participate in inflammatory processes, however, remains poorly understood, especially in the context of cocaine-mediated brain inflammation. Herein, we provided evidence that in pericytes, cocaine mediates induction of the proinflammatory chemokine CXCL10 at both the transcriptional and post-transcriptional levels. CXCL10 belongs to the C-X-C chemokine family that binds to the G protein–coupled receptor CXCR3, and plays an important role in the chemotaxis of immune cells including T cells, monocytes, and NK cells (Taub et al., 1993). Increased expression of various chemokines in the CNS has been implicated in HIV-associated dementia (Sanders et al., 1998). Specifically, CXCL10 has been detected in the cerebrospinal fluid and brains of HIV-1–infected individuals (McArthur et al., 1993; Kolson and Pomerantz, 1996; Sanders et al., 1998; Kolb et al., 1999). In line with these findings, CXCL10 has also been identified as a plasma inflammatory biomarker signature of immune activation in HIV patients on antiretroviral therapy (Kamat et al., 2012). In the current study, we demonstrate that in HBVPs, cocaine exposure resulted in enhanced transcriptional regulation and stability of CXCL10 mRNA. We also showed that release of CXCL10 from cocaine-exposed pericytes was critical for transmigration of monocytes across the brain endothelium in both in vitro and in vivo model systems.

σ-1R is an intracellular molecular chaperone (28 kD) that is predominantly located in the ER membrane (Nguyen et al., 2015) and has been shown to bind to cocaine. It has also been implicated in cocaine addiction and toxicity (Yasui and Su, 2016). Intriguingly, as a target receptor for cocaine, several studies have also explored it as a therapeutic target for the treatment of cocaine-mediated neuropathology (Yasui and Su, 2016). Previous studies have shown that cocaine exacerbated neuroinflammation via activation of σ-1R in endothelial cells, astrocytes, and microglia (Yao et al., 2010, 2011b; Yang et al., 2015). Based on our previous findings, cocaine-mediated translocation of σ-1R increased the permeability of BBB (Yao et al., 2011b), induced chemokine monocyte chemotactic protein-1 (CCL2) in microglia (Yao et al., 2010), provoked transcription of GFAP (Yang et al., 2010), and exacerbated neurotoxicity mediated by HIV viral proteins (Yao et al., 2009). In the current study, we demonstrated that σ-1R was also present in the cytoplasm of pericytes and, in the presence of cocaine, it was found to translocate from the ER lipid droplets to the plasma membrane. Moreover, we observed that σ-1R activation was critical for cocaine-mediated induction of CXCL10 expression since pretreatment of HBVPs with the σ-1R antagonist BD1047 or in cells transfected with si–σ-1R, cocaine failed to induce induction of CXCL10.

Further dissection of the signaling pathways involved in cocaine-mediated induction of CXCL10 using pharmacological approaches revealed sequential activation of c-Src kinase, a finding that is consistent with previous reports on the effects of cocaine in HBMECs (Yao et al., 2011b). Herein, for the first time, we provide evidence that in HBVPs, cocaine mediates a direct physical interaction between σ-1R and c-Src. An interaction of σ-1R with c-Src was further confirmed by cotransfection of GFP-tagged c-Src and RFP-tagged σ-1R (σ-1R–RFP) plasmid constructs in HEK293 cells. Additionally, an interaction was validated by assessing endogenous levels of p-Src and σ-1R following cocaine stimulation in HBVPs by double immunostaining. Modulation of MAPKs by σ-1R has been reported by various investigators (Hayashi and Su, 2005; Yao et al., 2011b; Du et al., 2017). In the current study, using both pharmacological and genetic approaches, we demonstrated the role of σ-1R in cocaine-mediated phosphorylation of c-Src.

Our findings also suggest that drugs of abuse such as cocaine could directly phosphorylate cellular receptors such as PDGFR-β as early as 5 min. Further evaluation of the functional role of activated PDGFR-β was conducted using both pharmacological and genetic approaches. Pretreatment of HBVPs with either a receptor tyrosine kinase antagonist, such as STI571 or si–PDGFR-β, resulted in abrogation of cocaine-mediated induction of CXCL10 release from HBVPs, thereby implicating the role of PDGF signaling in this process. In this study, we found that cocaine-mediated phosphorylation of PDGFR-β was dependent on upstream activation of c-Src, since inhibition of Src kinase significantly blocked cocaine-mediated phosphorylation of PDGFR-β. Interestingly, inhibiting PDGFR-β had no effect on cocaine-induced activation of c-Src.

The transcription factor NF-κB has emerged as a major regulatory transcription factor for a number of genes, including cytokines such as CXCL10 (Fitzgerald et al., 2003; Krinninger et al., 2011). Our findings demonstrated a time-dependent translocation of NF-κB in HBVPs following cocaine stimulation. Further dissection of NF-κB regulation using both the pharmacological and genetic approaches revealed that NF-κB nuclear translocation was dependent on sequential activation of σ-1R, c-Src, and PDGFR-β signaling. It is likely that NF-κB, the downstream effector of the σ-1R–c-Src–PDGFR-β signaling, could also via feedback regulation mediate translocation of σ-1R as well as phosphorylation of downstream mediators such as c-Src and PDGFR-β. To investigate, we pretreated the cells with the IKK-2 inhibitor SC514 to block NF-κB nuclear translocation and demonstrated that NF-κB did not regulate the upstream σ-1R–c-Src–PDGFR-β signaling via the feedback regulation.

In the present study, we report an interaction of pericyte-released CXCL10 with monocytes, leading to increased transmigration of these latter cells in both in vitro as well as in vivo model systems. Our findings demonstrate that PCM derived from cocaine-treated HBVPs (PCM-cocaine) induced monocyte transmigration in a concentration-dependent manner. The direct effect of cocaine on monocyte transmigration was eliminated by using a negative control conditioned medium (without the HBVPs) incubated with cocaine (CM-cocaine). It was found that CM-cocaine induced monocyte transmigration only at 100% concentration, indicating thereby that cocaine itself had an effect on monocyte transmigration. PCM-cocaine, on the other hand, induced monocyte transmigration at all the concentrations tested (even as low as 25% concentration), demonstrating thereby that the chemokines released from cocaine-stimulated HBVPs had a stronger effect on monocyte transmigration. Additionally, we assessed TEER as well as BBB permeability in the in vitro model. PCM-cocaine as well as CM-cocaine were found to increase TEER while also decreasing the permeability of the in vitro BBB model, albeit only at 100% concentration. At lower concentrations (25–75%), however, only PCM-cocaine was able to induce monocyte transmigration. Intriguingly, there was not change in either TEER or BBB permeability in the presence of PCM-cocaine, demonstrating thereby that PCM-cocaine–mediated monocyte transmigration was unrelated to change in permeability. To attribute the specificity of CXCL10 in the PCM-cocaine as contributing to enhanced monocyte transmigration, PCM-cocaine was treated with CXCL10 neutralizing antibody and assessed for monocyte migration. PCM-cocaine with was specifically able to block monocyte transmigration. Further validation of the role of CXCL10 and its signaling mediators in PCM-cocaine–mediated induction of monocyte transmigration was also determined by pharmacological and genetic approaches targeted at inactivation of σ-1R, PDGFR-β, and CXCR3 pathways.

To further unravel the effect of cocaine on CXCL10 induction in pericytes, MIVs from cocaine- or saline-treated mice were isolated and examined for expression of CXCL10 in pericytes. We observed increased localization of CXCL10 in NG2⁺/PDGFR-β⁺/Desmin⁺/TBX18⁺ pericytes in the MIVs isolated from cocaine-treated mice compared with vessels isolated form saline controls. Of note, four different pericyte markers were chosen to validate our findings. It should be mentioned that out of these markers, TBX18 is expressed in pericytes as well as vascular smooth muscle cells (Guimaraes-Camboa et al., 2017). Furthermore, in vivo validation of these findings also demonstrated that cocaine administration increased monocyte transmigration within the brain, and that this effect could be blocked in mice administered monocytes pretreated with the CXCR3 antagonist AMG487. These findings thus lend credence to a novel role of pericytes in cocaine-induced neuroinflammation mediated by CXCL10. Furthermore, we found increased fibrinogen leak and decreased pericyte coverage around the MIVs of mice administrated cocaine for 7 d, demonstrating thereby that cocaine increased the permeability of MIVs in vivo. Our findings are in agreement with previous reports describing the role of cocaine in mediating BBB breach (Sharma et al., 2009; Yao et al., 2011a). The interaction of pericytes with peripheral monocytes has ramifications for yet another novel mechanism by which cocaine abuse can exacerbate neuroinflammation specifically under conditions of HIV infection that are associated with BBB breach (Fiala et al., 1998; Napuri et al., 2013; Dahal et al., 2015).

In summary, our findings have delineated a detailed molecular pathway for cocaine-mediated induction of CXCL10 in pericytes with its ensuing effects causing exacerbation of neuroinflammation. Briefly, we have shown that cocaine-mediated induction of CXCL10 expression in HBVPs involves sequential activation and translocation of σ-1R with subsequent interaction of σ-1R with c-Src. Subsequent activation of c-Src leads to phosphorylation and activation of PDGFR-β, with downstream activation and nuclear translocation of NF-κB, leading ultimately to enhanced expression and release of CXCL10. Increased secretion of CXCL10, in turn, facilitates increased transmigration of monocytes across the BBB, thereby promoting enhanced inflammation in the brain. Strategies aimed at blocking cocaine-mediated signaling pathway could be developed as therapeutics to dampen cocaine-mediated neuroinflammation, with applications for other neurodegenerative disorders.

C57BL/6 mice (male, 6–8 wk) were purchased from Charles River Laboratories, Inc. CX3CR1-GFP homozygous mice were obtained from the Jackson Laboratory and have a targeted deletion of CX3CR1 that is replaced by a GFP reporter insertion. All animals were housed under conditions of constant temperature and humidity on a 12-h light/12-h dark cycle, with lights on at 7:00 a.m. Food and water were available ad libitum. All animal procedures were performed according to the protocols approved by the Institutional Animal Care and Use Committee at the University of Nebraska Medical Center.

Primary HBVPs were purchased from ScienCell and cultured in the pericyte medium (ScienCell). HBVPs were isolated from fetal human brain and were positive for the pericyte markers PDGFR-β, NG2, Desmin, and TBX18. Pericytes had negligible staining for αSM as well as the endothelial cell marker CD31 (>98% purification; Fig. S1, A and B). Cell culture dishes were coated with poly-l-lysine (2 µg/cm²; ScienCell), and cells were used in passages 2–5.

Primary HBMECs obtained from Dr. Monique Stins (Johns Hopkins University, Baltimore, MD) were cultured in RPMI 1640 medium (GE Healthcare Life Sciences) containing 10% heat-inactivated FBS (Atlanta Biologicals), 10% Nu-Serum (BD Biosciences), 2 mM l-glutamine (Invitrogen), 1 mM pyruvate (GE Healthcare Life Sciences), penicillin (100 U/ml), streptomycin (100 µg/ml; Gibco), essential amino acids (HyClone), and MEM vitamin solution (HyClone). Purified HBMECs were positive for endothelial makers DiI-AcLDL, ZO-1, and β-catenin and were found to be >99% pure after exclusion of staining for nonendothelial cell type markers (GFAP, smooth muscle actin, cytokeratin, and macrophage antigens), as described previously (Wen et al., 2011).

Human monocytes were obtained from HIV-1, HIV-2, and hepatitis B seronegative donor leukopacks and separated by countercurrent centrifugal elutriation, as previously described (Gendelman et al., 1988). Freshly elutriated monocytes were cultured in DMEM (Gibco) containing 10% heat-inactivated human serum (Thermo Fisher Scientific), 2 mM l-glutamine (Invitrogen), 100 mg/ml gentamicin, and 10 mg/ml ciprofloxacin (Sigma-Aldrich). Human monocytes were used for the in vitro study, while BMMs isolated from mice were used for the in vivo study.

CX3CR1-GFP homozygous mice (Jackson Laboratory), 6–8 wk of age, were used as BMM donors. Briefly, the femur was removed, and bone marrow cells were dissociated into single-cell suspensions and were cultured for 5 d supplemented with 1,000 U/ml of macrophage colony-stimulating factor.

Cocaine, σ-1R antagonist BD1047, IKK-2 inhibitor SC514, and FITC Dextran-4 were purchased from Sigma-Aldrich. Tyrosine kinase inhibitor STI571 was obtained from Novartis. Src kinase inhibitor (PP2) and its inactive orthologue (PP3) were purchased from Calbiochem. CXCR3 antagonist AMG487 was obtained from Tocris Bioscience. The concentrations of these inhibitors were based on the concentration curve study and our previous reports (Niu et al., 2014).

Cell Tracker Green 5-chloromethylfluorescein diacetate was purchased from Invitrogen. The human CXCL10/IP-10 DuoSet Kit was obtained from R&D Systems. CT-B conjugated to Alexa Fluor 488 was purchased from Invitrogen. Neutralizing human CXCL10/IP-10 antibody was purchased from R&D Systems.

The conditions for RT- and real-time PCR assays have been described previously (Niu et al., 2014). Real-time PCR primers for human CXCL10 and 18S were obtained from SA Biosciences. Total RNA was extracted with Trizol reagent (Invitrogen) according to the manufacturer’s instructions. Quantitative analyses of mRNA were conducted using the ABI 7500 Fast Real-Time PCR system (Applied Biosystems). Real-time PCR amplifications were performed for 40 cycles (denaturation: 30 s at 95°C; annealing: 1 min at 60°C) and RT-PCR for 28 cycles (denaturation: 30 s at 94°C; annealing: 30 s at 55°C; extension: 30 s at 72°C). Primers were as follows: σ-1R, forward primer: 5′-CCCATGGGAACAAATGAGACA-3′; reverse primer: 5′-CCAGATGGGTGTGAGTGCAT-3′; GAPDH (real-time PCR), forward primer: 5′-CTGTGGGCAAGGTCATCCCTG-3′; reverse primer: 5′-AGACGGCAGGTCAGGTCCACC-3′; GAPDH (RT-PCR), forward primer: 5′-CGTGGAAGGACTCATGACCA-3′; reverse primer: 5′-GCCTGCTTCACCACCTTCTT-3′.

HBVPs at passages 3 and 4 were used for collection of PCM. For serum starvation, HBVP medium was replaced with FBS-free pericyte medium followed by overnight starvation and exposure to cocaine (10 µM). After 24-h incubation, PCM from PCM-Con– and PCM-cocaine–exposed groups were collected and centrifuged at 1,000 rpm for 5 min to remove cell debris. Collected supernatant was stored at −80°C until ready for use in transmigration assays.

Supernatants were collected from HBVPs after 24 h of cocaine exposure, and the expression of cytokines/chemokines in supernatants was measured using the Milliplex MAP kit human Cytokine/Chemokine Magnetic Bead Panel Immunoassay (Millipore), which contains 29 different kinds of cytokines/chemokines as per the manufacturer’s instructions. Briefly, this is a bead-based suspension array using the Luminex MAP technology in which fluorescent-coded beads have cytokine capture antibodies on the bead surface to bind the proteins. The heatmap was created by R version 3.2.3 software. Data represent results obtained from three biological replicates.

Culture supernatants were collected from HBVPs exposed to cocaine or from untreated cells and assessed for expression of CXCL10 protein using a Human CXCL10/IP-10 DuoSet Kit (R&D Systems). Samples were analyzed for CXCL10 expression with three biological replicates, and each biological replicate had three or four technical replicates.

HBVPs were transfected with siRNA of si-Con, si–σ-1R, or si–PDGFR-β (Thermo Fisher Scientific). The knockdown efficiency of siRNAs was determined by Western blot 1 d after transfection. HEK293 cells were cultured in the coverslip and cotransfected with plasmid constructs containing GFP-tagged c-Src and plasmid constructs containing σ-1R–RFP for 24 h followed by assessment of colocalization of GFP-tagged c-Src and σ-1R–RFP using an inverted LSM 700 confocal microscope with a 40× oil objective (Zeiss). Images were processed with the Zeiss Zen microscope software.

The 972-bp full-length CXCL10 promoter construct, the 376-bp truncated CXCL10 promoter construct (sequence from −279 to +97), and point mutations in the putative AP-1, κB1, κB2, and ISRE recognition elements in the 972-bp full-length CXCL10 promoter were provided by Dr. David Proud.

HBVPs were transfected with the following luciferase reporter plasmids: full-length CXCL10, truncated CXCL10, AP-1 mut, κB1 mut, κB2 mut, or ISRE mut CXCL10 promoter plasmids. For normalization of transfection efficiencies, cells were also cotransfected with pRL-TK plasmid DNA containing the Renilla luciferase gene. 24 h after transfection, cells were treated with or without cocaine for 4 h, followed by assessed the luciferase activity using the Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s instructions. Data represent results obtained from three biological replicates, and each biological replicate comprised three technical replicates.

For mRNA stability assays, 10 µg/ml Act D (Sigma-Aldrich) was added to HBVPs cultured in the presence or absence of cocaine (10 µM) for 30, 60, 90, 120, and 180 min. At selected times following Act D treatment, total RNA was harvested, and CXCL10 mRNA expression was detected by real-time RT-PCR. The fold change in gene expression determined from the RT-PCR assay was then used to calculate the percentage of mRNA remaining following Act D treatment. Data represent results obtained from three biological replicates, and each biological replicate comprised three technical replicates.

Lipid rafts were isolated from confluent HBVPs treated with or without cocaine according to a previous study (Yao et al., 2011b). Briefly, lysates were mixed with 1 ml of 85% (wt/vol) sucrose and overlaid with 2 ml of 42.5% sucrose, 6 ml of 35% sucrose, and 6 ml and 5 ml of 5% sucrose. The gradient was centrifuged for 24 h at 39,000 rpm (SW 41 rotor; Beckman) at 4°C. 12 fractions were harvested from top to bottom (900 µl–1 ml per fraction) and frozen at −80°C until analysis. The concentration of protein in each fraction was analyzed with a bicinchoninic acid protein assay kit (Pierce).

CT-B conjugated to Alexa Fluor 488, which binds the ganglioside GM1 (a lipid raft marker), was used to label the lipid rafts. It has been known that CT-B colocalizes with flotillin, a protein that is known to be located in lipid rafts and ceramide. HBVPs were cultured onto coverslips and transfected with σ-1R–RFP plasmid for 24 h, followed by exposure to cocaine (10 µM) for 30 min at 37°C. CT-B conjugated to Alexa Fluor 488 (1 ng/ml) was added to the HBVPs during the last 10 min of incubation. The coverslips were then washed three times with ice-cold PBS, fixed with ice-cold 4% paraformaldehyde, and mounted with mounting medium (Prolong Gold Anti-fade Reagent; Invitrogen). Fluorescent images were acquired at RT on a Zeiss Observer using a Z1 inverted microscope with a 63×/1.4 oil-immersion objective. Images were processed using the AxioVs 40 Version 4.8.0.0 software (Zeiss). Data represent results obtained from three biological replicates, and each biological replicate comprised three technical replicates.

Treated cells or isolated MIVs were lysed using the Mammalian Cell Lysis kit (Sigma-Aldrich), as described previously (Niu et al., 2014). Equal amounts of protein were electrophoresed in a SDS-polyacrylamide gel under reducing conditions followed by transfer to PVDF membranes. Blots were blocked with 5% BSA in TBS-Tween, and Western blots were probed with antibodies specific for σ-1R (15168-I-AP; Proteintech Group), Src (2108; Cell Signaling), p-PDGFR-β (3161; Cell Signaling), NG2 antibodies (ab129051; Abcam), TBX18 (ab115262; Abcam), αSM (A5228; Sigma-Aldrich), and histone H3 (9715; Cell Signaling) at 1:1,000 dilution; NF-κB (ab16502; Abcam) at 1:2,000 dilution; ganglioside GM1 (bs-2367R; One World Lab) at 1:500 dilution; and p-Src (ab32078; Abcam), PDGFR-β (ab32570; Abcam), Desmin (ab32362; Abcam), and β-actin (A5316; Sigma-Aldrich) at 1:5,000 dilution. Secondary antibodies were alkaline phosphatase conjugated to goat anti-mouse/rabbit IgG, or rabbit anti-goat IgG (1:10,000; Jackson ImmunoResearch Labs). Signals were detected by SuperSignal West Dura Extended Duration or Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific). All experiments had at least four biological replicates, and representative blots are presented in the figures.

To investigate the interaction between c-Src with σ-1R, HBVPs were treated with cocaine (10 µM) for different time periods (5, 15, and 30 min) and then lysed in RIPA buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 5 mM EDTA, pH 8.0, 0.1% SDS, 1.0% NP-40, and 0.5% sodium deoxycholate) containing proteinase and phosphatase inhibitors. Cellular protein (500 µg) was incubated with c-Src antibody (1:100; 2108; Cell Signaling) overnight at 4°C and precipitated with protein A/G beads (Santa Cruz Biotechnology). The mixture was then centrifuged at 12,000 rpm for 1 min, and the cell pellets were rinsed twice with the lysis buffer (1.0% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 9.1 mM Na₂HPO₄, and 1.7 mM NaH₂PO₄) containing proteinase and phosphatase inhibitors. Finally, 30 µl of 2× Western blot loading buffer was added and boiled for 5 min. The protein complexes were detected using σ-1R antibody, while the input protein (without antibody addition) served as a control to show that equal amounts of total protein were used. To confirm the interaction of these two proteins, the σ-1R antibody (2 µg/150 µg total protein; sc-13705; Santa Cruz Biotechnology) was used to pull down the protein, and c-Src antibody was used to detect the signal.

To investigate the phosphorylation of PDGFR-β, HBVPs were treated with cocaine (10 µM) for various times and then lysed using RIPA buffer. For each sample, 150 µg of protein was used for immunoprecipitation. Cell lysates were incubated with 1 µl of anti-4G10 antibody (05-1050; Millipore) overnight at 4°C followed by the same procedure, as described above. Finally, 30 µl of 2× Western blot loading buffer was added and boiled for 5 min. After spinning at 12,000 rpm for 1 min, the supernatants were subjected to Western blot analysis and detected by p-PDGFR-β antibody, while the input protein (without antibody addition) served as a control to show that equal amounts of total protein were used. Data represent results obtained from three biological replicates, and representative blots are presented in the figures.

Boyden chambers (Corning Costar) were used to determine the transmigration of monocytes in vitro. Briefly, primary HBMECs were seeded (4 × 10⁴ cells/well) onto 6.5-mm polyester Transwell inserts (3-µm pore size) and grown for 5 d to achieve confluence. Human monocytes were fluorescently labeled with 10 µM cell tracker green for 10 min at 37°C. Labeled cells (10⁶ cells/ml) were added to the upper compartment of Transwell inserts in serum-free medium, while PCM was added to the basal side of the chamber. The Transwell plates were incubated for 18 h at 37°C, followed by quantification of monocyte transmigration by measuring the number of cells migrating across the insert with a Synergy Mx fluorescence plate reader (BioTek Instruments). Data represent results obtained from three biological replicates, and each biological replicate comprised two or three technical replicates.

Primary HBMECs were seeded (4 × 10⁴ cells/well) onto 6.5-mm polyester Transwell inserts (3-µm pore size) and grown for 5 d to achieve confluence. HBMEC monolayers were exposed to varying concentrations (50%–100%) of PCM-Con or PCM-cocaine for 18 h. FITC Dextran-4 (100 ng/ml) was then added to the upper chamber of the inserts and incubated for an additional 2 h. Aliquots (100 µl) were then collected from the lower chamber for measurement of fluorescence at 480 and 530 nm wavelengths for excitation and emission, respectively (BioTek Instruments). Permeability changes were expressed as fluorescence intensity of FITC-Dextran transported across the BBB into the lower chamber compared with PCM-Con. Data represent results obtained from three biological replicates, and each biological replicate comprised two or three technical replicates.

Brain MIVs were isolated as described previously (Niu et al., 2014). Briefly, the brains were removed and immediately immersed in ice-cold isolation buffer A (103 mM NaCl, 4.7 mM KCl, 2.5 mM CaCl₂, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄, and 15 mM Hepes, pH 7.4). The choroid plexus, meninges, cerebellum, and brain stem were removed followed by homogenization of the brain in 2.5 ml of isolation buffer B (25 mM NaHCO₃, 10 mM glucose, 1 mM Na⁺ pyruvate, and 10 g/liter dextran, pH 7.4) with complete protease inhibitors. 6 ml of Dextran (26%) was then added to the homogenates followed by centrifugation at 5,800 g for 20 min. Cell pellets were resuspended in isolation buffer B and filtered through a 70-µm mesh filter (Becton Dickinson). Filtered homogenates were repelleted by centrifugation, and part of it was used for staining by smearing on glass slides.

Brain MIVs smeared on glass slides and HBVP cultured on coverslips were fixed with 4% formaldehyde in PBS for 20 min at RT. The slides or coverslips were washed three times with PBS, permeabilized with 0.3% Triton X-100 for 30 min, rewashed three times, and blocked in 10% goat serum in PBS for 2 h at RT. The following antibodies were used for immunostaining: p-Src (1:100; ab32078; Abcam), σ-1R (1:100; sc-137075; Santa Cruz Biotechnology), NF-κB (1:100; ab16502; Abcam), CXCL10 (1:50; ab8098; Abcam), NG2 (1:200; ab129051; Abcam), PDGFR-β (1:100; ab32570; Abcam), Desmin (1:50; ab32362; Abcam), TBX18 (1:100; ab115262; Abcam), and CD31 (1:500; NB600-1475; Novus). The slides or coverslips were washed with PBS and incubated with Alexa Fluor 594–conjugated anti-mouse or anti-rabbit, Alexa Fluor 488–conjugated anti-rabbit or anti-mouse, Alexa Fluor 488–conjugated anti-mouse, or Alexa Fluor 594–conjugated anti-rabbit immunoglobulin G (Invitrogen) for 1 h at RT. After a final washing with PBS, the slides or coverslips were mounted with mounting medium (Prolong Gold Antifade Reagent; Invitrogen). Fluorescent images were acquired at RT on a Zeiss Observer, using a Z1 inverted microscope with a 40×/1.3 or 63×/1.4 oil-immersion objective. Images were processed with the AxioVs 40 Version 4.8.0.0 software (Zeiss). Photographs were acquired with an AxioCam MRm digital camera and were analyzed with ImageJ software.

Formalin-fixed, paraffin-embedded sections (5 µm) of frontal cortex from normal individuals or cocaine abusers were obtained from the Douglas-Bell Canada Brain Bank (see Table S1 for clinical data) and stained with antibodies specific for CD68 (1:100; ab955; Abcam), TMEM119 (1:100; ab185333; Abcam), and Desmin (10 µg/ml; AF3844; R&D Systems) overnight at 4°C. Frozen frontal cortex tissues with the same identification number were also obtained from the Douglas-Bell Canada Brain Bank and used for MIV isolations. Isolated MIVs were costained with anti-CXCL10 (1:50; ab8098; Abcam) and antibodies specific for pericyte markers (PDGFR-β [1:100; ab32570; Abcam], NG2 [1:200; ab129051; Abcam], Desmin [1:50; ab32362; Abcam], or TBX18 [1:100; ab115262; Abcam]) and CD31 (1:500; NB600-1475; Novus) overnight at 4°C. Brain sections or isolated MIVs were washed three times in PBS followed by incubation in Alexa Fluor 594–conjugated anti-mouse or anti-rabbit, Alexa Fluor 488–conjugated anti-rabbit or anti-mouse, and Alexa Fluor 647–conjugated anti-goat or anti-rat immunoglobulin G (Invitrogen) for 2 h at RT. After a final washing with PBS, sections were mounted with the mounting medium. Fluorescent images were acquired at RT on a Zeiss Observer using a Z1 inverted microscope with a 40×/1.3 or 63×/1.4 oil-immersion objective. Z-stacks were generated from images taken at 0.50–0.8-µm intervals. Images were processed using the AxioVs 40 Version 4.8.0.0 software (Zeiss). Photographs were acquired with an AxioCam MRm digital camera and were analyzed with ImageJ software. CD68⁺/TMEM119⁻ cells proximal to Desmin⁺ MIVs were counted. For MIVs, only those with a diameter <6 µm were used for quantification (Halliday et al., 2016).

8-wk-old C57BL/6N mice (male, n = 4) were microinjected with either saline (200 ng/µl) or CXCL10 (4 µl) into the brain using the microinjection parameters (coordinates +1.34 mm behind the bregma, +1.25 mm lateral from the sagittal midline at the depth of −4.0 mm to skull surface; flow rate: 0.5 µl/min). After 1 h, animals were injected with BMMs isolated from CX3CR1-GFP mice at a concentration of 10⁷ cells/100 µl through the tail vein. 24 h after cell infusion, the animals were euthanized and subjected to transcranial perfusion with saline to remove CX3CR1-GFP⁺ monocytes from tissue blood vessels. Brain tissues were removed and frozen at −80°C until cryosection. Sections were then stained with anti-GFP (1:500; ab13970; Abcam) and anti-F4/80 (1:100; ab6640; Abcam) antibodies and subjected to immunostaining as described above. Using four animals per group, different regions of the mouse brains were then quantified for double-positive cells in the cortex, hippocampus, thalamus, and striatum regions. For MIVs, only those with a diameter <6 µm were used for quantification (Halliday et al., 2016).

Assay of monocyte transmigration into the brain was performed in C57BL/6 mice. Animals were divided into three groups (n = 4): (1) saline, (2) cocaine, and (3) cocaine plus AMG487 (1 µM). Cocaine was injected once daily at a dose of 20 mg/kg intraperitoneally for 7 d. On the eighth day, animals were injected with BMMs isolated from CX3CR1-GFP mice with or without pretreatment of AMG487 at a concentration of 10⁷ cells/100 µl through the tail vein. 24 h after cell infusion, the animals were euthanized and subjected to transcranial perfusion with saline to remove CX3CR1-GFP⁺ monocytes from tissue blood vessels. Brain tissues were removed and frozen at −80°C until cryosection. Brain sections were then stained with GFP (1:500; ab13970; Abcam), F4/80 (1:100; ab6640; Abcam), and PDGFR-β (1:100; ab32570; Abcam), following the immunostaining protocol described above. Labeled cells were then quantified in the different regions including cortex, hippocampus, thalamus, and striatum. Only vessels with a diameter <6 µm were considered as MIVs (Halliday et al., 2016).

C57BL/6 mice were divided into two groups (n = 4): (1) saline and (2) cocaine. Cocaine was injected once daily at a dose of 20 mg/kg intraperitoneally for 7 d, and then animals were euthanized and subjected to transcranial perfusion with saline to remove blood from tissue blood vessels. Brain tissues were removed and frozen at −80°C until cryosection. Brain sections were stained with anti-fibrinogen (1:50; ab34269; Abcam) and/or antibodies specific for pericyte markers (PDGFR-β [1:100; ab32570; Abcam] or Desmin [1:100; ab8976; Abcam] and CD31 [1:500; NB600-1475; Novus]) overnight at 4°C. Brain sections were washed three times in PBS followed by incubation with Alexa Fluor 594–conjugated anti-mouse or anti-rabbit, Alexa Fluor 488–conjugated anti-rabbit or anti-rat, or Alexa Fluor 647–conjugated anti-rat immunoglobulin G (Invitrogen) for 2 h at RT. After a final washing with PBS, sections were mounted with the mounting medium. Fluorescent images were acquired at RT on a Zeiss Observer by a Z1 inverted microscope with a 40×/1.3 or 20×/0.8 oil-immersion objective. Z-stacks were generated from images taken at 0.50–0.8-µm intervals. Images were processed using the AxioVs 40 Version 4.8.0.0 software (Zeiss). Photographs were acquired with an AxioCam MRm digital camera and were analyzed with ImageJ software. For MIVs, only those with a diameter <6 µm were used for quantification for both pericyte coverage and fibrinogen leakage, which were described previously (Mehla et al., 2012; Nikolakopoulou et al., 2017). Briefly, for pericyte coverage, the percentage of Desmin- or PDGFR-β–positive surface area covering CD31⁺ capillary surface area was quantified. For fibrinogen leakage, the perivascular fibrinogen signals on the abluminal side of MIVs were analyzed.

Statistical analysis was performed using a two-tailed Student’s t test for comparison of the two groups and one-way ANOVA with a Bonferroni’s post hoc test for multiple comparisons. For comparison between the two groups, an F test was used to determine the equality of variances between groups. For comparison among multiple groups, a Brown-Forsythe test was used to determine the equality of variances among groups. Results were judged statistically significant if P < 0.05 by ANOVA for both Student’s t test and one-way ANOVA test. Data distribution was assumed to be normal, but this was not formally tested.

Table S1 shows clinical data for human brain tissue samples. Fig. S1 shows immunostaining of different cell markers in HBVPs and MTS assay of cell viability of HBVPs. Fig. S2 shows CXCL10 mRNA expression levels in HEK293 cells and representative Western blots of σ-1R in 10 lipid raft isolation fractions. Fig. S3 shows pretreatment of HBVPs with the IKK-2/NF-κB inhibitor SC514 failed to abrogate cocaine-mediated lipid raft translocation of σ-1R, Src phosphorylation, and PDGFR-β phosphorylation. Fig. S4 shows the effect of cocaine on permeability of BBB in vitro and in vivo. Fig. S5 shows CXCL10 expression levels in pericytes of MIVs isolated from brains of mice administrated either saline or cocaine. Fig. S6 shows pericyte coverage in MIVs from brains of mice administrated either saline or cocaine. Fig. S7 shows CX3CR1-GFP⁺ BMMs transmigration in the brain of mice administrated CXCL10.

We thank the Douglas-Bell Canada Brain Bank for providing the frozen and formalin-fixed brain samples of cocaine abusers; Dr. David Proud for providing all the luciferase reporter plasmids—full-length CXCL10, truncated CXCL10, AP-1 mut, κB1 mut, κB2 mut, or ISRE mut CXCL10 promoter plasmids; Dr. Naoto Yamaguchi (Chiba University, Chiba, Japan) for providing the GFP-tagged c-Src plasmid; and Dr. Tsung Ping Su (National Institute on Drug Abuse, Bethesda, MD) for providing us with the σ-1R–RFP plasmid.

This work was supported by National Institutes of Health, National Institute on Drug Abuse grants DA035203, DA040397, DA041751, DA044586 (to S. Buch); DA043138 (to S. Buch and G. Hu); and DA046831 and DA042704 (to G. Hu); and National Institute of Mental Health grants MH106425 (to S. Buch) and MH112848 (to S. Buch and G. Hu), as well as the Nebraska Center for Substance Abuse Research. The project was supported by National Institutes of Health, National Institute of Mental Health grant 2P30MH062261. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

The authors declare no competing financial interests.

Author contributions: F. Niu, G. Hu, and S. Buch designed the research; F. Niu, K. Liao, S. Sil, M. Guo, and L. Yang performed research; and F. Niu, G. Hu, K. Liao, S. Callen, and S. Buch wrote the manuscript.