Activation of Peroxisome Proliferator-Activated Receptor γ (PPARγ) Suppresses Rho GTPases in Human Brain Microvascular Endothelial Cells and Inhibits Adhesion and Transendothelial Migration of HIV-1 Infected Monocytes¹ Free

Adhesion of monocytes to activated brain endothelium plays an important role in the transmigration across the blood brain barrier (BBB)³ during inflammation of the CNS (1). Under physiological conditions, immune surveillance by lymphocytes crossing BBB is minimal and monocyte infiltration is practically undetectable; however, during the development of autoimmune disorders (i.e., multiple sclerosis) or virus driven diseases (i.e., HIV-1 encephalitis), leukocyte migration through the BBB increases considerably (2, 3, 4, 5). Enhanced leukocyte transendothelial migration is attributed to the activation of both the brain microvascular endothelial cells (BMVEC) and immune cells by proinflammatory cytokines. Cytokine activation promotes up-regulation of integrins in leukocytes and adhesion molecules (e.g., ICAM-1, and VCAM-1) on the BMVEC, which strengthen leukocyte-endothelial interactions leading to increase in transendothelial migration (6, 7). Thus, prevention of the initial leukocyte-endothelial interaction or reduction of the BMVEC controlled leukocyte passage could provide novel therapeutic avenues to lessen neuroinflammation.

Peroxisome proliferator-activated receptors (PPARs) are transcriptional regulators and members of a nuclear receptor family that include the closely related PPARα, PPARβ/δ, and PPARγ, (8). PPARs play important roles in regulation of adipocyte differentiation, insulin sensitivity, anti-inflammatory responses, and antiproliferative effects in certain tumors (9). In recent years, the lipid lowering fibrates, which activate PPARα and the insulin sensitizing thiazolidinediones (TZDs), which activate PPARγ have proven useful as potent anti-inflammatory agents (10). Ligand binding of PPAR promotes the heterodimerization with the retinoid x receptor inducing PPAR transactivation of target genes. In addition, activation of PPARs can also lead to transrepression, a process that has been shown to negatively interfere with key transcription factors (e.g., NF-κB, AP-1, STAT1, and sp1) involved in inflammatory responses. Stimulation of PPARs down-regulated the expression of proinflammatory genes such as IL-1β, IL-6, inducible NO synthase, cyclooxygenase-2, and various chemokines (11). Additionally, PPARs were found to decrease endothelial-leukocyte interactions in models of atherosclerosis (12, 13). Although these inhibitory effects on monocyte adhesion by PPAR activation have been in part attributed to suppression of adhesion molecules in the endothelial cells, the precise mechanism remains elusive. Recently, the effects of PPAR on the Rho pathway was identified in smooth muscle cells (14) in which PPARγ stimulation via the phosphatase SHP-2 inactivated RhoA/Rho kinase signaling. These results could provide an intriguing possibility for the modulation of GTPases by PPAR isotypes, because GTPases are important mediators of leukocyte-endothelial interaction.

Although much is known about key pathways involved in leukocyte extravasation, signaling events in the BMVEC actively participating in this process remain unclear (15). Small GTP-binding proteins such as RhoA and Rac1 regulate in transendothelial leukocyte migration (reviewed in Ref. 16). Rac1 was shown to mediate the clustering of adhesion molecules on the surface of endothelial cells essential for leukocyte docking. Consequently, firm adhesion of leukocytes to the endothelium signals to RhoA in endothelial cells leading to cytoskeletal rearrangement required for junctional modification and leukocyte paracellular transmigration. In inflammation, cytokine stimulation activates of the Rho family of GTPases, but it does not result in junctional disassembly unless an additional stimulus is provided like coculture with monocytes or Ab cross-linking of ICAM-1 and VCAM-1 (simulating leukocyte docking) (17, 18, 19). Therefore, the Rho family of GTPases provides a key molecular switch routing intracellular signaling events needed for transendothelial migration.

One area where PPAR anti-inflammatory effects have remained largely unexplored is in the context of neuroinflammation. Because enhanced leukocyte migration across the BBB is a major feature of neuroinflammation, we tested whether PPARγ activation in cytokine stimulated human BMVEC could control monocyte adhesion and migration. We found that PPARγ activation in BMVEC resulted in significant reduction of monocyte adhesion and migration in our BBB culture model. Analysis of the Rho family GTPases in cytokine stimulated and TZD treated BMVEC resulted in dose dependent inhibition of Rac1 and RhoA that paralleled diminished monocyte adhesion and migration across BMVEC monolayers. Also, we showed that TZD activation of PPARγ in stimulated BMVEC resulted in a significant decrease in the adhesion and transmigration of HIV-1 infected monocytes across brain endothelium. Together these observations provide a novel inhibitory mechanism on monocyte adhesion/migration by PPARγ ligands acting in endothelial cells forming the BBB.

Primary human BMVEC were isolated from tissue derived from surgical removal of epileptogenic cerebral cortex, as described previously (20) and provided by Drs. Marlys Witte and Michael Bernas (University of Arizona, Tucson, AZ). The procedures were approved by the University of Nebraska Medical Center Institutional Review Board. BMVEC were maintained in DMEM/F-12 media containing 10% horse serum, endothelial cell growth supplement (BD Biosciences), heparin (100 mg/ml; Sigma-Aldrich), amphotericin B (2.5 μg/ml; Invitrogen Life Technologies), and gentamicin (50 μg/ml, Invitrogen Life Technologies). For all experiments only BMVEC cultures of low passage (passage 2–5) were used. For barrier formation, the cells were plated confluent in a given culture vessel and maintained in media containing the above components but lacking the endothelial cell growth supplement, which allows BMVEC to display typical morphological and functional barrier characteristics after 3–5 days in culture. BMVEC cultures were routinely evaluated for expression of endothelial cell markers (von Willebrand factor, CD31/PECAM-1, and TJ proteins (ZO-1, occludin, claudin-5)), and the lack of macrophage (CD68 and CD163), astrocyte (glial fribrillary acidic protein) and pericyte (α-myosin) markers. Barrier formation by BMVEC was confirmed by measurement of transendothelial electrical resistance, TEER.

Primary human peripheral blood monocytes were obtained from HIV-1 and hepatitis B seronegative donors by leukopheresis and were purified by countercurrent centrifugal elutriation (21). The monocytes were maintained in DMEM media with 10% heat inactivated pooled human serum, penicillin (100 U/ml), streptomycin (100 U/ml), and l-glutamine (2 mM) and used within 24 h of isolation. Monocytes were infected with HIV-1_ADA at a multiplicity of infection of 0.01–0.1 virus/target cell for 4 h (21). Before infection, the HIV-1 cell-free stocks were treated with DNase I for 30 min at 37°C as described (22). All reagents were prescreened for endotoxin (<10 pg/ml; Associates of Cape Cod) and mycoplasma contamination (Gen-probe II; Gen-probe).

Brain tissues (cortex) from patients without neuro-cognitive impairment or neuropathologic changes were obtained from the National NeuroAIDS Tissue Consortium and our Center for Neurovirology and Neurodegenerative Diseases brain bank. The clinical histories of brain tissue donors were described in Ref. 19 . The isolation procedure of microvessels was performed as described in Ref. 23 . Protein was extracted from each sample for Western blot. For immunofluorescent microscopy, 50 μl of microvessel suspension was spread on poly-l-lysine- or fibronectin-coated glass coverslips, air-dried, fixed for 20 min with a 4% paraformaldehyde (pH 7.4), rinsed with PBS, air-dried, and used for immufluorescent staining with Abs to tight junction protein, claudin-5, von Willebrand Factor; CD163 (macrophage marker) was used to confirm the vascular identity of the isolated microvessels as described (24).

BMVEC were plated on 96-well plates coated with rat-tail collagen type I (BD Biosciences) at a density of 2.5 × 10⁴ cells/well. After formation of confluent monolayers, BMVEC were treated with PPARγ agonists or antagonists for 30 min before simultaneous incubation with recombinant human TNF-α (20 ng/ml; R&D Systems) for 4 h before the assay. The endothelial cells were rinsed of all treatments and fluorescently labeled monocytes (2.5 × 10⁵ cells/well loaded with calcein-AM (Invitrogen Life Technologies) at 5 μM/1 × 10⁶ cells for 45 min) were added to the endothelial monolayers and cocultured for 15 min at 37°C. After adhesion, the monolayers were washed three times with PBS and relative fluorescence was acquired on a fluorescence plate reader, Spectramax M5 (Molecular Devices). The data were calculated based on the standard curve derived from fluorescent intensity of known amounts of labeled monocytes. Results are represented as the fold difference, which is the number of monocytes that attached to endothelial cells with experimental condition over the number of monocytes that attached to endothelial cells without cytokine stimulation or drug treatment.

The transendothelial migration assay was a fluorescence-based assay. In brief, BMVEC were plated on rat-tail collagen type I coated FluoroBlok tinted tissue culture inserts (with 3-μm pores; BD Biosciences) at a density of 2.5 × 10⁴ cells/insert. Because monolayers were not visible on these inserts, manual readings of TEER were taken with a voltmeter (EVOM; World Precision Instruments) to confirm monolayer formation as described (19). Thereafter, endothelial cells were treated with PPAR acting compounds for 30 min and then coincubated for an additional 4 h with TNF-α (20 ng/ml). The inserts were washed of all treatments, and relevant β-chemokine, recombinant human monocyte chemotactive protein (MCP) −1 (CCL2/MCP-1, 30 ng/ml; R&D Systems) was introduced into the lower chamber to create a chemokine gradient present in neuroinflammatory disorders (5, 25). Labeled monocytes (2.5 × 10⁵ cells/insert loaded with calcein-AM at 5 μM/1 × 10⁶ cells for 45 min) were placed on the upper chamber, and migration was allowed to continue up to 2 h at 37°C. Migration was monitored by taking the relative fluorescence with a fluorescence plate reader (Molecular Devices) of the labeled monocytes found in the lower chamber. The number of migrated monocytes was then derived by plotting against standard curves of the relative fluorescence from known numbers of monocytes. The data are presented as the fold difference, which is the number of migrating monocytes from the indicated experimental condition over the monocytes that migrated across untreated (no drug and no cytokine) endothelial cells without the presence of chemoattractant (left y-axis). Additionally, the data are shown as percent of input (right y-axis), which indicates the percent of cells that migrated from the number of cells inputted.

Transfection of primary BMVEC was performed by nucleofection using the Nucleofector device (Amaxa System) according to protocol suggested by manufacturer. In brief, for each transfection reaction, 5.0 × 10⁵ trypsinized endothelial cells were resuspended in 100 μl of nucleofection reagent along with 1 μg of vector DNA. Each reaction was transferred to an electroporation cuvette and placed in the nucleofector device for electroporation. The electroporated cells were plated in the configuration necessary for migration assay, adhesion assay, or Western blot. BMVEC were transfected with DNA constructs containing coding sequences for the following proteins: wild type RhoA (WT), constitutively active RhoA (G14V), dominant negative RhoA (T19N), WT Rac1, constitutively active Rac1 (G12V), and dominant negative Rac1 (T17N). Constructs were obtained from the University of Missouri-Rolla cDNA Resource Center (www.cdna.org). Transfection efficiency with the Amaxa system typically reached ∼80%, as determined by flow cytometry.

The endothelial cells were also transfected with small hairpin RNA (shRNA) expressing vectors directed to PPARγ (GenBank accession number NM_138711). A scrambled sequence control and an empty vector control were also transfected. The silencing oligonucleotides were composed of the following: bases 1–21 antisense, 22–32 loop sequence, and 33–54 sense sequence. PPARγ shRNA sequence 1 (shPPARγ-1): 5′-TATTTGAGGAGAG TTACTTGGTTGATATCCGCCAAGTAACTCTCCTCAAATA-3′, PPAR shRNA sequence 2 (shPPARγ-2) 5′-TAATGTGGAGTAGAAATGCTGTTGATATCCGCAGCATTTCTACTCCACATTA-3′, and PPARγ shRNA sequence 3 (shPPARγ-3) 5′-TGTTCCGTGACAATCTGTCTGTTGATATCCGCAGACAGATTGTCACGGAACA-3′. All oligonucleotides were cloned into pRNAT-H1.1/Adeno vectors (Invitrogen Life Technologies) and were provided by GenScript.

Precipitations of the active form of Rho (RhoA-GTP) were performed from endothelial cell lysates by affinity purification using the EZ-Detect Rho activation kit from Pierce. In brief, each lysate (400 μg of total protein) was incubated with GST-rhotekin-RBD protein (400 μg) and introduced to an immobilized glutathione disc resin and incubated for 1 h at 4°C with gentle agitation. After incubation, the resin was washed 3 times with 1× Mg²⁺ lysis buffer followed by centrifugation at 14,000 × g for 30 s. The discs were then resuspended in Laemmli sample buffer containing 2-ME and boiled for 5 min at 95°C. Relevant controls such as guanosine 5′-O-(3-thio) triphosphate (GTPγS, for positive) and guanine diphosphate (GDP, for negative) were performed with untreated lysates and affinity precipitated as above.

Assays for active Rac1 (Rac-GTP) or Cdc42 (Cdc42-GTP) from cell lysates were also performed by affinity purification using the Rac1 or Cdc42 activation assays from Cytoskeleton. In brief, 2 mg of total protein from endothelial cellular lysates were incubated with 20 μg of PAK-PBD (p21 binding domain of p21 activated kinase 1) conjugated beads for 1 h at 4°C. After incubation, the PAK-PBD beads, which bind specifically to the active form of Rac1 or Cdc42, were washed twice with 1× wash buffer (25 mM Tris (pH 7.5), 30 mM MgCl₂, and 40 mM NaCl) by centrifugation at 5000 × g for 3 min at 4°C. The rinsed beads were then resuspended in 10 μl of Laemmli buffer and analyzed by Western blot using specific Abs to distinguish between Rac1 and Cdc42.

Total protein lysates (10 μg) or precipitated proteins (quantities indicated above) were resolved by SDS-PAGE on 12% gels (Pierce), then transferred to nitrocellulose membranes and incubated with Abs against RhoA (1/500, Pierce), Rac1 (1/250, Cytoskeleton), Cdc42 (1/250 Cytoskeleton), PPARγ, PPARβδ, PPARα (1/500, Abcam), and hemagglutinin epitope (1/1000, Covance). For detection of tight junction proteins, membranous and cytoplasmic extractions were performed from lysed BMVEC according to the manufacturer’s protocol included with the ProteoExtract subcellular proteome extraction kit (Calbiochem). Western blots were then performed using the following Abs: occludin (1/500, US Biological), claudin-5 (1/500, US Biological), and ZO-1 (1/500, US Biological). Bound Abs were detected with HRP-conjugated secondary Abs (1/5000, Pierce) and exposed to Supersignal West Pico chemiluminescent substrate (Pierce). Signal visualization was obtained using the gel documentation system, G:Box Chemi HR16 (Syngene).

BMVEC cells were treated as outlined in the figure legends, and analysis of PPARγ and PPARα DNA binding was performed as instructed by the manufacturer using the transcription factor ELISA kits for PPARα and PPARγ from Panomics. For GTPase ELISAs, BMVEC were treated as shown in the Fig. 1 legend and quantitation of RhoA and Rac1 GTPase activation was assessed following the manufacturer’s recommendations using the G-LISA small G-protein activation assays from Cytoskeleton.

BMVEC were treated with TNF-α (20 ng/ml, 4 h) and/or rosiglitazone. Cells were lifted with 0.5 mM EDTA at 4°C, washed with flow cytometry buffer (eBioscience) then incubated with allophycocyanin-conjugated VCAM-1 Ab (BD Biosciences) or PE-conjugated ICAM-1 Ab (BD Biosciences) for 45 min at 4°C. Cells were washed, fixed with 2% paraformaldehyde, acquired on a FACSCalibur (BD Biosciences), and analyzed using CellQuest Pro software (BD Immunocytometry System).

ICAM-1 Ab crosslinking was conducted as previously described (26). In brief, BMVEC were seeded at 1.3 × 10⁵ in six-well tissue culture plates and maintained under normal growth conditions for 5 days. Thereafter, the cells were incubated with TNF-α (20 ng/ml) for 4 h in the absence of growth factors but with serum, and then washed with 1× PBS. Treatment with PPARγ was performed for 30 min with or without TNF-α for 4 h as indicated. ICAM-1 crosslinking was conducted using mouse anti-human ICAM-1/CD54 Ab (R&D Systems) which was added to the endothelial cells for 10 min. Unbound Ab was removed by rinsing with PBS and then cells were incubated for 15 min with rabbit anti-mouse Abs (Santa Cruz Biotechnology). Cells were rinsed with PBS and then lysed for subsequent GTPase analysis (see above).

TEER was measured in real time using the ECIS system model 1600R, (Applied Biophysics). Primary BMVEC were grown to confluence on collagen type I coated ECIS cultureware arrays (8W10E). Measurements were recorded with the following settings: current (1-volt), frequency (400Hz), at 5 min intervals. Treatments were performed as indicated in the figure legends and data were represented as normalized resistance, which is the resistance measured post treatment over the resistance acquired before treatment introduction (typically ∼2600 Ω/cm²).

The experiments were independently performed multiple times (at least three times for all presented data) allowing statistical analyses. In each individual experiment, every condition was evaluated in three replicates. The data collected were analyzed using Prism v4.0 (GraphPad). The differences between the means were evaluated by Student’s t test, and statistical significance was considered at p values <0.01. All results were expressed as mean ± SEM.

Because PPAR isotype distribution varies depending on the cell or tissue type and because no prior isotype expression has been documented in human brain endothelial cells, we sought to identify the expression profile in isolated human brain microvessels as well as in primary cultures of human BMVEC. Additionally, given that it is possible for primary cells to change a particular gene expression profile due to passage and culturing conditions, it was important to rule out whether there was a difference in the PPAR isotype expression between culture cells and intact microvessels. To this end, Western blot analysis of lysates from isolated microvessels and primary cultures were probed for each of the known PPAR isotypes. Fig. 1,A shows the expression of PPARα, PPARβδ, and PPARγ in both sets of lysates. As seen in the Fig. 1, all three isotypes were expressed in brain microvessels and were also maintained in BMVEC cultures. Next, we examined PPARα and PPARγ activation by oligo binding ELISA using two PPARα agonists (fibrates) and three PPARγ agonists (TZDs). As shown in Fig. 1,B, PPARα oligo binding increased in a dose dependent manner with both fenofibrate and clofibrate, with fenofibrate inducing better binding (EC₅₀ = 5.6 μM) compared with clofibrate (EC₅₀ = 16.5 μM). Analysis of PPARγ activation by TZDs, rosiglitazone, pioglitazone, and ciglitazone resulted in rosiglitazone as a better activator (Rosi EC₅₀ = 2.0 μM, Pio EC₅₀ = 2.6 μM and Cig EC₅₀ = 2.3 μM; Fig. 1 C). Therefore, we used fenofibrate for PPARα activation and rosiglitazone for PPARγ activation in subsequent analyses.

Because monocyte adhesion to brain endothelial cells is the initial step in migration across the BBB during neuroinflamation, and both PPARα and PPARγ can inhibit inflammatory responses in the vasculature, we treated BMVEC with PPARα and PPARγ ligands and tested monocyte adhesion. First, BMVEC were treated with PPAR agonist for 30 min and then coincubated with TNF-α (20 ng/ml) for 4 h. Treatments were removed and adhesion assay was performed as indicated in the Materials and Methods. Monocyte adhesion increased 15-fold when endothelial cells were stimulated with cytokine as shown in Fig. 1,D. Pretreatment of BMVEC with various concentrations of PPARγ agonist led to a dose-dependent decrease in monocyte adhesion in both unstimulated and stimulated endothelial cells (Fig. 1 D). At maximum, the reduction in monocyte adhesion seen in TNF-α activated BMVEC after treatment with rosiglitazone resulted in an 8-fold decrease in adhesion (p < 0.001) as compared with control. In addition to rosiglitazone, pioglitazone also produced a decrease in monocyte adhesion, resulting in a 4-fold decrease at the highest concentration tested. Surprisingly, no effect on monocyte adhesion was seen with the PPARα agonist, fenofibrate. This observation is in contrast to previous analysis reporting that PPARα activation could decrease the attachment of monocytes to endothelial cells (27). However, these experiments were performed in non-brain-derived endothelial cells. Thus, our data demonstrate that in brain endothelial cells, PPARγ activation can decrease monocyte adhesion but not when PPARα is stimulated.

Next, we studied monocyte migration to determine whether PPARγ stimulation in BMVEC also could have an effect on monocyte transendothelial migration. In these experiments, we used CCL2/MCP-1 as a relevant chemokine produced during neuroinflammatory conditions, such as HIV-1 encephalitis (25). Fig. 1,E shows the results of monocyte migrations across BMVEC monolayers; the results are represented as normalized fold values to monocytes migrating across BMVEC without the presence of CCL2/MCP-1. Basal migration across resting endothelial cells without chemoattractact is indicated as 1-fold (∼6.5% from input), stimulation of endothelial cells with TNF-α and without CCL2/MCP-1 results in 1.3-fold increase (∼8.5% from input), endothelial cells with CCL-2/MCP-1 in lower chamber gives a 2-fold (∼12.6% from input) increase in migration, and BMVEC TNF-α stimulation with addition of CCL2/MCP-1 results in 4.3-fold (∼25.5% from input) increase monocyte migration. Pretreatment of BMVEC with increasing concentrations of rosiglitazone (two shown) in the above treatment conditions produced significant decreases in transendothelial migration. Particularly when CCL2 and TNF-α were present, rosiglitazone produced a 21% decrease for 5 μM concentration and 35% at 50 μM concentration (p < 0.01). Similar inhibitory results on migration were also obtained with other PPARγ agonist, albeit at higher concentration (data not shown). Also shown in Fig. 1 E, no significant change in migration of monocytes was observed with the PPARα agonist, fenofibrate.

To eliminate the possibility that the effects of rosiglitazone were not mediated through other cellular targets (28) or other PPAR family members, BMVEC were transfected with several PPARγ-specific shRNA. Three different shRNA sequences were evaluated for knocking down endogenous PPARγ. As shown in Fig. 2,A, transfected shPPARγ-2 lowered endogenous PPARγ by 85%, as compared with ∼60% for both shPPARγ-1 and shPPARγ-3. Because transfection with shPPARγ-2 showed the most prominent effect on PPARγ protein reduction, a control scramble sequence (shPPARγ-2-SC) of shPPARγ-2 was generated and transfected, which resulted in no observable PPARγ knock down (Fig. 2,A). Next, we performed adhesion assays using transfected shPPARγ-2 and shPPARγ-2-SC in BMVEC, which had been treated with rosiglitazone and stimulated with TNF-α (Fig. 2,B). The effect of rosiglitazone on monocyte adhesion was abolished in the shPPARγ-2 transfected BMVEC, but not in PPARγ-2-SC endothelial cells. These results indicated that endogenous PPARγ was indeed the primary target for the effect of rosiglitazone on monocyte adhesion because shRNA-mediated knockdown of PPARγ brought monocyte adhesion to control levels in the presence of the drug. The PPARγ-dependent mechanism was also observed in migration assays, where PPARγ knockdown resulted in the elimination of most of the migratory inhibition rendered by rosiglitazone (Fig. 2 C). To ensure that the effects exerted by PPARγ knock-down were not the result of compromised barrier functionality, separate experiments showed that no difference in “tightness” of the barrier was observed as evaluated by TEER (data not shown).

PPARγ ligands have been shown to suppress the transcriptional activity of factors mediating the up-regulation of adhesion molecules (i.e., NF-κB and STATs). Thus, we investigated whether this mechanism could be responsible for the observed inhibition on monocyte adhesion. Both ICAM-1/CD54 and VCAM-1/CD106 are critical for firm monocyte attachment to the endothelium. Representative histograms of BMVEC surface expression of ICAM-1 and VCAM-1 by flow cytometry are shown in Fig. 3 A. Unexpectedly, the up-regulation in surface expression of both ICAM-1 and VCAM-1 seen after 4 h stimulation with TNF-α was not decreased with prior pretreatment with 50 μM rosiglitazone. In addition, surface expression of ICAM-1 and VCAM-1 was also unchanged in resting cells. Of note, using natural PPAR ligands in human aortic endothelial cells, others have also shown the lack of regulation in adhesion molecule expression by PPARγ (29).

The paracellular (between cells) migration of monocytes across the BBB has been shown to be dependent in the orchestrated events occurring in endothelial cells that lead to tight junction (TJ) disassembly and permit migration. Thus, it is possible that PPARγ could be increasing barrier tightness via the TJ and therefore, limiting monocyte migration. To test this possibility, we analyzed changes in the TEER, and also on the expression and localization of TJ proteins after PPARγ treatment. Endothelial monolayers were cultured on coated electrode chamber arrays and incubated with the indicated compounds, and in some cases coincubated with TNF-α. As Fig. 3,B shows, the TEER measurements of treated BMVEC taken in real time for 6 h did not produce fluctuations in TEER of either rosiglitazone alone or when present with TNF-α. Also shown in Fig. 3 B, the immediate and sustained drop in TEER measurement was produced by the NO donor, linsidomine, used as control for inducing barrier leakiness. Of note, application of TNF-α at 20 ng/ml concentrations, which was the concentration used in the functional assays reported here, resulted in only a slight and a transient decrease in TEER.

To further confirm the results of the TEER analysis, the expression and localization of TJ proteins, ZO-1, occludin, and claudin-5 were investigated. Cytosolic and membranous fractions from treated BMVEC, as indicated in Fig. 3,C, were collected and probed for the TJ proteins shown. As presented in Fig. 3 C, we found no differences in the expression or intracellular distribution of any of the three TJ proteins. Thus, effects on monocyte migration cannot be explained by changes in TJ protein content or distribution in the BMVEC after PPARγ activation.

Central to monocyte-endothelial interactions is the modulation of cytoskeletal components and surface adhesion molecules by the Rho family of small GTPases (16, 30). During neuroinflammation, brain endothelial cells up-regulate adhesion molecules such as ICAM-1 and VCAM-1 leading to increased leukocyte-endothelial recruitment and trafficking (31). Proinflammatory cytokines including TNF-α, IL-1β, and IL-6 induce the activation of the GTPases in endothelial cells (32) leading to receptor clustering (Rac1), important for leukocyte adhesion and cytoskeletal and TJ rearrangement (RhoA), important in transendothelial migration (15, 19).

To evaluate whether PPARγ stimulation modulated the activation state of members of the Rho family of GTPases, we performed affinity purification of the active forms of RhoA, Rac1, and Cdc42 in TNF-α treated BMVEC. After treatment, cells were lysed and subjected to affinity purification-based pull-down of the GTP bound active forms of RhoA, Rac1, and Cdc42. The pull-down assays used the binding domain of the effector protein that targets the particular GTPase. In agreement with previous reports, TNF-α stimulation increased the level of both GTP-bound RhoA and Rac1 in BMVEC (Fig. 4, A and B). PPARγ agonist, rosiglitazone, significantly decreased binding of active RhoA and Rac1 in a dose-dependent manner (Fig. 4, A and B). It is interesting to note that PPARγ stimulation appeared to inhibit more effectively Rac1 (∼60% at 5 μM) than RhoA (∼20% at 5 μM). Although TNF-α induced Cdc42 activation/GTP binding, surprisingly rosiglitazone did not reduce Cdc42-GTP binding, suggesting that PPARγ could modulate the activation status of only certain GTPases. PPARγ stimulation of the GTPases did not change protein expression because total protein expression of GTPases was not affected. Thus, these results suggest that PPARγ via GTPase regulation could be the underlying mechanism preventing monocyte adhesion and migration across activated endothelium after PPARγ stimulation.

Although GTPases are activated by proinflammatory cytokines, as seen in Fig. 4, A and B, additional GTPase activation occurs after leukocyte-endothelial engagement. Ab crosslinking of adhesion molecules was used to emulate this interaction (26) leading to cytoskeletal rearrangement needed for leukocyte docking and extravasation. With Ab crosslinking to ICAM-1, we proceeded to investigate whether PPARγ mediated inhibition of RhoA and Rac1 could also be seen if further GTPase activation was induced. Using quantitative GTPase ELISA for Rac-1 and RhoA, we not only observed the induction in GTPase activation after addition of TNF-α (20 ng/ml for 4 h), but also found the further activation of the two GTPases triggered by ICAM-1 Ab crosslinking (Fig. 4, D and E). BMVEC exposed to rosiglitazone along with TNF-α stimulation and ICAM-1 crosslinking showed inhibition of Rac1 and RhoA. Lastly, application of monocytes to BMVEC monolayers resulted in Rac1 and RhoA activation (as seen with Ab crosslinking) (Fig. 4, G and F); rosiglitazone blocked these changes.

It is important to note that rosiglitazone inhibition of Rac1 and RhoA in these analyses, like the pull-down assays (Fig. 4, A and B), also appeared to be more efficient for Rac1. These effects on Rac1 may explain rosiglitazone effects in BMVEC inhibition of monocyte adhesion, because maximal inhibition is seen with 5 μM rosiglitazone similarly to that observed with Rac1 inhibition (Fig. 4,D). In contrast, RhoA inhibition by rosiglitazone in the endothelial cells appeared less sensitive (Fig. 4,E) at low concentrations, possibly explaining the effect on monocyte migration where the maximal inhibitory effect is seen with 50 μM (Fig. 1,E). Lastly, analysis of GTPase activity was observed to be similar to those with Ab crosslinking when actual monocytes were used (Fig. 4, F and G). Taken together, the analysis on the GTPases provides support for the notion that in BMVEC PPARγ negatively modulates Rac1 and RhoA, thus affecting monocyte adhesion and migration.

To further elucidate whether PPARγ inhibitory action on Rac1 and RhoA could influence monocyte adhesion, we examined monocyte adhesion to BMVEC expressing RhoA and Rac1 mutants. BMVEC were transfected with low concentrations (as outlined in the Materials and Methods) of DNA constructs expressing wild type (WT), constitutively active (CA) and dominant negative (DN) forms of Rac1 and RhoA. Inserts panels in Fig. 5, A and B show the expression of the various mutants: the first row denotes the detection of the 3xHA peptide tag fused to the mutants, the second row indicates the GTPase expression (endogenous plus mutant), and the third row shows the internal control (α-actin). After transfection, the cells were allowed to form monolayers and then treated as indicated (Fig. 5, A and B). Higher concentrations of rosiglitazone were omitted because maximal inhibition on adhesion was reached at 5 μM concentrations (as shown on Fig. 1). As shown in Fig. 5,A, the expression of Rac-CA resulted in a 4-fold increase in monocyte adhesion even in unstimulated BMVEC as compared with controls (transfected with Rac-WT). TNF-α stimulation alone in endothelial cells produced increased monocyte adhesion in Rac-WT as previously observed, and yet 20% higher in Rac-CA expressing cells. Rosiglitazone application led to a 23 and 35% decrease in monocyte adhesion for 2 μM and 5 μM, respectively, in Rac-WT; the Rac-CA transfectants completely counteracted the actions of rosiglitazone in activated BMVEC pointing to Rac1 as a critical mediator of monocyte adhesion in BMVEC. Interestingly, expression of Rac-DN led to a decrease in baseline adhesion to the cytokine-stimulated monolayers (∼36% as compared with Rac-WT), similar to the inhibition seen in rosiglitazone treated cells at the maximal effective dose (5 μM, Fig. 5 A).

In stark contrast to the functional analysis performed with the Rac1 mutants, the RhoA mutants had no effect on the response of rosiglitazone-treated endothelial cells, Fig. 5,B. TNF-α stimulation of BMVEC led to ∼10-fold increase in monocyte adhesion regardless of the RhoA mutant the endothelial cells were transfected with. The expression of Rho-CA and Rho-DN did not change the inhibitory effect of PPARγ agonist on monocyte adhesion as compared with control Rho-WT, suggesting that although TNF-α induced RhoA activation and PPARγ mediated its partial inhibition, PPARγ effects on RhoA did not contribute to monocyte adhesion to activated BMVEC. It is important to point out that all inhibitory effects on monocyte adhesion produced by rosiglitazone were reversed by coincubation with high affinity PPARγ antagonist ligand (GW9662), ensuring that over-expression of the mutants did not alter the specific action of the agonist (Fig. 5 A).

Previous studies identified the importance of RhoA in transendothelial migration (15). It was demonstrated that RhoA is a key mediator of TJ “gating”, which promotes leukocyte trafficking across brain endothelial cells (19, 33). We performed experiments to determine whether mutant GTPase expression in endothelial cells could affect PPARγ inhibitory function of monocyte migration. Monocyte migration toward CCL2/MCP-1 was enhanced across TNF-α-activated BMVEC expressing Rho-CA (5.1-fold) and was significantly diminished by Rho-DN (1.8-fold) as compared with cells transfected with the Rho-WT (4.1-fold, p < 0.001, Fig. 6 A). Although treatment with rosiglitazone diminished monocyte migration in BMVEC transfected with control Rho-WT in a dose-dependent fashion (21% with 5 μM or 35% with 50 μM), minimal inhibition (11%) was achieved in rosiglitazone-treated BMVEC expressing Rho-CA regardless of drug concentration. Similarly, no change in monocyte migration was found in BMVEC transfected with Rho-DN after PPARγ agonist treatment as compared with cells transfected with Rho-DN but without rosiglitazone application. Therefore, under these experimental conditions, transfection of cells with Rho-CA mostly counteracted the effects of PPARγ on monocyte migration, suggesting the role of PAPRγ in regulating the role of RhoA in transendothelial migration.

We next analyzed the impact of rosiglitazone on monocyte migration in BMVEC transfected with Rac1 mutants. As expected, CCL2/MCP-1-driven monocyte migration was enhanced across TNF-α stimulated BMVEC expressing Rac-CA (15%) as compared with cells transfected with the Rac-WT (p < 0.01, Fig. 6 B), while Rac-DN expression led to diminution of cell passage (23%). The range of these changes in migration was much less prominent as compared with results obtained with respective RhoA mutants. Monocyte migration across BMVEC expressing Rac-CA was reduced by rosiglitazone only at the higher concentration (50 μM), and the same trend existed for dominant negative Rac1. These data suggested that PPARγ effects on RhoA rather than Rac1-mediated changes of monocyte migration. Moreover, these results indicated that Rac1 inactivation occurred in a seemingly dose-sensitive manner, such that the effect of agonist at low concentration is more profound on Rac1 than for RhoA. These results underscored the dual modulation of PPARγ activation on RhoA and Rac1.

To evaluate whether activators of PPARγ could protect the brain against infiltration of HIV-1 infected monocytes, we infected monocytes with HIV-1_ADA, CCR5-tropic HIV-1 strain, for 4 h and then applied virus-infected monocytes to cytokine stimulated BMVEC pretreated with rosiglitazone. The infected monocytes were analyzed in adhesion and migration assays. As shown in Fig. 7,A, basal adhesion of infected monocytes to untreated BMVEC was increased by 5.5-fold when compared with the uninfected monocyte control. Consequently, adhesion was even higher when BMVEC monolayers were TNF-α stimulated, 16.5-fold vs 11.8-fold for uninfected monocyte control. Rosiglitazone treatment (PPARγ activation) of BMVEC readily attenuated the adhesion of HIV-1 infected monocytes ranging from 16.5-fold in untreated to 12.5, 9.5, and 5.8 for rosiglitazone concentrations 1, 5, and 50 μM, respectively (p < 0.01). Therefore, these data suggest that PPARγ stimulation in BMVEC prevented adhesion to HIV-1-infected monocytes to unstimulated as well as stimulated brain endothelium. Next, we examined rosiglitazone effects on transendothelial migration of infected monocytes. As shown in Fig. 7 B, migration of HIV-1-infected monocytes was significantly enhanced in response to CCL2/MCP-1 as compared with uninfected cells (p < 0.001). Furthermore, HIV-1-infected monocytes showed 1.7-fold enhanced migration across TNF-α activated endothelium as compared with uninfected cells. Addition of rosiglitazone to the cytokine-stimulated monolayers resulted in ∼50% inhibition in the migration of infected or uninfected cells. These observations point to novel therapeutic approaches in using PPARγ agonists to prevent infiltration of HIV-1-infected monocytes across the BBB.

The responses of the endothelium to neuroinflammatory conditions play a significant role in the regulation of immune cell infiltration into the brain parenchyma (34, 35). Whether CNS inflammation arises from an autoimmune (multiple sclerosis) or from a pathogen-driven disorder as in the case of HIV-1 encephalitis, immune cell migration has damaging effects to the BBB and to the neuronal environment (19, 36). Once immune cells cross the BBB, a cascade of events, in part orchestrated by astrocytes and other glia cells, contributes to potentiation of CNS inflammation. In this report, we explored the anti-inflammatory effects of PPAR ligands in preventing interactions between brain endothelium and monocytes. Using primary human BMVEC, we modeled the human BBB in vitro and simulated a condition of chronic inflammation (cytokine stimulated BMVEC) and then performed quantitative analysis of monocyte adhesion and transendothelial migration. Our analyses indicated that PPARγ activation by potent TZD, such as rosiglitazone led to a significant reduction in monocyte adhesion and migration in resting and stimulated BMVEC. This process was highly specific for PPARγ, as shown by knockdown of PPARγ in BMVEC and failure of selective PPARα agonist to exhibit similar effects. In this study, we demonstrated the inhibition of Rac1 and RhoA GTPases by PPARγ stimulation in brain endothelium and provided evidence that this is the mechanism preventing monocyte engagement and passage across brain endothelium. Lastly, as proof of concept, we showed that synthetic PPARγ agonist hindered adhesion and migration of HIV-1 infected monocytes. Taken together, our data suggest that PPARγ activation in brain endothelium may be a strategy for counteracting CNS inflammation contributed by immune cell infiltration.

It is recognized that PPARs agonists could be useful in attenuating inflammatory responses (13, 37). PPARs are nuclear hormone receptors, which are ligand activated and act as transcription factors. The three isotypes of PPARs (PPARα, PPARγ, and PPARβ/δ) are expressed in endothelium and have been studied as agents to curb the inflammation in atherosclerosis (13). Although up to date, no studies have been performed to evaluate the anti-inflammatory effect of PPARs activation in human brain endothelial cells, a recent study demonstrated the inhibitory effect of PPARγ stimulation on CD4⁺ lymphocyte adhesion and migration across monolayers of mouse brain endothelial cells (38). As CNS inflammation is attributed in part to the infiltration of immune cells across brain endothelium, we first evaluated whether activation of PPARα and PPARγ could prevent monocyte adhesion to endothelial cells and transendothelial migration. Using PPARγ and PPARα stimulators, we demonstrated that PPARγ stimulation and not PPARα exerted a striking dose-dependent inhibition of monocyte adhesion and migration across activated BMVEC. This observation differs from those reported for aortic vascular endothelium where either PPARα and PPARγ activation prevents monocyte adhesion (12, 39). Although rosiglitazone is a highly specific PPARγ agonist, it is possible that the observed effects occur via other intracellular factors independent of PPARγ as suggested by Chawla and colleagues (28). Using targeted shRNA down-regulation of endogenous PPARγ, we observed near complete reversal of monocyte adhesion/migration when the BMVEC were treated with rosiglitazone indicating specificity of PPARγ effects in BMVEC.

To address the underlying mechanism of diminished monocyte adhesion, we analyzed the expression of adhesion molecules ICAM-1 and VCAM-1 in TNF-α stimulated BMVEC. Although PPARγ stimulation was shown to down-regulate ICAM-1 and VCAM-1 in some experimental systems (38, 40), we did not find an appreciable effect on surface expression of adhesion molecules. Consistent with our data, others previously reported a lack of effect on ICAM-1 and VCAM-1 regulation after activation of PPARγ in TNF-α-stimulated human aortic endothelial cells using the natural ligands belonging to the conjugated linoleic acid class of fatty acids (29). It is possible that the effect on the expression of adhesion molecules by PPARγ could occur if treatment with PPARγ ligands was extended beyond the few hours performed in this study (38, 40). Similarly, the PPARγ agonist rosiglitazone did not change barrier function of BMVEC monolayers (measured by TEER) or distribution/level of TJ proteins in brain endothelium.

We next considered the possibility that PPARγ activation could inhibit the Rho family of GTPases, given that leukocyte adhesion and migration are tightly regulated by GTPase function in endothelial cells (15, 16). All Rho family members act as molecular switches by binding GDP, rendering inactivation, or GTP to shift into the activation state. GTP-bound GTPases act on effectors that then induce changes to cytoskeletal dynamics and gene regulation (41). In the endothelium, RhoGTPases are activated in response to inflammatory stimuli (such as TNF-α, IL-1β, and bacterial toxins), hypoxia, shear stress (33, 42, 43, 44, 45, 46), or endothelial-leukocyte interactions (19, 31). RhoA and Rac1 were activated in BMVEC after TNF-α treatment and further enhanced by monocyte-BMVEC engagement or simulation of this process by ICAM-1 Ab crosslinking. We found that PAPRγ activation markedly decreased the level of active GTP-bound Rac1 and RhoA. The maximal inhibition of Rac1 occurred at low concentrations of PPARγ agonist, whereas most prominent RhoA inhibition was achieved at higher doses. Using BMVEC-expressing CA and DN mutants of Rac1 and RhoA, we showed that PPARγ inhibition of Rac1 impacted monocyte adhesion; whereas, RhoA suppression affected monocytes transendothelial migration. Expression of the Rac-CA mutant had an overriding effect on monocyte adhesion even when PPARγ was activated. Conversely, Rho-CA did not counteract the inhibitory effect of the PPARγ agonist on monocyte adhesion. In the case of migration, the Rho-CA mutant showed significant increase in monocyte migration with a minimal inhibitory effect achieved by PPARγ activation. The Rac-CA mutant was able to supersede PPARγ-induced inhibition on monocyte migration only at the lower and not at the higher concentrations of rosiglitazone, while the Rho-CA showed an equipotent response to all doses. Taken together these results suggest a dose-dependent inhibition on Rac1 and RhoA GTPase activity. To our knowledge, this is the first time that PPARγ agonist are reported to negatively regulate both RhoA and Rac1 in brain endothelium or endothelial cells of any origin. The only available study on RhoA inhibition by PPARγ stimulation was performed in smooth muscle where PPARγ in aortic smooth muscle cells could induce the up-regulation of the protein tyrosine phosphatase SHP-2, leading to inhibition of the RhoA/Rho kinase pathway (14). Additional studies will be necessary to elucidate whether a similar mechanism is operational in the endothelium as wells as how PPARγ may regulate function of multiple GTPases and whether this mechanism occurs by transcription-dependent or independent mechanisms.

A number of neuroinflammatory disorders (such as multiple sclerosis or encephalitis) are associated with monocyte infiltration across BBB leading to neuronal injury and neurological demise (47). HIV-1 encephalitis is characterized by monocyte and macrophage accumulation in affected brain tissue (48). Perivascular macrophages serve as a source of neurotoxins and viral reservoir, and they are a driving force of chronic inflammation providing chemokines and homing cues for circulating HIV-1 infected cells (49, 50). A pool of perivascular virus-infected macrophages is derived from monocytes migrating across the BBB and suppression of monocyte infiltration may ameliorate neuronal demise (50). PPARγ stimulation significantly diminished adhesion and migration of HIV-1 infected monocytes across BMVEC monolayers. The evidences provided in this study of PPARγ-mediated inhibition on Rac1 and RhoA GTPases extend our previous findings showing that RhoA activation is required in the BMVEC during interactions with HIV-1 infected monocytes, and its suppression prevents monocyte migration and BBB dysfunction (19). Taken together, these findings suggest a novel role for PPARγ at the BBB, which offers a therapeutic strategy to diminishing infiltration of leukocytes into the brain parenchyma.

We thank Dr. Georgette Kanmonge for providing human microvessel lysates, Dr. Anuja Ghorpade for critical reading of this manuscript, and Robin Taylor and Debra Baer for excellent editorial support.

The authors have no financial conflict of interest.