Protein Kinase Cα-RhoA Cross-talk in CCL2-induced Alterations in Brain Endothelial Permeability*

Monocyte chemoattractant protein-1 (MCP-1 or CCL2) regulates blood-brain barrier permeability by inducing morphological and biochemical alterations in the tight junction (TJ) complex between brain endothelial cells. The present study used cultured brain endothelial cells to examine the signaling networks involved in the redistribution of TJ proteins (occludin, ZO-1, ZO-2, claudin-5) by CCL2. The CCL2-induced alterations in the brain endothelial barrier were associated with de novo Ser/Thr phosphorylation of occludin, ZO-1, ZO-2, and claudin-5. The phosphorylated TJ proteins were redistributed/localized in Triton X-100-soluble as well as Triton X-100-insoluble cell fractions. Two protein kinase C (PKC) isoforms, PKCα and PKCζ, had a significant impact on this event. Inhibition of their activity using dominant negative mutants PKCα-DN and PKCζ-DN diminished CCL2 effects on brain endothelial permeability. Previous data indicate that Rho/Rho kinase signaling is involved in CCL2 regulation of brain endothelial permeability. The interactions between the PKC and Rho/Rho kinase pathways were therefore examined. Rho, PKCα, and PKCζ activities were knocked down using dominant negative mutants (T17Rho, PKCα-DN, and PKCζ-DN, respectively). PKCα and Rho, but not PKCζ and Rho, interacted at the level of Rho, with PKCα being a downstream target for Rho. Double transfection experiments using dominant negative mutants confirmed that this interaction is critical for CCL2-induced redistribution of TJ proteins. Collectively these data suggest for the first time that CCL2 induces brain endothelial hyperpermeability via Rho/PKCα signal pathway interactions.

kinase activity prevents both the reorganization of the actin cytoskeleton and the redistribution of TJ proteins (14). Because of those results, the present study examined the signaling events triggered by CCL2 in order to induce morphological and biochemical alterations in tight junction complexes associated with reorganization of the endothelial cytoskeleton.
Our results show that CCL2 activates PKC (particularly PKC␣) in brain endothelial cells, which in turn induces phosphorylation of TJ proteins (occludin, ZO-1, ZO-2, claudin-5). However, PKC␣ activation in brain endothelial cells was strongly dependent on RhoA activation, indicating that RhoA is a crucial component in the cascade of signal events induced by CCL2.

EXPERIMENTAL PROCEDURES
Materials-CD-1 mice were obtained from Charles River (Portage, MI). Monocyte chemoattractant protein-1 (CCL2) was from Peprotech (Rocky Hill, NJ). All chemicals other than those listed below were from Sigma.
Aprotinin, leupeptin, pepstatin A, and antipain were purchased from Roche Applied Science. For permeability experiments, [ 14 C]inulin was obtained from PerkinElmer Life Sciences. Rho activation and PKC assay kits along with C3 exoenzyme were obtained from Upstate. The Rho kinase inhibitor Y27632 was purchased form Calbiochem. A Pierce assay kit was used for protein determination (Pierce Biotechnology). Western blots were visualized with a chemiluminescent HRP substrate kit (Pierce Biotechnology).
Brain Endothelial Cell Cultures-Mouse brain microvascular endothelial cells (mBMEC) were prepared using a previously described protocol (14,24). Briefly, 4 -6-week-old CD-1 mice were euthanized by decapitation under isoflurane anesthesia. The brains were minced in HBSS and gently homogenized in a Dounce-type homogenizer. The homogenates were suspended in 18% dextran solution and centrifuged to remove myelin. The resulting pellet were resuspended in HBSS, layered over a preformed Percoll gradient, and centrifuged at 2700 rpm for 11 min. After centrifugation, the top layer containing the microvessels was collected and then digested for 40 min at 37°C in HBSS solution containing 1 mg/ml collagenase/dispase, 10 units/l DNase I, and 1 g/ml TLCK. mBMEC were cultured in DMEM supplemented with 10% inactivated fetal calf serum, 2.5 g/ml heparin, 20 mM HEPES, 2 mM glutamine, 1ϫ antibiotic/antimycotic, and endothelial cell growth supplement and grown in 6-well plates coated with collagen type IV (BD Biosciences). Application of this protocol typically produces primary endothelial cell cultures that are ϳ99% pure (as determined by immunocytochemistry with an anti-PECAM-1 antibody).
Western Blotting-Protein concentrations were determined using a Pierce protein assay kit. Equal amounts of protein samples were loaded, separated using 7.5 and 15% SDS-polyacrylamide gel electrophoresis, and then transferred to Trans-Blot nitrocellulose membrane. Immunoblotting was performed with mouse anti-occludin, anti-ZO-1, anticlaudin-5, and anti-ZO-2 antibodies and rabbit anti-Rho, anti-PKC␣, PKC␤, PKC␥, PKC␦, PKC, PKC/, and PKC along with the corresponding anti-phospho-PKC antibodies. Immunoblots were then exposed to secondary anti-mouse or anti-rabbit HRP-conjugated antibody and visualized using a chemiluminescent HRP substrate kit. The relative densities/volumes of the bands on the film negatives were measured using Image J software (NIH Image, Bethesda, MD).
Fractional Analysis of Tight Junctions-mBMEC were subjected to modified extraction protocol using a ProteoExtract subcellular proteome extraction kit (Calbiochem) according to the manufacturer's suggestions. Utilizing different extraction buffers in the kit, four different fractions were separated: Triton-soluble cytosol fractions, Triton-insoluble membrane/organelle fractions, nuclear fractions, and actin cytoskeleton fractions. To prepare "total cell lysate" samples, cells were washed in phosphate-buffered saline, scraped, rinsed in 1 ml of radioimmunoprecipitation buffer, sonicated, and centrifuged briefly. After that, the supernatant was collected and stored at Ϫ70°C.
Immunoprecipitation-Immunoprecipitation was performed using a protein A size exclusion immunoprecipitation kit (Sigma) according to the manufacturer's suggestions.
Transendothelial Electrical Resistance (TEER)-Electrical resistance across endothelial cells monolayers was measured by Millicell ERS (World Precision Instruments). In these sets of experiments, mBMECs were plated in Transwell culture dishes (0.4-m pore size; Corning Inc.). CCL2 (100 ng/ml) was placed in the lower and upper compart-ment of the Transwell dual chamber system. TEER was measured between 15 min and 2 h. The resistance of blank filters was subtracted for calculation of final TEER values (ohm⅐cm 2 ).
mBMEC Monolayer Permeability-The effects of CCL2 on endothelial monolayer permeability was examined using [ 14 C]inulin, a tracer that crosses the endothelium by passive diffusion (27). Cells were cultured on Transwell culture dishes with 0.4-m pore size filters. The permeability experiments were initiated by the addition of 0.2 Ci of the isotope to the apical or donor chamber, which contained 0.4 ml of DMEM. The basal or receiving chamber contained 1.2 ml of DMEM. 0.2 ml of medium from the basal chamber was sampled and replaced with fresh DMEM at 15-min intervals from 0 to 120 min. Scintillation fluid was added to the samples, and radioactivity was counted using a Beckman 3801 liquid scintillation counter (Fullerton, CA). The permeability (P) (cm/min) of the monolayer during any time interval (T) was calculated using the following equation, where C(B) and C(B) T are, respectively, the concentrations of isotope in the basal chamber at the start and at the end of the time interval (in dpm/ml) and V(B) is the volume of the basal chamber (in ml). C(A) and C(A) T are, respectively, the concentrations of isotope in the apical or donor chamber at the start and at the end of the time interval (in dpm/ ml), and (C(A) T ϩ C(A) Tϩ15 )/2 is the average concentration over the time interval. T is the duration of the time interval in minutes, whereas A is the area of the filter (cm 2 ). Rho and Rho Kinase Activation Assay-Affinity precipitation of lysed mBMEC with agarose-bound recombinant Rhotekin protein was performed according to the manufacturer's instructions. After agarose bead removal, samples were resuspended in buffer and processed for Western blot using a rabbit polyclonal anti-Rho antibody. A Rho kinase activation assay (ROK␣/ROCK-II KinEASE TM FP fluorescein green assay; Upstate) was performed according to the manufacturer's instructions.
PKC Activation Assays-Because some studies have shown that phospho-PKC antibodies have limitations for detecting PKC activity, we also included functional assays for PKC activity and/or specific PKC isoform activity in total cell lysates after treatment with CCL2 (28). The following PKC assays were performed according to the manufacturer's suggestions: colorimetric PKC assay (PepTag; Promega, Madison, WI) and specific PKC␣ and PKC assays (PKC␣ KinEASE TM FP fluorescein green assay and PKC KinEASE TM FP fluorescein green assay, Upstate). Effect of CCL2 on redistribution and phosphorylation of TJ proteins in brain endothelial cells. mBMEC monolayers were exposed to CCL2 for 15 or 120 min or were not treated (control). Three cell fractions were prepared: Triton X-100-soluble cytosol fraction (CF), Triton X-100-insoluble membrane fraction (MF), and actin cytoskeleton fraction (ACF). A, Western blot analysis of TJ proteins (occludin, claudin-5, ZO-1, ZO-2) in the cell fractions. B, densitometric analysis was performed on the Western blots shown in A. The values are mean Ϯ S.D. for three independent experiments. C, the cytosol, membrane, and actin cytoskeleton fractions were immunoprecipitated (IP) with antibodies against the TJ proteins occludin, claudin-5, ZO-1, and ZO-2. These immune complexes were then immunoblotted for phospho-Ser. Data represent one of three independent experiments.
Statistics-All results are expressed as means Ϯ S.D. One-way analyses of variance were used to compare the mean responses among the experimental groups. A post-hoc Dunnett's test was used to determine significance between groups.

CCL2-induced Phosphorylation of Serine/Threonine Residues on
Tight Junction Proteins-To elucidate how CCL2 alters brain endothelial cell TJ structure, TJ protein phosphorylation was examined. mBMEC were exposed to CCL2 (100 ng/ml), and the phosphorylation status of TJ proteins in different cell fractions (denoted as Triton X-100soluble cytosol, Triton X-100-insoluble membrane, and actin cytoskeleton fractions) was determined. As shown in Fig. 1, CCL2 induced a redistribution of TJ proteins. In cells not treated with CCL2, occludin and claudin-5 were found in the membrane fraction. However, after CCL2 treatment only a small portion of these proteins remained in that fraction, with a movement to the cytosol fraction (within 15 min) and the actin cytoskeleton fraction within 15 min for occludin and 2 h for claudin-5 (Fig. 1, A and B). For ZO-1 and ZO-2, the proteins were predominantly in the cytosol fraction in control cells, and CCL2 caused a decrease in this fraction while causing an increase in the actin cytoskeleton fraction (Fig. 1, A and B). Interestingly, the total TJ protein levels did not change during CCL2 treatment, excluding the possibility of TJ protein degradation (data not shown).
In analyzing the phosphorylation status of TJ proteins in all fractions, immunoprecipitation was performed with anti-occludin, ZO-1, ZO-2, and claudin-5 antibodies followed by Western blotting analysis utilizing antibodies that specifically recognize phosphorylated Ser, Thr, and Tyr residues. As shown in Fig. 1C, this analysis indicated two important things. (a) TJ proteins (occludin, ZO-1 ZO-2, and claudin-5) underwent de novo phosphorylation in the presence of CCL2, (cytosol-and cytoskeleton-insoluble fractions but not the membrane fraction). (b) Phosphorylation of these proteins occurred on Ser residues that have been described as substrates for PKC activity (29 -31), suggesting that the PKC pathway could be also involved in the processing of CCL2induced tight junction redistribution.
CCL2 Activates Distinct PKC Isoforms in Brain Endothelial Cells-Using a PKC activation assay, we found that CCL2 (100 ng/ml) significantly increased PKC activity in brain endothelial cells (total cell lysate), with peak activity at 10 -30 min ( Fig. 2A). In particular, CCL2 activated three PKC isoforms, PKC␣/␤, PKC, and PKC/ (Fig. 2B). To examine whether activation of PKC␣/␤, PKC, and PKC/ by CCL2 was associated with tight junction redistribution, immunoprecipitation of Triton X-100-soluble cytosol and Triton X-100-insoluble actin cytoskeleton fractions was performed with anti-occludin, ZO-1, ZO-2, and claudin-5 antibodies followed by Western blot analysis of these samples using anti-phospho-PKC␣, -PKC, or -PKC antibodies. The results indicated that PKC␣ and PKC were mostly present in the cytosolic fraction (CF), whereas the actin-cytoskeleton-insoluble fraction (ACF) contained notably less amounts of PKC and very low levels of PKC␣ (Fig. 2C). On the other hand, we were not able to detect the presence of PKC in any of these fractions, suggesting that PKC is not associated with phosphorylation of TJ proteins in CCL2-induced brain endothelial hyperpermeability (data not shown). Our results, however, suggested that two PKC isoforms (PKC␣ and PKC) could participate in phosphorylation during TJ redistribution. To further evaluate the contribution of PKC␣ and PKC in CCL2induced brain endothelial hyperpermeability, experiments were performed in which the activity of PKC␣ and PKC in brain endothelial cells was diminished by transfection with dominant negative mutants for PKC␣ and PKC. At the functional level, in cells not treated with these mutants CCL2-induced increases in [ 14 C]Inulin permeability and decreases in TEER. Both of these effects were diminished when the activity of PKC␣ or PKC was inhibited (Fig. 3, A and B). These functional data were corroborated by biochemical analysis showing that reducing PKC␣ and PKC activity during CCL2 treatment reduced the redistribution of occludin, ZO-1, ZO-2, and claudin-5 as evaluated by Western blot analysis (Fig. 3C). At the morphological level, in cells not treated with these mutants CCL2 induced a fragmentation and loss of TJ protein immunostaining (Fig. 4). In contrast, CCL2 induced only slight changes in TJ protein distribution when PKC␣ or PKC activity was inhibited (Fig. 4). Taken together, the functional, biochemical, and morphological studies indicate that PKC␣ and PKC are involved in the redistribution of occludin, ZO-1, ZO-2, and claudin-5 by CCL2.
Interactions between Rho and PKC␣/PKC in CCL2-induced Tight Junction Opening-Prior evidence indicates that RhoA and Rho-associated kinase are involved in CCL2-induced TJ alterations as well as in the reorganization of the actin cytoskeleton (14,24). The data presented above indicate that two isoforms of PKC (PKC␣ and PKC) also play a prominent role in the phosphorylation/redistribution of TJ proteins. To further reconstruct the signaling pathways triggered by CCL2, potential Twenty-four hours later they were exposed to CCL2 for 120 min. The cells were then fixed and processed for immunofluorescent examination of occludin, claudin-5, ZO-1, and ZO-2. The samples were viewed by confocal microscopy (Zeiss LSM 510) and compared with control monolayers not exposed to CCL2. Arrows indicate fragmentation and loss of immunostaining for TJ proteins in the presence of CCL2. Arrowheads indicate continuous immunostaining along the cell border for TJ proteins in control mBMEC monolayer and mBMEC transfected with PKC␣-DN or PKC-DN. Scale bar, 20 m.

FIGURE 3. The involvement of PKC isoforms (PKC␣ and PKC) in CCL2-induced increases in brain endothelial permeability and redistribution of TJ proteins.
mBMEC monolayers were first transfected either with the dominant negative mutant PKC␣-DN or PKC-DN or just with vector. Twenty-four hours later, they were exposed to CCL2 (0 -120 min). Controls were not treated with CCL2. The cells were used to examine TEER at 2 h (A) and transendothelial permeability coefficient (PC) for [ 14 C]inulin from 0 to 120 min (B). Data represent mean Ϯ S.D. of five independent experiments. * and **, indicate significant differences from CCL2-treated group at the p Ͻ 0.01 and p Ͻ 0.001 level, respectively. Reducing PKC␣ and PKC activity reduced CCL2-induced barrier disruption as assessed by TEER and permeability measurements. C, reducing PKC␣ and PKC activity in mBMEC with PKC␣-DN or PKC-DN also diminished TJ (occludin, claudin-5, ZO-1, ZO-2) redistribution from membrane (MF) or cytosol (CF) fraction to the Triton X-100-insoluble actin cytoskeleton fraction (ACF) 2 h after CCL2 treatment.
interactions between the Rho/Rho kinase and PKC pathways were examined; i.e. do they act simultaneously and independently or is there "cross-talk" between Rho/Rho kinase and PKC␣/PKC, where one might be a downstream target of the other? Experiments were performed in which Rho and Rho kinase activity was blocked (by dominant negative mutant RhoAT19N or with siRNA Rho kinase (pKD ROK␣/ ROCK-II-v6)) and activity of PKC␣ and PKC was observed by specific PKC assay or the appearance of phosphorylated PKC in brain endothelial cells. Similar experiments were performed in the opposite direction where dominant negative mutants of PKC␣ and PKC were used to inhibit activity, and Rho and Rho kinase activity was observed. The results showed that diminishing the activity of PKC␣ or PKC did not interrupt any changes in activity of RhoA or Rho kinase in the presence of CCL2 (Fig. 5, A-D). However, diminishing the activity of RhoA significantly diminished PKC␣ activation, whereas PKC was unaffected. Inhibition of Rho kinase activity using the siRNA Rho kinase (pKD ROK␣/ROCK-II-v6) did not alter PKC␣ and PKC activity in mBMEC treated with CCL2 (Fig. 6, A-D).
Thus, it appears that in CCL2-induced brain endothelial hyperpermeability, RhoA acts as an upstream signal molecule that activates both PKC␣ (which causes phosphorylation and redistribution of TJ proteins) and Rho kinase (which is involved in the redistribution of TJ proteins FIGURE 5. Effect of inhibition of Rho or Rho kinase activity on PKC activation in brain endothelial cells. mBMEC were either transfected with RhoAT19N or its vector or were treated with siRNA for Rho kinase. Twenty-four hours later, the cells were exposed to CCL2 (100 ng/ml) for 0 -120 min. A-D, in total cell lysates, the presence of phospho-PKC␣ (A) and phospho-PKC (C) as well as the level of PKC␣ (B) and PKC activity (D). Blots represent one of three independent experiments, and the values in the bar graphs are means Ϯ S.D. of three independent experiments. Inhibition of Rho reduced phospho-PKC␣ levels and PKC␣ activity, but it did not affect PKC activity. Inhibition of Rho kinase did not affect PKC␣ or PKC. E, control study. The activity of Rho kinase and level of Rho-GTP was observed in cell lysates from mBMEC transfected by RhoAT19N or its vector or treated with siRNA for Rho kinase under the same condition as described in A-D. Data represent mean Ϯ S.D. of three independent experiments. and reorganization of the actin cytoskeleton). To confirm that the interaction of Rho and PKC␣ is critical for redistribution/phosphorylation of TJ proteins, we performed single and double transfection studies utilizing the RhoAT19N mutant and the dominant negative PKC␣-DN mutant. Absence of activity of both of these two signal molecules completely blocked the effect of CCL2 on brain endothelial permeability at the functional level (particularly the permeability coefficient for fluorescein isothiocyanate-albumin). At the morphological level, transfection with the two mutants prevented the fragmentation and loss of TJ protein immunostaining that was found in control brain endothelial cells treated with CCL2 (Fig. 7, A and B). Comparing double transfection data with results obtained from inhibiting Rho or PKC␣ activity alone indicated that Rho acts as a critical factor for PKC␣ activation. Thus, blocking Rho activity alone had an effect similar to that of blocking both Rho and PKC␣ activity and Rho acting upstream of PKC␣ in regulation of permeability brain endothelial barrier by CCL2.

DISCUSSION
Previous studies have shown that the chemokine CCL2 can increase blood-brain barrier permeability in vivo and brain endothelial barrier permeability in vitro (14,24). CCL2 alters TJ complex structure and induces actin cytoskeleton reorganization. The Rho/Rho kinase signal pathway has a prominent role in CCL2-induced hyperpermeability (14). The present study further examined the signaling pathways activated by CCL2 to enhance barrier permeability. The results demonstrate the following. (a) CCL2 causes TJ protein phosphorylation (occludin, ZO-1, ZO-2, and claudin-5) in brain endothelial cells. This occurred mostly at Ser/Thr residues that are specific substrates for PKC activity. (b) PKC isoforms (particularly PKC␣ and PKC) are activated by CCL2 and their active form is coupled to TJ proteins. (c) With respect to CCL2-induced hyperpermeability, the Rho/Rho kinase and PKC signal pathways interact at the level of Rho protein. (d) Rho proteins are key regulators of PKC␣ activity but do not influence PKC activity. The implications of these findings are discussed below.
CCL2 is the most commonly expressed chemokine during central nervous system inflammation. It is robustly expressed during the acute inflammatory response associated with ischemia/reperfusion (stroke) injury as well as in neuropathological conditions associated with socalled chronic central nervous system inflammatory responses, such as multiple sclerosis, human immunodeficiency virus infection, Alzheimer disease, and brain trauma (32)(33)(34)(35)(36). Mice lacking CCL2 (CCL2Ϫ/Ϫ) or the CCL2 receptor, CCR2 (CCR2Ϫ/Ϫ), have a decreased inflammatory response and leukocyte migration after middle cerebral artery occlusion and experimental allergic encephalitis, as well as reduced BBB disruption and brain edema formation (37,38).
In general, BBB disruption/disregulation during central nervous system inflammation is believed to result from the "loosening" of junctional complexes between brain endothelial cells. This leads to the formation of a paracellular route that facilitates the entry of leukocytes into the brain parenchyma. Although some other pathways for leukocyte transmigration have been proposed (e.g. transcytosis), evidence from experimental and clinical studies still point to the importance of the paracellular route in leukocyte entry and BBB disruption during central nervous system inflammation (39,40).
A number of experimental interventions and pathophysiological states can alter paracellular permeability; modify the expression, cellular distribution, and/or phosphorylation of TJ-associated proteins; and further change the functional interactions between TJ proteins and the cytoskeleton (1,2). TJ proteins (e.g. occludin, ZO-1, ZO-2, and claudin-5) are phosphoproteins, and regulation of their function is mostly via alterations in their state of phosphorylation (12,(41)(42)(43)(44). There is considerable controversy over whether additional phosphorylation or dephosphorylation of TJ proteins is linked to their redistribution during increases in endothelial barrier permeability (40,42). Changes in TJ phosphorylation and dephosphorylation status depend upon the type of cell (endothelial or epithelial), the type of stimulus (e.g. inflammatory cytokines, oxidative stress or growth factors), and the amino acid residues where phosphorylation is taking place (Ser, Thr, or Tyr). For example, vascular endothelial growth factor induces Ser/Thr phosphorylation and redistribution of occludin and ZO-1 in bovine aortic and retinal endothelial cells, but during calcium depletion, phorbol ester treatment, and bacterial infection, occludin undergoes dephosphorylation during TJ disruption (42,(45)(46)(47)(48). The current study suggests that CCL2 acts in a manner similar to vascular endothelial growth factor, inducing phosphorylation of Ser/Thr residues on occludin, ZO-1, ZO-2, and claudin-5. This phosphorylation may affect the detergent solubility of these proteins, resulting in a shift from the Triton X-100-soluble to the Triton X-100-insoluble fraction. Phosphorylation of occludin, ZO-1, ZO-2, and claudin-5 was found in the cytosol-soluble fraction soon after CCL2 exposure followed by a further shift to the Triton-X100-insoluble actin cytoskeleton fraction. Taking into consideration that CCL2 induced a loss and fragmentation of immunostaining for TJ proteins, the shift to the actin cytoskeleton-insoluble fraction could be an indication of a possible association of these proteins with some vesicular structure (calveole or pinocytotic vesicle). In epithelial cells, endocytosis appears to be involved in the redistribution of TJ proteins and barrier opening (49,50). The role of endocytosis in the redistribution of TJ proteins at the BBB requires further investigation.
Tight junctions are regulated by a diverse group of extracellular stimuli that initiate many intracellular signaling cascades (2,12,51,52). The mechanistic links between the signaling pathways and enhanced permeability have yet to be elucidated. Our previous study indicated that the CCL2-induced brain endothelial barrier "opening" is closely associated with activation of the Rho/Rho kinase axis (14). However, the current analysis of the phosphorylation status of occludin, ZO-1, ZO-2, and claudin-5 showed that CCL2 phosphorylated these proteins on Ser residues. Such residues are substrates for PKC action, suggesting that PKC could also be involved in TJ protein redistribution/phosphorylation. Our study found that two specific PKC isoforms, PKC␣ and PKC, are activated by CCL2. Both isoforms impacted upon the brain endothelial barrier permeability. PKC has a prominent role in endothelial TJ complex assembly/disassembly. Two classic forms of PKC, PKC␣ and PKC␤ (activated by H 2 O 2 , thrombin, and glucose), as well as atypical forms of PKC (PKC and PKC), which constitute part of tight junction complex, are thought to be mostly involved in TJ disassembly (18,31,54,56). In our model system, CCL2 induced activation of PKC␣ and PKC and through them caused morphological, biochemical, and functional alterations in the brain endothelial barrier. The inhibition study (specific exclusion of PKC␣ and PKC activation by transfection of brain endothelial cells with dominant negative mutants) clearly indicates the obligatory role of these two PKC isoforms in the regulation of permeability by CCL2. Under our experimental conditions, CCL2 exerts some of the effects and signaling patterns described for vascular endothelial growth factor, interleukin-6, thrombin, and bradykinin (54, 57-59). However, it FIGURE 7. RhoA and PKC␣ play a critical role in CCL2-induced increases in brain endothelial permeability. mBMEC monolayers were transfected with one of two mutants, RhoAT19N or PKC␣-DN, or with both of them, and 24 h later they were exposed to CCL2 (100 ng/ml) for 2 h. The TEER (A) and the permeability coefficient (PC) (B) of the mBMEC monolayers were then measured. CCL2, mBMEC monolayers treated with CCL2 and just vector; control, cells not treated with CCL2. Data represent the mean Ϯ S.D. of three independent experiments. * and **, indicate significant differences from control cells at the p Ͻ 0.01 and p Ͻ 0.001 levels, respectively. C, immunostaining for occludin, claudin-5, ZO-1, and ZO-2 2 h after CCL2 treatment. In contrast to vector-treated cells, which show fragmented staining along cell borders after CCL2 treatment, cells transfected with both RhoAT19N and PKC␣-DN showed continuous staining. Arrows indicate morphological alteration of TJ proteins in the presence of CCL2; arrowheads indicate continuous staining of TJ proteins in the mBMEC monolayer where PKC␣ and Rho activity were reduced by transfection with dominant negative mutants. Scale bar, 20 m. is important to note that different PKC isoforms may be activated in different model systems, making it difficult to extrapolate the role of PKC isoforms from one system to another.
The basic question addressed by our study is how does CCL2 alter TJ structure and BBB permeability? Taking into consideration the previously published data (14) and the current results, it appears that two signal pathways are involved, Rho/Rho kinase and PKC. Both pathways have the same end points, phosphorylation of TJ proteins and redistribution of those proteins away from the brain endothelial cell border. There is prior evidence that these two pathways can interact to regulate cell function. For example, in Jurkat cells, Rho interacts with PKC␣ and as a result activates transcription factor AP-1 (60). Similarly, an interaction of the Rho GTPase CdC42 with atypical PKC and PKC can cause stress fiber formation (56). In vitro protein-protein interaction studies have shown specific high affinity binding between Rho and PKC␣ and that Rho can regulate PKC␣ activation (53,55). Those findings are in accord with the current study, and they support our finding that these two pathways interact to increase brain endothelial barrier permeability.
Our findings suggest the following scenario for the enhancement of brain endothelial permeability by CCL2. CCL2, via CCR2 receptors on brain endothelial cells, activates RhoGTPase. In turn, this activates Rho kinase, which induces the reorganization of actin cytoskeleton and the redistribution TJ proteins. At the same time, RhoGTPase activates PKC␣, which acts on TJ proteins causing their direct phosphorylation and redistribution from the endothelial cell border. Thus, CCL2 affects TJ complexes between brain endothelial cells directly, by inducing phosphorylation of TJ proteins (which probably leads to conformation changes and redistribution), and indirectly, via alterations in the actin cytoskeleton, which may also lead to redistribution of TJ proteins (Fig.  8). An end result of these signaling events is increased brain endothelial barrier permeability. Thus, the presence of the chemokine CCL2 during inflammation will result in the loss of barrier integrity and facilitation of leukocyte movement into brain parenchyma.
In conclusion, this study provides a possible signal mechanism for regulation of brain endothelial barrier permeability. Although this study mostly highlights a possible mechanism for altering the junctional complexes between endothelial cells, which may regulate intercellular adhesion, we also would like to point out that part of the suggested signal mechanism also could be involved in the parallel process of actin cytoskeleton reorganization and increased intraendothelial contractility. However, in our opinion, Rho appears to be a critical molecule in the regulation of brain endothelial permeability, regulating both of these ongoing processes and directly modulating paracellular cleft formation between brain endothelial cells. Further studies of the specific signaling events associated with enhanced brain endothelial permeability may be of great potential therapeutic benefit, especially in conditions in which the blood-brain barrier is disrupted.