of June 20, 2018.This information is current as

CellsInsults in Brain Microvascular Endothelial Cannabinoids Inhibit HIV-1 Gp120-Mediated

Makriyannis and Shalom AvrahamSouvenir D. Tachado, Henry Koziel, Alexandros Tzong-Shi Lu, Hava Karsenty Avraham, Seyha Seng,

http://www.jimmunol.org/content/181/9/6406doi: 10.4049/jimmunol.181.9.6406

2008; 181:6406-6416; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/181/9/6406.full#ref-list-1

, 12 of which you can access for free at: cites 52 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists All rights reserved.Copyright © 2008 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

by guest on June 20, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Cannabinoids Inhibit HIV-1 Gp120-Mediated Insults in BrainMicrovascular Endothelial Cells1

Tzong-Shi Lu,2* Hava Karsenty Avraham,2* Seyha Seng,* Souvenir D. Tachado,†

Henry Koziel,† Alexandros Makriyannis,‡ and Shalom Avraham3*

HIV-1 infection has significant effect on the immune system as well as on the nervous system. Breakdown of the blood-brainbarrier (BBB) is frequently observed in patients with HIV-associated dementia (HAD) despite lack of productive infection ofhuman brain microvascular endothelial cells (HBMEC). Cellular products and viral proteins secreted by HIV-1 infected cells, suchas the HIV-1 Gp120 envelope glycoprotein, play important roles in BBB impairment and HIV-associated dementia development.HBMEC are a major component of the BBB. Using cocultures of HBMEC and human astrocytes as a model system for humanBBB as well as in vivo model, we show for the first time that cannabinoid agonists inhibited HIV-1 Gp120-induced calcium influxmediated by substance P and significantly decreased the permeability of HBMEC as well as prevented tight junction proteindown-regulation of ZO-1, claudin-5, and JAM-1 in HBMEC. Furthermore, cannabinoid agonists inhibited the transmigration ofhuman monocytes across the BBB and blocked the BBB permeability in vivo. These results demonstrate that cannabinoid agonistsare able to restore the integrity of HBMEC and the BBB following insults by HIV-1 Gp120. These studies may lead to betterstrategies for treatment modalities targeted to the BBB following HIV-1 infection of the brain based on cannabinoidpharmacotherapies. The Journal of Immunology, 2008, 181: 6406–6416.

T he blood-brain barrier (BBB),4 a regulated interface be-tween the peripheral circulation and the CNS, is com-prised of the cerebral microvascular endothelium, which,

together with neurons, astrocytes, pericytes, and the extracellularmatrix, constitutes a “neurovascular unit” (1–3).

Most forms of brain damage are associated with BBB disrup-tion, resulting in secondary damage to neural cells. The interen-dothelial space of the cerebral microvasculature is characterized bythe presence of a junctional complex that includes adherens junc-tions (AJs), tight junctions (TJ), and gap junctions (3–7). Whereasgap junctions mediate intercellular communication, both AJs andTJ act to restrict the permeability across the endothelium. AJs areubiquitous in the vasculature and mediate the adhesion of endo-thelial cells to each other, contact inhibition during vasculargrowth and remodeling, initiation of cell polarity, and the regula-tion of paracellular permeability (2). The primary component of

AJs is VE-cadherin. The TJ are the main components that conferlow paracellular permeability and high electrical resistance (2).Electrical resistance in vivo across the barrier usually is �1000ohm � cm2 due to the TJ (8, 9).

The zonula occludens (ZO) proteins are involved in the coordi-nation and clustering of protein complexes to the cell membraneand in the establishment of specialized domains within the mem-brane (10–13). ZO-1 links transmembrane proteins of the TJ to theactin cytoskeleton (10–13). The primary cytoskeletal protein, ac-tin, has known binding sites on all ZO proteins and on claudins andoccludin (10). Actin filaments serve both structural and dynamicroles in the cell. ZO-1 binds to actin filaments and to the C ter-minus of occludins and claudins (10), which couples the structuraland dynamic properties of perijunctional actin to the paracellularbarrier (10). Interestingly, ZO-3 expression is absent in endothelialcells (10–13).

The HIV invades the CNS early after viral exposure and causesprogressive cognitive, behavior, and motor impairments years laterwith the onset of immune deficiency (for review, see Refs. 14 and15). One of the major mediators of neuroAIDS is the transmigra-tion of HIV-infected leukocytes across the BBB into the CNS.Infected peripheral immune-competent cells, in particular macro-phages, appear to infiltrate the CNS and provoke a neuropatholog-ical response involving all cell types in the brain. The course ofHIV-1 disease is strongly influenced by viral and host factors, suchas the viral strain and the response of the host’s immune system(14, 15). In addition, HIV-1-dependent disease processes in theperiphery have a substantial effect on the pathological changes inthe CNS.

HIV infection alters BBB structure and function (14, 15), sug-gesting that altered BBB function is an early event in HIV-asso-ciated dementia (HAD), likely through the viral products activat-ing brain endothelial cells such as HIV-1 Gp120, resulting inchanges to the BBB by which substances pass through it. Alter-ations in the BBB can result from events occurring on either sideof the BBB, with a net effect of changing the ability of the BBB to

*Division of Experimental Medicine and †Pulmonary Division, Beth Israel DeaconessMedical Center and Harvard Medical School, and ‡Center for Drug Discovery andDepartment of Chemistry and Chemical Biology, Northeastern University, Boston,MA 02115

Received for publication June 9, 2008. Accepted for publication September 2, 2008.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported in part by National Institutes of Health Research GrantsHL80699 (to S.A.) and CA096805 (to H.K.A.) and the Blood Foundation (H.K.A.).2 T.-S.L. and H.K.A. share first authorship.3 Address correspondence and reprint requests to Dr. Shalom Avraham, Division ofExperimental Medicine, Beth Israel Deaconess Medical Center, Harvard Institutes ofMedicine, 4 Blackfan Circle, Third Floor, Boston, MA 02115. E-mail address:savraham@bidmc.harvard.edu4 Abbreviations used in this paper: BBB, blood-brain barrier; ACEA, arachidonyl-2�-chloroethylamide; 2-AG, 2-arachidonoylglycerol; AJ, adherens junction; BMEC,brain microvascular endothelial cell; EB, Evans blue; FAAH, fatty acid amide hy-drolase; HAD, HIV-associated dementia; HBMEC, human BMEC; JAM, junctionaladhesion molecule; SP, substance P; TEER, transendothelial electrical resistance; HC,�9-tetrahydrocannabinol; TJ, tight junction; ZO, zonula occludens.

Copyright © 2008 by The American Association of Immunologists, Inc. 0022-1767/08/$2.00

The Journal of Immunology

www.jimmunol.org

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

properly regulate the substances entering it. Indeed, HIV-infectedpatients showed changes in BBB integrity over a 2-year periodalthough there was no sign of neurological disease (16). Further-more, BBB leakage was greater in patients with AIDS than amongasymptomatic HIV� individuals (17, 18). Reduction in staining forthe TJ ZO-1 protein was also observed in basal ganglia and sub-cortical white mater in patients who died with HIVE along withaccumulation of HIV-infected brain macrophages (18, 19). Thus,TJ play an important role in the integrity and maintenance of hu-man brain microvascular endothelial cells (HBMEC) and the BBB.

The HIV-1 envelope protein Gp120, in particular, has been pro-posed as an etiologic agent of neuronal loss as well as for devel-opment of the HAD complex (20–23). There is a massive sheddingof the envelope Ags of HIV-1, such as Gp120, as detected byimmunoelectron microscopic studies (20, 21). Gp120, shed fromHIV-1 and HIV-1-infected cells, was shown to be present in theblood of HIV-1-infected patients (20, 21). Gp120 has also beendetected in the cerebrospinal fluid of HIV-1-infected patients (20,22) and is neurotoxic both in vitro and in vivo. The role of Gp120in contributing to HAD has been highlighted by the observationthat free Gp120 protein was found in the brains of AIDS patients(20, 22). HIV-1 Gp120 protein was reported to induce endothelialcell apoptosis and to down-regulate TJ proteins in monolayer cul-tures of HBMEC. These effects may lead to alteration of both thefunctional and the molecular properties of the BBB, which couldincrease trafficking of HIV and HIV-1 infected cells into the CNSand contribute to the pathogenesis of HAD. HIV-1 Gp120 fromboth CCR5 and CXCR4-tropic viruses were shown in a monolayerof HBMEC to diminish integrity and to increase permeability andmonocyte migration through CCR5/CXCR4 receptor-mediatedprotein kinase C activation and [Ca2

�]i 32 release (24, 25). Inaddition, Gp120 was shown to alter TJ protein expression andbrain endothelial cell permeability by a mechanism involving sub-stance P (SP) (26–30), although the molecular mechanisms ofthese Gp120-mediated effects are not well defined. It is widelyreported that SP and its receptor (neurokinin-1R) are potent mod-ulators of neuroimmunoregulation and HIV/AIDS infection. SPwas shown to induce and mediate inflammation, angiogenesis, in-fections, intestinal mucosal immunity, and stress. SP is able toactivate several immune cells, such as CD4� and CD8� T lym-phocytes, mast cells, NK cells, and macrophages.

Endocannabinoids are involved in neuroprotection through anumber of biochemical pathways (31). Neuroprotective effectshave been described for the 2-arachidonoylglycerol (2-AG) endo-cannabinoid in several neurotoxicity models (32–34). The identi-fication of cannabinoid receptors (CB1 and CB2) and their endog-enous lipid ligands has triggered an exponential growth of studiesexploring the endocannabinoid system and its regulatory functionsin health and disease. Cannabinoids (active ingredients in mari-juana) have been shown to be useful in the treatment of somecancer patients and to reduce pain and improve the quality of lifein patients with AIDS (31). However, whether cannabinoid ago-nists can initiate protective activities by shielding the BBB andprotecting HBMEC following Gp120-mediated damage and lossof integrity is not known.

The endocannabinoids 2-AG and anandamide are endogenouscannabinoids as well as members of the eicosanoid class of can-nabinoids, which are arachidonic acid derivatives. Anandamide isa natural endocannabinoid that undergoes rapid enzymatic degra-dation by the enzyme fatty acid amide hydrolase (termed FAAH).Inhibition of FAAH enzyme causes accumulation of anandamidein sites of its natural production (31–34).

The study of permeability across the BBB should use a repro-ducible in vitro model that recapitulates the functional and struc-

tural properties of the BBB in situ. During brain development,brain microvascular endothelial cells (BMEC) gradually acquirethe ability to form a highly selective barrier. Astrocytes play aninductive role in this process (35). BMEC are joined together byintercellular TJ that are responsible for the acquisition of selectivepermeability. This property is specific to brain TJ and probablydepends on the higher density of proteins such as occludins andZO-1 in the BBB TJ. Cocultures of BMEC with astrocytes estab-lished a tight permeability barrier across the BMEC monolayer aswas demonstrated by a significant increase in transendothelialelectrical resistance (TEER) of �1000 ohm � cm2 as comparedwith TEER of �350 ohm � cm2 for HBMEC monolayers, as wellas showed significant reduction in the paracellular permeability for[3H]inulin and [14C]sucrose, as compared with BMEC alone(35–39).

The BBB model consists of HBMEC and astrocytes coculturedon opposite sides of gelatin-coated tissue culture inserts that permitastrocyte processes to penetrate the insert and establish contactwith the HBMEC as described previously (35–39). In this model,HBMEC differentiate and express BBB markers as a result of con-tact with astrocytes, such as the glucose transporter GLUT-1 and�-glutamyl transpeptidase, and have enhanced expression of TJproteins. The dynamic in vitro model of the BBB mimics bothfunctionally and anatomically the brain microvascular endothelialcells, creating quasi-physiological conditions for the coculturing ofhuman endothelial cells and astrocytes.

In this study, using in vitro and in vivo BBB models, we dem-onstrate that CB1-based cannabinoid agonists and the FAAH in-hibitor URB597 prevented the down-regulation of the TJ ZO-1 andclaudin-5 expression, as well as inhibited Gp120-mediated damageof brain endothelium. These data support a potential therapeuticrole for cannabinoid agonists as protective agents that preserveBBB integrity and may improve the pathogenesis of HIV-associ-ated neurodegenerative diseases such as HAD.

Materials and MethodsReagents and materials

The following reagents were used: mAbs for Von Willebrand factor (SantaCruz, Biotechnology); actin (Chemicon); Gp120MN protein (Protein Sci-ence); anti-CB1, anti-CB2, anti-GAPDH, and NK-1R Abs (Santa CruzBiotechnology); anti-ZO-1 Ab and anti-claudin-5 (Chemicon); tetrameth-ylrhodamine isothiocyanate (TRITC)-conjugated anti-mouse IgG (Dako-Cytomation); Titanium One-Step RT-PCR kit (BD Biosciences); BD Ad-vantage 2 PCR kit (BD Biosciences); and Substance P ELISA kit (CaymanChemical Company).

HBMEC

HBMEC were purchased from Cell Systems. The cells were grown toconfluence in 100-mm plastic dishes (Corning) using CSC complete me-dium (Cell Systems) containing 10% FBS with a seeding density of 7.5 �104/cm2 at 37°C in an incubator containing humidified atmosphere and 5%CO2. The HBMEC used in our study formed tubular-like networks onmatrigel and had the ability to uptake acetylated low-density lipoprotein,indicating that these cells maintained the signature properties of brain en-dothelial cells. The cells also produced von Willebrand factor and GLUT-1transporter and �-glutamyl transpeptidase endothelial-specific markers(40–42). Cells were used only up to passage 6. HBMEC were recharac-terized regularly to confirm that the cells retained their CNS properties. Thecells were restained with Abs for Von Willebrand factor, a marker forendothelial cells. These cultures were analyzed routinely for astrocyte con-tamination by staining with anti-glial fibrillary acidic protein and found tobe free of such cells.

Cocultures of HBMEC and human astrocytes

Human astrocytes were obtained from Science Cell Research Laboratoriesand were cultured based on the protocol provided by the company. Astro-cytes were characterized to be �95% immunoreactive for glial fibrillaryacidic protein. Astrocytes were used from passages 3–7 for the experimentsas described below. The astrocyte-endothelial cocultures were established

6407The Journal of Immunology

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

as a BBB model system using the Transwell coculture system (CorningCostar) as described (36). Briefly, HBMEC and astrocytes were coculturedon opposite sides of poly-L-lysine-coated tissue culture inserts with 8-�mpores (Corning) that permit astrocyte processes to penetrate the insert andestablish contact with the endothelial cells as described previously (36). Inthis model, endothelial cells differentiate and express BBB markers, suchas the glucose transporter GLUT-1 and �-glutamyl transpeptidase, andhave enhanced expression of TJ proteins (36). Cocultures exhibit barrierresistance to [3H]inulin and albumin (36).

SP treatment

Confluent cultured endothelial cells were treated with SP (1 ng/ml) orcontrol vehicle (0.9% NaCl solution, normal saline) dissolved in growthmedium for the indicated times. Cells were then washed and harvested forimmunoblotting or immunoprecipitation assays or fixed with 3.7% para-formaldehyde for immunocytochemical study. SP concentration was mea-sured by the Substance P ELISA kit (Cayman Chemical).

Gp120 activation of HBMEC

HIV-1 Gp120MN protein was used in all experiments at a concentration of10 ng/ml or as indicated. As a control, we used heat-inactivated Gp120 (10ng/ml) and/or Gp120 Ab as indicated.

Cell preparation for electrophoresis and Western blot analysis

HBMEC were lysed using lysis buffer (20 mM HEPES, 0.42 NaCl, 1.5 mMMgCl2, 0.2 mM EDTA, 0.5 mM DTT, 0.5 mM PMSF, 2% Nonidet P-40,

FIGURE 1. The effect of HIV-1 Gp120 and TNF-� on the fluid phasepermeability of HBMEC. A, HBMEC were cultured to confluence onTranswell polycarbonate membranes and were incubated with HIV-1Gp120 (10 ng/ml) or with TNF-� (10 ng/ml) for 6 and 24 h, as indicated.After 4 h of exposure to Lucifer Yellow (LY), the amount of LY in thelower chamber was measured. Control cells, untreated for each time point,were also used in measuring LY. The results are the mean � SD of threeexperiments. ��, p � 0.01 B, Monocyte transmigration across HBMECmonolayers. Cells were untreated () (control) or treated (�) with Gp120(10 ng/ml). For treatment with Abs, mAbs against Gp120 (�-Gp120; usedat 20 �g/ml), CXCR4 Ab (�-CXCR4), or isotype control Abs (�-Control)were used (at a concentration of 30 �g/ml). Abs were added to the HBMECmonolayers for 2 h before fresh isolated monocytes were added to theupper chamber. Monocytes transmigrated across the HBMEC werecounted (n 4 per condition). The results are the mean � SD of threeexperiments. ��, p � 0.01.

FIGURE 2. CB1 cannabinoid receptor expression analysis in HBMEC.A, RNA expression by RT-PCR analysis. Total RNA was extracted fromHBMEC. Using the Titanium One-Step RT-PCR kit (BD Biosciences), theprimer sequences for CB1 were 5�-GCCTGGCGGTGGCAGACCTCC-3�(sense) and 5�-GCAGCACGGCGATCACAATGG-3� (antisense). The sizeof the cDNA fragment obtained by RT-PCR was 278 bp for CB1. Lane M,m.w. standards. Lane 1, CB1 cannabinoid receptor; B, Protein expressionanalysis by Western blot (WB) assay of HBMEC. Total cell lysates wereobtained from HBMEC that were either untreated or treated for 4 h with thecannabinoid agonists THC (10 nM) or the endogenous cannabinoid agonist2-AG (10 nM). In some lanes, HBMEC were pretreated with the CB1specific inhibitor AM251 (4 h at 10 nM). Cell lysates were analyzed byWestern blotting with specific Abs against CB1 (Santa Cruz Biotechnol-ogy). For equal loading, we used GAPDH as control. C, Protein expressionanalysis of CB1 and CB2 by Western blot assay in HBMEC. Total celllysates obtained from untreated HBMEC or HBMEC treated with exoge-nous cannabinoid agonist THC (4 h; 10 nM) or CP55940 (1 h; 1 ng/ml) orwith the endogenous cannabinoid agonist 2-AG (1 h at 10 nM), asindicated.

Table I. Effect of Gp120 on SP secretiona

Treatment (pg/ml)SP Secretion

(pg/ml)

Control-HBMEC 163 � 25HBMEC plus Gp120 255 � 17HBMEC plus Gp120 plus control Ab 244 � 22HBMEC plus Gp120 plus Gp120 Ab 183 � 18HBMEC plus Gp120 plus spantide (SP antagonist) 166 � 32

a HBMEC (2.5 � 103 cells) were cultured in the presence of Gp120 (10 ng/ml for6 h) alone or GP120 with control Ab (20 �g/ml), Gp120 Ab (20 �g/ml), or the SPantagonist spantide (1 ng/ml). The secretion of SP in the supernatant was measuredusing a Substance P ELISA kit (Cayman Chemical). The results are the means ofdeterminations from three experiments (each in quadruple) � SD.

6408 PROTECTIVE EFFECTS OF CANNABINOIDS ON BRAIN ENDOTHELIUM

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

25% glycerol), at 4°C. The mixture was centrifuged at 12,000 rpm at 4°Cfor 30 min and the supernatant was analyzed by 10% SDS-PAGE. Theproteins on the gel were transferred to polyvinylidene difluoride mem-branes (NEN Life Science Products, Boston, MA) for 4 h. The polyvinyli-dene difluoride membrane was soaked in 5% nonfat milk-Tween 20-TBS(t-TBS) blocking solution containing anti-CB1 Ab”(Santa Cruz Biotech-nology) or anti-ZO-1, anti-claudin-5, anti-junctional adhesion molecule(JAM)-1, or GAPDH Abs (Chemicon) as the primary Ab. Actin (Chemi-con) was also used for comparison of protein content in the differentgroups. The membranes were then washed with t-TBS for 5 min four timesand incubated with a t-TBS solution containing goat anti-mouse IgG oranti-rabbit IgG (Amersham Pharmacia) as the secondary Ab for 2 h. Targetproteins were detected using ECL (ECL kit, PerkinElmer Life Sciences)and autographed with x-ray film (Fuji Film).

RT-PCR analysis of CB1 mRNA

Total RNA was extracted from HBMEC with TRIzol (Invitrogen) and CB1primers individually. One microgram of RNA was reverse-transcribed andfollowed by PCR using a Titanium One-Step RT-PCR kit (BD Bio-sciences). The primer sequences for CB1 were 5�-GCCTGGCGGTGGCAGACCTCC-3� (sense) and 5�-GCAGCACGGCGATCACAATGG-3� (an-tisense); PCR products were electrophoretically separated on 1.2% agarosegels and visualized by ethidium bromide staining. The size of the cDNAfragments obtained by RT-PCR was 278 bp for CB1.

For genotyping of CB1/ and CB2/ mice, DNA was extracted fromthe tails followed by PCR analysis using an Advantage 2 PCR kit (BDBiosciences).

Calcium ion concentration assay

HBMEC were treated with the calcium-sensitive dye fluo-3-acetoxymethylester (fluo-3-AM) for 1 h. Cells were then washed twice with PBS andanalyzed using confocal microscopy analysis to detect the free cytosoliccalcium concentration at 488 nm. SP (1 ng/ml), calcium ionophore A23187(105 M), and EDTA (5 � 103 M) were applied individually and calciumion concentration alterations in cultured cells were determined.

Immunostaining and quantitation of tight junctions

HBMEC were fixed with 3.7% paraformaldehyde in PBS containing 0.2%Triton X-100 (T-PBS) for 20 min at room temperature. The fixed cells werewashed three times with T-PBS for 10 min each, then soaked in 10% nonfatmilk-T-PBS blocking solution at room temperature for 1 h. Cells were nextincubated overnight at 4°C with a T-PBS solution containing one of thefollowing Abs: anti-ZO-1, claudin-5, or control isotype Abs. The next day,cells were washed with T-PBS and incubated with a T-PBS solution con-taining tetramethylrhodamine isothiocyanate (TRITC)-conjugated anti-mouse IgG (DakoCytomation) as the secondary Ab for 2 h at room tem-perature. Cells were then washed three times with T-PBS for 15 min each,and cover sealed with fluorescence mounting medium (DakoCytomation).

Target proteins were visualized using fluorescence microscopy (Leica).The relative staining along intercellular borders (presumed sites of junc-tional contacts) was determined by quantitative fluorescence analysis. Thenumbers of cells expressing the labeled markers along the entire lateral cellmembrane were then determined by counting. At least five different fieldsderived from three independent experiments were counted for each markerand time point. Background fluorescence of the assay medium was sub-tracted from all values. The quantitation of TJ in the samples was analyzedand compared with control untreated HBMEC. In addition, Western blotanalysis using specific Abs for ZO-1, claudin-5, and actin was performedto quantitate the down-regulation of TJ. The intensity of immunoblottedbands of ZO-1 and claudin-5 in cell lysates was compared with that of actinusing the Image J program. All samples were tested in triplicate.

TEER measurements

Restriction of paracellular transport of small ions was determined by anal-ysis of TEER. For each filter, the electrical resistance system with current-passing and voltage-measuring electrodes was used. TEER (ohm � cm2)was calculated from the display electrical resistance on the readout screenby subtraction of the electrical resistance of a blank filter and a correctionfor filter surface area. The resulting TEER values represented the resistance(“tightness”) of the endothelial cell monolayers. The relative ratio betweenthe TEER value for the HBMEC samples and the TEER value of controluntreated HBMEC was calculated for each sample.

In vivo permeability of the BBB by D-mannitol

BBB “opening” in mice was induced by 1.6 M D-Mannitol infusion asdescribed (43). This “opening” of the BBB was shown to be mediated bySP (44, 45). Damage to the BBB was judged by extravasation of Evansblue (EB) dye (43, 46). Fifteen minutes after 1.6 M D-Mannitol infusion,EB dye was injected i.v. (20 mg/kg body weight). After 1 h, animals wereanesthetized with an i.p. injection of 100 mg/kg body weight ketaminehydrochloride and were transcardially perfused with normal saline to clearblood vessels of the dye. The brains were then removed, the hemispheresseparated, EB was extracted, and its fluorescence was measured using afluorospectrophotometer at a wavelength of 620 nm. Calculations werebased on external standards in the solvent (100–500 ng/ml). Eluted EB dyewas expressed as grams per milligram of brain tissue.

Statistical analysis

The values for the permeability assay, endothelial migration assay, andtransendothelial electrical resistance are presented as the mean � SD. Theone-way ANOVA test was used to compare the means of three groups ofdata. If the one-way ANOVA test indicated an overall significant difference( p � 0.05), Tukey’s multiple comparison test was used to determine sig-nificant differences in the mean between any two groups. The independentt test was used to analyze data from the Western blot assay. Differenceswere considered to be significant at p � 0.05.

FIGURE 3. The effects of cannabinoids on calcium ion flux alterations following SP treatment of HBMEC. A, A calcium ion flux alteration assay wasperformed following SP treatment in HBMEC. a, The calcium ion concentration in the HBMEC control; b, the calcium ion concentration of HBMECfollowing treatment with SP (5 min at 5 ng/ml); c, the calcium ion concentration of HBMEC treated with CP55940 (1 h at 1 ng/ml); d, the calcium ionconcentration of HBMEC treated with CP55940 (1 h; 1 ng/ml) followed by SP treatment (5 min at 5 ng/ml). B, Measurement of Ca2� concentration ratioin HBMEC samples.

6409The Journal of Immunology

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

ResultsEffects of HIV-1 Gp120 on HBMEC permeability

To analyze the effects of HIV-1 Gp120 on HBMEC permeability,the fluid phase permeability changes of HBMEC were measuredby the colorimetric method. An increase in transcytosis was ob-served after the HIV-1 Gp120 treatment of HBMEC (Fig. 1A). Theincrease in transcytosis was similar to that obtained upon treatmentof HBMEC with TNF-�, a well-characterized and potent cytokineknown to induce transcytosis (Fig. 1A) (47). Heat-inactivatedGp120 had no significant effect on the permeability of HBMEC(data not shown).

Next, to investigate the involvement of HIV-1 Gp120 in thetransendothelial migration of monocytes, cells were treated for 6 hwith Gp120 (10 ng/ml) or treated with heat-inactivated Gp120 ascontrols, as described previously (25). Freshly isolated monocyteswere isolated as described (25) and were added to the upper com-partment of Transwell tissue culture inserts containing HBMECmonolayers and allowed to migrate for 2 h (37°C in 5% CO2). Themigrated monocytes in the lower chamber were stained with a

macrophage marker (CD68) and counted as described (25). Eachtreatment condition was performed in triplicate. As shown in Fig.1B, treatment of HBMEC with Gp120 significantly enhancedmonocyte transmigration. Furthermore, the Gp120-induced mono-cyte migration was inhibited by specific Abs for Gp120 (�70%)and by specific Abs for CXCR4, whereas control isotype Ab hadno inhibitory effects. No effects on migration were observed withthe heat-inactivated Gp120 protein (data not shown). Polymyxin Balso did not affect the Gp120-induced transmigration of mono-cytes, indicating that the Gp120 effect was not due to a contami-nation with LPS (data not shown). Migration in response to MCP-1(10 ng/ml) served as a positive control (data not shown) (48).

Gp120 enhances SP expression in HBMEC

Because HIV-1 Gp120 was reported to induce SP secretion (26–28), we then examined the expression of SP in HBMEC. WhenHBMEC were cultured in the presence of HIV-1 Gp120, thesecells secreted high levels of SP (Table I). Secretion of SP wasblocked by Gp120 Ab or by the SP antagonist spantide.

FIGURE 4. Immunocytochemistry analysis of ZO-1 and claudin-5 expression in HBMEC following HIV-1 Gp120 treatment and inhibition of TJdown-regulation by cannabinoid agonists. A and B, Analysis of TJ protein (ZO-1) (A) and claudin-5 (B) expression by immunocytochemistry staining inHBMEC. HBMEC were fixed, permeabilized, and stained with specific Abs against ZO-1 and claudin-5 (as indicated by the yellow arrow). HBMEC wereeither incubated with PBS or treated with Gp120 (10 ng/ml for 6 h). In some cultures, HBMEC were pretreated with the exogenous cannabinoid agonistCP55940 (1 h at 1 ng/ml), as indicated. This is a representative experiment of nine. C, Quantitative analysis of TJ in HBMEC. The tight junctions of thesamples were analyzed, quantitated, and compared with control untreated HBMEC. The TJ in control HBMEC were defined as 100%. Changes in TJinduced by Gp120 or Gp120 plus CP55940 (Gp120 � CP55940) were compared with control HBMEC and quantitated as the percentage of change intreated samples as compared with control HBMEC. Data are presented as the mean � SD of five experiments. �, p � 0.05; ��, p � 0.01. D, Effects ofHIV-1 Gp120 on JAM-1, claudin-5, and ZO-1 expression in HBMEC in the presence of cannabinoid agonists by Western blot (WB) assay. Proteinexpression in HBMEC was examined by Western blot analysis. Total cell lysates were obtained from either untreated HBMEC or HBMEC treated withGp120 (6 h at 10 ng/ml) in the presence or absence of the cannabinoid agonist CP55940 (1 h at 1 ng/ml). The cell lysates were prepared and analyzed byWestern blotting with the indicated specific Abs. This is a representative experiment of three. �, p � 0.05; ��, p � 0.01; ��a p value between controland Gp120 groups; ��b p value between Gp120 and Gp120 plus CP55940 groups.

6410 PROTECTIVE EFFECTS OF CANNABINOIDS ON BRAIN ENDOTHELIUM

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

Expression of cannabinoid receptors CB1 and CB2 in HBMEC

CB1 mRNA expression was found in HBMEC by RT-PCR anal-ysis (Fig. 2A). Western blot analysis showed positive expression ofCB1 protein in HBMEC and this expression was up-regulated inthe presence of the cannabinoid receptor ligands �9-tetrahydro-cannabinol (THC) and 2-AG or the potent synthetic cannabinoidreceptor agonist CP55940 (Fig. 2, B and C). However, in the pres-ence of the CB1-specific inhibitor AM251, the expression of CB1was significantly decreased in the THC-treated cells but not in2-AG-treated cells (Fig. 2B). AM251 is a specific inhibitor forCB1-based cannabinoid agonists (such as THC), but not CB2-based cannabinoids (such as 2-AG). The increased expression lev-els of CB1 within 1 h strongly suggest that upon binding of acannabinoid agonist to CB1 receptor, the cannabinoid agonist pre-vents CB1 degradation. The down-regulation of CB1 by HIV-1Gp120 was on the level of protein (translation level) and not on thelevel of mRNA (transcription level) (data not shown). In addition,we examined the expression of CB2 in HBMEC. CB2 expressionwas detected by RT-PCR analysis (data not shown) and was up-regulated by the 2-AG and CP55940 agonists, but not by THC,which has low binding affinity to CB2 (Fig. 2C).

Cannabinoid agonist CP55940 inhibited Gp120-mediatedcalcium influx in HBMEC

The effects of HIV-1 Gp120 and SP on Ca2� influx in HBMECmonolayer cultures was reported recently (25). Gp120 mediated itseffects on Ca2� influx through SP secretion. SP significantly in-creased the calcium ion concentration in the cocultures of the

HBMEC/astrocyte model, which was blocked by the cannabinoidagonist CP55940 (Fig. 3).

Effects of cannabinoids on HIV-1 Gp120 mediateddown-regulation of ZO-1 and claudin-5 expression in HBMEC

The TJ proteins ZO-1 and claudin-5 are central molecules in main-taining the integrity of the TJ (2, 3). To examine the changes in TJprotein expression in HBMEC following treatment with HIV-1Gp120, cocultures of HBMEC were immunostained with specificAbs against ZO-1 and claudin-5. The localization of ZO-1 (Fig.4A) and claudin-5 (Fig. 4B) at junctional regions between endo-thelial cells is consistent with the expression of high-resistance TJ,a hallmark of microvascular endothelial cells of the brain that de-fine the BBB. The staining patterns of ZO-1 and claudin-5 inHBMEC demonstrate the integrity of the HBMEC system in sig-nificantly retaining the localization of these TJ-associated proteins.Upon exposure to Gp120, the expression of both ZO-1 and clau-din-5 at interendothelial junctional regions was markedly altered incultured HBMEC (see color Fig. 4). When the time of Gp120exposure was extended to 8 h, the effects were more significant forboth ZO-1 and claudin-5 (data not shown). For immunostainingand quantitation of TJ, HBMEC were fixed and immunostainedwith anti-ZO-1, claudin-5, or control isotype Abs. TJ proteins werevisualized using fluorescence microscopy (Leica). The relativestaining along intercellular borders (presumed sites of junctionalcontacts) was determined by quantitative fluorescence analysis.The numbers of cells expressing the labeled markers along theentire lateral cell membrane were then determined by counting. At

FIGURE 5. Inhibition of permeability alteration of brain microvascular endothelial cells following HIV-1 Gp120 treatment in a human BBB model bycannabinoids. A–C, HIV-1 Gp120 (A and C; 10 ng/ml for 6 h) and SP (B; 1 ng/ml for 5 min) induced alterations in the permeability of HBMEC, as indicatedby TEER. HBMEC cultures were treated with cannabinoid agonists such as CP55940 and ACEA (106 M for 1 h) in the presence (�) or absence () ofcannabinoid antagonist CB1-specific inhibitor AM251 (1 h at 10 nM) or with SP inhibitor (1 h at 1 ng/ml) or URB597 (1 ng/ml for 1 h), as indicated. Fortreatments of cannabinoids plus Gp120, the cultures of HBMEC were treated with Gp120 (10 ng/ml for 6 h) plus cannabinoid compounds (as indicatedabove). The relative ratio of TEER in Gp120-treated samples as compared with untreated control HBMEC was measured for each sample; n 8. The pvalues with asterisk (�, p � 0.05; ��, p � 0.01) shows significant differences from specific groups. This is a representative experiment of eight. The valueof TEER with medium alone was 937.7 � 9.4. The results are expressed as mean percent of controls � SD. D, Inhibition of SP secretion by cannabinoidsin a human BBB model following Gp120 treatment. Cannabinoid compounds (ACEA and CP55940), in the presence (�) or absence () of CB1-specificinhibitor AM251, were added to HBMEC in BBB cocultures as indicated. Bars represent the mean � SD; n 3. The p values with asterisk (��, p � 0.01)show significant differences from the indicated specific groups. This is a representative experiment of three.

6411The Journal of Immunology

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

least 25 different fields derived from five independent experimentswere counted for each marker and time point.

Next, we examined the potential of cannabinoid agonists to pre-vent Gp120-mediated damage of TJ in HBMEC. Addition of thecannabinoid agonist CP55940 prevented the down-regulation ofZO-1 and claudin-5 in HBMEC following treatment with Gp120,leading to continuous uniform and linear TJ structures that pre-served the integrity of the TJ in the HBMEC (Fig. 4, A and B).Approximately 70% of the expression levels of ZO-1 and 90% ofthe expression levels of claudin-5 were observed in the presence ofCP55940 cannabinoid agonist, as compared with control samples(Fig. 4C).

Cannabinoids prevent decrease in ZO-1, JAM-1 and claudin-5expression in HBMEC following exposure to HIV-1 Gp120

Because the reduced intensity of immunostaining of ZO-1 andclaudin-5 at interendothelial junctions could be derived from al-terations in subcellular distribution and/or the expression level ofthese proteins, Western blotting of whole cell lysates was per-formed to assess changes in TJ protein content. Western blot anal-ysis using specific Abs for ZO-1, claudin-5, JAM-1, and actin wasperformed to quantitate the down-regulation of TJ. The intensity ofimmunoblotted bands of ZO-1 and claudin-5 in cell lysates wascompared with that of actin. HIV-1 Gp120 treatment of HBMEC

resulted in a decrease in claudin-5, JAM-1, and ZO-1 content (Fig.4D), whereas addition of the cannabinoid agonist CP55940 pre-vented the HIV-1 Gp120-induced down-regulation of ZO-1,JAM-1, and claudin-5 (Fig. 4D). The effects of Gp120 on down-regulation of TJ expression was on the level of translation (proteinlevel) and not on the level of mRNA (transcription level) (data notshown). Thus, CP55940 has protective effects on HBMEC by pre-serving TJ expression in these cells.

Effects of cannabinoid agonists on the inhibition of permeabilitychanges induced by Gp120 in the human BBB model systemusing TEER assay

The permeability of the BBB coculture model system was ana-lyzed by measuring TEER (49). The TEER values represented theresistance (“tightness”) of the endothelial cell. Both Gp120 and SPinduced significant changes in HBMEC permeability (Fig. 5,A–C). The CB1-specific agonist arachidonyl-2�-chloroethylamide(ACEA) and the cannabinoid ligand CP55940 inhibited the per-meability changes of HBMEC induced by Gp120 and restored theintegrity of the HBMEC as compared with the untreated HBMEC(Fig. 5A). Furthermore, the SP inhibitor, alone or together withACEA, blocked the SP-mediated damage of HBMEC (Fig. 5B).Next, we analyzed the effects of endocannabinoids on HBMECpermeability. Treatment with the FAAH inhibitor URB597, which

FIGURE 6. Cannabinoids restore on ZO-1 localization in HBMEC following exposure to Gp120. A, HBMEC were fixed, permeabilized, and stainedwith specific Abs against ZO-1 and CB1. HBMEC were treated with Gp120 (6 h at 10 ng/ml) alone () or together (�) with the cannabinoid agonistCP55940 (1 h at 1 ng/ml), as indicated. The merged figure indicates the protein expression of ZO-1 (yellow arrow) and CB1. B, ZO-1/CB1 colocalizationanalysis at randomly selected spots in the HBMEC membrane. The green curve indicates the expression of CB1 and the red curve indicates the expressionof ZO-1. This is a representative experiment of three. C, Association of CB1 and ZO-1 is increased in the presence of cannabinoids. Total cell lysates wereobtained from either untreated HBMEC or HBMEC treated with Gp120 (6 h at 10 ng/ml) in the presence (�) or absence () of the cannabinoid agonistCP55940 (1 h; 1 ng/ml). Total cell lysates were immunoprecipitated with control Ab, CB1 Ab, or ZO-1 Ab. The immunoprecipitates were then analyzedby SDS-PAGE and blotted with CB1 or ZO-1 Abs as indicated (IP, immunoprecipitation; WB, Western blot analysis). D, 293T cells were transfected withZO-1, GFP-ZO-1 N terminus or GFP-ZO-1 C terminus and/or CB1 cDNA constructs as indicated. Cell lysates were prepared and immunoprecipitated asshown. The immunocomplexes were analyzed by SDS-PAGE and Western blotting with the indicated specific Abs.

6412 PROTECTIVE EFFECTS OF CANNABINOIDS ON BRAIN ENDOTHELIUM

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

blocks anandamide’s metabolic degradation and results in accu-mulation of anandamide levels, was also able to inhibit HIV-1Gp120-mediated permeability changes of HBMEC (Fig. 5C). Fi-nally, HIV-1 Gp120 induced SP secretion up to �350 pg/ml,which was inhibited by both cannabinoid agonists, ACEA andCP55940 (Fig. 5D). These HIV-1 Gp120 -mediated effects on SPsecretion were modulated via the CB1 receptor (Fig. 5D), becausethe CB1-specific inhibitor AM251 prevented these protective ef-fects of cannabinoids.

Analysis of endothelial cell integrity and ZO-1 localizationfollowing HIV-1 Gp120 treatment in a BBB coculture modeland its modulation by cannabinoids

Endothelial integrity is regulated by the cytoskeleton and TJ. Thelinkage between actin and ZO forms the strong micro-frameworkof the endothelial cells. To examine the changes in ZO-1 TJ pro-tein expression in HBMEC following treatment with HIV-1Gp120, HBMEC were immunostained with specific Abs againstZO-1. As shown in color Fig. 6A, ZO-1 expression was detected inTJ in HBMEC but was significantly decreased in HBMEC treatedwith HIV-1 Gp120. Treatment with HIV-1 Gp120 resulted in dis-continuous uniform and linear structures of TJ, and the TJ werecompletely damaged, forming unconnected tight junctional struc-tures. However, addition of CP55940 prevented the down-regula-tion of ZO-1 expression in HBMEC following treatment withHIV-1 Gp120, leading to continuous uniform and linear TJ struc-tures that preserved the integrity of the TJ in HBMEC. Interest-ingly, significant colocalization of ZO-1 and CB1 in the TJ inHBMEC was induced following treatment with CP55940 and wasquantitated by confocal analysis in the Gp120 plus CP55940-treated cells, as compared with untreated cells and cells treatedwith Gp120 (Fig. 6B).

Cannabinoids increased the association of CB1 with ZO-1 inHBMEC

Because ZO-1 and CB1 were colocalized in the presence ofcannabinoid agonists (Fig. 6), we then analyzed the associationof CB1 with ZO-1. Endogenous association of ZO-1 with CB1was observed in HBMEC in the presence of the cannabinoidagonist CP55940 (Fig. 6C). The association of ZO-1 with CB1was mediated via the NH2 terminus of ZO-1 but not the Cterminus of ZO-1 (Fig. 6D). Treatment of HBMEC led to down-regulation of ZO-1 and CB1 expression. However, in the pres-ence of cannabinoid agonist, both CB1 and ZO-1 expressionwere preserved.

Inhibition of the transmigration of human monocytes across theBBB by cannabinoid agonists

To further investigate the protective effects of cannabinoid ago-nists on HBMEC permeability, we first examined the transmigra-tion of normal human monocytes across the HBMEC/astrocytecultures following exposure to HIV-1 Gp120 in the presence orabsence of cannabinoid agonist CP55940 (which binds to bothCB1 and CB2) or the CB1-specific agonist ACEA. Normal humanmonocytes were isolated as described (25). As shown in Fig. 7A,Gp120 induced the migration of monocytes that was inhibited byACEA and CP55940, suggesting that CB1-based cannabinoidcompounds may mediate their protective effects on brain micro-vascular endothelium and prevent monocyte transmigration fol-lowing exposure to HIV-1 Gp120.

Cannabinoid agonists inhibited BBB permeability in vivo

We next used an in vivo model to examine the potential of can-nabinoids in preventing BBB permeability changes. We used a

known method of high osmotic solution (1.6M D-mannitol) to in-crease the permeability of the BBB in mice using Evans blue dyeas a macromolecular marker (43, 46). This opening of the BBBwas shown to be mediated by SP (43, 46). BBB “opening” in micewas induced by 1.6 M D-mannitol infusion as described (43, 46).Damage to the BBB was judged by EB dye (43, 46). Fifteen min-utes after 1.6 M D-mannitol infusion, EB dye was injected i.v. (20mg/kg body weight). After 1 h, animals were anesthetized andtranscardially perfused with normal saline to clear blood vessels ofthe dye. The brains were then removed, the hemispheres separated,

FIGURE 7. A, Inhibition of monocyte migration across the BBB modelconsisting of cocultures of HBMEC and human astrocytes by cannabi-noids. HBMEC were treated with Gp120 (10 ng/ml) or with Gp120 (10ng/ml) and cannabinoid agonists (CP55940 or ACEA) for 4 h as indicated.Human monocytes (200,000 cells) were applied to the upper wells of thecoculture of the BBB. The cells in the lower compartments were collectedand counted after 2 h. The y-axis represents the number of monocytes thatmigrated across the BBB model. Bars represent the means � SD; n 4.The p values with asterisk (�, p � 0.05; ��, p � 0.01) show significantdifferences from specific groups. This is a representative experiment offive. B, Effects of cannabinoids on the integrity of the BBB in vivo. Theextravasation of EB dye in the brain was eluted and measured by a spec-trophotometer at 620 nm, following 1.6 M D-mannitol infusion in mice.The means � SD in each group were analyzed statistically by one-wayANOVA and Tukey’s multiple comparison test. The p values with asterisk(�, p � 0.05) show significant differences between the specific groups asindicated; n 8 per group per treatment. C, Immunohistochemistry anal-ysis of TJ in mice. Brain sections were fixed, permeabilized, and stainedwith specific Abs against ZO-1 and claudin-5 after D-mannitol treatment inmice. These are representative immunostainings of one sample of 10 dif-ferent brain sections.

6413The Journal of Immunology

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

and EB was extracted and measured using a fluorospectrophotom-eter at a wavelength of 620 nm. Calculations were based on ex-ternal standards in the solvent (100–500 ng/ml). Eluted EB dyewas expressed as grams per milligram of brain tissue. As shown inFig. 7B, the high osmotic solution induced permeability changes inthe BBB. These permeability changes were correlated with theexpression levels of ZO-1 and claudin-5 and to the damaged TJ inthe BBB of these animals. Both SP inhibitor and cannabinoid ag-onist ACEA significantly reduced the BBB permeability changesin vivo and inhibited the opening of the BBB in mice (Fig. 7B).Next, frozen brain sections from untreated mice and mice treatedwith D-mannitol (to induce lesions) were fixed, permeabilized, andstained with specific Abs against ZO-1 and claudin-5. ZO-1 ex-pression in the normal brain endothelium was more abundant andcontinuous in control mice, whereas down-regulation of both ZO-1and claudin-5 were observed in the lesioned areas of treated mice(Fig. 7C).

Taken together, these results demonstrate that cannabinoid ago-nists prevented down-regulation of ZO-1, JAM-1, and claudin-5expression in HBMEC following HIV-1 Gp120-mediated insult,resulting in inhibition of the permeability changes and sustainingthe integrity of BMEC and the BBB, using cocultures of HBMEC/astrocytes and in vivo mice.

DiscussionTo date, there is little known about how the BBB could be mod-ulated to help prevent the development of HAD. In addition, thereis no effective therapy to stop or prevent damage of the BBB dur-ing the HAD disease cascade. The search for interventions to pro-tect against destruction of the BBB might help treat diseases thatfeature BBB damage. Using a human BBB model involving co-cultures of HBMEC and human astrocytes as well as an in vivomouse model, our results and those of others (30) showed that thesecreted HIV-1 protein Gp120, which plays a major role in HAD,induced SP secretion and increased BBB permeability. More spe-

cifically, both Gp120 and SP decreased ZO-1 and claudin-5 proteinexpression, leading to increased BBB permeability (Figs. 4 and 5).These permeability changes were prevented by the cannabinoidagonist CP55940 and the CB1-specific agonist ACEA (Figs. 4–6).These cannabinoids prevented the down-regulation of ZO-1 andclaudin-5 protein expression in HBMEC following Gp120 or SP-induced damage in HBMEC. Thus, cannabinoids are neuroprotec-tive compounds that prevent the degradation of CB1 and the down-regulation of ZO-1 and claudin-5 TJ protein expression inHBMEC following insult, resulting in preservation of endothelialcell structure and function.

The endocannabinoid 2-AG was shown to have neuron-protec-tive properties in animal models of ischemic brain injury (33).Endocannabinoids are involved in neuroprotection through numer-ous biochemical pathways. The endocannabinoid 2-AG was re-leased in mouse brain after closed head injury, and treatment withexogenous 2-AG exerts neuroprotection via the central cannabi-noid receptor CB1 (33). This process involves inhibition of in-flammatory signals that were mediated by activation of the tran-scription factor NF-�B. 2-AG decreased BBB permeability andinhibited the acute expression of the main proinflammatory cyto-kines TNF-�, IL-1�, and IL-6. It also augmented the levels ofendogenous antioxidants (33). In this study, we found thatCP55940 and ACEA protected HBMEC against HIV-1 Gp120-related toxicity through a different mechanism. Addition of can-nabinoid agonists such as ACEA and CP55940 inhibited SP se-cretion (Table I) and prevented the degradation of ZO-1 andclaudin-5 in HBMEC following treatment with either HIV-1Gp120 (Fig. 4) or SP, leading to continuous uniform and linear TJstructures that preserved the integrity of the TJ in the HBMEC(Fig. 4). The down-regulation of ZO-1 and claudin-5 was mediatedby SP directly. However, the possibility that this down-regulationis also mediated by proinflammatory cytokines induced by SP,such as TNF-�, IL-6 and IL-1�, needs to be investigated in futurestudies. The current finding suggests that the mechanism by which

FIGURE 8. Proposed mechanismsby which Gp120 affects the CNS.

6414 PROTECTIVE EFFECTS OF CANNABINOIDS ON BRAIN ENDOTHELIUM

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

cannabinoids exert their effects on the BBB may involve inhibitionof the early induction of SP by Gp120 through preservation of theTJ in HBMEC. These protective effects of cannabinoids are me-diated through CB1 receptors, but not CB2 receptors, becauseAM251 (the specific inhibitor for CB1), but not AM630 (CB2-specific inhibitor) (data not shown), specifically blocked these pro-tective effects on the HBMEC.

Circulating HIV-1 Gp120 might bind to anti-HIV-1 envelopeAbs synthesized during early infection or may act on the BBB byaltering the permeability of the brain microvascular endothelium,thus promoting the entry of HIV-1 into the CNS. This processmight account for the early CNS involvement before overt neuro-logical findings. Other proposed mechanisms, such as recruitmentof HIV-1-infected monocytes mediated by increased adhesionmolecule expression on the brain vessels, might be involved in thelater stages of infection and amplify the CNS invasion (50). SP, apotent modulator of neuroimmunoregulation, and its receptor, NK-1R, may be involved in the modulation of HIV-1 infection, both invivo and in vitro (26–29, 51) HIV-1-positive children have higherplasma levels of SP compared with HIV-1-negative children (51).SP was also found to play a critical role in the HIV-1 Gp120-induced increase in permeability of the rat brain microvascularendothelium, and this effect of SP was abrogated by the SP antag-onist (spantide) and/or by anti-SP polyclonal Ab (52). In addition,significant SP immunoreactivity was observed in HIV-1 Gp120transgenic mouse brain vessels, suggesting that SP is involved inHIV-1 Gp120-induced changes in the vascular component of theBBB (52). In the present study, we demonstrate that HIV-1 Gp120caused SP secretion in HBMEC (Table I) and that both HIV-1Gp120 and SP resulted in significant permeability changes in thesecells, using a human BBB culture model that enables selectivestudy of the physiology, pharmacology, and pathophysiology ofthe BBB. Thus, HIV-1 Gp120, through induction of SP, plays akey role in the disruption of the BBB and facilitates the transmi-gration of monocytes and leukocytes into the brain (Fig. 7A). Theincreased transmigration was shown to be directly correlated withincreased BBB disruption. In addition, using an in vivo micemodel, we showed that cannabinoid agonist inhibited permeabilitychanges induced by D-Mannitol through SP induction (Fig. 7B).

The TJ play a central role in sealing the intercellular space inadjacent endothelial cellular sheets (2, 3). Through these “barrier”and “fence” functions of TJ, endothelial cellular sheets establishvarious compositionally distinct fluid compartments. Therefore, TJare considered to be fundamental structures in multicellular organ-isms (4). HIV-1 Gp120 was shown to compromise BBB integrityand enhance monocyte migration across the BBB (49). HIV-1Gp120 was reported to cause disruption and down-regulation ofZO-1, ZO-2, occludin, and claudin-5 (50). Based on our results,both HIV-1 Gp120 and SP induced changes in ZO-1 and claudin-5TJ expression and the TJ were damaged, resulting in unconnectedtight junctional structures (Fig. 4).

Specific association of ZO-1 with CB1 through the NH2 termi-nus of ZO-1 was observed in the presence of cannabinoid agonists(Fig. 6). As shown in Fig. 6A, Gp120 induced the down-regulationof ZO-1 and CB1 expression in HBMEC. However, in the pres-ence of cannabinoid ligands, CB1 binds to its cognate cannabinoidagonist, which prevents CB1 degradation. This “activated” CB1 isthen able to bind ZO-1 and form a complex of ZO-1/CB1 inHBMEC, strongly suggesting that CB1, through its interactionwith ZO-1, protects ZO-1 from HIV-1 Gp120-mediated degrada-tion. In conclusion, the present study describes the mechanisms ofHIV-1 Gp120-induced permeability changes in HBMEC, throughSP, by down-regulation of ZO-1 and claudin-5 protein expression.Furthermore, we show that cannabinoid agonists are protective

agents that prevent this down-regulation and thus preserve the ex-pression of ZO-1 and claudin-5 in the BBB, resulting in sustainedBBB integrity. In addition, using an in vivo mouse model of BBBopening by D-mannitol through SP modulation (44, 45), we ob-served significant down-regulation of both ZO-1 and claudin-5 inthe lesion in vivo (Fig. 7C). Furthermore, we showed that thecannabinoid agonist ACEA as well as an SP inhibitor (Spantide)alone or in combination, significantly inhibited BBB permeabilityin vivo and prevented Gp120-induced down-regulation of TJ ex-pression in HBMEC (Fig. 7B). These results provide importantinformation on the therapeutic potential of cannabinoids and/or SPinhibitors to inhibit the transmigration of HIV-1-infected cellsacross HBMEC and to prevent damage of HBMEC following in-sult by HIV-1 secreted proteins such as HIV-1 Gp120 (Fig. 8).These studies should provide insights into preventive and/or ther-apeutic strategies to sustain BBB integrity in patients who devel-oped HAD, based on SP inhibitors and/or cannabinoidpharmacotherapies.

AcknowledgmentsWe thank Chiung-Pei Wang for typing assistance and Janet Delahanty andMakara Men for editing the manuscript. We thank Xin Li for technicalassistance. We thank Dr. Rick Rogers (Lab Imaging, Harvard MedicalSchool, Boston, MA) for help in immunostaining studies of tight junctions.

DisclosuresThe authors have no financial conflict of interest.

References1. Hawkins, B. T., and T. P. Davis. 2005. The blood-brain barrier/neurovascular unit

in health and disease. Pharmacol. Rev. 57: 173–185.2. Harhaj, N. S., and D. A. Antonetti. 2004. Regulation of tight junctions and loss

of barrier function in pathophysiology. Int. J. Biochem. Cell Biol. 36: 1206–1237.3. Wolburg, H., and A. Lippoldt. 2002. Tight junctions of the blood-brain barrier:

development, composition and regulation. Vascul. Pharmacol. 38: 323–337.4. Pardridge, W. M. 1983. Brain metabolism: a perspective from the blood-brain

barrier. Physiol. Rev. 63: 1481–1535.5. Morita, K., H. Sasaki, M. Furuse, and S. Tsukita. 1999. Endothelial claudin:

claudin-5/TMVCF constitutes tight junction strands in endothelial cells. J. CellBiol. 147: 185–194.

6. Musch, M. W., M. M. Walsh-Reitz, and E. B. Chang. 2006. Roles of ZO-1,occludin, and actin in oxidant-induced barrier disruption. Am. J. Physiol. 290:G222–G231.

7. Schneeberger, E. E., and R. D. Lynch. 2004. The tight junction: a multifunctionalcomplex. Am. J. Physiol. 286: C1213–C1228.

8. Bradbury, M. W. 1993. The blood-brain barrier. Exp. Physiol. 78: 453–472.9. Bradbury, M. W., and R. Deane. 1993. Permeability of the blood-brain barrier to

lead. Neurotoxicology 14: 131–136.10. Umeda, K., J. Ikenouchi, S. Katahira-Tayama, K. Furuse, H. Sasaki,

M. Nakayama, T. Matsui, S. Tsukita, and M. Furuse. 2006. ZO-1 and ZO-2independently determine where claudins are polymerized in tight-junction strandformation. Cell 126: 741–754.

11. Liebner, S., U. Cavallaro, and E. Dejana. 2006. The multiple languages of en-dothelial cell-to-cell communication. Arterioscler. Thromb. Vasc. Biol. 26:1431–1438.

12. Lee, J. F., Q. Zeng, H. Ozaki, L. Wang, A. R. Hand, T. Hla, E. Wang, andM. J. Lee. 2006. Dual roles of tight junction-associated protein, zonula occlu-dens-1, in sphingosine 1-phosphate-mediated endothelial chemotaxis and barrierintegrity. J. Biol. Chem. 281: 29190–29200.

13. Katsuno, T., K. Umeda, T. Matsui, M. Hata, A. Tamura, M. Itoh, K. Takeuchi,T. Fujimori, Y. I. Nabeshima, T. Noda, and S. Tsukita. 2008. Deficiency of ZO-1causes embryonic lethal phenotype associated with defected yolk sac angiogen-esis and apoptosis of embryonic cells. Mol. Biol. Cell 19: 2465–2475.

14. Toborek, M., Y. W. Lee, G. Flora, H. Pu, I. E. Andras, E. Wylegala, B. Hennig,and A. Nath. 2005. Mechanisms of the blood-brain barrier disruption in HIV-1infection. Cell. Mol. Neurobiol. 25: 181–199.

15. Banks, W. A., N. Ercal, and T. O. Price. 2006. The blood-brain barrier in neu-roAIDS. Curr. HIV Res. 4: 259–266.

16. Marshall, D. W., R. L. Brey, C. A. Butzin, D. R. Lucey, S. M. Abbadessa, andR. N. Boswell. 1991. CSF changes in a longitudinal study of 124 neurologicallynormal HIV-1-infected U.S. Air Force personnel. J. Acquired Immune Defic.Syndr. 4: 777–781.

17. Singer, E. J., K. Syndulko, B. Fahy-Chandon, P. Schmid, A. Conrad, andW. W. Tourtellotte. 1994. Intrathecal IgG synthesis and albumin leakage areincreased in subjects with HIV-1 neurologic disease. J. Acquired Immune Defic.Syndr. 7: 265–271.

6415The Journal of Immunology

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from

18. Andersson, L. M., L. Hagberg, D. Fuchs, B. Svennerholm, and M. Gisslen. 2001.Increased blood-brain barrier permeability in neuro-asymptomatic HIV-1-in-fected individuals–correlation with cerebrospinal fluid HIV-1 RNA and neopterinlevels. J. Neurovirol. 7: 542–547.

19. Dohgu, S., and W. A. Banks. 2008. Lipopolysaccharide-enhanced transcellulartransport of HIV-1 across the blood-brain barrier is mediated by the p38 mitogen-activated protein kinase pathway. Exp. Neurol. 210: 740–749.

20. Oh, S. K., W. W. Cruikshank, J. Raina, G. C. Blanchard, W. H. Adler, J. Walker,and H. Kornfeld. 1992. Identification of HIV-1 envelope glycoprotein in theserum of AIDS and ARC patients. J. Acquired Immune Defic. Syndr. 5: 251–256.

21. Gilbert, M., J. Kirihara, and J. Mills. 1991. Enzyme-linked immunoassay forhuman immunodeficiency virus type 1 envelope glycoprotein 120. J. Clin. Mi-crobiol. 29: 142–147.

22. Jones, M. V., J. E. Bell, and A. Nath. 2000. Immunolocalization of HIV envelopegp120 in HIV encephalitis with dementia. AIDS 14: 2709–2713.

23. Toggas, S. M., E. Masliah, E. M. Rockenstein, G. F. Rall, C. R. Abraham, andL. Mucke. 1994. Central nervous system damage produced by expression of theHIV-1 coat protein gp120 in transgenic mice. Nature 367: 188–193.

24. Sun, X., and S. K. Dey. 2008. Aspects of endocannabinoid signaling in periim-plantation biology. Mol. Cell. Endocrinol. 286: S3–S11.

25. Kanmogne, G. D., K. Schall, J. Leibhart, B. Knipe, H. E. Gendelman, andY. Persidsky. 2007. HIV-1 gp120 compromises blood-brain barrier integrity andenhances monocyte migration across blood-brain barrier: implication for viralneuropathogenesis. J. Cereb. Blood Flow Metab. 27: 123–134.

26. Ho, W. Z., and S. D. Douglas. 2004. Substance P and neurokinin-1 receptormodulation of HIV. J. Neuroimmunol. 157: 48–55.

27. Ho, W. Z., D. L. Evans, and S. D. Douglas. 2002. Substance P and humanimmunodeficiency virus infection: psychoneuroimmunology. CNS Spectr. 7:867–874.

28. Ho, W. Z., J. P. Lai, Y. Li, and S. D. Douglas. 2002. HIV enhances substance Pexpression in human immune cells. FASEB J. 16: 616–618.

29. Li, Y., S. D. Douglas, L. Song, S. Sun, and W. Z. Ho. 2001. Substance P enhancesHIV-1 replication in latently infected human immune cells. J. Neuroimmunol.121: 67–75.

30. Annunziata, P., C. Cioni, R. Santonini, and E. Paccagnini. 2002. Substance Pantagonist blocks leakage and reduces activation of cytokine-stimulated rat brainendothelium. J. Neuroimmunol. 131: 41–49.

31. Pacher, P., S. Batkai, and G. Kunos. 2006. The endocannabinoid system as anemerging target of pharmacotherapy. Pharmacol. Rev. 58: 389–462.

32. Croxford, J. L. 2003. Therapeutic potential of cannabinoids in CNS disease. CNSDrugs 17: 179–202.

33. Panikashvili, D., N. A. Shein, R. Mechoulam, V. Trembovler, R. Kohen,A. Alexandrovich, and E. Shohami. 2006. The endocannabinoid 2-AG protectsthe blood-brain barrier after closed head injury and inhibits mRNA expression ofproinflammatory cytokines. Neurobiol. Dis. 22: 257–264.

34. Marsicano, G., B. Moosmann, H. Hermann, B. Lutz, and C. Behl. 2002. Neuro-protective properties of cannabinoids against oxidative stress: role of the canna-binoid receptor CB1. J. Neurochem. 80: 448–456.

35. Kim, J. A., N. D. Tran, S. J. Wang, and M. J. Fisher. 2003. Astrocyte regulationof human brain capillary endothelial fibrinolysis. Thromb. Res. 112: 159–165.

36. Eugenin, E. A., K. Osiecki, L. Lopez, H. Goldstein, T. M. Calderon, andJ. W. Berman. 2006. CCL2/monocyte chemoattractant protein-1 mediates en-hanced transmigration of human immunodeficiency virus (HIV)-infected leuko-cytes across the blood-brain barrier: a potential mechanism of HIV-CNS invasionand neuroAIDS. J. Neurosci. 26: 1098–1106.

37. Schiera, G., S. Sala, A. Gallo, M. P. Raffa, G. L. Pitarresi, G. Savettieri, andI. Di Liegro. 2005. Permeability properties of a three-cell type in vitro model ofblood-brain barrier. J. Cell. Mol. Med. 9: 373–379.

38. Gaillard, P. J., L. H. Voorwinden, J. L. Nielsen, A. Ivanov, R. Atsumi,H. Engman, C. Ringbom, A. G. de Boer, and D. D. Breimer. 2001. Establishmentand functional characterization of an in vitro model of the blood-brain barrier,comprising a co-culture of brain capillary endothelial cells and astrocytes. Eur.J. Pharm. Sci. 12: 215–222.

39. Garcia, C. M., D. C. Darland, L. J. Massingham, and P. A. D’Amore. 2004.Endothelial cell-astrocyte interactions and TGF � are required for induction ofblood-neural barrier properties. Brain Res. Dev. Brain Res. 152: 25–38.

40. Lee, T. H., S. Seng, H. Li, S. J. Kennel, H. K. Avraham, and S. Avraham. 2006.Integrin regulation by vascular endothelial growth factor in human brain micro-vascular endothelial cells: role of �6�1 integrin in angiogenesis. J. Biol. Chem.281: 40450–40460.

41. Lee, T. H., H. K. Avraham, S. Jiang, and S. Avraham. 2003. Vascular endothelialgrowth factor modulates the transendothelial migration of MDA-MB-231 breastcancer cells through regulation of brain microvascular endothelial cell perme-ability. J. Biol. Chem. 278: 5277–5284.

42. Lee, T. H., H. Avraham, S. H. Lee, and S. Avraham. 2002. Vascular endothelialgrowth factor modulates neutrophil transendothelial migration via up-regulationof interleukin-8 in human brain microvascular endothelial cells. J. Biol. Chem.277: 10445–10451.

43. Lu, T. S., H. W. Chen, M. H. Huang, S. J. Wang, and R. C. Yang. 2004. Heatshock treatment protects osmotic stress-induced dysfunction of the blood-brainbarrier through preservation of tight junction proteins. Cell Stress Chaperones 9:369–377.

44. Garland, A., J. E. Jordan, J. Necheles, L. E. Alger, M. M. Scully, R. J. Miller,D. W. Ray, S. R. White, and J. Solway. 1995. Hypertonicity, but not hypother-mia, elicits substance P release from rat C-fiber neurons in primary culture.J. Clin. Invest. 95: 2359–2366.

45. Wang, M., J. Etu, and S. Joshi. 2007. Enhanced disruption of the blood brainbarrier by intracarotid mannitol injection during transient cerebral hypoperfusionin rabbits. J. Neurosurg. Anesthesiol. 19: 249–256.

46. Udaka, K., Y. Takeuchi, and H. Z. Movat. 1970. Simple method for quantitationof enhanced vascular permeability. Proc. Soc. Exp. Biol. Med. 133: 1384–1387.

47. Fiala, M., D. J. Looney, M. Stins, D. D. Way, L. Zhang, X. Gan, F. Chiappelli,E. S. Schweitzer, P. Shapshak, M. Weinand, et al. 1997. TNF-� opens a para-cellular route for HIV-1 invasion across the blood-brain barrier. Mol. Med. 3:553–564.

48. Stamatovic, S. M., P. Shakui, R. F. Keep, B. B. Moore, S. L. Kunkel,N. Van Rooijen, and A. V. Andjelkovic. 2005. Monocyte chemoattractant pro-tein-1 regulation of blood-brain barrier permeability. J. Cereb. Blood FlowMetab. 25: 593–606.

49. Ricardo-Dukelow, M., I. Kadiu, W. Rozek, J. Schlautman, Y. Persidsky,P. Ciborowski, G. D. Kanmogne, and H. E. Gendelman. 2007. HIV-1 infectedmonocyte-derived macrophages affect the human brain microvascular endothelialcell proteome: new insights into blood-brain barrier dysfunction for HIV-1-as-sociated dementia. J. Neuroimmunol. 185: 37–46.

50. Chaudhuri, A., F. Duan, B. Morsey, Y. Persidsky, and G. D. Kanmogne. 2008.HIV-1 activates proinflammatory and interferon-inducible genes in human brainmicrovascular endothelial cells: putative mechanisms of blood-brain barrier dys-function. J. Cereb. Blood Flow Metab. 28: 697–711.

51. Azzari, C., M. E. Rossi, M. Resti, A. L. Caldini, L. Lega, L. Galli, E. Fico, andA. Vierucci. 1992. Changed levels of substance P and somatostatin in HIV-positive children. Pediatr. Med. Chir. 14: 577–581.

52. Toneatto, S., O. Finco, H. van der Putten, S. Abrignani, and P. Annunziata. 1999.Evidence of blood-brain barrier alteration and activation in HIV-1 gp120 trans-genic mice. AIDS 13: 2343–2348.

6416 PROTECTIVE EFFECTS OF CANNABINOIDS ON BRAIN ENDOTHELIUM

by guest on June 20, 2018http://w

ww

.jimm

unol.org/D

ownloaded from