Lysophosphatidylcholine increases endothelial permeability: role of PKC�

and RhoA cross talk

Fei Huang,1 Papasani V. Subbaiah,2 Oksana Holian,3 Jihang Zhang,1

Arnold Johnson,4 Nancy Gertzberg,4 and Hazel Lum1

1Department of Pharmacology, Rush University Medical Center, Chicago; 2Department of Medicine,University of Illinois, Chicago; 3Department of Medicine, John H. Stroger Hospital of Cook County,Chicago, Illinois; and 4Research Service, Stratton Veterans Affairs Medical Center, Albany, New York

Submitted 4 January 2005; accepted in final form 4 March 2005

Huang, Fei, Papasani V. Subbaiah, Oksana Holian, JihangZhang, Arnold Johnson, Nancy Gertzberg, and Hazel Lum.Lysophosphatidylcholine increases endothelial permeability: roleof PKC� and RhoA cross talk. Am J Physiol Lung Cell Mol Physiol289: L176 –L185, 2005. First published March 11, 2005;doi:10.1152/ajplung.00003.2005.—Lysophosphatidylcholine (LPC)is a bioactive proinflammatory lipid that can be generated by patho-logical activities. We investigated the hypothesis that LPC signalsincrease in endothelial permeability. Stimulation of human dermalmicrovascular endothelial cells and bovine pulmonary microvascularendothelial cells with LPC (10–50 �M) induced decreases (withinminutes) in transendothelial electrical resistance and increase ofendothelial permeability. LPC activated (within 5 min) membrane-associated PKC phosphotransferase activity in the absence of trans-location. Affinity-binding analysis indicated that LPC induced in-creases (also by 5 min) of GTP-bound RhoA, but not Rac1 or Cdc42.By 60 min, both signaling pathways decreased toward baseline.Inhibition of RhoA with C3 transferase inhibited �50% of LPC-induced resistance decrease. Pretreatment with PKC inhibitor Go-6983 (concentrations selective for classic PKC), PMA-induced deple-tion of PKC�, and transfection of antisense PKC� oligonucleotideeach prevented 40–50% of the LPC-induced resistance decrease.Furthermore, these three PKC inhibition strategies inhibited 60–80%of the LPC-induced GTP-bound RhoA. These results show that LPCdirectly impairs the endothelial barrier function that was dependent, atleast in part, on cross talk of PKC� and RhoA signals. The evidenceindicates that elevated LPC levels can contribute to the activation ofa proinflammatory endothelial phenotype.

protein kinase C; signal transduction

LYSOPHOSPHATIDYLCHOLINE (LPC) belongs to a group of bioac-tive glycerol- or sphingosine-based lysophospholipids [i.e.,lysophosphatidic acid, sphingosine-1-phosphate, and sphingo-sylphosphorylcholine (SPC)] generated from membrane phos-pholipids as part of normal physiological activities or diseaseprocesses. It has been found to accumulate in pathologicaltissues such as in the ischemic myocardium, atheroscleroticaortas, and other inflammatory lesions of blood vessels (5, 33).LPC is also a major phospholipid component (40–50%) ofoxidized LDL (17) and is implicated as a critical atherogenicfactor of oxidized LDL. Several diseases such as endometriosis(22), asthma (18), and ovarian cancer (26) are associated withtwo- to threefold increased circulating levels of LPC. It is believedthat a primary source of pathological levels of LPC is through theaction of phospholipase A2 (PLA2) on membrane phosphatidyl-

choline, generating LPC concomitantly with arachidonic acid (2,16, 32, 35). Despite these documented elevations of LPC inassociation with pathological conditions, the pathophysiologicalrole of LPC in diseases remains to be established.

There is clear evidence that the bioactive activities ofLPC include activation of vascular endothelium. For exam-ple, extracellular LPC (10 –100 �M) is reported to upregu-late expression of adhesion molecules (8, 21, 40), produc-tion of cytokines (23), secretion of O2

� (7), and DNA-binding activity of NF-�B (34). In vivo studies show thatdirect LPC injection either subcutaneously (31) or into thespinal cord (27) causes inflammatory cell infiltration accom-panied by vascular leakage. Overall, evidence indicates thatLPC induces a significant proinflammatory endothelial phe-notype. Yet, to date, the effects of LPC on vascular endo-thelial permeability have not been systematically investi-gated, and the mechanisms are unknown. Therefore, the goalsof this study are to determine whether LPC induces endothelialbarrier dysfunction and to identify the signaling mechanismsresponsible for this regulation.

The critical findings are that LPC 1) impaired endothelialbarrier function that was sensitive to albumin concentration, 2)increased PKC phosphotransferase activity in the absence oftranslocation of either PKC� or PKC�, 3) activated RhoA, butnot Rac1 or Cdc42, and 4) induced barrier dysfunction that wasdependent in part on PKC� and RhoA cross talk. Thesefindings provide the first report that proinflammatory LPCdirectly increases endothelial permeability through activationof signaling mechanisms.

METHODS

Preparation of LPC

LPC (1-palmitoyl-2-hydroxy-sn-glycero-3-phosphocholine) waspurchased from Avanti Polar Lipids (Alabaster, AL), checked for fattyacid composition by gas-liquid chromatography, and found to be atleast 96% pure. It was dissolved in chloroform:methanol (2:1) andstored at �20°C. Aliquots were evaporated under nitrogen in glasstubes and resuspended in sufficient volume of Hanks’ balanced saltsolution to give a final concentration of 1 mM. The samples werevortexed at room temperature for 1 min (2�) to yield a cleardispersion, and the final concentration was confirmed by analysis oflipid phosphorus by the modified Bartlett procedure (15). The phos-pholipid dispersions were stored at 4°C and were used within 30 daysof the preparation.

Address for reprint requests and other correspondence: H. Lum, Dept. ofPharmacology, Rush Univ. Medical Center, 1735 W. Harrison St., CohnResearch Bldg., Rm. 416, Chicago, IL 60612 (e-mail: hlum@rush.edu).

The costs of publication of this article were defrayed in part by the paymentof page charges. The article must therefore be hereby marked “advertisement”in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Am J Physiol Lung Cell Mol Physiol 289: L176–L185, 2005.First published March 11, 2005; doi:10.1152/ajplung.00003.2005.

http://www.ajplung.orgL176

Cell Culture

Human dermal microvascular endothelial cells (HMEC) (1) weremaintained in culture in MCDB-131 medium (GIBCO BRL, Gaith-ersburg, MD) and supplemented with 5% fetal bovine serum (FBS;Hyclone, Logan, UT), 10 �g/l human epidermal growth factor (SigmaChemical, St. Louis, MO), 1 mg/l hydrocortisone, 1% penicillin-streptomycin, and 1% L-glutamine. HMEC is a well-characterized cellline that exhibits the expected morphological and functional endothe-lial phenotypes, expresses and secretes von Willebrand factor, takesup acetylated LDL, forms tubes when grown in Matrigel, and ex-presses CD31, CD36, ICAM-1, and CD44 (1). HMEC were passaged3–4 days when confluent and used for studies at passages 25–35.These cells form a highly restrictive barrier when evaluated byalbumin permeability and resistance (11, 12).

Bovine pulmonary microvessel endothelial cells (BPMEC) derivedfrom fresh calf lungs were obtained at passage 4 (VecTechnologies,Rensselaer, NY) and serially cultured from 4 to 12 passages forstudies (4, 11). The culture medium contained DMEM (GIBCO BRL;Grand Island, NY) supplemented with 20% FBS, 15 �g/ml endothe-lial cell growth supplement (Upstate Biotechnology, Lake Placid,NY), and 1% nonessential amino acids (GIBCO BRL). The BPMECwere identified as endothelial monolayers by 1) the characteristic“cobblestone” appearance using contrast microscopy, 2) the presenceof factor VIII-related antigen, 3) the uptake of acylated LDL, and 4)the absence of smooth muscle actin.

Transendothelial Electrical Resistance

The transendothelial electrical resistance was determined in realtime using the electric cell-substrate impedance sensor system (Ap-plied BioPhysics, Troy, NY) that measures electrical current flowthrough cells grown onto gold-plated electrodes (5 � 10�4 cm2) witha 500-�l well fitted above each electrode. For study, endothelial cells(2.5 � 105 cells/cm2) were grown to confluence on sterile, fibronec-tin-coated electrodes, and resistance was measured as previouslydescribed by us (11) and others (20). Initial baseline resistancetypically showed resistance �7,000 (11), and monolayers withlower resistances were rejected from study. We used the minimumbaseline electrical resistance of 7,000 (corresponds to the calculatedvalue of 3.5 cm2) to screen for relatively restrictive monolayerssince this resistance is within the reported range of values of 3–6 cm2 for endothelial cell monolayers. The endothelial cells werechallenged with reagents according to experimental protocol, andresistance was recorded continuously for up to 4 h. Reported valueswere normalized to initial baseline resistance of each monolayer.

Endothelial Permeability Assay

The permeability assay system measures the transendothelial clear-ance rate of albumin in the absence of hydrostatic and oncoticpressure gradients and has been described in previous studies (4, 12).In brief, the system consisted of two compartments separated by asterile gelatinized microporous polycarbonate filter (13-mm diameter,0.8-�m pore size; Corning Costar; Cambridge, MA) as previouslydescribed (12). BPMEC (2 � 105) were seeded onto the filters andcultured to confluence (3–5 days). The luminal (upper) compartment(0.7 ml) was suspended in the abluminal (lower) compartment (25ml), both of which contained Hanks’ balanced salt solution, 0.5%BSA, and 20 mM HEPES buffer. The lower chamber was stirredcontinuously for complete mixing, and the entire system was kept ina water bath at 37°C. Evans blue-labeled albumin (20 mg/ml) wasadded to the upper chamber, 300-�l samples were taken from thelower compartment at 5-min intervals for 60 min, and absorbance wasread at 620 nm. At the beginning of each study, an upper chambersample was diluted 1:100 to determine the initial absorbance of theEvans blue-albumin. The absorbance of free Evans blue in the luminaland abluminal compartments was always 1%, indicating that most

of albumin was bound by the dye. The clearance rate of Evansblue-albumin was determined by least-squares linear regression be-tween 10 and 60 min for the control and experimental groups.

PKC Phosphotransferase Activity

In vitro PKC activity was determined from cell fractions as previ-ously reported (13). In brief, HMEC were grown in 60-mm culturedishes and treated according to experimental protocol. All subsequentsteps were carried out on ice using ice-cold reagents. The cells werewashed with PBS, collected in extraction buffer (0.02 M Tris, 0.5 mMEDTA, 0.5 mM EGTA) containing protease inhibitors (2.5 mMPMSF, 25 �g/ml pepstatin A, 25 �g/ml leupeptin, 25 �g/ml aprotinin)and 10 mM 2-mercaptoethanol, and cytosolic and membrane fractionswere prepared. The phosphotransferase activity of cell fractions wasdetermined by incorporation of 32P into the PKC consensus peptidesubstrate in a reaction mixture of 33 �M unlabeled ATP plus[�32P]ATP, PKC peptide substrate [Ser25]PKC 19-31 (0.1 mg/ml),0.01% phosphatidylserine, 0.01% diacylglycerol, 5.4 mM MgCl2, 20mM Tris base, and 5 mM CaCl2. The mixture was incubated for 5 minat 30°C before being quenched with 75 mM ice-cold H3PO4, vacuumfiltered through ion-exchange cellulose discs, and counted in a scin-tillation counter. Values were calculated as specific PKC activity(pmol 32P incorporated �min�1 �mg protein�1).

Rho GTPase Affinity-Binding Assay

The activation of Rho was determined by affinity binding of thetarget protein rhotekin (binds RhoA-GTP) or CD-PAK (binds Rac1and Cdc42-GTP) as previously described (30). HMEC or BPMECwere grown in six-well dishes to confluence, treated according toexperimental protocol, and collected in glutathione-S-transferase(GST)-FISH buffer [50 mM Tris (pH 7.4), 10% glycerol, 100 mMNaCl, 1% Nonidet P-40, 2 mM MgCl2, 25 mM NaF, and 1 mMEDTA] plus protease inhibitor cocktail (10 �g/ml of pepstatin A, 10�g/ml each of aprotinin and leupeptin, and 1 mM PMSF). Cell lysateswere pelleted by centrifugation at 10,000 g at 4°C for 5 min, and equalvolumes of supernatant were incubated with purified GST-rhotekin orGST-CD-PAK coupled to glutathione Sepharose 4B beads (Amer-sham Pharmacia Biotech, Piscataway, NJ) at 4°C for 1 h. The GTPform of Rho bound to the rhotekin- or CD-PAK-sepharose beads waseluted by being boiled in 2.5� Laemmli sample buffer. The elutedsample and total cell lysate were electrophoresed on 12.5% SDS-PAGE, and Western blot analysis was made with affinity-purifiedantibodies directed against RhoA, Cdc42 (Santa Cruz Biotechnology),or Rac1 (BD Transduction Laboratories, San Jose, CA). The blotswere quantified by scanning densitometry.

PKC Translocation

HMEC were plated on glass coverslips coated with 1 �g/ml offibronectin and grown overnight. After being treated according to exper-imental protocol, the cells were washed with PBS twice, fixed with 3.7%formaldehyde in PBS for 20 min, and permeabilized with 0.5% TritonX-100 in PBS for 5 min at room temperature. After being blocked with5% donkey serum for 30 min, anti-PKC� (Transduction Laboratories,Lexington, KY) or anti-PKC�1 antibody (Sigma) was added for incuba-tion overnight at 4°C. The cells were washed with PBS three times,incubated with Cy2-conjugated anti-mouse IgG antibody (Jackson Im-munoresearch Laboratories, West Grove, PA) in PBS for 1 h at roomtemperature, and viewed on an Olympus AH3-RFCA fluorescent micro-scope (Olympus, Tokyo, Japan). Images were recorded with a Nikondigital camera (DXM 1200; Nikon, Kanagawa, Japan).

PKC� Antisense Oligonucleotides

Translation of PKC� RNA was inhibited using a phosphorothio-ated antisense oligonucleotide complementary to a region two basesupstream from the initiation codon of bovine PKC� [5�-GTC CCT

L177LYSOPHOSPHATIDYLCHOLINE INCREASES ENDOTHELIAL PERMEABILITY

AJP-Lung Cell Mol Physiol • VOL 289 • AUGUST 2005 • www.ajplung.org

CGC CGC CTC CTG-3� (38)]. The control oligonucleotide was ascrambled nonsense nucleotide (5�-TGC CTC CGC GCC TCC CGT-3�). Specificity of the sequence to bovine PKC� and scramblednonsense was verified by using the Entrez-GenBank data base of theNational Institutes of Health. The oligonucleotides were transfectedwith Lipofectin (GIBCO BRL) using a protocol previously described(4). BPMEC were incubated with either antisense or nonsense oligo-nucleotide at a final concentration of 0.2 nmol/ml in Lipofectin (6.6�l/ml) at 37°C for 4 h, a period over which the PKC� mRNA wasreduced 75–85% below baseline (4). The cells were then used forexperimental protocol as needed.

Cell Viability Assay

The assay is based on the reduction of the tetrazolium salt 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT; Sigma)by mitochondrial dehydrogenases in living cells. HMEC were plated at80,000 cells/well in a 96-well tissue culture dish, grown to confluence,and treated with LPC (0–50 �M) for 30 min. The cells were washed andincubated with 0.5 mg/ml of MTT for 3 h at 37°C, and 0.04 M HCl in�-isopropanol was added for reading of the optical density (OD) at 570nm in a plate reader (Molecular Devices, Sunnyvale, CA). The OD unitsprovided an index of enzymatic activity of living cells.

Statistics

Single-sample data were analyzed by the two-tailed t-test; a mul-tiple range test (Scheffe’s test) was used for comparisons of experi-mental groups with a single control group. Two-way ANOVA wasused for analysis to determine significant differences between groups.

RESULTS

LPC Decreases Transendothelial Resistance

We determined the direct effects of LPC on the in vitrobarrier function of endothelial cells. Stimulation of HMECwith LPC (10–50 �M) caused rapid decreases (within min-utes) of resistance from baseline (Fig. 1A). The inset shows anexpanded x-axis of the 25 �M LPC response, indicating aresistance decrease to near maximal by �15 min (Fig. 1A,inset). The resistance decrease in response to the lower con-centrations of 10–25 �M recovered to near baseline within30–60 min. However, at the higher LPC concentration of 50�M, the resistance drop was sustained and did not return tobaseline within the 3-h period of study. The extent of theresistance drop increased with increasing LPC concentrations,reaching a maximum decrease from baseline at 50 �M LPC(Fig. 1, A and B). Evaluation of the effects of LPC on cellviability indicated a 5–8% decrease of the mitochondrial de-hydrogenase activity at 20–50 �M LPC and was not statisti-cally significant from control (results not shown).

Because LPC is known to normally bind serum proteins,particularly to albumin, we also determined the effects ofaltering the relative proportion of albumin concentration withthat of LPC concentration on the resistance decrease. Resultsindicated that decreased albumin levels exacerbated the LPC-induced resistance decrease (Fig. 2). In the presence of 8mg/ml of albumin, 25 �M LPC did not decrease resistance(Fig. 2A). Under these experimental conditions, the ratio of[LPC] to [albumin] was �3, approximating the ratio in circu-lation. Reducing the albumin concentration 10-fold to 0.8mg/ml (thereby, the ratio is increased to 31), 25 �M LPCsignificantly decreased resistance from baseline (Fig. 2A).However, in the absence of albumin, only 2 �M LPC was

needed to reduce resistance (Fig. 2B) to a similar extent as 25�M in the presence of 0.8 mg/ml albumin. These resultssuggest that the bioactivity of LPC is inversely related tobinding by albumin or other serum proteins.

To determine whether LPC impairs barrier function of en-dothelial cells from other vascular beds, we evaluated theeffects of LPC on the transendothelial electrical resistance ofBPMEC. LPC stimulation resulted in a similar pattern ofresistance decrease as HMEC. That is, LPC induced a rapiddose-dependent decrease of the resistance that was reversible atonly the lower concentrations (Fig. 3).

Effects of LPC on PKC Activation

HMEC stimulated with LPC (25 �M) significantly increasedPKC phosphotransferase activity of the membrane fractionwithin 5 min, which slightly decayed with time but remained

Fig. 1. Dose response of lysophosphatidylcholine (LPC) on transendothelialelectrical resistance. Human dermal microvascular endothelial cells (HMEC)were grown onto electric cell-substrate impedance sensor (ECIS) electrodesuntil confluent and treated with LPC (10–50 �M; arrow), and resistance wasdetermined for up to 3 h. A: representative graph of resistance normalized tobaseline (Norm. Resist.) in response to LPC challenge (arrow); n 3–5; C,control buffer. Inset: expanded x-axis of the time course of the decrease andrecovery in response to 25 �M LPC. B: summary of the maximal resistancedecrease from baseline in response to LPC; n 3–5.

L178 LYSOPHOSPHATIDYLCHOLINE INCREASES ENDOTHELIAL PERMEABILITY

AJP-Lung Cell Mol Physiol • VOL 289 • AUGUST 2005 • www.ajplung.org

significantly elevated for up to 60 min (Fig. 4A). Surprisingly,LPC stimulation did not produce a corresponding decreasedPKC activity of the cytosolic fraction (Fig. 4B), which suggeststhat the increased PKC activity of the membrane fractionoccurred in the absence of translocation of the enzyme. We further investigated this possibility by evaluating the immuno-

fluorescent localization of the PKC� and PKC� isoforms inHMEC. We focused on these classic PKC isoforms since we(13, 24) and others (4, 19) have shown that PKC� and PKC�provide critical regulation of endothelial barrier dysfunctioninduced by a wide range of mediators. The immunofluorescentlocalization results showed that HMEC stimulated with LPC(25 �M for 15 min) did not translocate PKC� (Fig. 5A) orPKC� (data not shown) to the membrane, consistent with thePKC activity data. The positive control group in which HMECwere treated with PMA (100 nM, 15 min) showed decreasedPKC� cytoplasmic localization and increased localization atthe cell membrane (Fig. 5A). Further analysis by confocalgeneration of Z-sections from apical toward basal endothelialregions indicated absence of PKC� redistribution in responseto LPC stimulation (Fig. 5B).

LPC Activates RhoA, but not Rac-1 or Cdc42

We determined whether LPC activates RhoA, Rac-1, andCdc42 in endothelial cells by measuring the active GTP-boundform of Rho using the affinity-binding assay (see METHODS).

Fig. 2. Albumin prevents the biological activity of LPC. A: HMEC grown toconfluence on ECIS electrodes were stimulated with 25 �M LPC (arrow) in thepresence of 0.8 mg/ml or 8.0 mg/ml of albumin. B: HMEC were stimulatedwith 2 �M LPC (arrow) in the absence of albumin. Each is a representativegraph showing resistance changes in real time; n 3.

Fig. 3. LPC decreases resistance in bovine pulmonary microvascular endo-thelial cells (BPMEC). BPMEC were grown onto ECIS electrodes untilconfluent and treated with 10–50 �M LPC, and resistance was determined forup to 2 h. Shown is a representative graph of resistance normalized to baselinein response to LPC challenge (arrow); n 4.

Fig. 4. LPC increases PKC phosphotransferase activity. HMEC were treatedwith LPC (25 �M) for 1, 5, 15, and 60 min, and incorporation of 32P into thePKC consensus peptide substrate by the cell membrane (A) and cytosolicfractions (B) were determined in duplicates. Specific PKC activity,pmol �min�1 �mg�1 protein, is reported as mean experimented/control %(Exp/C %) � SE; n 4; *P 0.05 compared with control (0) group.

L179LYSOPHOSPHATIDYLCHOLINE INCREASES ENDOTHELIAL PERMEABILITY

AJP-Lung Cell Mol Physiol • VOL 289 • AUGUST 2005 • www.ajplung.org

Results indicated that in HMEC, LPC (25 �M) induced a twofoldincrease in RhoA-GTP within 5 min that was sustained at 15 minand decreased by 60 min (Fig. 6A). Total RhoA in whole celllysates (GTP-bound � GDP-bound forms) was similar amongexperimental groups, indicating that LPC did not increase de novogeneration of RhoA (results not shown). However, stimulationwith LPC did not increase Rac1-GTP and Cdc42-GTP (Fig. 6B).

Barrier Dysfunction Is Regulated by PKC and RhoA Cross Talk

Dependency on RhoA signaling. The effects of inhibition ofRhoA on resistance were determined by using C3 transferasetoxin (Clostridium botulinum; Biomol, Plymouth Meeting,PA), which inhibits RhoA, RhoB, and RhoC by ADP-ribosy-lation of Rho at Asn41 (9). Results indicated that in HMEC

Fig. 5. Immunofluorescent intracellular localization of PKC�. HMEC were untreated (control) and challenged with 25 �M LPC for 15 min or as positive control(100 nM PMA for 15 min). A: confocal laser microscopy showing the cellular distribution of PKC�; arrows indicate cell periphery; original magnification is�400; n 2. B: confocal serial Z-sections showing distribution of PKC� (control or stimulated with 25 �M LPC for 15 min). The cells were scanned fromapical to basal regions of the cells (a–d) to capture Z-sections at 1-�m intervals; original magnification is �600; n 2.

L180 LYSOPHOSPHATIDYLCHOLINE INCREASES ENDOTHELIAL PERMEABILITY

AJP-Lung Cell Mol Physiol • VOL 289 • AUGUST 2005 • www.ajplung.org

pretreated with C3 toxin (5 �g/ml) for 24 h, the LPC-inducedresistance decrease was significantly inhibited (�50%; Fig. 7,A and B). C3 transferase alone did not decrease resistance frombaseline. Higher concentrations of C3 transferase (10 �g/ml)did not further inhibit the resistance decrease (results notshown).

Dependency on PKC signaling. We used multiple ap-proaches to test the dependency of PKC signaling in theLPC-induced barrier dysfunction. Pharmacological pretreat-ment with the broad spectrum PKC inhibitor Go-6983 wasmade at selected concentrations to approximate its IC50 valuesfor PKC isoforms (IC50 for PKC�, PKC�, and PKC� 6–7nM; IC50 for PKC� 10 nM). Results indicated that pretreat-ment of HMEC with 8 and 20 nM resulted in 40–50%inhibition of the LPC-induced resistance decrease (Fig. 8A).Higher concentrations of Go-6983 (up to 70 nM) did notfurther inhibit the resistance decrease (IC50 for PKC� 60nM; results not shown). Go-6983 pretreatment at all theseconcentrations alone did not alter baseline resistance. Pretreat-ment with 10 �M rottlerin, a selective PKC� inhibitor (IC50

3–6 �M), also did not prevent the LPC-induced resistancedrop (results not shown). These results suggest that LPC-

induced barrier dysfunction is mediated by classic PKCisoforms.

To further investigate this possibility, HMEC were treatedwith PMA (1 �M) for 16 h to deplete PKC. Western blotanalysis showed selective loss of PKC�, but not PKC� (Fig.8B, top). This downregulation was associated with �50%inhibition of the LPC-induced resistance decrease (Fig. 8B,bottom), suggesting that PKC� may be the more importantisoform in the regulation.

The specific role of PKC� was tested by use of antisenseoligonucleotides. In these studies, the antisense PKC� oligo-nucleotides inhibited the LPC-induced resistance decrease(Fig. 9, A and B) as well as the LPC-induced increase inalbumin clearance rate of BPMEC (Fig. 9C). These observa-tions provide strong support that PKC� is critical in theregulation of LPC-induced endothelial barrier dysfunction.

RhoA is downstream target of LPC-activated PKC. Thesimilar time courses of the LPC-induced PKC activity (Fig. 4)with RhoA activation (Fig. 6) suggest possible cross talkbetween these two families of signaling molecules. We inves-tigated whether PKC regulates RhoA in response to LPCactivation in endothelial cells. For these studies, PKC activa-tion was inhibited by pretreatment with 20 nM Go-6983 ordownregulation by long-term pretreatment with PMA (1 �Mfor 16 h), both protocols shown to inhibit LPC-induced barrierdysfunction. Results showed that Go-6983 pretreatment pre-vented �60% of the LPC-induced RhoA activation, and PMA-

Fig. 7. Rho inhibitor blocks LPC-induced resistance drop. HMEC grown ontoECIS electrodes until confluent were pretreated with 5 �g/ml of C3 transferasefor 24 h and challenged with 25 �M LPC (arrow), and resistance was measuredas described. A: representative graph of resistance changes. B: summary ofresistance decrease from baseline in response to LPC; n 5; *P 0.01compared with LPC alone.

Fig. 6. Effects of LPC on activation of Rho GTPases. Rho activation wasdetermined by affinity binding of Rho target fusion proteins as described inMETHODS. A: confluent HMEC were treated with 25 �M LPC at 0, 1, 5, 15, and60 min. A representative Western blot shows the affinity-binding results ofRhoA. Densitometric analysis is shown as Exp/C % as means � SE of 5–10independent experiments; *P 0.01 compared with no LPC treatment. B:representative Western blot showed that LPC (25 �M, 5 min) did not induceincrease of Rac1-GTP and Cdc42-GTP.

L181LYSOPHOSPHATIDYLCHOLINE INCREASES ENDOTHELIAL PERMEABILITY

AJP-Lung Cell Mol Physiol • VOL 289 • AUGUST 2005 • www.ajplung.org

induced downregulation inhibited �80% (Fig. 10A). Similarly,pretreatment with antisense PKC� oligonucleotides prevented68% of the LPC-induced RhoA activation (Fig. 10B). Thecontrol nonsense oligonucleotide did not inhibit the LPC-induced RhoA activation (Fig. 10B).

DISCUSSION

The major findings in this study are that extracellular LPC 1)impaired endothelial barrier function that was sensitive to

albumin concentration, 2) increased PKC phosphotransferaseactivity in the absence of translocation of either PKC� orPKC�, 3) activated RhoA, but not Rac1 or Cdc42, and 4)induced barrier dysfunction that was dependent in part onPKC� and RhoA cross talk. These findings provide the first

Fig. 9. PKC� antisense (AS) oligonucleotides inhibit endothelial barrierdysfunction. A: BPMEC were grown on ECIS electrodes to confluence,transfected with PKC� AS oligonucleotide (0.2 nM/ml) for 4 h, and thenchallenged with 25 �M LPC. The negative control, PKC�-nonsense (NS)oligonucleotide (0.2 nM/ml), was similarly transfected into separate groups ofBPMEC, and resistance was measured as described. Shown is a representativegraph of resistance changes. B: summary of resistance decrease from baselinein response to LPC; n 5; *P 0.01 compared with LPC alone. C: BPMECgrown to confluence on gelatinized micropore filters were transfected withPKC�-AS oligonucleotide as described for resistance studies, and albuminclearance rate was determined; n 4; *P 0.05 compared with C (non-treated) group; **P 0.05 compared with LPC alone.

Fig. 8. Effects of PKC inhibition on LPC-mediated resistance. HMEC weregrown to confluence on ECIS electrodes and pretreated. A: cells were pre-treated with the PKC inhibitor Go-6983 (GO) from 8 to 20 nM and challengedwith 25 �M LPC, and resistance was determined for up to 3 h. Shown is arepresentative graph of the LPC-induced resistance change following pretreat-ment with GO. Control (C) indicates absence of GO and LPC; open arrowindicates time of inhibitor pretreatment; solid arrow indicates challenge withLPC; n 5. B: HMEC were treated with 1 �M PMA for up to 16 h todownregulate PKC. Control indicates absence of PMA. Western blot showsPKC�, but not PKC�1, was downregulated by 16 h (top). The downregulatedHMEC were stimulated with 25 �M LPC (arrow), and resistance was mea-sured. A representative graph shows the effects of downregulated PKC� onresistance decrease in response to LPC challenge; n 5.

L182 LYSOPHOSPHATIDYLCHOLINE INCREASES ENDOTHELIAL PERMEABILITY

AJP-Lung Cell Mol Physiol • VOL 289 • AUGUST 2005 • www.ajplung.org

report that the proinflammatory LPC directly induces increasein endothelial permeability through activation of signalingmechanisms.

Our findings support in vivo studies in which intraspinalcord injection of LPC into mice LPC induces recruitment ofleukocytes and widespread breakdown of the blood-brain bar-rier (27). Similarly, local subcutaneous injection of LPC inhuman subjects results in tissue edema and increased extrava-

sation of T lymphocytes, monocytes, and neutrophils (31).Although the LPC-induced vascular endothelial barrier dys-function observed in these in vivo studies is likely contributedby the associated activated leukocytic infiltration, our resultsprovide evidence that LPC directly and significantly inducesincreased permeability, which can further exacerbate inflam-matory tissue injury.

We showed that the LPC-induced increase in endothelialpermeability was regulated by two critical signaling pathways,PKC and RhoA. The inhibition of classic PKC by threedifferent approaches (Go-6983, PMA-induced downregulation,and antisense PKC� oligonucleotides) each prevented �50%of the LPC-mediated resistance drop. The LPC-induced barrierdysfunction occurred in the absence of translocation of PKC�to the endothelial cell membrane. The absence of translocationwas further supported by the observation that the LPC-in-creased PKC activity of the membrane cell fraction was notaccompanied by a corresponding decrease of activity in thecytosolic fraction, suggesting an increased intrinsic catalyticactivity of the membrane-associated enzyme. The positivecontrol, PMA, stimulated PKC� translocation to the membranein these cells, indicating that the translocation was stimulusselective. Several cell types (i.e., anterior pituitary cells, fibro-blasts) have been reported to show PKC enzyme activation thatwas temporally unrelated to translocation (10). Furthermore,PKC enzyme activity is known to be regulated through phos-phorylation of specific residues on the activation loop of thePKC kinase domain (28). Therefore, it is possible that inendothelial cells, a pool of membrane-bound PKC� exists thatcan be activated by LPC stimulation (28). We also observedthat the antisense PKC� oligonucleotides completely inhibitedthe LPC-induced albumin permeability. It is not clear thereasons why the extent of inhibition differed between the twobarrier assays, but they may be attributed to the inherentdifference of measuring ion flow and solute flux across a cellmonolayer.

We find that LPC-induced increase in endothelial permeabil-ity was also dependent on RhoA. LPC stimulation of endothe-lial cells increased the amount of RhoA-GTP, but not Rac1-GTP or Cdc42-GTP, the other primary Rho-GTPases in cells.RhoA activation occurred within 5 min, remained significantlyelevated at 15 min, and declined toward baseline by 60 min, atime period consistent with decreased resistance in response tothe same concentration of LPC. Inhibition of the LPC-activatedRhoA with the C3 exotoxin prevented �50% of the resistancedecrease, and increased C3 concentration (to 10 �g/ml) did notconfer additional inhibition, suggesting that other additionalmechanisms contribute to the overall regulation of the in-creased permeability. Interestingly, a similar degree of inhibi-tion of the thrombin-induced permeability increase by the C3toxin is reported by Van Nieuw Amerongen et al. (35a).Although definitive targets of regulation by RhoA in theLPC-mediated barrier dysfunction are not known, one likelycandidate is Rho kinase, which has been proposed for thethrombin response. Rho kinase is believed to inactivate myosinlight chain (MLC) phosphatase as part of a signaling networkthat controls MLC phosphorylation and endothelial cell con-traction (3), which is a critical mechanism underlying endo-thelial barrier dysfunction (20, 37).

The time course of activation of PKC and RhoA signals byLPC was similar, beginning at �5 min, remaining elevated for

Fig. 10. PKC inhibition prevents LPC-induced RhoA activation. A: HMECwere pretreated with or without 1 �M PMA overnight to induce downregula-tion (Dn) of PKC� or 20 nM PKC inhibitor GO for 15 min. Cells were thenstimulated with LPC (25 �M) for 5 min. Cell extracts were assayed forGTP-bound RhoA by affinity binding as described in METHODS. Shown is arepresentative affinity binding of RhoA-GTP. Densitometric analysis of RhoA-GTP is presented as means � SE of 4 separate experiments; *P 0.01compared with no LPC treatment (C); **P 0.05 compared with LPCtreatment alone. B: BPMEC were transfected with PKC�-AS oligonucleotide(0.2 nM/ml) for 4 h and stimulated with LPC, and RhoA activation wasdetermined as described above. The PKC�-NS oligonucleotide served ascontrol. A representative Western blot shows the affinity-binding results ofRhoA. Densitometric analysis is expressed as Exp/C % as means � SE of 4independent experiments; *P 0.05 compared with no LPC treatment;**indicate P 0.05 compared with LPC treatment alone.

L183LYSOPHOSPHATIDYLCHOLINE INCREASES ENDOTHELIAL PERMEABILITY

AJP-Lung Cell Mol Physiol • VOL 289 • AUGUST 2005 • www.ajplung.org

15 min, and decreasing toward baseline by 60 min. Impor-tantly, within 15 min of LPC stimulation, resistance decreasewas near maximal, and a subsequent reversal occurred duringthe next 30–60 min. These two coincident signaling pathwaysdisplayed cross talk in which RhoA was a downstream target ofPKC activation. Interestingly, thrombin stimulation of humanumbilical vein endothelial cells also demonstrates cross talkbetween PKC and RhoA (19). In this report, Mehta et al. (19)found that the PKC-induced activation of RhoA is throughphosphorylation, and, hence, inhibition of guanine nucleotidedissociation inhibitor. It remains to be determined whether asimilar mechanism of regulation of RhoA by PKC occurs in theLPC-induced endothelial barrier dysfunction. The signal trans-duction mechanisms by which LPC activates PKC and RhoAhave yet to be defined. There is emerging evidence linkingLPC-activated signals to possible orphan G protein-coupledreceptors, GPR4, G2A, and OGR1 (39). We recently reportedthat human brain and dermal microvascular endothelial cellsselectively express GPR4, but not G2A (14), and that GPR4may have a critical role in regulation of angiogenesis inresponse to SPC (6). Further studies are underway to defini-tively establish a role of GPR4 in the regulation of endothelialbarrier function.

We find that the presence of albumin greatly influenced thebioactivity of LPC. Albumin has been shown to contain one tothree high-affinity binding sites as well as multiple low-affinitybinding sites for LPC (29). Our results show that increasingalbumin concentration approximating the proportion in thecirculation (relative to LPC) prevented the LPC-induced resis-tance decrease. On the other hand, decreasing albumin (orincreasing LPC level) would induce a corresponding impair-ment of the endothelial barrier function. This finding is con-sistent with a recent report that perfusion of LPC into analbu-minemic rats resulted in a greater vasoconstriction responsethan in control normal rats (36). Furthermore, in the samestudy, infusion of albumin prevented LPC-induced vasocon-striction in both analbuminemic and control rats (36). Theseresults strongly suggest that effective proinflammatory activi-ties of LPC can occur in tissues either in which there iselevated LPC concentration that exceeds the binding capacityof albumin (and possibly other serum proteins) or under con-ditions of decreased albumin levels.

In summary, we show that LPC directly impairs the endo-thelial barrier function that is sensitive to albumin concentra-tion. This LPC-induced barrier dysfunction was dependent, atleast in part, on cross talk of PKC� and RhoA signals.

GRANTS

This work was supported by National Heart, Lung, and Blood InstituteGrants HL-71081 (H. Lum), HL-52597 (P. V. Subbaiah), HL-69585 (P. V.Subbaiah), and HL-59901 (A. Johnson) and the American Heart AssociationPostdoctoral Fellowship Award, Midwest (F. Huang).

REFERENCES

1. Ades EW, Candal FJ, Swerlick RA, George VG, Summers S, BosseDC, and Lawley TJ. HMEC-1: establishment of an immortalized humanmicrovascular endothelial cell line. J Invest Dermatol 99: 683–690, 1992.

2. Chakraborti S, Gurtner GH, and Michael JR. Oxidant-mediated acti-vation of phospholipase A2 in pulmonary endothelium. Am J Physiol LungCell Mol Physiol 257: L430–L437, 1989.

3. Essler M, Amano M, Kruse HJ, Kaibuchi K, Weber PC, and Aepfel-bacher M. Thrombin inactivates myosin light chain phosphatase via Rho

and its target Rho kinase in human endothelial cells. J Biol Chem 273:21867–21874, 1998.

4. Ferro T, Neumann P, Gertzberg N, Clements R, and Johnson A.Protein kinase C-� mediates endothelial barrier dysfunction induced byTNF-�. Am J Physiol Lung Cell Mol Physiol 278: L1107–L1117, 2000.

5. Katz AM and Messineo FC. Lipid-membrane interactions and the patho-genesis of ischemic damage in the myocardium. Circ Res 48: 1–16, 1981.

6. Kim K, Ren J, Jiang Y, Ebrahem Q, Tipps R, Cristina K, Xiao Y,Qiao J, Taylor KL, Lum H, Anand-Apte B, and Xu Y. GPR4 playscritical role in endothelial cells and mediates the effects of sphingo-sylphosphorylcholine. FASEB J. 19: 819–821, 2005.

7. Kugiyama K, Sugiyama S, Ogata N, Oka H, Doi H, Ota Y, and YasueH. Burst production of superoxide anion in human endothelial cells bylysophosphatidylcholine. Atherosclerosis 143: 201–204, 1999.

8. Kume N, Cybulsky MI, and Gimbrone MA Jr. Lysophosphatidylcho-line, a component of atherogenic lipoproteins, induces mononuclear leu-kocyte adhesion molecules in cultured human and rabbit arterial endothe-lial cells. J Clin Invest 90: 1138–1144, 1992.

9. Lerm M, Schmidt G, and Aktories K. Bacterial protein toxins targetingrho GTPases. FEMS Microbiol Lett 188: 1–6, 2000.

10. Liu JP. Protein kinase C and its substrates. Mol Cell Endocrinol 116:1–29, 1996.

11. Lum H, Hao Z, Gayle D, Kumar P, Patterson CE, and Uhler MD.Vascular endothelial cells express isoforms of protein kinase A inhibitor.Am J Physiol Cell Physiol 282: C59–C66, 2002.

12. Lum H, Jaffe HA, Schulz IT, Masood A, RayChaudhury A, andGreen RD. Expression of PKA inhibitor (PKI) gene abolishes cAMP-mediated protection to endothelial barrier dysfunction. Am J Physiol CellPhysiol 277: C580–C588, 1999.

13. Lum H, Podolski JL, Gurnack ME, Schulz IT, Huang F, and HolianO. Protein phosphatase 2B inhibitor potentiates endothelial PKC activityand barrier dysfunction. Am J Physiol Lung Cell Mol Physiol 281:L546–L555, 2001.

14. Lum H, Qiao J, Walter RJ, Huang F, Subbaiah PV, Kim KS, andHolian O. Inflammatory stress increases receptor for lysophosphatidyl-choline in human microvascular endothelial cells. Am J Physiol Heart CircPhysiol 285: H1786–H1789, 2003.

15. Marinetti GV. Chromatographic separation, identification, and analysisof phosphatides. J Lipid Res 3: 1–20, 1962.

16. McHowat J and Corr PB. Thrombin-induced release of lysophosphati-dylcholine from endothelial cells. J Biol Chem 268: 15605–15610, 1993.

17. Mcintyre TM, Zimmerman GA, and Prescott SM. Biologically activeoxidized phospholipids. J Biol Chem 274: 25189–25192, 1999.

18. Mehta D, Gupta S, Gaur SN, Gangal SV, and Agrawal KP. Increasedleukocyte phospholipase A2 activity and plasma lysophosphatidylcholinelevels in asthma and rhinitis and their relationship to airway sensitivity tohistamine. Am Rev Respir Dis 142: 157–161, 1990.

19. Mehta D, Rahman A, and Malik AB. Protein kinase C-� signalsrho-guanine nucleotide dissociation inhibitor phosphorylation and rhoactivation and regulates the endothelial cell barrier function. J Biol Chem276: 22614–22620, 2001.

20. Moy AB, VanEngelenhoven J, Bodmer J, Kamath J, Keese C, GiaeverI, Shasby S, and Shasby DM. Histamine and thrombin modulate endo-thelial focal adhesion through centripetal and centrifugal forces. J ClinInvest 97: 1020–1027, 1996.

21. Murohara T, Scalia R, and Lefer AM. Lysophosphatidylcholine pro-motes P-selectin expression in platelets and endothelial cells. Possibleinvolvement of protein kinase C activation and its inhibition by nitricoxide donors. Circ Res 78: 780–789, 1996.

22. Murphy AA, Santanam N, Morales AJ, and Parthasarathy S. Lyso-phosphatidyl choline, a chemotactic factor for monocytes/T-lymphocytesis elevated in endometriosis. J Clin Endocrinol Metab 83: 2110–2113,1998.

23. Murugesan G, Sandhya Rani MR, Gerber CE, Mukhopadhyay C,Ransohoff RM, Chisolm GM, and Kottke-Marchant K. Lysophos-phatidylcholine regulates human microvascular endothelial cell expressionof chemokines. J Mol Cell Cardiol 35: 1375–1384, 2003.

24. Nagpala PG, Malik AB, Vuong PT, and Lum H. Protein kinase C � 1overexpression augments phorbol ester-induced increase in endothelialpermeability. J Cell Physiol 166: 249–255, 1996.

26. Okita M, Gaudette DC, Mills GB, and Holub BJ. Elevated levels andaltered fatty acid composition of plasma lysophosphatidylcholine(lysoPC)in ovarian cancer patients. Int J Cancer 71: 31–34, 1997.

L184 LYSOPHOSPHATIDYLCHOLINE INCREASES ENDOTHELIAL PERMEABILITY

AJP-Lung Cell Mol Physiol • VOL 289 • AUGUST 2005 • www.ajplung.org

27. Ousman SS and David S. Lysophosphatidylcholine induces rapid recruit-ment and activation of macrophages in the adult mouse spinal cord. Glia30: 92–104, 2000.

28. Parekh DB, Ziegler W, and Parker PJ. Multiple pathways controlprotein kinase C phosphorylation. EMBO J 19: 496–503, 2000.

29. Portman OW and Illingworth DR. Lysolecithin binding to human andsquirrel monkey plasma and tissue components. Biochim Biophys Acta326: 34–42, 1973.

30. Qiao J, Huang F, and Lum H. PKA inhibits RhoA activation: aprotection mechanism against endothelial barrier dysfunction. Am JPhysiol Lung Cell Mol Physiol 284: L972–L980, 2003.

31. Ryborg AK, Deleuran B, Sogaard H, and Kragballe K. Intracutaneousinjection of lysophosphatidylcholine induces skin inflammation and accu-mulation of leukocytes. Acta Derm Venereol 80: 242–246, 2000.

32. Sametz W, Hummer K, Butter M, Wintersteiger R, and Juan H.Formation of 8-iso-PGF(2�) and thromboxane A2 by stimulation withseveral activators of phospholipase A2 in the isolated human umbilicalvein. Br J Pharmacol 131: 145–151, 2000.

33. Sobel BE, Corr PB, Robison AK, Goldstein RA, Witkowski FX, andKlein MS. Accumulation of lysophosphoglycerides with arrhythmogenicproperties in ischemic myocardium. J Clin Invest 62: 546–553, 1978.

34. Sugiyama S, Kugiyama K, Ogata N, Doi H, Ota Y, Ohgushi M,Matsumura T, Oka H, and Yasue H. Biphasic regulation of transcriptionfactor nuclear factor-�B activity in human endothelial cells by lysophos-phatidylcholine through protein kinase C-mediated pathway. ArteriosclerThromb Vasc Biol 18: 568–576, 1998.

35. Vadas P, Browning J, Edelson J, and Pruzanski W. Extracellularphospholipase A2 expression and inflammation: the relationship withassociated disease states. J Lipid Mediators 8: 1–30, 1993.

35a.Van Nieuw Amerongen GP, Draijer R, Vermeer MA, and van Hins-bergh VW. Transient and prolonged increase in endothelial permeabilityinduced by histamine and thrombin: role of protein kinases, calcium, andRhoA. Circ Res 83: 1115–1123, 1998.

36. Vuong TD, Braam B, Willekes-Koolschijn N, Boer P, Koomans HA,and Joles JA. Hypoalbuminaemia enhances the renal vasoconstrictoreffect of lysophosphatidylcholine. Nephrol Dial Transplant 18: 1485–1492, 2003.

37. Wysolmerski RB and Lagunoff D. Regulation of permeabilized endo-thelial cell retraction by myosin phosphorylation. Am J Physiol CellPhysiol 261: C32–C40, 1991.

38. Xia P, Aiello LP, Ishii H, Jiang ZY, Park DJ, Robinson GS, TakagiH, Newsome WP, Jirousek MR, and King GL. Characterization ofvascular endothelial growth factor’s effect on the activation of proteinkinase C, its isoforms, and endothelial cell growth. J Clin Invest 98:2018 –2026, 1996.

39. Xu Y. Sphingosylphosphorylcholine and lysophosphatidylcholine: G pro-tein-coupled receptors and receptor-mediated signal transduction. BiochimBiophys Acta 1582: 81–88, 2002.

40. Zhu Y, Lin JH, Liao HL, Verna L, and Stemerman MB. Activation ofICAM-1 promoter by lysophosphatidylcholine: possible involvement ofprotein tyrosine kinases. Biochim Biophys Acta 1345: 93–98, 1997.

L185LYSOPHOSPHATIDYLCHOLINE INCREASES ENDOTHELIAL PERMEABILITY

AJP-Lung Cell Mol Physiol • VOL 289 • AUGUST 2005 • www.ajplung.org