82

Insulin prevents oxidant-inducedendothelial cell barrier dysfunctionvia nitric oxide–dependent pathwaySiddhartha Rath, MD,a Theodore Kalogeris, PhD,a Nicholas Mai,a Gazi Zibari, MD,a J. StevenAlexander, PhD,b David Lefer, PhD,b and Richard H. Turnage, MD,a Shreveport, La

Background. The rigorous maintenance of normoglycemia by the administration of insulin is beneficialto critically ill patients. Because insulin induces endothelial nitric oxide (NO) release, and theconstitutive release of NO maintains normal microvascular permeability, the authors postulated thatinsulin would prevent peroxide (H2O2)-induced endothelial barrier dysfunction, an effect dependent onendothelial NO synthase (eNOS) activity.Methods. Murine lung microvascular endothelial cells (LMEC) grown to confluence on 8 l porepolyethylene filters were exposed to media (control), H2O2 (20 to 500 lmol/L), insulin (1 to 1,000nmol/L) or insulin (100 nmol/L) + H2O2 (10�4mol/L). Endothelial monolayer permeability wasquantitated by measuring the transendothelial electrical resistance at 15-minute intervals for 120minutes. Other cells were exposed to H2O2 and insulin after pretreatment with a NO scavenger (PTIO),an eNOS inhibitor (L-NIO), or a phosphoinositol-3-kinase inhibitor (LY-294002).Results. H2O2 caused a concentration- and time-dependent reduction in electrical resistance consistentwith an increase in monolayer permeability. This effect was prevented by insulin. Inhibiting NO release(L-NIO, LY-294002) or scavenging NO (PTIO) abolished this protective effect.Conclusions. These data suggest that insulin may modulate endothelial barrier function during oxidantstress by inducing the release of NO. (Surgery 2006;139:82-91.)

From the Departments of Surgerya and Physiology,b Louisiana State University Health Sciences Center,Shreveport, La

CLINICAL AND IN VIVO EXPERIMENTAL STUDIES haveshown a profound protective effect of insulin invarious acute inflammatory conditions, eg, tissuereperfusion injury, myocardial ischemia and infarc-tion, and critical illness.1-9 Notably, insulin’s pro-tective effects are not dependent on its metabolicactions, but rather may reflect a direct, anti-inflam-matory effect of this hormone.2,3 Several findingshave suggested that insulin has direct protectiveeffects on endothelial cells.10,11

The endothelial cell is the principal target ofthe pro-inflammatory cascades active in local andsystemic inflammatory states, and increased micro-vascular barrier permeability is a fundamentalphysiologic consequence of acute inflammation.Endothelial cell barrier function is determined, at

Accepted for publication June 2, 2005.

Reprint requests: Richard H. Turnage, MD, Chairman, Depart-ment of Surgery, LSU Health Sciences Center in Shreveport,1501 Kings Highway, Shreveport, LA 71130. E-mail: rturna@lsuhsc.edu.

0039-6060/$ - see front matter

� 2006 Mosby, Inc. All rights reserved.

doi:10.1016/j.surg.2005.06.056

SURGERY

least in part, by the actin cytoskeleton, whichcontrols cellular morphology and the proteins ofthe intercellular adhesion junctions that link con-tiguous cells thereby limiting the passage ofmacromolecules across the microvasculature. Pro-inflammatory mediators, including oxidants, in-crease microvascular permeability by promotingthe interaction of actin microfilaments and myosinleading to endothelial cell contraction and theformation of intercellular gaps.12-14 Actin-myosincontraction in endothelial cells is mediated bymyosin light chain (MLC) phosphorylation,12,15

which is promoted by a Ca2+-dependent myosinlight chain kinase (MLCK) or by the inactivationof MLC phosphatase by the Ras-related GTPase,Rho, and its downstream effector, p160 Rho ki-nase.16,17 (Reviewed in Ref. 18,19.)

In smooth muscle cells, insulin inhibits rho/rho kinase-mediated actin-myosin contraction bya mechanism dependent on the presence ofNO.20,21 Because insulin induces endothelial cellNO release,22-25 and the constitutive release of NOis a potent modulator of microvascular permeabil-ity,26-29 we postulated that insulin would preventoxidant-induced increases in endothelial cell

SurgeryVolume 139, Number 1

Rath et al 83

monolayer permeability by a mechanism depen-dent on NO.

MATERIAL AND METHODS

Chemicals. Dulbecco’s Modified Eagle Media(DMEM) were obtained from Sigma Aldrich Chem-icals Inc (St. Louis, Mo). Cell culture media supple-ments including fetal bovine serum, antibiotics, andvitamin mixtures were obtained from Mediatech,Inc (Herndon, Va). Hydrogen peroxide (H2O2) wasobtained from Fisher Scientific (Fairlawn, NJ). Insu-lin was obtained from Eli Lilly (Indianapolis, Ind). L-NIO (N5-[1-Iminoethyl]-L-ornithine dihydrochloride)was obtained from Acros Organic-Fisher Scientific(Fairlawn, NJ). PTIO (2-phenyl-4,4,5,5-tetramethyl-imidazoline-3-oxide-1-oxyl) was obtained from AlexisBiochemical- AXXORA, LLC (San Diego, Calif). Car-boxy-PTIO was obtained from Dojindo Laboratories(Tabaru, Japan). LY-294002 (2-[4-morpholinyl]-8-phenyl-4H-1-benzopyran-4-one) was obtained fromCell Signaling (Beverly, Mass), and spermine NON-Oate (SperNO, N-[4-(3-aminopropyl)-2-hydroxy-2-nitrosohydrazino]butyl-1,3-proanediamine) wasobtained from A.G. Scientific (San Diego, Calif).

Endothelial cell cultures. Murine lung micro-vascular endothelial cells (LMEC, LEII cells) orrabbit aortic endothelial cells (RAEC) were grownin 75-mm2 tissue culture flasks in DMEM supple-mented with 1% fetal bovine serum, 1% penicil-lin/streptomycin/amphoterecin B, 1% vitaminmixture, and a 1% mixture of MEM nonessentialamino acids (Invitrogen Co, Carlsbad, Calif). Thecells were maintained and incubated at 37�C in ahumidified 5% CO2 incubator and used at pas-sages 9 through 12.

Measurement of transendothelial electrical resis-tance.LMECs and RAECs were grown to confluencein DMEM on 8 l pore polyethylene terepthalateinserts in a Multiwell� 24-well tissue culture plate(Becton Dickinson, Franklin Lakes, NJ). Transen-dothelial electrical resistance (TEER) was measuredacross the monolayer using Ag/AgCl2 electrodes ashas been described previously.30,31 Electrical resis-tance was measured every 15 minutes (in ohms)and normalized as the ratio of measured resistanceto baseline resistance and plotted versus time. Thisassay of endothelial monolayer permeability corre-lates well with techniques quantitating the move-ment of labeled proteins across an endothelialmonolayer in response to tumor necrosis factor-alpha (TNF-a), oxidants, and thrombin.13,15,16

Effect of H2O2 and insulin on EC monolayerpermeability. The effect of H2O2 on TEER was as-sessed by exposing the confluent monolayers to 0,50, 100, or 500 lmol/L of H2O2 (n = 4 per group).

These concentrations have been shown previouslyto induce changes in endothelial cell morphology,actin cytoskeletal architecture, and monolayer per-meability without causing cell lysis.12-14 The effectof insulin on H2O2-induced increases in TEERwas assessed by pretreating the endothelial cellswith insulin (0 to 1,000 nmol/L)21-24 for 30 min-utes after which 100 lmol/L H2O2 was added tomedia (n = 4 per group). TEER was measured be-fore the addition of H2O2 or insulin (baseline) andat 15-minute intervals for 120 minutes. Resistancemeasurements were normalized to the average of 3measurements taken before the addition of H2O2

or insulin and expressed as a percentage of baseline.In a related set of experiments, the effect of

glucose concentration in the incubation media oninsulin’s effect on H2O2-induced barrier dysfunc-tion was tested by incubating the LMECs inDMEM containing either 1 or 4.5 g/L glucosefor 1 full passage, 7 days. The cells then were pre-treated with 100 nmol/L insulin for 30 minutesand then exposed to 100 lmol/L H2O2 for 90 min-utes. TEER was measured as described in the pre-vious paragraphs. At the conclusion of each ofthe experiments, the glucose concentration withinthe media was measured by a spectrophotometricassay in the clinical laboratory at LSU Health Sci-ences Center in Shreveport.

Effect of H2O2 on EC lysis. The effect of H2O2

on endothelial cell lysis was assessed by exposingconfluent monolayers of LMECs to either 0 or100 lmol/L of H2O2 (n = 4 per group) in the pres-ence or absence of 100 nmol/L insulin for 120minutes. The concentration of lactate dehydrogen-ase (LDH) in the incubation media was measuredby a colorimetric assay in the clinical laboratory atLSU Health Science Center in Shreveport. LDH isan intracellular macromolecule released into theextracellular milieu upon dissolution of the cellmembrane during cell death.32

The role of NO in insulin-mediated preserva-tion of endothelial barrier function. Inhibitors ofNO: TEER was measured in confluent LMECmonolayers pretreated with the NOS inhibitor, L-NIO (100 lmol/L, n = 8)33,34 or NO scavengers,PTIO (50 lmol/L, n = 8)34 and carboxy-PTIO(50 lmol/L, n = 4)35 in the presence or absenceof insulin (100 nmol/L). Thirty minutes later,the cells were exposed to 100 lmol/L H2O2, andTEER was measured every 15 minutes for 120minutes as described earlier. L-NIO inhibits boththe inducible and endothelial isoforms of NOS.In the current experiments, it is most likely to beacting through its effects on the latter isoform,because it is the predominant NOS isoform in

SurgeryJanuary 2006

84 Rath et al

endothelial cells and the one whose activity isacutely regulated within the time intervals in thisstudy.36,37 Preliminary studies in our laboratory(unpublished data) as well as studies by Montag-nani et al38 have shown that insulin activateseNOS via a Ca2+-independent mechanism involv-ing phosphorylation of Ser1179 by Akt within 2 min-utes of exposure.

Inhibitors of insulin/NO signaling: Insulin in-creases endothelial NO release by a signalingpathway that includes phosphoinositol-3-kinase(PI3K).23,25 We, therefore, pretreated confluentLMEC monolayers with the PI3K inhibitor, LY-294002 (100 lmol/L, n = 8)39 in the presence orabsence of insulin (100 nmol/L). Thirty minuteslater, H2O2 (100 lmol/L) was added to the media.TEER was measured before the addition of LY-294002 and every 15 minutes thereafter for 120minutes as described earlier.

NO donor: In these experiments, confluentLMECs were pretreated with the NO donorSperNO (100 lmol/L, n = 10)33,40 for 30 minutesafter which they were exposed to 100 lmol/LH2O2. TEER was measured before the addition ofSperNO and every 15 minutes thereafter for 120minutes and expressed as a percentage of baselinemeasurements.

Statistical methods. The data are expressed asthe mean + SEM. Time-dependent changes in TEER(from baseline) were analyzed using repeated

Fig 1. H2O2 caused a dose- and time-dependent de-crease in TEER in lung microvascular endothelial cells.TEER was measured before the addition of H2O2 (base-line) and at 15-minute intervals for 90 minutes. Resis-tance measurements were normalized to the average of3 measurements taken before the addition of H2O2

and expressed as a percentage of baseline. n = 4 pergroup; **P < .01, H2O2 (100, 500 lmol/L) versus base-line and controls; ***P < .001, H2O2 (100, 500 lmol/L)versus baseline and controls.

measures analysis of variance (ANOVA) with aDunnett’s post hoc test. Individual treatmentgroups were compared with one another at eachspecific time-point using a 1-way ANOVA with aTukey’s post hoc test. P value less than .05 was con-sidered statistically significant . There were at least4 measurements in each experimental group. Thedata were analyzed with GraphPad Instat (Version3.05 for Windows XP, GraphPad Software Inc, SanDiego, Calif, www.graphpad.com.)

RESULTS

The effect of H2O2 on EC monolayer barrierfunction. H2O2 caused a dose-dependent increasein endothelial cell monolayer permeability overthe 90 minutes of study (Fig 1). Within 30 minutesof exposure, 100 and 500 lmol/L H2O2 caused a40% to 50% decrease in TEER compared withbaseline measurements (P < .001) and controls ex-posed to media alone (P < .0001). A total of 20lmol/L H2O2 also decreased TEER by about40%, but this effect was not apparent until 75 min-utes after the addition of H2O2 (P < .05). Theseeffects persisted for the duration of the experimen-tal period (P < .0001). In control cells (ie, not trea-ted with H2O2) there was no significant change inthe TEER over the 90-minute experimental period.

Fig 2. Insulin attenuates the H2O2 (100 lmol/L)-induced decrease in TEER in lung microvascular endo-thelial cells in a dose-dependent manner. TEER wasmeasured before the addition of H2O2 or insulin (base-line) and at 15-minute intervals for 120 minutes. Resis-tance measurements were normalized to the average of3 measurements taken before the addition of H2O2 orinsulin and expressed as a percentage of baseline. n =4 per group; *P < .05, H2O2 versus baseline, controls,and insulin (100 nmol/L, 1,000 nmol/L) + H2O2; **P <.01, H2O2 versus baseline, controls, and insulin (100nmol/L, 1000 nmol/L) + H2O2.

SurgeryVolume 139, Number 1

Rath et al 85

The effect of insulin on H2O2-induced ECmonolayer barrier dysfunction. Pretreatment ofthe LMEC monolayer with insulin caused a dose-dependent attenuation of H2O2-induced permea-bility increase, with 100 and 1, 000 nmol/L insulincompletely abolishing H2O2-induced reductionsin TEER and 1 and 10 nmol/L having no effect(Fig 2). As in the previous experiment, the TEERof monolayers exposed to 10�4 mol/L H2O2 for120 minutes was about 30% less than that in baselinemeasurements (P = .001) or that of controls incu-bated in media alone (P = .0001). In contrast, theTEER of cells exposed to insulin (100 nmol/L) +H2O2 (10�4mol/L) was similar to baseline measure-ments and 20% to 30% greater than that of cells ex-posed to H2O2 alone at every time-point from 45 to120 minutes (P = .001). Insulin alone had no effecton TEER. Identical results were obtained when thisexperiment was performed using RAECs (Fig 3).

The effect of media glucose concentration oninsulin-mediated barrier protection. The concen-tration of glucose in the incubation media had noeffect on either H2O2-induced endothelial barrierdysfunction or insulin’s ability to prevent it. H2O2

caused a 35% to 40% reduction in TEER in cellsgrown in DMEM containing either 1.0 or 4.5 g/Lglucose, an effect prevented, in both groups, bypretreatment with insulin (Fig 4). The measuredconcentration of glucose in these groups is shownin the Table.

Fig 3. Insulin (100 nmol/L) attenuates the H2O2 (100lmol/L)-induced decrease in TEER in rabbit aortic en-dothelial cells. TEER was measured before the additionof H2O2 or insulin (baseline) and at 15-minute intervalsfor 120 minutes. Resistance measurements were normal-ized to the average of 3 measurements taken before theaddition of H2O2 or insulin and expressed as a percent-age of baseline. n = 4 per group; *P < .05, H2O2 versusbaseline, controls, and insulin + H2O2; **P < .01, H2O2

versus baseline, controls, and insulin + H2O2.

The effect of H2O2 and insulin on EC lysis.Exposure of the LMECs to 100 lmol/L H2O2 for120 minutes did not increase the concentrationof LDH in the incubation media when comparedwith control cells exposed to media alone (24 ± 4vs 20 ± 1 U/L, respectively). Similarly, there wasno difference in the LDH release of cells exposedto insulin alone or insulin + H2O2 (28 ± 2 vs 29 ±1 U/L, respectively).

The role of NO in insulin-mediated barrierprotection. Inhibition of NOS activity with L-NIOor NO scavenging with PTIO or carboxy-PTIOabolished insulin’s protective effect on H2O2-induced barrier dysfunction (Fig 5). Specifically,TEER measurements in LMEC exposed to H2O2

but pretreated with insulin + L-NIO, insulin +PTIO, or insulin + carboxy-PTIO were not differ-ent from that of cells exposed to H2O2 alone.These values were lower than those of control cells(media alone; P < .001) or cells exposed to insulin+ H2O2 (P < .001). TEER measurements in cells ex-posed to L-NIO, PTIO, or carboxy-PTIO alone (97 ±0.3, 94 ± 0.3, 97 ± 0.3, respectively) or in com-bination with insulin (97 ± 0.3, 100 ± 0.9, 94 ± 1,respectively) were not different from baseline mea-surements or that of cells exposed to media alone(98 ± 2).

Previous work in endothelial cells has shownthat insulin stimulates eNOS activity, hence, NOproduction via the PI3K/Akt pathway.23,25 Thus,

Fig 4. The concentration of glucose in the incubationmedia did not affect either H2O2 (100 lmol/L)-induceddecreases in transendothelial electrical resistance(TEER) or insulin’s (100 nmol/L) ability to prevent it.TEER was measured before the addition of H2O2 orinsulin (baseline) and at 120 minutes later. Resistancemeasurements were normalized to the average of 3 mea-surements taken before the addition of H2O2 or insulinand expressed as a percentage of baseline. n = 4 pergroup; *P < .01, H2O2 versus controls or H2O2 + insulin,1.0 g/L glucose; **P < .01, H2O2 versus controls or H2O2

+ insulin, 4.5 g/L glucose.

SurgeryJanuary 2006

86 Rath et al

we tested whether treatment of cells with the spe-cific PI3K inhibitor, LY-294002, would block insu-lin’s modulatory effect on H2O2-induced barrierdysfunction (Fig 6). The TEER of LMEC monolay-ers pretreated with LY-294002 + insulin and subse-quently exposed to H2O2 was not different thanthat of cells exposed to H2O2 alone (P > .05) andwas significantly less than baseline measurementsand TEER measurements in control cells (mediaalone; P < .001) or those exposed to insulin +H2O2 (P < .001). The TEER of cells exposed toLY-294002 alone or in combination with insulin(96 ± 1, 97 ± 2, respectively) was not differentfrom baseline measurements (90 ± 3) or that ofcells exposed to media alone (96 ± 1; P < .001),and was greater than that of cells exposed toLY-294002 + insulin + H2O2 (57 ± 1; P < .01).

If NO mediates insulin’s beneficial effect, thenpretreatment of cells with an NO donor shouldreproduce insulin’s prevention of H2O2-inducedendothelial barrier dysfunction. Therefore, LMECwere pretreated with the NO donor SperNO for 30minutes before the addition of H2O2, and TEERwas measured every 15 minutes (Fig 7). The TEERof cells exposed to H2O2 but pretreated with theSperNO was not different than that of control cellsexposed to media alone. SperNO alone had noeffect on TEER (89 ± 1), whereas the TEER of cellsexposed to H2O2 was 35% less than baseline mea-surements (P < .001) and that of time-matched cellsexposed to media alone, SperNO alone, or SperNO+ H2O2 (P < .001). As would be expected from Sper-NO’s action as an NO donor, inhibiting PI3K withLY-294002 did not affect SperNO’s protective effecton H2O2-induced barrier dysfunction (Fig 8).

DISCUSSION

Recent clinical reports have shown that rigorouscontrol of blood glucose by the administration ofinsulin has a profound protective effect in variousacute inflammatory conditions, including myocar-dial ischemia and infarction and critical ill-ness.1,2,5,6,8,9 In one of the first reports of this

Table. Glucose concentrations in the incubationmedia of cells grown in DMEM containing low (1.0g/L) or high (4.5 g/L) glucose and exposed toH2O2 (100 lmol/L) in the presence or absence ofinsulin (100 nmol/L)

(mg/dL) Control H2O2 InsulinInsulin +H2O2

Glucose, 1.0 g/L 74 ± 6 66 ± 1 72 ± 3 74 ± 1Glucose, 4.5 g/L 386 ± 4 387 ± 6 375 ± 6 389 ± 2

effect, Malmberg et al5 found that acute andchronic insulin administration reduced 1 year mor-tality of diabetic patients sustaining an acute myo-cardial infarction by 29% when compared withpatients receiving conventional therapy. van denBerghe et al1 extended this beneficial effect toother critically ill patients when they reportedthat intensive insulin therapy (targeting a blood glu-cose at or below 110 mg/dL) reduced in-hospitalmortality and overall 12-month mortality in criti-cally ill patients by 34% and 42%, respectively,compared with conventionally treated patients.1

Krinsley6 reported nearly identical results in a groupof 800 patients admitted to an intensive care unit.

The potential mechanisms underlying insulin’sprotective effect in critically ill patients has been thesubject of recent reviews.10,11 In addition to insu-lin’s obvious effects on glucose utilization and sub-strate oxidation, it is postulated that insulin mayhave systemic anti-inflammatory effects2,3,8 and di-rect protective effects on endothelial cells.10,11,41,42

Aljada et al41,42 reported that insulin inhibits the ac-tivity of the proinflammatory transcription factor

Fig 5. Inhibition of nitric oxide synthase with L-NIO(100 lmol/L) or scavenging NO with PTIO (50 lmol/L) abolished insulin’s protective effect on endothelialbarrier function. TEER was measured before the addi-tion of H2O2 or insulin (baseline) and at 15-minute in-tervals for 120 minutes. Resistance measurements werenormalized to the average of 3 measurements taken be-fore the addition of insulin and expressed as a percent-age of baseline. n = 8 per group; *P < .05, insulin + H2O2 +PTIO or insulin + H2O2 + L-NIO versus baseline, control,and insulin + H2O2; **P < .01, insulin + H2O2 + PTIO orinsulin + H2O2 + L-NIO versus baseline, control, and insu-lin + H2O2; ***P < .001, insulin + H2O2 + PTIO or insulin +H2O2 + L-NIO versus baseline, control, and insulin +H2O2. The concentrations of H2O2 and insulin were100 lmol/L and 100 nmol/L, respectively.

SurgeryVolume 139, Number 1

Rath et al 87

nuclear factor KB in endothelial cells as well as theexpression of intercellular adhesion molecule-1 and the proinflammatory cytokine monocyte che-moattractant protein-1. Insulin also induces NOrelease by increasing eNOS mRNA and protein ex-pression and eNOS phosphorylation in culturedendothelial cells as well as other cell types.22-25 Insu-lin activates eNOS through the IRS-1/ PI3-K/PDK-1/Akt signaling cascade leading to the phos-phorylation of eNOS on serine 1177.22-25 Othershave shown insulin-stimulated translocation ofeNOS from the membrane fraction to the cytosolof endothelial cells, associated with up-regulationof cGMP, the classic downstream mediator of NO-dependent cellular responses.43

In the current study, we found that insulinprevents oxidant-induced endothelial cell barrierdysfunction in a dose- and time-dependent man-ner. Furthermore, we found that inhibiting eNOSor scavenging NO abolishes insulin’s beneficialeffect, whereas the administration of an NO donorreproduces insulin’s effect. Together these datasuggest that NO mediates insulin’s protective ef-fect on oxidant-induced endothelial cell barrierdysfunction in vitro. This observation is consistent

Fig 6. Inhibition of PI3K with LY-294002 (100 lmol/L)prevents the protective effect of insulin on H2O2-induced endothelial cell barrier dysfunction. TEER wasmeasured before the addition of H2O2 or insulin (base-line) and at 15-minute intervals for 120 minutes. Resis-tance measurements were normalized to the average of3 measurements taken before the addition of H2O2 orinsulin and expressed as a percentage of baseline. n =8 per group; *P < .05, insulin + H2O2 + LY-294002 versusbaseline, control, and insulin + H2O2; **P < .01, insulin +H2O2 + LY-294002 versus baseline, control, and insulin +H2O2; ***P < .001, insulin + H2O2 + LY-294002 versusbaseline, control, and insulin + H2O2. The concentra-tion of H2O2 and insulin were 100 lmol/L and 100nmol/L, respectively.

with other in vivo studies relating endothelial NOproduction to the modulation of microvascularpermeability during basal states and acute inflam-mation.26-29

The lowest plasma concentration of insulin thatprevented H2O2-induced increases in endothelialmonolayer permeability was 100 nmol/L (or1420 lU/mL), which is above the normal physio-logic range of plasma insulin concentrations (15to 300 lU/mL).44 The insulin dose range weused was based on previous studies showing that100 nmol/L insulin (1) prevented rho-mediatedactin cytoskeletal reorganization in vascularsmooth muscle cells,21 (2) stimulated NO releaseby human vascular endothelial cells,25,39 and (3)attenuated TNF-induced endothelial cell nuclearfactor KB activity41 and intercellular adhesionmolecule-142 expression. A 10-fold higher dose ofinsulin (1, 000 lU/mL, 70.4 nmol/L) was found toimprove postischemic recovery of cardiac contrac-tility in an isolated working rat heart model.3

Oxidants are important mediators of endothelialbarrier dysfunction in acute inflammatory condi-tions associated with critical illness and multipleorgan dysfunction, including tissue reperfusioninjury, sepsis, and hemorrhagic shock (Reviewedin Ref. 45). Although the precise mechanisms

Fig 7. The NO donor SperNO (100 lmol/L) preventsH2O2 (100 lmol/L)-induced endothelial cell barrierdysfunction in a manner analogous to insulin. TEERwas measured before the addition of H2O2 or SperNOand at 15-minute intervals for 120 minutes. Resistancemeasurements were normalized to the average of 3 mea-surements taken before the addition of H2O2 or insulinand expressed as a percentage of baseline. n = 10 pergroup; *P < .05, H2O2 versus baseline, control, andSpermine NONOate + H2O2; **P < 0.01, H2O2 versusbaseline, control, and Spermine NONOate + H2O2;***P < .001, H2O2 versus baseline, control, and Sperm-ine NONOate + H2O2.

SurgeryJanuary 2006

88 Rath et al

remain under active investigation, there is substan-tial support for the idea that oxidants cause endo-thelial cell barrier dysfunction by activating cellsignaling pathways that control the organization ofthe actin cytoskeleton12-14 and the integrity of inter-cellular junctional complexes, ie, adherens andtight junctions.46

Preliminary observations suggest that oxidantsincrease microvascular permeability at least in partby rho kinase-- and MLCK-mediated increased MLCphosphorylation. Chiba et al47 found that inhibit-ing rho kinase prevents H2O2-induced pulmonaryedema in an isolated perfused lung model. More-over, preliminary studies in our laboratory suggestthat inhibitors of rho kinase or MLCK preventH2O2-induced increases in endothelial permeabil-ity in vitro (Rath et al, unpublished observations).These observations are consistent with many studiesrelating thrombin-induced rho kinase activation toMLC phosphorylation, actin stress fiber formation,endothelial cell contraction, and increased perme-ability.15,16 As discussed above, phosphorylated MLCresults from stimulation of Ca2+-dependent MLCKor Rho-mediated inactivation of MLC phospha-tase.17 The latter pathway involves the downstreameffector, p160 Rho kinase.16,17 (Reviewed inRef. 18,19) Others have related H2O2- and throm-bin-induced stress fiber formation and increasedendothelial permeability to p38 MAP kinase and

Fig 8. Inhibiting PI3K with LY-294002 (100 lmol/L) didnot affect Spermine NONOate’s (SperNO, 100 lmol/L)beneficial effect on H2O2 (100 lmol/L)-induced barrierdysfunction. TEER was measured before the addition ofH2O2, SperNO or LY-294002 and at 15-minute intervalsfor 120 minutes. Resistance measurements were normal-ized to the average of 3 measurements taken before theaddition of H2O2 or insulin and expressed as a percent-age of baseline. n = 4 per group; ***P < .001, H2O2

versus baseline, control, SperNO, SperNO + H2O2,SperNO + H2O2 + LY-294002.

ERK 1/2 activity.12-14,33,48,49 It is postulated thatthese signaling pathways elicit endothelial cytoskel-etal reorganization and barrier disruption by aMLCK-independent mechanism via a calmodulinkinase II–mediated phosphorylation of the actin-binding protein, caldesmon.33,48,49

The mechanism by which insulin-induced NOrelease protects endothelial cell barrier functionduring oxidant exposure is unknown. Studies invascular smooth muscle cells have related insulin-induced NO release to inhibition of rho/rhokinase--mediated actin cytoskeletal reorganiza-tion.20,21 In this model, it is postulated that NO-induced increases in cellular cGMP activate PKG,which directly phosphorylates rho A at Ser 188,leading to translocation of rho A from the mem-brane to the cytosol where it is inactivated. This,in turn, inhibits rho kinase activity and promotesMLC phosphatase-mediated dephosphorylation ofMLC, thus, inhibiting cell contraction.50,51 Alter-natively, NO-induced increases in cGMP may alsoactivate cGMP-phosphodiesterase, leading to anincrease in levels of cAMP. Elevated cAMP, in turn,inhibits p42/44 MAPK activity,52 which has beenimplicated in oxidant-mediated, MLCK-indepen-dent cytoskeletal reorganization and altered bar-rier function in endothelial cells.53

Adjacent endothelial cells are joined by tightjunctions and adherens junctions. Proinflammatorymediators, including oxidants, cause the dissolutionof adherens junctions and the redistribution of oneof its key components, vascular endothelial (VE)cadherin. VE cadherin is a transmembrane proteinin which extracellular domain mediates intercellu-lar homotypic adhesion and in which intracellulardomain is linked to the actin cytoskeleton. Thisassociation between VE cadherin and the actincytoskeleton is required for the stability of thecomplex and the full control of junctional perme-ability.54 Previous studies in our laboratory have re-lated junctional protein organization to the activityof Rho and members of the MAP kinase familydownstream from Rho, specifically p38 MAPK, andthe ERK1/ERK2 kinases.32,55,56 These latter 2 path-ways have been shown recently to mediate the effectof H2O2 on endothelial permeability14,46,49 and maycoordinate changes in the actin cytoskeleton to theintegrity of the adherens junctions. The effect of in-sulin or NO on adherens junctions and their proteinconstituents is unknown.

Hyperglycemia has an important effect on thephenotype and function of the endothelial cellphenotype and function. In the current study,endothelial responsiveness to H2O2 was similar re-gardless of the concentration of glucose in the

SurgeryVolume 139, Number 1

Rath et al 89

incubation mixture (1 g/L vs 4.5 g/L). Similarly,the ability of insulin to prevent H2O2-induced bar-rier dysfunction was no different between thosegroups incubated in high or normal concentrationsof glucose (Fig 4). Others, however, have shownthat exposure to hyperglycemic conditions may im-pair endothelium-dependent vascular relaxation,an effect postulated to be caused by the generationof superoxide anion.57 Ding et al58 found thatexposure of cultured human coronary endothelialcells to 25 mmol/L glucose for 7 days diminishedbasal and stimulated NO production and eNOSexpression, and Srinivasan et al59 reported similarobservations in cultured human aortic endothelialcells exposed to 7 days, but not 4 hours, of hypergly-cemic media. In our study, the cells were incubatedin normoglycemic or hyperglycemic conditions for7 days before exposure to H2O2 and insulin. De-spite this period of exposure to high glucose con-centrations, insulin’s ability to abolish the effectof H2O2 on endothelial monolayer permeabilityremained intact (Fig 4).

Current studies are underway in our laboratoryto examine the effect of prolonged hyperglycemiaon endothelial responsive to insulin and its effecton oxidant stress. The importance of these studiesresides in the observations that hyperglycemiaalters the intracellular reduction-oxidation (re-dox) state of the cell to promote the generationof oxygen free radicals.59 Hyperglycemia-inducedsuperoxide anion production has been implicatedin the reduction of endothelial eNOS mRNA andprotein expression59 and may react with and inac-tivate NO or form peroxynitrite, which can oxidizeand inactivate the eNOS cofactor tetrahydrobiop-terin60 or, in and of itself, produce endothelialcell barrier dysfunction.61

Our study suggests that insulin prevents oxi-dant-mediated endothelial barrier dysfunction byinducing the release of NO. We postulate that NOinhibits rho/rho kinase-mediated phosphorylationof MLC, thus inhibiting actin-myosin, and hence,cellular contraction. These data may provide evi-dence for a direct beneficial effect of insulin onthe microvascular endothelium and may provide apotential mechanism underlying insulin’s protec-tive effect in critically ill patients.

REFERENCES

1. van den Berghe G, Wouters P, Weekers F, Verwaest C, Bruy-ninckx F, Schetz M, et al. Intensive insulin therapy in criti-cally ill patients. N Engl J Med 2001;345:1359-67.

2. Jeschke MG, Klein D, Herndon DN. Insulin treatment im-proves the systemic inflammatory reaction to severe trauma.Ann Surg 2004;239:553-60.

3. Fischer-Rasokat U, Doenst T. Insulin-induced improvementof postischemic recovery is abolished by inhibition of pro-tein kinase C in rat heart. J Thorac Cardiovasc Surg 2003;126:1806-12.

4. Melin J, Hellberg O, Larsson E, Zezina L, Fellstrom BC. Pro-tective effect of insulin on ischemic renal injury in diabetesmellitus. Kidney Int 2002;61:1383-92.

5. Malmberg K, Ryden L, Efendic S, Herlitz J, Nicol P, Walden-strom A, et al. Randomized trial of insulin-glucose infusionfollowed by subcutaneous insulin treatment in diabetic pa-tients with acute myocardial infarction (DIGAMI study): ef-fects on mortality at 1 year. J Am Coll Cardiol 1995;26:57-65.

6. Krinsley JS. Effect of intensive glucose management proto-col on the mortality of critically ill adult patients. MayoClin Proc 2004;79:992-1000.

7. Lewis KS, Kane-Gill SL, Bobek MB, Dasta JF. Intensive insu-lin therapy for critically ill patients. Ann Pharmacother2004;38:1243-51.

8. Brix-Christensen V, Andersen SK, Andersen R, Mengel A,Dyhr T, Andersen NT, et al. Acute hyperinsulinemia re-strains endotoxin-induced systemic inflammatory response:an experimental study in a porcine model. Anesthesiology2004;100:861-70.

9. Furnary AO, Gao G, Grunkemeier GL, Wu YX, Zerr KJ,Bookin SO, et al. Continuous insulin infusion reduces mor-tality in patients with diabetes undergoing coronary arterybypass grafting. J Thorac Cardiovasc Surg 2003;125:1007-21.

10. van den Berghe G. How does blood glucose control with in-sulin save lives in intensive care? J Clin Invest 2004;114:1187-95.

11. Ritchie SA, Ewart M-A, Perry CG, Connell JMC, Salt IP. Therole of insulin and the adipocytokines in regulation of vas-cular endothelial function. Clin Sci (London) 2004;107:519-32.

12. Zhao Y, Davis HW. Hydrogen peroxide-induced cytoskeletalrearrangement in cultured pulmonary endothelial cells.J Cell Physiol 1998;174:370-9.

13. Kevil CG, Oshima T, Alexander JS. The role of p38 MAP ki-nase in hydrogen peroxide mediated endothelial solute per-meability. Endothelium 2001;8:107-16.

14. Nguyen A, Chen P, Cai H. Role of CaMKII in hydrogen per-oxide activation of ERK1/2, p38 MAPK, HSP27 and actinreorganization in endothelial cells. FEBS Lett 2004;572:307-13.

15. van Nieuw Amerongen GP, Draijer R, Vermeer MA, vanHinsbergh VWM. Transient and prolonged increase in en-dothelial permeability induced by histamine and thrombin:role of protein kinases, calcium, and rho A. Circ Res 1998;83:1115-23.

16. van Nieuw Amerongen GP, van Delft S, Vermeer MA, Col-lard JG, van Hinsbergh VWM. Activation of rho A by throm-bin in endothelial hyperpermeability: role of rho kinase andprotein tyrosine kinases. Circ Res 2000;87:335-40.

17. Verin AD, Patterson CE, Day MA, Garcia JGN. Regulationof endothelial cell gap formation and barrier function bymyosin-associated phosphatase activities. Am J Physiol 1995;269:L99-108.

18. Alexander JS. Rho, tyrosine kinase, Ca2+, and junctions inendothelial hyperpermeability. Circ Res 2000;87:268-71.

19. van Hinsbergh VWM, van Nieuw Amerongen GP. Intracellu-lar signaling involved in modulating human endothelialbarrier function. J Anat 2002;200:549-60.

20. Sandu OA, Ito M, Begum N. Insulin utilizes NO/cGMPpathway to activate myosin phosphatase via Rho inhibitionin vascular smooth muscle. J Appl Physiol 2001;91:1475-82.

SurgeryJanuary 2006

90 Rath et al

21. Begum N, Sandu OA, Duddy N. Negative regulation of Rhosignaling by insulin and its impact on actin cytoskeletonorganization in vascular smooth muscle cells: role of nitricoxide and cyclic guanosine monophosphate signaling path-ways. Diabetes 2002;51:2256-63.

22. Montagnani M, Chen H, Barr AV, Quon MJ. Insulin-stimu-lated activation of eNOS is independent of Ca2+ but re-quires phosphorylation by Akt at Ser1179. J Biol Chem2001;276:30392-8.

23. Zeng G, Nystrom FH, Ravichandran LV, Cong LN, Kirby M,Mostowski H, et al. Roles for insulin receptor, PI3-kinase,and Akt in insulin-signaling pathways related to productionof nitric oxide in human vascular endothelial cells. Circula-tion 2000;101:1539-45.

24. Kuboki K, Jiang ZY, Takahara N, Ha SW, Igarashi M, Yamau-chi T, et al. Regulation of endothelial constitutive nitric ox-ide synthase gene expression in endothelial cells and invivo: a specific vascular action of insulin. Circulation 2000;101:676-81.

25. Montagnani M, Ravichandran LV, Chen H, Esposito DL,Quon MJ. Insulin receptor substrate-1 and phosphoinosi-tide-dependent kinase-1 are required for insulin-stimulatedproduction of nitric oxide in endothelial cells. Mol Endocri-nol 2002;16:1931-42.

26. Persson J, Ekelund U, Grande P-O. Nitric oxide and prosta-cyclin play a role in the regulation of microvascular proteinand hydraulic permeability in cat skeletal muscle. Microcir-culation 2003;10:233-43.

27. Victorino GP, Wisner DH, Tucker VL. Basal release of nitricoxide and its interaction with endothelin-1 on single vesselhydraulic permeability. J Trauma 2001;50:535-9.

28. Jones SP, Greer JJM, Kakkar AK, Ware PD, Turnage RH,Hicks M, et al. Endothelial nitric oxide synthase overexpres-sion attenuates myocardial reperfusion injury. Am J Physiol2004;286:H276-82.

29. Ozaki M, Kawashima S, Hirase T, Yamashita T, Namiki M,Inoue N, et al. Overexpression of endothelial nitric oxidesynthase in endothelial cells is protective against ischemia-reperfusion injury in mouse skeletal muscle. Am J Pathol2002;160:1335-44.

30. Nwariaku FE, Rothenbach P, Liu Z, Zhu X, Turnage RH,Terada LS. Rho inhibition decreases TNF-induced endothe-lial MAPK activation and monolayer permeability. J ApplPhysiol 2003;95:1889-95.

31. Borbiev T, Verin AD, Birukova A, Liu F, Crow MT, GarciaJGN. Role of CaM kinase II and ERK activation in throm-bin-induced endothelial cell barrier dysfunction. Am J Phys-iol 2003;285:L43-54.

32. Grosser N, Schroder H. Aspirin protects endothelial cellsfrom oxidant damage via the nitric oxide-cGMP pathway.Arterioscler Thromb Vasc Biol 2003;23:1345-51.

33. Dedkova EN, Ji X, Lipsius SL, Blatter LA. Mitochondrial cal-cium uptake stimulates nitric oxide production in mito-chondria of bovine vascular endothelial cells. Am J Physiol2004;286:C406-15.

34. Lee PC, Kibbe MR, Schuchert MJ, Stolz DB, Watkins SC,Griffith BP, et al. Nitric oxide induces angiogenesis andupregulates avB3 integrin expression on endothelial cells.Microvasc Res 2000;60:269-80.

35. Shimizu S, Ishii M, Iwasaki M, Shiota K, Yamamoto T, KiuchiY. Carboxy-PTIO increases the tetrahydrobiopterin level inmouse brain microvascular endothelial cells. Jpn J Pharma-col 2001;87:51-60.

36. Dimmeler S, Fleming I, Fisslthaler B, Hermann C, Busse R,Zeiher AM. Activation of nitric oxide synthase in endothelial

cells by Akt-dependent phosphorylation. Nature 1999;399:601-5.

37. Fulton D, Gratton J-P, McCabe TJ, Fontanaq J, Fujio Y, WalshK, et al. Regulation of endothelium-derived nitric oxide pro-duction by the protein kinase Akt. Nature 1999;399:597-601.

38. Montagnani M, Chen H, Barr VA, Quon MJ. Insulin-stimu-lated activation of eNOS as independent of Ca2+ but re-quires phosphorylation by Akt at Ser1179. J Biol Chem2001;276(32):30392-8.

39. Fisslthaler B, Benzing T, Busse R, Fleming I. Insulin en-hances the expression of the endothelial nitric oxide syn-thase in native endothelial cells: a dual role for Akt andAP1. Nitric Oxide 2003;8:253-61.

40. Polte T, Oberle S, Schroder H. Nitric oxide protects endo-thelial cells from tumor necrosis factor alpha-mediated cyto-toxicity: possible involvement of cyclic GMP. FEBS Lett1997;409:46-8.

41. Aljada A, Ghanim H, Saadeh R, Dandona P. Insulin inhibitsNFkappaB and MCP-1 expression in human aortic endothe-lial cells. J Clin Endocrinol Metab 2001;86:450-3.

42. Aljada A, Saadeh R, Assian E, Ghanim H, Dandona P. Insulininhibits the expression of intercellular adhesion molecule-1 by human aortic endothelial cells through stimulation bynitric oxide. J Clin Endocrinol Metab 2000;85:2572-5.

43. Konopatskaya O, Whatmore JL, Tooke JE, Shore AC. Insu-lin and lysophosphatidylcholine synergistically stimulateNO-dependent cGMP production in human endothelialcells. Diabet Med 2003;20:838-45.

44. Tillil H, Shapiro ET, Miller MA, Karrison T, Frank BH, Gal-loway JA, et al. Dose-dependent effects of oral and intrave-nous glucose on insulin secretion and clearance innormal humans. Am J Physiol 1988;254:E349-57.

45. Maier RV. Pathogenesis of multiple organ dysfunctionsyndrome—endotoxin, inflammatory cells, and their medi-ators: cytokines and reactive oxygen species. Surgical Infec-tions 2000;1:197-205.

46. Kevil CG, Ohno N, Gute DC, Okayama N, Robinson SA,Chaney E, et al. Role of cadherin internalization in hydro-gen peroxide-mediated endothelial permeability. Free Rad-ical Biol Med 1998;24:1015-22.

47. Chiba Y, Ishii Y, Kitamura S, Sugiyama Y. Activation of rho isinvolved in the mechanism of hydrogen-peroxide-inducedlung edema in isolated, perfused rabbit lung. MicrovascRes 2001;62:164-71.

48. Borbiek T, Birukova A, Liu F, Nurmukhambetova S, Ger-thoffer WT, Garcia JGN, et al. P38 MAP kinase-dependentregulation of endothelial cell permeability. Am J Physiol2004;287:L911-8.

49. Usatyuk PV, Natarajan V. Role of mitogen-activated proteinkinases in 4-hydroxy-2-nonenal-induced actin remodelingand barrier function in endothelial cells. J Biol Chem2004;279:11789-97.

50. Sauzeau V, Le Jeune H, Cario-Toumaniantz C, Smolenski A,Lohman SM, Bertoglio J, et al. Cyclic GMP-dependent pro-tein kinase signaling pathway inhibits RhoA-induced Ca2+

sensitizatin of contraction in vascular smooth muscle.J Biol Chem 2000;275:21722-9.

51. Etter EF, Et M, Wardle RL, Brautigan DL, Murphy RA. Acti-vation of myosin light chain phosphatase in intact arterialsmooth muscle during nitric oxide-induced relaxation.J Biol Chem 2001;276:34681-5.

52. Jiao X, Gonzalez-Cabrera PJ, Xiao L, Bradley ME, Abel PW,Jeffries WB. Tonic inhibitory role for cAMP in alpha(1a)-adrenergic receptor coupling to extracellular signal-relatedkinases 1/2. J Pharmacol Exp Ther 2002;303:247-56.

SurgeryVolume 139, Number 1

Rath et al 91

53. Kevil CG, Oshima T, Alexander B, Coe LL, Alexander JS.H2O2-mediated permeability: role of MAPK and occluding.Am J Physiol 2000;279:C21-30.

54. Navarro P, Caveda L, Breviario F, Lampungnani MG, DejanaE. Catenin-dependent and independent functions of vascu-lar endothelial cadherin. J Biol Chem 1995;270:30965-72.

55. Nwariaku FE, Liu Z, Zhu X, Turnage RH, Sarosi GA, TeradaLS. Tyrosine phosphorylation of vascular endothelial cad-herin and the regulation of microvascular permeability. Sur-gery 2002;132:180-5.

56. Nwariaku FE, Chang J, Zhu X, Liu Z, Duffy SL, HalaihelNH, et al. The role of p38 MAP kinase in tumor necrosis fac-tor-induced redistribution of vascular endothelial cadherinand increased endothelial permeability. Shock 2002;18:82-5.

57. Beckman JA, Goldfine AB, Gordon MB, Creager MA. Ascor-bate restores endothelium-dependent vasodilation impaired

by acute hyperglycemia in humans. Circulation 2001;103:1618-23.

58. Ding Y, Vaziri ND, Coulson R, Kamanna VS, Roh DD. Effectsof simulated hyperglycemia, insulin, and glucagons on en-dothelial nitric oxide synthase expression. Am J Physiol En-docrinol Metab 2000;279:E11-7.

59. Srinivasan S, Hatley ME, Bolick DT, Palmer LA, Edelstein D,Brownlee M, et al. Hyperglycaemia-induced superoxideproduction decreases eNOS expression via AP-1 activationin aortic endothelial cells. Diabetologia 2004;47:1727-34.

60. Marletta MA. Nitric oxide synthase: aspects concerningstructure and catalysis. Cell 1994;78:927-30.

61. Marumo T, Noll T, Schini-Kerth VB, Harley EA, Duhault J,Piper HM, et al. Significance of nitric oxide and peroxyni-trite in permeability changes of the retinal microvascularendothelial cell monolayer induced by vascular endothelialgrowth factor. J Vasc Res 1999;36:510-5.

Surgery posts in-press articles online in advance of their appearance in the print edition of the Journal. They are available at the Surgery website www.mosby.com/surgery at the “Articles in Press” link, as well as at Elsevier’s ScienceDirect website www.sciencedirect.com. Each print article will acknowledge the e-publication date (the date when the article first appeared online). As soon as an article is published online, it is fully citable through use of its Digital Object Identifier (DOI). Please visit the Surgery website and view our hot-off-the-wire articles through the “Articles in Press” link.