Mechanisms of Pulmonary Edema Induced by TumorNecrosis ... filedata indicate that TNF induces neutrophil-dependent pulmonary edema associated with increased Pp, (mediatedbyTXA2andPAF),increasedKtc

68

Mechanisms of Pulmonary Edema Induced byTumor Necrosis Factor-a

Denise C. Hocking, Patricia G. Phillips, Thomas J. Ferro, and Arnold Johnson

We tested the hypothesis that human recombinant tumor necrosis factor-a (TNF) promotespulmonary edema by neutrophil-dependent effects on the pulmonary vasculature. The isolatedguinea pig lung was perfused with phosphate-buffered Ringer's solution with or without humanneutrophils. The infusion of neutrophils (9x 106 total) into lungs isolated after the in vivoadministration of TNF (3.2x 10' units/kg) resulted in weight gain (+1.951±0.311 g versus-0.053 0.053 g in control) and an increase in the lung (wet-dry)-to-dry weight ratio (8.3±0.5versus 6.0±0.2 in control), indicating the formation of pulmonary edema. The neutrophil-dependent pulmonary edema induced by TNF was associated with a combination of increasedcapillary permeability (capillary filtration coefficient [KJ,si 0.170+0.048 g/min/cm H20/g at 30minutes versus 0.118±0.008 g/min/cm H20/g at baseline) and increased pulmonary capillarypressure (Ppr, 12.8±0.8 cm H20 at 60 minutes versus 6.0±0.3 cm H20 at baseline). The PP,increase was mediated by thromboxane A2 (TXA2) because the TXA2 synthetase inhibitorDazoxiben (0.5 mM) prevented the effect (Pp,, 6.7±0.6 cm H20 at 60 minutes with Dazoxiben),and thromboxane B2 (TXB2) levels were increased in the pulmonary venous effluent (5,244±599pg/ml at 60 minutes versus 60±13 pg/ml at baseline). Studies using WEB-2086 (37 ,uM), aplatelet activating factor (PAF) receptor antagonist, indicated that PAF mediated the increasedvascular permeability (Kt,C, 0.107±0.014 g/min/cm H20/g at 30 minutes using WEB-2086) and,in part, the increased Pp, (Pp,, 8.4±0.7 cm H20 at 60 minutes using WEB-2086). In addition,alterations of endothelial peripheral actin bands were noted after TNF administration. Thedata indicate that TNF induces neutrophil-dependent pulmonary edema associated withincreased Pp, (mediated by TXA2 and PAF), increased Ktc (mediated by PAF), and changes inendothelial peripheral actin bands. (Circulation Research 1990;67:68-77)

T umor necrosis factor-a (TNF) is a monokineproduced by monocytes and macrophages inresponse to endotoxin, interleukin-2, and

mitogens.1-3 TNF induces many immune functions,including endothelial secretion of interleukin-1,4,5endothelial procoagulant activity,6,7 and neutrophil(PMN) activation.8,9 TNF is a mediator of endotox-emia because 1) TNF infusion induces an endotox-emia-like state in animals,' 2) passive immunizationagainst TNF can prevent the multiple organ failure

From the Research Service, Veterans Administration MedicalCenter (P.G.P., A.J., T.J.F.), and the Departments of Physiology(D.C.H., A.J.) and Medicine (T.J.F.), The Albany Medical Collegeof Union University, Albany, NY.

Supported by the Veterans Administration (A.J.), the AmericanHeart Association of New York State (A.J., P.G.P.), the AmericanLung Association of New York State (T.J.F.), and the PottsMemorial Foundation, Catskill, N.Y. (T.J.F.). D.C.H. is a gradu-ate student supported by predoctoral training grant HL-07194-13in the Department of Physiology of the Albany Medical College.Address for correspondence: Arnold Johnson, PhD, Research

Service (151D), Veterans Administration Medical Center, 113Holland Avenue, Albany, NY 12208.

Received October 2, 1989; accepted February 6, 1990.

seen after endotoxin challenge,' 3) genetic deficiencyof TNF production inhibits the development of gram-negative shock,10 and 4) the anatomic evidence ofinjury and edema seen in the lungs and other organs inTNF-treated mice is similar to that seen inendotoxemia.1The mechanism of TNF-induced pulmonary

edema is not clear. TNF directly increases pulmonaryendothelial permeability to protein in vivo"l andinduces the rearrangement of human endothelialmonolayers in vitro.12 TNF may also promote pulmo-nary edema by increasing pulmonary capillarypressure.11 Platelet activating factor (PAF) maymediate TNF-induced pulmonary edema because1) TNF stimulates the release of PAF from endothe-lial cells, macrophages, and PMN,13 and 2) PAF caninduce pulmonary vasoconstriction and increases invascular permeability.14 Arachidonic acid metabo-lites such as thromboxane A2 (TXA2) may play arole in the pathogenesis of TNF-induced pulmonaryedema because cyclooxygenase inhibition partiallyblocks the hemodynamic and other effects of TNFin vivo.11"15

by guest on June 24, 2017http://circres.ahajournals.org/

Dow

nloaded from

http://circres.ahajournals.org/

Hocking et al Tumor Necrosis Factor-Induced Pulmonary Edema 69

The role of PMN in the pathogenesis of TNF-induced pulmonary edema is also unclear, but severalof the in vitro activities of TNF indicate that PMNmay mediate this effect. TNF can 1) activate PMN torelease oxygen radicals,8 2) increase PMN adherenceto the endothelium,16 3) stimulate endothelial cells toproduce a PMN chemotactic factor,'7 and 4) induceendothelial cell lysis by activated PMN.18 Pulmonaryedema mediated by TNF in vivo in the guinea pig isprevented by prior PMN depletion,1" whereas pulmo-nary edema occurs despite prior PMN depletion insheep.'9We postulated that TNF promotes pulmonary

edema predominantly by PMN-dependent effects onthe pulmonary vasculature. We tested this hypothesisusing the isolated guinea pig lung perfused withphosphate-buffered Ringer's solution with or withouthuman PMN. The TNF (3.2x 105 units/kg) wasadministered in vivo 18 hours before study to allowsufficient time for presumptive endothelial-derivedPMN chemotactic and adherence factors to beexpressed.16 The role of arachidonic acid metabolitesand PAF was evaluated because these factors maymediate the effects of TNF."11'3,1 We also studiedthe effects of TNF on perfused lung endothelial actinfilament morphology because peripheral actin bandsare thought to play a role in paracellular permeabilityby stabilizing intercellular junctional proteins,20 andchanges in the peripheral actin bands coincide withincreases in vascular permeability.2'

Materials and MethodsIsolated Perfused Lung PreparationThe isolated perfused lung studies were performed

as previously described.2',22 In brief, Hartley guineapigs (300-500 g; Charles River Laboratories, Wil-mington, Mass.) were anesthetized with sodium pen-tobarbital (5 mg/100 g i.p.; Abbott Laboratories,Chicago, Ill.), and the trachea was cannulated. Thechest was opened by median sternotomy, and 700units/kg heparin (Invenex, Chagrin Falls, Ohio) wasadministered by intracardiac puncture injection. Thelungs and heart were excised and suspended from acounter-weighted beam balance attached to a forcetransducer (model 100DCD, Shaevitz, Pennsauken,N.J.). The airway pressure was maintained at 1 cmH20 with 95% 02 and 5% CO2 throughout theexperiment. The pulmonary artery and left atriumwere cannulated, and the left atrial pressure was setat 4 cm H20. Lung perfusion (6 ml/min/100 g bodywt) was maintained by a peristaltic roller perfusionpump (model 1215, Harvard Apparatus, Millis,Mass.). Recirculation of the perfusate (300 ml) wasbegun after the venous effluent was clear of cellularelements. The perfusate consisted of a phosphate-buffered Ringer's solution with 5.55 mM dextroseand 0.5 g% bovine serum albumin (fraction V, SigmaChemical Co., St. Louis, Mo.).22 The system wasmaintained at a constant temperature of 370 C, andthe pH was maintained between 7.30 and 7.40. The

pulmonary artery pressure (Ppa) was monitored usinga pressure catheter (PE-50) that was inserted intothe pulmonary artery cannula and connected to apressure transducer (model P1OEZ, Spectramed,Oxnard, Calif.). The pulmonary venous pressure(Pp,) was monitored similarly except that the pressurecatheter was inserted into the left atrial catheter. ThePpa,Ppv, and change in weight were recorded contin-uously (model 3000S recorder, Gould Instruments,Cleveland, Ohio). Rarely, lungs were excluded fromstudy because a spontaneous increase in Ppa, Ppv5 orweight occurred during the 15-minute baseline per-fusion period following lung isolation.

Pulmonary Capillary Pressure, Pulmonary ArterialResistance, and Pulmonary Venous ResistanceThe pulmonary capillary pressure (Pp) was esti-

mated using the double occlusion method.23 Thearterial inflow and venous outflow lines were simul-taneously occluded using in-line solenoid valves,and the Pp, was estimated by measuring the subse-quent equilibrium pressure. The Ppc estimate wasused to determine both the pulmonary arterialresistance (Ra) and the pulmonary venous resis-tance (Rv) from the equations Ra=(Ppa-Ppc)/Q andRv=(PpX-Ppv)/Q, where 0 equals flow.

Pulmonary Capillary Filtration CoefficientThe pulmonary capillary filtration coefficient (Kfc),

a measure of pulmonary vascular permeability towater, was estimated using the method of Drake etal.24 After an isogravimetric period, the PPV waselevated by 4 cm H20 for 5.5 minutes, and theresulting increase in lung weight was recorded. Toevaluate changes in interstitial volume, the late com-ponent (from 2 to 5.5 minutes after Ppv elevation) ofweight change was analyzed. The rate of change inweight was extrapolated back to the time of the initialPpV elevation using the anti-log10 of the y intercept ofthe linear regression line. The Kfc was calculated bydividing the extrapolated rate of weight change bythe change in Pp, and the dry lung weight.

Lung (Wet-Dry)-to-Dry Weight RatioAt the end of the experimental period, the lungs

were removed, cleared of all extrapulmonary tissue,and weighed (wet weight). After the lungs were driedat 750 C for 3 days, they were again weighed (dryweight), and the lung (wet-dry)-to-dry weight ratio(W/D weight ratio) was calculated.

Tumor Necrosis Factor-oaTNF (highly purified recombinant human TNF-a

from Escherichia coli, courtesy of Dr. Abla Creasey,Cetus Corporation, Emeryville, Calif.)25 having aspecific activity of 24 x 106 units/mg26 was used. Thepreparation contained less than 0.8 pg of endotoxinper 106 units of TNF activity by standard limulusassay. TNF (3.2x 105 units/kg) was injected intraperi-toneally 18 hours before lung isolation. To control forendotoxin contamination and possible effects of the


Dow

nloaded from


70 Circulation Research Vol 67, No 1, July 1990

TNF vehicle, TNF was heated to 900 C (a tempera-ture that does not inactivate endotoxin27) for 45minutes. After heating, no residual TNF activity wasnoted using the standard L929 fibroblast lysis assay.28

Preparation of NeutrophilsHuman PMN were isolated from venous blood

obtained from healthy donors as described.29 Theblood was collected in tubes containing EDTA (0.1M) and layered onto neutrophil isolation medium(NIM, Los Alamos Diagnostics, Los Alamos, N.M.).The blood-NIM was centrifuged at 400g for 30 min-utes at room temperature. The PMN-rich layer wasremoved and pooled in a 50-ml conical tube withequal volume of Hanks' balanced salt solution with-out calcium and magnesium (HBSS, GIBCO Labo-ratories, Grand Island, N.Y.). The PMN were thencentrifuged at 300g for 10 minutes at room temper-ature. The supernatant was discarded, and theremaining red blood cells were removed by gentlelysis by adding hypotonic (0.2% NaCl) phosphate-buffered saline to the cells for 15 seconds followed bythe immediate addition of hypertonic (1.6% NaCl)phosphate-buffered saline. After centrifugation(300g, 5 minutes, room temperature) and removal ofthe supernatant, PMNs were resuspended in 1 mlHBSS containing 5.5 mM glucose. PMNs werecounted and diluted in HBSS containing 5.5 mMglucose to give a final concentration of 3 x 106 PMN/ml. Greater than 98% of the cells counted werePMNs by light microscopy. PMN viability was alwaysgreater than 97% by trypan blue exclusion.

DazoxibenTo test the role of TXA2 in the response to TNF,

Dazoxiben (Pfizer, Groton, Conn.), a specific inhibitorof TXA2 synthetase,30 was used. The Dazoxiben wasadded to the perfusate before the baseline period toachieve a final concentration of 0.5 mM. We havepreviously shown this dose to inhibit TXA2-dependentincreases in Ppc in the isolated perfused lung.22

WEB-2086To assess the role of platelet activating factor

(PAF) in the response to TNF, WEB-2086 (Boeh-ringer Ingelheim, Ridgefield, Conn.), a specific inhib-itor of the PAF receptor, was used.3' WEB-2086 was

added to the perfusate before the baseline period toachieve a final concentration of 37 ,M. This dose ofWEB-2086 has been shown to block the hemody-namic responses to PAF.31

Control StudiesControl group (n=7). The lung isolation and 15-

minute baseline perfusion period were as describedabove. The Ppa5 Ppc change in weight, and Kfc were

measured at baseline and at 30 and 60 minutes.

Tumor Necrosis Factor-a In Vivo StudiesTNF group (n=8). TNF was administered as out-

lined above. The lungs were otherwise treated as inthe control group.TNF+Dazoxiben group (n=5). Guinea pigs were

treated as in the TNF group. After lung isolation,Dazoxiben was added to the perfusate as describedabove. The lungs were then treated as in the TNFgroup.Dazoxiben group (n=3). This group was studied to

control for possible independent effects of Dazoxi-ben. Dazoxiben was added to the perfusate beforethe baseline period as outlined above, and the lungswere treated as in the control group.TNF+WEB-2086 group (n=6). Guinea pigs were

treated as in the TNF group. After lung isolation,WEB-2086 was added to the perfusate as describedabove. The lungs were then treated as indicated inthe TNF group.WEB-2086 group (n=3). This group was studied to

control for possible independent effects of WEB-2086. WEB-2086 was added to the perfusate beforethe baseline period as outlined above, and the lungswere treated as in the control group.

Heat-inactivated TNF group (n=3). These experi-ments were performed identically to the TNF group,using TNF that had been heat-inactivated at 900 Cfor 45 minutes.28

Neutrophil StudiesPMN group (n=6). Lung isolation and baseline

measurements were obtained after a 10-minuteperiod of stable lung weight, and PMNs (9x 106suspended in 3.0 ml HBSS containing 5.5 mM glu-cose) were infused into the lung over 5 minutes viathe pulmonary artery port. Measurements weretaken as in the control group.TNF+PMN group (n=10). Guinea pigs were

administered TNF as in the TNF group. The lungswere then isolated and treated as in the PMN group.TNF+ WEB-2086+PMN group (n=11). The proto-

col for this group was similar to the TNF+WEB-2086group. PMN were then infused and measurementsobtained as in the PMN group.WEB-2086+PMN group (n=3). This group was

studied to control for possible independent effects ofWEB-2086. WEB-2086 was added to the perfusatebefore the baseline period as outlined above, and thelungs were treated as in the PMN group.TNF+Dazoxiben+PMNgroup (n=8). The protocol

for this group was similar to the TNF+Dazoxibengroup. PMN were then infused and measurementsobtained as in the PMN group.Dazoxiben+PMN group (n=3). This group was

studied to control for possible independent effects ofDazoxiben. Dazoxiben was added to the perfusatebefore the baseline period as outlined above, and thelungs were treated as in the PMN group.


Dow

nloaded from



Radioimmunoassay ofArachidonic Acid MetabolitesLung venous effluent samples were collected at 0, 30,

and 60 minutes for measurement of thromboxane B2(TXB2), the stable product of TXA2, and 6-ketopros-taglandin-F10 (6-keto-PGFia), the stable product ofprostacyclin. Concentrations were determined using adouble-antibody radioimmunoassay as described.22

Perfused Lung Pulmonary Artery Endothelial ActinFilament MorphologyThe perfusion fixation technique previously

described by Forssman et a132 was used. Isolatedlungs were prepared as indicated above. At the endof the experimental period, the lungs were perfusedfor 3 minutes with a rinse solution (0.9% NaCl, 2.5%polyvinylpyrrolidine, 0.025% heparin, and 0.5% pro-caine HCl), followed by a 15-minute perfusion withfixative (2.0% formaldehyde, 0.1% picric acid, and 50mM sodium cacodylate, pH 7.4). After fixation, thelungs were briefly rinsed with phosphate-bufferedsaline containing calcium. The pulmonary artery wasthen cut open lengthwise and pinned on dental wax.After three 5-minute washes with phosphate-buffered saline, the tissue was permeabilized with-20°C acetone for 5 minutes, followed by threeadditional 5-minute washes with phosphate-bufferedsaline. Specific staining of actin filaments was per-formed by incubating the tissue segment with rhoda-mine phalloidin (Molecular Probes, Eugene, Ore.)for 60 minutes at 370 C. Cells were visualized usingan Olympus IM2 inverted microscope equipped forepifluorescence and photographed on Kodak Tri-XPan film at x 1,235 magnification.

StatisticsThe two-way analysis of variance with repeated

measures for mixed design33 was used to compare theresults of different time points (columns) and dif-ferent experimental groups (rows). If variance amongrows or columns was noted, Scheffe's test with cor-rection for multiple comparisons33 was used to deter-mine significant differences between specific pointswithin the rows or columns.

ResultsPulmonary Effects of Neutrophils After Administrationof Tumor Necrosis Factor-a

Increases in both the lung weight and Ppc werenoted in the TNF+PMN group at 60 minutes com-pared with baseline and also compared with theheat-inactivated TNF+PMN, TNF, and PMN groupsat 60 minutes (Figure 1). The TNF and PMN weresynergistic, as opposed to additive, in promotingweight gain. Small increases in the Ppc and lungweight were noted in the TNF group at 60 minutescompared with baseline (Figure 1). Small increases inlung weight at 60 minutes compared with baseline,without significant increases in Ppc, were noted in theheat-inactivated TNF+PMN and PMN groups (Fig-

15,-

E

a-l

2.400

2.000

is 30 60

TIME (Min.)

FIGURE 1. The pulmonary effects of neutrophils (PMN)after administration oftumor necrosis factor-a (TNF). Valuesare expressed as mean±SEM for the TNF+PMN (0), heat-inactivated TNF+PMN (n), TNF (rQ), and PMN (o) groups.Pp, pulmonary capillary pressure; AW, change in lung weight.*Differentfrom respective baseline value ( p <0.05); tdifferentfrom the heat-inactivated TNF+PMN, TNF, and PMNgroups(p<O.OI).

ure 1). No differences in the baseline Pp, measure-ments were noted among the groups (Figure 1).The use of either Dazoxiben or WEB-2086 pre-

vented the increase in Pp. and lung weight seen in theTNF+PMN group (Figure 2). No differences in thebaseline Ppc measurements were noted among thegroups (Figure 2).

Pulmonary Effects ofAdministration of TumorNecrosis Factor-a

Increases in the Pp, and lung weight were noted inthe TNF group at 60 minutes compared with baselineand the heat-inactivated TNF group (Table 1). Theaddition of either Dazoxiben or WEB-2086 to theperfusate inhibited the increase in Pp, and lungweight seen in the TNF group (Table 1). No differ-ences in the baseline Ppc measurements were notedamong the groups (Table 1).

Lung (Wet-Dry)-to-Dry Weight RatioAn increase in the lung W/D weight ratio was

noted in the TNF+PMN group compared with theother groups (Table 2). The addition of either

Of

T Tm

1


Dow

nloaded from



TABLE 2. Effect of Tumor Necrosis Factor and Neutrophils onLung (Wet-Dry)-to-Dry Weight Ratio

Group W/D

Control 6.0±0.2TNF+PMN 8.3±0.5*Heat-inactivated TNF+PMN 6.0±0.3TNF+Dazoxiben+PMN 5.9±0.2TNF+WEB-2086+PMN 6.6±0.4TNF 6.5±0.3Heat-inactivated TNF 5.8±0.1PMN 6.4±0.5

Values are mean±SEM. W/D, (wet-dry)-to-dry weight ratio;TNF, tumor necrosis factor-a; PMN, neutrophil.

*Different from all other groups (p<0.05).

TIME (Min.)

FIGURE 2. The effect of Dazoxiben and WEB-2086 on

pulmonary responses in the tumor necrosis factor(TNF) +neutrophil (PMN) group. Values are expressed as

mean±SEM for the TNF+PMN (a), TNF+Dazoxiben+PMN (o), and TNF+WEB-2086+PMN (n) groups. P1,pulmonary capillary pressure; AW, change in lung weight.*Different from the respective baseline value (p<0.01);tdifferent from the TNF+PMNgroup (p<0.01).

Dazoxiben or WEB-2086 to the circulating perfusateprevented the increase (Table 2).

Pulmonary Arterial and Venous ResistanceAn increase in R, was noted in the TNF+PMN

group at 60 minutes compared with baseline and theother groups (Table 3). The use of either Dazoxibenor WEB-2086 prevented the increase in R, seen in theTNF+PMN group (Table 3). There were no signifi-cant changes in the Ra in any of the groups (Table 3).

Pulmonary Capillary Filtration CoefficientIncreases in the Kf,C were noted at 30 minutes

compared with baseline, but not at 60 minutes, inboth the TNF+Dazoxiben and TNF+Dazoxiben+PMN groups (Table 4). The KfC was unchanged inthe TNF+WEB-2086+PMN, TNF+WEB-2086,Dazoxiben+PMN, Dazoxiben, and control groups(Table 4). The Kf,C was measured only in the controlgroup and in groups in which Dazoxiben or WEB-2086 were used because prolonged periods of stabil-ity in lung weight (which are required for themeasurement24) were present only in these groups.

Lung Effluent Arachidonic Acid MetabolitesLevels of TXB2 were increased in the TNF+PMN

group at 60 minutes compared with baseline and theother groups (Table 5). Smaller increases were noted inTXB2 levels in the TNF, PMN, and INF+WEB+PMN groups at 60 minutes compared with baseline(Table 5). In the control, TNF+Dazoxiben+PMN, andTNF+WEB-2086+PMN groups, there were no

changes in TXB2 levels (Table 5). Levels of 6-keto-PGFla were increased in the TNF+PMN group at60 minutes compared with the other groups, and in allgroups at 60 minutes compared with baseline (Table 5).

Pulmonary Artery Endothelial Actin FilamentMorphologyThe characteristic appearance of normal pulmo-

nary artery endothelium stained with rhodamine

TABLE 1. Effect of Tumor Necrosis Factor on Pulmonary Capillary Pressure and Change in Lung Weight

Pulmonary capillary pressure (cm H20) Change in lung weight (g)

Group Baseline 30 min 60 min 30 min 60 minControl 5.9±0.4 5.4±0.3 5.9±0.3 -0.010±0.060 -0.053±0.053TNF 5.8±0.3 6.0±0.3 8.3±0.7* -0.012±0.066 0.409±0.179*Heat-inactivated TNF 4.9±0.1 5.1±0.1 5.4±0.3 -0.019±0.065 -0.073±0.074TNF+Dazoxiben 4.9±0.1 5.2±0.2 5.4±0.2 -0.080±0.040 -0.052±0.060TNF+WEB-2086 4.7±0.1 4.6±0.2 5.5±0.2 -0.047±0.066 -0.021±0.076

Values are mean±SEM. TNF, tumor necrosis factor-a.'Different from the respective baseline value (p<0.01) and from the heat-inactivated TNF, TNF+Dazoxiben,

TNF+WEB-2086, and control groups (p<0.01).

15T--

c

E-UctoU

4-

a.i

*t

t

j.--1


Dow

nloaded from


Hocking et al Tumor Necrosis Factor-Induced Pulmonary Edema

TABLE 3. Effect of Tumor Necrosis Factor and Neutrophils on Arterial and Venous Resistance

Arterial resistance Venous resistance

Group Baseline 60 min Baseline 60 minControl 0.072±0.012 0.073±0.007 0.056±0.006 0.086±0.015TNF+PMN 0.068±0.016 0.108±0.019 0.075±0.008 0.339±0.54*tHeat-inactivated TNF+PMN 0.043±0.018 0.052±0.021 0.073±0.008 0.117±0.004TNF+Dazoxiben+PMN 0.088±0.008 0.079±0.014 0.060±0.007 0.110±0.027TNF+WEB-2086+PMN 0.070±0.013 0.068±0.008 0.051±0.008 0.164±0.031*TNF 0.061±0.009 0.078±0.007 0.070±0.010 0.078±0.028Heat-inactivated TNF 0.075±0.015 0.086±0.024 0.040±0.007 0.061±0.027PMN 0.087±0.008 0.064±0.014 0.064±0.010 0.108±0.023

Values are mean±SEM in cm H20/ml/min. TNF, tumor necrosis factor-a; PMN, neutrophil.*Different from the respective baseline value (p<0.05).tDifferent from all other groups (p<0.01).

phalloidin, a cobblestone pattern with intact periph-eral actin bands, was seen in the control group

(Figure 3A). In the TNF+PMN group (Figure 3B),focal areas of abnormal morphology were noted,characterized by marked ruffling of peripheral actinbands, less regular orientation of endothelial cells,and occasional cell shape changes. Abnormalitiessimilar to those seen in Figure 3B were noted in theTNF group. The use of Dazoxiben or WEB-2086 didnot prevent these abnormalities (morphology similarto Figure 3B). No abnormalities were noted in thePMN, Dazoxiben, and WEB-2086 groups (morphol-ogy similar to Figure 3A).

DiscussionIn this study, we have shown that TNF adminis-

tered in vivo sensitizes the lung to the developmentof PMN-dependent edema, as increases were notedin both lung weight and lung W/D weight ratio in theTNF+PMN group compared with the TNF andPMN groups. The pulmonary edema was not due tononspecific effects of the TNF vehicle or contamina-tion with endotoxin because heat-inactivated TNFdid not cause similar changes. The increases in lungweight noted in the TNF and PMN groups were notassociated with increases in the lung W/D weightratio, and therefore may have been due to eitherincreases in vascular volume or pulmonary edema. Asmall amount of edema may have been present in

these groups because decreases in vascular volumemight occur in association with the increases in R,noted in this study, and because the lung W/D weightratio is a specific but insensitive indicator of pulmo-nary edema.34The pulmonary edema in the TNF+PMN group

was mediated in part by vasoconstriction because theincrease in lung weight was associated with anincrease in Pp,, and reduction of the increase in PpCusing Dazoxiben prevented the increase in lungweight despite an increase in the 1fc. The increasesin Pp, were predominantly due to venous constrictionbecause R, was increased, whereas Ra did not changesignificantly. TXA2 was a mediator of the increase inPpC in the TNF+PMN group because the TXB2concentration of lung effluent was increased, andtreatment of the lung with Dazoxiben inhibited theincreases in both Pp, and TXB2 levels. PAF alsopromoted increases in Pp, but probably did so byinducing TXA2 release35,36 because pretreatment ofthe lung with WEB-2086 inhibited the increases inboth PC and TXB2 levels. Increases in 6-keto-PGFlalevels were noted in all groups, indicating constitutiverelease by the isolated lung of prostacyclin. Theelevations in Pp, in the TNF+PMN group occurreddespite the increased release of prostacyclin, a

vasodilator.22Increased pulmonary vascular permeability was

also a factor in the pathogenesis of the edema noted

TABLE 4. Effect of Tumor Necrosis Factor and Neutrophils on Pulmonary Capillary Filtration Coefficient

Pulmonary capillary filtration coefficient

Group Baseline 30 min 60 min

Control 0.111±0.013 0.104±0.016 0.109±0.018TNF+Dazoxiben+PMN 0.118±0.008 0.170±0.048* 0.153±0.020TNF+WEB-2086+PMN 0.107±0.016 0.107±0.014 N.D.

TNF+Dazoxiben 0.122±0.025 0.220±0.079* 0.150±0.022TNF+WEB-2086 0.136±0.012 0.137±0.008 0.128±0.018Dazoxiben+PMN 0.126±0.018 0.131±0.020 0.151±0.020

Dazoxiben 0.119±0.022 0.098±0.014 0.111±0.028

Values are mean±SEM in g/min/cm H20/g dry lung wt. TNF, tumor necrosis factor-a; PMN, neutrophil; N.D., notdone because lungs were not isogravimetric.

*Different from respective baselines (p<0.05).

73


Dow

nloaded from



TABLE 5. Effect of Tumor Necrosis Factor and Neutrophils on Thromboxane B2 and 6-Ketoprostaglandin-F1,. inPulmonary Venous Effluent

TXB2 (pg/mi) 6-Keto-PGFIay (pg/mi)

Group Baseline 60 min Baseline 60 min

Control 49±25 49±30 112±12 612+21*TNF+PMN 60±13 5,244±599*t 184±27 1,101±17*tTNF+D3azoxiben+PMN 33-+21 76±15 108±24 686+ 198*TNF+WEB-2086+PMN 25+20 307+29* 98±21 441 +209*TNF 21+6 190±23* 86±13 305+10*PMN 34±4 284±135* 77±11 422±47*

Values are mean-+-SEM. TXB2, thromboxane B2; 6-keto-PGF1,, 6-ketoprostaglandin-Fia; TNF, tumor necrosisfactor-a; PMN, neutrophil.

*Different from respective baseline values (pcO.05).tDifferent from all other groups (p<0.05).

in this study because Kf c was increased in theTNF+Dazoxiben+PMN group. Kf, was increased at30 minutes but not at 60 minutes, indicating eitherreversible increases in vascular permeability ordecreases in lung compliance and/or vascular surfacearea, which could have prevented continuedincreases in Kf, despite persistent increases in vascu-lar permeability.34 PAF was a mediator of theincreases in vascular permeability because Kf, wasnot increased in the TNF+WEB-2086+PMN group.This result is consistent with previous studies indicat-ing that PAF can increase pulmonary vascularpermeability during endotoxemia.14 Despite theincreases in Kf , minimal weight gain was noted inthe TNF+Dazoxiben+PMN and the TNF+Dazoxi-ben groups. The lack of weight gain may be explainedby the absence of a significant driving pressure (i.e.,an increased Ppc) for edema formation, which mini-mized the effect of the increased Kf,.37,38

The cellular source of the PAF and TXA2 remainsunclear. Cooperation between the TNF-activated lungand circulating PMN may be required because PAFand TXA2 were more physiologically active in theTNF+PMN group than in either the TNF or PMNgroups. Likely sources of PAF in our system includeendothelial cells, macrophages, and PMN becausethese cells release PAF upon stimulation by TNF.13Possible sources of TXA2 include circulating PMN39and parenchymal lung cells such as macrophagesj40epithelial cells,41 and endothelial cells.42 The releaseof TXA2 may be stimulated from these cells by TNF,39PAFY40 or other factors. In our study, TXA2 may havebeen released predominantly by PAF because WEB-2086 markedly inhibited TXA2 release. However, asmall component of PAF-independent TXA2 secre-tion was present because the inhibition of TXA2release with WEB-2086 was not complete.

FIGURE 3. Changes in pulmonary artery endothelial actin filament morphology in the tumor necrosis factor (TNF) +neutrophil(PMN) group. Panel A: Rhodamine phalloidin staining in the control group shows the characteristic appearance of normalendothelium, a cobblestone pattern with intact peripheral actin bands. Panel B: The endothelium in the TNF+PMN group showsfocal areas of abnormal morphology characterized by marked ruffling of peripheral actin bands, less regular orientation ofendothelial cells within the monolayer, and occasional cell shape changes. Peripheral bands in an individual cell are denoted bytwin arrows; magnification X1,235. A pattem similar to panel B was seen in the TNF, TNF+Dazoxiben+PMN, andTNF+ WEB-2086+PMN groups (not shown). A pattem similar to panel A was seen in the PMN, Dazoxiben, and WEB-2086groups (not shown).


Dow

nloaded from



Alterations in the endothelial peripheral actinbands were noted in the TNF+PMN group. Theperipheral band abnormalities may play a role in theincreased capillary permeability seen in the TNF+PMN group because similar cytoskeletal changes areassociated with endothelial permeability increases invitro,21 and because the changes were not due toaltered vascular tone, as the use of Dazoxiben pre-vented increases in PPC but not the morphologicalchanges. The changes in peripheral actin bands mayact as a permissive cofactor that renders the endothe-lium susceptible to the permeability-increasing effectof PAF14 because PAF mediated the increased capil-lary permeability, and PAF has no independenteffect,43 or only a small one,38 on permeability innormal isolated perfused lungs.The in vivo effects of TNF played a role in the

pathogenesis of the pulmonary edema in theTNF+PMN group. Vasoconstriction mediated byboth TXA2 and PAF resulted from in vivo TNFtreatment because both Ppc and TXB2 levels wereincreased in the TNF group, and the increases in PpCwere inhibited using either Dazoxiben or WEB-2086.Increases in pulmonary vascular permeability werealso noted after in vivo TNF treatment. The perme-ability increases were mediated by PAF because Kf,Cwas significantly increased in the TNF+Dazoxibengroup but not in the TNF+WEB-2086 group. TNFmay have contributed to the permeability increasesby altering peripheral actin bands. The peripheralband abnormalities noted in the TNF+PMN groupwere not prevented using either Dazoxiben or WEB-2086 and were also seen in the TNF group. Thesedata suggest that TNF induces the changes in vivo,independent of ex vivo TXA2, PAF, or PMN. How-ever, a role for these potential mediators in vivo hasnot been eliminated.The mechanisms of the effects of TNF in vivo

remain uncertain. The effects could have resultedfrom the direct action of TNF on the endothelium orother tissues.1-'3 In addition, TNF may have primedthe lung for the actions of PMN, possibly as achemotactic,17 adherence-promoting,16,44,45 andactivating8 agent for PMN because TNF and PMNwere synergistic (as opposed to additive) in promot-ing lung weight gain in the TNF+PMN group. Theinflammatory milieu that may have been created byTNF in vivo could have resulted in pulmonary edemain vivo, with subsequent clearing by pulmonary lym-phatics and local and blood-borne anti-inflammatoryand tissue-protecting mechanisms, resulting in an"activated" but minimally edematous lung at thetime of excision.Our data indicate that PMNs promote pulmonary

edema in TNF-treated lungs because increases inlung weight and lung W/D weight ratio were noted inthe TNF+PMN group compared with the TNFgroup. PMN promoted edema by increasing the Pp,(probably by the release of TXA2) in the presence ofTNF- and PAF-mediated permeability increases,rather than by directly increasing permeability

because 1) PpC and TXB2 levels were increased in theTNF+PMN group compared with the TNF group,2) Kf,C was not significantly increased in theTNF+Dazoxiben+PMN group compared with theTNF+Dazoxiben group, and 3) changes in endothe-lial morphology occurred similarly in the TNF+PMNand TNF groups. In contrast, Varani et al'8 showedthat pretreatment (18 hours) with human recombi-nant TNF-a (50 ng/ml) in vitro enhanced the suscep-tibility of vascular endothelial cells to lysis by PMN.However, the PMNs used by Varani et al wereactivated by phorbol myristate acetate or C5a, mak-ing comparison with our data difficult.

Previous studies addressing the role of PMN in thepathogenesis of TNF-induced pulmonary edemahave been inconclusive. Stephens et al19,46 demon-strated that pulmonary edema occurs 8 hours afterthe injection of human recombinant TINF-a (1.4 x 106units/kg i.v.) in the guinea pig. The edema wasassociated with increases in the permeability index,peripheral leukopenia, and increases in the numberof PMN per alveolus. Cyclophosphamide-inducedPMN depletion prevented the edema response,19implicating PMN in the vascular effects of TNF.Horvath et all" showed that the infusion of TNF-a(6.5 x 106 units total, which is approximately 2 x iO0units/kg over 30 minutes) in sheep results in pulmo-nary edema associated with increases in Pp, andcapillary permeability. In contrast to the work ofStephens et al,'9 the vascular effects of TNF noted byHorvath et all" were not prevented by hydroxyurea-induced PMN depletion, suggesting that the effectsoccurred independent of circulating PMN. However,interpretation of the results of PMN depletion stud-ies is made difficult by the possible independenteffects of the agents used to deplete PMN and thepersistence of PMN in the lung despite bloodstreamPMN depletion.'9The results of this study indicate that PMNs medi-

ate edema in lungs isolated from TNF-treated guineapigs. The edema is associated with permeabilityincreases (by the pulmonary release of PAF) andvasoconstriction (by the pulmonary release of PAFand TXA2). TNF mediates changes in endothelialperipheral actin bands independent of ex vivo TXA2,ex vivo PAF, or exogenous PMNs.

AcknowledgmentsThe authors thank Dr. Min-Fu Tsan for his helpful

collaboration and Ms. Rebecca Kittell for her experttechnical assistance with the radioimmunoassay.

References1. Beutler B, Milsark IW, Cerami AC: Passive immunization

against cachectin/tumor necrosis factor protects mice fromlethal effect of endotoxin. Science 1985;229:869-871

2. Nedwin GE, Svedersky LP, Bringman TS, Palladino MA,Goeddel DV: Effect of interleukin-2, interferon-y, and mito-gens on the production of tumor necrosis factors a and P. JImmunol 1985;135:2492-2497


Dow

nloaded from



3. Beutler BA, Milsark IW, Cerami A: Cachectin/tumor necrosisfactor: Production, distribution and metabolic fate in vivo. JImmunol 1985;135:3972-3977

4. Dinarello CA, Cannon JG, Wolfe SM, Bernheim HA, BeutlerB, Cerami A, Figari IS, Palladino MA, Connor JV: Tumornecrosis factor (cachectin) is an endogenous pyrogen andinduces production of interleukin-1. J Exp Med 1986;163:1433-1450

5. Nawroth PP, Bank I, Handley D, Cassimeris J, Chess L, SternD: Tumor necrosis factor/cachectin interacts with endothelialcell receptors to induce release of interleukin-1. J Exp Med1986;163:1363-1375

6. Bevilacqua MP, Pober JP, Majeau GR, Cotran RS, GimbroneMA: Recombinant tumor necrosis factor induces procoagulantactivity in cultured human vascular endothelium: Character-ization and comparison with the actions of interleukin-1. ProcNatlAcad Sci USA 1986;83:4533-4537

7. Nawroth PP, Stern DM: Modulation of endothelial cell hemo-static properties by tumor necrosis factor. J Exp Med 1986;163:740-745

8. Shalaby MR, Aggarwal BB, Rinderknecht E, Svedersky LP,Finkle BS, Palladino MR: Activation of human polymorpho-nuclear neutrophil functions by interferon-y and tumor necro-sis factors. J Immunol 1985;135:2069-2073

9. Klebanoff SJ, Vadas MA, Harlan JM, Sparks LH, Gamble JR,Agosti JM, Waltersdorph AM: Stimulation of neutrophils bytumor necrosis factor. J Immunol 1986;136:4220-4225

10. Beutler B, Krochin N, Milsark IW, Luedke C, Cerami A:Control of cachectin (tumor necrosis factor) synthesis: Mech-anisms of endotoxin resistance. Science 1986;232:977-980

11. Horvath CJ, Ferro TJ, Jesmok G, Malik AB: Recombinanttumor necrosis factor increases pulmonary vascular permeabil-ity independent of neutrophils. Proc Natl Acad Sci USA1988;85:9219-9223

12. Stolpen A, Guinan E, Fiers W, Pober J: Recombinant tumornecrosis factor and immune interferon act singly and incombination to reorganize human vascular endothelial cellmonolayers. Am J Pathol 1986;123:16-24

13. Camussi G, Bussolino F, Salvidio G, Baglioni C: Tumornecrosis factor/cachectin stimulates peritoneal macrophages,polymorphonuclear neutrophils and vascular endothelial cellsto synthesize and release platelet activating factor. J Erp Med1987;166:1390-1404

14. Chang S-W, Feddersen CO, Henson PM, Voelkel NF: Plateletactivating factor mediates hemodynamic changes and lunginjury in endotoxin-treated rats. J Clin Invest 1987;79:1498-1509

15. Kettelhut I, Fiers W, Goldberg A: The toxic effects of tumornecrosis factor in vivo and their prevention by cyclooxygenaseinhibitors. Proc NatlAcad Sci USA 1987;84:4273-4277

16. Gamble JR, Harlan JM, Klebanoff SJ, Vadas MA: Stimulationof the adherence of neutrophils to umbilical vein endotheliumby human recombinant tumor necrosis factor. Proc Natl AcadSci USA 1985;82:8667-8671

17. Strieter RM, Kunkel SL, Showell HJ, Remick DG, Phan SH,Ward PA, Marks RM: Endothelial cell expression of a neu-trophil chemotactic factor by TNF-a, LPS, and IL-1-,B. Science1989;243:1467-1469

18. Varani J, Bendelow MJ, Sealey DE, Kunkel SL, Gannon DE,Ryan US, Ward PA: Tumor necrosis factor enhances suscep-tibility of vascular endothelial cells to neutrophil-mediatedkilling. Lab Invest 1988;59:292-295

19. Stephens KE, Ishizaka A, Wu Z, Larrick JW, Raffin TA:Granulocyte depletion prevents tumor necrosis factor-medi-ated acute lung injury in guinea pigs. Am Rev Respir Dis1988;138:1300-1308

20. Madara JL, Barenberg D, Carlson S: Effects of cytochalasin Don occluding junctions of intestinal absorptive cells: Furtherevidence that the cytoskeleton may influence paracellularpermeability and junctional charge selectivity. J Cell Biol1986;102:2125-2136

21. Johnson A, Phillips P, Hocking D, Tsan M-F, Ferro T: Aprotein kinase inhibitor prevents pulmonary edema inresponse to H202. Am J Physiol 1989;256:H1012-H1022

22. Ferro TJ, Johnson A, Everitt J, Malik AB: IL-2 inducespulmonary edema and vasoconstriction independent of circu-lating lymphocytes. J Immunol 1989;142:1916-1921

23. Linehan JH, Dawson CA, Rickaby DA: Distribution of vascu-lar resistance and compliance in a dog lung lobe. JApplPhysiol1982;53:158-168

24. Drake R, Gaar KA, Taylor AE: Estimation of the filtrationcoefficient of pulmonary exchange vessels. Am J Physiol 1978;234:H266-H274

25. Wang AM, Creasey AA, Ladner MB, Lin LS, Strickler J, VanArsdell JN, Yamamoto R, Mark DF: Molecular cloning of thecomplementary DNA for human tumor necrosis factor. Sci-ence 1985;228:149-154

26. Creasey AA, Doyle LV, Reynolds MT, Jung T, Lin LS, VittCR: Biological effects of recombinant human tumor necrosisfactor and its novel muteins on tumor and normal cell lines.Cancer Res 1987;47:145-149

27. White CW, Ghezzi P, Dinarello CA, Caldwell SA, McMurtryIF, Repine JE: Recombinant tumor necrosis factor/cachectinand interleukin-1 pretreatment decreases lung oxidized glu-tathione accumulation, lung injury, and mortality in ratsexposed to hyperoxia. J Clin Invest 1987;79:1868-1873

28. Ruff MR, Gifford GE: Rabbit tumor necrosis factor: Mecha-nism of action. Infect Immun 1981;31:380-385

29. Ferrante A, Thong YH: Optimal conditions for simultaneouspurification of mononuclear and polymorphonuclear leuco-cytes from human blood by the Hypaque-Ficoll method. JImmunol Methods 1980;36:109-118

30. Tyler HM, Saxton CAPD, Perry MJ: Administration to man ofUK-37-248.01, a selective inhibitor of thromboxane synthe-tase. Lancet 1981;1:629-632

31. Casals-Stenzel J, Muacevic G, Weber KH: Pharmacologicalactions of WEB 2086, a new specific antagonist of plateletactivating factor. J Pharmacol Exp Ther 1987;241:974-981

32. Forssman WG, Ito S, Weihe E, Aoki E, Dym M, Fawcett DW:An improved perfusion fixation method for the testis.AnatRec1977;188:307-314

33. Steel RGD, Torrie JH: Principles and Procedures of Statistics:ABiomedicalApproach, ed 2. New York, McGraw-Hill Book Co,1980, pp 173-177

34. Taylor AE, Parker JC, Granger DN, Mortillaro NA, Rutilli G:Assessment of capillary permeability using lymphatic proteinflux: Estimation of the osmotic deflection coefficient, in EffrosRM, Schmid-Shonbein H, Ditzel J (eds): Microcirculation.New York, Academic Press, Inc, 1981, pp 19-32

35. Bradley LM, Goldstein RE, Feuerstein G, Czaja JF: Circula-tory effects of PAF-acether in newborn piglets. Am J Physiol1989;25:H205-H212

36. Hamasaki Y, Mojarad M, Saga T, Tai H-H, Said SI: Platelet-activating factor raises airway and vascular pressures andinduces edema in lungs perfused with platelet-free solution.Am Rev Respir Dis 1984;129:742-746

37. Patterson CE, Barnard JW, Lafuze JE, Hull MT, Baldwin SJ,Rhoades RA: The role of activation of neutrophils andmicrovascular pressure in acute pulmonary edema. Am RevRespir Dis 1989;140:1052-1062

38. Sakai A, Chang S-W, Voelkel NF: Importance of vasoconstric-tion in lipid mediator-induced pulmonary edema. J ApplPhysiol 1989;66:2667-2674

39. Bachwich PR, Chensue SW, Larrick JW, Kunkel SL: Tumornecrosis factor stimulates interleukin-1 and prostaglandin E2production in resting macrophages. Biochem Biophys Res Com-mun 1986;136:94-101

40. O'Flaherty JT, Nishihira J: Arachidonate metabolites,platelet-activating factor, and the metabolism of proteinkinase C in human polymorphonuclear neutrophils. JImmunol1987;138:1889-1895

41. Yoneda K: Ultrastructural localization of phospholipases inthe Clara cell of the rat bronchiole. Am J Pathol 1978;93:745-752

42. Crutchley DJ, Ryan JW, Ryan US, Fisher GH: Bradykinin-induced release of prostacyclin and thromboxanes from bovinepulmonary artery endothelial cells. Biochim Biophys Acta1983;751:99-107


Dow

nloaded from



43. lmai T, Vercelloti GM, Moldow CF, Jacob HS, Weir EK:Pulmonary hypertension and edema induced by platelet-activating factor in isolated, perfused rat lungs are blocked byBN52021. J Lab Clin Med 1988;111:211-217

44. Bevilacqua M, Pober J, Mendrick D, Cotran R, Gimbrone M:Identification of an inducible endothelial-leukocyte adhesionmolecule. Proc NatlAcad Sci USA 1987;84:9238-9242

45. Pober JS, Gimbrone MA, LaPierre LA, Mendrick DL, FiersW, Rothlein R, Springer TA: Overlapping patterns of activa-tion of human endothelial cells by interleukin-1, tumor necro-

sis factor, and immune interferon. J Immunol 1986;137:1893-1896

46. Stephens KE, Ishizaka A, Larrick JW, Raffin TA: Tumornecrosis factor causes increased pulmonary permeability andedema: Comparison to septic acute lung injury.Am Rev RespirDis 1988;137:1364-1370

KEY WORDS * actin filaments * platelet activating factor :polymorphonuclear leukocytes * tumor necrosis factor-athromboxane A2


Dow

nloaded from


D C Hocking, P G Phillips, T J Ferro and A JohnsonMechanisms of pulmonary edema induced by tumor necrosis factor-alpha.

Print ISSN: 0009-7330. Online ISSN: 1524-4571 Copyright © 1990 American Heart Association, Inc. All rights reserved.is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Circulation Research

doi: 10.1161/01.RES.67.1.681990;67:68-77Circ Res.

http://circres.ahajournals.org/content/67/1/68World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://circres.ahajournals.org//subscriptions/

is online at: Circulation Research Information about subscribing to Subscriptions:

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

document. Permissions and Rights Question and Answer about this process is available in the

located, click Request Permissions in the middle column of the Web page under Services. Further informationEditorial Office. Once the online version of the published article for which permission is being requested is

can be obtained via RightsLink, a service of the Copyright Clearance Center, not theCirculation Researchin Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:


Dow

nloaded from

http://circres.ahajournals.org/content/67/1/68

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://circres.ahajournals.org//subscriptions/


Documents

Mechanisms of Pulmonary Edema Induced by TumorNecrosis ... filedata indicate that TNF induces neutrophil-dependent pulmonary edema associated with increased Pp, (mediatedbyTXA2andPAF),increasedKtc