Interleukin-2 Directly Increases Albumin Permeability of Bovine and Human Vascular Endothelium In Vitro Gordon H. Downie, Una S. Ryan, Brendan A. Hayes, and Mitchell Friedman Department of Medicine and Center for Environmental Medicine and Lung Biology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, and Department of Medicine, University of Miami School of Medicine, Miami, Florida

The direct effects of interleukin-2 (IL-2) on albumin permeability of cultured bovine pulmonary artery endothelial cell (BPAEC) and human arterial endothelial cell (HAEC) monolayers were studied. BPAEC were exposed to IL-2 (500 to 25,000 U/ml) for 4 h. The steady-state transfer rate of [J25I]albumin across the BPAEC monolayer was 3.3 ± O.4%/h (n = 10) in control BPAEC (diluent alone), was significantly increased in BPAEC exposed to 500 U/ml of IL-2 (72 ± 3% above control values, n = 6, P < 0.02), and further increased in BPAEC exposed to 5,000 U/ml (60 ± 2 % increase above 500 U/ml values, n = 5, P < 0.02). No further increase was noted after exposure to 25,000 U/ml of IL-2. Additionally, no further increase in ['25I]albumin transfer rates was noted in BPAEC exposed to 5,000 U/ml of IL-2 for 24 versus 4 h. Similar changes were found using HAEC. Preincubation of HAEC with an anti-IL-2low-affinity receptor antibody (anti-IL-2Ro:) inhibited the IL-2-induced permeability increase. Expression ofIL-2Ro: receptors in HAEC incubated with 5,000 U/ml of IL-2 for 4 h was also found. Thus, IL-2 appears to have a direct effect on cultural arterial endothelial monolayers not requiring the presence of other cell types or serum proteins. IL-2-induced increases in endothelial macromolecular permeability may play an important role in the pathogenesis of the IL-2-induced vascular leak syndrome seen in vivo.

Lymphokines are polypeptide products of activated lymphocytes that participate in a variety of cellular responses, including regulation of the immune system. Interleukin-2 (IL-2) is a potent glycoprotein lymphokine with a molecular mass of 15 kD that is a secretory product of activated T lymphocytes (1). IL-2 has been shown to mediate several immune reactions in vivo, including enhancement of the tumoricidal activity of peripheral blood mononuclear cells (2). The introduction of adoptive immunotherapy, involving the use of recombinant IL-2, either alone or in combination with lymphokine-activated killer cells (LAK cells), has been shown to have substantial antitumor activity (2). The pres-

(Received in original form March 26, 1991 and in revised form December 12, 1991) Address correspondence to: Mitchell Friedman, M.D., Division of Pulmonary Diseases, Critical Care and Occupational Medicine, Room 724 CB 7020, Burnett-Womack Building, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599. Drs. Ryan'sand Hayes' current address: Monsanto Company, 800 N. Lindbergh Blvd., St. Louis, MO 63167. Abbreviations: bovine pulmonary artery endothelial cell, BPAEC; fetal calf serum, FCS; Hanks' balanced salt solution, HBSS; human arterial endothelial cell, HAEC; interleukin-I, IL-I; interleukin-2, IL-2; Iymphokineactivated killer cells, LAK cells; lactate dehydrogenase, LDH; medium 199, M199; phosphate-buffered saline, PBS; tumor necrosis factor-a, TNF-a. Am. J. Respir. Cell Mol. BioI. Vol. 7. pp. 58-65, 1992

ence of IL-2 is thought to be essential to maintain the state of activation of the LAK cells (3). The application of adoptive immunotherapy offers an alternative treatment for solid tumors resistant to conventional therapy but is also associated with significant toxicity (2, 3, 6). The major toxicity limiting therapy associated with adoptive immunotherapy is the development of increased vascular permeability to macromolecules (''vascular leak syndrome"), which leads to fluid retention including pulmonary edema (3, 4, 6). Other effects after IL-2 infusion in patients include left ventricular dysfunction with hypotension, increased cardiac output, decreased systemic vascular resistance, oxygen-supply dependency of oxygen consumption, and renal dysfunction (5, 7). These IL-2-induced hemodynamic alterations are similar to those observed in human patients with sepsis or in animal models of septic shock (8, 9). The underlying molecular mechanism(s) responsible for the IL-2-induced vascular leak syndrome is not known (2-9). Of the several proposed mechanisms involved in the pathogenesis of the vascular leak syndrome, few have implicated IL-2 as a direct toxin to endothelial cells (6,9). Rather, it has been suggested that IL-2 is indirectly involved in the increased vascular permeability based on the following published observations. Endothelial cells can be activated in response to immune inflammation. Pober has demonstrated endothelial cell expression of class II major histocompatibility complex (MHC) antigens in response to allogeneic peripheral blood lymphocytes (10). Bevilacqua and colleagues

Downie, Ryan, Hayes et al.: IL-2 Directly Increases Endothelial Permeability to Albumin In Vitro

have shown that another Iymphokine, interleukin-l (IL-l), can initiate endothelial cell binding of leukocytes (11) and promote the extrinsic clotting pathway (12). It has also been demonstrated that IL-2 administration leads to the expression ofleukocyte adhesion molecules on endothelial cell surfaces (13). This IL-2-induced immune activation of vascular endothelium has been inferred to indirectly result in the vascular leak syndrome, possibly by.stimulating the endogenous production of other lymphokines, .e.g., IL-l or tumor necrosis factor (10, 13). It has also been suggested that the development of the vascular leak syndrome may be dependent on the presence of T lymphocytes or by the infiltration of lymphoblastoid cells into target organs (14). LAK cells have also been shown to exhibit strong adhesion to endothelium with subsequent endothelial injury including lysis (15). The development of routine and easily implemented methods for culturing endothelial cells in vitro allows for studies of possible direct involvement of endothelium in pathologic processes including those associated with increased vascular permeability (16-20). Cultured vascular endothelial cells form confluent monolayers, cease proliferation, and retain many of their in situ structural and functional characteristics (21). Methods have been established to measure the permeability to macromolecules of intact monolayers of endothelium in vitro grown on polycarbonate filters (16-20). Using these techniques, it has been possible to demonstrate increases in the permeability of endothelial monolayers to macromolecules, e.g., albumin, after exposure to agents that have been demonstrated to increase vascular permeability in vivo. The in vitro studies described in the present study were designed to determine if IL-2 can directly alter endothelial permeability. We have studied the effects of varying concentrations of IL-2 on radio labeled albumin permeability and morphology of bovine pulmonary artery endothelial cell (BPAEC) monolayers and also human arterial endothelial cell (HAEC) monolayers in vitro.

Materials and Methods Cells Main-stem pulmonary arteries of calves (donated by Dr. N. Olson, North Carolina State School of Veterinary Medicine) were removed aseptically, placed in Moscona's saline containing penicillin (300 U/ml), streptomycin (300 Itg/ml), and gentamicin (50 Itg/ml). The vessels were opened in a laminar flow hood and the BPAEC were obtained by gentle scraping of the intimal surface with a scalpel blade (21). The cells were plated in 25-cm2 culture flasks (Corning Glassworks, Corning, NY) and were grown to confluence in medium 199 (MI99) containing 10% fetal calf serum (FCS) (Hyclone Laboratories, Logan, UT), penicillin (100 U/ml), streptomycin (100 Itg/ml), and gentamicin (50 Itg/ml). BPAEC were identified by their characteristic morphologic appearance by inverted-phase microscopy, positive immunofluorescence staining for factor VIII-related antigen (22), and the presence of angiotensin-converting enzyme (21). Cells were subcultured by removal with gentle scraping. BPAEC from passages 8 through 13 were used for the experiments. No differences in response were noted between the passages of BPAEC used in the study.

59

HAEC were obtained from Transplantation Surgery, University of Miami School of Medicine, usually within 8 to 12 h of accidental death. Under sterile conditions, each vessel was trimmed of extraneous tissue and rinsed with Puck's saline (Ca2+/Mg2+ free, pH 7.4) containing 3x antibiotics (penicillin 300 Itg/ml, streptomycin 300 Itg/ml). After mounting on a dental wax platform, each vessel was cut open to expose the luminal surface and pinned taut and flat. The endothelial monolayer was scraped free of the luminal surface as described for the isolation of BPAEC. Cell harvests were seeded in primaria T-25 flasks using human medium (M199 with Earle's salts containing L-glutamine, 200 mg/liter; fetuin, 0.15 mg/ml; NaHC03 , 1.4 g/liter; Hepes, 2.5 mM; heparin, 10 U/ml; 20% FCS; penicillin, 200 U/ml; streptomycin, 100 Itg/ml; fungizone, 250 Itg/ml; and endothelial cell growth supplement, 200 Itg/ml [Sigma Chemical Co., St. Louis, MOD (23) and allowed to develop until cell colonies were firmly established (approximately 2 to 6 wk). Initially, subconfluent monolayers were passaged to free the primary culture of unwanted smooth muscle cells. Subsequently, confluent monolayers were passaged non-enzymatically using a rubber policeman and a 1:2 split. HAEC were characterized by Di-I-Ac-LDL uptake (24), by release of factor VIII surface antigen, and for the presence of angiotensinconverting enzyme activity (approximately 1 X 104 moll cell). Cell cultures at confluence displayed an elongated cobblestone morphology and contained a variable percentage of giant polyploid cells. Experimental Protocol Endothelial cells were plated on top of gelatin-impregnated polycarbonate filters (3-ltm pore size; Costar, Cambridge, MA), and the filters with cells were grown to confluence in Ml99 supplemented with 10% FCS and antibiotics (for BPAEC) or in complete human medium (for HAEC) in a 37° C incubator in 5 % CO 2 , 95 % air. The medium was changed every 2 to 3 days until the monolayers achieved confluence. The endothelial cell monolayers were then washed with phosphate-buffered saline (PBS), pH 7.4, and serum-free Ml99 supplemented with 50 mM Hepes, nucleosides, and antibiotics. Recombinant IL-2 (500 to 25,000 U/ml; Cetus Corp., Emeryville, CA or Hoffman-La Roche, Inc., Nutley, NJ) or its diluent, alone, was added to the media. The endothelial cell monolayers were then incubated at 37° C for 4 to 24 h with the various experimental media before permeability or morphologic studies were performed. Additional BPAEC monolayers were incubated with recombinant IL-l (100 U/ml; Collaborative Research Corp., Bedford, MA) or tumor necrosis factor-a (TNF-a) (2,000 U/ml) for 4 h. In additional studies, the BPAEC monolayers were coincubated with IL-2 (5,000 U/ml) and 62.5 Itg of a monoclonal antibody against IL-2 (anti-IL-2; Genzyme, Boston, MA). In other studies, HAEC monolayers were incubated with 2.5 Itg of a purified monoclonal antibody against the 55-kD glycoprotein IL-2 receptor (CD 25; Becton Dickinson, Mountain View, CA) for 30 min at 3r c. CD25 or, as it will be subsequently termed, anti-IL-2Ra is a monoclonal antibody that binds to the active site of the IL-2Ra receptor and blocks IL-2 actions (25).

60

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

In Vitro Endothelial Cell Permeability The method for measuring permeability of endothelial monolayers in vitro has been described in detail previously (16). After incubation with the various media described, the filters containing confluent endothelial cells were washed 2 times with PBS. The filters with attached endothelial cells were then mounted in modified flux chambers, and the chambers placed in a culture dish. The upper well of the chamber was filled with serum-free medium containing 50 mM Hepes. The dish was filled with the same medium. A stirring bar was added to the lower well, and the entire chamber placed on an electrical stirring device and incubated at 37° C. The chamber was incubated until the level of media between the upper well and the surrounding fluid in the beaker was equal. Thus, no hydrostatic pressure difference was present between the upper and lower wells. After this equilibration period, a small aliquot of medium in the upper well was removed and replaced with medium containing [' 251]bovine serum albumin (30,000 cpm/ml). The radiolabeled albumin had been extensively dialyzed against 1 M PBS immediately before use. Chromatographic monitoring of the dialyzed (1 251]albumin as well as the media in the lower well after the end of study has demonstrated > 95 % of the 1251 to cochromatograph with albumin (16). Small aliquots of media (in triplicate) were removed serially from both the upper and lower wells 10, 30, 60, 120, 180, and 240 min after the addition of the 1251 probe. The 1251 activity in each aliquot was measured in a gamma counter, and the average cpm/ml for the samples from the upper and lower wells determined. Appropriate corrections were made for background using the experimental media. The (1 251]albumin transfer rate of the BPAEC monolayers was expressed as the rate of appearance of counts in the lower well relative to the number of counts in the upper well/h over the 90 to 240-min period of steadystate clearance (16). Each albumin transfer rate point ("n") represents the average rate of duplicate filters within a group. Each group of filters included duplicate control filters (i.e., monolayers on filters incubated with diluent alone). In additional filters, nonradiolabeled bovine serum albumin (final concentration of 1%) was added along with (1 251]bovine serum albumin in the upper well. The (1 251]albumin transfer rate across the monolayer was determined using the previously described method. Average (1 251]albumin steady-state transfer rates for BPAEC in our laboratory are 2.8 ± 0.2 %/h (n = 18). These results correspond to a permeability coefficient for albumin of 0.7 cm/s X 10-6 (16). The HAEC exhibited similar (1 251]albumin transfer rates to BPAEC, but steady-state clearance of the radiolabeled albumin in HAEC was not reached until 120 min (data not shown). Therefore, albumin steady-state transfer rates for HAEC are reported for the 120- to 240-min period after addition of the 1251 probe. BPAEC Injury Assays Three independent methods of detecting BPAEC damage after IL-2 exposure were utilized: lactic dehydrogenase (LDH) release, cell detachment, and pH]deoxyglucose release. BPAEC-conditioned medium was assayed for LDH using a colormetric assay kit (Sigma). To measure cell detachment, BPAEC were grown to confluence in 24-well culture dishes

and exposed to IL-2 for 1 to 24 h. After exposure, the conditioned medium was removed and the number of cells in the conditioned medium (i.e., detached cells) counted by hemocytometry. The remaining BPAEC that were adherent to the culture dish were removed by trypsin and counted. The percent of detachment was computed as the number of detached cells divided by the total number of cells (detached cells plus attached cells). pH]deoxyglucose release was measured using a previously reported technique (26). Briefly, confluent monolayers of BPAEC in 24-well culture dishes were radiolabeled with 2 ILCi/well of pH]deoxyglucose (final concentration of 4.75 X 10-8 mmollliter) in 1 ml of Hanks' balanced salt solution (HBSS) containing 10% FCS for 16 h. This concentration is markedly below the concentrations that inhibit glycolytic metabolism and also below the K; value for carrier-mediated transport mechanism of human umbilical vein endothelial cells for 2-deoxyglucose (26). After labeling, the BPAEC was washed 4 times with HBSS, exposed to IL-2 (500 to 25,000 U/ml) for 4 h, and pH]deoxyglucose release computed. For all endothelial cell injury assays, comparison experiments were always performed using BPAEC from the same line and passage number. Other Studies In additional studies designed to measure IL-2 concentrations in BPAEC-conditioned media, six-well culture dishes (47-mm diameter; Costar) were plated with BPAEC and allowed to grow to confluence as previously described. The wells were either incubated with diluent alone or with IL-2 (500 to 5,000 U/ml) for up to 24 h. The conditioned medium was then removed, centrifuged, and decanted, and the supernatant used for determination of IL-2 using an enzymelinked immunoabsorbent assay (Genzyme). The conditioned medium was also assayed for the presence of IL-l (by an enzyme-linked immunoabsorbent assay technique) or TNF-a (using a biologic cytotoxic tumor assay). These latter studies were kindly performed by Dr. Suzanne Becker (c. E. Environmental, Research Triangle Park, NC). The possible presence of IL-2 receptors on HAEC was studied using confluent endothelial monolayers, incubated with either diluent alone or IL-2 (5,000 U/ml) in serum-free media for 4 h. The presence of soluble IL-2 receptors, either in the conditioned media or lysed cellular extract, was measured using a soluble IL-2 receptor immunoenzymometric assay kit (Immunotech S. A., Marseille, France). Statistics All data are shown as mean ± SEM. Statistical analysis was performed by using the Student's t test for paired variates and one-way ANOVA (27). Values of P < 0.05 were considered significant.

Results The effects of a 4-h exposure of confluent BPAEC monolayers to IL-2 (500 to 25,000 U/ml) on (1 251]albumin transfer rates across the BPAEC monolayers are shown in Figure 1 and expressed as the percent change above control BPAEC monolayers exposed to serum-free media alone. Incubation

Downie, Ryan, Hayes et al.: IL-2 Directly Increases Endothelial Permeability to Albumin In Vitro

,.

(n=5)

(n=6) o

I

100

[ 111111

i

(n=6)

t Ii I1II

1

1000 10000 IL-2 (Unitsjml)

I

I t t I [II

100000

Figure 1. [i25I]albumin steady-state transfer rates (computed over the 90- to 240-min period after addition of the radiolabeled albumin probe to the upper well) for bovine pulmonary arterial endothelial cells (BPAEC) grown to confluence on collagen-coated polycarbonate filters and previously exposed for 4 h in serum-free media to either 500 U/rnl of interleukin-2 (IL-2) (n == 6), 5,000 U/rnl of IL-2 (n == 5), or 25,000 U/rnl (n == 6) of IL-2. Data are shown as percent change from paired control BPAEC exposed to serum-free media alone (mean ± SEM). * Significant difference from control values, P < 0.02. t Significant difference from next lower IL-2 concentration, P < 0.05.

of BPAEC monolayers with diluent alone for 4 h resulted in (125I]albumin steady-state transfer rates of 3.3 ± O.4%/h (n = 10). These values were similar to the values in control BPAEC monolayers, i.e., exposed to serum-free media alone (data not shown). In contrast, there was a significant

concentration-dependent increase in (125I]albumin transfer rates of BPAEC monolayers after exposure to IL-2 (500 to 25,000 U/rnl) for 4 h. Incubation of BPAEC monolayers with 500 U/rnl ofIL-2 for 4 h resulted in a significant 66 ± 9% increase in (125I]albumin transfer rates above control values (n = 6, P < 0.02). Incubation of BPAEC monolayers with 5,000 U/rnl of IL-2 resulted in a 126 ± 4% increase above control values (n = 5, P < 0.02). Similar to the 5,000 U/rnl group values, incubation of BPAEC monolayers with 25,000 U/rnl ofIL-2 for 4 h also resulted in a significant 110 ± 20% increase above control values (n = 6, P < 0.02). This increase at the 25,000 U/rnl IL-2 level was not significantly different from the changes in (125I]albumin transfer rates found at the 5,000 U/rnl IL-2level. The addition of nonradioactive 1% albumin serum to the upper well did not alter the steady-state clearance rates of BPAEC monolayers exposed to serum-free media alone or of BPAEC monolayers exposed to 5,000 U/ml ofIL-2 (data not shown). In regards to effects of duration of incubation of BPAEC with IL-2 on [I25I]albumin transfer rates, no further increases in [125I]albumin transfer rates were found in BPAEC monolayers incubated with 5,000 U/ml of IL-2 for 24 h (80 ± 21 % increase above control values, n = 6) compared with the 4-h incubation values (P = NS). The increased (125I]albumin transfer rates, after exposure to 5,000 U/ml of IL-2, were associated with inverted-phase microscopy evidence of increased formation of gaps between cells after 4 h of incubation (Figure 2). Similar to the (125I]albumin transfer rate data, no further increases in monolayer disruption were noted after an additional 20 h of incubation with 5,000 U/ml of IL-2 (Figure 2). No detectable release of LDH from BPAEC or increase in percent BPAECdetachment « 0.1% detachment at any IL-2 concentration) was found after exposure of BPAEC to 500 to 25,000

CONTROL

4 Figure 2. Inverted phase-contrast micrograph (x150) of a BPAEC monolayer exposed for 4 or 24 h to diluent alone (control) or to 5,000 Ulrnl ofIL-2. Note that the increase in gap formation between cells seen after a 4-h exposure to 5,000 U/rnl of IL-2 does not further increase over the subsequent 20 h.

hours

24 hours

61

IL-2

62

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

Vlml of IL-2 for 4 h. Similarly, no detectable LDH release or increase in percent BPAEC detachment was noted in BPAEC incubated with 5,000 Vlml for 24 h versus 4 h. In regards to pH]deoxyglucose release, no significant change was found in BPAEC monolayers exposed to 500 Vlml of IL-2 for 4 h compared with controls (52 % release versus 48% release, respectively, n = 4, P = NS). In contrast to these data, significant increases in ['Hldeoxyglucose release was found after exposure of BPAEC to either 5,000 Vlml or 25,000 Vlml of IL-2 for 4 h (average release for each IL-2 concentration of 57 % compared with control values of 48 %, n = 4, P < 0.05). As shown in Figure 3, the concentration of IL-2 in the conditioned media from confluent BPAEC monolayers decreased by 64% (at the 5,000 Vlmllevel) and 58% (at the 500 Vlml level) during the first 4 h of incubation. At both IL-2 concentrations, there was a further small decrease in IL-2 concentration over the subsequent 20 h of incubation with IL-2. Thus, there were demonstrable amounts of IL-2 present during the period of study. In studies comparing addition of IL-2 to unconditioned serum-free media in a cellfree system for similar time periods, it appeared that the presence of the BPAEC accounted for 41% of the decrease in IL-2 seen (data not shown). Additional studies were conducted to determine if IL-2 exposure to BPAEC monolayers resulted in the synthesis by BPAEC of other cytokines that may have altered BPAEC monolayer permeability (i.e., IL-l or TNF-a). No significant amount of either TNF-a « 2 Vlml) or IL-l « 0.1 Vlml) was found in conditioned media of BPAEC that had been previously incubated with 5,000 Vlml of IL-2 for 4 h. To determine the specificity of the IL-2-induced increases in endothelial cell albumin permeability, we also examined the effects of two other cytokines, TNF-a and IL-1.

o

o o o

...

\l---------------iiI

5000 units

C\I

I

::!

I \ \ \ \

1- - - - - - - -- - __ - - - - - - - - f

500 units

:5+-----,--....,----.----,----..-----, ... o 4 B 12 16 20 24 Time (Hours)

Figure 3. Changes in IL-2 concentration (U/ml) in conditioned media from BPAEC over a 24-h period after addition of either 5,000 U/ml of IL-2 (solid line) or 500 U/ml of IL-2 (dashed line). The amount of IL-2 found in the media after addition was measured by an enzyme-linked immunosorbent assay. Data are shown as mean ± SEM (n = 4 at each time point).

t

I

...

lU

III

~

-..

""co lU

t

d

III

T

t:.~ d

·s

,gCll

T

::;;:

.I

N

~

0

Control (n-4)

IL-2

(n=4)

IL-2

+ I.nti-IL-2

(n=4)

Figure 4. ('25I]albumin steady-state transfer rates for BPAEC exposed for 4 h in serum-free media to either diluent alone (control, n = 4), 5,000 U/ml of IL-2 (n = 4), or coincubated with 5,000 U/ml of IL-2 and 62.5 /Lg of a monoclonal antibody against IL-2 (anti-IL-2) (n = 4). Data are shown as mean ± SEM. t Significant difference from control, P < 0.05; :j: Significant difference from IL-2, P < 0.05.

BPAEC monolayers, incubated with TNF-a (2,000 Vlml) for 4 h, had a 90 ± 24% increase in steady-state [t25I]albumin transfer rates compared with paired control monolayers (n = 4, P < 0.05). In contrast, BPAEC layers incubated with IL-l (100 Vlml) for 4 h resulted in no significant increase in [t25I]albumin transfer rates compared with paired control monolayer values (mean difference of 1.2 ± 0.1%, n = 4, P = NS). As shown in Figure 4, coincubation of the BPAEC monolayers with IL-2 (5,000 Vlml) and a monoclonal antibody against IL-2 resulted in a significant protection against the IL-2-induced increase in BPAEC [125I]albumin transfer rates. The mean ['25I]albumin steady-state transfer rates in BPAEC monolayers coincubated with 5,000 Vlml of IL-2 and 62.5p,g of anti-IL-2 was 3.7 ± 0.9%/h (n = 4), which was similar to the control BPAEC monolayer values of 2.2 ± 0.3%/h (n = 4, P = NS) but significantly different from the IL-2-treated BPAEC monolayer values of 6.5 ± 1.5%/h (n = 4, P < 0.05). When BPAEC were incubated with the anti-IL-2Ra receptor antibody for 30 min before exposure to 5,000 Vlml of IL-2 for 4 h, there was no significant change in albumin transfer rates compared with monolayers exposed to 5,000 Vlml of IL-2 alone (data not shown). However, because of the uncertainty of the ability of the IL-2Ra antibody used to cross react in bovine species as indicated by flow cytometric preliminary studies (data not shown), we exposed HAEC to either diluent alone for 4 h, 5,000 Vlml of IL-2 for 4 h, or to the anti-IL-2Ra antibody for 30 min and then to 5,000 Vlml of IL-2 for 4 h. As shown in Figure 5, IL-2 alone increased the [,25I]albumin transfer rates of HAEC monolayers to a magnitude similar to that observed using BPAEC

Downie, Ryan, Hayes et al.: IL-2 Directly Increases Endothelial Permeability to Albumin In Vitro

t

I

.,

....,

ttl

P::",

...

~

rn

I:: ttl ... co

....

I::

·s

::J

T

T t'J

.0

1

on

'" ~

a

Control

IL-2

(n=6)

(n=6)

Anti-TAC

+ IL-2

(n=5)

Figure 5. [i25I]albumin steady-state transfer rates for human arterial endothelial cells exposed for 4 h in serum-free media to either diluent alone (control) (n = 6),5,000 U/ml oflL-2 (n = 6), or incubated with 2.5 !Jog of a purified monoclonal antibody against the 55-kD glycoprotein human low-affinity receptor for IL-2 (antiTAC) for 30 min at 37° C and then exposed to 5,000 U/ml of IL-2 for 4 h (anti-TAC + IL-2) (n = 5). t Significant difference from control, P < 0.05.

(Figure 1). In contrast to BPAEC, preincubation of HAEC monolayers with the anti-IL-2Ra antibody protected the HAEC against the effects ofIL-2 in regards to changes in albumin permeability. Additionally, HAEC incubated with IL-2 (5,000 U/ml) for 4 h demonstrated, by immunoenzymometric assay, a low density of soluble IL-2 receptors. The conditioned media from HAEC monolayers incubated with 5,000 U/ml of IL-2 for 4 h demonstrated 378 pg/5 X 105 cells of soluble IL-2 receptors. The cellular extract from the HAEC monolayers demonstrated 462 pg/5 x lOS cells of membrane IL-2 receptors. These data are similar to the data obtained in control HAEC monolayers.

Discussion As previously shown by our laboratory, bovine BPAEC grown on gelatin-coated polycarbonate filters form continuous intact monolayers with normal-appearing intracellular junctional complexes (16). The BPAEC monolayers in the present study achieved steady-state albumin clearance rates that were similar to values reported from our laboratory as well as other laboratories (16-19). The albumin transfer rates of endothelial cell monolayers, studied in this manner, have been shown to increase after exposure to several agents which may be correlative to various in vivo conditions of lung injury (16-19, 28,29). The use of an in vitro endothelial cell permeability model, as utilized in the present study, has a number of advantages including direct access to luminal and abluminal fluid for analysis, limitation to a single cell type, and that the experimental media can be defined in terms of its chemical composition (30). Furthermore, we have shown that HAEC can also be grown on filters and used for measurements of albumin permeability. The HAEC also ap-

63

pear to have similar steady-state albumin transfer rates and response to IL-2 compared with BPAEC (Figure 5). Using an isolated perfused rat lung preparation, IL-2 infusion has been shown to result in a rapid (~ 5 min), concentration-dependent increase in [125I)albumin uptake in the lungs with subsequent increases in total lung water (4). Ferro and associates have demonstrated increases in pulmonary capillary filtration coefficients, an index of vascular permeability to water, in isolated guinea pig lungs after exposure to 2,000 to 10,000 Ufml ofIL-2 (31). Rapid increases in radiolabeled albumin vascular clearance have also been demonstrated using intact murine models (6). Glauser and co-workers demonstrated, using an intact in vivo sheep model, that IL-2 infusion (in concentrations similar to that used in humans) caused severe hemodynamic alterations and increased pulmonary vascular permeability to macromolecules (9). These investigators have also shown that IL-2 infusion resulted in infiltration of lymphoblastoid cells into the lung. Although the exact sequence of events are not completely known, it has been suggested that the development of the vascular leak syndrome depends on the presence of T lymphocytes and possibly the secondary production of other lymphokines with subsequent endothelial cell activation and injury (3, 32, 33). This local production of various cytokines could enhance the adherence oflymphocytes and neutrophils to endothelium (33). For example, it has been recently reported that IL-2 can stimulate the synthesis of TNF-a in vitro (34,35) and that TNF-a is toxic to endothelial cells in vivo and in vitro (36). We and others (28) have shown that TNF-a, exogenously administered, significantly increases [125IJalbumin permeability in BPAEC after 4 h of exposure. Similar to others (28), we were not able to demonstrate an increase in [I25IJalbumin permeability after 4 h of incubation with IL-I. In the present study, we have demonstrated that exposure of BPAEC to IL-2 for 4 h resulted in significant dosedependent increases in the transfer rates of [I25I)albumin across pulmonary endothelial monolayers in vitro without requiring the presence of other cell types or serum components. The increase in [125I]albumin transfer rates appears to be maximum at the 5,000 U/ml IL-2 concentration (Figure 1). The addition of albumin did not alter the response of the BPAEC to IL-2. Furthermore, there did not appear to be a further increase in either (125I]albumin transfer rates or morphologic evidence of increased gap formation between cells (Figure 2) when BPAEC monolayers were incubated with 5,000 U/ml of IL-2 for 24 h versus 4 h. We do not believe that these transfer rate changes are specific for the recombinant form of the IL-2 molecule because we found similar increases in BPAEC albumin permeability when biologic IL-2 (5,000 U/ml; Collaborative Research) was used in place of recombinant IL-2 (average increase in [125I]albumin BPAEC transfer rates of 59 % above paired control values, n = 2). The IL-2-induced changes in albumin permeability were associated with evidence of cell toxicity as demonstrated by increased release of pH]deoxyglucose at IL-2 concentrations ~ 5,000 U/ml. The increased IL-2-dependent release of pH]deoxyglucose was not associated with a demonstrable increase in BPAEC LDH release or cell detachment. These data are similar to the data of Andreoli and colleagues who concluded that the [3H]deoxyglucose

64

AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 7 1992

assay is a more sensitive index of endothelial cell damage than other assays including LDH or 51Cr release (26). When administered intravenously to human cancer patients, IL-2 has an estimated clearance half-life of approximately 30 to 60 min with peak serum levels of 5,000 U/ml shown to occur after administration of 106 U/kg body weight of IL-2 (2). Thus, the concentration of IL-2 found in the BPAEC-conditioned media after the 4-h incubation period to 5,000 U/ml of IL-2 (Figure 3) is similar to IL-2 concentrations measured in serum samples from patients receiving IL-2 infusions in vivo (2). These data suggest that this in vitro IL-2-induced endothelial cell injury model may have relevance to the vascular leak syndrome observed in vivo. Using a similar in vitro model using bovine arterial endothelial cells, Ferro and co-workers were unable to demonstrate an increase in radio labeled albumin permeability after incubation with 1,000 to 10,000 U/ml of IL-2 for 30 min (31). These data are consistent with our findings of no demonstrable morphologic changes in BPAEC monolayers after a 1-hexposure to 5,000 U/ml ofIL-2 (data not shown). The observation that definite gap formation between cells and increases in (1z5I]albumin transfer rates were found by 4 h suggest a time dependency for IL-2-induced changes in vitro. A recent report has demonstrated no increase in albumin permeability of human umbilical vein endothelial cell monolayers exposed in vitro to various cytokines, including IL-2, but did report a significant increase in permeability after incubation with LAK cells (35). The difference between that published study and the data we now present may relate to the use by the other investigators of a shorter (90 min) time of incubation (a time period in which we did not detect any IL-2-induced morphologic changes) or to the possible difference in response of endothelium from different sites (arterial versus venous) to IL-2. Data presented in the present study, as well as other reported data (28, 37), suggest that the injury (morphologic alterations or albumin permeability increases) that occurs in endothelial cell monolayers after exposure to various cytokines (e.g., IL-l, IL-2, TNF-a) appears to have both a concentration and time dependency. This injurious effect of cytokines on endothelial cells differs from other injuries (e.g., induced by thrombin or histamine) that also may involve specific cell surface receptors. Thrombin or histamine causes endothelial cell injury within minutes of exposure rather than hours, as seen with cytokine exposure (19, 38). The exact mechanisms responsible for the differences between these various endothelial injuries are unknown but may, for example, involve differences in response of specific receptors. De novo cellular expression ofIL-2 receptors and binding and release of IL-2 is an important event involved in T-cell activation and graft rejection. Although the regulation of the IL-2 receptor in endothelial cells is not known, regulation ofIL-2 receptor expression in T cells, for example, has been shown to be associated with transmembrane signaling events including changes in intracellular calcium fluxes, activation of the phosphoinositide cycle, and translocation of protein kinase C from the cytosol to the cell membrane (39). The IL-2 receptor (IL-2R) is present in three forms: high-, intermediate-, and low-affinity forms (39). IL-2 has been

shown to bind to distinct receptor molecules, the IL-2 receptor a (IL-2Ra, TAC antigen, p55) and a recently described IL-2 receptor {3 (IL-2R{3, p70-75). The IL-2Ra and the IL-2R{3 chains, although structurally distinct molecules, both bind IL-2 independently (39). The IL-2Ra form constitutes the low-affinity form. The IL-2R{3 form constitutes the intermediate-affinity form. The association of the {3 chain with the a chain seems to result in the high-affinity form of the IL-2 receptor in lymphoid cells (39). Whether the expression of the a and {3 chains in nonlymphoid cells results in the formation of the high-affinity receptor is not known. Immunoprecipitation and fluorescence-labeled antibody binding studies have shown that the CD 25 monoclonal antibody (used in the present study to protect against the permeability changes) reacts with the same molecule as the anti-IL-2Ra monoclonal antibody (40, 41). The vascular leak syndrome does not appear in vivo in humans without "priming" of the patient with IL-2 and the time required for this priming process appears to be correlated with the time required for the induction and expression of IL-2Ra receptors (2, 42, 43). The data presented in the present study, demonstrating reversal of IL-2-induced increases in albumin permeability in HAEC by pretreatment with an anti-IL-2Ra antibody and the expression of soluble IL-2R receptors on HAEC (which have homology with the IL-2Ra receptors) (43), suggests that the low-affinity IL-2Ra receptor molecule may playa role in the IL-2-induced permeability change observed in vitro and perhaps also in vivo. However, it is possible that the IL-2-induced permeability changes we observed are not mediated by an IL-2 receptor. IL-2 could exert a direct, non-receptor-mediated toxic effect on endothelial cells. This possibility is suggested by the IL-2-induced increases in [3H]deoxyglucose release after exposure to IL-2 concentrations ~ 5,000 U/mL However, the presence of demonstrable receptors on the HAEC (although not increased with IL-2 exposure) and the demonstration of a time and concentration dependency to the endothelial injury with a maximal response occurring after a 4-h exposure to 5,000 U/ml ofIL-2 (Figure 1) suggests that occupation of available receptors may have occurred. The protection afforded by the anti-IL-2Ra antibody (Figure 5) further supports the hypothesis of an IL-2 receptor-mediated endothelial injury. An alternative explanation is the possibility of the presence of a non-IL-2 receptor on the endothelial cell that shares some homology with a component molecule of the IL-2 receptor such as IL-2Ra. Urdal and associates have demonstrated the existence of transformed IL-2 receptor molecules that differ in molecular weight and structure but retain some function (44). Thus, it is possible that non-IL-2 receptors could be stimulated by IL-2 and blocked by anti-IL-2Ra antibodies but retain other unique functions. In summary, we have demonstrated that IL-2 has a direct effect on cultured bovine and human endothelial cell monolayers in vitro. The toxic effects of IL-2 were sublethal and did not require the presence of other cell types or serum proteins, and the permeability increase appeared, in part, to be IL-2Ra receptor-mediated. Obviously, the cause of the vascular leak syndrome in vivo is multifactorial, including endothelial cell activation, cellular infiltration, and endotoxinlike hemodynamic alterations. However, we suggest that IL-2-induced, IL-2Ra-mediated increases in endothelial

Downie, Ryan, Hayes et al.: IL-2 Directly Increases Endothelial Permeability to Albumin In Vitro

macromolecular permeability may play an important role in the pathogenesis of the vascular leak syndrome seen in vivo. Acknowledgments: The writers wish to thank Ms. Megan Corum and Ms. Anna McCrory for their expert technical assistance and Drs. Howard Ozer, David Harris, and Suzanne Becker for their support. We also wish to thank HoffmanLa Roche, Inc. for their generous donation of IL-2. This study was supported by Grants HL-39720, HL-07106, HL-33064, and HL-21568 from the National Institutes of Health. Dr. Downie was also supported by an Allen & Hanburys Pulmonary Fellowship Award. This work was presented in part at the Fall Meeting of the American Physiologic Society, Montreal, Quebec, Canada, October 8-13, 1988 and at the Annual Meeting of the American Thoracic Society, Cincinnati, Ohio, May 13-16, 1989.

References 1. Smith, K. A. 1988. Interleukin-2: inception, impact and implications.

Science 240:1169-1176. 2. Rosenberg, S. A., M. T. Lotze, and J. J. Mule. 1988. New approaches to the immunotherapy of cancer using interleukin-2. Ann. Intern. Med. 108:853-864. 3. Lotze, M. T., V. L. Matory, and A. A. Rayner. 1986. Clinical effects and toxicity of interleukin-2 in patients with cancer. Cancer 58:2764-2772. 4. Fairman, R. P., F. L. Glauser, R. E. Merchant, D. Bechard, and A. A. Fowler. 1987. Increase of rat pulmonary microvascular permeability to albumin by recombinant interleukin-2. Cancer Res. 47:3528-3532. 5. Silverman, H. S., J. Abrams, and L. J. Rubin. 1988. Effects of interleukin-2 on oxygen delivery and consumption in patients with advanced malignancy. Chest 94:816-821. 6. Rosenstein, M., S. E. Ettinghauser, and S. A. Rosenberg. 1986. Extravasation of intravascular fluid mediated by the systemic administration of recombinant interleukin-2. J. Immunol. 137:1735-1742. 7. Gaynor, E. R., L. Vitch, L. Stichlin et al. 1988. The hemodynamic effects of treatment with interleukin-2 and lymphokine-activated killer cells.

Ann. Intern. Med. 109:953-958. 8. Ognibene, F. P., S. A. Rosenberg, M. Lotie et al. 1988. Interleukin-2 administration causes reversible hemodynamic changes and left ventricular dysfunction similar to those seen in septic shock. Chest 94:750-754. 9. Glauser, F. L., D. E. Bechard, G. G. Deblois et al. 1989. Cardiorespiratory and cellular changes with interleukin-2 infusion in sheep. J. Appl.

Physiol. 66:128-134. 10. Pober, J. S. 1988. Cytokine-mediated activation of vascular endothelium. Am. J. Pathol. 133:426-433. II. Bevilacqua, M. P., J. S. Pober, M. E. Wheeler, R. S. Cotran, and M. A. Gimbrone. 1985. Interleukin-I acts on cultured human vascular endothelial cells to increase the adhesion of polymorphonuclear leukocytes, monocytes and related leukocyte cell lines. J. Clin. Invest. 76:20032011. 12. Bevilacqua, M. P., J. S. Pober, G. R. Majean, R. S. Cotran, and M. A. Gimbrone, Jr. 1984. Interleukin-I (IL-I) induces biosynthesis and cell surface expression of procoagulant activity in human vascular endothelial cells. J. Exp. Med. 160:618-623. 13. Cotran, R. S., J. S. Pober, M. A. Gimbrone, Jr. et al. 1987. Endothelial activation during interleukin-2 immunotherapy. J. Immunol. 139:18831888. 14. Damle, N. K., and L. V. Doyle. 1989. IL-2-activated human killer lymphocytes but not their secreted products mediate increase in albumin flux across cultured endothelial monolayers. J. Immunol. 142:2660-2669. 15. Mier, J. W., E. P. Brandon, P. Libby, M. W. Janicka, and F. R. Aronson. 1989. Activated endothelial cells resist Iymphokine-activated killer cell-mediated injury. J. Immunol. 142:2407-2414. 16. Friedman, M., U. S. Ryan, W. C. Davenport, E. L. Chaney, D. L. Strickland, and L. Kwock. 1986. Reversible alterations in cultured pulmonary artery endothelial cell monolayer morphology and albumin permeability induced by ionizing radiation. J. Cell. Physiol. 129:237-249. 17. Shasby, D. M., S. S. Shasby, J. M. Sullivan, and M. J. Peach. 1982. Role of endothelial cell cytoskeleton in control of endothelial permeability.

Circ. Res. 51:657-661. 18. Cooper, J. A., P. S. Del Vecchio, F. L. Minnearetal. 1987. Measurement of albumin permeability across endothelial cell monolayers in vivo. J.

Appl. Physiol. 62:1076-1083. 19. Meyrick, B., and K. L. Brigham. 1984. Increased permeability associated with dilation of endothelial cell junctions caused by histamine in intimal explants from bovine pulmonary artery. Exp. Lung Res. 6:11-25. 20. Shasby, D. M., S. E. Lind, S. S. Shasby, J. C. Goldsmith, andG. W. Hunninghake. 1985. Reversible oxidant-induced increases in albumin transfer

65

across cultured endothelium: alterations in cell shape and calcium homeostasis. Blood 3:605-614. 21. Ryan, U. S., M. Mortana, and C. Whitaker. 1980. Methods for microcarrier culture of bovine pulmonary artery endothelial cells avoiding enzymes. Tissue Cell 12:619-635. 22. Jaffe, E. A., L. W. Hoyer, and R. L. Nachman. 1973. Synthesis of antihemophilic factor antigen by cultured human endothelial cells. J. Clin.

Invest. 52:2757-2764. 23. Maciag, T., J. Cerundolo, S. Ilsley, P. R. Kelley, and R. Forand. 1979. An endothelial cell growth factor from bovine hypothalmus: identification and partial characterization. Proc. Natl. Acad. Sci. USA 11:5674-5678. 24. Voyta, J. C., P. C. Netland, D. P. Via, and B. R. Zetter. 1980. Specific labelling of endothelial cells using fluorescent acetylated low density lipoprotein. J. Cell Biol. 99:8Ia. (Abstr.). 25. Urdal, D. C., C. J. Maness, S. Gillis, A. Larsen, and S. K. Dower. 1984. Purification and chemical characterization of the receptor for interleukin-2 from activated human T-lymphocytes and from a human T-celilymphoma line. Proc. Natl. Acad. Sci. USA 81:6481. 26. Andreoli, S. P., R. L. Baehner, and J. M. Bergstein. 1985. In vitro detection of endothelial cell damage using 2-deoxY-D-JH-glucose: comparison with chromium 51, JH-Ieucine, JH-adenine, and lactic dehydrogenase. J.

Lab. Clin. Med. 106:253-261. 27. Snedecor, G. W., and W. G. Cochran. 1978. Statistical Methods. 6th ed. Iowa State University Press, Ames, IA. 28. Royall, J. A., R. L. Berkow, J. S. Beckman, K. M. Cunningham, S. MataIan, and B. A. Freeman. 1989. Tumor necrosis factor and interleukin la increase vascular endothelial permeability. Am. J. Physiol. 257:399-410. 29. Neufeld, G. F., J. J. Williams, D. J. Graves, L. R. Sona, and B. E. Marshall. 1975. Pulmonary capillary permeability in man and a canine model of chemical pulmonary edema. Microvasc. Res. 10:192-207. 30. Albelda, S. M., P. M. Sampson, F. R. Haselton et al. 1988. Permeability characteristics of cultured endothelial cell monolayers. J. Appl. Physiol. 64:308-322. 31. Ferro, T. J., A. Johnson, J. Everitt, and A. B. Malik. 1989. IL-2 induces pulmonary edema and vasoconstriction independent of circulating lymphocytes. J. Immunol. 142:1916-1921. 32. Pober,J. S., andM. A. Gimbrone, Jr. 1982. ExpressionofIa-likeantigens by human vascular endothelial cells is inducible in vitro: demonstration by monoclonal antibody binding and immunoprecipitation. Proc. Natl. Acad. Sci. USA 79:6641-6645. 33. Pober, J. S., M. A. Gimbrone, Jr., R. S. Cotranetal. 1983. Iaexpression by vascular endothelium is inducible by activated T-cells and by human gamma-interferon. J. Exp. Med. 157:1339-1353. 34. Safo, N., T. Goto, K. Haranchaetal. 1986. Actions of tumor necrosis factor on cultured vascular endothelial cells: morphologic modulation growth inhibition, and cytotoxicity. J. Natl. Cancer Inst. 76: 1113-1119. 35. Damle, N. K., and L. V. Doyle. 1988. IL-2-activated human killer lymphocytes but not their secreted products mediate increase in albumin flux across cultivated endothelial monolayers. J. Immunol. 142:2660-2669. 36. Le, J., and J. Vilcek. 1987. Tumor necrosis factor and interleukin-I: cytokines with multiple overlapping biological activities. Lab. Invest. 56:234-248. 37. Bucana, C. D., J. Trial, A. C. Papp, and K. K. Wu. 1988. Bovine aortic endothelial cell incubation with interleukin 2: morphological changes correlate with enhanced vascular permeability. Scanning Microsc. 2: 1559-1566. 38. Laposata, M., D. K. Dovnarsky, and H. S. Shin. 1983. Thrombin-induced gap formation in confluent endothelial cell monolayers. Blood 62: 549556. 39. Waldman, H. A. 1986. Structure, function and expression of interleukin-2 receptor on normal and malignant lymphocytes. Science 232:727-732. 40. Uchiyama, T., S. Broder, and T. A. Waldman. 1981. A monoclonal antibody (anti-tac) reactive with activated and functionally mature human T cells. 1. Production of anti-tac monoclonal antibody and distribution of tac (+) cells. J. Immunol. 126: 1393-1397. 41. Robb, R. J., and W. C. Greene. 1983. Direct demonstration of the identity of T cell growth factor binding protein and the tac antigen. J. Exp. Med. 158: 1332-1337. 42. Circardi, M., M. Gardinali, G. Bisiani, A. Rosti, P. Allavena, and A. Agostoin. 1990. The systemic capillary leak syndrome: appearance of IL-2-receptor-positive cells during attacks. Ann. Intern. Med. 113:475477. 43. Sayar, D., M. Ketzimel, L. Gerez, C. Silberberg, A. Reshef, and R. Kaempfer. 1990. Expression of the human IL-2 receptor on lymphocytes involves rapid turnover of its p 55 a-subunit. J. Immunol. 145:29462949. 44. Urdal, D. L., C. J. March, A. Larsen, andS. K. Dower. 1984. Purification and chemical characterization of the receptor for IL-2 from activated human T lymphocytes. Proc. Natl. Acad. Sci. USA 81:6481-6485.

Interleukin-2 directly increases albumin permeability of bovine and human vascular endothelium in vitro.

The direct effects of interleukin-2 (IL-2) on albumin permeability of cultured bovine pulmonary artery endothelial cell (BPAEC) and human arterial end...
984KB Sizes 0 Downloads 0 Views