711

Effects of Staphylococcal Toxic Shock Syndrome Toxin 1 on Aortic Endothelial Cells Peter K. Lee, Gregory M. Vercellotti, James R. Deringer, and Patrick M. Schlievert

Departments of Microbiology and Medicine. University of Minnesota Medical School. Minneapolis

Toxic shock syndrome (TSS) is an acute multisystem illness characterized by the sudden onset of fever, hypotension, rash, variable multiorgan involvement, and desquamation of the skin on recovery [1-3]. Since its first description in 1978 [1], the syndrome has been associated with infection or colonization by Staphylococcus aureus strains that produce pyrogenic toxins, notably, TSS toxin 1 (TSST-1) and enterotoxin types Band C [4-8]. These toxins belong to a larger family of pyrogenic toxins made by S. aureus and group A streptococcus and share many physicochemical and biologic properties [9]. In TSS, hypotension and shock are the most severe symptoms and often the cause of death. Clinically, the patients have massive nonpitting edema, especially involving the hands and feet, indicating significant extravasation or leakage offluid from inside the vessel to the interstitium [10]. In vivo animal model studies of TSS further suggested the importance of fluid leakage from vessels [11]. The mechanism for the leakage caused by TSS-associated toxins is unknown, but they may have direct toxic effects on endothelial cells, thereby compensating the capillary system and leading to leakage and hypotension. This is consistent with the observation by Kushnaryov et al. [12] that TSST-I binds to human umbilical vein endothelial cells.

Received 19 February 1991; revised 28 May 1991. Presented in part: annual meeting of the American Society for Microbiology (abstract B272). Dallas. May 1991. Grant support: National Institutes of Health (HL-36611. HL-33793. and [to P.K.L.] CA-09138). Reprints or correspondence: Dr. Peter K. Lee, Department of Microbiology. Box 196 UMHC, University of Minnesota Medical School, 420 Delaware St. S.E., Minneapolis. MN 55455-0312. The Journal ofInfectious Diseases 1991;164:711-9 © 1991 by The University of Chicago. All rights reserved. 0022-1899/91/6404-0011 $0 1.00

To examine the possibility that capillary leak results from the direct toxic effects ofTSST-1 (the representative TSS-associated toxin and most commonly associated with TSS), the effects of the toxin on endothelial cells isolated from porcine aorta were examined using a slCr-release assay. Furthermore, the mechanism of the toxic effects on endothelial cells was examined, and binding studies were done to examine for specific receptor(s) on endothelial cells for TSST-1. Last, to mimic in vivo capillary leakage, a transendothelial membrane permeability model was used to examine the ability of TSST-1 to cause leakage across the endothelium.

Materials and Methods Toxin preparation. TSST-l (from human strain MN 8 of S. aureus) was prepared by ethanol precipitation and isoelectric focusing [13], and ovine TSS toxin (TSST-ovine) was prepared by a similar method from ovine S. aureus RN5625. TSST-ovine served as the negative control toxin in the slCr-release assays and in the binding studies. Toxin concentrations were determined by Ouchterlony immunodiffusion [14] and protein assay (Bio-Rad, Richmond, CA) and estimation by SDS-PAGE [15]. Toxin preparations migrated as homogeneous bands when 20I1-g amounts were subjected to SDS-PAGE and stained with Coomassie brilliant blue. Antisera. Anti-TSST -1 serum from American-Dutch belted rabbits was prepared as described previously [14]. Anti-staphylococcal a-hemolysin (or a-toxin) serum was prepared in a similar manner using a-hemolysin isolated from strain MNBI by ethanol precipitation and isoelectric focusing [13]. Normal rabbit serum was obtained from a rabbit that had not been immunized. Isolation and culture of porcine aortic endothelial cells (PAEC). PAEC were isolated enzymatically from porcine aorta using type I collagenase (Worthington Biochemical, Freehold, NJ; 0.2% for 15 min at 37°C) as described previously [16]. Primary cells were grown in complete Dulbecco's modified Eagle's

Downloaded from http://jid.oxfordjournals.org/ at New York University on May 26, 2015

In staphylococcal toxic shock syndrome, hypotension and shock due to capillary leak may rapidly lead to death of the host. To investigate its pathogenesis, the cytotoxic effects of toxic shock syndrome toxin 1 (TSST-1) on porcine aortic endothelial cells (PAEC) were examined in vitro. TSST-l killed PAEC (as measured by S1Cr release) in a dose- and time-dependent fashion and was blocked by anti-TSST-1 antibodies. Receptor-mediated endocytosis may be critical for the cytotoxic effects of TSST-1, as killing was inhibited by cold (4°C) and by addition of chloroquine and methylamine. Furthermore, calcium and oxygen appeared necessary for TSST-1 effects on PAEC. Membrane receptor binding studies indicated PAEC bind TSST-1 with high affinity (Kd = 5.7 X 10-7 M) and had 2.2 X 104 receptors/cell. Last, as measured by 12sI-labeled albumin flux in a transendothelial permeability model, TSST-1 enhanced the permeability of PAEC monolayers in a dose- and time-dependent manner.

712

Lee et al.

TSST-I-polymyxin B mixture was incubated with monolayers for up to 1 h at 37°C, and then wells were harvested as described above. Endotoxin (l and 10 JLg/well) and TSST-I (up to 5 JLg/ well) were added to monolayers and incubated for up to 4 h at 37°C. Reaction mixtures were harvested as before. Endotoxin was obtained from Salmonella typhimuriurn using the method of Westphal et al. [20]. Effect oftemperature on TSST-l cytotoxicity. After 51Crlabeling and washing, confluent PAEC monolayers were incubated with 5 or 10 JLg ofTSST-1 per well in HBSS at 37 or 4°C for 0.5 and I h. To assay at lower temperature, the monolayers and the necessary reagents were preincubated on ice at 4°C. Excess 51Cr was washed from monolayers using ice-cold HBSS; the reaction solution was harvested at 4°C. Role of divalent cations in TSST-I cytotoxicity. After 51Cr labeling and washing, confluent PAEC monolayers were preincubated with 0.25 ml of EGTA (4 mM) for 20 min at 37°C. TSST-I (10 JLg in 0.25 ml) was then added to each well and incubated for 2 h. Reaction solution was harvested as described above. The 51Cr-release assay was also done with Ca/Mg-free HBSS. After 51Cr labeling, confluent PAEC monolayers were washed four times with Ca/Mg-free HBSS. TSST-I (10 JLg) in Ca/Mg-free HBSS was added to each well and incubated for 2 h. At the end of the incubation period, the reaction solutions were harvested as described above using Ca/Mg-free HBSS. Effects of chloroquine and methylamine. After 51Cr labeling and washing, confluent PAEC monolayers were preincubated with 0.25 ml of receptor-mediated endocytosis inhibitor chloroquine (diphosphate salt, 250 JLM; Sigma) or methylamine hydrochloride (50 mM; Sigma) for 20 min at 37°C. TSST -I (10 JLg in 0.25 ml ofHBSS) was added to each well and incubated for 2 h. Reaction solution was harvested as described above. Role of oxygen and oxidants in TSST-l cytotoxicity. All reagents necessary for the cytotoxicity assay under anaerobic conditions were preincubated in an anaerobic (85% N 2, 10% CO 2 , 5% H 2) system (model 1024; Forma Scientific, Marietta, OH) overnight. Confluent 51Cr-Iabeled PAEC monolayers were placed inside the anaerobic chamber and washed three times with anaerobic HBSS. TSST-I (5 or 10 JLg in anaerobic HBSS) was added per well and incubated for I h at 37°C. Reaction solutions were harvested as described above using anaerobic HBSS. Furthermore, deferoxamine was used in an attempt to protect PAEC from TSST-1. Methods were as described above, except that monolayers were preincubated for 20 min at 37°C with deferoxamine mesylate ( 10 mM) before toxin addition. Iodination of TSST-l and bovine serum albumin (BSA). The chloramine T procedure [21] was used to label TSST -1 and BSA (Sigma) with 1251. As described previously [22], 100 JLg ofTSST1 was shaken at room temperature for 2 min with 10 JLg of chloramine T (Sigma) and 1 mCi of carrier-free 1251 (Amersham). The labeling reaction was stopped by adding 30 JLg of sodium metabisulfite. After 2 min, KI was added to achieve a final iodide concentration of 5 X 10-5 M. The entire reaction took place in 300 JLI of0.1 M TRIS buffer, pH 7.4. The mixture was separated over a Sephadex G-25 (Sigma) column packed in a 12-ml syringe at room temperature and eluted with PBS containing 0.25% (wt/vol) gelatin. The 1251-labeled TSST-I C251TSST-I) was collected in 0.5-ml fractions and tested for imrnu-

Downloaded from http://jid.oxfordjournals.org/ at New York University on May 26, 2015

MEM (DMEM; GIBCO, Grand Island, NY) containing 16% (vol/vol) heat-inactivated fetal bovine serum (GIBCO), penicillin (100 units/rnl), and streptomycin (100 units/ml) supplemented with L-glutamine at 37°C in 5% CO 2 , Endothelial cell colonies were identified by their morphology and ability to take up acetylated low-density lipoproteins [17]. Cells were subcultured after resuspension by brief exposure to 0.05% trypsin (Sigma, St. Louis) plus 0.53 mM EDTA and then grown to confluence in 24-well (2-cm 2 wells, 2 X 10 5 cells/well) or 48-well (Ivcrrr' wells, I X 10 5 cells/well) culture plates (Costar, Cambridge, MA) for the 51Cr-release assay and for binding studies, respectively. The cultures were used from passages 4 to 8 and were studied within 48 h of reaching confluence. PAEC cytotoxicity assays. Confluent endothelial cells grown in 24-well tissue culture plates were washed twice with Hanks' balanced salt solution (HBSS; GIBCO) at 3rC, radiolabeled with 2 JLCijwell Na251Cr04 (Amersham, Arlington Heights, IL) for 3 h in HBSS, and then washed three times with HBSS. The monolayers were exposed to toxin (TSST-lor TSST-ovine) in HBSS (0.5 ml total volume) for the desired amount of time at 37°C in a humidified atmosphere of 5% CO 2 , For the timecourse assay, 10 JLg of TSST-1 (or TSST-ovine) per well was added to the PAEC monolayer and incubated for up to 4 h. In the dose killing assay, 0.1-100 JLg ofTSST-l per well was added to monolayers and incubated for 2 h. Total volume of reaction solution in all experiments was limited to 0.5 ml/well. At the end of each incubation period, the reaction solution was removed, the monolayer was washed twice with 0.5 ml of HBSS, and the combined washes from each well were centrifuged at 1000 g for 7 min. The radioactivity of the supernates, the pellets, and the NaOH-Iysed (adhered) endothelial cells were measured separately in a ')'-counter (Gamma 5500; Beckman, Palo Alto, CA), generating specific cytotoxic values calculated as previously described [18]. Each experiment was done in quadruplicate with spontaneous 51Cr release subtracted from each value. Spontaneous 51Crrelease was 10% ± 7% in all experiments. Data are ±SE. Cytotoxicity of PAEC was verified using the trypan blue viability assay [19]. After toxin incubation with PAEC, the endothelial cells were detached with brief exposure to 0.05% trypsin + 0.53 mM EDTA. The cells were pelleted and gently resuspended in calcium- and magnesium-free (Ca/Mg-free) HBSS. One part of trypan blue saline solution was added to one part of cell suspension. The cells were counted on a hemocytometer for unstained and stained cells. Effects ofantisera on PAEC cytotoxicity. TSST-I (5 JLg in 0.25 ml) was pre incubated with 0.25 ml of anti-TSST-l rabbit serum, normal rabbit serum, or anti-a-hemolysin serum for 15 min at 37°C. The entire reaction mixture was then added to 51Cr-labeled confluent PAEC monolayers and incubated for 2 h at 37°C in 5% CO 2 , Percentage of inhibition of cytotoxicity was calculated relative to the cytotoxicity due to TSST-1 alone: counts per minute (cpm) of 51Cr release from TSST-I with antiserum/cpm of 51Cr from TSST-l alone X 100%. Effect ofendotoxin on TSST-l cytotoxicity. TSST-l (5 JLg) in HBSS was preincubated with polymyxin B sulfate (2.5, 12.5, and 125 units, where 10 units = 1 JLg; Roerig-Pfizer, New York) for 30 min at 37°C before addition to PAEC monolayers. The

lID 1991; 164 (October)

11D 1991;164 (October)

Capillary Leak in Toxic Shock Syndrome

80 4)

r.n 60 as Q)

Q)

a:

...

40

~ 0

20

0

--ts-- TSST·1

---..-

TSST-o

...--"T"""----...----"'I"-===;====~

O+-...

o

2

3

4

Hours Figure 1. Time course of toxic shock syndrome toxin I (TSST1) cytotoxicity of porcine aortic endothelial cells (PAEC): Confluent PAEC monolayers in 2-cm 2, 24-well tissue culture plates were incubated with 0.5 ml of Hanks' balanced salt solution containing 10 J.Lg ofTSST-l or TSST-ovine (TSST-o) at src: At end of incubation, reaction solution was harvested and percentage of SlCr release was calculated. Results are shown as mean ± SE of quadruplicate samples in duplicate experiments.

of incubation, the plates were gently rotated. The permeability of 1251-BSA or radioactivity to the subluminal chamber was counted in the -y-counter. Calculating from specific activity, the counts were converted to picomoles of 1251-BSA, and the dosedependence data were presented as permeability (picomoles of 12sl-BSA/h) across the Transwell membrane [26]. All counts were subtracted for nonspecific permeability (0 J.Lg ofTSST-l in luminal chamber) of I25!-BSA across the membrane. All samples were run in quadruplicate and repeated. For time-course studies, 5 J.Lg ofTSST-l was added to the PAEC-containing Transwells and incubated for up to 195 min, with plates gently rotated for the last 15 min of every time point. As with the dose-dependence study, at each time point, all counts were adjusted for nonspecific permeability of 1251-BSA across the PAEC transmembrane. For time-course studies, all samples were run in quadruplicate. Data from the time-course assay were presented as permeability (picomoles of I2SI-BSA) across the Transwell membranes plotted against time.

Results Time course of PAEC cytotoxicity. Time-course studies demonstrated rapid PAEC cytotoxicity from TSST-l administration. Addition of TSST-1 at 10 ,ug/well had an immediate cytotoxic effect on PAEC (figure 1). After 15 min of incubation, 18.4% of 51Crwas released specifically in the reaction solution. The percentage of cytotoxicity or 51Cr release increased to a maximum of near 60% at 1.5-2 h. This represented nearly 100% cytotoxicity as observed by trypan blue viability assay. Most of the radioactivity from harvested reaction solutions resided in the supernatants, not in the pellets. PAEC detachment ranged from 5% to 12% of 51Cr release or cytotoxicity values. TSST-ovine served as the negative control toxin. It is immunologically identical to TSST-1 and

Downloaded from http://jid.oxfordjournals.org/ at New York University on May 26, 2015

nologic reactivity with hyperimmune serum to TSST -1 in Western blot analysis [23] and for biologic activity in a lymphocyte proliferation assay [4]. 1251-labeled BSA C25I-BSA) was made using the same procedure. Binding assays. Competitive binding assays, which used increasing concentrations of unlabeled, native TSST -1 and a constant amount of 1251_TSST-l, were done to determine the specificity of binding of TSST-l to PAEC. PAEC were grown to confluency in l-cm", 48-well plates (1 X 105 cells/well). Monolayers and all reagents were chilled on ice at 4°C. After twice washing the monolayers with cold HBSS, 0.2 Ilg of 1251_TSST-I (specific activity, 34 IlCi/llg) and increasing concentrations (0.1-500 Ilg) of unlabeled TSST-1 in a 100-1l1 total volume of HBSS with 1%(0.15 mM) BSA (HBSS-BSA) were added to each well and gently rotated (The Belly Dancer; Stovall Life Science, Greensboro, NC) for I h at 4°C. Monolayers were then washed five times with cold HBSS. 125 1_TSST-l bound to PAEC was determined by harvesting the cells with I M NaOH and counting the pellet in a -y-counter. All samples were assayed in quadruplicate. The results are reported after subtracting the nonspecific binding of 1251_TSST-1 (or cpm bound in the presence of 500 Ilg of unlabeled TSST-l) from each value. Data are presented as percentage of control binding (or cpm bound in the presence of unlabeled TSST-1 /cprn bound without unlabeled TSST -I X 100%) plotted against log of competitor, or unlabeled TSST-I, concentration. Data were analyzed according to the method of Scatchard [24]. To examine competitive binding of TSST-ovine to monolayers, unlabeled TSST-ovine at 0.1-500 Ilg was used as the competitor to 1251-TSST-l (0.2Ilg) binding. For TSST -ovine, all samples were run in triplicate. Binding studies were done as described above. Permeability assay. A modification of method described previously [25] was used. Briefly, PAEC were cultured and permeability assays were done in 24-well tissue culture plates containing microporous polycarbonate membrane-bottomed inserts set in the wells (Transwells: Costar). The membrane has OA-llm pores and 15%-20% porosity and is 6.5 mm in diameter. The inside of the Transwell insert is designated as the luminal chamber and the outer as the subluminal chamber. Transwells were preincubated with complete DMEM for 1 h and then with 0.25 ml of 50 Ilg/ml human fibronectin overnight at 37°C. Confluent PAEC monolayers were subcultured and seeded at 3.3 X 104 cells/well (I: I) onto Transwells. Culture medium was changed every 2-3 days, and the PAEC Transwell monolayers were used 9-12 days after seeding. To assess confluency of PAEC monolayers on Transwells, adjacent 2-cm 2 wells without Transwells were seeded concurrently and checked for confluency. In preparation for the permeability assay, PAEC Transwell monolayers were transferred to HBSS with 1%(0.151 mM) BSA (0.7 ml)-containing 24-well plates, and the luminal chambers were washed once with HBSS-BSA. To study dose dependence, TSST-l (0.1-10 Ilg/well) and 1251-BSA (5Ilg/well) as the tracer molecule in HBSS-BSA (100 J.Ll total reaction volume per luminal chamber) were added to the luminal chamber and incubated for 90 min at 37°C. The specific activity of 125I-BSA was 11 IlCi/llg, and the total BSA (labeled and unlabeled) concentration in luminal chamber was 0.152 mM. For the last IS min

713

Lee et al.

714

80 Q) l/) 60 co Q)

~

7

8

9

a

~

40

9 0~

A

~

Q3

c:

z

a

~ lJ.

20

a 0 0

2

3

4

5

TSST-1

6

10

~o

(J,lQ)/Well

shares significant amino acid and nucleotide sequence similarities with TSST-I. However, unlike TSST-I, TSST-ovine lacks pyrogenic or lethal activity; in the 5tCr-release assay, TSST-ovine had no cytotoxic effect on the monolayers (figure I). Dose dependence ofPAEC cytotoxicity. PAEC cytotoxicity correlated with the amount ofTSST-I added per well. Addition of < 1 J.Lg ofTSST-l zwell had little effect on 5tCr release from PAEC (figure 2). However, with increasing concentrations ofTSST-I, increases in percentage of 5tCr release were observed up to 65% when 7-9 J.Lg ofTSST-I was added. Once again, this maximum level represented nearly 100% cytotoxicity as observed by trypan blue viability assay. TSST-ovine treatment (as much as 100 J.Lg/well) caused no 5tCr release from monolayers (unpublished data). Inhibition of PAEC cytotoxicity by anti-TSST-J serum. To further illustrate that PAEC cytotoxicity was due to TSST-I, TSST-I was preincubated with rabbit anti-TSST-I serum. This successfully inhibited the cytotoxic effect on PAEC. Addition of 5 J.Lg of TSST-I/well exhibited 40.6% ± 1.6% 5tCr release. However, after preincubating TSST-I with the antiserum, only 2.4% ± 0.5% 5tCr release was observed. This was calculated as being 94.1 %inhibition ofcytotoxicity. Preincubation with normal rabbit serum had no inhibitory effect (40.3% ± 2.0% 5tCr release). Furthermore, rabbit antia-hemolysin serum had no inhibitory effect on TSST-I. This suggested that a-hemolysin, a cytotoxic agent produced by S. aureus [27], did not contaminate our TSST-l preparation. Role ofendotoxin in PAEC cytotoxicity. The possible role of endotoxin in TSST-I cytotoxicity of PAEC was investigated. First, before administration to PAEC, the TSST-I preparation was preincubated with polymyxin B sulfate (which inactivates lipid A ofendotoxin) to examine the possible presence and involvement of endotoxin. Polymyxin Bat

2.5, 12.5, and 125 units/well failed to protect PAEC from the cytotoxic effects of TSST-I. After a l-h incubation, 33.1 % ± 1.0%5tCr release was observed with 5 f-Lg ofTSST-1 alone. With 5 f-Lg ofTSST-1 plus 2.5, 12.5, or 125 units of polymyxin B, 34.1 % ± 0.4%, 33.4% ± 0.3%, and 32.7% ± 1.1% 5tCr release was seen, respectively. Polymyxin B alone had no cytotoxic effect. Similarly, polymyxin B did not affect cytotoxicity in experiments with 4-h incubations. Second, endotoxin was administered with TSST -1 to monolayers in an attempt to potentiate PAEC cytotoxicity. TSST -I (5 f-Lg) added with lipopolysaccharide (LPS; I or 10 f-Lg) to monolayers for I h of incubation caused 33.4% ± 0.7% and 34.1 % ± 1.4%5tCr release, respectively, compared with 5 f-Lg ofTSST-1 alone (33.1 % ± 1.0%). Incubation for 4 h gave similar results, with no increase in cytotoxicity observed with the addition of LPS with TSST-I. Endotoxin (up to 100 J.Lg/ well) alone had no cytotoxic effect (no 5tCr release) on PAEC monolayers when incubated for up to 4 h. Temperature dependence ofPAEC cytotoxicity. To examine possible mechanisms of cytotoxicity, the 51Cr-release assay was done at 4°C. At this lower temperature, endocytosis, but not binding, may be inhibited [28]. Thus, if endocytosis of TSST-I was important for PAEC cytotoxicity, then the lowered incubation temperature might prevent cytotoxicity. At 0.5 h of incubation, 5 and 10 J.Lg of TSST-I/well incubated at 37°C showed 19.4% and 34.2% 5tCr release, respectively (figure 3). However, when the incubation temperature was lowered to 4°C, only 2.8% and 3.9% 5tCr release were observed. Similarly, at 1 h of incubation at 37°C, 28.0% and 45.2% 5tCr release were seen with 5 and 10 J.Lg of TSST -I, respectively. With a reduced incubation temperature of 4°C, only 3.1%5tCr was released for 5 J.Lg ofTSST-1 and 6.4% for 10 f-Lg ofTSST-1. Calcium dependence of PAEC cytotoxicity. Previously,

50 Q) l/)

40

co Q) Q)

c:

5 J.lQ@37"C 10J.lQ @37"C 5J,JQ@4°C 10J.lQ@4OC

30

~

9 ~ 0

20 10 0 0.5

Hours Figure 3. Temperature dependence of toxic shock syndrome toxin I (TSST-I) cytotoxicity of porcine aortic endothelial cells (PAEC): Confluent PAEC monolayers were exposed to TSST-l (5 and 10 JLgjwell) at 37 or 4°C for 0.5 and I h. Percentage of 51Cr release was determined. Results are shown as mean ± SE of quadruplicate samples.

Downloaded from http://jid.oxfordjournals.org/ at New York University on May 26, 2015

Figure 2. Dose dependence of toxic shock syndrome toxin I (TSST-I) cytotoxicity of porcine aortic endothelial cells (PAEC): Confluent PAEC monolayers were exposed to different concentrations (0.1-100 ~g/well) of TSST-I for 2 h at 37°C. At end of incubation, reaction solution was harvested and percentage of 51Cr release was calculated. Results are shown as mean ± SE of quadruplicate samples in duplicate experiments.

lID 1991: I64 (October)

Capillary Leak in Toxic Shock Syndrome

JID 1991; 164 (October)

Table 1. Inhibition of toxic shock syndrome toxin I (TSST-I) cytotoxicity of porcine aortic endothelial cells using chloroquine, methylamine, EGTA, and calcium- and magnesium-free Hanks' balanced salt solution (HBSS).

Treatment TSST-I only (I 0 ~g/well) TSST-I + Chloroquine (250 ~M) + Methylamine (50 mM) + EGTA(4 mM) + Calcium- and magnesium-free HBSS

%51C r release ± SE*

% inhibition of cytotoxicity!

53.0 ± 0.2 18.5 ± 0.4 21.0 ± 0.3 27.1 ± 0.6

65.1 60.4 48.9

12.8± 1.8

75.6

* Percentage of 51Cr release was determined as described in Methods and presented as mean ± SE of quadruplicate samples. t Percentage of inhibition of cytotoxicity was calculated relative to cytotoxicity due to TSST-l alone (counts per minutes [cpm] ofTSST-1 with treatrnent/cpm of TSST-I alone X 100%).

50

fa Q)

40

ca Q) Qi

30

en

a:

U 0~

AEROBIC fj ANAEROBIC

20 10 0 5

10

TSST·1

(}Jg)/Well

Figure 4. Inhibition of toxic shock syndrome toxin I (TSST -1 ) cytotoxicity of porcine aortic endothelial cells (PAEC) under anaerobic conditions: Confluent PAEC monolayers were exposed to TSST-I (5 and 10 Jlg/well) and incubated for I h at 37°C under aerobic or anaerobic conditions. All reagents necessary for the anaerobic assay were preincubated in an anaerohic cham her. Percentage of 51Cr release was determined. Results are shown as mean ± SE of quadruplicated samples.

the cytotoxicity assay as possible. When the PAEC cytotoxicity assay was done under normal aerobic conditions (95% air, 5% CO 2), 28.5% and 48.9% 51Cr releases were observed for 5 and 10 J.Lg of TSST-1, respectively. However, under anaerobic conditions (85% N 2, 10% CO 2, 5% H 2) at 37°C, 0 and 3.9% 51 Cr release were seen for 5 and 10 J.Lg ofTSST -1, respectively (figure 4). Binding assay and analysis. TSST -1 was effectively labeled with 125} without detectable physical degradation of the molecule. The radiolabeled toxin displayed identical molecular weight as native protein by SDS-PAGE. 125}_TSST-l was immunologically and biologically reactive, with activity comparable to equal amounts of unlabeled TSST-l. 125 1_ TSST-l reacted with antibodies to TSST-l in Western blot analysis, was mitogenic, and caused proliferation of mouse lymphocytes to the level with unlabeled toxin. A competitive binding assay was done to characterize TSST-l binding to PAEC. Binding of 125}-TSST-l (0.2 J.Lg) to PAEC was inhibited in a dose-dependent manner by increasing amounts (0. 1-500 J.Lg) of unlabeled TSST -1 (figure 5). Nearly 50% inhibition of 1251_ TSST-l binding was observed with 5 J.Lg of unlabeled TSST -1, and near maximal inhibition occurred with 100 J.Lg of unlabeled TSST -1. In contrast, increased concentrations (1-500 J.Lg) of unlabeled TSST-ovine did not inhibit 125 1_ TSST-l binding to PAEC (figure 5). When analyzed according to the method of Scatchard [24], data from labeled and unlabeled TSST-l competitive binding experiments best fit a linear relationship (r = 0.87), a result suggesting a single class ofTSST -1 receptors. The calculated dissociation constant (Kd ) was 5.7 X 10- 7 M, and the number of receptors binding TSST -1 was 2.2 X 10 4 / cell.

Downloaded from http://jid.oxfordjournals.org/ at New York University on May 26, 2015

Ca2+ has been shown to be important for TSST-I-mediated cytotoxicity of rat renal tubular epithelial cells [29] and for endocytosis [30]. Whether Ca 2+ is necessary for TSST-l-mediated PAEC cytotoxicity was addressed. Adding EGTA, a Ca 2+-chelator, to the monolayers before incubation with 10 J.Lg ofTSST-l resulted in only 27.1% 51Cr release compared with 53% for TSST-l alone (table 1). This represented a 48.9% inhibition of cytotoxicity. Use of Ca/Mg-free HBSS had an even more dramatic inhibitory effect, as only 12.8% of 51Cr was released, which represented a 75.6% inhibition of cytotoxicity (table 1). With the readdition of'Ca?" to Ca/Mgfree HBSS reaction solution, the level of cytotoxicity returned to that obtained with TSST -1 alone in normal HBSS. Effects ofchloroquine and methylamine on PAEC cytotoxicity. Chloroquine and methylamine inhibit receptor-mediated endocytosis [29, 30]. Preincubation of PAEC with chloroquine before incubation with 10 J.Lg ofTSST -1 resulted in 18.5% 51Cr release, which was 65.1 % inhibition of cytotoxicity, compared with 53% 51Cr release by PAEC treated with TSST -1 alone (table 1). Methylamine had a similar inhibitory effect, as only 21.0% 51Crrelease (60.4% inhibition) was observed when PAEC monolayers were preincubated with methylamine before TSST-1 administration (table 1). Role of oxygen and oxidants in PAEC cytotoxicity. Previous studies examining TSST -1 cytotoxicity of rat renal tubular epithelial cells suggested that oxidants might playa role in the mechanism of cytotoxicity [29]. An antioxidant and iron chelator, deferoxamine, was used in an attempt to inhibit TSST-l cytotoxicity of PAEC. Preincubation of PAEC with deferoxamine ( 10 mM) for 20 min before 10 J.Lg TSST-l treatment resulted in 31.2% ± 0.2% 51Cr release or 41.1 % inhibition of cytotoxicity relative to values obtained from monolayers treated with TSST -1 alone. The role ofoxygen and possibly oxidants was further examined by attempting to eliminate as much of the O 2 present in

715

110 1991;164 (October)

Lee et al.

716 120 C)

c:

100

~

=ij

c:

iii 0

.:

80

6

0

0

~ 0

>-

HI

t

:cas

1.5

Q)

E ...

.c -..

Q)

Q.

60

~

c:

0

2.0

It

!

~

E fJ) c. 0.5 m Ill(

6

40

1.0

0

s

I

..

A

20

It)

N

.a

0.0 0

0 • 8

0

Unlabeled 1551-1 Unlabeled 1551-0

- 7 Log

• 6

• 5

- 4

2

• 3

[Competitor (M)]

unlabeled TSST -lor TSST-ovine (TSST-0): PAEC monolayers were incubated in the presence of 1251- TSST-1 (5 Jlg) and increasing concentrations of unlabeled TSST-I (0.1-500 Jlg) or unlabeled TSST -0 (1-500 Jlg). At end of I h of incubation at 4°C, cells were washed, monolayers were harvested with 1 M NaOH, and bound 1251_TSST-l was measured. Data points are mean specific bound TSST-I ± SE (total counts per minute [cpm] bound - cpm bound in presence of 500 Jlg of unlabeled TSST-1) of quadruplicate samples.

Transendothelial membrane permeability assay. A transendothelial membrane permeability model was used to examine effects of TSST-l on PAEC monolayer permeability. BSA was effectively labeled with 1251, as the radiolabeled BSA displayed identical molecular weight as native protein on SDS-PAGE. In a dose-dependence study, increased 1251_ BSA permeability through Transwell membranes correlated with the amount of TSST-l added to the luminal chamber (figure 6). Increased permeability of 1251-BSA (0.27 ± 0.03 pmol) through monolayers was seen at 0.1 J.Lg ofTSST-I/luminal chamber and further rose to 2.34 ± 0.11 pmol of 125 1_ BSA at 10 J.Lg of TSST-ljchamber. All values were subtracted for nonspecific permeability (0 J.Lg ofTSST-1) ofO. 73 ± 0.03 pmol of 1251-BSA. Time-course studies showed PAEC monolayer permeability of 0.71 ± 0.10 pmol of 1251-BSA after 45 min of incubation with 5 J.Lg ofTSST-I/chamber (figure 7). Near maximal permeability of 3.09 ± 0.18 pmol of 1251-BSA was demonstrated after 135 min of incubation. Nonspecific permeabilities (0 J.Lg ofTSST-I) of 1251-BSA at 45, 75, 135,and 195min were 0.49, 0.69, 1.49, and 1.84 pmol, respectively.

Discussion Recently, we provided evidence consistent with capillary leak being responsible for hypotension and subsequent shock in a TSS model system [11], but the mechanism of how pyrogenic toxins cause capillary leak is not understood.

4

5

6

7

8

9

10

(~g)/Chamber

Figure 6. Toxic shock syndrome toxin 1 (TSST-l) dose dependence and permeability of 1251-labeled bovine serum albumin (' 251_ BSA) through porcine aortic endothelial cell (PAEC)-monolayered Transwells. TSST-l (0.1-10 Jlg/well) and 1251-BSA (5 Jlg/well) in Hanks' balanced salt solution + 1%BSA were added to luminal chamber ofPAEC-monolayered Transwells and incubated for 90 min at 37°C. 1251-BSA samples in subluminal chamber were counted and recorded as permeability in picomoles of 125I-BSA. All counts were subtracted for nonspecific permeability of 1251-BSA (0 Jlg ofTSST-1 in luminal chamber). Results are given as mean ± SE of quadruplicated samples in duplicate experiments.

Tumor necrosis factor (TNF)-a may be released after pyrogenic toxin stimulation of monocytes [31, 32]. TNF-a has been implicated in endotoxin shock [33, 34] and may cause hypotension through its effects on the endothelium and inflammatory cells to mediate vascular leak [35]. TSST -I previously has been shown to cause sustained release of TNF in

4 ~

.a as Q) E ...

3

Q)

Q.

0

Ill(

E c.

2

fJ)

m I

II> N

0 0

50

100

150

200

Minutes Figure 7. Time course of toxic shock syndrome toxin I (TSST1)-induced permeability of '251-labeled bovine serum albumin (' 251-BSA) through porcine aortic endothelial cell (PAEC)-monolayered Transwells. TSST-l (5 Jlg/well) and 1251-BSA (5 Jlg/well) in Hanks' balanced salt solution + 1% BSA were added to luminal chamber and incubated for up to 195 min at 37°C. At each time point, all counts were adjusted for nonspecific permeability of 1251_ BSA (0 Jlg ofTSST-l). Results are given as mean ± SE of quadruplicate samples.

Downloaded from http://jid.oxfordjournals.org/ at New York University on May 26, 2015

Figure 5. Inhibition of '251-labeled toxic shock syndrome toxin 1 (' 251-TSST-I) binding to porcine aortic endothelial cells (PAEC) by

3

TSST-1

JID 1991; 164 (October)

Capillary Leak in Toxic Shock Syndrome

different organs may respond differently to TSST-I, much like the varied responses seen with endotoxin. Fourth, only a fraction of the TSST-l present in the assay may have been active or cytotoxic. Although pyrogenic toxins are regarded as being relatively resistant to denaturation, the toxin purification procedure involves many steps, including freezing and lyophilization. Indeed, in the cytotoxicity assay and other assays in our laboratory, some variation has been noted among different toxin batches. Finally, previous studies showed that pyrogenic toxins localize in certain organs of the body [22]; because of this, surprisingly high concentrations of the toxin may be present in specific organs. Thus, the toxin amount per cell in those affected organs may approach concentrations seen in this study. The mechanism of TSST-L-mediated PAEC cytotoxicity was examined. Some bacterial toxins, such as hemolysins, form pores in eukaryotic cellular membranes [27], leading to osmotic death of cells. Our results suggest that TSST-1 may not belong to this family of pore-forming bacterial toxins. Because cytotoxicity is inhibited by chloroquine, methylamine, incubation at 4°C, and the lack of Ca?", we propose that TSST-1 first binds to a specific receptor on the endothelial cell membrane and becomes internalized, possibly in a manner similar to receptor-mediated endocytosis. This is consistent with the report of Kushnaryov et al. [12], who showed by electron microscopy that TSST-1 is internalized in a receptor-mediated endocytosis-like manner by human umbilical vein endothelial cells. After internalization, the toxin may produce its cytopathic effects by the generation of oxidants, as suggested by anaerobic conditions and deferoxamine inhibition of cytotoxicity. This is consistent with the observation of Keane et al. [29] that TSST-l cytotoxicity of rat renal tubular cell may be due to oxidant damage. To investigate further the possibility that TSST-1 binds specifically to membrane receptors on PAEC, a competitive binding assay was done. The results suggest that TSST-l binds to a single class of receptors on PAEC with K d of 5.7 X 10- 7 M, and each PAEC has 2.2 X 104 receptors/cell. These values are close to previously reported results from studies of TSST-l binding to immune cells [40]. The identity of the receptor on endothelial cells is unknown but is likely different from the binding site for TSST-l found on T cells. Kushnaryov et al. [12] showed that radiolabe1ed TSST-l binding to endothelial cells, using whole human umbilical vein pre parations, had a «, of 6.5 X 10- IO M. This higher affinity may be attributed to several factors. First, the difference between the affinities may be due to experimental methods. Our binding study was done using isolated and cultured endothelial cell preparations, whereas Kushnaryov's study used whole umbilical veins. The variability may have occurred with the presence of other cell types or from incomplete washing of excess radiolabeled toxin. Second, the receptor for TSST -Lon PAEC may differ in affinity from

Downloaded from http://jid.oxfordjournals.org/ at New York University on May 26, 2015

vitro [31, 32], which may contribute to vascular leak as well. In addition, pyrogenic toxins may directly affect and possibly be lethal to the endothelium to alter capillary permeability. Kushnaryov et al. [12] have shown that TSST-I binds specifically to human umbilical cord vein endothelium and becomes internalized in a receptor-mediated endocytosis-like manner. In addition, Keane et al. [29] have reported that TSST-1 potentiates the cytotoxic effects of endotoxin on renal tubular epithelial cells. In this study, we showed that TSST-1 was directly cytotoxic to PAEC monolayers, consistent with the latter hypothesis. Using the 5lCr-release assay, this cytotoxic effect occurred rapidly and was dependent on the concentration of TSST-1 added to monolayers. PAEC were used mainly because of their availability and also because the porcine physiology is similar to that of humans. Preliminary experiments with cultured human umbilical vein endothelial cells also have shown a cytotoxic effect by TSST-l. To ensure that TSST-1 (and not possible minor contaminants of the TSST -1 preparation) was the cytotoxic agent, several experiments were done. First, anti-TSST-l serum successfully inhibited the cytotoxic effect of TSST -1. Second, the cytotoxicity was specific for TSST-1 and not for biologically inactive TSST-ovine, which appeared incapable of competing with TSST-1 for endothelial cell receptors. Third, we ensured that staphylococcal a-hemolysin did not contaminate the TSST-1 preparation. Anti-a-hemolysin serum had no effect on cytotoxicity. Furthermore, previous experiments have shown that menstrually related TSST-lproducing S. aureus strains typically do not make a-hemolysin [36], including the strain used in the current study as the source of TSST-l. Although endotoxin may enhance the severity of TSS through synergism with TSST-l [11, 37, 38], TSST-I and endotoxin failed to exhibit synergism in the cytotoxicity of PAEC. Previous studies showed variable cytotoxic responses to endotoxin by endothelial cells cultured from different species and organs [39]. In this study, PAEC monolayers appeared to be resistant to cytotoxic effect ofLPS from S. typhimurium. It is possible, however, that the effects of TSST-l and endotoxin on endothelial cells isolated from other species or organs may be different. The PAEC cytotoxicity assay required at least 1 I-Lg of TSST-l/well. Although this seems to be a large amount of toxin, several explanations can be proposed. First, the reaction volume of 0.5 ml/well was chosen as a matter ofconvenience. Smaller reaction volumes and thus lower concentrations of toxin per cell may be sufficient to cause cytotoxicity. Second, as evidenced from the transendothelial membrane model, lower concentrations of TSST-l may not cause SICr release but may still cause capillary leakage, possibly by causing retraction of the endothelium and thus increasing the spaces between the junctions of individual cells. Third, endothelial cells isolated from different species and possibly

717

718

Lee et al.

10. II.

12.

13.

14.

15. 16.

17.

18.

Acknowledgments

We thank Mary Hebert and Dong Tuong for technical help, and Leo Furcht for providing human fibronectin.

19.

20.

References

21.

I. Todd J, Fishaut M, Kapral F, Welch T. Toxic-shock syndrome asso-

ciated with phage-group-I staphylococci. Lancet 1978;2: 1116-8. 2. Davis JP, Osterholm MT, Helms CM, et al. Tri-state toxic shock syndrome study: evaluation of case definition and prevention of recurrence. Ann Intern Med 1980;303: 1429-35. 3. Shands KN, Schmid GP, Dan BB, et al. Toxic shock syndrome in menstruating women: its association with tampon use and Staphylococcus aureus and the clinical features in 52 cases. N Engl J Med 1980;303: 1436-42. 4. Schlievert PM, Shands KN, Dan BB, Schmid GP, Nishimura RD. Identification and characterization of an exotoxin from Staphylococcus aureus associated with toxic-shock syndrome. J Infect Dis 1981; 143:509-16. 5. Bergdoll MS, Cass BA, Reiser RF, Robbins RN, Davis JP. A new staphylococcal enterotoxin, enterotoxin F, associated with toxic-shock syndrome Staphylococcus aureus isolates. Lancet 1981; I:1017-21. 6. Crass BA, Bergdoll MS. Toxin involvement in toxic shock syndrome. J Infect Dis 1986;153:918-26. 7. Schlievert PM. Staphylococcal enterotoxin B and toxic-shock syndrome toxin-I are significantly associated with nonmenstrual TSS. Lancet 1986;1:1149-50. 8. Rizkallah MF, Tolaymat A, Martinez JS, Schlievert PM, Ayoub EM. Toxic shock syndrome caused by a strain of Staphylococcus aureus that produces enterotoxin C but not toxic shock syndrome toxin-I. Am J Dis Child 1989;143:848-9. 9. Bohach GA, Fast DJ, Nelson RD, Schlievert PM. Staphylococcal and

22.

23.

24. 25.

26.

27.

28. 29.

30.

31.

streptococcal pyrogenic toxins involved in toxic shock syndrome and related illnesses. Crit Rev Microbiol 1990; 17:251-72. Chesney PJ. Clinical aspects and spectrum of illness of toxic shock syndrome: overview. Rev Infect Dis 1989;II(suppl):SI-7. Lee PK, Deringer JR, Krieswirth BN, Novick RP, Schlievert PM. Fluid replacement protection of rabbits challenged subcutaneously with toxic shock syndrome toxins. Infect Immun 1991;59:879-84. Kushnaryov VM, MacDonald HS, Reiser RF, Bergdoll MS. Reaction of toxic shock syndrome toxin I with endothelium of human umbilical cord vein. Rev Infect Dis 1989; II(suppl):S282-8. Blomster-Hautamaa DA, Kreiswirth BN, Novick RP, Schlievert PM. Resolution of highly purified toxic-shock syndrome toxin I into two distinct proteins by isoelectric focusing. Biochemistry 1986;25:54-9. Schlievert PM, Blomster DA. Production of staphylococcal pyrogenic exotoxin type C: influence of physical and chemical factors. J Infect Dis 1983; 147:236-42. Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature (London) 1970;227:680-5. Ryan US, Maxwell G. Isolation, culture, and subculture of bovine pulmonary artery endothelial cells: mechanical methods. J Tissue Cult Methods 1986;10:3-5. Nagelkerke J, Barto KP, van Berkel TJC. In vivo and in vitro uptake and degradation of acetylated low density lipoprotein by rat liver endothelial, Kupffer, and parenchymal cells. J Bioi Chern 1983; 258:12221-7. Vercellotti GM, Yin HQ, Gustafson KS, Nelson RD, Jacob HS. Platelet-activating factor primes neutrophil responses to agonists: role in promoting neutrophil-mediated endothelial damage. Blood 1988; 71:1100-7. Mishell BB, Shiigi SM, Henry C, et al. Preparation of mouse cell suspensions. In: Mishell BB, Shiigi SM, eds. Selected methods in cellular immunology. San Francisco: WH Freeman, 1980:16-7. Westphal 0, Luderitz 0, Bister F. Uber die Extraktion von Bacterien mit Phenol/Wasser. Z Naturforsch 1952;7b: 148-55. Iodination of proteins: chloramine T methods. In: Hudson L, Hay FC, eds. Practical immunology. 2nd ed. Oxford, UK: Blackwell Scientific Publications, 1980:240-1. Poindexter NJ, Schlievert PM. Binding oftoxic-shock-syndrome toxinI to human peripheral blood mononuclear cells. J Infect Dis 1987; 156: 122-9. Blake M, Johnston K, Russell-Jones G, Gotschlich E. A rapid, sensitive method for detection of alkaline phosphatase-conjugated anti-antibody on Western blots. Anal Biochem 1984; 136: 175-9. Scatchard G. The attraction of proteins for small molecules and ions. Ann NY Acad Sci 1949;51660-72. Casnocha SA, Eskin SG, Hall ER, McIntire LV. Permeability of human endothelial monolayers: effect of vasoactive agonists and cAMP. J Appl Physiol 1989;67: 1997-2005. Goldblum SE, Sun WL. Tumor necrosis factor-alpha augments pulmonary arterial transendothelial albumin flux in vitro. Am J Physiol 1990;258: 157-67. Novick RP. Staphylococci. In: Davis BD, Dulbecco R, Eisen HN, Ginsberg HS, eds. Microbiology. 4th ed. Philadelphia: JB Lippincott, 1990:539-50. Gruenberg J, Howell KE. Membrane traffic in endocytosis: insights from cell-free assays. Annu Rev Cell Bioi 1989;5:453-81. Keane WF, Gekker G, Schlievert PM, Peterson PK. Enhancement of endotoxin-induced isolated renal tubular cell injury by toxic shock syndrome toxin I. Am 1 Pathol 1986; 122: 169-76. FitzGerald D, Morris RE, Saelinger CB. Receptor-mediated internalization of pseudomonas toxin by mouse fibroblasts. Cell 1980;21 :86773. Fast DJ, Schlievert PM, Nelson RD. Nonpurulent response to toxic

Downloaded from http://jid.oxfordjournals.org/ at New York University on May 26, 2015

that of human umbilical vein endothelial cell and may even correlate with the relative susceptibility of various species to the pyrogenic and lethal activities of TSST-1. Finally, the transendothelial membrane permeability model was used to demonstrate that TSST-l causes 125I-BSA leakage through PAEC monolayers, possibly mimicking the mechanism of vascular leakage in TSS. In both dose-dependence and time-course studies, maximal toxin effects in permeability assays were similar to amounts needed in cytotoxicity assays: maximal 51Cr release and 125I-BSA permeability were observed at > 5 J.Lg ofTSST -1 jwell and at ,...." 2 h of incubation for dose-dependence and time-course assays, respectively. No 51Crrelease from the cytotoxicity assay was noted at doses < 1 J.Lg of TSST-ljwell. However, in the Transwell study, some permeability of 125I-BSA was seen at 0.1 J.Lg of TSST-1 jluminal chamber. These data suggest that much smaller concentrations of TSST -1 may be sufficient in vivo to cause some vascular leakage without widespread cytotoxicity to vascular endothelium.

lID 1991; 164 (October)

lID 1991;164 (October)

32.

33.

34.

35.

36.

Capillary Leak in Toxic Shock Syndrome

shock syndrome toxin I-producing Staphylococcus aureus: relationship to toxin-stimulated production of tumor necrosis factor. J Immunol 1988; 140:949-53. Parson net J, Gillis ZA. Production of tumor necrosis factor by human monocytes in response to toxic-shock-syndrome toxin-I: J Infect Dis 1988; 158: 1026-33. Beutler B, Milsark IW, Cerami AC Passive immunization against cachectin/turnor necrosis factor protects mice from lethal effect of endotoxin. Science 1985;229:869-71. Mathison JC Wolfson E, Ulevitch RJ. Participation of tumor necrosis factor in the mediation of gram negative bacterial lipopolysaccharide-induced injury in rabbits. J Clin Invest 1988;81: I925-37. Sato N, Goto T, Haranaka K, et al. Actions of tumor necrosis factor on cultured vascular endothelial cells: morphologic modulation, growth inhibition. and cytotoxicity. J Natl Cancer Inst 1986;76:1113-21. Schlievert PM, Osterholm MT, Kelly JA, Nishimura RD. Toxin and

37.

38.

39.

40.

719

enzyme characterization of Staphylococcus aureus isolates from patients with and without toxic-shock syndrome. Ann Intern Med 1982;96:937-40. Schwab JH, Watson OW, Cromartie WJ. Further studies of group A streptococcal factors with lethal and cardiotoxic properties. J Infect Dis 1955;96: 14-8. Schlievert PM. Enhancement of host susceptibility of lethal endotoxin shock by staphylococcal pyrogenic exotoxin type C Infect Immun 1982;36: 123-8. Harlan JM, Harker LA, Striker GE, Weaver LJ. Effects oflipopoIysaccharide on human endothelial cells in culture. Thromb Res 1983;29: 15-29. See RH, Krystal G, Chow AW. Binding competition of toxic shock syndrome toxin I and other staphylococcal exoproteins for receptors on human peripheral blood mononuclear cells. Infect Immun 1990;58:2392-6.

Downloaded from http://jid.oxfordjournals.org/ at New York University on May 26, 2015