1001 Penicillin G-Induced Microbicidal Activity of Endothelial Cells Cultured on Gelfoam Blocks Bin Zhang, Michelle Centra, Guan Liang Cao, Robert M. Taylor, Roman E. Ratych, and Gerald M. Rosen Department of Pharmaceutical Sciences, Program in Pharmacology and Toxicology, University of Maryland School of Pharmacy; Research Service, VA Medical Center, and Department of Pathology and Dermatology, Johns Hopkins Medical Institutions, Baltimore, Maryland Neutrophils, monocytes, and macrophages are a group of leukocytes that are recognized to have a pivotal role in the control of bacterial and fungal infections by phagocytosing and killing these organisms. In contrast, endothelial cells are viewed as passive members of the immune system, acting as barriers to the influx of microbes and secreting factors, which bring professional phagocytes to sites of bacterial invasion. In recent years, however, a body of evidence has surfaced documenting the ability of endothelial cells to phagocytose bacteria [1-7]. However, unlike neutrophils, endothelial cells in these models did not kill bacteria [l, 3, 4, 6, 7]. Given the fact that endothelial cells may share a common genetic linkage with macrophages [8, 9], we questioned whether endothelial cells may, under certain conditions, have the ability to kill bacteria. The endothelial cell is unique in that its luminal surface is in contact with blood, while the other side is intimately associated with a subendothelial matrix containing collagen and laminin [10, 11]. This specific environment has spawned considerable efforts to develop in vitro matrices attempting to mimic the in vivo situation [10, 12]. Of particular interest is the study of Thompson et al. [13], in which the site-directed angiogenic properties of the three-dimensional matrix, Ge1foam, was well documented. The ability of Gelfoam to induce in vivo neovessel formation [13] suggested that this substrate might be an Received 30 October 1995; revised 19 June 1996. Grant support: NIH (HL-33550); Department of Veterans Affairs Research Service; Council for Tobacco Research-USA. Reprints or correspondence: Dr. Gerald M. Rosen, Dept. of Pharmaceutical Sciences, Program in Pharmacology and Toxicology, University of Maryland School ofPhannacy, 20 N. Pine St., Baltimore, MD 21201. The Journal ofInfectious Diseases 1996;174:1001-9 © 1996 by The University of Chicago. All rights reserved. 02-189!6745~$O 1.00 ideal matrix for in vitro culturing of endothelial cells. With this in mind, we demonstrated that endothelial cells grown on Gelfoam simulated the morphology ofthe endothelium in small vessels and capillaries [14], which was not observed when these cells were cultured on monolayer matrices [3, 4]. Staphylococcus aureus is a virulent gram-positive organism, causing infections in a variety of tissues [2]. Considering the reported phagocytosis of S. aureus by endothelial cells [1, 6, 7] and similarities of endothelial cells cultured on Ge1foam with their in vivo morphology [13, 14], we explored the possibility that endothelial cells might play a more active role in host immune defense than previously reported [1-7]. Material and Methods Chemicals. Medium Ml99, Hanks' balanced salt solution (HBSS), penicillin-streptomycin, Fungizone, L-glutamine, sodium hydrogen carbonate, and glucose were purchased from Life Technologies GIBCD BRL (Gaithersburg, MD). Collagenase type II was obtained from Worthington Biochemical (Freehold, NJ). Fetal calf serum was purchased from Hyclone (Logan, UT). Gelfoam was obtained from Upjohn (Kalamazoo, MI). Cytochalasins Band o and lysostaphin were obtained from Sigma (St. Louis). Benzyl C4 C]penicilIin potassium (54 rnCi/rnmol) was purchased from Amersham (Amersham, UK). All other reagents were purchased from commercial sources. S. aureus strain and culturing. S. aureus ATCC 292I3 and penicillin-resistant S. aureus ATCC 13301 (American Type Culture Collection, Rockville, MD) colonies were collected from 5% sheep's blood agar plates and grown to mid-logarithmic phase in 10 mL of trypticase soy broth at 37°C for 2 h. The organisms were harvested by centrifugation at 1000 g for 8 min and washed twice with O. 1% gelatin-PBS at pH 7.4. Bacterial concentrations were estimated spectrophotometrically and confirmed by plating on agar plates and counting colony-forming units. The susceptibility of S. aureus ATCC 29213 to penicillin G was tested during a 2-h incuba- Downloaded from https://academic.oup.com/jid/article/174/5/1001/804957 by guest on 29 September 2023 A body of evidence has surfaced documenting the ability of endothelial cells cultured on monolayers to phagocytose but not kill bacteria. Several years ago, a new three-dimensional endothelial cell culturing model was developed, which simulated the morphology of the endothelium in small vessels and capillaries. Given that endothelial cells may be derived from the same pluripotent stem cells as macrophages, the question of whether endothelial cells might phagocytose and kill bacteria was explored. Endothelial cells grown on Gelfoam blocks exhibited bactericidal activity towards Staphylococcus aureus, reaching maximal killing of >90% after 2 h. Evidence documents the involvementof bacterial adherence to the plasma membrane of the endothelial cell. This is followed by phagocytosisof S. aureus, leading to intracellular killing.Penicillin G, included in the endothelial cell growth medium, was found to be a critical factor in the bactericidal activity demonstrated by Gelfoam blocks laden with endothelial cells. 1002 Zhang et al. extracellular and adherent S. aureus. In each case, there were no colony-forming units of S. aureus from the lysostaphin-containing supernatant. The Gelfoam endothelial cell cultures were washed three times with 4 mL of HBSS per washing. Sterile 0.1% gelatin/ water was added to lyse the endothelial cells. The resultant bacterial suspension was vortexed, serially diluted, and plated on tryptic soy agar. The plates were incubated for 18 h in a dry incubator at 37°C, and the total number of viable S. aureus was quantitated by counting the colony-forming units. To evaluate the role of actin microfilaments in the killing of S. aureus by endothelial cells cultured on Gelfoam blocks, cytochalasin B (10 j.tg/mL) was added to endothelial cells 1 h before addition of bacteria and continued during the times of incubation. At specific time points, sterile 0.1 % gelatin/water was added to lyse Gelfoam-contained endothelial cells. The sample was diluted and plated on agar as described above. For some experiments with endothelial cells cultured on Gelfoam in M199 without antibiotics, either cytochalasin B (10 j.tg/mL) or cytochalasin D (10 j.tg/mL) was included in the medium, 1 h before the addition of S. aureus. Benzyl [14C}penicillin potassium-tracing study. Cultured M199, containing penicillin G (200 U/mL, 340 j.tM), was spiked with benzyl C4 C]penicillin potassium (0.06 j.tCi/rnL, 1.2 j.tM), and 2 mL was added to either Ge1foam laden with endothelial cells or Ge1foam blocks alone. Every 3 days, this medium was removed and replaced with fresh medium (2 rnL). On day 10, medium was removed, and Gelfoam blocks were washed as described above and dissolved with collagenase (0.63 mg/mL). Samples (0.5 mL) of medium from day 10 and of each of the four washings and of supernatant of dissolved Gelfoam blocks were transferred to glass vials. Endothelialcells were collected by filtering through a cell harvester (Brandel M-24; Life Technologies) and washed three times with buffer before filters were transferred to glass vials. The counting solution containing toluenePPO-POPOP was added to each vial and then placed in a liquid scintillation counter (LS 6800; Beckman, Brea, CA). Radioactive counts were expressed as disintegrations per minute. Transmission electron microscopy. Transmission electron micrographs of Gelfoam-contained endothelial cell cultures with S. aureus were made as follows. At l-h intervals, the reaction medium was decanted and the Gelfoam blocks containing the endothelial cells were immediately placed in cold 2.5% glutaraldehyde/a. 1 M Sorenson's phosphate buffer, pH 7.4, fixed for 3 h at 4°C, rinsed, and sectioned into 1-mm blocks [16]. Specimens were then postfixed in 1% osmic acid for 1.5 h, embedded in Spurr's plastic embedding compound, sectioned at 1 j.tm thickness, stained with toluidine blue, and examined by light microscopy to determine representative sites for ultrastructural analysis. Tissue blocks were then ultrathin-sectioned (60-90 nm) on an ultramicrotome (MT 5000; DuPont, Wilmington, DE), placed on 200-mesh copper grids, stained with uranyl acetate and lead citrate, and examined with a transmission electron microscope (Zeiss EM lOA) [17]. Statistical handling of data. Data for the Gelfoam and monolayer growth curves are the internal controls of bacterial growth taken from the experiments described above. Gelfoam and monolayer control incubations of S. aureus and medium were done simultaneously with their respective bacteria, medium, and endothelial cell culture incubations. Data are expressed as the mean ::!:: SD of three to five separate experiments with three to five replicate incubations per time interval. Downloaded from https://academic.oup.com/jid/article/174/5/1001/804957 by guest on 29 September 2023 tion using a dose range from a to 0.5 U/mL in antibiotic-free and serum-free M199. A 90% inhibition of bacterial growth at 2 h was observed at 0.2 U/mL penicillin G. Isolation and culturing of bovine pulmonary artery endothelial cells. Endothelial cells were isolated from bovine pulmonary arteries by brief collagenase digestion, grown in T-75 polystyrene flasks in the presence of antibiotics (200 U/mL penicillin G and 200 j.tg/mL streptomycin), and subcultured as described by Centra et al. [14]. All cultures used in this study were second to fifth passage. Confirmation of endothelial cell identity was obtained by staining for factor VIn antigen, as described by Jaffe et al. [15], and visual inspection of their typical morphology [3, 4]. For endothelial cells grown on monolayers, cell cultures (5 X 105 cells/ mL in 2 mL of M199 containing penicillin G, 200 U/mL, and streptomycin, 200 j.tg/mL) were seeded onto 3.8-cm 2 petri dishes until confluence was reached, usually within 3-4 days [14]. For endothelial cells grown on Gelfoam blocks, cells from T-75 polystyrene flasks (5 X 105 cells/rnL) were added to a precut 24 X 8 X 2 rnm autoclaved block of Gelfoam, previously placed in a 17 X 100 rnm polypropylene tube to a total volume of 2 rnL of M199. Four separate culturing conditions were then used: M199 containing penicillin G, 200 U/rnL, and streptomycin, 200 j.tg/mL; Ml99 without antibiotics;M 199containing only penicillin G (200 U/mL); and M199 containing only streptomycin (200 j.tg/rnL). For each of these conditions, the endothelial cells were allowed to grow undisturbed for several days, and thereafter appropriate media for each experimental design were changed frequently for a total culturing period of 10 days as described [14]. Cell-free controls were subjected to the identical culturing conditions used with their respective Gelfoam or monolayer cell cultures. At various time points throughout the different experiments, the viability of endothelial cells was closely monitored, using trypan blue dye exclusion. For up to 2 h of incubation, no endothelial cell death was observed. Microbicidal assay. Before the introduction of bacteria to endothelial cells, these eukaryotic cells, grown either on Gelfoam blocks (1 X 106 cells/block) or in monolayer cultures (5 X 105 cells/well), were extensively washed four times with 4 mL/washing of HBSS to remove antibiotics. At this point, the endothelial cell cultures were placed in 2 mL of an antibiotic-free and serum-free modified M199 buffer, containing sodium hydrogen carbonate (24 roM) and HEPES buffer (20 mM) to maintain pH 7.4. Gelfoam blocks, in the absence of endothelial cells, washed as for the Gelfoam laden with endothelial cells as described above, were used as controls. Bacteria were then added to endothelial cell cultures in a 10:1 endothelial cell-to-bacterium ratio. This dual cell system was incubated at 37°C, with a gas mixture of 95% air/5% CO2 , At defined time intervals, 0.1% gelatin-Hjf) was added to the cultures to lyse endothelial cells. The resultant bacterial suspension was vortexed, serially diluted, and plated on tryptic soy agar. The plates were incubated for 18 h in a dry incubator at 37°C, and the total number of viable S. aureus was quantitated by counting the colony-forming units. To determine the number of viable internalized bacteria, lysostaphin (0.5 j.tg/mL) was added to the samples at various time points after endothelial cell/bacteria incubation to kill extracellular and adherent S. aureus, as described by Hamill et al. [1]. Lysostaphin-containing samples were incubated at room temperature for 5 min. A portion ofthe medium from each sample was collected and plated to determine the efficiency of lysostaphin killing of JID 1996; 174 (November) Microbicidal Activity of Endothelial Cells JID 1996; 174 (November) 150 en '"~ '" • en III :cell 'S G IBI c ell . ,.. 100 0 ~ '0 "III 50 .c E c '" o 1.5 2 tim e (hour s) Figure 1. No. of viable S. aureus after incubationwith bovine pulmonary artery endothelial cells cultured on Gelfoam blocks and treated with antibiotic-containing medium 199 for 10 days. Bars reflect mean z SD of6 separateexperimentswith 3-5 replicateincubations per time interval. Ratio of endothelial cells to S. aureus in endothelialcellcultureswas 10:I. G = Gelfoam blockswithoutendothelial cells; C = Gelfoam blocks laden with endothelial cells. units of group B streptococci versus vortex treatment of these Iysates. Although our findings suggest that endothelial cells kill S. aureus, experiments were designed to demonstrate that the observed microbicidal activity was not due to antibiotics present in the growth medium. After M199 containing penicillin G and streptomycin was removed from Gelfoam containing endothelial cells, M 199 without antibiotics (4 mL) was added to this matrix. The matrix was then gently shaken. After 30 min, the medium was removed and the washing procedure was repeated three more times with 4 mL of M199 without antibiotics. The efficiency of the washing procedure was documented by observing increased S. aureus growth in each sequential washing, achieving control bacterial growth by the fourth wash (data not shown). This finding demonstrates that the final medium wash no longer possessed the ability to kill S. aureus. Next, we explored the possibility that antibiotics were trapped within this Gelfoam matrix or associated with the endo- 400 • Results When endothelial cells cultured on Gelfoam blocks for 10 days were incubated with nonopsonized S. aureus at an initial ratio of 10: I (endothelial cells to bacteria), we observed bacterial killing compared with Gelfoam alone in the absence of endothelial cells (figure 1). This is in line with similar microbicidal activity exhibited by neutrophils [18]. With time, this microbicidal activity increased, such that >90% bacterial killing was noted at 2 h (figure I) . In concurrence with earlier studies [1, 3, 4, 7], endothelial cells cultured on monolayers did not exhibit bacterial killing, even after 4 h of incubation (figure 2). In fact, a significantly faster S. aureus growth rate was observed with the Gelfoam controls at 1-4 h compared with growth curves for the monolayer controls in the absence of endothelial cells in each model (data not shown). This increased rate of bacterial growth on Gelfoam obviates the fact that this matrix may be contributing to aggregation (pseudokilling). To ensure that bacterial counts after the gel/water dilution represented single bacteria rather than clumps, fractions of the gel/water dilution were sonicated for 20 s on ice and plated on tryptic soy agar. We found no significant differences in counts between the vortex and sonication treatments at 3 h of incuba tion (2.28 X 104 S. aureus cfu for vortexing vs. 1.73 X 104 S. aureus cfu for sonication, n = 10). This demonstrates that less traumatic vortexing alone is sufficient to prevent bacterial clumping. This is consistent with the findings of Gibson et al. [19], who showed that sonication of endothelial cell Iysates did not increase the number of intracellular colony-forming M IIIl Me u:i :c'" := ~ .;OJ z, 300 100 "0 ;;; .0 E :J C o 3 time (hou r) Figure 2. No. of viable S. aureus after incubation with bovine pulmonary artery endothelial cells cultured on monolayers and treated with antibiotic-containing medium 199 for 3 days. Bars reflect mean :!: SD of 4 separate experimentswith 5 replicate incubationsper time interval. Ratio of endothelial cells to S. aureus in endothelial cell cultures was 10: I. M = monolayer without endothelial cells; Me = monolayer with endothelial cells. Downloaded from https://academic.oup.com/jid/article/174/5/1001/804957 by guest on 29 September 2023 o 1003 Zhang et al. 1004 150 til :::l ~ • :::l <1l en ~ .c <1l .:; '0 ~ Q) G II1:I cs 100 ... ... 0 >< 50 .c E :::l C 1.5 time 2 (hours) Figure 3. No. of viable S. aureus after incubation with bovine pulmonary artery endothelial cells cultured on Gelfoam blocks and treated with antibiotic-containing medium 199 for 10 days. G = Gelfoam blocks without endothelial cells; CS = endothelial cell suspension, initially cultured on Gelfoam blocks, then removed before introducing bacteria. Bar G is taken from figure 1. Other bars reflect mean ::t SD of 2 separate experiments with 3 replicate incubations per time interval. Ratio of endothelial cells to S. aureus in endothelial cell cultures was 10: I . thelial cells. We designed studies to explore these possibilities. In a first series of experiments, endothelial cells were separated from Gelfoam and each component tested to determine their role in the killing process. Extensively washed endothelial cell Gelfoam blocks were treated with collagenase (125 tlg /mL) for 10 min at 37°C and centrifuged to separate the cells from the dissolved matrix. Bacteria were then added to the dissolved matrix and incubated for 2 h. Bacterial growth was not inhib ited. The remaining isolated endothelial cells were son icated (model W-IO, power 3.5 and tune 4; Heat Systems Ultrasonics, Plainview, NY) for 30 s on ice and incubated for 2 h with bacteria. This cell lysate did not exhibit any bactericidal activity . To obviate the faint possibility that sonication destroyed the activity of the antibiotics, independent experiments were done in which penicillin G (0.1 U/mL) and streptomycin (0.1 tlg/mL) were sonicated for a comparable time and subsequently incubated with S. aureus. There was no decrease in antibiotic activity. To determine whether the endothelial cell lysate could inactivate penicillin G, this ant ibiotic (0.5 U/mL) was added to the endothelial cell lysate and incubated with S. aureus. The degree of bacterial killing in these experiments was unchanged compared with controls with penicillin G (0.5 U/mL) alone. In contrast, intact endothelial cells, removed from their Gelfoam matrix by coll agenase treatment, still retained their bactericidal activity (figure 3). A last series of experiments was designed to eliminate the possibility that antibiotics were responsible for the observed bacterial killing. This required the culturing of endothelial cells on Gelfoam for 10 days in the absence of penicillin G and streptomycin. To our surprise, the endothelial cells on Gelfoam were no longer able to kill S. aureus (figure 4). This unexpected finding suggested that the antibiotics may be playing a role in the killing process that is not directly related to their specific and well-described pharmacologic activity. This implies that endothelial cells grown on this matrix were activated by treatment with antibiotics, leading to the observed microbicidal activity. Interestingly, we found that the bacterial growth in the presence of Gelfoam treated with 1\1 199 without antibiotics was greater than when this matrix was placed in this medium with the antibiotics. One possible explanation for this phenomenon comes from the work of van den Broek et al. [20]. These authors reported that very low concentrations of penicillin G (0.0085-0.017 U/mL), levels below its bactericidal activity, retarded growth of S. aureu s. In fact , penicillin G, at 0.01 UI mL, inhibited S. aureus growth in a manner parallel to that observed when this bacterium was added to Gelfoam blocks pretreated with M 199 in the presence of antibiotics for 10 days, followed by the washing treatments described above (data not shown). Therefore, it is conceivable that our procedures did not remove all of the antibiotic but merely lowered the concentration of penicillin G to levels that slowed the growth rate of this bacterium. These experiments point to the fact that the microbicidal activity exhibited by endothelial cells cultured on Gelfoam is not due to contamination by the antibiotics added to the growth medium, although residual levels retard growth of S. aureus, but rather, bacterial killing requires viable, function ing endothelial cells . 300 e • G ·G <Ii rn -c til :::l II. Ilil c :::l <1l 200 Q) <1l .:; :c -e '0 ~ 0 ~ .c E 100 :::l C o 1 1.5 2 time(hours) Figure 4. No. of viable S. aureus after incubation with bovine pulmonary artery endothelial cells cultured on Gelfoam blocks and treated with antibiotic-containing or antibiotic-free (-) medium 199 for 10 days. G = Gelfoam blocks without endothelial cells; C = Gclfoam blocks laden with endothelial cells. Bars G and C were taken from figure I. Other bars reflect mean ::t SD of 3-6 separate experiments with 3-5 replicate incubations per time interval. Ratio of endothelial cells to S. aureus in endothelial cell cultures was 10: I. Downloaded from https://academic.oup.com/jid/article/174/5/1001/804957 by guest on 29 September 2023 o JID 1996; 174 (November) Microbicidal Activity of Endothelial Cells lID 1996 ; 174 (November) 250 1005 300 ._... '" • ~ C+P+S 111 -c 1m C+P r?:l C+S o '-' ... "' x ~ ~ 200 au) 100 ~ Ql 01 :J '" ~ '" ~ ~ '0 o 10 0 ti me ~ " .c E ~ c 1.5 (ho ur s) 2 Figure 6. No. of viable penicillin-resistant S. aur eus after incubation with bovine pulmonary artery endoth elial cells cultured on Gelfoam blocks and treated with antibiotic-containing medium 199 for 10 days. G = Gelfoam blocks without endothelial cells ; C = Gelfoam blocks laden with endothelial cells . Bars reflect mean :!: SD of 4 separate experiments with 3 replicate incubations per time interval. Ratio of endothelial cell cultures to S. aureus in endoth elial cell cultures was 10: I . o 1 I. S time (hours) No, of viable S aureus after incubation with bovine pulmonary artery endothelial cells cultured on Gelfoam blocks (C) and treated with antibiotic-containing (P = penicillin, 200 U/mL ; S = streptomycin, 200 ;.rg/mL) or antibiotic-free (-) medium 199 for 10 days. Bars reflect mean :!: SD of 3 separate experiments with 3 replicate incubations per time interval. Ratio of endothelial cells to S aureus in endothelial cell cultures was 10:1. Figure 5. Since penicillin G and strep tom ycin were in M 199, we undertook experim ents to determine which of these antibiotics was required for endotheli al cell - induced m icrobicidal activity. Endothelia l cells on Gelfoam were incubated with M199 in th e pre sence of either penicillin G (200 U/mL) or streptomycin (20 0 Ilg/mL). As shown in figure 5, only those endothelial cell s on Gelfoam cultured in M 199 containing penicillin G demonstrated bacterial killing. Therefore, penicillin G , and not streptomycin, appears to be the putative antibiotic responsible for the observed microb icidal act ivity. If re sidual penicillin G we re solely responsible for the observed bact erial killing , as shown in figure I , then substituting a penicillin-resistant strain of S. aureus should not alter the bacterial growth curve in the pr esence of endothelial cells cultured on Gelfoam blocks. A s shown in figure 6 , after 2 h the end othelial cell kill rate was > 90% compared with controls . Using radiolabeled penicillin G , we discovered that even after fou r washings, a sm all but sign ificant concentration of radiolabeled material was adherent to Gelfoam blocks, independent of the presence of endothelial cells (table 1). However, after separating endothelial cell s from the matrix, we were unable to detect cell-associated radioactivity (table I). These data further confirm tha t the penicillin G-dependent killing of S. aureus was not the result of either local accumulation or endothelial cell uptake of penicillin G. Since we have now observed a poss ibl e inductive event, it is important to focus on the endothelial cell to understand the killing process. Sequential ultrastructure examination of the interaction ofS. aureus with Gelfoam-contained endoth e lia l cell s is shown in figure 7. Adherence of the bacterium Localizat ion of C4 C]penicillin G in Gelfoam blocks and Gelfoam laden with endoth elial cells. Table I. Gelfoam control Medium Wash I Wash 2 Wash 3 Wash 4 Supernate Filter 299,728 :!:: 84 36 Gelfoam laden with endothelial cells 300,481 :!:: 80 52 44,202 :!:: 10,41 5 44,623 :!:: 12,498 866 9 :!:: 32 02 9221 :!:: 24 73 205 0 :!:: 455 225 2 :!:: 685 784 :!:: 159 984 :!:: 327 4477 :!:: 728 4021 :!:: 754 27 :!:: 4 41 :!:: 10 NOTE. Gelfoam blocks and Gelfoam laden with endothelial cells were cultured in 2 mL of medium 199 containing penicillin G (200 UlmL) and [1 4C]penicillin G (0.06 pCi/m L ) for 10 days. Every 3 days medium was changed with medium 199 of same composition. On day 10, Gelfoarn blocks, with and without endothelial cells, were washed and radioactivity from 0 ,5 mL of each sample and washes along with dissolved Gelfoam matrix and endothelial cells (collected on filters) were counted. Data are disintegrations per min, and each value represents mean :': SO of 20 different determinations. Downloaded from https://academic.oup.com/jid/article/174/5/1001/804957 by guest on 29 September 2023 :> c E ·!!! 150 ~ .c .!!! G IliII .0 - en ~ -'" • Zhang et al. 1006 r: degree of intracellul ar killing. Based on these data, > 90% of the tota l bacterial killi ng takes place with in the endothelial cell . We cannot overlook the po ssibility that the observe d bacterial killing takes pla ce within a phagocytic vacuole. To explore the feasibility of this hypot hesis, we incubated Gelfoam-co ntained endo thelial cell s, cultured in M 199 withou t antibiotic for 10 days, with S. aureus in the pres ence of either cytoc halasi n B (10 ttg /mL) or cytochalasin D ( 10 Jig/mL), folIowed by lysos taphin (0.5 ttg/mL ) treatm en t. As shown in figure 9, both cytochalasins demonstrated significant inhib ition of S. aureus internaliza tion, with cytochal asin D exhibiting considerably more pharmacologic activity. Becau se of these findin gs and the fact that cytoch alasin B interfere s with microfilament functi on in neutrophils [22, 23], we investigated the po ssibility that cytochalasin B may inhibi t the kilIing of S. aureus by our Ge lfoam-contained endothelial celI model. This was not observed . Cyt ochalasin B was found to only minimalIy, although nonsignifi cantly, affect the killing of S. aur eus by these euk aryot ic cells (figure 10). Discussion This paper describes the ability of endothelial cells, cultured on the three-dimensional mat rix , Ge lfoam, to kill S. aureus. This bactericidal act ivity is dependent on a series of events. As show n in figure 7, the adhere nce of S. aureus to the endothelial cells is the initial step . Then, phagocytosis interna lizes the bacterium , which even tua lly results in the killing of S. aureus. • Figure 7. Transmission electron micrographs showing sequential phagocyt osis of S. aureus by endothelial cells cultured on Gelfoam. Lett, Adherence of bacteria to endothelial cell plasma membrane: original magnification, X20,OOO. Middle, Pseudopodia! engulfment of bacteria by endothelial cells attached to Gelfoam: original magnification, X 15,750. Right, Endothelial cell phagosomal vacuole containing bacteria; original magnification, x 20,OOO. Downloaded from https://academic.oup.com/jid/article/174/5/1001/804957 by guest on 29 September 2023 to th e endo thel ial ce ll is th e initial step of th e process (figure 7, le ft ). T his is foll o wed by the ex tension of pseudop odia, whic h beg in to eng ulf th e bac te ria (figure 7, middle), si mi lar to wha t is obse rv ed wi th phagocytic cells [2 1]. At 3 h of inc u ba tio n, a representat ive electron ph ot o m icro gr ap h shows multi ple bacter ia within an apparent ph agocyt ic vac uole (figure 7, rig ht ). Alt houg h th e ult ras truc ture of the bacteria wit hi n this vacuole doe s not a llow us to co ncl ude th at S. aureus was killed, dat a from figure I un equi vocally de monstra te bacteri cid al activi ty by endo the lia l cells cultured on Ge lfoa m blo cks . These micrographs suggest, howe ver , that killing may be dep endent on the ph agocyto sis of S. a ureus. T he refo re, biochemical studies we re desi gned to dis criminate betw een sites of bacterial killing. Th e localization for endotheli al cell -mediated bactericidal activity was investigated usin g endothe lial cells cultured on Gelfoam blocks in the presence and absence of antibiotics. For these experiments, lysostaphin (0 .5 tt g/mL), which ha s been shown not to enter endothelial cells [I] , was added to quantify the number of interna lized S. aureus. W ith the antibiotic-free med ium - cultured endothelial ce lls, lysostaphin trea tment demonstrated an increased intracellular bact erial grow th with time of incub ation (figure 8). ln co ntrast to this finding , endothelial cells grown in the presen ce of antibiotics d id not perm it intracell ular bacterial grow th (figure 8) . Assuming that the pathway for internalization is the same , independen t of whet her medium co nta ins antibio tics, differences in viable S. aureus colonyformin g units in the two cell models a llow us to determine the lID 1996; 174 (November) Microbicidal Activity of Endothelial Cells lID 1996; 174 (November) thelial cells. A similar effect has been observed in the uptake of 20 A Nocardia asteroides by a pulmonary artery endothelial cell line 1/1 "e · C+LS • C+LS "co Q) co '> 10 "0 >< '0 Q; J:l E " c: o 2 1.5 ti me (hour s) B [25]. There are several possible explanations for why cytochal asin B did not inhibit the ability of endothelial cells to kill S. aureus. First, endotheli al cells are sensitive to cytochalasin B only when these cells are cultured on monolayers. Second, alterna tive, non-cytochalasin B- sensitive phagocytic pathways are responsible for the observed killing . Third, although cytochalasin B prevents phagosomal closure, microbic idal activity appears to takes place within an open phago cytic vacuole. With regard to the initial hypothesis, a number of studies have suggested that microv ascular endothelial cells in vivo are morphologically different in their cytoskel etal structure than when these cells are cultured on monolayer [26, 27]. Actin microfilam ents, which are of importance in regulating structural integrity of the endothelium, are located at central and periph- 1/1 "e c"o 20 ui ~ J:l .~ 0 ~ > >< • C+LS IIlI C+CB+LS mI C+CD+LS '0 ... Q) J:l E " 1/1 c: e" 0 ..L-- - - - o 1. 5 time 2 (hours ) "co ui Q) J:l Figure 8. No . of viable S. aureus after incubation with bovine pulmonary artery endothel ial cells cultured on Ge1foam blocks (C) and treated with antibiotic-free medium 199 (- ; A) or with antibioticcontaining medium 199 (B) for 10 days. At defined times, lysostaphin (LS, 0.5 ,ug/mL) was added and intracellular viabl e bact erial counts were determined. Bar s reflect mean ::+:: SD of 3 separate experiments with 3 replicat e incub ations per time interv al. Ratio of endothelial cell cultures to S. aureus in endothelial cell cultures was 10: 1. Assuming that the uptake of bacteria by endothelial cells is independent of whether these eukaryotic cells are cultured in medium containing antibiotics, these data clearly demonstrate that actin microfilament reorganization is an important component in the endothelial cell phagocytosis of S. aureus. Cytochalasins have been shown to prevent phagocyte-mediated uptake of particles, including microbes, when endothelial cells are grown in monolayer cultures. For example, Gibson et al. [19] have demonstrateda statistically significant cytochalasin D doseresponse inhibition of group B streptococcus invasion of human endothelial cells. Likewise, Vora and Karasek [24] have shown that cytochalasin B totally prevented the uptake of fluorescence paramagnetic latex beads by human dermal microvascular endo- '" > 10 . "0 ~ ~ '0 ... Q) J:l E e" 0 - ' - -- - o - 1.5 2 time (hours) Figure 9. No. of viable S. aureus after incubation with bovine pulmonary art ery endothelial cell s cultured on Gelfoam blocks (C) in antibiotic -free medium 199 for 10 days and pretreated with cyto chalasin B (CB) or cytochalasin D (CD) at 10 j.lg/mL for 1 h befo re inoculation of bacteria or left untreated. At defined times , lysostaphin (LS, 0.5 ,ug/mL) was added and intracellular viable bacterial counts were determined. Bar C + LS is taken from figure 8A. Other bars reflect me an::+:: SD of2 separate experiments with 3 replicate incubations per time interval. Ratio of endoth elial cells to S. aureus in endothelial cell cultures was 10: 1. Downloaded from https://academic.oup.com/jid/article/174/5/1001/804957 by guest on 29 September 2023 • ui :0 1007 Zhang ct a!. 1008 150 fIl :J • G C ill C+CB m e '" en :J 100 ~ .0 .~ 0 > "0 Q; ~ >< 50 .0 E :J C I. 5 lime (hours) Figure to. No. of viable S. aureus after incubation with bovine pulmonary artery endothelial cells cultured on Gelfoam blocks in antibiotic-containing medium 199 for 10 days and pretreated with cytochalasin B (CB, 10 J.1.g/mL) for I h before inoculation of bacteria or left untreated. G = Gelfoam blocks without endothelial cells; C = Gclfoam blocks laden with endothelial cells. Bars G and C were taken from figure I . Bar C+CB reflects mean ± SD of 3 separate experiments with 3 replicate incubations per time interval. Ratio of endothelial cells to S. aureu s in endothelial cell cultures was 10: I. cral sites within the cell [26-28]. Peripheral microfilamcnts, known as the dense peripheral band, participate in cell adhesion, whereas central microfilament bundles are important in extracellular matrix adhesion. In vivo, however, the dense peripheral band is not as well developed as observed in vitro [28]. Current knowledge suggests that the dense peripheral bands are sensitive to the effects of cytochalasin B, whereas central microfilament bundles appear to be nonresponsive towards cytochalasins [29]. The second hypothesis suggests that the observed bacterial killing occurs by a non -cytochalasin B -sensitive mechanism. Our data do not, however, discriminate between opposing routes. The concept of an alternative bacterial phagocytic pathway has been reported by Oelschlaeger et a1. [30] , who demonstrated that cytochalasin-mediated effects were both bacterial strain-dependent and target cell-dependent for Campylobact er jeju ni and Citroba cterfreundii. Similarly, Beaman and Beaman [25], using bovine pulmonary artery endothelial cells, showed that cytochalasin B increased the percentage of cellassociated N. asteroides. In fact , these authors termed this phe nomenon a " phagocytosis-inde p endent invasion factor. " In addition, macrophages, to which endothelial cells are gcnetically related, are insensitive to the actions of cytochalasin B, even though these cells continue to phagocytose microbes [31]. Our data support the third hypothesis, in that killing of S. aureus is almost exclusively intracellular, within a vacuole. Ev idence in support of this comes from the finding that in non penicillin G -treated endothelial cells, cytochalasins B and D, by pr eventing elosure of the phagosome, allowed lysostaphin to enter this cell and kill S. aureus. Since cytochalasin B did not inhibit bacterial killing, an enclosed phagosomal vacuole in the endothelial cells, unlike neutrophils, is not essential to the bactericidal activity exhibited by these eukaryotie cells . Penicillin G appears to pla y a dual role in the killing of S. aureus. First, we believe that at residual levels, penicillin G retards the growth of S. aureus. Support for this theory comes from our findings that the Gelfoam matrix, in either the presence or absence of endothelial cells , bound radiolabeled penicillin , even though no penicill in was detected associated with these eukaryotic cells (table I) . Since tracing studies cannot distinguish between the presence of penicillin or its inactive hydrolyzed products, there must undoubtedly still be some penicillin G remaining to retard the growth of S. aureus. Additionally, Root et a!. [32] demonstrated that sub-MIC doses of penicillin G impede staphylococcal growth by inhibiting crosswall form ation and causing the thickening of cell wall without subsequent separation. This pharmacologic activity enhances the susceptibility of S. aureus to be killed by phagocytic cells. Second, our findings suggest that penicillin G activates endothelial cells cultured on Gelfoam blocks to exhibit bactericidal activity, rather than mediating its microbicidal properties through the direct killing of S. aureus. Evidence in support of this hypothe sis comes from our data demonstrating a lack of either the internalization of penicillin G or the binding of this antibiotic to the surface of the endothelial cell (table I). Finally, penicillin-resistant S. aureus is equally killed by endothelial cells (figure 6). Classically, endothelial cells are thought to be passive participants in host immune response; however, our findings suggest a more active role for these cells , one in which they can be activated to phagocytose and kill bacteria. In some situations, this may provide an additional1ine of defense aga inst microbial penetration. Our results clearly demonstrate that endothelial cells cultured on Gelfoam blocks, which morphologically mimic these cells in the microvasculature, are activated by penicillin G to phagocytose and kill S. aureus. Future studies are aimed at elucidating the mechanism by which Gelfoamcontained endothelial cells kill S. aureus. Acknowledgments We appreciate the helpful suggestions and numerous discussions with Sovitj Pou, the continued technical assistance by Michael Gentry (University of Maryland School of Pharmacy) , and the photographic expertise of Milton D. Tudahl, Sr. (Director of Photographic Service s, Franc is Scott Key Medical Center, Baltimore). References I. Hamill RJ, Vann .1M , Proctor RA. Phago cytosis of Staphylococcus <llIreus by cultured bovine aortic endothelial cells: model for postadherence events in endo vascul ar infections. Infe ct Immun 1987; 54:833 - 6. 2. Ogawa SK, Yurberg fR, Ilatcher VB. Levitt M/\, Lowry FD. Bacterial adberence to hum an endothelial cells in vitro. Infect Immun 1985; 50: 218 - 24. Downloaded from https://academic.oup.com/jid/article/174/5/1001/804957 by guest on 29 September 2023 () .lID 1996 ; 174 (November) JID 1996; 174 (November) Microbicidal Activity of Endothelial Cells 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. Escherichia coli by human granulocytes. Immunology 1979; 37: 453-65. Gibson RL, Lee MJ, Soderland C, Chi EY, Rubens CEo Group B streptococci invade endothelial cells: type III capsular polysaccharide attenuates invasion. Infect Immun 1993;61:478-85. van den Broek PJ, Buys LFM, Mattie H, van Furth R. Effect of penicillin G on Staphylococcus aureus phagocytosed by human monocytes. J Infect Dis 1986; 153:586-92. Horowitz MA. Phagocytosis of microorganisms. Rev Infect Dis 1982; 4: 104-23. Britigan BE, Rosen GM, Chai Y, Cohen MS. Do human neutrophils make hydroxyl radical? Determination of free radicals generated by human neutrophils activated with a soluble or particulate stimulus using electron paramagnetic resonance spectrometry. J Biol Chern 1986;261: 4426-31. Cooper JA. Effects of cytochalasin and phalloidin on actin. J Cell Biol 1987; 105:1473-8. Vora M, Karasek MA. Retinoids upregulate phagocytosis by human dermal microvascular endothelial cells. J Cell Physiol 1994; 159:450-6. Beaman L, Beaman DL. Differences in the interactions of Nocardia asteroides with macrophage, endothelial, and astrocytoma cell lines. Infect Immun 1994;62:1787-98. Boor PJ, Gotlieb AI, Joseph EC, Kerns WD, Roth RA, Tomaszewski KE. Chemical-induced vascular injury. Toxicol Appl Pharmacol 1995; 132: 177-95. Gotlieb AI, Wong MKK. Current concepts on the role of the endothelial cytoskeleton in endothelial integrity, repair, and dysfunction. In: Ryan US, ed. Endothelial cells. Vol II. Boca Raton, FL: CRC Press, 1988: 81-101. Byers HR, White GE, Fujiwara K. Organization and function of stress fibers in cells in vitro and in situ. In: Shay J, ed. Cell and muscle motility. Vol 5. New York: Plenum Press, 1984:83-139. Wong MKK, Gotlieb AI. Endothelial cell monolayer integrity. I. Characterization of the dense peripheral band microfilaments. Arteriosclerosis 1986; 6:212-9. Oelschlaeger TA, Guerry P, Kopecko DJ. Unusual microtubule-dependent endocytosis mechanisms triggered by Campylobacter jejuni and Citrobacter freundii. Proc Natl Acad Sci USA 1993; 90:6884-8. Messick JB, Rikihisa Y. Characterization of Ehrlichia risticii binding, internalization, and proliferation in host cells by flow cytometry. Infect lmmun 1993;61:3803-10. Root RK, lsturiz R, Molavi A, Metcalf JA, Malech HL. Interactions between antibiotics and human neutrophils in the killing of staphylococci. Studies with normal and cytochalasin B-treated cells. J Clin Invest 1988; 67:247 -59. Downloaded from https://academic.oup.com/jid/article/174/5/1001/804957 by guest on 29 September 2023 3. Ryan US. Endothelial cell activation responses. In: Ryan US, ed. Pulmonary endothelium in health and disease. Vol 32. New York: Marcel Dekker, 1987:3-33. 4. Ryan US. Phagocytic properties of endothelial cells. In: Ryan US, ed. Endothelial cells. Vol III. Boca Raton, FL: CRC Press, 1988:33-49. 5. Ryan US, Ryan JW, Crutchley DJ. The pulmonary endothelial surface. Fed Proc 1985;44:2603-9. 6. Vann JM, Proctor RA. Ingestion of Staphylococcus aureus by bovine endothelial cells results in time- and inoculum-dependent damage to endothelial cell mono layers. Infect Immun 1987;55:2155-63. 7. Vann JM, Proctor RA. Cytotoxic effects of ingested Staphylococcus aureus on bovine endothelial cells: role of S. aureus a-hemolysin. Microb Pathog 1988;4:443-53. 8. Williams GM, Krajewski CA, Dagher FJ, Ter Haah AM, Roth JA, Santos GW. Host repopulation of endothelium. Transplant Proc 1971; 3: 869-72. 9. Robinson DH, Warren MK, Liang LT, Oprandy JJ, Nielsen TB, Kang YH. Retroviral transformation of cerebral microvascular endothelial cells: macrophage-like and microvascular endothelial cell properties. Blood 1991; 77:294-305. 10. Madri JA, Dreyer B, Pitlick FA, Furthmayr H. The collagenous components of the subendothelium: correlation of structure and function. Lab Invest 1980;43:303-15. 11. Madri JA, Pratt DM, Yannariello-Brown J. Endothelial cell-extracellular matrix interactions. Matrix as a modulator of cell function. In: Simionescu N, Simionescu M, eds. Endothelial cell biology in health and disease. New York: Plenum Press, 1988:167-88. 12. Bischoff J. Approaches to studying cell adhesion molecules in angiogenesis. Trends Cell Bioi 1995; 5:69- 74. 13. Thompson JA, Anderson KD, DiPietro JM, Zametta M, Anderson WF, Maciag T. Site-directed neovessel formation in vivo. Science 1988; 241: 1349-52. 14. Centra M, Raytch, RE, Cao GL, et al. Culture of bovine pulmonary artery endothelial cells on gelfoam blocks. FASEB J 1992;6:3117-21. 15. Jaffe EA, Hoyer LW, Nachman RL. Synthesis of antihemophilic factor by cultured human endothelial cells. J Clin Invest 1973; 52:2757 -64. 16. Sabatini DD, Bensch K, Barrnett RJ. Cytochemistry and electron microscopy. The preservation of cellular ultrastructure and enzymatic activity by aldehyde fixation. J Cell BioI 1963; 17:19-58. 17. Dvorak AM. Procedural guide to specimen handling for the ultrastructural pathology service laboratory. J Electron Microsc (Tokyo) 1987; 6: 255-301. 18. Leijh PCJ, van den Barselaar MT, van Zwet HL, Dubbeldeman-Rempt 1, van Furth R. Kinetics of phagocytosis of Staphylococcus aureus and 1009