Endothelial Cells Potentiate Oxidant-Mediated Kupffer Cell Phagocytic Killing

Free Radical Biology & Medicine, Vol. 24, Nos. 7/8, pp. 1217–1227, 1998 Copyright © 1998 Elsevier Science Inc. Printed in the USA. All rights reserved 0891-5849/98 $19.00 1 .00

PII S0891-5849(97)00453-X

Original Contribution ENDOTHELIAL CELLS POTENTIATE OXIDANT-MEDIATED KUPFFER CELL PHAGOCYTIC KILLING DOUGLAS A. POTOKA, SONSHIN TAKAO, TETSUHIRO OWAKI, GREGORY B. BULKLEY,

and

ANDREW S. KLEIN

Department of Surgery, Division of Liver Transplantation, The Johns Hopkins University School of Medicine, Baltimore, MD, USA (Received 14 April 1997; Revised 29 September 1997; Accepted 10 November 1997)

Abstract—Phagocytosis and killing of circulating organisms by Kupffer cells (KCs) are discrete, important components of host defense. However, the killing mechanism(s) are not fully understood, and the potential role of adjacent nonparenchymal cells such as hepatic endothelial cells has not been defined. Rat KCs 7 an hepatic endothelial cell enriched cellular fraction (HECEF) were incubated with Candida parapsilosis and assayed for phagocytosis and phagocytic killing by validated fluorochromatic vital staining. The role of reactive oxygen metabolites in KC phagocytic functions was examined by inhibition with superoxide dismutase and/or catalase. Diphenyleneiodonium and allopurinol were used to examine the potential roles of NADPH oxidase and xanthine oxidase, respectively, in generating these toxic oxidants. Coculture with HECEF increased KC phagocytic activity (from 75% to 88%) and candidacidal activity (from 20% to 31%). Superoxide dismutase, catalase, diphenyleneiodonium, or allopurinol caused inhibition of candidacidal activity, but did not affect phagocytosis, and did not block the potentiation of phagocytosis or of killing caused by coculture with HECEF. Reactive oxygen intermediates generated by both NADPH oxidase and xanthine oxidasedependent pathways are important in KC killing of Candida parapsilosis. In vitro, KC phagocytosis and killing are potentiated (via a non-oxidant-mediated mechanism) by coculture with a preparation of hepatic non-parenchymal cells composed primarily of endothelial cells. © 1998 Elsevier Science Inc. Keywords—Kupffer cell, Reticuloendothelial, Phagocytosis, Endothelial cells, Toxic oxidants, Free radical

INTRODUCTION

gested organisms is through the conversion of ambient oxygen into the superoxide ion (O22), hydrogen peroxide (H2O2), the hydroxyl radical (OHz), and singlet oxygen (1O2z).6,7 These reactive oxygen metabolites (ROMs) act directly and in concert with the myeloperoxidasehalide system and with lysosomal enzymes to kill both immediately adjacent and ingested microorganisms. The importance of ROMs in macrophage microbicidal activity is suggested by evidence that peritoneal macrophages produce ROMs upon activation or exposure to pathogens8,9 and that macrophage microbicidal activity is inhibited by free radical scavengers, such as superoxide dismutase (SOD), an enzymatic O22 scavenger, and catalase, an enzymatic scavenger of H2O2.8 –10 The role of a specialized, membrane associated NADPH oxidase in the generation of these ROMs by circulating macrophages is well-established.6,7 However, other enzyme systems may also contribute to ROM generation. In previous experiments, macrophage killing was inhibited not only by diphenyleneiodonium (DPI), an inhibitor of

Kupffer cells (KCs), the resident macrophages of the liver, comprise the most numerous population of macrophages in the body and are the major cellular component of the systemic reticuloendothelial system.1–3 A primary function of the pluripotent KC is to provide a defense against systemic infection through the phagocytosis and killing of circulating pathogens.4 Indeed, the KC’s strategic location in the hepatic sinusoids places it in an ideal position to clear the portal blood of gut-derived microorganisms and toxins in particular.5 The microbicidal activity of macrophages in general depends on their ability to both phagocytose organisms and to kill those pathogens once ingested. One mechanism by which other macrophages kill and degrade inAddress correspondence to: Andrew S. Klein, M.D., The Johns Hopkins Hospital, 600 N. Wolfe St., Baltimore, MD 21287– 8611; Tel: (410) 955-5662; Fax: (410) 614-2079; E-Mail: aklein@ welchlink.welch.jhu.edu. 1217

1218

D. A. POTOKA et al.

(“neutrophil”) NADPH oxidase,11,12 but also by micromolar concentrations of allopurinol, a very specific inhibitor of xanthine oxidase.13 This inhibition was additive (i.e., neither synergistic nor redundant), suggesting that both NADPH oxidase and xanthine oxidase are important for the generation of candidacidal oxidants by these macrophages.10 Given the common origin14 and similar function15 of KCs and other macrophages, KCs might be expected to exhibit similar microbicidal mechanisms. Like peritoneal macrophages, KCs have been shown to produce ROMs upon activation.16 –18 However, at least one prior study has reported that KCs have greater phagocytic activity, but release significantly less O22, than do peritoneal macrophages.19 Moreover, KCs are localized in situ within the specialized microenvironment of the hepatic sinusoid where they have a close structural and functional relationship with other non-parenchymal cells, including especially the hepatic endothelial cells.20 It seems likely that KC function might be modulated by interactions with other cell types within this sinusoidal microenvironment. If so, in vitro studies of isolated KCs alone could fail to reveal their full functional potential. This study focused upon in vitro oxidant-mediated phagocytosis and phagocytic killing of Candida parapsilosis by rat KCs alone, and by KCs in coculture with an hepatic endothelial cell-enriched fraction of non-parenchymal cells (HECEF). The acridine orange/crystal violet staining technique21 was used to differentially quantitate the phagocytic and candidacidal activities of these KCs isolated from rat liver. Experiments were conducted both in the presence and the absence of specific scavengers of ROMs and of specific inhibitors of known endogenous oxidant-generating enzyme systems to better define the mechanisms of this oxidant-mediated component of phagocytic killing. MATERIALS AND METHODS

Isolation and characterization of KC and of the HECEF fraction All experimental protocols were pre-approved by the Animal Care and Use Committee of the Johns Hopkins University School of Medicine. Separate rat KC and HECEF fractions were isolated using collagenase perfusion and pronase digestion as previously described22 with some modification. Briefly, male Lewis rats (175– 200 g) were anesthetized with methoxyflurane and adequate anesthesia was maintained throughout the procedure using a nose cone. Using sterile technique, the portal vein was exposed via a midline incision and cannulated with a 20G angiocatheter. The liver was first perfused with 200 ml of Ca21-free Hanks’ balanced salt

solution (HBSS; Sigma Chemical Co., St. Louis, MO) at a rate of 12–15 ml/min. HBSS perfusion was followed by perfusion with 250 ml of 0.05% Type IV collagenase (Sigma) in Ca21-rich Hanks’ balanced salt solution (HBSS1; Sigma) at a rate of 30 ml/min. The liver was then removed, placed in a sterile beaker with 200 –300 ml of HBSS1 containing 10 mg/ml DNase (Sigma), 8 mg/ml HEPES (Gibco BRL, Life Technologies, Inc., Grand Island, NY), and 50 U/ml penicillin 1 5 mg/ml streptomycin (Cellgro, Mediatech, Washington, DC), and mechanically disrupted to a paste-like consistency using sterile scissors. Pronase (Sigma), 2 mg/ml in HBSS1 with 8 mg/ml HEPES, was then added to achieve a pronase concentration of 0.02%, and digestion was allowed to proceed for 45 min at 37°C with stirring at 275 rpm while pH was maintained at 7.3–7.5 by the addition of sterile 1N NaOH, as necessary. The digest was then centrifuged at 1,100 rpm, 4°C for 10 min and the pellet was washed three times with 30 ml of cold HBSS1 containing 8 mg/ml HEPES and 10 mg/ml DNase (HBSS1HEPES/DNase). The final pellet was resuspended in 30 ml cold HBSS1HEPES/DNase and centrifuged at 800 rpm, 4°C for 1 min, and the resulting supernatant (containing most of the hepatic nonparenchymal cells) was layered on a sterile Histopaque (Sigma) gradient (10 ml 1.077 Histopaque over 10 ml 1.083 Histopaque) which was then centrifuged at 2,600 rpm, 4°C for 45 min. The interface cells were collected, resuspended in 30 ml cold HBSS1HEPES/DNase, and centrifuged at 1,100 rpm, 4°C for 10 min. The pellet was washed twice in 30 ml cold HBSS1HEPES/DNase, resuspended once more in 30 ml cold HBSS1HEPES/ DNase, and the nonparenchymal cells were collected in the supernatant after centrifugation at 800 rpm, 4°C for 1 min. The nonparenchymal cells were pelleted by centrifugation at 1,100 rpm, 4°C for 10 min, and the pellet was resuspended in 30 ml of room temperature RPMI media (Gibco BRL) with 10% FBS (Gemini Bioproducts, Calabasas, CA). KCs were then separated from the other hepatic nonparenchymal cells in this preparation by their differential adhesion to tissue culture flasks. Briefly, the suspension of nonparenchymal cells in RPMI 1 10% FBS was divided equally into three 75 cm2 polystyrene tissue culture flasks (Corning Inc., Corning, NY) which were then incubated at 37°C in 5% CO2 for 15 min. During this incubation KCs, confirmed by their distinctive morphology under phase-contrast microscopy, preferentially adhere to the flask, while other nonparenchymal cells, including hepatic endothelial cells, adhere less strongly. After this incubation the culture media was removed and added to new tissue culture flasks. The original flasks were washed with 2 ml of 0.05% trypsin, 0.53 mM EDTA solution (Cellgro) and twice with 4 ml

Kupffer cell phagocytic killing

HBSS1HEPES/DNase, and 10 ml of fresh RPMI 1 10% FBS was added to these flasks. The washes were added to the media from the first incubation step and the incubation/adhesion procedure was repeated twice. The media and washes from the final incubation were placed in new culture flasks and, following morphologic verification, used as an hepatic endothelial cell enriched fraction (HECEF). Cultures of the KC fraction contained .90% KCs, while the HECEF fraction contained 60 – 80% endothelial cells (the remainder being identified as fibroblasts or stellate cells, but not hepatocytes) as morphologically assessed by microscopic examination. The viability of KCs and endothelial cells was consistently .90% by trypan blue exclusion. Cultures of KCs and of HECEF were maintained at 37°C and 5% CO2 with daily replacement of the RPMI 1 10% FBS media. These cells were used for experiments within 2–5 days of isolation. Preparation of Candida parapsilosis Candida parapsilosis was chosen as the strain for these experiments because of its relatively low level of endogenous catalase activity and consequent sensitivity to oxidant killing.9 C. parapsilosis was passaged weekly on Sabouroud plates at 37°C. For each experiment, C. parapsilosis was grown in 40 ml of trypticase soy broth (Becton Dickinson, Cockeysville, MD) in a 50 ml centrifuge tube at 37°C for 24 h. The yeast were then pelleted by centrifugation at 1,600 rpm for 10 min. The pellet was washed twice with HBSS, resuspended in 15 ml of RPMI 1 10% FBS, and incubated at 37°C for 30 min to allow large clumps of yeast to settle to the bottom. After this incubation, the number of suspended C. parapsilosis in the supernatant was counted for each experiment using a hemocytometer. Incubation of KC with C. parapsilosis KCs were scraped from the culture flask using a sterile cell rubber policeman, counted using a hemocytometer (yield 5 104 to 105 KC/ml), and then dispensed into either four-chambered (500 ml per chamber) or eight-chambered ( 300 ml per chamber) Lab-Tex tissue culture slides (Miles Scientific, Naperville, IL). For some experiments, HECEF were prepared in a similar fashion and added to the chambers at a counted endothelial cell:KC ratio of 2:1. After the plating of the KCs 7 HECEF, the slides were incubated at 37°C, 5% CO2 for 2 h. The culture medium was then removed by gentle aspiration and replaced with fresh RPMI 1 10% FBS and the slides were incubated overnight for 20 h in fresh RPMI 1 10% FBS. C. parapsilosis were then added at a candida:KC ratio of 20:1. Experiments were also per-

1219

formed in the presence of superoxide dismutase (SOD; 100 –500 U/ml; Calbiochem, La Jolla, CA) and/or catalase (100 –500 U/ml; Calbiochem) and with diphenyleneiodonium (DPI; 10 –100 mM; Cookson Chemicals, Southampton, England) and/or allopurinol (10 –100 mM; Sigma). Catalase and/or SOD were added to the coculture at the time of addition of C. parapsilosis, while DPI and/or allopurinol were added 30 min or 20 h, respectively, prior to the addition of C. parapsilosis. All cocultures were then incubated for 90 min at 37°C, 5% CO2. Acridine orange/crystal violet staining The acridine orange/crystal violet staining assay provides an in vitro technique to quantitatively discriminate the phagocytic and intracellular killing functions of phagocytic cells.21 The fluorochrome acridine orange gives a green emission upon contact with doublestranded yeast DNA and a red/yellow emission upon contact with single-stranded (denatured) DNA. Viable Candida thus exhibit green orthochromatic fluorescence, whereas dead microorganisms exhibit bright red/orange metachromatic fluorescence.21 Crystal violet quenches most extracellular fluorescence, and therefore allows the discrimination of intracellular from extracellular organisms. The above-described tissue culture slides were stained with acridine orange/crystal violet immediately after 90 min coculture. Slides were gently washed twice for 30 s in HBSS, stained with 0.01% acridine orange (Sigma) in HBSS for 1 min, rinsed in HBSS for 30 s, and then stained with 0.05% crystal violet (Sigma) in HBSS. Excess crystal violet was then removed by rinsing in HBSS. The procedure was performed at 37°C and a pH of 7.2. Cover glasses were mounted on the slides and the edges were sealed with nail polish. Slides were either viewed immediately after staining or placed in the refrigerator on ice and viewed within 2 h of staining after pilot experiments had shown that this did not affect the results of the assay. Assessment of acridine orange/crystal violet staining For each individual experimental condition, 200 –350 KC and 1,000 –3,000 intracellular C. parapsilosis organisms were scanned using a fluorescence microscope (Carl Zeiss Inc., Germany) with fluorescence filters (excitation 485 nm, dichroic 510 nm). After acridine orange staining, viable C. parapsilosis appeared green, while dead C. parapsilosis appeared bright red/yellow. We have previously verified this relationship by quantitative culture.10 The observer was blinded to the various treatments at the time of slide viewing. Each slide, containing

1220

D. A. POTOKA et al.

Fig. 1. Phagocytic activity (A), phagocytic index (B), and candidacidal activity (C) of KCs in the absence and presence of HECEF. Results are the means 7 SD with n 5 18 for KCs only and for HECEF only, and n 5 14 for KCs 1 HECEF. The addition of HECEF clearly stimulated both phagocytosis and killing by KCs. (*p , 0.001 for KC 1 HECEF vs. KCs alone by Student–Newman–Keuls test). There was negligible phagocytic activity (,1%) in the HECEF only which precluded accurate determination of phagocytic index or candidacidal activity for this fraction.

four or eight wells, contained simultaneous control and treatment groups. Phagocytic Activity was defined as the percentage of KCs that had phagocytosed at least one C. parapsilosis organism (whether living or dead). Phagocytic Index was calculated as the average number of C. parapsilosis organisms ingested per phagocytosing KC. Candidacidal Activity was determined as the percentage of intracellular (phagocytosed) C. parapsilosis organisms that were dead. The calculations are summarized as follows: Phagocytic Activity ~%! 5

# phagocytosing KCs 3 100 total # KCs

Phagocytic Index # intracellular C. parapsilosis organisms # phagocytosing KCs Candidacidal Activity ~%! 5

5

# dead intracellular organisms 3 100 total # intracellular organisms

Analysis of data Values were expressed as means or percentages 7 1 SEM unless noted differently in the figure legend. Standard errors for percentages were calculated based on the binomial distribution. Differences between means were evaluated by ANOVA and the Student–Newman–Keuls

test, while apparent differences between percentages were evaluated for significance using the chi-square test. Values of p , .05 were considered to be significant. RESULTS

KC and HECEF phagocytic and killing function After 90 min incubation with C. parapsilosis under control conditions, 75% of KCs had phagocytosed at least one organism (phagocytic activity, Fig. 1A). When HECEF were also present during coculture, the phagocytic activity of KCs increased to 85% (p , .001). In contrast, the presence of HECEF had little effect upon KC phagocytic index (Fig. 1B). In the absence of HECEF, KCs killed 20% of phagocytosed C. parapsilosis organisms (Fig. 1C). The presence of HECEF significantly increased the candidacidal activity of KCs approximately 1.5-fold (p , .001). As a control, the phagocytic activity, phagocytic index, and candidacidal activity of HECEF alone was also determined by incubation of HECEF with C. parapsilosis under the same conditions used for KCs. There was no detectable phagocytic nor killing activity in the HECEF fraction. Effect of SOD and catalase on KC phagocytosis and killing SOD did not affect the phagocytic activity of KCs (Fig. 2A). However, SOD did inhibit the candidacidal

Kupffer cell phagocytic killing

1221

Fig. 2. Effect of SOD on KC phagocytic activity (A) and candidacidal activity (B) in the absence and presence of HECEF. Both phagocytic activity and candidacidal activity for KCs plus HECEF were significantly higher than for KCs alone at all concentrations of SOD (*p , .01 by ANOVA; data expressed as mean 7 SEM). Phagocytic activity was not affected by SOD. SOD significantly inhibited candidacidal activity of the KCs compared to control, both alone and in the presence of HECEF, but did not diminish the potentiation produced by the HECEF.

activity both of KCs alone and of KCs with HECEF (Fig. 2B). Maximal inhibition for KCs with or without HECEF was seen at SOD concentrations of 100 U/ml (p , .001 for each). SOD inhibited the candidacidal activity of KCs alone and of KCs plus HECEF in a parallel manner, such that candidacidal activity for KC plus HECEF remained 1.8 –2 times higher than for KCs alone at each concentration of SOD (p , .01). The effect of catalase on KC phagocytic and killing functions was similar to the effect of SOD. Catalase did not significantly affect KC phagocytic activity (Fig. 3A), but did inhibit KC candidacidal activity (Fig. 3B). The inhibition of candidacidal activity by catalase was parallel for KCs with or without HECEF. At each concentration of catalase, the candidacidal activity of KCs plus HECEF remained approximately 1.5 times higher than with KC alone (p , .01). To determine the relative contributions of O2 2 and H2O2 to KC function, phagocytic and candidacidal activities were also determined in the presence of both SOD and catalase. Concentrations of 100 U/ml of both catalase and SOD were chosen because each enzyme alone caused maximal inhibition of KC candidacidal activity at that concentration (Figs. 2B and 3B) and because higher concentrations of SOD may be less effective in scavenging superoxide.23,24 Neither phagocytic activity nor phagocytic index were significantly

affected by SOD, catalase, or SOD plus catalase (Fig. 4A, B). The combination of SOD plus catalase did, however, appear to have an additive inhibitory effect upon KC candidacidal activity (Fig. 4C) in both the absence and presence of HECEF, although these differences did not reach statistical significance. Effects of DPI and allopurinol on KC phagocytosis and killing To determine which enzyme system(s) may be important in the generation of candidacidal ROMs by KCs, phagocytic and candidacidal activities were determined in the presence of DPI, an inhibitor of NADPH-oxidase, and/or allopurinol, a highly specific inhibitor of xanthine oxidase. DPI did not affect KC phagocytosis in the absence or presence of HECEF (Fig. 5A). As expected, DPI did cause substantial inhibition of KC candidacidal activity (Fig. 5B). For KCs alone, maximal inhibition was about 50% at 30 mM DPI (p , .001). For KCs plus HECEF, inhibition was also maximal at 30 mM DPI, where candidacidal activity was 67% of control. The inhibition of candidacidal activity of KCs in the presence of HECEF appeared parallel to the inhibition seen with KCs alone. The candidacidal activity of KCs plus HECEF remained 1.8 –2.2 times higher than for KCs alone at all concentrations of DPI (p , .01).

1222

D. A. POTOKA et al.

Fig. 3. Effect of catalase on KC phagocytic activity (A) and candidacidal activity (B) in the absence and presence of HECEF. Both phagocytic activity and candidacidal activity for KCs plus HECEF were significantly higher than for KCs alone at all concentrations of catalase (*p , .01 by ANOVA; data expressed as mean 7 SEM). Phagocytic activity was not affected by catalase. Catalase significantly inhibited candidacidal activity of KCs compared to control, both alone and in the presence of HECEF, but did not diminish the potentiation produced by HECEF.

Fig. 4. Phagocytic activity (A), phagocytic index (B), candidacidal activity (C) for KCs or KCs plus HECEF in the presence of SOD (100 U/ml), catalase (100 U/ml), or SOD 1 catalase (100 U/ml of each). Results are the means 7 SD with n 5 3 or 4 for each condition. Once again, coculture enhanced phagocytic and candidacidal activities. SOD and catalase appeared to inhibit only killing, in an additive and parallel manner, but these differences did not reach statistical significance.

Kupffer cell phagocytic killing

1223

Fig. 5. Effect of diphenylineiodonium (DPI) on KC phagocytic activity (A) and candidacidal activity (B) in the absence and presence of HECEF. Candidacidal activity for KCs plus HECEF was significantly higher than for KCs alone for each concentration of DPI (*p , .01 by ANOVA; data expressed as mean 7 SEM), while the apparent increase in phagocytic activity for KCs plus HECEF did not reach statistical significance. Phagocytic activity was not affected by DPI. DPI significantly inhibited candidacidal activity of KCs compared to control, both alone and in the presence of HECEF. The potentiation produced by the HECEF was not diminished by DPI.

Like DPI, the xanthine oxidase inhibitor allopurinol had little effect on KC phagocytic activity (Fig. 6A), while KC candidacidal activity was inhibited by allopurinol (Fig. 6B). For KCs with or without HECEF, maximal inhibition of candidacidal activity was observed with 100 mM allopurinol (p , .001). Candidacidal activity of KCs plus HECEF remained 1.6 –1.9 times higher than candidacidal activity for KCs alone at all allopurinol concentrations (p , .01). To determine the relative contributions of NADPH oxidase and xanthine oxidase to the generation of candidacidal ROMs, KC functions were examined in the presence of both DPI and allopurinol. Neither KC phagocytic activity nor phagocytic index were affected by these inhibitors (Fig. 7A, B). Allopurinol and DPI did appear to have additive effects in the inhibition of KC candidacidal activity, although this was not statistically different from the inhibition observed with DPI alone or allopurinol alone (Fig. 7C). DISCUSSION

The microbicidal activity of macrophages, including KCs, is comprised of two principal physiologic processes, phagocytosis and intracellular killing. Phagocytosis is an energy requiring event involving the binding

of particles to the phagocyte cell surface with the subsequent ingestion of the particle.25,26 The phagocytic process results in a respiratory burst during which molecular oxygen (O2) is reduced to the superoxide anion (O22), which is converted through dismutation and ironcatalyzed Fenton reduction to more toxic intermediates, such as hydrogen peroxide (H2O2) and the hydroxyl radical (OHz).6,7 These ROMs promote lipid peroxidation, protein cross-linking and oxidation, and nucleic acid denaturation. Along with lysosomal enzymes and with hypochlorous acid (HOCI) generated by the myeloperoxidase-halide system, these ROMs function to kill and degrade phagocytosed microorganisms.27,28 In our experiments, 75% of KCs phagocytosed at least one C. parapsilosis after coculture for 90 min. This is comparable to the 70 – 80% phagocytosis previously reported in a murine peritoneal macrophage model using similar techniques.10 The phagocytic index determined for KCs in our experiments was 8.9 organisms per phagocytosing cell, which is comparable to the phagocytic index of approximately 7 organisms per phagocytosing cell previously observed for peritoneal macrophages.10 In the absence of HECEF, KCs killed 20% of ingested C. parapsilosis organisms in 90 min (candidacidal activity), which is lower than the 50% killing of C. parapsilosis by peritoneal macrophages observed

1224

D. A. POTOKA et al.

Fig. 6. Effect of allopurinol on KC phagocytic activity (A) and candidacidal activity (B) in the absence and presence of HECEF. Candidacidal activity for KC plus HECEF was significantly higher than for KCs alone for each concentration of allopurinol (*p , .01 by ANOVA; data expressed as mean 7 SEM), while the apparent increase in phagocytic activity for KCs plus HECEF did not reach statistical significance. Phagocytic activity was not affected by allopurinol. Allopurinol significantly inhibited candidacidal activity of KCs compared to control, both alone and in the presence of HECEF. Again, the potentiation produced by HECEF was not diminished by allopurinol.

previously,29 but consistent with the reported observation that KCs release significantly less O22 than peritoneal macrophages upon activation.19 KCs normally reside in close contact with the hepatic endothelial cells lining the hepatic sinusoids.20 We hypothesized that this specialized microenvironment influences KC function with respect to both phagocytosis and killing. The presence of HECEF in vitro potentiated both phagocytic and killing functions of KCs. The cellular subpopulation referred to in our studies as “the hepatic endothelial cell enriched fraction” contained 60 – 80% endothelial cells, with the remainder made up of other nonparenchymal cells, including fibroblasts and stellate (Ito) cells, but not hepatocytes. The fact that the HECEF alone showed no measurable phagocytic nor candidacidal activity argues against the possibility that the HECEF itself contained a significant population of KCs. It is therefore important to note that, while we attribute the effects of this fraction to the predominant hepatic endothelial cells, we cannot rule out the influence of other cellular components. We cannot exclude the possibility that during the coculture process, HECEF cells promote the growth of KC (or vice versa). However, given the short incubation time (,24 hrs) used in these experiments, it is unlikely that significant expansion of

either of these normally slow gowing cell populations occurred. Immunoglobulins, complement components, and fibronectin can serve as opsonins which promote KC phagocytosis.26,30 –32 Hepatic endothelial cells have been shown to secrete fibronectin,33 which promotes binding and phagocytosis of particles by KCs through interaction with high affinity fibronectin binding sites on the KC surface.32,34 One explanation of the observation that phagocytic activity increased but phagocytic index was unchanged in the presence of HECEF is that hepatic endothelial cells might up-regulate the number of actively phagocytosing KCs via secretion of an opsonin or other soluble mediator or by direct intercellular contact, while the maximal number of particles phagocytosed by each individual KC may be fixed by other factors (such as KC size). The presence of HECEF also caused a significant increase in KC candidacidal activity. It is interesting to note that inhibition of xanthine oxidase with low concentrations (10 mM) of allopurinol resulted in relatively more inhibition of KC candidacidal activity when HECEF were present. Xanthine oxidase resides on the surface as well as in the cytoplasm of endothelial cells35 and may play a role in up-regulation of KC killing by the

Kupffer cell phagocytic killing

1225

Fig. 7. Phagocytic activity (A), phagocytic index (B), and candidacidal activity (C) for KCs or KCs plus HECEF cells in the presence of DPI (30 mM), allopurinol (10 mM), or DPI plus allopurinol. Results are the means 7 SD with n 5 3 or 4 for each condition. No effect was seen on phagocytic activity. DPI and allopurinol inhibited killing in a parallel manner (*p , .05 vs. control), and although DPI plus allopurinol appeared to inhibit candidacidal activity more than either DPI or allopurinol alone, this did not achieve statistical significance.

release of O22 which may be locally active in killing or serve as a signal to upregulate KC killing. At least one previous study, in which latex beads were used as phagocytic targets, found endothelial cells to possess a significant phagocytic function when KC phagocytic function was impaired.36 In our experiments, we did not observe hepatic endothelial cells to have an appreciable phagocytic function in the presence or absence of KCs. These disparate observations may be related to characteristics of the different targets selected in each set of experiments. Alternatively, our utilization of pronase digestion, which has been shown to interfere with certain endothelial cell surface proteins during isolation may also account for the difference seen in our experiments. The ROM scavengers, SOD and catalase, and the enzyme inhibitors, DPI and allopurinol, did not affect KC phagocytic function in our experiments. Thus, although binding of particles to the KC surface initiates the respiratory burst leading to O22 generation,37 the subsequent formation of ROMs does not appear to be involved in the process of particle ingestion. Catalase and SOD each inhibited KC candidacidal activity, which is consistent with previous experiments showing that SOD inhibits macrophage O22 release and that both SOD and catalase inhibit killing.9,10 We have found that killing of C. parapsilosis by murine peritoneal macrophages is inhibited by SOD and by the combination of catalase and SOD, although catalase alone did not

inhibit candidacidal activity.10 In contrast, these present experiments found KC candidacidal activity to be significantly reduced by catalase alone as well as by SOD alone. We found maximal inhibition of KC candidacidal activity to occur at 100 U/ml SOD, with reduced inhibition of killing at greater SOD concentrations. This effect may be due to the ability of O22 to act as both an initiator and terminator of lipid peroxidation and the reduction of available terminator O22 molecules by the higher concentrations of SOD.23,24 Moreover, because SOD acts by accelerating the dismutation of O22 to H2O2, it can increase the concentration of H2O2, which is itself toxic, either directly or by interacting with O22 in the trace metal-catalyzed Fenton reduction to yield even more potent ROI.7 Indeed, the inhibition of KC candidacidal activity in our experiments by both SOD and catalase suggests that this might be an important component of the KC’s killing function. Traditionally, the generation of O22 by neutrophils and macrophages has been considered to be primarily a function of the enzyme “neutrophil” NADPH oxidase, which utilizes NADPH as an electron donor to reduce molecular oxygen.38 However, not only DPI, an inhibitor of NADPH oxidase,11,12 but also allopurinol, an inhibitor of xanthine oxidase,13 have previously been found to inhibit peritoneal macrophage candidacidal activity, suggesting that xanthine oxidase may also be involved in the generation of candidacidal O22 by macrophages.10 In the

1226

D. A. POTOKA et al.

present study, KC candidacidal activity was also inhibited by both DPI and allopurinol. However, the presence of biologically significant levels of xanthine oxidase in macrophages has been debated.39 Although enzymatic assays have shown low levels of xanthine oxidase activity in alveolar macrophages,40 more recent studies utilizing immunohistochemistry and a newer spectrofluorometric assay have shown substantial amounts of xanthine oxidase to be present in a murine peritoneal macrophage (IC-21) cell line.10 Moreover, KCs have recently been shown to possess significant xanthine oxidase activity when compared with hepatic parenchymal and endothelial cells.41 These data, along with our results showing inhibition of KC candidacidal activity by micromolar concentrations of the specific xanthine oxidase inhibitor allopurinol, suggest that xanthine oxidase does indeed play a significant role in the generation of candidacidal ROMs by macrophages including KCs. Although we have used allopurinol as a specific xanthine oxidase inhibitor, allopurinol and its most active metabolite, oxypurinol, can also function as scavengers of the highly reactive hydroxyl radical and of hypochlorous acid (HOCl) at very high concentrations, at least in vitro.42,43 The experiments which identified this scavenging activity for allopurinol used concentrations (0.5 to 1.5 mM) much greater than the concentrations of allopurinol (10 –100 mM) used in our experiments.42 We have found that specific inhibition of xanthine oxidase occurs at these concentrations of allopurinol without measurable scavenging activity.44 Therefore, the decreased candidacidal activity in the presence of allopurinol in our experiments is most likely to be due to the specific inhibition of xanthine oxidase activity. As well as ROMs, the production of nitric oxide (NOz) from L-arginine may be important in several macrophage and KC killing functions, including host defense and anti-tumor activity.45– 48 In addition to direct affects of NOz, NOz may combine with O22 to form peroxynitrite, which may exert direct cytotoxic effects by oxidizing sulfhydryl groups or by releasing the OHz radical upon decomposition.49,50 We have not attempted to directly examine the role of NOz in the candidacidal activity of KCs in the experiments presented here. However, DPI, which we used as an inhibitor of NADPH oxidase, has been found to also inhibit the formation of NOz by macrophage nitric oxide synthase (NOS).51 NOS exists in both constitutive and inducible forms, and it is thought that the inducible form, i-NOS, is responsible for most of the NOz-dependent killing by macrophages.48,52 Upon activation of KCs with LPS, a lag phase of 4 – 6 h occurs before the release of NOz increases significantly, and maximal i-NOS activity is not achieved until 12 h after activation.53 Therefore, we feel that during the 90 min coculture time used in our experiments, DPI is

unlikely to have acted by inhibiting NOz production by i-NOS. These observations do not, however, exclude the possibility that endothelial eNOS might be a source of NO which, in turn, could enhance phagocytic killing. In summary, the ROMs O22 and H2O2 play important roles in the phagocytic killing mechanism of KCs. The inhibition of KC candidacidal activity by both DPI and allopurinol suggests that both NADPH oxidase and xanthine oxidase are important sources of the candidacidal ROMs generated by KCs, at least in vitro. KC phagocytosis and killing were both potentiated by HECEF in vitro, which is consistent with their intimate structural and functional relationship within the hepatic sinusoids in situ. While the mechanism of this potentiation remains to be elucidated, it does not appear to be oxidant mediated. In any case, these findings clearly indicate that studies of Kupffer cells alone in in vitro experiments may fail to reveal their full functional capability.

REFERENCES 1. Wake, K.; Decker, K.; Kirn, A.; Knook, D. L.; McCuskey, R. S.; Bouwens, E.; Wisse, E. Cell biology and kinetics of Kupffer cells in the liver. Int. Rev. Cytol. 118:173–229; 1989. 2. Smedsrod, B.; De Bleser, P. J.; Braet, F.; Lovisetti, P.; Vanderkerken, K.; Wisse, E.; Geerts, A. Cell biology of liver endothelial and Kupffer cells. Gut 35:1509 –1516; 1994. 3. Wardle, E. N. Kupffer cells and their function. Liver 7:63–75; 1987. 4. Ruggiero, G.; Utili, R.; Andreana, A. Clearance of viable Salmonella strains by the isolated, perfused rat liver: a study of serum and cellular factors involved and the effect of treatments with carbon tetrachloride or Salmonella enteritidis lipopolysaccharide. J. Reticuloendothel. Soc. 21:79 – 88; 1977. 5. Yamaguchi, Y.; Yamaguchi, K.; Babb, J. L.; Gans, H. In vivo quantitation of the rat liver’s ability to eliminate endotoxin from portal vein blood. J. Reticuloendothel. Soc. 32:409 – 422; 1982. 6. Babior, B. M. Oxygen-dependent microbial killing by phagocytes. New Eng. J. Med. 298:659 – 668; 1978. 7. Klebanoff, S. J. Oxygen metabolism and the toxic properties of phagocytes. Ann. Intern. Med. 93:480 – 489; 1980. 8. Redmond, H. P.; Shou, J.; Gallagher, H. J.; Kelly, C. J.; Daly, J. M. Macrophage-dependent candidacidal mechanisms in the murine system. J. Immunol. 150:3427–3433; 1993. 9. Sasada, M.; Johnston, R. B. Macrophage microbicidal activity: correlation between phagocytosis-associated oxidative metabolism and the killing of candida by macrophages. J. Exp. Med. 152:85– 98; 1980. 10. Takao, S.; Smith, E. H.; Wang, D.; Chan, C. K.; Bulkley, G. B.; Klein, A. S. The role of reactive oxygen metabolites in murine peritoneal macrophage phagocytosis and phagocytic killing. Am. J. Physiol. 40:C1278 –C1284; 1996. 11. Ellis, J. A.; Mayer, S. J.; Jones, O. T. G. The effect of the NADPH oxidase inhibitor diphenyleneiodonium on aerobic and anaerobic microbicidal activities of human neutrophils. Biochem. J. 251: 887– 891; 1988. 12. Hancock, J. T.; Jones, O. T. G. The inhibition by diphenyleneiodonium and its analogs of superoxide generation by macrophages. Biochem. J. 242:103–107; 1987. 13. Massey, V.; Komai, H.; Palmer, G. On the mechanism of inactivation of xanthine oxidase by allopurinol and other pyrazolo[3,4d]pyrimidines. J. Biol. Chem. 245:2837–2844; 1970. 14. Crofton, R. W.; Diesselhoff, M. M. C.; van Furth, R. The origin, kinetics, and characteristics of the Kupffer cells in the normal steady state. J. Exp. Med. 148:1–17; 1978.

Kupffer cell phagocytic killing 15. Lasser, A. The mononuclear phagocytic system: a review. Hum. Pathol. 14:108 –126; 1983. 16. Bhatnagar, R.; Schirmer, R.; Ernst, M.; Decker, K. Superoxide release by zymosan-stimulated rat Kupffer cells in vitro. Eur. J. Biochem. 119:171–175; 1981. 17. Arthur, M. J. P.; Kowalski–Saunders, P.; Wright, R. Effect of endotoxin on release of reactive oxygen intermediates by rat hepatic macrophages. Gastroenterology 95:1588 –1594; 1988. 18. Uehara, K.; Maruyama, N.; Huang, C.; Nakano, M. The first application of a chemiluminescence probe, 2-methyl-6-[p-methoxyphenyl]-3,7-dihydroimidazo[1,2-a]pyrazin-3-one (MCLA), for detecting O2- production, in vitro, from Kupffer cells stimulated by phorbol myristate acetate. FEBS Lett. 335:167–170; 1993. 19. Laskin, D. L.; Sirak, A. A.; Pilaro, A. M.; Laskin, J. D. Functional and biochemical properties of rat Kupffer cells and peritoneal macrophages. J. Leukoc. Biol. 44:71–78; 1988. 20. Wisse, E. On the fine structure and function of rat liver Kupffer cells. In: Carr, I.; Daems, W. T., eds. The reticuloendothelial system: a comprehensive treatise. Vol. 1: Morphology. New York: Plenum Press; 1980; 361–379. 21. Pruzanski, W.; Saito, S. Comparative study of phagocytosis and intracellular bactericidal activity of human monocytes and polymorphonuclear cells: application of fluorochrome and extracellular quenching technique. Inflammation 12:87–97; 1988. 22. Seglen, P. O. Preparation of isolated rat liver cells. Methods Cell Biol. 13:29 – 83; 1976. 23. Nelson, S. K.; Bose, S. K.; McCord, J. M. The toxicity of highdose superoxide dismutase suggests that superoxide can both initiate and terminate lipid peroxidation in the reperfused heart. Free Radical Biol. Med. 16:195–200; 1994. 24. Omar, B. A.; Gad, N. M.; Jordan, M. C.; Striplin, S. P.; Russell, W. J.; Downey, J. M.; McCord, J. M. Cardioprotection by Cu,Znsuperoxide dismutase is lost at high doses in the reoxygenated heart. Free Radical Biol. Med. 9:465– 471; 1990. 25. Silverstein, S. C.; Steinman, R. M.; Cohn, Z. A. Endocytosis. Annu. Rev. Biochem. 46:669 –722; 1977. 26. Toth, C. A.; Thomas, P. Liver endocytosis and Kupffer cells. Hepatology 16:255–266; 1992. 27. Girotti, A. W.; Thomas, J. P.; Jordan, J. E. Xanthine oxidasecatalyzed crosslinking of cell membrane proteins. Arch. Biochem. Biophys. 251:639 – 653; 1986. 28. Davies, K. J. A. Protein damage and degradation by oxygen radicals. J. Biol. Chem. 262:9895–9901; 1987. 29. Brummer, E.; Stevens, D. A. Candidacidal mechanisms of peritoneal macrophages activated with lymphokines or g-interferon. J. Med. Microbiol. 28:173–181; 1989. 30. Griffin, F. M.; Bianco, C.; Silverstein, S. C. Characterization of the macrophage receptor for complement and demonstration of its functional independence from the receptor for the Fc portion of immunoglobulin G. J. Exp. Med. 141:1269 –1277; 1975. 31. Bianco, C.; Griffin, F. M.; Silverstein, S. C. Studies of the macrophage complement receptor: alteration of receptor function upon macrophage activation. J. Exp. Med. 141:1278 –1290; 1975. 32. Cardarelli, P. M.; Blumenstock, F. A.; McKeown–Longo, P. J.; Saba, T. M.; Mazurkiewicz, J. E.; Dias, J. A. High-affinity binding of fibronectin to cultured Kupffer cells. J. Leukoc. Biol. 48:426 – 437; 1990. 33. Rieder, H.; Ramadori, G.; Dienes, H.; Meyer zum Buschenfelde, K. Sinusoidal endothelial cells from guinea pig liver synthesize and secrete cellular fibronectin in vitro. Hepatology 7:856 – 864; 1987. 34. Cardarelli, P. M.; Blumenstock, F. A.; Saba, T. M.; Rourke, F. J. Fibronectin-enhanced attachment of gelatin-coated erythrocytes to isolated hepatic Kupffer cells. J. Leukoc. Biol. 36:477– 492; 1984. 35. Vickers, S.; Hildreth, J.; Kuhajda, F.; Madara, P.; Mather, I.; Baig, M.; Bulkley, G. Immunoaffinity localization of xanthine oxidase in the microvascular endothelial cells of porcine and human organs. Circ. Shock 31:45; 1990. 36. Steffan, A.; Gendrault, J.; McCuskey, R. S.; McCuskey, P. A.; Kirn, A. Phagocytosis, an unrecognized property of murine endothelial liver cells. Hepatology 6:830 – 836; 1986.

1227

37. Yamamoto, K.; Johnston, R. B. Dissociation of phagocytosis from stimulation of the oxidative metabolic burst in macrophages. J. Exp. Med. 159:405– 416; 1984. 38. Rossi, F. The O2-forming NADPH oxidase of the phagocytes: nature, mechanisms of activation and function. Biochim. Biophys. Acta 853:65– 89; 1986. 39. Jarasch, E. D.; Bruder, G.; Heid, H. W. Significance of xanthine oxidase in capillary endothelial cells. Acta Physiol. Scand. Suppl. 548:39 – 46; 1986. 40. Bruder, G.; Heid, H. W.; Jarasch, E. D.; Mather, I. H. Immunological identification and determination of xanthine oxidase in cells and tissues. Differentiation 23:218 –225; 1983. 41. Wiezorek, J. S.; Brown, D. H.; Kupperman, D. E.; Brass, C. A. Rapid conversion to high xanthine oxidase activity in viable Kupffer cells during hypoxia. J. Clin. Invest. 94:2224 –2230; 1994. 42. Moorhouse, P. C.; Grootveld, M.; Halliwell, B.; Quinlan, J. G.; Gutteridge, J. M. C. Allopurinol and oxypurinol are hydroxyl radical scavengers. FEBS Lett. 213:23–28; 1987. 43. Das, D. K.; Engelman, R. E.; Clement, R.; Otani, H.; Prasad, M. R.; Rao, P. S. Role of xanthine oxidase inhibitor as free radical scavenger: a novel mechanism of action of allopurinol and oxypurinol in myocardial salvage. Biochem. Biophys. Res. Commun. 148:314 –319; 1987. 44. Zimmerman, B. J.; Parks, D. A.; Grisham, M. B.; Granger, D. N. Allopurinol does not enhance antioxidant properties of extracellular fluid. Am. J. Physiol. 255:H202–H206; 1988. 45. Kurose, I.; Miura, S.; Fukumura, D.; Yonei, Y.; Saito, H.; Tada, S.; Suematsu, M.; Tsuchiya, M. Nitric oxide mediates Kupffer cellinduced reduction of mitochondrial energization in hepatoma cells: a comparison with oxidative burst. Cancer Res. 53:2676 –2682; 1993. 46. Liew, F. Y.; Millott, S.; Parkinson, C.; Palmer, R. M. J.; Moncada, S. Macrophage killing of Leishmania parasite in vivo is mediated by nitric oxide from L-arginine. J. Immunol. 144:4794 – 4797; 1990. 47. Granger, D. L. Macrophage production of nitrogen oxides in host defense against microorganisms. Res. Immunol. 142:570 –572; 1991. 48. Stuehr, D. J.; Nathan, C. F. Nitric oxide: a macrophage product responsible for cytostasis and respiratory inhibition in tumor target cells. J. Exp. Med. 169:1543–1555; 1989. 49. Beckman, J. S.; Beckman, T. W.; Chen, J.; Marshall, P. A.; Freeman, B. A. Apparent hydroxyl radical production by peroxynitrite: implications for endothelial injury from nitric oxide and superoxide. Proc. Natl. Acad. Sci. USA 87:1620 –1624; 1990. 50. Radi, R.; Beckman, J. S.; Bush, K. M.; Freeman, B. A. Peroxynitrite oxidation of sulfhydryls: the cytotoxic potential of superoxide and nitric oxide. J. Biol. Chem. 266:4244 – 4250; 1991. 51. Stuehr, D. J.; Fasehun, O. A.; Kwon, N. S.; Gross, S. S.; Gonzalez, J. A.; Levi, R.; Nathan, C. F. Inhibition of macrophage and endothelial cell nitric oxide synthase by dipheyleneiodonium and its analogs. FASEB J. 5:98 –103; 1991. 52. Green, S. J.; Meltzer, M. S.; Hibbs, J. B.; Nacy, C. A. Activated macrophages destroy intracellular Leishmania major ammastagotes by an L-arginine-dependent killing mechanism. J.Immunol. 144:278 –283; 1990. 53. Gaillard, T.; Mulsch, A.; Busse, R.; Klein, H.; Decker, K. Regulation of nitric oxide production by stimulated rat Kupffer cells. Pathobiology 59:280 –283; 1991.

ABBREVIATIONS

DPI— diphenyleneiodonium ROMs—reactive oxygen metabolites SOD—superoxide dismutase NOS—nitric oxide synthase