Eur. J. Biochem. 229, 349-355 (1995) 0 FEBS 1995

Hepatocyte-derived interleukin-6 and tumor-necrosis factor a mediate the lipopolysaccharide-induced acute-phase response and nitric oxide release by cultured rat hepatocytes Bashar SAAD’, Karl FREI’, Florence A. SCHOLL’, Adriano FONTANA’ and Peter MAIER’

* Section of Clinical Immunology, University Hospital, Zurich, Switzerland

(Received I February 1995) - EJB 95 0148/1

Institute of Toxicology, Swiss Federal Institute of Technology and University of Zurich, Switzerland

The regulation of acute-phase protein production and nitric oxide (NO) release in lipopolysaccharide- induced liver injury is thought to occur in response to monocytes/macrophages and Kupffer-cell-derived cytokines. In this study, we used primary cultured rat hepatocytes maintained as a differentiated phenotype to investigate the direct effects of endotoxin (lipopolysaccharide) on the production of the acute-phase proteins and on NO release.

Lipopolysaccharide (10 pg/ml) increased the production of a,-macroglobulin 2.5-fold compared to untreated cultures and decreased the production of albumin by 50%. The effect of lipopolysaccharide was mimicked by adding interleukin-6 (IL-6) and tumor-necrosis factor a (TNF-a), cytokines being induced by treatment of hepatocytes with lipopolysaccharide. Maximal TNF-a (600 pg/ml) and IL-6 (1800 pg/ml) concentrations were observed 4 h and 6 h after lipopolysaccharide stimulation, respectively. The lipopoly- saccharide-induced acute-phase protein response was blocked by anti-(IL-6) but not by anti-(TNF-a) IgG. The latter reduced the lipopolysaccharide-induced IL-6 production by 60%. Besides its effects on the acute-phase proteins, endotoxin caused a significant increase in NO production in cultured rat hepatocytes. Unlike anti-(IL-6) IgG, anti-(TNF-a) IgG reduced the lipopolysaccharide-induced NO production by 50 % indicating that endotoxin-induced NO production is partially mediated by TNF-a but not by IL-6. Precul- ture with gadolinium chloride (GdCl,), an inhibitor of Kupffer cells, did not change the response of hepatocytes to lipopolysaccharide indicating that the observed findings are direct endotoxin effects on hepatocytes. The data demonstrate that by their production of TNF-a and IL-6 rat hepatocytes respond to lipopolysaccharide treatment with an IL-6 mediated acute-phase protein and a TNF-a-mediated NO production. These features have previously been attributed to monocytes/macrophages and Kupffer cells.

Keywords. Lipopolysaccharides ; acute-phase proteins ; hepatocytes ; interleukin-6 ; tumor-necrosis factor a.

Lipopolysaccharides (LPS) present in the outer membranes of Gram-negative bacteria are known to be causally involved in the pathogenesis of septic shock and multiorgan failure (Mor- rison and Rayan, 1987). A characteristic aspect of the LPS-in- duced acute-phase response is a dramatic alteration in the he- patic synthesis of a group of plasma proteins known as acute- phase proteins (Fey and Gauldie, 1990). Many members of this group play important protective roles in the host defence against tissue damage and infections. In the LPS-induced liver injury, activated hepatic Kupffer cells play a key role (Toth and Thomas, 1992; Decker, 1990). The acute-phase response is reg- ulated by cytokines released by activated monocytes/macro- phages and Kupffer cells, notably interleukin-1 (IL-1), interleu- kin-6 (IL-6), and tumor-necrosis factor a (TNF-a) (Darlington et al., 1986; Callery et al., 1992). Among these cytokines, IL-6 is also a major inducer of the acute-phase response in human liver. IL-6 is produced not only by monocytes/macrophages and

Correspondence to P. Maier, Institute of Toxicology, Schorenstr. 16, CH-8603 Schwerzenbach, Switzerland

Fax: +41 1 825 04 76. Abbreviations. GalN, D-gdaCtOSamine ; LDH, lactate dehydroge-

nase; LPS, lipopolysaccharide(s) ; TNF-a, tumor-necrosis factor a; IL, interleukin; IFN, interferon.

Kupffer cells but also by other cell types including fibroblasts, keratinocytes, endothelial cells, mesangial cells, chondrocytes, astrocytes, T-cells and B-cells. Furthermore, Lotz et al. (1989) demonstrated that human hepatoma cells (HepG2) express IL-6 mRNA and secrete IL-6. The production was increased after stimulation with IL-1 or TNF-a. Hepatocytes also produce nitric oxide (NO) during chronic hepatic inflammation (Billiar et al., 1990) and in vitro in response to conditioned medium from acti- vated Kupffer cells (Curran et al., 1989) or in response to a mixture of LPS and TNF-a, IL-1 and IFN-y (Curran et al., 1990; Geller et al., 1993). Recently, Ohno and Maier (1995) showed that TNF-a alone can induce the production of NO in primary rat hepatocytes. In the liver, TNF-a production is not restricted to Kupffer cells. Using in situ hybridization and immunochemi- cal methods, Hunt et al. (1992) showed that TNF-a mFWA and protein were detectable in normal mouse hepatocytes.

The observations that (a) human hepatoma cells and primary mouse hepatocytes have the ability to produce IL-6 and TNF-a and (b) that rat hepatocytes have a high-affinity lectin-like recep- tor for the LPS inner core region on their plasma membrane (Parent, 1990), suggest that LPS can regulate the production of acute-phase proteins and NO production directly, independently

350 Saad et al. (Eul: J. Biochem. 229)

of Kupffer cells. Results obtained in the present study indicate that LPS affects the acute-phase response and the NO production via hepatocyte-derived IL-6 and TNF-a in an autocrine or para- crine loop.

MATERIALS AND METHODS

Antibodies, LPS and cytokines. Rabbit anti-(human IL-6) IgG and rabbit anti-(mouse TNF-a) IgG were obtained from Genzyme, peroxidase-conjugated rabbit anti-(rat albumin) IgG and Cy-3-conjugated goat anti-(mouse) IgG were purchased from Cappel, and rabbit anti-(rat a,-macroglobulin) IgG were a gift from Dr W. Ruch (Sandoz Forschungsinstitut, Berne, Swit- zerland). Mouse anti-(rat macrophage) IgG (KU-1) were a gift from Dr J. S. Reichner (Rhode Island Hospital and Brown Uni- versity).

Lipopolysaccharides from Escherichia coli (026 : B6, trichl- oroacetic acid extract) were obtained from Sigma. Recombinant human interleukin-6 (IL-6) was obtained from Genzyme. Re- combinant human interleukin-I0 (IL-10) was obtained from PeproTech. Inc. Recombinant rat interferon-? (EN-?) and re- combinant human TNF-a were obtained from Hoffman LaRoche (Basel, Switzerland).

Animals and hepatocyte cultures. For purification of hepa- tocytes, 8-week-old male Sprague-Dawley rats (ZUWSIV, Insti- tute of Laboratory Animal Sciences, University of Zurich, Swit- zerland, 240-270 g body mass, standard diet no. 890 Nafag, Gossau, Switzerland, water ad libitum) were used. The prepara- tion of the cells included a standardised two-step collagenase perfusion technique (Maier et al., 1994). The viability of the cells as assessed by the trypan-blue-exclusion test was more than 85%. The enriched hepatocytes were suspended in Williams E medium (BioConcept) supplemented with 100 nM dexametha- sone (Sigma), 10 nM insulin (Sigma), 30 nM selenium (Fluka), 1 pg/ml aprotinin (Sigma) and 2 mM L-glutamine (Gibco), 100 IU/ml streptomycin/penicillin in the absence of phenol red (culture medium). The cells were plated on crude membrane fractionkollagen type I (100: 1 by mass) coated tissue culture dishes (Petriperm) at a density of 3X106 cells/20 cmz (Saad et al., 1993a, b), in 3 ml culture medium, and incubated in an at- mosphere of 13% oxygen and 5% CO, at 37°C (Maier et al., 1994). After 4 h, the cells were washed and fresh medium was added. At 24 h, the cells were stimulated with LPS (10 pg/ml), IL-6 (100 ng/ml), IFF-y (100 U/ml), TNF-a (75 ng/ml, 500 U/ ml), IL-10 (20 ng/ml), anti-(IL-6) IgG (5 pg/ml), anti-(TNF-a) IgG (1 : 50 ; corresponding to 2X lo4 neutralizing units) or vari- ous combinations of LPS and cytokines diluted with fresh me- dium without dexamethasone. In the case of 48-h exposures, the medium including the corresponding LPS or cytokines was replaced after 24 h. The cytokine and LPS concentrations chosen showed the highest response in initial control experiments. Kupffer cells were isolated from rats as described by Inoue and Yamashita (1993) and cultured in hepatocyte culture medium supplemented with 10% fetal calf serum.

Treatment with gadolinium chloride (GdCl,) and D-ga- lactosamine (GalN). 4 h after cell plating, rat hepatocytes and Kupffer cells were incubated with 10 pg GdClJml. After 12 h, the culture medium was exchanged to include LPS instead of GdC1,. Hepatocyte and Kupffer cell cultures were treated for 48 h with 4 mM GalN (Gross et al., 1990) in the absence and presence of 10 pg/ml LPS.

Lactate dehydrogenase activity (LDH). The LDH activity in cell homogenates and in culture supernatants was determined spectrophotometrically (Cobas Fara) with a commercially avail- able kit (Boehringer Mannheim) following the manufacturer's

instructions. Protein concentrations were determined according to Bradford (1976) using BSA as standard.

Quantification of albumin secretion. The amount of albu- min in the culture supernatant was measured using an enzyme- linked immunoabsorbant assay (ELISA) as described previously (Saad et al., 1993b).

Measurements of a,-macroglobulin secretion. The amount of a,-macroglobulin in the culture supernatant was measured using an enzyme-linked immunoabsorbant assay (ELISA). Su- pernatants were diluted 1 :600 with NaCVP, (2.7 mM KC1, 1.5 mM KH2P04, 7.2 mM Na,HPO,, 137 mM NaC1, pH 7.4) and incubated in 96-well microtiter plates (Nunc) overnight at 4°C. After washing in NaCVP,, non-specific binding sites were blocked in NaCl/P, containing 0.5% BSA for 1 h at room tem- perature. After another washing step in NaClP,, rabbit-anti-(rat a,-macroglobulin) IgG (1 :500 dilution) was added in 100 pl NaCVP, containing 2% BSA (Sigma) for 2 h at room temper- ature. The microtiter plates were then washed and the second, alkaline-phosphatase-conjugated goat-(anti-rabbit) IgG (diluted 1 :5000) was added in 100 pl NaCVP, containing 2% BSA for 2 h at room temperature. After four washing steps, 100 p1 sub- strate Ip-nitrophenyl phosphate (Fluka) at 2 mg/ml in 0.1 M gly- cine, pH 10.4, containing 1 mM MgC1, and 1 mM ZnCl,] was added and the absorbance at 405 nm was measured in an ELISA reader. All washing steps were carried out with NaCVP, at room temperature. Background values were measured in the absence of primary antibody and subtracted from the experimental val- ues. All ELISA determinations were carried out in duplicate.

Determination of nitrite release. Nitrite determinations were made on 50-p1 aliquots of the culture supernatant mixed with 200 pl Griess reagent (Ding et al., 1988). The absorbance was measured at 540 nm after 10 min of reaction and NO; con- centration was determined with reference to a standard curve using 1-250 pM sodium nitrite in the culture medium. The ad- dition of a known concentration of sodium nitrite (0.5 mM) to hepatocyte cultures for 24 h resulted in a 70-74% recovery. Therefore, the nitrite concentrations determined represent an un- derestimation of the total NO synthesized.

Quantification of IL-6 production. The presence of IL-6 in supernatants of cultured hepatocytes was assessed by bioassay using an IL-6-dependent cell-line (7TD1) as previously de- scribed (Frei et al., 1988). The detection limit in the culture supernatants was 2 pg/ml.

Quantification of TNF-a production. TNF-a levels were determined in a TNF-a-specific bioassay using the WEHI cell line as previously described (Espevik and Nissen-Meyer, 1986). WEHI 164 clone 13 fibrosarcoma cells at 2X104 cells/l00 pl were incubated with serially diluted samples in a 96-well flat- bottom microtiter plate for 48 h at 37"C, 5% CO,. 10 p1 5 mg/ ml 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H-tetrazo~um bro- mide in NaCLIP, was added to the plate, which was further incu- bated for 4 h. The dye was removed and cells were lysed by addition of 100 p1 isopropanol/5 % formic acid. The absorbance was measured at 620 nm on a multiscan bichromatic ELISA reader (Flow Labs). The detection limit in the culture super- natants was 0.3 pg/ml.

Northern-blot analysis. Total RNA was extracted directly from the hepatocyte monolayer in 4 M guanidinium thiocyanate followed by phenollchloroform extraction according to Chom- czynski and Sacchi (1987). RNA was quantified by measuring the absorbance at 260 nm. Samples containing 7 pg RNA were fractionated on 0.9 % agarose/formaldehyde gels, transferred to Hybond-N filters (Amersham), and hybridized with 2X lo6 dpm/ ml 32P-labelled mouse TNF-a cDNA probe (Fransen et al., 1985) and mouse IL-6 cDNA probe (Chiu et al., 1988). Filters were washed at high stringency (15 mM NaC1, 1.5 mM sodium

Saad et al. ( E m J. Biochem. 229) 351

Fig.1. Immunofluorescence staining with KU-1 antibodies of cultured hepatocytes (A) or Kupffer cells (C) after three days in culture. B and D are the corresponding phase contrast micrographs to fluorescence images A and B, respectively. Magnifications were X 1000.

citrate, 0.1 % SDS, pH 6.5, 42°C) and exposed for autoradio- graphy for 1 day using intensifying screens.

Indirect immunofluorescence. Culture dishes were washed in NaCVP,, blocked in culture medium containing 10 % fetal calf serum, and incubated with the mouse anti-(rat macrophage) IgG (KU-1) for 2 h at room temperature in culture medium contain- ing 10% fetal calf serum. After washing in NaCUP,, Cy-3-conju- gated goat IgG binding to rabbit IgG was used to visualize the primary antibodies.

RESULTS

To evaluate whether hepatocytes can be affected directly in their expression of acute-phase response by endotoxin (LPS), primary rat hepatocytes cultured on crude liver membrane frac- tions/collagen in serum-free culture medium were used. The iso- lation procedure (3x20 g, 2 min each, 4°C) resulted in hepato- cyte fractions that were nearly 100% pure when examined by microscopic analysis and by indirect immunofluorescence with KU-1 antibodies reacting specifically with Kupffer cells (Fig. 1). In addition, the specificity of LPS effects on hepatocytes was measured in hepatocytes either treated with GdC13, an inhibitor of Kupffer cell functions, or GalN, a specific inhibitor of protein synthesis in hepatocytes.

The effect of LPS and cytokines on the production of a,- macroglobulin and albumin were quantified by ELISA. We chose those cytokines that are known to be of importance in the liver during the acute-phase response, namely IFN- y, TNF-a and

The total cellular protein content of cultured hepatocytes, an indirect estimate for the number of attached viable cells, and the extracellular LDH activity relative to the total LDH activity (intracellular and extracellular), a marker for plasma membrane integrity, were not affected by the LPS treatment, indicating that no significant cytotoxicity was observed following incubation with 10 pg LPS/ml (Fig. 2A). Likewise, higher concentrations of LPS (up to 100 pg/ml) also failed to lead to significant LDH release (data not shown). However, in combination, LPS treat- ment of hepatocytes with TNF-a, IFN-y , or both TNF-a and IFN-y produced a 2.5-fold higher LDH release. Treatment of

IL-6.

h

c E: 21

Fig. 2. Cytotoxic activity of LPS, cytokines or combinations. LDH release in the culture supernatants relative to total LDH (intracellular + extracellular, 24-h interval) was measured in cultured rat hepatocytes (1.5X105/cm2), 48 h after stimulation with various stimuli (A) or after treatment with GdCl, or GalN (B). The values represent the mean 2 SEM of four independent experiments carried out in duplicate.

hepatocyte cultures with GdCl, had no effect on LDH release. In contrast, the LDH release was threefold higher after treatment of hepatocytes with LPS in the presence of GalN compared to the cells treated with GalN in the absence of LPS (Fig. 2B). These results indicate that hepatocytes treated with GalN were more sensitive to the toxic effect of LPS.

Effects of LPS on NO production by cultured hepatocytes. Fig. 3A shows the time course of NO release into the culture

352 Saad et al. (EUK J. Biochem. 229)

16 T A1

- NGMA

0 8 24 48 LPS treatment (h)

n

Y f

Y f

Fig.3. Time course of LPS-induced (10 pg LPS/ml) NO release in cultured rat hepatocytes in the absence and presence of NG-mono- mehtyl-L-arginine (NGMA) (A), and the effects of LPS on GdCI, pre- treated hepatocytes (B) and Kupffer cells (C). Hepatocytes and Kupf- fer cells were treated with 10 pg GdClJml for 12 h and further incubated with 10 pg LPS/ml. Nitrite production was determined after 8, 24, and 48 h (24-h interval) (A) or after 48 h (B and C) using the Griess reaction. The values represent the mean? SEM from three independent experi- ments carried out in duplicate.

supernatant of LPS-treated hepatocytes. Increased levels of ni- trite, a stable oxidative metabolite of NO, were already observed 8 h after LPS stimulation, with further increases to 3.5 pM at 24 h and to 14 pM at 48 h (24-h interval). Therefore, the 48 h time point was used for the following experiments designed to characterize the effects of LPS on NO production by cultured hepatocytes.

After treatment with 1 mM NG-monomethyl-L-arginine, a competitive inhibitor of the nitric oxide synthetase (Fig. 3 A), or with GalN (Fig. 3B), NO production of the hepatocytes was completely inhibited. However, GdCI, treatment failed to inhibit the LPS-induced NO production by cultured hepatocytes (Fig. 3B). The same concentrations of GdCI, inhibited com- pletely the NO production from cultured Kupffer cells after LPS stimulation whereas the hepatocyte-specific protein synthesis in- hibitor GalN was ineffective (Fig. 3C). Anti-(IL-6) IgG had no effects on the LPS-induced NO production (Fig. 4A). These re- sults indicate that LPS-induced NO release is a direct effect of LPS on hepatocytes, and is independent of Kupffer cells or IL-6. In contrast, anti-(TNF-a) IgG reduced the LPS-induced NO production by 60% (Fig. 4A). These results indicate that the LPS-induced NO production may at least in part be medi- ated by TNF-a. This is supported by the fact that the addition of TNF-a to hepatocytes leads to enhanced NO levels (Fig. 4B).

Synergistic effects were observed after treatment with com- binations of LPS with either TNF-a or IFN-y or TNF-a and

Fig. 4. NO production in 106 hepatocytes/ml 48 h after stimulation with LPS in the absence and presence of TNF-a and IL-6 neutraliz- ing antibodies (A) and after stimulation with various inducers (B). The values are from three independent experiments carried out in dupli- cate. The values given represent the mean 2 SEM.

Table 1. Inhibitory effects of GdCI, and IL-10 on the LPS-induced TNF-a, IL-6, and nitrite release by cultured rat hepatocytes and Kupffer cells. The concentrations of TNF-a, IL-6, and NO were deter- mined in the culture supernatants after 4, 6, and 48 h, respectively, after treatment with LPS, GdCl,/LPS and LPSIIL-10, beginning 16 h after seeding (n.d., not detectable). Results represent mean values of three independent experiments.

Treatment Concentration of Cell type

TNF-a IL-6 NO

pg/106 cells PM

Hepatocy tes LPS 600 1800 14 GdCIJLPS 550 1800 14 LPSIIL-10 250 450 14

Kupffer cells LPS 2200 3500 12 GdCl,/LPS <1 <2 <0.3 LPS/IL-10 n.d. n.d. 3

IFN-y (Fig. 4B). Based on the elevated level of NO, the stimuli can be ranked in potency for the induction of NO release. For single stimuli, the order of decreasing potency was LPS, TNF-a, and IFN-y. LPS alone increased the NO release by about 45-fold. For the combination of stimuli (LPS + TNF-a + IFN- y , LPS + TNF-a, and LPS + IFN-y), LPS + TNF-a + IFN-y was the most effective combination producing a 400-fold higher NO release compared to untreated control cells. Unlike in Kupf- fer cells, IL-10 failed to inhibit the LPS-induced NO release in cultured rat hepatocytes even at 20 ng/ml (Table 1).

Effects of LPS on the acute-phase response of cultured hepa- tocytes. Hepatocytes cultured on liver crude membrane fraction/ collagen matrix under serum-free culture conditions were ex-

Saad et al. ( E m J. Biochem. 229) 353

P 4

2000 - 1800 - 1600 -

0 2 4 6 8 1 2 2 4 4 8

LPS treatment (h)

Fig. 5. Albumin (A) and a,-macroglobulin (B) secretion in cultured rat hepatocytes after 48 h (24-h interval) stimulation with 10 pg LPS/ml in the presence or absence of IL-6-neutralizing or TNF-a- neutralizing antibodies. The values given represent the mean 2 SEM of four independent experiments carried out in duplicate.

posed to 10 pg LPS/ml and the amount of albumin and a,-mac- roglobulin secreted into the culture medium was measured by ELISA using specific antibodies. Albumin secretion was de- creased by 50% after 2 days of LPS treatment when compared with control hepatocytes. This decrease was inhibited by the ad- dition of neutralizing anti-(IL-6) IgG but not by neutralizing anti-(TNF-a) IgG (Fig. 5A). GdC1, pretreatment failed to inhibit the LPS effect on albumin secretion (data not shown).

The effects of LPS on a,-macroglobulin production by cul- tured hepatocytes showed a similar pattern as with interleukin-6 and TNF-a (Fig. 5B). On day 2, a,-macroglobulin secretion by LPS-treated cells was elevated 2.5-fold when compared with that by untreated cells. Anti-(IL-6) but not anti-(TNF-a) IgG were able to block the LPS-induced a,-macroglobulin produc- tion by cultured hepatocytes (Fig. 5B). These results suggest that LPS regulates the albumin and a,-macroglobulin production via hepatocyte-derived and most likely TNF-a-mediated IL-6 re- lease.

Production of IL-6 and TNF-a by LPS-stimulated cultured rat hepatocytes. The production of IL-6 and TNF-a by cultured rat hepatocytes was tested in the culture supernatants using sen- sitive bioassays. It was found that hepatocytes produce detect- able amounts of IL-6 and TNF-a after stimulation with LPS. Fig. 6 A shows the time course of IL-6 production into the cul- ture supernatant by LPS-treated hepatocytes. IL-6 was detectable in the culture supernatants 4 h after stimulation, and reached a maximum at 6 h (1800 pg/ml) before decreasing at 24 h (650 pg/ ml). LPS treatment in the presence of neutralizing anti-(TNF-a) IgG reduced the production of IL-6 by 60% after 24 h (Fig. 6B). Pretreatment of hepatocyte cultures with GdC1, failed to inhibit the LPS-induced IL-6 production. In contrast, GdCl, treatment completely inhibited the production of IL-6 from cultured Kupf- fer cells after LPS stimulation (Table 1). Unlike NO production,

Fig. 6. Cytokine release after exposure to LPS. Time course of LPS- induced (10 pg LPS/ml) production of TNF-a and IL-6 was determined in cultured rat hepatocytes (A) by specific bioassays in supernatants from three independent experiments of lo6 hepatocytes/ml for the times indi- cated. The effects of GdCl,, GalN, and TNF-a-neutralizing antibodies on the production of IL-6 (B) were measured 24 h after LPS stimulation. The values represent the mean 5 SEM from three independent experi- ments carried out in duplicate.

GalN treatment of hepatocytes partially inhibited the LPS-in- duced IL-6 (Fig. 6B). This might be due to the fact that GalN differentially affects the protein synthesis of hepatocytes (Gross et al. 1990). These findings indicate that LPS-induced IL-6 pro- duction is independent of Kupffer cells and is mediated at least in part by hepatocyte-derived TNF-a.

The time course of secretion of LPS-treated hepatocytes showed that TNF-a release preceded the IL-6 release. TNF-a was detectable in the culture supernatant 2 h after LPS stimula- tion, became maximal at 4 h (660 pg/ml), decreased slightly by 6 h, and decreased rapidly after 12 h. No TNF-a was found in the supernatant at 24 h and 48 h after LPS stimulation. After preincubation of hepatocyte culture supernatants with neutraliz- ing anti-(IL-6) or anti-(TNF-a) IgG, no IL-6 and TNF-a were detectable in the bioassay, confirming the identity of the bioac- tive molecules with the respective systems.

Northern-blot analysis of IL-6 and TNF-a expression by LPS-stimulated rat hepatocytes. As a further confirmation that hepatocytes could be activated to produce TNF-a and IL-6, the production was analyzed at the mRNA levels using cDNA probes. Whereas control hepatocytes did not express TNF-a and IL-6 mRNA (Fig. 7), treatment with LPS for 1-8 h revealed 1.5-kb (TNF-a) and 1.3-kb (IL-6) messages, respectively, by Northern-blot analysis. When the filters were rehybridized with a /3-actin probe, a single band at 2 kb was visible (Fig. 7C).

354 Saad et al. ( E m J. Biochem. 229)

1.5 kb -

1.3kb-

2.0kb-

A

1 2 3 4 5 6 7

B

1 2 3 4 5 6 7

C

1 2 3 4 5 6 7

Fig. 7. Northern-blot analysis of RNA from untreated hepatocytes (lane 1) or LPS-treated hepatocytes for 1, 2, 3, 4, 6, and 8 h (lanes 2-7). Equal amounts of total RNA (7 pg/lane) were fractionated on aga- rose/formaldehyde gels, transferred to Hybond-N filters and hybridized with 3ZP-labelled in vitro transcripts of the TNF-a cDNA (A), IL-6 cDNA (B), or p-actin (C).

These data indicate that LPS induces the expression of both TNF-a and IL-6 mRNA.

DISCUSSION The results obtained in the present study indicate that hepa-

tocyte-derived IL-6 and TNF-a mediate the LPS-induced alter- ations of a,-macroglobulin and albumin production and that TNF-a partially mediates the LPS-triggered production of nitric oxide (NO) by cultured rat hepatocytes. At the cellular site, the specificity of the effects of LPS on hepatocytes was investigated in several ways. First, the production of IL-6 and TNF-a was not inhibited by treatment of the cultures with GdC13, an inhibitor of Kupffer cells (Husztnik et al., 1977; Koudstaal et al., 1991 ; Cla- vien et al., 1993). The same concentration of GdC1, inhibited completely the production of IL-6, TNF-a, and NO by cultured rat Kupffer cells. Secondly, treatment with LPS in the presence of GalN, an inhibitor of hepatocyte functions (Gross et al., 1990), abolished the observed LPS-induced a - 6 , TNF-a pro- duction, and NO production. In contrast, GalN had no effect on the production levels of IL-6, TNF-a, or NO by cultured rat Kupffer cells. Thirdly, LPS-induced NO production was inhib- ited by IL-10 treatment in Kupffer cells but not in hepatocytes. Fourthly, hepatocytes in the present study were cultured under serum-free culture conditions ; under these conditions, rat Kupf- fer cells and endothelial cells cannot survive (Scholl, F., unpub- lished results). In the present study, this was confirmed with immunochemical staining of cultured hepatocytes using Kupffer cells (KU-1 antibody) and in additional experiments endothelial- cell-specific markers (OX-18 mouse a rat MHC I ; Chemikon; Scholl, F., unpublished results). Fifthly, the induction of IL-6 and TNF-a production was observed by bioassay and Northern- blot analysis. Sixthly, the LPS-induced production of IL-6 and TNF-a on hepatocytes is functional, since antibodies reacting specifically with these molecules, were able to reduce the acute- phase response. These observations emphasize the functional significance of the previously unknown role of IL-6 and TNF-a produced by hepatocytes.

The LPS-induced alteration in the production of acute-phase proteins investigated in the present study in hepatocytes appears to involve LPS-regulated expression of IL-6. This could be shown by adding LPS to the cultures in the presence of antibod-

ies against IL-6, which leads to a complete abolishment of the LPS-induced increase of a,-macroglobulin and a decrease of al- bumin production. Furthermore, IL-6 mimics the effect of LPS on acute-phase protein synthesis. Indeed, LPS induced the pro- duction of IL-6 mRNA and protein in hepatocytes. The effective concentrations of the human recombinant cytokines were much higher than the concentration of the released rodent liver cyto- kines. The LPS-induced mRNA of hepatocytes hybridizing to the IL-6 probe displayed a size of 1.3 kb as reported in HepG2 cells and cultured peritoneal macrophages (Lotz et al., 1989).

The presently described effect of LPS on hepatocytes de- monstrates a physiologically relevant, direct effect of LPS on hepatocytes independent of Kupffer cells and macrophages. This appears to be due to the fact that hepatocytes were cultured un- der more physiological conditions. The cells, cultured on liver crude membrane fractiordcollagen (Saad et al., 1993a, b) under serum-free conditions and in combination with a defined tissue- equivalent oxygen tension (Maier et al., 1994), preserved their liver-specific functions, such as albumin production, xenobiotic metabolism, and carbohydrate metabolism, effectively for up to one week (Saad et al., 1994; Ohno and Maier, 1994). When hepatocytes were cultured on collagen-coated culture dishes in- stead of crude membrane fractionlcollagen, no or only very low acute-phase protein production could be measured after LPS treatment and NO production was quantitatively much less pro- nounced (unpublished results).

Besides the production of IL-6, cultured hepatocytes have also been found to secrete TNF-a, which was found to affect the IL-6 and the NO production, but not the acute-phase proteins investigated in the present study. Recently, TNF-a has been shown to enhance the secretion of L -6 by the HepG2 cell line (Lotz et al., 1989) and the production of NO by cultured rat hepatocytes (Ohno and Maier, 1995). These results are con- firmed by our observations that hepatocytes have the capacity to produce TNF-a, which influences the NO production. Immuno- histological staining and in situ hybridization studies showed both TNF-a immunoreactivity and TNF-a mRNA in hepatocytes in the liver of normal mice (Hunt et al., 1992). Interestingly, anti-(TNF-a) IgG affect IL-6 production but not the production of the acute-phase proteins of LPS-treated hepatocytes. Since anti-(TNF-a) IgG did not totally abolish IL-6 production, it is possible that residual amounts of IL-6 remaining in the culture supernatants are high enough to regulate the acute-phase re- sponse. In this context, it is interesting that TNF-a increases the production of IL-6 in hepatocyte cultures. However, the effect of TNF-a is less pronounced than that of LPS (unpublished re- sults). Besides TNF-a, other cytokines may be involved in the LPS-induced IL-6 production.

Previous studies (Biliar et al., 1989) showed that cocultures of rat hepatocytes and Kupffer cells stimulated with LPS pro- duce large amounts of NO. Furthermore, it was demonstrated that cultured hepatocytes also produce NO in response to condi- tioned Kupffer cell supernatant or to a mixture of LPS and the cytokines TNF-a, IL-1, and IFN-y (Curran et al., 1990; Geller et al., 1993). Human hepatocytes were also stimulated to pro- duce NO by the same combination of LPS and cytokines as rat hepatocytes (Nussler et al., 1992). However, no data are reported on the direct induction of NO by LPS. Results obtained in the present study show that in addition to the combination of LPS with TNF-a, and IFN-y (Fig. 4B), LPS alone is able to induce NO production on hepatocytes in the absence of Kupffer cells or other cytokines and that this induction is partially mediated by hepatocyte-derived TNF-a. In contrast to the inducible nitric oxide synthetase form of macrophages, the inducible nitric oxide synthetase of hepatocytes was unaffected by IL-10 treatment although both nitric oxide synthetases have 80% amino acid

Saad et al. (Eul: J . Biochem. 229) 355

sequence similarity (Geller et al., 1993). LPS present in the portal vein system may reach hepatocytes and stimulate NO and TNF-a production, which may alter hepatocyte function. Re- cently, NO was shown to inhibit protein production and mito- chondrial functions of cultured hepatocytes (Ohno and Maier, 1995; Kurose et al., 1993). In viral infections of hepatocytes, NO and TNF-a production by hepatocytes may interfere with viral replication (Karupiah et al., 1993), whereas TNF-a has been shown to inhibit viral replication and to synergize with interferons (Wong and Goedel, 1986). NO synthesis by hepato- cytes may contribute to host defence against a hepatic infection like malaria and to the hepatic dysfunction of sepsis (Mellouk et al., 1991).

In conclusion, besides Kupffer cells hepatocytes may re- spond to LPS with an acute-phase response and NO production, which are most likely mediated by IL-6 and TNF-a. Thus, the liver has at least a dual system of Kupffer cells and hepatocytes to respond to bacterial endotoxins.

The authors gratefully acknowledge the expert technical advice for Northem-blot analysis by Drs H. P. Eugster and A. N. Shakhov. This study was supported by the Swiss National Science Foundation, Beme, Switzerland (grant no. 31-9379.88).

REFERENCES Billiar, T. R., Curran, R. D., Stuehr, D. J., West, M. A., Bentz, B. G. &

Simmons, R. L. (1989) J. Exp. Med. 169, 1467-1472. Billiar, T. R., Curran, R. D., Harbrecht, B. G., Stuehr, D. J., Demetris,

A. J. & Simmons, R. A. (1990) J. Leukocyte Biol. 48, 565-569. Bradford, M. M. (1976) Anal. Biochem. 72, 248-254. Callery, P. M., Kamei, T. & Flye, M. W. (1992) Circ. Shock 37, 185-

Chiu, C. P., Mouldes, C., Coffmann, R. L., Rennick, D. & Lee, F. (1988)

Chomczynski, P. & Sacchi, N. (1987) Anal. Biochem. 162, 156-159. Clavien, P. A., Harvey, P. R., Sanabria, J. R., Cywes, R., Levy, G. A. &

Strasberg, S . M. (1993) Hepatology 17, 131-142. Curran, R. D., Billiar, T. R., Stuehr, D. J., Hofmann, K. & Simmons, R.

L. (1989) J. Exp. Med. 170, 1769-1774. Curran, R. D., Billiar, T. R., Stuehr, D. J., Ochoa, J. B., Habrecht, B. G.,

Flint, S. G. & Simmons, R. L. (1990) Ann. Surg. 212, 462-471. Darlington, G. J., Wilson, D. R. & Lachman, B. (1986) J. Cell Biol. 103,

787-793. Decker, K. (1990) Eul: J. Biochem. 192, 245-261. Ding, A. H., Nathan, C. F. & Stuehr, D. J. (1988) J. Immunol. 141,

188.

Proc. Natl Acad. Sci. USA 85, 7099-7103.

2407 -241 2.

Espevik, T. & Nissen-Meyer, J. (1986) J. Immunol. Methods 95, 99- 105.

Fey, G. & Gauldie J. (1990) in Progress in liver disease (Popper, H. & Schaffner, F., eds) vol. 9, p. 89, W. B. Saunders, Co., Philadelphia.

Fransen, L., Mueller, R., Marmenout, A., Tavemier, J., Van der Heyden, J., Kawashima, E., Chollet, A., Tizard, R., Van Heuverswyn, H., Van Vliet, A., Ruysschaert, M. R. & Fiers, W. (1985) Nucleic Acids Res.

Frei, K., Leist, T. P., Meager, A,, Gallo, P., Leppert, D., Zinkemagel, R. M. & Fontana, A. (1988) J. Exp. Med. 168, 449-453.

Geller, D. A., Nussler, A. K., Di Silvio, M., Lowenstein, C. J., Shapiro, R. A., Wang, S . C., Simmons, R. L. & Billiar, T. (1993) Proc. Narl Acad. Sci. USA 90, 522-526.

Gross, V., Ludolph, D., Vom Berg, D., Kreisel, W., Andus, T., Giffhom- Katz, S., Heinrich, C. P. & Gerok, W. (1990) Biochim. Biophys. Actu

Hunt, J . S . , Chen, H. L., Hu, X. L., Chen, T. Y. & Momson, D. C. (1992) Cytokine 4, 340-346.

Husztnik, E., Lazar, G. & Szilagyi, S . (1977) in Kupffer cells and other liver sinusoidal cells (Wisse, E. & Knook, D. L., eds) pp. 387-390, Elsevier, Amsterdam.

13, 4417-4429.

1036, 143-150.

Inoue, S. & Yamashita, A. (1993) Toxicol. Pathol. 21, 71-80. Karupiah, G., Xie, Q. W., Buller, R. M. L., Nathan, C., Duarte, C. &

Mac Micing, J. D. (1993) Science 261, 1445-1448. Koudstaal, J., Dijkhuis, F. W. J. & Hardonk, M. J. (1991) in Cells ofthe

hepatic sinusoids (Wisse, E., Knook, D. L. & McCuskey, R. S., eds) pp. 87-91, Kupffer cells Foundation, Leiden.

Kurose, I., Kato, S., Ishii, H., Fukumura, D., Miura, S., Suematsu, M. & Tsuchiya, M. (1993) Hepatology 18, 380-388.

Lotz, M., Zuraw, B. L., Carson, D. A. & Jirik, F. R. (1989) Ann. N. I: Acad. Sci. 557, 509-511.

Maier, P., Saad, B. & Schawalder, H. P. (1994) Toxicol. In vitro 8, 423- 435.

Mellouk, S . , Green, S. J., Nacy, C. A. & Hoffman, S. L. (1991) J. Imrnu- nol. 146, 3971 -3976.

Momson, D. C. & Rayan, J. C. (1987) Annu. Rev. Med. 38,417-432. Nussler, A. K., di Silvio, M., Billiar, T. R., Hofman, R. A., Geller, D.

A., Selby, R., Madariaga, J. & Simmons, R. L. (1992) J. Exp. Med. 176, 261-266.

Ohno, K. & Maier, P. (1994) J. Cell. Physiol. 160, 358-366. Ohno, K. & Maier, P. (1995) Eul: J . Pharmacol. 292, 205-214. Parent, J. B. (1990) J. Biol. Chem. 265, 345553461, Saad, B., Schawalder, H. P. & Maier, P. (1993a) In vitro Cell. & Dev.

Saad, B., Scholl, F. A., Thomas, H., Schawalder, H. P., Streit, V.,

Saad, B., Thomas, H., Schawalder, H. P., Waechter, F. & Maier, P. (1994)

Toth, C. A. & Thomas, P. (1992) Hepatology 16, 255-266. Wong, G. H. & Goeddel, D. V. (1986) Nature 323,819-822.

Biol. 29A, 32-40.

Waechter, F. & Maier, P. (1993b) Eul: J. Biochem. 213, 805-814.

Toxicol. Appl. Pharmacol. 126, 372-379.