Main

Interleukin-10 (IL-10) is a pleiotropic cytokine with a wide spectrum of biological effects on lymphoid and myeloid cells1,2. One of the known functions of IL-10 is its ability to inhibit the production of pro-inflammatory mediators, including tumor necrosis factor-α (TNF-α), IL -6, IL-1, granulocyte-macrophage colony stimulating factor, and the generation of nitric oxide (NO) by lipopolysaccharide (LPS)-activated monocytes/macrophages. However, the molecular mechanisms underlying the anti-inflammatory effects of this cytokine remain unknown. Earlier studies on IL-10 receptor-signaling pathways have revealed that the JAK1/STAT3-dependent pathway is not sufficient for the anti-inflammatory action3,4,5,6. Although IL-10 activates PI-3-kinase and p70 S6 kinase, these pathways are also not required for the anti-inflammatory effects of the cytokine7. Another elusive point is that the inhibitory activity of IL-10 on LPS-induced inflammation can be blocked by the protein synthesis-inhibitor cycloheximide, suggesting the involvement of newly synthesized protein(s) in its effects8,9.

Heme oxygenase (HO) is the rate-limiting enzyme in heme catabolism, which leads to the generation of biliverdin, free iron and carbon monoxide (CO)10,11,12. Three mammalian HO isoforms have been identified, one of which, HO-1, is a stress-responsive protein induced by various oxidative agents. Over the past decade, HO-1 has been implicated in the cytoprotective defense response against oxidative injury12. In addition to the antioxidant activities of biliverdin and its metabolite, bilirubin, there has been increasing interest in the potential effects of endogenous CO, which, like NO, activates the cGMP pathway and elicits neurotransmission in the central nervous system and vasodilation in the vascular system10. Otterbein et al.13 have shown that CO inhibits the expression of LPS-induced pro-inflammatory cytokines and increases LPS-induced expression of IL-10 in macrophages, suggesting that it is involved in the anti-inflammatory action of HO-1. However, more work is required to delineate the role of HO-1 in inflammation. Here we explored the potential interplay between IL-10 and HO-1 in the inhibition of LPS-induced inflammatory responses. We found IL-10 to be a potent inducer of HO-1 in mouse primary macrophages and the J774 cell line, with induction of HO-1 occurring as early as three hours after IL-10 treatment. Cotreatment with a HO inhibitor or a CO scavenger significantly suppressed the inhibitory effects of IL-10 on LPS-induced TNF-α and NO production, as well as the expression of matrix metalloproteinase-9 (MMP-9) in macrophages. HO-1 induction was also seen in mice receiving IL-10 administration. Furthermore, IL-10-mediated protection against LPS-induced septic shock in mice was decreased by cotreatment with a HO inhibitor, supporting the role of HO-1 in the anti-inflammatory action of IL-10 in vivo.

Induction of HO-1 expression by IL-10 in murine macrophages

To examine whether IL-10 affects HO-1 gene expression, we treated the murine macrophage cell line, J774, for 24 hours with various concentrations of IL-10. Western-blot analysis revealed dose-dependent induction of HO-1 by IL-10 (Fig. 1a). Concentrations of IL-10 as low as 10 ng/ml were capable of inducing maximal expression of HO-1 in this cell line. Using 10 ng/ml of IL-10, induction was evident as early as 3 hours, and reached a maximum after 24 hours of treatment with IL-10 (Fig. 1b). IL-6, which induces STAT-3 phosphorylation of IL-10 (ref. 14), did not induce HO-1 expression at concentrations up to 40 ng/ml (Fig. 1c). The induction of HO-1 by IL-10 was also demonstrated in primary peritoneal macrophages isolated from BALB/c mice (Fig. 1d).

Figure 1: IL-10 induces HO-1 protein expression in murine macrophages.
figure 1

a and b, Western blots of HO-1 and β-actin in J774 cells incubated (a) with the indicated concentrations of IL-10 and (b) with IL-10 for the indicated times. c, J774 cells were incubated with the indicated concentrations of IL-6 or IL-10, and the extent of STAT-3 phosphorylation (20 min) and HO-1 induction (24 h) after cytokine stimulation were determined by Western blot analysis. d, Protein expression of HO-1 and β-actin in mouse peritoneal macrophages incubated with the indicated concentrations of IL-10 or IL-6 for 24 h.

Full size image

p38 activation in IL-10-induced HO-1 gene expression

Northern-blot analysis showed that IL-10 (10 ng/ml) treatment of J774 cells for 12 hours resulted in an increase in HO-1 mRNA, which was completely blocked by the transcription inhibitor, actinomycin D (5 μg/ml) (Fig. 2a). This finding suggests that the induction of HO-1 resulted primarily from transcriptional activation. Cyclohexamide (2 μg/ml), which completely blocks the expression of A8 gene induced by the combination of LPS (1 μg/ml) and IL-10 (10 ng/ml)15, did not have significant effects on IL-10-induced HO-1 gene expression (Fig. 2a). This indicates that the induction of the HO-1 gene did not require new protein synthesis.

Figure 2: Involvement of p38 activation in IL-10-induced HO-1 gene expression.
figure 2

a, Northern blot showing HO-1 mRNA from J774 cells incubated with or without IL-10 in the absence or presence of actinomycin D (Act. D). The effect of cycloheximide (CHX) on the expression of HO-1 mRNA induced by IL-10, or A8 mRNA induced by LPS plus IL-10, was also examined. b, Western blots showing the extent of phosphorylation of ERK, JNK, and p38 MAPKs at the indicated times after IL-10 stimulation of J774 cells or primary peritoneal macrophages. c, p38 phosphorylation in J774 cells incubated with IL-10 alone, or with IL-10 pretreated with IL-10 neutralizing antibody (IL-10 Ab) or control IgG. d, HO-1 mRNA expression in the J744 cells incubated with IL-10 in the absence or presence of indicated concentrations of SB203580 or PD98059, then HO-1 mRNA expression measured. e, HO-1 protein expression in J774 cells incubated with IL-10 in the absence or presence of SB203580 for the indicated times. f, Effect of SB203580 on SOCS-3 protein expression in J774 cells incubated with IL-10.

Full size image

Recently, studies on HO-1 induction by stress stimuli have shown that pathways involving mitogen-activated protein kinases (MAPKs)16 are responsible for the transduction of signals to initiate gene activation17,18,19. To determine whether a similar signal mechanism is responsible for the upregulation of HO-1 gene expression by IL-10 in macrophages, we examined the activation states of three MAPK subfamilies, ERK, JNK and p38, in J774 cells. Treatment with IL-10 (10 ng/ml) resulted in rapid phosphorylation of p38, but not of ERK or JNK (Fig. 2b). p38 phosphorylation reached a peak at 5 minutes, then declined to baseline within 15 minutes. Similar results were observed in primary macrophages upon IL-10 stimulation (Fig. 2b). To exclude the possibility that IL-10 used in our experiments was contaminated with LPS, which has been shown to activate p38 as well as ERK and JNK (ref. 20), we carried out the same experiment using an anti-IL-10 neutralizing antibody. Western-blot analysis showed that the IL-10-induced p38 phosphorylation at 5 min was completely abolished by a 30 min preincubation at 37°C of IL-10 (10 ng/ml) with anti-IL-10 neutralizing antibody (1 μg/ml), but not with control IgG (1 μg/ml) (Fig. 2c). This result indicates that p38 activation is specifically induced by the cytokine. The IL-10-mediated increase in HO-1 mRNA level was completely blocked by SB203580, a specific inhibitor of p38, whereas similar concentrations of PD98059, a specific inhibitor of ERK, had no significant effect (Fig. 2d). The inhibitory effect of SB203580 (10 μM) on HO-1 protein induction was also observed early (3 h) after IL-10 treatment in J774 cells (Fig. 2e). In contrast, SB203580 at the same concentration did not affect the expression of suppressor of cytokine signaling-3 (SOCS-3), a known IL-10-inducible gene4, at 4 hours after cytokine stimulation (Fig. 2f).

HO-1 mediates suppression of LPS-activated TNF-α

Pretreatment of J774 cells with IL-10 significantly suppressed LPS-induced TNF-α production (Fig. 3a). To determine whether HO-1 mediates the inhibitory effect of IL-10, we transfected cells with antisense oligodeoxynucleotides (ODN) complementary to HO-1 mRNA prior to their sequential treatment with IL-10 and LPS. Western-blot analysis revealed that IL-10-induced HO-1 protein expression was substantially reduced by treatment with antisense ODN, but not by sense or scrambled ODN (Fig. 3a). In situ HO-1 immunostaining on transfected cells revealed that the HO-1 expression in 80% of the cells was completely abolished by antisense ODN treatment (data not shown). In parallel, IL-10-mediated suppression of LPS-induced TNF-α production was also significantly attenuated by antisense ODN treatment (Fig. 3a). To ensure that the inhibitory effect of HO-1 antisense ODN is not mediated through the induction of interferon and other cellular genes caused by the double-stranded RNA formed21, we performed additional experiments with antisense ODN to SOCS-3 mRNA and found that SOCS-3 antisense ODN abolished IL-10-induced SOCS-3 expression without affecting the inhibitory effect of IL-10 on LPS-induced TNF-α production (Fig. 3b). This result indicates that the ablation of HO-1 is responsible for the suppression of IL-10 effect by the HO-1 antisense ODN.

Figure 3: HO-1 mediates IL-10-induced suppression of TNF-α production in LPS-activated J774 cells.
figure 3

a, Cells transfected with antisense, sense or scrambled ODN were left untreated (□) or treated (▪) with IL-10 prior to LPS challenge. HO-1 and β-actin protein expression was examined by western blot (upper panel) and the amounts of TNF-α released into culture medium assayed by ELISA (lower panel). *, P < 0.005 compared with control cells treated with IL-10/LPS. b, Cells transfected without (□) or with (▪) SOCS-3 antisense ODN were treated with IL-10 and LPS. The SOCS-3 protein expression and TNF-α production were then determined. c–e, Cells treated with IL-10 (10 ng/ml, 4 h), followed by LPS (1 μg/ml, 2 h) in the absence or presence of ZnPP, CuPP, hemoglobin (Hb) or desferrioxamine (Def.). TNF-α released into culture medium was determined by ELISA. *, P < 0.005 compared with cells treated with IL-10/LPS. f, Cells treated with LPS in the presence or absence of bilirubin. Bilirubin had no effect on LPS-induced TNF-α production.

Full size image

The involvement of HO-1 in the anti-inflammatory action of IL-10 was also confirmed using a specific HO competitive inhibitor, zinc protoporphyrin IX (ZnPP), which at 1 μM significantly blocked IL-10-mediated inhibition of LPS-induced TNF-α production (Fig. 3c). In contrast, copper protoporphyrin IX (CuPP) (1 μM), which does not inhibit HO (ref. 22), was not effective. To determine whether CO or iron released from heme degradation by HO-1 was responsible for the action of IL-10, we examined the effects of hemoglobin (a scavenger of CO) and desferrioxamine (an iron chelator) on the IL-10-mediated inhibition of LPS-induced TNF-α production. We found that hemoglobin, at concentrations ranging from 2.5 μM to 10 μM, effectively and in a dose-dependent manner reversed the effect of IL-10 (Fig. 3d), whereas desferrioxamine (50–200 μM) had no significant effect (Fig. 3e). To determine whether bilirubin, the other degradation product of heme, had anti-inflammatory activity, we treated cells with LPS in the presence or absence of bilirubin (2.5–10 μM) and found that bilirubin did not significantly inhibit LPS-induced TNF-α production (Fig. 3f). These results suggest that CO, derived from heme degradation, mediates the inhibitory effect of IL-10 on TNF-α production. Using HO-1 antisense ODN and ZnPP, we also demonstrated that HO-1 is involved in IL-10-mediated suppression of LPS-induced TNF-α production in primary macrophages (Fig. 4a and b).

Figure 4: HO-1 inhibition reverses IL-10-induced suppression of TNF-α production in LPS-activated primary macrophages.
figure 4

a, Primary peritoneal macrophages were transfected with HO-1 antisense, sense or scrambled ODN, followed by stimulation without (□) or with (▪) IL-10 prior to LPS challenge. HO-1 and β-actin protein expression and TNF-α release are shown. *, P < 0.005 compared with control cells treated with IL-10/LPS. b, Cells were treated with or without IL-10 (10 ng/ml, 4 h), followed by LPS (1 μg/ml, 2 h) in the absence or presence of ZnPP (1 μM). *, P < 0.025 compared with cells treated with IL-10/LPS.

Full size image

HO-1 mediates suppression of NO and MMP-9

LPS-induced expression of inducible nitric oxide synthase (INOS) and production of NO provide important cytotoxic function in macrophages23. IL-10 inhibited the induction of INOS expression by LPS in J774 cells and primary macrophages, and ZnPP (1 μM) and hemoglobin (10 μM), which reversed the inhibition of TNF-α production by IL-10, markedly suppressed the IL-10-mediated inhibition of INOS expression and NO production, whereas desferrioxamine (200 μM) had no effect (Fig. 5a and b). Bilirubin (10 μM), however, did not show any significant effect on LPS-induced INOS expression and NO production. These results suggest that HO-1 and CO mediate the inhibitory effect of IL-10 on LPS-induced INOS expression.

Figure 5: HO-1 mediates IL-10-induced suppression of NO production and MMP-9 expression.
figure 5

a and b, J774 cells and peritoneal macrophages untreated or treated with IL-10 (10 ng/ml, 2 h) were incubated with or without LPS (1 μg/ml, 18 h). The culture medium was then replaced with medium without phenol red, and cells incubated for additional 6 h with or without ZnPP, hemoglobin (Hb), desferrioxamine (Def.) or bilirubin as indicated. HO-1 and INOS protein expression and nitrite accumulation were determined. In a, *, P < 0.005; in b, *, P < 0.05 and **, P < 0.025 compared with cells treated with IL-10/LPS. c, J774 cells treated with or without IL-10 (10 ng/ml, 6 h) were incubated for an additional 6 h in the absence or presence of the indicated agents. The culture medium was subjected to SDS–PAGE zymography to detect the activity of MMPs and a major band with a molecular size close to proMMP-9 was detected. Its identity as proMMP-9 was confirmed by western blotting. The HO-1 protein level in cell lysates was also determined. d, Peritoneal macrophages untreated or treated with IL-10 (10 ng/ml, 1 h) were incubated with or without LPS (1 μg/ml, 5 h), followed by a 6 h incubation in the presence or absence of the indicated agents. The expression of MMP-9 was detected by zymography.

Full size image

We next determined whether HO-1 plays a role in the suppressive effect of IL-10 on the expression of MMPs (refs. 24,25), which are important in the degradation and remodeling of the extracellular matrix at sites of inflammation26. We examined the constitutive expression of MMPs in J774 cells and found that IL-10 inhibited MMP-9 expression in these cells, as shown by both zymography and western blotting (Fig. 5c). This effect was again significantly attenuated by ZnPP and hemoglobin. The expression of MMP-9 in untreated primary macrophages was undetected but upregulated by LPS. Similar to that observed in J774 cells, MMP-9 expression as assessed by zymography in LPS-activated primary macrophages was markedly inhibited by IL-10 (Fig. 5d). ZnPP and hemoglobin again significantly reversed the effect of IL-10, supporting the involvement of HO-1 in this process.

Induction of HO-1 by IL-10 in vivo

We injected BALB/c mice intraperitoneally with IL-10 and measured HO-1 expression in various tissues and cells. A time-dependent increase was observed in peritoneal macrophages, liver and spleen, whereas no significant change was observed in circulating monocytes, lung or kidney up to 24 hours after IL-10 treatment (Fig. 6a and b). To examine whether the unresponsiveness of monocytes resulted from improper stimulation, we treated the fresh monocytes isolated from mouse peripheral blood with IL-10 for 24 hours in culture and found that the HO-1 was not induced under this condition. However, the expression of HO-1 was induced by IL-10 in macrophage-like cells derived from the peripheral monocytes and maintained in culture for one day prior to IL-10 treatment (data not shown). These observations suggest that macrophages respond more profoundly than monocytes to IL-10-induced HO-1 gene expression. When mice received a LPS challenge 3 hours after IL-10 treatment, the survival rate was markedly increased as compared with that of mice receiving LPS alone (80% versus 13% survival; P < 0.001) (Fig. 6c).

Figure 6: HO-1 is essential for IL-10-mediated protection against LPS-induced septic shock.
figure 6

a and b, BALB/c mice were injected intraperitoneally with IL-10. Changes in HO-1 protein expression in various tissues and cells were examined by western blot (a), and quantitative data was obtained by densitometry (b). c and d, Mice received different treatments as indicated and the lethality within 72 h was determined. In c, ▵, control (n = 15); ○, ZnPP (n = 5); ▴, IL-10 (n = 5); ● (with dashed line), LPS (n = 15); ⊗, IL-10/ZnPP (n = 5); ♦, IL-10/LPS (n = 15); ●, LPS/ZnPP (n = 10); ♦ (with dashed line), IL-10/ZnPP/LPS (n = 20); ⋄, IL-10/ZnPP/LPS/CO (n = 15). In d, □, CuPP (n = 5); ♦, LPS (n = 15); ●, IL-10/LPS (n = 15); ▵, IL-10/CuPP (n = 5); ♦ (with dashed line), CuPP/LPS (n = 10); ● (with dashed line), IL-10/CuPP/LPS (n = 10). *, P < 0.005 compared with mice receiving IL-10/LPS; #, P < 0.005 compared with mice receiving IL-10/ZnPP/LPS. e, Serum TNF-α levels in all mice were determined at 2 h after challenge with or without LPS. *, P < 0.01 compared with group treated with IL-10/LPS; #, P < 0.01 compared with group treated with IL-10/ZnPP/LPS.

Full size image

To further elucidate the role of HO-1, we treated animals with the HO inhibitor, ZnPP, and assessed its effect on IL-10-mediated protection. We noticed that mice receiving ZnPP seemed more susceptible to LPS-induced septic shock; however, the survival rate was not significantly different from that of littermates receiving LPS alone. Nevertheless, the protective effect of IL-10 was significantly attenuated and the survival rate reduced to 30% (P < 0.005) when the IL-10-treated mice received additional administration of ZnPP one hour before LPS challenge (Fig. 6c). To examine whether the inhibitory effect of ZnPP on IL-10-mediated protection can be countered by CO, we exposed animals receiving the treatment of IL-10/ZnPP/LPS to 250 p.p.m. CO for 12 hours immediately after LPS challenge and found that the survival rate returned to 86% (P < 0.005). To further confirm the specificity of ZnPP on HO inhibition in vivo, we conducted additional experiments using CuPP. ZnPP treatment significantly reduced the liver HO activity in mice receiving IL-10 for 12 hours (4.79 ± 2.27 nmol/mg protein/h (n = 8) versus 7.76 ± 2.07 nmol/mg protein/h (n = 8), P < 0.02); whereas CuPP treatment did not significantly affect the activity (8.31 ± 1.79 nmol/mg protein/h, n = 8). Likewise, CuPP treatment exhibited no significant effect on the protective effect of IL-10 (Fig. 6d). The serum TNF-α levels paralleled the severity of LPS-induced shock in these groups of animals (Fig. 6e). These observations strongly support the implication of HO-1 in the anti-inflammatory effect of IL-10 in vivo.

Discussion

Here we provide both in vitro and in vivo data in support of a role HO-1 in the anti-inflammatory effects of IL-10. Our data clearly show that IL-10 is a potent inducer of HO-1 in murine primary macrophages and a murine macrophage cell line. The induction of HO-1 was also observed in human peripheral monocyte-derived macrophages treated with IL-10 (data not shown), suggesting that the effect is not limited to rodents. Treatment with actinomycin D completely blocked the induction of HO-1 by IL-10, indicating that the regulation of HO-1 gene expression occurs at the transcriptional level. Although a putative STAT-responsive element is located in the promoter region of the murine HO-1 gene27, IL-6, which uses a similar STAT-3 signaling pathway to that used by IL-10 but lacks anti-inflammatory function14, did not induce HO-1 gene expression in murine macrophages. This result confirms an earlier report in human endothelial cells28, whereas a study on rabbit microvessel endothelial cells has yielded contradictory results29. We speculate that the differential effects of IL-6 on HO-1 gene expression is dependent on cell type and species. Regardless, our data is in agreement with previous reports by others showing that the STAT-3 dependent pathway is not sufficient to mediate the anti-inflammatory action of IL-10 (refs. 4–6). In fact, exposure of macrophages to IL-10 led to early but transient phosphorylation of p38, a MAPK that is commonly activated by inflammatory cytokines and stress stimuli30. We tested the role of p38 in HO-1 induction mediated by IL-10 and found that inhibition of p38 resulted in a complete blockade of this effect, supporting the role of p38 in this process. However, the detailed mechanism involved in the activation of HO-1 gene transcription by IL-10 in macrophages remains to be clarified.

The finding that the p38 signal pathway is involved in the IL-10-mediated anti-inflammatory effect is unexpected. Earlier studies on LPS-induced TNF-α biosynthesis have revealed the important role of the p38-dependent pathway in the regulation of TNF-α translation31,32,33. Furthermore, a recent report by Kontoyiannis et al.34 has demonstrated that the IL-10 suppression of TNF-α translation in LPS-stimulated macrophages is mainly through the inhibition of p38 activation. As IL-10 was coadministered in their study with LPS and the effect on p38 phosphorylation was examined at 15 minutes after challenge, it is unlikely that HO-1, which is induced by IL-10 at a later time period, has a role in mediating the inhibition of p38 activation observed under their experimental setting. In addition, we found that IL-10-induced p38 phosphorylation reaches a peak at 5 minutes, which is much earlier than the time point of LPS-induced maximal p38 phosphorylation (15 min)4,34. It will be intriguing to know whether the p38 signaling induced by IL-10 feeds back on the p38 activation later induced by LPS. Although research on p38 function has revealed the important role of its activation in inflammatory responses30, and inhibition of p38 is considered a potential therapeutic approach for anti-inflammatory therapy, some recent studies yield contradictory results. For example, treatment with a p38 inhibitor in mast cells results in enhanced production of TNF-α (ref. 35). A recent study by van den Blink et al.36 has also demonstrated that inhibition of p38 results in reduced cytokine production in most of the cell types tested, but increases cytokine release in LPS-activated macrophages in vitro and in murine models of endotoxemia and infection with pneumococcal pneumonia. The role of p38 in modulating the inflammatory responses therefore appears to depend on the cell type and stimulus. The observation that p38 mediates the induction of HO-1 by IL-10 in macrophages may explain part of the mechanism(s) by which p38 exerts anti-inflammatory effects in certain circumstances.

Involvement of HO-1 in the anti-inflammatory response has been shown in a few disease models13,37,38,39, including endotoxemia. Although biliverdin and bilirubin, produced by heme degradation, are antioxidants and may play a role in the protective response to tissue injury occurring during inflammation, recent studies have demonstrated that CO is the key molecule mediating the protective effect of HO-1 (refs. 13,40,41). Here we confirmed that CO mediates the inhibitory effects of IL-10 on LPS-induced inflammatory responses. Our data show that scavenging of CO by hemoglobin significantly reduced the inhibitory effect of IL-10 on LPS-induced TNF-α and NO production and MMP-9 expression in macrophages. Furthermore, in animal experiments we showed that CO exposure could fully restore the protective effect of IL-10, which was inhibited by additional ZnPP treatment in mice following LPS-induced septic shock. Because of the involvement of p38 in both IL-10-mediated HO-1 induction and the inflammatory responses elicited by LPS stimulation in macrophages20,31,32,33, the role of p38 in CO-mediated effects is difficult to prove in our experimental setting. Nevertheless, we suggest that IL-10 and HO-1 activate a positive-feedback circuit to amplify the anti-inflammatory capacity by upregulation of their respective expressions via CO and a p38-dependent mechanism.

In summary, the data presented here provide evidence to support the essential role of HO-1 in the anti-inflammatory function of IL-10 both in vitro and in vivo. CO appears to mediate most if not all of the protective effects of IL-10, although the detailed mechanism by which CO induces downregulation of inflammation-associated genes in macrophages remains to be determined. Identifying the cellular signal pathway and the responsive gene that are essential for the anti-inflammatory effects of IL-10 provides important information for the design of new therapeutic strategies to treat inflammatory diseases.

Methods

Materials.

LPS from Escherichia coli (serotype 055:B5) was from DIFCO (Detroit, Michigan). Recombinant mouse IL-10, IL-6 and anti-IL-10 neutralizing antibody were from R&D Systems (Minneapolis, Minnesota). Zinc protoporphyrin IX and copper protoporphyrin IX were from Porphyrin Product (Logan, Utah). SB203580 and PD98059 were from Calbiochem (La Jolla, California). HO-1 antibody was from StressGen (Victoria, Canada). INOS antibody was from Transduction Laboratories (San Diego, California). MMP-9, STAT-3, phospho-STAT-3 and SOCS-3 antibodies were from Santa Cruz Biotechnology(Santa Cruz, California). Phospho-JNK, phospho-p38 and phospho-ERK MAP kinase antibody kits were from New England BioLabs (Beverly, Massachusetts).

Cell culture.

Primary peritoneal macrophages were harvested from BALB/c mice and plated on a culture dish with RPMI 1640 medium containing 0.1% BSA. After 2 h of incubation in culture, nonadherent cells were removed by washing with serum-free medium. Adherent cells were then used for experiments. Murine J774 macrophages (ATCC TIB-67) were cultured in RPMI 1640 medium supplemented with 10% FBS. Cells at 60% confluency were changed to serum-free medium containing 0.1% BSA and subjected to various treatments.

Transfection with ODN.

The antisense phosphorothioated ODN complementary to murine HO-1 translation initiation codon and 6 base pairs on either side (5′-ACGCTCCATCACCGG-3′) was synthesized. The sense ODN (5′-CCGGTGATGGAGCGT-3′) and a scrambled ODN (5′-CACGTCACCTCAGCG-3′) were used as negative controls. The murine SOCS-3 antisense ODN (5′-GGTGACCATGGCGCA-3′) was used as an additional negative control. Cells (1 × 106/plate) were transfected with 1 μg ODN premixed with 10 μl LipofectAMINE reagent (Gibco BRL, Rockville, Maryland) according to the manufacturer's instructions. After incubation for 5 h, cells were washed once with serum-free medium, treated with 10 ng/ml IL-10 for 2 h, and followed by LPS (1 μg/ml) treatment for another 2 h prior to western-blot analysis and determination of TNF-α production.

Western-blot analysis.

Cells lysates were prepared and 50 μg of lysate proteins were electrophoresed on 8% or 10% SDS–polyacrylamide gel. Western blotting was carried out as described42 except that antigens were detected using enhanced chemiluminescence system (Pierce, Rockford, Illinois).

Northern-blot analysis.

Human HO-1 cDNA was obtained as previously described42. Murine A8 cDNA was prepared by RT-PCR using RNA isolated from J774 cells treated with LPS and IL-10 (ref.15). Northern blotting was performed as described42.

Determination of TNF-α concentration.

The TNF-α concentrations in culture media and mouse sera were determined using an IEA kit (Assay Designs, Ann Arbor, Michigan).

Determination of nitrite production.

Accumulated nitrite, a stable breakdown product of NO, in culture medium was determined using the Griess reagent43.

SDS–PAGE zymography.

Culture media (50 μl/sample) were electrophoresed at 4 °C in 8% SDS–PAGE gel containing 0.1% gelatin under a non-reducing conditions. The proteins in the gel were renatured by incubation with 2.5% Triton X-100 at room temperature for 1 h. The gelatinolytic activity was examined as described44.

Animal experiments.

To study the induction of HO-1 in vivo, BALB/c male mice (8-wk-old) received intraperitoneal administration of 1 μg IL-10/mouse. At the indicated times, the animals were killed and various cells and tissues collected, and HO-1 protein levels were determined by western blot. To assess the effect of IL-10 on endotoxin-induced shock, BALB/c mice received intraperitoneal injections of sterile PBS, ZnPP (25 mg/kg body weight), IL-10 (1μg/mouse), a lethal dose of LPS (20 mg/kg body weight), ZnPP for 1 h followed by LPS (ZnPP/LPS), IL-10 for 2 h followed by ZnPP (IL-10/ZnPP), IL-10 for 3 h followed by LPS (IL-10/LPS) or IL-10 for 2 h followed by ZnPP for 1 h and then followed by LPS (IL-10/ZnPP/LPS). In a separate experiment, ZnPP was replaced by CuPP (25 mg/kg body weight) for the treatment of animals. For CO exposure, animals were placed in a chamber (1.385 cubic ft) containing 250 p.p.m. CO, which was equilibrated by a flow of 0.5% CO (5,000 p.p.m.) mixed with compressed air, for 12 h. The concentration of CO in the chamber was continuously monitored by a gas monitor (Crowcon, Oxfordshire, UK). Serum TNF-α levels were determined 2 h after LPS challenge. The handling of animals was in accordance with the guidelines of the Institute of Biomedical Sciences, Academia Sinica.

HO activity assay.

To determine hepatic HO activity, liver was homogenized in 4 volumes of ice-cold 0.1 M potassium phosphate buffer pH 7.4, followed by centrifugation at 13,000g for 15 min at 4 °C. The supernatant (200 μg) was then incubated with 50 μM hemin, 1 mM NADPH, 2 mM glucose-6-phosphate and 1 unit glucose-6-phosphate dehydrogenase in 0.1 M potassium phosphate (pH 7.4) at 37 °C for 30 min in the dark. The bilirubin generated was estimated spectrophotometrically45.

Statistical analysis.

Each experiment was performed at least 3 times. The results are expressed as means ± s.d. Data on TNF-α and nitrite concentrations were analyzed using Student's t-test. Survival rate data were analyzed using Fisher's exact test. A value of P < 0.05 was considered statistically significant.