Nitric oxide inhibits the tumor necrosis factor alpha -regulated endocytosis of human dendritic cells in a cyclic GMP-dependent way.

Tumor necrosis factor-alpha (TNFalpha)-induced maturation of dendritic cells (DC), with down-regulation of their endocytic ability, has been reported to be mediated by the accumulation of the lipid messenger ceramide. We have now studied the effects and mechanisms of action of NO on endocytosis, investigated with fluorescein isothiocyanate-labeled dextran using human monocyte-derived DC, both immature and after treatment with TNFalpha. Exposure of DC to NO, released by either bystander phagocytes or NO donors, reversed the inhibition of endocytosis induced by TNFalpha. The intracellular accumulation of ceramide induced by TNFalpha was also inhibited by NO. In addition, NO was found to exert an inhibitory effect downstream of the TNFalpha-triggered ceramide accumulation, because NO donors reversed the inhibition of endocytosis induced by the cell-permeant C(2)-ceramide. These effects of NO were mimicked by the membrane-permeant cyclic GMP analogue, 8-Br cyclic GMP, and prevented by inhibition of the soluble guanylyl cyclase. At variance with rodents, the inducible isoform of the NO synthase was expressed neither in immature human DC nor after cell treatment with TNFalpha, interferon-gamma, and lipopolysaccharide, suggesting that regulation of these cells depends on exogenous NO. NO, working through cyclic GMP, might therefore prolong the ability of human DC to internalize antigens at the site of inflammation and thus modulate the initial steps leading to antigen-specific immune responses.

cells involved in the initiation of immune responses (1). Immature DC capture antigens at the site of inflammation and process and present them to T cells in secondary lymphoid organs where DC prime the immune response (1). These events are accompanied by a process of DC maturation, which includes down-regulation of endocytosis, and is orchestrated by proinflammatory signals generated at the site of infection (1). Among these signals, an important role might be played by NO, a gaseous messenger known to modulate specific functions of cell populations involved in the immune responses (2,3).
NO is generated intracellularly by both constitutive NO synthases (NOSs), as in the case of B and T lymphocytes (4), or by the inducible isoform of the enzyme (iNOS), expressed by macrophages after their activation with cytokines and bacterial products (5). NO may act either in an autocrine or paracrine fashion on neighboring cells, thus contributing to a coordinate action against pathogens (5)(6)(7). Mice with a targeted iNOS deletion are more susceptible to infections (8) and show enhanced T cell activity, characterized by strong cell-mediated immune responses and tissue damage (9). Indeed, the sustained generation of NO by iNOS endows activated macrophages and microglial cells with anti-microbial and cytotoxic activity and enhances the function of bystander T cells (2,10). Furthermore, NO regulates the generation of cytokines and chemokines at the site of infection (discussed in Ref. 11).
Evidence obtained in the murine system indicates that NO may also modulate DC function. In particular NO, either exogenous or produced by iNOS in DC themselves, appears to inhibit their antigen presentation function (12,13). So far, however, the mechanism of action by NO in the maturation process of DC has not been studied, nor have the intracellular targets of NO been identified.
In this study we have investigated the regulation by NO of the ability of human DC to endocytose extracellular antigens. We have used a well characterized model of DC maturation, i.e. human monocyte-derived DC exposed to tumor necrosis factor ␣ (TNF␣). This cytokine, working via its type I p55 receptor, triggers the maturation process of DC in vitro, with inhibition of their ability to endocytose soluble antigens (14 -16). Our results show that NO, generated either by bystander phagocytes expressing iNOS or by NO donors, prevents in a cyclic GMP (cGMP)-dependent way the down-regulation of endocytosis induced in DC by exposure to TNF␣. This effect of NO is due to inhibition of TNF␣-induced accumulation of ceramide, a lipid messenger known to play a key role in both antigen uptake and presentation by DC (17), as well as to additional effect(s) exerted downstream of ceramide accumulation.
Preparation of Immature and Mature Dendritic Cells-Peripheral blood mononuclear cells were isolated from the blood of healthy donors (kindly provided by the blood transfusion department of our institution) with a density Fycoll-Paque gradient as described (14). Mononuclear cells were resuspended in RPMI, 10% fetal calf serum, and allowed to adhere to 6-well plates (Costar, Cambridge, MA). After 2 h at 37°C nonadherent cells were removed. Monocytes were then cultured for 7 days in RPMI containing 10% fetal calf serum, 100 g/ml penicillin, 100 g/ml streptomycin, human GM-CSF (50 ng/ml), and IL-4 (1000 units/ ml) to derive immature DC. Mature DC cells were obtained from immature DC by a 48-h incubation in RPMI containing 10% fetal calf serum, 100 g/ml penicillin, 100 g/ml streptomycin in the presence of human TNF␣ (0.2-200 ng/ml). Similar conditions were used in the experiments in which C 2 -ceramide (80 M) was used. When the 48-h incubation with either TNF␣ or C 2 -ceramide was carried out in the presence of DETA-NO (100 M), SNAP (200 M), 8-Br-cGMP (3 mM), or ODQ (3 M), the compounds were added in various combinations 10 min before and were maintained throughout the incubation time. Control experiments in which DC were incubated with the various compounds in the absence of TNF␣ and C 2 -ceramide were carried out in parallel. In the experiments in which cells were exposed to DETA-NO and SNAP, they were dissolved immediately before addition to the cells. SNAP and DETA-NO prepared 7 days before the experiments were used in control experiments. Under these conditions they do not release any NO, as measured using a NO detecting electrode with a sensitivity of 1 nM (Mark-2 ISO NO, World Precision Instruments, Sarasota, FL). In the experiments in which DC were incubated with TNF␣, maturation was routinely assessed by flow cytometry, measuring the exposure on the plasma membrane of specific antigens known to be expressed by immature or mature DC, namely CD1a, a marker of human myeloid DC; MHC class I and class II molecules; CD80, CD86, and CD40, involved in T cell co-stimulation (14,18). Expression of these antigens was analyzed after staining with appropriate FITC-labeled Abs as described (18), using a fluorescence-activated cell sorter (FACStar Plus, Becton Dickinson, Sunnyvale, CA). Fig. 1 shows results from a typical analysis: GM-CSF, IL-4-treated (immature) DC express CD1a, MHC class I and class II, CD80, CD86, and CD40. TNF␣-treated (mature) DC show a significant up-regulation of MHC class I and class II, CD1a, CD80, CD86, and CD40. Expression of the macrophage marker CD14 was never observed. Viability and apoptosis in the various samples was assessed at different times by propidium iodide/annexin V staining exactly as described (19).
Co-culture of DC and NO-generating N9 Cells-Exposure of DC to a continuous flux of NO was achieved in vitro using the scavenger murine microglial brain N9 clone cells derived from embryonic mouse, which express iNOS upon activation (20). N9 cells were cultured in Iscove's modified DMEM containing 10% fetal calf serum, 100 g/ml penicillin, 100 g/ml streptomycin, 2 mM L-glutamine. Induction of iNOS by these cells requires IFN␥ together with a second signal (20). To induce iNOS expression, N9 cells (2 ϫ 10 5 cells/ml) were incubated for 24 h in the presence of polybed polystirene microspheres (microsphere/cell ratio ϭ 5) and mouse IFN␥ (10 units/ml) with or without the NOS inhibitors aminoguanidine (1 mM) and N -nitro-l-arginine methyl ester (L-NAME; 1 mM). NO production by N9 cells was measured by determining the nitrite accumulation in the culture medium using the Griess reaction (21). Standard curves with increasing concentrations of sodium nitrite were run in parallel. The co-colture of DC and N9 cells was carried out using a double chamber system (Costar, Cambridge, MA). A semipermeable polycarbonate membrane with a cut-off of 0.4 m separated the lower chamber, containing immature DC, from the upper chamber. DC were incubated for 48 h in the lower chamber at a density of 1 ϫ 10 6 cells/well either in the presence or absence of human TNF␣ (200 ng/ml). Activated, iNOS-expressing N9 cells were seeded in the upper chamber at a density of 2.5 ϫ 10 5 cells/well at the beginning of the incubation with TNF␣. DC were then collected, and endocytosis of FITC-dextran was measured as described below.
Analysis of Endocytosis of FITC-Dextran-DC, exposed to the various treatments described in the paragraphs above, were washed and resuspended in RPMI containing 10% fetal calf serum. 2 ϫ 10 5 cell samples were incubated at either 37 or 4°C with FITC-dextran (1 mg/ml). Uptake of the fluorescent dye was stopped at the indicated time points by the addition of ice-cold phsophate-buffered saline containing 1% fetal calf serum. Samples were then washed three times in the same buffer at 4°C and analyzed by flow cytometry using propidium iodide to exclude dead cells as described (22). FITC-dextran, reconstituted in FIG. 1. Phenotypic characterization of human DC exposed to TNF␣. Human monocytes were treated with GM-CSF (50 ng/ml) and IL-4 (1000 units/ml) for 7 days to obtain immature DC. These cells were then incubated with or without TNF␣ (200 ng/ml) for a further 48 h (right and left columns, respectively). Cell preparations were stained with FITC-conjugated Abs for surface antigens, as specified on the left-hand side, and analyzed by flow cytometry as described under "Experimental Procedures" (filled histograms). Their relative fluorescence intensity (RFI) was calculated versus negative controls (open histograms). The results shown are from one representative experiment.
RPMI and stored at 4°C, was centrifuged to remove aggregates before addition to the cells.
Measurement of cGMP Generation-Immature DC (1 ϫ 10 6 cells/ sample), incubated for 15 min at 37°C in phosphate-buffered saline with the phosphodiesterase inhibitor hydroxy buthyl methyl xanthine (0.6 mM), were incubated for an additional 15 min in the presence or absence of DETA-NO (100 M) with or without ODQ (3 M). The reaction was terminated by addition of ice-cold trichloroacetic acid (final concentration, 7.5%). After ether extraction, cGMP levels were measured using a radioimmunoassay kit and normalized on cellular proteins, determined by the bicinchoninic acid assay procedure (BCA protein assay; Pierce).
Measurement of Ceramide Concentrations-Immature DC cells (1 ϫ 10 6 cells/sample) were incubated in 80 l of phosphate-buffered saline with human TNF␣ (200 ng/ml) in the presence or absence of either DETA-NO (100 M), SNAP (200 M), or 8-Br-cGMP (3 mM) with or without ODQ (3 M) and then quickly shifted at 37°C. At the time points indicated, incubation was stopped by the addition of 300 l of ice-cold CH 3 OH/CHCl 3 (2/1, v/v). Samples were then supplemented with 100 l of CHCl 3 and 100 l of NaCl (1 M). The extracted phospholipids were incubated for 1 h at room temperature with 100 microunits diacylglycerol kinase in the presence of 5 mg/ml cardiolipin, 7.5% glucopyranoside, 1 mM diethylenetriamine pentaacetic acid, and 10 Ci of [␥-32 P]ATP (10 mCi/ml) as described (23). Under these conditions, diacylglycerol kinase is not rate-limiting, and full conversion of ceramide to ceramide phosphate is thus to be expected (24). The ceramide phosphates produced were separated by thin layer chromatography (Silca gel 60, Merck, Milan, Italy) using CHCl 3 /CH 3 OH/CH 3 COOH (65/15/5, v/v/v) as solvent. To determine the concentration of ceramide per sample, known amounts of ceramide standard were processed and loaded in parallel. The relevant spots were identified by autoradiography, and their radioactivity was estimated by microdensitometry using a Molecular Dynamics Imagequant apparatus (Buckinghamshire, UK).
Western Blotting-DC were incubated for 24 h in the culture medium with or without human TNF␣ (200 ng/ml), IL-4 (1000 units/ml), GM-CSF (50 ng/ml), IFN␥ (100 units/ml), and lipopolysaccharide (LPS, 10 g/ml). Cells were then collected, washed twice with cold buffer (150 mM NaCl, 1 mM EDTA, 2 mM Na 2 P 2 O 6 , 30 mM NaF, 20 mM Tris-HCl, pH 7.5), and lysed for 30 min in the same buffer containing 1% Triton X-100, 0.1 mM phenylmethyl sulfonylfluoride, 10 g/ml leupeptin, and 10 g/ml aprotinin. Protein content in the lysates was assayed by the bicinchoninic acid procedure. After addition of SDS and ␤-mercaptoethanol the samples were boiled, and 50 g of protein/lane were loaded into the slots of 10% SDS-polyacrylamide gels as described (25). High efficiency transfer of proteins onto nitrocellulose membranes was obtained at 200 mA for 18 h in a buffer containing 25 mM Tris, 192 mM glycine, 20% CH 3 OH, pH 8.3. After transfer, both the gels and the blots were stained with Ponceau red. For Western blotting, the nitrocellulose sheets were processed at room temperature, first for 1 h with phosphate-buffered saline containing 3% bovine serum albumin and then for 2 h with the anti-iNOS Ab in the same buffer, followed by five times washing for 5 min with 150 mM NaCl, 50 mM Tris-HCl, 0.05% Tween-20, 5% powdered milk, pH 7.4. The secondary Ab was then added for 30 min, after which the membranes were washed several times in the same buffer, and the signals were revealed with Enhanced ChemiLuminescence according to the manufacturer's instructions.
Statistical Analysis-The results are expressed as the means Ϯ standard error of the mean; n represents the number of individual experiments. Statistical analysis was performed using the Student's t test for unpaired variables (two-tailed). The marks *, **, and *** or ϩ, ϩϩ, and ϩϩϩ in the figures refer to statistical probabilities (p) of Ͻ0.05, Ͻ0.01, and Ͻ0.001, respectively, measured in the various experimental conditions as detailed in the legends to figures.

NO Modulates Endocytosis of DC Undergoing TNF␣-induced
Maturation-Human monocyte-derived DC retain their ability to uptake antigens. This ability is progressively lost when DC are treated with TNF␣ (14). We have investigated the role of NO in the modulation of the endocytic function of DC using the well characterized fluorochrome-labeled FITC-dextran as a marker of endocytosis. Two different sources of the gaseous messenger were used: the NO donors DETA-NO and SNAP and the N9 murine microglial cells activated with polystirene microspheres and IFN␥, a treatment that induces expression of iNOS and continuous generation of bioactive NO (10,20). In the first experimental approach, immature DC were incubated in a double chamber system with or without TNF␣ (200 ng/ml, 48 h) in the presence or absence of activated N9 cells. DC were then resuspended in cytokine-free medium and incubated with FITC-dextran, and the internalization of the latter was measured by flow cytometry. In the TNF␣-untreated, immature DC incubated at 37°C, uptake of FITC-dextran was found to proceed linearly for 15 min and then to level off progressively. In contrast, at 4°C no uptake was detected ( Fig. 2A). Treatment of DC with TNF␣ resulted in a statistically significant inhibition of endocytosis of FITC-dextran, which was prolonged throughout the period analyzed (1 h, Fig. 2A). This effect of TNF␣ was not due to cytotoxicity because cell viability was 94 Ϯ 3.8% after 48 h of treatment with the cytokine (n ϭ 5). Incubation with activated N9 cells prevented the inhibition of endocytosis induced by TNF␣ at all time points investigated ( Fig. 2A), an effect suppressed when incubation of DC with activated N9 cells was carried out in the presence of the NOS inhibitors aminoguanidine (1 mM) or L-NAME (1 mM) (Fig. 2B,  upper panel). The effect of these inhibitors was proportional to their ability to reduce NO release by the N9 cells, measured as nitrite accumulation in the culture medium (Fig. 2B, lower  panel).
In the second experimental approach, immature DC cells were incubated with TNF␣ (0.2-200 ng/ml) in the presence or absence of DETA-NO (100 M) or SNAP (200 M). TNF␣ induced a concentration-dependent inhibition of endocytosis (Fig. 3A) that was reversed by both NO donors (Fig. 3). Decomposed DETA-NO and SNAP, which are unable to release NO as measured by a NO-sensitive electrode, did not have any significant effect on endocytosis (Fig. 3B). Moreover, NO donors or activated N9 cells failed to induce any significant effect on endocytosis of DC that were not treated with TNF␣ (not shown). NO donors did not induce cytotoxicity (viability versus untreated controls was 95 Ϯ 5.0 and 93 Ϯ 4.3% after the 48-h incubation time with DETA-NO and SNAP, respectively, n ϭ 3).
The Effect of NO on Endocytosis of DC Undergoing TNF␣induced Maturation Is cGMP-dependent-NO effects are known to be mediated through both cGMP-dependent and -independent signaling pathways (5). cGMP generation in immature DC was increased by DETA-NO with respect to untreated controls (values were 2.39 Ϯ 0.23 and 0.35 Ϯ 0.05 pmol/mg/min, respectively), an effect that disappeared in the presence of the guanylyl cyclase inhibitor ODQ (3 M) (values were 0.42 Ϯ 0.07 pmol/mg/min) (n ϭ 3). Immature DC were treated with TNF␣ (48 h) in the presence or absence of NO donors, either alone or in the presence of ODQ or of the membrane-permeant cGMP analogue 8-Br-cGMP (3 mM). As shown in Fig. 4, DETA-NO and SNAP prevented the effect of TNF␣ on endocytosis of FITCdextran. This action of the NO donors was inhibited by ODQ and mimicked by 8-Br-cGMP. These results indicate that NO controls endocytosis regulated by TNF␣ through the activation of cGMP-dependent pathway(s).
Inhibition by NO of the Accumulation of Ceramide Induced by TNF␣ in DC Is Dependent on cGMP-Ceramide, a second messenger generated by the activation of the p55 receptor for TNF␣, has been shown to mediate some of the effects of the cytokine in DC (17). The kinetics of ceramide generation by DC treated with TNF␣ are shown in Fig. 5A. The cytokine induced a progressive accumulation of the lipid messenger, which reached a plateau after about 6 h and remained constant thereafter. No significant ceramide accumulation was observed in cells not exposed to TNF␣ (Fig. 5A). When DC were incubated with the cytokine in the presence of activated N9 cells, a negative correlation was found between the amount of NO gener-ated, measured as the concentration of nitrite released in the medium, and the intracellular accumulation of ceramide (Fig. 5B). A similar inhibition of ceramide accumulation by TNF␣ was observed in cells incubated with DETA-NO or SNAP (Fig. 5C). This effect of NO appeared to be cGMP-dependent inasmuch as it was prevented by incubation with the NO donors in the presence of ODQ and was mimicked by 8-Br-cGMP.

The NO/cGMP Action on Endocytosis Is Exerted Both Upstream and Downstream of the TNF␣-induced Generation of
Ceramide in DC-Incubation of DC with exogenous C 2 ceramide (80 M, 48 h) resulted in a statistically significant, persistent inhibition of FITC-dextran endocytosis with respect to that observed in untreated, control cells (Fig. 6A). When DC were incubated with exogenous C 2 ceramide in the presence of DETA-NO (Fig. 6) or SNAP (Fig. 6B), the inhibition of endocytosis by the lipid metabolite was reduced. This effect of the NO donors was prevented by incubation with ODQ and mimicked by 8-Br-cGMP (Fig. 6B). These results indicate that the action of NO is exerted not only via the inhibition of ceramide generation by TNF␣ but also via a mechanism(s) active downstream of it. This other effect of NO appears to be also dependent on cGMP generation. (26) and in purified DC after treatment with IFN␥ and LPS (13,27). We investigated whether exposure of human-derived DC to cytokines and LPS also induces expression of iNOS and generation of NO. Immature DC were treated for 48 h in the presence or absence of human TNF␣, human IFN␥ (100 units/ml), and LPS (10 g/ml), alone or combined as detailed in the legend to Fig. 7. NO generation was measured in the culture medium as nitrite formation, and the expression of iNOS was analyzed by Western blotting of DC lysates. Neither before nor after any of the treatments applied was any iNOS expression in DC or nitrite accumulation in the medium detected ( Fig. 7 and not shown).

FIG. 2. Effects of NO generated by iNOS-expressing N9 cells on the endocytic activity of DC treated with TNF␣.
A, immature DC were cultured without TNF␣ (squares), with TNF␣ (200 ng/ml, circles), or with TNF␣ and iNOS-expressing N9 cells (triangles) for 48 h. Cells were then washed and suspended in fresh culture medium with or without FITC-dextran (1 mg/ml), and endocytosis at 37°C was analyzed at the indicated time points as described under "Experimental Procedures." As a control, endocytosis of FITC-dextran was measured also at 4°C in cells that were not exposed to TNF␣ (diamonds). Endocytosis was calculated as a percentage of cells positive to FITC-dextran (DX) with respect to cells treated in the same way but not exposed to the fluorescent dye. B, immature DC were incubated for 48 h with or without TNF␣ (200 ng/ml), in the presence or absence of iNOS expressing N9 cells, the NOS inhibitors aminoguanidine (AG, 1 mM), and L-NAME (1 mM), as indicated in the key. The incubation medium was removed, and the nitrite concentration in it was measured as described under "Experimental Procedures." Cells were resuspended in fresh culture medium with or without FITC-dextran for 30 min. Endocytosis was calculated as described for A, and values were expressed as percentages of those measured in cells incubated without TNF␣ (100%). In both panels statistical probability versus cells treated with TNF␣ alone is indicated by the asterisks and calculated as described under "Experimental Procedures." ϩ in B refers to the statistical probability versus DC treated with TNF␣ in the presence of N9 cells (n ϭ 5).

DISCUSSION
We have investigated in DC the effects of NO and the mechanism of its action in the regulation of endocytosis. We have used immature human monocyte-derived DC that efficiently endocytose antigens in vitro (14). These cells were exposed to TNF␣, which triggered an in vitro maturation process, as confirmed by the concentration-dependent reduction of endocytic ability and by the up-regulation of molecules involved in T cell activation (1,14). The kinetics of endocytosis was studied by cell exposure to FITC-dextran. In immature DC, NO, generated either by activated N9 phagocytes or by two NO donors, DETA-NO and SNAP, did not modify endocytosis per se. In the presence of TNF␣, however, NO reversed the inhibitory effect of the cytokine, i.e. it maintained the ability of DC to internalize FITC-dextran. Because this effect of NO was prevented by ODQ, an inhibitor of the soluble guanylyl cyclase, and mimicked by the membrane permeant analogue of cGMP, 8-Br-cGMP, we conclude that NO acts via a cGMP-dependent mechanism.
The maturation process of DC is triggered by TNF␣ via activation of its p55 receptor (15), which in many cell types induces sphingomyelin breakdown with resulting accumulation of the lipid messenger ceramide (28,29). The involvement of latter in the inhibition of endocytosis was documented by previous studies with immature DC, where ceramide inhibited endocytosis of various substrates, namely lucifer yellow, horseradish peroxidase, and FITC-dextran, suggesting its role as the messenger by which TNF␣ down-regulates this process (17). We thus investigated whether the maintenance by NO of the endocytic ability in TNF␣-treated DC was due to inhibition of ceramide accumulation. Consistent with this possibility, we found that TNF␣ induced a time-dependent accumulation of ceramide, which was inhibited by both NO donors and NO released by activated N9 cells. cGMP generation accounted for this effect of NO, as demonstrated by experiments with ODQ and 8-Br-cGMP.
To analyze whether inhibition of ceramide accumulation by NO was the only event responsible for its ability to reverse the effect of TNF␣ on the endocytic ability of DC, experiments were carried out with the membrane permeant C 2 -ceramide. This lipid inhibited endocytosis of FITC-dextran in a persistent way, yet NO (but not NO plus ODQ) and 8-Br-cGMP were still able to reverse this effect. This finding indicates that the action of NO on endocytosis can be explained not only by its cGMP-dependent inhibition of ceramide accumulation but also by additional effects on the signal transduction pathway activated by TNF␣/ceramide, also mediated through cGMP. The molecular target(s) of this further action by NO/cGMP remain(s) to be established. These results indicate that the inhibition by NO, via cGMP, of the TNF␣-induced down-regulation of endocytosis is exerted at multiple levels along the signal transduction cascade triggered by this cytokine. NO and cGMP function therefore as wide inhibitors of the action of TNF␣ on endocytosis rather than as selective regulators of one single transductional event.
NO generation in peripheral tissues occurs as a consequence of various stimuli (2,(5)(6)(7)9). The events we describe, i.e. the regulation of endocytosis by NO through cGMP, might enable DC to prolong their antigen uptake function at the site of inflammation and therefore modify the ensuing immune responses in the lymphoid organs. So far, the role of NO in the maturation process of DC has been investigated in vitro in the murine system by measuring antigen presentation as well as the ensuing T cell proliferation after cell exposure to cytokines, which results in expression of iNOS (see e.g. 13). Both these functions were found to be impaired by the generation of NO by DC themselves, suggesting an overall inhibitory role for the messenger on DC maturation (see Ref. 30 for review). These data appear consistent with the mechanism of action by NO elucidated here for human DC, because maintenance of endocytosis, characteristic of an immature phenotype, is expected to be revealed in vitro as a reduced ability to stimulate preprimed T cells.
Regulation by NO of human DC function might, however, be different. To our knowledge, iNOS expression and NO generation by these cells have not been reported, except in primary biliary cirrhosis and hepatocellular carcinoma (31,32). Consistently, we could detect neither expression of iNOS nor NO generation by human DC exposed to various combinations of cytokines and LPS. Regulation of human DC might therefore depend on exogenous NO, generated at the site of infection by macrophages as a result of their activation by cytokine during the inflammatory response (33). Because the functional effects of TNF␣ on human DC appear to be reversible and reinducible (16), NO might maintain endocytosis of antigens as long as DC are confined to the inflammation site, tuning their response to this cytokine and possibly also to other maturative stimuli. After egress from the inflammation site, DC would no longer be exposed to NO and could therefore down-regulate their endocytic ability to prevent any interference by irrelevant selfantigens captured during migration to lymphoid organs (1). In conclusion, NO, acting on human DC in a paracrine fashion, may contribute to enhance immune responses, whereas disregulation of its homeostasis, with its generation under severe pathological conditions in DC (31,32), might instead impair the immunological function of these cells.