Renal cell carcinoma induces interleukin 10 and prostaglandin E2 production by monocytes

Immunotherapy with interleukin 2 (IL-2) is not an effective anti-cancer treatment in the majority of patients with renal cell carcinoma (RCC), suggesting that the activation of cytotoxic T cells or NK cells may be impaired in vivo in these patients. The production of immunosuppressive factors by RCC was investigated. Using immunohistochemistry, IL-10 was detectable in 10 of 21 tumour samples tested. IL-10 was undetectable in the supernatant of cell lines derived from these RCCs. However, these cell lines or their conditioned medium (RCC CM), but not normal renal epithelial cells adjacent to the RCC or breastcarcinoma cell lines, were found to induce IL-10, as well as prostaglandin E2 (PGE2) and tumour necrosis factor (TNF)α production by autologous or allogeneic peripheral blood mononuclear cells (PBMCs) and monocytes. IL-10 production induced by RCC CM was found to be dependent on TNF-α and PGE2 since an anti-TNF-α antibody (Ab) inhibited 40–70% of IL-10 production by monocytes, and the combination of anti-TNF-α Ab and indomethacin, an inhibitor of PGE2 production, inhibited 80–94% of RCC CM-induced IL-10 production by monocytes. The RCC CM of the five cell lines tested were found to induce a down-regulation of the expression of HLA-DR and CD86, as well as a strong inhibition of mannose receptor-dependent endocytosis by monocytes. The blockade of HLA-DR and CD86 expression was partially abrogated by indomethacin and anti-IL-10 Ab respectively, and completely abrogated by an anti-TNF-α Ab. The inhibition of mannose receptor-dependent endocytosis was partially abrogated by an anti-IL-10 Ab and completely abrogated by an anti-TNF-α Ab. These esults indicate that RCCs induce IL-10, PGE2 and TNF-α production by monocytes, which down-regulate the expression of cell-surface molecules involved in antigen presentation as well as their endocytic capacity. © 1999 Cancer Research Campaign


PBMC, monocytes and lymphocytes preparation
Total peripheral blood mononuclear cells (PBMCs) were isolated from heparinized blood obtained from healthy volunteers and from four patients with RCC in whom tumour cell lines (CAN, VER) or tumour cell cultures (LEC, DUF) were obtained, using Ficoll-Hypaque density-gradient centrifugation (Eurobio, Les Ulis, France). Monocytes and lymphocytes were further purified on a multistep Percoll gradient as previously described (Sallusto et al, 1994). Around 85-90% of the cells in the monocyte-enriched fraction were positive for CD14 expression using flow cytometry, compared with less than 3% of the cells in the lymphocyteenriched population.

Conditioned medium (CM) of renal cell carcinoma and other cell lines
Carcinoma cell lines were plated in 100-mm-diameter Petri dishes at a density of 5 × 10 5 cells ml -1 in complete medium. After 2 days of culture, supernatants were harvested, filtered through 0.22-µm mesh, aliquoted and stored at -20°C for further investigations. Each batch was tested for mycoplasma contamination and found to be negative.

Culture conditions
PBMCs, monocytes and lymphocytes were cultured in RPMI with 10% FCS in 24-well flat-bottomed plates (Falcon Labware, Oxnard, CA, USA) for the quantification of IL-10 production. PBMCs (10 6 ), 4 × 10 5 monocytes and 8 × 10 5 lymphocytes were plated in a final volume of 1 ml with 5 × 10 4 tumoral cells. In transwell (Costar, Brumath, France) culture conditions, monocytes were seeded in the upper compartment and tumoral cells in the lower compartment of 24-well plates. Cultures with conditioned medium (CM) were performed by addition of 5-80% or 40% RCC CM according as specified. The positive control for IL-10 production was obtained by addition of lipopolysaccharide (LPS), (1 µg ml -1 ) to PBMCs, monocytes or lymphocytes culture. Cells cultured in RPMI with 10% FCS alone were used as negative control.
Supernatants were collected after 48 h as this time has been reported to be the peak of IL-10 production by monocytes (de Waal Malefyt et al, 1991).

Immunohistochemistry
Immunochemical stainings were performed using an indirect threestep immunoenzymatic procedure with alkaline phosphatase (AP) (Combaret et al, 1989). Briefly, air-dried cryostat sections were fixed for 10 min in 4% paraformaldehyde at 4°C and incubated for 30 min with the primary antibody. After two washes in Trisbuffered, saline (TBS) containing 0.2% bovine serum albumin (BSA), slides were incubated for 30 min with AP-conjugated rabbit anti-mouse immunoglobulins (Ig) (Dako, Trappes, France), washed and incubated for 30 min with an AP-conjugated swine anti-rabbit Ig (Dako). Revelation was performed using the AP substrate (naphthol AS MX phosphate, dimethyl formamide, levamisole and fast red), before counterstaining with haematoxylin.

Flow cytometric analysis
Membrane staining Flow cytometric analysis was carried out by incubating 5 × 10 4 -10 5 cells for 20 min in 50 µl of phosphate-buffered saline (PBS) with 1% BSA and 0.1% sodium azide on ice with optimal concentrations of the above-mentioned antibodies coupled to phycoerythrin (PE). Cells were washed three times with PBS 1% BSA 0.1% sodium azide. The phenotype of the cells was analysed using a flow cytometer (FACScan).

Intracellular staining
Intracytoplasmic staining of IL-10 was performed after 48 h culture of PBMC with RCC cell lines in presence of Monensin (2 µM) (Sigma) to avoid cytokine secretion. Cells were permeabilized for 5 min in saponin buffer (0.33%) and labelled with an anti-IL-10 rat antibody coupled to PE (9D7) (Pharmingen) for 20 min according to the manufacturer's protocol. After three washes, intracytoplasmic IL-10 expression was analysed on a FACScan. Labelling specificity was assessed either with a rat IgG control antibody (R & D) staining or by preincubating 9D7 with a 100fold excess of recombinant IL-10 before PBMC staining.  Figure 2 Mean IL-10 production in co-culture of autologous PBMCs (10 6 cells ml -1 ) with four different RCC cell lines (CAN tum, VER, LEC, DUF) or with normal renal epithelial cells (CAN nor) (5 × 10 4 cells per well). Positive control was performed using LPS for each patient. This experiment is representative of three performed with different batch of PBMCs. IL-10 secretion by PBMCs treated with LPS or co-cultured with RCC CM is superior to that of PBMCs cultured with normal renal epithelial cells (Student's t-test P = 0.001)

Figure 3
Mean IL-10 production in co-culture of allogeneic healthy donor PBMCs (10 6 cells during 48 h) (s), purified monocytes (4 × 10 5 cells during 48 h) (s s) and lymphocytes (8 × 10 5 cells during 48 h) (s) with six different RCC cell lines (CAN, CHA, VER, Caki-2, ACHN, A-704), one normal renal epithelial primary culture (CAN nor) and one breast carcinoma cell line (MCF-7). Mean IL-10 production by epithelial cells alone (carcinoma cells or normal renal epithelial cells (s s). Positive control was performed using LPS (1 µg ml -1 ). This experiment is representative of five performed with different healthy donor PBMCs. IL-10 secretion by PBMCs (or monocytes) treated with LPS or cultured with RCC is significantly superior to that of PBMCs (or monocytes) co-cultured with breast carcinoma cell lines or normal renal epithelial cell cultures (Student's t-test, P < 0.01)

Figure 4
Intracytoplasmic IL-10 expression by allogeneic PBMCs cultured either in medium or with an RCC cell line (CHA coc as an example) in the presence of monensin (2 µM) was characterized by staining with a PEconjugated anti-IL-10 antibody (9D7) (B and D). The specificity of the labelling was assessed by preincubation of 9D7 with a 100-fold excess of IL-10 [9D7IL-10] (A and C). A control performed with a rat IgG1 control antibody was superimposable on the [9D7-IL-10] labelling (data not shown)

Figure 5
Mean IL-10 production by allogeneic purified monocytes (4 × 10 5 cells per ml) co-cultured with RCC cell lines (5 × 10 4 cells per well) ( ) or through a permeable membrane of transwell ( ) or with 40% RCC CM (s). This experiment was representative of five performed with different healthy donors

Cytokine ELISA
Cell supernatants, collected after 48 h of culture, were tested for the presence of IL-10 using a two-site sandwich ELISA method as previously described (Abrams et al, 1992). The detection limit of IL-10 ELISA is 0.04 ng ml -1 . TNF-α and PGE 2 levels in co-culture supernatants were measured using two-site sandwich immunoassays purchased from Immunotech (Marseille, France) for TNF-α and from Amersham (Les Ulis, France) for PGE 2 . The sensitivity of the tests was 10 pg ml -1 , and 16 pg ml -1 for TNF-α and PGE 2 respectively.

RT-PCR for IL-10
RT-PCR for IL-10 mRNA was performed as follows: precycle at 94°C for 3 min; cycles 1-40 at 94°C for 1 min (strand preparation), 58°C for 2 min (annealing) and 72°C for 3 min (primer extension). Then, the reaction was held at 72°C for 10 min. All reactions were performed with a Perkin Elmer DNA thermal cycler model 480. The primers for IL-10 mRNA were AGAAG-GCATGCACCAGCTCAGCA (3′) and TTTTGGAGACCTC-TAATTTATG (5′). Negative control was performed without cDNA adjunction in the master mix reagent. The positive control was the cDNA of the BJAB cell line (kindly provided by N Burdin, Schering Plough Corporation, Dardilly, France).

Endocytosis
Endocytosis was analysed using a previously described technique (Sallusto et al, 1995). After 48 h of culture, with or without RCC CM or the indicated antibodies and reagent, monocytes were washed three times and resuspended in 10% FCS medium buffered with 25 mM hepes at 37°C. After 10 min, fluorescein isothiocyanate (FITC)-dextran (DX-FITC, Molecular probe, Eugene, OR, USA) was added at the final concentration of 10 µg ml -1 for 30 min, at 37°C or at 4°C (control). The cells were then washed four times with cold PBS containing 1% FCS and 0.1% sodium azide and analysed on a FACScan. Mean IL-10 production by purified monocytes cultured with RCC CM (40%) in medium alone (s), with a polyclonal anti-TNF-α Ab (10 µg ml -1 ( ), with indomethacin (10 mM) ( ) or with the combination of indomethacin +anti-TNF-α Ab ( ). IL-10 production is significantly lower in the presence of anti-TNFα Ab (P < 0.01) but not indomethacin alone (P > 0.1). However, IL 10 secretion is significantly lower with anti-TNF-α+indomethacin compared with anti-TNF-α Ab alone in all cell lines (Student's t-test, P < 0.05)

Phagocytosis
Monocytes were cultured for 48 h with or without RCC CM or the indicated antibodies and reagent. During the last 4 h of the culture, 0.5 µm latex beads coupled to FITC (1:400) (Polysciences, Warrington, PA, USA) were added to the medium. To analyse the phagocytic capacity of monocytes, cells were recovered and washed three times with PBS. The phagocytosis of the latex beads-FITC was evaluated on a FACScan analyser.

Statistics
Statistical comparison between samples were performed using Student's t-test or Student's paired t-test.

IL-10 production is mediated by TNF-α and PGE 2
TNF-α and PGE 2 have been reported to induce IL-10 production by monocytes and macrophages (Wanidworanum and Strober, 1993;Kambayashi et al, 1995). Heat inactivation (100°C, 30 min) of RCC CM reduced partially (32% to 60%) IL-10 production by the eight different RCC CM tested, suggesting the co-involvement of a heat-insensitive molecule. Indeed, high levels of TNF-α and PGE 2 were detectable in the supernatant of monocytes cultured in the presence of RCC but not with breast carcinoma conditioned medium (Table 1). A polyclonal anti-TNF-α Ab blocked 40-70% of the production of IL-10 by monocytes incubated with the RCC CM of CAKI-1 and VER whereas indomethacin (an inhibitor of PGE-2 synthesis) induced a minor inhibition of IL-10 production ( Figure 6). Anti-TNFα Ab significantly reduced the production of PGE-2 by monocytes incubated with Caki-1 or VER CM (70.8 ± 12%; range = 57%-85%, P < 0.05 using Student's t-test in three experiments), indicating that PGE 2 production by monocytes is in part induced by TNF-α. The combination of anti-TNF-α Ab and indomethacin inhibited, at least additively, IL-10 production (80-94%) ( Figure 6). These results indicate that TNF-α and PGE 2 are responsible for the induction of IL-10 production by monocytes incubated in the presence of RCC CM. Importantly, the supernatant of the six RCC lines tested contained only low or undetectable TNF-α levels and no PGE 2 (Table 1). This suggests that RCC cell lines produce a soluble mediator which induces TNF-α and PGE 2 production by monocytes, both molecules acting additively to induce IL-10 production through an autocrine loop.

Phenotypic and functional alterations of monocytes by RCC: role of TNF-α, PGE 2 and IL-10
Phenotypic modifications Monocytes were collected after 2 days of culture with RCC CM and tested for CD14, CD54, CD80, CD86 and HLA-DR expression. The conditioned medium of the five RCC cell lines tested (Caki-1, Caki-2, CHA, VER, CAN), but not of breast carcinoma cell lines (T47-D, MCF-7, SK-BR3) (not shown), induced a significant decrease in HLA-DR, CD54 and CD86 expression, whereas CD14 an CD80 expression were unaffected (Figure 7). Anti-IL-10 Ab, but not a control polyclonal Ab, partially reversed the inhibition of CD86 expression induced by RCC CM, without affecting HLA-DR expression. Indomethacin (10 µM) partially reversed the decrease in HLA-DR expression without affecting CD86 expression ( Figure 8). The decrease in CD86 and HLA-DR expression induced by RCC CM could be completely abrogated by the addition of an anti-TNF-α neutralizing Ab (Figure 8) whereas the polyclonal control Ab had no effect (not shown).

Modulation of endocytosis and phagocytosis capacities by RCC CM
The endocytic capacity of monocytes cultured in RPMI-10% FCS for 48 h was evaluated using dextran-FITC incorporation ( Figure  9), which was inhibited by incubation at 4°C or with a solution of mannane (0.3 mg ml -1 ) (data not shown). The addition of RCC CM (VER and Caki-1 cell lines), but not breast carcinoma CM (data not shown), completely inhibited the endocytic capacity of monocytes ( Figure 9). This effect was partially reversed by an anti-IL-10 Ab, and completely reversed by an anti-TNF-α Ab, whereas polyclonal control Ab had no effect (Figure 9). In contrast, the capacity of monocytes to phagocyte 0.5 µm latex beads coupled to FITC was strongly enhanced in presence of the RCC CM (VER and Caki-1 cell lines) (Figure 10). Anti-IL-10 Ab alone partially abrogated this effect of RCC CM (Figure 10). Although indomethacin and anti-TNF-α alone had no effect, the combination of anti-IL-10 Ab with indomethacin + anti-TNF-α Ab synergistically blocked the increase in phagocytic capacity mediated by RCC CM.

DISCUSSION
The objectives of this study were to investigate the production by RCC tumours of immunosuppressive factors that may potentially affect the anti-tumour immune response. The results presented show that IL-10, an immunosuppressive cytokine, is detectable by immunohistochemistry in RCC tumours but not produced by RCC cells purified from these biopsies. The production of IL-10 by nontumoral cells in biopsies has also been reported in other tumour models such as non-Hodgkin's lymphoma (Voorzanger et al, 1996). However, RCC cell lines, but not normal renal epithelial cells or breast carcinoma cell lines, were found to produce (a) soluble factor(s) which triggers the production of IL-10 by autologous and allogeneic monocytes in vitro. These results are in agreement with previous observations showing that IL-10 transcripts are detectable in tumour-infiltrating leucocytes of RCC tumours by RT-PCR (Filgueira et al, 1993;Mauerer et al, 1995;Nakagomi et al, 1995;Wang et al, 1995). However, in this model, monocytes were found to be the major IL-10 producers among PBMCs, whereas purified lymphocytes failed to produce IL-10 in the same conditions. The observation that normal renal epithelial cells are not able to induce IL-10 production in vitro is consistent with the absence of IL-10 production by the normal adjacent renal parenchyma and indicates that the capacity to induce IL-10 production is highly correlated with a malignant phenotype for renal epithelial cells.
It is also important to notice that IL-10 production was significantly higher when monocytes and tumoral cells were in the same well as compared with culture in a transwell system, without contact between monocytes and tumoral cells. This observation could be explained by greater concentration of soluble factors at the contact of RCC cells, allowing the strongest activation of monocytes. However, we cannot exclude the possibility that a transmembrane molecule expressed at the surface of RCC cells may contribute to stimulate monocyte production of TNF-α and IL-10.
The results presented here show that RCC CM also elicited the production of PGE 2 and TNF-α by monocytes and that anti-TNF-α Ab and indomethacin, an inhibitor of PGE 2 production, at least additively inhibited IL-10 production induced by RCC CM. TNF-α and PGE 2 are well-known inducers of IL-10 production (Wanidworanum and Strober, 1993;Kambayashi et al, 1995;Huang et al, 1996) and recombinant TNF-α and PGE 2 were found to act synergistically in the induction of IL-10 production by monocytes in vitro (not shown). However, the capacity of indomethacin to affect IL-10 production has not been reported. Although a role for other metabolites of arachidonic acid cannot be excluded, these results strongly suggest that TNF-α and PGE 2 are involved in the induction of IL-10 production by monocytes elicited by RCC CM. An anti-TNF-α Ab was found capable of blocking RCC-mediated PGE 2 production by monocytes, in agreement with previous observations showing that TNF-α induces PGE 2 production (Bachwich et al, 1986;Lehmmann et al, 1988;Alleva et al, 1993). Taken together, these results indicate that TNF-α is a central mediator for the induction of PGE 2 and IL-10 production by monocytes in this model. The TNF-α/PGE 2 /IL-10 cascade is induced by RCC cells but not by normal renal epithelial cells or by the breast carcinoma cell lines tested.
In the present study, the addition of RCC CM potently affected the phenotype and function of peripheral blood monocytes cultured in vitro. RCC CM induced a down-regulation of both CD86 and HLA-DR expression on monocytes, these phenomena being reversed partially by an anti-IL-10 Ab and indomethacin respectively. This suggests that RCC CM-mediated inhibition of HLA-DR and CD86 expression is in part mediated by PGE 2 and IL-10 respectively. Indeed, PGE 2 and recombinant IL-10 respectively block HLA-DR and CD86 expression in vitro (data not shown), in agreement with previous reports (de Waal Malefyt et al, 1991). However, indomethacin may also block other arachidonic acid mediators that might play a role in this phenomenon. Although the anti-IL-10 Ab tested was capable of antagonizing the inhibition of CD86 induced by similar concentrations of recombinant IL-10 (100 U ml -1 ) (not shown), it failed to abrogate the effect of RCC CM, suggesting the existence of other inhibitory factors elicited by RCC CM. Indeed, an anti-TNF-α Ab abrogated the down-regulation of CD86 and HLA-DR expression induced by RCC CM much more efficiently than anti-IL-10 Ab or indomethacin. To our knowledge, this effect of an anti-TNF-α Ab has not been previously described. Anti-TNF-α may act here by inhibiting IL-10 and PGE 2 production by monocytes, which may result in a more efficient inhibition than the PGE 2 and IL-10 inhibitors used in this study.
In addition to these phenotypic alterations, the supernatant of RCC cell lines induced a strong down-regulation of endocytosis mediated by mannose receptor, a phenomenon that was also partly abrogated by an anti-IL-10 and completely inhibited by anti-TNF-α Ab. Endocytosis is an initial step allowing exogenous antigen processing within the endocytic compartment, followed by antigen presentation by MHC class II receptors (for review see Watts, 1997). The loss of CD86 and HLA-DR expression on monocytes induced by RCC CM may have important functional consequences for the immune response against RCC cells. In vivo, in man, the lack of expression of HLA class II antigens in RCC tumours has been associated with a lack of response to immunotherapy with IL-2 (Cohen et al, 1987;Rubin et al, 1989). The results presented here suggested that the loss of HLA-DR expression may be induced by RCC tumour cells themselves. The simultaneous inhibition of mannose receptor-mediated endocytosis by the TNF-α/PGE 2 /IL-10 cascade induced by RCC CM may further contribute to impair the antigen presentation properties of monocytes in vivo.
Finally, in contrast to these inhibitory properties, RCC CM were found to increase the phagocytic capacity of monocytes, an effect

Figure 11
A proposed scheme for the cytokine cascade elicited by RCC which was partially blocked by anti-IL-10 Ab. Although indomethacin (not shown) and anti-TNF-α had no effect, the combination of anti-IL-10 Ab, indomethacin and anti-TNF-α Ab synergistically blocked the increase in phagocytic activity. This synergistic effect may result from the inhibition of IL-10 production induced by anti-TNF-α and indomethacin. This is in agreement with recent reports showing that IL-10 increases the phagocytic capacity of macrophages (Capsoni et al, 1995;Hashimoto et al, 1997). The soluble factor(s) responsible for the activation of the TNF-α/PGE 2 /IL-10 cascade are not yet characterized. RCC CM contain only low or undetectable levels of these mediators, indicating that RCC cell lines produce (a) soluble mediator(s) capable of inducing the production of TNF-α by monocytes (Figure 11). The factor(s) responsible for the cytokine cascade induced by RCC CM is (are) currently under investigation since cytokines previously reported to be produced by RCC, i.e. TGF-β 1 FGFb, IL-6, IL-8, TGF-α and GM-CSF, were found unable to induce IL-10 production by monocytes (not shown).
Taken together, these results indicate that RCCs induce the production of a cytokine cascade TNF-α/PGE 2 /IL-10 which inhibits the expression of surface molecules involved in the antigen-presenting function of monocytes as well as their mannose receptor-mediated endocytic capacity. These phenomena may interfere with the spontaneous and therapeutic immune response against RCC tumours in vivo in patients.