Tumor cells prevent mouse dendritic cell maturation induced by TLR ligands

Abstract

Tumor cells can evade the immune system through several mechanisms, one of which is to block DC maturation. It has been suggested that signaling via Toll-like receptors (TLR) may be involved in the induction of prophylactic anti-cancer immunity and in the treatment of established tumors. In the present study we found that high numbers of tumor cells interfere with BMDC activation induced by the TLR ligands LPS and poly IC. Tumor cells blocked TLR3- and TLR4-mediated induction of MHCII and the co-stimulatory molecules CD40 and CD86, as well as the cytokines IL-12, TNF-α and IL-6. Importantly, tumor cells induced inhibitory molecules (B7-DC, B7-H1 and CD80) on spleen DC in vivo and on BMDC, even in the presence of TLR ligands. Moreover, after a long exposure with tumor cells, purified BMDC were unable to respond to a second challenge with TLR ligands. The failure of tumor exposed-BMDC to express co-stimulatory molecules and cytokines in the presence of TLR ligands has implications for the future development of DC-based cancer immune therapies using TLR ligands as adjuvants for the activation of DC.

Introduction

Dendritic cells (DC) are the main antigen-presenting cells (APC) capable of inducing immunity to foreign antigens [5]. In a normal functioning immune environment, DC reside in peripheral tissues in a resting (steady) state with high antigen uptake and processing ability. Upon activation, DC migrate from peripheral tissues to draining lymph nodes, displaying antigenic peptides in the context of MHC I or MHC II for presentation to CD8⁺ or CD4⁺ T cells, respectively. Thus, DC activation is critical for the effective stimulation of Ag-specific effector T cell responses [40]. Experimentally, a variety of agents have been used to induce DC activation, including monocyte-conditioned medium [58], a cytokine mixture consisting of TNF-α, IL-6, IL-1β and PGE2 [36], and TLR ligands [2, 3]. However, it is becoming increasingly clear that the DC maturation and function are influenced by the surrounding microenvironment [39].

Due to their unique capacity to stimulate resting T cells, DC are critically important for the induction and maintenance of anti-tumor responses and are the candidate cell type to use for immunization protocols [59]. It has been shown that bone marrow-derived DC (BMDC) are able to promote prophylactic anti-tumor immunity when pulsed with relevant tumor-associated T cell epitopes or apoptotic/necrotic tumor cells [10, 24, 55]. However, the degree of differentiation (mature vs. immature) may determine the DC function. Increasing evidence suggests that tumor cells can directly or indirectly influence the activity of DC as means of escaping anti-tumor immune responses. Some mechanisms associated with tumor escape result in impaired DC maturation [22, 62]. Several studies have demonstrated an accumulation of immature DC in tumor-bearing mice [21, 32] and in cancer patients [16, 52]. Importantly, whereas functionally mature DC can induce potent anti-tumor immunity in vivo [8, 14, 40], immature or partially differentiated DC induce either T-cell unresponsiveness [7, 28] or regulatory T cells (Treg) [15, 35]. Moreover, as well as inhibiting DC differentiation and maturation, tumor cells profoundly affect the functional capacity of DC to activate T cell-specific immunity by triggering the expression of inhibitory molecules, i.e. B7-H1 [13]. Since DC activation is essential for the induction of anti-tumor immunity, it is crucial to understand the interplay of tumor cells in the DC maturation process.

Because the maturation of DC in a tumor-bearing host is down-regulated as part of the anti-inflammatory responses, a pro-inflammatory stimulus may be necessary for the activation of an anti-tumor-specific immune response. It was shown that GM-CSF, TNF-α, IL-4 or CD40L could not induce B7 expression in rat tumor infiltrated-DC [11], suggesting that the lack of activation might be caused by a defect in DC differentiation in the tumor microenvironment. However, presently no data are available that conclusively address whether mice tumor differentiated-DC could respond to other known DC-activating signals. One possibility is the use of TLR ligands as activation signal. TLR, a family of receptors consisting of more than 10 protein members both in mouse and in human, have recently emerged as a critical component of the innate immune system for detecting microbial infection and activation of DC maturation programs to induce adaptive immune responses [33]. Importantly, several features of TLR stimulated-DC make these ligands attractive for their use in tumor immunotherapy. First, TLR ligation stimulates DC maturation, generating fully functional DC [33]. Second, TLR4 and TLR3 ligands have a crucial function in cross-priming leading to induction of MHC I presentation and CTL induction [1, 47]. Third, NK and NKT cells can be activated by mature DC that are pre-treated with TLR3 agonists [54]. Fourth, stimulation of TLR4 on DC blocks the inhibitory activity of regulatory CD4⁺CD25⁺ T cells [56]. Thus, these properties of TLR3 and TLR4 receptors may offer an opportunity to activate DC in the tumor microenvironment. Therefore, in this study, BMDC co-cultured with or without tumor cells were exposed to different TLR stimuli and compared with respect to phenotype and cytokine production.

We reported here that tumor cells regulate DC activation in vitro and in vivo inducing a state of differentiation characterized mainly by the expression of CD80, B7-H1 and B7-DC, and an absence of cytokine expression. We found that tumor cells efficiently prevent LPS- and poly IC-induced activation of BMDC via TLR4 and TLR3, respectively. Moreover, tumor cells inhibited cytokine production induced by both TLR4 and TLR3. Thus, it seems that DC differentiation induced by tumor cells cannot be reversed by TLR ligands; this might have important implications in the use of DC and TLR ligands for cancer therapy.

Materials and methods

Mice and cell lines

Six- to eight-weeks-old C57Bl/6 mice or Balb/c female mice were purchased from Harlan through their facilities at the National University of Mexico (UNAM). Mice were maintained in specific pathogen-free conditions, and studies were performed in accordance with local ethical guidelines. B16 is a poorly immunogenic spontaneously developed murine melanoma (American Type Culture Collection, ATCC, Manassas, VA, USA [18]). The tumor cell line TC-1 was originally generated by retroviral transduction of lung fibroblasts of C57Bl/6 origin by HPV-16 E6/E7 and c-H-ras oncogenes (ATCC [41]). J558 is a plasmacytoma of Balb/c origin (ATCC [27]). TC-1 and J558 cells lines were maintained in culture in RPMI 1640 (GIBCO, BRL) supplemented with 10% heat-inactivated FCS (GIBCO, BRL), 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin (GIBCO, BRL, R10 medium). The B16 melanoma cell line was maintained in cultured in DMEM (GIBCO, BRL) supplemented as described earlier. Cells lines were periodically verified to be mycoplasm-free by Hoechst staining.

Antibodies and reagents

FITC anti-CD86, CD40, CD80, CD69, I-A^b, I-E^k, CD11b, Rat IgG₂a, Mouse IgG₂a, APC-conjugated CD11b and CD11c and PE-conjugated CD11c, IL-12, IL-6, TNF-α and B7-DC and B7-H1 were all from Pharmingen-BD Biosciences (Mountain View, CA, USA). Additional reagents were CFSE (Molecular probes) ACK buffer (BioSource), Collagenase D (Roche) and EDTA 0.5 M (GIBCO, BRL). LPS from Salmonella typhimurium was generously donated by Dr. Rodolfo Pastelín (Faculty of Chemistry, UNAM). Poly IC was purchased from Amersham Biosciences.

Induction of tumor cell apoptosis

Tumor cells were harvested with 0.5 mM EDTA, washed twice with PBS (GIBCO, BRL), resuspended to 10⁷/ml in PBS and irradiated with 75 Gy as described previously [24, 42]. For detection of apoptotic tumor cells, Annexin V-FITC Apoptosis Detection kit (BD Biosciences) was used (data not shown). Within 24 h, 30–35% of the tumor cells were apoptotic, i.e. annexin V⁺ and PI⁻ (data not shown, [24, 42]).

Generation of bone marrow-derived DC

Bone marrow cells were harvested as previously described [45] from femurs and tibias of normal C57Bl/6 or Balb/c mice and washed with PBS. Cells were resuspended in RPMI 1640 supplemented with 10% heat inactivated FCS, 2 mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin and 10% supernatant of a GM-CSF producing cell line [31] and cultured in 100-mm diameter plates. On day 3 the cells were refreshed with medium containing GM-CSF. On day 5 of culture, DC were 60% of the suspension population, characterized by the expression of both CD11c and CD11b [45].

Culture of bone marrow-derived DC with tumor cells

On day 5 of bone-marrow progenitors culture, cell suspensions were harvested, counted and 5 × 10⁶ cells were incubated with irradiated tumor cells at a 1:1 ratio or 1:10 ratio (TC:DC), LPS (0.1, 1 or 10 μg/ml) or poly IC (25, 50 or 100 μg/ml) for 24, 48, and 96 h. The incubation was performed in 100-mm diameter plates with 10 ml of fresh R10. After culturing time, DC were harvested and stained with PE-conjugated anti-CD11c, APC-conjugated anti-CD11b and FITC-conjugated anti-CD40, CD86, CD69 and MHCII for analysis of the surface phenotype. In the case of B7-DC and B7-H1, the staining was performed with PE-conjugated antibodies.

When DC were differentiated in the presence of tumor cells, after 96 h of culture, the cells were harvested and positively selected with CD11c MACS microbeads (Miltenyi, Biotec, Auburn, CA, USA). The cells were then plated in the absence of tumor cells and re-stimulated with LPS (0.1 μg/ml) or poly IC (25 μg/ml) for 24 h after which, the flow cytometry analysis was performed. In some experiments, BMDC were stimulated with splenocytes from naive mice at a 1:1 ratio. Splenocytes were obtained after mechanical disgregation of spleens. Spleen suspensions were depleted of CD11c⁺ DC using MACS microbeads and then cultured for 3 days in R10. Flow cytometry data were analyzed with Flow Jo (Tree Star Software, San Carlos, CA, USA).

Cytokine production by DC

For intracellular cytokine staining, 2.5 μg/ml Brefeldine A (BFA, Sigma-Aldrich) was added in the last 12 h of co-culture. The cells were then harvested, stained for extracellular CD11c-APC and FITC-conjugated CD11b, and then stained for cytokines with the BD intracellular cytokine staining kit (BD Biosciences). The detection of cytokines was performed with PE-conjugated mAbs to IL-12, TNF-α or IL-6.

In vivo inoculation of tumor cells

For CFSE labeling, 10⁷/ml J558 tumor cells were incubated with 5 μM CFSE for 10 min at 37°C. The reaction was stopped by washing three times with PBS. CFSE labeled irrJ558 tumor cells of 2 × 10⁷ were injected i.v. into Balb/c mice. At 2, 12, 24 or 48 h after inoculation, animals were killed and the spleens were harvested. Single cell suspensions were prepared with 400 U/ml Collagenase D for 30 min. About 0.5 mM EDTA was added in the last 5 min of culture. Cells were then incubated with APC-conjugated CD11c, and PE-conjugated CD80, CD86, CD40 and I-E^k. CFSE⁺ CD11c⁺ cells indicate DC that have phagocytosed tumor cells in vivo.

Statistical analysis

Student’s t test was applied to reveal significant differences in induction of co-stimulatory or inhibitory molecules by BMDC treated under different cultured conditions. Our results were expressed as the mean ± SE. A P value of < 0.05 was accepted as the level of significance.

Results

Tumor cells induce a distinct DC differentiation phenotype in vitro in the absence of other stimuli

To induce an immune response, DC must undergo a maturation process that depends, in part, on the induction and/or up-regulation of co-stimulatory ligands, in addition to increased expression of MHC and other immunologically relevant molecules, such as CD40. In cancer patients and mouse tumor models, mature DC are found to be decreased in number [21, 32]. However, as human tumorigenesis is a slow process that, like chronic infection, might occur over a several years period, it is difficult to assess accurately immunological changes occurring at different stages. As an initial approach, we examined the effects of tumor cells on mostly immature BMDC in the absence of known maturation stimuli. We cultured bone marrow progenitors from C57Bl/6 mice for 5 days in the presence of GM-CSF, after which non-adherent cells were harvested, washed, counted and co-cultured at 1:1 or 1:10 (TC:DC) ratio with viable or IrrTC-1 transformed lung fibroblasts as a source of tumor cells. After 24 h, cells were harvested and stained with anti-CD11c-PE, anti-CD11b-APC and FITC-conjugated antibodies directed at one of the following: CD80, CD86, CD40, CD69 or MHCII.

Figure 1a shows that irradiated tumor cells induced up-regulation of CD69, CD40, MHCII and a marginal but consistent increase of CD86. In all cases the increase in CD86, MHCII and CD40 was to a lesser extent than the activation induced by the TLR4 and TLR3 ligands, LPS and poly IC, respectively. Viable tumor cells also induced phenotypic changes, but these were milder, with minimal increase in co-stimulatory molecules. Furthermore, BMDC treated with viable or IrrTC-1 tumor cells up-regulated the inhibitory molecules B7-DC and B7-H1, as well as CD80 (Fig. 1a, b). It was previously shown that CD80 has a higher affinity for CTLA4 than for CD28 [57], and CTLA4 is expressed in Treg cells [13]. Thus, the increase in CD80 may reflect in vivo the inhibition of the anti-tumor response by Treg cells. On the other hand, co-cultured of BMDC with non-tumor cells (i.e. syngeneic splenocytes) did not induce significant up-regulation of any co-stimulatory or inhibitory marker compared with non-treated cells (Fig. 1b). Unlike LPS-induced changes, which lasted longer, tumor cells-induced changes on BMDC were transient, returning to basal levels at 48 h post-stimulation (Fig. 1c).

Fig. 1.

Tumor cell-induced phenotype changes of mouse DC. C57Bl/6 bone marrow cell suspensions were cultured in GM-CSF for 5 days. Afterwards, viable, irrTC-1 tumor cells or syngeneic splenocytes, LPS (0.1 μg/ml) or poly IC (25 μg/ml) were added, up to 96 h. BMDC were stained with CD11c-PE and CD11b-APC, followed by FITC-labeled CD40, CD80, CD86, CD69, I-A^b, B7-DC, B7-H1 or Rat IgG_2a. a 5-day GM-CSF BMDC cultured for an additional 24 h alone or with the indicated stimuli. Results are representative of five independent experiments. b As in a, but the induction of inhibitory and co-stimulatory molecules is shown as fold induction of immature C57Bl/6 derived-BMDC (left panel, mean of five independent experiments) or immature Balb/c derived-BMDC (right panel, mean of three independent experiments). Results presented as mean ± SE. c Similar to a, but at 24, 48 and 96 h post-stimulation. An experiment representative of four is shown. d BMDC stimulated for 12 h with irrB16 tumor cells (C57Bl/6 derived-BMDC) or irrJ558 tumor cells (Balb/c derived-BMDC). An experimental representative of three with similar results is shown. In all cases, data represent MFI

Similar results were obtained with the addition of other sources of tumor cells, such as the murine melanoma B16 (Fig. 1b, d), as well as with the J558 plasmacytoma, performed with Balb/c-derived BMDC [Fig. 1b, d (bottom)]. The finding that irradiated cells induced more profound phenotypic changes on BMDC suggested the possibility that a few apoptotic cells in our living population were responsible for DC activation. However, this was not the case, because decreasing the number of either irradiated or viable cells added to the BMDC cultures resulted in loss of the stimulatory effect (Fig. 1a). Taken together, these results indicate that both living and apoptotic tumor cells induce BMDC differentiation, characterized by an increase of surface phenotypic markers, especially markers involved in the inhibition of the immune response.

Tumor cells induce phenotypic changes on phagocytic DC in vivo

Next, we examined whether the effects of tumor cells on DC in vivo were similar. Balb/c mice were injected i.v. with CFSE-labeled, irrJ558 tumor cells as described [42]. We inoculated only the J558 tumor cells but not the TC-1 or B16 tumor cells because they are smaller and easier to handle for i.v. inoculation. As irrJ558 tumor cells are selectively captured by spleen DC after i.v. inoculation [42], it was possible to compare the effect of tumor cells on DC that had phagocytosed the fluorescent tumor cells with those that had not.

Spleen DC were analyzed by flow cytometry for the expression of various cell surface molecules known to change during their maturation at 2, 12, 24 and 48 h after tumor cell inoculation. The total CD11c⁺ population in the spleens of mice inoculated with tumor cells did not show significant surface phenotypic changes at any time point (Fig. 2, middle row). However, those DC that had phagocytosed tumor cells (CD11c⁺/CFSE⁺ cells) up-regulated CD40, CD86 and MHCII as early as 2 h post-inoculation of irrJ558 tumor cells (Fig. 2, last row). Other surface molecules, such as CD80, as well as the inhibitory ligands B7-H1 and B7-DC, had slower kinetics of induction, which started 12 h after inoculation (Fig. 2). This is clearly different from the in vivo effects of LPS, which induced a marked increase in all the markers examined in all spleen DC (Fig. 2, first row). Thus, irradiated tumor cells inoculated in vivo are sufficient to confer on DC the capacity to up-regulate co-stimulatory and inhibitory markers but only in phagocytic DC, indicating the need for contact between DC and tumor cells for such an effect to occur. Of note, tumor cells in vivo induced an early but transient increase in CD86, CD40 and MHCII in phagocytic DC; and a later and more persistent increase on CD80, B7-DC and B7-H1.

Fig. 2.

Tumor cells alter the phenotype of phagocytic spleen DC in vivo. Balb/c mice were injected i.v. with PBS, LPS (1 μg) or with 2 × 10⁷ CFSE-labeled irrJ558 tumor cells. 2, 12, 24 and 48 h later, total spleen cells were isolated and stained with CD11c-APC and PE-conjugated anti-CD40, CD80, CD86, MHC II, B7-DC and B7-H1. Surface markers were analyzed in the total CD11c⁺ DC population from LPS-inoculated mice (first row) or mice inoculated with CFSE-labeled irradiated J558 (middle row). Expression of the surface markers of the double positive CD11c⁺/CFSE⁺ phagocytic population is shown in the last row. Data are representative of at least three independent experiments and numbers indicate MFI

Tumor cell differentiation of DC does not lead to cytokine expression

The concept of DC maturation is usually interpreted as a change in surface phenotype (high cell-surface levels of MHCII, CD40, CD80 and CD86) rather than a functional change (ability to prime an immune response). However, recent observations indicate that DC defined as mature solely on the basis of their surface phenotype do not always promote T-cell immunity and can even induce tolerance [51]. Therefore, parameters other than DC surface phenotypic changes need to be examined to probe the functional state of DC.

Besides surface phenotypic changes, common maturation stimuli prime DC to release large quantities of cytokines that play a role in T cell activation and differentiation [46, 69]. Figure 3 shows that at 12 and 24 h BMDC cultured in the presence of LPS produced high quantities of IL-12, TNF-α and IL-6. In contrast, despite the surface phenotypic changes induced, tumor cells did not induce the production of cytokines by BMDC. Moreover, BMDC co-cultured in the presence of syngeneic splenocytes did not produce any type of cytokines (data not shown). Thus, tumor cell-induced BMDC differentiation does not lead to the production and/or release of these cytokines, which might reflect a defect in DC differentiation.

Fig. 3.

Absence of cytokine expression by BMDC cultured with tumor cells. Five-day BM progenitors differentiated with GM-CSF were pulsed with 0.1 μg/ml LPS, IrrTC-1 or IrrB16 tumor cells (1:1 ratio) for 12 h (top three rows) or 24 h (lower three rows), and examined for intracellular IL-12, TNF-α and IL-6 (PE-label). BFA (2.5 μg/ml) was added to the cultures 12 h before staining. Dot plots are from gated CD11b⁺ cells. Results are representative of at least three independent experiments

Tumor cells alter DC differentiation even in the presence of TLR ligands

If tumor cells possess a mechanism to evade immune responses by inhibiting DC differentiation, it is necessary to examine if this also occurs in the presence of inflammatory stimuli known to activate DC to an immunogenic phenotype. This is an important consideration not only because the use of TLR ligands was suggested for DC-based immunotherapy [70, 72], but also because it is possible that the previous results might indicate a defect in DC differentiation in the presence of tumor cells and not only a lack of activating signals in the tumor microenvironment. In order to test this, B6 bone marrow progenitors were cultured in the presence of GM-CSF for 5 days and then for 24 h in the presence of the TLR ligands LPS or poly IC with or without the simultaneous addition of tumor cells. As shown in Fig. 4, those BMDC treated with LPS or poly IC displayed characteristic markers of mature DC. However, the presence of IrrTC-1 tumor cells in the culture partially impaired the up-regulation of several surface markers, including co-stimulatory molecules CD40 and CD86, CD69 and MHC II, all of which are necessary for the induction of adaptive immune responses (Fig. 4a, b). We also confirmed the inhibitory effect of tumor cells on TLR-activated BMDC using IrrB16 tumor cells (Fig. 4a). Although IrrB16 tumor cells did not alter significantly the expression of CD86 and CD69 on LPS-activated BMDC, TLR-induced up-regulation of CD40 and MHCII as well as poly IC-induced up-regulation of CD86 and CD69 were impaired. In contrast, TLR ligands-induced up-regulation of the inhibitory molecules B7-H1 and CD80 was not significantly reduced by IrrTC-1 or IrrB16 (Fig. 4a, b).

Fig. 4.

Inhibition of DC differentiation by tumor cells. a BMDC (5 days) were stimulated with 0.1 μg/ml LPS or 25 μg/ml poly IC in the presence or absence of the indicated tumor cells or syngeneic splenocytes (1:1, TC: DC ratio). After 24 h in culture, cells were harvested and stained with CD11c-PE, CD11b-APC and FITC-conjugated CD40, CD86, CD69, I-A^b, CD80 or B7-H1. Results are representative of at least four independent experiments and numbers represent MFI. b As in a, but the induction of CD40, CD80 or B7-H1 was measured by flow cytometry and mean values of MFI from three independent experiments were calculated as a percentage of control LPS (left panel) or poly IC (right panel) responses. Results presented as mean ± SE. ** Not significant. c As in a, but BMDC were stimulated for 12 h in the presence of BFA (2.5 μg/ml), after which intracellular IL-12, TNF-α and IL-6 were evaluated by flow cytometry as in Fig. 3. d As in c, but the stimulus was for 24 h with 0.1 μg/ml or 1 μg/ml LPS as indicated. For C and D, the experimental conditions were as described in Fig. 3. Results are from a minimum of three experiments

To further test the impact of tumor cells on DC maturation, we analyzed the production of cytokines by BMDC stimulated with TLR ligands in the presence of tumor cells. As shown in Fig. 4c, irrTC-1 and irrB16 tumor cells also impaired cytokine production by LPS-activated BMDC. This was observed even when the concentration of LPS in the cultures was raised 10 times above the original concentration (Fig. 4d). However, only high numbers of tumor cells (1:1, TC:DC ratio) had an inhibitory effect (data not shown). Non-tumor cells (syngeneic splenocytes) did not interfere with LPS- or poly IC-induced BMDC maturation (Fig. 4b, d). To examine whether tumor cells would exert cytotoxic effects or induce apoptosis in BMDC, we performed control experiments using propidium iodine staining. Neither tumor cells nor tumor cells in combination with TLR ligands induced significant cell death (data not shown). These data indicate that tumor cells-induced BMDC differentiation occurs even in the presence of TLR ligands and that such an effect is not induced by normal cells.

Prior exposure of DC to tumor cells alters the differentiation of DC affecting the activation through TLR receptors

Next it was important to examine whether prior exposure to tumor cells would affect the ability of BMDC to respond and become activated by a posterior stimuli with TLR ligands. We cultured DC in the presence of IrrTC-1, IrrB16 tumor cells or TLR ligands (i.e. LPS or poly IC). 96 h later CD11c⁺ cells were positively selected and cultured in the absence of tumor cells with LPS or poly IC. As shown in Fig. 5a, BMDC pre-cultured with tumor cells failed to respond to TLR stimuli. Moreover, as expected, BMDC pre-cultured in the presence of LPS did not respond to a second stimuli with LPS or with poly IC [25]. On the other hand, cells cultured with poly IC responded not only to poly IC but also to LPS (Fig. 5a). It is important to point out that BMDC were viable at the moment they were re-stimulated as assessed by trypan blue exclusion and flow cytometry after PI staining (data not shown).

Fig. 5.

BMDC differentiated in the presence of tumor cells failed to respond to TLR stimulation. a BMDC (5 days) were cultured 1:1 ratio (TC:DC) with irrB16 or irrTC-1 tumor cells (first stimuli, indicated in upper row). 96 h post-stimulation, CD11c⁺ cells were positively selected with MACS microbeads and cultured in the presence of 0.1 μg/ml LPS or 25 μg/ml poly IC for an additional 24 h (second stimuli, indicated in left column), after which cells were harvested and stained with CD11c-PE, CD11b-APC and FITC-conjugated anti-CD40 or CD86. b CD11c⁺ cells cultured with the indicated stimuli were selectived after 96 h and stimulated with LPS (0.1 μg/ml). Induction of CD40 was measured by flow cytometry and mean values of MFI from three independent experiments were calculated as a percentage of control LPS responses. Results presented as mean ± SE. ** Not significant. c CD11c⁺ cells incubated in the presence of BFA for evaluation of intracellular cytokine production 24 h after the addition of stimuli. Results represent a minimum of three experiments

Next we examined whether the inhibitory effect of tumor cells was dependent on the number of tumor cells in the culture. We found that tumor cells inhibited LPS-induced expression of co-stimulatory molecules in a dose-dependent manner (Fig. 5b). Pre-treatment of BMDC with high number of tumor cells (1:1 ratio) completely inhibited LPS-induced CD40 expression, whereas low dose tumor cells (1:10 ratio, TC:DC), did not interfere significantly with LPS-induced expression of CD40 (Fig. 5b). It is important to point out that a low dose of tumor cells (1:10, TC:DC ratio) was also insufficient to induce co-stimulatory or inhibitory molecules (Fig. 1).

Finally, we also examined cytokine expression after a second stimulus with LPS. As shown in Fig. 5c, after 96 h of pre-cultured in the presence of tumor cells, purified CD11c⁺ cells stimulated with LPS did not express IL-12 (Fig. 5c), TNF-α or IL-6 (data not shown). Thus, it appears that tumor cells impaired the subsequent BMDC activation through TLR stimuli. It is worth noting that pre-culturing of BMDC with syngeneic splenocytes did not interfere with LPS-induced expression of CD40 (Fig. 5b) or IL-12 production (Fig. 5c).

Importantly, the TLR ligands LPS and poly IC are potent inducers of CD86, CD40 and MHCII (Fig. 1) and cytokine production (Fig. 3). However, when BMDC were exposed to tumor cells before the addition of poly IC or LPS, there was no induction of these surface markers and BMDC resembled tumor differentiated-BMDC. Thus, the limited surface and cytokine phenotype of BMDC cultured with tumor cells do not appear to be due to an absence of activating signals from tumor cells. Rather, it appears that tumor cells actively alter the pathway of BMDC differentiation to a phenotype characterized predominantly by the expression of molecules involved in the negative control of immune responses.

Discussion

Adaptive immune responses are induced, coordinated and regulated by DC [4]. DC possess dual functions and can either induce/up-regulate or suppress/down-regulate immune responses. Mature DC induce strong T helper 1 (Th1)-type immune response and are considered potent inducers of tumor-specific immunity [38]. However, cancer cells can block DC maturation by a variety of mechanisms that includes the production of vascular endothelial growth factors VEGF [22], IL-6 [50], M-CSF [50], TGF-β [67], IL-10 [6], COX2 [34], PGE2 [37, 44] and glangliosides [60]. Immature DC presenting antigens to T cells in the absence of co-stimulation lead to T cell anergy/deletion [64] or promote the clonal expansion of CD4⁺CD25⁺ regulatory T cells [15]. In our current study, contrary to the immature phenotype reported before, we found that BMDC exposed to a variety of tumor cells acquired a phenotype characterized by predominant expression of inhibitory molecules, such as B7-DC, B7-H1, CD80, and the absence of cytokines related to T cell activation and differentiation. Moreover, we demonstrated that high numbers of tumor cells function as counter-regulators of BMDC activation induced by the TLR3 and 4 ligands, poly IC and LPS, respectively. These results reveal an unusual DC differentiation in the presence of tumor cells.

In the present study, tumor cells inhibited LPS-induced up-regulation of co-stimulatory molecules CD40 and CD86 (Figs. 4, 5); both necessary for initiation of T cell responses [4]. Moreover, LPS-induced surface expression of MHCII was significantly reduced by tumor cells. Tumor cells also strongly inhibited cytokine expression in LPS-treated BMDC. Furthermore, LPS-induced expression of cytokines and surface co-stimulatory molecules (data not shown) was still inhibited in tumor treated-BMDC even with higher concentrations of LPS (1 μg/ml; Fig. 4 or 10 μg/ml, data not shown). However, when the number of tumor cells was reduced in the co-culture, tumor cells were no longer capable of blocking LPS-induced signaling (Fig. 5b). These results strongly suggest that high numbers of tumor cells block LPS-induced maturation of BMDC.

Similar to the effects on LPS-induced DC maturation, co-cultured BMDC with tumor cells prevented induction of co-stimulatory molecules and production of cytokines in BMDC activated with poly IC (Figs. 4, 5), which signals through TLR3. Different from TLR4 that uses MyD88-dependent pathway, TLR3 uses a TICAM-1- and IFN-regulated factor 3-mediated pathways for the production of IFN-β in response to pathogen recognition [33]. In contrast, when BMDC were co-cultured with syngeneic splenocytes, neither LPS- nor poly IC-induced expression of co-stimulatory molecules and cytokine production was altered (Figs. 4, 5). These data identified tumor cells as strong negative regulators of TLR-induced maturation of BMDC.

Of note, tumor cells secrete a wide variety of factors [6, 22, 34, 37, 44, 60, 67] that potentially “modify” the ability of DC to response to TLR ligands. Given this observation, we sought to investigate the influence of tumor cells culture supernatant in our system. Surprisingly, we found that tumor cells seems to use soluble factors as well as cell–cell contact mechanisms since tumor cells supernatant was insufficient to completely inhibit LPS- and poly IC-induced maturation (data not shown). The soluble and surface tumor molecules participating in the inhibition of the DC maturation induced by TLR ligands in our system remain(s) unknown and requires further investigation.

The presence of functionally mature DC is rare in cancer patients [16, 52] and tumor-bearing mice [20, 32]. When we co-cultured BMDC with tumor cells, we did not observe typical signals of activation such as prolonged and stable induction of CD40, CD86 and cytokine production. Most importantly, BMDC cultured with tumor cells up-regulated several markers associated with the inhibition of immune responses, such as B7-H1, B7-DC and CD80 (see subsequently). Cancer is a chronic disease and by the time it is clinically noticed, it has been interacting with and affecting host cells for a long time to ensure its survival. Our results suggested that the first encounter between tumor cells and BMDC preferentially induced inhibitory molecules in DC. However, when BMDC were co-cultured with tumor cells for 96 h and then tested for the expression of surface markers, we did not observe appreciable expression of co-stimulatory or inhibitory molecules (Fig. 5a). This final result can be misunderstood as DC ignorance or a lack of activation signals in the tumor microenvironment. In contrast, if these 96 h-tumor exposed-BMDC were re-stimulated, neither LPS nor poly IC induced the expression of co-stimulatory molecules or cytokines production (Fig. 5). Our results strongly suggest that tumor cells induced and maintained a TLR non-responsive state in BMDC.

A considerable obstacle to the success of DC-based cancer vaccines might be the presence of T cells with regulatory function. Although regulatory T cells play a critical role in preventing autoimmune diseases by suppressing host immune responses against self- or non-self antigens, they might inhibit anti-tumor immunity and promote tumor growth. Several data are in agreement with this inhibitory role. An increased frequency of CD4⁺CD25⁺ T cells has been observed in the peripheral blood and tissues of patients with cancer [43, 73]. Mouse CD4⁺CD25⁺ T cells constitutively express CTLA4 [17] and administration of blocking antibodies against CTLA4 [65] or CD25 [53] in tumor-bearing mice improves immune-mediated tumor clearance and enhances the responses to immune-based therapy. Moreover, in mice bearing B16 melanoma cells, it was formally demonstrated that Treg cells are the major regulators of tumor immunity [71]. Notably, in our current studies, co-culturing BMDC with high numbers of tumor cells (1:1, TC:DC ratio) triggered, among other molecules, CD80 expression on DC (Fig. 1a). It was shown that CD80 has a much higher affinity for CTLA4 than for the co-stimulatory receptor CD28 [57]. Thus, in vivo these differentiated DC might stimulate Treg cells through CTLA4–CD80 interaction. In addition, CTLA4 can transduce signals in DC. It has been shown that CTLA4-Ig treatment, in vitro and in vivo, increases the expression of the enzyme indoleamine 2,3-dioxygenase (IDO) in DC [26]. Moreover, tumor cells may also induce the expression of IDO in DC. PGE2 has been described to up-regulate functional IDO during DC maturation [9], and many tumors have shown to be associated with elevated levels of PGE2. IDO activity results in the depletion of the essential amino acid tryptophan [66] and the accumulation of toxic downstream metabolites. Both of these processes inhibit T cell activation and induce T cell apoptosis [19, 49, 68]. Thus, IDO-expressing DC have the potential to be profoundly immunosuppressive [49].

Another molecule expressed in Treg cells is PD-1 [12], now accepted to be a receptor for two B7 family members, B7-H1 and B7-DC. Further, tumor factors stimulate B7-DC and B7-H1 expression in bone marrow-derived DC [13], and importantly, we showed here that these two inhibitory molecules are highly up-regulated in BMDC co-cultured with tumor cells (Fig. 1b). In vivo, DC production of IL-12 might be reduced after B7–PD1 interaction, reducing DC immunogenicity [13].

The presence of immature DC, triggered by different tumor factors, and the absence of inflammatory cytokines have been considered as the main cause of poor tumor immunity [48, 61]. Given that DC activation may be modified by pro-inflammatory stimuli, TLR-ligands have been suggested for cancer therapy [48, 61]. Recent studies demonstrated that TLR ligands can be used in the induction of anti-cancer immunity, although their use only appear to be effective in prophylactic settings [70]. Moreover, previously it was demonstrated that the presence of B16 tumor cells did not inhibit the ability of BMDC to up-regulate co-stimulatory molecules in response to a cocktail of TLR ligands [30]. However, those studies were performed with low numbers of tumor cells [30]. Our data clearly showed that neither LPS nor poly IC induced BMDC activation in the presence of tumor cells. However, the inhibitory effect was only observed at high numbers of tumor cells. Low numbers of tumor cells showed no ability to induce inhibitory molecules or to block the activation induced by TLR ligands.

LPS-TLR4 recruits both MyD88 and TICAM-1 adapter molecules. On the other hand, poly IC activates only through the adapter TICAM-1. Another possible approach is the use of cytosine-phosphate-guanosine (CpG). The recognition of the specific CpG motif is mediated by TLR9 [29] and activates only through MyD88 adapter molecule. In combination with irradiated tumor cells or tumor antigens, CpG showed anti-tumor efficacy but only in prophylactic settings [23]. However, patients with melanoma exhibited rapid and strong antigen-specific T cell responses when melanoma peptides are inoculated with CpG and incomplete Freund’s adjuvant [63]. Further studies are being continued in our system to clarify the possible use of TLR9 ligands for the induction of activated-BMDC in the presence of tumor cells.

Tumor cells actively develop different mechanisms to escape tumor immunity [74]. DC have an important role in the activation of tumor antigen-specific immunity. However, tumor cells inhibit DC activation and differentiation suppressing anti-tumor immunity. Therapeutically, agents that can selectively and efficiently induce the full-activation and function of DC are desired. According to our findings, BMDC failed to respond to the TLR ligands LPS and poly IC in the presence of high number of tumor cells, providing a possible explanation for the lack of success of therapeutic approaches based on DC maturation stimuli. Thus it might be crucial to understand the differentiation and activation of DC in the presence of tumor cells for designing future strategies using tumor antigen-pulsed DC for tumor immunotherapy.

Abbreviations

BMDC

Bone marrow-derived dendritic cells

Irr

Irradiated

TC

Tumor cells