Mechanisms of Ganglioside Inhibition of APC Function¹

It is well recognized that generation of immunosuppressive factors by tumor cells may contribute to the escape of tumor cells from host immune destruction. For example, within the tumor microenvironment suppression exists that disrupts the action of the tumor-infiltrating lymphocytes (2, 3, 4, 5). One class of molecules with a potential to interfere with the antitumor immune response is the gangliosides. Gangliosides consist of an oligosaccharide core with an attached sialic acid(s) and a ceramide and are found primarily in the outer leaflet of the cell membrane. Many tumors, such as neuroblastoma, medulloblastoma, and renal cell carcinoma, shed membrane gangliosides into the microenvironment. These biologically active molecules are efficiently bound to target cells and are immunosuppressive (6, 7, 8, 9, 10, 11, 12). In vivo, coinjection of gangliosides with poorly tumorigenic cells increases their tumorigenicity (8, 13).

APC and T cells engage in a series of complex and interconnecting signals to trigger a cellular immune response. Ag processing and presentation by APC (signal 1) allow T cells to recognize Ags. Several cytokines and costimulatory molecules are then up-regulated on both APC and T cells and interact to yield what is referred to as signal 2, or costimulation. APC cell surface molecules of central importance to costimulation are CD80 and CD86 (14, 15). Their up-regulation on APC is triggered through interaction of the constitutively expressed CD40 molecule on the APC with up-regulated CD40 ligand (CD154) on the T cell (16, 17). In addition, several cytokines, including IL-6, IL-12, and TNF-α, participate in these steps of activation, costimulation, and proliferation and are important in the induction of APC Ag uptake and processing, migration, lymphocyte recruitment, APC up-regulation of cell surface molecules, and effective APC-T cell interaction (18, 19, 20, 21). We previously found that exogenous gangliosides inhibit monocyte APC function and identified a direct effect on monocytes (12, 22).

To comprehensively examine how gangliosides directly affect APC, we studied two different human APC populations: monocytes isolated from peripheral blood and monocyte-derived dendritic cells (DC).³ We investigated the influence of preincubation with ganglioside G_D1a on monocyte stimulation in both T cell-dependent (tetanus toxoid (TT)) and T cell-independent (LPS) assays. We also studied the effects of G_D1a preincubation on human DC, both with and without Ag (TT) stimulation and with and without T cell addition. Under these multiple experimental conditions, the uniform finding was that G_D1a preincubation directly affected the APC in their ability to stimulate T cell proliferation, and strikingly, that the costimulatory molecules CD40 and CD80; the cytokines IL-6, IL-12, and TNF-α; and the nuclear translocation of NF-κB all were inhibited by preincubation of APC with G_D1a.

Heparinized blood was obtained from normal donors after they gave informed consent. The plasma was collected, and PBMC were enriched by Ficoll-Hypaque gradient centrifugation and resuspended in complete HB104 medium (Irvine Scientific, Santa Ana, CA) that includes 2 mM sodium pyruvate, 1 mM l-glutamine, penicillin (50 U/ml), streptomycin (50 mg/ml), and 1% protein supplement (albumin, insulin, and transferrin).

Adherent monocytes were obtained by incubating PBMC (2–4 × 10⁶/ml) in complete HB104 with 10% autologous plasma for 2 h at 37°C in a humidified 5% CO₂/95% air atmosphere. The nonadherent lymphocytes were removed and resuspended in complete HB104 containing 10% autologous plasma and 1% HEPES, and the adherent cells were recovered by incubation with 0.5 mM EDTA in PBS at 4°C.

CD14⁺ monocytes and CD4⁺ T cells were enriched by magnetic cell sorting negative selection (autoMACS; Miltenyi Biotec, Auburn, CA) according to the manufacturer’s protocol. Briefly, to negatively select CD14⁺ cells, PBMC were resuspended in PBS containing 2 mM EDTA; mixed with an Ab cocktail containing hapten-conjugated Abs against CD3, CD7, CD19, CD45RA, CD56, and IgE for 5 min at 8°C; washed; and mixed for 15 min with colloidal superparamagnetic MACS microbeads conjugated to an anti-hapten mAb. Then the cells were washed and applied to metal matrix columns in the autoMACS separation apparatus. Non-Ab-coated (negative) cells were collected and washed for further study. CD4⁺ T cells were similarly negatively selected using a cocktail containing hapten-conjugated Abs to CD8, CD11b, CD16, CD19, CD36, and CD56.

DC were generated by incubating CD14⁺ monocytes in complete HB104 with 10³ U/ml of IL-4 (BD PharMingen, San Diego, CA), 40 ng/ml of GM-CSF (R&D Systems, Minneapolis, MN) (23, 24), 1% HEPES, and 20% autologous plasma for 7 days. DC development was monitored by observation of characteristic morphological changes, including increases in cell size and dendrite formation.

Disialoganglioside G_D1a (≥99% pure by HPLC; Matreya, Pleasant Gap, PA) was dissolved in complete HB104 medium and sonicated to assure complete resuspension before use. Sonicated medium alone served as the control.

Monocytes or DC were preincubated in complete HB104 containing 1% HEPES and 0.5% autologous plasma with or without up to 50 μM G_D1a for 48 h. For the Ag-stimulated T cell-dependent assays, 0.9 limit of flocculation units/ml TT (Connaught Laboratories, Swiftwater, PA) was added to the wells during the last 24 h of the 48-h preincubation. To assess T cell-independent stimulation, monocytes were incubated with G_D1a for 48 h, washed to remove unincorporated ganglioside, and incubated with 5 ng/ml of LPS (Sigma-Aldrich, Natick, MA) for 24 h.

T cell proliferative responses were quantified after culture with G_D1a-preincubated, TT-exposed monocytes or DC, which were washed to remove unincorporated ganglioside and Ag. The autologous CD4⁺ T cells were obtained by negative selection. The two cell populations were then cocultured in complete HB104 containing 1% HEPES and 0.5% autologous plasma in 96-well plates at an APC/T cell ratio of 1/10. The cultures were incubated for 7 days (in the case of monocytes) or 5 days (in the case of DC) and then pulsed for 4 h with 0.5 μCi/well of tritiated thymidine (American Radiolabeled Chemicals, St. Louis, MO) and harvested onto glass-fiber filter paper, and cellular [³H]thymidine uptake was quantified by beta scintillation counting. All cultures were performed in triplicate. Inhibition of proliferation was determined by comparing the mean net counts per minute of ganglioside-preincubated cultures to that of stimulated cultures not preincubated in gangliosides (12, 25).

APC recovered by EDTA treatment were washed and stained using optimal concentrations of specific or isotype control Abs. Anti-mouse IgG1-FITC, anti-mouse IgG1-PE, anti-LFA-3 (CD58)-PE, anti-ICAM-1 (CD54)-PE, anti-HLA-DR-FITC, anti-CD14-FITC, anti-CD80-PE, and isotype controls were obtained from BD Biosciences (San Jose, CA). Anti-CD40-FITC and anti-CD86-PE were obtained from BD PharMingen. After incubation on ice in the dark for 30 min, the cells were washed twice in ice-cold HBSS containing 0.5% BSA and 0.1% sodium azide and resuspended in HBSS. 7-amino actinomycin D (BD PharMingen) was added as a viability stain. Cells were analyzed on a FACStar Plus flow cytometer (BD Biosciences). Monocyte cell surface Ag expression was assessed by gating the CD14-positive population. Changes in cell surface Ag expression are expressed either as the percentage of cells expressing an Ag or as the median fluorescence intensity, as indicated.

Supernatants from ganglioside- and/or TT-preincubated monocytes and DC cultures were harvested and analyzed for cytokine secretion by ELISA, using the BioSource kit protocol for IL-6, IL-12, and TNF-α (BioSource, Camarillo, CA). Cytokine data represent quantitative values or are based on OD readings of undiluted samples, comparing G_D1a-exposed cell cultures to control cultures, as indicated.

DC were incubated with 50 μM G_D1a for 72 h and/or 0.9 limit of flocculation units/ml of tetanus toxoid during the last 24 h. Using a modification (26) of the method described by Digman (27), the cells were then washed with PBS and incubated in lysis buffer (1 mM HEPES, 0.5 mM EDTA, 1 mM KCl, 1 mM DTT, 100 mM PMSF, and 100 mM sodium vanadate, adjusted to pH 7.9) on ice for 15 min. After centrifugation at 600 × g at 4°C for 10 min, the nuclear pellet was resuspended in nuclear lysis buffer (10 mM HEPES, 5 mM EDTA, 150 mM KCl, 0.05% SDS, 1% Triton, 20 mM NaF, 20 mM sodium pyrophosphate, 20 mM β-glycerophosphate, 20 mM sodium molybdate, 100 U/ml aprotinin, 50 μg/ml leupeptin, and 1 mM DTT) and incubated for 15 min on ice. The lysate was freeze-thawed three times in a dry ice-ethanol bath. The nuclear protein concentration was determined using a Bio-Rad protein quantification kit (Bio-Rad, Hercules, CA). Equal amounts (20 μg) of protein were loaded onto a 10% SDS-polyacrylamide gel, subjected to electrophoresis for 2 h, and transferred to a nitrocellulose membrane. After exposure to blocking buffer (50 mM Tris base, 150 mM NaCl, and 0.1% Tween 20 containing 5% BSA) for 1 h at room temperature, the membrane blots were incubated overnight at 4°C with Abs (1/1000) specific for the NF-κB proteins p50, p65, RelB, and C-Rel (Santa Cruz Biotechnology, Santa Cruz, CA), and then incubated with HRP-conjugated secondary Abs (1/2000) for 1 h. Specific Ab-conjugated protein bands were detected by chemiluminescence and exposure to x-ray film.

All results are reported as the mean ± SEM of two to six separate experiments unless otherwise indicated. The significance of differences was determined using Student’s paired comparison t test.

Monocyte preincubation with G_D1a ganglioside markedly reduced the ability of APC to induce a T cell proliferative response to TT (81 ± 8% inhibition; p = 0.009; Table I). This effect of purified G_D1a was similar to that previously observed using a mixture of purified total brain gangliosides (12) and provides us with the model homogeneous molecule for the present experiments to elucidate mechanisms of inhibition. Inhibition was observed in the absence of any toxic effect on the APC; the viability of monocytes preincubated with 50 μM G_D1a for 72 h was 98%, as assessed by Trypan Blue dye exclusion, also consistent with previous findings (12).

We also investigated the functional consequences of ganglioside exposure of human monocyte-derived DC to extend these findings. After being incubated with G_D1a for 48 h and TT during the last 24 h, DC also showed an impaired ability to induce memory T cell proliferation (mean inhibition, 83 ± 9%; p ≤ 0.03; Table II), showing that the function of even the most effective APC was directly inhibited by ganglioside exposure.

To delineate the effect of G_D1a on the three critical steps of interaction between the APC and T cell (adhesion, Ag presentation, and costimulation), we assessed the expression of adhesion, MHC class II, and costimulatory molecules. Monocytes preincubated in G_D1a showed very little difference from control monocytes in the expression of CD40, HLA-DR, or the cell adhesion molecules LFA-3 and ICAM-1 after exposure to TT and incubation with T cells (Table III). Strikingly, however, in five separate experiments a marked inhibition (92 ± 2%; p ≤ 0.0001) of up-regulation of the density of expression of the costimulatory molecule, CD80, was observed. Curiously, there was no reduction in the expression of CD86 (Fig. 1).

Several additional experiments further characterized the inhibition of CD80 expression. First, we investigated whether G_D1a might be masking the expression of CD80 induced by TT exposure. After a 44-h exposure of monocytes to TT, 38% of the cells were CD80 positive. In contrast, exposure to both G_D1a and TT resulted in marked inhibition, to 2% expression (Fig. 2 A), demonstrating the expected ganglioside inhibitory effect. However, there was virtually no inhibition of CD80 expression (35% of cells were positive) when G_D1a was added just before cell harvest for FACS analysis. Thus, the reduced expression of CD80 was not caused by masking of this cell surface molecule.

Secondly, to assure that G_D1a was not acting by altering the kinetics of CD80 expression, we added the ganglioside at the beginning of culture and monitored the degree of inhibition at later time points. There was no increase in CD80 expression at later points, indicating that the addition of G_D1a did not simply delay the onset of CD80 expression (Fig. 2 B).

To exclude that the effect of ganglioside preincubation on monocytes was solely dependent on an interaction with T cells (present in the previous experiments), we studied the effect of preincubation with 50 μM G_D1a on LPS-induced monocyte activation. A representative flow cytometric profile (Fig. 3) and composite data from three separate donors (Fig. 3, bar graph) are shown. CD14 expression was not reduced by monocyte preincubation in G_D1a (data not shown), but these LPS-stimulated monocytes revealed inhibition of both CD40 and CD80 expression as assessed by FACS (Fig. 3, bar graph). No significant effect on CD86 was observed. These studies with LPS-stimulated monocytes show that the reduction of expression of the costimulatory molecules CD40 and CD80 by ganglioside exposure was a direct effect on monocytes, occurring even in the absence of monocyte-T cell interactions.

The final type of APC studied was the DC. In contrast to monocytes, DC constitutively express CD80 and CD86 (28, 29), allowing us to determine whether ganglioside exposure affected the ability of the APC to constitutively express and/or up-regulate the expression of costimulatory molecules and to initiate a proliferative response. First, despite this constitutive expression, the proliferative response of T cells to ganglioside-preincubated, TT-pulsed DC was inhibited (Table II). Then, in assays parallel to those in Table II, incubation of DC with G_D1a and stimulation with TT produced a marked reduction of the median fluorescence intensity of CD80 (61 ± 9%; p ≤ 0.007) and a decrease in CD40 expression of 33 ± 6% (p ≤ 0.02) compared with control TT-pulsed DC (Fig. 4, histogram and bar graph, ▪). Thus, G_D1a exposure resulted in decreased expression of the constitutively expressed costimulatory molecules CD40 and CD80.

To determine whether interaction of DC with T cells might overcome the inhibition, T cells were added to TT-stimulated, ganglioside-preincubated DC. As also shown in Fig. 4 (bar graph, ▪), 24 h after the addition of T cells a similar pattern of alteration of the expression of costimulatory cell surface molecules was observed, i.e., a 45 ± 5% (p ≤ 0.01) decrease in CD40 expression, a 46 ± 11% (p ≤ 0.03) decrease in CD80 expression, and no decrease in CD86 expression. Together with the observed inhibition of T cell proliferation under these conditions, these results suggest that ganglioside exposure results in a significant reduction of the costimulatory molecule expression that is essential for mounting appropriate T cell responses by monocyte-derived APCs.

To avoid a potential contribution of APC-T cell interactions and to assure that we were measuring changes in cytokine release by the APC themselves, we studied whether the release of cytokines known to be involved in LPS-induced monocyte activation (30, 31, 32) was affected. The release of both TNF-α and IL-12 by purified monocytes exposed to LPS and varying concentrations of G_D1a was reduced in a concentration-dependent manner (Fig. 5). Thus, cytokine release linked to the expression of costimulatory molecules is inhibited in ganglioside-preincubated APC.

DC were the second system studied for cytokine release. Cultured DC constitutively produce and secrete IL-6, IL-12, and TNF-α (33, 34). Preincubation of DC in 50 μM G_D1a for 48 h decreased the constitutive secretion of all three cytokines (Fig. 6, left panel). Subsequent exposure of ganglioside-preincubated DC to TT did not reverse the inhibition of cytokine secretion (Fig. 6, right panel), indicating that antigenic stimulation of G_D1a-preincubated DC did not restore normal cytokine production. Thus, G_D1a exposure clearly reduces constitutive production and release of cytokines important to DC activation.

NF-κB is basic to the induction/maintenance of DC activation, including the expression of cell surface molecules and the production of cytokines that were shown to be reduced by exposure of APC to G_D1a. The expression of the NF-κB proteins p50, p65, RelB, and C-Rel in DC and their localization to the nucleus were examined after DC were incubated in G_D1a for 72 h, with TT added during the last 24 h (Fig. 7). As expected, nuclear binding of all four proteins (p50, p65, RelB, and C-Rel) was caused by TT stimulation of DC, in contrast to the detection of only traces of p50 and p65 in the nuclei of unstimulated DC. G_D1a exposure before TT activation resulted in reduced nuclear binding of all four NF-κB proteins (Fig. 7). This suppression of nuclear translocation of NF-κB proteins in DC suggests that G_D1a may affect the transcriptional regulation of genes critical to the immune response, resulting in the pleiotropic effects we observed.

The present study extends the understanding of the immunoregulatory role of gangliosides by demonstrating a significant inhibitory effect on the capacity of APC to initiate protective T cell responses. This may underlie the inhibitory effect of monocyte preincubation with G_D1a on Ag-dependent monocyte stimulation of T cells (Table I) (35, 36, 37, 38). In previous studies of Ag uptake and processing by monocytes, we found that despite priming of monocytes with TT to allow the full generation of MHC/Ag complexes, subsequent incubation in 100 μM HBG for 24 h inhibited monocyte triggering of the T cell proliferative response to the same degree as that of monocytes exposed first to gangliosides and then to TT. Since the Ag prepulse did not overcome ganglioside inhibition of Ag presentation by monocytes, the results suggest that gangliosides affect a step after the generation of MHC/Ag complexes (12). Together with the present data showing no interference with the expression of cell surface molecules important for adhesion, interference with subsequent steps is suggested. Thus, with respect to the paradigm that two distinct signals are essential to the induction of cellular immune responses, the present findings suggest that APC exposure to exogenous gangliosides leaves signal 1 intact (cell surface molecule expression required for Ag recognition and cell adhesion, e.g., MHC II, LFA-3, and ICAM-1), but significantly interferes with the potential of APC to deliver the costimulatory signal 2. Both the generation and maintenance of costimulatory properties of monocytes as well as of the even more potent APC, monocyte-derived DC, were profoundly altered by exogenous gangliosides. G_D1a that had become associated with the APC membrane 1) prevented the induction and maintenance of the expression of CD80 and CD40 on the APC cell surface; 2) reduced the production of cytokines IL-6, IL-12, and TNF-α; and 3) interfered with the nuclear binding of NF-κB, a key regulator of the costimulatory regulatory cycle.

At the level of cell surface molecule expression required to provide optimal costimulatory signals, our experiments showed that exogenous G_D1a had a direct and sustained effect on the expression (induction as well as preservation) of the costimulatory molecule CD80 by APC. The significance of the costimulatory signal in eliciting a T cell response is known (39, 40), and the expression of CD80 is considered essential in directing T cell responses toward initiating the effector arm of an immune response. Particularly with respect to antitumor immunity, a reduction or absence of CD80 in APC has been associated with increased tumorigenicity in in vivo tumor models (8).

With respect to the consistently observed effects on CD80, but not CD86, our current understanding of functional effects of selective engagement of CD80 and CD86 remains incomplete. While some studies suggest that the functions of CD80 and CD86 are overlapping (41), the present study, by showing that a selective decrease in CD80 expression by exposure to gangliosides was associated with a reduced capability of mounting a T cell proliferative response suggests that, in line with the observations of others (14, 42), the expression of CD86 could not substitute for an impaired expression of CD80. The fact that the expression of CD86 on monocytes and DC was not affected by exposure to G_D1a is, in fact, consistent with some previous observations. That is, costimulatory molecule expression can be regulated by a number of different pathways (43), and it is known that the expression of CD80 and CD86 can be independently regulated, even by the same stimulus (44). Moreover, previously suggested functional consequences of a selective deficiency of either CD80 or CD86 molecules include a correlation between CD86 expression and Th2-shifted immune responses (45, 46). Due to the substantially higher binding affinity of CD80 to CTLA-4, which down-regulates (in comparison with CD28, which up-regulates) T cell responses (47), the expression of only low levels of CD80 by APC (as could be caused by ganglioside exposure) may direct immune responses toward the induction of tolerance (48).

The expression of CD80, in turn, is regulated by ligation of CD40 to CD154 expressed on activated T cells (33, 49). Reduced expression of CD80 by ganglioside-exposed APC may therefore be linked to the reduced expression of CD40. In fact, triggering of CD40 has been found to be critical to enable APC to cross-prime CD8⁺ cells and to up-regulate costimulatory molecules (including costimulatory cytokines such as IL-12) in tumor settings, thereby turning tumor tolerance into effective antitumor immunity in vivo (50, 51). The reduced expression of CD40 and CD80 caused by ganglioside exposure may therefore cause an APC phenotype that leads to induction of tolerance rather than initiation of the effector arm of T cell responses. Chaux et al. (5) demonstrated reduced CD80 expression on tumor-infiltrating DC in a rat colon carcinoma and found that these DC were unable to stimulate T cells in vitro. Similarly, in the present study DC exposed to G_D1a expressed reduced levels of CD80 and were unable to induce normal Ag-dependent T cell proliferative responses.

To further trace the impaired costimulatory capacity caused by ganglioside exposure we investigated the release of related cytokines. These include IL-12 and TNF-α, which have been related to APC expression and function of costimulatory molecules (51, 52). TNF-α up-regulates both CD40 and CD80 (24, 52); IL-12 and TNF-α possess synergistic potential in enhancing antitumor responses (53, 54). IL-12 may cooperate with CD80 in causing immune-mediated tumor regression (55), since IL-12 induced tumor reduction of a CD80⁺, but not a CD80⁻, squamous cell carcinoma. IL-12 also directs the T cell response toward a Th1 subset and prevents a Th2 response (56). These observations underscore the significance of the G_D1a-induced reduction of IL-12 and TNF-α secretion by LPS-stimulated monocytes and of IL-6, IL-12, and TNF-α by DC.

The inhibition of TNF-α production by G_D1a exposure of DC, previously shown for LPS-stimulated monocytes (10), provides one possible explanation for the observed down-regulation of IL-12, CD40, and CD80, since TNF-α up-regulates the secretion of IL-12 (57) and the expression of CD40 and CD80 (24, 52), both of which were inhibited by G_D1a exposure of APC. Since the secretion of IL-6, IL-12, and TNF-α were all significantly reduced regardless of whether the DC were also Ag pulsed, it appears that the ganglioside action occurs at an early DC stage, influencing cytokines involved in vital maturation processes, rendering the APC unable to overcome the inhibition, even after antigenic stimulation.

The fact that multiple steps were affected by ganglioside exposure of APC also suggested that we examine an effect on NF-κB, since NF-κB plays an important role in regulating an immune response. The lack of NF-κB in the nucleus can disrupt the transcription of pertinent genes to APC maturation and activation. NF-κB activation has been shown to be involved in the transcription of TNF-α, IL-6, IL-12, CD40, and possibly CD80 (58, 59, 60, 61), all of which are affected by APC exposure to G_D1a. The data suggest that in APC, NF-κB could be a target of G_D1a. Indeed, the fact that the nuclear concentrations of all four NF-κB proteins that we studied were reduced in G_D1a-treated DC suggests that further study is warranted and that G_D1a exposure may affect transcriptional regulation of some of the genes critical and central to the pleiotropic effects we observed. This is consistent with analogous findings obtained in T cells from renal cell carcinoma patients (9) and with the lack of NF-κB activity in a monocyte-derived cell line exposed to G_D1a (10). Why all genes that encode for costimulatory, adhesion, and MHC class II molecules and have κB sites are not affected by G_D1a exposure of APC is not clear. The possibilities include that some of these, such as LFA-3, MHC class II, CD86, and ICAM-1, may be regulated by other factors as well, or their promoters are variably (and in this case less) sensitive to κB regulation.

Collectively, the findings are highly relevant to the hypothesis that there exists a ganglioside-related pathway of tumor escape from host immune surveillance. Exposure of APC to elevated concentrations of gangliosides shed by tumor cells into the tumor microenvironment may prevent APC from elaborating signals critical to the stimulation of normal T cell responses. This is particularly significant because such blunted T cell responses have been observed in vivo (8). Considering the essential role of appropriate costimulation in inducing effective antitumor immunity (62), impaired APC function by ganglioside exposure may result in Ag ignorance by tumor-infiltrating T cells. To this point, even those signals that were maintained after ganglioside exposure in our experiments (e.g., CD86 expression and IL-10 secretion) are dedicated to induce tolerance rather than an effector T cell response.