Dr M. Wölfl, Klinik für Kinderheilkunde, Zentrum für Kinderonkologie und -hämatologie, Joseph-Stelzmannstr.9, 50924 Köln, Germany. E-mail: address: Matthias.Woelfl@medizin.uni-koeln.de
Dendritic cell (DC) development and function is critical in the initiation phase of any antigen-specific immune response against tumours. Impaired function of DC is one explanation as to how tumours escape immunosurveillance. In the presence of various soluble tumour-related factors DC precursors lose their ability to differentiate into mature DC and to activate T cells. Gangliosides are glycosphingolipids shed by tumours of neuroectodermal origin such as melanoma and neuroblastoma. In this investigation we address the question of whether gangliosides suppress the development and function of monocyte-derived DC in vitro. In the presence of gangliosides, the monocytic DC precursors showed increased adherence, cell spreading and a reduced number of dendrites. The expression of MHC class II molecules, co-stimulatory molecules and the GM-CSF receptor (CD116) on the ganglioside-treated DC was significantly reduced. Furthermore, the function of ganglioside-treated DC was impaired as observed in endocytosis, chemotactic and T cell proliferation assays. In contrast to monocytic DC precursors, mature DC were unaffected even when higher doses of gangliosides were added to the culture. With regard to their carbohydrate structure, five different gangliosides (GM2, GM3, GD2, GD3, GT1b), which are typically shed by melanoma and neuroblastoma, were tested for their ability to suppress DC development and function. Suppression was induced by GM2, but not by the other gangliosides. These data suggest that certain gangliosides impair DC precursors, implying a possible mechanism for tumour escape.
There are many mechanisms as to how tumours escape immunosurveillance, one of which is defective antigen presentation due to tumour-associated factors . Dendritic cells (DC), the professional antigen-presenting cells, play a pivotal role in the development of antitumour immunity . There is ample evidence that tumours inhibit the differentiation, maturation and function of DC in vitro and in vivo[3,4]. DC that migrate into tumour lesions show an altered function with a reduced expression of MHC class II molecules and co-stimulatory molecules . These defective DC may not only fail to mount an effective antitumour response, but may even tolerize formerly activated tumour-specific T cells . Soluble tumour-derived factors such as IL-10, VEGF, IL-6 or M-CSF have been identified to be responsible at least in part for this mechanism [7–10]. However, additional unidentified tumour-related factors are likely to contribute to DC suppression.
Gangliosides are glycosphingolipids, consisting of a ceramide backbone connected to different carbohydrate and neuraminic acid residues, and exert multiple functions in various biological systems [11,12]. They can be detected in the serum of patients with neuroblastoma  and are correlated positively with tumour growth in cancer patients [14,15]. Gangliosides are mediators of immunosuppression in that they suppress lymphoproliferation and cytokine production [16–21] in various experimental settings. Gangliosides inhibit antigen presentation by monocytes  and by DC derived from CD34+ stem cells . In vivo, gangliosides inhibit a tumour-specific immune response, as demonstrated in a mouse model .
In this paper we sought to clarify the question of whether gangliosides have any influence on the in vitro differentiation of monocytes into DC. The major finding presented here is that the presence of gangliosides early in the DC culture inhibits the development of monocytes into DC and hampers the morphology, phenotype and function of the resulting DC. The data suggest that the inhibition of DC differentiation by gangliosides might be an important mechanism for tumours to escape immunosurveillance.
Materials and methods
Dendritic cell culture
DC were cultured according to the modified protocol reported by Jonuleit et al. . Briefly, monocytes were isolated from PBMC of healthy donors and incubated with X-Vivo15 (Bio Whittaker, Belgium), 0·5% autologous serum, 7 ng/ml granulocyte-macrophage colony stimulating factor (GM-CSF) (LeukomaxTM, Novartis, Switzerland) and 5 ng/ml interleukin 4 (IL-4) (Strathmann Biotech, Hannover, Germany). The first day of incubation with cytokines is referred to as day 1 of culture. Fresh medium and cytokines were added every other day. The maturation of immature DC was induced by adding tumour necrosis factor α (TNF-α) (10 ng/ml), IL-1β (10 ng/ml), IL-6 (10 ng/ml) (all from Strathmann Biotech) and prostaglandin E2 (1 µg/ml) (Sigma-Aldrich, Deisenhofen, Germany) on day 7. Mature DC were harvested on day 9. Standard culture results in>90% HLA-DR+, CD11c+, CD33+ cells with <5% of contaminating CD3+ or CD19+ or CD56+ cells.
Commercially available gangliosides were dissolved in PBS and sonicated briefly before being added to the DC cultures in different concentrations and at time-points, as indicated. The bovine brain ganglioside mixture (Matreya, Pleasant Gap, PA, USA and Calbiochem, Bad Soden, Germany) contains GD1a 55%, GM1 18%, GD1b 15%, GT1b 10% and others 2% (GQ1b, GM2, GD3, GD2) as indicated by the manufacturer. Bovine GM3 and GD3 were purchased from Matreya. GT1b and GM2 were purchased from Sigma-Aldrich, Deisenhofen, Germany. GD2 was purchased from Calbiochem. C18-Ceramide (Matreya) was used as a control. Concentrations are given in µm for GM2, GD2, GM3, GD3, GT1b and ceramide and in µg/ml for the ganglioside mixture.
Photographs were taken using an Axiovert25 microscope from Zeiss, Germany, and a Contax167MT camera. Film was purchased from Kodak.
Flow cytometric analysis
Phenotypic analyses were performed by flow cytometry using saturating concentrations of unconjugated monoclonal mouse-antihuman antibodies against the following antigens: CD1a, CD3, CD4, CD8, CD16, CD19, CD40, CD56, CD64, CD80, CD83, CD86 (all from BD Pharmingen, Heidelberg, Germany) and PE- or APC-conjugated antibodies against CD11c, CD14, CD33 HLA-DR (BD Pharmingen) and CD116 (GM-CSF-R) (Beckmann Coulter, Krefeld, Germany). After staining with the primary antibodies, cells were washed and stained with a FITC-labelled goat-anti-mouse secondary antibody (BD Pharmingen). Conjugated isotype-matched monoclonal antibodies were used as controls (BD Pharmingen). Viability was determined by double staining with propidium iodide (Sigma, Deisenhofen, Germany) and Annexin V-FITC (BD Pharmingen), as indicated by the manufacturer. For phenotype analyses dead cells and debris were excluded. Fluorescence analyses were performed on a FACScalibur flow cytometer (Becton Dickenson, Heidelberg, Germany).
PBMC, 4 × 105, were seeded to 96-well flat-bottom tissue culture wells (Costar). Monocytes were allowed to adhere for 2 h in X-Vivo15/0·5% autologous serum. After washing off the non-adherent fraction, the cells were incubated with X-Vivo/0·5% serum and cytokines as described for the DC culture. Various amounts of gangliosides were added. Assays were performed in triplicate. After incubation for 20 h, the medium was discarded and the cells were washed extensively three times with PBS. Cells were fixed with PBS/1% glutaraldehyde (Sigma) and stained with 0·1% crystal violet (Sigma) for 25 min. Plates were then washed extensively with H2O. Fifty µl of 0·2% Triton X-100 (Sigma) were added and absorption was measured at 565 nm using an enzyme-linked immunosorbent assay (ELISA) reader.
Mixed leucocyte reaction (MLR)
CD4+ T cells were isolated from PBMC by negative depletion using a T cell-negative depletion kit (Dynal, Hamburg, Germany) followed by further depletion of the CD8+ T cell subset with anti-CD8-conjugated beads (Dynal). The resulting CD4+ cells were>98% pure. The T cells were incubated in 96-well plates (2 × 105/well) with decreasing numbers of DC for 5 days; 0·5 µCi/well of [3H]-thymidine was added for the last 16 h. Then cells were harvested and the thymidine uptake was measured using a β-scintillation counter (Beckman, Germany).
Immature, 6-day-cultured DC were incubated with FITC-dextran (MW 30000, Molecular Probes, Leiden, Belgium) for 30–210 min at 37°C. For the assessment of background activity due to unspecific, external binding, control DC were incubated with FITC-dextran on ice. After incubation the cells were washed in PBS/5% FCS three times and fixed in PBS/0·1% PFA plus 0·005% trypan blue to quench unspecific binding. The cells were then analysed by flow cytometry. Inhibition was calculated as (MFI of untreated group – MFI of GLS-treated group)/MFI of untreated group.
Immature DC were harvested after 6–7 days of culture and washed extensively. The assay was performed in X-Vivo15/0·5% autologous serum. MIP-1α (Peprotech, Frankfurt, Germany) was added to 24-well plates at a concentration of 100 ng/ml. Approximately 1 × 105 DC in 100 µl medium were added to 5-µm pore size transwell inserts (Costar, Bad Soden, Germany) and incubated with the chemokine-containing wells for 4 h. After the removal of the transwell inserts, an aliquot of migrated cells was harvested and transferred to Truecount® tubes (Becton Dickenson) containing a defined number of fluorescent microspheres and analysed by flow cytometry. The analysis was stopped when 20% of the microspheres were counted. Gating on viable dendritic cells according to their appearance in the FSC/SSC channel revealed the number of migrated DC analysed per sample. Gating was validated using parallel samples of DC, stained with 7-AAD and antibodies against CD86 and HLA-DR. The absolute number of migrated cells was calculated as: (no. of cells acquired/no. of microspheres acquired) × no. of microspheres/sample. The absolute cell number was then normalized to 105 input cells.
For statistical analysis of FACS data the Mann–Whitney U-test for independent variables was used.
Gangliosides alter the morphology and adherence of monocytic DC-precursors
Monocytes were obtained from PBMC by plastic adherence and cultured in X-Vivo, supplemented with 0·5% autologous serum, IL-4 and GM-CSF as described in the Methods. When a single dose (1–40 µg/ml) of mixed gangliosides was added once on day 1 of culture and remained in culture, a dramatic change in cell morphology could be observed. After 12 h, the majority of the formerly loosely adherent cells became strongly adherent and exhibited cell spreading (Fig. 1a,b,e,f). The observed effect was dose-dependent as measured in adherence assays (Fig. 1g). These changes in morphology were also observed when the cells were incubated with GM2 (20 µm) (Fig. 1d), but not when GD2 was added (Fig. 1c). After 2–3 days the cells slowly regained their round shape, but the fraction of adherent cells remained higher and the cells showed smaller, irregular dendrites compared to the untreated group (not shown).
Gangliosides inhibit the up-regulation of MHC class II, co-stimulatory molecules and lineage-related molecules on DC
We next analysed the influence of gangliosides on DC phenotype. Cells were incubated with gangliosides on day 1 and examined by flowcytometry either as immature DC on day 7 or after cytokine-induced maturation on day 9. The expression of MHC class II molecules on immature, 7-day-cultured DC that were incubated with 40 µg/ml gangliosides on day 1 was reduced to more than 50% compared to the untreated group (median MFI 64 ± 19 versus 195 ± 58, P = 0·002, n = 6). Immature DC have a low expression of CD80 and CD86 and no CD83, therefore no significant differences in the expression of these markers at this early time-point could be detected (data not shown). However, these markers are usually strongly up-regulated after the induction of maturation with inflammatory cytokines for 2 days. Ganglioside-treated mature DC expressed reduced amounts of HLA-DR, CD80, CD83, CD1a and CD116 (GM-CSF-R). This was statistically significant for HLA-DR and CD116 and for the percentage of CD80+ cells (P = 0·001). There was a trend for a reduction in CD83 and CD1a+ cells; however, this did not reach statistical significance.The expression of CD86 and CD40 remained comparable to the untreated group (Fig. 2). Macrophage/monocyte markers such as CD14, CD16 and CD64 were negative in both groups. Lineage markers for CD3, CD19 and CD56 were consistently below 5% in all cultures. In all groups homogeneous DC populations according to their appearance in the FSC/SSC channel were observed and CD11c and CD33 as markers for monocytes/myeloid lineage were above 90% in all groups (data not shown).
Inhibition was not due to ganglioside-mediated toxicity, as viability assays with Annexin V and propidium iodide at different time-points gave similar results in all groups. For example, when gangliosides (40 µg/ml) were added on day 1 and analysed by flowcytometry on day 9, the fraction of Annexin-V-positive, propidium iodide-negative DC was 5·7% in the ganglioside-treated group versus 4·7% in the untreated control. The Annexin-V-/propidium iodide double-positive fraction was 4·5%versus 3·7%, respectively.
The presence of gangliosides during early DC differentiation results in an impaired ability of DC to stimulate T cells, whereas mature DC are resistant to the ganglioside-mediated effects
To test the functional abilities of ganglioside-treated DC, allogenic CD4+ T cells were stimulated with either untreated or ganglioside-treated DC to induce T cell proliferation. When monocytic precursors were co-incubated with gangliosides on day 1 of culture, they developed poor stimulatory abilities when tested as immature DC on day 7 (Fig. 3a). Even after the induction of maturation with proinflammatory cytokines, 9-day-cultured, ganglioside-pretreated DC still did not develop full stimulatory activity (Fig. 3b). This effect was not due to a toxic effect of the gangliosides, as the viability in all groups was similar. A direct inhibitory effect on the T cells due to contaminating gangliosides in the MLR itself can be excluded, as direct stimulation of T cells with PHA and simultaneous co-incubation with ganglioside-treated DC resulted in normal proliferative responses (data not shown).
To analyse whether the observed inhibitory effects were dependent on the stage of DC maturation, DC were incubated with gangliosides at different time-points of the DC culture. The addition of gangliosides on day 8 had little or no influence on the surface expression of HLA-DR, CD83, CD80 and CD86 (data not shown). On a functional level, data from the MLR demonstrate that mature DC are resistant to the ganglioside-mediated suppression, as the stimulatory capacity of mature DC that were treated with gangliosides on day 8 was similar to the untreated control (Fig. 3c).
Incubation with gangliosides during DC differentiation results in a decreased endocytotic and chemotactic activity
Endocytosis via macro- and micropinocytosis is one of the key functions of immature dendritic cells. We tested ganglioside-treated DC for their ability to take up FITC-dextran. Gangliosides were added to the DC culture on day 1. On day 6, immature DC were harvested and the DC were incubated with FITC-dextran at 37°C. After 30–210 min, endocytosis was halted and the cells were analysed by flowcytometry. The mean fluorescence intensity of the ganglioside-treated DC was up to 50% lower compared to the untreated control at all time-points tested, indicating that the endocytotic activity of ganglioside-treated DC was clearly reduced (median inhibition after 1 h: 31·4% ± 12·9%, after 3·5 h: 40·4% ± 17·8%) (Fig. 4).
Immature DC migrate in response to certain chemokines. We therefore tested the migratory activity of immature DC in a transwell system in response to MIP-1α, a chemokine known to be a strong attractant for immature DC. DC pretreated with gangliosides on day 1 of culture and tested for migration on day 6 revealed a reduced chemotactic activity, with only 0·6% of input DC migrated to the lower chamber, compared to 3·2% in the untreated group (Fig. 5a). This inhibition could also be observed in a dose-dependent manner, when gangliosides were added directly to the assay on day 6 (Fig. 5b).
Gangliosides differ in their suppressive activity dependent on their carbohydrate structure
Finally, we asked whether the observed inhibition was dependent on the carbohydrate structure of the individual gangliosides. For the neuroblastoma cell line LAN-5 the main ganglioside components have been reported to be GM2 (56%), GD2 (15%) and GT1b (11%) . For melanoma the shedding of GD3 and GM3 has been published . We analysed the influence of these gangliosides on dendritic cell differentiation. When added at varying concentrations (5–40 µm) on day 1 of culture, only the GM2-treated cells exhibited the typical cell spreading after 12 h (Fig. 1d), whereas GD2 (Fig. 1c), GD3, GM3 and GT1b did not alter DC morphology.
The expression of MHC class II, CD80 and CD116 was significantly reduced when GM2 (P < 0·03), but not when GD2, GD3, GM3 or GT1b were added to the DC culture (Table 1). GM2-treated DC stimulated allogenic T cells in MLR to a lesser degree than untreated DC (Fig. 3a,b). The incubation of DC with the other gangliosides (GD2, GD3, GM3, GT1b) did not significantly alter the T cell stimulatory activity (data not shown).
Table 1. Influence of various gangliosides on the DC phenotype
DC were incubated with specific gangliosides (20 µm) as indicated on day 1 of culture. On day 9, mature DC were analysed by flowcytometry. Numbers indicate the median MFI ± s.d.. of four experiments. *Statistical significance (P < 0·003) compared to untreated controls.
All the gangliosides mentioned above contain C18-ceramide as a fatty acid side chain. It is noteworthy that C18-ceramide itself did not affect DC differentiation. This implies that the inhibitory activity was due to the ganglioside structure rather than the ceramide residue alone (Fig. 3a, Table 1).
Efficient DC development and function is crucial in the initiation phase of any Ag-specific immune response against tumours. In order to prime naive T cells, immature DC must be able to migrate and process and present antigens [1,2]. The differentiation phase of DC is a critical period that can be disturbed by factors released by tumours such as IL-10 and VEGF [7–9].
In the present study we analysed the influence of gangliosides on the development of DC from human monocytes. Gangliosides are shed by many tumours and suppress a variety of cells such as PBMC, T cells [16–21], monocytes  and DC derived from CD34+ stem cells .
Our data demonstrate that early addition of gangliosides to the DC culture results in a strong morphological impairment, with increased adherence, cell spreading and reduced development of dendrites.
DC can be characterized by a set of surface expression markers such as MHC class II molecules, co-stimulatory signals (CD40, CD80, CD86), the GM-CSF receptor (CD116) and CD83. These markers are up-regulated during differentiation, and their expression can be enhanced by proinflammatory cytokines that induce DC maturation. In our assays, incubation of DC with gangliosides during their differentiation resulted in a clearly reduced MHC class II expression. Furthermore, the functionally important molecules CD80 and CD116 showed a lower expression on mature ganglioside-treated DC than in untreated controls. The yield of mature CD83+ DC was reduced consistently by about 20–30% when the cells were treated with gangliosides on day 1 of culture.
The gangliosides are most inhibitory in a low-protein environment, as proteins bind gangliosides [28,29]. This can be confirmed by our own observations. Therefore the protocol originally described by Jonuleit et al.  was modified by reducing the amount of autologous serum added to the culture. Another modification made in order to avoid any preselection of more mature DC was the complete harvest of non-adherent and adherent cells on day 9, while not transferring non-adherent cells into new wells on day 7, as described by Jonuleit et al. These culture conditions explain why our DC on day 9 consist of two populations, with about 40–70% mature CD83+ DC, expressing higher levels of HLA-DR, and 30–50% immature CD1a+ DC with lower levels of HLA-DR.
The concentration of gangliosides used in our system (1–40 µg/ml, which is approximately 0·5–20 µm) is similar to most in vitro experiments performed by others [16–19,21–23,30]. These in vitro concentrations are somewhat higher compared to in vivo concentrations; e.g. 0·026–1 µg/ml of GD2 in serum samples from neuroblastoma patients  or 2–4 µg/ml ganglioside-associated serum sialic acid in sera from patients suffering from head and neck cancer . Nevertheless, the use of higher ganglioside concentrations is relevant for the following reasons: (1) the in vitro culture conditions are chosen to obtain the highest yield of mature dendritic cells, thus using optimal cytokine concentrations to induce maturation and do not in itself reflect physiological conditions; (2) in vivo the tumour interstitium is a low-protein environment. Therefore even small amounts of gangliosides might be suppressive due to reduced protein binding; and (3) the in vitro and in vivo effects of substances may be comparable qualititatively, but not quantitatively due to different metabolic pathways (degradation in vitro, metabolism in vivo, etc.).
The ganglioside-mediated inhibition was most pronounced when gangliosides were added in the first days of culture, whereas mature DC were resistant to the described effects. This suggests that gangliosides disturb the early differentiation phase from monocytes to DC, and not the cytokine-induced maturation phase from immature to mature DC.
On a functional level, the capacity of the DC to stimulate allogenic T cells, to take up dextran via endocytosis and to migrate in response to MIP-1α was clearly reduced by preincubation with gangliosides.
Gangliosides differ in their carbohydrate structure and in the number of neuraminic acid residues. In malignant cells gangliosides are shed in characteristic patterns that differ from the ganglioside patterns of normal cells . Surprisingly, only GM2 had an inhibitory effect, whereas GM3, GD3, GT1b and GD2 did not have an inhibitory influence. C18-ceramide, the fatty acid chain of the gangliosides used in these assays, did not alter DC differentiation, thus the carbohydrate structure is important for the immunosuppressive activity of the gangliosides. Ladisch et al. examined the suppressive activity of different gangliosides with regard to their carbohydrate structure  and ceramide length . In their system all gangliosides, irrespective of their number of sialic acids, were shown to be inhibitory for lymphoproliferation. These experiments focused on the direct effects in the activation phase of PBMC. In contrast we describe the effects of gangliosides on developing DC, showing that only certain gangliosides are inhibitory.
Our data confirm data published recently by Shurin et al. . They observe the inhibition in the development of DC derived from CD34+ stem cells in co-culture experiments with neuroblastoma cell lines. The blocking of the ganglioside synthesis in the tumour cells leads to an improved DC development and function. The suppression of the erythroid lineage derived from CD34+ stem cells by gangliosides has also been reported by Sietsma et al. . Our data on ganglioside-mediated suppression of the differentiation and function of monocyte-derived DC demonstrate that gangliosides suppress cells at an even further differentiated stage in the mono/myeloid lineage. As reported by Shurin et al., DC development from CD34+ stem cells is inhibited by GD2 and GM3. For monocyte-derived DC GM2 is inhibitory, whereas GD2 and GM3 are not. These differences can be explained by the different DC populations grown under different conditions and reflect the complex and specific effects of gangliosides depending on their carbohydrate structure.
How gangliosides inhibit DC differentiation remains unclear. Gangliosides have been shown to interact with IL-2, IL-4 and GM-CSF and their receptors [18,19,33–35]. In our hands, however, the ganglioside-mediated inhibition cannot be overcome by high amounts (10×) of either IL-4 or GM-CSF (data not shown). The modulation of signal transduction by gangliosides might result in the suppression of Nf-κ-B and changes in the cytokine network, as described for T cells  and monocytes [36,37]. Furthermore, there is evidence that gangliosides alter many different signalling pathways initiated by receptor-associated tyrosine kinases or mediated by protein kinase C, MAP kinases and other kinases . This can be explained by interac-tions of gangliosides with the cell membrane, especially with ‘glycosphingolipid-enriched microdomains’, also called ‘lipid rafts’. These rafts have been implicated in glycosylphosphatidylinositol (GPI)-anchored protein signalling, immunoreceptor signalling and growth factor receptor signalling [38,39]. It has been demonstrated that exogenous gangliosides abolish clustering of GPI-anchored proteins and disrupt the lipid rafts . The potential role of these glycosphingolipid-enriched microdomains in DC development and signal transduction is unclear. It is tempting to speculate that exogenous gangliosides disrupt lipid rafts in DC, and hence disturb multiple signalling pathways affecting the cytoskeleton, membrane trafficking in endocytosis, migration and antigen presentation.
Tumor escape from immunosurveillance is a keystone in tumour immunology. With regard to immunomodulatory therapies such as vaccination against tumour antigens, it is important to understand how tumours escape immunosurveillance. Our data demonstrate that gangliosides inhibit the development and function of DC derived from monocytes, impairing endocytosis, migration and T cell stimulation. Ganglioside shedding might therefore be a significant mechanism as to how tumours such as neuroblastoma and melanoma escape immunosurveillance.
The authors thank U. Schlütter for excellent technical assistance. We are grateful to A.H. Enk for helpful discussions. This work was supported by a grant for young investigators by Köln Fortune, Germany and by the Fördergemeinschaft Kinderkrebsneuroblastom-Forschung for MW. WYB was supported by the Kind-Philipp-Stiftung, Germany. HB was supported by the Deutsche Forschungsgemeinschaft, SFB456, Germany.