Dendritic cells (DCs) are the most efficient APC and play a central role in initiating NK, T and B cell-mediated immune responses.1, 2 Circulating or peripheral resident DCs take up Ags and migrate into secondary lymphoid organs and activate immune cells. Immature DCs (iDCs) can be derived from several DC precursors in human, among which monocytes are the most frequently used. The exceptional ability of DCs to stimulate T cells in vitro and in vivo is attributed to their ability to present Ag, to secrete cytokines and to express high levels of immunostimulatory molecules including MHC class II, CD40, CD80, CD86 and CD54. The ability of DCs to activate naive T cells closely depends on the maturation stage of DCs.1 The maturation signals for DCs include CD40 ligation, LPS, inflammatory cytokines (TNF-α and IL-1) and immunostimulatory bacterial DNA (CpG-DNA).3, 4 The maturation signal induced by CD40 ligation on DCs has been actively studied before because it very often occurs in physiological situations.1, 3 Cross-linking of CD40 induces several signaling events and is followed by upregulation of costimulatory and cell adhesion molecules, secretion of IL-12 as well as proinflammatory cytokines in DCs.
DCs capture tumor cell-derived soluble protein, necrotic or apoptotic bodies of tumor cells using various Ag-uptake mechanisms such as receptor-mediated phagocytosis.5, 6, 7 Tumor Ag-loaded DCs induce MHC Class I- and Class II-restricted proliferation of autologous, tumor-specific CD4+ and CD8+ T cell and may thereby induce protection to tumor challenge. Previous reports suggest that DCs may play a direct role in elimination of tumor cells. The direct injection of DCs can inhibit the growth of tumor cells in mouse tumor models.3, 8 Furthermore, inhibition of tumor growth and significant early apoptosis of tumor cells were detected in breast tumors in mice injected with syngeneic DCs.9 It is thus suggested that DCs efficiently take up apoptotic tumor cells in vivo and induce anti-tumor immune responses. It remains unclear, however, whether DCs directly induce apoptosis of tumor cells. The fact that DCs can produce significant amounts of TNF-α, express membrane bound FasL and produce nitric oxide (NO) suggests that DCs may have intrinsic, apoptosis-inducing capabilities in certain conditions.10, 11 Recent reports demonstrated DCs can kill tumor cells through cell–cell contact-dependent mechanisms.12, 13
In our studies, monocyte-derived DCs were treated with or without soluble CD40L, LPS or both. Our data demonstrate that maturing DCs induce tumor-specific apoptosis through TNF- alpha-dependent and independent mechanisms. The data suggest the existence of an additional pathway of DC-mediated apoptosis in tumor cells different from those described previously.
MATERIAL AND METHODS
Breast cancer cell lines and normal breast fibroblasts
The human breast carcinoma cell lines, BT-20, MCF7, MDA-MB-436, MDA-MB-468, SK-BR-3, T-47D and leukemia cell line, K562 were obtained from the American Type Culture Collection (Rockville, MD). The cell lines were maintained using RPMI-1640 medium (complete medium; CM) supplemented with 10% FBS, L-glutamine, penicillin, streptomycin (all from Bio-Whittaker, Walkersville, MD). The cells were incubated at 37°C in a humidified, 5% CO2 atmosphere and subcultured every 3–5 days. Normal breast fibroblasts and PBMCs were used as control cells. Normal breast tissue was obtained from reduction mammoplasty specimens. The tissue (∼50 mg) was minced with a sterile scalpel, placed in 10 ml of DMEM supplemented with 10% FBS, L-glutamine, penicillin, streptomycin and amphotericin B (complete medium; DMEM-CM) and incubated with 10 μg/ml collagenase B (Boehringer Mannheim, Indianapolis, IN) at 37°C for overnight. The sample was spun and the cell pellet was resuspended in 5 ml of PBS and respun. The pellet was resuspended in 1 ml of PBS containing 0.025% trypsin and incubated at 37°C for 30 min with gentle agitation. The cell solution was spun and the pellet was resuspended in 5 ml of DMEM-CM and plated onto a 60-mm culture dish. Fibroblasts were cultured from this tissue extract and characterized by both morphology and mitogenesis induced by PDGF (platelet-derived growth factor).
Preparation of DCs
PBMCs were isolated from buffy coats of healthy donors (American Red Cross, St. Louis, MO). PBMCs were suspended in CM and allowed to adhere to 175 cm2-plastic flasks for 1 hr at 37°C. After removing non-adherent cells, adherent cells were incubated in Versene solution® (Life Technologies, Grand Island, NY) and harvested using cell scrapers. Cells were washed twice with CM and cultured in CM containing GM-CSF (800 U/ml) and IL-4 (400 U/ml) for 6 days. X-vivo 15 media (BioWhittaker) was used for some experiments. Culture media was changed after 3 days. After 6 days of culture, nonadherent cells were harvested as iDCs and purity was determined with FACS analysis using anti-CD1a, CD4, CD8, CD14, CD16, CD19 (all from BD Pharmingen, San Diego, CA) and CD83 monoclonal antibodies (MAbs, Immunotech, Westbrook, ME). The purity of cells was >90% CD1a+ and <3% CD8+, CD14+, CD16+ and CD19+.
Tumor growth inhibition assay
Growth inhibition was measured by thymidine incorporation, MTT assay or the trypan blue exclusion test. For the thymidine incorporation assay, tumor cells were cultured in 96-well plates with CM alone or in the presence of iDCs with soluble human trimeric CD40L (Alexis Co., San Diego, CA), LPS (Sigma, St. Louis, MO) or both. Cells were cultured for 72–96 hr and pulsed with 1 μCi/well [3H]-thymidine (ICN Pharmaceuticals, Irvine, CA) for the last 18 hr. To resuspend adherent tumor cells, trypsin was added for 20 min before harvesting. Thymidine incorporation was determined by a liquid scintillation counter (Wallace, Gaithersburg, MD). For the MTT assay, tumor cells were cultured in the presence of 50% (v/v) DC supernatants for 96 hr. To prepare DC supernatants, 6 day-cultured iDCs were used and supernatants were harvested at 24–48 hr after culture in medium alone or sCD40L+LPS. Cells were incubated with 20 μl/well of 5 mg/ml MTT solution (Sigma) for the last 4 hr. After incubation with MTT solution, culture medium was removed and stained cells were treated with 100 μl/well of absolute isopropanol in 0.1 N HCl. Absorbance was measured with an ELISA reader at 595 nm. The number and viability of tumor cells were determined by the trypan blue exclusion test. Rabbit anti-human TNF-alpha polyclonal antibody (anti-hTNF-alpha Ab, Biosource International, Camarillo, CA) was used in selected experiments.
For physical separation of DCs and tumor cells, transwell cultures were established in 24-well plates using transwell chambers (Corning, New York, NY). The transwell chambers with 6.5 mm diameter have 0.45 μm pore size membrane and approximately 0.3 ml volume capacity. IDCs were cultured in transwell chambers in the absence or presence of sCD40L, LPS or both and tumor cells were cultured on the bottom of the 24-well plates. To analyze the morphological changes of tumor cells, a light inverted microscope equipped with a camera was used to take photographs.
Fractionation of DC-derived supernatants
Supernatants were harvested from the culture of DCs incubated with sCD40L+LPS for 48 hr and concentrated 3-fold using Amicon centriprep filters (cutoff 10 kDa, Millipore, Bedford, MA). Fractionation was carried out by Fast-Performance Liquid Chromatography (FPLC, Amersham Pharmacia Biotech, Piscataway, NJ) on a Sephacryl S-200 HR® column (Amersham Pharmacia Biotech) with 0.05 M phosphate buffer (pH 7.0) containing 0.15 M NaCl. Flow rate was 0.5 ml/min and each fraction contained 2 ml. The protein content of each fraction was measured by a UV monitor and fractions were sterilized with 0.22 μm filters before use. Albumin, ovalbumin and chymotrypsinogen (all from Amersham Pharmacia Biotech) were used as standards to determine the molecular weight of proteins. ELISA kits were used for measuring the amount of TNF-α (Biosourse International) or soluble Fas ligand (Medical & Biological Laboratories, MBL International, Watertown, MA) in fractions, according to the manufacturer's instructions.
Flow cytometry analysis
For FACS analysis, cells were incubated with each MAb at a concentration of 1 μg/100 μl for 30 min at 4°C and washed twice with HBSS (BioWhittaker) containing 5% FBS and 0.1% sodium azide. Direct staining was carried out by incubation of FITC- or PE-labeled MAb. Indirect staining was carried out using primary Ab and FITC- or PE-labeled secondary Ab. Nonspecific binding was measured using FITC- or PE-conjugated isotype-matched Ab respectively (BD Pharmingen). Apoptosis analysis was carried out by using an Annexin V-FITC kit (Biosource International) following the manufacturer's instructions. Cells were stained with annexin V-FITC and propidium iodide (PI). For cell cycle analysis, cells were stained with PBS containing 50 μg/ml PI, 0.1% Triton X-100, 0.5 mM EDTA and 50 μg/ml RNase A (Sigma) for 60 min at room temperature. FACSCaliber flow cytometer (Becton Dickinson, Mountain View, CA) and CellQuest software were used for analysis.
In proliferation experiments, data represent the mean and SD of triplicate samples. The statistical significance of the experimental data was evaluated by the Student's t-test; p < 0.05 was considered as statistically significant.
CD40 ligation and LPS induce DC-mediated growth inhibition of tumor cells
PBMCs from normal donors, BT-20, MCF7, MDA-MB-438, MDA-MB-468, SK-BR-3 or T-47D breast tumor cells were cocultured with or without iDCs. Cells were treated with sCD40L, LPS or both and proliferation was measured using 3H-thymidine incorporation. PBMCs cultured without iDCs significantly showed less proliferation than those cultured with iDCs and addition of maturation signals, sCD40L or LPS, further enhanced proliferation in the presence of iDCs. In contrast, growth inhibition was measured in the cultures of tumor cells with iDC+sCD40L+LPS. Although sCD40L, LPS or both induced marginal effects, coculture with iDCs significantly inhibited tumor cell proliferation (p < 0.05, Fig. 1). To test the possibility that tumor cell-derived soluble factors changed the proportion or phenotype of DC, iDCs were cultured with sCD40L+LPS in the absence or presence of 50% (v/v) supernatants of SK-BR-3 tumor cells. IDCs were CD14 negative; CD1a was over 90% positive and CD83 was 15% positive. After culture for 48 hr, both DCs cultured in the absence or presence of tumor supernatants were still CD14 negative and CD83 was over 90% positive (Fig. 2).
DC-derived supernatants inhibit the growth of tumor cells
To determine whether cell–cell contact is required for the growth inhibition effect, we harvested iDCs, mDCs and iDC maturing supernatant (sCD40L+LPS) and tested the proliferation of BT-20, MDA-MB-468 and SK-BR-3 breast tumor cells (Fig. 3a). iDC maturing supernatant (sCD40L+LPS) most effectively inhibited the growth of tumor cells whereas iDCs or mDCs only showed marginal growth inhibitory effects. This suggests that inhibition of tumor growth by DCs is primarily mediated by one or more soluble factors secreted during the maturation process of iDCs. We further investigated the iDC maturing supernatant (sCD40L or LPS) on the growth of tumor cells. Supernatants were harvested from iDCs alone, iDCs+sCD40L, iDCs+LPS and iDCs+sCD40L+LPS cultures and tested on 3 different tumor cell lines (Fig. 3b). iDC maturing supernatant (sCD40L) significantly inhibited the growth of MDA-MB-468 and SK-BR-3 cells but not of BT-20 cells. iDC maturing supernatant (LPS) inhibited proliferation of all three tumor cell lines. Almost complete growth inhibition of the 3 cell lines was observed by iDC maturing supernatant (sCD40L+LPS). To confirm these results, proliferation was evaluated in MTT assays (Fig. 3c). iDC maturing supernatant (sCD40L+LPS) significantly inhibited the growth of 6 different tumor cells, but iDCs supernatant did not. To investigate whether the culture supernatants of already matured DCs in the absence or presence of sCD40L+LPS induced growth inhibition effects, we harvested the supernatants or DCs as described in Figure 3d and measured the growth inhibition effects (Fig 3e). iDC maturing supernatant (sCD40L+LPS) induced a stronger growth inhibition effect than mDC supernatant (sCD40L+LPS). Although iDCs did not induce a growth inhibition effect, all 3 kinds of mDCs induced a marginal, but significant growth inhibitory effect. Additionally, a 0–15% increase of the growth inhibition effect was observed after adding fresh sCD40L+LPS into each DC supernatant (data not shown).
Morphological changes of tumor cells exposed to soluble factors of activated DC
Tumor cells cocultured with iDCs+sCD40L+LPS (transwell) showed morphological changes (Fig. 4), primarily into spindle-shape, but tumor cells cocultured with sCD40L+LPS (data not shown) or iDCs alone did not (transwell, Fig. 4). The morphological changes were observed in SK-BR-3 cells after 12–18 hr coculture and in other breast tumor cell lines for 48–72 hr coculture. Normal human breast fibroblasts, however, did not show any morphological changes (Fig. 4).
DC-derived soluble factors induce G1 phase arrest and decrease of S-phase in tumor cells
To further investigate the DC-mediated growth inhibition effect, cell cycle analysis was carried out on tumor cells in transwell cultures. Tumor cells, MCF7 and SK-BR-3, consistently showed a decrease of cells in S-phase; an increase of cells in G1 phase and an increase of cells in hypoploid apoptotic phase at 48 hr after coculture with iDC+sCD40L+LPS (transwell). In contrast, normal PBMCs and breast fibroblasts did not show any changes in cell cycle (Fig. 5). Furthermore, the cell cycle changes, increase of G1 phase and decrease of S-phase, were observed in SK-BR-3 cells at 18 hr after coculture of iDC+sCD40L+LPS (transwell). It was suggested that DC-derived soluble factors induce cell cycle arrest of tumor cells.
DC-derived soluble factors induce the apoptosis of tumor cells
To determine the presence of apoptosis in tumor cells cocultured with iDC+sCD40L+LPS (transwell), annexin V-FITC/propidium iodide staining was carried out. Interestingly, SK-BR-3 cells treated with sCD40L+LPS showed marginal apoptosis, whereas MCF7 cells did not (Fig. 6a). This observation might be correlated with the expression of surface CD40 on SK-BR-3 cells, but not on MCF7 cells (Fig. 6b). The absence of CD40 on MCF7 was confirmed by a previous report.14 Apoptosis in both tumor cell lines was observed, however, after coculture with iDC+sCD40L+LPS (transwell) whereas few or no apoptotic tumor cells were observed after coculture with sCD40L+LPS or iDC alone (transwell, Fig. 6a). To test the possibility that DC-derived soluble factors have NK cell-like cytotoxicity, we treated K562 cells with DC supernatants. K562 cells treated with 50% (v/v) of iDC maturing supernatant (sCD40L+LPS) showed only marginal increase of apoptosis, whereas cells treated with 12.5% (v/v) of iDC maturing supernatant (sCD40L+LPS) or 50% (v/v) of iDC supernatant did not show any apoptosis (Fig. 7).
TNF-α-dependent and -independent growth inhibition effects of DC-derived soluble factors on tumor cells
To investigate the growth inhibitory effects further, we fractionated the supernatants of DCs using FPLC. X-vivo 15 medium containing 2% FBS was used to decrease the amount of serum proteins. iDC maturing supernatants (sCD40L+LPS) were harvested and concentrated 3-fold. One milliliter of concentrated supernatant was loaded onto the FPLC column and 150 ml of solution was collected into 75 fractions. Twenty fractions were evaluated for the presence of proteins and used for experiments. Growth inhibition was observed with fractions 8 and higher (Fig. 8a). To test whether TNF-α or soluble FasL were present in these fractions, ELISA was carried out. TNF-α was detected in Fractions 11–16, but soluble FasL was not (Fig. 8b). Interestingly, although Fractions 8–10 did not include TNF-α, those fractions significantly inhibited proliferation of tumor cells. To exclude the possibility that trace amounts of TNF-α affected the proliferation of tumor cells, we selected Fraction 9 containing undetectable TNF-α and Fraction 13 containing the highest concentration of TNF-α and cultured tumor cells with 50% (v/v) of Fraction 9 or 13 in the presence of 5 μg/ml rabbit anti-human TNF-α polyclonal antibody. Blocking with antibody significantly reversed the growth inhibition effect of Fraction 13 containing TNF-α, but not that of Fraction 9 containing undetectable TNF-α (Fig. 8c). Furthermore, apoptosis of tumor cells incubated with Fraction 9 was detected in the presence of anti-TNF-α antibody (data not shown). It is thus suggested that TNF-α-dependent and -independent mechanisms may contribute to the growth inhibitory effect of DC-derived soluble factors.
As the most efficient and critical APC in anti-tumor immune responses, DCs take up tumor Ag from apoptotic or necrotic bodies of tumor cells and present tumor antigens to elicit tumor-specific immune responses.5, 6, 7 Although apoptosis of tumor cells is a critical process for generating DC-mediated anti-tumor immune response, it has not been elucidated clearly whether DC directly induce apoptosis of tumor cells. Previous reports demonstrated that a subset of rat DCs and human CD11c+ DCs treated with IFNs killed NK-sensitive cells and some tumor cell lines in a cell–cell contact-dependent manner.15, 16 NK cell receptor protein 1 (NKR-P1)+ rat DCs reportedly were able to kill the NK-sensitive YAC-1 cells in a Ca2+ dependent, Fas-FasL interaction independent manner.15 Furthermore, human DCs induce cellular apoptosis via TNF-related apoptosis-inducing ligand (TRAIL) in some TRAIL-sensitive tumor cell targets but not TRAIL-resistant tumor cells or normal cell types. Human CD11c+ blood DCs express TRAIL after stimulation with either IFN-α or -γ and acquire the ability to kill certain tumor cells in a cell–cell contact-dependent manner.16
Two recent studies demonstrated that monocyte-derived DCs have the ability to kill tumor cells.12, 13 Vidalain et al.12 showed that DC stimulated with double-stranded RNA lysed the breast cancer cell line MDA231 through TRAIL. In contrast, CD40L stimulated DC lysed MDA231 cells in part through TNF-α and in part through an unidentified mechanism. Lysis in both cases was induced by coculture of DC and tumor cells at ratios of up to 50 DC:1 tumor cell. In contrast, Vanderheyde et al.13 observed DC-mediated lysis of a variety of tumor cell lines that was caspase-8-dependent. TNF-α, FasL, TRAIL and Fas-associated death domain (FADD) were not involved. Interestingly, these investigators observed similar levels of cytotoxicity by immature DC and LPS- or IFN (α, β or γ)-treated DC. Moreover, DC-derived soluble factors did not induce cytotoxicity. We demonstrated here that the apoptosis-inducing effects of DCs are mediated by soluble factors derived from DCs undergoing maturation.
Another study showed that DCs treated with LPS or IFN-γ, but not TNF-α or anti-CD40 MAb, induced a growth inhibitory effect on tumor cells.17 Our data showed that CD40 signaling induced growth inhibition of tumor cells that was enhanced by LPS. This discrepancy may be related to the reagent used to induce CD40 signaling, sCD40L vs. anti-CD40 mAb (ATCC clone G28-5) and the kind of tumor cells used, breast vs. colon adenocarcinoma cells. It is possible that the signal intensities induced by sCD40L and anti-CD40 mAb are different. Although it was reported that DCs did not induce apoptosis of Jurkat, THP-1 and K562 cells, those cell lines did show a significant growth inhibition.17 In contrast, we demonstrated that DCs induce apoptosis as well as cell cycle arrest of breast tumor cells, but not of normal breast fibroblasts, PBMCs or K562 cells. It is possible that there are differences in susceptibility or expression of receptors on target tumor cells that respond to the DC-derived soluble factors. In addition, various experimental conditions such as the purity or concentration of DCs may vary. It is not likely that the apoptosis of tumor cells are from potential artifacts because iDC maturing supernatant (sCD40L+LPS) did not induce apoptosis of K562 cells.
CD40 ligation is an important maturation-inducing signal for DCs and results among other things in the production of IL-12, upregulation of costimulatory molecules, an increased capacity of DCs to trigger proliferative responses and IFN-γ production by T cells.18 sCD40L and LPS were used in our experiments because they are among the major maturation-inducing molecules for DCs under physiologic and pathologic conditions. Although there are multiple maturation-inducing agents, it has not been fully elucidated if different maturation signals have a synergic effect on DCs and which function of DCs can be synergistically affected by those signals. A recent report demonstrated that CD40L and LPS enhanced the secretion of IL-12 by 5–60-fold and secretion of IL-10 by 5–15-fold by monocyte-derived DCs when compared to either stimulation alone.19 No synergistic upregulation of CD80 (B7-1), CD86 (B7-2) or CD83 molecules was observed. Our data demonstrated that sCD40L and LPS induce a synergic growth inhibitory effect on some tumor cell lines through DC-derived soluble factors. It is thus likely that different maturation signals of DCs may have synergic effects on the production of soluble factors that can inhibit the growth of tumor cells.
Interestingly, DC-derived soluble factors did not significantly affect the proliferation, cell cycle or apoptosis of normal PBMC or breast fibroblasts. Furthermore, iDCs enhanced the proliferation of normal PBMCs in the presence of sCD40L, LPS or both, compared to PBMCs grown in the absence of iDCs, in a cell-to-cell contact-dependent manner (Fig. 1). Our data showed that DCs undergoing maturation have opposite effects on normal PBMC and tumor cells with regard to proliferation. It is possible that DC-derived soluble factors do not significantly inhibit the growth of PBMCs due to the strong stimulatory signal mediated by allo MHC alleles on the DC.
To further investigate the growth inhibition effect, we fractionated DC-derived soluble factors by FPLC and measured the growth inhibition effects on tumor cells (Fig. 8). Fraction 8 and up induced growth inhibition whereas TNF-α was detected in Fraction 11 and higher. Significant growth inhibition even in the presence of saturating amounts of anti-TNF-α Ab suggests that the growth inhibition effect may be at least partially mediated by a TNF-α-independent mechanism. To test the involvement of TRAIL, Fractions 9 and 13 were used for Western blot analysis using a TRAIL-specific Ab, but TRAIL was not detected in both fractions (data not shown). Likewise, soluble FasL was not detected in any of the fractions. Taken together, the growth inhibition effects of DC-derived soluble factors may be mediated by TNF-α-dependent and -independent mechanisms. To identify the DC-derived, apoptosis-inducing soluble factor(s), further investigation is in progress.
CD40 has been demonstrated to be expressed at high levels on a variety of human carcinomas including breast and ovarian cancers. A recent report demonstrated that sCD40L significantly inhibited the proliferation of CD40+ human breast cancer cell lines and the inhibition could be augmented with IFN-γ.14 We observed consistently that sCD40L significantly inhibits the growth of SK-BR-3 breast cancer cells that express CD40, but not MCF7 breast cancer cells that do not express CD40. Our data that DC-derived soluble factors induce apoptosis and cell-cycle arrest of CD40− MCF7 cells as well as CD40+ SK-BR-3 cells suggest that sCD40L may have an indirect, DC-mediated anti-tumor effect as well as a direct anti-tumor effect. It is thus possible that sCD40L-based therapy may lead to additional therapeutic effects in combination with DC-mobilizing agents, such as GM-CSF.
DC-derived soluble factors induced morphological changes in tumor cells. The morphological changes were detected in 5/6 tumor cell lines including BT-20, MCF7, MDA-MB-468, SK-BR-3 and T-47D cells. It is suggested that soluble factors produced by activated DCs directly changed the expression of morphology-related genes or induced cell cycle-arrest and apoptosis of tumor cells that subsequently affected the morphology.
We addressed several points here. First, monocyte-derived, CD1a+ DCs undergoing maturation have the ability to induce apoptosis as well as cell-cycle arrest of breast tumor cells through secretion of soluble factors. Second, CD40 ligation induced DC-mediated growth inhibition of tumor cells and augmented growth inhibition in combination with LPS. Third, normal PBMCs, K562 and breast fibroblasts were not affected by DCs undergoing maturation. It is to be determined which soluble factors are responsible for the anti-tumor function of DCs and if cognate receptors on tumor cells transduce the growth inhibitory signals. Our findings suggest that DCs have the intrinsic ability to induce cell-cycle arrest and apoptosis of tumor cells, but not normal cells.