The cytokine FLT3 ligand (FL) enhances dendritic cell (DC) generation and has therefore been proposed as a means to boost antitumor immunity. Vascular endothelial growth factor (VEGF) is produced by a large percentage of tumors and is required for development of tumor neovasculature. We previously showed that VEGF decreases DC production and function in vivo. In this study, we tested the hypothesis that VEGF regulates FL effects on DC generation. In seven experiments, four groups of mice were treated with PBS, VEGF alone (100 ng/h), FL alone (10 μg/day), or with the combination of FL and VEGF. VEGF and PBS were administered continuously for 14 days via s.c. pumps. FL was given s.c. daily for 9 days, beginning on day 4. Tissues were collected and the number, phenotype, and function of lymph node, splenic, and thymic DCs were analyzed on day 14. As expected, treatment with FL resulted in a marked increase in the number of lymph node and spleen DCs and a smaller increase in thymic DC. Pretreatment of mice with VEGF inhibited these FL effects in lymph nodes and thymus by about 50%, whereas spleen DC numbers were undiminished by VEGF. VEGF treatment in vivo also inhibited the ability of FL to increase the number of hemopoietic precursor cells and the level of maturity exhibited by DC derived from these hemopoietic precursor cells in vitro. VEGF inhibited FL-inducible activation of transcription factor NF-κB. These data suggest that VEGF interferes with the ability of FL to promote dendritic cell differentiation from bone marrow progenitor cells in mice and therefore may decrease the therapeutic efficacy of FL in settings where increased numbers of DCs might provide clinical benefits.
Dendritic cells (DC)3 are the most potent professional APC and as such represent an attractive means to deliver specific disease-associated Ags to naive T cells in a variety of immunotherapy protocols (1). Successful cancer immunotherapy depends on adequate function of host APC, particularly DC. Therefore, increased generation of DC using growth factors or cytokines might boost the host immune responses. One clinical approach currently under investigation involves the use of cytokines such as FLT3 ligand (FL) to increase DC numbers in vivo. Studies in mice have demonstrated that FL enhances DC production in vivo (2, 3, 4) and is capable of inducing antitumor immune responses (5, 6, 7, 8, 9). The antitumor effects of FL are accompanied, in some cases, by the generation of protective, tumor-specific T cell responses (5), presumably as a result of enhanced Ag presentation. In other cases, FL may be acting directly or indirectly upon NK cell responses (5, 8, 9). However, the antitumor effect of FL has only been examined in animal models that involve relatively small tumor burdens (10). Thus, it is possible that factors produced by tumor cells and known to inhibit immune responses may interfere with the efficacy of FL in antitumor models.
Tumor cells appear to have developed mechanisms to avoid immune system recognition and elimination, and among these is the inhibition of DC generation and function. Defective function of DCs in cancer has been recently reported by several groups (11, 12, 13, 14, 15, 16). Previously, we demonstrated that one of the possible mechanisms of DC dysfunction in cancer is an abnormal functional maturation of these cells from hemopoietic progenitors (17, 18, 19). Others recently reported similar results (20, 21, 22). Several soluble factors have been implicated in defective DC maturation in cancer, including vascular endothelial growth factor (VEGF).
VEGF is produced in large amounts by most tumors and its production is closely associated with a poor prognosis (23, 24). VEGF stimulates the proliferation of endothelial cells and plays an important role in the formation of tumor neovasculature (reviewed in Ref. 25). We have demonstrated that anti-VEGF neutralizing Ab blocks the inhibitory effects of tumor cell supernatants on DC maturation in vitro (18). A direct association between increased levels of VEGF in tumor cells and low presence of DCs in the vicinity of the tumor has been described in 140 patients with gastric cancer (26). VEGF binds to CD34+ hemopoietic progenitor cells (HPC) through one of the VEGF-specific receptors, FLT1, and inhibits the activation of transcription factor NF-κB in these cells (27). Continuous VEGF infusion, at rates as low as 50 ng/h, resulted in a dramatic inhibition of DC development, associated with an increase in the production of B cells and immature Gr-1+ myeloid cells (28). Infusion of VEGF was also associated with inhibition of the activity of the transcription factor NF-κB in bone marrow progenitor cells (28). These data suggest that VEGF, at pathologically relevant concentrations in vivo, may exert effects on pluripotent stem cells that results in blocked DC development. We hypothesize that VEGF might interfere with FL effects on hemopoietic precursors and hence could be responsible for the decreased efficacy of FL observed in established tumor models. Investigation of VEGF/FL interactions might also help to clarify the mechanisms of action of these two growth factors on hemopoiesis. In this study, we demonstrate that VEGF partially blocks the ability of FL to stimulate DC production and maturation in mice in vivo. It appears that VEGF and FL may act on the same HPC. VEGF does not affect the expression of FLT3, the receptor for FL, on these cells, but rather may act through the inhibition of transcription factor NF-κB. These findings suggest that FL-based immunotherapy of cancer might be enhanced by the co-administration of agents that can block the inhibitory effects of tumor-derived factors like VEGF.
Materials and Methods
Six- to 8-wk-old female BALB/c, C57BL/6, and CBA mice were purchased from Harlan (Indianapolis, IN) and were housed in specific pathogen-free units of the Division of Animal Care at Vanderbilt University Medical Center.
Abs and reagents
VEGF was a generous gift from Genentech (South San Francisco, CA). FL was a generous gift from Immunex (Seattle, WA). The following Ab-producing hybridomas were obtained from American Type Culture Collection (Manassas, VA) and used as culture supernatants: anti-CD4 (L3T4, TIB-207), anti-CD8 (Lyt-2.2, TIB-210), anti-MHC class II (TIB-120). Mouse GM-CSF, IL-4, TNF-α, and VEGF165 were obtained from R&D Systems (Minneapolis, MN); polyclonal anti-mouse Ig was obtained from Sigma (St. Louis, MO). FITC- or PE-conjugated anti-CD11c, CD11b, CD86 (B7-2), IAd Abs were purchased from PharMingen (San Diego, CA). Isotype-matched FITC- and PE-conjugated IgG were used as a control of nonspecific binding. In some experiments, complete culture medium was used: RPMI 1640 (Life Technologies, Gaithersburg, MD) with 100 IU/ml penicillin, 0.1 mg/ml streptomycin, 5 × 10−5 M 2-ME, and 10% heat-inactivated FCS (HyClone, Logan, UT).
VEGF and FL administration
VEGF was delivered into BALB/c mice via Alzet osmotic pumps (Alza, Palo Alto, CA) for 14 days at a rate of infusion of 100 ng/h. The pumps were inserted s.c. into the back of the mice through a small skin incision. Wound edges were reapproximated with surgical clips. All procedures were performed in aseptic conditions and were approved by the Vanderbilt University Animal Care Committee. In controls, pumps were filled with PBS. On day 4 after initiation of VEGF infusion, administration of FL was started. FL was injected s.c. at a dose of 10 μg per mouse daily for 10 days.
Cell separation and analysis of cell-surface receptors
A single-cell suspension was prepared from inguinal, axillary, and brachial lymph nodes, spleens, and thymus by pressing the tissues through a wire mesh. Red cells were removed by hypotonic shock. No additional purification was performed. Cells were washed and used in additional experiments. For analysis of cell-surface receptors, cells were washed in PBS supplemented with 0.1% FCS and labeled with appropriate Abs for 30 min at 4°C. Cells were then washed and analyzed on FACScan flow cytometer (Becton Dickinson, Mountain View, CA). An enriched T cell population was obtained from the lymph nodes by removing adherent cells followed by 1 h incubation on plastic. Bone marrow was obtained from the femurs and tibias of BALB/c mice. In some experiments, marrow cells were cultured for 7 days with GM-CSF and IL-4 in complete medium to generate DCs as described elsewhere (17). In some experiments, bone marrow cells were enriched for HPC by depletion of lineage-specific cells via incubation with the mixture of Abs (TIB-207, TIB-210, TIB-120, and anti-mouse Ig) and complement (low-tox guinea pig complement; Cedarline, Hornby, Ontario, Canada) and subsequent gradient centrifugation on Lympholyte M (Cedarline) to remove dead cells.
Tissues from control and treated animals were snap-frozen in OCT compound and stored at −80°C. Then, 6-μm tissue sections were air-dried and fixed in acetone, rehydrated in PBS, and blocked for nonspecific staining with 2% normal goat serum in PBS. Primary anti-mouse CD11c, CD86 (PharMingen), or NLDC-145 (DEC 205) (Serotec, Oxford, England) Abs were applied for 60 min at room temperature. All slides were subsequently washed in PBS and incubated 30 min with biotinylated secondary Abs (Jackson ImmunoResearch, West Grove, PA). Bound Abs were detected with ABC reagent and Vectastain substrate kit (Vector Laboratories, Burlingame, CA). For the hematoxylin and eosin staining, murine tissues were fixed in 10% neutral buffered formalin and embedded in paraffin. Then, 4-μm sections were cut and stained with hematoxylin and eosin.
Assessment of colony formation
Colony formation by HPC was measured using semisolid 1% methylcellulose medium supplemented with recombinant cytokines (Epo, stem cell factor, IL-6, IL-3) supporting the optimal growth of burst forming unit (BFU)-erythrocyte (E), CFU-granulocyte-macrophage (GM), CFU-macrophage (M), CFU-granulocyte (G), and CFU-mixed erythrocyte-granulocyte-macrophage (GEMM) colonies (Methocult M3434; Stemcell Technologies, Vancouver, Canada). Bone marrow cells were seeded at 15,000 cells per plate. BFU-E colonies were scored on day 8–9 and all other colonies were enumerated on day 12–13.
T cell proliferation assay
Lymph node and spleen cells obtained from control, VEGF-, and FL-treated mice were irradiated (20 Gy), plated in triplicate, and incubated with enriched T cells obtained from CBA mice for a 4-day allogeneic MLR. Twenty-four hours before harvesting, all cultures were pulsed with 1 μCi of [3H]thymidine (Amersham, Arlington Heights, IL). Incorporation of [3H]thymidine was counted using a liquid scintillation counter and expressed as cpm.
Double-stranded oligonucleotide probes were prepared by annealing the appropriate single-stranded oligonucleotides at 65°C for 10 min in 10 mM Tris, 1 mM EDTA, 10 mM NaCl solution followed by slow cooling to room temperature. The probes were end-labeled with [32P]-labeled CTP by filling in 5′ overhangs with the Klenow fragment. The following murine intronic κ-chain κB site probe was used in this study: wild-type, 5′-AGTTGAGGGGACTTTCCCAGG; mutant κB site, 5′-AGTTGAGGCGACTTTCCCAGG.
Nuclear extract was obtained from the cells as described (29). Next, 10 μg of nuclear extract was incubated for 20 min with labeled probe (50,000 cpm) in the presence of 4 μg of poly(dI-dC) (Pharmacia) in binding buffer (20 mM HEPES, 5% glycerol, 0.2 mM EDTA, 1 mM DTT, 5 mM MgCl2). Competition assays were performed with a 100-fold excess of unlabeled probes. The DNA-protein complexes were separated on 4% polyacrylamide gels and visualized and analyzed on a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
RNA preparation and analysis of gene expression
A total of 3 × 106 bone marrow progenitor cells were cultured in complete culture medium in the presence or absence of GM-CSF, VEGF, and FL at 37°C for a different time. RNA was then extracted using the GlassMAX RNA microisolation spin cartridge system (Life Technologies). RT-PCR was performed on RNA samples following a DNA digestion. Reverse transcription was performed using SuperScript preamplification system for first-strand cDNA synthesis (Life Technologies). Briefly, 1 μg of RNA was incubated with 100 ng random hexamers for 10 min at 70°C. Then, 14 μl of the reaction mixture (10× PCR buffer, 25 mM MgCl2, 10 mM dNTP mix, 0.1 M DTT) were added to each sample followed by incubation at 25°C for 5 min. Next, 2 μl (400 U) of SuperScript II RT was added to each sample and incubated at 25°C for 10 min. Samples were then incubated at 42°C for 50 min. The following selected oligonucleotide PCR primer pairs for the FLT3 were used: forward primer, 5′-GTGACTGGCCCCCTGGATAACGAG-3′; reverse primer, 5′-TCCAAGGGCGGGTGTAACTGAACT-3′.
Hypoxanthine phosphoribosyltransferase (HRPT)-specific primers were used as a control and have been described elsewhere (17). Samples were denatured for 30 s at 94°C, annealed for 30 s at 53°C, and extended for 45 s at 72°C for 20 cycles of amplification. This number of cycles was selected to avoid saturation of PCR products and was determined after preliminary experiments. Products were visualized by staining with ethidium bromide after electrophoresis in 1% agarose gel. The PCR product sizes for flt3 and hprt were 509 bp and 164 bp, respectively. PCR products were transferred overnight in an alkaline transfer buffer (0.4 N NaOH, 1 M NaCl) onto Hybond N+ nylon transfer membranes (Amersham, Highland Park, IL), hybridized for 2 h in a rapid hybridization buffer (Amersham), and probed with [32P]-labeled oligonucleotide probes: FLT3, 5′-GCTGGGGGCATGCACACTGTCA-3′; HPRT, 5′-GTTGTTGGATATGCCTTGAC-3′. Membranes were then analyzed, and the expression of flt3 mRNA was quantified by Phosphoimager (Molecular Dynamics).
We have previously demonstrated that 4 wk of VEGF infusion at a rate of 50–100 ng/h resulted in splenomegaly, was associated with a block of DC production, and caused increased production of immature myeloid Gr-1+ cells (28). However, a shorter 2-wk infusion did not lead to an appreciable change in DC numbers in lymph nodes and spleen although the ability of the splenic DC population to stimulate allogeneic T cells was decreased. No morphological changes in lymph nodes and only minimal changes in spleens were observed after that short exposure (28). To investigate the possible interaction between VEGF and FL, the following protocol was employed. VEGF was delivered into BALB/c mice via a s.c. implanted ALZET osmotic pump for 14 days at a rate of infusion 100 ng/h. On day 4 after initiation of the infusion, FL injections were started. FL was delivered s.c. outside the area of the inserted pump at a dose of 10 μg per mouse daily for 9 days. In total, seven independent experiments (two to three mice per group per experiment) were performed. Four groups of mice were studied in each experiment. The control group was comprised of mice receiving continuous infusion of PBS and s.c. injections of PBS; the VEGF group included mice with continuous VEGF infusion and PBS injections; mice in FL group received FL combined with continuous infusion of PBS, and the FL plus VEGF group had mice received both VEGF infusion and FL. These four groups of mice were used in all experiments described below.
Effect of VEGF infusion on the ability of FL to stimulate DC production in vivo
Two weeks of VEGF infusions did not result in significant changes in the total number of either lymph node or spleen cells. FL increased the total number of lymph node cells almost 2-fold (p = 0.015) and spleen cells almost 8-fold (p < 0.001). Combined administration of VEGF and FL did not significantly alter FL-induced increases in spleen or lymph node cell number (data not shown). Thus, these results demonstrated that VEGF did not abrogate FL-mediated elevation of cell numbers in murine spleen or lymph nodes.
To investigate the effect of VEGF on FL-induced DC generation, lymph node cells, splenocytes, and thymocytes were labeled with anti-CD11c, anti-CD11b, and anti-CD86 Abs. FACScan analysis revealed that VEGF alone did not alter the number of DC in lymph nodes. FL dramatically increased the percentage of CD11c+ DC, but not CD11c−CD11b+ macrophages in the lymph node (Fig. 1). The percentage of CD11c+CD86+ lymph node cells, representing a subpopulation of mature DC, was increased up to 4-fold in mice treated with FL (p < 0.01). As shown in Fig. 1, the ability of FL to increase accumulation of CD11c+ DC in lymph nodes was significantly decreased by the coadministration of VEGF. The percentage of CD11c+, CD11c+CD11b−, and CD11c+CD86+ DC was ∼50% lower in mice treated with the FL/VEGF combination in comparison with mice treated with FL alone (p < 0.01). A similar, although less pronounced, effect of VEGF was observed for the population of CD11c+CD11b+ cells (p < 0.05) (Fig. 1). Similar results were obtained after staining of the lymph node tissue with anti-NLDC-145 Ab (Fig. 2). Treatment with FL increased the presence of NLDC-145+ DC in lymph nodes (Fig. 2,B). This effect was dramatically reduced by coadministration of FL with VEGF (Fig. 2 D). Thus, VEGF significantly decreased FL-induced accumulation of DC in murine lymph nodes in vivo.
Different results were observed in spleens obtained from FL- and/or VEGF-treated mice. Infusion of VEGF alone resulted in a decrease in the presence of the CD11c+CD86+ cell population (p < 0.05). Interestingly, this effect was associated with a moderate increase in the percentage of CD11c−CD11b+ macrophages (p > 0.05) (Fig. 1). As expected, FL dramatically elevated the presence of total CD11c+ DC in the spleen (p < 0.001). This increase was mostly due to the increased proportion of CD11c+CD11b− DC. A smaller increase was observed in the population of CD11c+CD11b+ cells, and almost no changes were found in the population of CD11c−CD11b+ macrophages (Fig. 1). FL did not affect the presence of the CD11c+CD86+ population of DC. VEGF slightly decreased the proportion of these FL-expanded cell populations. However, this effect did not reach statistical significance (p = 0.05) (Fig. 1). Similar effects were observed by immunohistological analysis. The dramatic increase in NLDC-145+ DC after FL administration was slightly reduced if FL was delivered together with VEGF (Fig. 2, H vs F).
Because our previous experiments have shown that 4-wk VEGF infusion induced a significant increase in the proportion of Gr-1+ cells, we investigated whether FL would affect the number of Gr-1+ cells in the spleen. In four independent experiments, FL alone did not change the presence of Gr-1+ cells in murine spleens. Two weeks of VEGF infusion resulted in slightly elevated levels of these cells (18.6 ± 3.6% vs 11.4 ± 3.2% in control, p > 0.05). Interestingly, the combined administration of VEGF and FL induced a more substantial increase in the proportion of splenic Gr-1+ cells (23.3 ± 3.6%; vs 11.4 ± 3.2% in control, p < 0.05). Thus, our data demonstrated that treatment of mice with FL caused DC accumulation in the spleen, which was not abrogated by the coadministration of VEGF.
Next, we studied the effect of FL and/or VEGF on thymic DC. Infusion of VEGF alone resulted in a significant decrease in the presence of a population of mature CD11c+CD86+ DC in the thymus (p < 0.05). FL administration increased primarily the proportion of CD11c+CD8α− DC up to 6-fold, whereas the population of CD11c+CD8α+ DC was increased only 1.5-fold. VEGF almost completely abrogated these effects of FL on DC accumulation in the thymus (Fig. 1).
Effect of FL and VEGF on the Ag-presenting ability of lymph node cells and splenocytes
We studied whether the alterations in the presence of DC in tissues following the FL and/or VEGF therapy would be accompanied by changes of the ability of lymph node and spleen cells to stimulate allogeneic T cells. Lymph node and spleen cells were isolated from control and treated mice, irradiated, and used as stimulators in allogeneic MLR. Analysis of T cell proliferation revealed that VEGF alone did not affect the ability of lymph node cells to stimulate allogeneic T cells. However, VEGF significantly inhibited the immunostimulatory activity of mixed spleen populations (p < 0.05) (Fig. 3). In contrast, FL dramatically increased the stimulatory capacity of both spleen and lymph node cells (p < 0.01). The effect of VEGF on FL activity was different in lymph node and spleen. VEGF significantly reduced the stimulatory effect of FL on lymph node, but not spleen, allostimulatory activity (p < 0.05), but did not change the ability of FL to increase the activity of spleen APC (p > 0.05) (Fig. 3). These data are consistent with the selective inhibitory effects that VEGF had on DC numbers in the lymph node relative to spleen.
Effect of VEGF and FL on bone marrow progenitor cells
In the next series of experiments, we examined the effects of VEGF treatment on the derivation of specific cell populations from hemopoietic progenitors. First, bone marrow cells were cultured in GM-CSF and IL-4-supplemented medium, and the presence of different populations of DC and their functional activity were compared between the four experimental groups. No difference between the groups was found in the total number of cells generated under these conditions. About 60% of cells expressed DC marker CD11c. Infusion of VEGF alone did not affect the proportion of any of the above described DC populations. In contrast, administration of FL stimulated the generation of mature CD11c+CD86+ DC. The percentage of these cells increased up to 2-fold in comparison with control values (p < 0.05). Infusion of VEGF in FL-treated mice completely abrogated this effect of FL (Fig. 4 A).
Similar results were obtained when the ability of these cells to stimulate allogeneic T cells was investigated. Bone marrow-derived DC from FL-treated animals were >2-fold more active in an allogenic MLR when compared with control BM-derived DCs (p < 0.05). The allostimulatory capacity of bone marrow DCs derived from VEGF plus FL-treated mice was lower than that of cells derived from FL-treated mice (p < 0.05) (Fig. 4,B). It is interesting to note that the infusion of VEGF alone resulted in a slight but statistically significant decrease in the stimulatory activity of cultured bone marrow-derived DC (p < 0.05) (Fig. 4 B).
To determine whether treatment with FL, VEGF, or their combination would regulate the accumulation of colony-forming progenitor cells, bone marrow cells were isolated from control and treated mice. Bone marrow cells were then cultured in semisolid methylcellulose medium supplemented with factors supporting the growth of myeloid and erythroid colonies. In agreement with previously published observations (28), VEGF infusion resulted in a moderate increase in number of CFU-GM and CFU-GEMM colonies (p < 0.05) (Fig. 4,C). FL treatment also moderately increased the number of mixed (CFU-GEMM) colonies, and FL dramatically increased the number of macrophage (10-fold) and granulocyte-macrophage (3-fold) colonies (p < 0.01) (Fig. 4,C). VEGF infusion almost completely abrogated these stimulatory effects of FL on myelopoiesis (Fig. 4 C). No significant effect on BFU-E was determined. These data confirmed previous findings that the administration of FL in mice significantly activates myelopoiesis, which in turn resulted in accumulation of a large number of DC in various lymphoid and nonlymphoid tissues (2, 30). Coadministration of VEGF decreased FL-induced DC expansion in certain tissues, which was a possible consequence of VEGF’s inhibitory effects on myelopoiesis.
Mechanisms of VEGF-FL interaction
The data described above indicate that FL and VEGF may act at similar stages of hemopoiesis and, perhaps, upon the same precursor cell population. We tested this hypothesis by first determining whether VEGF down-regulates the expression of FLT3, a receptor for FL, on the surface of HPC, thus causing these cells to be less responsive to FL. Bone marrow HPC were cultured in vitro with 3 ng/ml recombinant murine GM-CSF with or without of 100 ng/ml VEGF, 100 ng/ml FL, or with the combination of these factors. Our pilot experiments demonstrated that these concentrations of FL and VEGF provided an optimal functional effect in vitro (data not shown). Cultured cells were collected after different period of time, RNA was extracted, and RT-PCR and Southern blot analysis were performed. Expression of flt3 mRNA was normalized to the level of the housekeeper gene hprt in each sample. The results of three independent experiments showed no discernable effect of VEGF on the level of flt3 mRNA at any chosen time point or at any factor combination (the result of one of the experiment is shown in Fig. 5).
We have previously shown that VEGF inhibits TNF-α inducible activation of NF-κB in HPC, resulting in inhibition of DC differentiation (27). Here, we examined whether the same mechanism could be involved in the VEGF-mediated inhibition of FL activity. HPC were isolated from control bone marrow cells and treated with either FL or FL plus VEGF. Nuclear extracts were prepared and EMSA was performed as described in Materials and Methods. Analysis of these data revealed that FL induced a high level of NF-κB nuclear translocation and specific DNA binding. The presence of VEGF in the same cultures significantly inhibited this effect of FL (Fig. 6). These data support the notion that FL-mediated activation of NF-κB in hemopoietic precursors stimulates their differentiation into functionally active mature DC, which migrate to and accumulate in a variety of lymphoid and nonlymphoid tissues. VEGF may decrease this stimulatory effect of FL by inhibiting activation of transcription factor NF-κB in the same precursor cells.
FL is an effective stimulator of DC production in vivo (2). This effect is not due to redistribution and mobilization of the cells, but rather to the production of DC de novo from precursors (2). Because DC play a critical role in the initiation of specific immune responses, there is considerable interest in using this cytokine to generate large numbers of DC and, therefore, to enhance antitumor immune responses. However, tumor-derived factors, including VEGF, are known to affect maturation of DC (27, 28). Here, we examined whether VEGF might affect the ability of FL to stimulate DC production, using a relatively short 2-wk VEGF infusion combined with a 9-day treatment with FL. Relatively short, 2-wk infusion of VEGF did not affect DC populations in lymph nodes. However, it resulted in statistically significant decrease in the presence of mature DC in spleen and allostimulatory ability of spleen cells. This was consistent with previously reported observations (28). Apparently inhibitory effect of VEGF on DC differentiation was more evident in the organ with active hemopoiesis (spleen) than in lymph nodes. In agreement with previously published observations, FL treatment increased both major DC populations CD11c+CD11b− and CD11c+CD11b+ (2, 3). No substantial changes were seen in CD11c−CD11b+ cells. Because most of the latter cells also express the Gr-1 marker (2), it explains the lack of increase in Gr-1+ cells observed after the FL treatment. We investigated the effect of FL on lymphoid and myeloid populations of DC in the thymus. As previously reported, lymphoid CD11c+CD8α+ cells represent the majority of DC in the thymus (30). Interestingly, although FL moderately increased the proportion of these cells, a more profound effect was seen in population of CD11c+CD8α−. The increase in the proportion of DC in the spleen and lymph node was accompanied, as expected, by increased allostimulatory capacity of the total cell populations from these two organs, a function attributed specifically to DC. FL also stimulated the production of mature DC in the lymph node based upon a 3-fold increase in the proportion of cells expressing high levels of the costimulatory molecule B7-2. However, this effect was not observed in the spleens of the same animals. This difference might be explained by the fact that murine spleen is characterized as an active site of hemopoiesis, and spleen-derived DC may require additional incubation to reach full maturity (31, 32, 33). This fact could be demonstrated when we compared the proportion of CD11c+ DC in the spleen and lymph node in FL-treated mice and the ability of splenic and lymph node DC to stimulate T cells in an allogeneic MLR (Figs. 1 and 3). As a result of FL treatment, CD11c+ cells comprised almost 20% of the total population of nucleated lymph node cells, whereas in spleen, the proportion of these cells was >40%. However, when the same cells were tested for their ability to stimulate an allogeneic MLR, lymph node DC had a >2-fold higher activity than splenic DC. Presumably, the splenic microenvironment does not provide an adequate support for the final steps of DC maturation and activation.
VEGF reduced FL effects on DC generation in the lymph node by almost 50%. This was demonstrated by the analysis of DC phenotype and their ability to stimulate T cell proliferation in an allogeneic MLR. Similar effects of FL and VEGF were observed when thymic DC were analyzed. However, VEGF did not significantly change the FL effects in the spleen. The mechanism of this difference is not clear. An explanation may lie in the immature status of splenic DC. As mentioned above, our previous data demonstrated that 4 but not 2 wk of VEGF infusion drove cell differentiation toward Gr-1+ cells. Two weeks of exposure to VEGF only slightly elevated the level of Gr-1+ cells in spleen. The lack of a marked effect of VEGF after 2 wk of treatment could be explained by the rate of DC turnover and necessity to replenish the resident DC. However, if VEGF was administrated together with FL, it resulted in a significant increase in the proportion of Gr-1+ cells. A possible explanation is that FL induced the production of a large number of progenitor cells. VEGF was able to exert its effects on this amplified population, thus enabling the changes in cell populations to become more pronounced.
To study the possible mechanisms of interaction between FL and VEGF, we determined whether both factors affected hemopoietic progenitor cells. Bone marrow cells obtained from mice treated with VEGF and FL were cultured with GM-CSF and IL-4, a cytokine combination that supports the generation of DC in vitro. The majority of these cells do not express the CD86 (B7-2) marker. To achieve their terminal maturation stage, cells need to be activated additionally for 2–3 days with TNF-α, LPS, or CD40 ligand. In this study, we did not induce the final cell maturation to evaluate the effect of FL on this process. Neither VEGF, nor FL treatment in mice significantly changed the total number of DC generated under these conditions. However, FL substantially (almost 2-fold) increased the proportion of CD11c+B7-2+ cells (p < 0.05). This effect was associated with a significant (2- to 3-fold) increase in the stimulatory activity of these cells in an allogeneic MLR. However, VEGF dramatically reduced this effect of FL as was seen by the inhibition of both the expression of B7-2 and the APC activity. It is important to note that the combination of GM-CSF and IL-4 does not expand lymphoid subpopulations of DC. Although it is difficult to make a direct comparison between these results and the effects seen in mature DC populations in vivo, they demonstrate a similar trend in the effects of VEGF on FL-mediated expansion of DCs.
These data suggest that VEGF regulated FL activity on the level of progenitor cells. To further characterize this effect, the number of progenitor cells was studied in a colony-formation assay. We found that FL significantly increased the number of macrophage and granulocyte-macrophage colonies. VEGF alone only moderately increased CFU-GM, but not CFU-M colonies. These data are in agreement with previously published observation describing VEGF effects in vivo and in vitro (28, 34). However, VEGF abrogated the FL effects on both types of myeloid colonies.
Thus, VEGF interferes with FL at the level of progenitor cells by blocking the ability of FL to increase the number of DC progenitors. To study the mechanism of these effects, we hypothesized that VEGF down-regulated the expression of FL receptor FLT3 on the surface of HPC and hence decreased the response of the cells to FL. However, our results did not support this hypothesis, as VEGF did not affect the expression of flt3 mRNA at any time examined.
Next, we investigated the role of the transcription factor NF-κB in this phenomenon. This family of transcription factors is composed of five DNA-binding subunits, p65 (RelA), p50, p52, c-Rel, and RelB, which act as homo- or heterodimers. In the cytoplasm of quiescent cells, these subunits are associated with inhibitory molecules of the IκB family. Cell activation by various stimuli like TNF-α, LPS, IL-1, CD40L, etc. results in serine phosphorylation and degradation of IκB with subsequent nuclear translocation of NF-κB dimers (35, 36, 37). NF-κB, and specifically RelB, plays a critical role in differentiation of DC in vivo and in vitro (27, 38, 39). Wu et al. have recently shown that RelB knockout mice lack myeloid, but not lymphoid, DC, confirming different mechanisms of regulation of both lineages of DC (40). We have previously demonstrated the inhibition of TNF-α inducible NF-κB activation by VEGF in vitro (27) and in vivo in mice treated with VEGF (28). Inhibition of NF-κB activation was observed after the first week of infusion and preceded any morphological changes (28). To date, there is no information available on whether FL is able to induce NF-κB activation. To study this issue, we treated bone marrow HPC in vitro with FL and assessed NF-κB nuclear translocation and DNA binding using EMSA. We found that FL markedly induced NF-κB activation. The addition of VEGF dramatically reduced this effect. Thus, VEGF was able to block NF-κB activation. The consequences of this inhibition are not known yet. It is plausible to suggest that it might lead to a differential expression of genes responsible for DC differentiation. The genes possibly involved in this process are currently under investigation.
In conclusion, this study for the first time demonstrates that the FL effect on DC generation can be inhibited by VEGF. DC play a central role in induction of antitumor immune responses. Therefore, increased production of functionally active DC might be beneficial for cancer treatment. This study demonstrates that tumor-derived factors like VEGF might negatively affect the therapeutic efficacy of FL-based immunotherapies. VEGF inhibition might thus improve the efficiency of FL treatment. In support of this concept, we recently demonstrated that DC function is enhanced in tumor-bearing mice following administration of anti-VEGF Ab (19).4
We thank Immunex and Genentech for the continual support of our studies.
This work was supported in part by National Institutes of Health Grants CA76321 (to D.P.C.), CA80126 (to M.R.S.), and CA81101 (to D.I.G.).
Abbreviations used in this paper: DC, dendritic cell; FL, FLT3 ligand; VEGF, vascular endothelial growth factor; HPC, hemopoietic progenitor cells; BFU, burst-forming unit; E, erythrocyte; G, granulocyte; M, macrophage; GM, granulocyte-macrophage; GEMM, mixed granulocyte-erythrocyte-macrophage.
D. I. Gabrilovich, T. Ishida, S. Nadaf, J. E. Ohm, and D. P. Carbone. Antibodies to vascular endothelial growth factor enhance the efficacy of cancer immunotherapy by improving endogenous dendritic cell function. Submitted.