Cyclooxygenase-2 (COX-2) is a rate-limiting enzyme in the synthesis of prostaglandins from arachidonic acid.1 Unlike COX-1, which is constitutively expressed in normal cells and tissues,2 COX-2 is over-expressed in multiple cancers including breast, colorectal, lung, prostate, bladder, gastric, head and neck cancers.3, 4, 5, 6 Over-expression of COX-2 by tumors has been associated with enhanced angiogenesis,7 increased invasion and metastasis,8, 9 resistance to apoptosis and diminished tumor immunity.10, 11, 12 The COX-2-dependent effects on tumor immunity appear to be partly mediated by prostaglandin E2 (PGE2) and include polarization toward TH2 responses and suppression of dendritic cell (DC) functions.13, 14, 15
Dendritic cells are endowed with the ability to efficiently process and present antigens, secrete a variety of important immunostimulatory cytokines and express critical costimulatory molecules necessary for inducing an effective antitumor immune response.16, 17 These attributes have made DC-based vaccines popular candidates for cancer immunotherapy.
In this study, we have evaluated the effect of celecoxib, a selective COX-2 inhibitor, on the efficacy of DC-based cell vaccines in preventing and treating 4T1 murine mammary tumors. Celecoxib was delivered orally in the diet over a short period of time (˜ 20 days), as this represents the conventional route of administration in humans. We demonstrate that dietary celecoxib, administered prior to, or after primary tumor establishment, synergizes with tumor lysate-pulsed DC and GM-CSF to improve the antitumor immune response by suppressing primary tumor growth and markedly reducing the occurrence of spontaneous lung metastases. Unlike treatment with celecoxib alone or DC + GM-CSF, the triple combination treatment (celecoxib + DC + GM-CSF) results in increased secretion of IFN-γ and IL-4 by T lymphocytes and increased accumulation of immune cells at the tumor site. In addition, oral celecoxib therapy decreased serum PGE2 levels and was associated with inhibition of tumor angiogenesis evidenced by decreased serum vascular endothelial cell growth factor (VEGF) levels as well as decreased endothelial vascular structures in the tumor. It has been recently reported that long-term, high-dose celecoxib therapy is associated with increased risk of deleterious cardiovascular events in patients predisposed to colon cancer.18 In this study, we show that short-term (˜20 days) celecoxib treatment in combination with DC-based vaccines is effective in treating metastatic breast cancer without overt toxicity. This chemo–immunotherapy approach may present an important, effective short-term therapy option circumventing the harmful effects associated with long-term celecoxib therapy.
The 4T1 cell line is a poorly immunogenic, highly metastatic variant of 410.4, a tumor subline isolated from a spontaneous mammary tumor that developed in a BALB/cfC3H mouse.19 It was kindly provided by Dr. Fred Miller of the Michigan Cancer Foundation, Detroit, MI. All cell lines were maintained for a limited time in vitro by passage in Iscove's Modified Dulbecco's Medium (IMDM) (JRH Biosciences, Lenexa, KS), containing penicillin (100 U/ml), streptomycin (100 μg/ml), fungizone (0.75 μg/ml) and 10% fetal bovine serum (FBS) (Gemini Bio-Products, Woodland, CA).
Celebrex™ (celecoxib) capsules (Pfizer, New York, NY) were obtained from the University of Arizona Campus Health Center. The Celebrex™ powder was shipped to Harlan Teklad (Madison, WI) to manufacture the celecoxib diet 500 (500 mg celecoxib/kg). Both the control (AIN93G) and the celecoxib diet 500 are corn starch-based diets from Harlan Teklad.
Preparation of lysates for Western immunoblotting
To obtain tumor tissue lysates, tumors were generated by orthotopic injection of 1 × 104 4T1 cells into the right mammary fat pad of BALB/c mice. After 20 days, the primary tumors as well as nonmalignant tissue from a naïve mouse were removed. Tumor and nonmalignant tissue were minced, digested in collagenase buffer (PBS containing 200 U/ml collagenase, 1 mg/ml glucose and 25 mg/ml bovine serum albumin) for 2 hr at 37°C, subsequently forced through metal mesh filter, washed twice with PBS and centrifuged for 10 min at 200g. The resulting pellet was lysed using CelLytic MT mammalian tissue lysis reagent (Sigma, St Louis, MO) containing protease inhibitors (1 mM PMSF, 1 mM Na3VO4, 1 μg/ml aproptinin, 1 μg/ml leupeptin and 1 μg/ml pepstatin) and homogenized by forcing through a 21-gauge needle, incubated on ice for 10 min before clarification by centrifugation for 15 min at 14,000g (4°C). Protein in the supernatant was quantified using the BCA Protein Assay (Pierce, Rockford, IL). To obtain in vitro tumor cell line lysate, overnight cultured 4T1 cells were washed twice with ice-cold PBS, scraped into CelLytic lysis buffer containing protease inhibitors as described earlier and forced through a 21-gauge needle. The lysate was incubated on ice for 10 min and centrifuged at 14,000g for 15 min, and protein in the supernatant was recovered and quantified as described earlier.
Proteins (50 μg) from the lysates were resolved by 10% SDS-PAGE and electro-transferred to polyvinylidene difluoride (PVDF) membrane. Nonspecific binding sites were saturated by incubation in TBST/MLK (Tris-buffered saline containing 0.1% Tween-20 and 5% nonfat powdered milk). The membrane was immunoblotted using a COX-1 (1:800 in TBST/MLK) or COX-2 (1:1500 in TBS/MLK) specific antibody (both Cayman Chemical, Ann Arbor, MI) and visualized with a goat antimouse horseradish peroxidase (HRP)-conjugated secondary antibody (1:10000 in TBST/MLK) (Upstate Biotechnology, Lake Placid, NY) or goat antirabbit HRP-conjugated secondary antibody (1:10000 in TBST/MLK) (Pierce) using the Supersignal West Pico Chemiluminescent Substrate (Pierce). Subsequently, the membranes were stripped using Restore Western Blot Stripping buffer (Pierce) according to the manufacturer's instructions and reprobed with a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) specific antibody (1:500 in TBST/MLK) (Chemicon, Temecula, CA) and visualized using a goat antimouse-HRP-conjugated secondary antibody (1:1000 in TBST/MLK) (Upstate) and supersignal as described earlier. COX-1 and COX-2 electrophoresis standard were purchased from Cayman Chemical.
Generation of dendritic cells
For DC generation, bone marrow (BM) cells were harvested by flushing marrow cavities of femurs and tibia of BALB/c mice under aseptic conditions. The cells were cultured with 100 U/ml GM-CSF and 100 U/ml IL-4 at 106 cells/ml in RPMI 1640 (Mediatech, Herndon, VA) supplemented with 10% heat-inactivated FBS, as previously described.20 On day 6, the nonadherent and loosely adherent cells were collected, labeled with anti-CD11c microbeads (Miltenyi Biotec, Auburn, CA) and DCs were isolated by positive selection using the Midi MACS magnetic separation system (Miltenyi) according to the manufacturer's instructions. Eluted cells were centrifuged (250g, 5 min), washed with complete RPMI and plated at 106 cells/ml with 100 U/ml each of IL-4 and GM-CSF in a Petri dish overnight. DCs were pulsed for 24 hr with 4T1 tumor lysates at a ratio of 3 tumor cell equivalents per DC.20 Tumor (4T1) cell lysate for DC pulsing was prepared by repeated freeze-thaw cycles (6 cycles). After lysate pulsing, DCs were matured by incubation in monophosphoryl lipid A (MPL, 2 mg/ml) (Avanti Polar Lipids, Alabaster, AL) for 24 hr. These cells were 73% positive for CD11c expression, 91% positive for MHC class II (IAb) expression and 15, 88 and 62% positive for expression of the costimulatory molecules, CD40, CD80 and CD86, respectively.
Female (6–8-weeks-old) BALB/c mice were purchased from the NCI Frederick facility (Frederick, MD). The mice were housed in the University of Arizona animal facility in accordance with the principles of animal care (NIH publication No.85–23, revised 1985). For the prevention study, mice were injected with 104 viable 4T1 cells in 100 μl PBS into the right mammary fat pad and were fed with either control or celecoxib diet (total dose of 1.25 mg of celecoxib/mouse/day equivalent to 42 mg/kg body weight) beginning on the day of tumor injection until the end of the experiment. For the established disease study, mice were injected with 104 viable 4T1 cells and fed with control diet until tumors became palpable. On day 14 posttumor cell injection, the mice were switched to the celecoxib diet. DC vaccinations were started on day 14 posttumor injection and consisted of 3 subcutaneous (s.c.) injections each of 2 × 106 tumor-lysate-pulsed, matured DC given 7 days apart. Some mice received six s.c. injections of GM-CSF (10,000 U/injection) given on days 14 and 15, 21 and 22, and 28 and 29.21 Tumors were measured in 2 dimensions every 5 days using Vernier calipers, and the tumor volume was calculated using the formula V = (l × w2)/2, where V = volume (mm3), l = long diameter and w = short diameter. Metastatic nodules in the lungs of tumor-bearing mice were enumerated by staining with India ink and Fekete's solution as previously described.22
Pathological examination for celecoxib toxicity
Female BALB/c mice on the control diet or the celecoxib diet were euthanized by carbon dioxide inhalation and necropsied to determine whether there was evidence of toxicity associated with administration of celecoxib in the diet for 20 days. Heart, kidney, liver, lung, pancreas, spleen and gastrointestinal tract were fixed in 10% buffered formalin, processed and paraffin-embedded, and 5-micron sections were stained with hematoxylin and eosin (H&E) and evaluated by a veterinary pathologist. Compared to mice on the control diet, no gross or histologic lesions were observed in the mice maintained on the celecoxib diet.
IFN-γ and IL-4 production
Mice from the various treatment groups were killed and their spleens removed. Splenocytes were pooled from 3 animals from each treatment group. CD4+ and CD8+ T cells were separated from erythrocyte-depleted splenocytes by positive selection, using the Midi MACS magnetic separation system (Miltenyi Biotec, Auburn, CA) according to the manufacturer's instructions. One million CD4+ or CD8+ T cells were restimulated with 2.5 × 105 tumor lysate-pulsed, TNF-α matured DC20 in a 24-well tissue culture plate for 48 hr. To determine the specificity of the immune response, DCs were pulsed with lysates derived from 4T1 tumor cells or from 12B1 leukemia cells (syngeneic to BALB/c). The supernatant was then collected and assayed for IFN-γ and IL-4 production by ELISA (R&D Systems, Minneapolis, MN), according to the manufacturer's protocol.
Tumors were removed at day 33 posttumor injection. The excised tumors were bisected, and half of the mass was fixed in 10% buffered formalin for 24 hr and then processed for routine paraffin embedding. Five-micron sections of each lesion were stained with H&E or the endothelial cell specific marker CD31 (PECAM-1) by immunohistochemistry. The remaining half of each tumor was snap frozen for immunohistochemical staining for CD4 and CD8 leukocyte markers.
Frozen sections for each of the 4 treatment groups (control diet, celecoxib diet, control diet + DC + GM-CSF and celecoxib diet + DC + GM-CSF) were stained for CD4 (L3T4 rat monoclonal antibody, 1:20; BD Pharmingen, San Diego, CA) and CD8 (53-6.7 rat monoclonal antibody, 1:10; eBioscience, San Diego, CA). Rabbit antirat IgG biotinylated secondary antibody was used with a standard streptavidin HRP label and DAB substrate (DAKOcytomation, Carpinteria, CA). Formalin-fixed paraffin-embedded 5-micron tumor sections were stained for CD31 (PECAM-1, rat monoclonal antibody, 1:10; BD Pharmingen, San Diego, CA). Rabbit antirat IgG biotinylated secondary antibody was used with a standard streptavidin HRP label and DAB substrate (DAKOcytomation, Carpinteria, CA). Five fields of vision per tumor of 5 fixed tumors (control diet), 2 tumors (celecoxib diet), 3 tumors (control diet + DC + GM-CSF) and 2 tumors (celecoxib diet + DC + GM-CSF) were counted.
Serum prostaglandin E2 and vascular endothelial growth factor levels
Sera were pooled from 7 mice from each treatment group (control diet, celecoxib diet, control diet + DC + GM-CSF and celecoxib diet + DC + GM-CSF) and assessed for prostaglandin E2 (PGE2) and vascular endothelial growth factor (VEGF) levels using the PGE2 Metabolite ELISA kit (Cayman Chemical, Ann Arbor, MI) and the VEGF ELISA kit (R&D Systems, Minneapolis, MN) according to the manufacturers' recommended protocols, respectively.
Statistical significance of differences among data sets of treatment groups were assessed by one-way ANOVA including Tukey-Kramer post tests for multiple comparisons using the Prism software (GraphPad, San Diego, CA) and Excel (Microsoft Corp., Redmond, WA). Probability values of p < 0.05 were considered indicative of significant differences between data sets.
4T1 tumor cells and primary tumors over-express COX-2
Our first objective was to determine cyclooxygenase expression in 4T1 mammary tumors. Lysates were prepared from tumors that arose after transplantation of 4T1 cells in BALB/c mice and analyzed by Western immunoblotting using COX-1 or COX-2-specific antibodies. Figure 1 shows that COX-2 was markedly over-expressed in tumors resulting from implantation of 4T1 cells as compared to that in nonmalignant tissue. However, the possibility exists that the detected COX-2 in the tumor lysate may be derived from infiltrating host cells. A differential cell count of single cell suspensions from the tumors indicated that host cells constituted only a minor component of the tumor tissue. Thus, it is unlikely that the majority of the detected COX-2 protein is derived from infiltrating inflammatory cells (data not shown).
Figure 1 also demonstrates the expression of COX-1 in malignant, nonmalignant tissue as well as 4T1 tumor cells as COX-1 is constitutively expressed in most tissues.
Efficacy of COX-2 inhibition by celecoxib
Since COX-2 is a key regulator in the prostaglandin synthesis pathway, we determined PGE2 levels in the serum from celecoxib-treated and untreated mice so as to assess the efficacy of COX-2 inhibition by dietary celecoxib. Because PGE2 in the serum is rapidly converted to its 13,14-dihydro-15-keto metabolite, direct measurement of PGE2 serum levels is not possible. However, PGE2 metabolite levels can be used as an indirect measure of actual PGE2 levels.23, 24 Serum from tumor-bearing mice from the various treatment groups as well as serum from naïve (nontumor-bearing) mice was collected and pooled (7 per group), and the PGE2 levels were determined using a PGE2 metabolite ELISA. Figure 2 shows that 4T1 tumor-bearing mice had significantly higher PGE2 serum levels than naïve mice (p < 0.001). More importantly, the dietary celecoxib treatment significantly reduced (p < 0.001) serum PGE2 levels in the celecoxib alone and celecoxib + DC + GM-CSF treated mice 3–4 fold as compared to control diet-treated mice. In addition, the PGE2 levels in mice from both celecoxib-treated groups (celecoxib alone and celecoxib + DC + GM-CSF) were significantly lower (p < 0.001) than PGE2 levels in mice treated with DC + GM-CSF. This demonstrates that dietary celecoxib significantly inhibited COX-2 enzymatic activity in the celecoxib diet-fed mice.
Dietary celecoxib is more effective in the prevention setting than in the established disease setting to inhibit tumor growth
To assess the ability of celecoxib to inhibit 4T1 tumor growth, 104 viable tumor cells were injected into the mammary fat pad of BALB/c mice, and dietary celecoxib therapy was either initiated on the same day as the tumor cell injections (prevention model) or after tumors became palpable (established disease model).
The data from 2 independent experiments (Figs. 3a and 3c) indicate that when used as a prophylactic agent (prevention model), celecoxib consistently inhibited the growth of 4T1 tumors in both experiments. Commencing on day 25 (Fig. 3a) or day 20 (Fig. 3c) posttumor injection, there was a significant reduction (p < 0.05) in the growth rate of primary tumors in celecoxib-fed mice compared to animals on the control diet. Tumors of mice on the control diet grew progressively, ranging from an average tumor volume of 601 ± 89 mm3 (Fig. 3a) to 977 ± 218 mm3 (Fig. 3c) on day 30 and day 35, respectively. In contrast, the average tumor volumes in the celecoxib-fed mice on the corresponding days were 280 ± 86 mm3 (Fig. 3a) and 234 ± 81 mm3 (Fig. 3c), respectively. The effect of celecoxib treatment was more pronounced in the prophylactic model than in the established disease model, as celecoxib therapy initiated after tumor establishment did not consistently inhibit tumor growth with the same efficacy as celecoxib given at the time of tumor implantation.
Because spread from the primary tumor is a major cause of tumor recurrence and mortality, we were interested in the effect of celecoxib treatment on metastasis formation. For this purpose, we enumerated pulmonary metastatic lesions in the mice after termination of the experiment. Figures 3b and 3d demonstrate that the celecoxib diet-fed mice had significantly reduced numbers of lung metastases (p < 0.001) when celecoxib treatment was initiated at the same time as tumor cell injection. In the prevention setting, the average number of visible metastatic lung nodules in the celecoxib-fed groups was significantly reduced (19 ± 9, Fig. 3b and 12 ± 8, Fig. 3d) compared to animals on the control diet (63 ± 9, Fig. 3b and 40 ± 7 Fig. 3d). In contrast, mice that were switched to the celecoxib diet after tumor establishment displayed a reduction in the mean number of pulmonary metastases, but the reduction did not achieve statistical significance. These results demonstrate that dietary celecoxib therapy, when started simultaneously with tumor cell injection significantly inhibited primary tumor growth as well as metastasis to the lung. However, celecoxib treatment initiated after tumor establishment was less efficacious.
Dietary celecoxib in combination with DC vaccination suppresses growth of established tumors and reduces lung metastasis
Since celecoxib monotherapy did not consistently inhibit tumor progression when initiated after tumor establishment, we combined celecoxib treatment with DC-based immunotherapy in the presence of GM-CSF to improve the therapeutic outcome. To augment antigen presentation and facilitate DC recruitment in vivo, GM-CSF has been used as an adjuvant in antitumor immunotherapy.21, 25, 26, 27 For this experiment, mice were switched to the celecoxib diet on day 14 posttumor injection and received 3 vaccinations of tumor lysate-pulsed, matured DCs at 7-day intervals. To assess whether the addition of the adjuvant GM-CSF would improve celecoxib + DC treatment, some mice also received GM-CSF injections. The data in Figure 4a show that the combination of celecoxib and DC vaccination significantly reduced the growth rate of established primary tumors (p < 0.05). The average tumor volume on day 33 for this group was 202 ± 58 mm3 (Fig. 4a) compared to the average tumor volume of 506 ± 66 mm3 observed in untreated control animals (Fig. 4a). However, an even more marked reduction in primary tumor growth was observed in animals in the celecoxib group that received DC vaccination + GM-CSF (p < 0.001). The average tumor volume in this group on day 33 ranged between 104 ± 3 mm3 (Fig. 4a) and 127 ± 31 mm3 (Fig 4c) compared to the average tumor volumes of untreated mice, which ranged between 506 ± 66 mm3 (Fig. 4a) and 875 ± 165 mm3 (Fig. 4c). This is underscored by the observation that there was tumor regression in 2 animals from the experiment depicted in Figure 4c that received the celecoxib + DC + GM-CSF triple therapy. The 2 mice were tumor-free by day 28 and 35 posttumor injection, respectively. The animals were still tumor-free 8 weeks later when they received a tumor challenge of 1 × 105 4T1 cells (10-fold the original dose) but no other treatment. While two naïve control animals developed tumors of ˜1000mm3 by day 30 posttumor challenge, the 2 animals that had previously cleared the tumors remained tumor-free (data not shown). Because of the high primary tumor burden, the control animals in the experiment depicted in Figure 4a had to be sacrificed on day 33. However, the remaining treatment groups were followed until day 40 posttumor injection. By day 40 when the average tumor size in the celecoxib + DC treatment group was 357 ± 23 mm3, the average tumor size in the celecoxib + DC + GM-CSF group was 132 ± 10 mm3. In addition, after we had established that it is the combination of celecoxib + DC + GM-CSF that confers the inhibition of tumor progression (Figs. 4a and 4b), and so as to better examine the effect of the triple combination treatment (celecoxib + DC + GM-CSF) as well as to increase statistical significance, we omitted the celecoxib diet + DC, GM-CSF alone and the DC alone control groups in the subsequent experiments in favor of more test subjects per treatment group.
The goal of the combination therapy of celecoxib and DC vaccination was not only to reduce primary tumor growth but also to reduce the occurrence of lung metastasis. The data in Figure 4b show that the combination of celecoxib therapy plus DC vaccination reduced the incidence of lung metastasis. The average number of visible lung nodules in this group was 10 ± 4 compared to 56 ± 8 in the control group (p < 0.001). As with the effect on the primary tumor growth, the triple combination therapy (celecoxib + DC + GM-CSF) had the most marked effect on the formation of lung metastases. The average number of visible lung nodules in this group was 1 ± 1 (Fig. 4b, p < 0.001) or 3 ± 2 (Fig. 4d, p < 0.001).
These results demonstrate that the combination of dietary celecoxib and DC vaccination plus GM-CSF injections was successful in reducing primary tumor growth as well as lung metastasis even when therapy was initiated after the establishment of primary tumors.
Dietary celecoxib in combination with DC vaccination and GM-CSF enhances cellular infiltration and necrosis of tumors
Tumors were isolated from all treatment groups and evaluated by H&E staining for identification of tumor infiltrating immune cell subsets. Examination of the H&E stained sections revealed demonstrable differences among several groups (Figs. 5a–5h). The lesions from the control-fed group were composed of a solid growth of viable large neoplastic epithelial cells with large round to oval nuclei and small to moderate amounts of eosinophilic cytoplasm. There were 2–4 mitotic figures per high power field. The interface with the adjacent subcutaneous adipose tissue was abrupt with no evidence of a cellular infiltrate (Figs. 5a and 5b). However, sections of the tumor masses from the remaining groups revealed differences in degree of necrosis and cellular infiltrates. The lesions from the celecoxib only (Figs. 5c and 5d) or control-fed + DC + GM-CSF (Figs. 5e and 5f) groups showed marked necrosis within the tumor mass with a rim of viable tumor and inflammatory cells at the interface with the adjacent adipose tissue. The control-fed + DC + GM-CSF group appeared to have a more florid inflammatory infiltrate compared to the celecoxib only–fed group. The inflammatory cells were neutrophils, small lymphocytes and scattered plasma cells. The celecoxib-fed + DC + GM-CSF group (Figs. 5g and 5h) showed marked necrosis with a small rim of remaining viable tumor. There was a marked infiltrate of inflammatory cells at the periphery of the lesion. The inflammatory cells were neutrophils, small lymphocytes and scattered plasma cells. In addition, there were small clusters of small lymphocytes.
Immunohistochemical staining for subsets of CD4+ and CD8+ T lymphocytes revealed differences in the infiltration of the tumor mass by these cell populations (Figs. 6a–6h). The sections from each group were ranked for degree of CD8 and CD4 positive cell infiltrates. For both subsets of T lymphocytes, the celecoxib-fed + DC + GM-CSF treatment group displayed the greatest degree of infiltration (Figs. 6g and 6h), followed by the control-fed + DC + GM-CSF group (Figs. 6c and 6d) and the celecoxib only–fed treatment group (Figs. 6e and 6f), with the control-fed group (Figs. 6a and 6b) displaying the least infiltration.
Combination treatment with celecoxib plus DC and GM-CSF elicits increased cytokine production by CD4+ and CD8+ T lymphocytes
To determine whether the superior antitumor effect of celecoxib diet + DC and GM-CSF treatment is associated with enhanced immune responses, CD4+ and CD8+ T cells were isolated from spleens of tumor-bearing mice by immunomagnetic bead separation, restimulated in vitro with tumor lysate-pulsed, TNF-α matured DC and evaluated for IFN-γ and IL-4 production by ELISA.
Figure 7a shows that CD4+ T cells isolated from mice treated either with dietary celecoxib alone, DC + GM-CSF or celecoxib + DC + GM-CSF produced significantly higher amounts of IFN-γ than CD4+ T cells isolated from mice on the control diet (p < 0.001). The triple combination treatment resulted in the highest production of IFN-γ (6385 ± 337 pg/ml) followed in order by the DC + GM-CSF treatment (4547 ± 243 pg/ml) and the celecoxib only treatment (2164 ± 71 pg/ml). The combination of celecoxib + DC + GM-CSF increased the IFN-γ secretion significantly in comparison to celecoxib diet alone (p < 0.001) or DC + GM-CSF therapy (p < 0.01).
The same rank order was observed for the IFN-γ secretion by CD8+ T cells albeit at lower levels (Fig. 7b). Although both the DC + GM-CSF (p < 0.001) and the triple combination therapy (p < 0.001) resulted in a significant increase in IFN-γ secretion compared to the celecoxib therapy alone, there was no statistical difference in IFN-γ secretion between the DC + GM-CSF and the triple combination therapy.
The antitumor immune response was 4T1 specific, as stimulation with haplotype compatible (H-2d) irrelevant tumor lysates (12B1) did not result in substantial IFN-γ or IL-4 secretion by CD4+ or CD8+ T cells.
Interestingly, the data in Figure 7c show that CD4+ T cells isolated from the mice in the 3 treatment groups secreted significantly higher levels of IL-4 than CD4+ T cells isolated from mice on the control diet (p < 0.001). Figure 7c also shows that the IL-4 production by CD4+ T cells isolated from mice treated with either DC + GM-CSF (2732 ± 50 pg/ml) or the combination of celecoxib + DC + GM-CSF (4102 ± 757 pg/ml) was significantly higher than IL-4 production by CD4+ T cells from mice on the celecoxib diet alone (456 ± 11 pg/ml, p < 0.001). In addition, the combination of the celecoxib diet with DC + GM-CSF resulted in a significant increase (p < 0.001) in IL-4 secretion in comparison to DC + GM-CSF therapy alone.
Taken together, these results demonstrate that the immune response elicited in the treated mice was tumor-specific, and that the triple combination therapy of celecoxib + DC + GM-CSF was most effective in stimulating IFN-γ and IL-4 secretion.
Combination treatment of celecoxib plus DC and GM-CSF reduces angiogenesis
Angiogenesis enables solid tumors to progress from a growth-restricted state to a more invasive rapidly growing phenotype. Since celecoxib has been shown to be a potent antiangiogenic factor in vitro and in animal models,7, 28, 29, 30 we determined the effects of dietary celecoxib treatment on tumor angiogenesis. As a measurement of angiogenesis, we assessed the density of vascular structures within the primary tumor (Figs. 8a–8d) by staining tumor sections for the vascular endothelial cell marker CD31 (PECAM-1), and we determined the serum levels of VEGF. Tumors from the control group (Fig. 8a) displayed an average density of vascular structures of 14.4 ± 1.4 in comparison to tumors from mice on the celecoxib diet alone (Fig. 8b) that displayed an average density of 13.1 ± 1.8. The densities of PECAM-1 positive vascular structures per field of vision in tumors from mice on celecoxib diet + DC + GM-CSF (Fig. 8d) and the DC + GM-CSF group (Fig. 8c) were 8.4 ± 1.5 and 7.3 ± 0.8, respectively, and significantly lower (p < 0.05) in comparison to the vascular density in tumors from mice on the control diet. This finding was largely consistent with VEGF levels in sera from the various treatment groups (Fig. 8e). Serum VEGF levels were significantly reduced in the celecoxib-treated mice (celecoxib and celecoxib + DC + GM-CSF) in comparison to the control diet only-fed mice (p < 0.001). Interestingly, treatment with DC + GM-CSF reduced VEGF serum levels to a similar extent as treatment with celecoxib diet. However, the triple combination therapy reduced the serum VEGF level even further that was significantly different compared to serum VEGF level in the DC + GM-CSF group (p < 0.01).
In this study, we evaluated the effect of the selective COX-2 inhibitor, celecoxib on the growth of the COX-2-producing, poorly immunogenic and highly metastatic murine mammary tumor, 4T1. For this purpose, mice were fed with a nontoxic dose of celecoxib, in the diet prior to, and after establishment of orthotopically-injected 4T1 mammary tumors. Our results demonstrate that dietary celecoxib, by itself, significantly suppressed the growth of primary tumors as well as the incidence of lung metastases in the prophylactic setting. However, celecoxib was less effective at inhibiting the progression of large (˜30mm3) pre-established tumors, which is consistent with a report by DeLong et al. showing that rofecoxib therapy inhibited the growth of small but not large tumors in a murine mesothelioma model.31 Therefore, we sought to devise a more effective treatment option to combat established primary tumors as well as lung metastasis that are often associated with tumor recurrence and increased mortality. In this regard, we demonstrated that when used in combination with tumor lysate-pulsed DC and the biological adjuvant GM-CSF, celecoxib suppressed primary tumors and markedly reduced the incidence of spontaneously arising lung metastases. This triple combination therapy was superior to any of the other treatments (DC + GM-CSF, celecoxib + DC, celecoxib + GM-CSF, celecoxib alone and DC alone or GM-CSF alone). Our study extends the findings by DeLong et al. showing that the antitumor activity of the COX-2 inhibitor rofecoxib could be enhanced by combining with IFN-β therapy, and that rofecoxib treatment resulted in increased infiltration of CD4+ and CD8+ T cells of mesotheliomas.31 This is consistent with our finding of increased CD4+ and CD8+ T cell infiltration of the tumors in celecoxib + DC + GM-CSF-treated mice. In addition, another study combining immunotherapy with COX-2 inhibition showed that the combination of celecoxib with a carcinoembryogenic antigen (CEA) poxvirus-based vaccine in a CEA expressing multiple intestinal neoplasia mouse model (MIN)32 resulted in improved antitumor immunity and long-term survival. Although, the direct effect of PGE2 on DC and T cell function, particularly in the tumor microenvironment, remains to be evaluated, the salutary effect of celecoxib plus DC vaccines may be attributed in part to the reduction of the immune suppressive PGE2 production and retention of the antigen presenting functions of DC when tumor-derived COX-2 is suppressed as has been previously reported.13 Using a COX-2-producing 3LL Lewis lung model, Sharma et al.13 reported potent activity of peptide-pulsed DC in controlling the growth of pre-established tumors in which tumor-derived COX-2 was inhibited by antisense gene transfer. Our study also extends the findings of Connolly et al.29 in which they reported significant inhibition of established 4T1 tumors and spontaneous lung metastases following intraperitoneal administration of the selective COX-2 inhibitor, SC-236. Tumor suppression was ascribed to increased apoptosis and decreased angiogenesis in treated mice.29 This is consistent with our finding that the celecoxib combination treatment decreases VEGF serum levels and inhibits the proliferation of vascular endothelial structures in the tumor. The antiangiogenic actions of celecoxib may explain the high efficacy of the prophylactic celecoxib therapy in our study. In addition, we demonstrate that the short duration of administration of this dose of celecoxib did not give rise to overt pathologic abnormalities, suggesting that short-term celecoxib therapy may be used safely to treat mammary cancer if combined with DC immunotherapy. This chemo–immunotherapy approach is potentially promising when viewed in the context of the recently reported increased risks of cardiovascular events associated with high-dose, long-term celecoxib therapy.18 The superior inhibition of tumor growth by the triple combination therapy (celecoxib + DC + GM-CSF) in our study was characterized by increased IFN-γ secretion and infiltration of tumors by CD4+ and CD8+ T cells. The increased IFN-γ secretion by CD4+ and CD8+ cells is suggestive of polarization towards a TH1 cell-mediated antitumor immune response that is generally more efficacious in combating tumors.14, 15 This is consistent with experiments showing that the abrogation of COX-2 expression and consequential reduction of PGE2 levels alleviates the bias toward TH2 cytokine secretion resulting in a more effective antitumor response.12, 13 Interestingly, our data also revealed significant IL-4 secretion by CD4+ T cells isolated from mice in the treatment groups (celecoxib, 456 ± 11 pg/ml; celecoxib + DC + GM-CSF, 2732 ± 50 pg/ml and celecoxib + DC + GM-CSF, 4102 ± 757 pg/ml) versus control diet (41 ± 1pg/ml).
Our studies demonstrate that celecoxib treatment induces cellular responses in mice that may be involved in the control of established 4T1 tumors and suggest a potential role for IL-4 in the development of tumor immunity as previously reported by Schuler et al.33 They implicated IL-4 in the development of tumor immunity by demonstrating a reduction in IFN-γ production, CTL activity and tumor-reactive serum (IgG2a) in tumor-bearing IL-4 knock-out mice.33 In addition, we show that celecoxib therapy reduces tumor angiogenesis, demonstrating that the mechanism of tumor growth attenuation by celecoxib may be due to the combination of the enhancement of antitumor immune responses by lowering PGE2 levels and the inhibition of tumor angiogenesis.
In summary, we have demonstrated, the marked effect of short-term orally administered selective COX-2 inhibitor plus DC-based immunotherapy in suppressing the incidence of spontaneously arising lung metastases of a poorly immunogenic tumor. Oral administration of celecoxib in combination with DC-based immunotherapy represents a novel approach that can be exploited to generate a robust antitumor response capable of inhibiting primary tumor growth and controlling disseminated disease in the clinical setting.
We thank Barbara Carolus and Debbie Sakiestewa for flow cytometric analysis.