Increase in frequency of myeloid-derived suppressor cells in mice with spontaneous pancreatic carcinoma

Abstract

Pancreatic adenocarcinoma is one of the deadliest cancers with poor survival and limited treatment options. Immunotherapy is an attractive option for this cancer that needs to be further developed. Tumours have evolved a variety of mechanisms to suppress host immune responses. Understanding these responses is central in developing immunotherapy protocols. The aim of this study was to investigate potential immune suppressor mechanisms that might occur during development of pancreatic tumours. Myeloid-derived suppressor cells (MDSC) from mice with spontaneous pancreatic tumours, mice with premalignant lesions as well as wild-type mice were analysed. An increase in the frequency of MDSC early in tumour development was detected in lymph nodes, blood and pancreas of mice with premalignant lesions and increased further upon tumour progression. The MDSC from mice with pancreatic tumours have arginase activity and suppress T-cell responses, which represent the hallmark functions of these cells. Our study suggests that immune suppressor mechanisms generated by tumours exist as early as premalignant lesions and increase with tumour progression. These results highlight the importance of blocking these suppressor mechanisms early in the disease in developing immunotherapy protocols.

Introduction

Pancreatic cancer is the fourth leading cause of cancer-related death in the United States.¹ Despite numerous efforts in developing new therapies, the prognosis for patients with pancreatic cancer remains poor. One therapeutic approach that has shown some promise is immunotherapy.² Mouse models for spontaneous pancreatic cancer represent an ideal system to develop and validate immunotherapeutic approaches against pancreatic cancer.³

Previously, we have described a mouse model of spontaneous pancreatic cancer, which mimics closely the pathomorphological features and genetic alterations of the human disease.³ In this model, elastase-transforming growth factor-α (EL-TGF-α) transgenic mice are crossed with p53-deficient mice.^4,5 The EL-TGF-α/p53^−/− mice spontaneously develop acinar-cell-type pancreatic cancer 120 days after birth.⁶ In contrast, EL-TGF-α/p53^+/− mice have premalignant lesions, which develop into pancreatic neoplasms, when mice are 180 days of age. Previously, we have demonstrated spontaneous cellular and humoral responses in EL-TGF-α/p53^−/− mice. However, the immune responses seen were much lower than those in subcutaneously injected pancreatic tumour cell lines (derived from the pancreatic tumours) suggesting existing immune suppressor mechanisms in tumour-bearing mice.³ So far, the exact mechanism of immunosuppression in these mice has not been examined.

Tumours have evolved different suppressor mechanisms to evade the host’s immune response.⁷ Understanding the mechanisms behind this immune suppression is important in the success of immunotherapy protocols. Recently, myeloid-derived suppressor cells (MDSC) have been characterized as a population of cells that can negatively regulate T-cell function.⁸ The MDSC are a heterogeneous population of myeloid cells including macrophages, granulocytes and other cells that express both Gr-1 and CD11b in mice and suppress immune responses in vivo and in vitro.⁹ Although a number of studies have investigated the function of MDCS in tumour models, only limited data are available on these cells during the development of spontaneous tumours in mice. Our mouse model of pancreatic adenocarcinoma provides an optimal example to analyse MDSC at different stages of tumour development.

In this study, we have analysed the phenotype and frequency of CD11b⁺ Gr-1⁺ MDSC in spleen, mesenteric lymph nodes and pancreas of mice with pancreatic tumours, premalignant lesions and wild-type mice. In addition, MDSC were analysed in mice after subcutaneous injection of pancreatic tumours. Our results show that CD11b⁺ Gr-1⁺ cells can be detected in the early stages of pancreatic tumour development in premalignant lesions and expand upon tumour progression. Efforts to block these cells early in the tumour progression are important in developing immunotherapeutic protocols for pancreatic cancer.

Materials and methods

Mice

The EL-TGF-α/p53^−/− double transgenic mice and the EL-TGF-α/p53^+/− heterozygous mice have been previously described.³ These mice are on a C57BL/6 background and can specifically over-express TGF-α in the pancreas under the control of elastase promoter. Naive C57BL/6 and BALB/c mice were purchased from Charles River Laboratories (Sulzfeld, Germany). Where indicated, 10⁷ mPAC cells (a pancreatic tumour cell line)³ were injected subcutaneously into C57BL/6 mice.

Cell preparation and culture

Spleens, mesenteric lymph nodes and pancreatic tissues (naive pancreas, pancreatic tumour or premalignant lesion) were removed from the mice and lymphocytes were isolated. Blood was collected by heart puncture. Red blood cells were lysed using PharM Lyse™ lysing solution (BD Biosciences, Heidelberg, Germany). The lymphocytes were cultured in T-cell medium (RPMI-1640 with 10% fetal calf serum, 100 U/ml penicillin and streptomycin, 2 mm l-glutamine, 1 mm sodium pyruvate, 1% non-essential amino acids and 50 μmβ-mercaptoethanol) under standard conditions (37°, 5% CO₂).

Flow cytometry and cell sorting

Anti-mouse Gr-1 (RB6-8C5) was purchased from BioLegend (San Diego, CA); anti-mouse CD11b (M1/70.15) was from Immunotools (Friesoythe, Germany); anti-mouse H-2Kd (SF1-1.1), I-A/I-E (MHC-II; M5/114.15.2), CD86 (GL1) and CD11c (HL3) were from BD Pharmingen (Heidelberg, Germany); anti-mouse CD124 (hIL4R-M57) and F4/80 (BM8) were from eBioscience Inc. (San Diego, CA); and anti-mouse Ly6c (1G7.G10) was from Miltenyi Biotec (Bergisch Gladbach, Germany). Flow cytometry was performed on a Becton Dickinson FACSCalibur using Cell Quest Software (Becton Dickinson, Heidelberg, Germany). Data analysis was performed using FlowJo software (Tree Star Incorporation, Ashland, OR). The corresponding isotype antibodies were used with all the samples as controls. The cells were sorted with the BD FACSAria cell sorting system (Becton Dickinson). All the sorted cells were of purity above 95%.

Arginase activity assay

The arginase activity was tested as described previously.¹⁰ The urea concentration was measured at 540 nm. A standard curve of urea serial dilutions was performed in parallel.

MDSC suppression and depletion assay

C57BL/6 splenocytes (10⁵) were mixed with 10⁵ gamma-irradiated BALB/c splenocytes and cocultured with CD11b⁺ Gr-1⁺ cells at different ratios. For MDSC depletion assay, 10⁵ naive or MDSC-depleted EL-TGF-α/p53^−/− splenocytes were mixed with BALB/c splenocytes (50 μg/ml mitomycin treatment at 37°) at different ratios. Phytohaemagglutinin (0·1 μg/ml) was used for stimulation instead of BALB/c splenocytes as a positive control. After 48 hr of coculture, supernatants were collected and tested for interferon-γ (IFN-γ) by enzyme-linked immunosorbent assay (R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions. To measure the proliferation of the splenocytes, [³H]thymidine (Amersham, Freiburg, Germany) was added to the cultures after 48 hr of coculture. Thymidine incorporation was measured after 16 hr using a scintillation counter (Wallac, Turka, Finland).

Statistical analysis

All the statistical analyses were based on Student’s t-test. All P-values were two-tailed and a P-value < 0·05 was considered to be significant.

Results

EL-TGF-α/p53^−/− mice develop spontaneous pancreatic tumours within 120 days after birth

We have analysed EL-TGF-α/p53^−/−, EL-TGF-α/p53^+/− and wild-type mice at the age of 100–120 days. EL-TGF-α/p53^−/− mice develop pancreatic tumours within 120 days after birth as previously described in contrast to EL-TGF-α/p53^+/− mice, which develop premalignant lesions.^3,4 Histological analysis of pancreatic tissue from EL-TGF-α/p53^−/− mice revealed malignant pancreatic tumours with a complete loss of papillary structures, numerous mitotic figures and also high-grade atypia in stroma cells as expected (Fig. 1a). In contrast, several areas of cystic-type lesions lined by cuboidal epithelial cells with few of them showing moderate to severe cellular atypia are found in the pancreas from EL-TGF-α/p53^+/− mice (Fig. 1b). No fully developed malignant tumours were identified in EL-TGF-α/p53^+/− mice within the first 100–120 days of age in any of the analysed mice.

Figure 1.

Histological analysis of pancreas from elastase-transforming growth factor-α (EL-TGF-α)/p53^−/− (a), EL-TGF-α/p53^+/− (b) and wild-type (c) mice.

CD11b⁺ Gr-1⁺ MDSC are increased in mice with pancreatic tumours

In our previous study we were able to demonstrate spontaneous tumour-specific immune responses in EL-TGF-α/p53^−/− (spontaneous pancreatic tumours) and in wild-type mice after subcutaneous challenge with a pancreatic tumour cell line (mPAC) derived from the pancreatic tumours.³ However, tumours progressed in EL-TGF-α/p53^−/− mice despite tumour-specific immune responses, suggesting the development of possible immune suppressor mechanisms during carcinogenesis. CD11b⁺ Gr-1⁺ MDSC have recently been shown to suppress tumour-specific immune responses, so we decided to analyse the kinetics of these cells in this model of spontaneous pancreatic cancer. The frequency of CD11b⁺ Gr-1⁺ MDSC was analysed in mesenteric lymph nodes, peripheral blood, spleen and pancreatic tissue of tumour-bearing EL-TGF-α/p53^−/− mice, and mice with premalignant lesions (EL-TGF-α/p53^+/−). In addition, MDSC were analysed in wild-type mice after subcutaneous challenge with mPAC. In the spleen, the highest frequency of MDSC was detected in tumour-bearing mice (7·9 ± 4·6%). Similar frequencies of MDSC were detected in wild-type mice, mice with premalignant lesions and mPAC-challenged mice (Fig. 2a) as well as 6–8 weeks old EL-TGF-α/p53^−/− mice, which do not have any established tumours (data not shown). Analysis of peripheral blood and mesenteric lymph nodes demonstrated an increase in the frequency of MDSC from mice with spontaneous tumours (13·6 ± 7·34% and 1·01 ± 0·52%) and premalignant lesions (12·0 ± 6·43% and 0·71 ± 0·36%) (Fig. 2b,c). Finally, MDSC were analysed in the pancreas from mice with spontaneous tumours, premalignant lesions and wild-type mice. Again, the highest frequency of MDSC was observed in the pancreas from tumour-bearing mice (12·3 ± 6·8%) in comparison with mice with premalignant lesions in the pancreas (5·2 ± 4·1%) and wild-type mice (2·0 ± 1·7%) (Fig. 2d).

Figure 2.

Frequency of myeloid-derived suppressor cells (MDSC) in elastase-transforming growth factor-α (EL-TGF-α)/p53^−/− tumour-bearing mice, EL-TGF-α/p53^+/−, mice with premalignant lesions, age-matched wild-type mice and mPAC challenged wild-type mice (mPAC). The frequency of MDSC was determined in spleen (a), blood (b), mesenteric lymph nodes (c) and pancreas (d) and is represented as the percentage of CD11b⁺ Gr-1⁺ cells in the monocyte and lymphocyte populations. CD11b⁺ Gr-1^high and CD11b⁺ Gr-1^dull MDSC subtypes were analysed in spleen (e, f) and blood (g, h) of mice with established tumours, premalignant lesions and wild-type mice. (*) indicates a P-value of < 0·05.

Recently, it was shown that MDSC can be further divided into CD11b⁺ Gr-1^high and CD11b⁺ Gr-1^dull populations.^11,12 We analysed the subtypes of the MDSC in spleen and peripheral blood of the different groups of mice. A similar increase in CD11b⁺ Gr-1^high (3·0 versus 1·2%) and CD11b⁺ Gr-1^dull (2·7 versus 1·1%) splenic MDSC was observed in mice with established pancreatic tumours in comparison with wild-type mice (Fig. 2e,f). Analysis of MDSC subtypes in peripheral blood demonstrated an increase in CD11b⁺ Gr-1^high (5·4 and 4·3% versus 2·1%) and CD11b⁺ Gr-1^dull (7·5 and 5·4% versus 2·8%) MDSC for mice with established tumours and premalignant lesions in comparison with wild-type mice (Fig. 2g,h).

Phenotype of the CD11b⁺ Gr-1⁺ MDSC in spleen and pancreas of mice

To further characterize the CD11b⁺ Gr-1⁺ cells in mice with pancreatic tumours, we compared the phenotypes of CD11b⁺ Gr-1^high and CD11b⁺ Gr-1^dull cells from mice with pancreatic tumours, mice with premalignant lesions and mice after challenge with pancreatic tumour cells. Representative dot blots of the expression of different surface molecules are shown in Fig. 3. CD11c expression was higher on CD11b⁺ Gr-1^high MDSC derived from tumours, while Ly6C expression was lower in comparison with both mice with premalignant lesions and subcutaneously challenged mice.

Figure 3.

Phenotypic analysis of Gr-1^high and Gr-1^dull myeloid-derived suppressor cells (MDSC) derived from mice with pancreatic tumours, premalignant lesions and subcutaneous mPAC tumours. CD11b⁺ Gr-1^high and Gr-1^dull MDSC were gated and analysed for the indicated markers. The isotype control is shown for every staining.

CD124, interleukin-4 receptor α-chain (IL-4Rα), has been reported to be expressed on MDSC from tumour-bearing mice.¹³ Mice with spontaneous pancreatic tumours and premalignant lesions had a higher percentage of CD124⁺ CD11b⁺ Gr-1^dull MDSC (36·3% and 21·44%) than mice after subcutaneous challenge with pancreatic tumours (14·7%). More CD11b⁺ Gr-1^dull cells expressed CD124 than CD11b⁺ Gr-1^high cells in mice with subcutaneous tumours, premalignant lesions and established tumours. No significant differences were observed for the expression of F4/80 on CD124⁺ CD11b⁺ Gr-1^dull and CD124⁺ CD11b⁺ Gr-1^high MDSC as well as for the expression of MHC class I and II, CD80 and CD86 (data not shown).

Arginase activity of the MDSC

Arginase activity is a hallmark function of MDSC. The arginase activity of MDSC derived from mice with spontaneous pancreatic tumours and premalignant lesions was tested and compared with those from wild-type mice. As shown in Fig. 4, CD11b⁺ Gr-1⁺ cells derived from mice with pancreatic tumours have more than two-fold higher arginase activity (2·6 mm urea ± 0·2/10⁶ cells) than in naive wild-type mice (1·1 ± 0·05 mm urea/10⁶ cells) or mice with premalignant lesions (1·4 ± 0·001 mm urea/10⁶ cells). Further analysis demonstrated no differences in inducible nitric oxide synthase and arginase-2 expression between the MDSC subtypes from mice with premalignant lesions and pancreatic tumours (data not shown).

Figure 4.

Arginase activity of myeloid-derived suppressor cells (MDSC) from mice with established pancreatic tumours, mice with premalignant pancreatic lesions and wild-type mice. Data shown represent the mean value of four independent experiments. (*) indicates a P-value of < 0·05.

Suppression of T-cell responses by MDSC in pancreatic tumour-bearing mice

We analysed the suppressive function of CD11b⁺ Gr-1⁺ MDSC derived from mice with spontaneous pancreatic tumours in a mixed lymphocyte reaction. Gamma-irradiated treated splenocytes from BALB/c mice were coincubated with allogeneic splenocytes from a C57BL/6 mouse in the presence of increasing numbers of CD11b⁺ Gr-1⁺ MDSC derived from tumour-bearing mice. The IFN-γ concentration and proliferation of splenocytes were analysed in supernatants after 48 hr. A dose-dependent inhibition of proliferation and IFN-γ secretion was observed in the presence of CD11b⁺ Gr-1⁺ MDSC isolated from tumour-bearing mice (Fig. 5a,b). Finally, we compared the immunosuppressive activity of MDSC, derived from mice with spontaneous pancreatic tumours, mice with premalignant lesions or wild-type mice. Wild-type splenocytes from C57BL/6 mice were incubated with gamma-irradiated treated allogeneic splenocytes in the presence of equal numbers of CD11b⁺ Gr-1⁺ MDSC derived from the different mouse groups. The strongest inhibition was observed when MDSC from tumour-bearing mice were used (Fig. 5c,d). As expected, analysis of CD11b⁺ Gr-1^high and CD11b⁺ Gr-1^dull cells revealed a stronger suppression of cytokine release (Fig. 6a) and proliferation of T cells (Fig. 6b) with CD11b⁺ Gr-1^dull cells than with of CD11b⁺ Gr-1^high cells, but no difference was observed between MDSC subtypes from mice with premalignant lesions and pancreatic tumours.

Figure 5.

Analysis of the suppressive function of myeloid-derived suppressor cells (MDSC) derived from mice with established pancreatic tumours: (a, b) Increasing numbers of fluorescence-activated cell sorting (FACS)-analysed MDSC from elastase-transforming growth factor-α (EL-TGF-α)/p53^−/− mice were added to C57BL/6 splenocytes stimulated with irradiated allogeneic splenocytes. Interferon-γ (IFN-γ) secretion and proliferation of T cells was measured after 48 hr in cell supernatants and by thymidine incorporation. Data shown represent one of four independent experiments with similar results. (c, d) FACS-sorted MDSC from mice with premalignant lesions, established tumours and naive mice were added to C57BL/6 splenocytes stimulated with irradiated allogeneic splenocytes. IFN-γ secretion and proliferation of T cells was measured in cell supernatants and by thymidine incorporation after 48 hr. Data shown represent one of four independent experiments with similar results.

Figure 6.

Analysis of the suppressive function of CD11b⁺ Gr-1^high and CD11b⁺ Gr-1^dull myeloid-derived suppressor cells (MDSC) derived from mice with established pancreatic tumours and premalignant lesions: FACS-sorted CD11b⁺ Gr-1^high or CD11b⁺ Gr-1^dull MDSC from mice with premalignant lesions or established tumours were added to C57BL/6 splenocytes stimulated with irradiated allogeneic splenocytes. (a) IFN-γ secretion and (b) proliferation of T cells was measured in cell supernatants and by thymidine incorporation after 48 hr. Data shown represent the average of four mice.

Next, we depleted CD11b⁺ Gr-1⁺ MDSC from splenocytes of tumour-bearing mice and coincubated the MDSC-depleted splenocytes with mitomycin-C-treated allogeneic splenocytes. As expected, a dose-dependent increase in IFN-γ production was observed after incubation of splenocytes derived from tumour-bearing mice with increasing numbers of naive mitomycin-treated splenocytes from BALB/c mice. Depletion of CD11b⁺ Gr-1⁺ MDSC from splenocytes of tumour-bearing animals enhanced the production of IFN-γ (Fig. 7a) as well as proliferation (Fig. 7b).

Figure 7.

Depletion of myeloid-derived suppressor cells (MDSC) from splenocytes isolated from mice with established tumours enhances T-cell responses: (a) C57BL/6 splenocytes were stimulated with increasing numbers of allogeneic splenocytes derived from tumour-bearing mice (white bars) or after depletion (black bars) of MDSC. Interferon-γ (IFN-γ) secretion was determined in supernatants after 48 hr. Data represent one of four independent experiments with similar results. (b) BALB/c splenocytes were stimulated with increasing numbers of mitomycin-C-treated allogeneic splenocytes derived from tumour-bearing mice (white bars) or after depletion (black bars) of MDSC. Proliferation was measured by thymidine incorporation after 3 days. Data represent one of four independent experiments with similar results.

Discussion

Pancreatic cancer is one of the deadliest cancers and has limited treatment options. Immunotherapy represents an alternative and potent therapeutic option for pancreatic cancer. Indeed, promising initial results have been obtained from phase I and II clinical trials testing different immunotherapeutic approaches.² However, only limited information is available on the effect of immunotherapy on pancreatic tumours because of the lack of animal models that could mimic the patients’ disease.¹⁴ Recently, it has become clearer that tumours use a wide array of immunosuppressive strategies by which they not only dampen immune responses but also limit the effect of potent immunotherapeutic approaches.¹⁵ The expansion of MDSC with immunosuppressive activity has been described as an important mechanism of tumour evasion in mice with established tumours.⁸ So far, only limited information is available on MDSC in mice with spontaneously developing tumours and premalignant lesions.

A number of different genetically engineered murine pancreatic tumour models have been established to investigate pancreatic cancer. In addition, new models have been developed to investigate immune-based therapies.¹⁶ In these studies, a model antigen is expressed in pancreatic tumours, which are induced under the rat insulin promotor¹⁴ and therefore do not represent typical exocrine pancreatic tumours. Other immune-based models have been developed, which mimic the human exocrine pancreatic tumours more closely. Depending on the genetic alterations, these exocrine tumours vary in their histological phenotype.⁶ Our mouse model of EL-TGF-α transgenic mice forms cystic acinar neoplasms. Depending on the Trp53 genetic status, these mice develop tumours between 100 and 300 days,⁴ making this an ideal model to analyse immune responses in mice with developing tumours. However, in contrast to mice with endogenous oncogenic KrasG12D, TGF-α transgenic mice do not recapitulate the initiation and progression of PanIN lesions to metastasizing pancreatic ductal adenocarcinoma.⁶

Our mouse model has offered insight about the kinetics of MDSC subtypes and MDSC function during the progression from premalignant lesions to fully established tumours. In this study, we have made a number of important observations. First, an increase of MDSC was already detected in premalignant pancreatic tissue and peripheral blood from mice with premalignant lesions. However there was no difference in the frequency of splenic MDSC in mice with premalignant lesions and wild-type mice. Second, a similar increase in the frequency of both MDSC subpopulations, CD11b⁺ Gr-1^high cells, which correspond to Ly6G⁺ cells,¹¹ and CD11b⁺ Gr-1^dull was observed in peripheral blood from mice with pancreatic tumours.

Our data suggest that the increase in MDSC occurs before malignant tumours are established and proposes further that MDSC contribute to the immune suppressor network during tumour progression as previously seen by others in a mammary carcinoma model¹⁷ as well as in pancreatic cancer progression models.^18,19 Interestingly it was recently shown that treatment of mice with the cyclo-oxygenase-2 inhibitor celecoxib and low-dose gemcitabine was effective in preventing the progression of pre-neoplastic intraepithelial lesions to invasive pancreatic ductal adenocarcinomas. This type of treatment had multiple effects on the immune system including a reduction of the frequency of regulatory T cells and MDSC.²⁰ Based on the results from that study and our results presented here, it will be interesting to analyse whether this treatment will also be effective in patients with pancreatic cancer, who are treated with a cancer vaccine.

In our study, we also demonstrate a substantial increase of both MDSC subtypes as previously demonstrated in mice after subcutaneous challenge with tumour cells.¹² Only CD11b Gr-1^high cells from mice with established tumours have the typical CD11c⁺ Ly6C⁻ phenotype. As previously seen in mice with subcutaneous tumours, Gr-1^dull MDSCs expressed CD124, a marker which has been suggested to be expressed by MDSC that suppress antigen-activated CD8⁺ T-cell responses.¹³ Interestingly the expression of CD124 was highest on CD11b⁺ Gr-1^dull MDSC derived from mice with established tumours suggesting that MDSCs change their phenotype during tumour development and become more suppressive, which was also demonstrated in our study. Therefore, we suggest, that MDSC frequencies increase early during tumour development, but only MDSC from mice with established tumours demonstrate a strong immune suppressive function.

Different factors have been suggested as being responsible for the amplification of MDSC in tumour-bearing animals. We have analysed the presence of IL-6, IL-10, IL-12, vascular endothelial growth factor, IFN-γ, granulocyte–macrophage colony-stimulating factor and tumour necrosis factor-αin supernatants of tumour explants, however we could not detect significant differences in the cell supernatants from healthy pancreas or tumour tissue (data not shown). Recently, S100 A9 protein has been identified as a key factor for the accumulation of MDSC in cancer-bearing animals.²¹ Interestingly, we have previously shown that signal transducer and activator of transcription-3, which regulates the expression of S100 A9,²¹ is activated in the tubular complexes of the pancreas of TGF-α transgenic mice but not in wild-type mice.²² This could explain why an increase in the frequency of MDSC can already be detected very early during pancreatic tumour development in this mouse model and furthermore suggests that early genetic changes during tumour development already have an impact on tumour immunosurveillance.

Disclosures

The authors declare no financial interest.