Myeloid derived suppressor cells regulate growth of multiple myeloma by inhibiting T cells in bone marrow

Abstract

Myeloid-derived suppressor cells (MDSC) are one of the major factors limiting immune response in cancer. However, their role in bone marrow (BM), the site of primary localization of multiple myeloma (MM), is poorly understood. In this study we found a significant accumulation of CD11b⁺CD14⁻CD33⁺ immune suppressive MDSC in BM of patients with newly diagnosed MM. To assess the possible role of MDSC in MM, we used immune competent mouse models. Immune suppressive MDSC accumulated in BM of mice as early as one week after tumor inoculation. S100A9 knockout (KO) mice, which are deficient in their ability to accumulate MDSC in tumor-bearing hosts, demonstrated reduced MDSC accumulation in BM after injection of MM cells as compared to wild-type mice. Growth of the immunogenic MM cells was significantly reduced in S100A9KO mice. This effect was associated with the accumulation of antigen-specific CD8⁺ T cells in BM and spleens of S100A9KO, but not wild-type mice, and was abrogated by the administration of anti-CD8 antibody or adoptive transfer of MDSC. Thus, the accumulation of MDSC at early stages of MM plays a critical role in the MM progression and suggests that MDSC can be considered as a possible therapeutic target in this disease.

Introduction

Multiple myeloma (MM) is a hematologic cancer characterized by the accumulation of malignant plasma cells within the bone marrow (BM). BM microenvironment is known to be critically important for MM survival, growth, and chemosensitivity. While the majority of studies so far have focused on the contribution of BM stroma and osteoclasts in MM pathogenesis(1–4), less attention has been paid to the role of other cells that constitute the BM microenvironment, including cells involved in modulation of immune responses.

An impaired function of the immune system plays an important role in tumor growth(5). In MM, abnormalities in T cells, including a decreased number of peripheral blood (PB) CD4 and CD8 T cells, inversion of the CD4:CD8 ratio, abnormal Th1/Th2 CD4 ratio, down-regulation of signal transduction components, and abnormal T cell response have been reported(6, 7). In addition, cellular immune defects including abnormalities in macrophages, natural killer (NK) and dendritic cells have been described in MM. Despite the fact that MM localized preferentially in the BM, the majority of studies has been focusing on immunological alterations in PB of MM patients. At the same time very little is known regarding the function of the immune system in MM BM and particularly the ability of the immune system to generate an anti-MM immune response in the BM tumor microenvironment (8, 9).

In recent years, the important role of myeloid-derived suppressor cells (MDSC) in the regulation of immune responses in cancer has been established. This heterogeneous group of myeloid cells is comprised of pathologically activated myeloid progenitors and immature myeloid cells (IMC), with a potent immune suppressive activity (10). Under physiological conditions, IMC rapidly differentiate into mature myeloid cells. The accumulation of MDSC in cancer is the result of two sets of factors: one that promotes expansion of IMC and another that induces activation of these cells associated with a partial block in their differentiation. The first group of factors includes GM-CSF, M-CSF, VEGF and others, and signals primarily via STAT3 and STAT5 transcription factors; whereas, the second group consists of pro-inflammatory cytokines and signals via STAT1, S100A8/A9, and NF-kB (11–13). MDSC inhibit function of immune cells via a number of mechanisms involving nitric oxide (NO), arginase, reactive oxygen species (ROS), Cox-2 and others(12, 14). In mice, MDSC are characterized by the co-expression of Gr1 and CD11b molecules(15). In recent years two large groups of mouse MDSC were identified: CD11b⁺Ly6C^loLy6G⁺ polymorphonuclear MDSC (PMN-MDSC) and CD11b⁺Ly6C^hiLy6G⁻ monocytic MDSC (M-MDSC). These cells, although sharing immune suppressive activity, are distinct in their morphology and mechanisms of suppression(16, 17). While PMN-MDSC utilize ROS to mediate T cell suppression, M-MDSC have increased level of NO but undetectable level of ROS (18). In humans, the phenotype of MDSC depends on the type of tumor. In most tumors, immune suppressive MDSC are defined as CD11b⁺CD14⁻CD33⁺ or Lin⁻HLA-DR⁻CD33⁺ cells that can be further sub-divided into CD15⁺ PMN-MDSC and CD15⁻ M-MDSC. In some tumors, M-MDSC have been also defined as CD14⁺HLA-DR⁻/^lo (19, 20).

There is a wealth of information regarding the possible role of MDSC in the regulation of immune responses in solid tumors. However, little is known about the biology of MDSC and their possible immunosuppressive activity in hematologic malignancies, including MM. Although the presence of MDSC in PB of patients with MM was described(21), the possible role of immature myeloid cells in BM, the site most relevant for MM, has not been studied.

Present study is focused on defining the function of MDSC in the MM BM microenvironment. We demonstrated that immune suppressive MDSC are accumulated in the BM of patients with MM as well as in the BM of MM-bearing immunocompetent mice. To further understand the contribution of MDSC in regulation of the immune responses generated in BM of MM-bearing mice, we utilized an approach where T cell responses to the model antigen were monitored in vitro. Using a mouse model of MM and knockout mice with defective MDSC accumulation, we showed that MDSC could be directly responsible for the tumor-specific immune suppression observed in BM in this disease.

Materials and Methods

Isolation of human MDSC and functional assays

BM and PB samples were collected from newly diagnosed, non-treated patients with MM. The collection of samples was approved by the University of South Florida IRB. BM and PB samples from healthy donors were purchased from Lonza (Allendale, NJ) and Florida Blood Services (St.Petersburg, FL), respectively. Mononuclear cells were isolated by Ficoll-Paque density gradient centrifugation. CD11b⁺CD14⁻CD33⁺ populations of MDSC or IMC were isolated by flow sorting using FASCAria instrument (BD). Human T cells were purified from PB mononuclear cells (PBMC) obtained from healthy donors, using T cell enrichment columns (R&D systems). Dendritic cells were generated in vitro from PBMC obtained from a different donor. T cells (1×10⁵) were stimulated with LPS-matured dendritic cells (1.5×10⁴), with or without MDSC present in the co-culture. The number of IFN-γ-producing T cells was evaluated in an ELISPOT assay on automatic counter (Cellular Technology). Proliferation of T cells was measured by ³H-Thymidine incorporation.

Mice and cell lines

C57BL/6 and FVB/N mice were purchased from the National Cancer Institute (Frederick, MD) and were crossed to obtain mice of a mixed FVB/N×C57BL/6 background. F1 progeny (6–8 weeks old) were used. Mice were kept in pathogen-free conditions and handled in accordance to the requirements of the Guideline for Animal Experiments. S100A9 KO mice, on C57BL/6 background, have been previously described(23). S100A9KO mice, on a FVB/N background, were obtained by the backcrossing of C57BL/6 S100A9KO mice to FVB/N mice for 10 generations. F1 progeny of S100A9KO FVB/N×C57BL/6 mice (6–8 weeks old) were used for experiments.

MM cells BCM47BM (BCM), 38ATLN (ATLN), and DP42 were kindly provided by Dr. Van Ness (University of Minnesota) and described previously(24). DP42-OVA cells were established by transfection of DP42 cells with a pAc-Neo-OVA1 vector, expressing chicken ovalbumin protein (OVA), followed by the selection with 1.4 mg/mL G418 to create stably transfected clones.

MM cells were cultured in vitro in RPMI-1640 medium (BioSource International), supplemented with 10% fetal bovine serum, 5 mM glutamine, 50µM 2-mercaptoethanol, 1% antibiotics (all from Invitrogen) and 0.5 ng/mL recombinant mouse IL-6 (R&D Systems). MM tumors were established by i.v. inoculation of syngeneic MM cells into mice tail veins (10⁵ ATLN, 10³ BCM, 3 × 10³ DP42, and 10⁵ DP42-OVA cells). In survival studies mice were sacrificed when determined to be moribund according to IACUC criteria.

For CD8⁺ T cell depletion experiments mice were treated i.p. with 200µg of CD8 monoclonal antibody (Ab, Clone: 53-6.72, BioXCell, West Lebanon, NH) or control IgG2b (clone LTF-2, BioXCell) every 4 days, beginning 4 days after tumor cell injection. The mice received a total of 6 injections of anti-CD8 Ab or control IgG.

Mouse cell isolation and functional assays

BM cells were obtained by flushing mice femurs and tibias with ice cold serum-free RPMI-1640 medium. Single cell suspensions were prepared from spleens. Red blood cells were lysed using ACK buffer. Gr1⁺CD11b⁺ mouse MDSCs, or IMCs, were isolated by FACS-sorting on a FACSAria instrument (BD). The purity of the cell populations was 99%. As responder cells, splenocytes from Pmel-1 transgenic mice were used. The number of IFN-γ-producing cells, in response to stimulation with 0.1 µg/mL specific (SIINFEKL) or control (EGSRNQDWL) peptides was determined in ELISPOT assay performed as described earlier(18). The number of spots was counted in triplicate and calculated using an automatic ELISPOT counter (Cellular Technology). In parallel, T cell proliferation was evaluated using ³H-Thymidine incorporation, as previously described(17).

Adoptive transfer of MDSC

Gr1⁺CD11b⁺ MDSC were isolated from BM of DP42-bearing mice by flow sorting 10 days after tumor cell inoculation. MDSC (5×10⁶) were injected i.v. into tail vein of DP42-OVA bearing mice every 4 days, beginning on day 4 after DP42-OVA tumor cell injection. The mice received a total of 4 injection of MDSC.

Flow cytometry

Cells were labeled with specified Abs (all from BD Biosciences) for 30 min in the dark. Cells were washed twice with PBS, resuspended in PBS containing DAPI (Invitrogen) to exclude dead cells, and analyzed by flow cytometry. For detection of mouse antigen-specific T cells, K^b-SIINFEKL pentamers (ProImmune) were used. At least ten thousand DAPI-negative events were acquired using a LSRII flow cytometer (BD). Data was analyzed using FlowJo software (Tree Star).

MTT assay

Cells were cultured in 96-well plates with or without S100 proteins for 48 hrs. 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) dye was added for the last four hours of incubation. Insoluble formazan complexes were solubilized with DMSO and absorbance was measured at 540 nm using a Benchmark Plus microplate spectrophotometer (Bio-Rad, Hercules, CA).

Histochemistry

Femur bones were collected from MM-bearing and tumor-free mice, fixed, decalcified, and paraffin embedded. Slides were prepared and standard H&E staining performed.

Statistics

Statistical analysis was performed using GraphPad Prism 5 software (GraphPad Software Inc). Differences between the groups were calculated using a 2-tailed unpaired Student’s t-test. A statistically significant difference was determined at p < 0.05. A Log rank test was employed to evaluate the statistical significance in mouse survival experiments.

Results

MDSC in the bone marrow and peripheral blood of MM patients

The presence of different populations of MDSC was evaluated in BM and PB of MM patients: MDSC - CD11b⁺CD14⁻CD33⁺; PMN-MDSC - CD11b⁺CD14⁻CD33⁺CD15⁺; and M-MDSC - CD11b⁺CD14⁻CD33⁺CD15⁻ or CD11b⁺CD14⁺HLA-DR⁻/^low. Cells with the same phenotypes in control donors were called IMC. Since MM cells constitute a significant proportion of MM BM, the frequency of MDSC was evaluated as the proportion of these cells among the CD138⁻ non-myeloma cells of BM. Our data demonstrated a significant (p<0.05) accumulation of MDSC in BM of MM patients, as compared to healthy donors (Fig. 1A). MDSC represented 41.1±5.3% (range from 13.3% to 75.9%) of BM cells in MM, while IMC with the same phenotype only 22.9±2.8% (range from 7.7% to 33.3%). A significant accumulation of MDSC was also observed in PB of MM patients (Fig. 1A). PMN-MDSC represented the majority of MDSC in BM and was significantly increased in BM and PB of MM patients, as compared to healthy donors (Fig. 1B). In contrast, no differences in the proportions of CD11b⁺CD14⁻CD33⁺CD15⁻ M-MDSC were found (Fig. 1C). This was consistent with the lack of differences in the presence of CD11b⁺CD14⁺HLA-DR⁻/^low cells, which are considered to be a distinct population of M-MDSC (25, 26) (Fig. 1D). Similarly, no differences were observed in the presence of CD11b⁺CD14⁺HLA-DR^hi monocytes (Fig. 1E). No correlation was found between the proportion of MDSC and the extent of the disease evaluated by the proportion of MM cells in BM aspirates, core biopsies and concentration of heavy and light chain immunoglobulin in serum (data not shown).

Figure 1. Accumulation of MDSC in patients with MM.

BM and PB samples were obtained from patients with MM (n=15 and n=11, respectively) or healthy donors (n=6 and n=10, respectively). Mononuclear cells were labeled with specified antibodies and analyzed by flow cytometry (A–E). Proportion of the following cell populations was evaluated: (A) CD11b⁺CD14⁻CD33⁺ (B) CD11b⁺CD14⁻CD33⁺CD15⁺ PMN-MDSC, (C) CD11b⁺CD14⁻CD33⁺CD15⁻ M-MDSC, (D) CD11b⁺CD14⁺HLA-DR^-/lo M-MDSC, and (E) CD11b⁺CD14⁺HLA-DR^hi monocytes. Presented data show proportion of these cells calculated among CD138⁻ non-myeloma cell. * - statistical significance between groups (p<0.05).

To verify whether the population of cells with the phenotype of MDSC that accumulates in BM of MM patients, could be functionally defined as MDSC, we determined their immune suppressive activity. MDSC were sorted from BM of MM patients and their ability to inhibit T-cell response was tested in allogeneic mixed leukocyte reaction by IFN-γ ELISPOT assay and ³H-Thymidine incorporation (27). IMC isolated from BM of healthy donors were used as control. A significant inhibition of T-cell activity by MDSC was observed (Fig. 2A,C); whereas IMC lacked suppressive activity (Fig. 2B,D). These data indicate that MDSC with immune suppressive activity accumulated in BM of MM patients. Although proportion of myeloid cells with phenotype CD11b⁺CD14⁺HLA-DR⁻/^low has not been increased in BM of MM patients, these cells were able to suppress T cell responses while cells with similar phenotype isolated from BM of healthy donors lacked this ability (Fig. 2E,F).

Figure 2. Suppressive activity of MDSC isolated from patients with MM.

CD11b⁺CD14⁻CD33⁺ (A–D) or CD11b⁺CD14⁺HLA-DR^−/lo (E–F) cells were isolated by sorting of BM from MM patients (A,C,E) or healthy donors (B,D,F). Sorted cells were added at 1:1 ratio to purified T cells stimulated by allogeneic dendritic cells. (A,B) The secretion of IFN-γ was measured by ELISPOT after 48 hours of incubation. (C–F) T cell proliferation was measured by ³H-Thymidine incorporation. Each condition was set up in triplicate. Statistically significant differences between samples with and without MDSC: * - p<0.05; *** - p<0.005.

Syngeneic murine model of MM and MDSC accumulation

To better understand the role of MDSC in MM pathogenesis, we utilized syngeneic murine models of MM originally developed and characterized previously (24, 28). In these models, MM cells, injected i.v. into syngeneic mice, are homing to the BM and the growth of MM tumor closely resembles human disease (28). Since MM develops in BM of immunocompetent mice, these models allow investigation of MDSC in vivo in BM milieu. Three MM models were studied: BCM, DP42 and ATLN. Intravenous administration of these cells resulted in the development of MM in BM within 1–2 weeks (Fig. 3A). Three tumor models demonstrated different kinetics of tumor growth. BCM was the fastest growing tumor, with a median survival of 18 days. DP42 tumor grew slower, with a median survival of 21 days (range 20–24 days). All mice injected with BCM or DP42 cells developed tumors. ATLN was the slowest growing tumor in mice, with a median survival of 25 days (range 17–33 days). By day 40 after ATLN inoculation only about 80% of the mice developed this tumor (Fig. 3B). Mice were euthanized at different time points after tumor cell inoculation and the proportion of MM cells was evaluated in the BM and spleens. Since phenotypically MM cells are characterized by surface expression of syndecan (CD138), we utilized this marker to distinguish MM (Fig. 3C). In DP42 and ATLN models MM cells were detected in BM within a week after tumor cell inoculation and substituted the majority of hematopoietic cells in 2.5–3 weeks (Fig. 3D). In spleens MM was detected later, three weeks after tumor inoculation (Fig. 3E). In BCM model MM cells accumulated in both BM and spleens within seven days (Fig. 3D,E).

Figure 3. Mouse models of MM.

MM tumors were established by i.v. inoculation of BCM, DP42, or ATLN MM cells into syngeneic mice. (A) H&E staining of BM from control tumor-free or DP42 MM-bearing mice. Bars = 100 µm. (B) Survival of BCM (n=12), ATLN (n=13), and DP42 (n=9) mice. (C–E) Tumor-free (naïve) or MM-bearing mice were euthanized at indicated time points after tumor inoculation (3–4 mice per each time point). BM and spleens were collected; single cell suspensions were prepared and labeled with anti-CD138 antibody. Proportion of CD138⁺SSC^hi MM cells was determined by flow cytometry. (C) A representative flow cytometry analysis of BM from control tumor-free or BCM MM-bearing (day 7 after tumor inoculation) mice. Proportions of MM cells in (D) BM and (E) spleens of MM-bearing mice at different time points during disease progression. *- statistically significant differences from naïve mice (p<0.05).

We evaluated the presence and function of Gr-1⁺CD11b⁺ MDSC population in slower growing MM models: ATLN and DP-42. In ATLN model, a significant increase in the proportion and absolute number of MDSC in BM was observed as early as one week after tumor cell inoculation (Fig. 4A). From week two, the presence of MDSC in BM was gradually decreased; and, three weeks after tumor cell inoculation the proportion of MDSC in BM was significantly reduced as MM cells expanded and substituted all hematopoietic cells in BM (Fig. 4A). As in BM, the proportion and absolute number of MDSC in spleens was significantly increased one week after tumor cell inoculation. However, this cell population continued to grow during week two post tumor injection, reflecting the fact that MM cells accumulated in spleen at later time points and at less extent than in BM. Only at the end of week three, when mice became moribund and MM cells expanded, the presence of MDSC in spleens declined (Fig. 4B). The proportion and absolute number of MDSC in lymph nodes dramatically expanded by week three (data not shown). The kinetic of populations of PMN-MDSC and M-MDSC in BM was similar to that of total MDSC, with a significant increase one week post-inoculation and a gradual decrease during tumor progression (Fig. 4C). Interestingly, M-MDSC were responsible for the increase in total MDSC in spleens one week after tumor cell injection; whereas number of PMN-MDSC dramatically raised during week two post tumor inoculation (Fig. 4D). The DP42 model showed a similar pattern of MDSC accumulation (Fig. S1A,B).

Figure 4. MDSC in MM-bearing mice.

MM tumors were established by i.v. tail vein injection of 10⁵ ATLN cells into syngeneic mice. Mice were euthanized at indicated time points after tumor cell inoculation (3–4 mice per time point). As a control (naïve), tumor-free mice were used. Proportion (left panels) and absolute number (right panels) of Gr1⁺CD11b⁺ cells in (A) BM and (B) spleen, absolute number of CD11b⁺Ly6G⁻Ly6C^hi M-MDSC and CD11b⁺Ly6G^hiLy6C^lo PMN-MDSC in (C) BM and (D) spleen. (E,F) Gr1⁺CD11b⁺ cells isolated from BM of ATLN-bearing or naïve mice were cultured with splenocytes from Pmel-1 transgenic mice in the presence of control or specific peptides. Each condition was set up in triplicates. (E) MDSCs or IMCs were mixed with splenocytes in 1:1 ratio. INF-γ production by T cells was measured in ELISPOT assay. Shown are the numbers of spots calculated by subtracting background values (cells stimulated with control peptides) from specific values (cells stimulated with specific peptide). Three mice per group were used. *** - statistically significant difference between IMC and MDSC (p=0.000073). (F) Proliferation assay. Results are mean values obtained for 3 mice with each condition set up in triplicates. Proliferation was measured by ³H-Thymidine incorporation. ** - statistically significant difference (p<0.001).

The hallmark of MDSC is the ability of this cell population to suppress immune responses. Although immunosuppressive properties of MDSC isolated from lymphoid organs or blood of mice with solid tumors have been demonstrated in many studies, the function of MDSC in BM has been poorly investigated. In order to address this question we evaluated the ability of BM MDSC to inhibit antigen-specific T cell response. Gr-1⁺CD11b⁺ MDSC were sorted from BM of ATLN-bearing mice two weeks after the injection of tumor cells and Gr-1⁺CD11b⁺ IMC were sorted from BM of tumor-free mice. MDSC and IMC were added to Pmel-1 transgenic splenocytes (responders) in the presence of specific or control peptides. IFN-γ production and T-cell proliferation were evaluated. MDSC demonstrated potent suppressive activity; whereas IMC did not inhibit T-cell response (Fig. 4E,F). Taken together, our data demonstrated the rapid accumulation of immune suppressive MDSC in BM of MM-bearing mice, which was followed by MDSC accumulation in the spleens and lymph nodes.

MDSC promote MM growth in vivo

To address a possible role of MDSC in MM, we used S100A9KO mice (29). Under physiological conditions, these mice demonstrated normal myeloid cell differentiation. However, when challenged with tumor cells, S100A9KO mice showed a reduced accumulation of MDSC as compared to their wild-type (WT) counterparts(30). S100A9 was directly implicated in promoting MDSC accumulation(31).

Consistent with previously reported data, no differences were found in the presence of IMC in the spleens (Fig. 5A) and lymph nodes (data not shown) between WT and S100A9KO tumor-free mice. However, in BM of S100A9KO mice, the presence of IMC was significantly reduced (Fig. 5A). To evaluate the effect of S100A9 deficiency on MM growth and mice survival from MM, BCM and DP42 tumors were established in WT or S100A9KO mice. No difference in BM tumor burden was observed between WT and S100A9KO mice during MM progression (Fig. 5B). Although S100A9KO BCM-bearing mice demonstrated a slight survival advantage as compared to WT mice, the difference was not statistically significant (Fig. 5C). In DP42-bearing mice, no evidence of an improved survival was observed (Fig. 5D). These data raised the question as to what role if any MDSC could play in MM progression in BM. One of the major effects of MDSC is the inhibition of antigen-specific immune responses. Rapid growth of MM in BCM and DP42 models, 100% tumor take, and the lack of spontaneous rejection suggested that these tumor cells are likely to be poorly immunogeneic. To investigate the effect of MDSC in the model of a more immunogeneic tumor, we generated a DP42 cell line with a stable overexpression of OVA – DP42-OVA (Fig. S2A). DP42-OVA cells had similar kinetic of growth in vitro as compared to parental DP42 cells (Fig. S2B). However, S100A9KO mice, inoculated with DP42-OVA cells, demonstrated a significantly improved survival (p=0.009) (Fig. 6A) and a delayed tumor growth in BM (Fig. 6B), as compared to WT mice.

Figure 5. MM tumor growth in S100A9KO mice.

(A) BM and spleens were obtained from tumor-free S100A9KO or WT mice. Proportions of Gr1⁺CD11b⁺ IMCs were determined by flow cytometry and absolute numbers of IMCs were calculated. (B) DP42-bearing S100A9KO or WT mice were euthanized at indicated time points (3–4 mice per time point) and tumor burden in BM was determined by measuring the proportion of CD138⁺SSC^hi MM cells. (C–D) Survival of tumor-bearing S100A9KO and WT mice from BCM (n=12 for each group) (C) and DP42 (n=9 for each group) (D).

Figure 6. Growth of immunogenic MM tumors in S100A9KO mice.

(A–D) DP42-OVA MM tumors were established in S100A9KO or WT mice. (A) Survival of S100A9KO (n=6) and WT (n=8) mice was evaluated. Log rank analysis was used to determine statistical significance between groups (p=0.009). (B,C) S100A9KO or WT DP42-OVA-bearing mice were euthanized at indicated times (3 mice per time point) and tumor burden in BM (B) and spleen (C) was determined by measuring the proportion of CD138⁺SSC^hi MM cells. *** - indicates statistically significant difference with p<0.001. (D) Absolute number of Gr1⁺CD11b⁺ cells was determined in DP42-OVA S100A9KO and WT mice one week after MM cell inoculation. Three mice per group were analyzed. (E) Sorted BM Gr1⁺CD11b⁺ MDSCs from DP42-OVA-bearing S100A9KO or WT mice one week after tumor cell inoculation were mixed with Pmel-1 splenocytes at indicated ratio in the presence of specific or control peptides. Proliferation was measured using ³H-Thymidine incorporation. Statistically significant difference between no MDSC and WT MDSC groups is shown.

We compared the presence of Gr-1⁺CD11b⁺ MDSC in BM of WT and S100A9 mice one week after MM inoculation. At this point, the number of tumor cells in BM (Fig. 6B) and spleens (Fig. 6C) of both groups of mice was the same. MDSC in BM of DP42-OVA-bearing WT mice was increased, as compared to tumor-free mice (Fig. 6D compared with Fig. 5A). The presence of MDSC in BM of DP42-OVA-bearing S100A9KO mice was significantly (p=0.04) lower (Fig. 6D). MDSC isolated from BM of DP42-OVA-bearing WT mice one week after tumor inoculation suppressed antigen-specific T-cell response; whereas MDSC from S100A9KO MM-bearing mice lacked this ability (Fig. 6E).

We asked whether S100A9 protein in the form of homodimer or heterodimer with S100A8 protein could directly affect MM growth or the immune suppressive activity of MDSC. Recombinant S100A9, S100A8/A9, as well as S100A8 protein did not increase the survival and proliferation of MM cells (Fig. S3). There was no S100A9 mediated conversion of IMC from naïve tumor-free mice into immune suppressive MDSC observed (data not shown). S100 proteins also did not have direct effect on T cell function (Fig. S4). These results suggest that the loss of immune suppressive activity in S100A9KO mice was likely the result of a decreased proportion of immune suppressive MDSC among BM Gr-1⁺CD11b⁺ cells.

To investigate the involvement of the immune mechanisms in decreased tumor growth of MM in S100A9KO mice, we treated tumor-bearing mice with CD8 Ab. Administration of this Ab completely abrogated the improved survival of S100A9KO mice inoculated with DP42-OVA MM cells (Fig. 7A).

Figure 7. Tumor-specific T cell responses in MM.

DP42-OVA tumors were established in S100A9KO or WT mice. (A) Four days after tumor cell inoculation each group was split into two and treatment with anti-CD8 Ab or control IgG was initiated. Mice survival was evaluated. (B–D) DP42-OVA bearing mice were euthanized at indicated time points. BM and spleens were collected and labeled with anti-CD8 antibody and SIINFEKL-H2-K^b pentamer conjugated with FITC and analyzed by flow cytometry. (B) Typical result of one experiment. (C) Proportion of ova-specific CD8⁺ T cells in BM and spleens (D) * - statistical significance p<0.05; ** - statistical significance p<0.01 between S100A9KO and WT group of mice. BM (E) and spleens (F) were collected from DP42-OVA bearing S100A9KO and WT mice. Proportion of IFN-γ⁺CD4⁺ T cells was determined by flow cytometry. Three mice per group were analyzed. All experiments were repeated twice. * - indicate significant difference p<0.05. (G) MDSC were isolated from DP42-bearing WT mice and 5×10⁶ cells were injected i.v. into DP42-OVA-bearing S100A9KO mice on days 4, 8, 12 and 16 after tumor cell inoculation. Survival of DP42-OVA bearing S100A9KO mice with (n=5) or without (n=12) transferred MDSC was determined.

To investigate whether the decreased presence of MDSC in MM-bearing S100A9KO mice would result in an improved tumor-specific immune response, we measured the presence of OVA-specific CD8⁺ T cells in DP42-OVA-bearing mice, using SIINFEIKL-H2K^b pentamers (Fig. 7B). After one week S100A9KO mice had a higher presence of pentamer-positive CD8⁺ T cells in BM, however, the difference was not statistically significant. By week three, however, the presence of pentamer-positive CD8⁺ T cells was significantly higher in BM of DP42-OVA-bearing S100A9KO than in WT mice (Fig. 7C). Similar kinetic was observed in spleens of these mice (Fig. 7D). However, at all time points the proportion of pentamer-positive CD8⁺ T cells in spleens was more than 10-fold lower than in BM (Fig. 7C,D) indicating that BM was the primary site of accumulation of antigen-specific CD8⁺ T cells in S100A9KO MM-bearing mice. We also evaluated the presence of IFN-γ⁺ (Th1-type) CD4⁺ T cells in MM-bearing mice. A slight increase in proportion of these cells was observed one week after tumor inoculation in BM of S100A9KO mice. A week later the differences became significant (Fig. 7E). There was no increase in the proportion of IFN-γ positive cells in spleens (Fig. 7F).

To confirm the contribution of MDSC in MM tumor growth an adoptive transfer of Gr1⁺CD11b⁺ cells isolated from BM of WT DP42-bearing mice was performed into recipient S100A9KO DP42-OVA bearing mice. Transfer of WT MDSC into MM-bearing S100A9KO mice resulted in significantly decreased survival of these mice (Fig. 7G).

Discussion

In this study we report the direct role of MDSC in the regulation of BM antitumor immune responses and tumor progression in MM. MDSC are known to inhibit immune responses in solid tumors (32–34). Although the presence of MDSC in PB of MM patients has been demonstrated (21, 22), the involvement of these cells in the pathogenesis of hematological malignancies, and specifically MM, was not clear. We found that one of MDSC populations - CD11b⁺CD14⁻CD33⁺ cells, significantly increased in the BM of patients with MM. These cells demonstrated T-cell suppressive activity; whereas their control counterpart in healthy donors did not. Our data are consistent with a recent report showing the accumulation of MDSC in BM of patients with leukemia and lymphoma(26). In that study, most suppressive MDSC had phenotype Lin⁻CD11b^lo/− cells. Several populations of MDSC in humans were described (35). Although they share markers of myeloid cell lineage, they may differ in the extent by which some of the markers are expressed. It has now become evident that the specific phenotype of MDSC depends on the type of tumor. These populations most likely include cells with similar biology and function(25, 27). Our study revealed that in MM patients the major changes were observed in the population of PMN-MDSC. PMN-MSDC are the most abundant population of MDSC in PB of patients with many solid tumors(35). However, there is a report showing that M-MDSC more frequently accumulate in PB of patients with leukemia(36). Brimnes et al suggested that the CD14⁺HLA-DR^−/lo M-MDSC was increased in PB of MM patients, as compared to healthy donors. However, the variation between patients samples was high and the significance was not clear(37). Our data demonstrated no difference in the presence of CD14⁺HLA-DR⁻/^lo cells in BM and PB of MM patients as compared to healthy donors. However, the immune suppressive activity of these cells was similar to that observed in CD11b⁺CD14⁻CD33⁺ population. More studies are needed to identify the true nature of MDSC in MM patients.

In this study, we attempted to understand the possible role of MDSC in the regulation of antitumor immune responses in MM. This question could not be addressed only by using patients samples and, therefore, we employed MM models that were relatively recently established in double transgenic c-myc/Bcl-xL mice^18,19. These mice spontaneously develop MM tumors and several MM cell lines have been derived from those tumors. After i.v. injection in syngeneic mice, tumor cells home to the BM and form MM. In these mice, the accumulation of immune suppressive MDSC in BM was seen as early as one week after tumor inoculation. During disease progression, MM cells expanded dramatically and replaced other hematopoietic BM cells including MDSC, which may reflect the situation in patients with advanced disease. Such rapid accumulation of MDSC in BM is different from our observations in solid tumor mouse models, where accumulation of MDSC in BM takes place much later during tumor progression (data not shown). These observations may reflect the fact that BM is a primary tumor site in MM and therefore, the expansion of myeloid cells and their conversion into MDSC is directly affected by growing tumor.

In this study, we did not address the question of the specific mechanism responsible for MDSC-mediated immune suppression, since it has been extensively studied in many tumor models (12, 14, 22). The question we addressed was whether a rapid, but transient accumulation of MDSC during the initial phase of MM development was important for disease progression. We used mice with a defective ability to mount MDSC response to tumors – S100A9KO mice(30). These mice had a decreased presence of Gr-1⁺CD11b⁺ IMC in BM and did not respond to injections of MM cells with an accumulation of MDSC in BM. However, this did not translate into improved survival. We hypothesized that it could be the result of an inability of experimental MM cell line to generate a spontaneous immune response due to poor immunogenicity. To address this question, we overexpressed a model antigen (OVA) in a MM cell line. Growth of OVA-expressing MM tumors was significantly delayed in S100A9KO mice as compared to WT mice. Adoptive transfer of MDSC isolated from BM of WT MM-bearing mice into S100A9KO MM-bearing mice resulted in significantly reduced survival of these mice confirming the critical role played by MDSC in MM progression.

Our data indicated that in the absence of MDSC spontaneous expansion of antigen-specific T cells could be detected in BM as early as one week after tumor inoculation and increased further during the following two weeks. In contrast, the presence of these cells in spleen was barely detectable and was more than 10-fold lower than in BM. The delay of tumor progression was dependent on the presence of CD8⁺ T cells as administration of anti-CD8 antibody completely abrogated the improved survival of MM-bearing S100A9KO mice. Thus, our data pointed out that MDSC, present in BM tumor site of MM, can block the activity of antitumor cytotoxic CD8⁺ T cells in an antigen-specific manner and also decrease the presence of Th1 CD4⁺ T cells. Taken together, these data provide insight into the mechanism of immune defects in MM. Our results support the notion that immune therapeutic strategies aimed to improve antigen-specific T cell responses in MM need to be accompanied by depletion of MDSC.

Abbreviations used in this article

MDSC

myeloid-derived suppressor cells

IMC

immature myeloid cells

PMN-MDSC

polymorphonuclear MDSC

M-MDSC

monocytic MDSC

BM

bone marrow

MM

multiple myeloma

KO

knock out

WT

wild-type

OVA

ovalbumin

NK

natural killer

NO

nitric oxide

ROS

reactive oxygen species.