CD49d Is a New Marker for Distinct Myeloid-Derived Suppressor Cell Subpopulations in Mice

Increasing evidence shows that the host immune response is negatively regulated in cancer and inflammation through various mechanisms, one of which is expansion of myeloid-derived suppressor cells (MDSCs). MDSCs are a heterogeneous population of cells, composed of precursors of macrophages, granulocytes, dendritic cells, and myeloid cells at different stages of differentiation (1). MDSCs, characterized by the coexpression of Gr-1 and CD11b in mice, have been shown to inhibit T cell activation in different tumor models (2). Recently, we have identified a novel immunoregulatory pathway for MDSCs in a CD8⁺ T cell-mediated model of inflammatory bowel disease. A significant increase in the frequency of MDSCs was seen in mice that develop intestinal inflammation upon repeated transfer of CD8⁺T cells specific for an Ag expressed on enterocytes (3). The induced MDSCs suppressed Ag-specific CD8⁺ T cell responses and had arginase activity, a hallmark function of these cells. Despite the growing information about MDSC development and function, many questions remain unresolved. Owing particularly to heterogeneity of these cells, the different subpopulations within these cells need to be analyzed and characterized in more detail.

Several studies have already established the existence of distinct mononuclear and polymorphonuclear subpopulations within tumor-induced CD11b⁺Gr-1⁺ cells, based on the expression of Gr-1 (4, 5). Gr-1–specific Abs bind to two different epitopes—Ly6G and Ly6C. CD11b⁺Ly6G⁺Ly6C^low MDSCs have been described to have a more granulocytic phenotype, whereas CD11b⁺Ly6G⁻Ly6C^high cells display a more monocytic phenotype. Other markers such as CD80 (6), CD115 (7, 8), and CD124 (8) have also been used to identify additional subsets of suppressive MDSCs. However, most of these markers cannot distinguish MDSCs on the basis of their function.

The need for better markers prompted us to perform a gene-expression analysis and compare CD11b⁺Gr-1^high and CD11b⁺Gr-1^dull/int cells to each other. We found CD49d (VLA4), a member of the integrin α-chain family of proteins, to be exclusively expressed on CD11b⁺Gr-1^dull/int cells. CD49d is composed of a α4 (CD49d) and a β1 (CD29) chain and serves as a receptor for fibronectin and VCAM-1 (VCAM-1, CD106). It is normally expressed on monocytes, T cells, and eosinophils and mainly functions as a cell adhesion and signaling molecule (9).

In this study, we have performed a phenotypic and functional analysis of the CD49d⁺ and CD49d⁻ subsets of CD11b⁺Gr-1⁺ cells in two different mouse models (3, 10) to better characterize these subpopulations. We show that in both models, CD49d⁺ identifies a distinct subpopulation of MDSC that can suppress Ag-specific T cells responses through an NO-dependent mechanism and are similar to CD11b⁺Gr-1^dull/int cells in morphology, phenotype, and function. According to our results, CD49d can replace Gr-1 and can be used in combination with CD11b for better definition of different MDSC subpopulations. This newly discovered marker might help to understand the mechanisms of immune suppression by various MDSCs and reveal proper targets for immunotherapeutic approaches.

Female BALB/c and C57BL/6, 6–8 wk of age, were obtained from Charles River Laboratories (Sulzfeld, Germany). VILLIN-HA mice express the A/PR8/34 hemagglutinin (HA) from influenza virus A under control of the enterocyte-specific VILLIN promoter (11). Clone 4 (CL4)-TCR transgenic mice express the α/β-TCR that recognizes an epitope of the HA protein presented by MHC class I (H-2K^d:HA_512–520 complex) (12) and were bred on a BALB/c background.

Chronic intestinal inflammation was developed as previously described (3). A total of 5 × 10⁵ splenocytes from CL4-TCR transgenic mice were injected i.v. into naive VILLIN-HA recipients on day 0. Mice were injected again with 1 × 10⁷ splenocytes on days 12 and 27 for the second and third times, respectively. Mice were euthanized on day 31 for further analysis. For analysis of MDSCs from tumor-bearing mice, BALB/C and C57BL/6 were injected with CT-26 (colon carcinoma), EL-4 (thymoma), and Bl6 (melanoma) tumor cell lines. Spleens were harvested when tumors reached a maximum diameter of 15 mm, which occurred 2–3 wk after injection of 1 × 10⁶ tumor cells. All experiments were performed according to the institutional guidelines.

Flow cytometry was performed on a Becton Dickinson FACS Calibur using CellQuest software (Becton Dickinson, Heidelberg, Germany). Data analysis was performed using FlowJo software (Tree Star, Ashland, OR). Isotype-matched Abs were used with all the samples as controls. MDSCs were purified from spleens of mice using the BD FACSAria cell sorting system (Becton Dickinson). The purity of the MDSCs was shown to be ≥95%.

Sorted splenocytes (2 × 10⁴) were resuspended in 40 μl PBS and centrifuged onto a microscope slide using a Shandon Cytospin 3 (Thermo Scientific, Waltham, MA). Slides were then stained with May-Grünwald-Giemsa according to standard protocol.

Purified CD11b⁺Gr-1^high and CD11b⁺Gr-1^dull/int. MDSCs labeled with anti-CD11b–Alexa flour 488 and anti–CD49d-PE were attached to glass slides and mounted using Gold antifade mounting medium (Molecular Probes, Darmstadt, Germany). Confocal microscopy was performed on an Axiovert 200M inverted microscope using Zeiss LSM 510 Meta scan head (Carl Zeiss, Göttingen, Germany) equipped with a plan-Apochromat 63 × /1.4 NA oil-immersion objective lens; oil was used as the imaging medium.

Different subsets of purified MDSCs (1 × 10⁵ cells) were added to 1 × 10⁵ clone 4 TCR splenocytes (1:1 ratio) or, as indicated in the figures, in the presence of HA peptide (Biosynthan, Berlin, Germany). In some experiments, L-NMMA (Sigma-Aldrich, Munich, Germany) and L-NOHA (Sigma-Aldrich) were used as NO synthase (NOS) and arginase inhibitors, respectively. After 72 h [³H]thymidine (Amersham, Freiburg, Germany) was added to the cultures, and proliferation was measured by [³H]incorporation. Radioactivity was detected using a scintillation counter (Wallac, Turka, Finland). The production of NO was determined from the previous cocultured supernatant using the iT High-Sensitivity Nitrite Assay kit according to the manufacturer's protocol (Molecular Probes, Eugene, OR).

Production of H₂O₂ was quantified using Amplex Red Hydrogen Peroxide/Peroxidase Assay Kit (Invitrogen Darmstadt, Germany) as recommended by manufacturer. Ten × 10³ cells were resuspended in HBSS (Sigma-Aldrich). After addition of PMA (30 ng/ml), the absorbance at 560 nm was measured using a microplate plate reader (Bio-Rad, Hercules, CA). Absorbance results were normalized to a standard curve generated by serial dilutions of 20 mM H₂O₂.

The arginase activity of the MDSCs was determined as described before (13). Cells were lysed with 100 μl lysis buffer (0.1% Triton X+ 1 tablet of protease inhibitor mixture; Roche, Basel, Switzerland). After 30 min at 37°C, 100 μl of 25 mM Tris-HCl and 10 μl of 10 mM MnCl₂ were added. The arginase was activated by heating for 10 min at 56°C. The lysate was incubated with 100 μl of 0.5 M l-arginine (pH 9.7) for 1 h at 37°C. The reaction was stopped by the addition of 800 μl stop solution (96% H₂SO₄:85% H₃PO₄:H₂O ratio, 1:3:7). α-Isonitrosopropiophenone (40 μl) (dissolved in 100% ethanol) was added and heated at 100°C for 1 h. A standard curve consisting of serial dilutions of urea was run in parallel. The urea concentration was measured at 540 nm.

Lysates from purified cells were denatured at 95°C for 5 min and subjected to SDS-PAGE, and the gel was blotted onto nitrocellulose membrane. The membranes were incubated with a rabbit anti-mouse NOS2 Ab (BD Biosciences, Heidelberg, Germany) and rabbit anti-mouse GAPDH Ab (Sigma-Aldrich, Munich, Germany) Binding of the Abs was visualized using anti-rabbit IgG-HRP (Rockland Immunochemicals, PA) followed by ECL detection reagent (Amersham, U.K.).

Data are expressed as mean ± SD. Statistical analysis was performed using Student t test to assess differences between the different study groups. p < 0.05 was considered statistically significant.

In search for a specific marker expressed on monocytic (CD11b⁺Gr-1^dull/int.) and granulocytic (CD11b⁺Gr-1^high) MDSCs, we performed a gene-expression analysis on CD11b⁺Gr-1^high and CD11b⁺Gr-1^dull/int. cells isolated from spleens of VILLIN-HA mice after the third transfer of HA-specific CD8⁺ T cells. Gene-expression analysis revealed a differential expression of CD49d on CD11b⁺Gr-1^dull/int. and CD11b⁺Gr-1^high cells. FACS analysis for CD49d expression on CD11b⁺Gr-1^dull/int. and CD11b⁺Gr-1^high MDSCs derived from VILLIN-HA mice after three transfers of HA-specific CD8⁺ T cells and from tumor-bearing mice (Fig. 1A) confirmed this finding. In addition, CD11b⁺CD49d⁺ cells expressed Gr-1 at low levels, whereas CD11b⁺CD49d⁻ cells highly expressed Gr-1 (Fig. 1B). We further assessed the expression of CD49d on MDSCs isolated from different tumor models and locations (spleen, blood, and tumor). CD49d was expressed primarily on CD11b⁺Gr-1^dull/int. cells in all the tumor models studied (Fig. 1C), and no differences in CD49d expression on MDSCs isolated from spleen or blood were observed. Detailed characterization of MDSC subtypes has been hampered by a lack of markers to detect these cells in tissue sections. We therefore analyzed CD11b and CD49d expression on different MDSC subtypes by confocal microscopy. As shown in Fig. 1D, costaining of CD11b and CD49d was observed on mononuclear but not polymorphonuclear MDSCs.

We next investigated whether there is a difference in the suppressive ability of CD11b⁺Gr-1⁺CD49d⁺ and CD11b⁺Gr-1⁺CD49d⁻ MDSCs; 30.7% of CD11b⁺Gr-1⁺ cells were CD49d⁺, whereas 62.6% of cells did not express this marker (Fig. 2A). CD49d⁺ and CD49d⁻ cells were sorted from the tumor-bearing mice and cocultured with peptide-stimulated CL-4 TCR splenocytes to evaluate their ability to suppress T cell proliferation. CD11b⁺Gr-1⁺CD49d⁺ MDSCs were more potent suppressors of Ag-specific T cell proliferation than CD11b⁺Gr-1⁺CD49d⁻ cells (Fig. 2B). We found similar results with CD11b⁺Gr-1⁺CD49d⁺ MDSCs obtained from three transfer VILLIN-HA mice (data not shown). L-NMMA, a potent NOS inhibitor, reversed inhibition of T cell proliferation by CD11b⁺Gr-1⁺CD49d⁺ cells, but had no effect on CD11b⁺Gr-1⁺CD49d⁻ MDSCs (Fig. 2B). In addition, a significantly higher amount of NO was found in the supernatant of cocultures of CD11b⁺Gr-1⁺CD49d⁺ with CL-4 TCR splenocytes than that of CD11b⁺Gr-1⁺CD49d⁻ cocultures with T cells (Fig. 2C).

It has been suggested that MDSCs can be divided according to their Gr-1 expression into Gr-1^high and Gr-1^dull/int. cells (4). To address the role of CD49d as a possible marker, which can be used alternatively to identify MDSC subtypes, we performed morphologic characterizations via FACS and May-Grünwald-Giemsa staining. Our results showed that CD11b⁺CD49d⁺ cells demonstrated a low SSC profile similar to that in CD11b⁺Gr-1^dull/int. cells. Alternatively, CD11b⁺CD49d⁻ cells have a high SSC profile comparable to CD11b⁺Gr-1^high cells (Fig. 3A). In addition, CD11b⁺CD49d⁺ cells display a mononuclear morphology, and CD11b⁺CD49d⁻ cells have polymorphonuclear characteristics (Fig. 3B). We also analyzed the frequency and function of CD11b⁺CD49d⁺ cells and compared them with CD11b⁺Gr-1^dull/int. MDSCs. As shown in Fig. 3A, the frequency of CD11b⁺Gr-1^dull/int. MDSCs (3.13%) was similar to that of CD11b⁺CD49d⁺ cells (3.91%). Alternatively, the frequency of CD11b⁺CD49d⁻ subset was equal to the frequency of CD11b⁺Gr-1^high cells (Fig. 3A). Analysis of the frequency of CD11b⁺CD49d⁺ and CD11b⁺CD49d⁻ cell populations revealed that the later fraction is preferentially expanding in both tumor-bearing mice and mice with inflammatory bowel disease as compared with naive mice (Fig. 3C).

In an attempt to examine the possible mechanism by which CD11b⁺CD49d⁺ and CD11b⁺CD49d⁻ MDSCs suppress T cell function, we first analyzed NOS2 expression by Western blot analysis, because we had seen earlier that the suppressor function of CD11b⁺Gr-1^dull/int cells can be blocked by the addition of L-NMMA, a potent NOS2 inhibitor. As expected, NOS2 was detected only in CD11b⁺CD49d⁺ cells, but not in CD11b⁺ CD49d⁻ cells (Fig. 4A), and higher amounts of NO were detected in cocultures of CD11b⁺CD49d⁺ cells together with CL4 TCR splenocytes (Fig. 4B). The addition of L-NOHA to CD11b⁺CD49d⁻ cells did not impair their suppressor function, and ARG1 activity was lower in CD11b⁺CD49d⁻ than in CD11b⁺CD49d⁺ cells (Fig. 4C). This might be due to the low expression of arginase 1 by CD11b⁺CD49d⁻ MDSCs compared with CD11b⁺CD49d⁺ MDSCs (data not shown) To further elucidate the mechanism of suppression by CD11b⁺CD49d⁻ and CD11b⁺Gr-1^high MDSCs, reactive oxygen species (ROS) production by the different subtypes was evaluated. CD11b⁺CD49d⁻ MDSCs produced higher levels of ROS when stimulated with PMA, compared with CD11b⁺CD49d⁺ cells (Fig. 4D).

Furthermore, we compared the surface phenotype of CD11b⁺CD49d⁺, CD11b⁺CD49d⁻, CD11b⁺Gr-1^dull/int., and CD11b⁺Gr-1^high cells with each other (Fig. 5A). Our results confirmed that the two respective populations share a similar phenotype. CD11b⁺CD49d⁺ and CD11b⁺Gr1^dull/int. MDSCs express mainly monocytic markers and are Ly6C^high and F4/80⁺. Both CD11b⁺CD49d⁺ and CD11b⁺CD49d⁻ cells showed similar expression patterns for most of the other surface markers studied, including CD124, CD54, CD44, CD11a, CD31, CD62L, and CD16/32. Only a slight difference was observed regarding the expression of CD80, CD45, and MHC class II (Fig. 5A).

To ascertain this finding functionally, CD11b⁺CD49d⁺ and CD11b⁺CD49d⁻ MDSCs were purified from tumor-bearing mice (Fig. 5B). The purified CD11b⁺CD49d⁺, CD11b⁺CD49d⁻, CD11b⁺Gr-1^dull/int., and CD11b⁺Gr-1^high MDSCs were tested in suppression experiments with CL4 TCR Ag-specific splenocytes in the presence of HA peptide. CD11b⁺CD49d⁺ cells suppressed Ag-specific responses in a dose-dependent manner stronger than that in CD11b⁺CD49d⁻ cells (Fig. 5C). Similar to CD11b⁺Gr-1^dull/int. cells, CD11b⁺CD49d⁺ suppressed T cell responses through NO, whereas the NOS inhibitor L-NMMA had no effect on CD11b⁺CD49d⁻ MDSC (Fig. 5D). In addition, we compared the suppressive ability of CD11b⁺CD49d⁺ with that of CD11b⁺Gr-1^dull/int. MDSCs in parallel. We saw that CD11b⁺CD49d⁺ MDSC have a comparable inhibitory activity to CD11b⁺Gr-1^dull/int. MDSCs (Fig. 5E).

It has been shown previously that CD11b⁺Gr-1⁺ cells can be matured to functional dendritic cells (14). Therefore, we asked whether CD11b⁺Gr-1^high cells acquire CD49d upon maturation. For this CD11b⁺Gr-1^dull/int. and CD11b⁺Gr-1^high cells were sorted as described and cultured in the presence of GM-CSF and IL-4 for 3 d. Upon culturing the cells, monocytic CD11b⁺Gr-1^dull/int. MDSCs downregulated CD49d expression, but no changes were seen with CD11b⁺Gr-1^high cells (Fig. 6).

Ly6G⁺Ly6C^low granulocytic and Ly6G⁻Ly6C^high monocytic MDSCs have been described previously to be similar in their suppressive function (5). We examined whether CD49d staining can be used to distinguish Ly6G⁺Ly6C^low and Ly6G⁻ Ly6C^high subsets from each other. Indeed, CD49d is expressed mainly on Ly6G⁻Ly6C^high subset, but not on Ly6G⁺ Ly6C^low cells (Supplemental Fig. 1A). Therefore, we analyzed the suppressive function of CD11b⁺CD49d⁺, CD11b⁺Ly6C⁺, and CD11b⁺Ly6G⁻ cells in parallel. Our results showed that both CD11b⁺CD49d⁺ and CD11b⁺Ly6C⁺ cells were more suppressive than CD11b⁺CD49d⁻ and CD11b⁺Ly6C⁻ MDSC (Supplemental Fig. 1B). Furthermore, in both cases, suppression of T cells was dependent on NO. In contrast, both CD11b⁺CD49d⁻ and CD11b⁺Ly6G⁺ cells were only weak inhibitors of T cell proliferation, and this inhibition was not dependent on NO or arginase.

Numerous studies have shown the existence of counter regulatory mechanisms in inflammatory diseases and in tumors (15, 16). One of the recently identified mechanisms involves the recruitment of a heterogeneous population of myeloid-derived suppressor cells. These cells are widely studied in different mouse and human cancer models (17). We previously reported the accumulation of Gr-1⁺CD11b⁺ cells in the spleen and intestine of VILLIN-HA mice after repeated transfer of Ag-specific cells. In addition, we showed that cotransfer of Gr-1⁺CD11b⁺ together with CL4 TCR CD8⁺ T cells inhibited the occurrence of acute intestinal inflammation in VILLIN-HA mice (3). The heterogeneous nature of the cells and lack of a specific marker that defines potent suppressive cells certainly limits the full understanding of the biology of MDSCs.

Different approaches have been undertaken in the past to better define and characterize the different MDSC subpopulations within the Gr-1⁺CD11b⁺ with different functions. Monocytic and granulocytic MDSCs have been characterized previously based on Gr-1 or Ly6-G expression (4, 5), both recognizing the same epitope on MDSCs. Recently, Greifenberg et al. (18) described five different MDSC subtypes, which can be identified by CD11b and Gr-1 staining. By gene-expression analysis, we were able to identify CD49d (VLA-4) as a new marker for MDSCs. CD49d, an integrin α subunit, is normally expressed on monocytes, T cells, and epithelial cells and mainly functions as a cell adhesion and signaling molecule (19). In this study, we show that CD11b⁺CD49d⁺ represents the MDSC subpopulation with the strongest inhibitory function. CD49d staining can differentiate between monocytic and granulocytic MDSCs and can be used for immunohistologic analysis.

We tested CD49d expression in three different murine tumor models and in a colitis model. Furthermore, splenic, blood-derived and tumor-infiltrating MDSCs were analyzed, and CD49d⁺ and CD49d⁻ MDSCs were tested for the expression of a variety of surface markers, FSC/SSC pattern, May-Grünwald-Giemsa staining, and, most importantly, their function. Our data show that CD11b⁺CD49d⁺ cells represent a mononuclear, monocytic MDSC subpopulation that suppresses T cell function through inducible NOS, whereas CD11b⁺CD49d⁻ cells suppress T cells through ROS production as described previously (5).

VLA-4 (CD49d) has been shown to be constitutively expressed on rat neutrophils (20) and human neutrophils, when activated appropriately (21), and functions mainly in facilitating neutrophil adhesion and migration. Further studies are needed to examine whether CD49d will also support adhesion and migration of polymorphonuclear MDSCs and whether CD49d⁻ MDSCs migrate less.

A number of reports have suggested different markers, which can be used to define suppressive MDSC populations. Gr1⁺CD115⁺ MDSCs and CD80⁺Gr1⁺CD11b⁺ were shown to be more suppressive than Gr1⁺CD11b⁺ cells in colon and ovarian cancer, respectively (6, 7). IL-4Rα (CD124) is another marker that has been described to be expressed on CD11b⁺Gr-1⁺ MDSCs with immune suppressor function in tumor-bearing mice (8). Interestingly, we also demonstrated that IL-4Rα ^−/− (CD124) MDSCs are devoid of suppressive ability. In contrast, Youn et al. (5) found that IL-4Rα⁺ or CD115⁺ MDSCs inhibited Ag-specific responses similar to IL-4Rα⁻ or CD115⁻ MDSCs. However, none of these markers could be used as an alternative to Gr-1 staining. In contrast, our results indicate that it is feasible to use CD49d staining in combination with CD11b to separate CD11b⁺Gr-1^dull/int. from CD11b⁺Gr-1^high cells, which are different in their suppressor function and in the mechanism through which these cells suppress T cell function.

We describe CD49d as a new marker on MDSCs that can be used in combination with CD11b to identify phenotypically and functionally different MDSC subtypes. CD49d can be used instead of Gr-1 as a marker for MDSCs, thereby bypassing potential interferences that occur when Ly6C or Ly6G and Gr-1 are used in combination or to clearly distinguish between Gr-1^high and Gr-1^dull/int. cells. Furthermore, CD49d allows for selective elimination of the monocytic fraction. Finally, our data clearly indicate that CD49d analysis will help differentiate MDSC subpopulations and can be used in future studies to analyze MDSCs in paraffin-embedded tissue sections. Characterizations of MDSC subtypes will help further in understanding their mechanism of action and is essential when these cells will be targeted for immunotherapeutic approaches.

We thank Matthias Ballmeier and the cell sorting facility of Hannover Medical School for technical assistance with cell sorting and cytospin preparations.

Disclosures The authors have no financial conflicts of interest.