Purpose: Myeloid-derived suppressor cells (MDSC) are a heterogeneous group of pathologically expanded myeloid cells with immunosuppressive activity. In human disease, three major MDSC subpopulations can be defined as monocytic (M-MDSC), granulocytic [polymorphonuclear-MDSC (PMN-MDSC)], and early stage (e-MDSC), which lacks myeloid lineage markers of the former two subsets. The purpose of this study was to determine and compare the immunosuppressive capacity and clinical relevance of each of these subsets in patients with solid cancer.
Experimental Design: The frequency of MDSC subsets in the peripheral blood was determined by flow cytometry in a cohort of 49 patients with advanced head and neck cancer (HNC) and 22 patients with urological cancers. Sorted and purified MDSC subsets were tested in vitro for their T-cell–suppressive capacity. Frequency of circulating MDSC was correlated with overall survival of patients with HNC.
Results: A high frequency of PMN-MDSC most strongly correlated with poor overall survival in HNC. T-cell–suppressive activity was higher in PMN-MDSC compared with M-MDSC and e-MDSC. A subset of CD66b+/CD11b+/CD16+ mature PMN-MDSC displayed high expression and activity of arginase I, and was superior to the other subsets in suppressing proliferation and cytokine production of T cells in both cancer types. High levels of this CD11b+/CD16+ PMN-MDSC, but not other PMN-MDSC subsets, strongly correlated with adverse outcome in HNC.
Conclusions: A subset of mature CD11b+/CD16+ PMN-MDSC was identified as the MDSC subset with the strongest immunosuppressive activity and the highest clinical relevance. Clin Cancer Res; 24(19); 4834–44. ©2018 AACR.
In patients with cancer, three major subsets of circulating myeloid-derived suppressor cells (MDSC) exist. The suppressive capacity and clinical relevance of these three subsets (termed PMN-MDSC, M-MDSC, and e-MDSC) have not been compared so far. Using a cohort of patients with head and neck cancer, we found that a subset of CD16+/CD11b+ PMN-MDSC strongly correlates with poor outcome and has superior suppressive capacity compared with all other MDSC subpopulations. Our study will guide and improve the future immunomonitoring of MDSC in patients with cancer and suggests that CD16+ PMN-MDSC could be an attractive target for immunointervention in patients with solid tumors.
Myeloid-derived suppressor cells (MDSC) are a heterogeneous group of myeloid cells with immunosuppressive and tumor-promoting activity (1). Originally, MDSC were defined in tumor-bearing mice, where two major subsets exist in blood, lymphoid organs, and tumor tissue. These two subsets are now defined as Ly6G+ granulocytic [polymorphonuclear-MDSC (PMN-MDSC)] and Ly6C+ monocytic (M-MDSC) (2). Suppression of T-cell activity is the hallmark of MDSC function, and the cells exploit an arsenal of molecules and mechanisms to downregulate T-cell function. The relative usage and importance of this arsenal are species, tissue, cell-type, and context dependent (3).
For many years, research on human MDSC has been hampered by a lack of consensus on markers and immunophenotypic characterization. More recently, progress has been made, and several initiatives are ongoing to aid the characterization and monitoring of these cells in human diseases (refs. 4, 5; www.mye-euniter.eu). Although the immunophenotype and characteristics of human tissue MDSC are still largely elusive, minimal criteria to define the immunophenotype of circulating peripheral blood MDSC are already better established: Circulating PMN-MDSC express the neutrophil markers CD15 and/or CD66b, lack the monocyte marker CD14, and are positive for CD33. M-MDSC normally express higher levels of CD33 compared with PMN-MDSC, are positive for CD14, and characteristically have low or even absent levels of HLA-DR. Early-stage MDSC (e-MDSC) do express the myeloid markers CD33 and CD11b but lack the myeloid lineage markers CD14 and CD15/CD66b and have low or absent expression of HLA-DR (4, 6).
Based on their function and pathologic expansion, it was hypothesized that these human MDSC subsets may be of clinical relevance and influence disease progression and response to therapy. Although such evidence is slowly emerging, data are still surprisingly scarce, especially for PMN-MDSC and e-MDSC (7). Against this background, we saw an urgent need for a study directly comparing the clinical relevance of all three human peripheral blood MDSC subsets.
PMN-MDSC and M-MDSC are associated with T-cell suppression not only in mice but also in human cancer patients. Early-stage MDSC have been more recently suggested as a separate subset, and the functional properties of this subset still need to be defined (4). Therefore, in this study, we isolated all three MDSC subsets from patients with head and neck cancer (HNC) and directly compared their T-cell–suppressive capacity in different assays of polyclonal T-cell stimulation—the most frequently used readout system in the field. We found a clear hierarchy of suppression, with PMN-MDSC being the most suppressive subset and e-MDSC showing little or even no evidence of T-cell–suppressive activity.
PMN-MDSC are rarely detected in healthy donors but are significantly expanded in patients with cancer (8). A key feature of these cells is the fact that, based on a change in cellular density, PMN-MDSC copurify with mononuclear cells in density gradients. After earlier descriptions in different types of advanced adenocarcinomas (9), the presence of CD15/CD66b-positive cells has now been confirmed in many different types of cancer (8, 10, 11) and also in inflammatory and infectious diseases (12, 13). To further define these cells as MDSC, and to contrast them with inflammatory, so-called low-density granulocytes (LDG) observed, for example, in lupus and psoriasis (14), functional assays are needed. Although MDSC may also impair the function of non-T cells, the suppression of proliferation and cytokines in polyclonally stimulated T cells is currently the most often used and best accepted assay to monitor the functional activity of MDSC. To avoid potential pitfalls of a single test system, we used different systems of polyclonal T-cell stimulation and combined those with different readouts on proliferation and cytokine responses to comprehensively assess the suppressive activity of MDSC subsets in our study.
Originally, PMN-MDSC were mainly characterized by granulocyte lineage markers and considered as a homogenous cell population. More recently, evidence has accumulated to suggest that PMN-MDSC may display some heterogeneity in terms of differentiation and activation states (8, 15). Consequently, the presence of immature as well as activated subsets has been described (16, 17). Others have searched for markers to distinguish low-density PMN-MDSC from high-density “regular” neutrophils of the same patient. As a result of those transcriptomic-based analyses, LOX-1 was suggested as a novel marker to distinguish PMN-MDSC from nonsuppressive neutrophils (18).
We used CD16 and CD11b, two markers well recognized to define differentiation and activation states of neutrophils (19, 20), to define three immunophenotypically distinct subsets of PMN-MDSC. Those three PMN-MDSC subsets clearly differed in their capacity to limit the activity of stimulated T cells.
Materials and Methods
Study subjects and collection of blood
Pretreatment blood samples were collected prospectively from patients with advanced HNC and from healthy donors (see Supplementary Table S1 for blood donor characteristics). Patients with prior radiotherapy or chemotherapy, a synchronous carcinoma in another location, or a severe concomitant systemic infectious disease were excluded. Treatment strategies (surgery alone, surgery followed by radiotherapy or radiochemotherapy, primary radiochemotherapy) were developed by the multidisciplinary tumor board for HNCs, and treatments were conducted according to state-of-the-art technique and guidelines in place at the West German Cancer Center. Follow-up for all patients was retrieved for a period of at least 30 months. In addition to the HNC, 22 patients with urological cancers were included into the study. These patients had cancer of the ureter or the urinary bladder (10 patients), cancer of the prostate (5 patients), renal cell carcinoma (6 patients), or seminoma (1 patient). None of these patients presented with distant metastases. The cohort included four females (mean age, 58 years) and 18 males (mean age, 64 years).
Isolation of peripheral blood mononuclear cells and MDSC from the peripheral blood
We used an established protocol for isolation of MDSC (16). In brief, peripheral blood was drawn into 3.8% sodium citrate anticoagulant. Within 1 hour, blood was admixed with PBS (ratio 1:1 v/v) before separation by density gradient centrifugation (Biocoll, Merck-Biochrom). Mononuclear cells and cosedimenting MDSC were harvested from the interphase. Cells were used immediately after isolation for characterization and functional analysis.
Analytical flow cytometry
For characterization of MDSC subsets, the following antibodies were used: CD66b FITC (clone 80H3; Beckman Coulter), CD33 PE or APC (clone WM53), HLA-DR APC or HLA-DR PE-Cy7 (both clone G46-6), CD16 PE-Cy7 (clone 3G8), CD11b APC-Cy7 or V450 (clone Mac1; ICRF44; all BD Biosciences), CD14 PerCP-Cy5.5 (clone HCD14), CD16 BV510 (clone 3G8; both BioLegend), Arginase 1 APC (sheep polyclonal; Bio-Techne), and NOS2 (iNOS) Alexa Fluor 647 (clone c-11; Santa Cruz Biotechnology). Cells were analyzed with BD FACSCanto II using DIVA 8.01 software (BD Biosciences).
Isolation of MDSC subsets and PMN-MDSC subsets by FACS
Before sorting of MDSC subsets, the CD3+ cells were depleted from peripheral blood mononuclear cell (PBMC) using CD3 MicroBeads according to the manufacturer's instructions (Miltenyi Biotec) and later used in T-cell suppression assays. The CD3-depleted fraction was then labeled for isolation of MDSC subsets with CD66b FITC, CD33 PE, and HLA-DR APC or for isolation of PMN-MDSC subsets with CD66b FITC, CD11b APC-Cy7, and CD16 PE-Cy7. Cell suspensions were sorted on the lowest flow rate for HLA-DR−/CD33high (M-MDSC), HLA-DR−/CD33dim/CD66b+ (PMN-MDSC), HLA-DR−/CD33dim/CD66b− (e-MDSC) or CD66b+/CD11b−/CD16−, CD66b+/CD11b+/CD16−, and CD66+/CD11b+/CD16+ using BD FACSAria III (BD Biosciences; Figs. 1 and 4; Supplementary Fig. S1). Purity of sorted populations was >90% in all experiments.
Isolated PMN-MDSC subsets were resuspended in 200 μL PBS/60% FCS and centrifuged for 4 minutes at 60 x g on UNIVERSAL 320 centrifuge with cyto rotor (Hettich Lab Technology) for preparation of glass slide cytospins. Spun cells were dried and subjected to a Pappenheim staining. Staining consisted of 2 minutes with May-Grünwald solution (Merck), a washing step in distilled water, and 10-minute counterstaining with Giemsa solution (Merck) before mounting. Cells were examined using Axioskop 2 (Carl Zeiss MicroImaging) with objective lens EC Plan-Neofluar 63×/1.25 oil (ZEISS). Images were generated using AxioCam MRc (ZEISS) and AxioVision AxioVS40 software (ZEISS).
Measurement of arginase activity
PMN-MDSC subsets were sorted as described above. Cells were cultured for 24 hours in l-lysine and l-arginine–free RPMI medium (Thermo Fisher Scientific) supplemented with 150 μmol/L l-Arginine, 0.218 mmol/L l-Lysine hydrochloride (both Sigma-Aldrich), 2 mmol/L l-Glutamine, 100 IU/mL penicillin, 100 mg/mL streptomycin, and 10% (v/v) heat-inactivated FCS. Supernatant was collected and stored at −80°C until measurement. Determination of arginase activity was performed using an arginase assay kit (Abnova/Biomol) according to the manufacturer's protocol.
T-cell suppression assay (proliferation and secreted cytokines)
After isolation by CD3 microbeads (see above), T lymphocytes were labeled with 10 μmol/L Cell Proliferation Dye eFluor 450 (CPDye405) according to the manufacturer's instructions (eBioscience). For induction of T-cell proliferation, cells were stimulated in RPMI 1640 medium (Thermo Fisher Scientific) supplemented with 10% (v/v) heat-inactivated FCS, 100 IU/mL penicillin, and 100 mg/mL streptomycin (Thermo Fisher Scientific) in a 96-well round-bottom plate, either with CD2/CD3/CD28 beads (Miltenyi Biotec; regulatory T-cell suppression inspector kit, T-cell:bead ratio, 1:2), CD3/CD28/CD137 beads (Thermo Fisher Scientific; T-cell:bead ratio, 1:10) or coated CD3 (1 μg/mL, clone OKT-3; eBioscience) and CD28 (2 μg/mL, clone 28.2; Beckman Coulter). Autologous MDSC subsets or PMN-MDSC subsets were added in a T-cell:MDSC ratio of 2.5:1. To study mechanisms of suppression, the assays were performed in l-lysine– and l-arginine–free RPMI medium (Thermo Fisher Scientific) supplemented as described above. For inhibition of arginine, 10 μmol/L Nω-hydroxy-nor-Arginine (nor-NOHA); for ROS inhibition, 10 μmol/L N-Acetyl-L-cysteine (NAC; both Cayman Chemical/Biomol); and for iNOS inhibition, 100 μmol/L 1400 W hydrochloride (Bio-Techne) were added together with MDSC. CPDye405 intensity was analyzed by flow cytometry after 4 days of coculture and proliferation. Proliferation index calculation is based on dye dilution and was calculated with ModFit LT 3.3 (Verity Software House) according to an algorithm provided by the software. The index of the nonproliferated fraction was subtracted, and the index of T cells without MDSC was set as 100%. The proliferation index is the sum of the cells in all generations divided by the computed number of original parent cells theoretically present at the start of the experiment. The proliferation index thus reflects the increase in cell number in the culture over the course of the experiment. Supernatants of the coculture were collected, and IFNγ, IL10, and IL2 were measured by ELISA (Bio-Techne) according to the manufacturer's protocol.
Suppression of intracellular IFNγ from T cells
To determine intracellular IFNγ expression, the T cells were stimulated with CD3/CD28/CD137 beads (Thermo Fisher Scientific; T-cell:bead ratio, 1:10) or coated CD3 (2 μg/mL) and CD28 (2 μg/mL) for 12 hours with MDSC (T cells:MDSC ratio, 2.5:1) in the presence of BD GolgiStop (monensin). Afterward, cocultures were stained with CD3 FITC (clone BW264/56), followed by intracellular staining using BD Cytofix/Cytoperm plus kit and anti-IFNγ APC (clone 451-15; Miltenyi Biotec). CD3+/IFNγ+ cells were analyzed using BD FACSCanto II.
Data were analyzed using GraphPad Prism software (GraphPad Software). Statistical significance was assessed with Wilcoxon matched-pairs signed rank test. Results were considered significant at *, P ≤ 0.05; **, P ≤ 0.005; and ***, P ≤ 0.0005.
Clinical data were analyzed with the SPSS statistical software version 22 (SPSS). Survival curves (30-month follow-up period) were plotted according to the Kaplan–Meier method. Significance was initially tested in a univariate analysis using the log-rank test. The level of significance was set at P ≤ 0.05.
The use of human samples was approved by the ethics committee of the medical faculty of the University of Duisburg-Essen. Written informed consent was obtained from participants prior to inclusion in the study, and the study was performed in accordance with recognized ethical guidelines as specified in the Declaration of Helsinki.
High frequencies of PMN-MDSC are associated with poor survival
We used currently available and consensual marker combinations to derive a straight-forward gating strategy, which defines the three current human MDSC populations as nonoverlapping subsets (Fig. 1A). This strategy is based on previously suggested protocols (4, 6, 10) and defines M-MDSC as CD14+/CD66b−/CD33high/HLA-DRlow, PMN-MDSC as CD14−/CD66b+/CD33dim/HLA-DRlow or HLA-DRnegative, and e-MDSC as CD14−/CD66b−/CD33dim/HLA-DRlow or HLA-DRnegative. We then determined the relative frequency of these three MDSC subsets in the patients’ peripheral blood PBMC fraction, calculated the median for each subset, and compared it with frequencies in a control cohort of healthy donors (Fig. 1B). Kaplan–Meier curves of overall survival were plotted for patients, which had MDSC counts either below or above the median for each MDSC subset (see Supplementary Table S1 for patient characteristics). As shown in Fig. 1C, patients with a high frequency (above the median of 0.51% in total PBMC) of PMN-MDSC showed significantly decreased survival (P = 0.042). By tendency, also high levels (above the median of 0.31%) of M-MDSC were associated with poor outcome, although not reaching statistical significance (P = 0.055). Of note, no correlation between MDSC frequency and overall survival existed for e-MDSC (median is 0.88%). Frequencies of PMN-MDSC and M-MDSC, but not e-MDSC, were significantly increased in patients over healthy donors.
PMN-MDSCs are the most potent T-cell–suppressive MDSC subset
Suppression of T-cell activity is a hallmark of MDSC function and becomes increasingly important when defining the regulatory activity of myeloid cells in patients with cancer (4, 8). Polyclonal T-cell stimulation is most often used to assess this property, but exact technical guidelines and recommendations for these assays are still lacking. We used two different modes of stimulation (plate-bound antibodies and bead-bound antibodies) and three different readout systems (T-cell proliferation, cytokine ELISA, and cytokine intracellular flow cytometry) to comparatively assess the potential T-cell–suppressive capacity of all three MDSC subsets. Using this set-up, we obtained a comprehensive MDSC activity profile. We found that PMN-MDSC were the most potent suppressors of T-cell proliferation in these assays, and suppressive activity clearly exceeded the activity of M-MDSC. In contrast to the other two MDSC types, e-MDSC showed only little suppression of proliferation of polyclonally stimulated T cells (Fig. 2A and B). We next analyzed IFNγ secretion and expression. Similarly to T-cell proliferation, IFNγ production was most strongly suppressed by PMN-MDSC (Fig. 2C and D). M-MDSC reduced IFNγ release to about 50%, clearly inferior to PMN-MDSC, which suppressed IFNγ secretion by more than 80% (Fig. 2D). No statistically significant reduction of IFNγ responses was observed for e-MDSC. To substantiate these findings, two other cytokines (IL2 and IL10) were also evaluated. Again, statistically significant reduction of cytokine production was observed only for M-MDSC and PMN-MDSC, with PMN-MDSC showing superior suppressive activity also in these assays (Fig. 2E and F). We next sought to determine molecular mechanisms involved in this suppression. At first, we measured expression of arginase I and iNOS—two molecules previously identified as key factors in T-cell suppression by MDSC (1, 4, 17). Both, arginase and iNOS were strongly expressed in PMN-MDSC, and levels of expression exceeded expression in M-MDSC and e-MDSC (Fig. 3A and B). Mechanistically, inhibitors of arginase (Fig. 3C) and iNOS (Fig. 3D), but not ROS inhibition (Fig. 3E), restored T-cell proliferation in the presence of PMN-MDSC. Inhibition of iNOS was superior to arginase inhibition in restoring T-cell proliferation in the presence of M-MDSC and e-MDSC. Interestingly, the suppression of IFNγ was primarily affected by arginase inhibition (Fig. 3F), but not by the other two inhibitors (Fig. 3G and H).
In sum, these data establish PMN-MDSC as the strongest inhibitor of T-cell activity, whereas M-MDSC show intermediate activity and e-MDSC marginal or no inhibition of T-cell effector functions. Arginase I appears to be a major molecular pathway involved in inhibition by PMN-MDSC.
Identification of differential suppressive activity within the PMN-MDSC population
PMN-MDSC copurify with PBMC in density gradients and are phenotypically equivalent to low-density neutrophils (LD-PMN), a subset of neutrophils, which is induced in many pathologies including cancer, infection, and inflammation (8, 15). Previous studies have reported that these PMN-MDSC/LD-PMN fractions are heterogeneous and contain phenotypically diverse subsets of granulocytic cells (8, 15, 21). Based on previously published data (16), we used CD11b and CD16, two markers regulated during neutrophil differentiation (19, 20), to define three phenotypically distinct subsets within the PMN-MDSC fraction (Fig. 4A). Purification of CD16+/CD11b+, CD16−/CD11b+, and CD16−/CD11b− PMN-MDSC revealed that these phenotypes indeed displayed differential nuclear morphology, which is consistent with different maturation stages (Fig. 4B). Whereas the nucleus of double-positive PMN-MDSC mainly displayed the polymorphonuclear shape of mature cells, CD11b+/CD16− and double-negative cells consisted of band-shaped and less mature progenitor stages, respectively. The immature differentiation stage of the double-negative cells was confirmed by ex vivo differentiation. Exposure of sorted double-negative PMN-MDSC to G-CSF for 7 days resulted in strong upregulation of CD11b and modest upregulation of CD16 (Supplementary Fig. S1).
To more precisely decipher the functional composition of the heterogeneous PMN-MDSC population, we sorted all three PMN-MDSC subsets and performed a side-by-side comparison of their suppressive activity in four different T-cell test systems (Fig. 4C–F; test systems similar to Fig. 2). In three out of four test systems, all three PMN-MDSC subsets exerted suppressive activity. Only for the CD3/CD28/CD137-bead–induced IFNγ secretion (Fig. 4F) was the suppressive activity confined to the CD11b+/CD16+ subset only. Most importantly, it was this subset, which also consistently showed the strongest capacity to reduce T-cell effector activity in all four systems tested, with the strongest arginase expression (Fig. 4G) and activity (Fig. 4I). The CD11b+/CD16− subset had intermediate T-cell–suppressive activity, arginase activity (Fig. 4I), and arginase expression (Fig. 4G). Interestingly, and despite high iNOS expression, the double-negative subset showed the weakest capacity to reduce activated T-cell functions together with low arginase expression and activity. Thus, using a comprehensive set of test systems, we showed that CD66b+/CD11b+/CD16+ PMN-MDSC have strong arginase activity and are superior to the other PMN-MDSC subsets in terms of T-cell suppression.
To validate these findings further, we reanalyzed suppression assays performed with total PMN-MDSC fractions. Using a plate-bound and a bead-based stimulation system, we correlated the degree of T-cell proliferation and the level of CD16 expression/positivity. Here, we made use of the fact that the relative ratio of CD16+ and CD16− PMN-MDSC substantially varies among patients (see Fig. 5A for gating of PMN-MDSC and Fig. 5B for examples of CD16 heterogeneity). Confirming the previous findings, we observed a positive correlation between the levels of CD16 expression and suppression of T-cell proliferation (Fig. 5C and D). To test whether functional PMN-MDSC heterogeneity can also be observed in other tumor entities besides HNC, we embarked on a small pilot study in urological cancers. Similar to HNC, also in this group of patients, PMN-MDSC frequency was increased compared with healthy donors (Supplementary Fig. S2A). When we sorted those PMN-MDSC into the three subsets according to expression of CD11b and CD16, we confirmed data from HNC PMN-MDSC and found strongest expression of arginase I, low expression of iNOS, and strong T-cell–suppressive activity in the CD11b+/CD16+ subset (Supplementary Fig. S2B–S2D).
Clinical relevance of PMN-MDSC subsets
After we had found differential T-cell–suppressive activity of the three PMN-MDSC subsets, we finally tested their association with clinical outcome in a cohort of 49 patients with advanced HNC. To this end, we separated patients into groups with high or low levels for each of the three PMN-MDSC subsets. Interestingly, high levels of the CD11b+/CD16+ PMN-MDSC subset (above the median of 0.302% in total PBMC) were most strongly associated with poor survival (Fig. 6).
Over the last 10 to 20 years, the interest in MDSC, formerly also known as MSC or myeloid suppressor cells, has increased constantly (22). This has led to a better understanding of the identity, characteristics, and expansion mechanisms associated with these cells (1). A major advancement in the field has been the development of markers to define specific MDSC subsets, both in human patients and in murine disease models (4). Species-specific markers are now available to identify granulocytic and monocytic MDSC in mice and humans. Some years back, and based on existing data at that time, it was suggested to define a third putative circulating MDSC subset lacking granulocytic and monocytic lineage markers (6). More recently, recommendations for the characterization of those three major phenotypes of circulating human MDSC subpopulations were proposed (4). However, and despite considerable progress in the field, no comparative analysis of those three major human peripheral blood MDSC subsets is available to date, and the relative importance of these subsets for T-cell suppression and tumor progression in patients with cancer remains unclear. To address this issue, we compared the frequency, function, and clinical relevance of the three human MDSC subsets in a well-defined cohort of patients with advanced HNC.
Although functional testing is a major criterion in defining MDSC, at present, there is no consensus on the best test system, and numerous T-cell suppression assays are used and reported in the literature (23). In addition, MDSC may also inhibit natural killer cells, employing mechanisms distinct from suppression of T-cell activity (24, 25). To avoid reliance on a single system only, we applied different systems of T-cell proliferation and cytokine release in this study. Using this approach, we found a high suppressive capacity for PMN-MDSC, which exceeded those of the two other subsets. Consistent with this finding, a high frequency of PMN-MDSC in the circulation was most strongly associated with poor patient survival. M-MDSC also displayed substantial inhibition of T-cell functions, and high frequencies of this subset were likewise associated with poor outcome, albeit of borderline statistical significance. These data define PMN-MDSC as the most suppressive subset in this study and further emphasize the emerging clinical relevance of both PMN-MDSC and M-MDSC in human oncology (7).
In contrast to PMN-MDSC and M-MDSC, the origin, identity, and function of e-MDSC are very poorly understood, and little, if any, information on the characteristics and clinical relevance of these cells in human cancer patients is known. We found no or very little suppression of T-cell proliferation and cytokine expression by e-MDSC. In addition, the frequency of e-MDSC in the peripheral blood was not associated with patient outcome.
Although it is too early to generalize these findings, at least based on the data obtained in this study, it is thus debatable whether e-MDSC represent bona fide “functional” MDSC similar to PMN-MDSC and M-MDSC. Further studies in our lab are underway to test the hypothesis that these e-MDSC are potential MDSC precursors, which require additional expansion or activation steps to acquire suppressive activity.
PMN-MDSC copurify with MNC in density gradients and are thus phenotypically equivalent to so-called LD-PMN (8, 15). PMN-MDSC/LD-PMN are expanded in patients with cancer and largely absent in healthy donors (8, 15). In recent years, a remarkable plasticity and heterogeneity of PMN in cancer and other inflammatory diseases is emerging, and a large part of this can indeed be attributed to the diverse phenotypes and functions associated with LD-PMN (15, 26). Regardless of this plasticity, it is now clearly established that in both malignant and nonmalignant disease, the LD-PMN fraction contains immunosuppressive PMN-MDSC (16, 27–29). Yet, one of the current challenges in the field is to better define the characteristics and immunophenotypic markers associated with immunosuppressive PMN-MDSC within the pool of heterogeneous granulocytic cells in diseased patients. In a recent study, lectin-type oxidized LDL receptor 1 (LOX-1) was associated with the suppressive activity of PMN-MDSC in the peripheral blood of patients with cancer (18). Based on a transcriptomic approach, the authors identified LOX-1 as overexpressed in the LD-PMN/PMN-MDSC fraction compared with PMN of regular density. In their study, about 30% of LD-PMN in the peripheral blood expressed this marker, whereas only about 5% of total peripheral blood PMN were positive. In that study, no LOX-1+ cells from the LD-PMN fraction were sorted and functionally tested. When sorted from the total PMN population, LOX-1+ cells displayed a more immature morphology and were more suppressive than their LOX-1− counterparts.
In an alternative and partially contradictory approach, Marini and colleagues analyzed LD-PMN from G-CSF–mobilized donors (30). G-CSF injection resulted in a massive induction of the LD-PMN fraction, which contained mixtures of immature and mature stages of neutrophils and by far exceeds the induction observed in most patients with cancer. Marini and colleagues used CD10 to separate immature from mature stages and found that CD10+ mature cells were T-cell suppressive, whereas immature CD10− cells stimulated T-cell responses.
Based on previous findings in our group (16), we used CD11b and CD16 as markers to define subsets of PMN-MDSC. Both molecules are upregulated during granulopoiesis, and CD11b and CD16 are well-established markers to define differentiation stages of neutrophils (19, 20). Consistent with those reports, we confirmed that also in patients with cancer, the expression of CD11b/CD16 correlates with the nuclear morphology and thus identifies circulating mature neutrophils (CD11b+/CD16+) and immature stages (marker negative), respectively (Fig. 3A and B). In addition, immature stage of CD11b/CD16 double-negative PMN-MDSC was confirmed by G-CSF–driven in vitro differentiation of those cells, along with strong upregulation of CD11b and modest upregulation of CD16 (Supplementary Fig. S1).
One of the most important findings of our study is the fact that PMN-MDSC subsets defined by differential expression of CD11b and CD16 also displayed differential suppressive activity on T cells. CD66b+/CD11b+/CD16+ cells were the most suppressive subset, and this was true for both inhibition of proliferation and inhibition of IFNγ secretion. These mature PMN-MDSC also showed the highest expression of arginase I, which we identified as a major mechanism for suppression of T-cell proliferation and cytokine secretion. Immature CD11b−/CD16− PMN-MDSC were less suppressive compared with mature counterparts but still inhibited T-cell activity and should thus also be considered bona fide PMN-MDSC. Of note, suppression hierarchy among PMN-MDSC subsets was confirmed in patients with urological cancers (Supplementary Fig. S2).
All in all, our data confirm existing data suggesting that cancer progression induces and drives, among total MDSC, the expansion of PMN-MDSC, with induction of both immature and mature granulocytic cells. In addition, our data suggest that maturation of PMN-MDSC is associated with an increase in suppressive capacity and that maturation is required for PMN-MDSC to generate high regulatory activity. This is in line with data from G-CSF–mobilized donors in which mature LD-PMN were suppressive on T cells, whereas immature PMN were stimulatory (30). It is, however, important to note that in our patients with cancer, immature PMN-MDSC did not stimulate T cells. Instead, T-cell activity was still inhibited, albeit at a lower level as compared with the more mature PMN-MDSC counterparts.
Solito and colleagues reported on an in vitro system of MDSC generation, in which promyelocytic cells of early differentiation stages were shown to have the strongest T-cell–suppressive capacity (31). In this system, the authors used bone marrow cells to generate MDSC in vitro by stimulation with G-CSF and GM-CSF. In contrast to data from peripheral blood MDSC (this study and the one by Marini and colleagues), bone marrow–derived MDSC of promyelocytic nature and not more differentiated stages were T-cell suppressive in this in vitro model.
In an earlier study, we found that the frequency of PMN-MDSC in patients with HNC was associated with high serum levels of IL6. However, serum levels of colony-stimulating factors did not correlate with the levels of circulating PMN-MDSC (10). It thus appears that human in vitro systems of MDSC generation may serve as good models for specific molecular analyses. However, whether in vitro–generated MDSC truly reflect their human patient in in vivo counterparts remains to be tested.
In sum, for the first time, we have directly compared the frequency, T-cell–suppressive capacity, and clinical relevance of the three major circulating MDSC subpopulations in human cancer patients. We found that, among those subsets, PMN-MDSC showed the highest suppressive activity and most strongly correlated with survival. Suppression data were validated in various test systems and included T-cell proliferation and cytokine response. Among the PMN-MDSC, a subset double positive for CD11b and CD16 displayed the strongest inhibition of T-cell activity along with the strongest expression of arginase I. Our study should guide further research on immunomonitoring and targeting of MDSC in human oncology.
Disclosure of Potential Conflicts of Interest
B. Hadaschik is a consultant/advisory board member for Bristol-Myers Squibb and Janssen R&D. No potential conflicts of interest were disclosed by the other authors.
Conception and design: S. Brandau
Development of methodology: S. Brandau, K. Bruderek
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Lang, K. Bruderek, C. Kaspar, B. Höing, O. Kanaan, N. Dominas, T. Hussain, F. Droege, C. Eyth, B. Hadaschik, S. Brandau
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Bruderek, S. Brandau
Writing, review, and/or revision of the manuscript: S. Lang, K. Bruderek, B. Höing, O. Kanaan, F. Droege, C. Eyth, B. Hadaschik, S. Brandau
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Lang, K. Bruderek, C. Kaspar, B. Höing, O. Kanaan, C. Eyth, S. Brandau
Study supervision: S. Brandau
The authors thank Sandra Wasielewski, Ulrike Kaiser, and all clinicians of the ENT Department for supporting the retrieval of patient material for this study. The authors thank Sebastian Vollmer and Alexander Hackel for their kind help and the Imaging Center Essen (IMCES) for supporting the FACS studies. S. Brandau and K. Bruderek were supported by European Cooperation in Science and Technology (COST) Action Mye-EUNITER (www.cost.eu and www.mye-euniter.eu).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.