Yeast-Derived Particulate β-Glucan Treatment Subverts the Suppression of Myeloid-Derived Suppressor Cells (MDSC) by Inducing Polymorphonuclear MDSC Apoptosis and Monocytic MDSC Differentiation to APC in Cancer

It is well appreciated that tumor cells produce a plethora of immune-modulatory factors that constraint the tumor cytotoxic effects mediated by antitumor innate and adaptive immune responses (1–3). Not only do tumor-derived factors drive angiogenesis for nutrient supply, but they also disrupt the rhythm of differentiation of bone marrow–derived immune cells toward the accumulation and expansion of a heterogeneous population of immature immune-suppressive cells known collectively as myeloid-derived suppressor cells (MDSC) (4). In mice, two main subsets of MDSC have been identified according to their morphology and Gr-1, Ly6C, Ly6G, and CD11b expression: monocytic MDSC (M-MDSC) resemble monocytes and are Gr-1^low/intCD11b⁺(Ly6C^highLy6G⁻CD11b⁺) (5) and polymorphonuclear MDSC (PMN-MDSC) resemble polymorphonuclear granulocytes and are Gr-1^highCD11b⁺(Ly6G^highLy6C^lowCD11b⁺) (6). In humans, MDSC lack the Gr-1 homolog and are defined as CD14⁻HLA-DR⁻CD11b⁺CD33⁺ or CD14⁺HLA-DR⁻CD11b⁺CD33⁺ (7–10).

After the identification of MDSC as one of the major suppressors of T cell responses and inducers of T cell tolerance (11, 12), numerous studies have characterized their roles in cancer as suppressors of NK cells (13), inducers of regulatory T cells (14), and precursors of tumor-associated macrophages (7). MDSC-mediated T cell suppression is mainly attributed to the expression of Arginase 1, inducible NO synthase (iNOS), reactive oxygen species (ROS) (4), and cystine and cysteine deprivation (15). A main factor responsible for the accumulation of MDSC in cancer is the fact that MDSC are immature and do not subsequently differentiate to antitumor macrophages and dendritic cells (DC) under the influence of tumor-derived factors (16). Therefore, the importance of targeting MDSC expansion, suppression, and differentiation in combination with other therapies in cancer is being very well appreciated (17).

In an attempt to study a natural compound that targets MDSC, we studied the effect of the immunomodulator, particulate β-glucan on MDSC in tumor-bearing animals and non–small cell lung cancer (NSCLC) patients. Whole β-glucan particles (WGP) are microparticles of 1,3-β-glucan extracted from the yeast Saccharomyces cerevisiae, which has been shown to activate immune cells through the stimulation of C-type lectin receptor, dectin-1 (18, 19). Previous studies have shown that β-glucan treatment activates DC and induces T cell responses in vivo (20). Additionally, a recent report showed that WGP partially induces the differentiation of M-MDSC to F4/80⁺CD11c⁺ cells (21). However, the roles of particulate β-glucan in the modulation of the function of different MDSC subsets in mice and its implementation in human patients need further studies.

In the present study, we delineated the effect of particulate β-glucan on the function of both PMN-MDSC and M-MDSC in mice. We demonstrated that particulate β-glucan induced a cytotoxic phenotype and subsequent apoptosis in PMN-MDSC, whereas it converted M-MDSC to potent APCs that uptake, process, and present OVA Ag to OVA-specific CD4⁺ T cells and also cross-present Ag to prime OVA-specific CD8⁺ T cells. More importantly, NSCLC patients treated with particulate β-glucan for 2 wk had a decreased accumulation of CD14⁻HLA-DR⁻CD11b⁺CD33⁺ MDSC in their peripheral blood as compared with its frequency in the peripheral blood before treatment, which was correlated with an increased trend in peripheral monocytes and MDSC Ag presentation function, as well as a decreased trend in Arginase 1 mRNA expression in peripheral polymorphonuclear neutrophils. Overall, these findings provide a further step in our understanding of the mechanisms by which particulate β-glucan enhances antitumor immunity in mice and how these findings can be translated to the benefit of cancer patients.

Wild-type (WT) C57BL/6 mice were purchased from the National Cancer Institute. Dectin-1 knockout (KO) mice were described previously (22, 23). OT-I Rag1^−/− and OT-II CD4 OVA TCR-transgenic mice were purchased from Taconic.

Tumor models and particulate β-glucan therapeutic protocol were performed as described previously (20). Briefly, C57BL/6 WT or dectin-1 KO mice were implanted s.c. with Lewis lung carcinoma (LLC) (2 × 10⁵/mouse) or mammary cell carcinoma (E0771) cell lines (6 × 10⁵/mouse). On day 8 after implantation, mice with palpable tumors were orally administered daily by gavage with 100 μl particulate β-glucan (Biothera, 800 μg/mouse in PBS suspension) or with 100 μl PBS. Tumor diameters were measured with a caliper every 3–4 d, and mice were killed when the tumor diameter reached 15 mm. The tumor volume was calculated as (length × width²)/2. The protocols on murine tumor models were performed according to the institutional guidelines and laws, and were approved by the Institutional Animal Care and Use Committee at the University of Louisville.

NSCLC patients that were newly diagnosed received particulate β-glucan treatment at the James Graham Brown Cancer Center, University of Louisville. These patients did not receive any other treatment prior to or during β-glucan treatment. The clinicopathological features of those patients are summarized in Supplemental Table I. The study was approved by the Institutional Ethical Board, and blood samples were collected upon written informed consent. Each NSCLC patient received 500 mg particulate β-glucan for 10–14 d. Peripheral blood was collected before and after treatment. Whole blood cells were treated with ACK lysis buffer, washed twice with PBS, and stained with anti-human CD14, anti–HLA-DR, anti-CD11b, and anti-CD33 Abs (BioLegend) for 30 min on ice. Samples were then washed twice prior to acquisition on a FACSCalibur. Neutrophils were purified with a Histopaque density gradient and stored into TRIzol at −80°C for RNA extraction (24).

Mouse tumor tissues (12–15 mm in diameter) were excised and minced into small pieces. Additional mechanical digestion was performed with a gentleMACS dissociator (Mitenyi Biotec). Tumors were enzymatically digested in RPMI 1640 medium containing 10% FBS, type IV collagenase (1 μg/ml), and hyaluronidase (10 ng/ml) for 45 min at 37°C on a rotator. A second step of mechanical dissociation was performed after enzymatic digestion. The digested cells were then filtered, pelleted, and resuspended in complete RPMI.

Single-cell suspensions were treated with Fc Block for 10 min on ice and stained with the relevant fluorochrome-labeled mAbs for 30 min on ice. Cells were washed twice with PBS or staining buffer (PBS plus 0.1% FBS) (for FACS analysis) or running buffer (Miltenyi Biotec) (for cell sorting). For FACS analysis, cells were acquired using FACSCalibur or FACS Canto II (BD Biosciences). MDSC were sorted using BD FACSAria III cell sorter or MoFlo XDP (Beckman Coulter). The purity of sorted cells was >98% as assessed by flow cytometry. For intracellular staining of IFN-γ and granzyme B, cells were first stained with either anti-CD4 or anti-CD8 Abs, fixed, permeabilized with fixation/permeabilization buffer (BioLegend), washed with 1× permeabilization/washing buffer (BioLegend), and stained with the relevant cytokine Abs. Cells were then washed twice with 1× Perm/Wash buffer and analyzed by flow cytometry. Data analysis was performed using FlowJo software (Tree Star). The following fluorochrome-labeled mAbs were used: anti-mouse Gr-1, anti-mouse Ly6G, anti-mouse Ly6C, anti-mouse CD11b, anti-mouse CD45, anti-mouse F4/80, anti-mouse CD11c, anti-mouse CD40, anti-mouse CD86, anti-mouse CD80, anti-mouse IA/IE, anti-mouse H2-K^b, anti-CD8, anti-CD4, anti–IFN-γ, anti–granzyme B, and their corresponding isotype controls (purchased from BioLegend). The following anti-human mAbs were also purchased from BioLegend: anti-CD11b, anti-CD14, anti-CD33, anti-HLA-DR, anti-CD3, and anti–IFN-γ with their corresponding isotype controls.

Splenocytes from LLC-bearing C57BL/6 or dectin-1^−/− mice were stimulated with WGP for 3, 6, 14, or 18 h and then stained with anti–Gr-1 and anti-CD11b Abs (BioLegend) and 7-aminoactinomycin D and annexin V dyes (BD Biosciences). Cells were analyzed by flow cytometry.

Sorted PMN-MDSC from WT or dectin-1^−/− mice were stained with dihydrorhodamine 123 (DHR) as described previously (25, 26). Briefly, cells were incubated with 1μM DHR and catalase for 5 min at 37°C. Particulate β-glucan was then added for 30 min, 60 min and 120 min. Rhodamine 123 (RHO) was then detected on the FL-1 channel on a FACSCalibur or FACSCanto II.

Sorted M-MDSC or PMN-MDSC stimulated with or without particulate β-glucan (100 μg/ml) for indicated times were lysed in Triton X-100 lysis buffer in the presence of protease and phosphatase inhibitors. The whole-cell extracts were subjected to SDS-PAGE and electrotransferred to polyvinylidene difluoride membrane. The membranes were blocked and probed overnight at 4°C with the relevant primary Abs and then incubated with the secondary Abs. The blots were developed with ECL Plus Western blotting detection reagents (GE Healthcare). The primary Abs included: p-Erk1/2 (Thr²⁰²/Tyr²⁰⁴, Cell Signaling Technology). Erk1/2 (MK1, Santa Cruz Biotechnology), p-Stat3 (Tyr⁷⁰⁵, Cell Signaling Technology), p-AKT (Ser⁴⁷³, Cell Signaling Technology), p-p38 (Thr¹⁸⁰/Tyr¹⁸², Cell Signaling Technology), p-Zap/Syk (Tyr³¹⁹/Tyr³⁵², Cell Signaling Technology), STAT3 (C-20, Santa Cruz Biotechnology), p-SAPK/JNK (Thr¹⁸³/Tyr¹⁸⁵, Cell Signaling Technology), and β-actin (Sigma-Aldrich).

M-MDSC (CD11b⁺Ly6C^highLy6G⁻) were sorted from the spleens of LLC-bearing WT or dectin-1 KO mice and cultured for 7 d with 50 μg/ml particulate β-glucan in 48-well plates (Corning). For in vivo differentiation assays, M-MDSC were sorted from C57BL/6 LLC tumors (CD45.2) and treated with WGP (100 μg/ml) at 37°C for overnight. Freshly isolated and WGP-treated M-MDSC were intratumorally injected into SJL LLC tumor-bearing mice (CD45.1). The mice were sacrificed 7 d later and single-cell suspensions from tumors were stained with anti-CD45.2, F4/80, CD11c, and MHC class II mAbs. The cells were analyzed by flow cytometry.

For T cell proliferation assays, M-MDSC and PMN-MDSC sorted from the spleens or Gr-1⁺CD11b⁺ MDSC from tumors of LLC-bearing mice were cocultured with 1 μM CFSE-labeled splenocytes from OT-II or OT-I mice in the presence of OVA (100 μg/ml in OT-II cultures, 50 μg/ml in OT-I cultures, and 10 μg/ml in some splenic PMN-MDSC suppression experiments) and particulate β-glucan (50 μg/ml). Three days later, cells were harvested and stained. Additionally, some T cell proliferation assays were performed by coculturing sorted MDSC with CFSE-labeled splenocytes from C57BL/6 mice stimulated with plate-bound anti-CD3 (5 μg/ml) and soluble anti-CD28 (2 μg/ml).

For Ag-presentation assays, sorted M-MDSC from the spleens of LLC-bearing WT or dectin-1 KO mice were cultured in the presence or absence of particulate β-glucan (50 μg/ml) for 7 d. In some experiments, MEK1/2 inhibitor (PD98059) (30 ng/ml) or DMSO was added to cultures during differentiation. Cells were washed and cocultured with sorted and CFSE-labeled CD8⁺ or CD4⁺ T cells from OT-I and OT-II mice, respectively, in the presence or absence of whole OVA Ag (50 μg/ml). T cell proliferation and IFN-γ or granzyme B production were assessed 4–5 d later by flow cytometry.

To determine WGP treatment on Ag-specific T cell responses, WT mice were injected with OVA-expressing EG7 cells (3 × 10⁶/mouse). After palpable tumors formed, mice were adoptively transferred with purified OT-1 CD8 T cells (1 × 10⁶/mouse). Mice were treated with or without WGP treatment for 2 wk and sacrificed. Peripheral blood and tumor cells were used for tetramer staining. In brief, cells were blocked with Fc Block (anti-CD16/32) for 10 min and then incubated with Alexa Fluor 488–labeled OVA H2-K^b MHC class I tetramer obtained from the National Institutes of Health Tetramer Core Facility (1:1000 dilution) for 30 min at room temperature. Cells were then washed twice with PBS and stained with CD8 and CD19 mAbs and analyzed by flow cytometry. Total CD8⁺CD19⁻ cells were gated and analyzed for tetramer binding.

RNA was extracted with TRIzol reagent (Invitrogen) as described in the manufacturer’s protocols. Extracted RNA was transcribed to cDNA with a reverse transcription kit (Bio-Rad). Quantitative real-time PCR reactions were performed using SYBR Green Supermix (Bio-Rad) with the relevant primers (Supplemental Table I), and the reaction was detected on an MyiQ single-color RT-PCR detection system (Bio-Rad). The change in gene expression was quantified by measuring the change in threshold (ΔΔCt), where ΔCt = Ct_{target gene} − Ct_{housekeeping gene} and ΔΔCt = ΔCt_induced − ΔCt_reference. All primer sequences are listed in the Supplemental Table II.

Sorted splenic M-MDSC from WT or dectin-1 KO tumor-bearing mice were treated with or without particulate β-glucan for 18 h and mixed with LLC cells (1 × 10⁵) at a 1:1 ratio. Cells were mixed with Matrigel matrix basement membrane (Corning) and implanted s.c. in the flanks of C57BL/6 mice. Tumor diameter was measured every 2 d and mice were sacrificed on day 14 or day 25.

CD14⁺HLA-DR⁺CD3⁻ or CD14⁻HLA-DR⁻CD11b⁺CD33⁺CD3⁻ cells were sorted from the PBMC of NSCLC patients and cocultured with sorted and CFSE-labeled CD3⁺ cells from the PBMCs of allogeneic donors at 1:1 ratio for 5 d. Cells were then harvested and stained with human anti-CD3 and human anti–IFN-γ mAbs for subsequent flow cytometry analysis.

Data were analyzed using GraphPad Prism version 5.0 software. An unpaired Student t test was used to calculate significance. Significance was assumed to be reached at p < 0.05. All graph bars are expressed as mean ± SEM.

To delineate the effect of particulate β-glucan treatment on the composition of different MDSC subsets in spleens and tumors, C57BL/6 mice were challenged with LLC and mammary carcinoma E0771 cell lines s.c. Mice with palpable tumors were administered orally through gavage with PBS or particulate β-glucan (WGP) as previously established (20, 27). β-Glucan treatment significantly reduced tumor volumes (Fig. 1A), splenomegaly (Fig. 1B), and tumor weight (Fig. 1C) in both models. When analyzing the frequencies of MDSC subsets in the spleens and tumors, particulate β-glucan–treated mice had a significant decrease in PMN-MDSC frequencies in the spleen, but not M-MDSC, in both models (Fig. 1D). Additionally, particulate β-glucan treatment caused a significant decrease in the frequencies of PMN-MDSC and increase in the frequencies of M-MDSC in the tumors of E0771 mice and trending in the tumors of LLC-bearing mice (Fig. 1E).

Next, we determined whether particulate β-glucan treatment stimulates Ag-specific T cell responses. To this end, C57BL/6 mice were inoculated with OVA-expressing EG7 cells. Thus, we can track OVA-specific CD8 T cells using tetramer staining. As shown in Supplemental Fig. 1A, mice treated with WGP exhibited reduced tumor burden. OVA-specific CD8 T cells were significantly increased in the peripheral blood of WGP-treated tumor-bearing mice compared with those from untreated mice (Supplemental Fig. 1B). Furthermore, tumor-infiltrating OVA-specific CD8 T cells were also significantly increased in the tumor-bearing mice treated with WGP (Supplemental Fig. 1C). Those data suggest that particulate β-glucan treatment differentially regulate MDSC subsets with enhanced CD8 T cell responses.

Because suppression of CD8⁺ effector T cells in tumor-bearing mice is largely attributed to MDSC (28, 29), we thought to determine whether particulate β-glucan impacts T cell suppression mediated by splenic and tumor MDSC. To this end, we studied the effect of particulate β-glucan treatment on MDSC-mediated inhibition of OVA-specific CD4⁺ and CD8⁺ T cells or Ag nonspecific T cell proliferation induced by anti-CD3/CD28 stimulation. Suppression of OVA-specific CD4⁺ T cells mediated by splenic M-MDSC or PMN-MDSC was partially or completely obliterated upon particulate β-glucan treatment (Fig. 2A). Similarly, particulate β-glucan treatment significantly increased the proliferation of OVA-specific CD8⁺ T cells in the presence of suppressive M-MDSC or PMN-MDSC (Fig. 2B). Particulate β-glucan treatment also diminished the MDSC-mediated suppression of IFN-γ produced by OVA-specific CD4⁺ (Fig. 2C) and CD8⁺ T cells (Fig. 2D). Additionally, particulate β-glucan treatment decreased the ability of M-MDSC to suppress CD4⁺ T cells and CD8⁺ T cells stimulated with anti-CD3/CD28 (Fig 2E). However, splenic PMN-MDSC did not suppress the proliferation of T cells stimulated with anti-CD3/CD28, as previously reported (5). To generalize the efficacy of particulate β-glucan on the modulation of MDSC suppression, tumor Gr-1⁺CD11b⁺ cells were also sorted from LLC tumors and cocultured with OVA-specific CD4⁺ or CD8⁺ T cells in the presence of OVA at different ratios. Whereas Gr-1⁺CD11b⁺ MDSC suppressed the proliferation and IFN-γ production of CD4⁺ T cells (Fig. 3A, 3B) and CD8⁺ T cells (Fig. 3C), particulate β-glucan treatment significantly enhanced T cell proliferation and IFN-γ production on CD4⁺ and CD8⁺ T cells (Fig. 3). Taken together, these data emphasize the ability of particulate β-glucan to reverse MDSC-mediated T cell suppression.

Given that WGP treatment reduced PMN-MDSC–mediated T cell suppression and decreased PMN-MDSC frequency in the spleens and tumors of tumor-bearing mice, we tested whether particulate β-glucan treatment has a further impact on PMN-MDSC viability. Particulate β-glucan treatment enhanced PMN-MDSC apoptosis at 3, 6, and, more drastically, at 14 and 18 h poststimulation (Fig. 4A). PMN-MDSC apoptosis was mediated by dectin-1 receptor because particulate β-glucan did not induce PMN-MDSC apoptosis in PMN-MDSC sorted from the spleens of LLC-bearing dectin-1 KO mice (Fig. 4B). Because particulate β-glucan–induced apoptosis is more drastic at late time points, we asked whether particulate β-glucan treatment enhances PMN-MDSC respiratory burst, a hallmark of neutrophil-mediated cytotoxicity. We detected PMN-MDSC respiratory burst by DHR dye that oxidizes to fluorescent RHO in the presence of intracellular ROS (26). Upon particulate β-glucan stimulation, PMN-MDSC respiratory burst was significantly enhanced at 1 and 2 h poststimulation (Fig. 4C). To exclude the effect of dectin-1–independent phagocytosis on the induction of PMN-MDSC respiratory burst, we stimulated PMN-MSDC from dectin-1 KO mice with particulate β-glucan. As expected, particulate β-glucan stimulation did not induce respiratory burst in dectin-1 KO PMN-MDSC (Fig. 4C).

Previous studies have highlighted the importance of STAT3 phosphorylation in the cascade events mediating PMN-MDSC survival and suppression (4). Because particulate β-glucan treatment abrogated PMN-MDSC–mediated T cell suppression and induced PMN-MDSC apoptosis, we measured STAT3 phosphorylation at 15, 30, and 60 min poststimulation. STAT3 phosphorylation was inhibited after 15 min and completely abrogated after 30 min upon β-glucan stimulation (Fig. 4D). This effect was mediated by dectin-1 signaling, because particulate β-glucan stimulation enhanced Syk phosphorylation. Moreover, particulate β-glucan stimulation enhanced phosphorylation of molecules downstream of the dectin-1 receptor such as Akt, JNK, and Erk1/2 kinases but not p38 (Fig. 4D).

Previous studies have demonstrated the ability of particulate β-glucan to enhance DC Ag-presenting capability (27), and that 8–13% of M-MDSC cultured with particulate β-glucan and GM-CSF are F4/80⁺CD11c⁺ (21). To further determine the effect of particulate β-glucan on M-MDSC, we solely treated M-MDSC with particulate β-glucan in the absence of GM-CSF for 5 or 7 d and assessed the expression of F4/80, CD11c, CD11b, Gr-1, CD80, CD86, MHC class I, MCH class II, and CD40. Most M-MDSC cultured with particulate β-glucan for 7 d differentiated to F4/80^lowCD11c⁺CD11b⁺Gr-1⁻ cells and expressed CD86, CD80, MHC class II, MHC class I, and CD40 (Fig. 5A, 5B), and cells were viable (trypan blue exclusion) and 7-aminoactinomycin D⁻ (data not shown).

It has been proposed that MDSC can uptake, process, and present Ag, establishing a stable synapse with T cells required for Ag-specific T cell suppression (30). Freshly isolated splenic M-MDSC cultured with sorted OVA-specific CD4⁺ T cells in the presence of OVA did not induce CD4⁺ T cell proliferation or IFN-γ production (Fig. 5C). To test whether M-MDSC cocultured with particulate β-glucan for 7 d differentiated to potent APC, we cocultured M-MDSC with particulate β-glucan for 7 d and then cultured with sorted and CFSE-labeled OVA-specific CD4⁺ T cells in the presence of whole OVA Ag for 4–5 d. Differentiated M-MDSC induced CD4⁺ T cell proliferation and IFN-γ production (Fig. 5D). Moreover, differentiated M-MDSC cross-presented Ag to OVA-specific CD8⁺ T cells and immensely induced CD8⁺ T cell proliferation and molecules associated with effector functions such as IFN-γ and granzyme B (Fig. 5E). Taken together, these data clearly demonstrate that particulate β-glucan converts suppressive M-MDSC to potent APC that can promote Th1 differentiation and Ag cross-presentation to CD8⁺ effector T cells.

To exclude the possibility of any artifacts in M-MSDC differentiation that might be promoted by particulate β-glucan phagocytosis alone and to test whether it is directed by dectin-1 receptor, we performed Ag presentation assays with WT or dectin-1 KO M-MDSC pretreated with particulate β-glucan for 7 d. WT M-MDSC but not dectin-1 KO M-MDSC induced OVA-specific CD4⁺ and CD8⁺ T cell proliferation and IFN-γ production (Fig. 6A). Additionally, particulate β-glucan stimulation enhanced the phosphorylation of Syk, Akt, JNK, and Erk1/2 but not p38 (Fig. 6B). Surprisingly, STAT3 phosphorylation was enhanced after particulate β-glucan stimulation (Fig. 6B), which is different from PMN-MDSC (Fig. 4D). Pretreatment of M-MDSC with WGP in the presence of MEK1/2 inhibitor (PD98059) abrogated the acquired Ag-presenting capability of differentiated M-MDSC (Fig. 6C). Additionally, M-MDSC treated with particulate β-glucan for 7 d increased the expression of TNF-α, IL-12, iNOS, and IL-6 and decreased TGF-β compared with freshly sorted M-MDSC (Fig. 6D). Enhancement of TNF-α and IL-12 mRNA expression was completely abrogated in dectin-1 KO M-MDSC upon particulate β-glucan stimulation (data not shown).

Next, we examined whether particulate β-glucan treatment also promotes M-MDSC differentiation in vivo. M-MDSC sorted from tumors of C57BL/6 mice (CD45.2) were treated with or without WGP for overnight and then injected into tumors of congenic SJL mice (CD45.1). Tumors were excised 7 d after injection. As shown in Fig. 7A, transferred cells were readily seen in the tumors. WGP-treated M-MDSC had significantly more CD11c⁺ cells compared with untreated M-MDSC. Additionally, those CD11c⁺ cells expressed a high level of MHC class II molecules (Fig. 7B). They also coexpressed a low level of F4/80 (data not shown). To evaluate the effect of particulate β-glucan–treated M-MDSC on tumor development in vivo, we performed admixture experiments with M-MDSC/tumor cells. M-MDSC or M-MDSC treated with particulate β-glucan for 18 h were mixed with LLC cells at a 1:1 ratio and injected s.c. into C57BL/6 mice. Untreated M-MDSC induced enhanced tumor progression and growth compared with LLC alone (Fig. 7C). In contrast, particulate β-glucan–treated M-MDSC did not promote tumor growth. Mice implanted with LLC/particulate β-glucan–treated M-MDSC had significantly decreased tumor burden compared with LLC alone (Fig. 7C), suggesting that particulate β-glucan–treated M-MDSC stimulate an antitumor immune response. This effect was dependent on the dectin-1 receptor (Fig. 7D).

To imply the significance of the present findings into human patients, we conducted a particulate β-glucan clinical trial in patients with NSCLC that were newly diagnosed and had not received any other treatment such as chemotherapy. Particulate β-glucan was administered orally for 10–14 d at 500 mg dose and blood was withdrawn before and after treatment. Consistent with previous studies (8, 9), the frequency of HLA⁻CD14⁻CD33⁺CD11b⁺ MDSC was substantially increased in the peripheral blood of patients with NSCLC compared with those in age- and sex-matched healthy donors (Fig. 8A). Particulate β-glucan treatment significantly reduced the percentage of HLA-DR⁻CD14⁻CD33⁺CD11b⁺ MDSC in the peripheral blood of NSCLC patients when compared with its frequency in the peripheral blood before treatment (Fig. 8B).

Next, we sought to determine the Ag-presenting capability of CD14⁺HLA-DR⁺ monocytes and CD14⁻HLA-DR⁻CD11b⁺CD33⁺ MDSC in an MLR as previously described (31). Particulate β-glucan treatment induced an enhanced trend in proliferation and IFN-γ production by allogeneic T cells in both populations (Fig. 8C). Additionally, IFN-γ production by T cells from patients after treatment were also trending as enhanced although it did not reach statistically significance (data not shown).

In NSCLC, Arginase 1 expression was reported to be expressed by CD14⁻CD11b⁺CD33⁺SSC^high PMN-MDSC (9, 32). To assess the effect of particulate β-glucan treatment on the function of PMN-MDSC, we compared the expression of Arginase 1 mRNA in the polymorphonuclear neutrophils isolated from the peripheral blood of NSCLC patients before and after particulate β-glucan treatment. The expression of Arginase 1 mRNA was significantly decreased in a cohort of 15 patients, and it became comparable with Arginase 1 mRNA expression in healthy controls (Fig. 8D, upper panel), whereas it did not significantly change in the other cohort of 20 patients (Fig. 8D, lower panel). Overall, NSCLC patients have a decreased frequency of CD14⁻HLA-DR⁻CD11b⁺CD33⁺ MDSC with improved effector function after oral particulate β-glucan treatment.

In this study, we demonstrated that treatment with yeast-derived particulate β-glucan reduces tumor growth and differentially modulates PMN-MDSC and M-MDSC frequencies in tumor-bearing mice. This differential modulation was also reflected in its effect on the functions of both subsets. First, we demonstrated that particulate β-glucan reverses PMN-MDSC suppression, induces respiratory burst, and enhances apoptosis. On the second axis, particulate β-glucan skews the immature suppressive phenotype of M-MDSC toward a potent APC phenotype that drives the differentiation of Th1 CD4⁺ T cells and induces cytotoxic CD8⁺ T cells. On a third and most prevalent axis, particulate β-glucan was tested in patients newly diagnosed with NSCLC as an immunomodulator in cancer. To our knowledge, this is the first clinical trial with particulate β-glucan in patients newly diagnosed with NSCLC that have not yet been subjected to any other treatment, as chemotherapy significantly affects MDSC frequency and function. Importantly, note that particulate β-glucan is a natural compound with no reported off-target effects, with a low cost, and that can be administered orally.

Particulate β-glucan induced PMN-MDSC respiratory burst early after activation and largely enhanced PMN-MDSC apoptosis, which positively correlates with the decreased frequency of PMN-MDSC in tumors and spleens upon particulate β-glucan treatment. Although ROS have been implied as one of the main mechanisms for T cell suppression (33), which is regulated by STAT3 activation (34), other reports suggest that PMN-MDSC can acquire a tumor cytotoxic phenotype in the absence of TGF-β (35). This is in line with what we have previously shown that TGF-β mRNA expression is reduced in tumors of mice treated with particulate β-glucan (20). Additionally, activation of neutrophils with β-glucan enhanced neutrophil cytotoxicity against Candida albicans (36). Interestingly, STAT3 phosphorylation levels in PMN-MDSC decreased after dectin-1/Syk activation with particulate β-glucan and might have led to PMN-MDSC apoptosis, and this correlates with the findings that treatment of splenic Gr-1⁺CD11b⁺ from tumor-bearing mice with the JAK2/STAT3 inhibitor JSI-I24 in vitro for 7 d in the presence of GM-CSF and tumor-conditioned medium enhanced cell death compared with untreated cells (37).

Despite several studies reporting the antitumor effect of particulate β-glucan through the activation of innate immune cells and induction of Th1 T cell responses (20, 27, 38), little is known on the role of particulate β-glucan in the modulation of tumor-associated macrophages, MDSC, and regulatory T cells. It has been reported that particulate β-glucan induces the differentiation of M-MDSC to F4/80⁺CD11c⁺ cells with a decreased suppressive phenotype (21). However, this study showed that the differentiated F4/80⁺CD11c⁺ cells represent only 8–13% of the gated M-MDSC population cultured for 48 h in the presence of GM-CSF. However, Youn et al. (5) showed that GM-CSF alone could induce the expression of CD11c to ∼35% and F4/80 to ∼60% in splenic M-MDSC at day 3. Additionally, expression of F4/80 or CD11c surface markers does not necessarily imply a nonsuppressive phenotype, because it has been shown that MDSC do differentiate to suppressive F4/80⁺ cells in the tumor microenvironment (7, 39, 40). In our study, we showed that particulate β-glucan treatment induces differentiation of M-MDSC to potent APC both in vitro and in vivo. In vitro–differentiated M-MDSC by β-glucan stimulate Ag-specific Th1 and CD8 T cell responses. Furthermore, WGP-treated M-MDSC differentiate into CD11c⁺ cells with high MHC class II expression in vivo and induce reduced tumor growth, suggesting that the function of these cells would be skewed toward an antitumor phenotype. Our data thus provide likely a direct functional link between M-MDSC differentiation induced by particulate β-glucan treatment and reduced tumor growth.

We also demonstrated that dectin-1 receptor signaling is required for the acquisition of Ag-presenting function, because M-MDSC from dectin-1 KO mice treated with particulate β-glucan for 7 d did not acquire such capability. Although STAT3 phosphorylation has been shown to be a key mechanism for MDSC-mediated suppression (41), we found that activation of STAT3 in M-MDSC did not impact the reversal of suppression and acquisition of Ag-presenting phenotype, perhaps largely due to the shift in the cytokine profile toward an M1-like phenotype with enhanced expression of IL-12, TNF-α, iNOS, and inhibition of TGF-β, in addition to the upregulation of MHC class II, MHC class I, and the costimulatory markers such as CD86 and CD40. However, the Erk inhibitor completely abrogated the conversion of M-MDSC to APC mediated by particulate β-glucan.

Note, also, that several previous studies have shown that β-glucan–containing fungal cell wall extracts such as zymosan can promote regulatory innate immune responses and modulate autoimmunity (42, 43). However, most of those studies used zymosan as β-glucan preparation, and zymosan only contains ∼14% β-glucan. The particulate β-glucan (WGP) used in our study is a pharmaceutical grade and contains >99.9% β-glucan. Therefore, the data from those studies cannot fully ascribe to β-glucan’s effect. In fact, the zymosan-induced arthritis model has been widely used in the field (44). Furthermore, dose is another factor that may contribute to these seemingly contradictory data. In the study by Karumuthil-Melethil et al. (43), they concluded that a low dose of β-glucan induces regulatory innate immune responses.

It has been well documented that MDSC accumulate in different human cancers such as brain (45, 46), head and neck (47), breast (48), lung (49, 50), and others (reviewed in Refs. 10, 16, 17). In many of these studies, the accumulation of MDSC in patients was correlated with poor prognosis. In the present study, newly diagnosed patients with NSCLC were given particulate β-glucan orally for 10–14 d. Treated patients had a significant decrease in the percentage of CD14⁻HLA-DR⁻CD11b⁺CD33⁺ MDSC in the peripheral blood. Importantly, these patients were not exposed to any therapy during the administration of particulate β-glucan, which allowed the sole assessment of particulate β-glucan efficacy in patients. Additionally, an enhanced trend in T cell allogeneic responses to CD14⁺HLA-DR⁺ and CD14⁻HLA-DR⁻CD11b⁺CD33⁺ cells was observed after treatment. The administration of particulate β-glucan for only 10–14 d may not have been enough to modulate APC function in vivo; however, it was sufficient to reduce the percentages of CD14⁻HLA-DR⁻CD11b⁺CD33⁺ MDSC in the peripheral blood.

The expression of arginase I in PMN-MDSC has been reported in NSCLC patients and correlated with poor prognosis and CD8⁺ T cell suppression (9, 32). Interestingly, a cohort of 15 patients had a significant decreased expression of arginase I in polymorphonuclear neutrophils from the peripheral blood whereas no significant change was reported in the other 20 patients subjected to particulate β-glucan treatment. The patients with decreased arginase I after particulate β-glucan had a higher trend in arginase I expression compared with healthy controls before the treatment. Unlike the other patients, the level of arginase I was comparable to healthy controls, suggesting that patients’ variable responses to the treatment might have been due to different stages, tumor sizes, and progression. However, no correlation was found among those factors. This may be due to limited patient numbers in this trial. Nevertheless, this study, to our knowledge, provides a first insight toward introducing particulate β-glucan as an immunomodulatory therapy against solid tumors in humans. Because MDSC play a critical role in tumor-mediated immune evasion, the findings from the present study provide the rationale for a design to combine β-glucan treatment with other immunotherapeutic approaches such as cancer vaccines and immune checkpoint inhibitor therapies.

We thank M. Hall and A. Harper for recruiting human NSCLC patients. We also thank C. Worth for helping with operation of the MoFlo sorter. We thank Drs. J. Suttles, H. Shirwan, and S. Uriarte for constructive comments and suggestions for the study.

J.Y. declared a competing financial interest in Biothera who provided β-glucan for the study. The other authors have no financial conflicts of interest.