TLR8 ligation induces apoptosis of monocytic myeloid‐derived suppressor cells

Abstract

Myeloid‐derived suppressor cells (MDSCs) accumulate in tumors and the peripheral blood of cancer patients and demonstrate cancer‐promoting activity across multiple tumor types. A limited number of agents are known to impact MDSC activity. TLR8 is expressed in myeloid cells. We investigated expression of TLR8 on MDSC and the effect of a TLR8 agonist, motolimod, on MDSC survival and function. TLR8 was highly expressed in monocytic MDSC (mMDSC) but absent in granulocytic MDSC (gMDSC). Treatment of human PBMC with motolimod reduced the levels of mMDSC in volunteers and cancer donors versus control (P < 0.001). Motolimod did not impact levels of gMDSC. The reduction of mMDSC was due to induced cell death by TLR8 ligation. Pretreatment of PBMC with a FAS neutralizing antibody inhibited motolimod‐induced reduction of mMDSC (P < 0.001). Finally, we demonstrated that mMDSC impeded IL‐2 secretion by CD3/CD28‐activated T cells; IL‐2 secretion was partially restored when cells were cocultured with motolimod (142 ± 36 pg/ml vs. 59 ± 13 pg/ml; P = 0.03). There is increasing evidence that MDSCs contribute to the progression of cancer by inhibiting tumor‐directed T cells. TLR8 agonists may synergize with cancer immunotherapeutic approaches to enhance the antitumor effects of the adaptive immune response.

Abbreviations

FL

fluorescence

gMDSC

granulocytic MDSC

Lin

Lineage

mDC

myeloid DC

MDSC

myeloid‐derived suppressor cell

mMDSC

monocytic MDSC

pDC

plasmacytoid DC

1. INTRODUCTION

Myeloid‐derived suppressor cells (MDSCs) demonstrate a variety of functions that promote tumor growth as well as inhibit immune cell activity. Specific subtypes of MDSCs, monocytic MDSC (mMDSC) and granulocytic MDSC (gMDSC), can prevent the egress of activated CD8 T cells into the tumor,1 downregulate TCR‐zeta chain thus limiting T‐cell activation,2 and impede T‐cell differentiation.3 Elevated levels of MDSC indicate a poor prognosis or predict disease progression in many types of cancers.4, 5, 6 High levels of MDSC in the peripheral blood of cancer patients prior to therapy is associated with a lack of or incomplete response to treatment for several cancers.7, 8 Elevated blood MDSC has also been shown to be associated with a lack of clinical response to immune therapy in cancer patients9, 10, 11 There are few agents known that impair MDSC function or survival, which could be used in the treatment of cancer.

MDSCs modulate the adaptive immune response and can be activated via TLRs, including TLR2 and TLR4.12, 13, 14 Signaling through TLR4, for example, causes MDSC to secrete IL‐10 and promote a Th2 immune phenotype.14 TLR signaling in MDSC is an important pathway resulting in control of inflammatory processes.15

Motolimod is a small molecule selective TLR8 agonist and has been shown to stimulate type I cytokine production from antigen‐presenting cells resulting in T‐cell activation.16 We questioned whether TLR8 was expressed in MDSCs and, if so, what effects TLR8 ligation would have on MDSC differentiation and function.

2. MATERIALS AND METHODS

2.1. Peripheral blood mononuclear cells

Frozen PBMCs from volunteer donors were obtained from Astarte Biologics (Seattle, WA).17 Patient PBMCs (melanoma, prostate, and colon cancers treated to complete remission) were obtained under an IRB‐approved protocol.18 Frozen PBMCs were thawed and washed before use.19

2.2. Cell subset isolation

PBMCs were stained with fluorochrome‐conjugated monoclonal antibodies. Cell subtypes were isolated by BD FACS Aria II cell sorter. Myeloid DCs (mDCs; Lineage (Lin)‐HLA‐DR+, CD11C+) and plasmacytoid DCs (pDCs; Lin‐HLA‐DR+, CD11c–, CD123+) and mMDSC (HLA‐DR–CD14+) and gMDSC (Lin‐HLA‐DR–CD33+) were isolated. mMDSC, monocytes (HLA‐DR+CD14+) and CD3+ T cells were also isolated.

2.3. TLR mRNA expression analysis

Total RNA was extracted from FACS‐sorted mDC, pDC, mMDSC, and gMDSC. mRNA expression of TLR7, TLR8, and TLR9 was analyzed and normalized to b‐actin/HPRT.20

2.4. Cell culture and treatment with TLR agonists

PBMCs (2 × 10⁶/ml) were cultured with 167 or 500 nM motolimod (VentiRx Pharmaceuticals, Seattle, WA, USA) or PBS for 24 h. In some assays, the cells were cultured with 0.5 μM and 2 μM of TLR7/8 agonist CL‐075, 10 μM and 50 μM of TLR7 agonist Imiquimod, and 50 nM and 100 nM of TLR 9 agonist CpG ODN2006 with PBS control (all Invivogen, San Diego, CA). Doses were optimized as previously reported.16

2.5. Flow cytometry analysis

Cells were stained with antibodies and analyzed as percentage of mMDSC and gMDSC in PBMC. Cells were also stained with an annexin V PE, 7‐AAD apoptosis detection kit (BD Biosciences, San Jose, CA). All annexin V+ and 7‐AAD+ cells were counted as dead. Cell surface death receptor FAS (CD95) was added and mean fluorescence (FL) of FAS expression was compared between control and treated mMDSC. CD69+ and FAS‐L+ cells percentage of CD3+CD4+ and CD3+CD8+ cells was compared between groups.

2.6. MACS and labeling of mMDSC

HLA‐DR–CD14+ cells were enriched using microbeads (Miltenyl Biotec, Auburn, CA) and labeled with 3 μM CFSE for 15 min at 37°C using a Vybrant CFDA SE Cell Tracer Kit (Invitrogen, Carlsbad, CA).21 Labeled mMDSCs were mixed with donor PBMC and cultured with PBS and 500 nM motolimod for 24 h. Both HLA‐DR–and HLA‐DR+CD14+ cells percentage of CFSE‐labeled cells was analyzed.

2.7. FAS blocking

PBMCs were cultured with 2 μg/ml of a FAS neutralizing antibody (clone ZB4; EMDMillipore, Billerica, MA) and isotype control for 1 h. Cultured cells were incubated with or without 500 nM motolimod for 20 h before analysis.

2.8. Cytokine analysis

FACS‐sorted CD3+ cells (2.5 × 10⁶/ml) were stimulated with CD3/CD28 beads (Life Technologies, Carlsbad, CA, USA) at a 5:1 ratio. FACS‐sorted mMDSCs were added to the activated CD3 T cells at 1:1 ratio ± 750 nM motolimod. Both CD3+ T cells and mMDSCs alone and with motolimod were used as control. FACS‐sorted HLA‐DR+CD14+ monocytes were also cultured with or without motolimod for 20 h and supernatants harvested. Quantification of secreted cytokines was performed using a cytokine human magnetic kit (Life Technologies) by Luminex (Qiagen) as per manufacturer's instructions and reported as mean of picogram per milliliter of duplicated measurements.22

2.9. Statistical analyses

GraphPad Prism version 6 was used (GraphPad Software, La Jolla, CA). Paired t test compares the differences between control and treated cells. Differences in cytokine levels among groups were analyzed using a Mann‐Whitney U test. P value < 0.05 was considered statistically significant.

3. RESULTS AND DISCUSSION

3.1. mMDSCs, but not gMDSCs, express TLR8 mRNA

We have previously shown that TLR8 is expressed on monocytes and mDCs and the TLR8 agonist motolimod induces these cells to produce IL‐12 and TNF‐α.16 We questioned whether TLR8 was expressed on MDSC, a heterogeneic population of cells that contain at least two subsets, monocytic and granulocytic, both of which have variable immunosuppressive effects.3 We found that TLR8 was highly expressed on the mMDSC subpopulation (HLA‐DR–CD14+) but absent on gMDSC (Lin‐HLA‐DR–CD33+) (Fig. 1A). Further, the levels of TLR8 expression were higher in mMDSC than in mDC. mMDSCs, as compared to gMDSCs, are uniquely capable of inhibiting T cells that have become activated. mMDSCs prevent CD44 expression on activated T cells and limit the movement of T cells from the blood to tissues.23 mMDSCs also suppress expression of CD162 reducing T‐cell adherence.23 We evaluated the expression of TLR7 and TLR9 on MDSC in the same donors. We found that TLR7 mRNA was not expressed on either mMDSC or gMDSC (Fig. 1B). The expression of TLR9 on mMDSC and gMDSC was variable among donors (Fig. 1C).

Figure 1.

mMDSCs, but not gMDSCs, express higher levels of TLR8 mRNA. mRNA expression of TLR8 (A), TLR7 (B), and TLR9 (C) on mDC, pDC, mMDSC, and gMDSC in three donors. Columns and bars represents mean (±SE) of duplicate measurements. mMDSC, white column; other cell types, black columns

3.2. Motolimod treatment reduces the levels of mMDSC in PBMC

TLR8 is expressed on mDCs, which are some of the most effective antigen‐presenting cells in the tumor microenvironment and capable of activating cancer‐specific T cells.24 TLR8 ligation stimulates mDC to markedly upregulate costimulatory and MHC molecules, secrete type I cytokines such as TNF‐α and IL‐12, and can even restore defective type I cytokine secretion in the DC of patients with HIV.16, 25, 26 In contrast to TLR8 effects on DC, we found in vitro treatment of PBMC with varying doses of motolimod significantly decreased the number of mMDSCs in a dose‐dependent manner; 167 nM motolimod reduced mMDSCs to 42.9 ± 10.3% (mean ± SE: n = 15) of PBS treatment. When the cells were treated with 500 nM motolimod, the mMDSC fraction was further decreased to 23.1 ± 4.2% (Table 1). The majority of mMDSCs coexpress CD33. The HLA‐DR–CD14+CD33+ mMDSC in PBMC was also significantly decreased to 44.6 ± 14.8% of control (mean ± SE, n = 15) in cells treated with 167 nM and 16.3 ± 3.7% in cells treated with 500 nM motolimod (Table 1). Neither dose of motolimod affected Lin‐HLA‐DR–CD33+ gMDSC. The majority of gMDSCs coexpressed CD11b. The Lin‐HLA‐DR–CD33+CD11b+ gMDSC was not impacted by the treatment (Table 1). Representative flow cytometric dot plots are shown in Supplementary Fig. 1. We next questioned whether we would observe similar effects in the PBMC of cancer patients. PBMCs derived from two patients with colon cancer, one with prostate cancer, and one with melanoma were similarly treated. The mMDSC percentage in PBMC was decreased in all patients treated with motolimod (Fig. 2A).

Table 1.

Motolimod treatment reduces the levels of mMDSC in PBMC

	167 nM Motolimod	500 nM Motolimod
	Mean ± SE	Mean ± SE
	(n = 15)	(n = 15)
HLA‐DR–CD14+ mMDSC % among PBMCs	42.9 ± 10.3	23.1 ± 4.2
HLA‐DR–CD14+ CD33+ mMDSC % among PBMCs	44.6 ± 14.8	16.3 ± 3.7
Lin‐HLA‐DR–CD33+ gMDSC % among PBMCs	75.5 ± 27.0	57.6 ± 12.6
Lin‐HLA‐DR–CD33+ CD11b+ gMDSC % among PBMCs	84.1 ± 29.3	57.0 ± 15.0

The numbers above show the percentage reduction of MDSC among PBMCs as compared to PBS control.

Figure 2.

Motolimod treatment initiates apoptosis of mMDSC in cancer patients. (A) Percentage deduction of mMDSC among PBMCs in patients treated with 167 nM and 500 nM motolimod as compared to PBS control. Each line represents data from the same patient at varying doses. Red line: melanoma; green lines: colon cancer; blue line, prostate cancer. (B) Cell death percentage of mMDSC from above 4 patients. Lines are the means of each experimental group. *P < 0.05

3.3. The reduction in mMDSC was specific for treatment with a TLR8 agonist

We examined the effects of a TLR7/8 agonist CL075, a TLR7 agonist Imiquimod, and a TLR9 agonist ODN2006 on mMDSC. Similar to motolimod, the HLA‐DR–CD14+ mMDSC in PBMC was significantly decreased in cells treated with 0.5 μM CL075 (mean ± SE: 33.4 ± 7.1% of PBS control p = 0.0000) and 2 μM CL075 (22.8 ± 5.8%; p = 0.0000; n = 10; Fig. 3A). However, both doses of Imiquimod (P > 0.05, Fig. 3B) and ODN2006 (P > 0.05, Fig. 3C) did not impact the level of the mMDSC population.

Figure 3.

Reduction in mMDSC is specific for treatment with a TLR8 agonist. Percentage change of mMDSC among PBMCs in 0.5 μM and 2 μM CL075 (A), 10 μM and 500 μM Imiquimod (B), and 50 nM and 100 nM ODN 2006 (C) treated PBMC versus PBS controls. Lines are the means of the experimental groups (n = 10 volunteer donors). ***P < 0.000

Our results were unexpected as recent studies suggested incubation of mMDSC with a TLR7/TLR8 agonist could elicit significant IL‐12 secretion and promote the differentiation of the cells into M1 macrophage over a 3‐day period.27 We did find that motolimod treatment could promote limited type I cytokine secretion in mMDSC (Supplementary Fig. 2A and B) but without any increase in HLA‐DR+ cells (P > 0.05), which would indicate a shift in phenotype (P > 0.05; Supplementary Fig. 2C). These varying effects may be due to distinct functional differences between TLR8 and TLR7 agonists.28 TLR8 is present only in mDCs while TLR7 agonists will activate both mDCs and pDCs and B cells, which may result in differential stimulation of T cells. Moreover, studies demonstrate that type I cytokine levels produced by monocytes stimulated with a specific TLR8 agonist are 100 times greater than levels induced by a TLR7 agonist.28 Type I cytokine release mediated by motolimod could lead to direct activation of both CD4 and CD8 T cells resulting in upregulation of CD69 as well as FASL on the T‐cell surface.

3.4. Motolimod induces apoptosis of mMDSC, in part, through FAS ligation

FAS+ MDSCs are susceptible to FAS‐mediated killing by activated T cells.29 Investigations in mice have demonstrated that activated FASL+ T cells induce apoptosis in FAS‐expressing MDSCs.30, 31 We questioned whether the reduction of mMDSC in human PBMC was secondary to the same phenomenon—induced cell death. The induction of cell death of mMDSC within PBMC was significantly increased in cells treated with 167 nM motolimod (mean ± SE of cell death percentage of mMDSC: PBS, 45 ± 7%; 167 nM, 71 ± 9%; P = 0.0111). When PBMCs were treated with 500 nM motolimod, mMDSC death was further increased (84 ± 6%, vs. PBS, P = 0.0001; vs. 167 nM, P = 0.0455; Fig. 4A). In contrast, cell death among total PBMC was no different from the control with both doses of motolimod (P > 0.05; Fig. 4B; representative staining, Supplementary Fig. 3A and B). Motolimod treatment induced similar levels of cell death of the mMDSC in cancer patients (Fig. 2B).

Figure 4.

Motolimod induces apoptosis of mMDSC, in part through FAS ligation. (A and B) Cell death percentage of mMDSC (A) and PBMCs (B) following treatment with PBS or motolimod in doses of 167 and 500 nM (n = 6 volunteer donors). (C and D) FAS expression in mMDSC population following treatment with PBS or 500 nM motolimod; (C) mean FL of FAS, n = 4 volunteer donors, and (D) representative overlay FLOW histograms of FAS expression (black, PBS; red, motolimod). (E and F) mMDSC percentage of PBMC (E) and cell death percentage of mMDSC (F) in PBS alone, with FAS blocking Ab or IgG1 isotype control, and in 500 nM motolimod alone, with FAS blocking Ab or IgG1 isotype control. Lines are the means of the experimental groups. *P < 0.05; **P < 0.01; ***P < 0.001

We have previously shown that motolimod does not directly activate T cells.16 Treatment with the agent, however, indirectly activated T cells resulting in significant expression of CD69 on both CD4 (P = 0.0148) and CD8 (P = 0.0044) T cells compared to cells treated with PBS (Supplementary Fig. 4A and B). FASL was also significantly upregulated on CD4 and CD8 T cells with motolimod treatment (P = 0.0056 and 0.0257, respectively, as compared to PBS; Supplementary Fig. 4C and D). Murine MDSCs express the death receptor FAS and induce apoptosis upon exposure to activated T cells.30 We measured the FAS (CD95) expression on human mMDSC after motolimod treatment. The mean FL of FAS on mMDSCs was significantly elevated after the treatment (mean ± SE of mean FL: PBS, 994 ± 81%; motolimod, 1917 ± 88, P = 0.0003; Fig. 4C; representative histograms, Fig. 4D). The mMDSC population among PBMC was not affected by FAS‐neutralizing antibody or isotype control alone as compared to the PBS control (P > 0.05). The motolimod‐dependent reduction of the mMDSC population could be inhibited by pretreatment of the cells with FAS antibody (Fig. 4E) (mean ± SE of mMDSC percentage PBMC: motolimod, 0.04 ± 0.01; with FAS antibody, 0.59 ± 0.03; P = 0.0007) and not inhibited with the isotype control (P > 0.05). Representative analyses are shown in Supplementary Fig. 3C for the various culture conditions. Similarly, death of mMDSC induced by motolimod was significantly reduced by FAS antibody (mean ± SE of cell death percentage of mMDSC: motolimod: 94.1 ± 1.1%; with FAS antibody, 53.5 ± 10.1%, P = 0.0354) but not by the isotype control (P > 0.05, Fig. 4F).

3.5. Motolimod‐dependent depletion of mMDSC partially restores T‐cell activation

As shown in Fig. 5, mMDSC significantly inhibited IL‐2 secretion by CD3+ cells stimulated with CD3/CD28 antibody (mean ± SE of IL‐2 (pg/ml): CD3/CD28 antibody alone, 3213 ± 1172 pg/ml; with mMDSC, 59 ± 13 pg/ml; P = 0.029). IL‐2 secretion was partially restored when the cells were cocultured with motolimod (142 ± 36; P = 0.03). Motolimod incubation did not directly stimulate IL‐2 secretion from CD3+ cells or mMDSCs (Fig. 5).

Figure 5.

Motolimod‐induced depletion of mMDSC partially restores T‐cell activation. IL‐2 secretion from CD3+ cells treated with media, motolimod alone; and CD3+ cells stimulated with CD3/CD28, mMDSC, and mMDSC and motolimod; and mMDSC treated with media and motolimod. Columns and bars represent means (±SE) of 4 donors. *P < 0.05

4. CONCLUSIONS

Data presented here demonstrate that TLR8 is expressed in mMDSCs and that selective TLR8 ligation can significantly reduce the levels of mMDSCs in the peripheral blood via the induction of apoptosis. Reduction in the number of mMDSCs in PBMC, mediated by TLR8 agonist treatment, will significantly enhance type I cytokine secretion by activated T cells. There is increasing evidence that MDSC contribute to the progression of cancer by inhibiting tumor‐directed T cells and producing mediators that promote the growth and survival of tumor cells.12, 32, 33 Motolimod has a favorable safety profile when used as an immunotherapeutic in patients with cancer demonstrating limited toxicity and no evidence of cytokine storm.34 One strategy would be to precondition patients with systemic infusion of motolimod to deplete MDSC prior to therapy, especially an immune‐based therapy. Reducing MDSC, while activating T cells, may synergize with cancer immunotherapeutic approaches and enhance the antitumor effects of the adaptive immune response.

Supporting information

Supplemental Figure 1.

Supplemental Figure 2.

Supplemental Figure 3.

Supplemental Figure 4.

Dang Y, Rutnam ZJ, Dietsch G, et al. TLR8 ligation induces apoptosis of monocytic myeloid‐derived suppressor cells. J Leukoc Biol 2018;103:157–164. 10.1002/JLB.5AB0217-070R