Tumor-derived extracellular vesicles regulate tumor-infiltrating regulatory T cells via the inhibitory immunoreceptor CD300a

Although tumor-infiltrating regulatory T (Treg) cells play a pivotal role in tumor immunity, how Treg cell activation are regulated in tumor microenvironments remains unclear. Here, we found that mice deficient in the inhibitory immunoreceptor CD300a on their dendritic cells (DCs) have increased numbers of Treg cells in tumors and greater tumor growth compared with wild-type mice after transplantation of B16 melanoma. Pharmacological impairment of extracellular vesicle (EV) release decreased Treg cell numbers in CD300a-deficient mice. Coculture of DCs with tumor-derived EV (TEV) induced the internalization of CD300a and the incorporation of EVs into endosomes, in which CD300a inhibited TEV-mediated TLR3–TRIF signaling for activation of the IFN-β-Treg cells axis. We also show that higher expression of CD300A was associated with decreased tumor-infiltrating Treg cells and longer survival time in patients with melanoma. Our findings reveal the role of TEV and CD300a on DCs in Treg cell activation in the tumor microenvironment.

The mouse CD300 family molecules, which are encoded by 9 genes on 67 chromosome 11, are expressed on myeloid cells including macrophages, dendritic cells, 68 mast cells and granulocytes and either activate or inhibit innate immune responses 69 (Borrego, 2013;Voss et al., 2015). On the other hand, the human CD300 family 70 consists of 7 molecules encoded by genes located on chromosome 17 in a region 71 syntenic to mouse chromosome 11 (Clark et al., 2001). CD300a, one of the CD300 72 molecules in mouse, contains an immunoreceptor tyrosine-based inhibitory motif in its 73 cytoplasmic portion. It mediates an inhibitory signal via SHP-1 and SHP-2 by binding 74 to phosphatidylserine, which is exposed on the outer leaflet of the plasma membrane on To address whether CD300a is involved in tumor immunity, wild-type and 88 CD300a-deficient (Cd300a -/-) mice were transplanted intradermally with B16 melanoma 89 cells. The Cd300a -/mice showed larger tumor volume and shorter survival than did 90 wild-type mice ( Fig. 1A and B), indicating that CD300a suppresses the development of 91 melanoma. In contrast, Rag-deficient (Rag1 -/-) and Rag1 -/-Cd300a -/mice showed 92 comparable levels of tumor development and survival after injection of B16 melanoma 93 cells ( Fig. 1C and D). These results indicate that the suppressive effect of CD300a on 94 melanoma development is dependent on the adaptive immune response. However, we 95 also observed that CD300a was not expressed on tumor-infiltrating lymphocytes but 96 was broadly expressed on myeloid cells, including populations of Ly6G + neutrophils, 97 CD11c +~high DCs, and CD11c low CD11b + macrophages ( Fig. S1A-C). These results 98 suggest that CD300a expressed on myeloid cells suppresses melanoma development via 99 adaptive immune responses. To identify the CD300a-expressing myeloid cell population 100 that is involved in melanoma suppression, we used Cd300a fl/fl Itgax-Cre and Cd300a fl/fl 101 Lys2-Cre mice. Cd300a fl/fl Itgax-Cre mice expressed CD300a on Ly6G + cells and 102 CD11ccells, but not on CD11c +~high cells (Fig. S1A). In contrast, Cd300a fl/fl Lys2-Cre 103 mice express CD300a on CD11c +~high cells and the subpopulation of CD11c low cells, but 104 not on Ly6G + cells (Fig. S1A). Although tumor growth was comparable between 105 Cd300a fl/fl Lys2-Cre and Cd300a fl/fl mice, Cd300a fl/fl Itgax-Cre mice showed greater 106 tumor volume than did Cd300a fl/fl mice (Fig. 1E). These data implicated CD300a on 107 DCs, rather than on neutrophils or macrophages, in inducing the adaptive immune 108 response to inhibit tumor development. 109 (A-E) Tumor growth and survival curves of wild-type (WT, n = 5 in A and B), Cd300a -/-(n = 5 in A and B), Rag1 -/-(n = 11 in C and n = 6 in D), Rag1 -/-Cd300a -/-(n = 11 in C and n = 6 in D), Cd300a fl/fl (n = 7), Cd300a fl/fl Itgax-Cre (n = 13), and Cd300a fl/fl Lys2-Cre mice (n = 15) that were inoculated with 1 × 10 5 B16 melanoma cells on day 0. Data are given as means ± SEMs. *P<0.05, **P<0.01 and ***P<0.001. P values were obtained by using a two-way ANOVA followed by Bonferroni's post-test (A and C) and the log-rank test (B and D). Data were pooled from two (A, B and C) or three (D and E) independent experiments.

CD300a regulates tumor-infiltrating Treg cells. 111
Previous reports have demonstrated that the number of Treg cells in melanoma is 112 immunohistochemistry. The Treg cell population was larger in the tumor, but not the 117 draining lymph nodes, of Cd300a -/mice compared with that of wild-type mice ( Fig. 2A  118 and B), whereas the tumor-infiltrating CD8 + T cells in Cd300a -/mice produced 119 significantly less IFN-g than did those in wild-type mice (Fig. 2C). Furthermore, PD-1 120 expression on tumor-infiltrating CD8 + T cells in Cd300a -/mice was significantly 121 upregulated compared to that in WT (Fig. 2D), suggesting that tumor-infiltrating CD8 + 122 T cells in Cd300a -/mice display more exhausted state as previously described (Sawant 123 et al., 2019). To determine whether Treg cells were indeed involved in the exacerbated 124 tumor growth of Cd300a -/mice, we depleted Treg cells by using an anti-CD25 125 monoclonal antibody (mAb) (Onizuka et al., 1999) (Fig. S2A). After Treg cell 126 depletion, the tumor volume of the Cd300a -/mice decreased to a level comparable to 127 that seen in wild-type mice (Fig. 2E). Likewise, the tumor volume decreased in 128 Cd300a fl/fl Itgax-Cre mice to a comparable level to that in Cd300a fl/fl mice after depletion 129 of Treg cells (Fig. 2F). These results suggest that CD300a on DCs regulates the number  (n = 6 in each group) (D) Representative histogram of PD-1 expression from tumorinfiltrating CD8 + T cells 3 weeks after tumor inoculation (left). The MFI of PD-1 is shown (right). (n = 4 in WT, n = 6 in Cd300a -/mice) (E and F) Tumor growth curve of WT mice (control mAb, n = 7; anti-CD25 mAb, n = 5) and Cd300a -/-(control mAb, n = 8; anti-CD25 mAb, n = 6) or Cd300a fl/fl (control mAb, n = 4; anti-CD25 mAb, n = 3) and Cd300a fl/fl Itgax-Cre (control mAb, n = 3; anti-CD25 mAb, n = 5) mice that were treated with a mAb to CD25 or a control antibody 3 times (Days -6, -3, and 0) and then inoculated with B16 melanoma cells. Data are given as means ± SEMs. N.S.; not significant. *P<0.05, **P<0.01 and ***P<0.001. P values were obtained by using a two-way ANOVA followed by Bonferroni's post-test (A, C, D) and the student's t-test (B and D). Data were pooled from two (B, D and E) or three (A, C and D) independent experiments.

Supplemental figure 2. Tumor growth in Cd300a -/mice is independent of the microbiota.
(A) Flow cytometric analysis of Foxp3 + cells in the spleen and inguinal lymph node (iLN) of mice injected with isotype mAb and 300 µg of anti-CD25 mAb on Days -6 and -3 prior to analysis. (B) Representative fluorescence micrographs of tumor sections from germ-free (GF) wild-type (WT) and Cd300a -/mice and stained with an anti-Foxp3 monoclonal antibody (green) and the DNA-binding dye DAPI (left). The number of Foxp3 + cells was quantified from 4 high-power fields (LPF) (right). White arrows show Foxp3-positive cells. Scale bar, 200 µm. (C) Comparison of tumor growth of B16 melanoma between GF WT (n = 5) and Cd300a -/mice (n = 6). Data are given as means ± SEMs. *P<0.05 and **P<0.01. P values were obtained by using the student's test (B) and a two-way ANOVA followed by Bonferroni's post-test (C). Data were pooled from two independent experiments (A-C).

development. 135
We previously reported that a microbiota-mediated signal induces increased IFN-b 136 production by DCs and increased numbers of Treg cells in the barrier tissues such as the 137 intestine, skin, and airway of Cd300a -/mice relative to those of wild-type mice 138 (Nakahashi-Oda et al., 2016). In the current study, we found that the expression of 139 Ifnb was also higher in DCs in the tumor tissues of Cd300a -/mice than in those of 140 wild-type mice (Fig. 3A). To examine whether the microbiota is also involved in Treg 141 cell levels in the tumor and tumor growth, we used wild-type and Cd300a -/mice raised 142 under the germ-free (GF) conditions. In contrast to the barrier tissues, Cd300a -/mice 143 still showed larger numbers of Treg cells and a larger tumor volume than did wild-type 144 mice raised under GF conditions ( Fig. S2B and C). These results suggest that, unlike in 145 the barrier tissues, the microbiota-mediated signal was dispensable for the increased 146 numbers of Treg cells in the tumor and for the enhanced tumor growth in Cd300a -/-147 mice. 148 Quantitative RT-PCR analysis of Ifnb in WT-and Cd300a -/--derived BMDCs that received no treatment (0 h, n = 7) or B16 culture supernatants (2.5 h, n = 5; 4.0 h, n = 7). (C) Quantitative RT-PCR analysis of Ifnb in WT-and Cd300a -/--derived BMDCs that received no treatment (-) (n = 6 in each group) and were treated with HMGB-1 (n = 3 in each group) or B16-derived extracellular vesicles (EVs, n = 5 in each group). (D) The number of induced Foxp3-eGFP + cells (iTreg) generated from naïve T cells by using anti-CD3, anti-CD28, IL-2 and TGF-b. These iTreg cells were cocultured with EV-stimulated BMDCs in the presence of IL-2 and TGF-b for 5 days with a control mAb (n = 7) or an anti-IFN-b mAb (n = 5). (E) Tumor growth curves of WT (PBS, n = 6; GW4869, n = 9) and Cd300a -/mice (PBS, n = 7; GW4869, n = 9) that were treated with GW4869 or PBS 3 times (Days 14, 18, and 21). (F) Representative fluorescence micrographs of tumor sections from Foxp3-eGFP WT (PBS, n = 4; GW4869, n = 6) and Foxp3-eGFP Cd300a -/mice (PBS, n = 5; GW4869, n = 6) in the absence or presence of GW4869, and stained with an anti-GFP monoclonal antibody (green) and the DNA-binding dye DAPI (left). The number of Foxp3 + cells was quantified from 4 high-power fields (LPF) (right). White arrow shows Foxp3positive cells. Scale bar, 200 µm. Data are given as means ± SEMs. RQ; relative quantification. N.S.; not significant. *P<0.05, **P<0.01 and ***P<0.001. P values were obtained by using a one-way ANOVA (D and F) and a two-way ANOVA followed by CD300a with human IgG1 (Fig. S3B). Stimulation with the purified EVs induced higher 166 Ifnb expression in Cd300a -/-BMDCs than in wild-type BMDCs (Fig. 3C). In contrast, 167 neither wild-type nor Cd300a -/-BMDCs expressed IFN-b after stimulation with a 168 damage associated molecular patterns (DAMPs) high mobility group box-1 protein 169 (HMGB-1) (Fig. 3C), which can be released by damaged tumors. These results suggest 170 that CD300a suppresses TEV-induced IFN-b production in DCs. To further analyze how CD300a regulates TEV-mediated IFN-b production in DCs, we 189 cocultured pHrodo-or PKH-labeled exosomes with wild-type or Cd300a -/-BMDCs and 190 analyzed the localization of the TEVs in BMDCs by using confocal laser scanning 191 microscopy. We found that the TEVs were incorporated into endosomes, as identified 192 by the expression of endosome antigen (EEA)-1, in both genotypes of DCs (Fig. 4A). 193 The number of TEVs in the endosomes was comparable between wild-type and 194 Cd300a -/-BMDCs (Fig. 4B), suggesting that CD300a did not affect TEV incorporation 195 into the endosomes. Interestingly, we also found that CD300a was internalized from the 196 cell surface into the endosomes, an event that might be mediated by the tyrosine-based 197 sorting motif in the cytoplasmic region of CD300a (Yotsumoto et al., 2003), after 198 coculture of BMDCs with TEVs ( Fig. 4C and Fig. S4A). As a result, the TEVs 199 colocalized with CD300a at the endosomes ( Fig. 4A and C). Given that EVs expose 200 phosphatidylserine on their lipid bilayer, which is a CD300a ligand, these results 201 suggest that CD300a was activated via stimulation with TEVs at the endosomes. 202 Ifnb expression in Cd300a -/-BMDCs to a level comparable to that in wild-type BMDCs 210 ( Fig. 5D and Fig. S4B). In contrast, the TLR4 inhibitor TKA-242 did not affect the 211 expression of Ifnb in either BMDC genotype ( Fig. 4D and Fig. S4B). These results 212 suggest that CD300a inhibits TLR3-mediated signaling for IFN-b production. Moreover, 213 the expression of Ifnb in ticam-1 -/-Cd300a -/-BMDCs was also decreased to the 214 comparable level of that in ticam-1 -/-BMDCs after coculture with TEVs (Fig. 4E). In 215 addition, we found that the phosphorylation level of interferon regulatory factor 3 216 (IRF3), a downstream molecule of the TRIF signaling pathway, was increased to a 217 greater extent in EV-stimulated Cd300a -/-BMDCs than in wild-type BMDCs (Fig. 4F). 218 In vivo analyses also showed that, although tumor growth was significantly larger and 219 the survival rate was significantly shorter for B16 melanoma-injected Myd88 -/-Cd300a -/-220 mice compared with B16-injected Myd88 -/mice, tumor development and survival did 221 not differ between ticam-1 -/-Cd300a -/and ticam-1 -/mice (Fig. 4G-J). Taken together, 222 these data suggest that CD300a inhibits the TLR3-TRIF signaling pathway for IFN-b 223 production at the endosomes in DCs, resulting in the suppression of Treg cell activation 224 and tumor development.

CD300A expression associates with survival times in melanoma patients. 226
To examine the role of CD300A in tumor development in humans, we analyzed the data 227 on the single-cell RNA sequence of human melanoma tissues, which demonstrated that 228 CD300A is expressed on populations that express HLA-DR, ITGAX (CD11c), ITGAM 229 (CD11b), CD14, and CD163 (Fig. S5), consistent with the results of mouse melanoma. 230 We further analyzed the database of the Cancer Genome Atlas (TCGA) project and 231 found that skin cutaneous melanoma patients (SKCM) expressing low levels of 232 CD300A mRNA had shorter survival times than did those expressing higher CD300A 233 mRNA levels (Fig. 5A). We also found that the expression ratio of CD300A to ITGAX 234 is negatively correlated with that of FOXP3 to CD8A (Fig. 5B). These results suggested 235 that CD300A suppressed Treg cell proliferation and/or activation and tumor 236 development. Moreover, we found that patients with melanoma showed strong positive 237 correlation between FOXP3 and IFNB1 expression (Fig. 5C). Neutral   Mabs to CD63 (NVG-2), CD103 (2E7), XCR1 (ZET) were purchased from Biolegend. 323 Anti-IFN-b (7F-D3) was from Yamasa; control rat IgG (6130-01) was purchased from 324 Southern Biotechnology. Anti-PS antibody (1H6) was purchased from Merck Millipore. 325 The CD300a-specific mAb (EX42) was generated in our laboratory. Anti-CD25 (PC61) 326 was a gift from E. Nakayama (Okayama University). Cells were treated for 10 min with which is the phosphatidylserine receptors, were added to the culture medium of B16 400 melanoma containing 2 mM CaCl2, and the mixture was rotated for 3 h or overnight at 401 4 °C. The beads were washed three times with washing buffer and exosomes were 402 eluted with elution buffer (Fig. S3A and B)

Statistical analyses 452
Comparisons were performed using GraphPad Prism version 5.0 (GraphPad Software) 453 by one-way or two-way ANOVA, followed by Bonferroni's multiple comparisons test 454 or Student's unpaired t-test. Data are presented as means ± SEMs, and differences are 455