Targeting myeloid derived suppressor cells reverts immune suppression and sensitizes BRAF-mutant papillary thyroid cancer to MAPK inhibitors

Abstract

MAPK signaling inhibitor (MAPKi) therapies show limited efficacy for advanced thyroid cancers despite constitutive activation of the signaling correlates with disease recurrence and persistence. Understanding how BRAF pathway stimulates tumorigenesis could lead to new therapeutic targets. Here, through genetic and pathological approaches, we demonstrate that BRAF^V600E promotes thyroid cancer development by increasing myeloid-derived suppressor cells (MDSCs) penetrance. This BRAF^V600E-induced immune suppression involves re-activation of the developmental factor TBX3, which in turn up-regulates CXCR2 ligands in a TLR2-NFκB dependent manner, leading to MDSCs recruitment into the tumor microenvironment. CXCR2 inhibition or MDSCs repression improves MAPKi therapy effect. Clinically, high TBX3 expression correlates with BRAF^V600E mutation and increased CXCR2 ligands, along with abundant MDSCs infiltration. Thus, our study uncovers a BRAF^V600E-TBX3-CXCLs-MDSCs axis that guides patient stratification and could be targeted to improve the efficacy of MAPKi therapy in advanced thyroid cancer patients.

Fig. 1. Tbx3 is necessary for Braf^V600E-induced PTC formation.

a Representative images of TBX3-GFP positive cancer cells analyzed by GFP IF staining in mPTC/Tbx3^G/+ compared with TPO-cre; Tbx3^G/+ littermates at 5w. Scale bars, 50μm. b The expression of Tbx3 was measured by RT-qPCR in thyroid glands from wild-type (WT) and mPTC littermates at 5w, GAPDH was used as the loading control, n = 11 pairs of littermates. c The expression of TBX3 was measured by western blot with densitometric analyses in thyroid glands from WT and mPTC littermates at 5w, n = 12 pairs of littermates. d The image representing whole thyroid tissues from 2 pairs of WT, mPTC, mPTC/Tbx3^+/− and mPTC/Tbx3^−/− mice at 5w, and related thyroid weight was plotted, n = 58 mPTC, n = 95 mPTC/Tbx3^+/−, n = 59 mPTC/Tbx3^−/−. e Representative H&E and Ki67 IHC staining on thyroid tissues from WT, mPTC, and mPTC/Tbx3^−/− littermates, red arrows points to papillary formations. Scale bars, 50μm. f Mouse weight of mPTC, mPTC/Tbx3^+/−, and mPTC/Tbx3^−/− mice at 5w, n = 45 mPTC, n = 97 mPTC/Tbx3^+/−, n = 60 mPTC/Tbx3^−/−. g Survival of mice bearing mPTC, mPTC-Tbx3^+/− or mPTC-Tbx3^−/−, n = 44 mPTC, n = 48 mPTC/Tbx3^+/−, n = 48 mPTC/Tbx3^−/−. h RT-qPCR analysis of Tbx3 in thyroid tissues from TPO-creER; Braf^V600ECA with intraperitoneal tamoxifen injection at 1 m (mPTC-TAM) or oil treatment as control (mPTC-CON), each pair of tissue was obtained at same time, n = 9. i Representative H&E and Ki67 IHC staining on thyroid tissues from mPTC-CON, mPTC-TAM or mPTC-TAM/Tbx3^−/−. Scale bars, 50μm. j Representative images of whole thyroid tissues from mPTC-CON, mPTC-TAM or mPTC-TAM/Tbx3^−/− at 8 m after intraperitoneal tamoxifen injection, n = 5 WT, n = 10 mPTC-CON, n = 21 mPTC-TAM, n = 20 mPTC-TAM/Tbx3^−/−. Thyroid tissue weights were analyzed. n = 3 biological independent samples (a, e, i). Data are shown as the mean ± s.d. (b, d, f, h, j). P values were calculated by unpaired two-tailed Student’s t test (b, d, f, h, j) or Logrank (Mantel-Cox test) (g). Uncropped immunoblots and statistical source data are provided in Source Data.

Fig. 2. BRAF/MAPK/AP-1 axis promotes TBX3 transcription in PTC.

a, b Western blot of TBX3 in PTC or normal thyroid cells (Nthy-ori-3-1) with BRAF or BRAF^V600E over-expression (a) or K1 cells with BRAF knock-down (b). c K1 cells were treated with BRAF inhibitor PLX4032 or ERK1/2 inhibitor SCH772984 for 24 h, the level of TBX3 was analysis by western blot. d TBX3 and AP-1 factors levels in K1 cells over-expressing c-Jun, JunB, or c-Fos, and treated with PLX4032 or SCH772984. e TBX3 and AP-1 factors levels in K1 cells with BRAF^V600E over-expression and AP-1 knock-down. f AP-1 binding motifs were analyzed through TBX3 promoter region using Jasper database. g Construction of TBX3 promoter GLuc with potential AP-1-binding sites. The predicted -358bp and -362bp site were overlapped as -360bp site. Arrows represented primers were used for qPCR. h, i Promoter GLuc activities were analyzed in HEK293T cells co-transfected with TBX3 truncated promoter (h) or TBX3-149-MUT promoter (i) and AP-1 expression constructs. j Quantitative ChIP (qChIP) analysis with anti-IgG, c-Jun, JunB and c-Fos for -149 site in K1 cells, primers targeting GAPDH gene were used as control. k Luciferase activities of GLuc-149 were analyzed in K1 cells treated with PLX4032 or SCH772984 for 12 h, 24 h or 36 h. l THCA datasets were downloaded from TCGA database to analyze the correlation between TBX3 and BRAF, n = 501. Densitometric analyses of western blot were shown (a–e). Two independent experiments were carried out with similar results for each kind of cells (a–c), and n = 3 biological independent samples (d, e, h–k). Data are shown as the mean ± s.d. (h–k). P values were calculated by unpaired two-tailed Student’s t test (h–k). Uncropped immunoblots and statistical source data are provided in Source Data.

Fig. 3. Loss of TBX3 leads to CXCR2 ligands reduction.

a Heatmap representation of DEGs (fold change >1.5; P < 0.05) in tumor tissues from mPTC and mPTC/Tbx3^−/−. The right panel showed the KEGG pathway analysis of DEGs. Data were analyzed using DAVID 6.8 website. b Venn diagram analysis across two groups of DEGs from mPTC tissues versus human K1 cells due to TBX3 loss. The heatmap of regulated genes was analyzed according to venn diagram, the upper for mouse and the lower for human. Differentially expressed human genes: 706 up-regulated genes and 1004 down-regulated genes. c The expression of genes from mPTC RNA-seq was confirmed by RT-qPCR. d, e RT-qPCR confirmation of inflammatory factors among DEGs from mPTC tissue (d) or human K1 cells (e) RNA-seq. f Cytokine antibody array on cell culture media from K1 cells with TBX3 knock-down. g ELISA analysis of CXCL1, 2 and CXCL8 in cell culture media from K1 or TPC1 cells with TBX3 knock-down. h RT-qPCR analysis of CXCL1, 2 and CXCL8 in K1 and Nthy cells with TBX3 over-expression. i IHC staining of CXCL1 and CXCL2 in mPTC, mPTC/Tbx3^+/− and mPTC/Tbx3^−/−. Scale bars, 50μm. n = 3 biological independent samples (c–e, g–i). Data are shown as the mean ± s.d. (c–e, g, h). P values were calculated by unpaired two-tailed Student’s t test (c–e, g–h). Uncropped immunoblots and statistical source data are provided in Source Data.

Fig. 4. TBX3/TLR2-NF-κΒ axis induces chemokines expression.

a CXCL1 mRNA level in TBX3 over-expressed Nthy cells treated with small-molecule inhibitors against IKKβ (LY2409881 (LY), TPCA-1), CREB (KG-501), AP-1 (T-5224) or STAT3 (NSC74859 (NSC)) was measured by RT-qPCR, comparison between DMSO- and inhibitor-treated TBX3-over-expressed Nthy cells was used for statistical analyses. b RT-qPCR analysis of CXCL1 in TBX3 over-expressed Nthy cells infected with shRNAs against IKBKB or RELA, comparison between shctrl- and shRNA- TBX3-over-expressed Nthy cells was used for statistical analyses. c Western blot of NF-κB pathway members in K1, TPC1 with TBX3 knock-down, and Nthy cells with TBX3 over-expression. d Venn diagram analysis across two groups of genes. Down-regulated, shTBX3 versus control in K1 cells (n = 1004); Up-regulated, TBX3 over-expression versus control in 21NT breast cancer cells (n = 1410). e RT-qPCR analysis of TLR2 in K1 cells with TBX3 knock-down or over-expression. f RT-qPCR analysis of CXCL1, 2 and CXCL8 in TBX3 over-expressed K1 infected with shRNAs against TLR2. g Western blot of NF-κB pathway members in K1 cells as in (f). h HEK293T cells were co-transfected with TLR2-GLuc, TBX3 and TBX3 with activator domain deletion (TBX3^ΔA) assayed for GLuc activity using SEAP activity as control. i qChIP analysis using anti-IgG and anti-FLAG antibody was performed in K1 cells with Flag-TBX3 over-expression. j qChIP analysis using anti-IgG and anti-TBX3 antibody was performed in K1 cells with TBX3 knock-down. Densitometric analyses of western blot were shown (c, g). n = 3 biological independent samples (a, b, e–j), and two independent experiments were carried out with similar results for each kind of cells (c). Data are shown as the mean ± s.d. (a, b, e, f, h–j). P values were calculated by unpaired two-tailed Student’s t test (a, b, e, f, h, j). Uncropped immunoblots and statistical source data are provided in Source Data.

Fig. 5. CXCR2 ligands function downstream of TBX3 and promote tumor cell proliferation.

a, b PTC cells were infected with specific shRNA against CXCL1/2 or CXCL8, and subjected to CCK8 (a) or colony formation assay (b). c, d TBX3 knock-down PTC cells with or without CXCLs over-expression were subjected to cell growth assays of CCK8 (c) or colony formation (d). e K1 cells with TBX3 knock-down and Vector or CXCLs over-expression were transplanted into athymic mice subcutaneously. Tumor weights as well as curves were generated and statistically compared, n = 6. f HE and IHC staining of indicated factors were performed on tumor sections in (e). Scale bars, 50μm. n = 3 biological independent samples (a–d). Data are shown as the mean ± s.d. (a–e). P values were calculated by unpaired two-tailed Student’s t test (a–e). Statistical source data are provided in Source Data.

Fig. 6. TBX3 deficiency blocked tumor formation due to suppressed CXCR2-mediated MDSCs recruitment.

a Flow cytometric analysis of MDSCs (CD45⁺CD11b⁺Gr-1⁺) in thyroid glands from mPTC (n = 14), mPTC/Tbx3^+/− (n = 17) or mPTC/Tbx3^−/− (n = 11) littermates at 5w. b, c Percentage of G-MDSC (CD45⁺CD11b⁺Ly6C^-Ly6G⁺) and M-MDSC (CD45⁺CD11b⁺Ly6C⁺Ly6G^-) of CD45⁺ tumor-infiltrating leukocytes (TILs) in mPTC tumors compared with mPTC/Tbx3^−/− littermates at (b). Relative CD8⁺ T cells/MDSCs was plotted (c), n = 4. d IHC analysis of CD11b, MDSCs (Gr-1, S100A8), CD4⁺ T cells (CD4) and CD8⁺ T cells (CD8) in thyroid glands from mPTC/Tbx3^−/− compared with mPTC littermates. Scale bars, 50μm. e, f Flow cytometric analysis of G-MDSC and M-MDSC in mPTC-TAM or mPTC-TAM/Tbx3^−/− tumors induced with tamoxifen for 8 m. CD8⁺ T cells/MDSCs was plotted, n = 4. g Mice bearing mPTC and mPTC/Tbx3^−/− tumors were treated with 200 μg anti-CD8 antibody per mouse twice weekly at age of 3w for 20d. Tumor weights were counted, n = 6. h Mice bearing tumors received daily oral doses of PLX4032 10 mg/kg body weight or SB265610 2 mg/kg body weight at age of 3w for 20d. Tumor weights were counted, n = 7. i, j Flow cytometric analysis of mPTC tumors underwent above treatments. Percentage of positive cells relative to total cells was plotted. MDSCs (i) and CD8⁺ T cells/MDSCs (j), n = 3. A representative of three independent experiments was shown (d). Data are shown as the mean ± s.d. (a–c, e–j). P values were calculated by unpaired two-tailed Student’s t test (a–c, e–j). Statistical source data are provided in Source Data.

Fig. 7. Antagonizing MDSCs inhibits Braf^V600E-induced immune-suppression and tumor advancement.

a Percentage of tumor-infiltrating myeloid immune cells (CD45⁺CD11b⁺) of CD45⁺ TILs in mPTC at 5w compared with wild-type littermates, analyzed by FlowJo, n = 3. b Quantification of whole leukocyte populations in thyroid glands of mPTC at 1 m, 2 m and 3 m by FlowJo, n = 4. c Percentage of G-MDSCs and M-MDSCs of CD45⁺ TILs in mPTC tumors at 1 m and 3 m, n = 4. d Relative CD8⁺ T cells/G-MDSCs of mPTC at 1 m or 3 m was plotted, n = 4. e IHC analysis of CD11b, Gr-1, S100A8, CD4 and CD8. Scale bars, 50μm. f Quantification of whole leukocyte populations in thyroid glands of mPTC-TAM induced with tamoxifen for 6 m and 12 m, n = 5. g, h FlowJo analysis of G-MDSCs, M-MDSCs, CD4⁺ T cells and CD8⁺ T cells of CD45⁺ TILs in mPTC-TAM tumors induced with tamoxifen for 6 m and 12 m (g), CD8⁺ T cells/G-MDSCs was plotted (h), n = 5. i Mice bearing tumors were treated with anti-IgG antibody or anti-Gr-1 antibody three times a week at age of 3w for 20d, and received daily oral doses of PLX4032 or vehicle for 20d. Tumor volumes and tumor weights were counted, n = 7. j Flow cytometric analysis of tumors underwent above treatments. Percentage of positive cells relative to total cells was plotted. MDSCs (upper) and CD8⁺ T cells/MDSCs (lower), n = 4. k IHC staining in four groups of mPTC from (i). Scale bars, 50μm. A representative of three independent experiments was shown (e, k). Data are shown as the mean ± s.d. (a, c, d, g–j). P values were calculated by unpaired two-tailed Student’s t test (a, c, d, g–j). Statistical source data are provided in Source Data.

Fig. 8. BRAF^V600E promotes CXCLs level and MDSCs infiltration through TBX3-regulated pathway.

a Levels of CXCLs in PTC cells underwent PLX4032 or SCH772984 treatment were checked by RT-qPCR. b Levels of CXCLs in K1 cells stably over-expressing control Vector or BRAF^V600E and subsequently infected with lentiviruses expressing TBX3 shRNAs were analyzed by RT-qPCR. c, d IHC staining in PTC samples with BRAF^WT or BRAF^V600E paired with adjacent normal tissues. Representative images were shown (c), and the scores of the stained sections were plotted (d), n = 30 normal tissues, n = 66 grade low PTC tissues, and n = 44 grade high PTC tissues. Scale bars, 100μm. e Scatter diagrams of TBX3, CXCL1, CXCL2 and CXCL8 protein levels in BRAF^WT (n = 45) and BRAF^V600E (n = 65) PTC tissues. f A proposed model of the mechanism of re-activation of TBX3 by BRAF/MAPK constitutive activation leading to immune-suppression by recruiting MDSCs into tumor microenvironment and the rational of combination therapy using PLX4032 and SB265610 or anti-Gr-1 to treat thyroid cancer. n = 3 biological independent samples (a, b). Data are shown as the mean ± s.d. (a, b, d, e). P values were calculated by unpaired two-tailed Student’s t test (a, b, d, e). Statistical source data are provided in Source Data.