Monocytic and granulocytic myeloid derived suppressor cells differentially regulate spatiotemporal tumour plasticity during metastatic cascade

Abstract

It is widely accepted that dynamic and reversible tumour cell plasticity is required for metastasis, however, in vivo steps and molecular mechanisms are poorly elucidated. We demonstrate here that monocytic (mMDSC) and granulocytic (gMDSC) subsets of myeloid-derived suppressor cells infiltrate in the primary tumour and distant organs with different time kinetics and regulate spatiotemporal tumour plasticity. Using co-culture experiments and mouse transcriptome analyses in syngeneic mouse models, we provide evidence that tumour-infiltrated mMDSCs facilitate tumour cell dissemination from the primary site by inducing EMT/CSC phenotype. In contrast, pulmonary gMDSC infiltrates support the metastatic growth by reverting EMT/CSC phenotype and promoting tumour cell proliferation. Furthermore, lung-derived gMDSCs isolated from tumour-bearing animals enhance metastatic growth of already disseminated tumour cells. MDSC-induced 'metastatic gene signature' derived from murine syngeneic model predicts poor patient survival in the majority of human solid tumours. Thus spatiotemporal MDSC infiltration may have clinical implications in tumour progression.

Figure 1. Biochemical and functional properties of metastatic 4T1 and non-invasive EMT6 murine tumours.

(a) Luciferase-expressing 4T1 and EMT6 murine mammary tumour cells generate same size tumours in syngeneic BALB/c mice. Scale bar, 1 cm. (b–d) 4T1 tumour-bearing animals develop spontaneous pulmonary metastasis and show early pulmonary infiltration and enlarged spleen compared to the EMT6 tumour-bearing animals. Scale bar, 100 μm (lung sections); and 1 cm (gross organ pictures). (e,f) 4T1 tumour cells produce higher levels of inflammatory cytokines compared to the non-metastatic EMT6 or 67NR as assessed by cytokine antibody array. (g) Mouse transcriptome analyses of 4T1 and EMT6 tumours grown in BALB/c mice reveal 781 differentially expressed genes. (h) Top common genes that are upregulated in both EMT6 and 4T1 tumours. (i) Select genes (implicated in metastatic process) that are distinctly upregulated in 4T1 tumour compared to the EMT6 are. Results are presented as mean±s.d. (n=3). *P<0.05, **P<0.005, unpaired t-test.

Figure 2. Tumour-induced expansion and infiltration of MDSC subsets in BALB/c or C56BL/6J mice bearing murine tumours.

(a,b) 4T1 metastatic tumour in BALB/c mice compared to the less invasive EMT6 tumour induce a systemic induction and infiltration of mMDSCs (CD11b⁺Ly6C^hi) and gMDSCs (CD11b⁺Ly6C^int) in bone marrow (BM), spleen, lung and tumour. (c) A representative flow cytometry data of MDSC subsets is shown. (d,e) AT-3 murine tumour in syngeneic C56BL/6J mice induces the expansion and infiltration of mMDSCs and gMDSCs in BM, lung and tumour in similar manner. (f) GM-CSF and G-CSF potently induce the expansion of MDSCs under in vitro culture condition. (g,h) Expression of NOS2 and Arg1 in MDSCs from 4T1 tumour-bearing mice are verified by qPCR Results are presented as mean±s.d. (n=3). *P<0.05, **P<0.005, ***P<0.0005, unpaired t-test.

Figure 3. Monocytic-MDSCs localize at the invasive front of the primary tumour and induce epithelial–mesenchymal transition.

(a) Bright field (BF) images of co-cultures of EMT6 tumour cells (TC) (black arrow) with mMDSC (white arrows) or gMDSCs (red arrows) derived from 4T1 tumour-bearing mice. Morphological characteristic of EMT phenotype is visibly induced by the mMDSCs (top panel) which also show a strong affinity towards tumour cells. (b–e) mMDSC-induced EMT phenotype is assessed by enhanced expression of Vimentin and CK14. (f) Enhanced expressions of Vimentin and Twist induced by the mMDSCs from 4T1 tumour-bearing mice are verified by qPCR analyses. (g–j) mMDSCs from 4T1 tumour-bearing mice induce tumour cell invasion and CSC phenotype as shown by CD24-CD29 phenotype and tumour sphere forming assay. (k,l) In situ analyses of primary tumour and metastatic lesions by immunofluorescence staining reveal mMDSCs at the invasive edge of the primary tumour (TC, tumour center; TE, tumour edge) and gMDSCs around the pulmonary metastatic lesions in 4T1 tumour-bearing BALB/c mice (Met, metastasis; Inft, infiltrates). (m) Substantially higher number CD14-positive human mMDSCs were detected in metastatic human breast cancers compared to the indolent tumours. Results are presented as mean±s.d. (n=3). Scale bar, 50 μm; *P<0.05, **P<0.005; unpaired t-test.

Figure 4. Monocytic-MDSCs promote tumour cell invasion and CSC phenotype via upregulation of EMT-related genes.

(a) Mouse transcriptome analyses of MDSC subsets from 4T1 tumour-bearing mice reveal a distinct gene expression profile between mMDSCs and gMDSCs and (b) induce distinct gene signature in tumour cells when co-cultured with tumour cells. (c) EMT-related genes are upregulated in tumour cells when co-cultured with BM or tumour-derived mMDSCs from 4T1 tumour-bearing BALB/c mice. (d,e) Upregulation of the indicated genes was verified by qPCR showing several fold induction by mMDSCs compared to the gMDSCs. (f,g) These genes were also upregulated by BM or tumour-derived mMDSCs from AT-3 tumour model in C57BL/6J mice. (h,i) mMDSCs from both 4T1/BALB/c and AT-3/C57BL/6J mouse model stimulate strong activation of pStat1 and pStat3, overexpression of Vimentin and Twist1 while suppressing pERK1/2 activation in tumour cells upon co-culture. Results are presented as mean±s.d. (n=3). *P<0.05, **P<0.005, ***P<0.0005, unpaired t-test.

Figure 5. NOS2 production by MDSCs induce EMT/CSC phenotype.

(a) NOS2 induction by DPTA increased the expression of indicated cytokines and Vimentin in dose-dependent manner. (b) Inhibition of NOS2 by 1,400 W partially reduced the mMDSC-induced expression of IL1, IL6, TGFb and Vimentin. (c) NOS2 dependent pStat1 and pStat3 activation and EMT markers were confirmed by western blotting. Results are presented as mean±s.d. (n=3). *P<0.05, **P<0.005, unpaired t-test.

Figure 6. Lung-derived gMDSCs promote tumour cell proliferation via induction of a distinct gene expression signature called metastatic gene signature.

(a,b) Tumour cell proliferation was enhanced when they were co-cultured with lung- or tumour-derived gMDSCs compared to BM-derived mMDSCs and gMDSCs under serum-free culture condition. (c) There was also upregulation of EpCAM expression in tumour cells when they were co-cultured with lung-derived gMDSCs. (d,e) There are over 750 genes that are differentially regulated in tumour cells when co-cultured with gMDSCs either from BM or lungs of 4T1 tumour-bearing mice. (f) Top genes, highly upregulated in tumour cells upon co-culture with BM- or lung-derived gMDSCs are listed with their fold increase. We called 8 genes (indicated by red colour) ‘metastatic gene signature' since they were distinctly upregulated by lung-derived gMDSCs. (g–j) Upregulations of these genes as well as the proliferation marker PCNA were verified by qPCR in both 4T1 tumour-bearing BALB/c and AT-3 tumour-bearing C57BL/6J mouse models. Results are presented as mean±s.d. (n=3). Scale bar, 50 μm; *P<0.05, **P<0.005, ***P<0.0005 unpaired t-test.

Figure 7. Mouse metastatic gene signature correlates with proliferation cluster genes and predicts poor survival in human malignancies.

(a) TCGA (2015, Cell) data set (971 patients) shows overexpression and/or amplification of corresponding human genes; S100A8, S100A9, MMP8, CCL3, FPR1, TGFb2 predict poor survival in breast cancer patients. (b,c) Tumours with high expression/amplification of these genes were also enriched in the expression of PCNA as well as the proliferation cluster genes. (d,e) PCNA overexpression alone predicts poor survival, however, survival prediction improves by several multitudes when combined with signature expression. (f) Overexpression/amplification of metastatic gene signature also predicts poor survival in majority of solid tumours that are indicated in the figure. Results are presented as mean±s.d. P values (Log-rank test) are indicated in the figure.

Figure 8. Metastatic growth of EMT6-Luci tumour is enhanced by 4T1 tumour-secreted factors that modulate MDSC induction and infiltration.

(a–c) EMT6-Luci cells were either IV injected alone or in combination with a condition medium from 4T1 cells that lead to enhanced metastatic growth and induction of MDSC subsets. ***P<0.0005, two-way analysis of variance test (d) EMT6-Luci cells (50k per per injection) were either IV injected into naïve BALB/c mice or into the 4T1-primed mice (in which primary 4T1 tumours resected after 10 days) which resulted in enhanced metastatic growth. (e–g) Flow cytometry sorted tumour-derived mMDSCs (100k) or lung-derived gMDSCs (100k) from 4T1 tumour-bearing mice were co-injected with the 4T1-luci cells (50k per injection) and MDSC subsets injections were repeated 1 week later. (g,h) Animals injected with lung-derived gMDSCs showed accelerated metastatic growth and shortened survival compared to the control or mMDSCs co-injected group. Results are presented as mean±s.d. (5–10 mice in each group).

Figure 9. Depletion of gMDSCs suppress the pulmonary metastasis.

(a–c) Depletion of gMDSCs by anti-Ly6g antibody (200 μg/three times a week) in 4T1 tumour-injected mice resulted in suppression of pulmonary metastasis shown by optical imaging in life mice as well as ex vivo imaging of lungs. *P<0.05, ***P<0.0005, two-way analysis of variance test. (d) Reduced gMDSC infiltration in the lungs. Results are presented as mean±s.d. (5–10 mice in each group). *P<0.05, unpaired t-test.

Figure 10. Lung-derived gMDSCs induce metastatic growth of disseminated tumour cells.

(a,b) We developed a mouse model that shows no metastatic growth when primary tumours were resected at 1 week post implantation (c) despite the existence of disseminated tumour cells in regional lymph nodes and lungs. (d,e) All mice develop metastasis when primary tumours were resected at 2 weeks post implantation. (f) Illustration of the experimental design. (g) Primary 4T1-Luci tumours were resected after 1 week post implantation and mice were followed up for metastatic growth by bioluminescence imaging (BLI). There was no metastatic growth up to week 11. (h) After resection of primary tumours, mice were injected (via tail vein) with tumour-derived mMDSCs as indicated and followed up by BLI without any metastatic growth. (i,j) Mice injected with lung-derived gMDSCs showed metastatic growth in three out of four mice. (k) Our findings suggest a spatiotemporal regulation of tumour plasticity by MDSC subsets in primary site and in distant organs as illustrated. Results are presented as mean±s.d. (five mice in each group). Scale bar, 50 μm.