IRF5 suppresses metastasis through the regulation of tumor-derived extracellular vesicles and pre-metastatic niche formation

Abstract

Metastasis is driven by extensive cooperation between a tumor and its microenvironment, resulting in the adaptation of molecular mechanisms that evade the immune system and enable pre-metastatic niche (PMN) formation. Little is known of the tumor-intrinsic factors that regulate these mechanisms. Here we show that expression of the transcription factor interferon regulatory factor 5 (IRF5) in osteosarcoma (OS) and breast carcinoma (BC) clinically correlates with prolonged survival and decreased secretion of tumor-derived extracellular vesicles (t-dEVs). Conversely, loss of intra-tumoral IRF5 establishes a PMN that supports metastasis. Mechanistically, IRF5-positive tumor cells retain IRF5 transcripts within t-dEVs that contribute to altered composition, secretion, and trafficking of t-dEVs to sites of metastasis. Upon whole-body pre-conditioning with t-dEVs from IRF5-high or -low OS and BC cells, we found increased lung metastatic colonization that replicated findings from orthotopically implanted cancer cells. Collectively, our findings uncover a new role for IRF5 in cancer metastasis through its regulation of t-dEV programming of the PMN.

Figure 1

IRF5 is associated with better prognosis, decreased metastasis, and lower tumor grade in human OS. (a,b) Clinical and RNA-sequencing data from OS patients from the TARGET data matrix were stratified by IRF5 expression levels, metastases, and overall survival. Kaplan Meier curves were created for the bottom and top quartile of IRF5 expression via the UCSC Xena platform; n = 22 per group. (b) The top and bottom quartiles of IRF5 expression were compared for the presence of patients with metastasis at diagnosis or development of recurrence at a metastatic site via a Chi-square test (p = 0.0334), n = 22 per group (c,d) Representative immunofluorescence staining of IRF5 in a panel of human OS tumor samples and normal bone controls (c). Quantification of the proportion of samples staining positive for IRF5 was compared using a Chi-Square test. Normal bone N = 22, T1 N = 14, T1 Grade 2 N = 30, T1 Grade 3 N = 30 (d). *p < 0.05, **p < 0.01.

Figure 2

Murine intratumoral IRF5 is associated with decreased metastasis and predicts IRF5 transcript expression in t-dEVs (a) Representative Western blot of IRF5 expression in 4T1 (IRF5 low) and 4T1-IRF5 (IRF5-high) expressing cells, and K12 and K7M2 osteosarcoma cell lines. (b) Pulmonary metastatic burden differences between K12 and K7M2 implanted tumors, N = 6 per group. (c) Representative size distribution of EVs isolated from the four cell lines. t-dEVs are between 0 and 200 nm. (d) Representative images of Western blots for ALIX and CD63 from t-dEVs. (e,f) Quantitative differences in EVs released per cell in K7M2 versus K12 and 4T1 IRF5-high and -low. (g,h) Relative IRF5 transcript expression from qPCR analysis after normalization actin and to the cell with higher IRF5 expression (K12, 4T1 IRF5 high). Western blots are representative of 3 independent replicates. Full Western Blot images for (a,d) are attached in Supplemental Fig. 1. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 3

t-dEVs from OS and BC travel to the lungs in vivo, create a PMN, and increase metastatic burden. (a) Schema of in vivo EV tracking experiment. (b) Quantitation and representative image of total lung fluorescence after osteosarcoma EV injection. (c) Quantitation and representative image compared to other organs of total lung fluorescence after breast cancer EV injection. (d) Schema of EV pre-conditioning experiments. (e) Differences in metastatic burden of mice pre-conditioned with nothing, liposomes, K12 EVs, or K7M2 EVs prior to implantation of K7M2 tumors. (f) Representative H&E of metastatic burden in mice that were not pre-conditioned compared to K7M2 EV pre-conditioned mice. (g) Survival curves from the different EV pre-conditioning strategies over the time-period of our study, N = 8 per group. (h) Differences in metastatic burden of mice pre-conditioned with IRF5-high or IRF5-low EVs prior to IV injection of 4T1 cells. (i) Representative images of (h). (a,d) Created with Biorender. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 4

Immune cell reprogramming in the PMN is regulated by IRF5 t-dEVs. (a,b) Quantitation and representative flow cytometry gating of the CD11b^lo population during early stages of PMN formation from mice implanted with OS tumors (a) or BC implanted tumors (b). (c,d) Same as (a,b) except for CD11b⁺Ly6G⁻Ly6C^hi, or monocytic MDSCs during early stages of PMN formation from OS implanted tumors (c) or BC implanted tumors (d). (e,f) Quantitation and representative flow cytometry of CD11b⁺Ly6G⁺Ly6C^lo, or granulocytic MDSCs, during early stages of PMN formation from OS (e) and BC (f). (g,h) Representative flow histograms and quantitation of MFI for MHCII and CD206 during early stages of PMN formation in OS (g) and BC (h). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 5

EV packaged protein differs in IRF5-high and IRF5-low t-dEVs. (a) Differential expression of 2584 proteins expressed in IRF5-high and IRF5-low t-dEVs from 4T1 cells; red indicates differential expression with p > 0.05 of N = 3 individual replicates. (b) Representation of normalized enrichment scores of the GO Biological Process pathways differentially enriched in IRF5-high and IRF5-low t-dEVs with p < 0.001. (c) Normalized enrichment scores of the GO Molecular Function pathways differentially enriched in IRF5-high and IRF5-low EVs with p < 0.05. (d) Representative plots of Actin polymerization, Extracellular Matrix Structure, and Cytokine Receptor Binding showing enrichment in IRF5-low t-dEVs. (e) Representative plot of cytoplasmic ribosomal proteins showing enrichment in IRF5-high t-dEVs. (f) Relative mRNA expression of genes enriched in IRF5-low EVs for both BC and OS, VEGFc, EGFR, and CXCL5. (g) Relative mRNA expression of MMP9, enriched in IRF5-high EVs. (h) Relative mRNA expression of genes enriched in IRF5-low EVs from BC cells, but not enriched in OS t-dEVs, CAV1, PDGFa. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.