Differential prognostic impact of interleukin-34 mRNA expression and infiltrating immune cell composition in intrinsic breast cancer subtypes

Abstract

Interleukin-34 (IL-34) is a ligand for the CSF-1R and has also two additional receptors, PTPRZ1 and syndecan-1. IL-34 plays a role in innate immunity, inflammation, and cancer. However, the role of IL-34 in breast cancer is still ill-defined. We analyzed IL-34 mRNA expression in breast cancer cell lines and breast cancer patients and applied established computational approaches (CIBERSORT, ESTIMATE, TIMER, TCIA), to analyze gene expression data from The Cancer Genome Atlas (TCGA). Expression of IL-34 was associated with a favorable prognosis in luminal and HER2 but not basal breast cancer patients. Gene expression of CSF-1 and CSF-1R was strongly associated with myeloid cell infiltration, while we found no or only weak correlations between IL-34, PTPRZ1, syndecan-1 and myeloid cells. In vitro experiments showed that tyrosine phosphorylation of CSF-1R, ERK, and FAK and cell migration are differentially regulated by IL-34 and CSF-1 in breast cancer cell lines. Collectively, our data suggest that correlation of IL-34 gene expression with survival is dependent on the molecular breast cancer subtype. Furthermore, IL-34 is not associated with myeloid cell infiltration and directly regulates breast cancer cell migration and signaling.

Keywords: CSF-1; IL-34; PAM50 subclasses; breast cancer patients; gene expression.

Figure 1. IL-34 mRNA expression in normal tissue, cancerous tissue, and breast cancer cell lines

(A) RNA expression overview shows RNA-seq data from The Cancer Genome Atlas (TCGA). Datasets of normal and cancerous human tissues were obtained from the TCGA database. Boxplots show the distributions (median, spread and outliers) of the IL-34 mRNA levels (log2) by the RNAseq by Expectation-Maximization (RSEM) normalization across normal and cancerous tissue. (B) IL-34 mRNA expression across molecular subtypes of breast cancer cell lines and normal breast cell lines. IL-34 expression level reported as log2 values +/− SD according to the molecular subtype of cell lines; non-tu, non-tumorigenic cell lines.

Figure 2

(A) IL-34 mRNA expression in breast tumors. Association of relative IL-34 mRNA expression (log2) with established clinical and histopathological parameters was analyzed in breast tumors. Boxplots of IL-34 expression in patients with an age at breast cancer onset of <55 vs. ≥55 years (^*, p  =  0.002), in pre- vs. post-menopausal patients (^*, p  =  0.007), in pT1 vs. pT2–4 breast tumors, in grade 1-2 vs. 3 breast tumors, IL-34 expression in ductal vs. lobular tumor type, in normal breast tissue (non-tumorous; non-tu) vs indicated molecular breast cancer subtypes (^*, p = 0.005, ANOVA), in patients with the indicated tumor stages, in breast tumors from patients with a negative (neg, pN0) vs. a positive (pos, pN+) lymph node status, in estrogen receptor (ER) neg vs. pos tumors, in progesterone receptor (PR) neg vs. pos tumors, in HER2 neg vs. pos tumors, and in p53 mutant (mut) vs. wildtype (wt) tumors. Numbers in parentheses indicate the number of patients in each group. All p-values were determined via unpaired, two-sided t-tests except in molecular subtype and stage (ANOVA); neg, negative; pos, positive. Molecular subtype based on expression of the PAM50 gene set [71], determined with Affymetrix U133 Plus 2.0 GeneChips. (B) Association of IL-34 expression with survival of breast cancer patients. Kaplan–Meier analyses of the overall (left), disease-free (center) and metastasis-free survival (right) in breast cancer patients (n = 75) are shown. Patients were stratified into two groups according to IL-34 high and IL-34 low expression levels.

Figure 3. IL-34 expression and overall survival across molecular subtypes of breast cancer

(A) Kaplan–Meier plots of overall survival in IL-34-high and IL-34-low expressing TCGA BRCA patients (n = 1056). Median IL-34 expression was used as cut-off for group separation. Log rank tests were calculated. (B) IL-34 mRNA expression in normal breast tissue and in PAM50 breast cancer subclasses. The bar graph shows the IL-34 mRNA levels (log2) across normal breast tissue (n = 113) and the molecular subtypes of breast cancer (luminal A, n = 412; luminal B, n = 188; HER2-enriched, n = 64; basal, n = 140) of the TCGA BRCA dataset. Kruskal–Wallis and Dunn’s multiple comparison tests were calculated. Error bars indicate standard deviations. ^*, p < 0.05; ^**, p < 0.01; ^***, p < 0.001; ^****, p < 0.0001. (C) Kaplan–Meier plots of overall survival in IL-34-high and IL-34-low expressing TCGA BRCA patients stratified by PAM50 subclasses (luminal A, n = 412; luminal B, n = 188; HER2-enriched, n = 64; basal, n = 140). Median IL-34 expression was used as cut-off for group separation. Log rank tests were calculated.

Figure 4. Schematic representation of IL-34 and CSF-1 receptors in tumor tissue consisting of cancer cells, stromal cells, and immune cells

IL-34 and CSF-1 bind to CSF-1R. Heteromeric CSF-1/IL-34 can also form, which may differentially regulate activation/localization of CSF-1R. Additionally, IL-34 can bind to syndecan-1, which then regulates CSF-1R activity. Finally, IL-34 also binds to PTPRZ1.

Figure 5. Prognostic significance and tumor-specific changes of expression for the IL-34 system in breast cancer patients

(A) IL-34, CSF-1, CSF-1R, PTPRZ1, and syndecan-1 expression in TCGA breast cancer samples (n = 1102) vs. normal breast tissue (n = 113). Scatter plots show relative mRNA expression (log2). Each dot represents a single tissue sample. Black lines indicate means and standard deviations. P-values were determined via Mann–Whitney U tests (two-sided). (B) CSF-1, CSF-1R, PTPRZ1, and syndecan-1 mRNA expression in normal breast tissue and in PAM50 breast cancer subclasses. The bar graph shows the mRNA levels (log2) across normal breast tissue (n = 113) and the molecular subtypes of breast cancer (luminal A, n = 412; luminal B, n = 188; HER2-enriched, n = 64; basal, n = 140) of the TCGA BRCA dataset. Kruskal–Wallis and Dunn’s multiple comparison tests were calculated. Error bars indicate standard deviations. ^*, p < 0.05; ^**, p < 0.01; ^***, p < 0.001; ^****, p < 0.0001. (C) Kaplan–Meier plots of overall survival between CSF-1, CSF-1R, PTPRZ1, and SDC1 high and low expressing breast cancer patients (n = 1056) of the TCGA BRCA dataset. Median expression values were used as cut-offs for group separation. Log rank tests were calculated.

Figure 6. IL-34 ligand-receptor mRNA expression ratio

(A) Kaplan–Meier plots of overall survival between IL-34/CSF1R, IL-34/PTPRZ1, and IL-34/SDC1 mRNA ratio high and low breast cancer patients (n = 1056) of the TCGA BRCA dataset. Median expression values were used as cut-offs for group separation. Log rank tests were calculated. (B) IL-34/CSF1R, IL-34/PTPRZ1, and IL-34/SDC1 mRNA ratio median values in normal breast tissue and in PAM50 breast cancer subclasses. The bar graph shows the IL-34/receptor mRNA ratio across normal breast tissue (n = 113) and the molecular subtypes of breast cancer (luminal A, n = 412; luminal B, n = 188; HER2-enriched, n = 64; basal, n = 140) of the TCGA BRCA dataset. Kruskal–Wallis and Dunn’s multiple comparison tests were calculated. Error bars indicate standard deviations. ^*, p < 0.05; ^**, p < 0.01; ^***, p < 0.001; ^****, p < 0.0001.

Figure 7. Immune cell landscape of breast cancer compared with TCGA gene expression of IL-34, CSF-1, CSF-1R, PTPRZ1, and SDC1 (syndecan-1)

Scatter plots were generated using the online tool TIMER to identify different profiles of immune cells associated with investigated genes. Each dot represents a single tumor sample. (see also Table 5, Supplementary Table 1).

Figure 8. Association between mRNA expression of 5 genes and macrophage infiltration in breast tumors

(A) Scatter plots show the correlation between IL-34, CSF-1, CSF-1R, PTPRZ1 and SDC1 mRNA expression (log2 scale) and M0, M1, M2 and monocyte infiltration scores (obtained from TCIA database) in tumor samples from the TCGA BRCA dataset. Each circle represents a single tumor sample. Regression lines and confidence intervals are shown in red and grey, respectively. (B) Kaplan–Meier plots of overall survival between M1/M2 high and low infiltrated breast cancer patients (n = 1052) of the TCGA BRCA dataset. Median macrophage scores from TCIA database were used as cut-offs for group separation. Log rank tests were calculated.

Figure 9. IL-34 differentially regulates migration and signaling of human breast cancer cell lines

(A) Comparison of gene expression in MCF7, SK-BR-3, and MDA-MB-231 breast cancer cells with THP-1 macrophages. Graphs show results of qRT-PCR for IL-34, CSF-1, CSF-1R, PTPRZ1 and SDC1 performed on RNA from human MCF7, SK-BR-3 and MDA-MB-231 breast cancer cells as well as THP-1 macrophages. (B) Quantification of migrated MCF7, SK-BR-3 and MDA-MB-231 breast cancer cells from an in vitro migration assay are shown. Cells were either left unstimulated (control) or stimulated with IL-34, CSF-1 or pretreated with CSF1-R blocking antibody (antiCSF1R) prior to cytokine treatment. ^*, p < 0.05 vs. control;, p < 0.05 vs recCSF1. (C) Differential regulation of signaling upon IL-34 or CSF-1 treatment in human breast cancer cell lines. MCF7, SK-BR-3, and MDA-MB-231 breast cancer cells were stimulated with recombinant IL-34 or CSF-1 protein for the indicated times. Western blot images of indicated proteins in breast cancer cells are shown. p- indicates phosphorylated proteins.