CECR2 drives breast cancer metastasis by promoting NF-κB signaling and macrophage-mediated immune suppression

Abstract

Metastasis is the major cause of cancer-related deaths due to the lack of effective therapies. Emerging evidence suggests that certain epigenetic and transcriptional regulators drive cancer metastasis and could be targeted for metastasis treatment. To identify epigenetic regulators of breast cancer metastasis, we profiled the transcriptomes of matched pairs of primary breast tumors and metastases from human patients. We found that distant metastases are more immune inert with increased M2 macrophages compared to their matched primary tumors. The acetyl-lysine reader, cat eye syndrome chromosome region candidate 2 (CECR2), was the top up-regulated epigenetic regulator in metastases associated with an increased abundance of M2 macrophages and worse metastasis-free survival. CECR2 was required for breast cancer metastasis in multiple mouse models, with more profound effect in the immunocompetent setting. Mechanistically, the nuclear factor κB (NF-κB) family member v-rel avian reticuloendotheliosis viral oncogene homolog A (RELA) recruits CECR2 to increase chromatin accessibility and activate the expression of their target genes. These target genes include multiple metastasis-promoting genes, such as TNC, MMP2, and VEGFA, and cytokine genes CSF1 and CXCL1, which are critical for immunosuppression at metastatic sites. Consistent with these results, pharmacological inhibition of CECR2 bromodomain impeded NF-κB-mediated immune suppression by macrophages and inhibited breast cancer metastasis. These results reveal that targeting CECR2 may be a strategy to treat metastatic breast cancer.

Fig. 1.. Immune-related gene signatures differ between metastatic and primary breast cancer.

(A) Matched primary tumors and distal metastases from 13 breast cancer patients were collected and deregulated genes were analyzed by comparing distal metastases with matched primary tumors using RNA-sequencing (RNA-seq) analysis. (B) A heat map shows the expression of representative immune genes of tolerance mechanisms in 13 pairs of primary (blue) and matched metastatic (red) breast cancer tumor samples. HER2 and ER status are shown, “+” means the status is positive and “−” means the status is negative. na, not assessed. means the absence of patient information. (C) Tumor infiltrating lymphocyte- and macrophage-related gene expression was compared in matched pairs of metastatic and primary breast tumor samples. Orange lines mark the samples with increased expression in metastases and blue lines mark the ones with decreased expression. (D) A volcano plot of downregulated immune-oncology targets is shown for matched metastatic samples compared with primary breast tumors. Red dots denote the significantly changed targets. (E) RNA-seq data of matched primary tumor and distal metastases from 13 breast cancer patients were analyzed by CIBERSORTx and immune cell composition of complex tissues were characterized from their gene expression profiles. Populations of M1 macrophages and the ratio of M2 macrophages to total macrophages in primary and matched metastatic breast cancer samples are shown. Orange lines mark the samples with increased numbers in metastases and blue lines mark the ones with decreased numbers). The p-values were obtained using DESeq2 analysis of the counts (C and D) and Wilcoxon signed rank test (E).

Fig. 2.. CECR2 is highly expressed in breast cancer metastases and correlates with M2 macrophage abundance.

(A) The Venn diagram shows deregulated epigenetic genes with significantly changed mRNA expression (fold change >1.5) by RNA-seq in metastatic samples compared to primary samples. (B) The heat map shows the significantly deregulated epigenetic genes. CECR2 is highlighted in red. HER2 and ER status are shown, “+” means the status is positive and “−” means the status is negative. na, not assessed. (C) The plot shows the correlation between M2 ratios and CECR2 expression. RNA-seq data of matched primary tumor and distal metastases from 13 breast cancer patients were analyzed by CIBERSORTx and immune cell composition of complex tissues were characterized from their gene expression profiles. Pearson correlation coefficient and one-tailed probability p value are shown. (D) Kaplan-Meier analysis shows the association of CECR2 mRNA abundance with distant metastasis-free survival of breast cancer patients using the best cutoff. The cutoff value is 123 in the expression range of 2 to 1738. The hazard ratio (HR) and log-rank p values are shown. (E) CECR2 immunohistochemistry (IHC) staining is shown for a tumor tissue microarray with 59 pairs of matched primary and metastatic breast cancer samples. Representative figures are shown. Scale bars: 100 μm. (F) CECR2 IHC scores were quantified by multiplying the intensity of the signal and the percentage of positive cells. The IHC staining of tumors were scored as weak (score< 0.5), moderate (score between 0.5 and 2) and strong (score >2). Percentage of patient samples with strong CECR2 abundance in metastatic tumors versus that in primary tumor, p<0.05. Percentage of samples with weak CECR2 abundance in metastatic tumors versus that in primary tumor, p<0.05. The p values of unpaired two-tailed Students’ t test are shown. (G) CECR2 IHC staining of matched primary and multiple distant metastasis samples are shown for a single patient with breast cancer. Scale bars: 100 μm. (H) CECR2 IHC staining of MCF10A, MDA-MB-231 and its metastatic derivatives (MDA231-LM2, MDA231-BrM2 and MDA231-BoM) is shown. Scale bars: 100 μm.

Fig. 3.. CECR2 is required for migration, invasion, and metastasis.

(A) Western blot analysis shows control and CECR2 knockout (sg1 and sg2) LM2 cells. (B and C) Transwell migration (B) and invasion (C) assays were used to compare CECR2 knockout and control LM2 cells. (D) Normalized bioluminescence signals are shown for lung metastases in athymic nude mice after tail vein injection of control (n=8) or CECR2 knockout LM2 cells (n=7). Fold change at day 35 is shown. (E) Representative bioluminescence images of mice in (D) at week 5 are shown. Data are presented as mean ± SEM. (F) H&E staining of the lungs from mice in (D) at week 5 is shown. Scale bars: 500 μm for the upper panel and 100 μm for the lower panel. Arrowheads indicate metastatic tumors, and asterisks indicate vascular invasion of large tumor foci. (G) Metastatic tumors were scored based on the percentage of tumors in the lungs with the parameters described in fig. S3E. (H) Normalized bioluminescence signal is shown for lung metastases in immunocompetent wild-type (WT) BALB/c mice after tail vein injection of control 4T1 (n=10), Cecr2 knockout 4T1 (n=10) and Cecr2 knockout 4T1 with CECR2 reconstituted expression (n=10). Fold change at day 14 is shown. (I) Representative bioluminescence images of mice in (H) at week 2 are shown. (J) H&E staining is shown for lungs from mice in (H) at week 2. Scale bars: 200 μm. Arrows indicate metastatic tumors. (K) Schematic of metastasis assay using intracardiac (IC) injection. Mice were monitored for metastasis to the whole body, especially in brain, bone, and liver. (L to P) Normalized in vivo bioluminescence signals of whole-body metastases (L) are shown as well as ex vivo bioluminescence signals and representative pictures of brain metastases (M, N) and ex vivo bioluminescence signals of bone metastases (O) and liver metastasis (P) in WT BALB/c mice after IC injection of control (n=7) or Cecr2 knockout (sg1) (n=8) 4T1 cells. Fold change at day 21 is shown in L. The p values of unpaired two-tailed Students’ t test (B, C) and Mann-Whitney test (D, G, H, L, M, O and P) are shown. *p < 0.05; ** p < 0.01; *** p < 0.001, n.s., not significant. Representative data from triplicate experiments are shown, and error bars represent SEM.

Fig. 4.. CECR2 depletion downregulates NF-κB response genes.

(A and B) Gene set enrichment analysis is shown comparing transcriptomes of CECR2 knockout (CECR2 sg1 and CECR2 sg2) with control LM2 cells. The Venn diagram (A) shows the number of shared downregulated hallmark pathways and (B) the heatmap shows the 8 shared downregulated hallmark pathways. (C) RT-qPCR analysis of CSF1, CSF2 and CXCL1 expression in control and CECR2 knockout LM2 cells is shown. (D) RT-qPCR analysis of Csf1, Csf2 and Cxcl1 expression in control 4T1, Cecr2 knockout 4T1 and Cecr2 knockout 4T1 with CECR2 reconstituted expression is shown. (E) The heatmap shows ATAC-seq peaks for chromatin accessible sites decreased (top) or increased (bottom) by CECR2 depletion, with the aggregated reads within 1 kb of center of differentially accessible regions. (F) ATAC-seq signals around CSF1, CXCL1, CSF3 and IL1B genes showing promoter or putative enhancer regions that are less accessible in CECR2-deficient (sg1) LM2 cells are presented. (G) The Venn diagram shows genes that are downregulated and with decreased ATAC-seq signals in the promoter after CECR2 depletion in LM2 cells. All of the genes are significantly changed with the cutoff of adjusted p-value < 0.05 and fold change >1.2. (H) The top 10 hallmark pathways enriched for downregulated genes with decreased ATAC-seq signaling in the promoter are shown after CECR2 depletion in LM2 cells. The p values of unpaired two-tailed Students’ t test (C and D) are shown. *p < 0.05; ** p < 0.01; *** p < 0.001. Representative data from triplicate experiments are shown, and error bars represent SEM.

Fig. 5.. CECR2 interacts with acetylated RELA using its bromodomain to activate NF-κB response genes.

(A and B) Western blot analysis is shown of cell lysates (input) and immunoprecipitates (IP) from 4T1 (A) and LM2 (B) cells stimulated with 20 ng/ml TNF-α for 0.5 hour with the indicated antibodies. (C and D) ChIP-qPCR analyses are shown for the indicated proteins or histone mark at the CSF1 promoter, and a non-binding region downstream of CSF1 as the negative control. Control and CECR2 knockout (CECR2 sg1 and CECR2 sg2) LM2 cells (C), Control, CECR2 knockout (CECR2 sg1) and RELA knockout (RELA sg) LM2 cells (D) were stimulated with 20 ng/ml TNF-α for 0.5 hour. (E) Western blot analysis of cell lysates (Input) and anti-FLAG IP are shown for HEK293T cells transfected with the indicated combination of vectors expressing FLAG-CECR2, K310R mutated RELA and WT RELA. (F) Western blot analysis of cell lysates (Input) and anti-FLAG IP are shown for HEK293T cells transfected with the indicated combination of vectors expressing WT FLAG-CECR2, FLAG-CECR2 mutant with bromodomain deletion (ΔBRD) and T7-RELA. (G) Western blot analysis of cell lysates (input) and anti-RELA IP are shown for LM2 cells pretreated with control DMSO or CECR2 inhibitors (1 μM NVS-CECR2–1 or 1 μM GNE-886) for 2 days, and then stimulated with 20 ng/ml TNF-α for 0.5 hour. (H and I) RT-qPCR analyses of CSF1, CSF2 and CXCL1 expression in LM2 cells pretreated with the indicated concentration of NVS-CECR2–1 (H) or GNE-886 (I) for 2 days is shown. (J) Scratch migration assays are shown comparing the closure of wound healing distance in LM2 cells treated with DMSO, 1 μM NVS-CECR2–1, or 1 μM GNE-886 for 2 days. (K) Transwell invasion assays are shown comparing LM2 cells treated with DMSO, 1 μM NVS-CECR2–1, or 1 μM GNE-886 for 2 days. The p values of unpaired two-tailed Students’ t test (C, D, H to K) are shown. * p < 0.05, ** p < 0.01, *** p < 0.001. Representative data from triplicate experiments are shown, and error bars represent SEM.

Fig. 6.. CECR2 expression in breast cancer cells increases M2 macrophage proportions in tumor microenvironment.

(A) CCK8 cell proliferation assays are shown for macrophages cultured in RPMI-1640 medium with or without conditioned medium (CM) from control or Cecr2 knockout 4T1 cells. (B) Schematics of transwell co-culture experiments (left panel) and quantification of migrated macrophages (right panel) are shown. Macrophages were seeded into the top chamber (transwell size: 8 μm), and control or Cecr2 knockout (Cecr2 sg1) 4T1 cells were seeded into the bottom chamber. (C) RT-qPCR analysis is shown for M2 markers Arg1, CD206 and IL10 in macrophages cultured with or without conditioned medium (CM) from control or Cecr2 knockout 4T1 cells. (D and E) Flow cytometry analysis is shown for expression of the M2 marker, CD206, in macrophages cultured with or without conditioned medium (CM) from control or Cecr2 knockout 4T1 cells. Shown are representative plots (D) and quantification of the percentage of CD206 positive cells in total macrophages (E). (F) RT-qPCR analyses is shown for M2 markers Arg1, CD206 and Il10 in macrophages. Macrophages were seeded into 6-well plate and treated with conditioned media (CM) harvested from 4T1 cells treated with DMSO, GNE-886, or NVS-CECR2–1 at the indicated dosage for 2 days. (G and H) Flow cytometry analysis is shown for macrophages isolated from the lungs from immunocompetent WT BALB/c mice after tail vein injection of control (n=5) or Cecr2 knockout (sg1) 4T1 cells (n=5) at week 5. Shown are the percentages of total macrophages (G) and the ratios of M2 macrophages to total macrophages (H). (I and J) Flow cytometry analysis is shown for macrophages isolated from the lungs from immunodeficient BALB/c nude mice after tail vein injection of control (n=6) or Cecr2 knockout (sg1) 4T1 cells (n=7) at week 2. Shown are the percentages of total macrophages (I) and the ratios of M2 macrophages (J). The p values of unpaired two-tailed Students’ t test (A to C, E to J) are shown. *p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001. Representative data from triplicate experiments are shown, and error bars represent SEM.

Fig. 7.. CECR2 inhibition suppresses breast cancer metastasis through CSF1-mediated macrophage polarization and enhances anti-tumor immunity.

(A and B) BALB/c wild type mice were injected with control 4T1, Cecr2 knockout (sg1) 4T1 cells, or Cecr2 knockout 4T1 cells with CSF1 overexpression (n=8 for all the groups) through the tail vein. Metastatic lesions in the lungs at week 3 after tumor cell injection were stained by India ink. Shown are representative images (A) and quantification of metastases in the lungs (B). Arrows indicate tumor nodules. (C and D) H&E staining of the lungs from mice in (A) at week 3 is shown. Representative images (C) and quantification of tumor areas in the lungs (D) are presented. Scale bars: 200 μm. (E and G) Flow cytometry analysis is shown for lung lesions isolated from BALB/c wild type mice injected with control 4T1, Cecr2 knockout (sg1) 4T1 cells, or Cecr2 knockout 4T1 cells with CSF1 overexpression (n=8 for (E and F), n=3 for (G)) by tail vein at week 3. Shown are quantification of the percentages of total macrophages (CD45⁺F4/80⁺) (E), M2 macrophages (CD45⁺F4/80⁺CD206⁺) (F) and Granzyme B (GZMB)⁺ CD8⁺ T cells (CD45⁺CD8⁺GZMB⁺) (G). (H) Schematic illustration of NVS-CECR2–1 treatment. BALB/c mice were treated with intraperitoneal injection (i.p.) of NVS-CECR2–1 (10 μg/injection/mouse) or equal volume of PBS (n=5 for each group) every other day for 28 days one day after tail vein injection of 4T1 cells (1×10⁵ per mouse). All mice were euthanized on day 35 to collect lungs and H&E staining were performed. (I to K) Representative H&E staining (I), quantification of total tumor lesions per lung (J), and percentage of tumor area per lung (K) of lungs are shown for mice treated as described in (H). (L and M) Flow cytometry analyses is shown for total macrophages (L) and M2 macrophage ratio (M) isolated from the lungs from BALB/c mice treated as described in (H). The p values calculated by unpaired two-tailed Students’ t tests (E to G, L and M) or Mann-Whitney tests (B, D, J and K) are shown. * p<0.05; **p<0.01; ***p<0.001, n.s. not significant. Representative data from triplicate experiments are shown, and error bars represent SEM.