CD47 blockade reverses resistance to HDAC inhibitor by liberating anti-tumor capacity of macrophages

Abstract

Background: Targeting oncogenic histone modification by histone deacetylase inhibitors (HDACis) demonstrates promising prospects in clinical cancer treatment, whereas a notable proportion of patients cannot benefit from HDACi therapy. This study aims to explore how HDACi influences the tumor microenvironment, in order to identify potential targets for reversing the resistance to HDACi therapies.

Methods: Macrophage infiltration was compared between HDACi-responding and HDACi-nonresponding cancer patients. The impact of HDACis on the phagocytic capacity of macrophages was investigated through macrophage-tumor cell co-culture system. CD47 expression in tumor cell lines and patient-derived organoids was evaluated by quantitative polymerase chain reaction (QPCR) and flow cytometry. Mechanistic studies were conducted through co-immunoprecipitation (co-IP) and chromatin immunoprecipitation (ChIP). The synergistic effect of HDACis and CD47 neutralizing antibody was assessed in subcutaneous murine tumor models. Bioinformatics approaches were adopted to analyze how macrophage infiltration determines the prognostic significance of CD47 expression in cancer patients.

Results: High macrophage infiltration is a determinant of therapeutic non-response to HDACi, cancer patients who did not respond to HDACi exhibit massive infiltration of tumor-associated macrophages (TAMs). TAM depletion reversed the resistance to HDACi therapy. Mechanistically, HDACi impaired the phagocytic capacity of macrophages against tumor cells through epigenetically upregulating CD47 expression. Reciprocally, HDACi-upregulated CD47 polarized macrophages towards a pro-tumor M2 phenotype through SIRPα ligation. In tumor-bearing mice, HDACi monotherapy only marginally delayed tumor progression, while the concurrent neutralization of CD47 exhibited potent anti-tumor effect through re-educating TAMs towards a tumoricidal phenotype. In cancer patients, CD47 was found to determine the prognostic significance of TAMs.

Conclusions: Our study offers a rationale for targeting macrophage infiltration or blocking CD47 to sensitize HDACi therapies in cancer patients.

Keywords: CD47; Cancer; HDAC inhibitor; Macrophage; Tumor microenvironment.

Fig. 1

Macrophages determine the therapeutic outcome of HDACi therapy. A-C MC38 CRC cells were subcutaneously (s.c.) inoculated into control (Mac⁺) or clodronate liposome-treated (macrophage-depleted, Mac^△) C57BL/6 mice (n = 5–6/group), which were intraperitoneally (i.p.) administered with 60 mg/kg VOR. Tumor growth was measured (A); tumor weight was evaluated at experimental endpoint B; the capability of VOR in suppressing tumor volume and weight in Mac⁺ mice or Mac^△ mice were calculated C. D-F HCT116 CRC cells were s.c. inoculated into Mac⁺ or Mac^△ nude mice (n = 7/group), which were i.p. administered with 60 mg/kg VOR. Tumor growth was measured D; tumor weight was evaluated at experimental endpoint E; the capability of VOR in suppressing tumor volume and weight in Mac⁺ mice or Mac^△ mice were calculated F. G, H Tumor tissues from chidamide (CH)-treated breast cancer patients were collected before treatment, and stained with CD68 (G, created with http://BioRender.com), the percentages of CD68⁺ macrophages were compared between CH^R and CH^NR patients (H). (I-K) E0771 breast cancer cells were s.c. inoculated into Mac⁺ or Mac^△ C57BL/6 mice (n = 5/group), which were i.p. administered with 15 mg/kg chidamide. Tumor growth was measured I, tumor weight was evaluated at experimental endpoint J; the capability of chidamide in suppressing tumor volume and weight in Mac⁺ mice or Mac^△ mice were calculated (K). *p < 0.05; **p < 0.01; ***p < 0.001, unpaired, two-tailed Student’s t test

Fig. 2

VOR suppresses the phagocytosis of macrophages against CRC cells through CD47 induction. A Scheme illustrating the co-culture of CRC patient organoids with human primary macrophages (Created with http://BioRender.com). B The organoid-macrophage co-culture was treated with 0.5 μM VOR for 12 h, the expression of indicated genes was evaluated by QPCR. Blue fonts indicate fold change relative to control cells. C CRC patient organoids were stimulated with 0.5 μM VOR for 12 h, the expression of CD47 was evaluated by QPCR. D, E HCT116 cells were stimulated with VOR for 12 h, the expression of CD47 was evaluated by QPCR (D) and flow cytometry. NC = negative control, CT = control E. F, G Nude mice were s.c. inoculated with HCT116 cells, followed by the administration of VOR for 19 days (n = 5/group). The surface levels of CD47 in live tumor cells were evaluated by flow cytometry (F); The expression of CD47 in tumor tissues was evaluated by QPCR G. H HCT116 cells were treated with VOR for 24 h and labeled with Carboxyfluorescein succinimidyl ester (CFSE), followed by the co-incubation with mouse peritoneal macrophages overnight in the presence of IgG1 or α-CD47 (50 μg/mL). The phagocytosis of macrophages was evaluated by flow cytometry. I HCT116 cells were treated with the indicated HDAC subtype inhibitors (0.5 μM) for 24 h, the surface levels of CD47 were evaluated by flow cytometry. J HCT116 cells were treated with 0.5 μM VOR for 24 h and labeled with CFSE, followed by the co-incubation with human PBMC-derived primary macrophages overnight. The phagocytosis of macrophages was evaluated by flow cytometry. *p < 0.05; **p < 0.01; ***p < 0.001, unpaired, two-tailed Student’s t test; or paired, two-tailed Student’s t test for Fig. 2C

Fig. 3

Butyrate promote CD47-mediated phagocytic inhibition. A Fecal samples were collected from CRC patients. The concentrations of various SCFAs were measured by metabolomics in our previous study, the expression of CD47 in matched tumor tissues was evaluated by QPCR. The correlations between tumor CD47 expression and fecal SCFA levels were analyzed using Spearman’s correlation test. (n = 32. For isobutyrate or caproate, n = 31 or 25 respectively, as some patients have low levels of isobutyrate or caproate which were below the detection threshold). B HCT116 cells were stimulated with 2 mM various SCFAs for 24 h, the expression of CD47 was evaluated by flow cytometry. BR = butyrate. C HCT116 cells were pretreated with pertussis toxin (PT) or etomoxir (ETO) for 2 h, followed by butyrate stimulation for 24 h, CD47 expression was measured by flow cytometry. D Nude mice were s.c. inoculated with HCT116 cells, followed by the administration of 150 mM butyrate in drinking water for 18 days (n = 5/group). The expression of CD47 in tumor tissues was evaluated by QPCR. E Organoids from CRC patients were stimulated with 1 mM butyrate for 12 h, the expression of CD47 was evaluated by QPCR. F Mice were fed with antibiotic cocktails for four weeks. Stools were collected to prepare stool extracts, which was used to treat CRC organoids (left, created with http://BioRender.com). The expression of CD47 in organoids 12 h after stimulation was measured by QPCR (right). G HCT116 cells were treated with 1 mM BR for 24 h and labeled with CFSE, followed by the co-incubation with mouse peritoneal macrophages overnight in the presence of IgG1 or α-CD47 (50 μg/mL). The phagocytosis of macrophages was evaluated by flow cytometry. *p < 0.05; **p < 0.01; ***p < 0.001, unpaired, two-tailed Student’s t test; or paired, two-tailed Student’s t test for Fig. 3E

Fig. 4

HDACis facilitate M2 polarization through promoting SIRPα ligation of CD47. (A, B) HCT116 cells were treated with 0.5 μM VOR (A) or 1 mM BR (B) for 24 h, followed by co-incubation with mouse peritoneal macrophages overnight in the presence of IgG1 or α-CD47. The expression of CD206 on F4/80⁺ macrophages was evaluated by flow cytometry. C HCT116 cells were treated with 0.5 μM VOR or 1 mM BR for 24 h, followed by co-incubation with mouse peritoneal macrophages overnight in the presence of 50 μg/mL SIRPα blocking antibody (α-SIRPα). The expression of CD206 on F4/80⁺ macrophages was evaluated by flow cytometry. D The organoid-macrophage co-culture system was established and treated as in Fig. 2A, the expression of CD206 was evaluated by QPCR. E, F HCT116 cells were treated with 0.5 μM VOR (C) or 1 mM BR (D) for 24 h, followed by co-incubation with mouse peritoneal macrophages overnight. The expression of CD86 on F4/80⁺ macrophages was evaluated by flow cytometry. G The correlation between tumor CD206 expression and fecal butyrate levels in CRC patients was analyzed using Spearman’s correlation test. H The correlation between the expression of CD206 and CD47 in tumor tissues from CRC patients was analyzed using Spearman’s correlation test

Fig. 5

HDAC inhibition induces CD47 expression by preventing the binding of HDAC1 to Sp1. A HCT116 cells were treated with 0.5 μM VOR or 1 mM BR for 4 h, the acetylation of H3 histone was evaluated by Western blot. B Venn diagram showing proteins that can potentially interact with both CD47 promoter and HDAC1. C Sp1 binding sites within the CD47 promoter were predicted using the Jaspar database. D Sp1 binding to the CD47 promoter was evaluated by ChIP assay. E HCT116 cells were treated with 1 mM butyrate for 6 h, the interaction between Sp1 and HDAC1 was evaluated by Co-IP. F The interaction among Sp1, HDAC1, and p300 was predicted using STRING database (Created with http://BioRender.com). G HCT116 cells were treated with 0.5 μM VOR or 1 mM BR in the presence or absence of 20 μM AA for 12 h, the expression of CD47 was evaluated by QPCR. **p < 0.01; ***p < 0.001, unpaired, two-tailed Student’s t test

Fig. 6

CD47 blockade sensitizes HDACi therapy by re-educating TAMs. Nude mice were s.c. inoculated with HCT116-EGFP cells, and administered with VOR (60 mg/kg) or BR (150 mM) alone or in combination with α-CD47 (n = 5–7/group). A The growth of tumor in each group was monitored. B Tumor weight was measured at the endpoint. C, D The phagocytosis of HCT116-EGFP cells by F4/80⁺ macrophages in tumor tissues was analyzed by flow cytometry. Representative plots (C) and statistical results (D) were shown. E, F The expression of CD206 on tumor-infiltrating F4/80⁺ macrophages was analyzed by flow cytometry. Representative plots (E) and statistical results (F) were shown. *p < 0.05; **p < 0.01; ***p < 0.001, unpaired, two-tailed Student’s t test

Fig. 7

CD47 determines the prognostic significance of TAMs in cancer patients. A, D The OS (A) or PFS (D) of CRC patients were analyzed in TCGA colon adenocarcinoma (COAD)/rectum adenocarcinoma (READ) cohort based on CD47 expression. B, C, E, F The correlation between macrophage infiltration and the OS (B, C) or PFS (E, F) in CRC patients were analyzed in CD47^hi (B, E) and CD47^low (C, F) subgroups respectively. G, J The OS (G) or PFS (J) of breast cancer patients were analyzed in TCGA BRCA cohort based on CD47 expression. H, I, K, L The correlation between macrophage infiltration and the OS (H, I) or PFS (K, L) in breast cancer patients were analyzed in CD47^hi (H, K) and CD47^low (I, L) subgroups respectively. *p < 0.05; **p < 0.01; log-rank test

Fig. 8

Scheme illustrating the mechanism that HDACi therapy dampens the anti-tumor macrophage responses in the TME. Blockade of CD47-SIRPα interaction overcomes the resistance to HDACi therapy. Created with http://BioRender.com