Tumor immune microenvironment permissive to metastatic progression of ING4-deficient breast cancer

Abstract

Deficiencies in the ING4 tumor suppressor are associated with advanced stage tumors and poor patient survival in cancer. ING4 was shown to inhibit NF-kB in several cancers. As NF-kB is a key mediator of immune response, the ING4/NF-kB axis is likely to manifest in tumor-immune modulation but has not been investigated. To characterize the tumor immune microenvironment associated with ING4-deficient tumors, three approaches were employed in this study: First, tissue microarrays composed of 246 primary breast tumors including 97 ING4-deficient tumors were evaluated for the presence of selective immune markers, CD68, CD4, CD8, and PD-1, using immunohistochemical staining. Second, an immune-competent mouse model of ING4-deficient breast cancer was devised utilizing CRISPR-mediated deletion of Ing4 in a Tp53 deletion-derived mammary tumor cell line; mammary tumors were evaluated for immune markers using flow cytometry. Lastly, the METABRIC gene expression dataset was evaluated for patient survival related to the immune markers associated with Ing4-deleted tumors. The results showed that CD68, CD4, CD8, or PD-1, was not significantly associated with ING4-deficient breast tumors, indicating no enrichment of macrophages, T cells, or exhausted T cell types. In mice, Ing4-deleted mammary tumors had a growth rate comparable to Ing4-intact tumors but showed increased tumor penetrance and metastasis. Immune marker analyses of Ing4-deleted tumors revealed a significant increase in tumor-associated macrophages (Gr-1loCD11b+F4/80+) and a decrease in granzyme B-positive (GzmB+) CD4+ T cells, indicating a suppressive and/or less tumoricidal immune microenvironment. The METABRIC data analyses showed that low expression of GZMB was significantly associated with poor patient survival, as was ING4-low expression, in the basal subtype of breast cancer. Patients with GZMB-low/ING4-low tumors had the worst survival outcomes (HR = 2.80, 95% CI 1.36-5.75, p = 0.0004), supportive of the idea that the GZMB-low immune environment contributes to ING4-deficient tumor progression. Collectively, the study results demonstrate that ING4-deficient tumors harbor a microenvironment that contributes to immune evasion and metastasis.

Fig 1. ING4, NF-kB, and immune markers in breast tumors.

(A) Immunohistochemical staining images of tumors positive for: a) ING4 (IHC score +1), b) pp65/RelA(Ser276) (IHC score +2), c) CD68 (IHC score +2), d) CD4 (IHC score +3), e) CD8 (IHC score +2), f) PD-1 (IHC score +3); black boxes in the top panel are presented in higher magnification images in the bottom panel; black scale bars denote 100 mm in length. (B) Percentage of tumors positive for each marker (solid dark bar): ING4 (154/246, 63%), pp65 (84/245, 34%), CD68 (171/193, 89%), CD4 (50/193, 26%), CD8 (75/189, 40%), and PD-1 (24/132, 18%). (C) Venn diagram and table showing the number of tumors positive for CD68, CD8, CD4, and/or PD-1. (D) No association between immune markers with ING4-low vs ING4-high (left graph) or NF-kB-low vs NF-kB-high (right graph) tumors. (E) Increased CD8+ tumors in ING4-low/pp65-high tumors (solid dark bar, 59% vs 32–38%); p values were determined by Fisher’s Exact test; ns, not significant. (F) High lymph node positivity in ING4-low/pp65-high tumors (LN+, solid dark bar): All, all tumors; ER-, estrogen receptor-negative.

Fig 2. Aberrant activation of NF-kB in Ing4-deleted p53MT mouse mammary tumor cells.

(A) Schematic diagram of the Ing4 genome with exons (open boxes); CRISPR/CAS9 gRNAs R1 and M1 mapped to exon 1 and exon 3 (red line). (B) Nuclear accumulation of pp65/RelA in Ing4-deleted p53MT cells treated with cytokines: v2, the vector control cells; v2R1, R1 gRNA construct cells; v2M1, M1 gRNA construct cells; cells were treated with PBS (-), 10ng/ml TNFa (T), or 10ng/ml IL-1b (I), for 1 hour in serum-free media prior to cell fractionation; nuc, nuclear fraction; cyto, cytoplasmic fraction; Western blot for ING4 and phospho-p65/RelA (ser536); Histone H3 and a-tubulin antibodies were used as the loading controls. (C) Relative fold expression of the mouse Il6 gene in p53MT cells treated with (+) or without (-) TNFa (left panel) or IL-1b (right panel) for 4 hours in serum-free media. Numbers denote the average fold change of a minimum three replicates: open bar, v2; closed bar, v2R1; serrated bar, v2M1; RT-qPCR to quantify Il6 expression using Gapdh as the reference; p values were determined by two-tailed student t-test; ns, not significant. (D) Cytokine-induced migration of Ing4-deleted p53MT cells: PBS (-),TNFa (T), or IL-1b (I); v2, v2R1, or v2M1 cells migrated through 8mm pores in the transwell inserts were stained with DAPI and visualized under a fluorescent microscope. Cell numbers were determined by averaging a minimum of 4 images per experiment from at least 3 independent experiments; p values were determined by two-tailed student t-test; *, p < 0.001; **, p < 0.05.

Fig 3. Growth and metastasis of Ing4-deleted p53MT tumors in mice.

(A) Increased penetrance of Ing4-deleted tumors: Kaplan-Meier analysis of tumor-free mice during a 36 day follow up after 5 x 10⁴ cell implant into the #4 mammary fat pads. Black line, the v2 vector control (n = 10); red line, v2R1 (n = 10); blue line, v2M1 (n = 10); p values determined using two-tailed unpaired student t-test, v2 vs v2R1 p = 0.0007, v2 vs v2M1 p = 0.0043 and one-way ANOVA p = 0.0031 (B) Tumor growths in representative 4 mice in each group were graphed using an average tumor measurement of the longest axis at each time points, setting day 0 for each tumor when a 1–2 mm nodule is detected. Black circle, v2 (n = 4); red circle, v2R1 (n = 4); blue square, v2M1 (n = 4). (C) Tumor metastasis to the lung quantified by lung tissue harvest and dissociation at end point followed by plating in the media selecting for puromycin resistance encoded by the plasmid constructs. The results from WT (n = 1, no tumor implant), v2 (n = 3), v2R1 (n = 3), and v2M1 (n = 3) tumor mice are shown: puromycin-resistant foci (crystal blue-staining) were counted and presented in a Box plot with X marks the mean value and o median. p values were determined by student t-test comparing to v2.

Fig 4. Myeloid and natural killer cells in Ing4-deleted p53MT tumors.

(A) Gating strategy for myeloid and natural killer cells. Numbers indicate the percentage in the gate. v2-3T, a v2 tumor; MDSC, myeloid derived suppressive cells; TAM, tumor associated macrophages; NK, natural killer cells. (B) Average percentage of the marker positive cells in v2 (n = 4), v2R1 (n = 3), v2M1 (n = 5) tumors. Bar graphs display the percentage of CD45⁺ cells out of total live cells; percentage of Gr-1^hiCD11b⁺ (MDSC) out of total CD45⁺ cells; percentage of F4/80⁺ (TAM) out of CD45⁺Gr-1^loCD11b⁺ cells; and percentage of GZMB⁺ (NK) cells out of total CD45⁺NKp46⁺ cells. Open bar, v2; solid bar, v2R1; serrated bar, v2M1; error bars show minimum and maximum values; *, p<0.05; **, p<0.005. (C) Average percentage of PD-L1 positive cells in the CD45^- (predominantly tumor) or CD45⁺ (immune) compartments. ns, not significant.

Fig 5. T cells in Ing4-deleted p53MT tumors.

(A) T cell markers and the gating scheme. v2-3T, a v2 tumor; Treg, CD4⁺ regulatory T cells; Tc, CD8⁺ cytotoxic T cells; Th, CD4⁺ helper T cells. Numbers indicate the percentage in the gate. (B) Average percentage of the marker positive cells in v2 (n = 4), v2R1 (n = 3), v2M1 (n = 5) tumors. Bar graphs display the percentage of CD45⁺CD3⁺ T cells out of total live cells; CD4⁺ and CD8⁺ T cells out of all T cells; percentage of CD4⁺FoxP3⁺ (Treg) and CD4⁺FoxP3^- (Th) cells out of total CD4⁺ T cells; and percentage or GzmB⁺ cells out of total Treg, Th, or Tc populations, as indicated. Open bar, v2; solid bar, v2R1; serrated bar, v2M1; error bars show minimum and maximum values; *, p<0.05; **, p<0.01. (C) Average percentage of PD-1⁺ cells in the CD4⁺ or CD8⁺ T cell compartment. ns, not significant.

Fig 6. ING4 and GZMB expression associated with patient survival in breast cancer.

Kaplan-Meier survival analyses of the METABRIC data set using the mean gene expression values as the cut-off in the basal subtype breast cancer (n = 209). (A) ING4, GZMB (the granzyme B gene). (B) combination of ING4 and GZMB expression levels; black line, ING4-high/GZMB-high (n = 74); blue line, ING4-high/GZMB-low (n = 30); green line, ING4-low/GZMB-high (n = 74); red line, ING4-low/GZMB-low (n = 31). (C) FOXP3, ADGRE1 (the F4/80 gene), PDCD1 (the PD-1 gene), and CD274 (the PD-L1 gene). DFS, disease-free survival; HR, Hazard Ratio; p values determined by the log-rank (Mantel-Cox) test.