Invasive lobular and ductal breast carcinoma differ in immune response, protein translation efficiency and metabolism

Abstract

Invasive lobular carcinoma (ILC) is the second most common histological subtype of breast cancer following invasive ductal carcinoma (IDC). ILC differs from IDC in a number of histological and clinical features, such as single strand growth, difficulty in detection, and frequent late recurrences. To understand the molecular pathways involved in the clinical characteristics of ILC, we compared the gene expression profiles of luminal A ILC and luminal A IDC using data from TCGA and utilized samples from METABRIC as a validation data set. Top pathways that were significantly enriched in ILC were related to immune response. ILC exhibited a higher activity of almost all types of immune cells based on cell type-specific signatures compared to IDC. Conversely, pathways that were less enriched in ILC were related to protein translation and metabolism, which we functionally validated in cell lines. The higher immune activity uncovered in our study highlights the currently unexplored potential of a response to immunotherapy in a subset of patients with ILC. Furthermore, the lower rates of protein translation and metabolism - known features of tumor dormancy - may play a role in the late recurrences of ILC and lower detection rate in mammography and PET scanning.

Figure 1

LumA ILC is enriched for immune cell infiltration and high immune-checkpoint gene expression. (a) 853 up-regulated genes and 602 down-regulated genes (LumA ILC, n = 159 vs LumA IDC, n = 311, FDR < 0.05) in TCGA were validated in METABRIC (marked in red, the direction of the changes for DE genes were matched). (b) Proportion of immune phenotypes in LumA ILC (n = 157) and LumA IDC (n = 303). Tumors were classified into 6 immune-phenotypes (immune-phenotype 1–6) by Tamborero et al. and those in immune-phenotype 1–3 and 4–6 were defined as low immune tumors and high immune tumors, respectively. Chi-square test, ***p < 0.0005. (c) Expression of CD274 (PD-L1), PDCD1 (PD-1) and CTLA4 in LumA ILC and LumA IDC of different immune phenotypes. High Immune LumA ILC and IDC have similar PDCD1, and CTLA4 expression as Basal and HER2 subtypes. Low immune (LumA ILC, n = 77, vs LumA IDC n = 221), high immune (LumA ILC, n = 80, vs LumA IDC n = 82), all (LumA ILC, n = 157 vs LumA IDC, n = 303). Two-way ANOVA for the effect of histological subtype on immune checkpoint gene expression, *p < 0.05, **p < 0.005, ***p < 0.0005. The effect of immune phenotype on immune checkpoint gene expression, p < 0.0005 for all genes. No significant interaction (p > 0.05) between histology and immune phenotype. (d) Proportion of high immune tumors in ILC subtypes (Proliferative n = 18, Reactive-like n = 34, Immune-related n = 40). Chi-square test for equality of proportions, ***p < 0.0005.

Figure 2

Immune signature difference is not a reflection of normal breast contamination. (a) Tumor purity score (CPE) of LumA ILC (n = 157) and LumA IDC (n = 307). Mann-Whitney U test, **p < 0.005. (b) LumA high immune tumors (n = 162) have different immune cell profile than normal female breast tissue (n = 90). Immune cell types with median GSVA difference >0.2 between normal breast tissue and LumA high immune tumors are marked in red: CD56dim Natural Killer cells (NK dim), activated dendritic cells (aDC), effector memory T-cells (Tem), gamma delta T-cells (Tgd). (c) Tumors and normal breast can be differentiated based on immune cell expression. Tumors/normal breast tissues in heatmap were sorted by sum (Treg+mast_cell + Nkdim + aDC) – sum(Tem, Tgd).

Figure 3

LumA ILC has lower protein translation efficiency than LumA IDC. (a) Protein/mRNA ratio in LumA ILC (n = 115) and LumA IDC (n = 246). Mann-Whitney U test, *p < 0.05. (b) Protein levels of translation regulators in LumA ILC vs LumA IDC. Protein expression data were from TCGA RPPA. Limma was used to compare the protein expression of LumA ILC to LumA IDC with CPE correction. Significant DE proteins (Benjamini-Hochberg method adjusted p-value < 0.05) were marked in red (up-regulated in LumA ILC) or blue (down-regulated in LumA ILC). (c) Regulation network of protein translation regulators in Fig. 2b. Modified from. (d) Protein synthesis rate of ILC and IDC cell lines. O-propargyl-puromycin (OPP) labeled the newly synthesized proteins. Fluorescence representing the amount of OPP indicated the protein synthesis rate of cells. Cells without OPP labeling or pre-treated with cycloheximide (CHX) to inhibit protein synthesis served as negative controls. Representative data of two independent experiments were presented. Data are mean ± s.d. of 3 replicates. Two-way ANOVA, ***p < 0.001. (e) Dose response and IC50 of translation inhibitors in ILC and IDC cell lines. 4EGI-1 to inhibit the binding of eIF4E and eIF4G, cycloheximide to inhibit the tRNA translocation, salubrinal to inhibit eIF-2α were used. Representative data of at least two independent experiments were presented. Data in dose response curves are mean ± s.d. of 6 replicates. Data in bar graphs of IC50 are mean + upper limit of 95% confidence intervals. Two-tailed t-test was performed to compare the IC50s between ILC and IDC cell lines. The p-values for cycloheximide, salubrinal and 4EGI-1 are 0.15, 0.17 and 0.42, respectively.

Figure 4

LumA ILC is more bioenergetically quiescent as compared to LumA IDC. (a) Top 15 inhibited pathway in LumA ILC compared to LumA IDC. –log10(0.05) is marked with red line. (b) The basal OXPHOS and glycolysis rate of ILC and IDC cell lines. Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR), which are indicators of OXPHOS and glycolysis rates respectively were measured with Seahorse XF^e96 analyzer. Representative data of two independent experiments were presented. Data are mean ± SEM of 3 repeated measurements. Each measurement measured 6 or 8 biological replicates. Two-way ANOVA, p-values for ECAR[p(ECAR)] and OCR[p(OCR)] between ILC and IDC cell lines were calculated independently.