Microenvironment-Mediated Mechanisms of Resistance to HER2 Inhibitors Differ between HER2+ Breast Cancer Subtypes

Abstract

Extrinsic signals are implicated in breast cancer resistance to HER2-targeted tyrosine kinase inhibitors (TKIs). To examine how microenvironmental signals influence resistance, we monitored TKI-treated breast cancer cell lines grown on microenvironment microarrays composed of printed extracellular matrix proteins supplemented with soluble proteins. We tested ∼2,500 combinations of 56 soluble and 46 matrix microenvironmental proteins on basal-like HER2+ (HER2E) or luminal-like HER2+ (L-HER2+) cells treated with the TKIs lapatinib or neratinib. In HER2E cells, hepatocyte growth factor, a ligand for MET, induced resistance that could be reversed with crizotinib, an inhibitor of MET. In L-HER2+ cells, neuregulin1-β1 (NRG1β), a ligand for HER3, induced resistance that could be reversed with pertuzumab, an inhibitor of HER2-HER3 heterodimerization. The subtype-specific responses were also observed in 3D cultures and murine xenografts. These results, along with bioinformatic pathway analysis and siRNA knockdown experiments, suggest different mechanisms of resistance specific to each HER2+ subtype: MET signaling for HER2E and HER2-HER3 heterodimerization for L-HER2+ cells.

Keywords: HER2+ breast cancer subtypes; HER3; HGF; NRG1β; crizotinib; drug resistance; lapatinib; microenvironment; neratinib; pertuzumab.

Figure 1. MEMA Studies Reveal Multiple Protein Combinations that Confer Lapatinib Resistance to Otherwise Sensitive HER2+ Breast Cancer Cell Lines

(A) Plots of median cell count versus median EdU-positive ratio for AU565 and HCC1954 cells on MEMAs following 72 hr of 750 nM lapatinib or DMSO treatments. Microenvironment pertubagens (MEPs, combination of ECM and ligand) are color coded by ligand and treatment. Non-EGF, FGF, and HGF family ligands are shown in gray. (B and C) Isolated plots from (A) of AU565 and HCC1954 cells exposed to NRG1β and HGF, respectively, following lapatinib treatment shows ECM or adhesion proteins influencing ligand-mediated drug resistance. Error bars display SEM, n = 13–15. (D and E) Mean cell count (n = 2, biological replicates [BR] = 3) derived from live-cell imaging of nuclear-GFP-expressing SKBR3 and HCC1954 cells treated with DMSO, 500 nM lapatinib, 25 ng/mL NRG1β or HGF, and 500 nM lapatinib plus 25 ng/mL NRG1β or HGF over a 96-hr time course. Growth factors were added at time 0, and lapatinib was spiked in at the 24-hr time point. Cell counts normalized to counts at time 0 for each treatment condition. (F and G)Mean cell count and SEM (n = 3)for AU565 and HCC1954 cells treated for 72 hr with combinations of DMSO, 500 nM lapatinib, 500 µM capecitabine, and 50 ng/mL NRG1β or HGF. Lapatinib significantly decreased cell counts compared with DMSO (****AU565, p < 0.0001; ****HCC1954, p < 0.0001), and ligand added to lapatinib significantly increased cell count compared with lapatinib alone in both cell lines (***AU565, p = 0.0009; ****HCC1954, p < 0.0001). Capecitabine significantly decreased the cell count compared with DMSO (**AU565, p = 0.0091; ***HCC1954, p = 0.0002), but addition of ligand did not significantly alter cell counts. The combination of lapatinib and capecitabine significantly decreased cell counts compared with DMSO (****AU565, p < 0.0001; ***HCC1954, p = 0.0002), and the addition of ligand significantly increased cell counts compared with the combination alone in both cell lines (**AU565, p = 0.0044; **HCC1954, p = 0.0033). ns, not significant.

Figure 2. HER2+ Cell Lines Are Sub-classified into Luminal-like L-HER2+ and Basal-like HER2E Phenotypes

(A) mRNA expression clustered heatmap of genes identified by TCGA as significantly different between HER2E and luminal HER2+ patient tumors in a panel of HER2+ breast cancer cell lines. Gene expression was sorted for variance across cell lines; the top 10% (66 genes) are used to cluster the panel. (B) Representative images of L-HER2+ and HER2E lines immunofluorescently labeled with DAPI (blue), KRT14 (green), and KRT19 (red).

Figure 3. HER2+ Breast Cancer Cell Lines Exhibit a Subtype Intrinsic Proliferative Response to NRG1β and HGF under Lapatinib or Neratinib Treatment

(A) Heatmap of mean cell count (n = 3) of 8 HER2+ cell lines exposed to a dose range of neratinib and three concentrations of NRG1β and HGF. Each value is normalized to the mean cell count of the corresponding DMSO-treated control. Scale to the right indicates the relative cell count ratio between drug-treated and untreated controls. (B) Mean percentage of EdU-positive cells and SEM (n = 3) in four L-HER2+ cell lines treated in (A) with DMSO, 100 nM neratinib, and 100 nM neratinib plus 25 ng/mL NRG1β. Neratinib plus NRG1β treatment results in significantly increased EdU positivity compared with DMSO treatment in 3 out of 4 cell lines (****AU565, p < 0.0001; **SKBR3, p = 0.0068; *BT474, p = 0.0397). (C) Mean absorbance measurements and SEM (n = 2, BR = 3) of alamar blue stains from four cell lines in 3D Matrigel assays following 96 hr treatments of 200 or 500 nM neratinib, with and without 50 ng/mL NRG1β and HGF. Absorbance values are shown as the ratio of drug-treated to untreated control. (D and E) Representative images and quantification of xenograft tumor response to local delivery of drugs alone and in combination with NRG1β and HGF proteins. Reservoirs loaded with pure PEG polymer served as a control. Sectioned tissue surrounding the implantable nanodosing device (red boxes) is stained for CC3 and Ki67 to assess apoptosis and proliferation, respectively. Graphs show mean and SEM normalized signal intensity of the tumor region adjacent to treatment reservoirs (BT474, n = 4; JIMT1, n = 3).

Figure 4. HER2E and L-HER2+ Lines Show Differential Reliance on MAPK and PI3K Signaling

(A) Log-scale mRNA expression of ERBB3 and MET in a panel of HER2+ cell lines. Enlarged spots indicate cell lines used in Figure 3A. (B) Quantification of western blot protein analysis of HER3 and MET levels in eight cell lines (BR = 3). Average HER3 levels are significantly reduced in HER2E compared with L-HER2+ (*p = 0.012). (C) Ratio of ERBB3 and MET mRNA expression in human breast cancer tumors defined as HER2E and L-HER2+ by TCGA. (D) Mean log mRNA counts and SEM (n = 8) of eight L-HER2+ cell lines and eight HER2 lines and expression of FOXA1 and EGFR. (E) RPPA time course of AU565 and HCC1954 cells treated with 250 nM lapatinib. The y axis shows mean (n = 3) signal intensity of each phospho-protein versus its respective total protein, with each signal normalized to its DMSO-treated control cohort at each time point, representing change in protein activity over 72 hr of treatment. The top three proteins are canonical constituents of the PI3K/MTOR pathway, the bottom three are canonical constituents of the MAPK pathway. (F) GI 50 graphs and SEM (n = 3) of CTG assays from HER2E and L-HER2+ cell lines treated for 72 hr with a dose range of lapatinib, trametinib, and the combination.

Figure 5. L-HER2+ and HER2E Lines Differ in Their Resistance Mechanisms to HER2 Inhibition

(A) Mean cell count and SEM (n = 3) of eight HER2+ cell lines treated for 72 hr with combinations of 500 nM lapatinib, 500 nM crizotinib, 30 µg/mL pertuzumab, 50 ng/mL NRG1β, 50, and ng/mL HGF. PBS, DMSO, and human IgG isotype control were used as controls for growth factors, TKIs, and pertuzumab, respectively. Addition of NRG1β to lapatinib results in significantly increased cell counts in each L-HER2+ line compared with drug alone (****AU565, p < 0.0001; ***SKBR3, p =0.0003; ***BT474, p =0.0002; ****EFM192A, p< 0.0001). Addition of HGF to neratinib results in significantly increased cell count in each HER2E line compared with drug alone (**HCC1954, p = 0.0037; ***HCC-3153, p = 0.0002; ***JIMT1, p = 0.0001; **21-MT1, p = 0.0045). Addition of pertuzumab to lapatinib plus NRG1β results in significantly decreased cell counts in each L-HER2+ line compared with lapatinib plus NRG1β alone (***AU565, p = 0.0005; ***SKBR3, p = 0.0008; **BT474, p = 0.0013; **EFM192A, p = 0.0001). Addition of crizotinib to neratinib plus HGF results in significantly decreased cell counts in each HER2E line compared with neratinib plus HGF alone (***HCC1954, p = 0.0006; ****HCC-3153, p < 0.0001; ****JIMT1, p < 0.0001; ****21-MT1, p < 0.0001). (B) Maximum projection fluorescent images of SKBR3 and HCC1954 cells treated for 48 hr with combinations of 500 nM lapatinib and 12.5 ng/mL NRG1β. Cell nuclei imaged with DAPI (blue), β-tubulin (green), and HER2-HER3 heterodimers (red) imaged by PLA. (C) Mean and SEM of PLA spot counts for SKBR3 (n = 79, 93, 150, 54) and HCC1954 (n = 52, 60, 54) treated for 48 hr with DSMO, 12.5 ng/mL NRG1β, 500 nM lapatinib, and the combination. p value shows unpaired t test of significance. NS, not significant.

Figure 6. Association of PI3K Subunits with HER3 Regulated by NRG1 and HER2 Targeting Drugs

(A) Heatmap of detection counts from mass spectrometry proteomic analysis of HER3 immunoprecipitation from AU565 cell lysate following 48 hr of treatment with 500 nM lapatinib, 50 ng/mL NRG1β, 25 µg/mL pertuzumab, and selected combinations. Each count is normalized to the detected quantity of HER3 in each treatment sample. Scale shows the ratio of individual protein count to HER3 count (BR = 3). (B) A simplified model of HER2 and HER3 dimerization on the cell surface when activated by NRG1β. This structure alters the conformation of the ATP-binding pocket on the HER2 cytoplasmic kinase domain to one to which lapatinib cannot readily bind. (C) Pertuzumab inhibits cis-phosphorylation of HER3 via separation of the cytoplasmic kinase domains. This structure returns the HER2 kinase domain conformation to one to which lapatinib can bind. Note that receptors remain linked under pertuzumab treatment. (D) Higher-order receptor structures overcome inhibition by pertuzumab via trans-phosphorylation of HER3 (orange dot). (E) The proposed mechanism for how pertuzumab and lapatinib combine to overcome cis- and trans-phosphorylation of HER3, respectively.