Introduction

Breast cancer is the most frequently diagnosed type of cancer in women worldwide [1]. Among all subtypes of breast cancer, the aggressive human epidermal growth factor 2-positive (HER2-positive) breast cancer, which is characterised by overexpressed or amplified HER2 receptor, accounts for approximately 20–30% of all annually diagnosed cases [2]. In the past century, patients diagnosed with HER2-amplified or -overexpressed (also referred as HER2-positive) breast cancer were frequently associated with the poorest prognosis and with the highest incidence of brain metastasis [3,4,5,6,7]. However, this is no longer the case following the relatively recent development of novel HER2-targeted therapies, including monoclonal antibody and small-molecule tyrosine kinase inhibitors (TKI), that more precisely target HER2-positive cells to inhibit the activation of downstream signalling while triggering cell death pathways [8].

Orally taken ATP-competitive TKIs, such as the pan-HER-TKI neratinib, have garnered considerable attention in recent years as an effective extended adjuvant HER2-targeted therapy for patients with HER2-positive breast cancers after receiving a combination of chemotherapy and monoclonal antibody trastuzumab [9]. Neratinib treatment significantly improves invasive disease-free survival rate while reducing long-term toxicity and complication rate. Mechanistically, neratinib forms a covalent bond with cysteine residues in the ATP-binding pocket of HER1 (also known as EGFR), HER2 and HER4 receptors to inhibit their downstream canonical Ras/ERK and Akt/mTOR pathways, suppressing cell proliferation, and triggering cell death [10, 11]. A previous study using in vitro HER2-positive breast cancer cell lines, which expressed wild-type oestrogen receptor, such as SKBR3, showed that neratinib treatment repressed the activity of nuclear factor-erythroid factor 2-related factor 2 (NRF2), a master regulator of antioxidant defence, to then promote oxidative-stress-dependent cell death [12]. Oxidative stress is a condition of sustained elevation of intracellular reactive oxygen species (ROS) that potentially leads to reversible and irreversible damage to biomolecules such as lipids, proteins, and nucleic acids [13]. ‘ROS’ is an umbrella term describing a collection of oxygen-derived radicals, such as singlet superoxide (O2• −) and hydroxyl radical (OH), and species, such as hydrogen peroxide (H2O2) and lipid peroxide. Strikingly, a recent report from Nagpal and colleagues revealed that, rather than apoptosis as induced by other TKIs, such as lapatinib, neratinib promoted non-apoptotic cell death called ferroptosis in both human (SKBR3) and mouse (TBCP1) HER2-positive breast cancer cell lines [14]. Ferroptosis is a newly discovered type of cell death characterised by perturbed iron and redox homeostasis with excessive lipid peroxidation at phospholipid membrane [15]. Unless the rise of endogenous lipid peroxide is averted such as by the scavenging enzymatic activity of glutathione peroxidase 4 (GPX4), the accumulation of lethal membrane lipid peroxide results in the loss of membrane integrity and ultimately leads to cell death [16].

Despite therapeutic effectiveness against HER2-positive breast cancers, patients receiving neratinib can experience mild to severe symptoms of gut toxicity, such as abdominal pain and diarrhoea, which can lead to early dose reduction or treatment discontinuation [17, 18]. Interestingly, a recent randomised controlled trial conducted by Sonnichsen and colleagues reported that though serum levels of neratinib were higher in patients receiving 240 mg of neratinib once daily (QD) than those receiving 120 mg of neratinib twice daily (BID), patients in the QD group experienced less severe diarrhoea [18]. This suggested that diarrhoea is potentially a consequence of local exposure to neratinib rather than systemic exposure. In support, starting patients on a reduced dose of neratinib for the first two weeks is also associated with greatly reduced levels of diarrhoea and treatment discontinuations [19]. Furthermore, an escalating dose of neratinib resulted in more severe symptoms of gut toxicity observed in our preclinical rat models and other clinical trials [20, 21]. We therefore postulate that inhibition of locally expressed HER receptors in the intestinal epithelial cells which subsequently triggers epithelial cell death is likely a cause of gut toxicity. This is plausible due to the gut epithelium naturally expressing HER receptors and being strongly dependent on HER downstream signalling pathways for cell survival, differentiation, and proliferation [22,23,

Table 2 List of primers used for RT-qPCR experiments

RNA-sequencing analysis

RNA-sequencing analysis on previously published datasets was carried out to examine which biological pathways were significantly enriched following neratinib treatment. Processed count tables for protein encoded genes from neratinib-treated mouse TBCP-1 breast cancer and human glioblastoma SF268 and TS895 cell lines were kindly provided by Dr Normand Pouliot (Olivia Newton-John Cancer Research Institute) and Dr Colin Tang (Weill Cornell Medicine) [14, 31]. Differential gene expression between neratinib and vehicle control groups was performed using DESeq2 (v1.38.3) statistical package in R [35]. Unless otherwise stated, a master list of differentially expressed (DE) genes was generated when the P-adjusted value less than 0.05. Subsequently, a list of significantly upregulated genes with log2FoldChange values above 0.1 was generated from the master DE gene list with P-adjusted values or false discovery rate (FDR) below 0.05. Subsequently, this gene list was inserted into the online database for annotation, visualisation, and integrated discovery (DAVID) bioinformatic resource website for gene ontology (GO) terms and KEGG pathway analyses as previously described [36,37,38]. The computational algorithm used to determine statistical significance of individual pathways was detailed in the original papers [36, 39]. Here, GO terms and KEGG pathways were deemed as significantly enriched with P values below 0.05.

CRISPR-screening analysis

As previously described in Tang et al. [31], a pool CRISPR screen was performed on neratinib-treated SF268 human glioblastoma cell line to identify genes contributing to neratinib sensitivity. A count table of differentially expressed single-guided RNA and an output file of DrugZ were kindly provided by Dr Colin Tang from Weill Cornell Medicine. Using R script publicly available generated by Dr Colin Tang, drugZ rank plot was generated for normalised Z score against sgRNA rank 95. The positive or negative normalised Z score of a gene with FDR below 0.05 confers with either neratinib sensitivity or resistance, respectively.

Data analysis and statistics

Descriptions for sample size and statistical tests are detailed in figure legends. Unless otherwise specified, all statistical analyses were carried out using unpaired Student’s t-test in GraphPad Prism 9 (version 9.2.0; USA). Statistical values were reported in mean ± standard error of the mean (mean ± SEM). Statistical significance, which was reported as exact P values, was considered as followed – for all histological assessment, RT-qPCR, gene ontology and KEGG pathway analyses, P values below 0.05 were deemed significant while for differential gene expression and Drug-Z analyses, P-adjusted values, or false discovery rate (FDR) below 0.05 were deemed significant.

Data availability

The data generated in this study are available within the article and its supplementary data files, with further raw data available upon request. Further information and requests for resources should be directed to and will be fulfilled by the corresponding authors, Joanne M. Bowen (joanne.bowen@adelaide.edu.au) and Triet PM. Nguyen(phuminhtrietthomas.nguyen@unimelb.edu.au).