Targeted dual degradation of HER2 and EGFR obliterates oncogenic signaling, overcomes therapy resistance, and inhibits metastatic lesions in HER2-positive breast cancer models

Abstract

Aims:

Human epidermal growth factor receptor 2 (HER2) is an oncogenic receptor tyrosine kinase amplified in approximately 20% of breast cancer (BC). HER2-targeted therapies are the linchpin of treating HER2-positive BC. However, drug resistance is common, and the main resistance mechanism is unknown. We tested the hypothesis that drug resistance results mainly from inadequate or lack of inhibition of HER2 and its family member epidermal growth factor receptor (EGFR).

Methods:

We used clinically relevant cell and tumor models to assess the impact of targeted degradation of HER2 and EGFR on trastuzumab resistance. Trastuzumab is the most common clinically used HER2 inhibitor. Targeted degradation of HER2 and EGFR was achieved using recombinant human protein PEPD^G278D, which binds to the extracellular domains of the receptors. siRNA knockdown was used to assess the relative importance of EGFR and HER2 in trastuzumab resistance.

Results:

Both HER2 and EGFR are overexpressed in all trastuzumab-resistant HER2-positive BC cell and tumor models and that all trastuzumab-resistant models are highly vulnerable to targeted degradation of HER2 and EGFR. Degradation of HER2 and EGFR induced by PEPD^G278D causes extensive inhibition of oncogenic signaling in trastuzumab-resistant HER2-positive BC cells. This is accompanied by strong growth inhibition of cultured cells, orthotopic patient-derived xenografts, and metastatic lesions in the brain and lung of trastuzumab-resistant HER2-positive BC. siRNA knockdown indicates that eliminating both HER2 and EGFR is necessary to maximize therapeutic outcome.

Conclusions:

This study unravels the therapeutic vulnerability of trastuzumab-resistant HER2-positive BC and shows that an agent that targets the degradation of both HER2 and EGFR is highly effective in overcoming drug resistance in this disease. The findings provide new insights and innovations for advancing treatment of drug-resistant HER2-positive breast cancer that remains an unmet problem.

1. Introduction

Excessive signaling by human epidermal growth factor receptor 2 (HER2), a receptor tyrosine kinase (RTK), which may result from its overexpression or mutation, drives cancer development and progression. HER2 is overexpressed in approximately 20% breast cancer (BC), known as HER2-positive BC, which results primarily from amplification of the ERBB2 gene which encodes HER2 (Memon et al., 2022). Targeting HER2 is critical for treatment of HER2-positive BC. Indeed, since 1998, nine anti-HER2 therapies have been developed, which include three monoclonal antibodies (trastuzumab, pertuzumab, and margetuximab), two antibody drug conjugates (trastuzumab emtansine, and trastuzumab deruxtecan), and four tyrosine kinase inhibitors (lapatinib, neratinib, pyrotinib, and tucatinib) (Swain et al., 2023). While these therapies have significantly improved disease outcome, primary and acquired drug resistance is common. Trastuzumab, also known as herceptin, is the most commonly used anti-HER2 therapy. However, as a monotherapy, it only achieves objective response rate (ORR) and median progression free survival (MPFS) of 26% and 3.5-3.8 months, respectively, in patients with metastatic disease (Vogel et al., 2002). Adding trastuzumab to chemotherapy increases ORR from 32% to 50% and MPFS from 4.6 months to 7.4 months (Slamon et al., 2001). Trastuzumab deruxtecan, also known as DS-8201a or enhertu, appears to be the most efficacious among the anti-HER2 drugs and achieves ORR and MPFS of 60.9% and 16.4 months, respectively (Modi et al., 2020).

Mechanism of resistance to HER2 inhibitors in HER2-positive BC is complex and not fully known. However, there is growing evidence that the inability of current HER2 inhibitors to induce its degradation may contribute significantly to treatment resistance (Zhang, 2021). Overexpressed HER2 initiates oncogenic signaling primarily through homodimerization and heterodimerization with other RTKs. HER2 is the preferred dimerization partner for its family members, including epidermal growth factor receptor (EGFR), HER3, and HER4 (Tzahar et al., 1996). HER2 was also shown to bind to additional 35 RTKs (Kennedy et al., 2019). Many of the RTKs that bind to HER2 have been implicated in drug resistance and disease progression in HER2-positive BC, including but not limited to AXL (Goyette et al., 2018), EGFR (DiGiovanna et al., 2005), ephrin type-B receptor 4 (EPHB4) (Ding et al., 2020), fibroblast growth factor receptor 2 (FGFR2) (Fernandez-Nogueira et al., 2020), HER3 (Watanabe et al., 2019), insulin-like growth factor 1 receptor (IGF1R) (Nahta et al., 2005), MET (Shattuck et al., 2008), and platelet-derived growth factor receptor (PDGFR) (Kim et al., 2023). That HER2 may partner with a large variety of RTKs to orchestrate oncogenic signaling poses a significant challenge to pharmacological suppression of HER2. The HER2 signaling partners may attenuate the impact of inhibitors of HER2 kinase activity. In this context, it is noteworthy that HER3 is kinase impaired but exerts strong oncogenic activity by forming signaling complexes with other RTKs (Lyu et al., 2018). Moreover, because a large number of RTKs may bind to HER2 to diversify its signaling, inhibiting one or two of these RTKs is unlikely to achieve an adequate and durable impact.

We recently discovered that PEPD^G278D, a recombinant and enzymatically inactive mutant of human peptidase D, binds to the extracellular domains of HER2 and EGFR and directs them to lysosomes for degradation (Yang et al., 2022; Yang et al., 2016a, 2019; Yang et al., 2014). PEPD^G278D binds to subdomain 3 in HER2 extracellular domain (Yang et al., 2014), whereas trastuzumab and pertuzumab bind to HER2 in subdomain 4 and 2, respectively (Cho et al., 2003; Franklin et al., 2004). PEPD^G278D binds to subdomain 2 in EGFR extracellular domain (Yang et al., 2016a), whereas EGFR-specific therapeutic antibodies, e.g., cetuximab (Li et al., 2005), bind to EGFR in subdomain 3. PEPD^G278D is a homodimeric protein with 493 amino acids in each subunit. We also found that PEPD^G278D targets overexpressed HER2 and EGFR, but not those expressed low in normal cells, because the formation of PEPD^G278D complex with each RTK requires high abundance of the RTK on cell surface (Yang et al., 2015, 2019). There is no evidence that PEPD^G278D binds directly to other RTKs. However, by inducing the degradation of HER2 and EGFR, PEPD^G278D may inactivate RTKs that heterodimerize with HER2 or EGFR. PEPD^G278D showed promising antitumor activity in our recent studies using cell lines and cell line-derived xenografts of HER2-positive BC (Yang et al., 2015, 2019). In the present study, we investigated the extent to which PEPD^G278D inhibits oncogenic signaling in trastuzumab-resistant HER2-positive BC cells and its activity against orthotopic patient derived xenografts (PDXs) resistant to trastuzumab. Orthotopic tumors locate within a physiological microenvironment comparable to that of the human disease, and PDX is well known to conserve the heterogeneity of original tumor and best predicts disease response to therapeutic intervention. We further investigated the ability of PEPD^G278D to inhibit metastatic lesions of trastuzumab-resistant HER2-positive BC. We used siRNA to discern the relative importance of HER2 and EGFR for growth of trastuzumab-resistant HER2-positive BC cells. We found that PEPD^G278D causes extensive inhibition of HER2-orchestrated signaling, overcomes resistance to trastuzumab, inhibits the growth of orthotopic PDXs, and inhibits the growth of metastatic lesions in brain and lung. We also found that eliminating EGFR is equally if not more important than eliminating HER2 for inhibiting trastuzumab-resistant HER2-positive BC cells. Our study demonstrates that drug-resistant HER2-positive BC cells and tumors are highly vulnerable to induction of simultaneous degradation of both HER2 and EGFR, and that PEPD^G278D is a highly promising agent for targeting this vulnerability.

2. Materials and Methods

2.1. Antibodies

The following antibodies were purchased from Cell Signaling: anti-EGFR (Cat # 2232), anti-pY1173-EGFR (Cat #4407), anti-HER2 (Cat #2165), anti-pY1221/1222-HER2 (Cat # 2243), anti-HER3 (Cat # 12708), anti-pY1328-HER3 (Cat # 14525), anti-IGF1R (Cat # 9750), anti-PY1131-IGF1R (Cat # 3021), anti-MET (Cat # 8198), anti-pY1234/1235-MET (Cat # 3077), anti-AXL (Cat # 8661), anti-pY779-AXL (Cat # 96453), anti-PDGFRα (Cat # 3174), anti-pY1018-PDGFRα (Cat # 4547), anti-FGFR2 (Cat # 11835), anti-pY653/654-FGFR2 (Cat # 3471), anti-AKT (Cat # 4691), anti-pS473-AKT (Cat # 4060), anti-ERK (Cat # 9102), anti-pT202/Y204-ERK (Cat # 9101), anti-focal adhesion kinase (FAK) (Cat # 3285), anti-pY397-FAK (Cat # 8556), anti-Janus kinase 1 (JAK1) (Cat # 3344), anti-pY1034/1035-JAK1 (Cat # 74129), anti-SRC (Cat # 2123), anti-pY416-SRC (Cat # 6943), anti-signal transducer and activator of transcription 3 (STAT3) (Cat # 4904), anti-pY705-STAT3 (Cat # 9145), anti-EPHB2 (Cat # 83029), anti-cleaved caspase 3 (Cat # 9661), anti-histone H3 (Cat # 4499), anti-voltage-dependent anion channel (VDAC) (Cat # 4661), anti-β-tubulin (Cat # 86298). Anti-EPHA4 (Cat # 37-1600), anti-PY594/604-EPHB1/EPHB2 (Cat # PA5-40236), and anti-mucin 4 (MUC4) (Cat # 35-4900) were purchased from Thermo Fisher Scientific. Anti-pY602-EPHA4 (Cat # EP2731) was purchased from ECM Biosciences. Anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (Cat # MAB374) was purchased from Millipore. Anti-mouse IgG conjugated to horseradish peroxidase (IgG-HRP) (Cat # NA931V) and anti-rabbit IgG-HRP (Cat # NA934V) were purchased from GE Healthcare.

2.2. Chemicals, biochemicals, and enzymes

Recombinant PEPD^G278D was generated in our lab as previously described (Yang et al., 2013) with minor modifications. Briefly, the full length human PEPD^G278D was subcloned from pBAD/Thio-PEPD^G278D (Yang et al., 2013) to pET21a, to generate expression vector pET21a-PEPD^G278D. The PEPD^G278D sequence was amplified by PCR using NdeI-forward 5’-GGCTACCTACATATGGCGGCGGCCACCGG-3’) and XhoI-reverse primers (5’-GGTGCTCGAGCTAATGGTGATGGTGATGATGCTTGGGGCCAGAGAAGGGGGTAAAGG-3’). Amplified PCR products were digested by NdeI and XhoI (New England BioLabs, Cat # R111 and R0146) and ligated into pET21a which was digested with the same restriction enzymes. The new construct was sequenced to confirm the integrity of the entire coding sequence. PEPD^G278D was expressed with C-terminal 6xhis tag in BL21(DE3) competent E. coli (Thermo Fisher Scientific, Cat # EC0114), purified by nickel chromatography, and concentrated in phosphate-buffered saline (PBS) using Ultracel YM-30 Centricon (Millipore, Cat # UFC803024). The high purity and dimer status of PEPD^G278D was confirmed using non-reducing sodium dodecyl-sulfate (SDS) polyacrylamide gel electrophoresis (PAGE) and protein staining. The PEPD^G278D preparations tested negative for endotoxin, using the E-TOXATE Kit (Sigma) which has a detection limit of 0.005 endotoxin unit per 0.1-ml sample. Garadacimab (GD) was purchased from ProteoGenix (Cat # PX-TA1560) or MedChemExpress (MCE, Cat # HY-P99631). The ProteoGenix product was used in dose-finding experiments, but for cost considerations, the MCE product was used later in the antitumor experiments. Following reagents were purchased from Sigma-Aldrich: Enoxaparin (EP) (Cat # 1235820), BSA (Cat # 9048-46-8), a protease inhibitor cocktail (Cat # P8215), dimethyl sulfoxide (DMSO) (Cat # M81802), phenylmethanesulfonyl fluoride (PMSF) (Cat # 329–98-6), phosphatase inhibitor cocktail 2 (Cat # P5726), phosphatase inhibitor cocktail 3 (Cat # P0044), and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (Cat # M2128). Lipofectamine RNAiMAX (Cat # 13778075) was purchased from Thermo Fisher Scientific. FuGENE 4K (Cat # E5911) was purchased from Promega. Phosphate buffer with 4% paraformaldehyde was purchased from Wako Chemicals (Cat # 163-20145). D-luciferin was purchased from Gold Biotechnology (Cat # LUCK-1G). Protein G-sepharose beads were purchased from Sigma (Cat # P3296). SDS (Cat # 161–0301) and 30% acrylamide/bis-acrylamide solution 29:1 (Cat # 1610156) were purchased from Bio-Rad. Cell lysis buffer (10x) was purchased from Cell Signaling (Cat # 9803).

2.3. Assay kits

The human RTK Phosphorylation Array C1 Kit was purchased from RayBiotech (Cat # AAH-PRTK-1-8). The Mitochondria Isolation Kit for Mammalian Cells (Cat # 89874) and the NE-PER Nuclear and Cytoplasmic Extraction Kit (Cat # 78833) were purchased from Thermo Fisher Scientific. The BCA Protein Assay Kit (reagent A: Cat # 23228; reagent B: Cat # 1859078) was purchased from Pierce. The Immobilon Classico Western HRP Substrate was purchased from Millipore (Cat # WBLUC0100). The SuperSignal West Pico PLUS Chemiluminescent Substrate was purchased from Thermo Fisher Scientific (Cat # 34577). The Clarity Western ECL Substrate was purchased from Bio-Rad (Cat # 170-5060).

2.4. Cell lines and cell culture

HCC1954 cell (Cat # CRL-2338) and MCF-7 cell (Cat # HTB-22) were purchased from the American Type Culture Collection. JIMT-1 cell was purchased from the Deutsche Sammlung von Mikroorganismen und Zellkulturen (Cat # ACC-589). MDA-IBC3 cells were generously provided by Dr. Wendy A. Woodward of the University of Texas MD Anderson Cancer Center. MCF-7/HER2 cells were generated by transfecting MCF-7 cells with pcDNA3-ERBB2 (Yang et al., 2014) and selection under G418 (Thermo Fisher Scientific, Cat # 10131035). Additional detail for generating MCF-7/HER2 cells is provided in the supplementary method. All cell lines were authenticated by short tandem repeat analysis and tested negative for mycoplasma. MCF-7 cells, MCF-7/HER2 cells and JIMT-1 cells were cultured in high-glucose Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS). HCC1954 cells were cultured in RPMI-1640 medium supplemented with 10% FBS. MDA-IBC3 cells were cultured in Ham’s F12 medium supplemented with 5 μg/ml insulin, 1μg /ml hydrocortisone, and 10% FBS. All cell lines were cultured in humidified incubators at 37 °C with 5% CO₂. High glucose DMEM (Cat # 10–013-CV), Ham’s F12 medium (Cat # 10–080-CV), and RPMI-1640 medium (Cat # 10–040-CV) were purchased from Corning Cellgro. FBS was purchased from Gibco (Cat # 10437). Insulin was purchased from Thermo Fisher Scientific (Cat # 12585014). Hydrocortisone was purchased from Sigma (Cat # H4881).

2.5. siRNA transfection

HCC1954 cells and JIMT-1 cells were grown in six-well plates (1 × 10⁵ cells/well with 2 ml medium) for 24 h and then transfected with nonspecific scramble siRNA, EGFR siRNA or HER2 siRNA (25 nM) using Lipofectamine RNAiMAX. All siRNAs were purchased from Origene. The siRNA sequences are as follows: rCrGrUrUrArArUrCrGrCrGrUrArUrArArUrArCrGrCrGrUA T (scramble siRNA, Cat # SR30004), rGrGrArUrArUrUrCrUrGrArArArArCrCrGrUrArArArGrGA A (EGFR siRNA, Cat # SR320009A), and rUrGrUrArArUrUrUrUrGrArCrArUrGrGrUrUrGrGrGrArCrUrCrUr U (HER2 siRNA, Cat # SR320064A).

2.6. Measurement of cell growth

Cell growth was measured by the MTT assay in 96-well plates. Each well was seeded with 3 x 10³ cells (MCF-7, MCF-7/HER2, JIMT-1, MDA-IBC3) or 4 x 10³ cells (HCC1954) with 100 μl culture medium overnight, treated with solvent, PEPD^G278D (5, 25 or 250 nM), or trastuzumab (250 nM) in 200 μl medium per well for 72 or 96 h, and then incubated with medium containing 9.2 mM MTT (200 μl/well) at 37 °C for 3 h. The cells were then washed with PBS and mixed with DMSO (150 μl per well). Cell density was determined by measuring the reduction of MTT to formazan spectroscopically at 570 nm using a Synergy HTX Multi-Mode Microplate Reader (BioTek).

2.7. Measurement of phosphorylation of tyrosine kinases

Tyrosine phosphorylation of human RTKs and non-RTKs was measured using the Human RTK Phosphorylation Antibody Array C1 Kit from RayBiotech, following manufacturer’s instructions. Briefly, cell lysates or tumor tissue homogenates (300-400 μg protein per sample) were added to each well containing the antibody array membrane and incubated overnight at 4 °C. After wash with wash buffer, 1 ml of the prepared biotinylated antibody cocktail was added to each well and incubated overnight at 4 °C. After another wash with wash buffer, 2 ml of 1x HRP-streptavidin was added into each well and incubated for 2 h at room temperature. The membrane was washed with wash buffer and incubated with 500 μl of detection buffer for 2 min at room temperature. The membrane was then imaged by the ChemiDoc MP Imaging System (Bio-Rad). Phosphorylation signal of each protein was quantified by ImageJ (National Institutes of Health).

2.8. Preparation of cell lysates, subcellular fractions, and tumor tissue homogenates

To prepare whole cell lysates, cells were washed with PBS twice, mixed with 1x cell lysis buffer from Cell Signaling supplemented with 2 mM PMSF, a protease inhibitor cocktail, phosphatase inhibitor cocktail 2, and phosphatase inhibitor cocktail 3, placed on ice for 10 min, and sonicated at 0-4 °C to enhance cell lysis using a Branson Model 450 sonifier. The lysates were centrifuged at 14,000 g for 10 min at 4 °C, and the supernatant fraction was collected as whole cell lysate. Mitochondria fraction and cytosolic fraction were prepared using the Mitochondria Isolation Kit, following the manufacturer’s instructions. Nuclear fraction was prepared using the NE-PER Nuclear and Cytoplasmic Extraction Kit, following the manufacturer’s instructions. Tumor samples were mixed with CelLytic MT Cell Lysis Reagent (Sigma, Cat # 3228), which was supplemented with 2 mM PMSF, the protease inhibitor cocktail, phosphatase inhibitor cocktail 2, and phosphatase inhibitor cocktail 3 at 14.3 μl buffer per mg tissue, and homogenized in a Dounce homogenizer. The homogenates were cleared by centrifugation at 14,000 g for 15 min at 4 °C.

2.9. Western blotting (WB) and immunoprecipitation (IP)

Protein concentrations in whole cell lysates, tissue homogenates, and other samples were measured by the BCA assay. For WB, each sample was resolved by SDS-PAGE (8 to 12.5% polyacrylamide). Proteins were transferred to polyvinylidene fluoride membrane, probed with specific antibodies, and detected using either Luminata Classico, SuperSignal West Pico or Clarity Western ECL. For IP, whole cell lysates (0.5-1 mg protein per sample) were incubated overnight at 4°C with the required antibody in 500 μl of volume, followed by incubation with 30 μl of protein G sepharose beads (2 mg/ml) for 1 h at room temperature. The beads were washed with IP buffer, suspended in 2x SDS loading buffer, boiled for 5 min, and analyzed by WB.

2.10. Mouse studies

All mouse experiments were approved by the Institutional Animal Care and Use Committee at Virginia Commonwealth University, under protocol AD10003095. NSG mice were either purchased from the Jackson Laboratory or the Cancer Mouse Models Core at the Massey Comprehensive Cancer Center of Virginia Commonwealth University. SCID/Beige mice (C.B-17/IcrHsd-Prkdc scid Lystbg-J) were purchased from Envigo. Mice were randomized cage-wise into treatment groups using Research Randomizer (www.randomizer.org) once significant tumor growth was detected. Tumor size was measured using length x width² ÷ 2.

2.10.1. Orthotopic HBCx-73 PDX

Female NSG mice (7-8 weeks of age) were used. The HBCx-73 PDX was first expanded in donor mice and then implanted to experimental mice. Approximately 20 mm³ tumor fragment was implanted into the inguinal 4th mammary fat pad using a trocar. The drinking water was supplemented with 1 μM 17β-estradiol (Sigma, Cat #50-28-2) to support PDX growth, which was changed twice weekly. The PDX was used in three experiments. In the first experiment, mice bearing the PDX were treated with EP, trastuzumab, or EP plus PEPD^G278D. Treatment with trastuzumab or PEPD^G278D was started once tumor size reached approximately 100 mm³, and EP treatment was started 2 days earlier to prime the mice for PEPD^G278D. Trastuzumab was administered at 10 mg/kg weekly. PEPD^G278D was administered at 5 mg/kg three times weekly (Monday, Wednesday, and Friday). EP was administered at 0.5 mg/kg daily. Drug treatments lasted 14-16 days. In the second experiment, mice bearing the PDX were treated with vehicle, GD, or GD plus PEPD^G278D, when their tumors reached approximately 110 mm³. A single dose of GD (40 mg/kg) was administered to the mice 3 h before the first dose of PEPD^G278D which was administered to the mice at 5 mg/kg three times weekly as described above. PEPD^G278D treatment lasted 19 days. Because tumors either completely disappeared or were too small to provide tissues for molecular analysis in the mice that were treated with PEPD^G278D plus either EP or GD in the above experiments, a third experiment was carried out, in which mice bearing the PDX were treated for a short period of time with EP, trastuzumab, or EP plus PEPD^G278D, and treatments were not initiated until tumor size reached approximately 230 mm³. EP was administered at 0.5 mg/kg daily, which was started two days before PEPD^G278D. Trastuzumab was administered at 10 mg/kg once weekly for a total of two doses. PEPD^G278D was administered at 5 mg/kg three time weekly for a total of 4 doses. All mice were closely monitored for sign of adverse effects and were euthanized 24 hours after the last treatment, at which point the tumors were promptly removed, snap-frozen, and stored at −80 °C for later analysis. Necropsy was performed on all the mice.

2.10.2. Orthotopic HBCx-5 PDX

Female NSG mice (7-8 weeks of age) were used. HBCx-5 PDX was first expanded in donor mice and then implanted to experimental mice. Approximately 20 mm³ tumor fragment was implanted into the inguinal 4th mammary fat pad using a trocar. The drinking water was supplemented with 1 μM 17β-estradiol to support tumor growth, which was changed twice weekly. Mice bearing the PDX were used in two experiments. In the first experiment, mice bearing the PDX were treated with vehicle or trastuzumab. Trastuzumab was administered at 10 mg/kg weekly for 3 weeks and was started once tumor size reached approximately 135 mm³. In the second experiment, mice bearing the PDX was treated with GD, or GD plus PEPD^G278D. Treatments were started when tumor size reached approximately 100 mm³. PEPD^G278D was administered at 5 mg/kg three times weekly for 4.6 weeks. Two doses of GD at 40 mg/kg were administered to the mice, with the first dose administered 3 h before the first dose of PEPD^G278D, and the second dose administered 3 weeks later. All mice were closely monitored for sign of adverse effects and were euthanized 24 h after the last treatment; the tumors were promptly removed, snap-frozen, and stored at −80°C for later analysis. Necropsy was performed on all the mice.

2.10.3. MDA-IBC3 tumor metastasis model

Female SCID/Beige mice were used for the MDA-IBC3 tumor model. Approximately 1 x 10⁶ MDA-IBC3 cells stably expressing LUC were injected intravenously via tail vein to the mice (5 weeks of age) in 50 μl PBS. Mice were monitored by bioluminescence 1-2 times each week, which was started 4 weeks after cell injection to assess tumor burden. Mice were injected with D-luciferin (150 mg/kg), anesthetized with isoflurane, and imaged using the IVIS Spectrum In Vivo Imaging System (PerkinElmer) within 4-8 min of luciferin injection. Tumor-bearing mice were treated with EP, or EP plus PEPD^G278D. EP was administered at 0.5 mg/kg daily, which was started 33 days after cell injection and ended 25 days later. PEPD^G278D was administered at 5 mg/kg three times each week, which was started 35 days after cell injection and ended 23 days later. At the end of the experiment, the mice were injected with luciferin, anesthetized with isoflurane, imaged for bioluminescence, and then immediately euthanized, from which whole brains were rapidly removed and imaged for bioluminescence. Necropsy was performed on all the mice.

2.10.4. Dosing of study agents

All agents were administered to mice by intraperitoneal injection (ip). EP and PEPD^G278D were administered to mice in PBS. Trastuzumab was first prepared in water at 21 mg/ml and diluted with PBS for administration. GD purchased from MCE was used, which was supplied at 7.92 mg/ml in 100 mM proline-acetic acid and 20 mM arginine, pH 5.0. It was diluted in PBS for administration. Each agent was administered to mice in 60-100 μl volume per 20 g body weight. When a mouse was given both EP and PEPD^G278D on the same day, the agents were dosed at approximately 30 min interval. When GD and PEPD^G278D were given to the same mice on the same day, GD was dosed 3 h before PEPD^G278D.

2.11. Statistical analysis

All statistical analyses were carried out using GraphPad Prizm. For two-group comparison, we used paired two-tailed t test or the non-parametric Mann-Whiney U test when the assumptions of t test are not met. For multigroup comparisons, we used ANOVA followed by Tukey test. P value of 0.05 or lower was considered statistically significant.

3. Results

3.1. PEPD^G278D causes extensive inhibition of tyrosine kinases in HER2-positive BC cells despite resistance to trastuzumab

We used four human BC cell lines to assess the potential impact of PEPD^G278D on the activities of a large number of protein tyrosine kinases, including HCC1954, JIMT-1, MCF-7, and MCF-7/HER2. PEPD^G278D was compared to trastuzumab in HCC1954 cells and JIMT-1 cells. HCC1954 and JIMT-1 are HER2-positive BC cells and overexpress both HER2 and EGFR (Supplementary Fig. 1). Both cell lines also carry activating mutation in PIK3CA and have low PTEN (Yang et al., 2019). PEPD^G278D strongly inhibited the growth of HCC1954 and JIMT-1 cells, but neither cell line was sensitive to trastuzumab (Fig. 1A). These results are consistent with our previous finding that PEPD^G278D but not trastuzumab depletes HER2 and EGFR in these cells and that PEPD^G278D but not trastuzumab inhibits the growth of JIMT-1 tumors in mice (Yang et al., 2019). Both HCC1954 cells and JIMT-1 cells overexpress cell surface protein MUC4 (Supplementary Fig. 1), which is known to prevent trastuzumab from binding to HER2 (Nagy et al., 2005). Indeed, trastuzumab binding is undetectable in HCC1954 cells and JIMT-1 cells (Supplementary Fig. 2). MUC4 is overexpressed in a high percentage of human HER2-positive BC cases and is associated with poor disease prognosis (Mercogliano et al., 2017). MCF-7 cells which are estrogen receptor-positive have negligible levels of HER2 and EGFR (Supplementary Fig. 1), and as expected, were sensitive to neither PEPD^G278D nor trastuzumab (Fig. 1A). However, stably overexpressing HER2 sensitizes MCF-7 cells to both PEPD^G278D and trastuzumab, indicating oncogene addiction. Still, PEPD^G278D is more potent than trastuzumab in inhibiting MCF-7/HER2 cells, including inhibition of cell growth and HER2 signaling (Supplementary Fig. 3A–B), even though these cells do not express MUC4 (Supplementary Fig. 1) and trastuzumab binding is readily detected (Supplementary Fig. 2).

Fig. 1.

PEPD^G278D inhibits the phosphorylation of a diverse group of protein tyrosine kinases in trastuzumab-resistant HER2-positive BC cells. (A) Effects of PEPD^G278D and trastuzumab on cell growth measured by MTT assay. Cells were treated by each agent for 72 h. Each value is mean ± SD (n=3). P values are based on ANOVA, compared to control (Tukey test), **** P <0.0001. (B) Profiling of tyrosine kinase phosphorylation using an antibody array. Cells were treated by PEPD^G278D or trastuzumab for 48 h. Each value is an average of two independent measurements (See Supplementary Table 1 for raw data). * Kinases whose phosphorylation is undetectable, including EPHB2, EPHB6 and PDGFRβ in both JIMT-1 cells and MCF-7 cells, EPHB4 in MCF-7 cells, and HER4 in JIMT-1 cells.

We assessed the impact of PEPD^G278D on the activities of a large number of tyrosine kinases using an antibody array, which simultaneously measures tyrosine phosphorylation (activation) of 44 RTKs and 27 non-RTKs (Supplementary Table 1). PEPD^G278D was compared to trastuzumab in HCC1954 cells and JIMT-1 cells, while only PEPD^G278D was evaluated in MCF-7 cells and MCF-7/HER2 cells. Cells were treated with solvent, 25 nM PEPD^G278D, or 250 nM trastuzumab for 48 h. PEPD^G278D inhibited tyrosine phosphorylation of 29 RTKs (66% of them) and 15 non-RTKs (56% of them) by at least 50% in HCC1954 cells and/or JIMT-1 cells, but did not inhibit the phosphorylation of any of them in MCF-7 cells (Fig. 1B; Supplementary Table 1). PEPD^G278D inhibited tyrosine phosphorylation of a diverse group of RTKs and non-RTKs in the HER2-positive BC cell lines. The RTKs that are inhibited by PEPD^G278D encompass 10 subfamilies, including 1) anaplastic lymphoma kinase (ALK), 2) AXL, and its family member DTK, 3) HER2 and its family members, including EGFR, HER3, and HER4, 4) ephrin type-A receptor 1 (EPHA1) and its family members, including EPHA2, EPHA3, EPHA4, EPHA6, EPHA7, and EPHA8, 5) ephrin type-B receptor 1 (EPHB1) and its family members, including EPHB2, EPHB3, EPHB4, and EPHB6, 6) FGFR1, and its family members, including FGFR2, and FGFR2α, 7) MET, also known as hepatocyte growth factor receptor (HGFR), 8) IGF1R, and its family member insulin receptor (INSULIN R), 9) PDGFRβ, and 10) TYRO10 (Fig. 1B). The majority (73%) of the RTKs inhibited by PEPD^G278D have been previously shown to dimerize with HER2 (Zhang, 2021). The non-RTKs that are inhibited by PEPD^G278D include ABL1, the SRC family kinases (BLK, FGR, FYN, HCK, and LCK), BMX, Bruton’s tyrosine kinase (BTK), C-terminal SRC kinase (CSK), focal adhesion kinase (FAK), FER, Fyn-related kinase (FRK), and Janus tyrosine kinases (JAK1, JAK2, and JAK3) (Fig. 1B). The extensive inhibitory impact of PEPD^G278D on the tyrosine kinases is apparently mediated by HER2 and EGFR, because in MCF-7 cells, PEPD^G278D had no effect on tyrosine phosphorylation of any of the kinases described above (Fig. 1B), but in MCF-7/HER2 cells, it inhibited tyrosine phosphorylation of 75% of the RTKs analyzed and 59% of the non-RTKs analyzed by 50% or more (Supplementary Fig. 3C). Notably, PEPD^G278D inhibited EPHA3 in JIMT-1 cells but activated EPHA3 tyrosine in HCC1954 cells (Fig. 1B), and in MCF-7 cells, PEPD^G278D inhibited ROS and TRKB and activated LCK (Supplementary Table 1). The reason for PEPD^G278D activation of EPHA3 in HCC1954 cells as well as its activation of LCK and inhibition of ROS and TRKB in MCF-7 cells is unknown, but it seems to be unrelated to targeting of HER2 or EGFR.

Only four of the 71 tyrosine kinases analyzed were significantly inhibited by trastuzumab, including HER4 and EPHB6 in HCC1954 cells, and JAK2 and TYRO10 in JIMT-1 cells (Fig. 1B), even though trastuzumab concentration was 10 times higher than that of PEPD^G278D. Trastuzumab increased tyrosine phosphorylation of PDGFRβ and TRKB in HCC1954 cells and EPHB3 in JIMT-1 cells by more than 2-fold (Fig. 1B; Supplementary Table 1). The reasons for the stimulating effects of trastuzumab on EPHB3, PDGFRβ and TRKB described above are not known but they are cell line-specific.

3.2. PEPD^G278D disrupts the broad signaling network orchestrated by HER2 that is nonresponsive to trastuzumab

Because HCC1954 cells and JIMT-1 cells respond to PEPD^G278D similarly (Fig. 1), we next used one of them, HCC1954 cells, to assess specific changes in key signaling proteins induced by PEPD^G278D. Trastuzumab was included for comparison. Cells were treated by PEPD^G278D at 25 nM or trastuzumab at 250 nM for 48 h. As expected, PEPD^G278D caused profound loss of both EGFR and HER2 in both expression and tyrosine phosphorylation, whereas trastuzumab failed to do so (Fig. 2A). Several other RTKs that have been implicated in drug resistance in HER2-positive BC were also examined, including AXL, EPHB4, FGFR2, HER3, IGF1R, MET, and PDGFR. We were unable to adequately detect the expression or phosphorylation of AXL, EPHB4 and PDGFRβ. Therefore, we included PDGFRα, EPHA4, and EPHB2, whose phosphorylation levels in HCC1954 cells were reduced by PEPD^G278D by 46.5%, 82.5%, and 100%, respectively, in the antibody array assay described before (Supplementary Table 1). PEPD^G278D caused marked loss of tyrosine phosphorylation in HER3, IGF1R, MET, FGFR2, EPHA4, and PDGFRα without altering their expression level (Fig. 2A). PEPD^G278D decreased both expression and phosphorylation of EPHB2 (Fig. 2A), but loss of EPHB2 expression is cell line-specific, as it does not occur in JIMT-1 cells (Supplementary Fig. 4) or in the PDX described later. Trastuzumab showed no effect on these RTKs (Fig. 2A). Loss of phosphorylation but not expression of the RTKs in response to PEPD^G278D apparently resulted from disruption of their binding to HER2 or EGFR. While HER3, IGF1R, MET, FGFR2, PDGFRα, EPHA4 and EPHB2 each bind to HER2, PEPD^G278D caused marked disruption of each complex (Fig. 2B). Notably, EGFR was previously shown to bind to HER3, IGF1R, MET, FGFR2, and PDGFRα (Zhang, 2023), and such complexes, if present in the study cells, may also be disrupted by PEPD^G278D. We next measured various signaling proteins, downstream to RTK, that represent key oncogenic signaling pathways, including AKT, ERK, FAK, JAK1, SRC, and STAT3. PEPD^G278D decreased the phosphorylation but not expression of all of these proteins, whereas trastuzumab had no effect on them (Fig. 2C). The impact of PEPD^G278D on various RTKs and downstream signaling proteins is summarized in Fig. 2D. Both HER2 and EGFR have been shown to translocate to and exert oncogenic functions in the mitochondria and nucleus (Zhang, 2021, 2023). PEPD^G278D-induced degradation of membrane HER2 and EGFR in HCC1954 cells was accompanied by loss of both RTKs in mitochondria and nucleus (Fig. 2E). This suggests that PEPD^G278D may also inhibit the oncogenic activities of HER2 and EGFR in noncanonical locations.

Fig. 2.

PEPD^G278D obliterates oncogenic signaling orchestrated by HER2 in trastuzumab-resistant HER2-positive BC cells. HCC1954 cells were treated by PEPD^G278D at 25 nM or trastuzumab at 250 nM for 48 h. (A) Whole cell lysates were analyzed by WB for expression and phosphorylation of various RTKs. (B) Whole cell lysates were subjected to IP followed by WB. Note: Sample volume was adjusted for EPHB2 in order to load similar amount of EPHB2 between the two samples. (C) Whole cell lysates were analyzed by WB for expression and phosphorylation of various non-RTKs. (D) Graphic summary of the impact of PEPD^G278D on RTKs and non-RTKs. (E) Subcellular fractions, including cytosol plus membrane, mitochondria, and nuclear extract, were prepared and analyzed by WB. GAPDH, β-tubulin, VDAC, and histone H3 were used as loading controls. Whole HCC1954 cell lysates were used as a positive control for histone H3, β-tubulin, and VDAC in different cellular fractions. The specific phosphorylation sites measured include pY1221/1222-HER2, pY1173-EGFR, pY1328-HER3, pY1131-IGF1R, pY1234/1235-MET, pY653/654-FGFR2, pY1018-PDGFRα, pY602-EPHA4, pY594/604-EPHB2, pS473-AKT, pT202/Y204-ERK, pY397-FAK, pY1034/1035-JAK1, pY416-SRC, and pY705-STAT3.

Because PEPD^G278D induces the degradation of both HER2 and EGFR, we used siRNA to assess the relative importance of degradation of each protein. Knockdown (KD) of HER2 and/or EGFR by siRNA in HCC1954 cells and JIMT-1 cells inhibited cell growth (Fig. 3A–B). EGFR KD was as effective as HER2 KD in JIMT-1 cells and more effective than HER2 KD in HCC1954 cells. The finding that HCC1954 cells rely more on EGFR than HER2 for proliferation may be related to EGFR expression level, as EGFR level in these cells is much higher than in JIMT-1 cells (Supplementary Fig. 1). The above result is also consistent with existing understanding that EGFR plays an important role in drug-resistant HER2-positive BC cells and suggests that targeting the degradation of HER2 alone may not be sufficient for killing drug-resistant HER2-positive BC cells.

Fig. 3.

Depleting both HER2 and EGFR is important for maximal inhibition of the growth of trastuzumab-resistant HER2-positive BC cells. JIMT-1 cells and HCC1954 cells were treated by scramble siRNA, EGFR siRNA, HER2 siRNA, or combination of EGFR siRNA and HER2 siRNA at 25 nM each for 7 days. (A) Western blot analysis of EGFR and HER2. (B) Cell growth measured by MTT assay. Each value is mean ± SD (n=3). P values are based on ANOVA, compared to control (Tukey test), **** P < 0.0001.

3.3. PEPD^G278D is highly effective against trastuzumab-resistant HER2-positive orthotopic PDXs

We next evaluated the therapeutic activity of PEPD^G278D in vivo using two HER2-positive PDXs which were previously shown to be resistant to trastuzumab, including HBCx-73 (BC911LJ) and HBCx-5 (BC111VINC). These PDXs were established at Institut Curie and have been previously reported (Lefort et al., 2017; Marangoni et al., 2018). Briefly, HBCx-73 PDX was derived from a patient with HER2-positive BC treated with trastuzumab in the neoadjuvant setting. HBCx-5 PDX was derived from a patient with HER2-positive BC received neoadjuvant therapy, docetaxel as first line, and trastuzumab as second line. Both PDXs overexpress MUC4 (Supplementary Fig. 1) and harbor amplification of HER2, EGFR and SRC (Supplementary Table 2). In addition, HBCx-5 also harbors amplification of AKT, CCND1 which encodes cyclin D1, HER3, and IGF1R, and HBCx-73 also harbors amplification of CCNE1 which encodes cyclin E1 (Supplementary Table 2). While the significance of cyclin D1 overexpression is unclear in HER2-positive BC, cyclin E1 overexpression is linked to poor response to trastuzumab (Scaltriti et al., 2011). AKT, EGFR, HER3, IGF1R, MUC4, and SRC are also well known to contribute to drug resistance and poor prognosis in HER2-positive BC. Trastuzumab was included for comparison with PEPD^G278D. Tumor fragments from donor mice were implanted to the mammary fat pads of experimental female NSG mice. We first evaluated PEPD^G278D in HBCx-73 PDX. Treatment of tumor-bearing mice was started when tumor size reached approximately 100 mm³. Mice were treated with PEPD^G278D at 5 mg/kg three times weekly based on our previous studies. Mice were treated with trastuzumab at 10 mg/kg once weekly, as previously reported (Lefort et al., 2017; Marangoni et al., 2018). PEPD^G278D was used in combination with EP which was administered at 0.5 mg/kg daily. PEPD^G278D is degraded by coagulation proteases in blood circulation in vivo, but EP, a clinically used anticoagulant, blocks the degradation by inhibiting the coagulation proteases (Supplementary Fig. 5A–B) (Yang et al., 2015, 2016b). EP itself has no antitumor activity (Yang et al., 2016a, 2015). While tumors grew rapidly in mice treated by EP alone, and trastuzumab was ineffective, tumor growth was inhibited by 98% in both tumor volume and weight after two weeks of treatment with PEPD^G278D plus EP, and 36% of the tumors showed complete regression (Fig. 4A–B). In another experiment, we replaced EP with GD, a monoclonal antibody targeting activated FXII which initiates the PEPD^G278D degradation pathway (Supplementary Fig. 5C). GD is currently being developed by CSL Behring for treatment of hereditary angioedema by targeting activated FXII (Craig et al., 2023). We hypothesized that GD not only blocks PEPD^G278D degradation but also may not require frequent dosing, and therefore may be more attractive than EP for combination with PEPD^G278D in potential clinical evaluation. Indeed, we found that a single dose of GD provides strong protection against PEPD^G278D degradation for three weeks, and 40 mg/kg is the optimal dose for GD (Supplementary Fig. 5D). A single dose of GD at 40 mg/kg had no effect on HBCx-73 growth, but combining GD with PEPD^G278D resulted in complete remission of all tumors after less than 3 weeks of treatment (Fig. 4C–D). We confirmed that plasma concentration of PEPD^G278D was high at the end of the experiment. Specifically, the mice were first treated with a single dose of GD at 40 mg/kg shortly before the first dose of PEPD^G278D (5 mg/kg), and PEPD^G278D treatment was continued three times weekly for 20 days. The mean plasma level of PEPD^G278D at 24 h after the final PEPD^G278D dose was 269.2 nM (Supplementary Fig. 5E). Notably, plasma PEPD^G278D was measured by ELISA which detects homodimeric PEPD^G278D, confirming that its natural state is maintained in vivo.

Fig. 4.

PEPD^G278D strongly inhibits trastuzumab-resistant orthotopic HER2-positive HBCx-73 PDX. (A and B) Mice carrying orthotopic HBCx-73 PDX were treated with EP, trastuzumab, or EP plus PEPD^G278D. EP (0.5 mg/kg) was administered by ip daily (days 10-26). Trastuzumab (10 mg/kg) was administered ip weekly (days 12-26). PEPD^G278D (5 mg/kg) was administered ip three times weekly (days 12-26). Each value is mean ± SEM, n=12-14. **** P < 0.0001, t test. (C and D) Mice were treated with vehicle, GD, or GD plus PEPD^G278D. A single dose of GD (40 mg/kg) was administered ip on day 13. PEPD^G278D (5 mg/kg) was administered ip three times weekly (days 13-32). Each value is mean ± SEM (n=12). **** P < 0.0001, t test. (E and F) Mice were treated with EP, trastuzumab, or EP plus PEPD^G278D, using the same doses as described above. EP was administered daily (days 18-28). Trastuzumab was administered on days 21 and 28. PEPD^G278D was administered three time weekly (days 21-28). Each value is mean ± SEM (n=6), * P < 0.05, ** P < 0.01, t test. (G-I) WB analysis (G and H) and IP followed by WB (I) of signaling proteins in tumor tissues. Tumors from two mice per group were analyzed, which were obtained 24 h after the final treatments in (E and F). See Fig. 2 legend for protein phosphorylation sites.

A new experiment with a short treatment time for EP, trastuzumab, and EP plus PEPD^G278D was carried out, in order to obtain tumor tissues for molecular analysis. Treatment with PEPD^G278D and trastuzumab was started when tumor volume reached approximately 250 mm³. Daily EP was started two days before PEPD^G278D. The mice were treated with EP, trastuzumab, and EP plus PEPD^G278D for about one week. Mice treated with PEPD^G278D showed 57% and 58% decrease in tumor weight and volume, respectively, while trastuzumab was again ineffective (Fig. 4E–F). Tumors treated by PEPD^G278D showed profound loss of both expression and phosphorylation of HER2 and EGFR, loss of phosphorylation but not expression of other RTKs analyzed, including FGFR2, IGF1R, MET, PDGFRα, EPHA4, EPHB2, and AXL (Fig. 4G). Notably, unlike in HCC1954 cells where AXL and p-AXL could not be detected adequately, both of which could be measured in HBCx-73 PDX. Interestingly, both phosphorylation and expression of HER3 was lost in the PEPD^G278D-treated tumors. PEPD^G278D also induced the loss of HER3 expression in the other PDX described later. The reason for HER3 downregulation by PEPD^G278D in the PDXs is unknown, although this is a positive finding. PEPD^G278D did not reduce HER3 expression in HCC1954 cells as described before. PEPD^G278D does not directly bind to HER3 (Yang et al., 2014) and does not downregulate HER3 expression in other HER2-positive tumor models (Yang et al., 2015, 2019) and a colorectal cancer PDX (Yang et al., 2022). None of the above-described changes were detected in tumors treated by trastuzumab (Fig. 4G). Tumors treated by PEPD^G278D also showed loss of phosphorylation but not expression of all other signaling proteins analyzed that are downstream to RTK, including AKT, ERK, FAK, JAK1, SRC, and STAT3, whereas trastuzumab-treated tumors did not show any of such changes (Fig. 4H). Moreover, PEPD^G278D, but not trastuzumab, induced caspase 3 cleavage, indicative of tumor apoptosis (Fig. 4H). We also measured the binding of PDGFRα, IGF1R, MET, AXL, FGFR2, EPHA4, and EPHB2 to HER2 in tumor tissues. Each RTK bound to HER2, but each hetero-complex was disrupted by PEPD^G278D (Fig. 4I). The impact of PEPD^G278D on the RTKs and non-RTKs in the tumors shown above closely resembles that in HCC1954 cells shown before.

We next evaluated PEPD^G278D in mice bearing orthotopic HER2-positive HBCx-5 PDX. In the first experiment, tumor-bearing mice were treated with vehicle or trastuzumab (10 mg/kg) weekly for three weeks. Trastuzumab showed no inhibitory impact on tumor growth (Fig. 5A–B), consistent with published data mentioned before. In the second experiment, tumor-bearing mice were treated with GD, or GD plus PEPD^G278D. As mentioned before, GD was used to inhibit PEPD^G278D degradation in vivo. GD was administered to mice at 40 mg/kg on two occasions, first on day 14 and then on day 35 after tumor implanting. PEPD^G278D was administered to mice at 5 mg/kg three times each week, which was started on day 14. The experiment was stopped after 34 days of PEPD^G278D treatment, due to marked tumor growth inhibition in the combination group and the need to obtain some tumor tissues for molecular analysis. At the end of the experiment, average tumor volume and weight in the combination group were only 10.5% and 16.2% of that in the GD group, respectively i.e., inhibiting tumor growth by 83.8-89.5% (Fig. 5C–D). Tumors in mice treated by GD plus PEPD^G278D never showed any growth and instead appeared to show a trend of slow regression. We were able to analyze only some of the signaling molecules due to limited sample availability. Tumors treated by PEPD^G278D showed profound loss of both expression and phosphorylation of HER2, EGFR and HER3, loss of phosphorylation but not expression of other RTKs analyzed, including IGF1R, MET, and AXL, loss of phosphorylation but not expression of downstream signaling proteins including AKT, ERK, SRC, and STAT3, and caspase 3 activation (Fig. 5E). We also measured HER2 binding to other RTKs, including AXL, IGF1R, and MET, and all bindings were disrupted in tumors treated by PEPD^G278D (Fig. 5F). These changes are not different from those shown in HBCx-73 PDX. As mentioned before, it is not known why PEPD^G278D reduces HER3 expression level. Neither trastuzumab nor GD had any impact on the signaling proteins and caspase 3 described above (Fig. 5E).

Fig. 5.

PEPD^G278D strongly inhibits trastuzumab-resistant orthotopic HER2-positive HBCx-5 PDX. (A and B) Mice bearing HBCx-5 PDX was treated with vehicle or trastuzumab. Trastuzumab (10 mg/kg) was administered to the mice ip weekly (days 21-40). Each value is mean ± SEM (n=12). (C and D) Mice bearing HBCx-5 PDX was treated with GD, or GD plus PEPD^G278D. GD (40 mg/kg) was administered to the mice ip on days 14 and 35. PEPD^G278D (5 mg/kg) was administered to the mice by ip three times weekly (days 14-47). Each value in is mean ± SEM (n=13-14), **** P < 0.0001, t test. (E and F) WB analysis (E) and IP followed by WB (F) of signaling proteins in tumor tissues. Tumors from two mice per group were analyzed, which were obtained 24 h after the final treatments in (A-D). See Fig. 2 legend for protein phosphorylation sites.

3.4. PEPD^G278D inhibits metastatic lesions of trastuzumab-resistant HER2-positive BC

We next evaluated PEPD^G278D in a metastatic lesion model of HER2-positive BC. MDA-IBC3 cells were previously isolated from pleural effusion fluid from a patient with HER2-positive inflammatory BC (Klopp et al., 2010). Prior treatment of this patient is not known. MDA-IBC3 cells overexpress HER2 and metastasize to brain and lung after intravenous injection (Debeb et al., 2016). Inflammatory BC is an aggressive subtype associated with rapid onset, early metastasis and poor survival (Giordano and Hortobagyi, 2003). Many oncogenic drivers are amplified in MDA-IBC3 cells, including MYC, FAK1, ATAD2 (ATPase family AAA domain-containing protein 2), MTDH1 (metadherin), PVT1 (plasmacytoma variant translocation 1), and CD44 (Fernandez et al., 2013). MDA-IBC3 cells overexpress both HER2 and EGFR but do not express MUC4 (Supplementary Fig. 1). PEPD^G278D strongly inhibited the growth of MDA-IBC3 cells in vitro in a dose-dependent manner, whereas trastuzumab was ineffective (Fig. 6A). PEPD^G278D abolished the expression and phosphorylation of both HER2 and EGFR in these cells, whereas trastuzumab was ineffective (Fig. 6B). Trastuzumab binding was readily detected in these cells (Supplementary Fig. 2), indicating that resistance to trastuzumab is not due to lack of its binding to HER2. We also analyzed several major signaling proteins downstream to HER2 and EGFR, including SRC, AKT, and ERK. PEPD^G278D caused the loss of phosphorylation but not expression of SRC, AKT and ERK, whereas trastuzumab had no effect on either phosphorylation or expression of these proteins (Fig. 6B). Because the impact of PEPD^G278D on these signaling proteins in MDA-IBC3 cells is virtually identical to that in HCC1954 cells and the PDXs, we did not measure the effect of PEPD^G278D on other signaling proteins in these cells. We next injected MDA-IBC3 cells, which stably express firefly luciferase (LUC), to female SCID/Beige mice via tail vein and monitored tumor growth by bioluminescence (Fig. 6C). The mice showed significant tumor growth 30 days after cell injection and were treated with either EP, or EP plus PEPD^G278D EP daily at 0.5 mg/kg and PEPD^G278D at 5 mg/kg three times weekly were started on day 33 and 35 following MDA-IBC3 cell injection, which lasted 25 and 23 days, respectively. The experiment was stopped due to progressive mouse body weight loss, mainly in the EP only group (Supplementary Fig. 6), which apparently resulted from disease progression. At the end of the experiment, PEPD^G278D reduced total tumor burden by 95%, based on whole body bioluminescence imaging (Fig. 6D). Bioluminescence imaging indicated tumor growth in the head and elsewhere (Supplementary Fig. 7A). However, it was not possible to measure bioluminescence from the head without the interference of signal from the chest area in a live mouse. Therefore, each whole brain was imaged for bioluminescence immediately after removal from the mouse which was injected with luciferin, underwent full body bioluminescence imaging and then euthanized. Bioluminescence signal indicates tumor presence in every brain, with total flux ranging from 2 x 10⁷ to 5.3 x 10⁸ photons/second in the EP group, and 1.6 x 10⁶ to 2.1 x 10⁸ photons/second in the EP plus PEPD^G278D group (Supplementary Fig. 7B–C). Brains from age-matched mice that did not previously receive MDA-IBC3 cells showed total flux of < 1 x 10⁵ photons/second (Supplementary Fig. 7D). PEPD^G278D treatment reduced brain tumor burden by 64% (Fig. 6E). The bioluminescence result mentioned above suggests that brain tumor response to PEPD^G278D may vary significantly among the mice, which could be related to the location of the tumor or tumor size at treatment start. Macroscopic metastatic brain lesions were not seen. Representative microscopic brain tumors are shown by hematoxylin and eosin (H/E) staining (Supplementary Fig. 8). Macroscopic metastatic lung lesions were not seen during necropsy, but microscopic lung tumors were detected (Supplementary Fig. 9A). Necropsy did not reveal morphological abnormalities in other organs and tissues in any of the mice. To estimate lung tumor burden, total photon flux of the chest, neck and head which was measured shortly before euthanasia (Supplementary Fig. 9B) was subtracted by the total photon flux of the whole brain removed from the mouse, to calculate total chest photon flux as estimate of lung tumor burden. The total chest photon flux was high in each mouse treated with EP only, ranging from 3.2 x 10⁷ to 3.7 x 10¹⁰ photons per second, but it could be detected only in two of the mice treated with EP plus PEPD^G278D, both of which showed low signal (total chest flux of 1.4 x 10⁶ and 9.3 x 10⁶ photons per second, respectively) (Fig. 6F). Thus, PEPD^G278D appears to markedly inhibit lung metastatic lesion. Because macroscopic lesions were not detectable in the brain and lung, analysis of potential impact of PEPD^G278D on signaling proteins in tumors in these sites were not possible.

Fig. 6.

PEPD^G278D inhibits metastatic lesions of trastuzumab-resistant HER2-positive BC. (A) MDA-IBC3 cells were treated with PEPD^G278D or trastuzumab for 72 h and then analyzed for cell growth by MTT assay. Each value is mean ± SD (n=3). P values are based on ANOVA, compared to control (Tukey test), **** P < 0.0001. (B) MDA-IBC3 cells were treated by PEPD^G278D or trastuzumab for 48 h, and whole cell lysates were prepared and analyzed for various signaling proteins by WB. See Fig. 2 legend for protein phosphorylation sites. (C) The metastatic tumor lesion model and experimental treatment. Mice were injected intravenously with MDA-IBC3 cells labeled with LUC, monitored for tumor growth by bioluminescence imaging, and treated by EP, or EP plus PEPD^G278D. EP (0.5 mg/kg) was administered ip daily (33-58 days after cell injection). PEPD^G278D (5 mg/kg) was administered ip three times weekly (35-58 days after cell injection). (D) Whole body tumor burden measured by bioluminescence. Each value is mean ± SEM (n=7), * P < 0.05, t test. (E) Brain tumor burden measured by bioluminescence after removing from the mice. * P <0.05, t test. (F) Putative lung tumor burden (bioluminescence signal of chest and above subtracted by brain bioluminescence). ** P < 0.01, Mann-Whitney U test. Both individual values and mean ± SEM are shown in (E and F).

4. Discussion

Numerous mechanisms of primary and secondary resistance to HER2-targeted therapies in HER2-positive BC have been shown. In the case of trastuzumab, which has been the mainstay of anti-HER2 treatment, its binding to HER2 is hindered by MUC4 which is overexpressed in up to 60% of HER2-positive BCs (Mercogliano et al., 2017). MUC4 may also block binding of other therapeutic antibodies and antibody-drug conjugates to HER2. Even in the absence of MUC4, trastuzumab is a poor inducer of HER2 downregulation (Mohsin et al., 2005), which is also shown in our present study. Many RTKs have been shown to confer resistance to trastuzumab and other HER2 inhibitors, as mentioned before. Our present study shows that these RTKs may confer resistance to HER2 inhibitors in HER2-positive BC primarily by forming signaling complexes with HER2. Various molecular changes downstream to HER2 and other RTKs have also been shown to confer resistance to trastuzumab and other anti-HER2 therapies in this disease, including, but not limited to, compensatory signaling (e.g., PI3K pathway activation) (Jensen et al., 2012), dysregulation of cell cycle regulators (e.g., cyclin E1 overexpression) (Scaltriti et al., 2011), and activation of non-RTKs (e.g., SRC) (Zhang et al., 2011). Many of these resistance mechanisms are present in the experimental models used in our present study. Also implicated in trastuzumab resistance is aberrant immune response, such as HLA-G desensitization of cancer cells to trastuzumab by inhibiting NK cell mediated antibody-dependent cellular cytotoxicity (Zheng et al., 2021). Moreover, HER2-positive BC cells may lose HER2 expression or HER2-negative BC cells may be selected in a heterogenous tumor during a prolonged anti-HER2 treatment (Mittendorf et al., 2009). Given the diverse drug resistance mechanisms, the challenge is to ascertain which mechanism is most important to target. To date, the relative importance of these drug resistance mechanisms has been poorly understood.

Our present study sheds light on key vulnerability of trastuzumab-resistant HER2-positive BC, i.e., the most important resistance mechanism, and also shows a highly effective therapeutic strategy that targets this vulnerability. We demonstrate that eliminating HER2 and EGFR overcomes resistance to trastuzumab. This was shown in all the experimental models used and with both PEPD^G278D and siRNA. It was also previously shown that HER2 knockdown by siRNA inhibits HER2-positive BC cells that exhibit either primary or secondary resistance to HER2 inhibitors and that a nanoparticle carrying HER2 siRNA inhibits the growth trastuzumab-resistant tumors in mice (Gu et al., 2016). Moreover, it was shown that HER2-positive BC cells do not develop resistance even after long-term (7 months) HER2 siRNA treatment (Gu et al., 2018). Thus, therapeutically eliminating HER2 in HER2-positive BC may not drive emergence of HER2-negative BC cells. Targeted degradation of HER2 by PEPD^G278D obliterates both kinase-dependent and -independent activities of the RTK, in contrast to current HER2 inhibitors that mainly inhibit its kinase activity (Zhang, 2021) and therefore only partially inhibit its oncogenic activity. The kinase-independent functions of HER2, which appear to contribute significantly to its oncogenic activities, may be mediated largely by the signaling complexes it forms with other RTKs. Our present study shows that by inducing HER2 degradation, PEPD^G278D inactivates other RTKs that form complex with HER2. It has also become increasingly clear that both the kinase-dependent and -independent activities of EGFR contribute to its oncogenic activities and cancer progression (Zhang, 2023). EGFR also heterodimerizes with all its family members and many other RTKs (Kennedy et al., 2016). Targeted degradation of EGFR undoubtedly eliminates both kinase-dependent and -independent activities of the RTK, whereas current EGFR inhibitors inhibit almost exclusively its kinase activities (Zhang, 2023). EGFR is overexpressed in about 35% of HER2-positive BCs, and EGFR overexpression is associated with poor prognosis (DiGiovanna et al., 2005; Tsutsui et al., 2003). EGFR is overexpressed in all the HER2-positive BC models used in our present study, and EGFR KD by siRNA is as effective if not more so than HER2 KD in inhibiting trastuzumab-resistant HER2-positive BC cells. Our study shows that targeted degradation of both HER2 and EGFR is necessary for maximizing inhibition of trastuzumab-resistant HER2-positive BC cells and tumors. The profound inhibition of trastuzumab-resistant HER2-positive PDXs by PEPD^G278D is particularly encouraging, because PDX is well known to conserve the heterogeneity of the original tumor. HBCx-73 PDX is most responsive to PEPD^G278D, showing complete remission in 36% tumors after PEPD^G278D plus EP treatment and in 100% tumors after PEPD^G278D plus GD treatment. We did not investigate if PEPD^G278D-induced complete remission of HBCx-73 tumors is durable or if residual tumors after PEPD^G278D treatment are able to regrow. However, we previously showed that 42% of HER2-positive BT-474 BC tumors totally disappeared after PEPD^G278D treatment, which did not return during 40 days of follow-up, while partial remission tumors regrew after PEPD^G278D withdrawal but were still exquisitely sensitive to PEPD^G278D re-treatment (Yang et al., 2015). Also, PEPD^G278D achieved complete remission in all HER2-positive JIMT-1 BC tumors, and no tumor returned during 60 days of follow-up (Yang et al., 2019). Overall, our results suggest that HER2 and EGFR are principal oncogenic drivers in heterogenous HER2-positive BC even after developing resistance to trastuzumab. PEPD^G278D also strongly inhibits metastatic lesions in the lung and brain. The inhibition of brain tumor growth by PEPD^G278D suggests that it can cross blood-brain barrier. It is worth nothing that PEPD^G278D is the first-in-class dual degrader of HER2 and EGFR and its mechanism of action is distinct from the current inhibitors of HER2 and EGFR and those under development (Swain et al., 2023).

Proteolysis targeted chimera (PROTAC) is a well-known platform for developing therapeutics for targeted degradation of oncoproteins. pROTAC links a chemical entity that binds to the target to another entity that binds to an E3 ligase and triggers ubiquitination and degradation of a target when it is introduced into cells. A lapatinib-based PROTAC degrades both EGFR and HER2 in cultured cells and is more effective than kinase inhibition in inhibiting cell proliferation in vitro (Burslem et al., 2018). However, in vivo activity of such agent is not known. PROTAC may induce the degradation of its target in both cancer cells and normal cells, although its impact on normal cells could potentially be minimized by selecting an E3 ligase that is only present or overexpressed in cancer cells. In contrast, PEPD^G278D targets HER2 and EGFR that are overexpressed in cancer cells without attacking the RTKs expressed low in normal cells. This important selectivity apparently results from the requirement for high level of the RTK molecules on cell surface to form complex with PEPD^G278D (Yang et al., 2015, 2019).

One minor weakness of PEPD^G278D as a potential cancer therapeutic agent is that it is degraded in vivo by a plasma proteolysis system (Yang et al., 2016b), and an anticoagulant is needed to block the degradation. We have shown that anticoagulant Ep is highly effective in protecting PEPD^G278D, which is administered to mice at a low dose, albeit daily, and has not shown any adverse effects in our studies. Notably, anticoagulants including EP are frequently used to treat thromboembolic disease in cancer patients (Letai and Kuter, 1999; Mosarla et al., 2019). In the present study, we also demonstrate that a single dose of monoclonal antibody GD which targets FXIIa is sufficient for blocking PEPD^G278D degradation for at least 3 weeks. In view of its infrequent dosing, GD may be more appealing than EP as a blocker of PEPD^G278D degradation for potential clinical translation of the latter. GD has recently been approved by the US Food and Drug Administration as an investigative therapy for hereditary angioedema by targeting FXIIa.

In conclusion, this study shows that trastuzumab-resistant HER2-positive BC is highly vulnerable to targeted degradation of both HER2 and EGFR and that PEPD^G278D is highly effective in targeting this vulnerability. Our data suggest that PEPD^G278D is highly promising for treating patients with HER2-positive BC resistant to current HER2 inhibitors. Collectively, the findings advance understanding of drug resistance in HER2-positive BC and point a new direction for research and development of new agents for treatment of HER2-positive BC.