A Novel HER2-Selective Kinase Inhibitor Is Effective in HER2 Mutant and Amplified Non–Small Cell Lung Cancer

The HER2 is a member of the ErbB receptor tyrosine kinase family along with EGFR, HER3, and HER4. Upon dimerization, activated HER2 triggers intracellular signaling cascades through PI3K/protein kinase B (AKT) and MAPK/extracellular xignal-regulated kinase 1/2 (ERK1/2) that are under tight regulation in physiological context (1). Constitutive hyperactivation of HER2 serves as an oncogenic driver by inducing cell transformation, proliferation, and progressive metastasis (2). Gene amplification and activating mutations of HER2 are detected in multiple cancers (3). Amplification of HER2 is one of the most prevalent oncogenes found in approximately 20% of all breast cancers (4). HER2 overexpression and mutations are associated with poor clinical outcomes of other cancers such as esophagogastric, ovarian, bladder, and colorectal cancers (5–7). With the advent of anti-HER2 antibodies, the FDA approved trastuzumab, pertuzumab, trastuzumab emtansine (T-DM1), trastuzumab deruxtecan (T-DXd), and margetuximab-cmkb for HER2-positive breast cancer. Despite favorable responses initially, a significant number of patients experience relapses during treatment with the aforementioned antibodies (4). Thus, more efficacious and novel interventions are needed in HER2-positive cancers for prolonged disease-free survival.

Approximately 1.4% of patients with lung adenocarcinoma have HER2 amplification as an exclusive oncogenic driver (8). HER2-activating mutations are found in approximately 2.3% of all lung adenocarcinomas, the majority of which localize to the tyrosine kinase domain, particularly within exon 20 (9). In-frame insertions in exon 20 add a few amino acids to the C-terminus of the αC-helix (residues 753–768) that are thought to constitutively activate the kinase domain by promoting the active conformation (10). Among exon 20 insertions, Y772dupYVMA (YVMA) is the most common type comprising about 34% of HER2 mutant lung cancers, followed by G776delinsVC (VC; 5.7%) and, G778dupGSP (GSP; 3.4%; ref. 9).

Although several thousand new cases are confirmed every year worldwide, there are currently no approved targeted therapies for lung cancers harboring HER2 amplification and mutations (11, 12). New HER2 antibody–drug conjugates (ADC) such as T-DM1 and T-DXd demonstrated promising activities in lung cancer showing 44% and 72.7% ORR, respectively (13, 14). However, acquired drug resistance develops to ADCs and new therapeutic options are still needed. Recently reported covalent compounds, including poziotinib and TAK-788, showed modest inhibition against EGFR/HER2 exon 20 mutations (11, 15). However, many tyrosine kinase inhibitors (TKI) against exon 20 variants, display significant toxicity from inhibition of the wild-type (WT) EGFR (16). Thus, there is an unmet need for novel agents for patients with lung cancer that have activity toward a broad spectrum of HER2 mutations while sparing inhibition of WT EGFR kinase.

In the current article, we describe the identification and study of JBJ-08–178–01, a covalent HER2 inhibitor that is selective for HER2 over WT EGFR thus potentially overcoming some of the limitations of the existing compounds. We further describe the biological basis for the efficacy of JBJ-08–178–01, which stems from both inhibition of the kinase and degradation of the receptor.

JBJ-08–178–01 was synthesized as described previously in the patent (WO2019241715). Other inhibitors are listed in the Supplementary Materials. Inhibition assays were performed using the HTRF KinEASE tyrosine kinase assay kit (Cisbio) according to the manufacturer's protocol. Inhibitors were dispensed into 384-well plates using a D300e dispenser (Hewlett-Packard). Assay buffer containing purified HER2 or EGFR (at 5 nmol/L) was dispensed and incubated at room temperature for 30 minutes. Reactions were initiated with 100 μmol/L ATP and allowed to proceed for 30 minutes at room temperature before being quenched using the detection reagent. FRET signal was measured using a PHERAstar microplate reader (BMG LABTECH). Data were processed using GraphPad Prism and fit to a three-parameter dose–response model with a Hill slope constrained to −1. The assay was performed three independent times in triplicate.

HER2 kinase domain at 4–5 mg/mL was incubated with 2 mmol/L TCEP and 0.25 mmol/L JBJ-08–178–01 for 30 minutes at room temperature. Hanging drop crystallization was performed with drops containing 1 μL protein+1 μL well solution over wells containing 400 μL of 0.1 mol/L PIPES pH 6.1, 27% (w/v) PEG 3,350, and 0.2 mol/L diammonium tartrate. Small clusters of needles grew in regions of precipitated protein in 1–2 days at room temperature. Crystal clusters were crushed with a seed bead and 0.1 μL of seed solution was added to hanging drops over wells containing 0.1 mol/L PIPES pH 6.5, 25% (w/v) PEG 3,350, and 0.3–0.4 mol/L diammonium tartrate. Thin single needles grew in 1–3 days at room temperature and were macroseeded into hanging drops over wells containing 0.1 mol/L PIPES pH 6.5, 23% (w/v) PEG 3,350, and 0.2 mol/L diammonium tartrate to obtain single needles with approximate dimensions of 15 × 15 × 120 μm. Crystals were cryoprotected in 0.1 mol/L PIPES pH 6.5, 23% (w/v) PEG 3,350, 0.2 mol/L diammonium tartrate, and 20% ethylene glycol and flash-frozen in liquid nitrogen. Diffraction data were collected at the Advanced Photon Source at Argonne National Laboratory using NE-CAT beamline 24-ID-C at 100 K. Data were integrated, scaled, and merged using xia2, DIALS, and Aimless and the structure determined by molecular replacement in Phenix using 3PP0 from the Protein Data Bank (PDB) as a model (17–21). The best diffracting dataset has excessively high R_merge values due to the presence of severe translational non-crystallographic symmetry. The final structure was obtained via iterative rounds of refinement in Phenix and manual model building in Coot. The HER2 JBJ-08–178–01 structure has been deposited in the PDB with the accession code, 7JXH.

The incorporations of various HER2 mutants and WT EGFR into Ba/F3 cells were previously established and characterized (22, 23). Similarly, HER2 mutations of S310F, L755S, V777L, and V842I were generated in pDNR-dual vector (Clontech) using the Quikchange II XL Site-directed Mutagenesis Kit (Agilent Tech) based on the manufacturer's protocols. Confirmed mutation was shuttled into the lentiviral vector, JP1722. After infection, transformed Ba/F3 cells were selected by puromycin and cultured in RPMI-1640 supplemented with 10% FBS and 1% streptomycin/penicillin.

MCF7, MDA-MB-453, A-431, H1781, H1819, Calu-3, BT-474, SKBR3, and NCI-N87 cell lines were purchased from the ATCC. MCF7, MDA-MB-453, and A-431 were cultured in DMEM; H1781, H1819, Calu-3, and NCI-N87 in RPMI-1640; BT-474 in ATCC Hybri-Care Medium with 1.5 g/L sodium bicarbonate; SKBR3 in the ATCC-formulated McCoy's 5a Medium Modified. To make the complete medium, all media were supplemented with 10% FBS and 1% streptomycin/penicillin. All cells used in this study were periodically tested negative for Mycoplasma using the Mycoplasma Plus PCR Primer Kit (Agilent). All patient-derived cancer cells or tumor spheroids in this study were collected and studied according to the Declaration of Helsinki and DFCI IRBs. DFCI 429 was established from a pleural effusion sample. Next-Generation Sequencing confirmed de novo HER2 amplification. DFCI315 and DFCI359 were characterized to harbor HER2 exon20 V777_G778insGSP and HER2 exon19 755_757LREdelinsRP, respectively (23). Xenograft tumors were processed for short-term ex vivo as previously described (23). The spheroids in a range of 40 to 70 μmol/L were resuspended in type I collagen (Corning) and loaded into the DAX-1 3-D cell culture chip (AIM Biotech). After solidification, the media were applied to the outer channels with indicated drugs.

At least overnight after seeding, Ba/F3 cells and cancer cell lines were treated with 10-dose titration in triplicates. After 3 days, 50% of CellTiter-Glo (v/v; Promega) was added and incubated for 20 minutes on a shaker before luminescent reading (BMG Labtech). The data were analyzed using GraphPad Prism 8.0. Tumor spheroids were dissociated into single cells using TrypLE Express (Invitrogen). The cells were resuspended in complete RPMI/10% Matrigel and plated into 384-well ultra-low attachment microplates (Corning). On the next day, cells were treated as indicated for 6 days. The viability assay was performed using CellTiter-Glo 3D (Promega) according to the manufacturer's protocol. The statistical significance was accessed using ANOVA and Tukey post-test for multiple comparison. For Live/Dead staining, ViaStain AO/PI (acridine orange/propidium iodide) solution (Nexcelom) was applied to tumor spheroids for >15 minutes. The whole image of device containing spheroids were obtained by a Nikon Eclipse 80i fluorescence microscope with automated motorized stage (Proscan) and Zyla 5.5 sCMOS camera (Andor). AO-positive and PI-positive areas were automatically quantified using the NIS-Elements AR software (Nikon). All experiments were repeated more than twice.

HER2 copy number alteration (CNA) in each cell line was obtained from the Cancer Cell Line Encyclopedia (CCLE) database (www.broadinstitute.org/CCLE). The copy number for H1819 was obtained from a previous publication (24). The correlation of CNA and average IC₅₀ value in each cell line was evaluated by GraphPad Prism 8.0, for which two-tailed Spearman rank correlation coefficient was used for statistical analysis.

Cells were plated at 4,000 cells/well into 96-well plates. The following day, cells were treated with fresh media containing CellEvent Caspase-3/7 Green ReadyProbes Reagent (Invitrogen) as the manufacturer's instructions. After adding drugs, real-time confluence and GFP fluorescence detected for activated caspase-3/7 in each well were monitored using the Incucyte S3 live cell analysis system (Essen Bioscience). The fluorescent areas were normalized with confluences at each time point.

Cells were seeded into 96-well plate with 30%–40% confluency. Next day, the cells were treated with indicate drugs and pre-imaged before adding conjugated antibodies. HER2 antibody (sc-08, Santa Cruz Biotechnology) was conjugated with fab-pHrodo fragments (Essen Biosciences) using 1:3 molar ratio for 20 minutes in the dark, +37°C. The conjugated antibody was administrated to the cells, and the imaging was continued immediately. Imaging and analysis were performed using Incucyte S3 imager (Essen Biosciences) more than twice.

Detailed information of each antibody is listed in the Supplementary Materials. The cells were plated a half day before the treatment. After drug treatment for indicated time, the cells were lysed with RIPA buffer (Boston BioProducts) containing protease and phosphatase inhibitors (Sigma-Aldrich). Equivalent amount of proteins was run in 4%–12% Bis-Tris NuPAGE (Invitrogen) and transferred into polyvinylidene difluoride membrane (Millipore). The membrane was blocked with 5% non-fat milk (Bio-Rad Laboratories) and blotted with primary antibodies in the cold room overnight. After washing, the membranes were incubated with secondary antibodies with a dilution of 1:5,000 and developed in ECL solution (PerkinElmer).

For coimmunoprecipitation, the cells were lysed with NP-40 buffer (Boston BioProducts Inc.) with protease/phosphatase inhibitors. From BCA assay, 400 to 800 μg of lysates were incubated with 20 μL of A/G-PlusAgarose beads (Santa Cruz Biotechnology) and anti-HER2 antibodies (1:200) for 4 hours at 4°C. Additional 5 μg of protein lysates were saved for the input blots. The beads were spun down and washed three times with the lysis buffer. The bead-bound proteins were mixed with 20 μL of 2X sample buffer and denatured at 99°C for 7 minutes. The samples ran at 4%–12% Bis-Tris gels and followed the same procedure as Western blotting.

After deparaffinization and rehydration, tissue sections went antigen retrieval using sodium citrate buffer (10 mmol/L, pH 6.0) before staining. The cells plated on tissue-culture–treated slides or tumor spheroids on the 3D-culture chips were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. After blocking with 5% goat serum, all samples were incubated with primary antibodies for overnight at 4. Primary antibodies further bound with secondary antibodies for 1 hour at room temperature. DAPI were treated for nuclear staining before mounting.

TUNEL staining was performed using ApopTag Fluorescein In Situ Apoptosis Detection Kit (Millipore Sigma) according to the manufacturer's protocol. The stitched images of whole tumors were obtained by a Nikon Eclipse 80i fluorescence microscope. Avoiding necrotic lesions, green fluorescence for cell apoptosis was automatically quantified using the NIS-Elements AR software (Nikon). TUNEL-positive counts on each image were normalized by total number of DAPI-positive cells.

All xenograft studies were conducted at Dana-Farber Cancer Institute with the approval of the Institutional Animal Care and Use Committee in an AAALAC accredited vivarium. Female NSG and NCr nude mice were purchased from the Jackson Laboratory and Taconic Biosciences, Inc., respectively. Five million of Calu-3 cells with 50% Matrigel were implanted subcutaneously in the right flank of the NCr nude mice. In vivo passaged fragments of DFCI 315 were implanted in NSG mice. For pharmacodynamics, mice were randomly assigned when tumor volumes reached 200–300 mm³ and started dosing for 3 days. Plasma, tumor, and lung samples were collected 3 hours post final dosing (n = 3 per group). For efficacy, when tumor volumes reached 130 to 230 mm³, mice randomized into treatment groups (8 mice per group). JBJ-08–178–01 and tucatinib were administered orally using 0.5% hydroxypropyl methylcellulose (HPMC; Sigma) with 0.05N HCl in water as vehicle, whereas neratinib was formulated with 0.5% HPMC and 0.4% Tween 80 in water. Control vehicle treated mice received 0.5% HPMC with 0.05N HCl orally. Tumor volumes were determined from caliper measurements by using the formula, Tumor volume = 1/2(length × width²). Tumor volumes and body weights were measured twice weekly.

We developed JBJ-08–178–01 as a HER2-selective inhibitor that binds at the ATP-binding site to form a potency-conferring covalent bond with Cys805 located at the edge of ATP-binding pocket. JBJ-08–178–01 combines the 3-cyanoquinoline-based scaffold of neratinib with the [1,2,4]triazolo[1,5-a]pyridine substituent of tucatinib, an FDA-approved reversible HER2 inhibitor (Fig. 1A). In biochemical inhibition assays, JBJ-08–178–01 potently inhibited WT HER2 with approximately 3-fold selectivity over WT EGFR (Fig. 1A). We assessed the efficiency of JBJ-08–178–01 in Ba/F3 cells harboring various HER2 exon 20 insertions or amplification. JBJ-08–178–01 suppressed the cell growth across all exon 20 mutants or amplification (IC₅₀ < 50 nmol/L; Fig. 1B and Table 1). In addition, JBJ-08–178–01 maintained potency against non-insertion mutations, including S310F, L755S, V777L, V842I, and exon 19 indel (755_757LREdelinsRP) of HER2 (Table 1). JBJ-08–178–01 exhibited a much greater degree of HER2-selectivity in Ba/F3 cell lines than in biochemical assays, owing to weak growth inhibition against Ba/F3 cells with WT EGFR (IC₅₀ = 368.1 ± 39.0 nmol/L; Table 1; Supplementary Fig. S1A). Although afatinib, poziotinib, neratinib, and TAS0728 were all potent against HER2-mutated or amplified Ba/F3 cells, they also potently inhibited WT EGFR Ba/F3 cells, often more potently than they did the HER2 variant cells (Fig. 1B; Supplementary Fig. S1B). Although tucatinib was as mutant selective as JBJ-08–178–01, it was less potent at inhibiting HER2 compared with the covalent TKIs (Fig. 1B).

The IC₅₀ value ratio between WT EGFR and the targeted HER2 or EGFR variant can be used as a selectivity index for inhibitors of ErbB-family kinases (25). Afatinib, poziotinib, neratinib, and lapatinib had low selectivity indexes (≤1.0) due to their high potency against WT EGFR (Fig. 1C and Table 1). On the other hand, JBJ-08–178–01 and tucatinib showed much higher selectivity across all HER2 mutations and amplification. The average selectivity index of JBJ-08–178–01 and tucatinib was 6.83 and 8.91, respectively (Table 1).

To further understand the molecular basis for the observed HER2 selectivity, we determined the crystal structure of JBJ-08–178–01 in complex with the WT HER2 kinase domain (Fig. 1D; Supplementary Table S1). The compound binds as expected in the ATP-site, and we observe clear electron density for the covalent bond with Cys805 (Supplementary Fig. S1C–S1F). Consistent with our design hypothesis, we observe a hydrogen bond between the triazolopyridine ring of the inhibitor and Ser783 (Fig. 1D). This interaction likely accounts for much of the HER2-selectivity of the compound as the equivalent position in EGFR is a cysteine (Cys775), which cannot form a hydrogen bond or polar interaction with the inhibitor due to its different side-chain orientation. The covalent bond with Cys805 is clearly important for the potency of JBJ-08–178–01; a reversible analog of JBJ-08–178–01 was more than 100-fold less potent in the Ba/F3 assay with the YVMA insertion mutant (Supplementary Fig. S1G). Furthermore, JBJ-08–178–01 did not inhibit the growth of HER YVMA Ba/F3 cells containing a concurrent C805S mutation (Supplementary Fig. S1G).

Next, we assessed inhibition of HER2 and its downstream signaling by JBJ-08–178–01. In Ba/F3 cells harboring common HER2 exon 20 insertions (YVMA, GSP, and VC) and amplification, JBJ-08–178–01 reduced phosphorylation of HER2, AKT, and ERK1/2, in a dose-dependent manner (Fig. 1E). On the other hand, JBJ-08–178–01 inhibited phosphorylation of WT EGFR and ERK1/2 comparably with tucatinib but it was not as potent as neratinib (Fig. 1E). Selectivity of JBJ-08–178–01 was further investigated in A431 epidermoid cells that harbor amplification of WT EGFR. Afatinib and poziotinib potently inhibited the growth of A431 with IC₅₀ values of <100 nmol/L (Table 2). JBJ-08–178–01 was markedly less potent, similar to the mutant selective EGFR inhibitors, osimertinib and TAS6417 (Fig. 1F and Table 2). In addition, Western blotting confirmed JBJ-08–178–01 inhibited phospho-EGFR and phospho-ERK1/2 to a lesser extent than poziotinib and neratinib (Fig. 1F). Collectively, covalent binding to Cys805 and hydrogen bonding to Ser783 enables JBJ-08–178–01 to potently inhibit a diverse set of HER2 mutants with a wide selectivity window over WT EGFR.

The efficiency of JBJ-08–178–01 was tested in H1781 (HER2 exon 20 insertion VC), H1819, and Calu-3 (both with HER2 amplification) NSCLC (non–small cell lung cancer) cell lines. In all cell lines, JBJ-08–178–01 displayed robust growth inhibition with IC₅₀ values of ≤200 nmol/L (Fig. 2A and Table 2). Live-cell imaging analysis revealed that JBJ-08–178–01–treated cells induced rapid and greater degree of apoptosis compared with cells treated with poziotinib, neratinib, or tucatinib (Fig. 2B). Withdrawal of JBJ-08–178–01 halted apoptosis, indicating continuous treatment of JBJ-08–178–01 is necessary for its continuous efficacy. However, after stopping treatment with JBJ-08–178–01, cell regrowth was much slower than in those treated with tucatinib, possibly due to the covalent nature of JBJ-08–178–01 (Fig. 2C). JBJ-08–178–01 treatment led to decreases in phosphorylation of HER2, AKT, and ERK1/2 in a dose-dependent manner across all NSCLC cell lines (Fig. 2D). We further examined the ability of JBJ-08–178–01 to inhibit growth of other cancer cell types. JBJ-08–178–01 was tested in breast (BT474, SKBR3) and gastric (NCI-N87) cancer cell lines that harbor HER2 amplification. All three lines favorably responded to JBJ-08–178–01 (Table 2). A similar degree of target engagement and downstream inhibition was also observed (Supplementary Fig. S2A). Because all cells bear different CNA of HER2, we evaluated whether HER2 CNA is associated with therapeutic response to JBJ-08–178–01. Our analysis revealed that higher copy numbers of HER2 correlated with increased sensitivity to JBJ-08–178–01 (r = −0.903; Fig. 2E). Despite a low CNA (2.3 relative to ploidy+1), H1781 still responded better to JBJ-08–178–01 compared with the cell lines with adjacent scores such as A431 and SNU-16. This is likely because H1781 harbors a HER2 exon 20 insertion that is known to be more tumorigenic than amplification (26).

The effectiveness of JBJ-08–178–01 was further tested in patient-derived lung cancer lines. DFCI 429 with de novo HER2 amplification, showed favorable response to JBJ-08–178–01 (Fig. 3A and Table 2). Treatment with JBJ-08–178–01 resulted in better inhibition of p-HER2 and pERK1/2 than same dose of afatinib and tucatinib, and more proteolytic cleavage of PARP, which is an early indicator of cell apoptosis (Fig. 3B). DFCI 315, a PDX-derived lung cancer organoid harboring a HER2 exon 20 insertion GSP, also responded to JBJ-08–178–01 ex vivo (Fig. 3A). After 2-day treatment of DFCI 315 with JBJ-08–178–01, immunofluorescence assay displayed reduced p-HER2 signal and size of organoids compared with vehicle control (Fig. 3C; Supplementary Fig. S2B). Cytotoxic effect of JBJ-08–178–01 was visually quantified by staining with AO and PI. DFCI 315 did not show notable changes in the number of AO-positive live cells following treatment by any of the HER2 inhibitors (data not shown). However, JBJ-08–178–01 significantly increased PI-positive dead cells as dose-matched neratinib (P < 0.05; Fig. 3D). In a second PDX-derived tumoroid model, DFCI 359 (HER2 exon 19 LREdelinsRP), JBJ-08–178–01 induced significant decrease of live cells and increase of dead cells similar to other TKIs (Fig. 3E).

During in vitro assays, we consistently observed that JBJ-08–178–01 not only repressed HER2 kinase activity but also reduced total HER2 levels (Figs. 2D and 3B). Reduction of HER2 was unlikely the result of transcriptional regulation, as JBJ-08–178–01 treatment increased HER2 transcription in Calu-3 and H1781 cells (Supplementary Fig. S3A). We thus hypothesized that this phenomenon occurred at the post-translational level. In the presence of a protein synthesis inhibitor, cycloheximide, JBJ-08–178–01 treatment decreased protein stability of HER2 compared with vehicle in Calu-3 cells (Fig. 4A). With a darker exposure, we observed JBJ-08–178–01 also downregulated p95 expression, a truncated variant of HER2. JBJ-08–178–01 had little effect on the decrease of EGFR and HER3 levels. Similar trends were found in H1781 and H1819 cells (Supplementary Fig. S3B). Next, we sought to determine whether JBJ-08–178–01 could promote internalization of HER2 from the cell surface using pH-sensitive fluorescence dye conjugated with HER2 antibodies (27). Upon JBJ-08–178–01 treatment, we observed there was a dramatic increase of fluorescence signal in the cytoplasm, which is indicative of HER2 endocytosis (Fig. 4B). It occurred in a dose-dependent manner and was more dramatic than neratinib, which was previously shown to induce HER2 endocytosis and subsequent degradation (28). Many studies have shown that HER2 levels are steady on the cell membrane, whereas other ErbB members are easily undergo endocytosis and recycling (29). Constitutive interaction of HER2 with HSP90/cdc37 is critical for the receptor stability (30). When HER2 is released from HSP90, HSP70 and an E3 ubiquitin ligase facilitate endocytosis and degradation of the receptor. Our coimmunoprecipitation results showed that JBJ-08–178–01 rapidly dissociated HSP90 from HER2 in NSCLC and breast cancer cells (Fig. 4C; Supplementary Fig. S3C). Conversely, JBJ-08–178–01 facilitated binding of HSP70 and ubiquitination of HER2.

Next, we evaluated the mechanism by which JBJ-08–178–01 induces HER2 degradation. In a coimmunofluorescence study, we observed HER2 colocalized with S20 alpha, a subunit of proteasome, after JBJ-08–178–01 treatment (Fig. 4D; Supplementary Fig. S3D). Another HER2-degrading TKI, neratinib did not induce colocalization of HER2 and S20 (Fig. 4D). Instead, as previously reported (28, 31), neratinib-bound HER2 colocalized with lysosome marker, LAMP-1 (Supplementary Fig. S4). Only minor level of HER2 and LAMP-1 co-localization was observed following JBJ-08–178–01 treatment (Supplementary Figs. S3D and S4). Also, JBJ-08–178–01 treatment led to a minor amount of HER2 to localize with LC-3, a marker for autophagy (Supplementary Fig. S5). To determine whether JBJ-08–178–01 promotes HER2 degradation mainly through the proteasome, the cells were pre-treated with two proteasome inhibitors, MG-132 and bortezomib, before JBJ-08–178–01. Both proteasome inhibitor(s) attenuated HER2 degradation by JBJ-08–178–01 in NSCLC (Fig. 4E; Supplementary Fig. S6A). Finally, we investigated whether JBJ-08–178–01 could synergize with a selective HSP90 inhibitor, tanespimycin, to enhance HER2 degradation. We found that tanespimycin alone increased HER2 degradation in a dose-dependent manner (Fig. 4F), demonstrating the importance of HSP90 in HER2 stability. When combined with JBJ-08–178–01, HER2 degradation occurred to a greater extent than either monotherapy. The combined effect between JBJ-08–178–01 and tanespimycin was further evaluated in cell viability assays. Compared with single treatment of JBJ-08–178–01, combination with tanespimycin resulted in greater levels of growth inhibition (Fig. 4G). Consistently, we found that Bim expression, a pro-apoptotic protein, was increased following combined treatment in H1781 cells (Fig. 4F). A reciprocal assay using JG-98, an allosteric HSP70 activator, produced similar results (Supplementary Fig. S6B and S6C). Overall, our results demonstrate that JBJ-08–178–01 promotes endocytosis of HER2 from the cell surface and ubiquitination, leading to its degradation, which might also contribute to growth inhibition in HER2 mutant cells (Fig. 4H).

To test the efficiency of JBJ-08–178–01 in vivo, two xenograft models, a patient-derived DFCI 315 and Calu-3, were implanted into mice. JBJ-08–178–01 led to a dose-dependent inhibition of HER2 and downstream signaling (Fig. 5A and B). High doses of JBJ-08–178–01 (100 mg/kg/d in DFCI 315; 50 mg/kg twice per day (BID) and 100 mg/kg/day in Calu-3) achieved comparable levels of inhibition to neratinib (40 mg/kg/d) and tucatinib (100 mg/kg BID). Consistent with the in vitro findings, JBJ-08–178–01 showed strong HER2-degrading effects on both xenograft models (Fig. 5A–,C). Next, we checked tissue apoptosis using TUNEL assay in Calu-3 tumors. As shown in Fig. 5D, JBJ-08–178–01 at 50 mg/kg BID significantly increased cell apoptosis compared with vehicle.

In the efficacy study using DFCI 315, high doses of JBJ-08–178–01 at 50 and 75 mg/kg BID resulted in significant and more rapid inhibition of tumor growth compared with the neratinib and tucatinib (Fig. 5E). Similar results were observed in Calu-3 tumors where JBJ-08–178–01 at 50 mg/kg BID yielded comparable tumor regression with neratinib (Fig. 5F). JBJ-08–178–01 at 100 mg/kg QD yielded better target engagement than 50 mg/kg BID (Fig. 5B) but failed to induce more apoptosis and better efficacy (Fig. 5D and F). The non-covalent HER2 inhibitor, tucatinib, was less efficacious than other covalent inhibitors in both models (Fig. 5E and F).

The activity of JBJ-08–178–01 represents a promising therapeutic to target a wide range of HER2 exon 20 mutations in lung cancer. It has been challenging to treat HER2 exon 20 insertions using available ERBB TKIs (32–34). The phase II trial of dacomitinib showed 12% overall response rate in patients with lung cancer harboring HER2 exon 20 mutations (34). Only one out of 26 patients with HER2 mutant NSCLC had an objective response to neratinib in a clinical trial (35). It is intriguing that all IC₅₀ values of JBJ-08–178–01 were lower than 50 nmol/L in Ba/F3 expressing different HER2 exon 20 variants (Fig. 1B and Table 1). JBJ-08–178–01 has activity in NSCLC cells harboring clinically relevant exon 20 mutations both in vitro and in vivo (Figs. 2–4). Furthermore, JBJ-08–178–01 is also effective against the S310F, L755S, V777L, and V842I, suggesting it may be broadly efficacious across other HER2 mutant cancers (Figs. 1B and 3C).

Although potent pan-ERBB TKIs are already available on the market or undergoing evaluation in clinical trials, many agents are not as effective clinically as compared with their efficacy in preclinical studies. One possibility for this disconnect is due to tolerability, which may limit the ability to achieve therapeutically efficacious doses in humans (11, 36). Cutaneous and gastrointestinal toxicities are observed in patients treated with HER2 TKIs that are also potent inhibitors of WT EGFR, including afatinib, poziotinib, and neratinib. Thus, developing novel inhibitors, ones that are less effective on EGFR than HER2, is one approach to improve the therapeutic window of this class of agents. In our studies, JBJ-08–178–01 has an average of 6.8-fold lower IC₅₀ value against HER2 mutants than WT EGFR (Fig. 1C and Table 1). On the other hand, afatinib, poziotinib, and neratinib had low selectivity scores (<0.7; Table 1), consistent with their known toxicity profiles. Whether the preclinical selectivity of JBJ-08–178–01 will translate into an improvement in clinical tolerability will remain to be determined and will need to be studied in future clinical trials. We observed some weight loss in NSG mice but not in NCr nude strain (Supplementary Fig. S7). This maybe a strain-specific effect, or suboptimal bioavailability in mice (Table 3), and whether it will pose challenges to treating humans awaits evaluation in future clinical studies.

Our study provides structural insights to explain how JBJ-08–178–01 drives high selectivity and potency on HER2. The crystal structure of JBJ-08–178–01 covalently bound to the HER2 kinase domain suggests the selectivity for HER2 stems from a hydrogen bond between the triazolopyridine ring of JBJ-08–178–01 and Ser783. Although no crystal structure of tucatinib with HER2 is available, structural modeling indicates that the equivalent interaction is formed with its triazolopyridine substituent. The covalent bond with Cys805 is essential for JBJ-08–178–01 potency (Fig. 1D), and consistent with this the C805S mutation conferred resistance to JBJ-08–178–01 (Supplementary Fig. S1C).

We noted a discrepancy between the biochemical and cellular assays of HER2 selectivity. Cellular selectivity between HER2 and EGFR was greater than suggested by enzymatic IC₅₀ values (6.8-fold vs. 3-fold; Fig. 1A and Table 1). We speculate that the greater selectivity in cells is the result of receptor and kinase dimerization, which does not readily occur in assays with low concentrations of purified kinase domain. From our crystal structure, JBJ-08–178–01 is not expected to bind well to the active conformation that is driven by dimerization, which is likely why both EGFR and HER2 were similarly sensitive to JBJ-08–178–01 in biochemical assays that employ the monomeric kinase domain.

A unique feature of JBJ-08–178–01 was that it both inhibited kinase activity and decreased levels of HER2 protein in vitro and in vivo (Fig. 4H). JBJ-08–178–01 also decreased one of the variants of HER2, p95 that lacks the extracellular domain and is associated with drug resistance and poor clinical outcomes (37). This dual effect may be responsible for its enhanced efficacy in HER2 mutant cancers. JBJ-08–178–01 disrupted HER2-HSP90 binding, resulting in HER2 endocytosis and subsequent degradation in HER2-amplified and -mutant cells (Fig. 4C; Supplementary Fig. S3B).

Dual inhibitory and degradative effects have been reported for other covalent inhibitors. Afatinib reduces total levels of EGFR and HER2 in NSCLC cell lines (38). Neratinib also decreases EGFR and HER2 levels on the cell surface mainly through lysosomal degradation (28, 31). The degradative effect of JBJ-08–178–01 seems distinct from these previous findings. First, JBJ-08–178–01 has minimal effect on other ERBB receptors (Fig. 4A and Supplementary Fig. S3B). Unlike neratinib-induced degradation, endocytosed HER2 by JBJ-08–178–01 mainly undergoes proteasomal rather than lysosomal degradation (Fig. 4D–E; Supplementary Figs. S3D and S4). Questions remain as to whether HER2 degradation contributes to the efficacy of JBJ-08–178–01 and other inhibitors, beyond their ability to inhibit the catalytic activities of HER2 and/or EGFR. A recent publication shows that neratinib enhances T-DM1-driven HER2 instability on the cell surface, highlighting the translational value of TKI-induced HER2 internalization and degradation (39). Intriguingly, this combination resulted in more durable responses in a patient with lung cancer who was refractory after T-DM1 monotherapy. We also found that JBJ-08–178–01 exerted a greater antitumorigenic effect in NSCLC cells when coadministered with tanespimycin or JG-98 in vitro highlighting the contribution of HER2 instability to cause more cell death in HER2-addictied cancers (Fig. 4F and G; Supplementary Fig. S6C).

In summary, JBJ-08–178–01 is a novel HER2-selective inhibitor effective in preclinical models of HER2 mutant cancers. JBJ-08–178–01 has a broad therapeutic window and potently inhibits HER2 amplification and other activating mutations in multiple cancers while sparing WT EGFR. Understanding whether these features translate to a more effective and better tolerated therapeutic for HER2-driven cancer must await clinical development and clinical studies in patients with cancer.

J. Son reports a patent for WO 2019/241715 A1 issued. J. Jang reports grants from HER2 Research, LLC, during the conduct of the study, as well as reports a patent for WO2019241715 issued. T.S. Beyett reports other support from NCI/NIH during the conduct of the study. N.P. Kwiatkowski reports grants from Arbella Therapeutics during the conduct of the study and reports employment of Kymera Therapeutics, Inc. P.C. Gokhale reports grants from Marengo Therapeutics, Epizyme Inc., Daiichi Sankyo, Foghorn Therapeutics, 28-7 Therapeutics, and Moderna outside the submitted work. M.J. Eck reports grants from Arbella Therapeutics during the conduct of the study, grants, personal fees, and non-financial support from Novartis, and grants from Takeda and Springworks Therapeutics, and personal fees from H3 Biomedicine outside the submitted work. N.S. Gray reports grants from Arbella during the conduct of the study, personal fees from C4 therapeutics, personal fees from Syros, personal fees from Jengu, personal fees from Inception, personal fees from EoCys, personal fees from Allorion, personal fees from GSK, personal fees from Larkspur, and personal fees from B2S outside the submitted work; in addition, N.S. Gray reports a patent for patents covering compounds pending. P.A. Jänne reports grants and personal fees from AstraZeneca, Boehringer Ingelheim, and personal fees from Pfizer, Roche/Genentech, Chugai, and grants and personal fees from Eli Lilly, personal fees from SFJ Pharmaceuticals, Voronoi, grants and personal fees from Daiichi Sankyo, personal fees from Biocartis, Novartis, Sanofi Oncology, grants and personal fees from Takeda Oncology, and personal fees from Mirati Therapeutics, Transcenta, Silicon Therapeutics, Syndax, Nuvalent, Bayer, Esai, Allorion Therapeutics, Accutar Biotech, AbbVie, grants from Revolution Medicines and PUMA outside the submitted work; and reports a patent for EGFR mutations issued and licensed to LabCorp. No disclosures were reported by the other authors.

J. Son: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. J. Jang: Conceptualization, resources, data curation, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. T.S. Beyett: Conceptualization, resources, data curation, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. Y. Eum: Validation, investigation, visualization, writing–original draft. H.M. Haikala: Formal analysis, validation, investigation, visualization, methodology, writing–original draft. A. Verano: Formal analysis, validation, investigation. M. Lin: Formal analysis, validation, investigation. J.M. Hatcher: Formal analysis, validation, investigation, visualization. N.P. Kwiatkowski: Resources, funding acquisition, validation, project administration. P.Ö. Eser: Conceptualization, software, formal analysis. M.J. Poitras: Formal analysis, investigation. S. Wang: Formal analysis, validation, investigation. M. Xu: Investigation. P.C. Gokhale: Data curation, supervision, validation, writing–review and editing. M.D. Cameron: Formal analysis, investigation. M.J. Eck: Conceptualization, resources, supervision, funding acquisition, validation, investigation, project administration, writing–review and editing. N.S. Gray: Conceptualization, resources, data curation, supervision, funding acquisition, project administration, writing–review and editing. P.A. Jänne: Conceptualization, resources, data curation, supervision, funding acquisition, project administration, writing–review and editing.

The study was supported by the American Cancer Society (CRP-17-111-01-CDD; to P.A. Jänne), HER2 research LLC (to P.A. Jänne and N.S. Gray), Schlauch Family Fund for Lung Cancer Research (to P.A. Jänne), James B. Gillen Thoracic Oncology Research Fund (to P.A. Jänne), and Miracles Foundation and the Robert A. and Renée E. Belfer Family Foundation. T.S. Beyett is supported by a Ruth L. Kirschstein National Research Service Award (1F32CA247198-01). This work was based upon research conducted at the Northeastern Collaborative Access Team Beamlines (P30 GM124165, P41 GM103403) using resources of the Advanced Photon Source at the Argonne National Laboratory (DE-AC02-06CH11357). The study was supported in part by NIH grants RO1 CA116020 and R35 CA242461 (to M.J. Eck).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.