Computational and Functional Analyses of HER2 Mutations Reveal Allosteric Activation Mechanisms and Altered Pharmacologic Effects

Abstract

Amplification of HER2 can drive the proliferation of cancer cells, and several inhibitors of HER2 have been successfully developed. Recent advances in next-generation sequencing now reveal that HER2 is subject to mutation, with over 2,000 unique variants observed in human cancers. Several examples of oncogenic HER2 mutations have been described, and these primarily occur at allosteric sites outside the ATP-binding site. To identify the full spectrum of oncogenic HER2 driver mutations aside from a few well-studied mutations, we developed mutation-allostery-pharmacology (MAP), an in silico prediction algorithm based on machine learning. By applying this computational approach to 820 single-nucleotide variants, a list of 222 known and potential driver mutations was produced. Of these 222 mutations, 111 were screened by Ba/F3-retrovirus proliferation assays; 37 HER2 mutations were experimentally determined to be driver mutations, comprising 15 previously characterized and 22 newly identified oncogenic mutations. These oncogenic mutations mostly affected allosteric sites in the extracellular domain (ECD), transmembrane domain, and kinase domain of HER2, with only a single mutation in the HER2 orthosteric ATP site. Covalent homodimerization was established as a common mechanism of activation among HER2 ECD allosteric mutations, including the most prevalent HER2 mutation, S310F. Furthermore, HER2 allosteric mutants with enhanced covalent homodimerization were characterized by altered pharmacology that reduces the activity of existing anti-HER2 agents, including the mAb trastuzumab and the tyrosine kinase inhibitor lapatinib. Overall, the MAP-scoring and functional validation analyses provided new insights into the oncogenic activity and therapeutic targeting of HER2 mutations in cancer.

Significance: This study identified new oncogenic HER2 allosteric mutations, including ECD mutations that share covalent dimerization as a mechanism of oncogenicity, suggesting the need for novel inhibitors to treat HER2-mutant cancers.

Figure 1.

HER2 mutation prevalence and protein domain architecture. A, Prevalence of somatic HER2 mutations observed throughout the protein domains. Foundation Medicine, Inc. data used in this study contain 2,616 unique SNV HER2 mutations. HER2 domains: The protein architecture consists of L domains 1 and 2 (L1, L2), cysteine-rich domains 1 and 2 (CR1, CR2), TMD, JMD, KD, and CTD. The L domains and the CR domains collectively form the ECD. Prediction: Oncogenicity of cancer-associated mutations are predicted as either oncogenic mutations (dark blue) or neutral mutations (gray). Validation: Select HER2 SNVs are biologically validated as driver mutations (magenta) or neutral (green) mutations, based on Ba/F3 cell proliferation assay results. Refined collection of verified HER2 driver mutations with a varying degree of prevalence is shown on the bottom plot. The most prevalent mutation, S310F, is located in the ECD (CR1), whereas an intracellular mutational hotspot, containing L755S, Exon20ins, V777L, and V842L, is located in the KD. Other notable mutations are labeled. B,HER2 SNVs and oncogenic mutations by prevalence range. Quantification analysis of somatic HER2 mutations in three prevalence groups.

Figure 2.

Distribution of HER2 mutations in solid tumors. Compositions of representative HER2 mutations, exon 20 insertions, S310F/Y, R678Q, and the KD hotspot mutations (L755/L777/V842) differ among all solid tumors, breast cancer, lung cancer, bladder cancer, and colorectal cancer.

Figure 3.

Protein structure analysis of HER2 oncogenic mutations. A,HER2 mutations mapped on a full-length dimer structure model as driver mutations (magenta) or neutral (green) mutations. Individual protein domains are indicated by different colors. B, Radar charts of HER2 mutations on key proteomic features. In general, driver mutations are associated with well-conserved residues that are likely involved in conformational changes. The ECD driver mutations are in close proximity to the disulfide bonds, whereas the KD driver mutations are located near the ligand binding and protein interaction sites. C–E, Close-up views of prevalent HER2 mutations: S310F in CR1 (C), R678Q in JMD (D), and V842I in KD (E), and other nearby driver mutations. F, Western blots of Ba/F3-HER2 oncogenic mutants inducing covalent homodimerization in the absence and presence of reducing agents (indicated by − and +). Phospho-HER2 blots and total HER2 blots are shown.

Figure 4.

HER2 variants drive oncogenicity through reinforcement of dimerization. A, Altered domain structures of HER2 variants, S310F, p95-M611, and Δ16. The presumed free cysteines associated with these mutations are indicated by yellow lollipops. Trastuzumab binding site (HER2 WT residues 579–625) is altered in HER2 p95-M611and HER2 Δ16. B, The S310F missense mutation disrupts the β-hairpin motif (orange) and the adjacent disulfide bond formed by C299 and C311. Structural comparison of HER2 WT (S310) and HER2 S310F mutant in oxidizing and in reducing conditions. Structural disorder induced by S310F likely results in increased distance between C311 and C299 to interfere with the disulfide bond formation. C, U87MG cells expressing HER2 S310F result in the marked formation of activated covalent dimers. Reduced dimer formation is observed with the introduction of Cys-to-Ser substitutions (C299S and C311S). Residual covalent dimer formation suggests the involvement of other cysteines in intermolecular disulfide bonds. D, U87MG cells individually expressing HER2 variants (S310F, Δ16, and p95-M611) form activated homodimers under nonreducing conditions. E, Trastuzumab effectively reduces the active dimer population of HER2 WT expressed in U87MG cells, but does not reduce the amount of covalently dimerized HER2 variants (S310F, Δ16, and p95-M611) expressed in U87MG cells. Cells were treated with control (C; PBS) or decreasing concentrations of trastuzumab (1.0, 0.5, 0.25, 0.125 µg/mL) for 24 hours. F, Proliferation of BT474 cells expressing HER2 WT is suppressed with 1 µg/mL trastuzumab, whereas Ba/F3 cells expressing HER2 variants (p95-M611 or Δ16) required higher trastuzumab concentration for growth reduction.

Figure 5.

TKIs reveal altered pharmacology of HER2 variants S310F, p95-M611, and Δ16. A, Concentration-dependent inhibition of HER2 WT (cell line, BT474) and variants (S310F, p95-M611, and Δ16; cell line Ba/F3) by lapatinib. B and C, Lapatinib-induced dimerization of HER2 p95-M611 variant.

Figure 6.

Common mechanism of oncogenicity and unique pharmacology of HER2 allosteric mutations. HER2 WT monomers in an extended state can either homodimerize or heterodimerize with other ERBB family members in a dynamic manner to facilitate the regulation of normal ERBB signaling activities (yellow). HER2 dimer interfaces mainly involve the cysteine-rich domains, CR1 and CR2, of the ECD, the TMD, the JMD, and the intracellular KD. The asymmetrical dimerization of KD is required for HER2 activation. WT HER2 is targeted by the HER2-selective antibody trastuzumab, which readily binds to the ECD to reduce the cell surface expression of HER2, and the tyrosine kinase inhibitor lapatinib, which shows high efficacy toward WT HER2. In comparison, allosteric HER2 oncogenic mutants display one of several features, such as free cysteine and increased hydrophobicity (the S310F position and the dimer arm region are shown in purple), that promote covalent or stable HER2 homodimerization and constitutive activation of HER2 signaling (red and yellow circles). Intracellular KD mutations that increase its asymmetrical dimerization contribute to HER2 activation through inside-out signaling. Conformational changes induced by allosteric mutations have resulted in reduced pharmacologic effects due to altered antibody binding sites and the TKI-induced dimerization of HER2 mutants.