INVITED REVIEWS
Multiple Mutations within Individual Oncogenes: Examples and Clinical Implications
2023 Volume 72 Issue 3 Pages 88-92
Details
2023 Volume 72 Issue 3 Pages 88-92
Gain-of-function mutations had been believed to function as a single mutation in oncogenes, although some secondary mutations, such as EGFR T790M mutations, are frequently acquired in patients that are resistant to tyrosine kinase inhibitor treatment. Recently, we and other investigators have reported that multiple mutations (MMs) frequently occur in the same oncogene before any therapy. In a recent pan-cancer study, we identified 14 pan-cancer oncogenes (such as PIK3CA and EGFR) and 6 cancer type-specific oncogenes that are significantly affected by MMs. Of these, 9% of cases with at least one mutation have MMs that are cis-presenting on the same allele. Interestingly, MMs show distinct mutational patterns in various oncogenes relative to single mutations in terms of mutation type, position, and amino acid substitution. Specifically, functionally weak, uncommon mutations are overrepresented in MMs, which enhance oncogenic activity in combination. Here, we present an overview of the current understanding of oncogenic MMs in human cancers and provide insights into their underlying mechanisms and clinical implications.
In human cancers, oncogenes are activated by multiple genetic mechanisms, such as gain-of-function mutations, copy number amplification, and gene fusion. Among these, a gain-of-function mutation is the most common, which results in an altered gene product with intensified activity or a new biological function (neomorphic mutation).1 Tumor suppressor genes (TSGs) are commonly affected by multiple loss-of-function mutations, typically in a biallelic manner.2,3 However, gain-of-function mutations had been believed to function as a single mutation in individual oncogenes, although certain secondary mutations, such as EGFR T790M mutations, are frequently acquired in patients resistant to tyrosine kinase inhibitor treatment.4,5 Recently, we and other investigators have reported that multiple mutations (MMs) frequently occur in the same oncogene prior to therapy.6,7,8 Specifically, we identified 14 oncogenes that were significantly affected by MMs in pan-cancer analysis, including PIK3CA, EGFR, MTOR, and ERBB2.8 In these oncogenes, 9% of cases with at least one mutation have MMs, which are cis-presenting on the same allele.6,8 This review summarizes recent advances in oncogenic MMs and their biological relevance and clinical implications, primarily focusing on those in PIK3CA, EGFR, and NOTCH1.
Phosphatidylinositol 3-kinase (PI3K) activation initiates a signal transduction cascade that promotes cancer cell growth, survival, and metabolism via direct phosphorylation and activation of AKT, a serine-threonine kinase.9 PI3K is composed of a 110-kDa catalytic subunit (p110) and an 85-kDa regulatory subunit (p85). The catalytic subunit consists of an adaptor binding domain, a RAS-binding domain, a protein-kinase-C-homology-2 (C2) domain, a helical domain, and a kinase domain. Somatic mutations in PIK3CA, encoding class IA p110 (p110α), occur in up to 50% of common epithelial cancers, such as breast, head and neck, gastrointestinal, and endometrial cancers.10 Approximately half of the PIK3CA mutations are present in one of the three major hotspots; E542 and E545 in the kinase domain and H1047 in the helical domain, whereas most of the remaining mutations are located in minor hotspots, such as R88, K111, G118, N345, C420, E453, Q546, E726, and M1043.11 In the three major hotspots, specific amino acid substitutions (namely E542K, E545K, and H1047R) are preferentially selected. These mutant p110α subunits enhance in vitro lipid kinase activity, maintain PI3K-AKT signaling under conditions of growth factor deprivation, and transform cells.9 The two classes of major hotspot PIK3CA mutations augment constitutive PI3K-AKT signaling through different mechanisms. In wild-type PI3K holoenzyme, p85 inhibits p110α through intermolecular interactions. Molecular modeling and X-ray crystal structure analyses showed that the helical domain mutants E542K and E545K disrupted this inhibitory intermolecular interaction between p85 and p110.12,13 In contrast, the kinase domain mutant H1047R induced a conformational change in the lipid-binding surface, resulting in a higher affinity for lipids.14,15
Among oncogenes, MMs are most frequently observed in PIK3CA in multiple cancer types,6,8 with more missense mutations than in-frame indels. The frequency of PIK3CA MMs varies depending on the mutational positions (Fig. 1a). Specifically, MMs occur less frequently in the three major hotspots (E542, E545, and H1047) than in the minor hotspots. In particular, E453 and E726 are frequently affected by MMs, where more than half of the mutated cases harbor another PIK3CA mutation. In addition, in MM-harboring cases, less frequent (minor) amino acid substitutions are more prevalent than common (major) amino acid substitutions, such as E542K, E545K, and H1047R, even in major hotspots. Among PIK3CA MMs, combinations of major and minor hotspot mutations, such as E542–E726 and E726–H1047 mutations, are overrepresented, followed by a combination of minor and minor hotspot mutations. Analysis of allele frequencies of these MMs suggests that most mutational combinations occur in the same clones. In cases harboring a combination of major and minor hotspot mutations, the allele frequency of major hotspot mutations is sometimes higher than that of minor hotspot mutations, although allele frequency is comparable between mutations in most cases. These observations suggest that the major hotspot mutation is an early genetic event that can induce clonal expansion to a certain extent.
. Distribution of mutations in oncogenes.
Distribution of mutations in samples with single and multiple mutations in recurrently mutated cancer types: (a) PIK3CA mutations; (b) EGFR mutations. (c) Distribution of NOTCH1 mutations in samples with single and multiple mutations in T-ALL.
Estimating the functional activity of PIK3CA mutations from in vitro and in vivo assays using mutant-transduced cell lines suggested that in-frame indels had stronger oncogenic capacity than missense mutations.8 In missense mutations, functional activity was inversely associated with the proportion of MMs in mutated cases. These findings suggest that functionally weak mutations are selected as MMs in PIK3CA. In addition, cell lines transduced with major–minor double mutant of PIK3CA, such as E453Q–E545K and E545K–E726K, exhibited increased cell proliferation in vitro and enhanced tumor growth in vivo relative to single mutant-transduced cell lines.6,8 Furthermore, analysis of clustered regularly interspaced short palindromic repeat (CRISPR) loss-of-function screening data using the Cancer Cell Line Encyclopedia (CCLE) cell lines showed that cell lines with PIK3CA MMs were highly dependent on PIK3CA and AKT genes. This implies that PI3K-AKT signaling activation is essential in cancers with PIK3CA MMs.8
Notably, analysis of drug sensitivity data using CCLE cell lines revealed that various PI3K inhibitors are more effective against cell lines with PIK3CA MMs than against other cell lines, suggesting that MMs can serve as a biomarker for predicting the efficacy of molecularly targeted therapy.8 In addition, cell lines transduced with PIK3CA double mutants were synergistically more sensitive to PI3Kα inhibitors (alpelisib and GDC-0077) than those transduced with single mutants.6 More importantly, in a phase 3 clinical trial investigating the efficacy of the PI3Kα/γ/δ inhibitor taselisib in combination with the estrogen receptor antagonist fulvestrant versus placebo and fulvestrant in metastatic estrogen receptor-positive breast cancer patients, patients harboring PIK3CA MMs detected by circulating tumor DNA analysis achieved a significantly higher response rate than those in the placebo arm.6 In contrast, single PIK3CA mutant patients in the taselisib arm had a slightly but not significantly better response rate than those in the placebo arm. These findings suggest that patients with multiple mutant tumors may achieve a higher clinical benefit from PI3Kα inhibition than those with single mutant tumors.
Increased PI3K-AKT signaling functions as the molecular mechanism of enhanced oncogenic capacity of PIK3CA MMs. Specifically, major–minor double mutants of PIK3CA strongly enhance the phosphorylation and activation of AKT and its downstream mTOR, p70S6K, PRAS40, and GSK-3β.6,8 Analysis of reverse-phase protein array data from CCLE cell lines and The Cancer Genome Atlas (TCGA) patient samples demonstrated elevated AKT phosphorylation in PIK3CA MM-harboring samples.8 In addition, in vitro functional assays revealed that these double mutants abrogate the p85–p110 interaction, increasing the lipid binding and kinase activity.6 Molecular dynamics simulation of the synergistic mutants R88Q–H1047R, a mutational combination with the strongest oncogenic activity in vitro, suggested a coordinated structural change underlying the enhanced downstream pathway activation by PIK3CA MMs.8 R88Q and H1047R had different effects on the overall structure and residue–residue contacts. Specifically, the R88Q mutation disrupted the R88–D746 salt bridge between the ABD and kinase domains, promoting the rotation of the iSH2 domain and contributing to the exposure of the kinase domain. In contrast, the H1047R mutation distorted the orientation of the kinase domain, increasing substrate accessibility. Interestingly, the R88Q–H1047R double mutant cleaved the R38–D743 salt bridge, resulting in detachment of the interface between the ABD and kinase domains, even though R88Q and H1047R single mutants had almost no effect on this salt bridge.
Epidermal growth factor receptor (EGFR) is a receptor tyrosine kinase that transduces the necessary growth factor signals from the extracellular milieu to the cell16 and is an oncogene frequently affected by MMs. EGFR is frequently mutated and/or amplified in lung cancer and glioblastoma; however, the distribution of EGFR mutations varies between these tumor types.10,17 In lung cancers, EGFR mutations mainly reside in the intracellular kinase domain, with a prominent hotspot at L858, and include both small indels (particularly exon 19 deletions and exon 20 insertions) and missense mutations.16 In glioblastoma, EGFR mutations mainly consist of missense mutations and cluster in the extracellular domain, forming a hotspot at A289, while relatively large deletions involving several exons, such as exon 2–7 deletions generating EGFR variant III transcripts, are also common.18 EGFR is an important therapeutic target for these tumors, and EGFR inhibitors are especially effective against those harboring gain-of-function mutations in the kinase domain.16
Approximately 10% of EGFR-mutated cases harbor MMs in primary tumors.8 Even though EGFR T790M mutations are known to be acquired as secondary events in patients resistant to EGFR inhibitors,5 mutational combinations involving T790M, such as T790M–L858R, are most common, followed by those involving E709 and G719 (Fig. 1b).8 Similar to PIK3CA MMs, missense mutations have a higher fraction of MMs than in-frame indels, and minor hotspot mutations contain a higher fraction of MMs than major hotspot mutations in EGFR. In addition, minor amino acid substitutions are more prevalent than major amino acid substitutions in major hotspots, such as A289. The distribution of EGFR MMs differs between lung cancers and glioblastomas. Specifically, the fraction of MMs in major hotspots and minor positions in the kinase domain is comparable between the two tumor types. However, mutations affecting minor hotspots in the extracellular domain occur more frequently in lung cancer, suggesting lineage specificity for oncogenic MMs.
Even though the EGFR T790M mutation confers resistance to EGFR inhibitors, the T790M–L858R double mutant also promotes autophosphorylation in vitro and accelerates the development of lung adenocarcinoma in vivo compared with single mutants.19,20 Importantly, a high-throughput functional evaluation of EGFR mutations revealed that both T790M-containing combinations and other mutational combinations decreased sensitivity to the EGFR inhibitors gefitinib and erlotinib.21
Apart from the 14 genes identified in the pan-cancer analysis, an additional 6 genes were identified as those in which significant enrichment of MMs was observed in a specific cancer type.8 These include FLT3 mutations in acute myeloid leukemia, JAK2 mutations in B-cell acute lymphoblastic leukemia (B-ALL), CARD11 mutations in non-Hodgkin lymphoma, JAK3 and NOTCH1 mutations in T-cell acute lymphoblastic leukemia (T-ALL), and CHD4 mutations in uterine endometrial carcinoma. Among these, NOTCH1 mutations in T-ALL show the highest frequency of MM in the mutated cases (up to 30%). NOTCH1 encodes a member of the Notch family that plays vital roles in the developmental processes by regulating cell fate decisions. In the hematopoietic system, Notch signaling promotes T-cell development by driving thymus-seeding progenitors into the T-cell lineage.22 NOTCH1 is activated by gain-of-function mutations in over 50% of T-ALL patients.23NOTCH1 missense mutations and in-frame indels are mainly located in the heterodimerization (HD) domain, and C-terminal truncating mutations preferentially occur in the proline/glutamic acid/serine/threonine enriched motif (PEST) domain. In NOTCH1 MMs, missense mutations and in-frame indels in the HD domain significantly co-exist with C-terminal truncating mutations in the PEST domain (Fig. 1c).8 Analysis of CRISPR loss-of-function screening data in the CCLE revealed that the highest relative dependency on the NOTCH1 gene was observed in a NOTCH1 MM-harboring cell line among hematopoietic and lymphoid cell lines. Luciferase assay showed that single missense mutations in the HD domain increased the transcriptional output of NOTCH1 compared to the wild type. However, HD domain mutations further increased its transcriptional activity in combination with the PEST domain truncating mutation.
Accumulating evidence suggests that MMs are common in oncogenes, in which less common (minor) positions and amino acid substitutions are preferentially selected. In oncogenic MMs, an initial mutation is considered to exert a cis-acting effect on the selection of a following mutation in the same oncogene during tumor evolution, and the two mutations synergistically enhance the oncogenic capacity. Moreover, MMs can serve as the underlying mechanisms for clonal selection of functionally weak, uncommon mutations. Remarkably, PIK3CA MMs can help predict a better response to PI3K inhibitors. In contrast, several secondary mutations, including EGFR T790M, confer resistance to tyrosine kinase inhibitors. Given that MMs occur frequently in oncogenes encoding kinases, future prospective studies should examine the utility of oncogenic MMs as biomarkers.
This study was supported by a Grant-in-Aid from the Japan Agency for Medical Research and Development (JP21cm0106575) and The Uehara Memorial Foundation.
