Systematic Functional Interrogation of Rare Cancer Variants Identifies Oncogenic Alleles

Abstract

Cancer genome characterization efforts now provide an initial view of the somatic alterations in primary tumors. However, most point mutations occur at low frequency, and the function of these alleles remains undefined. We have developed a scalable systematic approach to interrogate the function of cancer-associated gene variants. We subjected 474 mutant alleles curated from 5,338 tumors to pooled in vivo tumor formation assays and gene expression profiling. We identified 12 transforming alleles, including two in genes (PIK3CB, POT1) that have not been shown to be tumorigenic. One rare KRAS allele, D33E, displayed tumorigenicity and constitutive activation of known RAS effector pathways. By comparing gene expression changes induced upon expression of wild-type and mutant alleles, we inferred the activity of specific alleles. Because alleles found to be mutated only once in 5,338 tumors rendered cells tumorigenic, these observations underscore the value of integrating genomic information with functional studies.

Significance: Experimentally inferring the functional status of cancer-associated mutations facilitates the interpretation of genomic information in cancer. Pooled in vivo screen and gene expression profiling identified functional variants and demonstrated that expression of rare variants induced tumorigenesis. Variant phenotyping through functional studies will facilitate defining key somatic events in cancer. Cancer Discov; 6(7); 714-26. ©2016 AACR.See related commentary by Cho and Collisson, p. 694This article is highlighted in the In This Issue feature, p. 681.

Figure 1. Project pipeline and summary of alleles included in this study

(A) Project pipeline. (B) Distribution of incidence of the alleles included in the project. 73.8% of the 474 alleles included in this study were found to be mutated only once. (C) Alleles mutated frequently were also found to be mutated in larger number of lineages. The size of dots corresponds to the number of overlapping dots.

Figure 2. Pooled in vivo screen identifies novel transforming alleles

(A) Pooled in vivo screen design. (B) Tumor formation over an 18 week timeframe per pool. (C) Pool 1, a positive control pool, showed consistent tumor composition across tumors. Each tumor is represented as a bar. The compositions of tumors were shown as different colors. (D) KRAS^D33E induced tumor formation in pool 5. (E) NFE2L2^G31R and POT1^G76V induced tumor formation in pool 4. (F) NFE2L2^G31R and PIK3CB^E497D induced tumor formation in pool 9. (G) Summary of the in vivo pooled screen. X-axis shows penetrance, which was calculated to be (times each allele was more than 0.01% of tumor reads) / (number of sites the allele was implanted). Since tumor size cannot exceed 2cm in the longest dimension, not all sites were observed for the full length of time. Y-axis shows maximum enrichment, which was calculated to be (maximum percentage of allele in any tumor) – (percentage of the allele in pre-injection cell pellet). Positive controls (colored in grey) had penetrance of around 80%, and low maximum enrichment due to competition against each other.

Figure 3. Gene expression profiling differentiates functional variants

(A) Expression signatures were analyzed by pairwise Spearman correlation to identify similar or dissimilar alleles to the allele of interest. (B) KRAS^G12V induces similar gene expression changes as other known activating alleles of KRAS and NRAS. (C) NRAS^Q61H induces similar gene expression changes as other known activating alleles of NRAS. However, the signature from the novel Y64D allele had a lower correlation, similar to wild type. (D) IDH1/2 alleles were correlated to known activating mutant IDH2^R172K. Other known activating alleles of IDH1/2 are highly correlated to IDH2^R172K. (E) When correlated to the PTEN wild type, F90S, R233Q, K6N, R173H correlated strongly with the wild type PTEN. The known loss-of-function, dominant negative allele G129E showed a lower correlation. G127R, G129V, G127V also showed low correlation to the wild type. (F) When alleles were correlated against SPOP^F102C, a loss-of-function, dominant negative SPOP allele, other known loss-of-function, dominant negative alleles W131G, F133S, K134N, and W131C were highly correlated. On the other hand, E50K, K101I, E47A had lower correlation to F102C. (G) FBXW7 wild type, R658Q, I347M, S462Y, and R689Q, were strongly anti-correlated to MYC. Known dominant negative alleles (R505C, R465C, R465H) no longer were anti-correlated to MYC. BRD4 wild type was the most closely correlated to MYC.

Figure 4. Validation of rare oncogenic alleles

(A) Individual tumor validation of KRAS alleles. The KRAS^D33E and KRAS^A59G alleles formed tumors robustly. E62K did not form tumors in the pooled assay but formed tumors in individual assays, at a later time point. (B) Individual tumor validation of NFE2L2 alleles. In the pooled assay, only G31R scored in multiple tumors. In the individual assay, G31V, G31A, T80K formed tumors as well. N160S formed tumors at a later time point. NFE2L2 wild type formed one small tumor by the end of the experiment. (C) Individual tumor validation of PIK3CB alleles. E497D and the wild type formed tumors after long latency. PIK3CB^A1048V formed tumors with shorter latency at the majority of injection sites. (D) Individual tumor validation of POT1 alleles. The G76V allele formed tumor at a later time point. One of the POT1^G76V mice died of unknown cause. (E) The structure of KRAS (PDB: 4EPV) shows that all four of the mutants are in close spatial proximity. Mutated residues are shown in red, GDP bound to the substrate pocket is shown in blue. (F) Immunoblot of KRAS alleles (including other positive control alleles) shows increased phospho-ERK and phospho-AKT1 levels in KRAS^D33E, and KRAS^A59G overexpressed cells. (G) RAF binding domain pull down assay shows increased GTP bound KRAS in D33E and A59G mutants. (H) Quantitative PCR of NFE2L2 mRNA expression shows all alleles were expressed. (I) Immunoblot of NFE2L2 alleles show increased NFE2L2 protein level in G31A, G31R, G31V and T80K overexpressed cells. There was no change in phospho-ERK or phospho-AKT1 levels.