Mutational Analysis of the Tyrosine Kinome in Colorectal Cancers

See allHide authors and affiliations

Science  09 May 2003:
Vol. 300, Issue 5621, pp. 949
DOI: 10.1126/science.1082596

Tyrosine kinases (TKs) are central regulators of signaling pathways that control differentiation, transcription, cell cycle progression, apoptosis, motility, and invasion (1). Although a few TK genes have been shown to be mutationally altered in specific human cancers (1), it is not known how many or how often members of the TK gene family are altered in any particular cancer type. In this study, we have used high-throughput sequencing technologies and bioinformatics from the human genome project to address this question.

A recent analysis organized the protein kinase complement of the human genome (the “kinome”) into a dendrogram containing nine broad groups of genes (2). We selected one major branch of this dendrogram, containing three of the nine major groups, for mutational analysis. These included the 90 tyrosine kinase genes (TK group), the 43 tyrosine kinase–like genes (TKL group), and the 5 receptor guanylate cyclase genes (RGC group).

To evaluate whether these genes were genetically altered in colorectal cancer, we initially analyzed all exons encoding their predicted kinase domains. A total of 819 exons from all annotated TK, TKL, and RGC genes containing this domain were extracted from genomic databases (3). These exons were polymerase chain reaction (PCR)-amplified from DNA derived from 35 colorectal cancer cell lines and were directly sequenced (3).

From the ∼4 megabases (Mb) of sequence information obtained, we identified 14 genes that contained somatic (i.e., tumor-specific) mutations within their kinase domains. These genes were then analyzed for mutations in another 147 colorectal cancers (3). We identified 46 mutations in the 14 genes, 2 of which were synonymous; the rest were either nonsynonymous or splice site alterations (Table 1; table S1). All of these mutations were shown to be somatic in the cancers that could be assessed by sequencing of DNA from matched normal tissue.

Table 1.

Mutations observed in the tyrosine kinome (5). Genes with more than one nonsynonymous or splice site mutation observed in a panel of 182 colorectal cancers. The amino acid change resulting from each mutation is highlighted in blue if the residue is evolutionarily conserved or in orange if mutations of equivalent residues in other kinases are known to be pathogenic. Nucleotide position of mutations and gene accession numbers are available in table S1.

Two independent observations support the hypothesis that the mutations found in the seven genes mutated in more than one tumor in our cohort (NTRK3, FES, KDR, EPHA3, NTRK2, MLK4, GUCY2F) were functional rather than coincidental. First, the ratio of nonsynonymous to synonymous mutations is a reliable indicator of selection, as synonymous alterations are unlikely to exert a selective growth advantage. Only one synonymous somatic mutation was identified in the seven genes, yielding a N:S (nonsynonymous: synonymous) ratio of 37:1, far higher than the N:S ratio of 2:1 predicted for nonselected passenger mutations (P < 1 × 10-4). Second, the prevalence of nonsynonymous alterations in the kinase domains of these seven genes was 55 alterations per Mb (95% confidence interval, 33 to 85 alterations per Mb). This was more than 50 times the prevalence of synonymous mutations identified in our study, as well as previous estimates of the frequency of nonfunctional alterations in the colorectal cancer genome (∼ 1 per Mb, P < 0.001) (3). Though the effect of the mutations on kinase function has not yet been experimentally tested, their positions within each protein suggest that many of them may be activating in nature. For the genes mutated more than once, 17 of the 37 nonsynonymous mutations occurred in conserved residues in key regions in the kinase domain, including the autoinhibitory activation loop (Table 1; table S1). Six of these occurred at the precise amino acids previously shown to be pathogenic in other protein kinase genes, and another four occurred just one residue away from such residues (table S1 and fig. S1) (3). To confirm these predictions, investigation of the pathways through which the mutant kinases act are likely to yield fresh insights into colorectal cancer pathogenesis.

This study represents the first systematic mutational analysis of any gene family in any human cancer type. Despite decades of research on tyrosine kinase genes, few of the genes we identified had been previously linked to human cancer. The large-scale sequencing-based approach we used to find previously unknown gene mutations could readily be applied to other enzyme-encoding genes in any common tumor type. For example, our data suggest that a minimum of 30% of colorectal cancers contain at least one mutation in the tyrosine kinome (3), and it is possible that other kinase groups will reveal additional mutations. One reason for analyzing kinase genes is that they provide attractive targets for therapeutic intervention. This has been convincingly demonstrated for BRC-ABL tyrosine kinase in patients with leukemia (4). Our results thereby provide new opportunities for drug development. One could envision personalized therapeutics based on the kinases that are mutationally altered in individual cancers.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Tables S1 to S3


  • * These authors contributed equally to this work.

  • On leave from the University of Torino, Institute for Cancer Research, Torino, Italy.

References and Notes

Navigate This Article