Pathology of lung adenocarcinoma analyzed in this study. (A) A paraffin section stained by hematoxylin and eosin from a primary lung cancer tissue obtained by CT-guided biopsy (400×). In the cancer tissue, poorly differentiated tumor cell nests were present in the desmoplastic stroma. In addition, the cancer cells had plump cytoplasm and large pleomorphic nuclei. (B) Immunohistochemical analyses of the cancer (from metastatic tumor in the cervical spine). From left to right, CK7 (positive), TTF1 (positive), and CK20 (negative). The results highly suggest that the origin of this cancer is lung adenocarcinoma.

Discovery of novel transforming KIF5B-RET fusion gene in lung adenocarcinoma. (A) Graphical representation of whole-genome and transcriptome sequencing data from the liver metastatic lung cancer tissue. Chromosome ideograms are shown in the outer layer. Coverage of cancer whole-genome sequencing is shown in the first middle layer. Expression level of genes is shown in the second middle layer using heatmap. Intra- and interchromosomal fusion genes are shown in the central layer. The thickness of lines shows the amount of evidence (number of spanning reads). The KIF5B-RET fusion gene is shown in red. (B) Detection of KIF5B-RET fusion gene from transcriptome sequencing. We identified 34 “discordant paired-end reads” and 60 “spanning reads” across the exon-junction. A discordant paired-end read is defined as a read whose end-sequences are aligned to each of the fusion partner genes. A spanning read is a read, one of whose end-sequences is aligned across the junction of the predicted fusion transcript. In this analysis, the fusion occurred between the 16th exon of KIF5B and 12th exon of RET. (C) Validation of KIF5B-RET fusion transcript in RNA (cDNA) from liver metastatic cancer tissue by PCR amplification and electrophoresis. The fusion gene is only detected in the liver metastatic lung cancer tissue of AK55. The negative control cDNA (normal) were extracted from the blood of a healthy Korean individual (AK1) (Kim et al. 2009). (D) Validation of the fusion gene breakpoint using Sanger sequencing in cDNA. (E) RNA expression level of each RET exon. Active expression is observed from the 12th exon, downstream from the junction of the predicted KIF5B-RET fusion gene. This suggests that the RET oncogene is expressed exclusively from the fusion gene, rather than the natural RET gene.

Molecular characteristics of KIF5B-RET fusion kinase. (A) Functional domains of KIF5B-RET fusion kinase. The fusion kinase consists of 638 N-terminal residues of KIF5B and 402 C-terminal residues of RET kinase. As a result, the fusion protein consists of a protein kinase domain together with a coiled-coil domain. The coiled-coil domain induces dimerization of the fusion kinase, which activates the oncogenic protein tyrosine kinase domain by autophosphorylation. (B) The three-dimensional structure of the KIF5B-RET chimeric oncogene, as predicted by the PHYRE2 algorithm (Kelley and Sternberg 2009). The N- and C-terminal of the fusion protein are colored in red and blue, respectively.

A chromosomal rearrangement for generating KIF5B-RET fusion in the lung cancer tissue of AK55. (A) Detection of a 10.6-Mb-long inversion event in chromosome 10 from the whole-genome sequencing of the liver metastatic lung cancer. KIF5B is generally expressed with its universal promoter. By the inversion event, this promoter activates global expression of the KIF5B-RET fusion gene. (B) Validation of the KIF5B-RET fusion gene DNA by inversion-specific PCR amplification and electrophoresis. The fusion gene is only detected in the cancer tissues of AK55 (primary lung cancer, liver and bone metastatic lung cancer tissues). The negative control DNA sample was extracted from a healthy Korean individual (AK1) (Kim et al. 2009). (C) Identification of the fusion gene and inversion breakpoint using Sanger sequencing. The inversion breakpoints were located in the introns of KIF5B and RET as predicted. Two bases downstream from the breakpoint (chr10: 42,931,604, hg18), a 1-bp deletion was identified, suggesting error-prone DNA repair mechanisms might contribute to this inversion event after double-strand DNA breaks.

Replication studies of the KIF5B-RET fusion gene in an additional five triple-negative lung adenocarcinomas. (A) cDNA PCR targeting KIF5B-RET fusion transcripts and gel electrophoresis in the liver metastatic lung cancer of AK55 and five additional triple-negative lung adenocarcinomas. cDNA from AK55 and LC_S2 shows clear evidence of the fusion transcript. Because the fusion transcript in AK55 contains one more exon of KIF5B (exon 16) compared with that in LC_S2 (exon 15), the size of the PCR product in AK55 is longer than that in LC_S2. (B) cDNA PCR targeting KIF5B-RET fusion transcripts and gel electrophoresis in 15 double-negative lung adenocarcinomas. LC_S6 shows clear evidence of the fusion transcript. The fusion transcript in LC_S6 contains seven more exons of KIF5B (exons 17–23) compared with that in AK55. (C) Comparison of schematic KIF5B-RET fusion transcripts between AK55, LC_S2, and LC_S6. Each rectangle indicates an exon of KIF5B (blue) and RET (red) gene.