KIF5B-RET fusion kinase promotes cell growth by multilevel activation of STAT3 in lung cancer

We performed RT-PCR assay for the KIF5B-RET fusion genes in 100 Chinese LAD (Lung adenocarcinoma) patients. One of samples had PCR products, suspected as KIF5B-RET fusion genes, while matched adjacent normal lung tissue had no band, suggesting that the translocation was somatic (Additional file1: Figure S1.A). KIF5B-RET transcripts were detected in 1 of 100 tumors (about 1.0%). None of the 7 lung cancer cell lines that we examined harbored the fusion transcript (data not shown). From the direct sequencing of the PCR products, it showed junctions between exon 15 of the KIF5B gene and exon 12 of the RET gene (Additional file1: Figure S1.B). We subsequently measured the protein expression of KIF5B-RET in the positive tumor tissue. Based on the mRNA sequence of KIF5B-RET, the predicted protein is approximately 108 kDa. Western blot analysis using an anti-RET antibody revealed a protein band, of which the size corresponded to the predicted value, inferring the expression of KIF5B-RET fusion protein (Figure 1A). Imunohistochemical analysis performed on the paraffin sections of the positive tumor tissue revealed moderate, cytoplasmic, non membranal, KIF5B-RET expression in tumor cells (Figure 1B). Since RET protein had low expression in this sample, the staining implicated expression of the fusion proteins.

This positive case was a poorly differentiated nodule subtype LAD. The patient was a countrywoman aged 58 years, and had no smoking history. The mass measured 4 cm in greatest dimension, and the right lung pleura and the bronchus were involved, but no lymph node metastasis or distant metastasis (pathologic stage I b). RET expression was observed as moderately intense cytoplasmic staining but no obvious perinuclear or paranuclear accentuation. The tumor cells also had strong expression of TTF1 and Napsin A. Characteristics are listed in Additional file1: Table S1 and Additional file1: Table S2. The clinicopathological characteristics of this case were consistent with previous researches[29], namely, more poorly differentiated tumors, younger (<65 years), non-smokers, and solid subtype.

It was predicted that the N-terminal portion of the KIF5B coiled-coil region, retained in all variants, has the ability to dimerize through two coiled-coil structures, similar to the KIF5B-ALK fusions[30]. Consistently, when the KIF5B-RET variant 1 and 2 were exogenously expressed in A549 human lung cancer cells, RET Tyr905 was phosphorylated in the absence of serum or GDNF stimulation, indicating an aberrant activation of RET kinase by fusing with KIF5B (Figure 2A). To determine whether KIF5B-RET directly affects the proliferative and colony-forming abilities of NSCLC cells, human lung cancer A549 cells were infected with KIF5B-RET-expressing, wild type RET expressing, or control lentivirus. After blasticidin selection, the cell proliferation was measured with cell counting and the number of colonies was scored. The KIF5B-RET transfected cells showed a significantly increasing proliferation rate when compared with the RET transfected or control cells (Figure 2B). Similarly, compared with control and RET-expressing cells, the enforced expression of KIF5B-RET in A549 cells also caused an increase in colony number and size (Figure 2C). Consistent with in vitro observations, we also confirmed that the enforced expression of KIF5B-RET caused a significant increase in A549 xenograft tumor weight in nude mice compared with control (KIF5B-RET group vs control group: 0.53 ± 0.2 g vs 0.22 ± 0.15 g, ^***P < 0.001; Figure 3). All of these findings corroborate that the KIF5B-RET fusion kinase promotes the growth of lung cancer cells both in vitro and in vivo.

It has been shown that KIF5B-RET fusion kinase promotes cell proliferation of lung cancer, then the phosphorylation levels of proliferation related signaling molecules was investigated by measuring the enforced expression of KIF5B-RET in A549 and Beas-2b cells. We found that ERK and STAT3 signaling pathways were aberrantly activated in enhanced KIF5B-RET-expressing Beas-2b (Figure 4A) or A549 cells (Figure 4B). Moreover, the KIF5B-RET-induced phosphorylation of ERK and STAT3 was also reduced by ZD6474 treatment in a similar dose response to RET phosphorylation (Figures 4C&4D). Taken together, ERK and STAT3 signaling pathways may be involved in mediating the effect of KIF5B-RET in promoting cell proliferation.

To further investigate the role of STAT3 in mediating the function of KIF5B-RET in cell growth, we analyzed STAT3 phosphorylation in 293 T cells expressing KIF5B-RET. As shown in Figure 5A, the expression of KIF5B-RET fusion kinase increased STAT3 phosphorylation on both Tyr⁷⁰⁵ and Ser⁷²⁷. As the JAK and c-Src tyrosine kinases are well known as upstream kinases for the phosphorylation of the Tyr⁷⁰⁵ residue in STAT3, we next asked whether these kinases were still involved in KIF5B-RET-induced Tyr⁷⁰⁵ phosphorylation of STAT3. Indeed, both a JAK2 inhibitor (AG490) and a c-Src inhibitor (PP1) inhibited STAT3 on Tyr⁷⁰⁵ phosphorylation in various degrees. However, more interestingly, after treating with AG490 or PP1, the KIF5B-RET-expressing A549 cells still expressed a higher level of STAT3 Tyr⁷⁰⁵ phosphorylation compared to the control (Figure 5B), indicating that apart from the upstream JAK and c-Src kinases, KIF5B-RET may also directly phosphorylate and activate STAT3 Tyr⁷⁰⁵. We next performed a co-immunoprecipitation assay to confirm whether KIF5B-RET binds to STAT3. KIF5B-RET-FLAG expressing 293 T cell lysates were incubated with FLAG antibody, and the immunecomplexes were then purified, separated by SDS-PAGE, and analyzed with western blot. We found that STAT3 was present in the protein complex immunoprecipitated by the FLAG antibody (Figure 5C). To determine whether STAT3 is the direct substrate of KIF5B-RET, purified GST-STAT3 protein was incubated with purified activated KIF5B-RET protein from immunoprecipitation, and the result showed that activated KIF5B-RET phosphorylated STAT3 at Tyr⁷⁰⁵ site (Figure 5D). Taken together, these observations suggested that STAT3 is likely a direct substrate of KIF5B-RET in positive NSCLC cells, and that KIF5B-RET induced cell proliferation may be mediated, at least in part, through its phosphorylation of STAT3 Tyr⁷⁰⁵ directly. In addition, our previous experiments show KIF5B-RET co-activated ERK and STAT3 signaling pathways in positive cells, and the MEK_1/2 inhibitor (U0126) attenuated the KIF5B-RET-induced ERK_1/2 and STAT3 Ser⁷²⁷ phosphorylation, but not Tyr⁷⁰⁵ phosphorylation (Figure 5E), suggesting that KIF5B-RET-induced STAT3 Ser⁷²⁷ phosphorylation was partly mediated through the Ras/Raf/MEK_1/2/ERK_1/2 pathway. Since cyclinD,VEGF,and ICAM-1 are major targets of STAT3, and are involved in cell proliferation and invasion, we analyzed whether KIF5B-RET affected expression of these genes. As shown in Figure 5F, the expression of cyclinD1, VEGF and ICAM-1 was higher in KIF5B-RET positive cells than in parental A549 and BEAS-2B cells. Together, KIF5B-RET fusion protein induces multilevel activation of STAT3 which may target cyclinD1, VEGF and ICAM-1 and play a key role in oncogenesis.

The important role of KIF5B-RET-STAT3 in promoting the proliferation of lung cancer cells led us to wonder what would happen after inhibition at any step of these signalings. To date, a few tyrosine kinase inhibitors have been shown to inhibit oncogenic RET activity, for example, ZD6474, which is a selective inhibitor of the vascular endothelial growth factor (VEGF) receprot-2 tyrosine kinase. We first assessed the IC50 values of ZD6474 in four different NSCLC cell lines (Beas-2b, NCI-H1299, A549, and KIF5B-RET-expressing A549 cells) (Figure 6A). The IC50 value was 14.20 μM in KIF5B-RET-expressing A549 cells, whereas in other three cell lines they were all higher than 64 μM. Western blot were then analyzed in the presence of ZD6474, the phosphorylation of KIF5B-RET was reduced by ZD6474 in a concentration-dependent manner (Figure 6B). Consistent with these results, ZD6474 significantly inhibited the cell proliferation and colony formation of the KIF5B-RET-expressing A549 cells, and it exerted less effects on the control A549 cells (Figures 6C-6E). These findings suggest that the suppression of the RET kinase activity reduces the proliferation of KIF5B-RET positive cells in vitro. Since STAT3 pathway plays an important role in mediating the KIF5B-RET promoted cell growth, we used some related inhibitors to investigate the effect on cell growth after down regulation of STAT3 signaling. U0126 is a MEK inhibitor which inhibits STAT3 Ser⁷²⁷ phosphorylation mediated through the Ras/Raf/MEK1/2/ERK1/2 pathway. PP1 is an inhibitor of Src-family tyrosine kinases which inhibits upstream kinases for the STAT3 Tyr⁷⁰⁵ phosphorylation. S3I-201 is an inhibitor of STAT3 transcription factor activation, dimerization, and gene transcription. Colony formation assays were performed in the presence or absence of the inhibitors U0126, PP1, or S3I-201. Significantly, all these inhibitors reduced the number and sizes of colonies of A549 cells carrying KIF5B-RET (Figure 7A). And the cell proliferation of A549 cells with KIF5B-RET also markedly decreased after the treatment of the STAT3 inhibitors S3I-201(Figure 7B).

A549, H1299, Beas-2b, and 293 T cell lines were all from the cell bank of Chinese academy of sciences. A549 and H1299 cells were cultured at 37°C in RPMI-1640 supplemented with 10% heat-inactivated FCS. Beas-2b and 293 T cells were cultured in DMEM with 10% FCS.

Different inhibitors of specific signal transduction pathways, including Vandetanib (ZD6474), U0126, PP1, AG490 and S3I-201, were purchased from Selleck. Phosphor-Ret(Tyr⁹⁰⁵), Ret, phospho-STAT3 (Tyr⁷⁰⁵), Phospho-STAT3(Ser⁷²⁷), STAT3, phospho-ERK1/2(Thr²⁰²/Tyr²⁰⁴), ERK_1/2, glyceraledehyde-3-phosphatedehydrogenase (GAPDH), and anti-Flag antibodies were purchased from Cell Signaling Technology. STAT3 recombinant protein was purchased from Abnova.

Primary lung cancers tissues were from Chinese patients who did not receive neoadjuvant therapy and who underwent resection at Zhejiang Provincial Cancer Hospital, Hangzhou, between 2008 and 2010. The corresponding non-neoplastic lung tissues were immediately frozen and stored at −80°C until assayed. Informed consent and ethics approval was obtained for research purposes. Ethics committee of the hospital approved the study.

Total RNA was extracted from lung cancer tissues or cultured cells with TRIzol Reagent (Invotrogen). Revert Aid First Strand cDNA Synthesis Kit (Fermentas) was used to construct the template cDNA for realtime PCR. The primer sequences for screening the KIF5B/RET fusion gene were as follows: forward primer, 5′-AGGAAATGACCAACCACCAG-3′, and reverse primer, 5′-TCCAAATTCGCCTTCTCCTA-3′. PCR was performed with initial denaturation at 95°C for5 min, followed by 40 cycles of amplification (at 98°C for 10 sec, 60°C for 15 sec, and 72°C for 3 min), and final extension at 72°C for 5 minutes.

Total protein lysates were obtained from cultured cells or tumor tissues using radio-immunoprecipitation assay. Cell extracts were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (8% polyacrylamide gels) and then transferred to nitrocellulose membranes. The membranes were blocked overnight with TBS containing 0.1% Tween 20 (TBST) and 5% nonfat milk, and probed with the primary antibodies overnight at 4°C. After washing with TBST, the membranes were incubated with horseradish peroxidase (HRP)–conjugated secondary antibody for 1 h at room temperature, washed with TBST 3 times, and reacted with Super Signal West Pico chemiluminescent substrate (Pierce).

Tumor samples were fixed in 10% neutral-buffered formalin for 24 h and embedded in paraffin. For hematoxylin-eosin stain, sections were reacted with hemalum for nuclear staining, which is followed by counterstaining with eosin for staining of other eosinophilic structures. For immunohistochemical stain, sections were reacted with primary antibodies for 24 h at 4°C, reacted with HRP–conjugated secondary antibody for 1 h, and diaminobenzidine (DAB) was used as the substrate to produce an observable brown color.

A full-length KIF5B-RET cDNA was cloned into lentiviral vector for constitutive gene expression (pLenti6.3-MCS-IRES2 vector, Invitrogen). Lentiviral vector was cotransfected with pLP1, pLP2, and pLP/VSVG packaging vectors (Invitrogen) into 293 T cells. A549 and Beas-2b cells were infected with empty, KIF5B-RET-FLAG -expressing lentiviruses, and were treated with Blastcidin (10 μg/ml) for 2 weeks.

300 cells were cultured in a 6-well culture plate for 14 days. At the end of incubation, cells were fixed with methanol for 10 min and stained with crystal violet for 10 ~ 15 min. Three wells of cell colonies were scored (≥40 cells).

Four-week-old athymic nude mice (BALB/c nude) were used in these studies with the approval of the institution’s ethics committee. 1 × 10⁶ control and A549 cells, carrying KIF5B-RET, were separately subcutaneously injected in the left and right flanks of mice. The tumors were measured 1 month later. The weight of the xenograft tumors was recorded and analyzed.

Flag and STAT3 antibodies were used for Co-IP. Cell lysates were incubated with primary antibodies or control IgG overnight at 4°C and the immune complex was precipitated by the ProteinG Magnetic Beads (Millipore).The beads were then washed, boiled, and subjected to SDS-PAGE.

Values are shown as means ± SEM. Statistical analysis was performed by Student’s t-test, and P value less than 0.05 was considered statistically significant.