A novel BMX variant promotes tumor cell growth and migration in lung adenocarcinoma

The non-receptor tyrosine kinase BMX has been reported in several solid tumors. However, the alternative splicing of BMX and its clinical relevance in lung cancer remain to be elucidated. Exon1.0 array was used to identify a novel alternative splicing of BMX, BMXΔN, which was confirmed by rapid amplification of cDNA ends and reverse transcription-polymerase chain reaction. BMXΔN, resulting from exon skipping with excluding exon 1 to exon 8 of BMX gene, was found in 12% human lung adenocarcinoma specimens. BMXΔN is not found in paired pathologically normal lungs and positively correlated with EGFR mutation in lung adenocarcinomas. Moreover, BMXΔN increases cell proliferation, neoplastic transformation, and migratory property of human non-small cell lung cancer cells. The function of BMXΔN in lung cancer might be presumably due to enhanced ERK signaling.


INTRODUCTION
Alternatively spliced proteins are particularly relevant in oncology since they have been linked to cancer progression and drug resistance [1,2]. They may provide selective drug targets, or serve as a marker set for cancer diagnosis as well. For example, the tumor suppressor gene p53 is subject to alternative splicing and p53 splice variants are frequently expressed in primary ovarian cancers [3]. The p53δ, encoding a C-terminally truncated protein, was demonstrated to be associated with impaired response to primary platinum-based chemotherapy and might serve as an adverse prognostic marker for recurrence free and overall survival in ovarian cancers [3]. A splicing variant of Merlin, Δ2-4 Merlin, promotes tumor metastasis by interfering with the tumor suppression role of wild type Merlin [4]. There is substantial evidence that primary metabolism is altered in cancer cells, and the pyruvate kinase M1 and M2 splicing isoforms control the balance between aerobic and anaerobic glycolysis during tumor progression [5,6]. These observations emphasized the importance of investigation of alternative splicing genes in cancer for improving targeted therapy.
Affymetrix Exon1.0 array detects gene expression at single exon level. This facilitates the identification of alternative splicing isoform of certain genes as well as the gene fusions. Our previous study has analyzed the Exon1.0 array from 76 Chinese lung adenocarcinomas and identified CCDC6-RET fusion as novel oncogenic driver [7]. Certainly, except for the gene fusion, detection of alternative splicing is another outcome of this dataset.

Research Paper
BMX (bone marrow tyrosine kinase gene in chromosome X), which encodes a non-receptor tyrosine kinase belonging to BTK (Bruton's tyrosine kinase) family. BMX has been shown to play a pivotal role in the regulation of various cellular processes including proliferation, differentiation, transformation, apoptosis, and cell motility. Previous study described BMX as a direct substrate for caspases and the resulting truncated molecule contains an intact SH2 domain and kinase domain which has an enhanced kinase activity [8]. BMX acts upstream of RhoA and activates RhoA by releasing GDI from the RhoA-GDI complex through the interaction between the PH domain of BMX and RhoA [9]. BMX directly associates with Pak1 via its N-terminal pleckstrin homology domain and also phosphorylates Pak1 on tyrosine residues [10]. Study has also shown that BMX interacts with p53 in response to DNA damage and that such interaction leads to bidirectional inhibition of the activities of both proteins in LNCaP human prostate carcinoma cells [11]. Studies also illustrated some of the upstream activator for BMX. For example, BMX activity is modulated by FAK through an interaction between the PH domain of BMX and the FERM domain of FAK and the activation of BMX by FAK promotes cell migration [12]. In addition, BMX can be induced by growth factors, cytokines [13], the extracellular matrix, and possibly by hormones [14]. More importantly, BMX mediates various signaling pathways including STAT signaling pathway [15,16], PI-3K signaling pathways [17][18][19], and GPCR signaling pathway [20].
BMX expression is altered in a number of different cancers, including those of the breast and prostate [10,[21][22][23], suggesting BMX may play roles in cancers. For example, BMX expression level is up-regulated in hormoneresistant prostate cancer and positively correlated with tyrosine phosphorylation of AR conditions. Overexpression of BMX in androgen-sensitive LNCaP cells promotes tumor growth while knocking down BMX expression in hormoneinsensitive prostate cancer cells inhibits tumor growth under androgen-depleted conditions [24].
Here we describe the discovery of a novel spliced variant of BMX, designated as BMXΔN, which results from the skipping exon 1 to exon 8 in BMX gene. BMX∆N is strongly associated with EGFR mutation in clinical samples. Moreover, this isoform promotes lung cancer cell growth, migration, and neoplastic transformation.

Identification of a novel BMX skipping isoform in lung adenocarcinoma
Through bioinformatics analyses of Exon1.0 array data from Chinese lung adenocarcinoma and 5′ RACE, we identified a novel BMX skipping variant ( Figure  1A, 1B). We called this novel BMX isoform, BMX∆N, which lacked the N-terminal sequence from exon 1 to exon 8 ( Figure 1C). We further found that BMX∆N was absent in all the 14 paired non-cancerous lung tissues. Representative reverse transcription-PCR analysis showed that BMX∆N was detectable in lung adenocarcinomas but not in paired non-cancerous lung samples ( Figure 1D). Then, we expanded the study of BMX∆N in a cohort with 174 adenocarcinoma samples and identified a total of 21 lung adenocarcinomas harboring this isoform (12%, 21/174) ( Figure 1E).

Detection of BMX∆N translation start codon
The sequence of the BMX∆N gene contains four putative start codons (ATG 1 -ATG 4 ). We detected at which ATG codon BMX∆N translation initiates. We constructed a series of plasmids with different ATGs and then transfected the plasmid into HEK-293T cells (Figure 2A). Western blot analysis of total protein from HEK-293T cells showed that BMX∆N was translated from plasmid carrying ATG 3 ( Figure 2B), indicating that the ATG located in exon 13 is the start codon for BMX∆N.

The relationship between BMX∆N expression and EGFR mutation
We further analyzed the relationship between BMX∆N expression and clinicopathological features in human lung adenocarcinomas (Tables 1 and 2). BMX∆N expression was not significantly correlated with age, gender, pathological stage ( Table 2) and metastasis (Table 1). However, we found that BMX∆N was tightly associated with EGFR mutation (p = 0.002). Indeed, 20 out of 21 samples harbor EGFR mutation (Table 2).

Low expression of BMX in lung adenocarcinomas
On the basis of previous studies showing three transcript variants of BMX, a pair of primers (F8/R12) was designed encompassing exons 8 to 12 of the BMX open reading frame for detection of wild type BMX and other two variants. Another pair of primers (F16/R17) was also designed encompassing exons 16 and 17 of the BMX open reading frame to detect all BMX isoforms including BMXΔN. Using F16/R17 primers to observe the levels of BMX mRNA in lung adenocarcinomas and adjacent non-tumour specimens by quantitative PCR, we found that there were no significant differences in expression. However, the transcript of BMX was different between BMXΔN positive lung adenocarcinomas and adjacent non-tumour specimens ( Figure 3). Interestingly, when we use F8/R12 primers to detect the levels of BMX mRNA in BMXΔN positive lung adenocarcinomas, we found very low expression of BMX or even no expression of BMX in these lung adenocarcinomas (data not shown). These www.impactjournals.com/oncotarget findings indicate that BMXΔN is the dominant isoform in these specimens. Because wild type BMX functions as an oncogene in prostate cancer [24], we decided to explore the role of BMXΔN in our study of lung carcinogenesis.

Effect of BMX∆N on NSCLC cell growth
To investigate the role of BMX∆N in cell growth, we performed cell proliferation assay on A549, CRL-5872, and PC9 cells ( Figure 4). Expression of BMX∆N efficiently promoted cell growth in A549 ( Figure 4B). The wide type BMX elevated cell growth as well. To examine whether BMX∆N is involved in cell transformation, we performed soft agar colony formation assays on these cells. Figure 4C shows that more colonies were formed in BMX∆N expressing cells compared with control. Similar growth promotion effect and transformation activity were also observed in CRL-5872 cells ( Figure    Only patients with detailed pathological data were compared in statistical analysis. formation in soft agar ( Figure 4J, 4K). These data suggested that BMX∆N promoted lung cancer cell growth in vitro.

BMX∆N facilitates tumor cell migration and enables cell transformation
We further studied the role of BMX∆N in lung cancer cell migration ( Figure 5A, 5B). The wound-healing assay showed BMX∆N transfected PC9 cells obtained quicker closure of the scratched "wound" compared with control cells (Figure 5A). Migration was also examined using transwell assays where the cells were incubated in serum-free DMEM medium in the upper compartment and allowed to migrate towards the lower compartment containing 15% FBS. The result showed that enforced BMXΔN expression greatly increased the migration ability of A549 cells ( Figure 5B). To evaluate the transformation capacity of BMX∆N, we introduced BMX∆N and mutant EGFR into Ba/F3 cells. BMX∆N-transfected Ba/  F3 cells showed accelerated growth rate compared with mock transfectants, whereas no difference in growth was observed between the BMX∆N-and EGFR L858Rtransfected cells. These results indicate that BMX∆N was capable of transforming Ba/F3 cells in vitro ( Figure 5C). The above findings demonstrated an important role of BMX∆N in lung cell carcinogenesis.

BMX∆N activated ERK in lung cancer cells
Previous studies have shown that BMX expression could activate several signaling pathways, including PI3-AKT pathway and STAT pathway. We tested if these pathways are also involved in BMX∆N function in lung cancer. We detected the phosphorylation of STAT3, ERK, AKT and FAK in lung cancer cells expressing either wild type BMX or BMX∆N and control cells followed by EGF stimulation. As shown in Figure 6, the much higher level of expression of phosphorylated ERK could be detected in BMX∆N transfected cells, suggesting MAPK pathway might contribute to the role of BMX∆N in lung cancer.
Our data indicated that other tested pathways were not affected by BMX∆N expression in A549 cells.

DISCUSSION
The discovery of alternative splicing variants in cancers has been paid much attention recently and their   detection potentially increases with the use of innovative approaches. We here identified a BMX skipping isoform through the analyses of Exon1.0 array profiling of human lung adenocarcinoma samples in combination with RACE method. BMX∆N lacks the N-terminal sequence from exon 1 to exon 8. BMX∆N only expresses in lung cancer but not in paired non-cancerous tissues. Through screening a large collection of NSCLC patient samples, we identified a total of 21 lung adenocarcinomas expressing BMX∆N (12%, 21/174). Interestingly, BMX∆N is strongly associated with EGFR mutation in our study. However, the clinical relevance is not known yet.
A few studies have correlated BMX function with tumor growth, metastasis or poor prognosis in cancer. Overexpression of BMX in androgen-sensitive LNCaP cells promotes tumor growth while knocking down BMX expression in hormone-insensitive prostate cancer cells inhibits tumor growth under androgen-depleted conditions [21]. BMX is up-regulated in bladder cancer and predicts poor prognosis in patients with cystectomy [25]. A study has shown that BMX could maintain selfrenewal and tumorigenic potential of glioblastoma stem cells by activating STAT3 [26]. Here, we present evidence that BMX was not up-regulated in lung adenocarcinomas. It might not contribute to lung carcinogenesis despite knowing it could promote cell proliferation of NSCLC. Importantly, our results reveal a previously unknown splicing skipping form of BMX. The studies suggested that BMX∆N might play roles in lung tumorigenicity, with expression of BMX∆N promoting cell growth, cell migration, and cell transformation. Future studies are necessary to clarify the mechanism by which BMX∆N activates ERK1/2.
Collectively, this study discovered a novel BMX skipping with crucial function in lung cancer cells. Future studies into this novel BMX variant might provide a better understanding of lung tumorigenesis and clinical implication for therapeutics.

Specimen collection
The study was approved by the ethics review board at Fudan University Shanghai Cancer Center, Shanghai, China. 174 cases of lung adenocarcinomas with paired pathological normal lungs were collected consecutively with written informed consents from all patients. Fresh surgical specimens were snap-frozen and stored in liquid nitrogen upon resection until use. The pathology of each tumor sample was determined by pathologists. All these specimens were with a minimum of 70% of tumor cellularity, and all patients did not receive neoadjuvant chemotherapy. The status of EGFR mutations and other drive mutations in these specimens was determined as previously described [27,28]. The correlation of BMX∆N expression and patients' clinical characteristics were illustrated in 146 lung adenocarcinoma samples, a subset of 174 cases, containing 19 BMX∆N positive samples (Table 1).

Cell culture, DNA constructs and plasmid transfection
A549, CRL-5872 and PC9 cells were purchased from the ATCC. Cells were cultured in DMEM, supplemented with 8% Fetal Bovine Serum (FBS), 100 μg/ml streptomycin and 100 U/ml penicillin, at 37ºC in 5% CO2 incubator. A lentiviral construct expressing wild type BMX or BMX∆N were generated by cloning a DNA fragment corresponding to BMX full length or BMX residues 384-675 (NP-001712.1) into the NheI and NotI sites of pCDH-CMV-copGFP vector (SBI). Viral particles were produced in HEK-293T cells co-transfected with pCDH constructs and packaging plasmids pCMV-VSVG/ delta8.2 (System Biosciences) in DMEM media. The progeny viruses released from HEK-293T cells were filtered, collected and used to infect cells.

Gene functional assays
For cell proliferation assay, cells were seeded in 96well plates at a density of 3×10 3 cells per well, and cell growth rate was assessed with the 3-(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) kit (Roche Diagnostics). The MTT assays in each cell line repeated three times, respectively. For soft agar colony formation assay, 8×10 3 cells were seeded in 6-well plates, and after three weeks of culture cell colonies were counted by crystal violet staining. The results are expressed as the mean ± SD of three independent experiments. For woundhealing assay, PC9 cells were cultured on a 12-well plate and maintained in DMEM. At 80% to 90% confluence, the cells were starved for 12 hours cultured in DMEM without FBS. A 10 µl pipette tip was used to create a linear scratch. The cells were then washed with PBS to remove floating cellular debris and fed for an additional 24 hours with full DMEM. Migration photos were captured immediately after scratching and at 24 hours after scratching by a digital camera. Cell migration was also assessed using 12well transwell chambers (Corning Costar) with a pore size of 8 μm. A549 cells (1 × 10 5 ) were seeded in serum-free medium in the upper chamber and incubated at 37 °C for 24 h. Afterward, the cells remained in the upper chamber were carefully removed with a cotton swab, whereas the cells having traversed to reverse face of the membrane were fixed with 5% acetic acid and stained with 0.4% crystal violet. Three random fields were counted at x20 magnification. The results represent the average of samples from three independent experiments. Oncogenic transformation assay was performed in Ba/F3 cells. Ba/ F3 cells were infected with lentivirus containing a control vector or a BMX∆N/EGFR-L858R plasmid. Infected cells were incubated with IL-3 (0.5 pg/mL) to support Ba/F3 marginal growth for approximately 72 hours. 3000 Ba/ F3 cells per well were plated in quadruplicate in 96-well plates and cultivated for 3 days without IL3. Cell viability was measured daily. The experiments were repeated independently three times.

Exon array analysis and 5′ RACE
Exon array analysis and 5′ RACE were performed as previously described [7]. Briefly, Affymetrix Human Exon 1.0 microarray was used and the Robust Multichip Average method was applied to perform background correction, normalization and exon-level probe set summarization. 5′ RACE-PCR was performed using SMARTer TM RACE cDNA Amplification Kit from Clontech Laboratories Inc (Mountain View, CA) according to the manufacturer's instructions. In brief, 1 µg RNA extracted from lung cancer patient tissues was reverse transcribed using primers 5′-RACE CDS primer A and SMARTer II A Oligonucleotide supplied by SMARTer TM RACE cDNA Amplification Kit. PCR was performed with BMX gene specific primer 5′-GGTTCAAGTCCTTTTC CGTGACTCCTCA-3′/5′-CCCGAAGTGGTTCAATGGA AGACAGGA-3′ in conjunction with RACE universal primer A mix (UPM): 5′-CTAATACGACTCACTATA GGGCAAGCAGTGGTATCAACGCAGAGT-3′ and 5′-CT AATACGACTCACTATAGGGC-3′. PCR products were purified for direct sequencing as well as cloning into pGEM-T vector (Promega, Madison, WI) for sequencing.

Statistical analysis
The statistical analysis was conducted in SPSS 16.0 (SPSS Inc, Chicago, IL, USA). Pearson's chi-squared test was used on categorical variables. Two group comparisons were analyzed by the two-tailed Student's t test. A p value less than 0.05 was considered statistically significant.

Authors' contributions
YW and JX designed and performed the experiments, interpreted the data and discussed the manuscript. ZF and FL analyzed and interpreted the Exon array data to identify alternative spliced genes, DL, ZW and YF participated in performing the experiments, JZ reorganized clinical information, HC provided clinical information and samples, HJ designed the experiments, interpreted the data and wrote the manuscript. HL designed and performed the experiments, interpreted the data and wrote the manuscript. All authors read and approved the final manuscript.