Introduction

The traditional subdivisions of the standard cell cycle are divided into four main phases designated as G1, S, G2, and M [1]. In addition, a fifth state, called G0, is a quiescent state that accounts for the majority of non-proliferating cells in the human body. The cell cycle is controlled by numerous mechanisms that ensure correct cell division [2]. Previous studies have demonstrated that cell cycle abnormalities are closely related to many cancers [3]. Consequently, targeting components of the cell cycle, such as cyclin-dependent kinases (CDKs), presents unique opportunities for drug discovery, and a number of potential molecular targets for novel anticancer agents have been identified in cell cycle control mechanisms [4].

Imatinib is a tyrosine kinase inhibitor that specifically targets the breakpoint cluster region-Abelson murine leukemia (BCR-ABL) fusion protein and was originally developed in the 1990s for treating chronic myeloid leukemia (CML) and some rare forms of stomach cancer. Imatinib has been shown to potently inhibit proliferation and induce apoptosis in BCR-ABL+ cells. In addition, a number of studies have reported the effects of imatinib on cell cycle-related genes in other types of cancers [5–8]. However, to date, no studies have comprehensively investigated the effect of imatinib on cell cycle-related genes.

Most human genes produce multiple mRNA isoforms through alternative splicing, which produces multiple different transcripts from one gene. Recent studies suggest that up to 94 % of all human genes undergo alternative splicing, and aberrant pre-mRNA splicing is a critical component of the pathogenesis of neoplasias [9]. Not surprisingly, genes involved in the cell cycle also undergo alternative splicing and may play important roles in tumor development; however, little is known about the role of precise splicing in cell cycle progression [10, 11].

In our previous study, we have used the Affymetrix Human Exon 1.0 ST array which contains almost 5.4 million probes groups into 1.4 million probe sets for interrogating over 1 million exon clusters to investigate the different splices of apoptotic genes between K562 cells and imatinib-treated K562 cells. As the same, this powerful tool has used to identify that genes and exons are differentially expressed in normal and imatinib-treated K562 cells in this study. Through combining exon array and several special commercial softwares for gene level expression analysis, we complete this study conveniently and persuasively.

Materials and methods

We have done similar work in our previous study, so a brief introduction of methods of this study was given in this paper. First, human myeloid leukemia cell line K562 were cultured in Roswell Park Memorial Institute 1640 (RPMI-1640; Hyclone, USA) media supplemented with 10 % fetal bovine serum (FBS; Hyclone) in a 5 % CO2 humidified atmosphere at 37 °C, and divided into two groups which include imatinib-treated cells and untreated controls. Second, total RNA was extracted from 107 K562 cells of treatment and control groups using the TRIzol reagent (Tiangen, China) following the manufacturer’s instructions. RNA samples were labeled according to the GeneChip Whole Transcript (WT) Sense Target Labeling Assay as provided by the manufacturer (http://www.affymetrix.com) and hybridized to the Human Exon 1.0 ST arrays for 16 h according to the Affymetrix protocol. Third, data was analyzed by easyExon software (http://microarray.ym.edu.tw/easyexon) and the Affymetrix Power Tools (APT) package. Differentially expressed genes (p < 0.05) between the two cell groups were functionally annotated and classified by using the functional annotation tool of database for annotation, visualization, and integrated discovery (DAVID) http://david.abcc.ncifcrf.gov/ [12], which provides integrated solutions for the annotation and analysis of genome-scale datasets derived from high-throughput technologies [13]. Last, reverse transcription-PCR (RT-PCR) was used to confirm the differentially expressed and alternatively spliced genes detected by the Human Exon 1.0 ST arrays. The primers and reaction conditions for each transcript are shown in Tables 1 and 2. Images of the gels were analyzed by Quantity One 4.6.2 software (Bio-Rad, USA). For gene analysis, ratios of gray values between the target gene and β-actin were calculated, and for exon analysis, the ratios of gray values between splice variants were calculated in untreated and imatinib-treated cells, respectively. For the detail of those methods, refer to our previous study [1].

Table 1 The primers and amplification conditions used in this study
Full size table
Table 2 Functional classification of differentially expressed genes
Full size table

Results

Gene level expression differences induced by imatinib treatment

Using the easyExon software, we conducted a t test filtration at the gene level of the two K562 cell groups. No other information was needed. For selecting differentially expressed genes, the criteria for probe set filtration were set as follows: fold change (FC) ≥2 or ≤0.5 with a statistical significance of p < 0.05 (calculated by a student’s t test). Based on these criteria, we identified 2193 transcripts with expression patterns that were significantly different between control and imatinib-treated K562 cells.

The transcript cluster filtration tool of easyExon was used to reduce the number of positive hits based on (i) GO term and (ii) gene name or accession number. Using GO term “cell cycle” for filtration, we obtained a gene list that contained 185 cell cycle-related genes, of which 147 (79.46 %) were down-regulated and 38 (20.54 %) were up-regulated. The number of genes exhibiting decreased expression was far greater than those having increased expression. Based on FC values, the 10 most up-regulated and the 10 most down-regulated genes are shown in Table 3.

Table 3 Cell cycle-related genes that were significantly changed in K562 cells in response to imatinib treatment (the 10 most up-regulated and the 10 most down-regulated genes)
Full size table

Previous studies have demonstrated that BCR-ABL can directly promote cell cycle entry in the absence of growth factors in both primary leukemic cells and BCR-ABL transfected cell lines [14]. In addition, BCR-ABL expression is sufficient to induce G1-to-S phase transition, DNA synthesis, and activation of cyclin-dependent kinases (Fig. 1). Therefore, in our study, we classified the 185 genes according to their specific functions by using the DAVID functional annotation tool (Table 2). The group enrichment score, which is the geometric mean in log scale of p values from the corresponding annotation cluster, was used to rank the biological significance of the genes. The p values associated with each annotation terms inside each clusters are exactly the same meaning/values as p values (Fisher exact/EASE score) shown in the regular chart report for the same terms (http://david.abcc.ncifcrf.gov/helps/functional_annotation.html#E4).

Fig. 1
figure 1

The cell cycle stages and differentially expressed cell cycle genes of imatinib-treated K562 cells. BCR-ABL can accelerate K562 cell cycle, which can be specifically inhibited by imatinib. Some genes involved in key steps of the K562 cell cycle are listed in the chart in red (up-regulation) and green (down-regulation)

Full size image

Exon-level expression differences induced by imatinib treatment

Through using the easyExon analysis which with a limitation of probe sets in a transcript cluster from 4 to 40, a total of 331 genes that represented by one or several of the probe sets is obtained. Those alternatively spliced genes are different expressions in the imatinib-treated K562 cells contrast with the K562 cells, and most of these probe sets had a change in expression of less than threefold. We set the easyExon “GO Biological Process” parameter to cell cycle, which was related to the transcript-affiliated gene name or accession number, and identified that 190 genes from the original set of 331 are cell cycle-related genes. Each of these genes contained at least one probe set meeting the above criteria, which represented a total of 227 probe sets. Table 4 lists the 10 most up-regulated and the 10 most down-regulated probe sets.

Table 4 Alternative splicing probe sets of cell cycle-related genes that were significantly changed in K562 cells in response to imatinib treatment (the 10 most up-regulated and the 10 most down-regulated probe sets)
Full size table

In this circumstance, there are three statistical methods available for exon-level filtration in the easyExon software, including Affymetrix Microarray Detection of Alternative Splicing (MIDAS), Alternative Splice Analysis of Variance (Partek AS ANOVA) and pattern-based correlation (PAC), we used the default Affymetrix MIDAS. The criteria for the probe set filtration for selecting alternatively spliced genes were set as follows: fold change (FC) ≥2 or ≤0.5 with a statistical significance of p < 0.05.

Detection of alternative splicing events in response to imatinib treatment

From the graphic presentation of analysis results, we were able to identify whether a specific isoform was up- or down-regulated in the imatinib-treated cells [15] (Fig. 2). After identifying probe sets that were differentially spliced in response to imatinib treatment (Supplemental Table 3), we detected changes of two randomly chosen internal exons (T-box transcription factor (TBX3) and phospholipase C, beta 1 (PLCB1)) between untreated and imatinib-treated K562 cells with using the easyExon program for experimental validation. For TBX3 (NM_016569), the probe set (ID3472775, indicated in red) that targeted exon 3 was down-regulated in imatinib-treated K562 cells. The expression fold change of this probe set was larger than 2 (labeled in red) and was statistically significant (marked by an asterisk). These results indicated the presence of two TBX3 isoforms in K562 cells: one with exon 3 and another without exon 3. In addition, the isoform lacking exon 3 was up-regulated in imatinib-treated K562 cells (Fig. 2). For PLCB1 (NM_015192), the probe set (ID 3875823, indicated in red) targeting exon 30 was up-regulated in imatinib-treated K562 cells. In addition, two spliceosomes were detected in K562 cells, which suggested that alternative splicing was affected by imatinib treatment (Fig. 3). Using this same process, we identified other transcripts that undergo alternative splicing as well.

Fig. 2
figure 2

The transcript cluster of TBX3. The transcript with exon 3 was down-regulated in the imatinib-treated K562 cell line (red line). This exon was targeted by probe set 3472775 and found to be differentially expressed using the MIDAS test (labeled in red). The fold change of this probe set between untreated and imatinib-treated cells was significant and greater than 2

Full size image
Fig. 3
figure 3

The transcript cluster of PLCB1. The 30th exon is up-regulated in the imatinib-treated K562 cell line (red line). This exon was targeted by probe set 3875823 and was identified by MIDAS (labeled in red)

Full size image

RT-PCR validation of differentially expressed and alternatively spliced genes

We confirmed the transcript level of CENPE and SLC4A1 on the basis of the largest and smallest fold changes in imatinib-treated K562 cells, respectively. Three independent samples were treated with 1 μM imatinib or dimethyl sulfoxide (DMSO) to assess differential expression by RT-PCR. As shown in Fig. 4, the mRNA levels of CENPE and SLC4A1 had the decreased and increased tendency in the imatinib-treated K562 cells compared to the controls respectively, which basically confirmed and validated the results obtained from the exon array.

Fig. 4
figure 4

RT-PCR validation of differentially expressed genes in imatinib-treated K562 cells. Representative bands of SLC4A1 and CENPE mRNA expression in K562 cells. β-actin was used as an internal control. Three independent samples were treated with 1 μM imatinib or DMSO at various time points. An increase and decrease in the relative expression of SLC4A1 and CENPE was observed in a time-dependent manner, respectively

Full size image

Using primers targeting the exons that flank the exon of interest, we also carried out RT-PCR validation of differential splicing in TBX3 and PLCB1 mRNA transcripts from imatinib-treated K562 cells. The results indicated a difference in the ratio of isoforms in response to imatinib treatment for probe sets 3472775 (TBX3) and 3875823 (PLCB1) (Fig. 5). Importantly, the relative abundance of TBX3 and PLCB1 transcripts was consistent with the trend of SI values calculated from the exon array data (Fig. 5).

Fig. 5
figure 5

Imatinib regulates TBX3 and PLCB1 splicing in K562 cells. K562 cells were treated with 1 μM of imatinib, total RNA was extracted, and RT-PCR was used to validate the alternative splicing of TBX3 and PLCB1. a A schematic representation of alternative splicing of TBX3 exon 3 and PLCB1 exon 30 as well as the resulting mRNAs. The RT-PCR primer positions are schematically presented (black arrows). b RT-PCR validation of alternatively spliced TBX3 and PLCB1. Agarose gel electrophoresis showing larger (exon-included) and smaller (exon-skipped) isoforms. An decrease and increase in the variant L/variant S ratio of TBX3 and PLCB1 was observed, respectively, in a time-dependent manner. β-actin was used as an internal control. The lane designated Marker contains a 100-bp ladder

Full size image

Discussion

Imatinib is considered to be the most effective drug for the treatment of chronic phase CML with a relatively good safety profile. In this study, we employed the Human Exon 1.0 ST array together with easyExon software to assess the comprehensive profile of differentially expressed genes and alternative splicing events in imatinib-treated K562 cells, with a focus on cell cycle-related genes. A total of 185 differentially expressed genes and 227 alternative splicing events were identified from all analyzed cell cycle genes, several of which were successfully validated by RT-PCR. Our results revealed that imatinib treatment induces aberrant gene expression and alternative splicing events during various points in the cell cycle of K562 cells (Fig. 1). Our results also demonstrated that the exon array technology is reliable and sensitive enough to detect differential expression at the gene and exon level in this cell line.

It is known that the CDK family of serine/threonine kinases is at the core of mammalian cell cycle pathways and controls cell cycle progression by phosphorylating proteins that function at specific cell cycle stages (Fig. 6) [16, 17]. This current understanding has provided a collective rationale for targeting CDKs and associated regulatory molecules as a therapeutic strategy for cancer [18]. The selective interruption of cell cycle regulation in cancer cells by interfering with CDKs provides a novel and potentially effective strategy for killing the cells. Flavopiridol (alvocidib) was the first CDK inhibitor to be tested in clinical trials after being identified from an anticancer agent screen in 1992 [19]. More specific second-generation CDK inhibitors that target single CDKs have since been developed and are currently being investigated in preclinical and clinical studies (reviewed in reference [18]). In the present study, we found that some CDK family members (i.e., CDK1, CDK2, and CDK6) and CDK regulators (i.e., cyclin A, cyclin B, and CDC25A) were markedly down-regulated in imatinib-treated K562 cells (Fig. 6), including key regulators in the G1/S transition, G2/M transition, and S phase progression. Our observations are consistent with findings by Du et al. [20] showing that the combination of arsenic trioxide (ATO) and imatinib can induce programmed cell death in K562 cells more effectively and efficiently than imatinib alone. The direct down-regulation of CDKs and CDK-related genes may abolish the cell cycle machinery, which may be a prerequisite for apoptosis in imatinib-treated CML cells. Based on these findings, we hypothesize that imatinib also inhibits CDK family members that directly regulate the cell cycle.

Fig. 6
figure 6

Regulation of CDK activity in the cell cycle. The four major mechanisms of CDK regulation are cyclin binding, CAK phosphorylation, regulatory inhibitory phosphorylation, and binding of CDK inhibitory subunits (CKIs). Genes involved in these mechanisms that are regulated by imatinib are listed in boxes. Up-regulated genes are indicated in the red box and down-regulated genes are indicated in the green box

Full size image

Cell cycle driving and surveillance are the two key biological mechanisms of cell cycle control. External stimuli from growth factors or mitotic signals (i.e., platelet-derived growth factor (PDGF) and Ras) initiates the transcription of early response genes (i.e., Juu, Fos, and Myc) followed by delayed-response genes (i.e., cyclins and CDKs), which allow cells to pass through the cell cycle checkpoints and proliferate. In our study, we found that the expression of several genes related to these two mechanisms underwent a significant change in imatinib-treated K562 cells (Fig. 7). From the gene expression profile, we concluded that imatinib can prevent cells from entering the cell cycle and can inhibit DNA repair in cells (Fig. 7), which is consistent with the findings by Majsterek et al. [21]. It has been suggested that BCR-ABL may contribute to the resistance of CML cells to imatinib by stimulating DNA repair [22]. Therefore, additional studies exploring the influence of imatinib on DNA repair will be very useful for understanding drug resistance in human leukemia.

Fig. 7
figure 7

Ideogram illustration of dynamic changes underlying imatinib-mediated cell cycle regulation in K562 cells. Cell cycle driving and surveillance are the two key biological mechanisms of cell cycle control. Some genes involved in these mechanisms and modulated by imatinib are marked in red (up-regulation) and green (down-regulation). Genes that have no change in expression are indicated in black. Note: the expression changes of CDK and cyclin families are shown in Fig. 6

Full size image

Another major finding of our study was that treatment of K562 cells with imatinib markedly shifted the number of alternative splicing events of cell cycle-related genes. Alternative splicing involves differential exon selection of a gene transcript to generate mRNA and protein isoforms with structural and functional diversity [23]. In addition, it has been demonstrated that alternative splicing is closely related to cell cycle progression [24, 25]. However, little is known about the detailed functional role of precise splicing control in cell cycle progression. Based on our study, aberrant splicing of such a large number of cell cycle-related genes suggests that alternative splicing of cell cycle genes is a prerequisite for imatinib-induced apoptosis. This conclusion is in agreement with the findings by Ahn et al., which suggested that cell cycle defects may be manifested by specific splicing defects [11]. Previous studies have also suggested that abnormal alternative splicing is associated with malignant phenotypes, such as chemoresistance and invasive properties [23]. As mentioned previously, BCR-ABL may contribute to the resistance of CML cells to chemotherapy drugs, such as imatinib, through the stimulation of DNA repair, which is part of the cell cycle surveillance mechanism. Based on these analyses, we speculate that the abnormal splicing of cell cycle-related genes may play a significant role in the resistance of CML cells to imatinib, which could serve as a target for developing novel therapeutic strategies for treating CML.

In conclusion, using exon arrays, we have discovered that imatinib induces wide-ranging effects on the expression level and alternative splicing of cell cycle-related genes in K562 leukemic cells. This information may help improve the imatinib treatment strategies in patients with CML and may be useful for imatinib resistance research. Although several of the imatinib-induced alternative splicing events were successfully validated by RT-PCR in our study, a more comprehensive validation is necessary to provide a complete analysis of the current findings.