Modification of Mcl-1 alternative splicing induces apoptosis and suppresses tumor proliferation in gastric cancer

No significant change in Mcl-1 expression in GC tissues was observed

To investigate the characteristics of Mcl-1 distribution in GC, data on Mcl-1 expression in GC were collected from the Cancer Genome Atlas (TCGA) database and summarized. Mcl-1 expression showed a statistically non-significant increasing trend related to poor survival and higher tumor grades and cancer stages (Supplementary Figure 1). These results indicated that the Mcl-1 expression, without distinguishing Mcl-1L and Mcl-1S isoform expression, presented no substantial changes in GC tissues and no obvious association with GC development.

Mcl-1L expression was increased while Mcl-1S expression was decreased in GC

Given that the isoforms of Mcl-1L is pro-apoptotic and the S form anti-apoptotic, respectively, we examined Mcl-1L and Mcl-1S expression in GC. Descriptive statistics of the enrolled patients, 59 with gastric adenocarcinoma and 31 with gastritis, are summarized in Supplementary Table 1. All GC patients showed similar Mcl-1S/Mcl-1L mRNA levels with respect to age (p = 0.597), sex (p = 0.927), and pathological grades (p = 0.334), and lower ratio was positively correlated with TNM staging (p <0.001). This result showed that Mcl-1L and Mcl-1S expression in GC was independent of age, sex, and pathological grades, but was related to GC progression.

Next, to confirm the role of Mcl-1L and Mcl-1S in human GC, a series of histological and molecular analyses were performed. Quantitative reverse transcription-polymerase chain reaction (RT-qPCR) results showed that Mcl-1L mRNA expression was significantly higher (p <0.05) and Mcl-1S mRNA expression was lower (p <0.01) in tumor specimens than in normal gastric mucosal tissues. Consistent with this difference, the Mcl-1S/Mcl-1L mRNA ratio was significantly decreased (p <0.001) in GC (Figure 1A). Moreover, this decrease was accompanied with progression of T stages, instead of N stages (Figure 1B). This result indicated that Mcl-1L over-expression and/or Mcl-1S under-expression was closely related to tumor size increase but not to lymph node metastasis [31]. Only one patient in the M stage developed distant metastases and showed invalid statistics. Kaplan–Meier survival analysis showed that the overall survival of patients with lower Mcl-1S/Mcl-1L levels was decreased compared with that of patients with higher levels (Figure 1C). These results suggested that high Mcl-1L and low Mcl-1S expression levels in GC tissue were correlated with tumor proliferation and prognosis.

Subsequently, we validated Mcl-1L and Mcl-1S expression using differently differentiated GC cell lines, including MKN-28 (well-differentiated), SGC-7901 (moderately-differentiated), and MKN-45 (poorly-differentiated). As shown in Figure 1D, higher Mcl-1L and lower Mcl-1S mRNA expression levels were observed in all three GC cell lines than in human gastric epithelial cells (GES-1 cells). Additionally, compared with GES-1 cells, GC cell lines showed significantly reduced Mcl-1S/Mcl-1L ratio. However, there was no difference in the Mcl-1L and Mcl-1S mRNA levels among the GC cell lines, demonstrating that up-regulated Mcl-1L and down-regulated Mcl-1S expression patterns, similar to those in GC tissues, were observed in the GC cell lines. However, they were not correlated with GC cell differentiation.

Consistent with this difference at the mRNA level, the Mcl-1 protein expression pattern was characterized by markedly higher Mcl-1L levels, lower Mcl-1S levels, and a relatively similar Mcl-1S/Mcl-1L ratio in both GC tissues and cell lines, compared with those in normal tissues and cells (Figure 1E and 1F). These results further confirmed the up-regulated Mcl-1L and down-regulated Mcl-1S expression in GC and the reversed pattern in normal tissues and cells.

Taken together, these findings suggested prominent Mcl-1L and reduced Mcl-1S expression in GC tissues and cell lines. Moreover, the lower Mcl-1S/Mcl-1L ratio contributed to gastric tumor proliferation and poor prognosis.

Modification of Mcl-1 pre-mRNA alternative splicing using steric-blocking oligonucleotides (SBOs) in GC cell lines

Given the effects of the Mcl-1 isoforms on apoptosis, we speculated that blocking Mcl-1L expression and inducing Mcl-1S expression would cause the cancer cells to switch to a pro-apoptotic state and restrain tumor progression. Therefore, we transfected GC cell lines with fluorescein-tagged Mcl-1-specific SBOs to shift the Mcl-1 pre-mRNA splicing pattern from Mcl-1L to Mcl-1S. Fluorescein-tagged SBOs were delivered into the GC cell nuclei. They bound to complementary fragments with different transfection efficiencies depending on the SBO dosage (Supplemental Figure 2). After 48 h of transfection of 5 or 10 μM SBOs, Mcl-1L mRNA expression was down-regulated and Mcl-1S mRNA expression was up-regulated in a dose-dependent manner in the three GC cell lines (Figure 2A). Similar to the quantitative change in the mRNA levels, a decrease in the Mcl-1L protein expression level and an increase in the Mcl-1S protein expression level were observed in SBO-treated GC cell lines (Figure 2B). These results demonstrated that the Mcl-1-specific SBOs shifted the Mcl-1 splicing pattern from Mcl-1L to Mcl-1S efficiently, resulting in decreased anti-apoptotic Mcl-1L and increased pro-apoptotic Mcl-1S expression levels.

Effects of the shift in Mcl-1 pre-mRNA splicing from Mcl-1L to Mcl-1S on regulation of apoptosis in vitro

To validate whether the shift in the Mcl-1 splicing pattern from Mcl-1L to Mcl-1S enhanced GC cell apoptosis, the apoptotic cells were quantified by flow cytometry (FCM). Annexin V and 7-amino-actinomycin (7-AAD) were used to identify the early and late apoptotic cells, respectively. The results indicated that 48 h post-transfection of SBOs, the number of apoptotic cells increased in the three GC lines (Figure 3A). The apoptosis rates are summarized in Figure 3B, which shows that SBOs result in a significant dose-dependent increase in the number of both early and late apoptotic cells. Western blotting (WB) was performed to evaluate apoptosis by detecting the key markers of apoptosis (Bak, cleaved caspase 9, and cleaved caspase 3) (Figure 3C). The integrated density values indicated that the expression of apoptotic factors in SBO-treated GC cell lines showed statistically significant acceleration with a clear dose-response relationship (Figure 3D). Collectively, these data verified that modification of the Mcl-1 splicing pattern from Mcl-1L to Mcl-1S using SBOs strikingly stimulated GC cell apoptosis.

Modification of Mcl-1 pre-mRNA alternative splicing from Mcl-1L to Mcl-1S in vivo

The tumorigenicity assay in male severe combined immune-deficient mice was performed to investigate whether modification of Mcl-1 splicing via SBO treatment affected tumor cell apoptosis and growth in vivo. Mcl-1S mRNA was up-regulated while Mcl-1L mRNA was prominently down-regulated, depending on the SBO dosage in the mouse xenograft models of both MKN-45 and HGC-27 cells (Figure 4A). As expected, Mcl-1L and Mcl-1S protein levels were altered in parallel with their mRNA expressions (Figures 4B and 4C). Mcl-1S/Mcl-1L ratio was also significantly increased in the SBO-treated MKN-45 cells (Figure 4C). These results indicated that Mcl-1-specific SBOs succeeded in blocking Mcl-1L expression and inducing Mcl-1S expression in mice.

The role of altered Mcl-1 pre-mRNA splicing in promoting apoptosis and suppressing GC in vivo

To validate whether Mcl-1 pre-mRNA alternative splicing using SBOs could induce tumor cell apoptosis in vivo, apoptotic cells were detected in the xenograft models. In the mouse xenograft models of both MKN-45 and HGC-27 cells, hematoxylin and eosin (HE) staining revealed a dose-dependent increase in the dead cell area of the SBO-treated tumor tissue, compared with that of the control (Figure 5A). Immunofluorescence assay of the HGC-27 xenograft tumors demonstrated that SBO treatment resulted in a dose-dependent increase in the number of apoptotic cells (annexin V-positive) (Figure 5B); identical results were observed in the MKN-45 xenograft models (not presented in current study). Correspondingly, quantitation by FCM showed that treatment with 0, 6.25, and 12.5 mg/kg SBOs led to apoptosis rates of 2.76, 14.44, and 28.75% in the MKN-45 xenograft models and 2.49, 31.26, and 50.57% in the HGC-27 xenograft models, respectively. The results showed that the early and late apoptosis rates increased in a statistically significant dose-dependent manner (Figure 5C). Additionally, after Mcl-1-specific SBO treatment for 3 d, Bak, activated caspase 9, and caspase 3 expression levels in the tumor tissues were markedly elevated in a dose-dependent manner (Supplementary Figure 3). These results indicated that the shift in Mcl-1 splicing from Mcl-1L to Mcl-1S induced significant cell apoptosis in gastric xenograft tumors.

Next, we examined the effects of Mcl-1-specific SBO treatment, which caused a shift in Mcl-1 splicing, on tumors in vivo. The volumes of the SBO-treated tumors remained stable, while those of the Con-SBO-treated tumors experienced growth rates of over 30%. In the mouse HGC-27 xenograft models, the volumes of the tumors treated with 6.25 and 12.5 mg/kg SBOs reduced by 1.32% and 2.21%, respectively, while in the mouse MKN-45 xenograft models, the volumes increased by 2.85% and 0.85%, respectively (Figure 6A). Additionally, immunohistochemical detection of Ki-67, performed to evaluate tumor viability and proliferation, demonstrated a significant dose-dependent decrease in the Ki-67 expression rate in the SBO-treatment groups (Figure 6B). These results indicated that Mcl-1S over-expression and Mcl-1L under-expression, regulated by Mcl-1-specific SBO treatment, inhibited tumor proliferation in vivo in this model.

Gene set enrichment analysis (GSEA)

Mcl-1 mRNA data used for GSEA are accessible from TCGA database (https://tcga-data.nci.nih.gov/docs/publications/tcga/). The correlations among Mcl-1 mRNA level, tumor grade, and cancer stage were analyzed using the software Gene Expression Profiling Interactive Analysis (GEPIA) (http://gepia.cancer-pku.cn/). The median Mcl-1 expression level was used as the parameter to divide the high and low groups of clinical GC specimens. Statistical significance (false discovery rate) was set at 0.05.

Patients and clinical data

Fifty-nine gastric adenocarcinoma and thirty-one gastritis patients were recruited for this study after admission at the Gansu Provincial Hospital. The diagnosis, grading, and staging of gastric adenocarcinoma were established according to the NCCN Clinical Practice Guidelines in Oncology: Gastric Cancer (Version 2.2018). Kaplan–Meier survival curves for GC patients with low and high Mcl-1S/Mcl-1L mRNA ratios (grouped according to the median value of the Mcl-1S/Mcl-1L ratio) were generated by retrospective follow-up. The protocol of this study followed the ethical guidelines of the 1975 Declaration of Helsinki, and the study was approved by the Ethics Review Committee of Gansu Provincial Hospital. All participants provided their written informed consent to participate in this study.

Cell culture

Authenticated and differently differentiated GC cell lines (MKN-28, SGC-7901, and MKN-45) were obtained from Beijing Fenghui Biotechnology Co. Ltd. (China). The cells were incubated at 37°C in 5% CO₂ for 4–6 h and then immediately cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (9:1) and penicillin/streptomycin (100 U/mL) (Invitrogen Corporation, Carlsbad, CA, USA) at 37°C in 5% CO₂. Cells in the logarithmic growth stage were digested, seeded into 6-well plates (4×10⁵ cells/mL in each well), incubated at 37°C for 24 h, digested with pancreatin, and collected for further research.

Modification of Mcl-1 pre-mRNA alternative splicing

SBOs were used to cause down- and up-regulation of Mcl-1L and Mcl-1S expression, respectively. The SBOs were synthesized, and the Endo-Porter delivery system was purchased from Gene Tools (Philomath, OR, USA). A pair of SBOs was designed such that they could bind to the 3'-acceptor and 5'-donor-splicing site of exon 2 of Mcl-1 pre-mRNA and splice out exon 2, thereby shifting the splicing pattern from Mcl-1L to Mcl-1S. The SBO sequences were 5'-CGAAGCATGCCTGAGAAAGAAAAGC-3' and 5'-AAGGCAAACTTACCCAGCCTCTTTG-3'. The SBOs blocked the 5'-donor and 3'-acceptor sites of exon 2 to skip exon 2, thereby shifting the splicing pattern from Mcl-1L to Mcl-1S mRNA. A non-targeting oligonucleotide sequence was cloned as the control: 5'-CCTCTTACCTCAGTTACAATTTATA-3' (Con-SBO). The Endo-Porter delivery system was used to optimize the conditions, according to the manufacturer’s instructions. GC cell lines were treated with 5 and 10 μM Mcl-1-specific SBOs for a verified optimum period of 48 h [29]. Localization and transfection efficiency of fluorescein-tagged SBOs were defined by laser confocal microscopy. RT-qPCR and WB analyses were performed to confirm Mcl-1L and Mcl-1S expression. FCM was used for counting the apoptotic cells.

Xenotransplantation and tumor analysis

All animal experiments were performed according to the Gene Tools and ARRIVE guidelines [56]. The protocol of this study was approved by the Ethics Review Committee of Gansu University of Chinese Medicine. Briefly, MKN-45 or HGC-27 cells (1×10⁷ cells) in serum-free medium (100 μL) were implanted into the subcutaneous tissue of the antedorsal walls of immunodeficient mice (4–5-week-old males, one tumor per mouse, n = six mice per group). Mice were grouped after tumor formation and then injected with vivo-morpholino-modified Mcl-1-specific SBOs or a non-targeting vivo-morpholino-modified oligonucleotide (as the control) (Gene Tools, LLC, Pacific Grove, CA) by local multipoint administration. Dosing (1.25 or 6.25 mg/kg SBOs) was repeated daily for 3 days, and the mice were sacrificed on day 4. The tumor volumes were measured with calipers and calculated using the following equations: volume = length × (width)² × 0.5; change in tumor volume = (terminal volume – initial volume)/initial volume. Each tumor sample was snap-frozen in liquid nitrogen or fixed immediately in paraformaldehyde and then processed for further blind histological analysis.

RT-qPCR

Total RNA was extracted using TRIzol reagent, and cDNA was synthesized using the RNeasy mini kit (Qiagen, Germany). cDNA was used for RT-qPCR using Taqman universal PCR master mix (ABI, 4444556, USA). Mcl-1L and Mcl-1S mRNA was detected with Mcl-1 isoform-specific gene expression probes (ABI, Mcl-1L: Hs00172036_m1; Mcl-1S: Hs00766187_m1) and normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA (ABI, Hs99999905_m1). RT-qPCR reactions were performed at 95°C for 10 min with 40 cycles at 95°C for 15 s, 60°C for 30 s, and 70°C for 30 s. Mcl-1L and Mcl-1S expression was analyzed using the comparative threshold cycle (Ct) method of relative quantification.

WB

Proteins were extracted from tumor tissues and purified using total protein extraction kits (KeyGEN, China). Protein concentration was detected and normalized by the bicinchoninic acid method. Denatured samples were resolved on 12% (w/v) SDS-PAGE gels and transferred onto PVDF membranes (Millipore, USA) by the wet transfer-coating finishing method. The PVDF membrane was immersed in 5% skimmed milk in tris-buffered saline at room temperature for 1 h and then incubated overnight with rabbit anti-human Mcl-1 antibody (1:1,000) (Bio-Rad, AHP998, USA), which can recognize both the Mcl-1L (41 kDa) and Mcl-1S (35 kDa) proteins, at 4°C. Rabbit polyclonal antibody against GAPDH (Bio-Rad, USA) was loaded as the control; the secondary antibody was horseradish peroxidase-conjugated IgG (1:5,000) (Bio-Rad, USA). Densitometry analysis was performed using ImageJ software (Bio-Rad, USA).

Apoptosis detection by FCM

GC and transplanted tumor cell apoptosis was detected using the Annexin V-APC/7-AAD apoptosis detection kit (Multisciences Biotech, China). Cells (10,000 cells per sample) were re-suspended in a binding buffer, stained with annexin V-APC and 7-AAD for 15 min in the dark, and diluted with the binding buffer to obtain a final volume of 500 mL. Data acquisition and analysis were performed using a flow cytometer (FACS Caliber, BD, USA).

Immunofluorescence

The paraffin sections of the tumor samples were stained with rabbit anti-annexin V antibody, goat anti-rabbit IgG H&L (Alexa Fluor® 488) (Abcam, USA), and DAPI (BD, USA) to determine tumor tissue apoptosis. Images were captured using a confocal microscope (Leica Microsystems, SP8, Germany) on standard settings.

Immunohistochemistry

Paraffin tumor sections (6 μm) were HE stained to detect dead cells and immunostained with antibodies against Ki-67 (Protein Tech Group, China) to analyze tumor proliferation. The dead cell area and Ki-67 positive rate were analyzed using KF-PRO-020 (KFBIO, China).

Statistical analyses

Statistical analysis was performed using SPSS 18.0 software. Data of at least three independent experiments for each of the cellular and animal groups were presented as the mean ± standard deviation. The differences in the measurement data among the groups were analyzed by the homogeneity test of variances. The significance of differences among individual groups was analyzed by Student’s t-test (two-tailed). One-way ANOVA was performed for multiple group comparisons. Survival curves were plotted using the Kaplan–Meier method and compared using the log-rank test. We defined a high expression level as that above the median and a low expression level as that below the median. All statistical tests were two-sided, and P <0.05 indicated statistical significance.