Roles of eIF3m in the tumorigenesis of triple negative breast cancer

Background Without targets, triple negative breast cancer (TNBC) has the worst prognosis in all subtypes of breast cancer (BC). Recently, eukaryotic translation initiation factor 3 m (eIF3m) has been declared to be involved in the malignant progression of various neoplasms. The aim of this study is to explore biological functions of eIF3m in TNBC. Methods Multiple databases, including Oncomine, KM-plotter and so on, were performed to analyze prognosis and function of eIF3m in TNBC. After transfection of eIF3m-shRNA lentivirus, CCK-8, colony formation assay, cell cycle analysis, wound healing assay, transwell assays, mitochondrial membrane potential assay and cell apoptosis analysis were performed to explore the roles of eIF3m in TNBC cell bio-behaviors. In addition, western blotting was conducted to analyze the potential molecular mechanisms of eIF3m. Results In multiple databases, up-regulated eIF3m had lower overall survival, relapse-free survival and post progression survival in BC. EIF3m expression in TNBC was obviously higher than in non-TNBC or normal breast tissues. Its expression in TNBC was positively related to differentiation, lymph node invasion and distant metastasis. After knockdown of eIF3m, cell proliferation, migration, invasion and levels of mitochondrial membrane potential of MDA-MB-231 and MDA-MB-436 were all significantly suppressed, while apoptosis rates of them were obviously increased. In addition, eIF3m could regulate cell-cycle, epithelial–mesenchymal transition and apoptosis-related proteins. Combined with public databases and RT-qPCR, 14 genes were identified to be modulated by eIF3m in the development of TNBC. Conclusions eIF3m is an unfavorable indicator of TNBC, and plays a vital role in the process of TNBC tumorigenesis.

of carcinogenesis in order to improve the diagnosis and treatment of TNBC.
Protein synthesis has been considered as a critical step from gene expression to cell biological behavior [5]. Abnormal translations of oncogenes, tumor suppressors and eukaryotic translation initiation factors (EIFs) are essential for tumorigenesis [5]. Among eIFs, eukaryotic translation initiation factor 3 (eIF3) is the most complex eukaryotic translation initiation factor [6]. This complex composed of 13 subunits (eIF3a to m) is necessary for the initiation of protein synthesis [7]. Among these subunits, 5 core subunits (eIF3a, b, c, g and i) are conserved in all eukaryotes, while the rest only exist in mammal [8].
Recent studies have reported that several subunits are essential for the development of various tumors, including BC [9][10][11]. Notably, as the last defined subunit, eIF3m plays an indispensable role on maintaining the integrity of the eIF3 complex by stabilizing the core subunits [12]. Therefore, the ablation of eIF3m severely suppresses the translation and cancer cell proliferation [13]. In addition, eIF3m is a critical factor for cell cycle progression via regulating cell division cycle 25A (CDC25A) [13]. However, its function in BC or TNBC is not yet characterized.
In this research, eIF3 subunits were detected in breast tumors and corresponding adjacent normal breast tissues as well as public databases. After selection of suitable breast cancer cell lines with higher eIF3m expression, stable transfected cells with eIF3m knockdown were constructed for investigating the significance of eIF3m in vitro and exploring the potential pathogenesis of TNBC tumorigenesis. This article revealed possible roles of eIF3m in TNBC and provided a novel and reliable biomarker for clinic.

IHC
Paraffin-embedded consecutive sections were subjected to IHC staining for the expression of eIF3m by the rabbit polyclonal anti-eIF3m antibody (bs-9033R, Beijing BIOSS, China) diluted at 1:100 in phosphate-buffered saline (PBS, GIBCO Life Technologies, USA), with a SP Rabbit & Mouse HRP Kit (CWBIO, China). PBS without primary antibodies was used as negative control. Two pathologists independently evaluated the scores of the staining intensity of eIF3m expression (x = 0, no staining of cells; 1, mild staining; 2, moderate staining; and 3, marked staining). Any disagreement was resolved by discussion. Scores of the percentage of cells were measured by ImageJ 1.52r (y = 0%-100%). The total score (immunoreactivity score, IRS = x×y × 100) ranged from 0 to 300.

Cell transfection
To knock down eIF3m expression, pLKO.1-eIF3m-shRNA-puro was used, and three shRNAs were designed as listed in Table 1. The titer of lentivirus was 1x10 8 TU/ ml. MDA-MB-231 and MDA-MB-436 were cultured for 24 h in a 24-well plate, and then cells were transfected with shRNAs-Puro lentivirus (100 μl/ml) in a medium of 8 mg/ml polybrene (Sigma-Aldrich, USA) when a confluence of 30-50%. After 24 h, transfected cells were cultured in a new medium with 1.0 μg/ml puromycin (TargetMol, China) for about 12 days. To select the most efficient shRNA, the transfection efficiency of each shRNA was analyzed through western blotting and RT-qPCR. Cells transfected with this shRNA were named as "eIF3m-shRNA" groups, and cells transfected with empty vector were named as "Vector" groups.

RT-qPCR
Cells and tissues were collected to isolate total RNA through Trizol regent (Thermo Fisher Scientific, USA). A total of 2 µg RNA of each sample was reverse transcribed using the SuperScript II RNase-Reverse Transcriptase system (Thermo Fisher Scientific, USA). cDNA was subjected to quantitative PCR using primers specific for eIF3 subunits, co-expressed genes and GAPDH. PCR primers were designed as listed in Table 2 and Additional file 1: Table S1. The PCR cycling conditions were as follows: (1) 94 °C for 4 min; (2) 40 cycles of 95 °C for 1 min; (3) 60 °C for 1 min; and (4) 72 °C for 1 min. Amplifed DNA was measured by the SYBR Premix Ex Taq ™ kit (Takara Bio, Japan), and qPCR was performed using an iQ5 real-time PCR detection system (Bio-Rad, USA). Then, we used 2 −ΔΔCt value to calculate the relative expression. The formula of ΔΔCt could refer to the previous method [14].

Cell viability assay
To assess cell proliferation, 2000 stable transfected cells per well were seeded in 96-well plates for 1-5 days, and then were treated with CCK-8 (cell counting kit-8, Immunoway, USA) for 2 h at the incubator. The absorbance of each group was read at 450 nm of the microplate reader (OD value). Cell viability was calculated by the formula: Cell viability (%) = OD value of "eIF3m shRNA"/OD value of "Vector".

Colony formation assay
Cells were seeded in 6-well plates at a density of 200 cells. They were cultured for about 10 days, until the clones were visible (≥ 50 cells). The colonies were fixed for 15 min with methanol and then stained with crystal violet for 15 min after airing. Finally, numbers of colonies were counted.

Cell cycle analysis
Centrifugated cells were washed 2 times with precooling PBS. And then, they were fixed in 80% ethanol Table 1 The design of shRNA-eIF3m. SC shRNA: the control shRNA

Wound healing assay
Transfected cells were planted into 6-well plates to complete confluence in order to assess the role of eIF3m on the migration. Cells were wounded by scratching with a sterile plastic 200 µL micropipette tip and the loose cells were removed by PBS. Subsequently, Opti-MEM medium (GIBCO Life Technologies, USA) was added to the wells for incubation at 37 °C with 5% CO 2 for 24 h. Finally, images were captured by the inverted microscope (IX73, Olympus, Shanghai, China). The relative proportion of wound = the width of the wound/ the initial width.

Transwell assays
8 mm transwell inserts (JETbiofil, China) were placed into 24-well plates, separating upper and lower chambers. The upper side of the membrane was precoated with Matrigel (Corning, USA) for gel formation. New medium was added to the lower chamber and 10 5 cells/ well in Opti-MEM were added to the upper chamber. After 72 h at 37 °C, the number of invading cells was counted through a counting chamber under the inverted microscope.

Mitochondrial membrane potential assay
This assay could be divided into four parts:

Cell apoptosis analysis
Centrifugated cells were washed 2 times with PBS and stained with PE and 7-AAD for 15 min at room temperature in dark (KeyGen Biotech, China). After diluted in PBS, stained cells were analyzed with the flow cytometer.

Western blot analysis
Proteins of tissues or cells were extracted by RIPA extraction buffer with PMSF (Beyotime Biotechnology, China). 20 µL per sample were loaded in pre-cast gels

Statistical analysis
All experiments were done in triplicates. Continuous variables were expressed as the‾x ± SD and experiments were performed in triplicate. All results were analyzed by t test through SPSS 20.0 software with a p < 0.05 considered statistically significant. The differences of eIF3 subuntis expression between breast tumors and matched normal tissues were analyzed by a paired-t test, while the differences of biological behaviors between "Vector" group and "eIF3m-shRNA" group were analyzed by an unpaired-t test. An ANOVA followed by Tukey's multiple comparisons test was used to compare eIF3m expression in cell lines. Kaplan-Meier analysis and the log rank test were also conducted to analyze survival data of public databases. All graphs were generated by GraphPad Prism 6.0.

Abnormal up-regulation of eIF3m had worse prognosis in BC in multiple databases
Oncomine analysis of tumor versus. normal samples revealed that eIF3m was significantly elevated in invasive breast carcinoma, such as invasive ductal breast carcinoma, mixed lobular and ductal breast carcinoma, invasive lobular breast carcinoma, and so on (Table 3). In addition, there were no substantial differences of eIF3m between normal breast tissues and benign breast neoplasm (p > 0.05, Table 3). KM-plotter and BCIP database revealed that high eIF3m was correlated with a poorer prognosis, including OS, RFS and PPS in BC (Fig. 1).

eIF3m became a unique and valuable indicator for TNBC
To analyze differences of eIF3m expression among different types of breast cancer, one GEO database (GEO accession: GSE45827) showing all subunits expression in TNBC, non-TNBC and normal-like tissues was performed (Fig. 2a). In Table 4, expression levels of eIF3b, d, f, g, i, j, k, l and m in breast tumors were significantly higher than in normal-like tissues (p < 0.05). Among these subunits, there were no differences of eIF3b and k expression between TNBC and normal-like tissues, or between TNBC and non-TNBC (p > 0.05). Only eIF3m expression in TNBC was obviously higher than in non-TNBC tissues (p = 0.000253). In addition, similar results were found in 158 BC cases via RT-qPCR ( Fig. 2b-d). These results reflected that eIF3 was a vital protein molecule in BC, and eIF3m was involved in the occurrence and development of TNBC.
Expression of eIF3m was detected in the cytoplasm through IHC (Fig. 3). There were no differences of eIF3m expression between tumors and corresponding adjacent normal breast tissues in non-TNBC. But, its expression levels in TNBC were significantly higher compared with adjacent normal tissues (p < 0.001). In addition, expression of eIF3m in TNBC was also higher than non-TNBC (p < 0.001).
Poor differentiation group of TNBC had a higher level of eIF3m in comparison of well-moderate differentiation group (p < 0.01, Fig. 3g). Cases with lymph node metastasis positive or distant metastasis positive group also had a higher level of eIF3m compared with metastasis negative group (p < 0.05, p < 0.01, Fig. 3g).

eIF3m expression in breast cell and breast cancer cell lines
Among five BC cell lines and MCF-10A, MDA-MB-231 and MDA-MB-436, two TNBC cell lines, showed highest expression of eIF3m (Fig. 4a, b). Therefore, these two cell lines were chosen to be transfected with lentivirus. The transfection efficiency was accessed by western blotting (Fig. 4c) and RT-qPCR (Fig. 4d). Among shRNAs, shRNA2# had the lowest eIF3m expression. Thus, we used shRNA2# for the next assays.

Knocking down eIF3m inhibited cell proliferation in TNBC cells
As shown in Fig. 5a, reduced expression of eIF3m obviously suppressed cell viability (p < 0.05). The number of colonies formed on 6-well plates in "eIF3m-shRNA" groups was significantly fewer than in "Vector" groups (p < 0.01, Fig. 5b). In addition, cells transfected with eIF3m-shRNA obtained a higher proportion at G1 phase ( Fig. 5c). At the same time, expression of CyclinD1, CyclinE, CDK2, CDK4 and CDK6 decreased obviously in "eIF3m-shRNA" groups in comparison of "Vector" groups; but P21 and P27 in "eIF3m-shRNA" groups were up-regulated (Fig. 5d). These results confirmed that knockdown of eIF3m suppressed proliferation of TNBC cells.

Knocking down eIF3m suppressed migration and invasion in TNBC cells
By knocking down eIF3m, the width of scratch area after 24 h was obviously longer compared with "Vector", (Fig. 6a). Also, the transwell assay showed fewer cells going through chambers when decreasing expression of eIF3m in TNBC (Fig. 6b). Knockdown of eIF3m debased the levels of N-cadherin, Vimentin, Snail and Twist, while it elevated E-cadherin expression (Fig. 6c). Thus, low expression of eIF3m made a sharp decline in the abilities of migration and invasion.

Knocking down eIF3m promoted apoptosis in TNBC cells
The brightness of JC-1 red in MDA-MB-231 and MDA-MB-436 was reduced by using eIF3m-shRNA, while JC-1 green's brightness was increased obviously (Fig. 7a).The ratios of red/green fluorescence were also significantly decreased (Fig. 7b, c). Through extraction of mitochondrial proteins and cytosol, knockdown of eIF3m down-regulated Cyto-C obviously in mitochondria, but elevated its expression in cytosol (Fig. 7d). In addition, rates of cell apoptosis in "eIF3m-shRNA" groups were increased along with the regulation of Bcl-2 family and the activation of Caspase-3 and Caspase-9 (Fig. 7e, f ). These results revealed that knockdown of eIF3m elevated apoptosis rates through releasing Cyto-C from mitochondria into cytosol and activating Caspases.

Functions of eIF3m on the development of breast and TNBC analyzed by databases
PDBsum entry showed the structure of eIF3, and eIF3m was a key subunit maintaining the integrity of the eIF3 complex by stabilizing other subunits (Fig. 8a) [15,16]. KEGG pathways displayed the vital role of eIF3 on RNA transport in which eIF3 and other eukaryotic translation initiation factors must be combined with the ribosome and mRNAs in order to complete the translation (Fig. 8b). In addition, eIF3m was involved in the development of mammary epithelium and gland (Fig. 8c).

Discussion
It is well known that breast tumor occurs in the process of continuous and cumulative improper regulation of gene expression. This abnormal regulation is caused by various exterior and internal carcinogenic factors. In other words, molecular mechanisms of the BC tumorigenesis are not thoroughly understood. EIFs are involved in translation and protein synthesis in order to meet the demand of successful embryonic development and maintenance of normal tissue homeostasis [37]. Among them, several proteins or subunits of EIFs are essential for tumorigenesis [5]. Recently, oncologists have focused on the biggest eukaryotic translation initiation factor, eIF3, which is constituted by 13 subunits. This molecule participates in the 43S preinitiation complex formation and facilitates recruiting GTP-eIF2-tRNA-methionine ternary complex [38]. The complex stimulates mRNA binding with 43S pre-initiation complex, promotes the process of translation and influences cell biological behaviors and functions [39]. Not only each subunit plays an important role in the synthesis and metabolic activities, but also several ones are implicated in carcinogenesis. As the largest subunit, eIF3a has vital biological functions in various carcinomas, including BC [40]. Other core subunits, eIF3c, g and i are also associated with BC tumorigenesis, especially in drug resistance [11,41,42]. In addition, several "noncore" subunits (eIF3b, d, e and f ) correlated strongly with the tumorigenesis of breast neoplasm [43][44][45][46].
Based on public databases, a unique advantage of eIF3m could be found in TNBC. Among 13 subunits, only eIF3m met the three points: 1. a higher level in breast tumor than in normal tissue; 2. a higher level in TNBC than normal-like tissues; and 3. a higher level in TNBC than in non-TNBC. These results revealed that eIF3m might become a critical molecule in the development and progression of BC, especially in TNBC. One previous study demonstrated that eIF3m was required for cell-cycle and cell proliferation [14]. Silencing eIF3m almost completely turned a caspase substrate, PARP1, into activated form, which led to promoting cell death of HCT-116 [13]. In this research, we verified functions of eIF3m on cell-cycle and cell apoptosis of TNBC. Through regulation of CyclinD1 and other cell-cycle proteins, knockdown of eIF3m arrested cell-cycle at G1 phase and inhibited cell proliferation. Decreased eIF3m also caused a sharp decline of mitochondrial membrane potential, followed by release of Cyto-C from mitochondria to cytosol. As a result, instable mitochondrial membrane activated Caspases and then induced cell apoptosis. In addition, knocking down eIF3m suppressed migration and invasion of TNBC through regulation of epithelial-mesenchymal transition (EMT) related proteins. EMT is the process of epithelial cells transformed into mesenchymal cells. It elevates motility, migration and invasion of cancer cells, leading to metastasis of tumors, especially in TNBC [47]. Thus, we argue that overexpressed eIF3m promotes proliferation and activate EMT in TNBC.
Results of IHC uncovered that high eIF3m was involved in poor differentiation, lymph node metastasis and distant metastasis of TNBC. In addition, high expression of eIF3m had a poorer prognosis of BC patients both in two public databases, KM-plotter and BCIP database. Due to no enough survival data of this cohort, further follow-ups and researches will be performed to analyze the prognostic role of eIF3m in TNBC patients. Thus, we demonstrate that eIF3m is a novel and reliable bio-marker eIF3m maintained the integrity of the eIF3 complex in order to complete the process of translation which was essential for the development of normal breast tissues and tumors. Compared with other subtypes and normal tissues, TNBC had the highest expression level of eIF3m. This might be caused by complex genetic imbalance involving activation of oncogenes and inhibition of tumor suppressors. This imbalance was influenced by various unfavorable factors from daily life, society and environment [48]. When eIF3m was up-regulated, more oncogenes were expressed and vital signaling pathways related to tumor growth and invasion were activated. As Fig. 6 Influence of eIF3m knockdown on migration and invasion in TNBC. a Wound healing assay measured relative proportion rates at 24 h for comparing migration abilities of different cells (× 40); b Transwell assay detected the number of cells passing through chambers in each group by image analysis (× 100); c Epithelial-mesenchymal transition related proteins of transfected cells were investigated via western blotting. *p < 0.05, **p < 0.01, ***p < 0.001 a result, eIF3m expression was elevated again, and then improve oncogenes and pathways one more time. This complex cycle accelerated the carcinogenesis of TNBC. As for specific reasons or factors that led to eIF3m upregulation, we still need to perform further investigations and explain potential molecular mechanisms in the near future. Due to no animal experiment temporally as a result of the limitation of our funding, the next step will include in vivo assays to identify its molecular mechanisms and roles in regulating expression of oncogenes.
In addition, our research discovered a discrepancy that mRNA expression of different oncogenes was not showing the similar kind of response in these two TNBC cells. This difference might be as a result of the specificities of different cell lines or experimental errors. Therefore, further researches should choose the oncogenes with the similar response to investigate their interaction with eIF3m. Fig. 7 Influence of eIF3m knockdown on mitochondrial membrane potential and cell apoptosis in TNBC. a JC-1 green and JC-1 red of cells were observed in the fluorescence microscope (× 400); b JC-1 green and JC-1 red distributions were detected by flow cytometric analysis; c ratios of red/ green fluorescence in mitochondria were measured through the microplate reader; d Blots and relative protein expression of Cyto-c expression in mitochondria and cytosol of stably transfected cells; e Cell apoptotic rates were detected by flow cytometry; f Apoptosis-related proteins of transfected cells were investigated via western blotting, Cle-Cas3: Cleaved-Caspase-3; Cas3: Caspase-3; Cle-Cas9: Cleaved-Caspase-9; Cas9: Caspase-9. *p < 0.05, **p < 0.01, ***p < 0.001 Conclusion eIF3m was an indispensable subunit of eIF3 and played a vital role in the process of TNBC tumorigenesis. High expression of eIF3m predicted poor prognosis in BC patients. Further researches should be conducted to reveal its possible molecular mechanisms and explore eIF3m-related oncogenes, in order to provide clues for confirming potential therapeutic targets in TNBC. Schematic model of the role of eIF3 on RNA transport analyzed by KEGG pathways; c BCIP database showed co-expressed genes of eIF3m in the development of breast and TNBC; d RT-qPCR investigated co-expressed genes in different groups. *p < 0.05, **p < 0.01, ***p < 0.001  Fig. 9 Schematic model of the carcinogenic mechanism of eIF3m in TNBC