Translation initiation is a pivotal step in protein synthesis and a crucial cellular process in regulating cell proliferation under normal physiological conditions (Hershey, 2010). The complex, multi-level process of translation involved >10 eukaryotic translation initiation factors (eIFs) composed of at least 30 subunits (Malina et al., 2011). The initiation step is predominantly targeted by oncogenic signal transduction pathways despite the strict protein translation regulation, causing cell transformation (Bilanges and Stokoe, 2007).

The translation process is frequently deregulated during tumor development and progression wherein specific mRNA translation involved in tumor cell transformation, survival, angiogenesis, invasion, and metastasis are aberrantly activated and/or inactivated. In particular, eIF2Bε, eIF3, eIF4E, and eIF4G-1 function as putative oncogenes, whereas eIF4G-2 is considered a tumor suppressor.

The eIF2 complex, consisting of eIF2α, eIF2β, and eIF2γ, is a key component in turning on general protein synthesis among the translation initiation machinery. The first evidence associating eIF2 to oncogenesis came from the observation that eIF2α phosphorylation abrogation through its serine-51 residue mutation promoted the malignant transformation of NIH3T3 cells (Donze et al., 1995). Subsequent studies have revealed that several anticancer therapies reduce tumor cell growth by inducing eIF2α phosphorylation or blocking eIF2α dephosphorylation.

Very little is known about the role of eIF2β in carcinogenesis, whereas the roles of various eIFs in cancer have been widely reported. The eIF2β, alternatively known as eIF2S2, has been up-regulated in human colorectal cancer. A recent study indicated that eIF2S2 deletion attenuated the development of testicular germ cell carcinomas (Heaney et al., 2009). Gene ontology analysis of human papillomavirus (HPV)-infected human cervical cancer SiHa cells revealed the elevated eIF2S2 mRNA expression.

Reportedly, deregulation of translation initiation contributes to carcinogenesis, partly through various tumor suppressor gene function inactivation. In particular, the functions of p53 and programmed cell death-4 (Pdcd4) tumor suppressor proteins are compromised due to deregulated translational control (Suzuki et al., 2008). The mothers against decapentaplegic homolog-4 (SMAD4) is a tumor suppressor protein considered a signaling molecule in the transforming growth factor-β (TGFβ)-induced anti-proliferative cell signaling pathway. Genes that encode various tumor suppressor proteins, such as p15, p27, and plasminogen activator inhibitor (PAI)-1, are transcriptionally activated by SMAD4 that binds to the promoter regions of aforementioned genes (Chiao et al., 1999). SMAD4 expression is either reduced or completely lost in various cancers. Several SMAD4 inactivation mechanisms in cancer have been reported, including homozygous deletion, gene mutation, promoter hypermethylation, and post-translation modification of SMAD4 (Miyaki and Kuroki, 2003; Wang et al., 2007). To date, several SMAD4 binding protein identifications, which determine the cellular status and function of SMAD4, have simplified the understanding of SMAD4 function as a tumor suppressor, as well as the mechanisms of SMAD4 turnover in cancer. We performed a protein microarray and revealed eIF2S2 as a new SMAD4-binding protein to identify novel SMAD4-interacting proteins (Rajasekaran et al., 2021). Therefore, this study aims to investigate the role of eIF2S2 in regulating SMAD4 function and its consequences in cervical carcinogenesis.

Our results reveal overexpressed eIF2S2 in human cervical cancer, and eIF2S2 stable knockdown inhibits cervical cancer cell proliferation and migration. Additionally, stable eIF2S2 overexpression promotes NIH3T3 cell transformation. Further, we demonstrate that eIF2S2 interacts with SMAD4 in cervical cancer cells and decreases SMAD4 expression, thereby reducing antiproliferative proteins p15 and p27 expression. Thus, aberrant eIF2S2 induction may promote cervical carcinogenesis through SMAD4 signaling inactivation.

Tissue specimens preparation

Specimens of cervical intraepithelial neoplasia (CIN)-I, CIN-II, CIN-III, invasive cervical carcinomas, and carcinoma-associated adjacent cervical epithelium along with normal cervical epithelial tissue were collected from Seoul National University Hospital (Seoul, Korea).

Immunohistochemistry (IHC)

Tissue sections of 5-μm thick were attached to glass slides for IHC analysis and heated overnight at 60°C to ensure proper adherence. The sections were deparaffinized with xylene, hydrated in serial dilutions of alcohol, and then immersed in 3% hydrogen peroxide solution to neutralize endogenous peroxidase activity. Sections were then microwaved in 10-mM citrate buffer for antigen retrieval. Slides were incubated with polyclonal antibodies against eIF2S2 (Proteintech, IL, USA) at 1:50 dilution for 1 h at 25°C. After washing, the slides were incubated for 30 min at room temperature with anti-rabbit polymer (Envision K4003, DAKO, Glostrup, Denmark) as the secondary antibody. The slides were washed and the chromogen was developed for 5 min with liquid 3,3‘-diaminobenzidine (Envision K4003, DAKO) after 3 rinses with distilled water with 0.1% Tween 20. Experienced pathologists specializing in gynecologic pathology blindly reviewed the slides and assessed IHC data. Immunostaining in tumor and control epithelial cells was estimated by light microscopy and graded as 0, 1+, 2+, and 3+ based on the staining intensity.

Chemicals and reagents

Daeil Lab Service (Seoul, Korea) provided water-soluble tetrazolium salts (WST, as a component of the EZ-Cytox kit). Proteintech (RSM, USA) supplied purified TGFβ. Thermo scientific (MA, USA) provided Dulbecco’s Modified Eagle Medium (DMEM), Roswell Park Memorial Institute (RPMI)-1640, and fetal bovine serum (FBS). Santa Cruz Biotechnology (CA, USA) supplied primary antibodies for SMAD4 (B8 clone, sc-7966), SMAD7, p27, and MMP9. Abcam (Cambridge, UK) provided the primary antibody for p15. Proteintech (IL, USA) provided the eIF2S2 antibody. Invitrogen (NY, USA) and Thermo scientific supplied horseradish peroxidase (HRP)-conjugated anti-mouse and anti-rabbit antibodies. Sigma Chemical Co. (MO, USA) provided flag-HRP conjugated as well as flag-M2 antibodies.

Plasmid constructions

Human eIF2S2 cDNA, encoding the open reading frame region, was polymerase chain reaction (PCR)-amplified from the cDNA of peripheral blood mononuclear cells (PBMC) using the forward primer TTAAGCTTTCTGGGGACGAGATGAT and reverse primer TTGGATCCTTAGTTAGCTT TGGCAC. A 1012-bp PCR product was cloned into the BamHI and KpnI restriction sites of the mammalian expression vector p3xFlag-CMV-10 (Invitrogen, CA, USA). Further, the sequences of the constructs were verified by sequencing. Sequences encoding full-length eIF2S2, eIF2S2-Nt (amino acids [aa] 1-141), and eIF2S2-Ct (aa 141-333) were fused to pFlag-VN173 vector for BiFC experiments. Sequences encoding full-length SMAD4, SMAD4-MH1 (aa 18-142), SMAD4-MH2 (aa 323-552), SMAD4-linker (aa 143-322), SMAD4-MH1-linker (aa 1-309), and SMAD4-MH2-linker (aa 144-552) as well as MH2 domain (aa 274-467) or MH1 domain (aa 9-176) of SMAD2, were fused to pHA-VC155 vector. Dr. Chang-Deng Hu (Purdue University, West Lafayette, IN, USA) kindly gifted both pFlag-VN173 and pHA-VC155 vectors. Full-length HA-VC155-SMAD7 (aa 1-426), HA-VC155-SMAD7-MH1 (aa 64-207), and HA-VC155-SMAD7-MH2 (aa 260-426) were received from Enzynomics (Daejeon, Korea). Genolution Pharmaceuticals (Seoul, Korea) provided control shRNA and eIF2S2-shRNA for stable transfection. Dr. Joan Massague (Howard Hughes Medical Institute, Memorial Sloan-Kettering Cancer Center, NY, USA) kindly gifted the p15^INK4b promoter reporter construct (pGL2-p15 ^INK4b promoter -751/+70).

Cell culture and transfection

The Korean Cell Line Bank (KCLB, Seoul, Korea) supplied HPV18-positive human cervical cancer cell line HeLa. American Type Culture Collection (Manassas, VA, USA) provided the HPV16-positive human cervical cancer cells (SiHa and CaSki), non-HPV infected cervical cancer cells C33A, mouse epidermal JB6 cells, and mouse embryonic fibroblasts NIH3T3. Dr. Michael Birrer (National Cancer Institute, Bethesda, MD, USA) gave human immortalized ovarian surface epithelial cells IOSE80-NIH. Cells were grown at 37°C in a humidified incubator under 5% CO₂ in DMEM and RPMI containing 10% FBS and combined antibiotics (Thermo scientific). Cells were routinely examined for mycoplasma contamination. SiHa and HeLa cells were transfected with a p3xFlag-CMV-10-eIF2S2 plasmid (2.5 μg) in each well of a 6-well plate using 2.5 μl of lipofectamine LTX (Invitrogen). HeLa cells were co-transfected with the expression vectors indicated in each experiment (1 µg each) using Polyexpress (Excellgen, Gaithersburg, MD, USA) for BiFC analysis. NIH3T3 cells were transfected with a control vector or p3xFlag-CMV-10-eIF2S2 using lipofectamine LTX (Invitrogen) for stable transfection of eIF2S2, following the manufacturer’s protocol. G418 (Invitrogen) and NIH3T3-vector and NIH3T3-eIF2S2 clones were isolated and grown in RPMI enriched with 10% FBS to confirm stable transfection. Additionally, SiHa cells were transfected with 4 μg of control shRNA or eIF2S2-shRNA using lipofectamine LTX (Invitrogen), following the manufacturer’s protocol. Stable transfection was confirmed using zeocin antibiotic (Invitrogen), and control shRNA or eIF2S2-shRNA clones were isolated and grown in DMEM enriched with 10% bovine serum.

Western blot analysis

Cells were harvested and lysed with RIPA buffer (150 mM of NaCl, 10 mM of Tris [pH: 7.2], 0.1% sodium dodecyl sulfate [SDS], 1% Triton X-100, 1% deoxycholate, and 5 mM of ethylenediaminetetraacetic acid [EDTA]) enriched with a complete protease inhibitor cocktail tablet (Roche Diagnostics, Mannheim, Germany), and then incubated on ice for 30 min with regular vortexing before centrifuging at 14,000 rpm at 4°C for 15 min. Protein concentration was determined using a bicinchoninic acid (BCA) protein assay kit (Pierce Biotechnology, Rockford, IL, USA). The protein samples were boiled in 1X SDS sample buffer for 5 min for complete denaturation and were resolved on 10%-15% SDS-polyacrylamide gel, following the protocol described earlier (Jung et al., 2011). After electrophoresis, proteins were transferred onto a polyvinyl difluoride membrane, blocked with 5% nonfat dry milk in 1X TBST (Tris-buffered saline with 0.1% Tween-20) and incubated with primary antibody at the appropriate final concentration followed by hybridization with horseradish peroxidase-conjugated anti-rabbit or anti-mouse secondary antibodies (1:5000). The membrane was washed with 1X TBST three times for 10 min for each step. Finally, Western blot images were developed on photographic film using enhanced chemiluminescence reagents.

Quantitative real-time reverse transcription–PCR (qRT-PCR)

Total RNA was extracted from cells transfected using trizol (Invitrogen, NY, USA) and reversed transcribed to cDNA using the Superscript^TM II First-Strand Synthesis System (Invitrogen), following the manufacturer’s protocol. After cDNA synthesis, qRT-PCR was conducted as previously described (Kwon et al., 2009) in a dual system LightCycler (Roche Applied Science, Mahhheim, Germany) using the primers. Supplementary Table 1 lists Universal Probe Library (UPL, Roche Applied Science) probe sequences, with the HPRT Taqman probe (TIB MOLBIOL, Berlin, Germany) used as a “reference gene” to normalize gene expression.

Bimolecular fluorescence complementation (BiFC) analysis

The BiFC assay is based on an observation that interaction between proteins fused to the fragments facilitated the association of fluorescent protein fragments. It enables visualization of interacted proteins at high spatial resolution without requiring other exogenous agents (Tom.K. Kerpolla). HeLa cells were co-transfected with the expression vectors indicated in each experiment (1 µg each) using Polyexpress (Excellgen) in each well of the 6-well plate. The fluorescence emissions were observed 24-30 h after transfection using an Olympus IX71 fluorescence microscope (OLYMPUS, Tokyo, Japan) with a TH4-200 camera.

Immunoprecipitation (IP)

Cells were harvested in IP lysis buffer containing 20 mM of Tris (pH: 7.5), 15% glycerol, 1% TritonX-100, 1 mM of EDTA, 8 mM of MgSO4, 150 mM of NaCl, 10 mM of sodium fluoride, 10 mM of glycerophosphate, 0.5 mM of sodium orthovanadate, 0.1 mM of phenylmethylsulfonyl fluoride, and a mixture of protease inhibitors. Cell lysates (2000 μg) were clarified by centrifugation before incubation overnight at 4°C in the presence of SMAD4 monoclonal antibody (2 μg) in the same aforementioned lysis buffer. Protein G agarose (Invitrogen, 50% slurry) was washed and 50 μg of bead slurry was then added and the incubation continued for 2 h 20 min at 4°C. Immunoprecipitates were recovered by centrifugation, washed thrice with IP wash buffer (20 mM of Tris [pH: 7.5], 15% glycerol, 1% Triton X-100, 1 mM of EDTA, 8 mM of MgSO4, 150 mM of NaCl) and resolved by Western blotting using specific antibody. Rabbit true blot (eBioScience, MA, USA) was utilized as a secondary antibody in the case of immunoblotting with eIF2S2.

Soft agar colony formation assay

NIH3T3 (vector- or eIF2S2-transfected) and SiHa (control-shRNA- or eIF2S2-shRNA-transfected) cells were collected by trypsinization, suspended in a culture medium, and counted. The agar mix (0.33% BME and 10% FBS in culture medium) was placed in a 6-well plate as the bottom agar. After bottom agar solidification, 1 ml of cells suspended in agar solution (1.2 ml of cell suspension in culture medium+0.33% agar mix) was poured over the bottom layer. NIH3T3 cells (2.4×10⁴/ml) stably transfected with vector or eIF2S2 were treated with TPA (10 ng/ml) before adding to the bottom agar layer. SiHa cells (5×10⁴/ml) stably transfected with control or eIF2S2 shRNA was poured onto the bottom layer. The cell colonies were scored with a microscope equipped with the Image-Pro PLUS computer software program (Media Cybernetics, Silver Spring, MD, USA).

Cell proliferation assay

The WST method (EZ-Cytox kit; Daeil Lab Service) was used to measure the effect of eIF2S2 siRNA on cell proliferation. SiHa and HeLa cells (1×10⁵) transfected with either control si-RNA or eIF2S2 si-RNA (10 or 50 nM) were grown in triplicate in 6-well plates for 24, 48, and 72 h at 37°C. EZ-Cytox solution (200 μl) was added to each well and incubated for 80 min. The number of viable cells was measured in a 96-well plate at a 492-nm optical density on a Sunrise reader (Tecan Trading AG, Männedorf, Switzerland). Cell viability was the percentage of control siRNA-transfected cells. SiHa or HeLa cells (1×10⁵ cells/well) transfected with control si-RNA or eIF2S2 si-RNA (10 or 50 nM) were plated in 6-well plates for the trypan blue dye exclusion assay. Cells were harvested by trypsinization after 48 h, resuspended in serum-free media, and counted after treatment with 0.04% trypan blue.

siRNA transfection

Shamchully Pharm, Co, LTD (Seoul, Korea) supplied four siRNA against eIF2S2, including si-eIF2S2-01, si-eIF2S2-02, si-eIF2S2-03, and si-eIF2S2-04, with sequences of GGU AAU AAC CAA CUU GUA ATT, GGG UAC AAG UGG UUC UAU ATT, GAC AUU AUG CUU GGC AAU ATT, and CCA ACC AUC UCC UUG CAU UTT, respectively. SiHa cells were transfected with all four siRNA against eIF2S2 and further experiments were conducted using eIF2S2-02-siRNA. Dharmacon (IL, USA) provided the control siRNA (non-specific control siRNA VIII, cat. No-D-001206-08-20). Cells (5×10⁴) were seeded into 6-well plates for transfection with siRNA. Cells were transfected with siRNA after 24 h using oligofectamine (Invitrogen, Carlsbad, CA, USA), following the manufacturer’s protocol with minor modifications. The final siRNA concentration in the cell culture medium was 10 or 50 nM. Western blot analysis with an eIF2S2-specific antibody and qRT-PCR revealed that eIF2S2-02 siRNA caused the most significant eIF2S2 inhibition, as illustrated in Supplementary Fig. 1.

Cell migration assay

Cells (5×10⁴), transfected with eIF2S2 si-RNA (10 nM) or control si-RNA, were subjected to Millipore’s (MA, USA) 24-well CHEMICON QCM™ cell migration assay. Cells were starved in DMEM containing 0.5% FBS for an additional 24 h after incubation at 37°C for 24 h. Cell number was detected with a GENios Pro microplate reader (Tecan Trading AG) using a 485/535 nm filter set (Gildea et al., 2000). All migration and invasion assays were performed in triplicate in at least three independent experiments. Values are presented as percentages of control.

Wound healing assay

In vitro wound healing assay was conducted to investigate the migration of SiHa and Hela cells transfected with either control si-RNA or eIF2S2 si-RNA. Transfected cells were grown on 6-well plates until they reached the desired confluence, wounds were then prepared by a single scratch on the monolayer using a yellow pipette tip, and wounded layers were washed with PBS to remove cell debris. We measured the closure or filling of the wounds at 0, 12, or 24 h with an Olympus IX71 fluorescence microscope with a TH4-200 camera. All experiments were conducted in triplicate.

Luciferase reporter gene assay

Cells were seeded into 12-well plates at a density of 1×10⁵ cells per well before transfection. Cells were transfected with p15 promoter construct (a kind gift from Dr. Joan Massague, Howard Hughes Medical Institute, Memorial Sloan-Kettering Cancer Center, NY, USA) with an empty vector or 3xFlag-CMV-10-eIF2S2 with genefectin transfection reagent. A normalization control involved pRL-TK (Promega, Madison, WI, USA). Culture cells were starved for 1 h after a further 24 h, and the luciferase activity was measured with the Dual-Luciferase^® Reporter Assay System (Promega) in the presence or absence of TGFβ (5 ng/ml).

Tumor xenograft assay in nude mice

The Institutional Animal Care and Use Committee (IACUC) of Seoul National University approved the animal experiment (SNU-110210-5). Female BALB/c (nu/nu) mice (5-7 weeks old, females, n=3 or 6 per group) were caged in polycarbonate chambers and housed in a pathogen-free isolation facility with a light/dark cycle of 12/12 h. Animals were fed with rodent chow and water ad libitum. The Seoul National University Ethics Research Board approved all animal studies. The animals were subcutaneously injected on the right hind flank with eIF2S2 shRNA and control shRNA stably transfected SiHa cells (2.2×10⁷ cells in 100 μl PBS). Tumor volume was identified by measuring the length×width×Depth×0.52 of the tumors using a caliper. Animals were sacrificed at the end of the experiment, and tumor xenografts were excised for further analysis.

Statistical analysis

The significance of differences between groups was determined using a two-way ANOVA. GraphPad Prism 7 (GraphPad Software) was utilized for data analysis, and p<0.05 was considered statistically significant.

Stage-dependent eIF2S2 overexpression in human cervical cancer

We examined eIF2S2 expression in different stages of human cervical cancer to investigate the eIF2S2 expression during cervical cancer development. Immunohistochemical staining was conducted on the cervical cancer tissue section (n=10), including specimens of CIN-I, CIN-II, CIN-III, invasive cervical carcinomas, and carcinoma-associated adjacent cervical epithelium, along with normal cervical epithelial tissue. A gradual increase in eIF2S2 expression was found in CIN-I, CIN-II, CIN-III, and carcinomas. In contrast, eIF2S2 was undetectable and weakly expressed in 70% and 30% of normal cervical epithelium, respectively (Fig. 1, Table 1). Strong immunoreactivity (scored as 3+), primarily in the cytoplasm, was detected in 10%, 20%, and 70% of CIN-II, CIN-III, and invasive carcinomas, respectively. The eIF2S2’s expression pattern demonstrated an increasing tendency in ovarian (SKOV3), breast (MDA-MB-231), and cervical cancer cell lines (SiHa and HeLa), compared to normal cell lines, such as mouse epidermal (JB6) cells, human ovarian epithelial (IOSE80-NIH) cells, and mouse embryonic fibroblasts (NIH3T3) (Supplementary Fig. 2). Our results indicate a potential correlation between eIF2S2 expression and cancer development.

Table 1 The expression of eIF2S2 in normal cervical squamous epithelium, low-grade CINs, high-grade CINs and cervical carcinomas

Cervical specimen	Degree of immunoreactivity
	0	1+	2+	3+
Normal	70 (7/10)	30 (3/10)	0 (0/10)	0 (0/10)
CIN-I	10 (1/10)	50 (5/10)	40 (4/10)	0 (0/10)
CIN-II	0 (0/10)	30 (3/10)	60 (6/10)	10 (1/10)
CIN-III	0 (0/10)	30 (3/10)	50 (5/10)	20 (2/10)
Carcinomas	0 (0/10)	0 (0/10)	30 (3/10)	70 (7/10)

Values are presented as percentage (n). p<0.001 (normal versus CIN-I, CIN-II, CIN-III or carcinomas). p<0.01 (CIN-II or CIN-III versus carcinomas; CIN-I versus CIN-III). CIN-I versus CIN-II, insignificant.

Figure 1. Increased eIF2S2 expression in human cervical cancer tissues (n=10). Representative immunohistochemical staining of eIF2S2 in various tissues: normal cervical epithelial tissue, cervical intraepithelial neoplasia CIN-I, CIN-II, CIN-III, invasive cervical carcinomas, and cervical squamous epithelium adjacent to cancer (original magnification ×100). The intensity of immunoreactivity was scored 0 (undetectable), 1+ (weak), 2+ (moderate), and 3+ (strong). Data are presented as a percentage of immunoreactivity scores. Scale bar=100 µm

eIF2S2 promotes the growth of cervical cancer cells

We initially assessed the effects of downregulating eIF2S2 on the proliferation and growth of cervical cancer cells to investigate the role of eIF2S2 in cancer development. Transient SiHa and HeLa cell transfections were conducted with eIF2S2 siRNA, which reduced eIF2S2 expression that correlated with the amount of siRNA treatment and exposure duration (Supplementary Fig. 3). The eIF2S2 depletion in SiHa and HeLa cells caused a significant reduction in both cell viability (Fig. 2A, 2B) and proliferation (Fig. 2C, 2D) compared to cells transfected with control siRNA or mock cells.

Figure 2. Effect of eIF2S2 knockdown on viability and growth in vitro and in vivo. (A) SiHa and (B) HeLa cells were transfected with control or eIF2S2 siRNA for 24, 48, or 72 h. The survival fractions of cells were obtained by the water-soluble tetrazolium salts (WST) assay. (C) SiHa and (D) HeLa cells were transfected with control or eIF2S2 siRNA consecutively for 1 and 2 days and were left growing until 6 days to assess the long-term effect of eIF2S2 siRNA on cell proliferation. Cell proliferation was evaluated by trypan blue dye-excluding method. All experiments were conducted in triplicate. *p<0.001, compared to control siRNA. (E, F) Female BALB/c (nu/nu) mice were subcutaneously injected with SiHa-sh-control cells (n=3) or SiHa-sh-eIF2S2-clone-4 cells (n=6). Animals were sacrificed after 107 days of inoculation. *p<0.001, significantly different from the control group.

We stably transfected SiHa cells with either control shRNA (sh-control) or eIF2S2 shRNA (sh-eIF2S2) to evaluate the association of eIF2S2 knockdown with tumor growth in vivo, and we subsequently selected SiHa-sh-eIF2S2 clone-4 for further analysis (Supplementary Fig. 4). Thereafter, we subcutaneously injected SiHa-sh-eIF2S2 clone-4 and SiHa-sh-control cells into nude mice. Xenograft tumor development was detected 53 days after inoculation, and we measured the tumor volume every 2 weeks until 107 days. Remarkably, the group administered with SiHa-sh-eIF2S2 cells exhibited a significantly diminished tumor volume compared with those injected with SiHa-sh-control cells (Fig. 2E, 2F).

eIF2S2 promotes cervical cancer cell invasion and migration

We then assessed the effect of eIF2S2 expression knockdown on cervical cancer cell migration and invasion using soft agar colony-forming assays, transwell migration assays, and wound healing assays. Anchorage-independent growth assay involved SiHa-sh-eIF2S2 clone-3 and -4 were selected for (Supplementary Fig. 4). Clones harboring eIF2S2 shRNA exhibited a significantly decreased number of colonies in the soft-agar colony formation assay compared to cells stably transfected with control shRNA (Fig. 3A, 3B). Additionally, the transwell migration assay (Fig. 3C, 3D) and wound healing assay (Fig. 3E, 3F) demonstrated a significant decrease in SiHa or HeLa cells after transient transfection with eIF2S2 siRNA compared to control siRNA.

Figure 3. Knockdown of eIF2S2 inhibits invasion and migration in cervical cancer cells. (A) SiHa cells that were stably transfected with either control or eIF2S2 shRNA (clone-3 and clone-4) underwent the anchorage-independent cell transformation assay. Data represent three independent experiments. (B) The average number of colonies in each group was calculated and the statistical analysis was conducted to measure the significance test. (C) SiHa and (D) HeLa cells were transiently transfected with control or eIF2S2 siRNA and cell migration assay was conducted using a 24-well transwell after 24 h of serum starvation. *p<0.001, compared to control siRNA. (E) SiHa and (F) HeLa cells were transfected with either control or eIF2S2 si-RNA for various duration. A wound healing assay was performed as described in the Materials and Methods section.

Furthermore, we established an eIF2S2 overexpression stable cell line in NIH3T3 cells to investigate whether increased expression of eIF2S2 contributes to the tumorigenic process (Supplementary Fig. 5). NIH3T3-vector or NIH3T3-eIF2S2 clones (eIF2S2-CI2 and CI3) were then subjected to the anchorage-independent cell transformation assay in the presence or absence of TPA. The average number of colonies was significantly higher in NIH3T3-eIF2S2 cells treated with TPA compared with the treatment that caused TPA-induced anchorage-independent growth in the NIH3T3-vector (Fig. 4).

Figure 4. Stable eIF2S2 overexpression promotes TPA-induced cell transformation. (A) NIH3T3-vector (control) or NIH3T3-eIF2S2 (eIF2S2-CI2&3) cells underwent an anchorage-independent cell transformation assay in the presence or absence of TPA. Data represent three independent experiments. (B) The average number of colonies in each group was calculated and statistical analysis was conducted to measure the significance test. All experiments were performed in triplicate.

The N-terminal domain of eIF2S2 interacts with the MH1 domain of SMAD4

We focused on the eIF2S2 and SMAD4 interaction because eIF2S2 could be a binding partner of SMAD4 following bioinformatics-based protoarray analysis in our previous study to comprehend the role of eIF2S2 in tumorigenesis (Rajasekaran et al., 2021). We initially conducted an immunoprecipitation assay with SiHa cells to confirm the potential interaction between eIF2S2 and SMAD4. Anti-SMAD4 or mouse control IgG was used for the immunoprecipitation, followed by Western blotting with anti-eIF2S2 antibody. Fig. 5A shows endogenous eIF2S2 co-immunoprecipitated with endogenous SMAD4, thereby confirming the interaction between eIF2S2 and SMAD4. We used BiFC with different deletion constructs of eIF2S2 and SMAD4 as illustrated in Supplementary Fig. 6 to further validate this interaction. The fluorescence signal generated in cells co-transfected with plasmids containing the full length of eIF2S2 and SMAD4 indicates the binding between eIF2S2 and SMAD4 (Fig. 5B, left panel). A series of truncated domains, named N-terminal eIF2S2 (aa 1-140, Nt eIF2S2) and C-terminal eIF2S2 (aa 141-333, Ct eIF2S2), were co-transfected with full length-SMAD4 to further determine the SMAD4-interacting region of eIF2S2. Cells co-expressing N-terminal eIF2S2 and SMAD4 demonstrated a strong fluorescence signal, whereas C-terminal eIF2S2 and SMAD4 co-expression failed to produce a fluorescence signal (Fig. 5B, left panel). A series of truncated domains, named SMAD4-MH1 (aa 18-142), SMAD4-MH2 (aa 323-552), SMAD4-linker (aa 143-322), SMAD4-MH1-linker (aa 1-309), and SMAD4-MH2-linker (aa 144-552), were co-transfected with N-terminal-eIF2S2 to determine the binding region of SMAD4 to the N-terminal eIF2S2. The N-terminal domain (aa 1-137) of eIF2S2 produced fluorescence signals with MH1 as well as MH1-linker domains of SMAD4, but not with plasmids expressing SMAD4-MH2, SMAD4-linker, or SMAD4-MH2-linker domains (Fig. 5B, right panel). These results indicate that the N-terminal domain of eIF2S2 interacts with either the MH1 region, but not the linker, MH2- or MH2-linker regions of SMAD4.

Figure 5. Interaction of eIF2S2 and SMAD4 in SiHA and HeLA. (A) Immunoprecipitation in SiHa was performed with anti-SMAD4 monoclonal antibody (S4) or mouse control IgG. Whole lysate (WCL), supernatant (Sup), and immunoprecipitate (IP sample) were separated by SDS-PAGE and immunoblotted with either eIF2S2 or SMAD4 antibody. (B) HeLa cells were co-transfected with plasmids that harbor full-length eIF2S2, N-terminal-eIF2S2 (aa 1-140), or C-terminal-eIF2S2 (aa 141-333) in presence of full-length SMAD4 for BiFC assay (left panel). Full-length SMAD4 co-transfected with SMAD2-MH1 and -MHC2 was used as the negative and positive control, respectively. The N-terminal domain of eIF2S2 (aa 1-140) was co-transfected with plasmids that harbor each SMAD4 domain, and their interaction was visualized by BiFC assay (right panel). All experiments were performed in triplicate.

eIF2S2 inhibits p15 and p27 expression and downregulates p15 promoter activity

We investigated the effect of eIF2S2-SMAD4 binding via the MH1 domain on the expression and promoter activity of SMAD4-target genes because the MH1 domain of SMAD4 plays a crucial role in DNA binding of SMAD4 as a transcriptional factor. The expression of p15 and p27, which are transcriptionally regulated by SMAD4 and inhibit cell proliferation, was markedly decreased in SiHa cells that overexpress eIF2S2 compared to cells harboring an empty vector (Fig. 6A), whereas eIF2S2 knockdown by siRNA increased p15 and p27 expression in SiHa cells (Fig. 6B). Additionally, transient eIF2S2 overexpression in these cells diminished the mRNA expression of p15 (Fig. 6C). Moreover, we revealed that eIF2S2 overexpression dose-dependently inhibited TGFβ-induced p15 promoter activity in SiHa cells (Fig. 6D). Thus, eIF2S2 may be involved in cervical cancer carcinogenesis by suppressing SMAD4 function as a transcription factor. We investigated the association of eIF2S2 with SMAD7 because SMAD7 could play a role in interfering with the DNA binding activities of SMAD4. Immunoprecipitation and BiFC analysis revealed that the SMAD7-MH2 domain interacts with eIF2S2 (Supplementary Fig. 8). Transient eIF2S2 overexpression in SiHa cells exhibited a marked SMAD7 expression induction (Fig. 6E). Conversely, transfecting cells with si-eIF2S2 decreased SMAD7 expression in SiHa cells (Fig. 6F). Thus, this result indicates that SMAD7 upregulation by eIF2S2 may inhibit the ability of SMAD4 to bind to DNA as a transcription factor.

Figure 6. eIF2S2 downregulates p15 and p27 expression in cervical cancer cells. The p15 and p27 expression level was detected in SiHa cell lysate, where cells were (A) overexpressed with p3x-Flag-CMV10-eIF2S2 transfection or (B) knocked down with eIF2S2 siRNA transfection. The mRNA expression of p15 was detected by qRT-PCR (C) with p3x-Flag-CMV10-eIF2S2 transfection. HPRT was used as an internal control. *p<0.001, compared to vector alone. (D) SiHa cells were transiently transfected with eIF2S2 or control vector and incubated with or without TGFβ (5 ng/ml) for 24 h. The p15 promoter activity from transfected cells was measured as described in the Materials and Methods section. (E) SiHa cells were transiently transfected with p3x-Flag-CMV10-eIF2S2 or control vector. Protein lysates underwent Western blot analysis to detect the expression level of Smad7. Flag and eIF2S2 were measured to confirm the transfection efficiency, and actin was used as the loading control. (F) SiHa cells were transfected with control siRNA or eIF2S2 siRNA, and the Smad7 expression was detected by Western blot analysis. All experiments were performed in triplicate.

eIF2S2 negatively regulates SMAD4 expression

We then attempted to investigate if the cellular status of eIF2S2 modulated the tumor suppressor gene SMAD4 expression. This was confirmed by co-transfecting SiHa cells with plasmid constructs harboring eIF2S2 or SMAD4. Additionally, eIF2S2 overexpression decreased SMAD4 expression in cells that ectopically induced SMAD4 expression (Fig. 7A). In contrast, SMAD4 expression inhibition by eIF2S2 was further confirmed by siRNA-based eIF2S2 knockdown, which demonstrated increased SMAD4 expression in SiHa (Fig. 7B) We investigated SMAD4 expression from NIH3T3 cells that are stably transfected with eIF2S2 and revealed reduced SMAD4 expression in cells harboring ectopically expressed eIF2S2 (Fig. 7C).

Figure 7. Modulation of SMAD4 expression by eIF2S2. (A) SiHa cells were transfected with p3x-Flag-CMV10-eIF2S2 or Flag-Smad4 or their combination, and the Smad4 and Flag expressions were detected by immunoblot analysis. (B) SiHa cells were transiently transfected with eIF2S2 siRNA (10 nM) or control siRNA (10 nM) and cell lysates were separated by SDS-PAGE and immunoblotted with Smad4 and eIF2S2. GAPDH expression was used as the loading control. (C) Protein extracts from clones of NIH3T3 cells stably transfected with vector alone or p3x-Flag-CMV10-eIF2S2 were subjected to Western blot analysis to detect Smad4 expression. Flag and eIF2S2 were detected to confirm transfection efficiency. (D) SiHa cells were incubated with the proteasomal inhibitor MG132 before the transfection of cells with p3x-Flag-CMV10-eIF2S2 or empty vector. Smad4 expression was detected by Western blot analysis. All experiments were performed in triplicate.

We performed qRT-PCR analysis of SMAD4 mRNA expression from cells transfected with either p3x-Flag-CMV10-eIF2S2 or a control vector to determine the transcriptional regulation of reduced SMAD4 expression upon eIF2S2 overexpression. Transient transfection with eIF2S2 did not cause a significant change in SMAD4 mRNA levels in SiHa and HeLa cells compared to respective vector-transfected cells (Supplementary Fig. 7).

SiHa cell incubation with proteasomal inhibitor MG132 before transfection with p3x-Flag-CMV10-eIF2S2 or control vector abrogated eIF2S2-mediated SMAD4 expression downregulation (Fig. 7D). However, SMAD4 ubiquitination was not confirmed with increased eIF2S2 expression (data not shown); thus, eIF2S2 expression may induce proteasomal degradation of another factor that could affect SMAD4 expression, thereby indirectly downregulating SMAD4 expression.

Cervical cancer is the second leading cause of cancer among females globally (Jemal et al., 2011), with a prevalence of 11.7%, particularly high in Sub-Saharan Africa, Eastern Europe, and Latin America, where approximately 85% of cases occur (Bruni et al., 2010). HPV infection is widely acknowledged as a major contributor to cervical cancer development despite substantial efforts over the past three decades to elucidate its etiology (zur Hausen, 1996). However, noteworthily, a significant proportion of HPV-infected women do not develop cervical cancer, indicating the involvement of additional etiologic factors in cervical carcinogenesis. The biochemical basis of cervical cancer includes specific oncogene activation, including HPV E6 and E7 oncogenes (Cone et al., 1992; Hsu and McNicol, 1992), HER2/neu (Mitra et al., 1994), N-and H-Ras (Mammas et al., 2005), and c-Myc (Brychtova et al., 2004), and tumor suppressor gene inactivation, including p53 (Shai et al., 2008), pRb (Jones and Wells, 2006), and SMAD4 (Maliekal et al., 2003).

In recent years, dysregulated protein translation has been implicated in the neoplastic transformation of cells, with several eIFs determined as putative oncogenes in various cancers (Bilanges and Stokoe, 2007). Notably, eIF4E has been recognized as an oncogenic marker in cervical cancer, demonstrating a gradual increase in expression correlated with histopathologic grade, while remaining undetectable in normal cervical epithelium (Lee et al., 2005; Matthews-Greer et al., 2005). Another eIF2 complex subunit, eIF2S2, is upregulated in human colorectal cancer (Staub et al., 2006), and studies demonstrating that eIF2S2 deletion suppresses the incidence of testicular germ cell carcinomas further supported its involvement in carcinogenesis (Heaney et al., 2009). Gene ontology analysis of HPV-positive SiHa cells revealed a significantly elevated eIF2S2 mRNA expression (Ahn et al., 2005). Consistently, the present study demonstrated increased eIF2S2 expression across histopathologic grades of cervical neoplasia, including CIN and cervical cancer, whereas eIF2S2 expression remained undetectable in normal squamous epithelial tissue. The stable eIF2S2 knockdown attenuated anchorage-independent growth and xenograft tumor growth in SiHa cells, and transient eIF2S2 silencing reduced proliferation and migration in both SiHa and HeLa cells, emphasizing eIF2S2 involvement in cervical carcinogenesis. Furthermore, a significant increase in phorbol ester-induced soft agar colony formation was found in NIH3T3 cells stably transfected with p3xFlag-CMV10-eIF2S2 compared to cells harboring an empty vector, thereby providing additional confirmation of the pro-tumorigenic role of eIF2S2 in cervical carcinogenesis.

Several studies have emphasized the loss of the tumor suppressor function of SMAD4 in several human cancers, including cervical cancer (Maliekal et al., 2003; Kloth et al., 2008). Mechanisms that contribute to the loss of SMAD4 function involved homozygous deletion, gene mutation, promoter hypermethylation, and post-translational modification (Miyaki and Kuroki, 2003; Wang et al., 2007). Additionally, certain binding partners of SMAD, such as Jab1, a c-Jun oncoprotein co-activator, and NRF1, directly interact with SMAD4, thereby inducing ubiquitination and proteasomal degradation (Wan et al., 2002; Rajasekaran et al., 2021; Song et al., 2022). In a protein microarray designed to determine novel SMAD4 interacting proteins, eIF2S2 appeared as a new SMAD4 binding protein. The co-immunoprecipitation assay and BiFC analysis of the present study confirmed this novel interaction. The N-terminal domain of eIF2S2 interacted with the MH-1 domain of SMAD4. This N-terminal domain of eIF2S2, exclusive to eukaryotic eIF2S2, contains lysine-rich boxes binding to eIF2Bε (Asano et al., 1999) and eIF5 (Das et al., 1997) during protein translation activation. Considering that eIF2Bε promotes cell proliferation and transformation (Gallagher et al., 2008), investigating the association of endogenous occupancy of the N-terminal domain of eIF2S2 by SMAD4 with eIF2S2 interaction with eIF2Bε, and thereby contributing to cervical carcinogenesis, presents an intriguing avenue for investigation. Another intriguing aspect is the phosphorylation of serine-2 and residues 67 at the N-terminal domain of eIF2S2 by casein kinase-2, which phosphorylates eIF5 and promotes cell growth (Homma and Homma, 2005). Understanding the effect of eIF2S2 interaction with SMAD4 on casein kinase-2 activity, eIF5 phosphorylation, and eIF2Bε concerning the oncogenic potential of eIF2S2 warrants further exploration.

Conversely, the MH1 domain of SMAD4 is indispensable for the DNA binding of SMAD4 and subsequent SMAD4-regulated gene transcription (Jones and Kern, 2000). The observation that eIF2S2 overexpression decreased p15 and p27 expression, and conversely, siRNA-mediated eIF2S2 silencing induced mRNA expression of p15, indicates that the interaction of eIF2S2 at the MH1 domain of SMAD4 may down-regulate SMAD4-mediated transactivation of genes that encode cell growth inhibitory proteins. The reduced SMAD4 protein expression upon eIF2S2 overexpression may decrease p15 and p27 expression. These results indicate that eIF2S2 improves cell proliferation and transformation, at least in part, by blocking SMAD4-mediated anti-tumor signaling.

The present study revealed that eIF2S2 interacts not only with SMAD4 but also with SMAD7, as exhibited by BiFC analysis. Another noteworthy result indicative of eIF2S2’s role in cervical carcinogenesis is that the transfection of cervical cancer cells with eIF2S2 induced the constitutive SMAD7 level.

SMAD4 functions as a mediator in the TGFβ-induced signal transduction pathway; thus, the role of eIF2S2 in TGFβ-induced anti-proliferative gene expression was assessed. The dose-dependent TGFβ-induced p15 promoter activity inhibition upon transient eIF2S2 overexpression indicates that eIF2S2 may influence the TGFβ-induced signaling pathway at stages other than SMAD4. Moreover, the known function of the eIF2S2-inducible protein SMAD7 in the negative regulation of TGFβ signaling strengthens this possibility. Thus, additional studies are warranted to confirm the role of eIF2S2 in modulating the TGFβ signaling pathway and its consequences in cervical carcinogenesis.

In summary, the present study convincingly reveals that eIF2S2 functions as a putative oncogene in cervical carcinogenesis, at least in part by interacting with SMAD4 and inhibiting SMAD4 expression, thereby reducing expression of SMAD4 target genes p15 and p27. Moreover, eIF2S2 binds with SMAD7 and upregulates SMAD7 protein expression in cervical cancer cells, indicating the possibility of eIF2S2-dependent cross-regulation of SMAD4 signaling by SMAD7, which has inhibited the TGF signaling pathway, and hence cancer progression. However, the elucidation of the detailed mechanism of SMAD4 regulation by eIF2S2, as proposed in Fig. 8, merits further investigation.

Figure 8. A proposed mechanism by which eIF2S2 may contribute to cervical carcinogenesis. The dotted lines indicate the mechanisms yet to be investigated.

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (No. 2022R1C1C1009883, No. 2021R1F1A1057411), as well as by the Priority Research Centers Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science, and Technology (2016R1A6A1A03007648). Additionally, it received support from the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education, Science, and Technology (NRF-2014R1A1A).