OVCAR-3 and SK-OV-3 cell lines were purchased from the American Type Culture Collection (HTB-161 and HTB-77, respectively; Rockville, MD, USA). These cell lines are derived from human ovarian cancer cell tissue and are often used to generate solid tumor xenografts (10,14–18). The cells were cultured in RPMI-1640 medium (Sigma-Aldrich, St Louis, MO, USA) at 37°C in a water-saturated atmosphere containing 5% CO2/95% air. The culture medium was supplemented with 10% fetal bovine serum (NICHIREI Biosciences, Tokyo, Japan), 100 U/ml penicillin, and 100 µg/ml streptomycin.

The pCMV6-AC-GFP-hCBR1 plasmid (RG204950, OriGene Technologies, Inc., Rockville, MD, USA) was used to generate hCBR1-overexpressing cells. This plasmid encodes hCBR1 fused to turbo green fluorescent protein (hCBR1-tGFP), which is expressed under the control of the CMV promoter. Plasmid pCMV6-AC-GFP (PS100010, OriGene Technologies, Inc.), which expresses tGFP, was used to generate negative control cell lines.

Ovarian cancer cells were plated into a 10 cm dish and cultured to 70–80% confluence. Cells were then transfected with 24 µg of pCMV6-AC-GFP-hCBR1 or pCMV6-AC-GFP using Lipofectamine 3000 (Life Technologies, Rockville, MD, USA) for 24 h at 37°C. Then, G418 (NACALAI TESQUE, Inc., Kyoto, Japan) was added (0.4 mg/ml) to the culture medium to select cell lines possessing the plasmid DNA in their genome.

Cells were homogenized on ice in radio-immunoprecipitation assay (RIPA) buffer (Fujifilm Wako, Osaka, Japan) containing cOmplete Mini EDTA-free Protease Inhibitor Cocktail Tablets provided in EASY packs (Roche, Basel, Switzerland). Protein measurements were performed using Bio-Rad Protein Assay Dye Reagent Concentrate (Bio-Rad Laboratories, Hercules, CA, USA). Proteins (8 µg) were loaded into the wells of 5–20% SDS-polyacrylamide gels (Fujifilm Wako), electrophoresed, and transferred to polyvinylidene difluoride membranes. The membranes were blocked for 1 h at room temperature with Tris-buffered saline (TBS) containing 0.05% Tween-20/1% skim milk or 2% bovine serum albumin [for the anti-phospho-translation initiation factor 2 (eIF2)α blot] and then incubated overnight at 4°C with a polyclonal rabbit anti-hCBR1 antibody (Ab) (1:1,000; ab-186825; Abcam, Cambridge, UK), an anti-β-actin Ab (1:5,000; M177-3; MBL, Tokyo, Japan), an anti-eIF2α Ab (1:5,000; 9722, Cell Signaling Technology, Danvers, MA, USA), an anti-phospho-eIF2α (p-eIF2α) Ab (1:5,000; 9721, Cell Signaling Technology), and an anti-activating transcription factor 4 (ATF4) Ab (1:1,000; ab-216839; Abcam). After washing with TBS/0.05% Tween-20, the membranes were incubated for 1 h at room temperature with an appropriate horseradish peroxidase (HRP)-conjugated secondary Ab [either anti-mouse IgG-HRP (NA9310V; GE Healthcare, Pittsburgh, PA, USA) or anti-rabbit IgG-HRP (NA9340V; GE Healthcare)]. Protein bands were detected using enhanced chemiluminescence (Image Quant LAS4000 system; GE Healthcare). The detected bands were quantified using Image J (19).

Wild-type, negative control, and hCBR1-tGFP-overexpressing ovarian cancer cells were plated in 6 cm dishes (1.0×104 cells/dish) and cultured at 37°C for 24, 48, 72, 96, or 120 h. Cells were then washed with 1 ml of Dulbecco's phosphate-buffered saline and detached with 0.5 g/l-Trypsin/0.53 mmol/l-EDTA Solution (NACALAI TESQUE, Inc., Kyoto, Japan). Trypan blue solution (0.1%; NACALAI TESQUE, Inc.) was used to discriminate living from dead cells; cells were counted using a hemocytometer (Bio Medical Science, Tokyo, Japan). The rate of cell growth was calculated as the slope of a linear fit created by plotting the natural logarithm of the cell number against culture time.

Wild-type OVCAR-3 and five lots of OVCAR-3-derived cells overexpressing CBR1 were cultured and sampled in triplicate. Cell lysates were prepared as described for immunoblot analysis. Label-free whole cell proteomic analysis was performed as described previously (20). Briefly, cell lysates containing 20 µg of total protein, as determined by the bicinchoninic acid (BCA) method, were denatured with acetone. The precipitates were then denatured with 50% trifluoroethanol. Disulfide bonds were reduced with dithiothreitol (DTT) and alkylated with iodoacetamide followed by trypsin digestion. After desalting and purification of the resulting peptides using MonoSpin C18 columns (GL Sciences Inc., Tokyo, Japan), the samples were subjected to LC-MS/MS using a nanoLC system (Eksigent 400, AB Sciex, Framingham, MA, USA) coupled online to a mass spectrometer (TripleTOF6600, AB Sciex). Peptide separation was performed using LC on a nano C18 reverse-phase capillary tip column (75 µm × 125 mm, 3 µm, Nikkyo Technos CO., Tokyo, Japan) at 300 nl/min with a 90 min linear gradient of 8–30% acetonitrile in 0.1% FA, and then, with a 10 min linear gradient of 30 to 40% acetonitrile in 0.1% FA. The mass spectrometer was operated in information-dependent acquisition (IDA) and data-independent acquisition (SWATH) while in positive ion mode, scanning full spectra (400-1,500 m/z) for 250 msec, followed by up to 30 MS/MS scans (100-1,800 m/z for 50 msec each), for a cycle time of 1.8 sec. Candidate ions with a charge state between +2 and +5 and counts above a minimum threshold of 125 counts per second were isolated for fragmentation, and one MS/MS spectrum was collected before adding those ions to the exclusion list for 12 sec. For SWATH acquisition, the parameters were set as follows: 100 msec TOF MS scan, followed by 200 variable SWATH windows each at 50 msec accumulation time for m/z 400-1,250. MS/MS SWATH scans were set at 5 Da window overlapping by 1 Da for m/z 400-1,250 and varied on each side of the mass range. The total cycle time was 9.6 sec. A rolling collision energy parameters script was used to automatically control the collision energy. Database searching for acquired spectra was performed using ProteinPilot 5.0.1 software (AB Sciex). The positive identification threshold was set at a false discovery rate of 1% or less. The resulting library file and SWATH (data independent acquisition) files were processed by PeakView (ver. 2.2.0, AB Sciex) and exported to MarkerView (ver. 1.3.0.1; AB Sciex). The peak area of individual proteins was normalized relative to the sum of the peak areas of all detected proteins.

Experiments were performed in triplicate for immunoblot analysis and evaluation of ovarian cancer cell proliferation. Normal distribution was tested by Shapiro-Wilk test. Statistical differences between the three groups (hCBR1-tGFP-overexpressing, negative control, and wild-type cells) were analyzed using one-way ANOVA followed by Tukey's test (parametric data) or Kruskal-Wallis test (non-parametric data). P<0.05 was considered to indicate a statistically significant difference. All statistical analyses were performed using IBM SPSS ver. 29.0 statistics software (IBM, Chicago, Illinois).

Regarding the statistical analysis of the proteome data, principal component analysis using SIMCA software (version 15.0.2, Infocom Corp., Tokyo, Japan) confirmed that there was no outlier proteome in any of the samples. Since the cell lines used show endogenous CBR1 expression, proteins covarying with CBR1 were subjected to Spearman's rank correlation analysis and Pearson's correlation analysis (R version 4.1.0 software) to capture proteomic variations due to CBR1 transfection. Proteins with FDR-adjusted P<0.05 were identified as proteins covarying with CBR1 and subjected to pathway enrichment and network analysis by Ingenuity Pathway Analysis (IPA) software (Qiagen, Venlo, Netherland).

To investigate the role of CBR1 on the malignancy of ovarian cancer cells, we first generated OVCAR-3 and SK-OV-3 cell lines stably expressing hCBR1-tGFP. We also generated ovarian cancer cells expressing only tGFP as a control (negative control cells). Immunoblot analysis using an anti-CBR1 Ab revealed stable expression of hCBR1-tGFP (Figs. 2A and 3A). A band corresponding to hCBR1-tGFP was not observed in the negative control and wild-type cell samples. We measured CBR1 protein expression levels (i.e., the combined expression levels of hCBR1-tGFP and CBR1) in OVCAR-3 and SK-OV-3 cells using image analysis software (Figs. 2B and 3B). The results confirmed that expression of CBR1 protein in hCBR1-tGFP-overexpressing OVCAR-3 cells was significantly higher than that in wild-type and negative control cells (Fig. 2B). As to SK-OV-3 cells, hCBR1-tGFP-overexpressing cell #1 and #2 showed the highest and relatively high expression of CBR1 protein, respectively (Fig. 3B). Although all hCBR1-tGFP-overexpressing SK-OV-3 cells showed hCBR1-tGFP expression (Fig. 3A), the expression was slight in the cell line #4. Therefore, statistical significance on the total CBR1 amount was not observed among cell groups (Fig. 3B). Nevertheless, hCBR1-tGFP was visibly expressed in the cell line #4, and the other hCBR1-tGFP-expressing cells show a tendency of upregulation of total hCBR1 protein. To examine the importance of the CBR1 protein in ovarian cancer cells, these established SK-OV-3 cells were also used for subsequent experiments.

To evaluate the effect of CBR1 overexpression on tumor cell growth, we measured proliferation of OVCAR-3 and SK-OV-3 cells expressing hCBR1-tGFP. As shown in Fig. 4A and B, growth of OVCAR-3 and SK-OV-3 cells expressing hCBR1-tGFP was slower than that of wild-type or negative control cells. These results suggest that overexpression of hCBR1 strongly inhibits growth of ovarian cancer cells, although it should be noted that overexpression of tGFP alone moderately affected growth of SK-OV-3 cells (Fig. 4B). hCBR1-tGFP-overexpressing OVCAR-3 cells showed significant suppression of cell proliferation (P<0.05, vs. negative control or wild-type) (Fig. 4A). hCBR1-tGFP-overexpressing SK-OV-3 cells also showed significant suppression of cell proliferation (P<0.05, vs. negative control or wild-type, Fig. 4B). As to SK-OV-3 cells, differences in cellular growth were also observed between wild-type and negative control cells (P<0.05, Fig. 4B). Because tGFP is potentially cytotoxic (21), expression of tGFP may also affect the growth of SK-OV-3 cells.

In hCBR1-tGFP-overexpressing OVCAR-3, we confirmed an inverse correlation between hCBR1 overexpression and cell proliferation (Fig. 4C). In the hCBR1-tGFP-overexpressing lines (#1, #2), CBR1 expression levels were high, and the cell proliferation rate was reduced. By contrast, negative control cells (#3-#5) and wild-type cells expressed low levels of CBR1 and showed higher proliferation rates. In these experiments, CBR1 expression is the sum of endogenous CBR1 and exogenous hCBR1-tGFP, and the higher levels of CBR1 protein in hCBR1-tGFP-overexpressing lines are consistent with the results shown in Fig. 1B. Thus, these results support the above data showing that hCBR1-tGFP-overexpressing lines grow more slowly than negative control and wild-type cells (Fig. 4A). Previously, we showed that transient expression of CBR1 in a xenograft mouse model may induce apoptosis through activation of caspase pathways (6). Thus, stable overexpression of hCBR1-tGFP may induce apoptosis in OVCAR-3 and SK-OV-3 cells.

To elucidate the CBR1-mediated molecular mechanism underlying growth inhibition, we performed LC-MS/MS of OVCAR-3 cell lysates, followed by IPA. Whole cell proteomics analysis using a label-free quantification method plus MS resulted in quantification of 939 proteins, with no samples showing outliers upon principal component analysis (Fig. S1). Of these, the abundance of 155 proteins correlated significantly with expression of hCBR1 (FDR-adjusted P-value <0.05, Spearman's rank correlation, Fig. S2). We then performed IPA to identify enriched canonical pathways associated with hCBR1-correlated proteins; the top 20 pathways are shown in Table I. Expression of proteins involved in various signaling pathways, whose role in controlling tumor progression is suggested, correlated with expression of hCBR1. Among these, the eIF2 signaling pathway was the most enriched.

The z-score for the eIF2 signaling pathway, as calculated by IPA, was +1.387. The z-score indicates whether the detected signal is biased toward activation or inactivation, with positive values indicating activation and negative values indicating inactivation (Fig. S3). Table I shows that eIF2 signaling is highly affected by overexpression of hCBR1. Proteins associated with eIF2 signaling are listed in Table II. Of the 23 proteins associated with eIF2 signaling, 17 had positive correlation coefficients.

We also used IPA to perform a protein-protein interaction network analysis of hCBR1-correlated proteins. The top 35 out of the 155 proteins whose expression correlated with that of CBR1 were extracted in the order of interaction network robustness and presented as a subnetwork (Fig. 5). Among the 35 proteins, 15 formed a subnetwork related to eIF2 signaling, as shown by the cyan borders. These results suggest that the eIF2 signaling pathway may be involved in suppressing cell growth.

Next, we investigated whether the eIF2 signaling pathway is actually altered by overexpression of hCBR1-tGFP. Considering the magnitude of changes in eIF2 signaling, and the critical importance of phosphorylation status of eIF2 during regulation of the pathway (Fig. 6A, see also the Discussion section) (22), we performed immunoblot analysis to examine phosphorylation of eIF2α, a subunit of eIF2. A clear band corresponding to p-eIF2α was observed in lysates prepared from hCBR1-tGFP-expressing cells (#1, #2) (Fig. 6B-a), confirming that overexpression of hCBR1 induces phosphorylation of eIF2α. Expression and phosphorylation of eIF2α protein in OVCAR-3 cells were quantified using image analysis software. The results confirmed that the phosphorylation levels of eIF2α in hCBR1-tGFP-overexpressing cells were higher than those in wild-type and negative control cells (Fig. 6B-b).

Phosphorylation of eIF2α downregulates general translation and upregulates expression of ATF4 (Fig. 6A) (22). Although we examined involvement of ATF4 downstream of eIF2α, we found that expression of ATF4 was not upregulated in hCBR1-tGFP-expressing cells (Fig. 6C). Therefore, overexpression of hCBR1 may induce downregulation of general translation via phosphorylation of eIF2α.