The human breast carcinoma cell lines MCF-7 (expressing wild-type p53) and MDA-MB-231 [bearing mutated p53 (R280K)] were obtained from the European collection of cell cultures (ECACC) and maintained in Dulbeccos modified Eagles medium (Sigma-Aldrich, UK) supplemented with 10% foetal bovine serum (Gibco) and 1% penicillin/streptomycin (Lonza) at 37°C in a humidified atmosphere containing 5% CO₂ until they reached 70% confluency (48 h). Where indicated cells were treated with 10 ng/ml interferon-γ (IFN-γ) (Sigma-Aldrich) for 24 h or 10 µM etoposide (ETOP) for 24 h (Sigma-Aldrich).

Cellular extracts from MCF-7 and MDA-MB-231 were collected in ice-cold TNN buffer containing 1 mM DTT, 1 mM PMSF, and 1 µg/ml protease inhibitors. Protein concentrations were determined by the Bradford assay (Sigma Aldrich) and 30 µg of protein per sample were resolved on a 20% precast polyacrylamide gel and transferred to a PVDF membrane. The membranes were blocked using 5% fat-free milk in PBS (v/v) for 1 h at 25°C. Membranes were then incubated in 2.5% milk in PBS-0.1% Tween-20 (v/v) with anti-P4HB antibody (Santa Cruz Biotechnology; sc-136230) (dilution 1:500) or β-actin (Sigma Aldrich; A1978) (dilution 1:10,000) overnight at 4°C. The membranes were washed three times with PBS-0.1% Tween-20 (v/v) for 5 min and then incubated with secondary anti-mouse immunoglobulin G conjugated to horsedish peroxidase (GE Healthcare) (dilution 1:1,000) in 2.5% milk in PBS-0.1% Tween-20 (v/v) for 1 h at 25°C. Protein bands were then visualized using the ChemiDoc MP imaging system (Bio-Rad).

A concentration of 5 µM of the siGENOME P4HB siRNA and 5 µM of the siGENOME non-targeting siRNA pool was added to each well containing 2×10⁵ cells in DMEM and incubated for 72 h according to the suppliers instructions (Dharmacon) as described previously (28). Immediately after the 72 h transfection cells were used for subsequent experimentation. The sequences of the siRNA pools against P4HB and scramble siRNA were: P4HB: ACAGGACGGUCAUUGAUUA, GGACGGUCAUUGAUUACAA, CCAAGAGUGUGUCUGACUA, CAGAGAGGAUCACAGAGUU and scramble: UAGCGACUAAACACAUCAA, UAAGGCUAUGAAGAGAUAC, AUGUAUUGGCCUGUAUUAG, AUGAACGUGAAUUGCUCAA.

ROS generation was measured using the dye CM-H₂DCF-DA (29). Briefly, cells were grown in 6-well plates at a density of 1×10⁶ cells/well and treated as indicated. Cells were centrifuged for 3 min at 400 × g at 25°C, the supernatant was removed and the pellet resuspended in cold PBS. Then, 1 µM of CM-H₂DCF-DA was added and the extracts were incubated for 30 min in the dark at 25°C. Subsequently, 30 µl of samples were transferred to A2 slides (ChemoMetec) and ROS were measured using the NucleoCounter NC-3000™ (ChemoMetec).

The NucleoCounter NC-3000™ system was used to detect changes in the intracellular level of (reduced) thiols. Following the indicated treatments cells were dissociated from the culture plates using 500 µl dissociation buffer (ChemoMetec). Then the cells were stained with solution 5 according to the manufacturers instructions (ChemoMetec), loaded onto 8-chamber NC-slides and samples were analysed using the NucleoCounter NC-3,000™.

Mitochondrial transmembrane disruption was measured using the cationic dye JC-1 (5,5,6,6-tetrachloro-1,1,3,3-tetraethyl benzimidazol carbocyanine iodide) (ChemoMetec) and the NucleoCounter NC-3,000™. Following the indicated treatment, the cells were stained with JC-1 and DAPI according to the manufacturers instructions (ChemoMetec). Cellular JC-1 monomers and aggregates were detected as green and red fluorescence, respectively. Mitochondrial depolarization and apoptosis were revealed as a decrease in the ratio of red/green fluorescence. Staining with the blue fluorescent dye (DAPI) was used to detect necrotic and late apoptotic cells. After stained cells were loaded on an 8-chamber NC-Slide A8™ and samples were analysed quantifying the amount of blue, green, and red fluorescence using the NucleoCounter NC-3,000™.

ATP levels were measured using the ViaLight plus kit (Lonza), based on the amount of bioluminescent ATP generated in cells. ATP monitoring reagent (AMR plus) was prepared by adding assay buffer into the vial containing the lyophilized AMR and incubated at room temperature for 15 min for complete rehydration. Cells were lysed in 50 µl and the cell lysate was added to a luminometer plate together with 100 µl of cell lysis reagent for 10 min. A total volume of 100 µl of AMR plus was added to the appropriate well. The plate was then incubated at room temperature for 2 min and luminescence values were obtained using the Fluostar OPTIMA microplate reader (BMG Labtech).

Breast cancer cell lines were seeded in a 6-well plate at a concentration of 1×10⁶ cancer cells/well for 24 h in a CO₂ incubator. Cells were then transfected with P4HB siRNA or scramble siRNA for 48 h at 37°C. Following treatment with IFN-γ or ETOP for 24 h at 37°C, the cells were detached from the wells with 500 µl dissociation buffer and the cell lysates were transferred to a 15 ml falcon tube and centrifuged at 300 × g for 5 min at 4°C. The supernatant was removed and 3×10⁵ cells in 100 µl (10% FBS in PBS) were transferred to a fresh tube. APC-anti human HLA-G (Thermo Fisher) was added and the cells were incubated for 30 min on ice in the dark. The stained samples were analysed by flow cytometry (Becton Dickinson).

The METABRIC breast cancer dataset was downloaded from cBio Cancer Genomics Portal (http://cbioportal.org) for 1,900 breast cancer patients (26). The distribution of the mRNA expression levels of PDIA1 and HLA-G was measured across samples. PDIA1 (P4HB) and HLA-G mRNA expression levels were considered high if their z-scores were higher or equal to the 75th percentile of the distribution (30) and low if their z-scores were lower than the 75th percentile of the distribution (z-score high PDIA ≥ 0.572, z-score high HLA-G≥ 0.517).

Graphs were plotted using GraphPad PRISM 7 (GraphPad software Inc.). The Mann-Whitney test was used to assess the significance of the difference between two groups, whereas the Kruskal-Wallis test was used to assess the significance of the differences between more than two groups. Log-rank was used to test the significance of the difference between Kaplan-Meier survival curves. The mean was calculated as ± SEM. P<0.05 indicated statistical significance.

To investigate the role of the PDIA1 in the modulation of diverse hallmarks of cancer such as resistance to cell death, regulation of cellular energy production pathways and evasion of immune destruction, the gene expression of this oxidoreductase was silenced in the ERα-positive (MCF-7) and the ERα-negative (MDA-MB-231) breast cancer cells. The efficiency of silencing of the PDIA1 gene expression was confirmed for each assay performed. The representative western blot analysis shown in Fig. 1, which shows the efficient silencing of the PDIA1 gene expression only in MCF-7 and MDA-MD-231 cells transfected with the siRNA-PDIA1 untreated or treated with IFN-γ or ETOP, but not in those cells transfected with the scramble siRNA.

Type II interferon IFN-γ induces both carcinogenic and cytotoxic effects (31) playing an important role in cell growth and survival in a manner involving the generation of ROS (32). The topoisomerase II inhibitor ETOP has also been shown to induce ROS generation in brain cancer cells (33). Additionally, IFN-γ regulates the expression of genes encoding NADPH oxidases (Nox) (34) and evidence indicating functional interaction between PDI and Nox family members (35) suggests that IFN-γ may indirectly regulate PDIA1 levels and activity under diverse oxidative stress conditions. The association between overexpression of NADPH oxidases and resistance to ETOP treatment as a result of ROS-mediated induction of senescence has also been reported (36), suggesting that PDIA1 is a potential indirect modulator of ROS mediated effects exerted upon ETOP treatment. To investigate the potential involvement of PDIA1 in altering the IFN-γ and ETOP mediated redox state in breast cancer cells ROS levels were measured in MCF-7 and MDA-MB-231 cells treated with these compounds in the presence or absence of PDIA1.

In MCF-7 cells silencing of PDIA1 did not affect ROS generation in the untreated or IFN-γ treated cells compared to scramble-transfected cells (Fig. 2A, compare bars 1 and 3 to bars 2 and 4, respectively). Decreased ROS levels were recorded in PDIA1-silenced MCF-7 cells treated with ETOP compared to scramble-transfected cells under the same conditions (Fig. 2A, compare bar 5 to bar 6). Decreased ROS levels were measured in MCF-7 cells treated with IFN-γ in the presence of PDIA1 compared to the untreated cells (Fig. 2A, compare bar 3 to bar 1) whereas in the presence of PDIA1 ETOP did not affect ROS generation in MCF-7 cells (Fig. 2A compare bar 5 to bar 1). In the absence of PDIA1 significantly decreased ROS generation was observed in the ETOP-treated MCF-7 cells compared to untreated cells (Fig. 2A, compare bar 6 to bar 2). Significantly increased ROS levels were observed in PDIA1-silenced MDA-MB-231 cells treated with either IFN-γ or ETOP compared to the scramble-transfected cells under the same conditions (Fig. 2C, compare bars 3 and 5 to bars 4 and 6, respectively). Significantly decreased ROS levels were observed in the IFN-γ treated MDA-MB-231 cells in the presence of PDIA1 compared to untreated cells under the same conditions (Fig. 2C, compare bar 3 to bar 1). ETOP treatment of PDIA1-silenced MDA-MB-231 cells resulted in a significant increase of ROS generation compared to untreated cells (Fig. 2C, compare bar 6 to bar 2). PDIA1 protein levels were not significantly affected by any treatment (Fig. 1). Representative histograms showing the ROS levels in scramble or siRNA-PDIA1-transfected MCF-7 or MDA-MB-231 cells untreated or treated with IFN-γ or ETOP are provided in Fig. 2B and 2D, respectively.

The ratio of glutathione (GSH) vs. glutathione disulfide (GSSG) is important for cellular physiology since a decreased GSH/GSSG ratio leads to increased susceptibility to oxidative stress whereas increased GSH levels induce resistance of cancer cells to oxidative stress (37). Inhibition of PDIA1 isomerase activity has been shown to abrogate glutathione depletion (8) indicating the essential role of PDIA1 in the process of disulfide bond formation and therefore the regulation of the ratio of GSH vs. GSSG (38). To investigate the role of PDIA1 in the regulation of the GSH homeostasis in breast cancer cells the GSH/GSSG ratio was investigated in MCF-7 and MDA-MB-231 cells in which the PDIA1 expression had been silenced.

Significant upregulation of GSH levels was observed in the untreated PDIA1-silenced MCF-7 cells compared to scramble-transfected cells under the same conditions (Fig. 3A, compare bar 1 to bar 2). There were no other changes evident in GSH levels in MCF-7 cells in any of the conditions studied (Fig. 3A). By contrast, in the MDA-MB-231 cells significant downregulation of GSH levels was observed in the untreated PDIA1-silenced MDA-MB-231 cells compared to the untreated and scramble-transfected cells (Fig. 3C, compare bar 1 to bar 2). Downregulation of GSH levels was also recorded in the ETOP-treated MDA-MB-231 cells in the presence of PDIA1 compared to the untreated cells (Fig. 3C, compare 5 to bar 1). Finally upregulated GSH levels were measured in PDIA1 silenced MDA-MB-231 cells treated with IFN-γ compared to the siRNA-PDIA1 transfected untreated cells (Fig. 3C, compare bar 4 to bar 2). Representative histograms showing the GSH levels in scramble or siRNA-PDIA1 transfected MCF-7 or MDA-MB-231 are provided in Fig. 3B and 3D respectively.

The communication between ER and mitochondria is critical for several important cellular functions including the balance between survival and death (39). The role of PDIA1 in the regulation of this communication has been shown in cells in which blocking the PDIA1 activity prevents stimulation of the mitochondrial outer membrane potential (MOMP) and inhibits apoptosis (15). In addition, the combination of ETOP with a PDI inhibitor has been shown to trigger increased ER-associated calcium influx and loss of mitochondrial membrane integrity (40). These findings led us to investigate the effect of PDIA1 on the integrity of the mitochondrial membrane of breast cancer cells.

The mitochondrial membrane potential was investigated in MCF-7 and MDA-MB-231 cells in which the PDIA1 expression had been silenced and cells were left untreated or treated with IFN-γ or ETOP. Silencing of PDIA1 in MCF-7 cells downregulated mitochondrial membrane potential in the untreated cells compared to the scramble-transfected cells under the same conditions (Fig. 4A, compare bar 1 to bar 2). Increased mitochondrial membrane potential was observed in the presence of PDIA1 in MCF-7 cells treated with IFN-γ compared to the untreated cells (Fig. 4A, compare bar 3 to bar 1), whereas decreased mitochondrial membrane disruption was measured in the presence of PDIA1 in MCF-7 cells treated with ETOP compared to the untreated cells (Fig. 4A, compare bar 5 to bar 1). PDIA1-silenced MCF-7 cells treated with IFN-γ exhibited increased mitochondrial membrane disruption compared to PDIA1-silenced untreated cells (Fig. 4A, compare bar 4 to bar 2). Silencing of PDIA1 in MDA-MB-231 cells upregulated mitochondrial membrane disruption in ETOP-treated cells compared to the scramble-transfected cells under the same conditions (Fig. 4C, compare bar 5 to bar 6). Increased mitochondrial membrane disruption levels were also measured in MDA-MB-231 cells treated with IFN-γ in the presence of PDIA1 compared to the untreated cells (Fig. 4C, compare bar 3 to bar 1). In ETOP-treated PDIA1-silenced MDA-MB-231 cells the mitochondrial membrane disruption was upregulated compared to the untreated cells under these conditions (Fig. 4C, compare bar 6 to bar 2). Representative histograms showing the intensity of JC-1 (%) in scramble or siRNA targeting PDIA1-transfected MCF-7 and MDA-MB-231 cells are provided in Fig. 4B and 4D, respectively.

The communication between ER and mitochondria by a network of proteins residing in the interface between the two organelles is critical in controlling vital physiological functions including energy metabolism and cellular death/survival decisions (41). Structural and functional changes of the ER-mitochondria contact proteins result in the deregulation of calcium (Ca²⁺) homeostasis and consequently mitochondrial energy production and cell death (42). PDI plays an important role in the cross talk between cellular redox state and Ca²⁺ homeostasis in the ER (43) suggesting its involvement in the regulation of energy metabolism (44).

To test this hypothesis ATP production was investigated in MCF-7 and MDA-MB-231 cells treated with IFN-γ or ETOP in the presence or absence of PDIA1 (Fig. 5). Silencing of PDIA1 in MCF-7 cells upregulated ATP levels in the untreated cells (Fig. 5A, compare bar 1 to bar 2). Decreased ATP levels were recorded in MCF-7 cells treated with IFN-γ or ETOP in the presence of PDIA1 compared to the untreated cells (Fig. 5A, compare bars 3 and 5 to bar 1). Decreased ATP levels were also observed in PDIA1-silenced MCF-7 cells treated with IFN-γ (Fig. 5A, compare bar 3 to bar 4) or ETOP compared to scramble-transfected cells under the same conditions (Fig. 5A, compare bar 5 to bar 6). Overall, no significant changes in ATP levels were evident in untreated, IFN-γ- or ETOP-treated PDIA1-silenced MCF-7 cells compared to those measured in the untreated, IFN-γ or ETOP treated cells in the presence of PDIA1 (Fig. 5B)

Decreased ATP levels were measured in MDA-MB-231 cells treated with IFN-γ or ETOP in the presence of PDIA1 compared to the untreated cells (Fig. 5C, compare bars 3 and 5 to bar 1). Silencing of PDIA1 in MDA-MB-231 cells downregulated ATP levels in the untreated cells (Fig. 5C, compare bar 1 to bar 2), IFN-γ-treated cells (Fig. 5C, compare bar 3 to bar 4), and ETOP-treated cells (Fig. 5C, compare bar 5 to bar 6). ATP generation was downregulated significantly in untreated, IFN-γ- or ETOP-treated MDA-MB-231 cells in the absence of PDIA1 vs. the presence of PDIA1 (Fig. 5D).

In recent years, the connection between ER stress signalling pathways, UPR induction, deregulation of energy metabolism, and immune responses has been unravelled (45,46). The link between the cellular redox state with several ER-resident chaperones and in particular PDIA1 in mediating the processing, optimal selection and antigen loading to the MHC class I during the process of antigen presentation have been demonstrated (23,24,47). Apart from the peptide antigen processing, loading and stabilization of the early MHC class I complex (22), PDIA1 has also been shown to modulate the MHC class I expression (48). Two types of the MHC class I molecules have been described; the classical (human leukocyte antigen HLA-A, HLA-B, and HLA-C alleles) and the non-classical (HLA-E, HLA-F, HLA-G) proteins (49). Human breast cancer tissues and breast cancer cells express HLA-G whereas in normal epithelial mammary cells HLA-G mRNA expression has not been detected indicating the involvement of the non-classical MHC class I molecules in the evasion of the immune surveillance by breast cancer cells and their important role in determining the prognosis of breast cancer patients (50–52).

To investigate any potential role of PDIA1 in modulating the surface expression of the non-classical MHC class I HLA-G molecule and thus in the process of evasion of immune surveillance (51,52) the HLA-G surface levels were followed in untreated MCF-7 and MDA-MB-231 cells or treated with IFN-γ or ETOP in the presence or absence of PDIA1 (Fig. 6). Significantly lower HLA-G surface levels were observed in both MCF-7 and MDA-MB-231 cells in the absence of PDIA1 compared to MCF-7 and MDA-MB-231 cells in the presence of PDIA1 (Fig. 6A and B).

To explore the correlation between PDIA1 and HLA-G gene expression and identify any potential clinical implications of the PDIA1 and HLA-G ratio in breast cancer patient data obtained from the METABRIC dataset available in the cBio Cancer Genomics Portal (http://cbioportal.org) (26) were analyzed to follow the PDIA1 and HLA-G mRNA levels in various stages of ERα-positive and ERα-negative breast cancer patients. Table I indicates the number of ERα-positive and ERα-negative patients and the stage classification of their disease. Results shown in Fig. S1 indicate the PDIA1 and HLA-G mRNA expression levels in ERα-positive (Fig. S1A and B) and ERα-negative (Fig. S1C and D) patients in each stage of the disease. The correlation between PDIA1 and HLA-G mRNA levels in ERα-positive stage 1 (Fig. S2A), stage 2 (Fig. S2B) and stage 3 (Fig. S2C) patients indicates statistically significant results for stage 2 and 3 patients. A similar analysis of the correlation between PDIA1 and HLA-G mRNA levels in ERα-negative stage 1 (Fig. S2D), stage 2 (Fig. S2E) and stage 3 (Fig. S2F) patients did not show any statistical significance. Kaplan-Meir survival curves were plotted to explore whether the PDIA1/HLA-G mRNA ratio was associated with the overall survival of stage 2 breast cancer patients (Fig. 7). Statistically significant association between low PDIA1/high HLA-G mRNA ratio and longer overall survival was observed in ERα-negative stage 2 patients (Fig. 7D). Analysis of the subgroup of the living breast cancer patients at stage 2 exhibited low PDIA1 and high HLA-G mRNA ratio indicated that in the ERα-positive patients longer survival is associated predominantly with high PDIA1 and low HLA-G mRNA levels (Fig. 7B) whereas in the ERα-negative patients longer survival is associated mainly with low PDIA1 and high HLA-G mRNA levels (Fig. 7E). A similar analysis carried out in the subgroup of the living breast cancer patients at stage 2 exhibiting high PDIA1 and low HLA-G mRNA levels indicated that longer survival was associated mainly with high PDIA1 and low HLA-G mRNA levels in the ERα-negative patients (Fig. 7F) but there was no correlation in the ERα-positive patients (Fig. 7C).