Altered expression of anti-apoptotic protein Api5 affects breast tumorigenesis

Using the GENT2 website, the subtype-based expression profile was plotted by selecting the subtype profile tab and providing “Api5” as the gene symbol. The survival plot is presented as provided by the tool with the overall survival of the patients. TCGA data (Legacy dataset) was downloaded from Xena Browser (UCSC Xena) by selecting IlluminaHiSeq dataset along with clinical data. Graph Pad Prism (Graph Pad Software, La Jolla, CA, USA) is used for plotting the extracted data. Further Kaplan Meier analysis was carried out using the kmplot online tool. Under breast cancer, “API5” was given as a gene symbol and “201687_s at” dataset was selected and using median cut-off, the survival plot was generated. Mutation data were analysed using the TCGA portal online tool. TCGA BRCA was selected as a dataset, and Api5 was given as a gene symbol.

The MCF10A cell line was a generous gift from Prof. Raymond C. Stevens (The Scripps Research Institute, La Jolla, CA), while MCF10AT1 and MCF10CA1a were purchased from ATCC. These cells were grown as described earlier [1]. HEK 293 T cell line was a generous gift from Dr Jomon Joseph (National Centre for Cell Science, Pune, India). The cells were grown in DMEM with high glucose and sodium pyruvate (Invitrogen) containing 10% foetal bovine serum (Invitrogen), and 100 units/ml penicillin–streptomycin (Invitrogen). Stable cell lines overexpressing Api5 was prepared using lentiviral-mediated transduction. Briefly, HEK293T cells were transfected with 1 µg CSII-EF MCS mCherry Api5 vector having a mCherry-Api5 sequence, along with 0.5 µg pCMV-VSV-G-RSV-Rev and 1 µg pCAG-HIVgp for viral particle preparation using Lipofectamine 2000 (Invitrogen)-mediated transfection. Opti-MEM® used for transfection was obtained from Invitrogen. DMEM containing 15% horse serum was added to the cells 24 h post transfection. 48 h post transfection, viral supernatant was collected and filtered through a 0.45 µm filter to remove cell debris. The viruses were then used to transduce MCF10A. 4 µg polybrene was added to the cells to increase the transduction efficiency. As control, a stable cell line expressing only mCherry was also prepared. The transduced cells were sorted using BD FACS Aria (BD Biosciences) to get a pure population with maximum number of transduced cells. Api5 KD stable cells were generated in MCF10AT1 and MCF10CA1a in a similar manner. The shRNA was cloned in pLKO.1-EGFP vector, and packaging plasmids pMD2.G and pPax2 were used for lentiviral preparation. Cells were then sorted in BD FACS Aria.

3D breast acinar cultures were set up in an 8-well chamber cover glass plates (Nunc Lab-Tek, Thermo Fisher Scientific) or 12-well plates (Eppendorf) using standard protocols [14, 15]. Cultures were grown in a humidified incubator with 5% CO2 and maintained at 37 °C (Eppendorf). The medium was changed every four days. For lysate collection on different days, higher cell density was seeded for Day 4 and Day 8.

For dissociation of acinar culture, Dispase™ (Corning, Sigma-Aldrich) was used. After the addition of Dispase™, the culture was incubated for 20 min. The dislodged acini were spun down at 900 rpm for 10 min, followed by 2 rounds of 1X PBS wash before plating in 12-well petri plates.

3D spheroid cultures were immune-stained with specific antibodies using established protocols [15]. For MCF10CA1a and MCF10AT1 spheroid staining, the 1X PBS and 1X IF buffer washes were added with 0.5% Triton X to aid for better penetration of antibodies. Images were captured using Leica SP8 or Zeiss LSM 710 laser scanning confocal microscope.

Lysates from 2D or 3D cultures were collected in lysis buffer containing 50 mM Tris–HCl, pH 7.4, 0.1% Triton X-100, 5 mM EDTA, 250 mM NaCl, 50 mM NaF, 0.1 mM Na₃VO₄ and protease inhibitors. Immunoblotting against specific proteins was performed as per established protocols [16]. Images were acquired in ImageQuant LAS4000 (Cytiva, USA). The blot images represented shows all the bands that were captured using the imaging system. The entire blots were cut as per experimental requirements and probed for different antibodies as our model provides limited protein load and our blotting setup cannot carry out multiple antibody probing together. There are no images where two or more blots were merged.

Formalin-fixed and paraffin-embedded breast cancer patient samples were collected from Prashanti Cancer Care Mission, Pune. Staining was carried out as per standard protocol [17]. The analysis was carried out by observing slides at 20X magnification of the compound microscope.

H-score for IHC is calculated as follows:

The intensity score and percentage positivity score ranges from 0 to 3, and the maximum H score is 9. Three individuals calculated the scores independently by observing ten different positions on the slide corresponding to the tissue ID. The median value from this observation was plotted on the graph.

Single cell migration analysis was performed as described earlier [18]. The data was processed using Fasttracks software [19] (FastTracks (https://​www.​mathworks.​com/​matlabcentral/​fileexchange/​66034-fasttracks), MATLAB Central File Exchange. Retrieved June 1, 2021) and parameters such as speed, displacement, distance, and persistence were calculated and plotted using Graph Pad Prism (Graph Pad Software, La Jolla, CA, USA).

Soft agar colony formation assay was performed as detailed earlier [15]. Images were acquired using the 10X objective of a Nikon Eclipse TS-100 microscope. Ten randomly selected fields were imaged, and the colonies were manually counted.

6 × 10⁶ cells were injected subcutaneously in the flanks of athymic mice (Foxnl ^nu / Foxnl ^nu, 6–8 weeks old) mixed with 1:1 diluted Matrigel® PBS mixture. Tumour size was measured with a vernier calliper. Furthermore, after 8–12 weeks, mice were sacrificed, and the tumour was dissected out. The tumour was then fixed with 10% formaldehyde and embedded in paraffin wax. The tumour area was calculated by multiplying the major and minor axis of the tumour as measured using a vernier calliper. The study is reported in accordance with the ARRIVE guidelines.

Different morphometric parameters were tested for significance using the Mann–Whitney test. The significance test for the in-silico data was performed using either Mann–Whitney or Kruskal- Wallis test. The Mann–Whitney U-test was used to analyse the statistical significance of the relative Golgi area changes and the relative fluorescence intensity following immunostaining in the 3D cultures. The different parameters for analysing cell migration was tested for significance using the Mann–Whitney test. Statistical analysis for mice tumour area was performed using the 2-way ANOVA followed by Sidak’s multiple comparison test. P < 0.05 was considered statistically significant. Graph Pad Prism software (Graph Pad Software, La Jolla, CA, USA) was used to analyse data.

To investigate the expression pattern of API5 in breast cancers, in silico analysis was performed to compare API5 expression between normal and tumour tissues of the breast using the GENT2 database [20]. The plot shows Api5 transcript levels are up-regulated in breast cancer tissue samples when compared to normal breast tissues (Fig. 1A and B). Further Kaplan–Meier survival analysis demonstrated poor survival of patients with high Api5 expression (Fig. 1C).

When similar analyses were performed using the TCGA dataset, the basal subtype showed higher expression of API5 when compared to the other molecular subtypes and adjacent normal tissues (Fig. 1D-G).

Survival analysis in KM plotter [21] suggested that higher API5 expression is associated with poor breast cancer patient survival (Fig. 1H). Using TCGA portal [22] online tool, we found higher Api5 expression is associated with mutations in PIK3CA (PI3-kinase catalytic subunit alpha), TP53 and CDH1 (E-cadherin), suggesting the possible deregulation of critical pathways associated with high API5 levels (Fig. 1I).

Furthermore, immunohistochemical analyses on breast cancer tissue samples showed that tumour tissues showed significantly higher Api5 expression when compared to the adjacent normal or reduction mammoplasty tissues. Also, Api5 expression levels were high in Stage 2 breast cancer samples in comparison to Stage 1 (Additional file 1A-D).

Elevated expression of Api5 in tumour tissues supports the possibility that Api5 may be a tumour promoter in breast malignancies. To investigate this further, we decided to overexpress Api5 in a non-tumorigenic breast epithelial cell line and study its effects on acinar phenotypes.

To study the effect of Api5 overexpression on the cellular transformation of the breast acini, Api5 was overexpressed in MCF10A cells (will be called Api5 OE henceforth) (Additional file 2A). Following 16 days of culturing, it was observed that Api5 OE acini (Fig. 2A) have a larger surface area (Fig. 2B), volume (Fig. 2C), increase in the number of cells per acini (Fig. 2D) and filled lumen phenotype (Fig. 2E). There was no difference in the sphericity of Api5 OE acini when compared to the control acini (Additional file 2B). Similarly, Api5 overexpression did not produce protrusion-like structures in the 3D cultures (Additional file 2C). Api5 OE acini were not growth-arrested by day 16 and continued to proliferate as increased Ki67 protein expression was observed when compared to the control (Fig. 2F-G). More than 80% of Api5 OE acini showed greater than six Ki67 positive nuclei per acini (6 cells per acini = approx. 33% of cell/acini) as shown in Fig. 2H-I. On checking for the expression of PCNA, another proliferation marker, in the Api5 OE acini, an increase in PCNA levels was also observed in Api5 OE when compared to the control (Fig. 2J-K), thus further confirming that Api5 overexpression leads to increased proliferation. To investigate whether Api5 OE resulted in cellular transformation, anchorage-independent growth of Api5 OE cells dissociated from the 16-day acini was investigated. Api5 OE cells formed colonies on soft agar (Fig. 2L), indicating that overexpression of Api5 leads to the transformation of non-tumorigenic breast epithelial cells grown as acinar cultures.

The transformation of epithelial cells can also lead to increased migratory potential. When single-cell migration analysis was carried out, a significant increase in the distance travelled (Fig. 2M), velocity (Fig. 2N), decrease in persistence (Fig. 2O) while no change in displacement (Fig. 2P) was observed in 3D dissociated Api5 OE cells when compared to the controls. The track data was further used for plotting the cellular movement on an XY plot using ImageJ (Fig. 2Q). The tracking parameters suggest that Api5 OE cells move faster and cover a longer distance than the control. Further Api5 OE cells showed lower persistence suggesting that the cells were moving in a zigzag fashion rather than taking a relatively straight direction as was observed in the control cells. To understand the transformative potential of Api5, Api5 OE MCF10A cells dissociated from the acinar cultures were injected subcutaneously into the flanks of athymic mice and followed for eight weeks to check their in vivo tumorigenic potential. The Api5 OE cells did not form tumours (data not shown) suggesting that overexpression of Api5 is capable of partially transforming breast epithelial cells grown as acinar cultures with diminished tumorigenic potential.

Further, we studied the effect of Api5 OE on polarity in 3D acinar cultures. Api5 OE cells were cultured for 16 days and immunostained for the different polarity markers. α6-integrin and Laminin V, which marks the basal region of the acini [14] were observed to be mis-localised in around 80% of the Api5 OE acini compared to the control (Fig. 3A - D). Similarly, 83% of the Api5 OE acini had mis-localised GM130 (Fig. 3E and F), where GM130 was not observed to be apically positioned to the cell nucleus [14]. In addition, 63% of the Api5 OE acini showed loss of E-cadherin at cell–cell junctions (Fig. 3G and H). Interestingly, β-catenin, another cell–cell junction marker remained unaffected in the Api5 overexpressed acinar cultures (Additional file 2D and E). Taken together, our data suggest that overexpression of Api5 leads to polarity disruption in the breast acinar cultures, which can be associated with oncogene-mediated transformation. These results indicate that Api5 overexpression results in several characteristic changes in the epithelial cell line, possibly through epithelial to mesenchymal transition (EMT).

To study whether overexpression of Api5 leads to EMT, Api5 OE cells were cultured for 16 days, lysates collected, and immunoblotted. Vimentin, twist, slug and fibronectin showed 1.5-fold upregulation in Api5 OE acini when compared to the control (Fig. 4A and Additional file 3A-D). However, there was no change in β-catenin and N-cadherin expression levels (Fig. 4A and Additional file 3E and F). Interestingly the epithelial marker, E-cadherin showed a twofold increase in Api5 OE cells in comparison to the control (Fig. 4A, and Additional file 3G). This was confirmed by calculating the corrected fluorescence intensity of E-cadherin in day 16 acini (Fig. 4C) where a significant increase in the fluorescence intensity was observed, thereby suggesting an increase in E-cadherin levels. To further investigate whether the EMT-like phenotype that was observed upon overexpression of Api5 was a transient phenomenon or not, cells dissociated from the 16-day breast acinar cultures were grown as monolayer cultures and analysed. Similar to the phenotypes observed in the acinar cultures, an increase in vimentin, slug, fibronectin and E-cadherin were observed in the Api5 OE dissociated cells (Fig. 4B and Additional file 3H-K). N-cadherin expression remained unaltered in the Api5 OE dissociated cells (Fig. 4B and Additional file 3L). Both the epithelial cell markers cytokeratin 14 and 19 showed differential regulation in the Api5 OE dissociated cells. Cytokeratin 14 was downregulated while cytokeratin 19 was up-regulated (Fig. 4B and Additional file 3 M and N). The upregulation of vimentin was further confirmed when the 16-day Api5 OE breast acini, as well as the dissociated cells, were immunostained for vimentin and increased fluorescence intensity was demonstrated for the mesenchymal marker (Fig. 4D-F and Additional file 3O). Thus, taken together our data suggests that overexpression of Api5 leads to a partial EMT-like phenotype in the breast epithelial cells.

Glandular epithelial structures like breast acini have intact Golgi localised towards the apical region while some transformed or cancerous cells have dispersed Golgi [23]. Api5 OE cells were dissociated from 3D cultures and immunostained for the cis-Golgi marker GM130. It was observed that Api5 OE dissociated cells showed a significant increase in Golgi area when compared to the control, thereby suggesting an aberrant Golgi phenotype (Fig. 4G and H). Api5 OE in MCF10A cells grown as acinar cultures resulted in significant characteristic changes in the epithelial cell line thus resulting in cellular transformation.

Since overexpression of Api5 was leading to a transformed phenotype, we wanted to explore whether down-regulation of Api5 could alter the cancerous phenotype in pre-malignant and malignant breast cancer cells. On investigating the protein expression of Api5 in the MCF10 cell line series [24‐26], Api5 protein expression was observed to be up-regulated in the malignant MCF10CA1a cells in comparison to the non-tumorigenic MCF10A or pre-malignant MCF10AT1 cells (Fig. 5A-B). shApi5 knock-down stable cells were prepared in both MCF10AT1 and MCF10CA1a (Additional file 4A and G). When Api5 KD MCF10CA1a cells were cultured for 8 days, they formed smaller spheroids as was observed by a significant reduction in the surface area and volume when compared to MCF10CA1a control (Fig. 5C-E). This reduction in size was further corroborated with a significant reduction in the number of cells forming the Api5 KD MCF10CA1a spheroids (Fig. 5C and F). Interestingly, 58% of Api5 KD MCF10CA1a spheroids showed a reduction in proliferation as was analysed where less than 50% cells per spheroid were Ki67 positive (Fig. 5G and Additional file 4B). The levels of PCNA was also reduced upon knockdown of Api5 (Fig. 5H and Additional file 4C). Furthermore, Api5 KD MCF10CA1a cells formed fewer colonies on soft agar thereby indicating that a reduction in Api5 expression negatively affected anchorage-independent growth (Fig. 5I and Additional file 4D). When these Api5 KD MCF10CA1a cells were injected into the flanks of athymic mice they formed tumours, although the tumour volume was significantly reduced compared to the control MCF10CA1a cells. It was observed that while the MCF10CA1a control tumours continued to grow, Api5 KD MCF10CA1a tumours stopped growing and maintained a size similar to that of week 2 tumours (Fig. 5J and K, Additional file 4E and F).

Similarly, knock-down of Api5 in the pre-malignant cell line MCF10AT1 also led to a reduction in the surface area and volume of the spheroids (Fig. 5L-N). 80% of spheroids formed by Api5 KD MCF10AT1 had less than 100 cells per spheroid (Fig. 5O). Interestingly, knockdown of Api5 in MCF10AT1 spheroids did not affect proliferation as there was similar Ki67 expression in control and Api5 KD cells (Additional file 4H-I). Similar to the results observed in Api5 KD MCF10CA1a cells, knock-down of Api5 in the MCF10AT1 cells also resulted in a reduced ability to grow on soft agar suggesting a partial reversal of the malignant phenotype (Additional file 4 J-K).

To further understand the importance of Api5 in breast carcinogenesis, it is essential to delineate the molecular signalling involved in the various processes. Initially, we investigated the effect of Api5 OE on apoptosis. Overexpression of Api5 led to a reduction in Bim, and active caspase 9 on day 12 of morphogenesis (Fig. 6A-C). Since ERK-mediated Bim degradation has been reported to be through Api5 signalling [5], we checked whether overexpression of Api5 could lead to an alteration in ERK and MEK activity. Phosphorylation of ERK and MEK kinases were observed upon Api5 OE during day 12 of acinar growth (Fig. 6A, D and E). Also, an increase in FGF2 protein expression was observed from days 4 to 12 in Api5 OE acini when compared to the control. This increase was a little over threefold on day 4 while it was 1.5-fold on day 12. By day 16, FGF2 was marginally lower in Api5 OE cells compared to the control (Fig. 6A and F). Earlier studies have reported API5-induced immune resistance to be dependent on FGF2-mediated activation of FGFR1 receptor [5]. Similarly, in our study, we observed increased FGFR1 phosphorylation on days 4 and 8 that also coincided with higher FGF2 levels (Fig. 6A and G). Although an increase in FGF2 levels was observed on day 4, an increase in ERK signalling was observed only from day 12. FGF2 is known to activate tyrosine kinase receptors that then activate the PI3K signalling cascade [27]. Overexpression of Api5 increased phosphorylation of Akt at T308 on days 4 and 8, while the Akt S473 phosphorylation was similar to that of the control cells (Fig. 6A, H and Additional file 5A). Therefore, it can be inferred that PDK1, which phosphorylates Akt at T308, is activated, which may lead to further activation of its downstream signalling molecule, cMYC. Upon probing for cMYC in these lysates, a significant increase in cMYC expression was observed in days 4 and 8 Api5 OE lysates, which coincided with the phosphorylation of Akt at T308 residue (Fig. 6A and I). Further studies confirmed PDK1 to get activated on day 4 in Api5 overexpressing cells which coincided with both phosphorylation of Akt at T308 residue and cMYC upregulation (Fig. 6A and J). Thus, the FGF2-mediated activation of PDK1-Akt/cMYC signalling during the initial days of growth supports the increased proliferation, protein synthesis and transformation of MCF10A cells, while later activation of the ERK signalling cascade leads to reduced apoptosis and lumen filling. FGF2- PDK1/ ERK signalling is very often linked through RAS and a recent study has shown that Api5 interacts with KRAS [28], thus KRAS expression pattern was also studied. Interestingly, KRAS was up-regulated in Api5 OE 3D lysates from day 4 to day 12 and by day 16, KRAS levels were similar to that of the control (Fig. 6A and K). Taken together, we propose Api5 to activate FGF2-mediated Ras-ERK and Akt signalling cascades.

Furthermore, Api5 KD resulted in a decrease in ERK phosphorylation (Fig. 6L and M). Also, reduction in both FGF2 protein and mRNA levels was observed in Api5 KD cells when compared to control (Fig. 6L and N, Additional file 5B and C). Immunoblotting of lysates from 3D spheroid cultures showed decreased FGFR1 activation when Api5 is knocked down in MCF10CA1a cells (Fig. 6L and O). Akt activation was also perturbed upon Api5 KD, as was demonstrated by diminished Akt phosphorylation at both S473 and T308 sites (Fig. 6L and P).

To further investigate the existence of this regulation in breast cancer patients, co-expression analysis was performed using the TCGA database. Interestingly, BAX and CASPASE-9 negatively correlated with API5 transcript levels, but BIM mRNA levels positively correlated with API5 (Additional file 5G-I). We observed a similar trend in the protein expression levels in Api5 OE MCF10A lysates collected on day 16. ERK2, FGF2, PDK1, KRAS and cMYC transcript levels also showed a positive correlation with API5 transcript (Additional file 5 J-M), supporting the data obtained from the 3D acinar cultures. These data indicate that Api5 is regulating the FGF2-mediated PDK1 and ERK signalling in breast cancer and can possibly be used as a target for therapy.