3.1. Evaluation of etoposide activity on in vitro PCa models with different AR functionality
In order to investigate etoposide clinical relevance in PCa treatment, we tested this drug in LNCaP, 22Rv1, PC3, DU145, PDB and MDB cell lines, which were selected among in vitro models that could reflect different PCa progression stages according to AR activity and androgen sensitivity (Table S1).
The treatment of PCa cells for 120h with scalar concentration of etoposide elicited significant dose-dependent inhibition of cell growth for all concentrations tested (Figure 1A and 2B). Etoposide had similar inhibitory effect on LNCaP and 22Rv1 (Rln ≈1), whereas increased cytotoxicity was observed in DU145, PC3, PDB and MDB (Rln <1) (Table 1). The invasion assay was performed in presence of 1mM etoposide for 48h, and we observed that etoposide affected aggressiveness in all resistant cell lines, with an increased effect in the AR-negative PC3 and DU145 cell lines (Figure 1C).
Subsequently, we compared the activity of etoposide to that of other antineoplastic agents used for PCa therapy. The above cell lines were treated with enzalutamide, docetaxel and the PARP1 inhibitor olaparib (Figure 2A and 3B; Table 1). The cytotoxic effect of etoposide was higher compared to the other treatments investigated in almost all cell lines, except for LNCaP and 22Rv1 that showed slightly higher sensitivity to docetaxel compared to etoposide (Ret value 0.92 and 0.77, respectively). Of note, olaparib showed activity irrespective of PARP1 cellular content, which was equally or more expressed in resistant cell lines compared to androgen-responsive LNCaP (Figure 1B). Interestingly, etoposide show higher cytotoxicity compared to the others compounds tested (Ret >1) in DU145, PC3, PDB and MDB cell lines, which had impaired or absent AR (Table S1).
We also assessed the activity of etoposide and of other treatments in PDB and MDB, which are bicalutamide-resistant, LNCaP-derived, cell lines that show significant plasticity after antiandrogen withdrawal[16]. In addition to bicalutamide, the normal culture medium of PDB includes 0.1 nM DHT, whereas MDB grow in absence of DHT. In our experiment, PDB and MDB were grown in absence of bicalutamide and DHT (PDB w/o Bic and MDB w/o Bic, respectively) for 1 or 2 months. The WB analysis of protein extract from these cells after 1 month from withdrawal showed significant modulation of PSA and AR expression (Figure 2A). Both cell lines had increased PSA expression, suggesting augmented AR activity. However, PDB also showed downregulation of AR expression, revealing a possible compensative mechanism to maintain homeostasis following antiandrogen withdrawal.
Using the IC50 ratio of treatments under investigation (calculated as the ratio of IC50 in cells grown without bicalutamide compared to that of cells grown with bicalutamide) (Rbic, Table S2), we found that etoposide, docetaxel and enzalutamide treatments showed reduced cytotoxicity (Rbic<1) 1 month after bicalutamide withdrawal compared to parental cell lines grown in the antiandrogen regimen (Figure 2B).
Table 1. Evaluation of etoposide effect on cell viability in different in vitro PCa models in respect to LNCaP or in comparison with other antineoplastic drugs used for PCa therapy
|
ETOPOSIDE
|
DOCETAXEL
|
ENZALUTAMIDE
|
OLAPARIB
|
Cell Line
|
IC50* (mM)
|
se
|
Rln**
|
Ret***
|
IC50 (nM)
|
se
|
Rln
|
Ret
|
IC50 (mM)
|
se
|
Rln
|
Ret
|
IC50 (mM)
|
se
|
Rln
|
Ret
|
LNCaP
|
9.24
|
0.57
|
1.00
|
1.00
|
8.49
|
0.55
|
1.00
|
0.92
|
20.06
|
0.29
|
1.00
|
2.17
|
13.47
|
0.17
|
1.00
|
1.46
|
22RV1
|
10.31
|
0.82
|
1.12
|
1.00
|
7.96
|
0.30
|
0.94
|
0.77
|
nd
|
nd
|
nd
|
nd
|
24.62
|
3.81
|
1.83
|
2.39
|
DU145
|
5.05
|
0.51
|
0.55
|
1.00
|
8.47
|
0.03
|
1.00
|
1.68
|
56.06
|
0.95
|
2.79
|
11.10
|
16.14
|
1.37
|
1.20
|
3.20
|
PC3
|
7.04
|
0.36
|
0.76
|
1.00
|
11.86
|
0.52
|
1.40
|
1.69
|
74.61
|
9.13
|
3.72
|
10.60
|
15.10
|
0.91
|
1.12
|
2.15
|
PDB (with Bic)
|
5.96
|
0.17
|
0.65
|
1.00
|
8.89
|
1.45
|
1.05
|
1.49
|
22.24
|
1.43
|
1.11
|
3.73
|
16.94
|
2.65
|
1.26
|
2.84
|
MDB (with Bic)
|
7.48
|
0.99
|
0.81
|
1.00
|
26.59
|
4.47
|
3.13
|
3.55
|
31.86
|
2.41
|
1.59
|
4.26
|
16.25
|
1.77
|
1.21
|
2.17
|
* IC50 is the drug concentration required for 50% inhibition in vitro
** Rln =IC50 (Cell Line evaluated)/IC50 LNCaP; Rln<1 treatment more effective than in LNCaP cell line; Rln>1 treatment less effective than in LNCaP; Rln=1 effect equal to LNCaP
*** Ret = IC50 (drug evaluated)/IC50 etoposide; Ret<1 treatment more effective that with etoposide; Ret>1 treatment less effective that with etoposide treatment; Ret=1 effect equal to etoposide treatment
3.2. TOP2A and TOP2B expression in prostate cancer cell lines
Western blot analysis was performed on whole protein extract from LNCaP, 22Rv1, DU145, PC3, MDB and PDB, in order to investigate the expression of TOP2 isoforms in these cell lines (Figure 3). Although the isoform expression of TOP2A and TOP2B was different among these cell lines, the total content of TOP2 was comparable (Figure S3). Full-length TOP2A (TOP2A-FL) was poorly expressed in LNCaP, PDB and MDB cell lines, whereas it was highly expressed in the castration-resistant 22Rv1, DU145 and PC3. Cell lines with very aggressive phenotype (DU145 and PC3) showed significant lower expression of TOP2A proteolytic fragments compared to LNCaP. Of note, etoposide induced significant inhibition of invasion on these cell lines (Figure 1C). A statistically significant overexpression of TOP2B was found in 22Rv1 and PDB compared to LNCAP
3.4. Consistency between TOP2 mRNA and protein expression
In order to investigate the consistency between mRNA and protein expression, we analyzed the data reported in Human Protein Atlas. We found that TOP2A and TOP2B expression profiles, evaluated by immunohistochemistry, were consistent with mRNA levels assessed in 36 and 44 normal tissues, respectively. This feature was also confirmed in the TCGA-Prostate Adenocarcinoma dataset of PCa tissues; despite significant data dispersion from both approaches, mRNA distribution analysis by GEPIA was consistent with immunohistochemistry data from HPA (Figure S4).
3.5. Relevance of TOP2A and TOP2B genetic alterations in mCRPC
Using the mutation and copy-number alterations data provided by cBioPortal, we found that TOP2A and TOP2B genetic alterations were rarely reported in primary PCa, whereas they were more frequently observed in metastatic samples. TOP2A and TOP2B amplifications were the most common alterations and were found in 5% of patients included in the SU2C-Prostate Cancer Foundation (PCF) Dream Team metastatic dataset[20]. In the Neuroendocrine Prostate Cancer (NEPC) dataset, TOP2A and TOP2B amplifications were found in 19% and 20% of patients with adeno-mCRPC, respectively, and in 30% and 27% of those with neuroendocrine-mCRPC[21]. We investigated the relationship between TOP2A and TOP2B amplifications and copy-number variations in genes commonly involved in PCa biology and NEPC differentiation. We identified significant correlations between TOP2A/TOP2B amplifications and AURKA, MYCN, SRRM4, FOXA1 and FOXA2 amplifications (Figure S5). Of note, concurrent TOP2A amplification was found in 3 of 4 NEPC patients with RB1 loss in the NEPC dataset. However, we did not identify any correlation or association between TOP2A or TOP2B amplifications and clinical data. We also found a significant correlation between these amplifications and genomic burden in both datasets[20, 21].
3.6. Relevance of TOP2A and TOP2B mRNA expression in mCRPC
We investigated possible association between TOP2A or TOP2B mRNA expression and clinical data. Insufficient mRNA data were available in the Neuroendocrine Prostate Cancer dataset[21]. Using the data on metastatic tumors from the SU2C/PCF dataset[20], we found a statistically significant association between high TOP2A mRNA expression and poor overall survival (Figure 4A). In the same dataset, we identified a significant correlation between TOP2A mRNA expression and neuroendocrine features, as well as an inverse correlation with AR score (Figure 4C). This observation was consistent with our observation of increased TOP2A expression in aggressive castration-resistant cell lines, such as PC3 and DU145 (Figure 3). No correlation was identified in patient overexpressing TOP2B[20]. We investigated possible associations between TOP2A overexpression and presence of alterations in genes that are commonly altered in metastatic PCa and in NEPC. We identified that patients with higher TOP2A mRNA expression were significantly more likely to harbor RB1 homologous deletions compared to those with lower TOP2A mRNA expression (38.2% vs. 5.5%) (Figure 4B). We also observed that patients with higher TOP2A mRNA expression were less likely to harbor AR amplifications (not statistically significant). In addition, we found that TOP2A mRNA was significantly overexpressed in the 43 metastatic patients with RB1 loss compared to the whole population (q value = 0.007), and RB1 mRNA was significantly underexpressed in TOP2A-overexpressed metastatic PCa (q value < 0.001).
Using GEO2R analysis tool, we observed that the TOP2A mRNA was significantly overexpressed in mCRPC compared to primary PCa and normal tissues (Figure S6). In addition, TOP2A mRNA expression was higher in non-AR driven PCa compared to AR-driven PCa, as well as in small-cell or neuroendocrine PCa compared to adeno-PCa. Regarding TOP2B mRNA values, no substantial differences between primary PCa and mCRPC were found; however, significant lower expression of TOP2B was observed in primary PCa compared to normal tissues and in small-cell or neuroendocrine PCa compared to adeno-PCa.
3.7. TOP2A interaction network and functional enrichment analysis
In order to identify potential interactors of TOP2A, we analyzed the mRNA enrichments in patients with high TOP2A mRNA expression both in primary and metastatic PCa datasets[20, 22]. Separately for each cohort of datasets, we selected the first 150 mRNA with the highest statistically significant enrichment in patients with high TOP2A mRNA compared to patients with lower TOP2A mRNA expression. In patients with TOP2A-overexpressed tumors, we identified an interaction network of 63 mRNA that were significantly co-expressed with TOP2A both in primary and metastatic datasets. This network was visualized by STRING database and showed low protein-protein interaction (PPI) enrichment p-value (<1.0e-16) (Figure 5). The gene ontology enrichment analysis revealed a gene set tightly involved in the cell cycle and mitotic cell processes, such as AURKA, AURKB and Cell Division Cycle Associated gene family.