Increased expression of PLK1 suppresses HR in the TCGAdataset.
Before investigating the effect of PLK1 expression on HR, we examined PLK1 expression patterns in various cancers using the TCGA dataset. PLK1 expression patterns vary depending on the type of cancer (Fig. 1). This finding suggests that the threshold of PLK1 overexpression in one type of malignancy may not be the same as overexpression in other tumor types. Consequently, we investigated the correlation between PLK1 expression and HRD.
We analyzed the correlation between PLK1 expression and HRD in malignancies associated with shorter survival when PLK1 is overexpressed 18. The Pearson correlation coefficient was used to analyze the correlation between PLK1 expression levels and HRD scores. Notably, there was a positive correlation between PLK1 expression and HRD scores, indicating that increased PLK1 expression leads to HR inhibition (Fig. 2). A statistically significant positive correlation was observed in most cancer types, except for colorectal and cervical cancers. However, these exceptions may be due to the limited sample sizes, so we further analyzed pan-cancer data to expand our understanding. In this analysis, we saw that increased PLK1 expression levels were again linked to high HRD scores (Fig. 2).
The impact of PLK1 expression on HR and the sensitivity to PARP inhibition in the CCLEdataset.
To ensure the accuracy and reliability of the results obtained from the TCGA dataset analyses, we also examined the relationship between PLK1 expression and HRD score in the Cancer Cell Line Encyclopedia (CCLE) dataset. The CCLE dataset comprises genetic information from 1,739 cancer cell lines 19. PLK1 expression and HRD score data are available for 930 cancer cell lines in the CCLE dataset 20. Using the Pearson correlation coefficient analysis, we found a positive correlation between PLK1 expression and HRD score (Fig. 3A).
The CCLE dataset also includes drug sensitivity data. For 566 cell lines, both PLK1 expression and sensitivity to Olaparib, a commonly used PARP inhibitor in clinical settings are available. Given that HR defects result in increased sensitivity to PARP inhibition, we examined the effect of PLK1 expression on sensitivity to PARP inhibitors. In the analysis of these 566 cell lines, there is a negative correlation between PLK1 expression and Olaparib Ic50 value. This suggests that increased PLK1 expression is associated with heightened sensitivity to the PARP inhibitor (Fig. 3B).
Increased expression of PLK1 suppresses HR in cell-based experiments.
To validate the reproducibility of the bioinformatics results, we performed cell line-based experiments. For homologous recombination to occur, Rad51 must be targeted to DNA damage sites by BRCA2 21. Therefore, we investigated whether Rad51 focus formation is impaired in cells with overexpression of PLK1. We created PLK1-stably expressing cells in the U2OS cell line (Fig. 4A). PLK1 expression does not influence the induction of γH2AX, a marker of DNA double-strand breaks, following PARP inhibitor treatment (Fig. 4B). However, cells overexpressing PLK1 show a marked decrease in Rad51 focus formation (Fig. 4C) under the same conditions. These findings indicate that while PLK1 overexpression does not interfere with the generation of DNA double-strand breaks or the subsequent cellular signalling pathways leading to γH2AX focus formation, it does inhibit homologous recombination 22.
The efficiency of HR has traditionally been measured using the DR-GFP system, where HR restores a functional promoter followed by a GFP sequence in response to an I-SceI restriction enzyme treatment. This results in a detectable GFP signal in HR-proficient cells 23. However, it is known that the DR-GFP system often does not correlate well with sensitivity to genotoxic agents, including PARP inhibitors, in cells with certain gene mutations, such as BRCA1 missense mutations 24.
Recently, a new system called ASHRA has been developed to measure the integration of a nucleic acid sequence of interest, such as a GFP sequence, into an endogenous locus using the CRISPR/Cas9 system. The efficiency of HR measured with ASHRA is highly correlated with cellular sensitivity to genotoxic agents, suggesting that ASHRA is the best method for this study, which investigates the effect of PLK1 overexpression on both HR efficiency and cellular sensitivity to PARP inhibition 24. We used a donor sequence to integrate the GFP sequence into the ß-actin genome (ACTB) using the CRISPR/Cas9 nuclease, targeting the ß-actin locus. This system was designed to create a fusion transcript of ß-actin followed by in-frame integration of the GFP sequence in HR-proficient cells. Both parental and vector-integrating cells show the induction of the fusion transcript of β-actin-GFP with gRNA targeting ACTB, indicating successful integration of the GFP sequence into the β-actin genome, as compared to samples expressing non-target scramble gRNA (Fig. 4D). In contrast, PLK1-overexpressing cells do not exhibit the induction of β-actin-GFP production with either ACTB-targeting or non-targeting gRNA (Fig. 4D). This suggests that the overexpression of PLK1 suppresses HR. It is known that cells with impaired HR are sensitive to PARP inhibition 13. Strikingly, U2OS cells overexpressing PLK1 also display increased sensitivity to PARP inhibition (Fig. 4E).
PLK1 expression correlates with increased sensitivity to PARP inhibitor ex-vivo, in clinical samples of ovarian cancer.
To expand the clinical significance of our findings, we assessed the relationshipbetween PLK1 levels and PARP inhibitor sensitivity in clinical samples, using an ex vivo drug sensitivity analysis method 25,26. Fresh biopsies of ovarian cancers (n = 18) were obtained from participating patients at St. Marianna University Hospital, between September 2012 and August 2014 (Table 1). We dissociated ovarian cancer cells through collagenase treatment of the biopsy sample and subjected them to drug sensitivity analysis in vitro 27. In parallel, we also performed immunostaining in these samples to measure the protein expression level of PLK1. PLK1 expression level was measured by a H-score, a scoring system widely used in medicine 28 (Fig. 5A).
Table 1
Age | Median (range) | 54 (41–77) |
Pathological Stage | I | 6 |
| II | 0 |
| III | 5 |
| IV | 7 |
Pathological diagnosis | Serous adenocarcinoma | 8 |
| Mucinous adenocarcinoma | 1 |
| Clear cell carcinoma | 3 |
| Endometrioid adenocarcinoma | 4 |
| Carcinosarcoma | 2 |
There was a negative correlation between PLK1 expression and sensitivity to PARP inhibitor treatment ex vivo (Fig. 5B). In this analysis, Spearman’s correlation was used due to a limited sample number, and the correlation was statistically significant. Moreover, samples with high PLK1 expression, defined by expression levels above the median, were sensitive to PARP inhibition (Fig. 5C). The effect of PLK1 expression levels remained significant in a multivariate analysis using a Gaussian model to account for other possible factors like age, staging or pathological diagnosis (Table 2).
Table 2
Gaussian multivariate analysis for clinical samples
| crude | | adjusted |
| OR | p-value | | OR | p-value |
PLK1 expression | 1.04E-05 | *1.04E-05 | | 2.6E-08 | *1.56E-02 |
Age | 5.27E + 06 | 4.53E-01 | | 1.55E + 06 | 4.33E-01 |
Stage | 8.78E + 22 | 5.73E-01 | | 2.40E + 55 | 8.14E-01 |
Pathology | 1.14E + 19 | 6.22E-01 | | 2.15E-65 | 1.60E-01 |
OR: odds ratio |
C.I.: confidence interval |
Sensitivity (Ic50), PLK1 expression, age and stage were evaluated as a continuous variable. |