Introduction
Large tumor suppressor (LATS) family proteins LATS1 and LATS2 have been proposed to be tumor suppressors. They have been reported to govern cellular homeostasis by preventing cell proliferation and migration, by inducing cell death and senescence, and by regulating cell cycle checkpoints to maintain genetic stability (Visser and Yang
2010; Furth and Aylon
2017). Consistent with their proposed tumor-suppressive function, LATS proteins have been reported to be down-regulated in various cell types including breast cancer (Morinaga et al.
2000), non-small lung cancer (Lin et al.
2014), and gastric cancer (Son et al.
2017). However, LATS was reported to be overexpressed in nasopharyngeal cancer (Zhang et al.
2010), suggesting that the function of LATS remains controversial and may be tissue- and cancer-type dependent.
Large tumor suppressor1 and LATS2 are the core kinases of the Hippo pathway, in which LATS1 and LATS2 become activated by upstream kinases in response to stimuli and then phosphorylate and thereby repress the nuclear activity of two transcriptional co-factors YAP and TAZ by their cytoplasmic retention, eventually preventing the transcription of tumor-promoting genes. The Hippo signaling pathway plays a critical role in organogenesis, tumorigenesis, metastasis, stem cell differentiation and renewal, and mechanotransduction (Furth and Aylon
2017; Meng et al.
2016; Janse van Rensburg and Yang
2016; Yu et al.
2015). Accordingly, YAP/TAZ are activated, whereas LATS1/2 are inactivated in many human malignant tumors (Plouffe et al.
2015; Zanconato et al.
2016).
Clinically, LATS expression can have prognostic value. LATS1 and LATS2 expressions have been reported as significant markers for good prognosis in patients with gastric cancer (Son et al.
2017), reduced LATS1 correlated with poor outcome with breast cancer patients (Takahashi et al.
2005), and reduced LATS2 correlated with poor survival in acute lymphoblastic leukemia (Jiménez-Velasco et al.
2005).
The role of LATS in ovarian cancer remains poorly understood. Epithelial ovarian cancer is the leading cause of death in gynecologic cancer. In patients with advanced FIGO stage serous ovarian cancer, the most common, aggressive and deadly type, the 5-year survival rate is < 30%. This poor outcome is due to the lack of early disease-specific symptoms and reliable tools for early diagnosis, as well as ineffective therapy for advanced disease (Ozols
2006; Bowtell et al.
2015). Recently, a study evaluated immunohistochemistry data from Chinese ovarian cancer samples and showed that LATS1 and LATS2 expression was reduced in serous ovarian cancer patients associated with shorter survival and increased recurrence, while LATS1 and LATS2 were highly expressed in fallopian tube and LATS1 was expressed in normal ovarian tissue (Xu et al.
2015).
To validate these previous data, we investigated LATS expression (1) and clinical outcome in ovarian cancer using various publicly accessible transcriptomic data sets and (2) in our Swiss ovarian cancer patient cohort and related the results to various clinicopathological parameters. As LATS expression has also been shown to regulate epithelial-mesenchymal transition (EMT), a cellular program that promotes invasion and metastasis during cancer development, and modulate drug responses (Nozaki et al.
2019; Zhang et al.
2012; Takahashi et al.
2007; Kawahara et al.
2008), we determined LATS protein expression in a panel of ovarian cancer cell lines and investigated (3) whether LATS expression in these cell lines was an indicator for the EMT state of the cells and for the sensitivity to chemotherapeutic drugs.
Materials and methods
Transcriptomic database analysis
Transcriptomic ovarian databases were analyzed using the Bioconductor package “curatedOvarianData” which represents a manually curated data collection for gene expression meta-analysis of patients with ovarian cancer (Ganzfried et al.
2013). The results are presented as hazard ratios (
HR) of
LATS1 and
LATS2 with corresponding 95% CI using forest plots. In addition, TCGA RNASeqV2 data were analyzed in more detail, presenting results from Cox regression using the two quantiles (0.05, 0.95) for
LATS1 and
LATS2.
Patient cohort, tissue microarray immunohistochemistry, and ethical approval
The patient cohort was previously recruited at the University Hospital Zurich, Switzerland from 1990 to 2007 and the respective tissue microarray (TMA) was used for the immunohistochemistry (IHC) experiments (described in: Jacob et al.
2012). Deparaffinized and stained sections were incubated with anti-LATS1/2 antibody (GTX87014; Lucerna Chem, Lucerne, Switzerland) as per standard laboratory and manufacturer’s protocols. This antibody recognizes endogenous levels of total LATS 1/2 protein. Tissue slides were counterstained with hematoxylin. Sections were dehydrated and coverslipped. Immunostaining was scored by the weighted average score (intensity: 0–3, coloring: 0–100% of LATS1/2 expression) by three trained scientists (CM, GS, VHS) independently and discrepancies were resolved by consensus. The cohort consisted of 710 samples on three TMAs from initially 271 patients. There were 200 samples with incomplete data or with non-evaluable scores. 510 samples had evaluable scores, whereof 122 had no LATS expression and 388 had LATS expression. For patients with more than one sample, the scores were averaged. From the resulting 205 patients, a miscellaneous group of 9 patients was excluded due the low patient numbers and unclear malignancies. The final overall cohort consisted of 196 ovarian tumor patients: 85 (benign) cystadenomas, 32 borderline, and 79 carcinoma (Table
1).
Table 1
Overall ovarian tumor cohort description
Tumor type | | | | | 196 |
Cystadenoma | 85 (43.4%) | 85 (100%) | | | |
Borderline | 32 (16.3%) | | 32 (100%) | | |
Carcinoma | 79 (40.3%) | | | 79 (100%) | |
Histological subtype | | | | | 196 |
Clear cell | 14 (7.2%) | | 1 (3.12%) | 13 (16.5%) | |
Endometrioid | 12 (6.1%) | | | 12 (15.2%) | |
Mucinous | 43 (21.9%) | 31 (36.5%) | 10 (31.2%) | 2 (2.53%) | |
Serous | 127 (64.8%) | 54 (63.5%) | 21 (65.6%) | 52 (65.8%) | |
Stage | | | | | 107 |
I + II | 48 (44.9%) | | 25 (83.3%) | 20 (27.0%) | |
III + IV | 59 (55.1%) | | 5 (16.7%) | 54 (73.0%) | |
Grade | | | | | 110 |
1 | 4 (3.6%) | | | 4 (5.1%) | |
2a | 23 (20.9%) | | | 23 (29.5%) | |
3 | 51 (46.4%) | | | 51 (65.4%) | |
Borderline | 32 (29.1%) | | 32 (100%) | | |
Scoreb | 0.25 [0.06;0.63] | 0.10 [0.00;0.26] | 0.73 [0.42;0.98] | 0.38 [0.15;0.66] | 196 |
Medsplitc | | | | | 196 |
Low expression | 100 (51.0%) | 63 (74.1%) | 5 (15.6%) | 32 (40.5%) | |
High expression | 96 (49.0%) | 22 (25.9%) | 27 (84.4%) | 47 (59.5%) | |
Ethics approval was obtained from the Swiss Ethical Cantonal Department SPUK (approval #StV06/2006) and the Ethical Committee of Nordwest- und Zentralschweiz, Switzerland (EKNZ 2015 ± 436). Neither written nor oral consent was necessary for this retrospective study and data accession was anonymous. The whole study was performed according to the Declaration of Helsinki and local laws and regulations.
Cell lines and cell culture
A2780, BG-1, CaOv3, IGROV-1, Kuramocchi, OAW42, OVCAR-3, OVCAR-4, OVCAR-8, SKOV-3, TOV112D and TYK-nu (parental) ovarian cancer cell lines; FT190 and FT194 fallopian tube cell lines; and HOSE 6.3 and HOSE 17.1 human ovary surface epithelial cells were cultured in RPMI (R8758, Sigma-Aldrich, Buchs, Switzerland) supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich) and penicillin/streptomycin (100 U/mL/100 μg/mL; Sigma-Aldrich) at 37 °C in a 95% humidified atmosphere containing 5% CO
2. All cell lines were routinely tested for mycoplasma infection. Cisplatin-resistant A2780/CP and TYK-nu(R) cell lines were obtained from the National Cell Bank of Iran (NCBI) and the JCRB Cell Bank Japan, respectively, and these cell lines had acquired cisplatin resistance generated by stepwise incubation of the parental cells with inclining cisplatin concentrations (Masuda et al.
1988; Yoshiya et al.
1989). Paclitaxel-resistant IGROV1-PXL cells were generated in our laboratory by stepwise exposure of parental IGROV-1 cells to increasing concentrations of paclitaxel (Kohler et al.
2017). They were also cultured as described above.
Western blot analysis
Western blotting was used to determine the protein expression in the cell lines and was performed according to standard laboratory protocols. Briefly, cell lysates were obtained from subconfluent cultures at the time of harvest. Cells were lysed with RIPA buffer (9806, Cell Signaling; BioConcept, Allschwil, Switzerland). Protein concentration was determined by the BCA Protein Assay (23227; Pierce, Perbio Science, Switzerland). Twenty micrograms of protein was loaded and separated using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), followed by blotting onto polyvinylidene difluoride (PVDF) membranes (162-0177, BioRad, Crissier, Switzerland). Proteins were detected with specific primary antibodies and appropriate secondary antibody (HRPO-conjugated anti-mouse (7076, Cell Signaling) or HRPO-conjugated anti-rabbit (7074, Cell Signaling). The primary antibodies were rabbit anti-LATS1 (9153, Cell Signaling), rabbit anti-LATS2 (ab110780, Abcam, Lucerne Chem), rabbit anti-E-cadherin (3195, Cell Signaling), mouse anti-vimentin (MA5-11883, Invitrogen, Thermo Fisher Scientific, Wohlen, Switzerland), and mouse anti-MDR1 (sc-13131, Santa Cruz, Lab Force, Muttenz, Switzerland) antibodies. Rabbit anti-tubulin antibody (2148, Cell Signaling) was used as a sample loading control. Complexes were visualized by enhanced chemiluminescence (Dura West, Pierce, Perbio Science) and autoradiography. Quantitative analysis of the complexes (intensity on autoradiogram) was performed by densitometry (normalized against tubulin) using Image J software.
MTT assay
The MTT assay was used to determine the sensitivity of the cell lines to chemotherapeutic drugs and was performed as follows: cells were seeded into 96-well plates and treated on the next day with chemotherapeutic drugs for 72 h, followed by the addition of 20 μl of MTT dye (Sigma-Aldrich) in PBS (final concentration: 0.5 mg/ml). Chemotherapeutics were purchased from the following suppliers: carboplatin (Labatec SA, Geneva, Switzerland) and cisplatin, doxorubicin, and paclitaxel (Sigma-Aldrich). After 3 h, the medium was removed and the purple crystals were dissolved in 200 μl DMSO. Optical density (absorbance at 540 nm) was measured with a SynergyH1 Hybrid Reader (Biotek, Zurich, Switzerland). Data (mean ± SD of at least four independent experiments performed in quadruplets) are presented as function of drug concentration. IC50 values were calculated by linear extrapolation.
Statistical analysis
Descriptive statistics is presented as median (IQR) or counts and percentages as appropriate. To compare weighted average scores between subgroups, Dunn’s test following Kruskal–Wallis tests was performed. To analyze the influence of LATS1 and LATS2 on time-to-event data, the expression values were split by medians providing high/low groups. Kaplan–Meier statistics were presented as graphs and median times with 95% confidence intervals. Time to event curves were compared using Log-rank tests presented as
p values: a
p value < 0.05 is considered as statistically significant. All evaluations were done using the statistical software R version 3.5.1 (
https://www.R-project.org/).
P values were considered as exploratory and therefore not adjusted for multiple comparisons. For comparisons regarding drug sensitivity and LATS expression in cell lines, the mean ± SD values were calculated and statistical analysis was performed using the two-tailed Student’s
t test, where
p values < 0.05 were considered statistically significant.
Discussion
We investigated LATS expression in publicly accessible transcriptomic data sets and in tissue samples from a Swiss ovarian cancer patient cohort with regard to clinicopathological characteristics and outcome and based on the intriguing results then determined in a panel of ovarian cancer cell lines whether LATS expression is an indicator for the EMT status of these cells and for drug sensitivity. Based on the results, we may conclude that LATS expression (a) did not associate with better outcome in ovarian cancer, (b) was not reduced in SOC cell lines, and (c) was not associated with EMT status and drug sensitivity.
Several interesting findings emerged from this study: Firstly, the analysis of 15 databases indicated that neither
LATS1 nor
LATS2 expression was associated with favorable survival in ovarian cancer and that
LATS2 may even have a negative role for survival. They neither support the putative function of LATS proteins as tumor suppressors nor are they consistent with the previous study of ovarian cancer patients, where elevated LATS1 and LATS2 expression was associated with better outcome in serous ovarian cancer patients (Xu et al.
2015). Rather,
LATS2 may function as a possible tumor promoter. However, it is worthwhile mentioning that the interpretation of the available transcriptomic data may be done with care because whether an association or its reverse was present depended on the data set selected for the analysis.
Secondly, LATS1/2 expression was not favorable for overall survival in both the overall cohort and the serous carcinoma sub-cohort. It even tended to have a negative effect on relapse-free survival, but the informative value is limited by the low number of events. These results are inconsistent with a tumor-suppressive function of LATS family proteins, but support the findings from the transcriptomic data sets analysis. Histologically, LATS1/2 expression was significantly higher in the serous than in the mucinous subtype, whereas it was comparable among the endometrioid, clear cell, and mucinous subtypes. LATS1/2 expression was lowest in cystadenoma, intermediate in carcinoma, and highest in borderline. LATS1/2 expression was not different among the different FIGO stages for both the overall tumor patient cohort and the serous carcinoma sub-cohort. Of note, LATS1/2 expression was significantly higher in grade 3 than grade 2 in the overall cohort: a statement not meaningful for the serous carcinoma sub-cohort, because only one single case was available in the low-grade group.
These results are discrepant to those from the previous study by Xu et al. who reported that LATS1 and LATS2 expression was elevated in mucinous compared to serous ovarian carcinomas, correlated with better outcome (longer survival and less recurrence), decreased with higher FIGO stage, and was elevated in high-grade serous carcinomas (Xu et al.
2015). The reasons for this discrepancy are unclear, but may be at least partly explained if LATS had a function different from that of a tumor suppressor. It is unlikely that the use of an antibody recognizing total LATS 1/2 protein instead of individual antibodies for LATS1 and LATS2 accounted for the discrepant immunohistochemistry results. On the other hand, our immunohistochemistry results are largely in line with those from transcriptomic data sets: those did not display an unfavorable outcome for ovarian cancer patients with low
LATS expression as expected if LATS was tumor suppressive. Our data are, however, consistent with a previous study reporting LATS overexpression in nasopharyngeal cancer (Zhang et al.
2010).
Thirdly, we also report on issues not yet addressed in ovarian cancer, such as the possible relationship between LATS protein expression in ovarian cancer cell lines and their EMT status and their drug sensitivity. Interestingly LATS1 and LATS2 proteins were higher or at least comparably expressed in ovarian cancer cell lines compared to “normal” HOSE and FT cell lines, but in no case LATS protein expression was decreased relative to the “normal” cell lines. These results may be surprising: assuming a tumor-suppressive function of LATS and considering the decreased LATS expression in SOC compared to healthy tissue reported previously (Xu et al.
2015), LATS protein expression would have been expected to be lower in HOSE and FT cells than in SOC cells. On the other hand, they seem consistent with the transcriptomic data set results. Also, expression of LATS1 and LATS2 is not necessarily interdependent, meaning that high LATS1 expression does not necessarily mean high LATS2 expression in ovarian cancer cell lines. Taken together, the results do not support the proposed function of LATS proteins as tumor suppressors.
Likewise, LATS1 and LATS2 expression was comparable among epithelial, intermediate, and mesenchymal ovarian cancer cell lines, indicating that LATS protein expression was not associated with the cells’ EMT status. This is at odds with the view that LATS expression declines when cells transition from epithelial to mesenchymal (Nozaki et al.
2019; Moroishi et al.
2015; Lei et al.
2008). Opposed to these studies is a report showing that LATS can potentiate tumor-promoting factors and augment EMT (Zhang et al.
2012).
In addition, the expression of LATS1 and LATS2 was not associated with drug sensitivity, meaning that the extent of LATS expression in SOC cell lines did in general not indicate whether cells responded better or poorer to chemotherapeutic drug exposure. However, it may do so in individual cases: although lacking statistical significance, cells with higher LATS1 tended to be drug resistant and those with higher LATS2 tended to be drug sensitive, suggesting that LATS1 and LATS2 have opposing functions in this context. Notably, the absence of an association between LATS1 and LATS2 expression and drug sensitivity does not imply that LATS proteins are not implicated. Indeed, LATS1/2 does modulate cellular responses to chemotherapeutic drug exposure. Low
LATS expression improved therapy response in advanced and recurrent breast cancer patients, possibly through the LATS downregulation-induced disruption of cell cycle checkpoints (Takahashi et al.
2007). Conversely, loss of LATS1 rendered HeLa cells resistant to paclitaxel-induced cell death (Yeung et al.
2018) and loss of LATS2 resulted in doxorubicin and etoposide resistance in leukemic cell lines (Kawahara et al.
2008). The molecular mechanisms by which LATS governs drug sensitivity or resistance are, however, not understood. We also wondered whether acquired drug resistance involved alterations in LATS protein expression. Indeed, altered LATS protein expression was found, but an obvious pattern was missing: this means that acquired resistance was not generally associated with elevated LATS1 and LATS2. Whether the de novo expression of multidrug resistance transporter MDR1 and the altered LATS expression in the paclitaxel-resistant IGROV1-PXL cells are connected cells is unknown.
In summary, our results suggest a function of LATS proteins different from that of a tumor suppressor and may even point to an opposed function of LATS proteins in ovarian cancer. The role of LATS proteins hence remains controversial and possibly is context dependent and cancer type dependent.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.