Relationship between polycomb‐group protein BMI‐1 and phosphatases regulating AKT phosphorylation level in endometrial cancer

Abstract The PI3K/AKT pathway is frequently activated in endometrial carcinoma. BMI‐1 (B‐lymphoma Mo‐MLV insertion region 1) protein affects expression of PTEN (phosphatase and tensin homolog) in some cancers, but its significance for endometrial tumorigenesis is not known. The objective of this study was to determine the relationship between BMI‐1 and expression of factors affecting AKT (protein kinase B) phosphorylation level in endometrial cancer. The expression of proteins and mRNAs was investigated in endometrial cancer specimens and samples of non‐neoplastic endometrial tissue by Western blot and RT‐PCR, respectively. The impact of BMI‐1 down‐regulation on AKT phosphorylation and expression of genes coding for several phosphatases were studied in HEC1A cells. The results showed that BMI‐1 depletion caused increase in PHLPP1 and PHLPP2 (PH domain and leucine‐rich repeat protein phosphatases 1/2) expression and decrease in phospho‐AKT (pAKT) level. In more advanced tumours with higher metastatic potential, the expression of BMI‐1 was lower compared to tumours less advanced and without lymph node metastasis. There were significant inverse correlations between BMI‐1 and PHLPPs, especially PHLPP1 in normal endometrial samples. The inverse correlation between BMI‐1 and PHLPP1/PHLPP2 expression was observed in PTEN positive but not PTEN negative cancers. Low PHLPP2 expression in tumours predicted poorer overall survival. BMI‐1 impacts on AKT phosphorylation level in endometrial cells by regulation of PHLPP expression.


| BACKG ROU N D
B-lymphoma Mo-MLV insertion region 1 (BMI-1) is a component of polycomb repressive complex 1, which is involved in the regulation of genes activity. Several studies have shown that BMI-1 expression is frequently altered in various types of human cancers, including gastric, ovarian, breast, head and neck, pancreatic and lung cancers. 1,2 It has been found that this protein plays an important role in cancer initiation and progression by promotion of cell growth and proliferation, inhibition of apoptosis as well as regulation of invasion and metastasis. 1,2 BMI-1 is related to poor prognosis of non-small cell lung cancer, ovarian cancer, head and neck cancer and haematological malignancies. 2 It is suggested that one of the possible mechanisms by which BMI-1 promotes tumorigenesis is the activation of PI3K/AKT pathway which is due to inhibition of PTEN expression. [3][4][5] It has been shown that in nasopharyngeal cancer cells, PTEN inhibition and activation of the PI3K/AKT/GSK3β pathway by BMI-1 causes stabilization of the transcription factor Snail1, which together with PRC1 binds to E-cadherin promoter and down-regulates its expression. It leads to reduction in the ability of cells to adhere and in consequence enhances their invasive potential. 3 BMI-1 promotes invasion and metastasis of pancreatic cancer stem cells by activating PI3K/AKT signalling. 5 BMI-1 is probably also involved in endometrial carcinogenesis, but data are limited and its exact function is unknown.
Honig et al 6 found that endometrial carcinomas express higher level of BMI-1 than benign controls. Moreover, it has been found that miR-194 and miR-200 by targeting BMI-1 inhibit epithelial to mesenchymal transition (EMT) of endometrial cancer cells. 7,8 In contrast to results suggesting function of BMI-1 in EMT and metastasis, Engelsen et al 9 demonstrated that low, rather than high BMI-1 expression is associated with an aggressive phenotype in endometrial carcinomas.
Alterations of PI3K/AKT signalling pathway play essential role in endometrial carcinogenesis and currently a number of clinical trials are underway to assess the effectiveness of therapies using PI3K/ Akt pathway inhibitors. 10 The PI3K/AKT pathway is frequently activated in endometrial carcinoma, often due to PTEN loss. However, the significance of PTEN loss for endometrial cancer progression is not clear. Karnezis  Several other phosphatases negatively regulate the PI3/AKT pathway. Two isoforms of PH domain leucine-rich repeat protein phosphatase (PHLPP), PHLPP1 and PHLPP2 have been shown to directly dephosphorylate AKT. 12 There are no data concerning expression of PHLPPs in endometrial cancer. Thus, the further studies concerning regulation of PI3K/AKT pathway in endometrial cancer are needed.
The aim of this study was to estimate the BMI-1 expression in normal and cancerous endometrial tissues and determine if there is any association between BMI-1, PTEN and phosphorylation level of AKT. Moreover, we analysed the impact of BMI-1 silencing on expression of several phosphatases involved in direct or indirect activation of AKT. The correlation of BMI-1 and PHLPPs in endometrial normal and cancer tissues was analysed.  Cell Signaling Technology). The blots after stripping were reprobed with anti-β-actin antibodies (1:1000; Santa Cruz Biotechnology).

| Western blot and densitometric analysis
The bands corresponding to proteins were analysed in Gel Pro 3.0 Analyzer software (Media Cybernetics) by measuring integrated optical density (IOD) of the bands. To avoid the errors due to transfer and handling differences, we applied in lane normalization using βactin as an internal reference. Moreover, the same reference sample was applied to each blot (external control). This normal endometrial sample was chosen in the first Western blotting experiments since it showed median expression of studied proteins. The results are presented as a relative IOD, that is the ratio of specific protein IOD to β-actin IOD and reference sample. This procedure allowed us to compare the bands on different blots and avoid the errors due to possible differences in ECL signal. The AKT phosphorylation level is expressed as phospho-AKT/AKT ratio.

| Statistical analysis
Statistical analysis of the results was performed using the statistical program GraphPad Prism 5.0 (GraphPad Software Inc). The nonparametric Mann-Whitney U test was used when two groups were compared. Comparisons between more than two groups were done using Kruskal-Wallis test. For pairwise multiple comparisons, Dunn's post hoc test was used. Spearman correlation coefficient was calculated for correlation analysis. The Student's paired -test was used to compare the differences between treated and untreated cells.
A P-value <.05 was considered to indicate a statistically significant difference.

| Down-regulation of BMI-1 in HEC1A cells
To assess the impact of BMI-1 on PTEN and AKT, the expression the BMI1 was depleted in endometrial cancer cell line HEC1A by siRNA. In cells treated with siRNA, the BMI-1 transcript level was substantially reduced (by 70%-80%). The protein level of BMI-1 was also very low ( Figure 1A Table S1.  of Western blot analysis are shown in Figure 2. The results of densitometric analysis of all samples are shown in Table 2. There were no statistically significant differences between expression of BMI-1 in non-tumour tissues and low stage tumours (Table 2).

Mean ± SD
However, in more advanced tumours, classified as stage III and IV, the expression of BMI-1 was lower than in less advanced tumours, corresponding to the first and second stage according to FIGO classification ( Table 2). The expression of BMI-1 was also lower in cancers that exhibited the ability to metastasize to regional lymph nodes compared to non-metastatic cancers ( There were no significant differences in expression level of AKT in samples of normal endometrial tissue and cancer samples or between cancers with different stage or grade (Table 2).
Phosphorylation levels of AKT in endometrial tissue samples were expressed as pAKT to AKT ratio. In normal endometrial tissues, pAKT/AKT ratio was significantly lower than in the endometrial cancer tissues (Table 2). However, stage I tumours showed tendency to have higher pAKT/AKT ratio compared to higher stage tumours (P = .07). Lower phosphorylation was observed in cancers with ability to lymph node metastasis compared to those without metastasis (Table 2). No significant differences in pAKT/AKT ratio were observed between tumours of different histological grade.

| Expression of BMI1, PTEN, PHLPP1 and PHLPP2 mRNA in normal and endometrial cancer samples
There were no statistically significant differences between BMI1 mRNA expression in normal and cancer samples (Table 3). Less differentiated tumours showed a tendency to have higher BMI-1 mRNA expression level than tumours of G1/G2. Similarly, to the protein expression results, there was a significantly lower BMI1 mRNA expression in cancers that exhibited the ability to metastasize to regional lymph nodes compared to non-metastatic cancers.
As BMI-1 protein is suggested to be a negative regulator of  between samples depending on FIGO stage or histological grade and lymph node metastasis status (Table 3).

| Correlation between BMI-1, PTEN, PHLPP1/2 expression and AKT phosphorylation level
The correlations between BMI-1, PTEN, PHLPPs and phosphorylation level of AKT were analysed in order to check potential association of BMI-1 with PTEN/AKT pathway. We did not observe correlation between BMI-1 and PTEN proteins expression levels (Table 4) (Table 4).
Since relationship between PTEN and PHLPP expression was suggested in some cells, 14 (Table 4).

| PTEN, BMI-1 and PHLPPs status and endometrial cancer prognosis
Kaplan-Meier survival analysis was used to estimate the predictive effect of PTEN, BMI-1 and PHLPPs expressions on overall survival ( Figure 3). Overall survival differences in case of BMI-1 and PHLPP1 were not statistically significant. However, there was a tendency that patients in the low BMI-1 expression group had a shorter overall survival than those in the high BMI-1 group (log-rank test P = .08, Figure 3B). The Kaplan-Meier analysis showed that patients in the low PHLPP2 expression group had a significantly shorter overall survival than those in the high PHLPP2 group (log-rank test, P = .014, Figure 3D). Moreover, patients in the positive PTEN expression group had a significantly shorter overall survival than those in the negative PTEN expression group (log-rank test P = .016, Figure 3A). The results of our studies did not show significant difference between the low-stage endometrial cancers and normal tissue.

| D ISCUSS I ON
However, we found lower expression of BMI-1 in samples of more advanced endometrial carcinomas (stage III/IV) compared to less advanced tumours (stage I and II). Moreover, the expression of BMI-1 was lower in tumours that exhibited the ability to metastasize to regional lymph nodes compared to non-metastatic cancers. Thus, our results similarly to results of Engelsen et al 9 suggest that low expression of BMI-1 correlates with a more aggressive phenotype of endometrial carcinoma. We used Kaplan-Meier survival analysis to estimate the predictive effect of BMI-1 expression on overall survival. There was a tendency that patients in the low BMI-1 F I G U R E 3 PTEN, BMI-1 and PHLPP1/2 expression status and endometrial cancer prognosis. Kaplan-Meier curve showing overall survival for endometrial cancer patients with positive and negative expression of PTEN (A); low and high levels of BMI-1 protein expression (B); low and high PHLPP1 or PHLPP2 mRNA expression (C and D, respectively). Significance was estimated with the log-rank test. Overall survival was defined as the interval from the date of initial surgical resection to the date of last known contact of death expression group had a shorter overall survival than those in the high BMI-1 group.
In many studies, PTEN protein level in endometrial cancers have been analysed by immunohistochemistry, but the results regarding PTEN expression impact on cancer progression and prognosis are inconsistent. Some studies showed that lack or low PTEN level is associated with worse prognosis, [19][20][21] but the others suggest that loss of PTEN expression is associated with favourable prognosis. [22][23][24] Gao et al 24 found that the level of PTEN expression in patients with endometrial carcinoma was significantly related to differentiation (the expression of PTEN in high differentiated endometrial carcinoma was significantly higher than that in middle-low differentiated one and clinical stage). In our study, PTEN protein expression was much lower in tumours compared to controls, but we did not find Also, we found the strong inverse correlation between BMI-1 and PHLPP1/2 in normal endometrial tissues. Weak inverse correlation has been found only in cancers with expression of PTEN. These results may suggest for the first time that BMI-1 impact on PHLPP and at least partially it is dependent on PTEN. PHLPP1 and PHLPP2 have been identified as tumour suppressors because they can suppress of pro-survival signalling, such as PI3K/Akt signalling pathway. According to that both PHLPP1 and PHLPP2 expressions are lost in diverse cancers, 31 but their role in progression of cancers has not been identified. Our results are the first results concerning PHLPP1 and 2 expressions in endometrial cancer. We found decreased expression of PHLPP2 in cancer samples compared to non-tumoral samples, but there were no significant differences in PHLPP expression between samples depending on FIGO stage or histological grade and lymph node metastasis status. However, PHLPP2 seems to be a promising predictive factor of patients overall survival (Figure 3, log-rank test P = .014). Thus, further analysis of PHLPPs expression especially on protein level including higher number of cases should be performed. Unfortunately, the commercially available antibodies against PHLPP should be used with caution since it is suggested that some PHLPP1 antibodies recognize a signal in tissues that may represent a non-specific band unrelated to PHLPP1, that cover the specific one. 32 In summary, BMI-1 affects AKT activity by regulation of In summary, our results for the first time showed that down-regulation of BMI-1 in cancer cells might affect expression of PHLPP1 and PHLPP2. Moreover, decreased PTEN and increased phospho-AKT levels are more likely associated with an early event in endometrial tumorigenesis, while low expression of BMI-1 and low level of AKT phosphorylation may be involved in endometrial cancer progression.
BMI-1 seems to impact on AKT phosphorylation level in endometrial cells by regulation of PHLPP expression. However, some limitations of our work need to be considered. First one is the small sample size, especially the small number of more advanced and aggressive cancer samples. Further studies are needed for larger group of patients with stage III and IV cancer. The second limitation is that analyses of association between studied genes and proteins were performed using only one endometrial cancer cell line. Although, we obtained the similar results for breast cancer cells, to verify the universality of the mechanism in endometrial cancer and thus its significance, other endometrial cancer cell lines should be investigated. Taking into account that relationship between BMI-1 and PHLPP seems to be dependent on PTEN expression it is necessary to use in future studies both PTEN positive and PTEN negative cells. Third, we have not taken into account mutations in studied genes, especially those, that do not affect expression but affect the activity of proteins. It should be done in future to find out whether relationship between BMI-1 and PHLPP depends on PTEN enzymatic activity.

ACK N OWLED G EM ENTS
This study was supported by statutory funds of University of Lodz.

CO N FLI C T O F I NTE R E S T
The authors confirm that there is no conflict of interest.

AUTH O R CO NTR I B UTI O N S
AZ, AK and PJ involved in study conception and design. ŁC, KWK and PC involved in acquisition of data. AZ, PC, EF and PJ involved in analysis and interpretation of data. AK and AZ drafted the manuscript. MB and AB critically revised the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.