Introduction

Endometrial cancer is the fourth most common cancer in the female population. It was estimated that 43,470 endometrial carcinoma cases would be newly diagnosed in 2010 in the USA [1]. Estimated number of new endometrial cancer cases diagnosed in 2008 in Europe equals 82,530 [2]. Deregulation of various molecules and signaling pathways has been implicated in endometrial oncogenesis; however, its exact pathogenesis has not been completely elucidated [3]. Discovery of microRNAs has brought a new insight into the pathogenesis of many diseases including endometrial cancer [4]. MicroRNAs are non-coding, single stranded RNAs, consisting of 22 to 26 nucleotides, which regulate gene expression on the posttranscriptional level. Studies of the past few years have established microRNAs as important molecules in the pathogenesis of various solid tumors, suggesting their possible role and utilization in cancer screening, treatment, and prediction of outcome [4, 5]. Deregulation of microRNAs expression was also observed in endometrial cancer samples as reported by few studies [6–8]. However, the reasons of microRNAs expression alterations observed in this tumor type have not been elucidated so far. Deterioration of microRNA biogenesis machinery has been implied as one possible mechanism among other suggested causes like DNA mutations, single nucleotide polymorphism or epigenetic changes [9–14].

Drosha and Dicer are two major components of microRNA biogenesis machinery. MicroRNAs are transcribed by polymerase II as stem-loop molecules consisting of hundreds to thousands of nucleotides called pri-microRNAs which are sequentially cleaved, first by Drosha, which produces so-called pre-microRNAs, and then by Dicer [15]. Drosha belongs to the ribonuclease III superfamily of double-stranded RNA-specific endoribonucleases, and cleaves pri-microRNAs into approximately 70 nucleotides (nt) hairpin pre-microRNAs. Interaction with exportin-5 and Ran-GTP facilitates transport of pre- microRNAs to the cytoplasm. Within the cytoplasm pre-microRNAs are further processed by a multidomain protein Dicer into short double-stranded molecules [15]. Apart from Dicer and Drosha, which are main regulators of microRNA biogenesis, the complicated microRNA machinery involves other proteins including DGCR8 and Ago2, and recently discovered co-regulators KSRP, hnRNP A1, p53, p63, p73, which are thought to be involved in the tissue specific microRNA processing or connected with particular cellular conditions [16–20].

Two Dicer protein isoforms were discovered by Potenza et al. in neuroblastoma cells, the long named Dicer1, which was also found in various other tissues and the short one, called t-Dicer [21]. In 2005, Irvin-Wilson and Chaudhuri reported transcription of Dicer from an alternative promoter as well as alternative splicing of the 5′-exons of Dicer transcript in breast cancer cell lines, and recently, Grelier at al. described 14 putative variants of human Dicer mRNA in breast cancer. The three variants were found to encode a full-length protein (variants a, b, and c), whereas two shorter variants d and e encoded proteins of 113 and 92.7 kDa, respectively [22–24].

Alterations of Dicer and Drosha expression levels have been found to accompany microRNA deregulations in various malignancies including breast and ovarian cancers [25, 26]. An importance of an intact expression of microRNA biogenesis genes was demonstrated in Dicer depleted mouse models, in which global decrease in microRNAs levels was observed and was connected with severe anatomical and functional alterations. Dicer depleted or mutant mice presented with developmental abnormalities and increased lethality, as well as axonal degeneration [27, 28]. Loss of Dicer expression in mice uterus and oviducts resulted in dramatic anatomical and histological abnormalities [29–31].

The present study aimed to investigate Dicer and Drosha genes expression in endometrial cancer samples by quantitive real-time PCR using validated reference genes and to analyze impact of clinicopathological parameters on their expression.

Material and methods

Patients

Altogether, forty-four patients were included in the study. All patients were explained the study purpose and informed consent was obtained from each study participant. The Medical University of Lublin Ethical Committee has approved the study design (decision # KE-0254/22/2009).

Twenty-six patients diagnosed with endometroid endometrial cancer by endometrial biopsy performed prior to the operation were included in the study. All patients were scheduled for the primary surgical treatment. None of the patients underwent any neoadiuvant therapy prior to the operation. Upon obtaining the full pathology report, FIGO (the International Federation of Gynecology and Obstetrics) staging of the disease progression was performed for each patient. Control samples were obtained from 18 females operated due to benign gynecological diseases other than of endometrial origin. Clinicopathological characteristics of the patients were summarized in Table 1.

Table 1 Clinicopathological characteristics of the patients
Full size table

Sample collection and storage

Endometrial cancer samples were collected from histopathology confirmed endometrial endometrioid carcinoma bulk tumors obtained from patients undergoing hysterectomy. Normal endometrial samples were collected from patients undergoing hysterectomy due to benign gynecological diseases other than of endometrial origin. All tissue samples were collected no more than 15 min after resection of the uterus and were immediately immersed in RNAlater™ solution and incubated in that solution for another 24 h in 4°C (Ambion, USA). All samples were stored in −80°C until RNA isolation. The samples were examined microscopically to ensure tumor cellularity greater than 70%.

RNA isolation and quality control

Isolation of total RNA was performed with mirVana Isolation Kit (Ambion) according to the manufacture's protocol. Forty to eighty milligrams of tissue were processed per sample. In order to reduce the amount of the residual genomic DNA, treatment with DNAase I was performed using the Turbo DNA-free™ Kit (Ambion) according the manufacturer's protocol. Concentration and purity of RNA were assessed using BioPhotometer plus (Eppendorf, Germany) and a TrayCell (Hellma GmbH&Co, Germany). Integrity of the RNA was evaluated by electrophoresis using Agilent RNA Nano kit and the Agilent Bioanalyzer (Agilent Technologies, USA). The RIN (RNA integrity number) values obtained for endometrial cancer and control samples ranged between 6.6 and 9.4. The total list of RNA samples, which includes the RIN values and 260/280 ratios of all samples, was presented in Online Resource 1.

Reverse transcription and real-time PCR

Either 2 μg or 500 ng of total RNA was reverse transcribed using hexamer random primers and a High Capacity cDNA Reverse Transcritpion Kit with RNase Inhibitor (Ambion, USA). Reverse transcription was performed in the final volume of 20 μL. cDNA samples were tenfold diluted before real-time PCR. Amplifications of Dicer and Drosha were performed utilizing specific TaqMan® probes and primers purchased from Applied Biosystems, USA (assays IDs Hs00229023_ml and Hs01095033_m1, respectively). Dicer isoform 1 is encoded by two transcript variants 1 and 2. Primers and probe for Dicer1 were chosen to amplify both 1 and 2 transcript variants. In brief, each 20 μL reaction consisted of 10 μL TaqMan® Gene Expression Master Mix, 1 μL assay, 5 μL nuclease-free water and 4 μL cDNA.

Real-time PCR protocol consisted of pre-incubation in 95°C for 10 min and 45 quantification cycles, each consisting of denaturizing at 95°C for 15 s and annealing/extension at 55°C for 1 min.

Amplification of four candidate reference genes, ACTB (assay ID Hs99999903_m1), eIF6 (assay ID Hs00158272_m1), RLPO (assay ID Hs99999902_m1), and HPRT (part number 4326321E) was additionally performed in all forty-four samples using TaqMan® probes and primers (Applied Biosystems, USA). RT-minus, no template control and inter-plate calibrators were included in every run. All real-time PCR reactions were carried out in triplicates in a Rotor Gene thermal cycler (Corbett Research, Australia).

Efficiencies of primer/probe sets as well as a dynamic range of the assays were determined by performing standard qPCR with seven 4-fold dilution of the commercially available cDNA. Efficiencies for all assays were calculated using the equation E = 10(−1/slope) and were between 0.9 and 0.98 (data for each assay was included in the Online Resource 2).

Statistical analysis

Data analysis was performed using GenEx 5.2.7 software (MultiD Analyses AB, Sweden). EIF6 and HPRT were chosen to normalize for RNA quality and cDNA input, basing on the Normfinder analysis of the expression of the four candidate reference genes [32]. Normalization with inter-plate calibrators was performed to validate for technical variations between the runs. Mean and confidence interval were utilized for descriptive statistics. Differences between the groups were tested with either a student t test, analysis of variance (one-way ANOVA with the Bonferroni post hoc test) or Mann–Whitney tests depending on the Kologomorov–Smirnov p value results. All tests used for statistical analysis were two-sided. Correlations between Dicer and Drosha expressions and clinicopathological characteristics were assessed with Pearson test. P value of less than 0.05 was established to denote significance in all statistical analyses performed in the study.

Results

The mean age of patients in the endometrial cancer group was higher comparing to control group; however, no correlation was found between age and Dicer and Drosha mRNA expression levels. The cancer and control groups did not differ significantly regarding BMI (p = 0.25).

Prior to statistical analysis, raw qPCR data of Dicer and Drosha mRNA expression were normalized to two reference genes, eIF6 and HPRT. The genes were chosen from the four candidate genes. The variability for the best combination of genes calculated by Normfinder was 0.189, which was also confirmed by GeNorm [32, 33].

Our study revealed that both Dicer and Drosha mRNA expressions were significantly down-regulated in endometrial cancer samples comparing to control group (Figs. 1 and 2). Expression of Dicer was down-regulated in cancer group comparing to control group by the factor of 1.54 (p = 0.009). Drosha mean expression value was 1.4 times lower in endometrial cancer group versus control group with the p value of 0.008. Additionally, the correlation between down-regulation of Dicer and Drosha was found with Pearson correlation coefficient value of 0.75 (Fig. 3).

Fig. 1
figure 1

Dicer expression in endometrial cancer samples and in the control group (values presented in the log scale); *p = 0.009

Full size image
Fig. 2
figure 2

Drosha expression in endometrial cancer samples and in the control group (values presented in the log scale); *p = 0.008

Full size image
Fig. 3
figure 3

Dicer and Drosha expression in endometrial cancer (red) and normal endometrial (green) samples

Full size image

To evaluate the influence of the clinical stage on Dicer and Drosha expression, endometrial cancer samples were divided into two groups. One group comprised samples staged as FIGO 1A and the other group consisted of all-remaining samples, which were classified as FIGO stages 1B–3C. Statistical analysis did not reveal significant differences between two groups. Similarly, no significant differences were encountered when myometrial invasion was taken into account.

Analysis of variance (one-way ANOVA with the Bonferroni post hoc test) revealed the influence of histological grade on Drosha expression, with the significantly higher difference detected between controls and higher-grade cancers (p = 0.038). No influence of histological grade was detected in regards to expression of Dicer (Fig. 4).

Fig. 4
figure 4

The influence of histological grading on Dicer and Drosha expressions (values are presented in the log scale): one-way ANOVA with the Bonferroni post hoc test revealed the influence of histological grading on Drosha expression, with the significantly higher difference detected between controls and higher-grade cancers; *p = 0.038

Full size image

The influence of BMI on Dicer and Drosha expression was investigated separately in the endometrial cancer and control groups. The analysis of variance did not reveal any influence of BMI on Dicer gene expression in endometrial cancer group. The expression of Drosha, on the other hand, was influenced by BMI and the post hoc tests showed significant differences between normal weight (BMI < 25 kg/m2, n = 8) and overweight (25 kg/m2 ≤ BMI < 30 kg/m2, n = 9) patients as well as between overweight and obese (BMI ≥ 30 kg/m2, n = 9) individuals (p = 0.003 and p = 0.02, respectively). It is also worth to note that the expression of both investigated genes was the lowest in the group consisting of overweight patients, although not significantly different in the case of Dicer. At the same time the number of high-grade tumors did not differ significantly between the three BMI groups. No influence of BMI on Dicer and Drosha mRNA levels was found in the control group.

Analysis of the influence of parity revealed significantly lower Dicer expression in cancerous samples collected from nulliparous and uniparous females comparing to those obtained from multiparous individuals (p = 0.002). No influence of parity was found in control samples.

A comparison of the data in regard to the menopausal status performed separately for cancer and control samples did not reveal any significant differences.

Discussion

Alterations of microRNAs expression profiles were reported in endometrial cancer by a number of studies; however, the underlying mechanism of the observed aberrations is still largely unresolved [6–8, 34–37]. Altered expression of the genes involved in the microRNA biogenesis has been implicated as a possible cause of the differences in the microRNA profiles between normal and cancerous tissues [14, 25, 26]. In addition, Dicer and Drosha expressions were investigated in regard to survival prediction in breast and ovarian cancers as well as in other malignancies [23, 38–44]. To the best of our knowledge, this is the first study to investigate Dicer and Drosha expressions in endometrial cancer. Using a quantitive real-time PCR approach and validated reference genes, we found a decreased expression of Dicer and Drosha transcripts in endometrial cancer samples in comparison to healthy controls. In addition, the down-regulation of Dicer significantly correlated with Drosha down-regulation. These results are in concordance with a number of studies, which investigated Dicer and/or Drosha gene expression levels in ovarian and breast cancers. Merritt et al. observed down-regulation of both Dicer and Drosha, and Pampalakis et al. observed down-regulation of Dicer in ovarian cancer samples [39, 45]. Moreover, in the study conducted by Merritt et al., low expression of Dicer comprised an independent predictor of the poor clinical outcome and reduced disease-specific survival. The same study revealed a correlation between Dicer and Drosha transcripts and protein expression [39]. In contrast, Flavin et al. observed increased Dicer mRNA and protein expression levels in ovarian cancer samples [46]. Several studies performed in breast cancer samples indicated down-regulation of Dicer and/or Drosha expression in malignant comparing to normal tissues [23, 38]. Most of those studies could not however demonstrate any concordance with the proteins expression [47, 48]. The results obtained by Grelier et al. suggested down-regulation of Dicer mRNA to be related with greater invasiveness and metastatic spread of breast carcinoma and to have an independent prognostic value in metastatic disease [23].

Higher histological grade was connected with greater invasiveness and worse prognosis in endometrial cancer [49, 50]. Interestingly, our study revealed a correlation between higher-grade tumors and down-regulation of Drosha expression suggesting its possible connection with greater invasiveness and worse prognosis of endometrial cancer. Due to a short follow-up period, we could not however assess the prognostic value of Drosha in our sample group. The data revealed by our study warrants therefore further investigation with longer observation period and a larger group of patients.

The literature referring to the expression of Dicer and Drosha in other tumor types is inconsistent. Increased expression of Dicer mRNA and protein was found in prostate and esophageal carcinomas as well as in precursor lesions of lung adenocarcinoma, whereas decreased levels of both Dicer and Drosha were characteristic for high-risk neuroblastoma tumors. The down-regulation of Dicer mRNA was reported in hepatocellular carcinoma [41–43, 51, 52]. This may suggest tissue- and tumor-specific alterations in microRNA biogenesis genes.

Obesity comprises a well-known risk factor for endometrial cancer [53]. Therefore we analyzed the influence of BMI on Dicer and Drosha expression separately in cancer samples and in normal endometrium. We found that the expression of Drosha, but not Dicer, was influenced by BMI and the Bonferroni post hoc test revealed significant differences between the normal weight and overweight patients as well as between overweight and obese individuals. It is also worth to note that the expression levels of both investigated genes were the lowest in the group consisting of overweight patients, although not significantly different in the case of Dicer. As the histological grade influenced Drosha expression, we evaluated three BMI groups for the number of high-grade tumors and found no significant difference. In the study performed by Merritt et al., no significant associations were present between gene expression levels and age and tumor grade; however, low Dicer mRNA level was correlated with advanced tumor stage [39]. Other investigators demonstrated similar associations in ovarian and breast carcinomas [23, 38, 45]. In our study, there was no significant association between down-regulation of Dicer and Drosha and more advanced tumor stages; however, there was a trend toward lower expression levels in FIGO stages higher than 1A.

The reasons for alterations in Dicer and Drosha expression observed in our study and other reports published to date are largely unknown. One possible hypothesis might be connected with the Dicer gene location at the subtelomeric region on the chromosome 14 (14q32.13), which was found affected by allelic deletion in various tumors [54, 55]. Interestingly, Fujino et al. reported a high percentage of endometrial cancers presenting loss of heterozygosity (LOH) on 14q chromosome and defined a minimal region of deletion for these tumors to region 14q32. In addition, investigators observed a strong association between 14q LOH and poor clinical outcome [56].

Biocomputational analysis performed by Pampalakis et al. indicated DNA methylation as a possible mechanism of Dicer down-regulation based on the presence of the strong CpG island spanning the first exon of the gene [45, 50]. Although methylation of Dicer gene was not detected in a lung adenocarcinoma, this possibility warrants further investigation [40].

Recent studies suggested that regulation of Dicer expression could appear at the posttranscriptional level involving both mRNA and protein stage [57, 58]. In silico algorithms revealed Dicer mRNA to bear several binding sides for a number of various microRNAs. It was also suggested that target sites located within the coding region, which were recently found for let-7, could differ in mechanism of posttranscriptional repression. Thus, microRNAs by targeting various sites could potentially lead not only to translational inhibition but also cause mRNA instability [57, 58].

Interestingly, our search performed with Diana-microT 3.0 and TargetScan revealed that a number of microRNAs previously reported to be up-regulated in endometrial cancer targeted Dicer mRNA [59, 60]. As an example, Boren et al. reported a significant up-regulation of miR-103, miR-107, and let-7c, which are strongly associated with 3′UTR region of Dicer mRNA [35]. Additionally, miR-200a and miR-141 as well as miR-9, predicted by TargetSan, were up-regulated in the microarray study performed by our team (data not published, article under review).

Deregulation of p53 protein has been well established in the pathogenesis of endometrial cancer. P53 alterations have been mostly connected to the non-endometrioid adenocarcinoma. Still, mutations in p53 gene were reported in 4–32% of endometroid endometrial cancers and were recently proposed as prognostic factors [61, 62]. P63 and p73 protein alterations were also observed in endometrioid endometrial tumors [61, 62]. At the same time, recent studies suggested involvement of the p53, p63, and p73 in regulation of microRNA processing components including Dicer and Drosha. According to that hypothesis Dicer and Drosha could be regulated either at the transcription step or indirectly by p53/p63/p73-dependent microRNAs at the posttranscriptional level [63].

In summary, our study revealed for the first time that expression alterations of main regulators of microRNAs biogenesis are present in endometrial cancer tissue and could be potentially responsible for altered microRNAs profiles observed in this malignancy. This interesting observation warrants further studies, which would explore correlations with protein expression as well as functional sequel and consequences of Dicer and Drosha expression deregulation in endometrial cancer cell lines.