Dysregulated microRNAs and their predicted targets associated with endometrioid endometrial adenocarcinoma in Hong Kong women

Abstract

The objective of this study, a parallel study to global gene expression profiling, was to identify dysregulated microRNAs (miRNAs) associated with endometrioid endometrial adenocarcinoma (EEC), examine their correlation with clinico-pathological characteristics and identify predicted target genes of the dysregulated miRNAs. Using real-time quantitative reverse transcription-polymerase chain reaction (qRT-PCR), profiling of miRNA expression was performed in 30 EECs and 22 normal counterparts in which genome-wide gene expression had been previously profiled and reported. Clustering analysis identified 30 miRNAs which were significantly dysregulated in EEC. The expression of a sub-group of miRNAs was significantly correlated with clinico-pathological characteristics including stage, myometrial invasion, recurrence and lymph node involvement. By searching for predicted miRNA targets that were linked to the dysregulated genes previously identified, 68 genes were predicted as candidate targets of these 30 dysregulated miRNAs. miR-205 was significantly overexpressed in EECs compared with normal controls. After transfection of a miR-205 inhibitor, the expression of miR-205 in endometrial cancer cell line RL95-2 cells decreased whereas its predicted target gene, JPH4, showed increased protein expression. JPH4 seems to be a real miR-205 target in vitro and in vivo, and a candidate tumor suppressor gene in EEC. Based on this study in EEC, miRNAs predicted to be involved in tumorigenesis and tumor progression have been identified and placed in the context of the transcriptome of EEC. This work provides a framework on which further research into novel diagnosis and treatment of EEC can be focused.

Endometrial cancer is the third most common gynecologic malignancy and the ninth most common malignancy for females overall in Hong Kong. Although it has a relatively low mortality, some are aggressive. About 80–90% of these cancers are endometrioid endometrial adenocarcinoma (EEC). The search for novel molecular markers for early detection and predicting outcomes has been going on in most cancers with a view to identifying molecular targets for therapeutic agents. Previously we have established global gene expression profiles in a large set of EEC and normal counterparts using oligonucleotide microarrays. The study revealed a panel of genes significantly dysregulated as well as potential molecular pathways involved in carcinogenesis and progression of this major subtype of endometrial cancer.¹

MicroRNAs (miRNA) are a class of single-stranded noncoding RNA molecules of about 21–25 nucleotides in length that are cleaved from 70–100 nucleotides hairpin-shaped precursors.^2,3 Such small RNAs are able to regulate the function of multiple target genes. Complementarities between a 6–7 nucleotide “seed” in the 5′ of the miRNA and target sites in the 3′ untranslated region of messenger RNA (mRNA) recruit ARGONAUTE silencing complexes that repress mRNA translation and lead to mRNA destabilization and degradation through mechanisms that are not yet quantitatively understood.² It is becoming clear that miRNA play critical roles in various biological regulation pathways, including organ development, cell differentiation, proliferation and apoptosis.⁴ Previous miRNA expression profiling in some organs have suggested that miRNA expression may be tissue specific and that individual miRNAs can possess context specific functions in different organs or cell types.^5,6 Recent reports have also shown that miRNAs may play an important role in carcinogenesis and tumor progression. Calin et al.⁷ reported a downregulation of miRNA-15 and miRNA-16 in chronic lymphatic leukemia. Aberrant miRNA expression has been reported in a variety of human solid tumors cancers, including lung, breast, liver, thyroid and ovarian cancer.^8–13 These studies indicate miRNA are a newly recognized class of genes involved in human tumorigenesis and tumor progression, and represent potential, important diagnostic markers for tumor progression and potential therapeutic targets.

We analyzed miRNA expression profiles in EEC to find dysregulated miRNAs associated with endometrial cancer and to determine their correlation with clinico-pathological characteristics. We also explored the predicted target genes of dysregulated miRNAs in EEC and performed a preliminary functional study of one overexpressed miRNA in an endometrial tumor derived cell line. In this study, miRNAs predicted to be involved in tumorigenesis and tumor progression in EEC have been distinguished and placed in the context of the transcriptome of EEC. This work provides a framework from which further research into the novel diagnosis and treatment of EEC can be focused.

Material and methods

Patients and samples

Study subjects were recruited at the Department of Obstetrics and Gynaecology, The Chinese University of Hong Kong, Prince of Wales Hospital, Hong Kong. Tumor tissue specimens obtained from a total of 38 primary EEC tumors were embedded in OCT compound (Tissue-Tek® OCT™ Compound, Sakura Finetek, Europe, B.V., Netherlands), snap-frozen in liquid nitrogen, and stored at −80°C immediately after excision. Twenty-six normal endometrium tissue specimens were collected from women who underwent hysterectomy for nonmalignant conditions and were processed in the same manner. Of the 26 normal endometrial specimens, 8 cases were in the proliferative phase, 9 in secretory phase and 9 were postmenopausal. Frozen sections from each specimen were stained by H&E to verify the quality of the tissue sample and the extent of inflammatory infiltrate. Histologic typing of each tumor was done according to World Health Organization criteria, whereas clinical staging followed International Federation of Gynecology and Obstetrics standards. All specimens and their corresponding clinico-pathological information were obtained under protocols approved by the Clinical Research Ethics Committee of The Chinese University of Hong Kong (Table I). An informed consent form was obtained from each patient.

TABLE I –

CLINICO-PATHOLOGICAL CHARACTERISTICS OF ENDOMETRIOID ENDOMETRIAL ADENOCARCINOMA

Case	Age	Grade	Stage	Myometrium invasion (%)	Lymph node involvement	Recurrence
U-196	37	1	IB	<50	No	No
U-197	75	1	IC	>50	No	No
U-199	53	1	IB	<50	No	No
U-200	63	1	IB	<50	No	Yes
U-202	62	2	IIIC	>50	Yes	No
U-207	77	2	IB	<50	No	No
U-208	47	2	IB	<50	No	No
U-216	46	1	IB	<50	No	No
U-218	51	2	IB	<50	No	No
U219	47	2	IB	>50	No	No
U-226	52	1	IB	<50	No	No
U-277	49	2	IIB	<50	Yes	No
U-229	47	1	IA	<50	No	No
U-231	49	1	IA	<50	No	No
U-232	47	1	IB	<50	No	No
U-233	72	1	IIIA	>50	No	No
U-237	40	1	IB	<50	No	No
U-238	53	2	IIIC	>50	Yes	Yes
U-240	50	1	IB	<50	No	No
U-244	51	1	IB	<50	No	No
U-266	55	1	IB	<50	No	No
U-269	79	1	IIB	<50	No	No
U-282	52	1	IC	>50	No	No
U-285	87	1	IIIA	<50	No	Yes
U-299	53	2	IB	<50	No	No
U-305	46	1	IIB	<50	No	No
U-327	82	2	IB	<50	No	Yes
U-339	40	2	IIIA	>50	No	No
U-340	59	2	IB	<50	No	No
U-348	72	1	IA	<50	No	No

Microdissection and RNA extraction

Endometrial cancer can present with a paucity of neoplastic cells embedded in dense stroma that has a significant lymphocytic infiltrate. Normal endometrium can be the same. Therefore, microdissection either manually or using a laser capture microdissection system (MicroBeam MicroLaser Systems, PALM Microlaser Technologies GmbH, Bernried, Germany) was performed to obtain relatively homogeneous populations of endometrial target cells. OCT compound embedded frozen tissue was cut into 8-μm cryosections, which were placed onto slides treated with DEPC-water to prevent RNA degradation. Slides were stained with hematoxylin for 30 sec, washed with DEPC-water for 15 sec, dehydrated with absolute ethanol for 15 sec, and then dried in air for 1 hr. Microdissection was then performed. It is estimated that the epithelial components in both cancer and control samples reached 90% or more, and were comparable.

Total RNA was extracted from microdissected endometrial target cells using TRIzol total RNA isolation reagent (Invitrogen, Carlsbad, CA), according to the manufacturer’s protocol. In brief, 750 μl of TRIzol® was added to each sample of dissected cells. Samples were vortexed briefly and incubated at room temperature for 5 min. Then 200 μl of chloroform was added and the samples were vortexed vigorously for 15 sec. After incubation at room temperature for 3 min, samples were centrifuged at 12,000g and 4°C for 15 min. The upper aqueous phase was transferred to a new tube and 500 μl of isopropanol was added for RNA precipitation. Samples were vortexed briefly and allowed to sit at room temperature for 10 min. They were then centrifuged at 12,000g and 4°C for 10 min. The supernatant was removed and 1 ml of 75% ethanol in DEPC-water was used for washing of the pellet. Samples were then centrifuged at 7,500g and 4°C for 5 min. The supernatant was removed and the pellet was allowed to dry for 10 min at 58°C. The pellet was dissolved in 8 μl of DEPC-water. The concentration and integrity of RNA was assessed by using Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA).

Quantitative real-time reverse transcription-polymerase chain reaction

For measurement of miRNA expression, reverse transcription (RT) was performed using the TaqMan® MicroRNA Reverse Transcription Kit and TaqMan® Human MicroRNA Assays (Applied Biosystems, Foster City, CA). Fifteen nanograms of total RNA was used for RT using stem-loop RT primers for each miRNA. For each reaction, 0.5 μl of 10× reaction buffer, 0.05 μl of 100 mM dNTP, 0.35μl of MultiScribe™ enzyme (50U/μl), 0.063 μl of RNase inhibitor (20U/μl) and 1 μl of 5× RT primer were mixed. DEPC-water was added to reach the final volume of 5 μl. The reaction mixture was incubated at 16°C for 30 min, 42°C for 30 min and 85°C for 5 min in GeneAmp® 9700 PCR system (Applied Biosystems). qPCR of cDNA obtained from miRNA was performed using TaqMan® Universal PCR Master Mix and TaqMan® Human MicroRNA Assays (Applied Biosystems). The RT product for each miRNA was diluted by 2.4-fold using DEPC-water. For each PCR reaction, 2.5 μl of 2× master mix, 1 μl of diluted RT product, 0.5 μl of 10× primers and probe mix for each miRNA assay and 1 μl of milli-Q water was mixed. The reaction mixture was incubated in a 384-well optical plate at 95°C for 10 min, and 40 cycles of 95°C for 15 sec and 60°C for 1 min in 7900HT Real-Time PCR system (Applied Biosystems). The threshold cycle (Ct) data was determined using default threshold settings. The Ct is defined as the fractional cycle number at which the fluorescence passes the fixed threshold. Relative quantification of miRNA expression was calculated with the 2^−ΔΔCt method (Applied Biosystems User Bulletin N°2, P/N 4303859). We used endogenous control to normalize the expression levels of miRNA by correcting differences in the amount of cDNA loaded into PCR reactions. In this study of miRNA expression, hsa-let-7a was used as an endogenous control as suggested by Applied Biosystems. Our own data from endometrial cancer and normal tissues showed that hsa-let-7a exhibited relatively even expression levels. The data were presented as log₁₀ of relative quantity of target miRNA, normalized with respect to hsa-let-7a. To remove any non specific variation that might be caused by hsa-let-7a used for normalization in endometrial tissue, we also tested another control, U6 that was used in many studies of miRNAs recently, in a set of 4 study samples. There was no significant bias in comparison of normalization between hsa-let-7a and U6.

Effects of anti-miR-205 miRNA inhibitor on endometrial cancer cells in vitro

The endometrial cancer cell line RL95-2 was purchased from ATCC (American Type Culture Collection, Manassas, VA) and maintained according to the ATCC’s protocol. All transfections were carried out in triplicate. To transfect, 5 pmol of anti-miR-205 miRNA inhibitor (Ambion, Austin, TX) was diluted into 10 μl Opti-Mem into each well. Then 0.3 μl NeoFx (Ambion) was diluted into 10 μl Opti-MEM for each sample, incubated for 10 min at room temperature and 10 μl of diluted transfection mixture was added to wells that already contained the inhibitor, and incubated for another 10 min at room temperature. Next, 100 μl of diluted cell suspension mixture containing 8,000 cells were added on top of the complex. After 24 hr, the medium was changed and the samples were assayed after 48 hr. An oligonucleotide negative control (Control #1, Ambion) was also included in the study.

To determine miR-205 expression in the RL95-2 cells after transfection of anti-miR-205 miRNA inhibitor, total RNA was extracted from the cells with and without treatment of anti-miR-205 miRNA inhibitor. Quantitative real-time reverse transcription-polymerase chain reaction (qRT-PCR) was carried out in the same way as mentioned above.

To determine the protein expression of one predicted miR-205 target gene, JPH4 in the RL95-2 cells with and without transfection of miR-205 inhibitor, Western blot analysis was performed. The cell samples were first denatured at 100°C for 2 min with addition of 3× sample loading buffer. Fifty micrograms of samples were loaded into the wells. SDS-PAGE was performed with the Mini-PROTEAN III electrophoresis cell (Bio-Rad, Hercules, CA) and blotting was carried out by Mini Trans-Blot electrophoretic transfer cell (Bio-Rad) according to manual instructions. The membrane was washed by 1× TBS buffer briefly after the blotting, followed by incubating in 5% (w/v) skimmed milk in 1× TBST blocking solution at room temperature for 1 hr with gentle shaking. The membrane was incubated with primary antibody at 4°C overnight and secondary antibody at room temperature for 2 hr. The antibodies were diluted to optimal dilutions with 5% skimmed milk in 1× TBST blocking solution. Primary antibody used was rabbit anti-JPH4 (1:200; Chemicon, Billerica, MA). Secondary antibody used was affinity purified donkey anti-rabbit IgG horseradish peroxidase conjugated (1:2000; GE Healthcare, Piscataway, NJ).

Data analysis

Bioinformatic analysis was conducted to determine hierarchical clustering and to identify miRNAs most significantly aberrantly regulated using dChip software (http://www.dChip.org). To analyze correlation between miRNA expression and clinico-pathological characteristics in EEC, Pearson correlation analysis was performed to detect association between two continuous variables. Independent t-test was carried out to compare the mean difference between two groups. Analysis of variance was performed to compare the mean differences among three groups. SPSS for Windows version 14.0 was used for all statistical analysis. All tests were 2-sided and p value denotes significance if its value was less than 0.05. As there were 10 cancer patients with less than 5 years of follow-up in the study population, analysis of survival was not carried out. Predicted targets of miRNAs were searched from web database: http://diana.cslab.ece.ntua.gr/microT/

Results

Deregulated miRNAs in EEC compared with normal endometirum

Initially, profiling of 157 miRNA was completed in 30 EECs including 25 stage I–II and 5 stage III cases, and 22 normal endometrium controls including 7 which were in the proliferative phase, 7 in secretary phase and 8 were postmenopausal. All of these endometrial samples have been previously studied for global mRNA gene expression profiling by us.¹ To disclose differentially expressed miRNAs, which might contribute to the development or progression of EEC, a clustering analysis was performed. Using dChip program, 51 miRNAs satisfying the filtering criteria were differentially related in EEC versus normal control at a p value <0.01 and that possessed a mean fold-change >2.0. These included 48 upregulated and 3 downregulated miRNAs (Fig. 1). When defined by a stringent criteria of 90% lower confidence bound fold-change >2.0 with a mean difference ≥100, a total of 30 miRNAs were found to be significantly dysregulated in EEC compared with normal controls. The permutation analysis returned a 0.5% median false discovery rate. These 30 miRNAs were all upregulated in this set of EEC. Table II shows these 30 highly differentiated miRNAs and their chromosomal location. The lower bound fold-change observed in EEC in relation to normal controls varied between 2.0- and 27.2-fold upregulation.

Figure 1 –

A dendrogram plotted from hierarchical clustering of 51 miRNAs in 52 endometrial samples shows miRNAs deregulated in primary EEC (C) compared with normal endometrium (N). Red represents the upregulated miRNA expression whereas blue represents the downregulated miRNA expression.

TABLE II –

THIRTY miRNAs DIFFERENTIALLY EXPRESSED IN EEC VERSUS NORMAL ENDOMETRIAL CONTROL (LOWER BOUND FOLD-CHANGE > 2.0), THEIR CHROMOSOME LOCATION AND PREDICTED TARGET GENES

miRNA	Lower bound of fold-change	Chromosome localization	Putative target genes associated with EEC (lower bound fold-change)1	Reported in other tumors
hsa-miR-205	27.2	1q32.2	PEG3 (−9.08), P2RY14 (−4.82), JPH4 (−4.59), ECM2 (−4.18), S100A2 (5.73)	Ovary,13 lung,14 bladder,15 head and neck16
has-miR-182	8.11	7q32.2	OLFM1 (−7.4), LMOD1 (−5.65), TCEAL7 (−5.65), PDGFRA (−5.24), DCN (−5.0), DKFZP564O0823 (−4.93), EDRNB (−4.26), COL3A1 (−4.1), S100A2 (5.73)	Nil
hsa-miR-325	7.92	Xq21.1	F10 (−7.05), C2orf32 (−5.49), DCN (−5.0), EFEMP1 (−4.86), EHD2 (−4.45)	Nil
hsa-miR-183	6.8	7q32.2	FLRT2 (−6.41), FBN1 (−5.26), FOXL2 (−4.39), TCF8 (−4.18), COL3A1 (−4.1), HP (4.09), DKK4 (46.9)	Bladder, colonrectal15,17
hsa-miR-203	6.24	14q32.33	KLRC2 (−11.01), OLFM1 (−7.4), ALDH1A2 (−7.01), ATP8A2 (−6.58), FLRT2 (−6.41), JPH4 (−4.59), EFHA2 (−4.19), ZIC1(30.8)	Bladder15
hsa-miR-210	5.23	11q15.5	PAK3 (−7.06), SRPX (−6.99), C2orf32 (−5.49), GAS6 (−5.35), EHD2 (−4.45), SIN3B (−4.03), SERPINA3 (10.76)	Endometrium18
hsa-miR-223	4.9	Xq12	OLFM1 (−7.4), RBP7 (−6.39), OGN (−5.79), ADAM33 (−5.1), CTSW (−5.05), TNXB (−4.76), PITX1 (4.68)	Bladder15
hsa-miR-194	4.11	1q41	PDGFRA (−5.24), FOXC1 (5.21), FOXA1 (5.99), ZIC1(30.8)	Nil
hsa-miR-95	4.09	4p16.1	DCN (−5.0), S100A2 (5.73), SERPINA3 (10.76)	Cervix19
hsa-miR-155	3.87	21q21.2	ZNF537 (−5.94), OGN (−5.79), EFHA2 (−4.19), TFAP2A (6.57)	Breast, lung, pancreas, chronic lymphocytic leukemia9,14,20–22
hsa-miR-200a	3.69	1p36.33	SLC18A2 (−9.19), OLFM1 (−7.4), ATP8A2 (−6.58), TRO (−6.3), C2orf32 (−5.49), TCF8 (−4.18), FOXC1 (5.21), FOXA1 (5.99)	Ovary13
hsa-miR-301	3.58	17q23.2	SLC24A3 (−7.28), SRPX (−6.99), MUM1L1 (−6.92), RORB (−5.88), FBN1 (−5.26), PDGFRA (−5.24), JPH4 (−4.59), TFAP2A (6.57), ZIC1 (30.8)	Pancreas20,21
hsa-miR-141	3.11	12p13.31	SLC18A2 (−9.19), OLFM1 (−7.4), ATP8A2 (−6.58), FLRT2 (−6.41), TRO (−6.3), C2orf32 (−5.49), ZNF537 (−4.94), TNXB (−4.76), FBXO32 (−4.46), TCF8 (−4.18), CXCL12 (−4.07), FOXC1 (5.21), FOXA1 (5.99), KIF1A (20)	Ovary13
hsa-miR-215	3.03	1q41	MUM1L1 (−6.92), FOXA1 (5.99)	Nil
hsa-miR-103	3	5q34	OLFM1 (−7.4), PAK3 (−7.06), SRPX (−6.99), ADAMTS5 (−5.86), TCEAL7 (−5.65), PDGFRA (−5.24), ADAM33 (−5.1), TFAP2A (6.57), CXCL5 (7.72)	Endometrium, bladder, pancreas15,18,23
hsa-miR-106a	2.86	Xq26.2	OLFM1 (−7.4), ADAMTS5 (−5.86), TCEAL7 (−5.65), PDGFRA (−5.24), ZNF537 (−4.94), TBX3 (−4.81), FOXL2 (−4.39), FOXA1 (5.99), KIF1A (20), ZIC1 (30.8)	Endometrium, neuroblastoma, T-cell leukemia18,24,25
hsa-miR-151	2.85	8q24	FBN1 (−5.26), TCF8 (−4.18), SERPINA3 (10.76)	Nil
hsa-miR-107	2.84	10q23.31	OLFM1 (−7.4), PAK3 (−7.06), SRPX (−6.99), TRPC1 (−6.16), ADAMTS5 (−5.86), PDGFRA (−5.24), ADAM33 (−5.1), TFAP2A (6.57), ZIC1 (30.8)	Endometrium, pancreas, acute promyelocytic leukemia18,23,26
hsa-miR-330	2.76	19q13.32	KLRC2 (−11.01), EFEMP1 (−4.86), IMP-3 (17.54)	Nil
hsa-miR-25	2.57	7q22.1	SLC24A3 (−7.28), FLRT2 (−6.41), C2orf32 (−5.49), FBN1 (−5.26), PDGFRA (−5.24), TBX3 (−4.81), CXCL5 (7.72), ZIC1 (30.8)	Nil
hsa-miR-28	2.4	3q28	GNLY (−19.72), OLFM1 (−7.4), GAS6 (−5.35), ADAM33 (−5.1), EMILIN1 (−4.67), CXCL12 (−4.07)	Nil
hsa-miR-130b	2.36	22q11.21	OLFM1 (−7.4), SLC24A3 (7.28), PAK3 (−7.06), SRPX (−6.99), MUM1L1 (−6.92), RORB (−5.88), PDGFRA (−5.24), JPH4 (−4.59), TCF8 (−4.18), S100A2 (5.73), FOXA1 (5.99)	Nil
hsa-miR-17-5p	2.36	13q31	OLFM1 (−7.4), FLRT2 (−6.41), TCEAL7 (−5.65), ADAMTS5 (−5.53), PDGFRA (−5.24), ADAM33 (−5.1), ZNF537 (−4.94), TBX3 (−4.81), FOXL2 (−4.39), PITX1 (4.68), FOXA1 (5.99), ZIC1 (30.8)	Bladder and lung15,27
hsa-miR-10a	2.34	17q21.32	FLRT2 (−6.41), TRO (−6.3), TNXB (−4.76), RAMP1 (−4.27), ECM2 (−4.18), TFAP2A (6.57)	Acute myeloid leukemia28
hsa-miR-191	2.16	3p21.31	SLC24A3 (−7.28), HGF (−7.08), OGN (−5.79), DCN (−5.0), S100A2 (5.73), SERPINA3 (10.76)	Bladder, leukemia15,28
hsa-miR-184	2.13	15q25.1	TRO (−6.3), GAS6 (−5.35), CTSW (−5.05), EHD2 (−4.45), BST2 (5.1), TFF3 (9.67), KIF1A (20)	Neuroblastoma29
hsa-miR-326	2.13	11q13.4	SLC24A3 (−7.28), FBN1 (−5.26), ADAM33 (−5.1), LTBP4 (−4.96), EMILIN1 (−4.67), RAMP1 (−4.27), KIF1A (20)	Nil
hsa-miR-34a	2.13	1p36.22	FLJ20701 (−9.16), PDGFRA (−5.24), P2RY14 (−4.82), LTF (4.79)	Ovary, lymphocytic leukemia13,30
hsa-miR-200c	2.1	12p13.31	C2orf32 (−5.49), NDN (−5.39), TCF8 (−4.18), TFAP2A (6.57)	Ovary13
hsa-miR-23a	2.01	19p13.13	PAK3 (−7.06), ALDH1A2 (−7.01), SRPX (−6.99), PLEKHH2 (−5.82), FBN1 (−5.26), PDGFRA (−5.24), DCN (−5), ZNF537 (−4.94), FBXO32 (−4.39), FOXL2 (−4.39), EFHA2 (−4.19), TCF8 (−4.18), SFRP4 (−4.09), CXCL12 (−4.07), FOXC1 (5.99), ALDH1A2 (7.01), CXCL5 (7.72), Z1C1 (30.8)	Lung14

¹Bold genes are upregulated in EEC displayed in previous global gene expression profiling.

Validation

To validate the global miRNA expression profiling, 14 upregulated miRNAs (miR-95, miR-103, miR-106a, and miR-151, miR-155, miR-182, miR-183, miR194, miR-200a, miR-200c, miR-203, miR-205, miR-210, miR-223) that displayed significantly differential expression between the tumor and normal control specimens and showed relatively even level of expression across each set of endometrial samples (tumor and control), were selected for further qRT-PCR analysis in an independent new set of 8 primary cancer (6 at stage I–II and 2 at stage III) and 4 normal (1 in proliferative phase, 1 in secretory phase and 2 postmenopausal) samples. We noted that miR-325 showing a 7.92-fold lower bound change in tumor, but did not include this for validation. This was because only a few of samples in both tumor and control sets had detectable expression of this miRNA. For all 14 miRNAs, changes in expression pattern observed in the validation set agreed favorably with that in the training set (Fig. 2). Because of limited number of new specimens available, only a total of 12 endometrial samples were used in the validation. However, the result of validation would still be very indicative as the same method of qRT-PCR and protocol were used to determine the miRNA expression.

Figure 2 –

Validation of 14 miRNAs of interest in an independent set of EEC.

Correlation between miRNA deregulation and clinico-pathological characteristics of EECs

We collected information on clinico-pathological parameters to search for any possible associations with expression of the 30 highly dysregulated miRNAs. The mean age of patients with the cancer was 60.9 (40.2–84.6) years, whereas for the controls it was 63.8 (47.2–81.6) years, which was not significant. The correlation of dysregulated miRNA expression with clinico-pathological characteristics showed the aberrant expression of miR-200a and miR-205 was correlated with advanced stage (p < 0.05); the aberrant expression of miR-205 was correlated with myometrium invasion and recurrence (p < 0.05); the aberrant expression of miR-10a, miR-34a and miR-95 was positively correlated with lymph node involvement (p < 0.05).

Predicted gene targets of deregulated miRNAs in EEC

By web searching for predicted miRNA targets that were linked to the dysregulated genes previously revealed in global gene expression profiling of these EECs using oligonucleotide microarray, a total of 68 aberrantly regulated genes were predicted to be target genes of these 30 miRNAs of interest in ECC (Table II). As shown in Figure 3, the 68 genes seem to be regulated by one or more of the 30 miRNAs.

Figure 3 –

Number of miRNAs that regulate 68 predicted target genes in EEC.

Effect of dysregulated miR-205 on protein expression of its target gene in endometrial cancer cells in vitro

miR-205 was significantly overexpressed in EEC with a 27.2-fold increase compared with normal control. After the transfection of miRNA 205 inhibitor for 48 hr, the expression of miR-205 in RL95–2 cells was decreased by 64.9% whereas its predicted target gene JPH4 showed increased protein expression by 14.2% when studied using Western blot assay.

Discussion

miRNAs represent a new class of small noncoding RNAs (ncRNAs) that can regulate gene expression by targeting mRNAs of protein coding genes. miRNAs are ubiquitously expressed and have been observed to be dysregulated in almost all human cancers examined, characterized usually by abnormal levels of expression for mature and/or precursor miRNA transcripts in comparison with the corresponding normal tissues.^31,32 Global miRNA profiling allows the identification of signatures associated with diagnosis and prognosis as well as therapeutic targets of human tumors and thus miRNAs fingerprinting represents a new additional tool for the study of cancer.

To conduct a global assessment of differential miRNA expression in EEC in Chinese Hong Kong women, we first used a real-time qRT-PCR method to determine the expression of 157 mature miRNAs using total RNA extracted from microdissected EECs and their normal counterparts. Real-time RT-PCR is a sensitive and specific for detecting expression of mature miRNAs. It is particularly well suited to the study of low abundance RNAs, and samples from limited amounts of tissue such as that provided by laser capture microdissection.

The first objective of this study was to reveal changes in miRNA expressions that can provide insight into events characteristic of endometrial malignant transformation. However, the counterpart of endometrial cancer, i.e., normal endometrium, is dynamic in that it undergoes cyclic phenotypic changes. One may conceptualize these cyclic changes as a reciprocal interplay between the mitogenic effects of estrogen during the (follicular) proliferative phase and a dominant progesterone effect in the (luteal) secretory phase. Therefore, we chose to analyze both of these phases in the miRNA expression profiling of “normal endometrium.” In addition, postmenopausal “atrophic” endometrium was also collected for miRNA expression profiling. This aims to reduce the bias of the control profiles influenced by menstrual cycle.

Initial profiling of global miRNA expression indicated 30 miRNAs significantly upregulated between normal and EEC samples (>2-fold). Very recently Boren et al.¹⁸ published their study in which they measured expression of 335 miRNAs in 37 endometrial cancers, 20 normal endometrium, and 4 complex atypical hyperplasia samples and identified 13 miRNAs (miR-let7i, miR-let7c, miR-30c, miR-103, miR-106a, miR-107, miR-152, miR-181a, miR-185, miR-193, miR-210, mir-223 and miR-423) to be associated with endometrial cancer development. Among the 13 miRNAs with dysregulation in the endometrial cancer, miRNA 103, miR-106a, miRNA 107 and miRNA 210 were also found to be dysregualted in our set of EEC. In the set of their predicted target genes of these 4 miRNAs, only C2orf32 gene, as the predicted target of miRNA 210, was shown to be dysregulated in both Boren’s report and ours. The aberrant expression of these 4 miRNA shown in both studies suggests that these be given high priority for evaluating their functional role in the initiation and progression of endometrial cancer. Meanwhile we also found another 9 of the 13 miRNAs which did not appear in both profiles of endometrial cancers. It is reasonable to postulate that the following may explain some of the discrepancy: different methods and not same miRNA sets were used in the measurement of miRNA expression; different ethnic populations and Boren’s study included a variety of histologic types of endometrial cancer whereas we included endometrioid type cancer only. In all, 20 of the 30 differentially expressed miRNAs have previously been reported as being differentially expressed in other cancers including pancreatic, bladder, head and neck, ovarian, cervix and various leukemias (Table II). These 20 miRNAs are widely expressed in many tissues and thus may have a more general function in tumor initiation and progression. The remaining 10 miRNAs including miR-182, miR-325, miR-194, miR-215, miR-151, miR-330, miR-25, miR-28, miR-130b and miR-326 have not been reported to be associated with any human tumors yet. More studies need to be conducted to confirm whether these 10 miRNAs are differentially modulated in other cases of EEC.

miR-205 was found to be highly overexpressed in tumor relative to nontumor control samples, and was also related to advanced stage and myometrium invasion, suggesting this miRNA may be involved in both carcinogenesis and acquisition of a more aggressive phenotype. MiR-182 and miR-183 are expressed from a common transcript, and were both found to be upregulated ~7- to 8-fold in tumor samples, providing a degree of confidence and internal validation of the accuracy of our RT-PCR quantification of different miRNA species. In addition, this study also revealed overexpression of miR-34a, miR-95 and miR-200a were associated with tumor progression, suggesting these miRNAs are candidate prognostic markers or therapeutic targets. There is some correlation of dysregulated miRNAs with clinico-pathological characteristics of EEC evident in this study suggesting it may be possible to identify more convincing miRNA signatures in EEC using a larger cohort. It will also be interesting to further analyze miRNAs association with clinical outcomes.

miRNAs repress translation and modulate stability of their target mRNAs. We predicted that mRNA level of target genes for overexpressed miRNAs would be lower in tumor samples. Putative target genes of these miRNAs were identified using the web database http://diana.cslab.ece.ntua.gr/microT/. We have previously established global mRNA expression profiles in EEC.¹ In the present study we performed parallel profiling of global miRNA expression that facilitated integrated analysis of potential molecular signatures in EEC at both miRNA and mRNA levels. We only included highly differentially regulated genes, which distinguished EEC from normal endometrium for analysis. We were able to generate a list of 68 potential mRNA binding partners for 30 miRNAs that were observed to be differentially expressed in EEC. Notably these miRNAs potentially regulate a number of genes that govern cell signaling, proliferation and apoptosis. The significance of such a correlation between the data sets is speculative at this point but provides a platform from which to design further studies into the relationship between miRNAs and gene expression profiles and EEC.

Although a significant number of predicted miRNA targets are downregulated, some mRNAs are also elevated in the tumor samples as shown in this study. The later mRNAs may not be true actual targets of these miRNAs, because they may be expressed as splice or UTR variants that do not contain the miRNA target site, or they may have acquired mutations in their miRNA binding sites during tumor progression, allowing them to avoid regulation. Furthermore, the protein level of these genes may still not be elevated, despite the increase in mRNA. The avoidance of miRNA repression by mRNAs is an area of significant interest and will be explored further in these EEC samples.

To validate the function of an upregulated miRNA in EEC, we used specific antisense miRNA inhibitors to knock-down the expression of particular miRNAs and monitored protein expression changes in a candidate target. miR-205 has been found to be significantly upregulated in lung, bladder, ovarian, head and neck and cancer cell lines.^13–16 Very recently, two groups (Gregaory et al.³³ and Park et al.³⁴) identified miR-205 in epithelial-mesenchymal transition (EMT) though the direct regulation of ZEB1 and ZEB2 that may control expression of E-cadherin related to tumor metastasis. However, the case for a potential role of the miR-205 in metastasis is less convincing so far. It will be of great interest to explore the role of miR-205 in tumor metastasis further.³⁵ In this study, miR-205 expression was found to be significantly elevated in EEC when compared with controls. In endometrial cancer RL 95-2 cell line, expression of miR-205 was decreased by 64.9% after transfection of 5 pmol miR-205 inhibitor when compared with control, whereas expression of a predicted target gene JPH4 was increased by 14.2% at 48 hr. Although this de-repression of JPH4 is moderate, one should note that only 64.9% miR-205 inhibition was achieved. The protein half-life may be longer than the assay time-point, and that other miRNAs may be expressed in cells that contribute to JPH4 repression. Nevertheless, our previous study showed JPH4 downregulated by 4.59-fold in EECs, and thus our experiments suggest the JPH4 gene is a real miR-205 target in vitro and in vivo, and thus a candidate tumor suppressor gene.¹ Surveying further predicted target genes may provide a clearer picture of the targets of miR-205, and in particular the ability to observe effects of the treatment on cell phenotypes such as proliferation and apoptosis. Combination treatment with other miRNA inhibitors and/or mimics may enhance this effect. In addition, this study has also identified novel miRNA candidates that have not previously been associated with cancer. The investigation of their roles in cellular proliferation is warranted.

This study demonstrated the distinct miRNA dysregulation profiles in endometrial cell transformation and progression in Hong Kong women with EEC. Some of the miRNAs have been previously found to be dysregulated in other cancers while a subset of them has not been reported before. These profiled miRNAs serve as an important starting point for the identification of novel clinical markers and therapeutic targets. Unique miRNA signatures, novel molecular markers and candidate therapeutic targets may be further identified through expanded miRNA expression profiling. Using enhanced miRNA libraries of >400 sequences, we are now expanding this dataset in the hope of identifying further dysregulated miRNAs. The incorporation of mRNA and miRNA expression profiles that we report here is a significant advance towards generation of comprehensive transcriptome signatures of mRNAs and miRNAs in EEC.