Deregulation of RAD21 and RUNX1 expression in endometrial cancer

Abstract

Cohesins and cohesin-regulated genes are deregulated in numerous types of human cancer. However, data concerning their status and role in endometrial cancer are scarce. This study aimed to determine the clinical significance of double-strand-break repair protein rad21 homolog (RAD21) and runt-related transcription factor 1 (RUNX1) gene dosage and mRNA expression in endometrial cancer. RAD21 is a component of the cohesin complex, crucial for chromosome segregation and DNA repair. RUNX1 is the transcription factor implicated in RAD21 regulation. The study group included 144 endometrial cancer patients. RAD21 and RUNX1 expression profiles were measured by reverse-transcription quantitative PCR. RAD21 gene dosage was determined by quantitative PCR. RAD21 gene dosage was associated with RAD21 mRNA expression (ϱ=0.22; p=0.009). Furthermore, RAD21 expression strongly correlated with RUNX1 expression (ϱ=0.43; p<0.0000001). Increased RAD21 gene dosage correlated with more advanced tumor stage (p=0.021), higher grade (p=0.021), cervical involvement (p=0.01) and the absence of obesity (p=0.025), while RAD21 mRNA expression correlatd with cervical involvement (p=0.027). The mRNA expression of RAD21 and RUNX1 was found to be deregulated and co-dependent in endometrial cancer. RAD21 gene dosage is associated with unfavorable tumor characteristics. However, elucidating the role of these molecular markers in endometrial oncogenesis requires further investigation, including functional studies and survival analysis.

Introduction

Endometrial cancer is the most frequent malignancy of the female genital tract, with an estimated 46,470 cases and 8,120 mortalities expected to be recorded in 2011 in the USA (1). Despite such high prevalence, the understanding of its molecular background in terms of genesis, growth and progression remains insufficient. Furthermore, little is known concerning factors which would allow for the differentiation between types I and II endometrial cancer, which differ substantially in prognosis.

In view of the poor understanding of the molecular background of endometrial cancer we have attempted to identify new markers which may: i) correlate with patients’ clinico-pathological features; ii) further elucidate molecular pathways of endometrial carcinogenesis; iii) aid the differentiation of types I and II.

One of the key elements to be investigated in this particular context are cohesins: multisubunit protein complexes which are highly conserved and play canonical roles in processes such as chromatin regulation, chromosome segregation and DNA damage response (2–4). Is has been shown that cohesin-defective cells possess features known to be crucial drivers of oncogenesis. These features include genomic instability, impaired DNA repair and anomalies concerning gene expression (4–7). The deregulation of cohesin expression and cohesin-regulated genes is common in numerous types of human cancer (8–13), including endometrial cancer (14). Furthermore, cohesin-defective cells have been discovered to be sensitive to ionizing radiation and DNA-damaging drugs (8,15).

RAD21 (double-strand-break repair protein rad21 homolog), a mammalian ortholog of Mcd1p, is one of the four core proteins comprising a cohesin ring in sister chromatid cohesion (SSC), a physical linkage between sister chromatids. SSC allows for cell cycle checkpoint control and homologous repair of DNA double-strand breaks (16). Experiments performed on zebrafish revealed rad21 to be a regulator of runx1 (6). RUNX1/AML1 (runt-related transcription factor 1/ acute myeloid leukemia 1) belongs to the family of RUNX transcription factors which, when complexed with other proteins, activates or represses the transcription of regulators involved in cell differentiation, growth and survival. The RUNX genes function as tumor suppressors and dominant oncogenes, depending on the context (17).

RAD21 and RUNX1 actions are crucial for sustaining basic functions in healthy cells. These two markers have been found to be deregulated in different types of tumors, including endometrioid, prostate, breast and oral squamous carcinoma together with acute lymphoblastic leukemia (8,9,11,12,14,18–25). The present study was designed to address the hypothesis of cohesin deregulation in endometrial cancer. It aimed to investigate RAD21 and RUNX1 mRNA expression profiles with the use of reverse transcription quantitative PCR in endometrial cancer tumors. Additionally, RAD21 mRNA expression was compared with RAD21 gene dosage measured by quantitative PCR, as it has been shown that copy number variations (CNVs) are common in various types of cancer (26) and that some of them may contribute to aberrant cohesin expression in cancer (9,16).

Materials and methods

Patients and tissues

The retrospective study encompassed 144 frozen tumor samples collected from a cohort of endometrial cancer patients treated at the Department of Gynaecology, Gynaecological Oncology and Gynaecological Endocrinology (Medical University of Gdańsk, Gdańsk, Poland) between 2005 and 2011. The inclusion criteria were operable endometrial cancer confirmed by histological examination and a signed consent form. The characteristics of the patients are summarized in Table I. The mean age was 63.3 years (range, 30–87). The study was accepted by the Ethics Committee of the Medical University of Gdańsk.

Table I.

Clinicopathological data (n=144).

Variable	Number of cases (%)
Menopausal status
Premenopausal	9 (6.3)
Postmenopausal	135 (93.7)
Obesity
Absent	43 (29.9)
Present	54 (37.5)
Missing data	47 (32.6)
Ca-125 status
Negative	86 (59.7)
Positive	11 (7.6)
Missing data	47 (32.6)
Histology
Endometrioid	135 (93.7)
Nonendometrioid	9 (6.3)
Stage (FIGO)
IA–IB	107 (74.3)
II	19 (13.2)
IIIA–IIIC	14 (9.7)
IVA–IVB	3 (2.1)
Missing data	1 (0.7)
Grade
I	53 (36.8)
II	61 (42.4)
III	22 (15.3)
Missing data	8 (5.6)
Lymph node status
Negative	39 (27.1)
Positive	8 (5.6)
Missing data	97 (67.4)
Myometrial infiltration
≤1/2	81 (56.3)
>1/2	62 (43.1)
Missing data	1 (0.7)
Cervical invasion
Absent	104 (72.2)
Present	39 (27.1)
Missing data	1 (0.7)
Metastases
Absent	102 (70.8)
Cervix	18 (12.5)
Cervix and other organs	13 (9)
Other organs	9 (6.3)
Missing data	2 (1.4)
ESR1 status
Positive	37 (25.7)
Negative	103 (71.5)
Missing data	4 (2.8)

FIGO, International Federation of Gynecology and Obstetrics; ESR1, estrogen receptor 1 gene.

Tumor samples were collected by surgical excision prior to any systemic treatment and were immediately frozen and stored at −80°C. Tissue samples covered the spectrum of pathological stages of endometrial carcinoma, from noninvasive IA to metastatic IVB cancer according to the staging by FIGO in 2009 (International Federation of Gynecology and Obstetrics) (27). A Ca-125 level between 0 and 35 U/ml was considered normal (28). Patients with a body mass index >30 were classified as obese (29).

DNA and RNA isolation

Prior to nucleic acid isolation, tissue specimens (25 mg per sample) were homogenized (1 min, 6,000 rpm) with the use of MagNA Lyser (Roche, Basel, Switzerland). DNA and RNA were isolated with AllPrep DNA/RNA Mini kit (Qiagen, Hilden, Germany) using the tissue protocol, in accordance with the manufacturer’s instructions. After the isolation, DNA/RNA concentration and purity were determined by Spectrophotometer ND-1000 (NanoDrop Technologies, Wilmington, DE, USA). Good quality DNA was defined as an A_{260 nm/280 nm} ratio between 1.70 and 1.90. Good quality RNA was defined as an A_{260 nm/280 nm} ratio of ∼2.

RNA was subsequently reverse transcribed to cDNA with the Transcriptor First Strand cDNA Synthesis kit (Roche), according to the manufacturer’s instructions, with the use of random hexamer primers. There was 1000 ng of total RNA per reaction.

Quantitative PCR

Control DNA and RNA from five frozen samples of healthy donors were isolated, pooled and used for qPCR assay optimization as well as a calibrator. Analysis was performed with StepOnePlus™ Instrument (Applied Biosystems, Carlsbad, CA, USA). Each new set of the master mix was verified by a standard curve. The thermal profiles used were the default settings of the manufacturer, dedicated to either SYBR-Green or TaqMan probe assays. Results were analyzed and reported with the use of StepOne Software v2.1.

Gene dosage analysis

RAD21 and ESR1 (estrogen receptor 1) gene copy numbers were determined by qPCR with Power SYBR-Green Master mix (Applied Biosystems), using the APP (amyloid precursor protein) gene as a reference. APP was selected as a reference gene upon a search performed in the Atlas of Genetics and Cytogenetics in Oncology and Haematology (http://www.atlasgeneticsoncology.org/). Its stability against 3P (RNA, U4 small nuclear pseudogen) and SOD2 (superoxide dismutase 2) genes was verified using geNorm software (30). The primer sequences were as follows: APP F, 5′-AGC CCA GAA GGT GTC AAA CA-3′; APP R, 5′-CAT CTT CAT GTC CGT TGC AT-3′; RAD21 F, 5′-GGC ACT GTT ACC ACA AAC CTT TGG-3′; RAD21 R, 5′-GGG GAC ATT TGA ATG CTG ACT GGC-3′; ESR1 F, 5′-ACA TGG ACA CCT CCC AGT C-3′; ESR1 R, 5′-ACA GAC TAA CAC AGC CCA TC-3′. The quantity of DNA used per well was 100 ng.

RAD21 and ESR1 copy number was calculated using the ΔΔCt quantification method (31), which relates the gene dosages of a studied gene and a reference gene in the tumor tissue and a calibrator. The reactions were performed in duplicate on 96-well plates; a negative control for each gene and three calibrators were included on each plate.

We used experimentally determined cut-off values calculated using the critical difference parameter, as described previously (32). The amplification of RAD21 and ESR1 was classified as a relative quantity >1.36 and 1.14, respectively.

mRNA expression analysis

RAD21 and RUNX1 RNA expression levels were determined by qPCR with TaqMan® Universal PCR Master mix (Applied Biosystems), using HPRT1 (hypoxanthine phosphoribosyltransferase 1) as a reference. HPRT1 gene expression stability was verified against the expression of GAPDH (glyceraldehyde 3-phosphate dehydrogenase) and ACTB (β-actin) genes. TaqMan^® Expression Assays (Applied Biosystems) used were as follows: HPRT1 Endogenous Control Hs99999909_m1; RAD21 Gene Expression Assay Hs01085854_mH and RUNX1 Gene Expression Assay Hs01021967_m1. The quantity of cDNA per well was 75 ng.

RUNX1 and RAD21 expression was also calculated using the ΔΔCt quantification method. Reactions were performed in triplicate on 96-well plates; on each plate two negative controls for each gene and four calibrators were included. RUNX1 and RAD21 overexpression was classified as a value 2-fold higher than the value in the calibrator sample.

Statistical analysis

All statistical analyses were performed using the STATISTICA software, version 10. Logarithmized relative quantities of RAD21 gene dosage together with RUNX1 and RAD21 expression levels were assessed by Spearman correlation and Crosstabs statistics with Pearson’s chi-square test. Various comparisons of the results and clinicopathological data were performed with the nonparametric statistics, including the Mann-Whitney U test (Table II). P<0.05 was considered to indicate a statistically significant result.

Table II.

RAD21 and RUNX1 status with regard to clinicopathological data.

Variable	RAD21 gene dosage			RAD21 expression			RUNX1 expression
	n	Average ± SD	P-value	n	Average ± SD	P-value	n	Average ± SD	P-value
Menopausal status
Premenopausal	9	1.15±0.27	0.569	9	1.90±1.21	0.269	9	5.10±5.42	0.286
Postmenopausal	132	1.10±0.35		135	1.50±0.91		132	3.56±9.00
Obesity
Absent	41	1.18±0.30	0.025	43	1.39±0.82	0.069	40	4.81±15.34	0.611
Present	53	1.08±0.44		54	1.66±1.09		54	3.07±3.90
Ca-125
Negative	83	1.09±0.26	0.284	86	1.57±1.04	0.629	84	3.87±10.90
Positive	11	1.41±0.86		11	1.28±0.35		10	3.26±4.73	0.718
Histology
Endometrioid	133	1.11±0.35	0.969	135	1.51±0.91	0.975	132	3.61±9.05	0.073
Nonendometrioid	7	1.11±0.26		8	1.76±1.37		8	4.83±3.68
Stage
I, II	125	1.07±0.23	0.021	126	1.55±0.97	0.564	124	3.77±9.28	0.527
III, IV	15	1.41±0.77		17	1.35±0.53		16	2.92±4.00
Grade
I, II	113	1.09±0.35	0.021	114	1.44±0.71	0.589	112	3.72±9.69	0.691
III	20	1.22±0.32		22	1.92±1.68		21	3.57±4.19
Lymph node status
Negative	38	1.10±0.24	0.197	39	1.61±1.31	0.543	37	2.77±2.72	0.083
Positive	8	1.56±1.03		8	1.28±0.45		7	2.41±4.99
Myometrial infiltration
≤1/2	79	1.07±0.23	0.353	81	1.53±0.79	0.432	80	4.23±11.32	0.934
>1/2	59	1.14±0.46		60	1.53±1.12		59	2.78±3.23
Cervical invasion
Absent	102	1.05±0.22	0.01	104	1.60±0.97	0.027	104	3.94±9.99	0.306
Present	38	1.24±0.54		39	1.32±0.82		36	2.65±3.76
ESR1 status
Negative	103	1.09±0.38	0.269	103	1.40±0.64	0.085	102	3.73±10.06	0.213
Positive	37	1.13±0.26		37	1.87±1.45		35	4.59±4.20

SD, standard deviation; ESR1, estrogen receptor 1 gene; RAD21, double-strand-break repair protein rad21 homolog; RUNX1, runt-related transcription factor 1. Significant P-values are presented in bold. P-values were calculated using the Mann-Whitney U test.

Results

An increased level of RAD21 gene dosage was identified in 18/141 samples. The average gene dosage was 1.103±0.345. RAD21 overexpression was found in 23/144 samples, with an average of 1.524±0.931. Increased RUNX1 expression was observed in 58/141 samples, with an average of 3.656±8.805. RAD21 gene dosage was significantly associated with RAD21 mRNA expression (ϱ=0.22; p=0.009; Fig. 1A). Furthermore, RAD21 expression markedly correlated with RUNX1 expression (ϱ=0.43; p<0.0000001; Fig. 1B).

Figure 1.

(A) Correlation of RAD21 expression and RAD21 gene dosage. (B) Correlation of RAD21 and RUNX1 gene expression. RAD21, double-strand-break repair protein rad21 homolog; RUNX1, runt-related transcription factor 1.

Increased RAD21 gene dosage correlated with more advanced tumor stage (p=0.021), higher grade (p=0.021), cervical involvement (p= 0.01) and the absence of obesity (p= 0.025), while RAD21 mRNA expression was correlated with cervical involvement (p=0.027; Table II). In the case of RUNX1 mRNA expression only a trend was observed, with a higher expression level in the nonendometrioid histological type (p=0.073) and in the tumors with negative lymph node status (p=0.083; Table II). Menopausal status, level of Ca-125, myometrial infiltration and ESR1 status did not correlate with any of the examined molecular markers.

Discussion

Although endometrial cancer is well characterized at the level of clinicopathological features, its molecular background to date has received far less attention. As RAD21 and RUNX1 actions are crucial for sustaining basic functions in healthy cells and their expression tends to be deregulated in various types of cancer (2–4,8,9,11,12,14,18–25), we analyzed the role of these two markers in the context of endometrial cancer formation using the qPCR method. The examined markers correlated with each other, partially unraveling the network of genes involved in endometrial tumorigenesis. Findings of previous studies have suggested that the decreased expression of cohesins results in an inappropriate increase in homologous recombination which may drive tumorigenesis through the promotion of genomic instability, such as loss of heterozygosity (15,33,34).

In the present study, a marked correlation between mRNA expression of RAD21 and RUNX1 was observed. Furthermore, RAD21 copy number variations were significantly associated with RAD21 mRNA expression and RAD21 and RUNX1 status correlated with the clinicopathological features of the endometrial cancer patients.

Aberrant expression of RAD21 in cancer has been documented by several authors (8,9,11,18,19). RAD21 was found to be especially overexpressed in undifferentiated cancers of the breast, lung, bladder, brain and ovaries (18). Its suppression by siRNA reduced the proliferation of breast cancer cells (8). RAD21 overexpression has also been reported to confer poor prognosis in breast cancer patients (9,19). Notably, RAD21 was downregulated in oral squamous cells with high metastatic potential (11).

The aberrant expression of RUNX1 in cancer has also been documented in the literature (14,25), in particular RUNX1 amplification has been reported to be implicated in the development of leukemia (22,35,36). The upregulation of RUNX1, as measured by qPCR, has been reported in invasive endometrioid carcinoma (14). On the contrary, in breast cancer RUNX1 may act as a tumor suppressor gene. RUNX1 down-regulation is a component of a 17-gene signature predicting metastasis (37). It has also been shown that RUNX1 expression decreases as the breast tumor grade increases (38). Notably, experiments performed on neuroblastoma cell lines have shown that high and low RUNX1 levels disrupt proliferation, inducing cell death (23).

Our analyses did not reveal any link between RAD21/RUNX1 gene expression status and the stage of the tumor, however, RAD21 gene dosage was found to correlate with more advanced tumor stage, grade and cervical involvement. This suggests that RAD21-positive status is correlated with unfavorable clinical characteristics. Therefore, RAD21 amplification may serve as a marker of poor prognosis in endometrial cancer, as in breast cancer (9,19). Furthermore, similarly to breast cancer (9), we have observed a significant association between RAD21 expression and RAD21 gene dosage. This suggests that in case of endometrial cancer CNVs contribute to the deregulation of RAD21 expression. Comparative genomic hybridization revealed RAD21 to be within the region which is prone to high-level chromosomal gains (www.progenetix.net/progenetix).

The qPCR analyses of Abal et al revealed RUNX1 upregulation in endometrial cancer. The authors postulated that RUNX1 plays a crucial role during early stages of endometrial carcinogenesis and is responsible for the switch to myometrial infiltration (39). The findings of Doll et al indicated that RUNX1 overexpression, measured by immunohistochemistry and RT-qPCR, is associated with distant metastasis in an orthotopic endometrial cancer model in nude mice in which the endometrial cancer cell line HEC1A was used (21). Our results do not confirm these two particular hypotheses, showing a correlation only between RUNX1 mRNA expression and tumor histological type. This, however, is in agreement with the findings of Planagumà et al who measured the expression levels of 53 genes, including RUNX1, with cDNA array hybridization. RUNX1, additionally verified with RT-qPCR, was reported to be the most upregulated gene among those studied in endometrioid carcinoma (14). Unfortunately, the authors did not perform such analyses for nonendometrioid carcinoma.

Our results clearly demonstrate that the mRNA expression of RAD21 and RUNX1 is deregulated and co-dependent in endometrial cancer cells. This is in accordance with analyses performed by Horsfield in zebrafish, in which rad21 gene dosage reduction resulted in the decrease of runx1 transcription, suggesting that rad21 is a regulator of runx1 (6). This reveals another gene to be dependent on RAD21 function. Furthermore, the correlation between RUNX1 and ERM/ETV (40) as well as p21WAF1/CIP1 has been reported (41), partially unraveling the network of molecular interactions occurring in endometrial cancer. Nevertheless, the exact molecular background of these changes requires further elucidation in a broader context which would include a larger number of potential molecular markers whose role could be additionally investigated at the protein level, measured by immunohistochemistry and through correlation with patients’ outcome.

Given the possibility of RAD21 status being a prognostic factor, we assume that a correlation between the gene amplification and patients’ outcome is worth investigating. This is to be performed as soon as we gather necessary information concerning the patients’ survival. The role of RAD21 should also be verified in the context of therapy selection as in vitro experiments have demonstrated that cohesin depletion leads to higher sensitivity to DNA-damaging agents and ionizing radiation (5,8,42,43). As RAD21 downregulation increases the sensitivity to certain drugs used in breast cancer therapy, the inhibition of this gene may facilitate the more effective eradication of cancer cells (8,9). This may allow prognostic and predictive analyses of endometrial tumor response to radiation and drugs.