Abstract
Expression profiling studies using microarrays and other methods have shown that microRNAs (miRNAs) are dysregulated in a wide variety of human cancers. The up-regulation of miR-221 has been reported in carcinomas of the pancreas, breast, and papillary thyroid, as well as in glioblastoma and chronic lymphocytic leukaemia. In prostate cancer, however, down-regulation of miR-221 has been repeatedly confirmed in miRNA expression studies. Also unique to prostate cancer, and found in more than 50% of patients, is the aberrant expression of a known oncogene, the TMPRSS2:ERG fusion. To date, there has been no published study describing miRNA associations in prostate tumours that overexpress the ERG oncogene from the TMPRSS2:ERG fusion transcript. Herein we report that in a large and diverse cohort of prostate carcinoma samples, miR-221 is down-regulated in patients with tumours bearing TMPRSS2:ERG fusion transcripts, thus providing a link between miRNA and gene fusion expression.
Prostate cancer exists along a biological continuum that ranges from clinically insignificant to extremely aggressive disease. Although clinically localized prostate cancer is largely manageable by surgery, with patients rarely developing clinical recurrence, recurrent disease remains essentially incurable. Because of this difference in treatment options, it is critical to establish specific biomarkers to differentiate between different stages of disease not only to determine presence of disease, but also to differentiate between indolent and aggressive cancer. The most common class of prostate cancer biomarker thus far are gene fusions resulting from chromosomal rearrangements. Numerous recurrent chromosomal rearrangements have been identified that are generally characterized by the fusion of various 5’ regulatory elements to E twenty-six (ETS) transcription factors, leading to high expression of these oncogenic transcription factors. Transmembrane protease serine 2: ETS-related gene (TMPRSS2:ERG), present in over 50% of all prostate cancers, is the most commonly identified fusion gene (1).
TMPRSS2 is an androgen-responsive, prostate-specific serine protease of unknown function, and ERG is a member of the ETS transcription factor family and is rarely detected in normal prostate tissue (2). The consequences of ERG overexpression, and its correlation to the progression of prostate cancer remains unclear. What is known is that androgen stimulation induces the overexpression of an mRNA containing the ERG ORF and 3’ UTR when ERG is fused to the TMPRSS2 5’ UTR. In vitro studies have shown that overexpression of ERG stimulates cell migration and invasion, while its knockdown decreases the invasive properties of VCaP cells (3). In addition, it has been shown by our laboratory and others that TMPRSS2:ERG fusion mRNAs are present in prostate tumours but seldom in normal prostate cells. Thus, detection of abnormally high ERG expression could, at least theoretically, be a potential diagnostic and/or prognostic marker for prostate cancer. Although the chromosomal alterations seem to be important in the development of prostate neoplastic development, they alone may not be sufficient to induce cancer formation (2).
An active area of prostate cancer research is to find biomarkers that are predictive of recurrence in patients in order to aid oncologists in treatment or non-treatment decisions. TMPRSS2:ERG has been repeatedly, but not unanimously, associated with a poorer prognosis in prostate cancer patients and our laboratory, in collaboration with other groups, has previously demonstrated that prostate cancer patients with the TMPRSS2:ERG gene fusion have a higher risk of recurrence (4), while others have reported no association between this chromosomal rearrangement and clinical outcome (5). The reasons for this lack of congruity between findings are unclear and the value of TMPRSS2:ERG as an independent prognostic biomarker of prostate cancer remains contentious; however, it is possible technical discrepancies may be an important factor, such as institutional inconsistencies in disease staging, or statistical variations, such as disparate cohort sizes (1). In fact, studies that show an absence of clinical correlation between fusion and prognosis highlight the importance of finding out how the fusion may be epigenetically regulated.
MicroRNAs (miRNAs) have received considerable attention in recent years as possible biomarkers not only in prostate cancer, but also in various cancer subtypes. They are small, single-stranded, noncoding RNA molecules that regulate mRNA function by binding to the 3’ UTR of mRNAs to which they are partially complementary, thereby repressing translation (6). MiRNAs are known to be involved in almost every cellular function, including early development (7), differentiation (8), apoptosis (9), and cell cycle regulation (10); as such, it is not surprising that miRNAs have also been linked to cancer, since misregulation of any of these important cellular functions can lead to cancer (11). Aberrant expression of miRNAs has been found in prostate cell lines, xenografts, and clinical tissues. Since the majority of cancer deaths are caused by complications from metastasis, miRNAs that specifically regulate cancer metastasis (metastamirs) are of particular interest. The study of miRNAs as biomarkers and their exact involvement in the formation and/or progression of prostate cancer is still at its early stages, and more research is needed to evaluate the potential use of miRNAs as diagnostic and prognostic markers of prostate cancer.
Given that miRNAs and fusion genes have been independently linked to prostate disease and progression, we set out to study the possible connection between miRNA regulation, prostate cancer recurrence, and TMPRSS2:ERG gene fusion status. We used a large cohort of men with clinically localized prostate cancer who were treated with radical prostatectomy and had long-term follow-ups. We assessed whether miR-221, a metastasis-promoting miRNA (12) located on the X chromosome that is differentially expressed in recurrent prostate cancer, is also associated with TMPRSS2:ERG fusion gene. Thereby, our study may prove applicable to future use of these regulators as surrogate biomarkers of prostate cancer.
Materials and Methods
Study participants and prostate sample collection. Tumour samples were obtained after radical prostatectomy from prostate cancer patients who had surgery at Sunnybrook Health Sciences Centre (Toronto, Canada) between 1998 and 2006. As described by Nam et al., following radical prostatectomy, a midsection of the specimen was snap-frozen in liquid nitrogen, and stored at −80°C until extraction of RNA (4). Most tumours were not visible within the prostatectomy specimen, and thus, the samples obtained from the prostate were considered to be random. The banked slices of specimens were photocopied, oriented (anterior, posterior, right and left), quadrisected and cut into 5 mm sections on a cryostat. The sections were stained with haematoxylin and eosin (H&E) and then reviewed by the pathologist. The areas of tumour were marked on the stained slides and on the photocopied diagram. The marked areas were used to extract the tissue for total RNA extraction. All research was conducted with the approval of Sunnybrook Health Sciences Centre Research Ethics Board.
Patient follow-up. In 1998, a prostate tumour tissue bank was established at Sunnybrook hospital. Clinical data and follow-up information were collected prospectively. The medical records were thoroughly reviewed using standardised data entry forms by trained data abstractors and stored within a prostate cancer-specific database. Clinical follow up consisted of four assessments in the year following surgery, two assessments in the second year and one assessment every year thereafter. At each follow-up, patients had a prostate-specific antigen (PSA) test, and clinical evaluation. Biochemical recurrence was defined as a rise in blood levels of PSA in prostate cancer patients on two consecutive measurements after radical prostatectomy. Data on the following characteristics were available for each patient: Age, family history of prostate cancer, PSA score, Gleason grade, tumour stage, seminal vesicle invasion, surgical margins (categorized as positive or negative), metastasis (categorized as absence or presence), and TMPRSS2:ERG translocation (categorized as presence or absence). The clinical demographics of the patients used in the study are summarized in Table I.
Quantitative real-time polymerase chain reaction (PCR). Total RNA from prostate tumours was extracted with TRIzol reagent (Invitrogen Corporation, Carlsbad, CA, USA) according to the manufacturers' instructions. One microgram of total RNA was reverse-transcribed using the QuantiTect Rev. Transcription Kit (Qiagen GmbH, Hilden, Germany). Quantitative real-time PCR was performed in triplicate by using QuantiTect SYBR Green PCR Kit (Qiagen GmbH, Hilden, Germany) on LightCycler Real-time PCR system (Roche Applied Science, Mannheim, Germany). The miRNA level was normalised by house keeping gene RNU6B. Optimized miRNA-specific primers for miR-221, as well as for the endogenous control RNU6B, are also commercially available (miScript Primer Assays; Qiagen). The relative amount of miR-221 in each sample was calculated based on the crossing-point analysis (Relquant, version 1.01).
RT–PCR and direct DNA sequencing. Total RNA was extracted from the frozen prostate cancer tissue by homogenization in Trizol (Invitrogen Corporation) followed by ethanol precipitation. RNA pellets were dissolved in RNase-free H2O and quality determined using 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, USA). The presence of TMPRSS2:ERG was assayed using RT-PCR as previously described (4). All reactions were performed with two primer sets that yield a 125 bp (F-TAGGCGCGAGCTAAGCAGGAG, R-GTAGGCACACTCAAACAACGACTGG) and 595 bp (F-CAGGAGGCGGAGGCGGA, R-GGCGGTTGTAGCTGGGGGTG AG) product.
Statistical analysis. Results were statistically analysed using Prism v4.0 software (GraphPad Software Inc., La Jolla, CA, USA). Scatter plots were analysed using Student's t-tests which were two-tailed and unpaired. P-values less than 0.01 were considered statistically significant.
Results
Patient demographics. RNA was extracted from and TMPRSS2:ERG status analyzed in 170 radical prostatectomy samples. The distribution of clinical characteristics such as PSA, Gleason grade, pathological stage, and surgical margin status are described in Table I. Of these 170 patients, we opted to eliminate 17 from our study because certain clinical information was missing. Table II summarizes the fusion gene status and complete clinical information of the 153 remaining patients.
The age range at diagnosis of the 153 patients was 31-75 years. The average follow-up lasted 5.4 years in the fusion-negative group, and 4.69 years in the fusion-positive group. Among the cohort, 54.2% of the patients had tumours confined to the prostate gland and 69.3% of tumours were of Gleason score 7.
miR-221 is down-regulated in TMPRSS2:ERG fusion gene-positive prostate tumours. Due to the high prevalence of TMPRSS2:ERG in prostate cancer and its association with higher chances of recurrence and poorer prognosis, we hypothesized that miR-221, an miRNA previously linked to metastasis and recurrence in prostate cancer, may be associated with the fusion gene. To test our hypothesis, TMPRSS2:ERG status was determined for 153 radical prostatectomy samples by using RT–PCR and sequencing using random and oligo-dT primers. Prostate samples from 83 out of 153 (54.2%) patients were found to be positive for transcripts of TMPRSS2:ERG, while 70 (45.8%) lacked the fusion gene (Table II). TMPRSS2:ERG fusion-positive samples produced the expected 125 or 595 bp bands depending upon which primer set was used, as previously described (4). To analyse whether miR-221 expression was associated with the presence of the fusion gene, quantitative RT-PCR was performed, which showed that the mean expression level of miR-221 was significantly lower (p<0.01) in fusion-positive (4.52±0.34) compared with fusion-negative (7.70±1.09) tumours (Figure 1). Gleason grade and surgical margin status were not different between fusion-positive and fusion-negative populations (Table II).
Low miR-221 expression is associated with metastasis and biochemical recurrence of prostate tumour. To further analyse the association of miR-221 levels with the aggressiveness of prostate cancer, we categorized the tumour samples into two subgroups: those from patients who had metastasis at the time of radical prostatectomy and/or had biochemical recurrence in follow-up years and those from patients with non-metastatic and non-recurrent disease. Long-term follow-up information on all 153 patients was not available, as some had moved to other hospitals or did not follow-up with their appointments. Of the 153 tumours, 99 were from patients with sufficient follow-up, more than 5 years on average (Table III). These 99 patients were divided into two groups: those with recurrent and/or metastatic tumours, and those with non-recurrent and non metastatic tumours.
Quantitative RT-PCR was carried out in order to analyse the expression levels of miR-221 in prostate tumours. The qRT-PCR analysis using the 55 recurrent and/or metastatic samples and 44 non-recurrent and non-metastatic samples confirmed that the mean expression level of miR-221 was down-regulated (p<0.01) in the tumours with metastasis and/or recurrence (3.89±0.39) compared to tumours with no metastasis or recurrence (6.32±0.73) (Figure 2).
miR-221 levels are lower in recurrent/metastatic TMPRSS2:ERG fusion-negative tumours. We wanted to further analyse the differential expression of miR-221 in relation to both genetic aberrations and clinical parameters: specifically, TMPRSS2:ERG fusion status, metastasis and recurrence. All 99 tumour samples from patients with long-term follow-ups were studied. We found that tumours positive for the oncogenic TMPRSS2:ERG had down-regulated miR-221 levels with or without metastasis and biochemical recurrence (4.20 and 4.22, respectively) (Figure 3). In TMPRSS2:ERG fusion-negative tumours, on the other hand, miR-221 expression differed with clinical status: miR-221 levels were significantly lower in patients with recurrence and/or metastasis (3.52) than in those with no recurrence or metastasis (8.62) (Figure 3).
Discussion
The potential utility of the TMPRSS2:ERG fusion product as an independent prognostic marker for patients with clinically localised prostate cancer is becoming clearer every year. Initial studies comparing clinico-pathological parameters (4, 13-16) and prognostic significance (4, 13, 14, 17-19) of this fusion gene showed conflicting results. Some studies showed no correlation between histological grade (Gleason score) (13), while others found positive associations (16, 17), and yet others demonstrated correlations between fusion status and tumour stage (14, 15). Wang et al. examined 119 patients for fusion status from a case-control approach and found significant correlations with tumour stage, but no associations were found with early recurrence (20). Furthermore, Lapointe et al. in another case-control study found no correlations with any clinico-pathological parameter and recurrence-free survival (13). However, two cohort studies of men with clinically localised prostate cancer who did not undergo treatment (i.e. watchful waiting) showed that men who had TMPRSS2:ERG fusion had lower prostate cancer-specific survival compared to men without fusion expression (17, 18). Patients managed and selected for watchful waiting from these cohorts have different baseline distributions in grade, stage and PSA level to patients treated with surgery and may not be comparable. Initially, it was unclear whether the TMPRSS2:ERG gene fusion is only a surrogate marker for established prognostic factors of grade and stage, or whether it is an independent molecular-based marker for disease recurrence with no association with grade or stage, particularly for patients who are candidates for surgery for clinically localised prostate cancer.
Since that time, other groups, including our own, have found robust associations. More than two dozen high-impact reports have found clinically significant links to fusions in greater than 50% of over 1500 samples of clinically localized prostate cancer (1). Furthermore, a recent publication by Carver et al. showed that prostatic intraepithelial neoplasia is induced in host prostates by transgenic overexpression of ERG (21), suggesting that high frequency of ETS genetic rearrangements and subsequent overexpression of ETS factors may represent a crucial event in prostate tumourigenesis (22).
In addition to TMPRSS2:ERG, there are increasing reports of miRNAs or specific miRNA signatures that correlate with a wide range of clinico-pathological features in prostate cancer. Many miRNAs are found to be predictive of patient clinical outcome and/or response to treatment, suggesting that miRNAs can be used as diagnostic or prognostic/predictive biomarkers (23). Investigations of miRNA dysregulation in prostate cancer typically compare miRNA expression profiles in prostate cancer versus normal/benign tissues. Two studies have included prostate cancer among other human cancer types, finding overall down-regulation of miRNAs in solid tumours compared with normal tissues (24, 25). A study comparing prostate tumours with benign prostate tissue found 37 down- and 14 up-regulated miRNAs (26). Others have found either overall down-regulation of miRNAs (27) or general up-regulation of miRNAs (28). Most significantly for prostate cancer, Spahn et al. assessed miRNA expression profiles in lymph node metastasis of prostate carcinoma and found that miR-221 down-regulation was a hallmark of metastasis (29). In a larger patient group, they found that the expression of miR-221 was associated with prostate cancer progression and clinical recurrence.
miR-221 is known to be overexpressed in tumours of the breast (30), pancreas (31), glioblastoma (32), papillary thyroid carcinoma (33), and chronic lymphocytic leukaemia (34). Up-regulation of miR-221 in chronic lymphocytic leukaemia was found to be associated with a poor prognosis. Functional studies have found that miR-221 induces down-regulation of p27 and that inhibition of miR-221 impairs tumour formation in xenografted mice (35). Indeed, high levels of miR-221 are required in many different cancer types to inhibit the expression of p27, and stimulate proliferation.
However, none of these studies have looked for miRNA associations in prostate tumours that overexpress the ERG oncogene from the TMPRSS2:ERG fusion transcript. To that end, we investigated miR-221 expression in tumours bearing the TMPRSS2:ERG fusion transcript. Using a large and diverse cohort of prostate carcinoma samples, including patients with or without the fusion gene, as well as patients with different clinical progressions, we found that miR-221 is down-regulated in both patients with TMPRSS2:ERG fusion gene and in patients with more aggressive tumours.
As a major centre for prostate cancer, we assembled a large database and tissue bank from patients undergoing a prostate biopsy to determine the presence of prostate cancer and surgery for clinically local prostate cancer. This resource has been used for study of multiple genetic and serological markers for prostate cancer diagnosis and prognosis (4, 14, 36-40). The resource includes men who underwent a prostate biopsy because of an abnormal PSA or digital rectal examination and preserves outcome data, along with DNA, plasma, and paraffin-embedded tumour samples. TMPRSS2:ERG status was determined as presence or absence by RT-PCR of RNA for samples used in this study and Table I shows the range of clinical and molecular characteristics.
Almost half of the samples (54.2%) bear TMPRSS2:ERG fusion transcripts (Table II). We found that miR-221 expression in fusion-positive tumours is 1.7-fold lower than in fusion-negative samples, with a tighter distribution (Figure 1, Table II). Overexpression of the ERG transcription factor is a powerful inducer of prostate tumourigenesis, as we and others have reported. Similarly, one would also expect to see down-regulation of tumour suppressors, such as miR-221 in at least some prostate tumours. It is therefore possible that the ERG transcription factor or one of its target genes could directly down-regulate miR-221.
A subset of 99 cases from the full cohort had enough clinical follow-up with which to examine the long-term consequences of miR-221 dysregulation in prostate cancer patients (Table III). Average miR-221 expression was 1.6-fold lower in samples from tumours that recurred or metastasized after surgery (Figure 2). Previously we found that the TMPRSS2:ERG fusion is also strongly associated with recurrence. That report and our finding here that miR-221 is reduced in fusion-positive tumours suggests that they have a cumulative effect on tumour aggressiveness. It may be that TMPRSS2:ERG fusion has the stronger effect as we found that tumours positive for the oncogenic TMPRSS2:ERG have lower levels of miR-221 regardless of recurrence status (Figure 3).
In numerous other cancer types, miR-221 is up-regulated and acts to induce proliferation and tumour formation by down-regulating the p27 tumour suppressor. The role of miR-221 in prostate cancer appears to be different than in other cancer types, as reported here and by others (29). Although the Spahn group did not examine the mechanisms, they found that miR-221 is down-regulated in metastatic tumours (29). We also found down-regulation of miR-221 in prostate tumours and that such down-regulation can occur in the absence of TMPRSS2:ERG. Although results shown in Figure 1 suggested a link between fusion status and miR-221 status, direct dependence of miR-221 expression on TMPRSS2:ERG in every tumour seems unlikely given the independence of miR-221 down-regulation in fusion-negative tumours.
This is the first study to reveal that miR-221 down-regulation in prostate cancer is associated with the presence of the oncogenic TMPRSS2:ERG fusion transcript. Understanding how this association leads to greater metastasis and biochemical recurrence will facilitate understanding of prostate cancer biology. More practically, a deeper molecular understanding of the mechanisms of prostate cancer genesis and maintenance will eventually provide therapeutic interventions designed to affect the unique characteristics of individual prostate tumours.
Acknowledgements
This work was funded by the Canadian Cancer Society Research Institute, grant no. 019038.
- Received November 23, 2010.
- Revision received January 13, 2011.
- Accepted January 14, 2011.
- Copyright© 2011 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved