Fax: +65-6226-3696 or +65-6372-0161
Overexpression of RB1 transcript is significantly correlated with 13q14 allelic imbalance in colorectal carcinomas
Article first published online: 28 MAR 2006
Copyright © 2006 Wiley-Liss, Inc.
International Journal of Cancer
Volume 119, Issue 5, pages 1061–1066, 1 September 2006
How to Cite
Lai, P.-S., Cheah, P. Y., Kadam, P., Chua, C. L.-M., Lie, D. K. H., Li, H.-H., Eu, K.-W., Seow-Choen, F. and Lee, A. S.-G. (2006), Overexpression of RB1 transcript is significantly correlated with 13q14 allelic imbalance in colorectal carcinomas. Int. J. Cancer, 119: 1061–1066. doi: 10.1002/ijc.21945
- Issue published online: 5 JUN 2006
- Article first published online: 28 MAR 2006
- Manuscript Accepted: 2 FEB 2006
- Manuscript Received: 3 AUG 2005
- National Medical Research Council of Singapore
- colorectal cancer;
- allelic imbalance;
RB1 gene expression has been reported to be upregulated in colorectal carcinomas (CRC) at both the mRNA and protein levels when compared to normal colonic mucosa. However, allelic loss at the genomic level has been detected in CRC with widely differing frequencies ranging from 11.5% to 50%. To determine whether there is indeed a correlation between RB1 allelic imbalance (AI) and expression, a consecutive series of 55 CRC from Singapore patients were analysed by microsatellite analysis, real-time RT-PCR and immunohistochemistry. Microsatellite analysis using 3 RB1 intragenic microsatellite markers and 2 markers flanking RB1 detected AI in 32.7% (18/55) of the cases, in at least 1 locus. The highest AI frequency (22.9%) was observed at the microsatellite marker D13S137 (Cu13), which maps 5 cM distal to RB1. AI was present in both early and late Dukes stages. Real-time RT-PCR revealed that all 40 cases analysed expressed RB1 mRNA, with mRNA overexpression in 37.5% (15/40) and pRB protein expression in 88.2% (30/34) of cases. Notably, a statistically significant correlation was found between AI of RB1 and mRNA overexpression of RB1 (p < 0.001, Fishers exact test). These findings provide evidence that despite AI, RB1 expression is not abrogated. Thus, our data suggests that RB1 may play a role in colorectal tumorigenesis through functional regulation of the transcript and protein rather than through its tumour suppressor role by gene inactivation. © 2006 Wiley-Liss, Inc.
The retinoblastoma gene, RB1, is a well-known tumour suppressor gene that encodes a nucleoprotein (pRB) with a critical role in cell cycle regulation, proliferation, differentiation and apoptosis. Numerous studies have shown frequent loss of RB1 function and either low or no pRb expression in different tumour types, such as pituitary adenomas,1 glioblastomas,2, 3 hepatocellular carcinomas4 and sporadic retinoblastomas.5 In contrast, RB1 gene expression has been reported to be upregulated in colorectal carcinomas (CRC) at both the mRNA and protein levels when compared to normal colonic mucosa.6, 7, 8, 9, 10, 11, 12, 13 Consistent with this increased expression, RB1 transcripts and pRb are not truncated or absent in colorectal cancers, suggesting that Rb is functional in CRC.12, 14
At the genomic level, allelic loss detected by loss of heterozygosity (LOH) has also been reported in the RB1 region in CRC with widely differing frequencies ranging from 11.5%15 to as high as 50% of cases studied.16, 17 These differences in frequencies may be because LOH assays actually determine allelic imbalance which may indicate either loss or gain. Increased allelic copy number at polymorphic loci of RB1 has also been documented by densitometric comparisons of Southern blots in 32% of colorectal tumours,18 as well as by fluorescence in situ hybridization analysis in 44% of colorectal adenomas.19 The presence of RB1 allelic loss would suggest that its expression would be decreased, contradictory to previous expression studies in colorectal cancer. There has been no previous study correlating LOH or allelic imbalance (AI) at RB1 with its expression. Furthermore, past mRNA expression studies were semiquantitative and to our knowledge, none were assayed quantitatively using real-time RT-PCR.
The present study aims to clarify the role of RB1 in CRC, by determining whether there is indeed a correlation between RB1 AI and quantitative mRNA expression. A consecutive series of 55 colorectal cancers from Singapore patients were analysed by microsatellite analysis, real-time RT-PCR and immunohistochemistry. Statistical correlations between these analyses as well as with clinicopathological parameters were performed.
Material and methods
Patients and samples
CRC samples and corresponding normal colonic mucosa were snap-frozen in liquid nitrogen upon resection and stored at −70°C or in liquid nitrogen. Peripheral blood samples were collected in EDTA tubes and stored frozen at −80°C. This study was conducted in accordance to guidelines of the institution, and was approved by the Institutional Review Board (IRB) of the Singapore General Hospital. The median age of the patients was 60.11 years at the time of surgery. There were 36 male and 19 female patients. Forty-nine were Chinese, 4 were Malay and 2 of other ethnic groups.
DNA and RNA extraction
DNA was extracted from blood samples using the standard high-salt and ethanol precipitation method.20
Genomic DNA and RNA were isolated from microdissected colorectal tumour tissues as previously described.21 In brief, H & E stained frozen sections were examined to determine the percentage of tumour cells present. All tumour samples in this study were trimmed to enrich for tumour cells and had at least 70% tumour cells present. Samples that did not meet this requirement were not used in this study. Total RNA was isolated from tumour tissues using a column-based method (Qiagen, RNeasy; Qiagen, Hilden, Germany). To avoid contamination with genomic DNA, extracted RNA was then subjected to RQ1 DNAseI digestion (5U; Promega, Madison, WI) in a buffer containing 40 mM Tris-HCl (pH 7.9 at 25°C), 10 mM NaCl, 6 mM MgCl2, 10 mM CaCl2 for 30 min at 37°C. The RNA was then purified using phenol chloroform (5:1) at pH 4.5. RNA concentration was determined spectrophotometrically and RNA quality was determined by gel electrophoresis on 1% agarose gels, stained with ethidium bromide.
Matched normal (blood) and tumour DNA from each patient was analysed using 3 microsatellite markers within the RB1 gene, D13S153, Rbi4 and Rb1.20, and 2 markers 15 cM centromeric (D13S218) or 5 cM telomeric (D13S137) to the RB1 gene. Primer sequences were obtained for each marker from the Genome Database (http://www.gdb.org/) or were as described previously.22, 23 Each PCR reaction was carried out in a total volume of 25 μl, containing reaction buffer (50 mM KCl, 10 mM Tris-HCl pH 9.0, 0.1% Triton X-100), 50 ng of DNA template, 0.5–1.5 mM MgCl2, 0.4 mM dNTPs, 1.25U of Taq polymerase (Promega), 0.4 μM of each primer pair with either the forward or reverse primer labelled with 6-FAM fluorescent dye. Briefly, PCR was performed using the cycling conditions of initial denaturation at 96°C for 5 min followed by 28–35 cycles of denaturation at 96°C for 1 min, annealing at temperatures 53–63°C for 1 min, and extension at 72°C for 1 min; followed by a final extension at 72°C for 15 min. The PCR products were resolved on a 4% denaturing polyacrylamide gel using the ABI PRISM model 377 DNA Sequencer, and the data was analysed using the GENESCAN analysis software, version 2.1 (Applied Biosystems). The whole procedure was repeated twice for each patient. A patient was considered informative at a locus if the normal DNA showed heterozygosity at that locus. AI was calculated using the formula [N2/N1]/[T2/T1], where N2 and N1 are areas of normal allele 2 and 1, respectively, and T2 and T1 are areas of tumour allele 2 and 1, respectively. Peak areas rather than peak heights were used as they allowed for provision of 50% contaminating normal cells in tumour samples, and were thus better indicators for AI determination.24, 25 Samples were deemed to have AI if the ratio obtained was less than 0.5 or more than 1.5, indicating that one of the alleles had decreased by more than 50%. Presence of novel allele peaks consistently exhibited in the tumour sample in contrast to the corresponding normal sample was scored as microsatellite instability. Such a sample was considered as noninformative. Duplicate microsatellite analysis was done in all the samples using a second round of analysis involving independent separate PCR amplification for each marker.
Bisulphite treatment of 1 μg colorectal tumour DNA was done using the CpGenome DNA modification Kit (Intergen, Purchase, NY) according to the manufacturer's instructions. Treated tumour DNA was purified using the QIAquick PCR purification kit (Qiagen, Germany) and eluted in a volume of 40 μl. PCR was performed in a 50-μl reaction volume containing 4 μl of bisulfite-treated DNA, 1.5 mM MgCl2, 200 μM of each dNTP, 2.5 unit of HotStarTaq (Qiagen, Germany), and 2 μM of each primer. Primer sequences of RB1 for the methylated and unmethylated reactions were as previously described.1 After initial denaturation at 95°C for 15 min, 35 cycles each of denaturation at 94°C for 45 sec, annealing at 64°C (methylated reaction) or 60°C (unmethylated reaction) for 45 sec and extension at 72°C for 45 sec were performed, followed by a final extension at 72°C for 10 min. Methylation-specific PCR products were then visualised on 3% agarose gels, stained with ethidium bromide. Universally Methylated DNA (Intergen, Purchase, NY) and normal lymphocyte DNA were used as controls for methylation and unmethylation respectively.
Real time quantitative reverse-transcription PCR
Single stranded cDNA was generated from 100 ng of total RNA using 4.5 units of AMV-reverse transcriptase and 500 ng random hexamer primers (Promega, Madison, WI) in a 20-μl reaction incubated for 10 min at 25°C, 45 min at 42°C and 5 min at 65°C. Quantitative assessment of RB1 gene expression was detected using the Assay-on-Demand™ Gene Expression product (Applied Biosystems, Foster City, CA). Real-time PCR was performed in a 25-μl reaction volume containing 1 μl of cDNA, 900 nM of each primer, 250 nM of TaqMan® MGB probe (FAM™ dye labelled) (Applied Biosystems), 200 μM of each dNTP, 2.5 mM MgCl2, 1× Gold buffer and 5 units of AmpliTaq Gold DNA polymerase (Applied Biosystems). For thermal cycling, the following conditions were applied. Initial incubation of 2 min at 50°C and 10 min at 95°C, followed by 40 cycles of 15 sec at 95°C and 1 min at 60°C. The reaction was performed in triplicate and was monitored in an ABI PRISM 7700 Sequence Detection System (Applied Biosystems). The GAPDH housekeeping gene was chosen for endogenous control and the expression of RB1 gene was calculated relative to GAPDH. All experiments included mRNA from normal colonic mucosa (C24N). The level of RB1 mRNA in the tumour samples were expressed relative to C24N, designated as the calibrator. In addition, the RB1 mRNA level of a commercial normal colon RNA control was evaluated (Ambion).
Frozen sections (5 μM) were cut and immediately fixed in 100% acetone at room temperature for 10 min. Immunostaining was performed using the DAKO EnVision+ System, HRP (DAKO, CA), with signals visualised using 3,3′-diaminobenzidine substrate (DAKO) and counterstained with hematoxylin. Briefly, anti-pRb antibody (G3-245, BD PharMingen, CA) was applied to sections at a dilution of 1:50. For negative controls, primary antibodies were omitted. Positive controls included immunostaining with the cytokeratin 5/8 antibody which reacts with cells of epithelial origin (sc-8021, Santa Cruz Biotechnology, CA) applied at a dilution of 1:800, and secondly, using tissue sections known to be positive for pRb expression, for every experiment. Slides were scored by 2 independent investigators as previously described.26, 27, 28 When a nucleus displayed brown staining, it was considered positive. The intensity of staining was graded as 0 (negative), 1+ (weak), 2+ (moderate) and 3+ (strong) and the percentage of tumour cells stained was graded as 0 (0–24%), 1 (25–49%), 2 (50–74%) or 3 (>75%). The cumulative scores were ascertained with a score of 0 being negative, scores of 1–3 classified as low (L) and those from 4 to 6, as high (H). Frozen sections from each specimen were also stained with hematoxylin and eosin.
The Fisher's exact test was used to determine any associations between AI and clinicopathological parameters or experimental data. P-values below 0.01 were considered as significant.
Fifty-five paired DNA samples from blood lymphocyte and tumour of colorectal patients were assessed for AI (Table I, Fig. 1). All 5 microsatellite markers were informative for at least 1 marker in our samples with heterozygosity rates ranging from 0.545 to 0.782 (Table I). High incidence of AI was observed in the region between D13S153 (Rbi2) and D13S137 (Cu13) with the highest AI frequency of 23% observed in the latter (Table I).
|Microsatellite marker||AI frequency (cases with AI/informative cases)||Heterozygosity rate of marker (heterozygous samples/total samples)|
|D13S218 (210zb)||9% (4/43)||0.782 (43/55)|
|D13S153 (Rbi2)||18% (7/40)||0.727 (40/55)|
|Rbi4||17% (5/30)||0.545 (30/55)|
|Rb1.20||17% (6/35)||0.636 (35/55)|
|D13S137 (Cu13)||23% (8/35)||0.636 (35/55)|
Eighteen tumours (33%) exhibited AI in at least 1 locus (Fig. 1). Four tumours (C11, C48, C54 and C64) showed AI at every informative locus analysed, and 1 tumour (C54) particularly showed AI at both extragenic flanking markers proximal and distal to RB1 gene suggesting a chromosomal deletion at 13q14 locus spanning ∼20 cM. Two tumours (C10 and C14) clearly had allele loss spanning 15 cM while 3 tumours (C47, C48 and C66) had allele loss spanning 5 cM with all the AI in these 5 tumours overlapping loss at the RB1 gene locus.
D13S137 (Cu13) maps 5 cM distal to RB1. The higher AI frequency at this locus raises the possibility of the presence of another putative tumour suppressor locus which may contribute towards tumorigenesis in colorectal cancers. The occurrence of AI was observed both in early and late Dukes' stages (Fig. 1).
Expression of RB1 gene in colorectal cancer
Forty colorectal cancer samples were available for real-time RT-PCR analysis (Fig. 2). Six (15%) had a lowered mRNA expression of less than 50%, compared to normal colonic tissue, and were considered to be underexpressed. Fifteen of 40 (38%) colorectal tumour samples had increased mRNA expression of greater than 2-fold when compared to normal colon tissue, and were considered to be overexpressed. Two samples, C8 and C9, were considered not to have expression 2-fold higher than normal because the lower limit of the SD error bar was below 2-fold (Fig. 2).
The expression of pRb was also examined by immunohistochemistry in 34 available CRC. pRb staining was specific only to cells that were positively staining for the epithelial-specific cytokeratin CK 5/8 antibody (Fig. 3). Normal stromal tissue adjacent to the tumour was negative for both CK5/8 and pRb staining. Of the 34 samples, 30 expressed pRB protein, and 4 did not.
Correlations between RB1 expression data, microsatellite analysis and clinicopathological features
There was no statistical correlation between RB1 mRNA expression and patients' age at diagnosis (p = 0.324), tumour staging (p = 0.315) or protein expression (p = 0.847, Table II).
|A + B||8||7||1|
|C + D||7||11||5||0.315|
More than 70% (11/15) of the specimens that overexpresssed RB1 RNA also exhibited AI. There was thus a statistically significant correlation between AI at RB1 and mRNA overexpression of RB1 (p < 0.001, Fisher's exact test; Table III). No correlation was found between the presence of AI and the patients' age at diagnosis (p = 1.000), or tumour (Dukes) staging (p = 0.239, Table III). However, there was a trend of inverse correlation of AI with protein expression (p = 0.016, Table III).
|A + B||8||10|
|C + D||10||26||0.239|
In this study, we examined the correlation between AI at the RB1 gene and the expression of pRB in a series of 55 CRC. AI was detected in 33% (n = 55) of our samples, suggesting that RB1 may play a role in CRC. Two other studies by Cawkwell et al.16 and Tomlinson et al.17 reported LOH frequencies of 50% (n = 14) and 16% (n = 49), respectively. Both studies utilised a single microsatellite marker within the RB1 gene. We believe that the data from our current study may better reflect the true incidence of AI in CRC for the following reasons. (i) Five markers at the 13q14 locus were utilised to analyse a larger series of cases than the 2 previous studies; (ii) strict cut-off thresholds were observed in defining AI as described in our methodology24, 25; (iii) high quality DNA were available from snap-frozen tumour tissues; (iv) tissue heterogeneity was minimized by enrichment of tumour cells, with at least 70% tumour cells present.29
RB1 mRNA expression as determined by quantitative real-time RT-PCR was found in all of the tumours, suggesting that the RB1 gene is transcriptionally active (Fig. 2). Overexpression of pRB at the mRNA level was observed in 38% (15/40) of our tumours. Notably, 73% (11/15) of these overexpressed tumours had AI (Table III).
It is difficult to confidently distinguish between true allelic loss and allelic gain in microsatellite studies. If allelic loss had occurred, the single remaining allele of RB1 would be expected to lead to decreased mRNA expression instead of the observed overexpression. If AI is indicative of allelic loss, the observed overexpression of mRNA trancripts could be due to upregulation of the remaining allele either by dosage compensation or homeostatic mechanisms.13 Furthermore, protein expression was also observed in all samples with AI. The presence of RB1 mRNA and protein expression suggests that the remaining allele in these samples has no inactivating mutations and is fully functional.
Conversely, if AI is not indicative of allelic loss, overexpression could be due to whole chromosomal gains such as aneuploidy.30 Although aneuploidy is a common occurrence in CRC, we believe that true allelic loss had occurred. All cases were informative at a minimum of 1 locus, with the extent of AI defined across the 5 markers used in this study in 51 of the 55 cases. In 4 cases (C11, C14, C48 and C64), the markers most proximal or distal were not informative, and hence, the extent of AI could not be established (Fig. 1). With the exception of 1 case (C54), the region of AI in the majority (50/55; 91%) of the cases did not exceed the entire 20 cM at the 13q14 locus in this study. If aneuploidy accounted for the AI observed, we would expect all 5 markers within the same sample to have shown AI. None of the 50 samples, for which the extent of AI was clearly defined, exhibited AI for all 5 markers. These results suggest that aneuploidy of chromosome 13 may be uncommon in CRC.
Allelic gain by aneuploidy has been excluded in our tumours based on our results on the extent of AI (see above). However, allelic gains due to local DNA amplification cannot be ruled out.31 Nevertheless, mutation analysis in RB1 indicate such exonic/sequence duplication is rare (http://www.d-lohmann.de/Rb/mutations.html).32 In any case, partial duplication of RB1 is unlikely to lead to overexpression as compared to whole gene amplification, which has been ruled out as AI did not span the entire RB1 gene in the majority of the tumours (except for C11, C14, C48 and C64).
Both normal and increased expression of RB1 has been demonstrated at the mRNA12 and protein levels in CRC.8, 9, 13 The majority (88.3%) of our specimens also expressed pRB protein as determined by immunohistochemistry. In addition, gene silencing by promoter inactivation was not observed in any of our samples using MSP (data not shown). Hence, our study demonstrates that the RB1 gene is not silenced at either the mRNA or protein levels, thus corroborating with previous studies.
Our current study demonstrates that the expression of RB1 is present in all CRC specimens suggesting that RB1 is functional in CRC. In contrast, LOH in RB1 has been documented in CRC previously16, 17 implying that RB1 should be inactivated or underexpressed. To our knowledge, expression of RB1 mRNA has never been positively associated with LOH or AI in CRC previously. Our study shows that even in samples with AI, RB1 is expressed. In fact none of the samples with AI showed underexpression of RB1 (Table III). Our data thus provides evidence to show that despite AI, RB1 expression is not abrogated.
A statistically significant correlation was observed between overexpression of RB1 and AI in this study (Table III). This overexpression may be due to upregulation by the tumour cells to maintain a homeostatic balance between feedback control mechanisms that coordinate levels of proteins regulating cell proliferation and apoptosis. Thus, taking into consideration that RB1 transcripts are present in all tumours, our data suggest that RB1 plays a role in colorectal tumorigenesis through functional regulation of the transcript and protein rather than through its tumour suppressor role by gene inactivation.
We thank Danny CT Ong for excellent technical assistance.