Materials and methods
Tissue collection and patients
Paired surgical specimens from primary bladder tumors and their ANTs were collected from 25 unselected patients who underwent surgery in the University Hospital of Alexandroupolis, Greece, after obtaining informed consent. The Ethics Committees of the University of Thessaly, Department of Pathology, Medical School of Larissa, Larissa, Greece and the Democritus University of Thrace, Departments of Urology and Pathology and University Hospital of Alexandroupolis, Alexandroupolis, Greece approved this study.
The clinical material was frozen at −80°C and further subdivided for standard histological evaluation, DNA and RNA extraction. Tumor content >80% was recorded in all specimens studied. The histological review according to conventional guidelines (WHO/ISUP classification) revealed 24 UCCs and one PUNLMP in our clinical material. The UCCs further revealed 13 low grade tumors (6 in stage pTa and 7 in pT1) and 11 high grade tumors (1 in stage pTa, 9 in pT1 and 1 in pT2; Table I).
The cohort of patients included 20 males and 4 females with UCC and 1 male with a PUNLMP with an age range of 50–90 years (median, 74; Table I).
Quantitative analysis of mRNA expression
We used Purescript® RNA isolation and SuperScript First Stand Synthesis System (Invitrogen®, Life Technologies, Paisley, UK) for cDNA synthesis, by reverse transcription (RT), as described previously (27). qPCR analysis of hMSH2 and hMLH1 mRNA was performed as previously described (16). qPCR analysis of hMSH6 and hPMS2 was performed using specific primers: hMSH6 sense, 5′-AACAAGGGGCTGGGTTAG-3′; hMSH6 antisense, 5′-CGTTGCATTGCTCTCAGTATTTC-3′; hPMS2 sense, 5′-GAGTCAAGCAGATGTTTGCCTC-3′; hPMS2 anti-sense, 5′-TGTGTCTCATGGTTGGCCTT-3′; and fluorescent hybridization probes hMSH6 - F L, 5′-TATACA GGTTCAAAATCAAAGGAAGCCC-FL; hMSH6-LC, 5′-LC640-GAAGGGAGGTCATTTTTACAGTGCAAG-PH; hPMS2-FL, 5′-GGGTGATCAGTTTCTTCATCTCGCTTGT-FL; hPMS2-LC, 5′-LC640-TTAAGAGCAGTCCCA ATCATCACCGACTT-PH designed for Light Cycler instrument 3.5 (TIB® Molbiol, Berlin, Germany). All reactions included Porphobilionogen deaminase (hPBGD) housekeeping gene primers as internal controls (Roche Diagnostics, Mannheim, Germany). Following an initial denaturation step at 95°C for 10 min, hMSH6-PCR assays underwent 45 cycles of denaturation at 95°C for 10 sec, annealing at 54.2°C for 15 sec and extension at 72°C for 6 sec and hPMS2-PCR assays underwent 45 cycles of denaturation at 95°C for 10 sec, annealing at 57°C for 10 sec and extension at 72°C for 6 sec. The mRNA expression of each MMR gene was expressed as a ratio of MMR gene mRNA to control hPBGD mRNAs (MMR/control mRNAs) and defined two major phenotypic groups, the reduced (r) for mRNA ratios <1 and the normal or elevated (R) for ratios ≥1, as previously described (16). Additionally, the MMR gene expression of tumor samples was compared with that of the corresponding ANT samples. This value is indicated as relative mRNA expression of MMR genes between tumor and ANT (tumor/ANT) of each patient (Table I).
Genomic instability analysis
Genomic instability analysis was performed for the following polymorphic regions of the hMSH2 and hMLH1 loci: D3S1234 (3p14), D3S1612 (3p21.3–22) and D3S1768 (3p21.3–22) distal to the hMLH1 locus on chromo-some 3p and D2S1788 (2p22.3) proximal to the hMSH2 locus on chromosome 2p, to compare the possible loss of mRNA expression with allelic imbalance of the chromosomal regions that contain the genes (28). The primer sequences for each microsatellite copy were obtained from the National Center for Biotechnology Information database. Nucleotide repeat markers, stretches within non-coding repeats such as BAT26 in intron 5 of hMSH2 and BAT25 in intron 16 of c-kit were used as established mononucleotide markers for determining MSI status (29,30).
MSI analysis was performed as previously described (31). Briefly, following DNA extraction from bladder tissue specimens (Puregene® Cell and Tissue extraction kit; Gentra), genomic DNA samples were stored at −20°C until use. PCR analysis included amplification of the β-globin gene in order to qualify and normalize the amount of DNA in each sample. The primers used for the amplification of the β-globin gene, PCR, qPCR and melting curve analysis conditions were as previously described (31). All samples were run in duplicate and two non-template-controls (NTCs) were included in the reactions. qPCR amplifications and melting curve analyses were repeated twice. The conditions of reactions were 95°C for 15 min, 36 cycles of 95°C for 15 sec, annealing temperature for each set of primers (55°C for D3S1768, D2S1788, BAT25; 56°C for D3S1612; 56.5°C for BAT26; and 58.5°C for D3S1234) for 30 sec and 72°C for 30 sec (acquiring for SYBR). Continuously Melting Curve analysis performed ramping 65–95°C (raising by 0.2°C each step) and finishing at 72°C for 5–10 min. Following completion of the amplification melting curve, analysis was performed as previously described (31).
Statistical analysis
We used the paired Student’s t-test to compare ratios of hMSH2, hMLH1, hMSH6 and hPMS2 alterations between tumor and matched ANT for different patient characteristics, including age, gender and clinicohistopathological parameters such as tumor type, grade and stage. The correlation between the mRNA expression ratios of hMSH2, hMLH1, hMSH6 and hPMS2 in bladder tumors and their ANTs for different patient and tumor characteristics was examined by Pearson test. The χ2 test was also used to examine the distribution of MMR mRNA phenotypes (r, R, rr, rR, Rr and RR) in tumor and ANT specimens at different tumor histopathological grades or stages. Statistical significance was considered for values of p<0.05.
Results
hMSH2, hMLH1, hMSH6 and hPMS2 mRNA quantification in bladder tumors
We evaluated hMSH2, hMLH1, hMSH6 and hPMS2 mRNA levels in primary UCCs and their corresponding ANTs relative to the reference hPBGD control gene by qPCR (Fig. 1). These data are summarized in Table I with patient age, gender, tumor type, stage and grade.
The urothelium adjacent to UCCs revealed reduced mRNA ratios (<1) of the hMSH2, hMSH6 and hPMS2 genes in 47.8 (11/23), 43.5 (10/23) and 13% (3/23) of samples, respectively, compared with 21.7 (5/23), 29.2 (7/24) and 12.5 (3/24) of UCC tumors (P= 0.027284). We observed a statistically significant difference between the proportions of reduced mRNA ratios of hMSH2/control and hMSH6/control genes observed in ANTs of low grade UCCs relative to their tumors (P=0.025347) that was more pronounced than those observed between ANTs and high grade UCCs (P= 0.000; Table I). Moreover, high grade UCCs exhibited higher proportions of reduced mRNA ratios of hMSH2/control, hMSH6/control and hPMS2/control (40, 40 and 20%, respectively) compared with low grade tumors (7.6, 15.3 and 7.6%, respectively; P=0.000).
The hMLH1/control mRNA ratios were ≥1 in UCCs and their ANTs, contrary to PUNLMP and its corresponding ANT that exhibited reduced mRNA ratios (<1). PUNLMP also showed reduced (<1) hMSH2/control and hMSH6/control mRNA ratios, while its corresponding ANT exhibited elevated or normal (≥1) ratios.
hMSH2, hMLH1, hMSH6 and hPMS2 mRNA relative expression
Calculation of tumor/ANT MMR mRNA ratios from quantification data (Table I) revealed different proportions of UCC ratios ≤0.8 with 26.1% (6/23) for hMSH2, 21.7% (5/23) for hMLH1, 47.8% (11/23) for hMSH6 and 26% (6/23) for hPMS2, respectively. Calculation of PUNLMP/ANT MMR mRNA revealed ratios ≤0.8 of hMSH2, hMSH6 and hPMS2 (Table I). The reduction in hMSH2, hMLH1, hMSH6 and hPMS2 mRNA expression between primary bladder tumors and their matched ANTs, male and female, pTa stage and pT1, high and low grade UCC was not statistically significant by paired Student’s t-test analysis (Table II).
Bladder tumors revealed mRNA overexpression ratios (≥1.8) of hMSH2, hMLH1, hMSH6 and/or hPMS2 compared with their corresponding ANTs in 47.8 (11/23), 35 (8/23), 30 (7/23) and 30% (7/23) of specimens, respectively (Table I).
The statistically significant difference of hMSH2 transcriptional levels between UCC tumors and their matched ANTs (P=0.019165; Student’s t-test) was more pronounced in males compared with females (males, P=0.020265; females, P= 0.169501; Student’s t-test) and was independent of stage (pTa, P=0.127745; pT1, P=0.089642; Student’s t-test; Table II). Low grade UCCs exhibited a statistically significant elevation of hMSH2 mRNA expression relative to their matched ANTs (low grade, P=0.048544; high grade, P=0.130441; Student’s t-test; Table II).
We observed significant difference in hMLH1 transcriptional activation between bladder tumors and their matched ANTs (P= 0.000; Student’s t-test; Table II). Notably, males showed a higher hMLH1 transcriptional activation in tumors compared with females (males, P=0.000366; females, P=0.5; Student’s t-test; Table II). We also observed statistically significantly higher levels of hMLH1 mRNA expression between UCC tumors and their matched ANTs of pT1 stages than pTa (pT1, P= 0.012455; pTa, P=0.081273; Student’s t-test; Table II). Low grade UCCs compared with their matched ANTs exhibited relatively higher mRNA expression levels of hMLH1 than high grade UCCs (low grade, P=0.012; high grade, P=0.032; Student’s t-test; Table II).
The statistically significantly elevated mRNA expression of hPMS2 observed between UCC tumors and their matched ANTs (P= 0.005896; Student’s t-test) was also identified in males compared with females (males, P=0.005159; females, P=0.474152; Student’s t-test), pT1 UCCs (pT1, P= 0.038387; pTa, P=0.143226; Student’s t-test) and was pronounced in high grade compared with low grade tumors (high grade, P=0.028778; low grade, P=0.05819; Student’s t-test; Table II).
Correlation between mRNA expression of the hMSH2, hMLH1, hMSH6 and hPMS2 MMR genes in bladder tumors
We observed a statistically significant correlation between hMSH2/hPBGD and hMSH6/hPBGD mRNA ratios in bladder tumors (r=0.795; Pearson test; Table II).
We also observed a statistically significant association between the relative mRNA expression of hMSH2 and hMLH1 (tumor/ANT) expression (r=0.415684; Pearson test) that was more pronounced in females compared with males (females, r= 0.873155; males, r= 0.37; Pearson test), slightly more intense in pTa than pT1 stages (pTa, r= 0.778713; pT1, r=0.61326; Pearson test) and in low grade than high grade UCCs (low grade, r=0.700724; high grade, r=0.598678; Pearson test; Table II). Only females exhibited a statistically significant association of relative mRNA expression between hMSH2 and hMSH6 (r=0.735679, Pearson test), hMLH1 and hPMS2 (r= 0.560746; Pearson test) and hMSH6 and hPMS2 (r= 0.801165; Pearson test; Table II). In addition, females and pTa UCCs exhibited a significant association between hMLH1 and hMSH6 relative (tumor/ANT) levels of mRNA expression (female, r=0.728012; pTa, r=0.567703; Pearson test; Table II).
Phenotyping MMR sorting
We used the ratio of MMR mRNA expression relative to reference mRNA control to adopt a functional unified assessment for our findings, as previously described (16). We classified our specimens into two major phenotypic groups, one with reduced (r) and the other with regular or enhanced (R) ratios of expression (Materials and methods) and subdivided our study group into eight phenotypes, r2 and R2 for hMSH2, r1 and R1 for hMLH1, r6 and R6 for hMSH6 and finally p2 and P2 for hPMS2 DNA repair system components (Table III) or their combined phenotypes R2R1, R2R6, R2P2, R1R6, R1P2, R6P2 and R2r1, R2r6, R2p2, R1r6, R1p2, R6p2 and r2R1, r2R6, r2P2, r1R6, r1P2, r6P2 and r2r1, r2r6, r2p2, r1r6, r1p2, r6p2 by descending MMR system activity.
Clinical and biological evaluation of single and combined MMR phenotypic distributions
We tested the ability of our phenotypes to distinguish our study group into distinct bladder tumors and ANTs subtypes. We thus examined the significance of the differences between the subgroups created according to our phenotypic criteria (Table III, Fig. 1).
The distribution of individual r and R MMR mRNA phenotypes was significantly different between ANTs and UCCs (P=0.021751; χ2 test), particularly r2 vs. R2 (P= 0.012261; χ2 test). Additionally, a marked difference of r and R phenotypic distribution was observed between high and low grade UCCs (P= 0.00013; χ2 test; Fig. 1A), particularly r2 vs. R2 (P=0.00053, χ2 test) and r6 vs. R6 (P<0.04, χ2 test; Fig. 1B). The frequencies of independent r2, r1, r6, p2, R2, R1, R6 and P2 phenotypes in UCCs and their corresponding ANTs are shown in Table III. The reduced r1 phenotype was not identified in any UCC or corresponding ANTs, in contrast to the PUNLMP and its ANT (Table III).
The distribution of combined MMR mRNA phenotypes in UCCs and their matched ANTs are shown in Table IV. We observed two r2R1 and R2R1 combined hMSH2 and hMLH1 mRNA phenotypes, two R1p2 and R1P2 combined hMLH1 and hPMS2 phenotypes and two R1r6 and R1R6 combined hMLH1 and hMSH6 mRNA phenotypes in UCCs and/or their ANTs with same frequencies between observed and calculated frequencies of combined loci. We also observed four R2R6, r2R6, R2r6 and r2r6 combined hMSH2 and hMSH6, three R2P2, r2P2 and r2p2 combined hMSH2 and hPMS2 and three R6P2, r6P2 and r6p2 combined hMSH6 and hPMS2 mRNA phenotypes in UCCs and/or their ANTs with different frequencies between observed and calculated frequencies of combined loci. There was a statistically significant difference between high and low grade tumors in r2R1 vs. R2R1 (P=0.00019; χ2 test), r2r6 vs. R2R6 (P=0.000786; χ2 test), r2p2/P2 vs. R2P2 (P=0.00053, χ2 test) and less marked in R1r6 vs. R1R6 (P<0.04; χ2 test) and in r6p2/P2 vs. R6P2 (P<0.04, χ2 test). The histogram in Fig. 2 shows the association of reduced (homozygous or heterozygous) r2R1 or r2r6 or r2p2/P2 or R1r6 or r6p2/P2 and normal or elevated (homozygous) R2R1 or R2R6 or R2P2 or R1R6 or R6P2 to high and low grades, respectively (Table IV and Fig. 2).
A statistically significant difference was also observed between R2R6 and r2r6 phenotypic distribution in UCCs relative to their ANTs (P<0.02; χ2 test) that was pronounced in low compared with high grade tumors (P=0.0000006; χ2 test; Table IV and Fig. 3).
Genomic instability
We examined 6 genetic markers for genomic instability [MSI and/or loss of heterozygosity (LOH)] in bladder tumors (Table V) distal or close to hMLH1 and hMSH2 to determine the correlation between possible loss of mRNA expression and allelic imbalance of the chromosomal regions that harbor the genes. D2S1788 (2p22.3) and BAT26 are located distal to and in the hMSH2 locus, respectively, while D3S1612 (3p21.3–22), D3S1768 (3p21.3–22) and D3S1234 (3p14) are distal to the hMLH1 locus. Additionally, BAT25 stretches within the c-kit gene and was included as it has been previously correlated with a DNA repair mechanism (29,30).
We observed genetic instability (MSI and/or LOH) in 3 (15%) of the 20 analyzed bladder tumors vs. their ANTs. Two polymorphic regions at the 3p loci had been affected (distal to hMLH1). D3S1768 locus exhibited MSI in one of the 12 (8.3%) analyzed cases, which was characterized as MSI-L (one unstable marker), while the remaining cases were not informative for LOH (Table V). D3S1612 locus was affected by LOH in two out of 7 (28.57%) informative cases (Table V). None of the MSI-analyzed cases were informative for LOH at the D3S1234, D3S1788, BAT25 or BAT26 polymorphic regions (Table V).
The MSI-L bladder tumor was a high grade UCC with stage pT1, while LOH was noted in a low grade pTa stage and a high grade pT1 stage UCC. The 2 UCCs which showed LOH at the D3S1612 (3p21.3–22) locus exhibited normal or elevated MMR phenotypes but reduced (≤0.8) hMSH6 mRNA tumor/ANT ratios (Tables I and V).
Discussion
To date, a series of studies have attempted to determine the expression of MMR proteins, mainly MSH2 and MLH1, in bladder cancer, the majority using IHC methods (17–24). Only two previous studies have determined MMR mRNA levels in bladder cancer by qPCR analysis and even in a few series of clinical specimens with different percentages from IHC analysis (25,26). The current study presents for the first time a quantification analysis of MMR mRNA transcripts in paired bladder tumors and their ANTs.
It is known that MSH2/MSH6 proteins form heterodimers that act as a complex (MutSα). This complex function is to detect single base-base mismatches and insertion-deletion loops and bind to the side of the DNA error (11,12,14). Our data showed that unaffected urothelia adjacent to UCC tumors (mainly adjacent to low grade UCC tumors) express low ratios of hMSH2 and hMSH6 mRNA levels (r2r6 phenotype), implying a low activity of DNA damage recognition of single mismatches and insertion-deletion loops errors. The ANTs of high grade UCCs also exhibited a reduced r2r6 phenotype, leaving the urothelium at high risk of cancer. Moreover, urothelia adjacent to high grade UCCs showed statistically higher percentages of the reduced r2r6 phenotype, approaching the levels of high grade tumors, in contrast to ANT of low grade UCCs, which showed a significant difference between the corresponding tumors. However, the hMLH1 gene was found to have elevated mRNA ratios (R1 phenotype) both in UCCs and their ANTs, indicating either high requirements for DNA repair of the progressively increasing errors in cancerous or precancerous urothelium or the involvement of hMLH1 in another tumorigenesis pathway (33). The counterpart of hMLH1, hPMS2, was also overexpressed in a percentage of pT1–2 and high grade UCCs, to cooperate with MLH1 as complex (MutLα) due to the demanding repair or another function (10,14,34). Nevertheless, a percentage of UCCs presented reduced levels of hMLH1 and hPMS2 mRNA expression relative to their ANTs which indicates low DNA repair activity in a large proportion of UCCs and therefore accumulation of replication errors in the abnormal proliferating malignant cells.
The unbalanced mRNA levels of MMR genes, including overexpression of hMSH2, hMLH1 and hPMS2 and reduction of mRNA levels of hMSH6, in the urothelium of UCC, mainly in males, was correlated with tumor progression. A recent study implicates MutLα as a general stimulating factor for miRNA biogenesis, giving the complex an additional function in tumorigenesis (34). In our cohort of specimens we observed that a proportion of tumors exhibited mRNA overexpression of hMSH2, hMLH1 and hPMS2. For hMSH2 this was more frequent in low grade pTa tumors; for hMLH1 in low grade pT1 tumors; and for hPMS2 for high grade pT1–2 tumors relative to the ANTs that may indicate the tumor progression. An explanation may be that from low to high grade tumors or from pTa to pT3 histological stages additional DNA errors take place, e.g., small and larger insertion-deletion loops (12,13). The need for recognition of these errors by other MMR complexes and enzymes, such as MutSβ (MSH2-MSH3), is indicated by the significant reduction of the hMSH6 counterpart of hMSH2(25).
We analyzed a case of PUNLMP and its ANT for MMR mRNA expression. The normal urothelium adjacent to PUNLMP revealed regular or elevated (≥1) mRNA levels of MutSα complex which detects single base-base mismatches and insertion-deletion loops while the mRNA levels of hMLH1, a crucial component of MutLα that is responsible for repairing the DNA errors (10,14), were <1. This is in agreement with the results of a previous study which showed that MLH1 is expressed at a lower level than MSH2 and MSH6 in human cells (35), suggesting a regular proliferation of urothelium cells and a limited DNA repair requirement. hMSH2, hMSH6 and hPMS2 mRNAs were reduced in PUNLMP compared with its ANT, probably due to a low rate of apoptosis (36).
The correlation of our results with clinical data revealed the statistically significant association of hMSH2 and hMLH1, hMSH2 and hMSH6, hMSH6 and hPMS2, hMLH1 and hMSH6 tumor/ANT mRNA expression ratios in females. We derive the conclusion that the urothelium of females has a better balance in the expression DNA MMR genes compared with males, who exhibited imbalance. Most likely, the MMR mechanisms are biologically differently regulated in the two genders. Additionally, a significant association was also found between the changes in hMSH2 and hMLH1 mRNA expression levels in UCCs compared with their ANTs, indicating that hMSH2 and hMLH1 cooperation in DNA repair (10,11,14) requires an associated mechanism for regulating hMSH2 and hMLH1 gene expression.
The biological significance of these findings is indicated by the association between hMSH2, hMLH1, hMSH6 and hPMS2 mRNA expression in our tissue cohort. We identified a significant association between reduced mRNA expression levels of hMSH2/control and hMSH6/control, indicating a common mechanism of hMSH2 and hMSH6 suppression of transcriptional activation that is in accordance with their biological function, as components of the MutSα complex act cooperatively (11–13). The interdependence of the four genetic loci was shown by the observed and calculated frequencies of their combined phenotypes (Table IV) (31). hMSH2 and hMSH6 revealed different frequencies and were considered as depended loci, as were hMSH2 and hPMS2 or hMSH6 and hPMS2(37). Besides, hMSH2 and hMLH1, hMHL1 and hMSH6 or hMHL1 and hPMS2 exhibited identical observed and calculated frequencies in UCCs and/or their ANTs and were considered as independent loci (16).
The identification of MSI in bladder tumors vs. their ANTs and correlation with MMR mRNA expression or MMR phenotypes showed that MSI-H was absent, MSI-L was rare in our study group and LOH was found in a small proportion of informative UCC samples. This result is in agreement with those of previous studies which reported the absence or low frequencies of MSI in bladder cancer (32,38). LOH and MSI-L were observed in a region distal to the hMLH1 locus. The two UCCs affected by LOH at 3p loci exhibited regular or elevated MMR phenotypes. Therefore, allelic imbalance at these chromosomal regions which harbor hMLH1 was not correlated with loss of hMLH1 mRNA expression; this is in agreement with a previous study of non-small cell lung tumors (28). However, the UCCs affected by LOH showed reduced hMSH6 mRNA tumor/ANT ratios, which may mean that genetic instability in the bladder, distal to the hMLH1 locus, is correlated with a reduced expression of hMSH6.
In conclusion, this is the first study to quantify MMR mRNA expression in bladder tumors and adjacent normal urothelium. Reduced (r) mRNA phenotypes of hMSH2, hMSH6 and hPMS2 were found to be correlated with precancerous or cancerous urothelium and a previously unrecognized reduced r2r6 (hMSH2 and hMSH6) phenotype with a precancerous urothelium. Additionally, we did not identify a reduced r1 phenotype of hMLH1, a crucial component of MutLα complex, in UCCs or their ANTs and hMLH1 was overexpressed in a significant proportion of UCCs. Therefore, the hMLH1-elevated (R1) mRNA phenotype and mRNA overexpression was correlated with urothelium with malignant potential. The correlation of our results with clinical data revealed that in males the MMR mechanism appears to be unbalanced relative to females and gradually elevated mRNAs expression levels of hMSH2, hMLH1 and hPMS2 in males show a progression from low to high grade and from pTa to pT1–2 tumors. Biologically, we demonstrated that hMSH2, hMSH6 and hPMS2 are interdependent loci; particularly, hMSH2 and hMSH6 were indicated to have a common mechanism of suppressing transcriptional activation. hMLH1 was independent, but requires an association with a hMSH2 mechanism, frequently in low grade tumors, for regulation of mRNA expression. Finally, reduction of hMSH2 and hMLH1 mRNA expression in UCCs is unlikely to be correlated with allelic imbalance at polymorphic regions which harbor the genes; however, LOH distal to hMLH1 may be correlated with hMSH6 reduction.