Genetic and epigenetic alterations underlie the pathogenesis of cancer. In particular, the disruption of epigenetic mechanisms leads to abnormal development of cells and is involved in malignant transformation . Variations in DNA methylation are important epigenetic modifications which may affect gene expression by modifying the DNA structure without altering the native nucleotide sequence.
In the present study, we have analyzed the mRNA expression of MMR genes and their promoter methylation in CRC tissues in patients from the Czech Republic, a country that has the second highest CRC incidence and mortality among 38 European countries . With the exception of MLH1 and MSH2 genes, the methylation status of the other MMR genes was globally less studied or even never previously analyzed.
In our study, we observed the promoter methylation of MLH1 gene only. Notably, this modification was observed in MSI-H colon tumors only, as it is generally observed in CRC . The incidence of MLH1 promoter methylation observed in our study is in concordance with another Czech study of Vasovcak et al. , where mutational profiles of CRC high risk genes together with methylation of MLH1 gene were analyzed. Similarly to the work of Vasovcak et al. , any MLH1 promoter methylation was not detected in rectal tumors. Rare MLH1 promoter methylation in rectal cancers was described also in the study of Samowitz et al. , but it was accompanied by high degree of MMR protein deficiency, possibly due to the inclusion of Lynch associated tumors. The reason for the differential MLH1 promoter methylation and tumor localization is still unknown and can be caused by dietary habits, different environment (e.g. varying pH) in different parts of the colon, or by the combination of both aspects  or by other factors which could affect the presence of promoter methylation, like presence of bacterial flora . In our study, the presence of promoter methylation of MLH1 gene was also not related to the mRNA levels. This lack of association could be due to the small size of the population. An inverse correlation to MLH1 expression was observed in a previous study of Oster et al.  and may have different explanation. Methylation of MLH1 gene might be explained by the fact that only few CpG sites were interrogated, and the interrogated sites may not be the sites involved in regulation of the gene. In addition, the presence of alternative transcription start sites may also be involved. Recently Jones  summarized that genes silenced by Polycomb complexes are much more likely than other genes to become methylated in cancer and thus a silent state could even precede methylation. Thus, the evidence regarding the timing of DNA methylation could be consistent with the idea that methylation adds an additional level of stability to epigenetic states.
In patients where the tumor and normal tissues presented MLH1 promoter methylation, blood samples were also analyzed to confirm potential germline hypermethylation. However, in our study, we did not observe any MLH1 promoter methylation in the DNA from blood (data not shown). This result pointed to somatic origin of MLH1 promoter methylation in our study.
Although rather small (1.16-fold), we also observed different expression levels for EXO1 gene when compared tumor and adjacent mucosal tissues. Higher expression levels for EXO1 in tumor tissues are in agreement with the study of Ioana et al. . Although, data in that study were normalized to a different reference gene, GAPDH, and the investigated population was also smaller than ours. Recently, Caradec et al.  suggested not to use GAPDH as a reference gene for normalization in CRC experiments, since it appears to be among the most variable. Other authors showed that GAPDH expression varies according to oxygen tension and hypoxia, critical factors in cancer development, especially in CRC . On the other hand, Ide et al.  observed a lower mRNA level of MMR genes in tumor samples as compared with the normal tissue. In our study, MSH3 gene had significantly higher expression levels in colon tumors when compared to adjacent mucosa. Tentori et al.  observed that defective expression of the protein MSH3 is frequently detected in colon cancer. Higher expression levels were found in tumors of the colon when compared to those in the rectum. These differences were more pronounced in EXO1, MSH2, MSH3, MSH6, and PMS2 genes. The same tendency was observed in adjacent mucosa for MSH2 and PMS2 genes. Our results may suggest different mechanisms in the genesis of colon and rectal cancers as it was already postulated by . The reason for higher mRNA levels of MMR genes in colon could be actually due to the fact that stools are kept in the colon for a longer time than in the rectum. In this way, colon is more exposed to various carcinogens from the food, and thus needs more protection against the carcinogenic events. Higher expression levels of DNA repair genes could be one of these mechanisms of protection. As tumor localization was the major factor influencing gene expression, location-specific analysis may identify location-associated pathways and enhance the accuracy of class prediction.
In the present study, we have also observed a strong relationship between between EXO1 expression and those of genes involved in the MutSα heterodimer (MSH2-MSH6). Previously, Jiricny  also noticed that decreased activity of EXO1 is accompanied with the low concentrations of genes involving in the MutSα heterodimer.
We observed a strong correlation between expressions of MSH2 and MSH6 genes. Vageli et al.  recently demonstrated that reduction of MSH6 mRNA levels is a frequent event in bladder tumorigenesis and reflects a common mechanism of suppression with MSH2. Another MMR heterodimer, MutLα, consisting of MLH1 and PMS2, positively correlated, but the strength of such correlation was considerably lower than that for the MutSα heterodimer. Interestingly, the strongest correlations were observed between MLH3 and PMS1 and PMS2. This is the first time that a correlation between above genes is reported. Previous observations indicated that PMS2 gene is required for the correction of single-base mismatches, and PMS2 and MLH3 contribute both to the correction of insertion-deletion loops resulting from DNA replication, DNA damage or from recombination events between non-identical sequences during meiosis . The role of PMS1 in MMR still awaits further clarification, but it is assumed that coordinates the downstream processes after mismatch recognition by MutSα heterodimer together with MLH1.
DNA repair pathways are a part of a multistep, multifactorial process to remove the damaged DNA sequence and to resynthesize particular part of the DNA strand. Thus, interplay exerted by multiple genes is crucial and more informative for identifying genes responsible for human cancer. Analyzing the difference in expression, individual variability, and co-expression in our study has provided an initial characterization of the MMR pathway and can help in further understanding of the cellular DNA repair system in human CRC.
A lack of association between methylation in MMR genes (representing rather low-frequency events) and their corresponding expressions could be due to the small size of the population. Above aspect emerges therefore as a main limitation of the present study.