There is increasing evidence showing that expression of miRNAs is deregulated in cancer. This deregulation is thought to occur via a number of different pathways, including transcriptional [27
] and epigenetic [29
] alterations. Additionally, mutations in the coding regions of miRNAs, as well as DNA copy number abnormalities, are thought to contribute to miRNA deregulation [31
]. Finally, dysfunctional or deregulated proteins in the miRNA biogenesis pathway may also play a role in human tumorigenesis [34
The current study demonstrates that two non-synonymous mutations (rs11544382 (M1115T) and rs34324334 (S241N)) in an important mi-croRNA biogenesis gene, XPO5
, have significant associations with breast cancer risk. These results are consistent with findings from two previous studies, which showed a borderline significant association between renal cell carcinoma and rs11070, a SNP in the 3′ untranslated region of XPO5
, in addition to an association with esophageal cancer risk [12
]. This SNP, however, showed no association with bladder cancer risk in another recent study [37
]. Nevertheless, these studies suggest a potential role for the microRNA biogenesis gene, XPO5
, in human cancers.
The two nsSNPs examined in the current study may influence XPO5 functionality and may therefore be causally related to breast cancer development. Our bioinformatic searches indicate that both SNPs are located in XPO5 functional domains, and are strongly suggestive of the deleterious impact of an rs11544382 mis-sense variant. Interestingly, the effect of rs34324334 missense variation was predicted to be benign. However, the fact that rs34324334 is located in a highly conserved Exportin-1/Importin-b-like region is suggestive of its potential functional importance. Alternatively, it is possible that the rs34324334 mis-sense variant is itself not causal, but is linked to one or more polymorphisms associated with breast cancer risk. Unfortunately, the lack of supporting experimental evidence precludes the drawing of any definitive conclusions with respect to rs34324334's functional role in breast cancer at this point. Further functional analysis is required to determine with greater certainty whether either of these SNPs has a true functional impact on XPO5 or whether they are merely indicative of the influence of other polymorphisms.
It should be noted that a portion of the controls used for our genotypic analysis consists of women who underwent surgery for benign breast conditions. It is possible that some of these women may be at an increased risk of developing breast cancer, and assuming that the observed genetic associations are true, the inclusion of these women into the control population might bias our estimates toward the null (i.e., the true effect sizes are stronger than those observed).
In addition to genetic associations, our epigenetic analysis further demonstrates that alterations in the methylation pattern of XPO5′s promoter region could be a risk biomarker. To our knowledge, there have been no studies linking genetic or epigenetic variations in key components of the miRNA biogenesis pathway to breast cancer tumorigenesis. One potential concern regarding these results is whether the observed epigenetic changes in the surrogate tissue, peripheral blood lymphocytes (PBLs), accurately reflect the changes in the breast tissue. A study examining whether methylation status of IGF2 in PBLs was representative of the status in colon tissue found high agreement between the two tissue types (kappa statistic = 86.5%, p < 0.0001) [38
]. Additionally, a large case-control study of breast cancer reported significant associations between methylation status of genes measured from PBLs and cancer risk, demonstrating that, in principle, methylation in PBLs may be useful biomarkers for risk prediction [39
]. Although both studies support the concept that methylation status of PBLs may serve as a surrogate measure for epigenetic alterations at the target tissue, without paired RNA and DNA from patient tissue, it is difficult to definitively determine the phenotypic impact.
Despite the absence of RNA in our study, our methylation findings are consistent with the XPO5
tissue expression levels present in the ArrayExpress database (www.ebi.ac.uk/arrayexpress
]). Since hypermethylation is generally associated with decreased gene expression, and we found decreased methylation at the XPO5
promoter to be associated with increased breast cancer risk, we would expect greater expression of XPO5
in breast tissue from individuals with breast cancer. The tissue expression array data was consistent with this prediction, as breast tumor tissue had the highest level of XPO5
expression, followed by adjacent tissue from breast cancer patients, and tissue from healthy controls. This significant gradient in gene expression lends further support to the idea that aberrant XPO5
expression may play a role in early tumor development.
Furthermore, altered expression of XPO5
has been previously detected in human cancers. In a study of low-grade lung adenocarcinoma, expression of XPO5
was 2.8-3.0-fold down-regulated [34
]. On the contrary, expressions of XPO5
was 1.6-fold up-regulated in high-grade prostate cancer [41
], which may explain an almost global increase of microRNA expression in prostate adenocarcinoma. Overexpression of XPO5
has been shown to enhance not only miRNA expression, but also the inhibition of target gene expression [9
]. Likewise, reduced XPO5
expression has been shown to limit miRNA biogenesis [10
]. Alterations in miRNA levels resulting from aberrant XPO5
expression may have a similar impact on a cell's prolifera-tive rate, offering a possible mechanism by which altered XPO5
promoter methylation patterns could increase an individual's risk of developing breast cancer.
As previously mentioned, a key limitation in our investigation is the absence of an experimental functional component to complement our genetic association and bioinformatics analyses. The lack of empirical information on the functional nature of rs11544382 and rs34324334 variants precluded the possibility of corroborating our genetic association results with findings from our methylation and transcriptional profiling analyses. A more in-depth examination of the functional impact of these genetic variants is thus warranted to check for consistency between the impact of genetic and epigenetic XPO5 variations on breast cancer risk. Moreover, the small size of the subjects eligible for our methylation analysis prevented an examination of potential interactions between XPO5 polymorphisms and promoter methylation in modulating XPO5 function. As such, further efforts should be made to investigate whether, and to what degree, XPO5 methylation differs under disparate genotypic settings, and whether these differences can be informative in predicting XPO5 function and breast cancer risk. Conversely, future studies may also need to consider how different methylation settings can enhance or attenuate the phenotypic effects conferred by XPO5 polymorphisms.