The KLF4 protein contains both activator and repressor domains, and can function as a positive or negative regulator of gene expression
[27–29]. Oncogenic effects of KLF4 have been reported in breast, skin, and lung cancers
[14, 15, 30]. However, KLF4 also acts as a tumor suppressor in human colon cancer development
. Information on tumor stage was available for some, but not all, of the human primary colon cancers examined here, and as a consequence we could not corroborate an early report suggesting KLF4 as a prognostic predictor of colon cancer
. Prior studies in human colon cancer cell lines and primary colorectal cancers implicated hypermethylation of the KLF4 gene promoter, loss of heterozygosity, or mutation of the open reading frame
. Recently, a KLF4/miR-206 autoregulatory feedback loop was reported to regulate protein translation reciprocally in normal and cancer cells
. In this investigation of rat colon tumors, human primary colon cancers, and human colon cancer cell lines, the data supported an inverse trend between KLF4 and miR-206.
Located on chromosome 6p12.2, miR-206 is similar in expression and function to miR-1, but its sequence differs by four nucleotides
. Studies in breast cancer models have reported tumor suppressive effects of miR-206 due to its pro-apoptotic properties, via the inhibition of notch3 signaling and cell migration
[9, 33] or proliferation
. Furthermore, miR-206 was one of five miRNAs that exhibited stage-dependent differential expression in human colorectal cancers
. Interestingly, the latter report noted, in contrast to the current study, that miR-206 levels were more typically attenuated whereas let-7a was increased in the cancers. Given that let-7 family members are normally ascribed a tumor suppressor function, the authors speculated that a high let-7a/low miR-206 ‘signature’ might designate colon tumors with a unique phenotype in terms of cancer progression, compartmentalization, or microenvironment
. This subset of cancers would differ from the more typical scenario involving increased miR-206 and reduced let-7a expression.
Loss of let-7a, and of other let-7 family members, coincided with changes in other high-abundance miRNAs in the heterocyclic amine-induced rat colon tumors examined here
. Interestingly, however, when the dataset of 679 miRNAs was taken in its entirety, the low-abundance miR-206 was identified as the most significantly upregulated (up to 73-fold) in rat colon tumors relative to normal colonic mucosa. In a separate study using azoxymethane as the initiating agent, miR-206 was also significantly upregulated (~100-fold) in rat colon tumors
. The latter report did not pursue miR-206 further, owing to its low abundance relative to other miRNAs. Nonetheless, we were intrigued that two quite different colon carcinogens increased miR-206 so dramatically in colon tumors, relative to normal colonic mucosa.
Thus, we next examined a panel of human primary colon cancers and detected increased miR-206 levels in ~50% of the cases. As predicted by computational modeling, a significant inverse relationship was noted between miR-206 and KLF4 (Figure
3A, inset). Other known colon cancer miRNA signatures were confirmed, such as an increase in miR-21 and a decrease in miR-98, but no concordance was observed for these miRNAs and KLF4 (data not shown). To provide further proof-of-concept, we examined a panel of human colon cancer cell lines and noted that SW480 and SW48 cells provided the best evidence for an inverse association between miR-206 and KLF4. In cells with intermediate constitutive levels, forced expression or knockdown of miR-206 resulted in the expected reciprocal changes in KLF4, and miR-206 ectopic upregulation increased cell proliferation kinetics in real-time monitoring assays. Interestingly, Caco-2 cells resembled non-transformed CCD841 cells, as well as certain human primary colon cancers (Cases 9 and 15, Figure
3), in having low expression levels of both miR-206 and KLF4. The latter signature appeared to involve slower cell proliferation and doubling time, although further work is needed to clarify this possibility, especially in vivo. Among the other factors that might contribute in vitro, miR-206 expression was influenced by cell confluency, being about threefold higher at 70% versus 20% confluency (MAP et al., unpublished data).
In contrast to these observations, cancers of the breast and lung have been characterized as exhibiting low miR-206 expression levels, suggesting a possible tumor suppressor function
[9, 10, 12]. Changes in the expression of KLF4 also can be site specific, being upregulated in cancers of the breast, skin, and lung, but attenuated in colon and gastric tumors
[14, 15, 19, 31]. The prior investigation of high-abundance miRNAs in rat colon tumors highlighted a role for c-Myc, Oct-3/4, and Sox2
, whereas the present work has implicated a fourth ‘defined factor’ for pluripotency, namely KLF4. Further studies are in progress on the role of pluripotency factors, cancer stem markers, and other potential molecular targets of the miR-206/KLF4 axis (Figure