Here, we demonstrate that abrogation of the Wnt signaling pathway in colon carcinoma cells lead to up-regulation of CLU. By further dissecting the molecular signaling cascade responsible for increased CLU levels, we find that changes in CLU expression are specifically controlled by the branch of the Wnt pathway mediated by the transcription factor TCF1.
Our finding, that CLU expression can be regulated by the Wnt signaling pathway, add to the plethora of stimuli which can regulate CLU levels. The CLU promoter has already been shown to be responsive to a variety of external stress-inducing agents such as oxidative stress, ionizing radiation, heat shock, and chemotherapeutics[2
]. In addition several types of growth factors including TGF-β, NGF, EGF, PDGF, βFGF, and interleukin-1,-2 and -6 have been demonstrated to influence CLU expression in cultured cells[2
]. Although some of these signals may ultimately converge on identical downstream effectors directly responsible for regulating CLU promoter activity, several unique trans
- and cis
-factors have been demonstrated to influence CLU levels[2
]. Thus, CLU expression can be modulated by many factors and their impact on CLU levels may in some cases be context dependent. For example, TGF-β and EGF treatment elevate CLU levels in some cell types whereas the very same growth factors suppress CLU expression in others[2
]. Also, interleukin-6 induced CLU expression in hepatoma cells (HepG2) whereas CLU was down-regulated in rat glial cultures exposed to the same cytokine[27
The Wnt signaling pathway is known to be involved in the development of colorectal cancer and is also of major importance in colon physiology[29
]. In particular, this pathway controls the maintenance of colon tissue architecture by controlling epithelial cell behaviour in the crypts of Lieberkühn. The pathway is activated in epithelial cells at the crypt bottom and decreases in a gradient toward the crypt top. In colonic epithelial cells, it may therefore be expected that CLU is expressed in an inverse pattern to that of Wnt signaling. This, however, does not seem to be the case in the adult human colon as it has previously been reported that CLU only stains a small fraction of scattered colonic epithelial cells with neuroendocrine differentiation[12
]. In this tissue it therefore appears as if the Wnt pathway is not the dominant force controlling CLU expression. Rather, CLU expression is likely the result of a complicated interplay between various pathways. The notion that the complexity of CLU regulation in vivo
is not always being fully recapitulated by in vitro
model systems, was also recently emphasized by Patel et. al who unexpectedly observed decreased CLU levels with rat donor age in glial cultures, despite CLU in vivo
shows an opposite adult age trend: CLU expression increases during aging in select brain regions and in neurodegenerative disorders[27
]. Although Wnt signaling may not be the primary regulator of CLU expression in human colonic epithelial cells, other studies indicate that this might be the case in epithelia of the murine intestine. Lars E. French and colleagues used in situ hybridization (ISH) to investigate CLU expression during murine embryogenesis, and found CLU to be weakly expressed in dividing epithelial cells at the base of the villi in the small intestine but strongly expressed in more differentiated cells that line the rest of the villus[30
]. Similarly, Suh E. and colleagues used immunohistochemistry (IHC) to show that epithelia of the developing murine intestine was positive for CLU protein staining, and exhibited a gradient of activity with higher expression in the villus compartment than in the crypts[9
]. At least in the developing murine small intestine, it may therefore be that Wnt signaling contributes to modulate CLU expression in epithelial cells.
Although we identified which TCF factor was responsible for CLU up-regulation, namely TCF1, it remains to be demonstrated whether TCF1, or other downstream effectors, bind directly to the CLU promoter and regulate its activity. A previous study reported the Wnt target, c-MYC, to repress CLU expression in murine colonocytes although no direct interaction between c-MYC and the CLU promoter was demonstrated[14
]. In agreement with these observations we observed a decrease in c-MYC levels before CLU was induced upon over-expression of dnTCF1 in LS174T cells. However, considering that c-MYC levels also dropped when we over-expressed dnTCF4 and CLU levels remained unchanged, we do not believe c-MYC to regulate CLU levels in our model system. Notably, other investigators found no changes in CLU levels in murine fibroblast cells stably transfected with a c-MYC construct compared to mock transfected cells[31
]. Therefore, other factors might be more likely to regulate CLU in our model system, possibly, AP-1 or Egr-1 transcription factors which can bind directly to the CLU promoter[6
] and have been demonstrated to be suppressed by Wnt signaling[32
It has become clear that the CLU gene encode several mRNA variants. At least three CLU mRNA variants are expressed in colon tissue, CLU34, CLU35, and CLU36. It has previously been reported that only one of these variants, CLU35, was down-regulated in human colorectal tumors samples compared to normal tissue indicating that the CLU mRNA variants may be differentially regulated[12
]. This view was supported by a recent study by Cochrane et al. who used real time PCR to demonstrate that two CLU transcripts could be differentially regulated in prostate cells by androgens[13
]. Interestingly, our results indicate that the Wnt signaling pathway, via TCF1, specifically regulate the expression levels of CLU34. Our 5'-end RACE results indicate that the NM_001831.2 sequence is not present in colon carcinoma cells and this observation is further supported by the failure to detect the full NM_001831.2 sequence by RT-PCR in normal and cancerous colon tissue samples. Instead, both RT-PCR[12
] and real time PCR can successfully detect a transcript, CLU34, which is shorter at the 5'-end and matches the most predominant CLU mRNA isoform previously reported[21
]. This shorter sequence has implications for the nature of the predicted protein because the ATG of the RefSeq NM_001831.2 exon 1 is not part of the CLU34 transcript and consequently the ORF of the predicted protein starts in exon 2. Hence identical proteins are predicted from the CLU35 and CLU34 transcripts. Importantly, this common protein is predicted to be secreted which fits exactly with what we find in our model system, as concomitantly with the induction of CLU34 mRNA an abrupt increase in cytoplasmic and secreted CLU protein species is observed. Thus, the CLU34 mRNA species apparently produce these protein variants.
The existence of two human CLU transcripts (CLU34/35) probably encoding identical CLU proteins is paralleled by previous findings in quail which demonstrated that two CLU transcripts with different, mutually exclusive, non-coding 5' exons can be produced from the avian gene. However, whereas the avian CLU transcripts are produced from two promoters often being co-regulated as suggested by correlating expression levels of the two CLU transcripts in tissue samples from a variety of organs[11
], the human CLU transcripts are clearly differentially regulated.