β-catenin plays a crucial role in embryogenesis, tissue homeostasis and carcinogenesis. Because of its role in regulating homeostasis of the intestinal tract as well as being a key player in sporadic forms of colorectal cancer, inhibition of β-catenin is not a suitable strategy to treat patients with colorectal cancer. For this reason, identifying and characterizing β-catenin target genes could result in better understanding of colorectal carcinogenesis and development of new therapies. Here, we used two non-isogenic colorectal cancer cell lines, DLD1, and SW480, and identified 335 commonly regulated β-catenin target genes. 193 of these genes were also differentially regulated in LS174T cells. (Genes that are differentially expressed due to the genetic background of a given cell should be deregulated to a similar extent irrespective of the siRNA treatment and hence should not appear as being differentially regulated in the DNA microarray experiment.) Compared to the number of differentially regulated genes in LS174T and DLD1 cells (786 genes) as well as in LS174T and SW480 cells (677 genes), the number of commonly regulated genes in DLD1 and SW480 (335 genes) was low, resulting in a low number of commonly regulated genes in DLD1, SW480, and LS174T cells (193 genes). While DLD1 and SW480 cells both express a mutated form of APC, they differ with respect to the underlying genetic disturbance: DLD1 cells show microsatellite instability (MSI), whereas SW480 cells are derived from a tumor with chromosomal instability (CIN). However, SW480 cells (CIN) share a lot of commonly regulated genes with LS174T cells (MSI), suggesting that other factors apart from genetic instability contribute to the observed differences in gene expression between DLD1 and SW480 cells. Looking at the mutational status, DLD1, SW480, and LS174T cells differ with respect to the status of the genes APC, PI3K, and TP53, but they are identical with respect to status of the genes KRAS and PTEN: DLD1 (APC: Mutant; TP53: WT; PI3K: Mutant; KRAS: Mutant; PTEN: WT), SW480 (APC: Mutant; TP53: Mutant; PI3K: WT; KRAS: Mutant; PTEN: WT), and LS174T cells (APC: WT; TP53: Mutant; PI3K: Mutant; KRAS: Mutant; PTEN: WT) [82, 83].
Deregulation of PI3K/Akt signaling (as observed in DLD1 and LS174T) has been implicated in the phosphorylation of β-catenin at Ser552 , thereby promoting nuclear accumulation and enhancing β-catenin dependent transcriptional activity even in the absence of aberrant Wnt signaling [85, 86]. This observation suggests that in particular LS174T cells that express wild-type APC could benefit from a mutation in the PI3K gene. But even DLD1 cells (with mutant APC) could profit from PI3K mutations: Deming et al. demonstrated that expression of a dominant active form of PI3K in Apc
Min/+ mice resulted in an increased tumor number and size and implicated the CCND1 gene as one transcriptional target that contributes to the observed phenotype of these mice . In our experiments, CCND1 mRNA expression levels were up regulated in LS174T cells, but not in DLD1 cells.
The tumor suppressor p53 has been implicated in down-regulating β-catenin expression and/or activity by Siah-1 dependent (and GSK-3β independent) degradation [88, 89], reduction of TCF4 mRNA and protein levels  as well as reduction of β-catenin mRNA by p53 dependent regulation of miRNA miR-34 . Accordingly, inactivation of the transcriptional activity of p53 by mutations –as observed in SW480 cells– should result in increased β-catenin mRNA and protein levels and transcriptional activity as well as increased mRNA and protein levels of TCF4 (TCF7L2). Interestingly, LS174T cells showed a strong increase in the fold change value for CTNNB1 (encoding β-catenin) (4,56 versus 1,59 (DLD1) and 1,18 (SW480), respectively). All three cell lines showed slightly reduced fold change values for TCF4 (TCF7L2). The effect of p53 inactivation on the development of intestinal adenomas was also analyzed in Apc
Min/+ mice: Despite the fact that p53 protein expression is deregulated as a consequence of APC loss (and probably other factors) and p53 is known to regulate β-catenin expression via a negative feedback loop, loss of p53 did not promote the initial steps of intestinal neoplasia in Apc
Min/+ mice , suggesting that p53 has only a limited role in this mouse model.
Apart from regulating β-catenin protein expression levels via proteasomal degradation, (nuclear) APC has been implicated in affecting β-catenin activity by recruiting co-repressors of β-catenin to promoter regions, as well as sequestering and enhancing the nuclear export of β-catenin [93–96]. Mutated forms of APC still show a nuclear localization, but they are more frequently observed in the cytoplasm when compared to wild-type APC, suggesting that mutation of APC is not only contributing to higher β-catenin protein expression levels, but also contribute to enhanced β-catenin activity . When Zeineldin et al. generated mice expressing APC with mutated nuclear localization signals (mNLS) and compared these Apc
mNLS/mNLS mice with Apc
+/+ mice, they found up regulation of the mRNA for the genes AXIN2, MYC, and CCDN1 as well as down regulation of HATH1 mRNA in response to stimulation with Wnt . Accordingly, we expected an increase in the fold change values for AXIN2, MYC, and CCDN1 transcripts as well as a reduction of HATH1 transcripts in DLD1 and SW480 cells (APC mutated) compared to LS174T (APC wild-type). In our experiments, DLD1 and SW480 showed a modest increase in the fold change values for MYC (see Additional file 1, tab “DLD1_SW480_LS174T_only”) compared to LS174T cells. However, LS174T (and SW480) cells responded with a higher fold change for AXIN2 compared to DLD1 cells. CCND1 transcript levels were increased only in LS174T cells, whereas HATH1 transcript levels did not changed in any of three cell lines.
The differential expression of the genes AXIN2 and CCND1 in the three colorectal cancer cell lines highlights the fact that the genetic background of a given cell has a major impact on the expression of a specific gene. However, linking a specific somatic mutation in a tumor suppressor gene or proto oncogene, e.g. APC, PI3K, or TP53, to these differences in expression is difficult because the expression of a specific gene is affected by multiple transcription factors (along with their cofactors and corepressors). Therefore, we are not able to attribute the observed differences in the gene expression profiles for DLD1 and SW480 to a single mutation.
Compared to LS174T cells, the total number of differentially regulated β-catenin target genes was much lower in DLD1 and SW480 cells. This could be either specific to the genetic background of the cell lines used or due to different experimental approaches. While we transiently transfected DLD1 and SW480 cells with β-catenin siRNA and analyzed the gene expression 48 hours after transfection, Mokry et al.  stably transfected LS174T cells with a doxycycline inducible expression vector encoding β-catenin shRNA and stimulated the cells for 72 hours with doxycycline before analyzing the gene expression profile. Both research groups used Affymetrix Human Genome U133 Plus 2.0 DNA microarrays for their experiments. The longer incubation period used to identify potential β-catenin target genes in LS174T cells might contribute to secondary effects like the differential expression of indirect β-catenin target genes, thus explaining why the absolute number of differentially regulated genes was higher in LS174T cells when compared to DLD1 and SW480 cells. For this reason, we were particularly interested in genes that are up and down regulated in the non-isogenic colorectal cancer cell lines DLD1 and SW480 or DLD1, SW480, and LS174T cells, assuming that these genes are commonly and directly regulated by β-catenin.
Remarkably, the number of β-catenin regulated genes was very similar in the three cell lines when focusing on the list of 66 previously described β-catenin target genes. Here, DLD1, SW480 and LS174T cells differentially regulated 21, 30 and 34 of these genes, respectively. There are at least two reasons why the number of detected β-catenin target genes did not include (almost) all 66 β-catenin target genes presented in Table 1. First, the detection of β-catenin target genes is dependent on the cellular context. DLD1, SW480, and LS174T represent colorectal cancer cell lines. Therefore, the expression pattern of β-catenin target genes in these cells is likely to differ from β-catenin target genes found in healthy colon cells, colonic adenomas, or metastatic colonic cells that are also listed in Table 1. Second, the gene expression profile was analyzed 48 and 72 hours after beginning with the RNAi treatment, respectively. Different periods of time influence the identification of β-catenin target genes as well, since some of these genes might be regulated only for a short period of time immediately after induction of β-catenin activity. This phenomenon has been described by Van de Wetering et al. (2002): They transfected LS174T cells with doxycycline (Dox) inducible expression vectors encoding dnTCFs and analyzed the changes in gene expression using DNA microarrays. 11 h after treatment with Dox, 2411 genes where differentially regulated. After an incubation period of 23 h, they identified 1199 differentially regulated genes. Comparison of the two sets of genes revealed that the expression of 1971 genes was abrogated between 11 and 23 h after Dox treatment, while 759 additional genes appeared in this time frame . Similarly, another group analyzed the gene expression profiles of several known β-catenin target genes in a time-course experiment and classified these genes according to their expression profiles . Based on this criterion, they identified genes that were “immediately up (or down) regulated”, “early up (down)”, or “late up (down)” in response to the activation of the Wnt/β-catenin signaling cascade. When comparing their results to the regulation of the same genes in different cellular contexts, this group found that the regulation of these genes was similar, but not necessarily identical in different cell lines. This observation suggests that the expression levels and activities of a given transcription factor and its cofactors (or co-repressors) influence the starting point and duration of the gene expression of a specific target gene. Therefore, it is recommendable to identify target genes of a given transcription factor in different cell lines at least at the same time points, better yet in a time-course experiment.
Using the KEGG pathway database, we identified 11 signaling pathways or cellular functions in DLD1 and SW480 cells that are differentially regulated in response to treatment of the colorectal cancer cell lines with β-catenin siRNA. In addition, the steroid hormone biosynthesis pathway was similarly regulated in all three cell lines analyzed. Whereas the steroid hormone biosynthesis pathway [101, 102] as well as the endocytosis pathway [103, 104], the insulin signaling pathway [105, 106], apoptosis [107, 108], regulation of the actin cytoskeleton , and focal adhesion , have been associated with Wnt/β-catenin signaling, there are no publications in the PubMed database linking β-catenin to the remaining six pathways in colorectal cancer cells. Interestingly, the list of KEGG pathways/functions did not include the canonical Wnt signaling pathway. This was probably due to the definition of the “KEGG Wnt signaling pathway” that summarized the canonical, the planar cell polarity (PCP), and the Wnt/Ca2+ signaling pathway under one database entry, despite different functions of these signaling pathways. While the canonical Wnt pathway has been implicated in regulating the activity of the β-catenin protein, PCP signaling results in remodeling of the cytoskeleton that is a prerequisite for migration and cell polarization via the GTPases RHOA and RAC1. Wnt/Ca2+ signaling, however, plays a role in the regulation of cell adhesion and cell movements during gastrulation and results in the activation of protein kinase (PKC), calcium calmodulin mediated kinase II (CAMKII) and calcineurin .
Furthermore, all three Biocarta pathways that were enriched in DLD1 and SW480 cells have been associated with Wnt/β-catenin signaling based on PubMed publications: the m-calpain pathway [111, 112], the Creb pathway [26, 113], and the IGF1R pathway [114, 115].
Since 6 out of the 66 previously described β-catenin target genes encode proteins that belong to the bHLH protein family (Table 1), we focused our analysis on this group of genes. Our analysis revealed that the genes ASCL2, ID1, ITF2, and MYC were regulated in a β-catenin dependent manner in DLD1 and SW480 cells. Interestingly, the proteins ASCL2 and ITF-2B as well as ID1 and ITF-2B are known to interact with each other [116, 117], suggesting that these bHLH proteins form a functional network in colorectal carcinoma cells.