Colon cancer accounts for more than 10% of all cancer deaths annually. Our genetic evidence from Drosophila and previous in vitro studies of mammalian Atonal homolog 1 (Atoh1, also called Math1 or Hath1) suggest an anti-oncogenic function for the Atonal group of proneural basic helix-loop-helix transcription factors. We asked whether mouse Atoh1 and human ATOH1 act as tumor suppressor genes in vivo. Genetic knockouts in mouse and molecular analyses in the mouse and in human cancer cell lines support a tumor suppressor function for ATOH1. ATOH1 antagonizes tumor formation and growth by regulating proliferation and apoptosis, likely via activation of the Jun N-terminal kinase signaling pathway. Furthermore, colorectal cancer and Merkel cell carcinoma patients show genetic and epigenetic ATOH1 loss-of-function mutations. Our data indicate that ATOH1 may be an early target for oncogenic mutations in tissues where it instructs cellular differentiation.
Like most cancers, colon cancer displays a loss of differentiation, and the stronger this property, the more aggressive the cancer. This suggests that the loss of the capacity to differentiate may be a critical and possibly early event during the formation of these tumors. The key gene instructing secretory cell fate differentiation in the epithelium of the colon, namely Atonal homolog 1 (ATOH1), is highly conserved in flies, mice, and humans. We asked whether ATOH1 could be a pivotal factor in causing colon cancer in mice and humans. Our studies show that colon-specific loss of ATOH1 in mice is sufficient to trigger colon cancer and that the majority of human colon cancers also have an inactivated ATOH1. Reactivating ATOH1 in cultured human colon cancer cells causes these cells to stop dividing and to commit suicide. Since reactivation of this epigenetically silenced gene can be achieved using small chemical compounds, studying how ATOH1 acts may offer therapeutic avenues in the future.
Citation: Bossuyt W, Kazanjian A, De Geest N, Kelst SV, Hertogh GD, Geboes K, et al. (2009) Atonal homolog 1 Is a Tumor Suppressor Gene. PLoS Biol 7(2): e1000039. https://doi.org/10.1371/journal.pbio.1000039
Academic Editor: Nicholas Hastie, Western General Hospital, United Kingdom
Received: September 4, 2008; Accepted: January 12, 2009; Published: February 24, 2009
Copyright: © 2009 Bossuyt et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by VIB, a Stichting Emmanuel Van der Scheuren predoctoral fellowship to WB, Fonds voor Wetenschappelijk Onderzoek–Vlaanderen (FWO) postdoctoral fellowship to SA, the European Molecular Biology Organization (EMBO) Young Investigator Program, Impuls, CREA, and Geconcerteerd Onderzoeksacties (GOA) grants from K.U. Leuven, and FWO grants G.0542.08N and G.0543.08N to BAH, the Foundation against Cancer, foundation of public interest to JC and Foundation for Digestive Health and Nutrition Research Scholars Award, National Institutes of Health (NIH) K01 DK071686, and American Cancer Society (Ohio Affiliate) Pilot Award to NFS. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: AOM, azoxymethane; CGH, comparative genomic hybridization; CRC, colorectal cancer; GALT, gut-associated lymphoid tissue; JNK, Jun N-terminal kinase; MCC, Merkel cell carcinoma; NTRK, Neurotrophic tyrosine kinase receptor type 1; RTK, receptor tyrosine kinase; RT-qPCR, quantitative reverse-transcriptase PCR
The Atonal (Ato) proneural transcription factors form a highly conserved group of key developmental regulators in multiple neural and neuroendocrine tissues. The mammalian Ato (CG7508) ortholog, ATOH1 (Ensembl accession number: ENSG00000172238), is essential for cell fate commitment of mechanoreceptive Merkel cells in the skin  and the secretory goblet, Paneth, and enteroendocrine cells in the intestine [2,3] in addition to multiple neuronal lineages [4,5]. The functional conservation between the fly and mammalian proteins is underscored by the fact that Drosophila ato can fully rescue the Atoh1 (Ensembl: ENSMUSG00000073043) null mutant mouse .
Genetic analyses in Drosophila  suggest that ato regulates the formation and progression of tumors in fly retina, where it acts as a master regulator of cell fate specification. In mammals, two aggressive human cancers derive from tissues where ATOH1 instructs cell fate commitment, namely Merkel cell carcinoma (MCC) and colorectal cancer (CRC). MCC is a rare, but very aggressive, neuroendocrine cancer of the skin with approximately 40% mortality . CRC is a highly prevalent cancer with high mortality (36%) representing 11% of all cancer deaths annually . Recent studies in colon cancer cell lines suggest that ATOH1 can inhibit tumor cell growth in vitro . If the anti-oncogenic function of Drosophila ato is conserved in its mammalian counterparts, one would predict that the loss and gain of function of ATOH1 would enhance and suppress tumor formation, respectively, in MCC and CRC models. In addition, the ATOH1 should be subject to loss-of-function mutations in a significant number of human cancer patients.
We tested this prediction in two different mouse models for colon cancer, as well as human MCC and CRC cell lines. In addition, we examined the status of the ATOH1 locus in primary tumor samples from human MCC and CRC patients. Our data show that loss of ATOH1 strongly enhances the formation and progression of tumors in mice and human cell lines. Conversely, gain of ATOH1 function strongly inhibits the oncogenic phenotypes in human cell lines. Furthermore, we find genetic and epigenetic loss-of-function mutations with very high frequency in primary human tumors derived from ATOH1-dependent tissues and provide biochemical insight into how these epigenetic mutations may arise. Finally, we describe a highly conserved anti-oncogenic molecular signaling pathway that links ATOH1 activity to the stress sensor Jun N-terminal kinase (JNK) pathway mediated via cell type–specific differential co-option of receptor tyrosine kinases (RTKs). Together with the genetic analysis in Drosophila , these data support a novel, highly conserved tumor suppressor function for the Atonal group of transcription factors.
Loss of Atoh1 Promotes Tumor Formation in Two Colorectal Cancer Mouse Models
In the mouse, Atoh1 is a master regulator of secretory cell fate commitment in the intestinal epithelium [2,11]. To investigate whether Atoh1 plays a role in intestinal tumors, we analyzed the function of the mammalian ato homolog, Atoh1, in colon tumorigenesis. We assessed the tumor susceptibility of mice with an intestine-specific deletion of Atoh1 (Atoh1Δintestine)  in two different established mouse models of CRC.
We first treated Atoh1Δintestine mice with azoxymethane (AOM), a chemical carcinogen that preferentially induces colon tumors. We found a significant enhancement in polyp formation in the large intestines of Atoh1Δintestine compared to wild-type littermate mice, characterized by increased incidence (33% [4/12] vs. 100% [8/8], p < 0.005; Figure 1A), multiplicity (0.75 vs. 10.3 polyps/colon, p < 0.0008; Figure 1C), and size (2.1 vs. 4.1 mm/polyp, p < 0.0004; Figure 1E) when examined under the dissecting microscope. Histological analysis of colons from AOM-treated Atoh1Δintestine mice confirmed these findings and showed a range of histological phenotypes, from severely hyperplastic mucosa with embedded multifocal adenomas, to large, highly dysplastic adenomas, with one animal having invasive adenocarcinoma (Figure 2A, 2A′, and 2A′′). In contrast, histological examination of AOM-treated Atoh1wt littermates showed none with adenomatous changes; with all polyp-like structures shown to be gut-associated lymphoid tissue (GALT; Figures 2B and S1). These data support the notion that loss of Atoh1 can be an initiating event in mammalian cancer formation.
(A) shows the incidence of polyps in AOM-treated mice.
(B) shows the incidence of polyps in the Atoh1wt and Atoh1Δintestine mice in an APCmin background.
In (A) and (B), the two-tailed Fisher exact test was used.
(C) Bar graph shows the average number of macroscopically visible polyps (>1 mm) in the colons of AOM-treated Atoh1wt (WT; n = 9 polyps in 12 mice, white bars) and Atoh1Δintestine (n = 82 polyps in 8 mice, black bars).
(D) The average number of polyps in the colons of APCmin (n = 27 polyps in 27 mice) and APCmin; Atoh1Δintestine (n = 123 polyps in 10 mice).
(E and F) The average maximum diameter of each polyp is shown as a bar graph for Atoh1wt or Atoh1Δintestine mice. (E) Comparison of polyp size from AOM-treated Atoh1wt and Atoh1Δintestine mice; and (F) from APCmin and APCmin; Atoh1Δintestine mice.
For (C–F), the two-tailed Student t-test was used to measure significance. Error bars indicate the standard error of the mean. Double asterisks (**) indicate p <0.01; triple asterisks (***) indicate p <0.001.
(A) Hematoxylin and eosin staining of a representative adenoma in AOM-treated Atoh1Δintestine colon (5× magnification). The section is in the rectal region and is characterized by cystic structures with severe crypt hyperplasia with multifocal embedded adenomas and severe lymphocyte infiltration of the submucosa. (L, lumen; M, muscle). (A)′ and (A)′′ show higher magnification (20×) of two areas of the adenoma, indicating the aberrant branched crypt structures that are imbedded in the submucosa.
(B) The colons of AOM-treated Atoh1wt mice showed large GALT (highlighted by arrows) that were macroscopically counted as polyps. Magnification is 5×. Note the relative size of the GALT compared to adjacent normal-appearing crypts.
(C) Histological analysis of colon in APCmin.
(D) Histological analysis in APCmin; Atoh1Δintestine mice show highly dysplastic adenomas with cystic structures and lymphocyte infiltration.
(E) Proliferation was measured by determining the percentage of BrdU-positive epithelial cells in AOM-treated Atoh1wt and Atoh1Δintestine normal-appearing colonic crypts. The white bar represents crypts from Atoh1wt mice; the gray bar represents nondeleted Atoh1wt crypts, and the black bar Atoh1-null crypts in Atoh1Δintestine mice.
(F) Apoptosis was measured by determining the percentage of cleaved caspase-3 (c-Caspase 3)-positive epithelial cells in AOM-treated Atoh1wt and Atoh1Δintestine normal-appearing colonic crypts. Bar shading indicates crypt genotype as in (E).
For each graph in (E and F), the two-tailed Student t-test was used to measure significance: a single asterisk (*) indicates p < 0.05; double asterisks (**) indicate p < 0.01. Error bars indicate the standard error of the mean.
In previous work, we found that ablation of Atoh1 leads to an increase of proliferation but has no effect on apoptosis . We therefore examined proliferation and apoptosis in the “preneoplastic” normal-appearing epithelium within the colons of AOM-treated Atoh1wt and Atoh1Δintestine mice. As in the Drosophila model, we find more proliferation of epithelial cells in Atoh1Δintestine crypts compared to nonrecombined wild-type crypts within the same animals, or compared to crypts from Atoh1wt littermates (Figures 2E and S2D–S2F) but no obvious difference in apoptosis (Figures 2F and S2G–S2I).
We extended our tumor analysis with an independent genetic mouse model for CRC by crossing Atoh1Δintestine mice to APCmin mice (APC: ENSMUSG00000005871), in which Wnt signaling is constitutively activated [12,13]. APCmin mice develop spontaneous adenomas primarily in the small intestine, with only occasional polyps in the colon . Interestingly, Wnt signaling is thought to interact negatively with ato during sense organ formation in the developing Drosophila epithelium . Comparing the large intestines of APCmin; Atoh1wt mice to APCmin; Atoh1Δintestine littermates at 16 wk, we find significantly more polyps in the colons of double-mutant mice, characterized by increased incidence (52% [14/27] vs. 100% [10/10], p <0.007; Figure 1B), multiplicity (1.1 vs. 12.3 polyps/colon, p <0.0001; Figure 1D), and size (3.3 vs. 4.3 mm/colon, p <0.0002; Figure 1F). Histological examination confirmed these polyps as adenomas in both APCmin; Atoh1wt and APCmin; Atoh1Δintestine mice, with two polyps in APCmin; Atoh1Δintestine mice having progressed to invasive adenocarcinoma (Figure 2C and 2D). The polyps in the APCmin; Atoh1Δintestine mice originated from the Atoh1 mutant crypts as indicated by the absence of goblet cells in the polyps (Figure S3A and S3B).
(A) ATOH1 expression normalized to GADPH and control colon samples: ATOH1 expression is lower in tumor compared to control samples. Additionally, adenocarcinomas have significantly lower ATOH1 expression than adenomas. Box plot indicating 25–75 percentiles, central line indicates the median, red cross indicates the mean. Error bars indicate next data point. Outliers are represented as dots. Double asterisks (**) indicate p < 0.01 (t-test).
(B) Deletions in the ATOH1 locus. Upper dashed line indicates the upper limit of single-deletion detection set at 0.70 of the ratio of ATOH1 locus to control locus. White and gray bars indicate different primer sets.
(C) Endogenous ATOH1 expression in different MCC14.2-derived cell lines and human keratinocytes (HK) in response to Atoh1 expression.
(D) Detection of methylation at the ATOH1 locus using pull-down assay of methylated DNA. Bands in “M-CpG pulldown” indicate positive for ATOH1 methylation. “Input” shows DNA input before the pull-down. Samples were processed blindly. Internal methylated and unmethylated controls are shown in the last four lanes.
(E) Inhibition of methyltransferase activity with 5-aza-deoxycytosine for 7 d leads to an increase of ATOH1 expression in the Ht29 cell line. Error bars indicate the standard deviation.
(F) Graph representing observations on the genomic and mRNA levels. First column: percentage of patients showing deletions or duplications (black: deletions, gray: two copies, and white: duplications). Second column: percentage of patients showing methylation using the pull-down of methylated DNA assay (black: methylated, and white: not methylated). Third column: mRNA expression of CRC samples versus control colon (black: low expression, grey: normal expression, and white: high expression).
In summary, thus far, loss-of-function analysis of Atoh1 in two mouse models of colon tumorigenesis supports a role for Atoh1 as a key switch in tumor formation and progression. This role appears to be mediated by increased cell proliferation in the absence of Atoh1. Importantly, the mouse tumors lack the secretory cell types that depend on Atoh1 for their formation. Thus, the role of Atoh1 in tumor formation is likely linked to its function as a master regulator of differentiation.
Loss of ATOH1 Expression in Primary Human CRC and MCC Tumors
Given the remarkable evolutionary conservation of ato/Atoh1 function in cancer development, we decided to analyze whether loss of the human ortholog, ATOH1, might be selected for during malignant transformation in human cancer. To this end, we investigated the role of ATOH1 in CRC and MCC. In vitro overexpression experiments in CRC cell lines suggest that ATOH1 gain of function decreases the population growth potential of CRC cells . Similarly, expression analysis in MCC cell lines suggests an inverse correlation between ATOH1 levels and the population growth of MCC cells (Figure S4A–S4C) . These observations hint at a potential role for ATOH1 in these cancers.
To test this possibility in vivo in primary human tumors, we began by asking whether the expression of ATOH1 is down-regulated in primary tumor samples from 42 CRC and four MCC patients. As MCC is a very rare cancer, only four samples were available for analysis. Seventy percent of the CRC samples show a significant decrease of ATOH1 mRNA expression compared to tissue-matched colon samples from normal controls as analyzed by quantitative reverse-transcriptase PCR (RT-qPCR), suggesting that loss of ATOH1 expression is a highly common feature of CRC oncogenesis (Figures 3F and S4). Furthermore, ATOH1 mRNA levels are significantly lower in adenocarcinoma samples compared to adenomas (t-test: p = 0.017), indicating progressive loss of ATOH1 expression levels with increasing tumor severity (Figure 3A). Similarly, in MCC, the two samples with lower ATOH1 expression levels were derived from the two patients showing metastases (Figure S5).
CRC and MCC Patients Show Deletions in the ATOH1 Locus
One mechanism to explain loss of gene expression during oncogenesis is the deletion of the locus, which we tested for using quantitative PCR on genomic DNA in all 46 patient samples, one MCC cell line (MCC14.2), and one CRC cell line (Ht29). At least one deletion in the locus was detected in 57% of the samples (Figure 3B and 3F). A second primer set yielded similar results (49% deletion rate, Figure 3B). One allele was also deleted in both Ht29 and MCC14.2. Comparative genomic hybridization (CGH) array experiments  on three samples did not show a deviation of the clones flanking the ATOH1 locus (Figures S5 and S6), indicating that the deletions observed in patients are likely to be ATOH1-specific microdeletions. When only one copy is deleted, complete loss of gene function could be achieved by point mutations in the remaining allele. Sequencing of the ATOH1 open reading frame of 24 samples, however, did not reveal any mutations in any of the samples (unpublished data). Together with the loss of ATOH1 mRNA expression in patients, these results suggest that an epigenetic silencing mechanism may be involved—a possibility we tested in further detail.
The ATOH1 Locus Is Methylated in CRC and MCC Patients
To address a putative epigenetic transcriptional silencing mechanism, we began by asking whether transcription could, in principle, be initiated from the ATOH1 locus in cancer cells. Drosophila ato and mouse Atoh1 are both known to be autoregulatory [18,19]. This provides the opportunity to test whether transcription could be activated from the ATOH1 locus by ATOH1 itself in normal versus cancer cells. We took advantage of expressing the mouse ortholog in human cells to distinguish the expression of endogenous ATOH1 from that of Atoh1. For the tests in human cancer cell lines, we used the MCC14.2 cell line, which is derived from MCC and has strongly reduced expression of ATOH1. Surprisingly, Atoh1 overexpression failed to activate endogenous human ATOH1 expression in the MCC14.2 cell line (Figure 3C), despite the virtually complete conservation of the two proteins. This may be due to a key difference between the two proteins, or to a possible disruption of the autoregulatory loop during oncogenesis. To test whether Atoh1 can activate ATOH1 expression in normal noncancerous skin cells, we transduced normal human primary keratinocytes with an Atoh1 expression vector and tested the expression of endogenous ATOH1 40 h after lentiviral transduction. In contrast to the cancer cell lines, endogenous ATOH1 is up-regulated 38-fold by Atoh1 (Figure 3C). This indicates that the expressability of the ATOH1 locus is inhibited during oncogenesis.
Genomic database searches show that all known ATOH1 orthologs, from human to Drosophila, reside in a CpG island covering at least the promoter and the transcription start site, suggesting that CpG methylation may be a mechanism of ATOH1 loss of function. We therefore assayed all patient samples and both cell lines for ATOH1 methylation using three independent assays: pull-down of methylated DNA, methylation-sensitive restriction digest, and methylation-specific PCR for bisulfite DNA modification. We found that up to 81% of the patients, as well as both cell lines, show methylation of the ATOH1 locus in both the coding sequence and the promoter sequences directly upstream of the ATG. In contrast, none of the control samples show ATOH1 methylation (Figures 3D, 3F, and S7A–S7C). This suggests that methylation is likely causal to the transcriptional silencing of the locus during oncogenesis as this is accompanied by lower ATOH1 expression. In a few cases, however, the correlation was not observed. In two cases (samples 27 and 30), this may be due to the presence of a putative duplication, which is yet to be methylated. Alternatively, but not exclusively, it may be due to heterogeneity of the degree of methylation in different cells comprising the tumor samples analyzed. To provide evidence that methylation silences the ATOH1 locus, we inhibited DNA methyltransferases (Dnmts) with 5-azadeoxycytidine in the Ht29 cell line. This results in an approximately 8-fold increase in ATOH1 expression (Figure 3E). Thus, the ATOH1 locus is methylated in cancer patients and derived cell lines, and this methylation can be reversed, resulting in the transcriptional reactivation of the locus. The autoregulatory function of ATOH1 combined with the methylation of its regulatory sequences might hint to the involvement of ATOH1 in the silencing of its own promoter. To gain some insight into whether this may indeed be the case, we asked whether Atoh1 can physically interact with Dnmt proteins. We tested the binding of Atoh1 to Dnmt1 (ENSG00000130816), Dnmt3a (ENSG00000119772), and Dnmt3b (ENSG00000088305) using GST pull-down assays. We find that Atoh1 binds to all three Dnmts. This is further supported by the ability of GST-Atoh1 to pull down DNA methyltransferase activity from nuclear extracts, similar to the positive control (EED: embryonic ectoderm development, ENSG00000074266) . Focusing on Dnmt1, the maintenance DNA methyltransferase, we confirmed this observation using coimmunoprecipitation experiments in 293T cells transfected with HA-tagged Atoh1. Lastly, mapping experiments identified several regions of Dnmt1 as mediating the association with Atoh1 (Figure S8A–S8D).
In summary, deletions and epigenetic silencing via methylation combine to cause loss-of-function mutations of the human ATOH1 locus in primary human cancers. Together with loss-of-function evidence in five independent cancer models in human cells, Drosophila, and mouse, these data strongly support a role for the Ato/ATOH1 transcription factors as key modulators of oncogenic transformation.
ATOH1 Functions via JNK-Dependent Inhibition of the Cell Cycle and Induction of Apoptosis
To better understand the role that ATOH1 plays in cancer, we sought to determine the molecular mechanism by which it acts to suppress the formation and progression of tumors. Gain- and loss-of-function analyses point to Atoh1-dependent regulation of proliferation in mouse colon tumors (Figures 2E and S3D–S3F). Analysis of the role of Drosophila ato in fly retinal tumors  shows that this function is mediated by the JNK signaling pathway. We hypothesized that the Ato-JNK-p21 pathway may mediate the tumorigenic phenotype in our mouse models of CRC as well. If loss of ATOH1 were indeed an initiating event in tumor formation, we reasoned that the effects of Atoh1 function would have to be detectable in preneoplastic tissue as this is where oncogenesis begins. We therefore first examined the expression levels of Atoh1, Cdkn1a (p21, ENSG00000124762), Cdkn1b (p27, ENSG00000111276), and Cdkn1c (p57, ENSG00000129757) and the JNK target cJun (ENSG00000177606) in colon crypts isolated from Atoh1wt and Atoh1Δintestine mice (Figure 4A). We observe a 6.8-fold reduction in Atoh1 expression and an approximately 2.8-fold reduction in expression of all three Cdkn1 isoforms in colon crypts from Atoh1Δintestine mice. We also observe a similar reduction in cJun mRNA levels. We further examined the Ato-JNK-Cdkn1 pathway by immunoblotting proteins from colon polyps and normal-appearing colon tissue from APCmin/+; Atoh1wt and APCmin/+; Atoh1Δintestine mice. Colon polyps show higher levels of cJun and p21waf protein compared to normal colonic tissue. Importantly, however, we find a reduction in cJun levels in APCmin/+; Atoh1Δintestine polyps compared to APCmin/+; Atoh1wt polyps, consistent with the mRNA analysis showing less cJun (Figure 4B). Importantly, we find a reduction in p27 protein and a trend toward less p21 in Atoh1-mutant colon tissues and polyps, consistent with our analysis of mRNA levels (Figure 4B). We also find a specific reduction in pJNK1 levels (Figure 4C) in APCmin/+; Atoh1wt versus APCmin/+; Atoh1Δintestine tissues, in agreement with earlier reports that JNK1 (MAPK8: ENSG00000107643) may have anti-oncogenic activity in the mouse intestine , whereas JNK2 (MAPK9: ENSG00000050748) deficiency enhances tumorigenesis in other epithelia . Finally, we examined the pattern of JNK activation in the colon of Atoh1wt and Atoh1Δintestine mice, and observe identical numbers of pJNK1/2-positive cells in Atoh1-null compared to control crypts (Figure S9). Together with western blot analysis demonstrating selective reduction in pJNK1, but not pJNK2, our data suggest that the level of pJNK1 in individual crypt cells is decreased upon loss of Atoh1. These data are consistent with a preneoplastic function of Atoh1 in regulating proliferation, by a JNK-dependent induction of cell cycle inhibitors.
(A) Gene expression analysis in colon crypts from Atoh1Δintestine and Atoh1wt mice. Quantitative RT-PCR was used to assess gene expression in isolated colon crypts. The ratio of gene expression is shown, in which negative numbers indicate reduced expression in Atoh1Δintestine crypts. t-tests determined p-values shown for each gene.
(B) Western analysis of p27kip, p21waf1, and c-JUN in APCmin; Atoh1wt, and APCmin; Atoh1Δintestine colonic tissues and polyps. Representative colonic tissue and polyp lysates of APCmin; Atoh1wt, and APCmin; Atoh1Δintestine were used for western analysis of p27kip, p21waf1, and c-JUN. Actin was used a loading control. p21waf1 was significantly up-regulated in polyps compared to nonneoplastic colon tissue. c-Jun protein levels were significantly reduced in colonic polyps upon Atoh1 loss. p27 is down-regulated in preneoplastic colonic tissue upon loss of Atoh1. Quantifications are shown in Figure S12A–S12C.
(C) Representative tissues were used for Western analysis of phosphorylated JNK1 and JNK2. Total JNK1 and JNK2 are shown as control. Actin loading control is shown below. pJNK1, but not pJNK2, was significantly reduced in colon tissue from Atoh1Δintestine compared to Atoh1wt mice. Quantifications are shown in Figure S12D1–S12D2′.
Next we asked whether the molecular mechanism of ATOH1 function in human cancer is similar to the mouse. We took advantage of several established MCC cell lines [23,24] with either high ATOH1 expression (MCC1 and MCC6, derived from less aggressive tumors) or low ATOH1 expression (MCC13, MCC14.2, and MCC26, derived from highly aggressive metastatic tumors; Figure S4A and S4C). We find that the growth rate of cell lines, measured as their doubling time, correlates inversely with levels of ATOH1 expression (Figure S4B). To determine whether the reduction in ATOH1 levels might be causal to decreased doubling time, we restored ATOH1 function by creating stable cell lines expressing Atoh1 using lentiviral vectors (Figure 5A and 5B). Stable Atoh1-expressing cell lines (MCC14.2-Atoh1.1a, MCC14.2-Atoh1.1b, MCC14.2-Atoh1.2a, and MCC14.2-Atoh1.2b) have a significantly slower population doubling time compared to control cell lines (MCC14.2 and MCC14.2-GFP; p <0.0001; Figure 5C). To assess whether this increase in doubling time reflects decreased malignancy, we tested these cell lines for growth in soft agar. Cell lines with high levels of Atoh1 expression display a marked decrease in growth in soft agar compared to control cell lines (Figure 5D, p < 0.001). The change in population doubling time could be due to a slower cell cycle, an increased apoptotic rate, or both. Although the distribution of cells in the cell cycle appears unaltered (Figure S10A), the speed of the cell cycle is 25% lower in Atoh1-expressing cells, as assayed by BrdU pulse-chase experiments (Figure 5E, p <0.01).
(A) Constructs used for creating lentiviral and transfection vectors.
(B) Western blot analysis for Atoh1 and chromogranin on untransduced MCC14.2 cells (lane 1), GFP-transduced MCC14.2 cells (lane 2), and two independently derived MCC14.2 cell lines transduced with Atoh1-IRES-eGFP (lane 3: MCC14.2-Atoh1.1a, and lane 4: MCC14.2-Atoh1.2a). Quantifications in Figure S12E and S12F.
(C) Doubling time in hours for MCC14.2 (lane 1), MCC14.2-GFP (lane 2), and four lines independently transduced with Atoh1-IRES-eGFP (lanes 3–6: MCC14.2-Atoh1.1a, MCC14.2-Atoh1.1b, MCC14.2-Atoh1.2a, and MCC14.2-Atoh1.2b, respectively).
(D) Assay for growth in soft agar. Colonies per view with a 10× lens, lane 1: MCC14.2, lane 2: MCC14.2-GFP, lane 3: MCC14.2-Atoh1.1a, and lane 4: MCC14.2-Atoh1.2a. Error bars indicate the 25th and 75th percentiles.
The triple asterisks (***) in (C and D) indicate a significant difference from MCC14.2 (t-test: p < 0.001).
(E) Percentage of cells positively labeled for BrdU and past S-phase in a BrdU pulse-chase experiment; double asterisks (**) indicate p < 0.01 (t-test).
In addition to slower proliferation, we find a specific and strong increase in apoptotic cell death, as measured by annexin-V and cleaved caspase-3 (ENST00000308394), in MCC (MCC14.2 series) and CRC (Ht29) cell lines transduced or transfected with Atoh1 (Figures 6A, 6B, and S10B). This increase in cell death is mediated by the intrinsic apoptosis pathway, as suggested by enhanced caspase-9 (ENST00000333868) cleavage (Figure 6A and 6B).
(A) Western blot analysis for p21waf1, cleaved caspase-3, cleaved caspase-9, and phosphorylated JNK of lysates of MCC14.2 cells, MCC14.2-GFP, and two MCC14.2 cell lines transduced with Atoh1-IRES-eGFP (MCC14.2-Atoh1.1a and MCC14.2-Atoh1.2a). The corresponding actin loading controls are shown under each blot. Quantifications are shown in Figure S12G–S12J.
(B) Western blot analysis for p21waf1, cleaved caspase-3, cleaved caspase-9, and p-JNK of lysates of Ht29 cell line transfected with pCLIG-eGFP (left lane) or pCLIG-Atoh1-IRES-eGFP (right lane); actin loading controls are shown under the respective blots. Quantifications are shown in Figure S12K–S12N.
(C) The molecular changes are specific to functional ato: western blot of lysates of Ht29 cells transfected with CMV-ato, CMV-ato1, or empty vector. Actin loading controls are shown below the respective blots. Quantifications are shown in Figure S12O–S12Q.
(D) Graph expressing ratio of cell numbers of MCC14.2-derived cell lines transfected with pMSCV-ATOH1ERD-IRES-eGFP versus cell number of cell lines transfected with pMSCV-IRES-eGFP (vector). Statistical analysis was done using t-test under different conditions compared to MCC14.2. Single asterisk (*) indicates p < 0.05; double asterisks (**) indicate p < 0.01.
(E) MCC1 cells transfected with dominant-negative ATOH1ERD fusion (lanes 1 and 2) and with empty vector control (lanes 3 and 4). Western blot analysis for p21waf1, phosphorylated JNK, and cleaved caspase-3 on lysates of MCC1 cell line. The actin loading control of each blot is shown below. Quantifications are shown in Figure S12R–S12T.
Data from mouse models indicate an involvement of JNK-mediated regulation of cell proliferation for the tumor suppressor effect of Atoh1. To assess whether the same mechanisms are operating in human cancer, we tested the expression levels of the caspase-3 and caspase-9, p21waf1, and p-JNK in MCC and CRC cells lines. We note a clear and specific up-regulation of p21waf1 and p-JNK levels upon Atoh1 expression (Figures 6A, 6B, and S10C–S10H). Similarly, transfection of wild-type Drosophila Ato results in the up-regulation of cleaved caspase-3, p21waf1, and p-JNK (Figure 6C), in contrast to a null-mutant form of the protein that completely fails to bind DNA , indicating the conservation and specificity of the Ato/Atoh1 effect.
Next, we tested the expression of these proteins in a loss-of-function setting. We generated and expressed the transcriptional repressor form of ATOH1 by fusing it to the engrailed repressor domain (ATOH1ERD) , which specifically inhibits the Atoh1-induced effects on MCC cell lines (Figure 6D). In the MCC1 cell line, which shows higher endogenous ATOH1 expression compared to MCC14.2 (Figure S4C), expression of ATOH1ERD leads to a down-regulation of p21waf1 and p-JNK as well as inhibition of caspase-3 cleavage (Figure 6E).
In summary, gain- and loss-of-function studies in mouse colon cancer models and human cancer cells support a conserved antitumor function for ATOH1 mediated by JNK and p21.
ATOH1 Activates RTK Expression in Human Cancer Cells
How does ATOH1 expression lead to JNK activation in cancer cells? ATOH1 and its orthologs are known to exert their developmental functions by modulating Notch; Atonal orthologs inhibit Notch signaling to induce differentiation during development, whereas Notch inhibits ato expression by its target gene Hes1 (ENSG00000114315) [27–29]. Notch signaling has also been described to be upstream of JNK . However, three lines of evidence suggest that the Atoh1-mediated effects are not Notch-dependent. First, Atoh1 expression levels do not influence expression of HES1, a Notch signaling target gene (Figure S11A). Second, we do not detect cleaved intracellular Notch in the MCC14.2 cell line, which expresses low levels of ATOH1 (Figure S11B). Finally, blocking of Notch activation by selective inhibition of γ-secretase  has no effect on the growth of MCC cells (Figure S11C). Other known upstream activators of JNK are RTKs . As Atoh1 is a transcription factor, we checked mRNA expression of all 90 human RTKs upon Atoh1 expression in both the MCC14.2 and the Ht29 cell line compared to control cells. We observed a significant and specific Atoh1-dependent up-regulation of Neurotrophic tyrosine kinase receptor type 1 (NTRK1: ENSG00000198400), a hallmark for differentiation in Merkel cells [33,34] in MCC14.2, and of FGF receptors in Ht29 (Figure S11D). The increase in NTRK1 expression in MCC14.2 seen on the mRNA level was confirmed using RT-qPCR (Figure S10E). We also observed elevated NTRK1 levels in the endogenously ATOH1-expressing MCC1 cells (Figure S10E). This was accompanied by higher protein levels of both the NTRK1 receptor and one of its ligands, Neurotrophin-3 (NT3: ENSG00000185652; Figure S10E–S10H), also a marker for Merkel cell differentiation. Therefore, Ato gain of function results in tumor type–specific elevation of RTK levels.
ATOH1 Functions by RTK-Mediated JNK Activation
To test whether the RTKs may be functionally linked with ATOH1, we incubated the various MCC cell lines with K252a, a narrow-specificity RTK inhibitor (NTRK, FGFR, PDGFR, and IGFR) [35–37]. This results in a dose-dependent decrease in doubling time of the MCC14.2-derived, Atoh1-expressing cell lines, indicating that the Atoh1-induced change in doubling time is RTK-dependent (Figure 7A). In addition, the induction of apoptosis by Atoh1 in the MCC14.2 and the Ht29 cell line is blocked by RTK inhibition (Figure 7B and 7C). These effects are accompanied by RTK inhibitor–dependent decrease in the expression of p21waf1 and p-JNK (Figure 7D), suggesting that both p21waf1 and p-JNK are downstream effectors of Atoh1-induced RTK signaling. Next, we asked whether p-JNK is required for p21waf1 up-regulation and apoptosis by treating MCC cells with SAPK Inhibitor II, a specific JNK inhibitor . We observed a decrease in the Atoh1-induced expression of p21waf1 and cleaved caspase-3 (Figure 7E). Similarly, transfection of dominant-negative c-jun (TAM67)  leads to a significant increase in cell number compared to mock-transfected cells (t-test: p = 0.04), specifically in the MCC14.2-Atoh1.2a, which has high Atoh1 expression levels (Figure 7F).
(A–D) show that ATOH1′s anti-oncogenic function acts through receptor tyrosine kinases. (A) Doubling times in hours of MCC14.2 and MCC14.2 transduced with eGFP (lane 2) or Atoh1-IRES-eGFP (lanes 3 and 4) with increasing concentrations of K252a. (B) AnnexinV-positive signals normalized to transfected MCC14.2 cells (GFP). Cells transduced with eGFP (lanes 1 and 2) and Atoh1-IRES-eGFP (lanes 3 and 4) with 0.33 μM K252a (lanes 2 and 4) or DMSO as a control (lanes 1 and 3). The double asterisks (**) indicate significant difference from GFP transfected without K252a (t-test: p < 0.05). (C) AnnexinV-positive signals normalized to transfected Ht29 cells (GFP). Cells transfected with eGFP (lanes 1 and 2) and Atoh1-IRES-eGFP (lanes 3 and 4) with 0.33 μM K252a (lanes 2 and 4) or DMSO as control (lanes 1 and 3). The double asterisks (**) indicate significant difference from GFP transfected without K252a (t-test: p < 0.05). (D) Inhibition of RTKs inhibits p21waf1 expression and JNK phosphorylation. Western blot analysis for p21waf1 and pJNK of MCC14.2 (lanes 1 and 5), MCC14.2 transduced with eGFP (lanes 2 and 6), and two MCC14.2-derived cell lines transduced with Atoh1-IRES-eGFP (lanes 3, 4, 7, and 8) with 0.3 μM K252a (lanes 1–4) and with DMSO as a control (lanes 5–8). Actin loading controls are shown below each blot. Quantifications are shown in Figure S12U and S12V.
(E) JNK regulates apoptosis and p21waf1 expression. Western blot analysis for p21waf1 and cleaved caspase-3 of MCC14.2 (lanes 1 and 5), MCC14.2 transduced with eGFP (lanes 2 and 6), and two MCC14.2-derived cell lines transduced with Atoh1-IRES-eGFP (lanes 3, 4, 7, and 8) with 1 μM JNK inhibitor (lanes 1–4) and with DMSO as a control (lanes 5–8). Actin loading control is shown below the blots. Phospho-JNK (P-JNK) blot shows the effect of the JNK inhibitor on JNK phosphorylation status Quantifications are shown in Figure S12W–S12Y.
(F) Relative cell numbers (the number of Tam67-transfected cells divided by the number of mock-transfected cells) for MCC14.2 and MCC14.2 transduced with eGFP (lane 2) or Atoh1-IRES-eGFP (lanes 3 and 4) with dominant-negative c-jun (TAM67).
In summary, taken together, evidence from mutation analysis in human patients, as well as gain- and loss-of-function analysis in mouse and human cells, support a model (Figure 8) in which ATOH1 modulates JNK activity, possibly via co-option of context-specific RTK signaling, to induce apoptosis and up-regulate p21waf1 expression, keeping tumor growth in check. Loss-of-function mutations in ATOH1 prevent JNK-mediated apoptosis and p21-mediated cell cycle arrest, leading to enhanced tumor progression.
In preneoplastic tissue, Atoh1 keeps malignant transformation in check in a JNK-dependent mechanism by the induction of apoptosis and the inhibition of cell cycle progression. When Atoh1 is lost due to deletion or methylation, these brakes on oncogenesis fail, and malignant transformation can progress.
Our data support an evolutionarily conserved tumor suppressor role for ATOH1 in CRC and MCC. Loss of ATOH1 promotes tumor formation and progression, and mutations in the ATOH1 locus are found with relatively high frequency. Given the high deletion and methylation rate of ATOH1 in human tumor samples, loss of ATOH1 function is likely to be an early event in these tumors. We therefore propose that ATOH1 acts as a key switch regulating the transformation of pre-oncogenic epithelia to neoplastic and metastatic tumors.
Genetic analysis of the function of Drosophila ato in fly eye tumor suggests that its anti-oncogenic function is linked to its activity as a regulator of cell fate commitment and differentiation . This is similar to what we observe in the different mouse models, where the adenomas and adenocarcinomas do not contain secretory cells (Figure 2) . Interestingly, this also appears to be the case in human cancer, where the majority of human CRCs do not contain differentiated secretory cells. The loss of ATOH1 in most human CRC patients likely explains this observation. In addition, cell type–specific RTK differentiation genes are up-regulated upon overexpression of Atoh1 in CRC and MCC cell lines. Importantly, these markers of differentiation are necessary for the anti-oncogenic effect of ato/ATOH1.
It is tempting to speculate that in other tissues, similar loss of differentiation factors is involved in oncogenesis. In this sense, the loss of differentiation factors has already been implicated in late stages of tumor progression as with GATA-3 (ENSG00000107485) in breast cancer . Although loss of GATA-3 in early tumor leads to an inhibition of tumor formation, loss of GATA-3 in later stages leads to the acquisition of metastatic potential. We, therefore, wonder whether loss of other differentiation factors, perhaps bHLH proteins, might play a role in earlier stages of breast cancer development.
Tumor suppressor genes are defined by the fact that (1) loss-of-function mutations make the cells more prone to malignant transformation, (2) overexpression leads to inhibition of the malignant phenotype, and (3) spontaneous somatic mutations are found in patients with cancer. “Classical tumor suppressor genes,” such as p53 (ENSG00000141510), are special in the sense that when mutated, the cell is more prone to accumulate additional mutations, and thus actively drive malignant progression as opposed to just “taking away the brakes” . It is notable that ato/Atoh1 shows most of the hallmarks of a tumor suppressor gene. Because silencing of ato/Atoh1 is not sufficient to drive oncogenesis, we suggest that ato/Atoh1, and similar genes, are important brakes on malignant transformation. Therefore, the role that differentiation factors might play as key switches in malignant transformation in different tissues is not different from the classical definition of tumor suppressor in a functionally relevant sense.
The RTK and JNK signaling pathways, which are essential for Atoh1's tumor suppressor activity, have been suggested as context-dependent oncogenes or tumor suppressors . Our data indicate that Atoh1 is important in deciding this context by the up-regulation of cell type–specific RTKs. The activation and co-option of RTK and JNK signaling by Ato/ATOH1 in this context suggests that the status of differentiation of the tumor-initiating cell may be the key determinant of the specific role of various signaling pathways in cancer. This may have important clinical implications: treatment of CRC or MCC patients with RTK or JNK inhibitors might have an adverse effect on tumors where ATOH1 is still expressed.
In this medical context, our data suggest that screening for ATOH1 expression, deletion, and methylation may be a useful diagnostic tool for early detection and treatment decision of MCC and CRC. Similarly, treatment of CRC and MCC patients whose tumors show epigenetic silencing of ATOH1 with DNA methyltransferase inhibitors might prove a powerful avenue for therapy, because it appears to be sufficient to restore ATOH1 expression and induce cancer cell death. Furthermore, such treatment in combination with Notch inhibitors may enhance re-expression of the ATOH1-driven differentiation program [43,44] and synergistically inhibit cancer growth. Therefore, elucidation of the basic mechanisms of ato/ATOH1 function, as well as their target genes and interacting proteins, might offer potential avenues for future therapeutic intervention.
Materials and Methods
Cloning of ATOH1ERD expression construct.
The ATOH1 open reading frame was PCR amplified and fused in frame with the engrailed repressor domain after XhoI restriction digest. This fusion product was then blunt ligated in the MSCV-IRES-GFP vector (Clontech), giving rise to the MSCV-ATOH1ERD-IRES-GFP construct.
Mouse models and treatments.
Atoh1Δintestine mice are a conditional deletion of Atoh1 and are described in more detail elsewhere [3,45]. These mice were generated using the loxP/cre system in which Cre-mediated deletion of Atoh1 is mosaic and is restricted to the distal ileum and large intestine (80%–90% deletion). APCmin mice were purchased from The Jackson Laboratory and mated with Atoh1Δintestine mice to generate APCmin; Atoh1Δintestine mice. Eight-week-old male Atoh1wt and Atoh1Δintestine littermates were injected intraperitoneally with AOM (Midwest Research Institute of the National Cancer Institute's chemical carcinogen repository) at 10 mg/kg body weight. AOM was injected weekly for a total of six injections, and the mice were sacrificed 20 wk after the first AOM injection. Two hours prior to sacrifice, mice were injected with 50 mg/kg BrdU. The large intestine was isolated and flushed with PBS and then fixed in 10% buffered formalin at room temperature for 16–24 h. Colons were placed in 70% ethanol before macroscopic analysis of polyps, and then embedded into paraffin blocks. Similarly, 16–19-wk-old APCmin and APCmin; Atoh1Δintestine colons were isolated and fixed as for the AOM-treated colons. For macroscopic analysis, the position and diameter (millimeters) of each polyp were recorded using a dissecting microscope. The protocol for use of animals was approved by the Cincinnati Children's Hospital Institutional Animal Care and Use Committee.
Immunohistochemistry and microscopic analysis.
The colons were imbedded in paraffin blocks and sectioned for hematoxylin and eosin staining by the University of Cincinnati pathology core facility. Tumor phenotypes were determined as in Boivin et al. (2003) . Paraffin-embedded colons of AOM-treated mice were sectioned at 5 μm and used for BrdU and cleaved caspase-3 staining. Mouse anti-BrdU antibodies were obtained from the Developmental Studies Hybridoma Bank maintained by the Department of Biological Sciences of the University of Iowa. The sections were de-paraffinized, rehydrated, and antigen retrieval was performed in citric acid buffer (pH = 6) using a microwave. Endogenous peroxidase activity was blocked with hydrogen peroxide/methanol solution, and Avidin/Biotin block was performed according to the manufacturer's recommendations (Vector Laboratories). Endogenous immunoglobulins block and primary and secondary antibody incubations were performed using the M.O.M. kit following the manufacturer's recommendations (Vector Laboratories). Anti-BrdU antibody was incubated for 16 h at 4 °C in a humidified chamber. Color was developed using the DAB peroxidase substrate kit (Vector Laboratories) followed with hematoxylin staining and dehydration of the tissues. Nuclei from well-oriented crypts in Atoh1wt and Atoh1Δintestine colons were counted at 40× magnification, followed by counts of BrdU-positive cells. The percent of BrdU-positive Atoh1wt or Atoh1-null cells was determined for each animal (at least 1,000 cells per genotype were counted for each animal). Student two-tailed t-test was performed to measure significance. Cleaved caspase-3 staining and counting were performed similarly to BrdU staining and analysis. Polyclonal rabbit anti–cleaved caspase-3 antibodies (1:100) were obtained from Cell Signaling Technology. The Rabbit IgG VECTASTAIN ABC kit was used according to the manufacturer's recommendations for blocking, antibody incubations, and signal amplification (Vector Laboratories).
Crypts from wild-type and Atoh1Δintestine mice were isolated using a modification of the Evans method [47,48]. Mice were sacrificed and the colons removed and flushed with ice-cold PBS, opened flat, and then placed in cold PBS containing protease and phosphatase inhibitors. After a brief vortexing, the colon was cut into four pieces and placed into a 15-ml tube containing shaking solution (1.5 mM KCL, 96 mM NaCl, 27 mM Na Citrate, 8 mM KH2PO4, 5.6 mM Na2HPO4, 15 mM EDTA, and 1 mM dithiothreitol supplemented with protease and phosphatase inhibitors). The tubes were agitated at 4 °C on a vortexer holding the tubes at 180° until the solution appeared cloudy (5–8 min). The colon pieces were transferred into new tubes with shaking buffer and agitated until most of the crypts were released (enrichment was assessed by phase contrast microscopy at different time points). The solution containing the crypts was filtered through a 100-μm cell strainer to isolate the crypts, and sorbitol (2% final concentration) was slowly added and mixed immediately. The solution was centrifuged at 160g for 8 min at 4 °C. The supernatant (containing single cells) was removed, and the pellet (containing purified crypts) was snap frozen in liquid nitrogen, to be used for protein and RNA analyses.
Quantitative reverse-transcriptase polymerase chain reaction.
RNA was purified from colon crypts isolated from wild-type and Atoh1Δintestine mice using Trizol (Invitrogen) according to the manufacturer's recommendations. Trizol-purified RNA (100 μg) was subjected to DNase digestion and further purification (RNeasy Mini; Qiagen); 2 μg of total RNA was reverse transcribed (Superscript III, Invitrogen), and cDNA equivalent to 100 ng of RNA used for SYBR Green–based real-time PCR using an Mx3005 (Stratagene). For each gene assessed, colon crypt RNA from nine wild-type and eight Atoh1Δintestine mice was compared using the standard curve method of relative quantification. All gene expression levels were normalized to the expression of GAPDH. t-Tests measured significant differences between the average normalized expression levels in wild-type versus Atoh1Δintestine mice.
Cecal and colonic tissues from wild-type and Atoh1Δintestine mice were homogenized in lysis buffer (1× Phosphate Buffered Saline, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 0.7 mM EDTA supplemented with protease and phosphatase inhibitors), sonicated, and the lysates used for quantitation and western blots. Similarly, lysates of colonic tissues and polyps from APCmin, and APCmin; Atoh1Δintestine double-mutant mice were used for whole lysate preparation and western blots. The following antibodies were used for western blot analysis: p21waf1 mouse monoclonal antibody (1:500, cat #556431; BD Pharmingen), actin mouse IgM antibodies (1:100, JLA20; Developmental Studies Hybridoma Bank), mouse monoclonal p27kip1 antibody (1:250, cat #610241; BD Transduction Laboratories), goat anti-p57 polyclonal antibodies (1:100, cat # sc-1039; Santa Cruz Biotechnology), rabbit polyclonal c-Jun antibodies (1:1000, cat # sc-1694; Santa Cruz Biotechnology), and pJNK antibody (1:1000, cat #559309; EMD-Calbiochem).
Formalin-fixed tissues from AOM-treated mice were used for phosphorylated-JNK (pJNK) immunohistochemistry. Rabbit polyclonal pJNK antibody was used (1:100, cat #559309; EMD-Calbiochem). The sections were de-paraffinized, hydrated, and the antigen retrieval was performed in citric acid buffer using a microwave. Endogenous peroxidase activity was blocked with hydrogen peroxide/methanol solution, a short remobilization step was included (0.2% Triton X 100 in PBS), and Avidin/Biotin block was performed according to the manufacturer's recommendations (Vector Laboratories). The Rabbit IgG VECTASTAIN ABC kit was used according to the manufacturer's recommendations for blocking and antibody incubations (Vector Laboratories). Primary antibody was incubated at 4 °C for 14–16 h; color was developed using the DAB peroxidase substrate kit (Vector Laboratories) followed by hematoxylin staining and coverslipping. Cells of well-oriented crypts in wild-type and Atoh1Δintestine colons were counted at 40× magnifications, followed by counts of pJNK-positive cells. The total number of cells and pJNK-positive cell numbers were added for each animal, and the average numbers were compared across crypts and animals. A Student t-test was performed to measure significance.
MCC cell lines were cultured in RPMI medium supplemented with 15% FCS (Perbio). The Ht29 cell lines (obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen [DSMZ]) were cultured in McCoy medium supplemented with 10% FCS. Primary human keratinocytes were isolated and pooled from foreskins of three different donors (less than 6 y). Fourth passage cells were used in the experiments. The procedure has been approved by the ethical committee of the University of Leuven. Experiments performed adhered to the Declaration of Helsinki Principles. Keratinocytes were seeded in serum-free and growth factor–containing medium (Keratinocyte-SFM; Invitrogen), which contains several growth factors (5 μg/ml insulin, 74 ng/ml hydrocortisone, 6.7 ng/ml triiodo-l-thyronine, 50 μg/ml bovine pituitary extract, and 5 ng/ml human recombinant EGF). All culture experiments were done at 37 °C in 5% CO2.
A fixed amount of cells was seeded in standard culture conditions, and the number of cells was counted after 3 to 5 d. The doubling times (T2X) were calculated using the formula T2X = LN(2)/((LN(nΔt)-LN(n0))/Δt) with Δt, time in culture, n0, number of seeded cells, and nΔt the number of cells after Δt
The lentiviral vector HIV-CMV-Atoh1-IRES-GFP was constructed by cloning a PCR fragment of Atoh1-IRES into the HIV-CMV-GFP vector , which was used as control vector. The HIV-CMV-GFP vector was first restricted with XbaI and Age1, and ligated to the Atoh1-IRES fragment, which was spanned between Xba1 and Age1. The Atoh1 and GFP were expressed from a bicistronic vector under the control of the CMV promoter. These lentiviral vectors were produced as described in [50,51]. Stable cell lines were created by transducing the cells in normal medium, supplemented with 8 nM polybrene and a multiplicity of infection between 100 and 150). Four stable cell lines were made with the lentiviral vectors expressing Atoh1 and eGFP from a bicistronic Atoh1-IRES-eGFP construct (MCC14.2-Atoh1.1a, MCC14.2-Atoh1.1b, MCC14.2-Atoh1.2a, and MCC14.2-Atoh1.2b), and one with eGFP alone (MCC14.2-GFP) as a negative control, under the control of a CMV promoter (Figure 5A). We failed to created stable cell lines from the Ht29. Cells were heavily selected against, and FACS isolation did not yield surviving cells.
Colony formation in soft agar.
A total of 2,500 cells/ml were resuspended in 0.6% agarose (Invitrogen) in culture medium. A 2-ml layer of 0.35% agarose in culture medium was added on top of the 2-ml 0.6% layer. After 1 wk, 2 ml of 0.35% agarose in culture medium was added. Cells were cultured for 2 wk in standard conditions, and the number of colonies was analyzed under an inverted microscope with a 10× magnification. The experiment was done in triplicate.
Cell cycle distribution and progression.
Cells were grown up to 40% confluency under standard conditions and were trypsinized. Cells were fixed and permeabilized using 70% ice-cold ethanol for 2 h. Cells were washed in PBS, and cells were stained for 30 min using 0.1% (v/v) Triton X-100 (Sigma) in PBS and 0.2 mg/ml DNase-free Rnase A (Sigma) and 20 μg/ml PI (Sigma). Cells were analyzed on a FACSCalibur cytometer (BectonDickinson). To analyze the cell cycle progression, cells were pulsed for 20 min with 10 μM BrdU (Sigma). Medium was then substituted with normal medium for 6 h. Cells were collected and fixed in 70% ice-cold ethanol. DNA was denatured using 2 M HCl for 20 min. Cells were stained using anti-BrdU and IgG-alexa555 and analyzed in a FACSCalibur (BD Bioscience).
Annexin-V staining was performed using Annexin-V-biotin (Roche) and streptavidin-PercP (BD biosciences) as described by the manufacturer. Cells were imaged with the Leica TCS Sp2 confocal microscope, and images were analyzed and quantified using LCS software.
Lysates were obtained by scraping cells in medium and centrifuging the medium. The pellet was resuspended in lysis buffer containing PBS with 1 mM EDTA, 1 mM EGTA, 50 mM NaF, 1% TritonX, 5 mM Na3VO4, 20 μM PAO, and Complete Protease Inhibitor. The crude extract was separated by SDS-PAGE in a 4%–12% NuPAGE novex bis-tris gel (Invitrogen) and electroblotted onto Hybond-ECL membrane (Amersham). Antibodies were diluted in the appropriate concentrations in 5% BSA in TBS-tween20. The antibodies used are beta-actin (clone AC-15; Sigma), anti–cleaved caspase-3 (Asp-175; Cell Signaling Technology), polyclonal antibody to caspase-9(active) (ALEXIS), chromogranin A+B (Abcam), cyclinA1 (Santa Cruz Biotechnology), pJNK (Thr183/Tyr185; BioSource), p27 (BD Bioscience), p21waf1 (DCS60; Cell Signaling Technology), PCNA (clone PC10; Sigma), TRK (Santa Cruz Biotechnology), Atoh1 (1/50; Developmental Studies Hybridoma Bank), NT3 (Santa Cruz Biotechnology), and Cleaved Notch (Val1744; Cell Signaling Technology). All western blot analyses are quantified in Figure S11.
Tyrosine kinase inhibitor, SAPK inhibitor II, and γ-secretase inhibitor.
K252a diluted in DMSO, from VWR, and DMSO as control were used. Final concentrations ranged from 0 μM to 0.5 μM; when no concentration is mentioned, 0.33 μM was used. SAPK inhibitor II, from Calbiochem, was used at a concentration of 10 μM. We used inhibitor X as a γ-secretase inhibitor (Calbiochem). Concentrations used are mentioned in the figure legend.
RT-PCR and qPCR.
mRNA was amplified using the Superscript II One-Step RT-PCR system with Platinum Taq-100 reactions (Invitrogen) and the ATOH1 primers CAGCCAGTGCAGGAGGAAAA and GAAAATTCCCCGTCGCTTCT, and the Hes1 primers GGACATTCTGGAAATGACAGTGAA and AGCGCAGCCGTCATCTG, according to the manufacturer's protocol. Quantitative RT-PCR was performed on a ABI prism 7000 (Applied Biosystems). The primers were designed using Primer Express (Applied Biosystems). The primers used are ATOH1: CAGCCAGTGCAGGAGGAAAA and GAAAATTCCCCGTCGCTTCT; Atoh1: GCTGTGCAAGCTGAAGGG and TCTTGTCGTTGTTGAAGG. Primers to check copy number of the ATOH1 locus: ATOH1 locus set 1: CCCCGGGAGCAATCTTG and GGGACCGAGGCGAAGTT; control locus set 1: TCTGGGACCTGAGCTAATGGA and GGCCATAATTAGGACCATGAAAGA; and ATOH1 locus set 2: GCCAGTGCAGGAGGAAAACA and GAAAATTCCCCGTCGCTTCT. Control locus set 2: GGGTTCAGCCTCAACTTGTATCC and CCCACCACCTGGCATCTCT.
The 90 kinase RT-PCR assay.
RNA was diluted to 300 μg/μl and treated with TURBO DNA-free (Ambion) following the manufacturer's protocol. cDNA was synthesized using random primers and SuperscriptII reverse transcription (Invitrogen) using the manufacturer's protocol. cDNA was amplified using gene-specific primers (concentration 5 μM) for the different human tyrosine kinases.
Methylation of the DNA was detected using ApaI enzyme. DNA (2 μg) was dissolved in 50 μl of the appropriate buffer and 10 U of ApaI, and incubated at 30 °C overnight. A mirror condition was done in which the ApaI was exchanged by glycerol as a control for unspecific degradation. The resulting DNA was analyzed using two primer sets: one set spanning the restriction site (AATAAGACGTTGCAGAAGAG and TCGCAGAGCAAAAATTAAAGGGTGC) and another set next to the restriction site (CCCCGGGAGCATCTTGCAGCCA and TCGCAGAGCAAAAATTAAAGGGTGC).
Pull-down of methylated DNA fragments (restricted using EcoRI) was performed using the Methylcollector kit (Active Motif), and bisulfite modification of DNA was done using the EZ DNA Methylation-Gold kit (Zymo Research) according to the manufacturers' protocols.
We carried out array CGH using Code Linked Slides (AP Biotech) containing the 3,527 BAC clones from the Wellcome Trust Sanger Institute 1 Mb Clone Set, a gift from N. P. Carter (The Wellcome Trust Sanger Institute). Array CGH was performed as described .
The ATOH1 open reading frame was amplified with PCR using AATAAGACGTTGCAGAAGAG and TCGCAGAGCAAAAATTAAAGGGTGC and AmpliTaq Gold DNA polymerase and the GeneAmp PCR System 2400 (Applied Biosystems). The PCR products were purified and sequenced in both directions on the ABI Prism BigDye (Terminator Cycle Sequencing Kit version 1.1) on an ABI PRISM 3100 Genetic Analyser (Applied Biosystems).
Figure S1. Gut-Associated Lymphoid Tissues in AOM-Treated Atoh1wt Mice
The colons of AOM-treated Atoh1wt mice showed large GALT that were macroscopically counted as polyps; shown is 5× magnification of representative colonic GALTs. GALT indicated by an arrow.
(726 KB PDF)
Figure S2. Histology of Polyps in APCmin Background
(A) Polyps in the APCmin background still have goblet cells, indicating that Atoh1 is still active.
(B) The polyps in the APCmin; Atoh1Δintestine mice originate in Atoh1 mutant tissue as seen by the absence of goblet cells.
(4.70 MB PDF)
Figure S3. Representative Normal-Appearing Crypts in AOM-Treated Atoh1wt and Atoh1Δintestine Colons
AOM-treated colon sections with well-oriented crypts were used for BrdU and cleaved caspase-3 (c-Caspase 3) counting. The genotypes of the representative slides are indicated on the left side of the figure. The specific stain is identified at the top part of the figure. Atoh1wt (WT) crypts were distinguished from Atoh1-null crypts by the lack of the secretory goblet cells in the null crypts.
(A–C) Hematoxylin and eosin (H & E) staining of Atoh1wt crypts in Atoh1wt mice (A); and nondeleted Atoh1wt (B) and Atoh1-null (C) in Atoh1Δintestine mice.
(D–F) Representative BrdU staining of normal-appearing crypts in Atoh1wt mice (D) and nondeleted Atoh1wt (E) and Atoh1-null (F) in Atoh1Δintestine mice.
(G–I) Representative cleaved caspase-3 staining of normal-appearing crypts in Atoh1wt mice (G); and nondeleted Atoh1wt (H) and Atoh1-null (I) in Atoh1Δintestine mice. Arrows point to positive cells at the surface of the crypts. The images were captured at 20× magnification.
(4.96 MB PDF)
Figure S4. ATOH1 Expression Correlates with Population Doubling Time
(A) ATOH1 mRNA transcripts in five independent MCC cell lines. The name of the cell line is indicated above each lane. RT-PCR was done with 100 ng of RNA under nonsaturating conditions.
(B) Doubling times in hours of the five MCC cell lines. Error bars indicate the standard deviation.
(C) RT-qPCR for ATOH1 in MCC1 and MCC14.2 cell lines. ATOH1 mRNA levels standardized to GADPH mRNA levels.
(760 KB PDF)
Figure S5. Summary Table of All Patient Data
First column indicates patient numbers. Next, the relative ratios for ATOH1 genomic DNA (gDNA) over control locus is presented for each of the two primer sets as analyzed by qPCR. The next two columns indicate the classification of the deletion/duplication status of the ATOH1 locus. The mRNA expression is shown relative to control colon samples, with next to it, the classification of the expression. Clinical data are given in the third panel, namely cancer stage and metastasis. In the last panel, methylation of the ATOH1 locus is shown using three different methods.
(760 KB PDF)
Figure S6. Array CGH Analysis of Three Patient Samples, Demonstrating No Aberration of the ATOH1 Locus
Each bar represents the log2 of the value for affected individuals (ind) versus a control sample (reference) for each probe, ordered based on the probes' chromosomal location. The region between 131 Mb and 141 Mb is shaded. The location of the ATOH1 locus is shown with an arrow. The abnormalities were confirmed by dye swap experiments.
(4.55 MB PDF)
Figure S7. Methylation of the ATOH1 Locus
(A) Schematic representation of ATOH1 locus. The white box indicate the ATOH1 ORF, and the gray box the position of the CpG island. The primers are indicated as arrows.
(B) Detection of methylation at the ATOH1 locus using the ApaI methylation-sensitive restriction enzyme. PCR fragments generated using primers spanning the ApaI restriction site from ApaI restricted genomic DNA; presence of a band indicates methylation of the ATOH1 CpG island.
(C) Detection of methylation using methylation-sensitive PCR: upon bisulfite modification, presence of a band indicates methylation of the ATOH1 CpG island.
(159 KB PDF)
Figure S8. Atoh1 Binds the Dnmt Proteins and Associates to a DNA Methyltransferase Activity
(A) GST Pull-down assays using Atoh1 protein fused to GST (GST-Atoh1) and in vitro translated Dnmts (IVT-Dnmt1, IVT-Dnmt3a, or IVT-Dnmt3b).
(B) A GST-fused Atoh1 protein was used to purify DNA methyltransferase activity from nuclear extracts. After incubation, the beads were washed and assayed for DNA methyltransferase activity read as c.p.m. of S-adenosyl-l[methyl-3H] methionine incorporated into an oligonucleotide substrate. GST-tagged embryonic ectoderm development protein (GST-EED) was used as a positive control.
(C) Coimmunoprecipitation experiments shows that Atoh1 associated with Dnmt1. The 293T cells were transiently transfected in culture dishes (10-cm diameter) with 3 μg of HA-Atoh1 plasmid
(D) Mapping of Atoh1 binding to Dnmt1. GST Pull-downs were performed with Dnmt1 fragments fused to GST and in vitro–translated Atoh1. The upper part is a schematic representation of the Dnmt1 sequences used.
(1.04 MB PDF)
Figure S9. Immunohistochemical Analysis of Phosphorylated JNK in Atoh1Δintestine and Atoh1wt Mice
(A) Representative pJNK1/2 staining in Atoh1wt crypt (dashed white line). The arrows indicate pJNK-positive cells.
(B) pJNK-positive cells (arrows) in wild-type (dashed gray line) and Atoh1-null (dashed black line) crypts in Atoh1Δintestine mice.
(C) Bar graph showing the percentage of pJNK-positive cells in wild-type mice (white); and wild-type (gray) and Atoh1-null crypts in Atoh1Δintestine mice. Error bars indicate the standard error of the mean. No significant differences between genotypes were detected.
(4.18 MB PDF)
Figure S10. ATOH1 Interacts with the Cell Cycle and Apoptosis
(A) Cell cycle distribution of MCC14.2-derived cell lines without (MCC14.2 and MCC14.2-GFP) and with Atoh1 expression (MCC14.2-Atoh1.1a and MCC14.2-Atoh1.2a). No significant change in distribution throughout the cell cycle can be observed.
(B) Maximal projection image of AnnexinV staining (red) on cells transduced with lentiviral vectors expressing GFP (left panel) or Atoh1-IRES-GFP (right panel). GFP is in green.
(C) Western blot analysis for CyclinA1, PCNA, p27kip, c-myc, phospho-H3 and p21waf1 of lysates of MCC14.2 cells, MCC14.2-GFP and two MCC14.2 cell lines transduced with Atoh1-IRES-eGFP (MCC14.2-Atoh1.1a and MCC14.2-Atoh1.2a). The corresponding actin loading controls are shown under each blot.
(D–H) Quantification of expression levels of cyclinA1 (D), PCNA (E), p27 (F), c-myc (G), and phospho-histoneH3. (H) Representative blots are shown in (C).
(2.90 MB PDF)
Figure S11. ATOH1 Acts Independently of Notch but Modulates RTK Expression
(A) RT-PCR for target of Notch signaling HES1 on mRNA isolated from MCC14.2 cells, MCC14.2 cells transduced with GFP, and two MCC14.2 cell lines transduced with Atoh1-IRES-eGFP (MCC14.2-Atoh1.1a and MCC14.2-Atoh1.2a). GADPH loading control is shown below.
(B) Western blot analysis for cleaved intracellular Notch (NICD). Different concentrations of presinilin inhibitor X were used; a plus sign (+) indicates a positive control for cleaved NICD.
(C) Different concentrations of γ-secretase inhibitor (inhibitor X) on MCC14.2 do not influence the proliferation rate. First lane: 10 μM inhibitor X, second lane: DMSO control of previous lane, third lane: 1 μM inhibitor X, fourth lane: DMSO control of previous lane, and fifth lane: untreated.
(D) RT-PCR for expression of 90 tyrosine kinases scored from undetectable (white) over orange (expression) to red (strong expression) on mRNA or on untransduced MCC14.2 cells (lane 1), GFP transduced MCC14.2 cell line (lane 2), and two independent MCC14.2-derived cell lines transduced with Atoh1-IRES-eGFP vectors (lane 3: MCC14.2-Atoh1.2a) and on untransfected HT29 cells (lane14: HT29), GFP-transfected HT29 cells (lane5: HT29-GFP), and HT29 cells transfected with the Atoh1-IRES-eGFP construct (lane 6: HT29-Atoh1).
(E) RT-qPCR for ATOH1 mRNA levels (compared to GADPH mRNA levels) in MCC cell lines. MCC1 (first lane: MCC cell line with endogenous high ATOH1 expression), MCC14.2 (lane 2), GFP-transduced MCC14.2 cell line (lane 3), and two independently made MCC14.2-derived cell lines transduced with Atoh1-IRES-GFP vectors (lane 4: MCC14.2-Atoh1.1a, and lane 5: MCC14.2-Atoh1.2a).
(F) Western blot analysis for NTRK1 and Neurotrophin-3 (NT3) of untransduced MCC14.2 cells (lane 1), GFP-transduced MCC14.2 cell line (lane 2), and two independently made MCC14.2-derived cell lines transduced with Atoh1-IRES-GFP vectors (lane 3: MCC14.2-Atoh1.1a, and lane 4: MCC14.2-Atoh1.2a); actin loading controls are represented below each blot.
(G and H) Quantifications of western blots in (F), normalized to actin.
(2.04 MB PDF)
Figure S12. Quantifications of Western Blot Analysis in Main Figures (Minimum of Two Blots per Quantification)
The quantifications are performed with USI software. The signal for the protein of interest was standardized to its respective actin loading control. Antibody is stated above each graph, representative experiments are shown in main figures: (A) Figure 4B, (B) Figure 4B, (C) Figure 4B, (D1) Figure 4C, (D1′) Figure 4C, (D2) Figure 4C, (D2′) Figure 4C, (E) Figure 5B , (F) Figure 5B, (G) Figure 6A, (H) Figure 6A, (I) Figure 6A, (J) Figure 6A, (K) Figure 6B, (L) Figure 6B, (M) Figure 6B, (N) Figure 6B, (O) Figure 6C, (P) Figure 6C, (Q) Figure 6C, (R) Figure 6E, (S) Figure 6E, (T) Figure 6E, (U) Figure 7D, (V) Figure 7D, (W) Figure 7E, (X) Figure 7E, and (Y) Figure 7E.
(365 KB PDF)
We thank H. Van Miegroet and N. Mentens for performing and scoring the 90 tyrosine kinase experiment; Chelsea Combs and Jefferson Vallance for technical assistance; S. Claerhout for providing primary keratinocytes; and S. Reeve, S. Claerhout, P. Vanderhaeghen, P. Verstreken, J. C. Marine, and P. Carmeliet for critical reading of the manuscript.
WB, AK, FF, NFS, and BAH conceived and designed the experiments. WB, AK, NDG, SVK, and JL performed the experiments. WB, AK, JL, FF, JC, NFS, and BAH analyzed the data. GDH, KG, GPB, MC, TV, PM, and JC contributed reagents/materials/analysis tools. WB, AK, JC, NFS, and BAH wrote the paper.
- 1. Ben-Arie N, Hassan BA, Bermingham NA, Malicki DM, Armstrong D, et al. (2000) Functional conservation of atonal and Math1 in the CNS and PNS. Development 127: 1039–1048.N. Ben-ArieBA HassanNA BerminghamDM MalickiD. Armstrong2000Functional conservation of atonal and Math1 in the CNS and PNS.Development12710391048
- 2. Yang Q, Bermingham NA, Finegold MJ, Zoghbi HY (2001) Requirement of Math1 for secretory cell lineage commitment in the mouse intestine. Science 294: 2155–2158.Q. YangNA BerminghamMJ FinegoldHY Zoghbi2001Requirement of Math1 for secretory cell lineage commitment in the mouse intestine.Science29421552158
- 3. Shroyer NF, Helmrath MA, Wang VY, Antalffy B, Henning SJ, et al. (2007) Intestine-specific ablation of mouse atonal homolog 1 (Math1) reveals a role in cellular homeostasis. Gastroenterology 132: 2478–2488.NF ShroyerMA HelmrathVY WangB. AntalffySJ Henning2007Intestine-specific ablation of mouse atonal homolog 1 (Math1) reveals a role in cellular homeostasis.Gastroenterology13224782488
- 4. Ben-Arie N, Bellen HJ, Armstrong DL, McCall AE, Gordadze PR, et al. (1997) Math1 is essential for genesis of cerebellar granule neurons. Nature 390: 169–172.N. Ben-ArieHJ BellenDL ArmstrongAE McCallPR Gordadze1997Math1 is essential for genesis of cerebellar granule neurons.Nature390169172
- 5. Bermingham NA, Hassan BA, Price SD, Vollrath MA, Ben-Arie N, et al. (1999) Math1: an essential gene for the generation of inner ear hair cells. Science 284: 1837–1841.NA BerminghamBA HassanSD PriceMA VollrathN. Ben-Arie1999Math1: an essential gene for the generation of inner ear hair cells.Science28418371841
- 6. Wang VY, Hassan BA, Bellen HJ, Zoghbi HY (2002) Drosophila atonal fully rescues the phenotype of Math1 null mice: new functions evolve in new cellular contexts. Curr Biol 12: 1611–1616.VY WangBA HassanHJ BellenHY Zoghbi2002Drosophila atonal fully rescues the phenotype of Math1 null mice: new functions evolve in new cellular contexts.Curr Biol1216111616
- 7. Bossuyt W, De Geest N, Aerts S, Leenaerts I, Marynen P, et al. (2009) The Atonal proneural transcription factor links differentiation and tumor formation in Drosophila. PLoS Biol 7: e1000040. W. BossuytN. De GeestS. AertsI. LeenaertsP. Marynen2009The Atonal proneural transcription factor links differentiation and tumor formation in Drosophila.PLoS Biol7e1000040
- 8. Brenner B, Sulkes A, Rakowsky E, Feinmesser M, Yukelson A, et al. (2001) Second neoplasms in patients with Merkel cell carcinoma. Cancer 91: 1358–1362.B. BrennerA. SulkesE. RakowskyM. FeinmesserA. Yukelson2001Second neoplasms in patients with Merkel cell carcinoma.Cancer9113581362
- 9. O'Connell JB, Maggard MA, Ko CY (2004) Colon cancer survival rates with the new American Joint Committee on Cancer sixth edition staging. J Natl Cancer Inst 96: 1420–1425.JB O'ConnellMA MaggardCY Ko2004Colon cancer survival rates with the new American Joint Committee on Cancer sixth edition staging.J Natl Cancer Inst9614201425
- 10. Leow CC, Romero MS, Ross S, Polakis P, Gao WQ (2004) Hath1, down-regulated in colon adenocarcinomas, inhibits proliferation and tumorigenesis of colon cancer cells. Cancer Res 64: 6050–6057.CC LeowMS RomeroS. RossP. PolakisWQ Gao2004Hath1, down-regulated in colon adenocarcinomas, inhibits proliferation and tumorigenesis of colon cancer cells.Cancer Res6460506057
- 11. Radtke F, Clevers H (2005) Self-renewal and cancer of the gut: two sides of a coin. Science 307: 1904–1909.F. RadtkeH. Clevers2005Self-renewal and cancer of the gut: two sides of a coin.Science30719041909
- 12. Su LK, Kinzler KW, Vogelstein B, Preisinger AC, Moser AR, et al. (1992) Multiple intestinal neoplasia caused by a mutation in the murine homolog of the APC gene. Science 256: 668–670.LK SuKW KinzlerB. VogelsteinAC PreisingerAR Moser1992Multiple intestinal neoplasia caused by a mutation in the murine homolog of the APC gene.Science256668670
- 13. Moser AR, Pitot HC, Dove WF (1990) A dominant mutation that predisposes to multiple intestinal neoplasia in the mouse. Science 247: 322–324.AR MoserHC PitotWF Dove1990A dominant mutation that predisposes to multiple intestinal neoplasia in the mouse.Science247322324
- 14. Colnot S, Niwa-Kawakita M, Hamard G, Godard C, Le Plenier S, et al. (2004) Colorectal cancers in a new mouse model of familial adenomatous polyposis: influence of genetic and environmental modifiers. Lab Invest 84: 1619–1630.S. ColnotM. Niwa-KawakitaG. HamardC. GodardS. Le Plenier2004Colorectal cancers in a new mouse model of familial adenomatous polyposis: influence of genetic and environmental modifiers.Lab Invest8416191630
- 15. Niwa N, Hiromi Y, Okabe M (2004) A conserved developmental program for sensory organ formation in Drosophila melanogaster. Nat Genet 36: 293–297.N. NiwaY. HiromiM. Okabe2004A conserved developmental program for sensory organ formation in Drosophila melanogaster.Nat Genet36293297
- 16. Leonard JH, Cook AL, Van Gele M, Boyle GM, Inglis KJ, et al. (2002) Proneural and proneuroendocrine transcription factor expression in cutaneous mechanoreceptor (Merkel) cells and Merkel cell carcinoma. Int J Cancer 101: 103–110.JH LeonardAL CookM. Van GeleGM BoyleKJ Inglis2002Proneural and proneuroendocrine transcription factor expression in cutaneous mechanoreceptor (Merkel) cells and Merkel cell carcinoma.Int J Cancer101103110
- 17. Fiegler H, Carr P, Douglas EJ, Burford DC, Hunt S, et al. (2003) DNA microarrays for comparative genomic hybridization based on DOP-PCR amplification of BAC and PAC clones. Genes Chromosomes Cancer 36: 361–374.H. FieglerP. CarrEJ DouglasDC BurfordS. Hunt2003DNA microarrays for comparative genomic hybridization based on DOP-PCR amplification of BAC and PAC clones.Genes Chromosomes Cancer36361374
- 18. Sun Y, Jan LY, Jan YN (1998) Transcriptional regulation of atonal during development of the Drosophila peripheral nervous system. Development 125: 3731–3740.Y. SunLY JanYN Jan1998Transcriptional regulation of atonal during development of the Drosophila peripheral nervous system.Development12537313740
- 19. Helms AW, Abney AL, Ben-Arie N, Zoghbi HY, Johnson JE (2000) Autoregulation and multiple enhancers control Math1 expression in the developing nervous system. Development 127: 1185–1196.AW HelmsAL AbneyN. Ben-ArieHY ZoghbiJE Johnson2000Autoregulation and multiple enhancers control Math1 expression in the developing nervous system.Development12711851196
- 20. Viré E, Brenner C, Deplus R, Blanchon L, Fraga M, et al. (2006) The Polycomb group protein EZH2 directly controls DNA methylation. Nature 439: 871–874.E. ViréC. BrennerR. DeplusL. BlanchonM. Fraga2006The Polycomb group protein EZH2 directly controls DNA methylation.Nature439871874
- 21. Tong C, Yin Z, Song Z, Dockendorff A, Huang C, et al. (2007) c-Jun NH2-terminal kinase 1 plays a critical role in intestinal homeostasis and tumor suppression. Am J Pathol 171: 297–303.C. TongZ. YinZ. SongA. DockendorffC. Huang2007c-Jun NH2-terminal kinase 1 plays a critical role in intestinal homeostasis and tumor suppression.Am J Pathol171297303
- 22. Chen N, Nomura M, She QB, Ma WY, Bode AM, et al. (2001) Suppression of skin tumorigenesis in c-Jun NH(2)-terminal kinase-2-deficient mice. Cancer Res 61: 3908–3912.N. ChenM. NomuraQB SheWY MaAM Bode2001Suppression of skin tumorigenesis in c-Jun NH(2)-terminal kinase-2-deficient mice.Cancer Res6139083912
- 23. Leonard JH, Bell JR, Kearsley JH (1993) Characterization of cell lines established from Merkel-cell ("small-cell") carcinoma of the skin. Int J Cancer 55: 803–810.JH LeonardJR BellJH Kearsley1993Characterization of cell lines established from Merkel-cell ("small-cell") carcinoma of the skin.Int J Cancer55803810
- 24. Leonard JH, Dash P, Holland P, Kearsley JH, Bell JR (1995) Characterisation of four Merkel cell carcinoma adherent cell lines. Int J Cancer 60: 100–107.JH LeonardP. DashP. HollandJH KearsleyJR Bell1995Characterisation of four Merkel cell carcinoma adherent cell lines.Int J Cancer60100107
- 25. Jarman AP, Grell EH, Ackerman L, Jan LY, Jan YN (1994) Atonal is the proneural gene for Drosophila photoreceptors. Nature 369: 398–400.AP JarmanEH GrellL. AckermanLY JanYN Jan1994Atonal is the proneural gene for Drosophila photoreceptors.Nature369398400
- 26. John A, Smith ST, Jaynes JB (1995) Inserting the Ftz homeodomain into engrailed creates a dominant transcriptional repressor that specifically turns off Ftz target genes in vivo. Development 121: 1801–1813.A. JohnST SmithJB Jaynes1995Inserting the Ftz homeodomain into engrailed creates a dominant transcriptional repressor that specifically turns off Ftz target genes in vivo.Development12118011813
- 27. Ligoxygakis P, Yu SY, Delidakis C, Baker NE (1998) A subset of notch functions during Drosophila eye development require Su(H) and the E(spl) gene complex. Development 125: 2893–2900.P. LigoxygakisSY YuC. DelidakisNE Baker1998A subset of notch functions during Drosophila eye development require Su(H) and the E(spl) gene complex.Development12528932900
- 28. Hassan BA, Bellen HJ (2000) Doing the MATH: is the mouse a good model for fly development? Genes Dev 14: 1852–1865.BA HassanHJ Bellen2000Doing the MATH: is the mouse a good model for fly development?Genes Dev1418521865
- 29. Li Y, Baker NE (2001) Proneural enhancement by Notch overcomes Suppressor-of-Hairless repressor function in the developing Drosophila eye. Curr Biol 11: 330–338.Y. LiNE Baker2001Proneural enhancement by Notch overcomes Suppressor-of-Hairless repressor function in the developing Drosophila eye.Curr Biol11330338
- 30. Kim JW, Kim MJ, Kim KJ, Yun HJ, Chae JS, et al. (2005) Notch interferes with the scaffold function of JNK-interacting protein 1 to inhibit the JNK signaling pathway. Proc Natl Acad Sci U S A 102: 14308–14313.JW KimMJ KimKJ KimHJ YunJS Chae2005Notch interferes with the scaffold function of JNK-interacting protein 1 to inhibit the JNK signaling pathway.Proc Natl Acad Sci U S A1021430814313
- 31. Shearman MS, Beher D, Clarke EE, Lewis HD, Harrison T, et al. (2000) L-685,458, an aspartyl protease transition state mimic, is a potent inhibitor of amyloid beta-protein precursor gamma-secretase activity. Biochemistry 39: 8698–8704.MS ShearmanD. BeherEE ClarkeHD LewisT. Harrison2000L-685,458, an aspartyl protease transition state mimic, is a potent inhibitor of amyloid beta-protein precursor gamma-secretase activity.Biochemistry3986988704
- 32. Schlessinger J (2000) Cell signaling by receptor tyrosine kinases. Cell 103: 211–225.J. Schlessinger2000Cell signaling by receptor tyrosine kinases.Cell103211225
- 33. Fundin BT, Silos-Santiago I, Ernfors P, Fagan AM, Aldskogius H, et al. (1997) Differential dependency of cutaneous mechanoreceptors on neurotrophins, trk receptors, and P75 LNGFR. Dev Biol 190: 94–116.BT FundinI. Silos-SantiagoP. ErnforsAM FaganH. Aldskogius1997Differential dependency of cutaneous mechanoreceptors on neurotrophins, trk receptors, and P75 LNGFR.Dev Biol19094116
- 34. Szeder V, Grim M, Kucera J, Sieber-Blum M (2003) Neurotrophin-3 signaling in mammalian Merkel cell development. Dev Dyn 228: 623–629.V. SzederM. GrimJ. KuceraM. Sieber-Blum2003Neurotrophin-3 signaling in mammalian Merkel cell development.Dev Dyn228623629
- 35. Tapley P, Lamballe F, Barbacid M (1992) K252a is a selective inhibitor of the tyrosine protein kinase activity of the trk family of oncogenes and neurotrophin receptors. Oncogene 7: 371–381.P. TapleyF. LamballeM. Barbacid1992K252a is a selective inhibitor of the tyrosine protein kinase activity of the trk family of oncogenes and neurotrophin receptors.Oncogene7371381
- 36. Nye SH, Squinto SP, Glass DJ, Stitt TN, Hantzopoulos P, et al. (1992) K-252a and staurosporine selectively block autophosphorylation of neurotrophin receptors and neurotrophin-mediated responses. Mol Biol Cell 3: 677–686.SH NyeSP SquintoDJ GlassTN StittP. Hantzopoulos1992K-252a and staurosporine selectively block autophosphorylation of neurotrophin receptors and neurotrophin-mediated responses.Mol Biol Cell3677686
- 37. Ruggeri BA, Miknyoczki SJ, Singh J, Hudkins RL (1999) Role of neurotrophin-trk interactions in oncology: the anti-tumor efficacy of potent and selective trk tyrosine kinase inhibitors in pre-clinical tumor models. Curr Med Chem 6: 845–857.BA RuggeriSJ MiknyoczkiJ. SinghRL Hudkins1999Role of neurotrophin-trk interactions in oncology: the anti-tumor efficacy of potent and selective trk tyrosine kinase inhibitors in pre-clinical tumor models.Curr Med Chem6845857
- 38. Shin M, Yan C, Boyd D (2002) An inhibitor of c-jun aminoterminal kinase (SP600125) represses c-Jun activation, DNA-binding and PMA-inducible 92-kDa type IV collagenase expression. Biochim Biophys Acta 1589: 311–316.M. ShinC. YanD. Boyd2002An inhibitor of c-jun aminoterminal kinase (SP600125) represses c-Jun activation, DNA-binding and PMA-inducible 92-kDa type IV collagenase expression.Biochim Biophys Acta1589311316
- 39. Alani R, Brown P, Binétruy B, Dosaka H, Rosenberg RK, et al. (1991) The transactivating domain of the c-Jun proto-oncoprotein is required for cotransformation of rat embryo cells. Mol Cell Biol 11: 6286–6295.R. AlaniP. BrownB. BinétruyH. DosakaRK Rosenberg1991The transactivating domain of the c-Jun proto-oncoprotein is required for cotransformation of rat embryo cells.Mol Cell Biol1162866295
- 40. Kouros-Mehr H, Bechis SK, Slorach EM, Littlepage LE, Egeblad M, et al. (2008) GATA-3 links tumor differentiation and dissemination in a luminal breast cancer model. Cancer Cell 13: 141–152.H. Kouros-MehrSK BechisEM SlorachLE LittlepageM. Egeblad2008GATA-3 links tumor differentiation and dissemination in a luminal breast cancer model.Cancer Cell13141152
- 41. Harris H (2004) Tumor suppression: putting on the brakes. Nature 427: 201.H. Harris2004Tumor suppression: putting on the brakes.Nature427201
- 42. Heasley LE, Han SY (2006) JNK regulation of oncogenesis. Mol Cells 21: 167–173.LE HeasleySY Han2006JNK regulation of oncogenesis.Mol Cells21167173
- 43. Searfoss GH, Jordan WH, Calligaro DO, Galbreath EJ, Schirtzinger LM, et al. (2003) Adipsin, a biomarker of gastrointestinal toxicity mediated by a functional gamma-secretase inhibitor. J Biol Chem 278: 46107–46116.GH SearfossWH JordanDO CalligaroEJ GalbreathLM Schirtzinger2003Adipsin, a biomarker of gastrointestinal toxicity mediated by a functional gamma-secretase inhibitor.J Biol Chem2784610746116
- 44. van Es JH, van Gijn ME, Riccio O, van den Born M, Vooijs M, et al. (2005) Notch/gamma-secretase inhibition turns proliferative cells in intestinal crypts and adenomas into goblet cells. Nature 435: 959–963.JH van EsME van GijnO. RiccioM. van den BornM. Vooijs2005Notch/gamma-secretase inhibition turns proliferative cells in intestinal crypts and adenomas into goblet cells.Nature435959963
- 45. Kiesslich R, Goetz M, Angus EM, Hu Q, Guan Y, et al. (2007) Identification of epithelial gaps in human small and large intestine by confocal endomicroscopy. Gastroenterology 133: 1769–1778.R. KiesslichM. GoetzEM AngusQ. HuY. Guan2007Identification of epithelial gaps in human small and large intestine by confocal endomicroscopy.Gastroenterology13317691778
- 46. Boivin GP, Washington K, Yang K, Ward JM, Pretlow TP, et al. (2003) Pathology of mouse models of intestinal cancer: consensus report and recommendations. Gastroenterology 124: 762–777.GP BoivinK. WashingtonK. YangJM WardTP Pretlow2003Pathology of mouse models of intestinal cancer: consensus report and recommendations.Gastroenterology124762777
- 47. Flint N, Cove FL, Evans GS (1991) A low-temperature method for the isolation of small-intestinal epithelium along the crypt-villus axis. Biochem J 280(Pt 2): 331–334.N. FlintFL CoveGS Evans1991A low-temperature method for the isolation of small-intestinal epithelium along the crypt-villus axis.Biochem J280Pt 2331334
- 48. Taylor JA, Martin CA, Nair R, Guo J, Erwin CR, et al. (2008) Lessons learned: optimization of a murine small bowel resection model. J Pediatr Surg 43: 1018–1024.JA TaylorCA MartinR. NairJ. GuoCR Erwin2008Lessons learned: optimization of a murine small bowel resection model.J Pediatr Surg4310181024
- 49. Van Damme A, Thorrez L, Ma L, Vandenburgh H, Eyckmans J, et al. (2006) Efficient lentiviral transduction and improved engraftment of human bone marrow mesenchymal cells. Stem Cells 24: 896–907.A. Van DammeL. ThorrezL. MaH. VandenburghJ. Eyckmans2006Efficient lentiviral transduction and improved engraftment of human bone marrow mesenchymal cells.Stem Cells24896907
- 50. VandenDriessche T, Naldini L, Collen D, Chuah MK (2002) Oncoretroviral and lentiviral vector-mediated gene therapy. Methods Enzymol 346: 573–589.T. VandenDriesscheL. NaldiniD. CollenMK Chuah2002Oncoretroviral and lentiviral vector-mediated gene therapy.Methods Enzymol346573589
- 51. VandenDriessche T, Thorrez L, Naldini L, Follenzi A, Moons L, et al. (2002) Lentiviral vectors containing the human immunodeficiency virus type-1 central polypurine tract can efficiently transduce nondividing hepatocytes and antigen-presenting cells in vivo. Blood 100: 813–822.T. VandenDriesscheL. ThorrezL. NaldiniA. FollenziL. Moons2002Lentiviral vectors containing the human immunodeficiency virus type-1 central polypurine tract can efficiently transduce nondividing hepatocytes and antigen-presenting cells in vivo.Blood100813822