Physiol. Genomics Fuel your research with LabChart
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Physiol. Genomics 13: 57-68, 2003. First published January 14, 2003; doi:10.1152/physiolgenomics.00152.2002
1094-8341/03 $5.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
13/1/57    most recent
00152.2002v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (19)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Fleet, J. C.
Right arrow Articles by Edenberg, H. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Fleet, J. C.
Right arrow Articles by Edenberg, H. J.
Received 6 November 2002; accepted in final form 14 January 2003.
Physiological Genomics 13:57-68 (2003)
1094-8341/03 $5.00 © 2003 American Physiological Society

Gene expression profiling of Caco-2 BBe cells suggests a role for specific signaling pathways during intestinal differentiation

James C. Fleet1, Liyong Wang1, Olga Vitek2, Bruce A. Craig2 and Howard J. Edenberg3

1 Interdepartmental Nutrition Program, Purdue University, West Lafayette 47907
2 Department of Statistics, Purdue University, West Lafayette 47907
3 Center for Medical Genomics, Indiana University School of Medicine, Indianapolis, Indiana 46202


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 EXPERIMENTAL DESIGN AND METHODS
 RESULTS
 DISCUSSION
 References
 
We examined the pattern of gene expression resulting from spontaneous differentiation of Caco-2 BBe cells to gain insight into the molecular changes necessary for enterocyte differentiation. RNA was prepared from cells harvested at three cell stages: proliferating (50% confluent, 2 days in culture), postproliferative nondifferentiated (8 days), and differentiated (15 days). Gene expression profiles were determined using Affymetrix Human Genome U95A GeneChips. Differentially expressed genes were identified following statistical analysis (i.e., ANOVA, bootstrapping adjustments to P values, false detection rate criterion). We identified 1,150 unique genes as differentially expressed; expression of 48.6% fell and 46% increased from 2 to 15 days, while 5.4% had expression that either peaked or dipped at 8 days. Genes expressed during differentiation included several small-intestine-specific genes involved in nutrient transport/metabolism, e.g., DCT1, hephaestin, folate receptor 1, sucrase-isomaltase, and apolipoproteins CI, CIII, B100, H, and M, indicating that this colonic adenocarcinoma cell line has a hybrid colonocyte/enterocyte phenotype. Patterns of gene expression based upon functional classification suggest a role for cell-cell/cell-matrix interactions, suppression of Wnt signaling, and activation of TGFß and phosphatidylinositol 3-kinase pathways during enterocyte differentiation.

microarray; Wnt; transforming growth factor ß; phosphatidylinositol 3-kinase


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
EXPERIMENTAL DESIGN AND METHODS
 RESULTS
 DISCUSSION
 References
 
THE INTESTINE IS A LARGE, complex organ with multiple functions distributed segmentally across the proximal to distal axis. Active absorption of calcium occurs in the duodenum, absorption of vitamin B12 occurs in the ileum, and fermentation of fiber as well as absorption of water occurs in the large bowel. Disruption of these essential processes occur during infections, as a result of inflammatory bowel syndromes (e.g., Crohn’s Disease, inflammatory bowel disease), and following removal of intestinal segments (59). Thus understanding the processes that lead to the development of functionally distinct segments is a critical goal of gastroenterological and nutrition research.

Within each segment of the intestine is the crypt-villus axis. In the crypt, stem cells in the lower portion proliferate and migrate up toward the villus and downward to the nadir of the crypt. As the cells migrate, they receive signals that stimulate them to commit and differentiate into one of the several intestinal cell phenotypes, e.g., paneth cells, enteroendocrine cells, goblet cells, M cells, caveolated tuft cells, and absorptive epithelial cells (9). The mechanism by which crypt stem cells of the mature intestine commit and differentiate into each of these cell lineages is not fully clear.

A number of cells systems have been developed to study the differentiation of proliferating crypt-like cells into functioning enterocytes (12; 45). The colonic adenocarcinoma cell line Caco-2 is unique among these models in that upon contact inhibition of proliferation, cultures of the parental Caco-2 cell line, as well as a number of clonal lines, spontaneously differentiate and acquire the phenotype of a mature absorptive epithelial cell, i.e., cellular polarization, development of tight junctions and a well-developed brush-border membrane, and expression of a number of brush-border hydrolases (e.g., sucrase-isomaltase, lactase) that are characteristic of the small intestine (12, 49). Although investigators have examined the activity and role of specific signaling pathways across the crypt-villus axis as well as the transcription factors controlling the villus-specific expression of specific genes, we do not have a complete picture of the molecular changes that lead to absorptive epithelial cell differentiation.

DNA microarrays offer researchers a unique opportunity to conduct a broad survey of the changes in gene expression that accompany a biological process such as absorptive epithelial cell differentiation. Microarrays have been used to analyze whole intestine gene expression in mice (2), postconfluent differentiation of the parental Caco-2 cell line (37), and other processes (35, 39). However, a systematic examination with rigorous statistical analysis of the expression data has not yet been reported in this field. Here we examine the changes in gene expression that occur as the BBe subclone of Caco-2 progresses from proliferation to a differentiated enterocyte with features of mature, small-intestinal epithelial cells. Our results demonstrate the value of a detailed statistical approach to evaluating microarray data and they provide us insight into the pathways that may be critical to the cell during this process.


   EXPERIMENTAL DESIGN AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL DESIGN AND METHODS
RESULTS
 DISCUSSION
 References
 
Cell Culture and Experimental Protocol
The BBe clone of Caco-2 (CRL-2102) was obtained from American Type Culture Collection (Rockville, MD). These cells were previously described by Peterson and Mooseker (48) and were cultured as previously described (20). For the experiment, cells were seeded into six-well dishes at a density of 250,000 cells per well and fed with DMEM containing 10% FBS every 2 days until day 10 and then daily until the end of the experiment. Cells were harvested into TriReagent at 2 days (50% confluent, proliferating), 8 days (4 days postconfluent; postproliferative, nondifferentiated), and 15 days (differentiated) in culture, and total cellular RNA was isolated according to the manufacturer’s directions (Molecular Research Center, Cincinnati, OH). The RNA was examined by visual inspection on a 1.5% agarose gel to confirm that there was no degradation. RNA samples were then sent to the Center for Medical Genomics at the Indiana University School of Medicine for analysis.

Microarray Analysis
RNA quality was further tested using a bioanalyzer (Agilent Technologies, Palo Alto, CA) and by measuring absorbance from 200 to 350 nm. Four separate RNA preparations, each from an independent dish of cells, were analyzed for each time point. Starting from 10 µg of total RNA for each sample, cDNA was synthesized and biotinylated cRNA was generated by in vitro transcription, following the standard Affymetrix protocols (http://www.affymetrix.com/support/technical/manual/expression_manual.affx). Biotinylated cRNAs were fragmented, and each sample was hybridized to an Affymetrix U95A GeneChip (12,363 sequences) at 42°C for 17 h, then washed, stained, and scanned following the standard Affymetrix protocol. Expression data were generated using the Affymetrix Microarray Suite software version 4.0.

Statistical Evaluation of the Microarray Data
A rigorous statistical analysis is essential to determine which genes are differentially expressed relative to the noise inherent in microarray analysis (42). For this work, the following steps were taken.

First, the number of genes evaluated was reduced by including only those genes called "present" by the Affymetrix software for at least three of four replicates in at least one stage of differentiation. A second reduction was based on the findings of Mills and Gordon (40), who showed that false positives are more frequent in genes at the lowest level of expression. As a result, we calculated the first quartile of expression level (i.e., 590 arbitrary units) and excluded genes whose mean expression at all three stages of culture was less than this value.

Next, we utilized a linear mixed model (i.e., random effect ANOVA or split plot design; Ref. 50) to explain known sources of variation and test the hypothesis that there is a change in expression of a gene between any two stages with respect to the average expression of all genes at the same stage. Since this ANOVA model assumes constant error variance, the "log average ratio" was chosen as the measure of gene expression that most closely approximates this assumption. In addition, a bootstrapping scheme was applied to minimize the potential consequences of nonnormal data (19). Overall, this approach includes data standardization to account for systematic noise due to the experiment and within the technology, accounts for correlation between the expression measures (i.e., correlation of expression measures within a GeneChip), and allows for an inference procedure that determinates differentially expressed genes.

Finally, three specific pairwise comparisons were made between the stages of culture for each gene: 2 days vs. 8 days, 8 days vs. 15 days, and 2 days vs. 15 days. Given the large number of comparisons (3 comparisons x 4,612 genes after filtering = 13,836 comparisons), the hypothesis tests were adjusted to balance the type I (false positive) and type II (false negative) error rates. In principle, two major alternatives are available: procedures controlling the familywise error rate detection (e.g., Bonferroni correction; {alpha} divided by the number of comparisons) and procedures controlling the false discovery rate (5). Conclusions presented here are based on the false discovery rate criterion due to its better control of the type II errors.

After identifying the differentially expressed genes, the genes were clustered by self-organizing maps (58) using GeneCluster version 2.0 from the Whitehead Institute/MIT Center for Genome Research (http://www-genome.wi.mit.edu/cancer/software/software.html). We used a 4 x 3 matrix and the default settings of the software (50 epochs, 1 seed, random vectors method of initialization, bubble neighborhood definition, {alpha}i =0.1, {sigma}i =5, {alpha}f = 0.005, {sigma}f = 0.2).

Cross-Validation of Microarray Data
Three methods were used to provide validation for our microarray results. First, we examined the internal consistency of our data using the genes on our list of differentially expressed genes that were represented at least twice on the Affymetrix U95A GeneChip. We examined whether the duplicates for these genes were placed into identical or similar clusters. In addition, the average 8-day and 15-day expression values were expressed relative to the average 2-day value (as maximum/minimum value) to define a fold change, and we plotted the fold changes at 8 and 15 days from replicate 1 against those from replicate 2; a linear regression was calculated. Second, we determined whether the experimental approach identified genes whose expression had previously been reported to change during normal small intestinal differentiation. Finally, we validated the differential expression of seven genes by reverse transcription PCR (RT-PCR). The RNA samples that had been analyzed by microarray were pooled by stage (5 µg of RNA from each of the four replicates within a stage). cDNA samples were prepared from the pooled RNA samples, and PCR was conducted on the cDNA samples as we have described previously (20). PCR primers were designed in Jellyfish (LabVelocity; http://jellyfish.labvelocity.com/), and eight gene targets were examined: glyceraldehyde phosphate dehydrogenase (GAPDH = control gene) (GenBank ID no. X02231), forward primer 5' CCATGGAGAAGGCTGGGG 3', reverse primer 5' CAAAGTTGTCATGGATGACC 3', annealing temperature (Ta) = 55°C, 17 cycles; sucrase-isomaltase (no. X63597), forward primer 5' GGTGGTCACATCCTACCATGTCAAG 3', reverse primer 5' CCAGTTGATTTCTATTGGTTCTTCT 3', Ta = 55°C, 25 cycles; TGFßII receptor (no. D50683), forward primer 5' CCTACTCTGTCTGTGGATGACCT 3', reverse primer 5' GATCTCTCAACACGTTGTCCTTC 3', Ta = 61°C, 29 cycles; CSF receptor I (no. M33210), forward primer 5' TACCCCAAGAAGGATGTGAGAG 3', reverse primer 5' GGTAACGTATTGAGAACCCACTC 3', Ta = 61°C, 31 cycles; aryl hydrocarbon receptor (no. L19872), forward primer 5' CCACAACATTCCAAATGTACAGA 3', reverse primer 5' AGTGGCTGAAGATGTGTGGTAGT 3', Ta = 61°C, 31 cycles; PDGF receptor (no. J03278), forward primer 5' GATGAGGAGTTTCTGAGGAGTGA 3', reverse primer 5' GTTGAGGAGGTGTTGACTTCATT 3', Ta = 61°C, 31 cycles; c-fos (no. V01512), forward primer 5' GGTGCATTACAGAGAGGAGAAAC 3', reverse primer 5' CCTGGCTCAACATGCTACTAACT 3', Ta = 61°C, 31 cycles; and jagged (no. U77914), forward primer 5' ATGACTGTAATACCTGCCAGTGC 3', reverse primer 5' TCCGTAGTAAGACCTGGTGACAT 3', Ta = 61°C, 31 cycles.

After the PCR reaction, the PCR products were separated on 3% agarose gels, then stained with ethidium bromide, and a digital image was recorded under UV light using the FluorS Imaging system (Bio-Rad).

Gene Classification and Functional Group Analysis
Genes were categorized into one of 26 different functional categories based upon bioprocess and molecular functions described in the annotation tables available for each target at the NetAffx analysis center (http://www.affymetrix.com/analysis/index.affx). These categories were further refined by the target of the bioprocess or molecular action (e.g., catabolism, amino acid; proliferation, induction; signaling, MAPK pathway). The clusters and groups of related clusters were examined for the functional categories present. These categories, as well as the Affymetrix annotation tables, were used to identify genes that are associated with specific signaling pathways. Genes involved in specific signaling pathways were listed and examined for coordinate changes in regulation.

Access to the Data
These data have been submitted to the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL DESIGN AND METHODS
 RESULTS
DISCUSSION
 References
 
Differential Gene Expression Across Stages of BBe Cell Differentiation
Of the 12,363 targets on the array, 5,546 were expressed in at least one of the cell stages (44.9%), and 4,612 of these had an expression value of >590 in at least one of the stages (37.3% of the genes on the array). Table 1 summarizes the number of genes that meet various criteria for defining differential expression. Using the false discovery rate criterion to balance type I and type II error rates, we determined that P < 0.007 was the significance cutoff for defining differential gene expression in our three comparisons (2 days vs. 8 days; 8 days vs. 15 days; 2 days vs. 15 days). We found that 1,238 targets were differentially expressed across the three stages of culture based upon this P value. Multiple appearance of 74 genes accounted for 162 of these entries. Thus 1,150 unique genes were identified as differentially expressed during BBe cell differentiation.


View this table:
[in this window]
[in a new window]
  		Table 1. Summary of comparisons between stages of BBe cell differentiation

 
Validation of Microarray Results
Seventy-four genes were found to appear more than once in our list of 1,238 differentially expressed genes (60 genes in duplicate, 12 genes in triplicate, and 2 genes in quadruplicate). Forty-six of the duplicated genes ended up in the same cluster, 27 of the duplicated genes were placed in similar clusters, and only 1 of the duplicated genes reversed the direction of its change in gene expression (histone H2A.F/Z variant, from 2 days = 756, 8 days = 730, 15 days = 1,043, to 2 days = 9,841, 8 days = 6,123, 15 days = 4,844). Figure 1 demonstrates the correlation between replicates of 71 of the duplicated gene targets. The correlation between these replicates was strong [replicate 2 = 1.22(replicate 1) - 0.3, r2 = 0.84]. Three genes were excluded that did not fit this relationship: histone H2A.F/Z variant, {alpha}2-thiol proteinase inhibitor, and cytochrome P(1)-450 family I. The later two entries had very low expression values at 2 days for one of the replicates that resulted in a very high expression ratio in one replicate but a less dramatic fold change in the other. However, the replicates of these two genes ended up in identical clusters.



View larger version (15K):
[in this window]
[in a new window]
  		Fig. 1. Duplicate occurrences of genes are highly correlated. Seventy-four genes were found to be represented more than once in the list of differentially expressed genes we identified. A fold change relative to the 2-day value was determined for these genes at 8 and 15 days (higher/lower value to give a value >1), and these fold changes were compared across the replicate samples. A simple linear correlation was calculated for 71 of these genes. The regression line was defined by the following equation: replicate 2 = 1.22(replicate 1) - 0.3; r2 = 0.84.

 
Table 2 is a summary of how our microarray data compares to previously published work on small intestine gene expression across the crypt-villus axis. A number of well-known marker genes were present on our array and were found to increase as BBe cells differentiated in culture, e.g., sucrase-isomaltase, integrin-{alpha}3, and {alpha}1-antitrypsin. Several other genes related to the phenotype of small intestinal nutrient transport were also expressed at high levels in differentiated BBe cells, e.g., iron absorption genes hephaestin, DCT1, transferrin receptor, the folate receptor, and aquaporin 3. Several genes that others have found to be expressed in Caco-2 cells were absent in our microarray analysis, i.e., dipeptidyl peptidase IV (18), lactase (61), and calbindin D9k (20). This suggests that the level of these mRNAs is below the detection limit of the microarray, a condition that is likely to be true for BBe expression of calbindin D9k (20) or that the probe sets for these mRNA species are poorly designed.


View this table:
[in this window]
[in a new window]
  		Table 2. Relationship of microarray-based gene expression data on BBe cell differentiation to previously published findings on villus associated gene expression in small intestine

 
We examined seven genes by RT-PCR to confirm the changes we observed in the microarray analysis. All of the genes examined showed a pattern of expression similar to that found in the array data (Fig. 2). The differential expression of several of these genes was modest (2-fold or less), demonstrating the ability of the array to detect subtle changes in expression.



View larger version (45K):
[in this window]
[in a new window]
  		Fig. 2. Validation of microarray results for 7 genes by RT-PCR. RNA was prepared from BBe cells cultured for 2, 8, or 15 days (n = 4 per stage). Individual samples were analyzed for gene expression using the Affymetrix U95A GeneChip. Individual samples were pooled by stage, and the pooled samples were analyzed for gene expression by RT-PCR as described in the EXPERIMENTAL DESIGN AND METHODS.

 
Patterns of Gene Expression Across Stages of BBe Cell Differentiation
Figure 3 shows the 12 clusters that were identified from our data using self-organizing maps. Cluster formation was assessed under several different parameters (e.g., size of the matrix, number of iterations of the algorithm). The 4 x 3 cluster matrix selected had the best balance between three parameters: variability around the parent curves, the number of unique clusters, and the number of empty clusters. A reduced model was also examined whereby the 12 clusters were grouped based upon similar cluster patterns: group 1 = clusters 1, 5, 9 (downregulated from 2 to 8 days); group 2 = cluster 2 (downregulated after 8 days); group 3 = cluster 10 (nadir at 8 days); group 4 = cluster 3 (peak at 8 days); group 5 = clusters 4, 7, 8 (strongly upregulated from 2 to 8 days); group 6 = cluster 11, 12 (strongly upregulated from 8 to 15 days). There were 614 genes (48.6%) in clusters that showed a decline in expression from 2 to 15 days (clusters 1, 2, 3, and 9); 562 genes (46%) were in clusters that showed an increase in expression from 2 to 15 days (clusters 4, 7, 8, 11, and 12); and 63 (5.4%) genes had expression that either peaked or dipped at 8 days in culture (clusters 3 and 10).



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 3. Self-organizing map-based clusters of genes identified to be differentially expressed during BBe cell differentiation. Statistical analysis identified 1,238 total genes (1,150 unique genes) that were differentially expressed during BBe cell differentiation. These genes were subjected to cluster analysis using self-organizing maps using a 4 x 3 matrix in GeneCluster version 2.0 from the Whitehead Institute/MIT Center for Genome Research (http://www-genome.wi.mit.edu/cancer/software/software.html). The number of genes within a cluster is listed next to the cluster number (c#:# of genes). Within the manuscript the clusters are also combined into groups based upon similar patterns of change: group 1 = clusters 1, 5, 9; group 2 = cluster 2; group 3 = cluster 10; group 4 = cluster 3; group 5 = clusters 4, 7, 8; group 6 = clusters 11 and 12.

 
Functional Category Analysis
Examination of the gene functions within the clusters revealed interesting patterns in the functional categories represented by the clusters and groups; these patterns are shown for the groups in Table 3.


View this table:
[in this window]
[in a new window]
  		Table 3. Summary of the distribution of genes that are differentially expressed during BBe cell differentiation by functional classifications and group

 
Downregulated genes.
Group 1 was characterized by a large number of genes involved in proliferation (e.g., cyclins A2, B1, B2, D1, and D2; cdc proteins; signaling molecules like v-myb, Mad2, and Tc4; E2F-4 and E2F-5; and markers like PCNA, Ki-67 antigen, and ornithine decarboxylase), nucleic acid metabolism (e.g., 13 genes involved in DNA repair, 18 genes controlling purine and pyrimidine metabolism), and protein synthesis (e.g., 4 genes for ribosomal proteins, 35 genes for mRNA processing, 6 for translation, and 13 genes mediating protein folding). Thirteen of the 15 differentially expressed genes encoding proteins of the ubiquitin-mediated protein degradation pathway were found in group 1.

Transitional genes.
The number of genes present in the two clusters characterized by a spike or nadir of expression at 8 days in culture is small (63 total in clusters 3 and 10), making global statements about the shared function of these genes difficult. However, a greater proportion of genes in these clusters were involved in signaling or transcriptional control (21/62 = 33.9%) compared with either up- or downregulated genes (17.6 and 16.4%, respectively). Cluster 3 contained genes encoding TGFßII receptor {alpha}, PDGF receptor, CSF-I receptor, and TNF receptor superfamily member 11b (osteoprotegerin).

Upregulated genes.
The groups showing upregulation of gene expression with growth arrest and differentiation (groups 5 and 6) have a greater proportion of genes involved in metabolism (15.5% of genes in differentially expressed genes in these groups vs. 7.1% in groups 1 and 2). These genes include those for lipid metabolism (e.g., apolipoproteins CI, CIII, B100, H, and M; apolipoprotein B mRNA editing protein; LXR {alpha}; and SREBP-1) vitamin and mineral metabolism (e.g., retinal binding protein 4, selenium binding protein 1, selenophosphate sythetase 2, biotinidase, hephaestin, DCT1, transferrin receptor, and folate receptor), nutrient transport/digestion (Fig. 4C), and xenobiotic metabolism (12/14 xenobiotic metabolizing genes found in our list of differentially expressed genes were in groups 5 and 6, Fig. 4B). In addition, 8 of the 12 unique genes controlling oxidative stress are found in cluster 5 (Fig. 4A). While 27 genes in the "proliferation" category were upregulated, 16 of these genes were clearly, functionally linked to the inhibition of cell cycle. Similarly, 11 of 22 upregulated proteolysis-related genes encode for inhibitors. Finally, more genes with unknown or ambiguous function were found in groups 5 and 6 (141 genes or 26.7% of the total in those groups) than in groups 1 and 2 (83 genes or 14.8%).



View larger version (38K):
[in this window]
[in a new window]
  		Fig. 4. Differential expression of genes involved in protection from oxidative stress (A), xenobiotic metabolism (B), or nutrient metabolism or digestion (C). BBe cells were cultured for 2 days (proliferating, 50% confluent), 8 days (postproliferative, nondifferentiated), and 15 days (differentiated), and RNA was harvested and isolated. Gene expression in the RNA samples was analyzed using the Affymetrix U95A GeneChip. Following statistical analysis to define the differentially expressed genes, genes were classified based upon their functional characteristics and plotted based upon these functional classifications.

 
Signals important for intestinal differentiation are likely to be due to either paracrine signaling or through direct cell-cell/cell-matrix interactions. Figure 5 shows the impact of differentiation on the expression of genes encoding proteins involved in cell adhesion (Fig. 5A), cell-cell contact (Fig. 5B), and cell-matrix interactions (Fig. 5C). Distinct groups of genes within these categories were associated with proliferating cells [e.g., zyxin, BAP2{alpha}, and NK cell transcript 4 (Fig. 5A)], whereas other groups of genes were upregulated during the transition from proliferation to differentiation [e.g., occludin and fibrinogen-{alpha}, -ß, and -{gamma} (Fig. 5A)] or after differentiation [e.g., contactin and transmembrane 4 superfamily member 6 (Fig. 5A)].



View larger version (38K):
[in this window]
[in a new window]
  		Fig. 5. Differential expression of genes involved in cell adhesion (A), cell-cell contacts (B), and cell-matrix interactions (C). See Fig. 3 for a brief description of the study protocol.

 
Several signaling pathways have been associated with intestinal differentiation. Figure 6A shows that genes in the Wnt pathway are downregulated during differentiation. In contrast, phosphatidylinositol 3-kinase (PI3-kinase) pathway genes (Fig. 6B) are increased during the transition phase from proliferation to differentiation as well as into the differentiation phase. Distinct members of the TGFß signaling pathway family are associated with proliferation (e.g., BMP-2, 4, and 7; inhibin-ß B and C), whereas others are associated with the transition stage (e.g., TGFßII receptor {alpha}, TGFß family member, and fibromodulin) or the differentiated phenotype (latent TGFß binding protein 4, TGFß early growth response gene, and activin A receptor II) (Fig. 6C).



View larger version (29K):
[in this window]
[in a new window]
  		Fig. 6. Differential expression of genes involved in the Wnt signaling pathway (A), the PI-3 kinase signaling pathway (B), and the TGFß signaling pathway (C). See Fig. 3 for a brief description of the study protocol.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL DESIGN AND METHODS
 RESULTS
 DISCUSSION
References
 
Researchers are excited about the number of genes that can be examined simultaneously using a DNA microarray. However, the large number of comparisons resulting from microarray analysis means that there is a very strong likelihood of identifying false-positive differences (42). Our analysis shows that compared with independent t-tests, which identified 3,518 of 5,546 "present" genes as differentially expressed at P < 0.05 (Table 1), our approach dramatically reduced the false-positive rate (2,280 potential false positives excluded from consideration). This is only marginally improved by filtering out the genes in the lowest quartile of expression. Application of conservative correction procedures like the Bonferroni correction to the t-test results ({alpha} divided by the number of comparisons) severely reduced the number of genes identified to 227. Although most of these genes were also on our list of 1,150, the difference between the Bonferroni results and our own indicates that this comes at the expense of a high false-negative rate (>1,000 differentially expressed genes excluded). As a simple alternative to statistical analysis of microarray data, some investigators have utilized a threefold change in expression as a rigorous cutoff for defining a meaningful difference. However, our statistical analysis shows that 35% of the 601 "present" genes identified as differentially expressed using the threefold cutoff were not in our list of differentially expressed genes. This indicates that the fold-change approach, which does not account for variability in the expression of the genes, is also subject to a high false-positive rate. Although this was improved by filtering out genes in the lowest quartile of expression, the fold-difference approach still ignores more modest changes that may have biological significance.

Previously, Mariadason et al. (37) studied the spontaneous differentiation of the parental Caco-2 cell line using spotted cDNA microarrays and confluent cultures of cells as a reference point. They observed that Caco-2 cell differentiation downregulated the expression of 70% of the genes on their arrays, and they suggested that their data supported the use of Caco-2 cells as a model for studying colonocyte differentiation. In the data reported by Mariadason et al. (37), 322 genes were also found on our list of differentially expressed genes (i.e., exact name matches). When we compared these data [using the expression ratio of our data from 8-day and 15-day cultures of BBe cells to the expression ratio of their data from 5 days postconfluent (9 days in culture) and 14 days postconfluent (18 days in culture)], the correlation between these ratios was very poor (r2 = 0.01). Although this lack of correlation could be due to using comparisons with different baselines (i.e., confluent cells vs. 50% confluent cells), we think that the main cause of this poor correlation is differences in our approach. In contrast to the strong correlation that we observed between duplicates within our data set (r2 = 0.84), the correlation between 97 duplicates that we found in the Mariadason et al. data set of named, differentially expressed genes was generally weak (r2 = 0.15), suggesting their approach (spotted array, no sample replicates) was likely to identify only genes with robust changes in gene expression.

In addition to the report by Mariadason et al. (37), others have used DNA microarrays to evaluate the impact of butyrate, tricostatin A, sulindac, and curcumin on gene expression profiles in Caco-2 (56) or SW620 cells (39). These studies also focus on the expression changes associated with growth arrest and the downregulation of cell cycle. In addition, Tadjali et al. (57) used a filter array spotted with 18,149 expressed sequence tags (ESTs) and found that the number of genes expressed in differentiated cultures of parental Caco-2 cells (7 days postconfluent) was reduced compared with undifferentiated cells (3 day cultures). In contrast to these reports, our microarray analysis reveals that the BBe clone of Caco-2 expresses a small intestinal phenotype upon differentiation (Table 2), consistent with what others have found for the parental line and several Caco-2 clones (12, 49). It is this characteristic that has made Caco-2 cells a valuable tool for the study of differentiation-induced gene expression in enterocytes, e.g., sucrase-isomaltase (8), lactase-phloridzin hydrolase (31), calbindin D9k (1, 33), and dipeptidyl peptidase IV (18). However, it is clear that the Caco-2 cell and its clones are not small intestinal enterocytes and that the acquisition of this phenotype likely reflects a regression to a fetal phenotype. During weeks 14–28 of human fetal development, the colon develops villus-like projections and expresses the small intestinal marker enzyme, sucrase-isomaltase (32). Bates et al. (2) have conducted a survey of mouse tissues and found 571 genes were expressed to a greater degree in at least one segment of the intestine. Forty-nine of the genes identified in that screen were present in the BBe cell cultures. While 35 of these genes were expressed predominantly in mouse small intestine, 14 were expressed predominantly in the lower intestine. Thus it appears that the Caco-2 cell differentiation leads to a hybrid cell with both colonocyte and enterocyte characteristics.

The major strength of the microarray approach is that it can identify relationships between phenotypes and gene expression that can serve as the basis for future hypothesis testing. Although many potential stories can be told with microarray data, we feel that the most interesting one that arose from our analysis relates to the regulation of genes that are involved in, or utilize, specific signaling pathways, i.e., cell-cell and cell-matrix interactions, Wnt, PI3-kinase, and TGFß signaling. Modulation of these pathways may be important for the development of the enterocyte (or colonocyte) phenotype.

Cell-cell/cell-matrix interactions.
The role of interaction between integrins and extracellular matrix during enterocyte differentiation is well characterized (4, 54) and involves the coordinated production of proteins from both the intestinal epithelium and mesenchymal cells (46). Our data suggest that distinct families of genes involved in cell-cell or cell-matrix interactions are expressed in proliferating as opposed to postproliferative or differentiated cells. Expression of genes for the cell adhesion molecules bystin-like protein, NK cell transcript 4, and zyxin, the junction-associated proteins vinculin and brain-heart protocadherin, and the integrin/matrix proteins NAG-2, laminin-{gamma}1 and -ß1, and the proto-oncogene integrin-linked kinase genes was highest in proliferating cells. However, as cells stopped proliferating, the mRNA level for a large number of other genes in this category increased (Fig. 5); few of these genes changed further from 8 to 15 days in culture. This suggests that cell-cell, cell-matrix interactions are more likely involved with the early process of differentiation (i.e., growth arrest and commitment) than postproliferative acquisition of the differentiated phenotype. Several interesting examples suggesting this are given in the sections below.

Wnt signaling.
The Wnt signaling pathway has been implicated in the control of intestinal proliferation; deregulation or mutation within members of this pathway leads to excessive proliferation and has been implicated in colon cancer (e.g., mutations in the inhibitory factor APC) (9). In undifferentiated Caco-2 cells, overexpression of APC or ß-catenin, and expression of a dominant negative form of TCF-4, a transcription factor that mediates Wnt action, significantly increased promoter activity of alkaline phosphatase and intestinal fatty acid binding protein, but not sucrase reporter genes (38). This suggests that blocking Wnt signaling is a critical step for the development of at least some of the differentiated phenotype in Caco-2. Our microarray data found that a number of TCF gene targets (c-myc, cyclin D1) and Wnt signaling components (frizzled, PP2A subunits, PKC iota) are coordinately downregulated by differentiation. The downregulation of Wnt signaling may result from specific cell-cell or cell-matrix interactions. Perreault et al. (47) recently showed that the transcription factor Fox1 activates the Wnt pathway by increasing the expression of extracellular proteoglycans which act as coreceptors for Wnt, i.e., syndecan-1. The downregulation that we observed in the expression of the desmosome component plakophilin-2 could be an essential step in suppression of Wnt signaling. Expression of plakophilin-2 in SW480 colon tumor cells upregulates endogenous ß-catenin signaling activity, and this can be abolished by ectopic expression of E-cadherin (14).

PI3-kinase.
The role of the PI3-kinase signaling pathway in enterocyte differentiation is controversial. Laprise et al. (34) showed that inhibition of PI3-kinase with LY294002 inhibits differentiation of the Caco-2 2/15 clone (e.g., sucrase expression, formation of a well-developed brush-border membrane). In their study, PI3-kinase was recruited to, and activated by, E-cadherin-mediated cell-cell contacts, leading to the assembly of adherens junctions and enterocyte differentiation. In contrast, Wang et al. (62) found that inhibition of PI3-kinase by wortmannin enhanced basal and sodium butyrate-induced proliferation of HT-29 cells and the parental line of Caco-2. They later showed that PI3-kinase inhibition or PTEN activation activates the cdx-2 gene promoter through a pathway that includes NF-{kappa}B activation (28). While we also see an upregulation of cdx-2 mRNA levels during differentiation in our microarray data (2.2-fold from 2 to 15 days in culture), our data appear to support the studies by Laprise et al. (34). We find that genes encoding PI3-kinase pathway-related genes are generally upregulated during the transition stage between proliferation and differentiation (Fig. 6B). PAK1 (p21 activated kinase 1) is a kinase activated by PI3-kinase through stimulation of the PDGF receptor in NIH 3T3 cells (52). In mast cells, PDGF receptor activation stimulates cell-matrix adhesion through fibronectin by the independent activation of PI3-kinase and phospholipase C {gamma}1 (29). In our study, fibronectin mRNA levels were high at 2 and 8 days in culture but fell by 70% afterward. This suggests that activation of PI3-kinase (and PAK1) through the PDGF receptor could be involved in matrix-induced differentiation but is not needed thereafter. Finally, we found that differentiation was associated with the expression of the erbB2 (HER2), erbB3 (HER3), and TOB (transducer of erbB2). Increased expression of erbB2 and erbB3 is seen after maturation of mammary epithelium in rats (17). In 15-day-old cultures of the T84 colonic epithelial cell line, stimulation of these receptors can activate PI3-kinase and lead to inhibition of calcium-dependent chloride secretion (27). These data suggest that erbB2 and erbB3 may utilize the PI3-kinase pathway to control epithelial cell differentiation and differentiated function of intestinal epithelium.

TGFß-related signaling.
Our data suggest that activation of TGFß signaling through the TGFßII receptor {alpha} could be involved in the transition from proliferation to differentiation (2-fold upregulation of mRNA at 8 days). Suppression of TGFß type II receptor by ras transformation in intestinal epithelial cells leads to resistance to the growth inhibitory actions of TGFß (10), and colon cancer in mice may be due to PKC ßII induced suppression of the TGFßII receptor (41). Similarly, disruption of the gene encoding latent TGFß binding protein 4 leads to epithelial cells with reduced levels of phosphorylated Smad2, overexpression of c-myc, and uncontrolled proliferation (53). Increased levels of TGFß early growth response gene (also known as TGFß inducible early gene, TIEG) mRNA indicates that the TGFß signaling pathway has been activated by differentiation in BBe cells; TIEG is a zinc finger transcription factor family member whose expression is rapidly induced in cells treated with TGFß (55). In MG-63 osteosarcoma cells, TIEG mRNA overexpression leads to changes that mimic the effect of TGFß treatment (23), indicating that TIEG is the mediator of the prodifferentiating effects of TGFß. The downregulated TGFß family member genes we identified include members of the inhibin/activin system (6) and several BMP family members; these have not been extensively studied in the context of intestinal biology. However, activin A is known to promote growth arrest and differentiation in a number of tissues (13, 15) and inhibins work in antagonism of activin action (7). Thus our observation that inhibin-ß B and C mRNA levels are downregulated while activin A receptor II mRNA levels increase during differentiation are consistent with a role for these TGFß family members in differentiation.

Conclusion.
Our data demonstrate the value of a rigorous statistical approach to the analysis of microarray data. By using procedures to balance type I and II errors we created a more reliable list of genes that are differentially expressed during BBe cell differentiation (i.e., compared with independent t-tests or fold change). This analysis revealed that the BBe cell expresses the molecular signature of both a colonocyte and an enterocyte. As such, researchers should used caution when extrapolating data from these cells to explain the biology of the normal colon or small intestine.

Cluster analysis and functional assessment of the differentially expressed genes has identified interesting patterns of change in the expression of genes associated with several signaling pathways. These changes provide a foundation for future hypothesis based experiments on the role of specific family members within the TGFß, Wnt, and PI3-kinase signaling pathways during enterocyte differentiation. An unresolved issue is how the patterns of change we observed in these (or other) pathways activates the expression of genes that characterize the differentiated phenotype of an absorptive epithelial cell, e.g., iron transport genes (DCT1, hephaestin), sucrase-isomaltase, and MRP3. Additional studies on the posttranslational modification of transcription factors implicated in intestine-specific gene expression [i.e., HNF-1{alpha}, cdx-2, and GATA-4 (8, 31)], similar to those that demonstrate the importance of p38 kinase in cdx-2-mediated gene transcription (25), will be necessary to clarify this issue.


   ACKNOWLEDGMENTS
 
We thank Ronald E. Jerome, Charles Nicholson, and Jeanette N. McClintick for technical assistance during the microarray analysis.

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-54111 (to J. C. Fleet). The Center for Medical Genomics at Indiana University School of Medicine is supported in part by a grant from the Indiana 21st Century Research and Technology Fund (to H. J. Edenberg) and by the Indiana Genomics Initiative. The Indiana Genomics Initiative of Indiana University is supported in part by the Lilly Endowment.


   FOOTNOTES
 
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).

Address for reprint requests and other correspondence: J. C. Fleet, 1264 Stone Hall, Purdue Univ., West Lafayette, IN 47907 (E-mail: fleetj{at}cfs.purdue.edu).

10.1152/physiolgenomics.00152.2002.


   References
TOP
 ABSTRACT
 INTRODUCTION
 EXPERIMENTAL DESIGN AND METHODS
 RESULTS
 DISCUSSION
 References
 

  1. Barley NF, Prathalingam SR, Zhi P, Legon S, Howard A, and Walters JRF. Factors involved in the duodenal expression of the human calbindin-D9k gene. Biochem J 341: 491–500, 1999.[Medline]
  2. Bates MD, Erwin CR, Sanford LP, Wiginton D, Bezerra JA, Schatzman LC, Jegga AG, Ley-Ebert C, Williams SS, Steinbrecher KA, Warner BW, Cohen MB, and Aronow BJ. Novel genes and functional relationships in the adult mouse gastrointestinal tract identified by microarray analysis. Gastroenterology 122: 1467–1482, 2002.[ISI][Medline]
  3. Beaulieu JF. Differential expression of the VLA family of integrins along the crypt-villus axis in the human small intestine. J Cell Sci 102: 427–436, 1992.[Abstract/Free Full Text]
  4. Beaulieu JF. Integrins and human intestinal cell functions. Front Biosci 4: D310–D321, 1999.[Medline]
  5. Benjamini Y and Hochberg Y. Controlling false discovery rate: a practical and powerful approach to multiple testing. J Royal Stat Soc B 57: 289–300, 1995.
  6. Bernard DJ, Chapman SC, and Woodruff TK. An emerging role for co-receptors in inhibin signal transduction. Mol Cell Endocrinol 180: 55–62, 2001.[Medline]
  7. Bernard DJ, Chapman SC, and Woodruff TK. Mechanisms of inhibin signal transduction. Recent Prog Horm Res 56: 417–450, 2001.[Abstract]
  8. Boudreau F, Rings EH, Van Wering HM, Kim RK, Swain GP, Krasinski SD, Moffett J, Grand RJ, Suh ER, and Traber PG. Hepatocyte nuclear factor-1 alpha, GATA-4, and caudal related homeodomain protein Cdx2 interact functionally to modulate intestinal gene transcription. Implication for the developmental regulation of the sucrase-isomaltase gene. J Biol Chem 277: 31909–31917, 2002.[Abstract/Free Full Text]
  9. Brittan M and Wright NA. Gastrointestinal stem cells. J Pathol 197: 492–509, 2002.[ISI][Medline]
  10. Bulus NM, Sheng HM, Sizemore N, Oldham SM, Barnett JV, Coffey RJ, Beauchamp DR, and Barnard JA. Ras-mediated suppression of TGFßRII expression in intestinal epithelial cells involves Raf-independent signaling. Neoplasia 2: 357–364, 2000.[Medline]
  11. Chandrasena G, Sunitha I, Lau C, Nanthakumar NN, and Henning SJ. Expression of sucrase-isomaltase mRNA along the villus-crypt axis in the rat small intestine. Cell Mol Biol 38: 243–254, 1992.[ISI][Medline]
  12. Chantret I, Rodolosse A, Barbat A, Dussaulx E, Brot-Laroche E, Zweibaum A, and Rousset M. Differential expression of sucrase-isomaltase in clones isolated from early and late passages of the cell line Caco-2: evidence for glucose-dependent negative regulation. J Cell Sci 107: 213–225, 1994.[Abstract]
  13. Chen L, Chen D, Block E, O’Donnell M, Kufe DW, and Clinton SK. Eradication of murine bladder carcinoma by intratumor injection of a bicistronic adenoviral vector carrying cDNAs for the IL-12 heterodimer and its inhibition by the IL-12 p40 subunit homodimer. J Immunol 159: 351–359, 1997.[Abstract]
  14. Chen X, Bonne S, Hatzfeld M, van Roy F, and Green KJ. Protein binding and functional characterization of plakophilin 2. Evidence for its diverse roles in desmosomes and beta-catenin signaling. J Biol Chem 277: 10512–10522, 2002.[Abstract/Free Full Text]
  15. Chen YG, Lui HM, Lin SL, Lee JM, and Ying SY. Regulation of cell proliferation, apoptosis, and carcinogenesis by activin. Exp Biol Med (Maywood) 227: 75–87, 2002.[Abstract/Free Full Text]
  16. Colnot S, Romagnolo B, Lambert M, Cluzeaud F, Porteu A, Vandewalle A, Thomasset M, Kahn A, and Perret C. Intestinal expression of the calbindin-D9K gene in transgenic mice. Requirement for a Cdx2-binding site in a distal activator region. J Biol Chem 273: 31939–31946, 1998.[Abstract/Free Full Text]
  17. Darcy KM, Zangani D, Wohlhueter AL, Huang RY, Vaughan MM, Russell JA, and Ip MM. Changes in ErbB2 (her-2/neu), ErbB3, and ErbB4 during growth, differentiation, and apoptosis of normal rat mammary epithelial cells. J Histochem Cytochem 48: 63–80, 2000.[Abstract/Free Full Text]
  18. Darmoul D, Lacasa M, Baricault L, Marguet D, Sapin C, Trotot P, Barbat A, and Trugnan G. Dipeptidyl peptidase IV (CD 26) gene expression in enterocyte-like colon cancer cell lines HT-29 and Caco-2. Cloning of the complete human coding sequence and changes of dipeptidyl peptidase IV mRNA levels during cell differentiation. J Biol Chem 267: 4824–4833, 1992.[Abstract/Free Full Text]
  19. Davison A and Hinkley D. Bootstrap Methods and Their applications. Cambridge, UK: Cambridge University Press, 1997.
  20. Fleet JC, Eksir F, Hance KW, and Wood RJ. Vitamin D-inducible calcium transport and gene expression in three Caco-2 cell lines. Am J Physiol Gastrointest Liver Physiol 283: G618–G625, 2002.[Abstract/Free Full Text]
  21. Frazer DM, Vulpe CD, McKie AT, Wilkins SJ, Trinder D, Cleghorn GJ, and Anderson GJ. Cloning and gastrointestinal expression of rat hephaestin: relationship to other iron transport proteins. Am J Physiol Gastrointest Liver Physiol 281: G931–G939, 2001.[Abstract/Free Full Text]
  22. Hadjiagapiou C, Giannoni F, Funahashi T, Skarosi SF, and Davidson NO. Molecular cloning of a human small intestinal apolipoprotein B mRNA editing protein. Nucleic Acids Res 22: 1874–1879, 1994.[Abstract/Free Full Text]
  23. Hefferan TE, Reinholz GG, Rickard DJ, Johnsen SA, Waters KM, Subramaniam M, and Spelsberg TC. Overexpression of a nuclear protein, TIEG, mimics transforming growth factor-ß action in human osteoblast cells. J Biol Chem 275: 20255–20259, 2000.[Abstract/Free Full Text]
  24. Hong WJ, Petell JK, Swank D, Sanford J, Hixson DC, and Doyle D. Expression of dipeptidyl peptidase IV in rat tissues is mainly regulated at the mRNA levels. Exp Cell Res 182: 256–266, 1989.[Medline]
  25. Houde M, Laprise P, Jean D, Blais M, Asselin C, and Rivard N. Intestinal epithelial cell differentiation involves activation of p38 mitogen-activated protein kinase that regulates the homeobox transcription factor CDX2. J Biol Chem 276: 21885–21894, 2001.[Abstract/Free Full Text]
  26. Jiang L, David ES, Espina N, and Ferraris RP. GLUT-5 expression in neonatal rats: crypt-villus location and age-dependent regulation. Am J Physiol Gastrointest Liver Physiol 281: G666–G674, 2001.[Abstract/Free Full Text]
  27. Keely SJ and Barrett KE. ErbB2 and ErbB3 receptors mediate inhibition of calcium-dependent chloride secretion in colonic epithelial cells. J Biol Chem 274: 33449–33454, 1999.[Abstract/Free Full Text]
  28. Kim S, Domon-Dell C, Wang Q, Chung DH, Di Cristofano A, Pandolfi PP, Freund JN, and Evers BM. PTEN and TNF-alpha regulation of the intestinal-specific Cdx-2 homeobox gene through a PI3K, PKB/Akt, and NF-{kappa}B-dependent pathway. Gastroenterology 123: 1163–1178, 2002.[ISI][Medline]
  29. Kinashi T, Escobedo JA, Williams LT, Takatsu K, and Springer TA. Receptor tyrosine kinase stimulates cell-matrix adhesion by phosphatidylinositol-3 kinase and phospholipase c-gamma-1 pathways. Blood 86: 2086–2090, 1995.[Abstract/Free Full Text]
  30. Koyama Y, Yamamoto T, Tani T, Nihei K, Kondo D, Funaki H, Yaoita E, Kawasaki K, Sato N, Hatakeyama K, and Kihara I. Expression and localization of aquaporins in rat gastrointestinal tract. Am J Physiol Cell Physiol 276: C621–C627, 1999.[Abstract/Free Full Text]
  31. Krasinski SD, Van Wering HM, Tannemaat MR, and Grand RJ. Differential activation of intestinal gene promoters: functional interactions between GATA-5 and HNF-1{alpha}. Am J Physiol Gastrointest Liver Physiol 281: G69–G84, 2001.[Abstract/Free Full Text]
  32. Lacroix B, Kedinger M, Simon-Assmann P, Rousset M, Zweibaum A, and Haffen K. Developmental pattern of brush border enzymes in the human fetal colon. Correlation with some morphogenetic events. Early Hum Dev 9: 95–103, 1984.[ISI][Medline]
  33. Lambert M, Colnot S, Suh E, L’Horset F, Blin C, Calliot ME, Raymondjean M, Thomasset M, Traber PG, and Perret C. cis-acting elements and transcription factors involved in the intestinal specific expression of the rat calbindin-D9k gene. Binding of the intestine-specific transcription factor Cdx-2 to the TATA box. Eur J Biochem 236: 778–788, 1996.[ISI][Medline]
  34. Laprise P, Chailler P, Houde M, Beaulieu JF, Boucher MJ, and Rivard N. Phosphatidylinositol 3-kinase controls human intestinal epithelial cell differentiation by promoting adherens junction assembly and p38 MAPK activation. J Biol Chem 277: 8226–8234, 2002.[Abstract/Free Full Text]
  35. Lee SY, Madan A, Furuta GT, Colgan SP, and Sibley E. Lactase gene transcription is activated in response to hypoxia in intestinal epithelial cells. Mol Genet Metab 75: 65–69, 2002.[Medline]
  36. Li W, Wang J, Coluccio LM, Matsudaira P, and Grand RJ. Brush border myosin I (BBMI): a basally localized transcript in human jejunal enterocytes. J Histochem Cytochem 48: 89–94, 2000.[Abstract/Free Full Text]
  37. Mariadason JM, Arango D, Corner GA, Aranes MJ, Hotchkiss KA, Yang W, and Augenlicht LH. A gene expression profile that defines colon cell maturation in vitro. Cancer Res 62: 4791–4804, 2002.[Abstract/Free Full Text]
  38. Mariadason JM, Bordonaro M, Aslam F, Shi L, Kuraguchi M, Velcich A, and Augenlicht LH. Down-regulation of ß-catenin TCF signaling is linked to colonic epithelial cell differentiation. Cancer Res 61: 3465–3471, 2001.[Abstract/Free Full Text]
  39. Mariadason JM, Corner GA, and Augenlicht LH. Genetic reprogramming in pathways of colonic cell maturation induced by short chain fatty acids: comparison with trichostatin A, sulindac, and curcumin and implications for chemoprevention of colon cancer. Cancer Res 60: 4561–4572, 2000.[Abstract/Free Full Text]
  40. Mills JC and Gordon JI. A new approach for filtering noise from high-density oligonucleotide microarray datasets. Nucleic Acids Res 29: E72, 2001.[Medline]
  41. Murray NR, Weems C, Chen L, Leon J, Yu W, Davidson LA, Jamieson L, Chapkin RS, Thompson EA, and Fields AP. Protein kinase C ßII and TGFßRII in omega-3 fatty acid-mediated inhibition of colon carcinogenesis. J Cell Biol 157: 915–920, 2002.[Abstract/Free Full Text]
  42. Nadon R and Shoemaker J. Statistical issue with microarrays: processing and analysis. Trends Genet 18, 265–271. 2002.[ISI][Medline]
  43. Nagy LE, Pittler A, and Kretchmer N. Development of glutaminase along the villus-crypt axis in the jejunum of rat. J Pediatr Gastroenterol Nutr 7: 907–913, 1988.[Medline]
  44. Oates PS, Thomas C, and Morgan EH. Transferrin receptor activity and localisation in the rat duodenum. Pflügers Arch 440: 116–124, 2000.[ISI][Medline]
  45. Pageot LP, Perreault N, Basora N, Francoeur C, Magny P, and Beaulieu JF. Human cell models to study small intestinal functions: recapitulation of the crypt-villus axis. Microsc Res Tech 49: 394–406, 2000.[ISI][Medline]
  46. Perreault N, Herring-Gillam FE, Desloges N, Belanger I, Pageot LP, and Beaulieu JF. Epithelial vs. mesenchymal contribution to the extracellular matrix in the human intestine. Biochem Biophys Res Commun 248: 121–126, 1998.[ISI][Medline]
  47. Perreault N, Katz JP, Sackett SD, and Kaestner KH. Foxl1 controls the Wnt/ß-catenin pathway by modulating the expression of proteoglycans in the gut. J Biol Chem 276: 43328–43333, 2001.[Abstract/Free Full Text]
  48. Peterson MD and Mooseker MS. Characterization of the enterocyte-like brush border cytoskeleton of the C2BBe clones of the human intestinal cell line, Caco-2. J Cell Sci 102: 581–600, 1992.[Abstract/Free Full Text]
  49. Pinto M, Robine-Leon S, Appay MD, Kedinger M, Triadow N, Dussaulx E, Lacroix B, Simon-Assmann P, Haffer K, Fogh J, and Zweibaum A. Enterocyte-like differentiation and polarization of the human colon carcinoma cell line Caco-2 in culture. Biol Cell 47: 323–330, 1983.[ISI]
  50. Rao PV. Statistical Methods in the Life Sciences. Pacific Grove, CA: Duxbury, 1998.
  51. Said HM, Nguyen TT, Dyer DL, Cowan KH, and Rubin SA. Intestinal folate transport: identification of a cDNA involved in folate transport and the functional expression and distribution of its mRNA. Biochim Biophys Acta 1281: 164–172, 1996.[Medline]
  52. Sells MA, Pfaff A, and Chernoff J. Temporal and spatial distribution of activated Pak1 in fibroblasts. J Cell Biol 151: 1449–1458, 2000.[Abstract/Free Full Text]
  53. Sterner-Kock A, Thorey IS, Koli K, Wempe F, Otte J, Bangsow T, Kuhlmeier K, Kirchner T, Jin S, Keski-Oja J, and von Melchner H. Disruption of the gene encoding the latent transforming growth factor-ß binding protein 4 (LTBP-4) causes abnormal lung development, cardiomyopathy, and colorectal cancer. Genes Dev 16: 2264–2273, 2002.[Abstract/Free Full Text]
  54. Stutzmann J, Bellissent-Waydelich A, Fontao L, Launay JF, and Simon-Assmann P. Adhesion complexes implicated in intestinal epithelial cell-matrix interactions. Microsc Res Tech 51: 179–190, 2000.[Medline]
  55. Subramaniam M, Hefferan TE, Tau K, Peus D, Pittelkow M, Jalal S, Riggs BL, Roche P, and Spelsberg TC. Tissue, cell type, and breast cancer stage-specific expression of a TGF-ß inducible early transcription factor gene. J Cell Biochem 68: 226–236, 1998.[ISI][Medline]
  56. Tabuchi Y, Arai Y, Kondo T, Takeguchi N, and Asano S. Identification of genes responsive to sodium butyrate in colonic epithelial cells. Biochem Biophys Res Commun 293: 1287–1294, 2002.[ISI][Medline]
  57. Tadjali M, Seidelin JB, Olsen J, and Troelsen JT. Transcriptome changes during intestinal cell differentiation. Biochim Biophys Acta 1589: 160–167, 2002.[Medline]
  58. Tamayo P, Slonim D, Mesirov J, Zhu Q, Kitareewan S, Dmitrovsky E, Lander ES, and Golub TR. Interpreting patterns of gene expression with self-organizing maps: methods and application to hematopoietic differentiation. Proc Natl Acad Sci USA 96: 2907–2912, 1999.[Abstract/Free Full Text]
  59. Thomson AB, Keelan M, Thiesen A, Clandinin MT, Ropeleski M, and Wild GE. Small bowel review: diseases of the small intestine. Dig Dis Sci 46: 2555–2566, 2001.[ISI][Medline]
  60. Trinder D, Oates PS, Thomas C, Sadleir J, and Morgan EH. Localisation of divalent metal transporter 1 (DMT1) to the microvillus membrane of rat duodenal enterocytes in iron deficiency, but to hepatocytes in iron overload. Gut 46: 270–276, 2000.[Abstract/Free Full Text]
  61. van Beers EH, Al RH, Rings EHHM, Einerhand AWC, Dekker J, and Buller HA. Lactase and sucrase-isomaltase gene expression during Caco-2 cell differentiation. Biochem J 308: 769–775, 1995.[Medline]
  62. Wang Q, Wang X, Hernandez A, Kim S, and Evers BM. Inhibition of the phosphatidylinositol 3-kinase pathway contributes to HT29 and Caco-2 intestinal cell differentiation. Gastroenterology 120: 1381–1392, 2001.[Medline]
  63. Westcarr S, Farshori P, Wyche J, and Anderson WA. Apoptosis and differentiation in the crypt-villus unit of the rat small intestine. J Submicrosc Cytol Pathol 31: 15–30, 1999.[ISI][Medline]



This article has been cited by other articles:


Home page
 Am. J. Physiol. Gastrointest. Liver Physiol.Home page
M.-L. Wang, S. A. Keilbaugh, T. Cash-Mason, X. C. He, L. Li, and G. D. Wu
Immune-mediated signaling in intestinal goblet cells via PI3-kinase- and AKT-dependent pathways
Am J Physiol Gastrointest Liver Physiol, November 1, 2008; 295(5): G1122 - G1130.
[Abstract] [Full Text] [PDF]

Home page
 J. Biol. Chem.Home page
W. V. Graham, F. Wang, D. R. Clayburgh, J. X. Cheng, B. Yoon, Y. Wang, A. Lin, and J. R. Turner
Tumor Necrosis Factor-induced Long Myosin Light Chain Kinase Transcription Is Regulated by Differentiation-dependent Signaling Events: CHARACTERIZATION OF THE HUMAN LONG MYOSIN LIGHT CHAIN KINASE PROMOTER
J. Biol. Chem., September 8, 2006; 281(36): 26205 - 26215.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow All Versions of this Article:
13/1/57    most recent
00152.2002v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science