![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
* Albert Einstein Cancer Center, Bronx, NY, Faculte de Medicine, U539 INSERM, Nantes, France, ** American Health Foundation, Valhalla, NY, Strang Cancer Prevention Center, New York, NY, New York Medical College, Valhalla, NY and # North Shore-LIJ Research Institute, Manhasset, NY
3 To whom correspondence should be addressed. E-mail: augen{at}aecom.yu.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT LITERATURE CITED |
|---|
KEY WORDS: • microarrays • gene profiling • intestinal cancer • p21 • Muc2
The development of high-throughput methods for analysis of expression of each of thousands of sequences simultaneously has provided powerful tools for investigation of mechanisms underlying cell and tissue phenotype. This includes identification of markers for recognizing response, and equally important, probability and nature of response to exogenous stimuli of cells, whether physiological (e.g., growth factors, nutritional factors and their derivatives) or pharmacological (e.g., chemotherapeutic drugs and chemopreventive agents). Microarrays, Affymetrix (Santa Clara, CA) technology and other approaches such as serial analysis of gene expression (SAGE)4 were developed and rapidly advanced in the last decade, which made the approach accessible to many investigators. However, the concept that gene expression profiling could provide considerable insight into mechanisms of differentiation and development, and could be a means for defining markers of response, and important clinical phenotypes such as probability of tumor response to drug and invasive or metastatic potential, was described by us a decade earlier. Moreover, the concept was reduced to practice.
In 1982, we published a comparison of gene-expression patterns among a mouse colon tumor and normal colon and other tissues based on arrays of 400 sequences ( 1). This analysis included an attempt to cluster the genes based on level of expression in the tissues. We then extended the approach to the analysis of 4,000 arrayed sequences in human colon tumors and normal tissues, developing the first computerized scanning and image-processing systems to digitize the data, which permitted numerical and statistical analyses of the results. The first such digitized arrays were produced at the Argonne National Laboratories in 1983 and 1984, and our first report appeared in 1987 ( 2). This paper identified the progression of changes in expression from flat mucosa through benign adenoma to malignant carcinoma, and compared this to genes that were altered in expression as carcinoma cells in culture were induced by the short-chain fatty acid, butyrate, back to a more normal phenotype. The data identified alterations in expression of mitochondrial genes in both differentiation and transformation of intestinal epithelial cells, and led to a series of studies by Heerdt that determined the importance of mitochondrial function in colon cell maturation pathways, especially in the coordination of apoptosis and proliferation ( 3– 8). In a number of review articles, we discussed the potential of such gene profiling in clinical management of patients and in providing insight into tumor behavior and response to therapy ( 9– 13). In 1991, this potential was realized by our description of a pattern of expression of 30 genes, selected from 4,000 screened in defined arrays, that successfully distinguished high-risk flat colonic mucosa from familial polyposis and hereditary nonpolyposis colon cancer patients from the flat mucosa of subjects at low risk for tumor development ( 14). Four years later, the Brown group at Stanford University reported technological achievements that permitted the arraying of sequences as microspots on glass slides for their simultaneous analysis ( 15). Although these first microarrays consisted of only 48 genes, the concept quickly captured the imagination of the scientific community, and rapid improvements in technology soon increased the number of sequences that could be screened in parallel. The history of the development of these approaches is described by Zweiger ( 16). The different methodologies that have been developed, each with its own strengths and limitations, were reviewed on many occasions and will not be dealt with here.
There is now clear recognition of the utility of gene expression profiling in a number of areas. The accessibility of the techniques to many investigators has rapidly increased the number of reports that use these methods, either as a beginning to gene identification or as an approach for defining phenotype. There are, however, many difficulties and pitfalls involved in the use of such data. We will attempt to describe methods that we have found useful in addressing a number of different biological and clinical questions under investigation in our laboratory. This is not meant in any sense to be a manual of methods, but rather an illustration of the kind of iterative and interactive approaches that can be used to analyze these complex databases and questions.
The role of k-ras gene mutations in colon cell transformation: the butyrate paradox
Gene expression profiling can provide immediate insight into the mechanism by which an altered phenotype is generated. This is the case in our investigations into the role of k-ras mutations in colonic cell transformation. Isogenic colon carcinoma cell lines have been generated in which the mutated k-ras gene in the parental HCT116 cell line has been specifically deleted, leaving the derived hkh2 and hke-3 cell lines with a normal k-ras allele, but no mutated allele ( 17). This is sufficient to reduce their ability to form tumors. Klampfer recognized that these isogenic lines offered an excellent opportunity to address how the presence of a ras gene mutation participates in the generation of the transformed phenotype in colon cells. A striking observation was that the parental cells that contain a mutant k-ras were highly sensitive to induction of apoptosis by the short-chain fatty acid (SCFA) butyrate, but the cells in which the mutant k-ras had been deleted did not respond by undergoing apoptosis when treated with butyrate. This is reminiscent of the butyrate paradox in which the fact that SCFAs such as butyrate, derived by fiber fermentation in the intestinal lumen, are the principal energy source for normal colonic epithelial cells and stimulate their proliferation, but in transformed cells they stimulate apoptosis ( 18). To explore this further, Klampfer performed microarray analysis using arrays prepared by the Albert Einstein core facility that contained >9,000 sequences. The most downregulated sequence in the mutant k-ras containing parental line compared to the hkh2 and hke-3 cell lines devoid of a k-ras mutation was gelsolin, an actin-binding and cleavage protein that has been implicated in pathways of cell differentiation and apoptosis ( 19). The downregulation was confirmed by Western blot analysis. To determine if it was a direct effector of mutant k-ras in altering cell phenotype, Klampfer used a retroviral infection strategy to introduce expression of an antisense vector to gelsolin into the Caco-2 colon cancer cell line (L. Klempfer, unpublished data). This antisense effectively reduced gelsolin expression and also greatly sensitized the cells to butyrate-induced apoptosis. Thus, the altered levels of expression of gelsolin, as well as the ability of forced downregulation to sensitize to apoptosis, provided presumptive evidence that mutations in k-ras, which constitutively activate MAPK signaling, encompass downregulation of gelsolin as an important mechanism in the sensitization of transformed colonic epithelial cells to butyrate-induced apoptosis. Further, in the nonsensitive cells that have higher levels of gelsolin, the molecule was shown to bind to voltage dependent anion channel protein (VDAC), an important element in regulating the release of cytochrome C and other factors from mitochondria that trigger apoptosis. The working hypothesis is that the downregulation of gelsolin expression caused by mutant k-ras eliminates the interaction of gelsolin and VDAC, thereby allowing the cascade of events that results in cell death to proceed. Thus, in this case, microarray analysis may have identified an important molecule regulated by ras signaling that plays a clear role in modulation of a critical phenotype: sensitivity to butyrate-induced apoptosis.
The role of p21WAF1/cip1 in intestinal tumor formation and response to sulindac
A second fairly direct use of gene expression profiling was our investigation of genes, particularly the cyclin-dependent kinase inhibitor p21WAF1/cip1, that are involved in the response of colonic carcinoma cells to the nonsteroidal antiinflammatory drug, sulindac. Although it is known that sulindac is an inhibitor of cyclooxygenase I and II, and that cyclooxygenase is a key target in its activity as a chemopreventive agent for colon tumor formation, it is not known how this translates into reduced tumor formation and growth ( 20). Mariadason et al. investigated the effects of sulindac on profiles of gene expression of colonic carcinoma cells in vitro, and found the effects to be complex, involving many genes that encode proteins involved in signaling and cell cycle regulation and progression ( 21). Interestingly, although both butyrate and sulindac induced a G0/G1 cell cycle arrest in cell culture, the profiles of gene expression induced by the two were remarkably different, suggesting that there were multiple paths to cell cycle arrest. Moreover, the pathway that had evolved to respond to the physiological regulator butyrate was quite different from that triggered by the comparatively and recently developed pharmacological agent sulindac. Indeed, examination of the response to the two agents after treatment revealed that there was a monotonic increase in the number of genes altered in expression by butyrate as a function of time: that is, a progressive expansion of the number of genes recruited into the response as a function of time, which was not seen after sulindac treatment. We interpreted this as an indication of a preprogrammed response to the physiological agent, with each phase of the response triggering a subsequent expansion ( 21). In contrast, sulindac did not trigger this program, and instead initiated a burst of change that did not expand in a regular manner as with butyrate, thereby leading to unexpected side effects and toxicities. Other data support this and will be presented later in the review.
We then repeated this experiment on human subjects, comparing RNA isolated from biopsies taken before and after 1 mo of daily oral treatment with 300 mg of sulindac. For the three subjects investigated, the alterations of gene expression seen with the use of the drug were again complex and highly heterogeneous ( 22). We have no data regarding pharmacokinetics for the subjects in this initial study, and because each subject was investigated only once, it is possible that variability in the array data contribute to the heterogeneity in response. However, we were able to compare the in vivo data (rectal biopsies of the subjects before and after sulindac) and in vitro data (the SW620 colonic carcinoma cell line in tissue culture) to gain insight into the response to sulindac. We first identified a subset of sequences that were altered in expression in vivo, but either not expressed or not altered in vitro ( 22). All of these sequences were genes normally expressed in lymphocytes. Interestingly, all of the sequences were downregulated in the subjects. This is likely due to the activity of sulindac as an antiinflammatory agent, therefore decreasing the number of lymphocytes, and hence lymphocyte expressed genes, in the subject biopsies. These sequences were either not expressed, or not altered in expression by sulindac, in the cells in culture because of the absence of lymphocytes. Thus, this confirms the general reliability of the array analysis to detect true differences in gene expression in biopsy tissue, regardless of etiology.
The more interesting set of genes were those altered in expression similarly by sulindac in both the subjects and the tissue culture cells. This amounted to only eight sequences, 
0.1% of the number on the array (
22). There were a number of interesting genes among this subset, which are under investigation. We first focused on the fact that p21WAF1/cip1, a cyclin dependent kinase inhibitor (cdki), was induced in both the cells in culture and the subjects treated with sulindac. This was of particular interest for several reasons: first, p21 is expressed in the intestinal crypt as cells exit the proliferative compartment, undergoing lineage-specific differentiation concomitant with their migration toward the lumen (
23). Thus, the distribution of expression of p21 in the crypt suggested it plays a roll in cell maturation. Second, the ability of p21 to inhibit cdk activity was consistent with a role in cell maturation. Third, it had already been demonstrated that p21 was induced by sulindac in cells in culture (
24). The question was, therefore, whether the induction of p21 was simply a marker of response to the drug, or important in the mechanism of response?
To address this, Yang et al. turned to mouse genetic models. A p21 knockout mouse had been generated, but other than cell cycle checkpoint defects in mouse embryo fibroblasts derived from these mice, there was no major phenotypic effect of the inactivation of p21 ( 25). We reasoned that p21 might modulate tumor formation initiated by other factors, even though it was insufficient for tumor initiation on its own. We therefore generated mice that were all Apc+/- to initiate tumor formation in the intestinal mucosa, but were either p21+/+, +/- or -/- ( 26). We found that the loss of p21 was effective in increasing tumor formation initiated by Apc, and that this was p21 gene dosage dependent. Moreover, we discovered that the loss of p21 was additive with the tumor-promoting effects of a Western-style diet (high fat and phosphate; low calcium and vitamin D) and this was apparent as additive effects on tumor number, size and decrease in mouse lifespan ( 26). The critical question then was whether the inactivation of p21 would affect the ability of sulindac to inhibit Apc-initiated tumor formation. We first showed, consistent with the work of others, that sulindac was effective in reducing tumor formation in mice that inherit a mutant Apc allele. However, the inactivation of even a single p21 allele completely eliminated the ability of sulindac to decrease small intestinal tumors in these mice ( 22). Yang has subsequently found that the remaining wild-type p21 allele in the flat mucosa of the Apc+/-, p21+/- mice is not induced by sulindac, consistent with the fact that sulindac cannot inhibit tumor formation in these mice (W. C. Yang, unpublished data). Currently, we are investigating whether there are epigenetic and/or genetic changes in the remaining wild-type p21 allele, in either the flat mucosa or in the tumors that develop, which may be responsible for the lack of induction of this allele by sulindac.
Absorptive cell differentiation
One of the principal lineages of epithelial cell differentiation in the intestinal mucosa is the absorptive cell. Although a number of markers that characterize absorptive cells have been described, the overall gene program necessary for this lineage to develop had not been investigated. One of the difficulties in approaching this question is that normal colonic epithelial cells have thus far proved to be refractory to growth in culture. As an alternative, several model systems have been used. In particular, we and others have demonstrated that the Caco-2 human colon carcinoma cell line undergoes cell maturation during a 2- to 3-wk period after contact inhibition of cell growth ( 27). During this time, DNA synthesis and cell division cease and the cells undergo differentiation along the absorptive cell lineage, expressing markers of differentiation such as carcinoembryonic antigen (CEA), sucrase-isomaltase, dipeptidylpeptidase and intestinal fatty acid binding protein ( 28). Mariadason et al. also have shown that this differentiation is partly under the regulation of ß-catenin–TCF signaling, leading to the conclusion that Apc mutations that perturb this signaling pathway cause tumor formation in part by disrupting normal lineages of differentiation in the intestine ( 28). Importantly, the Caco-2 cells do not undergo significant apoptosis during or after their cell cycle arrest and differentiation. This, too, mimics the process in vivo because the levels of frankly apoptotic cells in the intestinal crypts and villi are on the order of only a few percent, indicating that most cells do not undergo, or at least do not complete, apoptosis during the several days it takes them to migrate from the crypt to the lumen ( 29).
As expected, microarray analysis of >17,000 sequences by Mariadason showed extensive changes in gene expression concomitant with Caco-2 cell maturation ( 30). To analyze this, he used a method known as Functional Group Analysis ( 31). In this approach, 26 general functional groups were identified that encompassed the named genes on the microarray. A statistical analysis was then used to determine whether alterations in gene expression were overrepresented in particular functional groups, taking into account both the extent of change in gene expression overall, as well as the number of genes in each functional group.
Mariadason identified several functional groups that showed particularly high enrichment in the number of genes that were significantly altered in expression during Caco-2 cell differentiation ( 30). For example, sequences in the groups that encompassed cell cycle progression and nucleic acid metabolism were generally downregulated, reflecting the decreased cell proliferation and DNA synthesis as the cells matured. In the genes involved in protein translation, many translation initiation factors and tRNA synthetases were downregulated, leading to a progressive decrease in protein synthesis with cell maturation. Coordinate with this drop in protein synthesis, there also was a decrease in expression of genes that encode functions of protein processing, transport and degradation. Genes involved in drug metabolism changed to a large extent, but in this group there were both sequences that increased in expression and others that decreased. Overall, this resulted in increased resistance to effects of a number of chemotherapeutic drugs. However, the more general conclusion may be that as cells mature in the upper part of the crypt and villi of the intestine, they shift their ability to deal with xenobiotics in a number of ways, reflecting their exposure to the mix of environmental stimuli in the colonic lumen. Similarly, there were substantial changes in genes that encode and modify extracellular matrix components, again perhaps reflecting alterations in cell-cell and cell-substratum interactions that accompany the migration of cells as they differentiate. In this group, several genes related to angiogenesis also were altered in expression. This encompassed both up- and downregulated genes, with the overall result being a decrease in the ability of differentiated cells to stimulate endothelial cell migration.
Thus, in addition to the induction of genes that encode specific functions associated with the absorptive cell phenotype, differentiation along this lineage involved extensive and complex changes that shifted the overall physiology of the cells. Underlying this also were complex alterations of genes involved in signaling pathways and transcriptional regulation, some of which are undoubtedly involved in the generation of the more extensive alterations and the coordinate regulation of gene sets ( 30).
Goblet cell lineage
A second major lineage in the intestine is the secretory, or goblet cell, lineage. This lineage may play a major role in colon tumor development, in that aberrant crypt foci, early preneoplastic lesions in the colons of both mice and rats treated with colon carcinogens, mice with an inherited Apc mutation and humans at elevated risk for tumor development, all show a decrease in cells of this lineage and of the mucins they produce and secrete ( 32). Velcich was able to test this hypothesis because work she had done on the structure and regulation of the human and mouse MUC2 gene, the gene that encodes the major colonic mucin that is synthesized and secreted by goblet cells, had provided the reagents to target Muc2 for inactivation ( 33– 35). Velcich generated a mouse that was null for expression of Muc2, and showed that there was no compensatory overexpression of any other of a number of mucin genes ( 36). In these Muc2-/- mice, the intestinal mucosa was devoid of recognizable goblet cells, was distorted in shape, exhibited elevated proliferation and cell migration, and exhibited decreased apoptosis. Most important, in the absence of any other chemical or genetic insult, the mice developed intestinal adenomas and adenocarcinomas in the small and large intestine, and in the rectum ( 36). This was accompanied by deregulation of the c-myc gene, but unlike the Apc model, there was no detectable alteration in ß-catenin–TCF signaling.
Although the Muc2-/- mice do not have recognizable goblet cells, the lineage is not ablated, as demonstrated by the fact that they still express intestinal trefoil factor (Itf), another goblet cell-specific marker, in the same positions in the crypt and at similar intensity as in wild-type mice. To understand the mechanisms responsible for tumor formation in the Muc2-/- mice, it is necessary to understand the extent to which the gene program that is responsible for the lineage is perturbed by the targeted inactivation of Muc2, the gene that encodes the marker of terminal differentiation for this cell type. There may be no, or only modest, alterations beyond the elimination of Muc2, or the changes may be more extensive due to direct effects of inactivation of the gene. Alternatively, there may be a loss of the characteristic shape of the cells due to the absence of mucin, and hence altered cell-cell interactions.
To approach this, Velcich has set out to define the gene program that is responsible for differentiation of this cell type, in much the same way that Mariadason has done for the absorptive cell lineage (above). This investigation of goblet cell lineage makes use of HT29 clone 16E cells, a cell line that differentiates into normal appearing goblet cells upon contact inhibition of cell growth, replete with Muc2 synthesis and secretion ( 37).
Sensitivity to chemotherapy
Intertumor heterogeneity is evident at every level, including pathology, histology, probability of invasive or metastatic behavior and responsiveness to drugs. We postulated that this heterogeneity was a reflection of disparate profiles of gene expression among tumors, and consequently, that gene expression profiling could be an important tool for the clinical management of cancer patients ( 10). The advent of microarray technology provides the tools necessary to develop practical approaches to these questions.
In the case of colorectal cancer, 5-fluorouracil (5FU)-based treatment has been the chemotherapy of choice. However, the development of other effective drugs, such as oxaliplatin and camptothecin, combined with the relatively low rate of response and improved outcome with 5FU, has established the importance of developing informative markers that can guide treatment decisions. We have provided evidence, both from mechanistic studies in vitro and the use of patient material from phase-III clinical trials, that the responsiveness to 5FU is determined principally by two genes. These are c-myc, which must be amplified, coupled with wild-type p53. Both c-myc amplification and p53 function are necessary to establish high sensitivity to 5FU, with absence of either of these conditions rendering the cells insensitive and with a corresponding lack of improved outcome in response to the drug given in an adjuvant setting ( 38– 40). However, both c-myc and p53 are transcription factors that can have highly pleiotropic effects. Thus, it is important to understand how more extensive profiles of gene expression are linked to drug sensitivity and resistance.
To approach this, we have characterized the sensitivity of each of 30 colon carcinoma cell lines to 5FU using three different assays, and 5FU at a spectrum of concentrations. The assays are growth inhibition (measured as the Gi50, the concentration necessary for 50% inhibition of cell growth); extent of induced apoptosis; and clonogenicity expressed as a percent of untreated cells. These data were complemented by determining the gene expression profile for each cell line, each in duplicate, using a >9,000 member cDNA array made by the Albert Einstein core facility (http://sequence.aecom.yu.edu/genome/).
This complex data set was analyzed using a "jackknife" strategy, a statistical method used for crossvalidation when the sample size is relatively small, especially in relation to the number of variables (i.e., >9,000 sequences). In this approach, for each assay and for each drug level, the data for cell line n is dropped and, extracting 10 principal components from a list of the 50 genes (of 9,000) whose expression profile best correlates with response of the remaining 29 lines, a predictor is developed and tested on the nth cell line. The process is repeated 30 times, dropping each of the 30 cell lines in succession, and the correlation of the 30 predicted values is compared to the actual values. Highly statistically significant correlations were obtained for the 5-µM dose of 5FU for each of apoptosis and clonogenicity. The genes that provide the predictor set for the drug treatment are quite distinct for the two different assays, representing 165 and 191 genes for the apoptosis and clonogenicity assays, respectively. As expected from the rigorous statistical method used to identify these gene groups, cluster analysis for these gene subsets demonstrate their ability to accurately distinguish the five most sensitive cell lines from the five most resistant. These gene subsets, as well as others identified by other approaches to the data set, are being analyzed more completely by functional group analysis, real-time quantitative RT-PCR, and as input for development of predictive artificial neural networks. The most important follow-up will be their ability to predict drug response and outcome for patients. A similar analysis is underway for response to a number of other agents, including oxaliplatin and campothecin.
Novel mechanisms of c-myc regulation by nutritional factors
Mutations and inactivations of the APC gene are the initiating event in most human colon cancer (
41). APC encodes a multidomain protein of 300 KD (
42). Although many functions and pathways are influenced by APC, one of the most important in terms of transformation of colonic epithelial cells is the effect of APC on ß-catenin–TCF signaling. APC protein, in conjunction with glycogen synthase kinase 3ß and axin, targets ß-catenin for degradation, thus maintaining proper levels of ß-catenin in the cell cytoplasm. In the absence of functional APC, ß-catenin accumulates, forming a functional complex with the transcription factor TCF-4, which migrates to the nucleus and targets the regulation of a number of genes, including upregulation of c-myc and cyclin D1(
42). These genes likely play an important role in the increased proliferation that is characteristic of the mucosa at elevated risk for tumor development (
43). Butyrate, a short-chain fatty acid that is a physiological regulator of colonic cell maturation, induces a GO/G1 cell cycle arrest, and this is accompanied, as expected, by downregulation of c-myc expression. Sulindac, a nonsteroidal antiinflammatory drug that is highly effective as a chemopreventive agent for colon cancer, also induces what appears to be a similar GO/G1 arrest, but our microarray data showed an increase in c-myc expression, rather than a decrease (
21). Moreover, we showed that both butyrate and sulindac elevated ß-catenin–TCF signaling (
44). This would be incompatible with the fact that c-myc is a direct transcriptional target of this signaling pathway, and that c-myc steady-state levels decreased in response to butyrate.
Therefore, we investigated this issue more closely using a novel method of fluorescent in situ hybridization (FISH) that permits detection of transcription sites for individual genes ( 45, 46). Using fluorescent probes that interrogated the 5' end of c-myc transcripts, and spectrally distinct probes that interrogated the 3' end, Wilson could distinguish between c-myc transcription sites in situ in the nuclei of SW837 colon carcinoma cells that were initiated for mRNA transcription and those in which initiated molecules were fully transcribed ( 47). Consistent with elevation of ß-catenin–TCF signaling stimulated by butyrate and sulindac, both agents elevated the number of nuclei that had active c-myc transcription sites. However, in response to butyrate, the percent of these sites that were successfully transcribed to the 3' end decreased, leading to an overall decrease in c-myc mRNA steady-state levels ( 47). Thus, as has been reported by others ( 48), butyrate recruited a block to transcriptional elongation of the c-myc gene that effectively abrogated the increased transcriptional initiation. In contrast, this block was not recruited in response to sulindac, read through of the gene was efficient and the increased transcriptional initiation led to increased c-myc steady-state levels, as we detected in our microarray experiments.
There are three important implications of this experiment. First, butyrate, a physiological regulator of colonic cell maturation effectively recruited the block to c-myc transcriptional elongation, as did 1,25-dihydroxyvitamin D3, another physiological regulator, but the pharmacological regulator sulindac did not. This may be a specific example of the coordinated and integrated program that the cell has evolved to respond to such physiological regulators to which it is continually exposed, rather than alterations that were triggered in response to recently developed pharmacological agents. This may be one cause of the toxicity of such pharmacological agents. Second, a recent report has demonstrated that the transformed phenotype can be reverted when c-myc expression is even briefly inhibited ( 49). Although this report was in osteogenic sarcoma cells, it suggests that the ability of butyrate and vitamin D-3 to recruit the block to c-myc transcription may be a key pathway by which they can act as chemopreventive agents for colon tumor development. Finally, this methodology provides a new method for gene profiling on a single cell level in situ, in a manner that preserves the architectural integrity of the tissue and the spatial relationships among cells within the tissue. Using spectrally distinct fluors as probes for different genes, multiple genes can be imaged simultaneously on a single cell basis. Moreover, because the spectral characteristics of the emission from the probes hybridized to any transcript site can be analyzed, various combinations of fluors can be used to prepare probes that interrogate dozens, or even hundreds, of genes simultaneously, again, on a single cell basis. Proof of principle for this has recently been published by Levsky et al. ( 46), who detected the transcriptional activity of 11 different genes simultaneously.
Nutritional/genetic modulation of tumor formation
Great advances have been made in our understanding of the underlying molecular and biochemical defects that cause colon cancer. The major genetic risk groups for colon cancer, familial adenomatous polyposis and hereditary nonpolyposis colon cancer, involve mutations in the APC gene and genes involved in DNA mismatch repair, respectively (
41). However, these syndromes account for only 5% of colon cancer incidence in the U.S. Although there may be other genetic factors that will be shown to have a major impact on probability of colon cancer development, a major modulator in the "sporadic" development of colon cancer is diet. The challenge that we face is to understand how dietary factors interact with and modulate the pathways that we already understand are key to tumor development and progression.
We have adopted a strategy that utilizes mouse genetic models of intestinal cancer and their modulation by nutritional factors. An example of this was discussed above, in which Yang demonstrated the modulation of Apc-initiated tumorigenesis by both p21 and a Western-style diet ( 22, 26). We also have embarked on a long-term program in which the interaction of a number of mouse genetic models with nutritional factors will be dissected in detail at the histopathological, biochemical and molecular levels.
The nutritional factors we are focused on are defined by a new Western diet (nwdiet#1) developed by Newmark and Lipkin ( 50– 52). Like their original Western diet (WD), the nwdiet#1 is increased in fat and phosphate, and decreased in calcium and vitamin D. However, the nwdiet#1 also is decreased in methionine, choline, folate and fiber, which should deplete the intracellular methyl donor pool ( 52). Strikingly, although long-term feeding of the original WD to wild-type mice produced premalignant lesions in the intestine, the nwdiet#1 produced frank tumors ( 52). We have done microarray analysis, utilizing a new 27,000 mouse cDNA microarray, with intestinal RNA from each of four wild-type mice in each of the following groups: control AIN76A diet; the nwdiet/#1; and the nwdiet#1 with each of the components added back in separately (i.e., calcium and vitamin D, methionine, choline, folate or fiber). Preliminary analyses of the data demonstrate sets of genes that clearly distinguish the control fed mice from those fed the nwdiet#1. Intriguingly, the addition of calcium and vitamin D back to the nwdiet#1 shifted the profile of expression back toward the control. This is significant because it is clear that the addition of calcium and vitamin D back to the WD was highly effective in shifting tumor formation back toward control diet levels in three mouse genetic models (K. Yang and M. Lipkin, personal communication). Although the tumor experiments are ongoing to determine if this also is true for calcium and vitamin D added back to the nwdiet#1, the gene expression data predict that this will be the case. Moreover, the gene expression data for the mice in which each of the methyl group donors is added back to the nwdiet#1 separately show that each group clusters on a different part of the dendrogram as compared to the control diet, nwdiet#1, or the nwdiet#1 plus calcium and vitamin D. This predicts that the addition of each of these methyl group donors individually will not be effective in reducing tumorigenesis. It will be important to determine how the histopathology of the mucosa and the tumors are altered in each dietary group.
Future directions
Our goal now is to extend these dietary and gene profiling experiments from wild-type mice to each of three genetic models of intestinal tumor formation: the Apc1638+/- mouse, the Muc2-/- mouse, and a new mouse model being developed by Winfried Edelmann of our institution that has a point mutation in the Msh2 gene that mimics a pathogenic allele found in a human HNPCC family. This will provide a matrix of data that will permit us to understand the effects of genetic initiation and diet separately, as well as their interactions. Moreover, with Martin Lipkin and Peter Holt of the Strang Cancer Prevention Center, we will extend these analyses into the effects of calcium and vitamin D on modulation of intestinal gene expression in human subjects maintained in a General Clinical Research Center, where dietary factors can be tightly controlled. Finally, with Judith Christman at the University of Nebraska Medical School, we will pursue the question of how the diets, especially the alteration of methyl donor levels, alter DNA methylation and its link to the altered transcriptional patterns that are seen. These data, coupled with the potential to investigate sets of sequences by the novel transcriptional imaging methodology described above, as well as the potential of investigation of function of important sequences through molecular approaches, mouse genetic models, and the use of a high-throughput structural proteomics program at Albert Einstein Cancer Center, will permit us to make definitive progress on understanding the basis of nutrient-gene interaction in the development of colorectal cancer.
|
FOOTNOTES |
|---|
2 This project was supported in part by grants CA81328, CA88104, CA87559, CA96605, CA90808 and P30 13330 from the National Cancer Institute, and a fellowship to J. M. M. from the American Institute for Cancer Research.
4 Abbreviations used: CEA, carcinoembryonic antigen; FISH, fluorescent in situ hybridization; SAGE, serial analysis of gene expression; SCFA, short chain fatty acid; VDAC, voltage dependent anion channel protein; WD, Western diet.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTLITERATURE CITED |
|---|
1. Augenlicht, L. H. & Kobrin, D. (1982) Cloning and screening of sequences expressed in a mouse colon tumor. Cancer Res. 42: 1088–1093.
2. Augenlicht, L. H., Wahrman, M. Z., Halsey, H., Anderson, L., Taylor, J. & Lipkin, M. (1987) Expression of cloned sequences in biopsies of human colonic tissue and in colonic carcinoma cells induced to differentiate in vitro. Cancer Res. 47: 6017–6021.
3. Heerdt, B. G., Halsey, H. K., Lipkin, M. & Augenlicht, L. H. (1990) Expression of mitochondrial cytochrome c oxidase in human colonic cell differentiation, transformation, and risk for colonic cancer. Cancer Res. 50: 1596–1600.
4. Heerdt, B. G. & Augenlicht, L. H. (1991) Effects of fatty acids on expression of genes encoding subunits of cytochrome c oxidase and cytochrome c oxidase activity in HT29 human colonic adenocarcinoma cells. J. Biol. Chem. 266: 19120–19126.
5. Heerdt, B. G., Houston, M. A. & Augenlicht, L. H. (1994) Potentiation by specific short-chain fatty acids of differentiation and apoptosis in human colonic carcinoma cell lines. Cancer Res. 54: 3288–3294.
6. Heerdt, B. G., Houston, M. A., Rediske, J. J. & Augenlicht, L. H. (1996) Steady-state levels of mitochondrial messenger RNA species characterize a predominant pathway culminating in apoptosis and shedding of HT29 human colonic carcinoma cells. Cell Growth Differ. 7: 101–106.[Abstract]
7. Heerdt, B. G., Houston, M. A. & Augenlicht, L. H. (1997) Short chain fatty acid-initiated cell cycle arrest and apoptosis of colonic epithelial cells is linked to mitochondrial function. Cell Growth Differ. 8: 523–532.[Abstract]
8. Heerdt, B. G., Houston, M. A., Anthony, G. M. & Augenlicht, L. H. (1998) Mitochondrial membrane potential in the coordination of p53-independent proliferation and apoptosis pathways in human colonic carcinoma cells. Cancer Res. 58: 2869–2875.
9. Augenlicht, L. H. (1988) Gene expression in human colonic biopsies. In: Basic and Clinical Perspectives of Colorectal Polyps and Cancer (Steele, G., Burt, R., Winawer, S. J. & Karr, J. P., eds.), pp. 195–202. Alan R. Liss, New York.
10. Augenlicht, L. H. (1989) Gene structure and expression in colon cancer. In: Cell and Molecular Biology of Colon Cancer, 1st ed. (Augenlicht, L. H., ed.), pp. 165–186. CRC Press, Boca Raton, FL.
11. Augenlicht, L. H. & Heerdt, B. G. (1992) Modulation of gene expression as a biomarker in colon. J. Cell. Biochem. 16G: 151–157.
12. Augenlicht, L. H., Corner, G., Molinas, S. & Heerdt, B. G. (1992) Genetic biomarkers. In: Cancer Chemoprevention (Wattenberg, L., Lipkin, M., Boone, C. W. & Kelloff, G. J., eds.), pp. 559–569. CRC Press, Boca Raton, FL.
13. Augenlicht, L. H., Velcich, A. & Heerdt, B. G. (1995) Short chain fatty acids and molecular and cellular mechanisms of colonic cell differentiation and transformation. In: Diet and Cancer: Molecular Mechanisms of Interactions (AICR, ed.), pp. 137–148. Plenum Press, New York.
14. Augenlicht, L. H., Taylor, J., Anderson, L. & Lipkin, M. (1991) Patterns of gene expression that characterize the colonic mucosa in patients at genetic risk for colonic cancer. Proc. Natl. Acad. Sci. USA 88: 3286–3289.
15. Schena, M., Shalon, D., Davis, R. W. & Brown, P. O. (1995) Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 270: 467–470.
16. Zweiger, G. (2000) Transducing the Genome: Information, Anarchy and Revolution in the Biomedical Sciences, pp. 149–160. McGraw-Hill, New York.
17. Shirasawa, S., Furuse, M., Yokoyama, N. & Sasazuki, T. (1993) Altered growth of human colon cancer cell lines disrupted at activated Ki-ras. Science 260: 85–88.
18. Gibson, P. R., Moeller, I., Kagelari, O., Folino, M. & Young, G. P. (1992) Contrasting effects of butyrate on the expression of phenotypic markers of differentiation in neoplastic and non-neoplastic colonic epithelial cells in vitro. J. Gastroenterol. Hepatol. 7: 165–172.[Medline]
19. Kwiatkowski, D. J. (1999) Functions of gelsolin: motility, signaling, apoptosis, cancer. Curr. Opin. Cell Biol. 11: 103–108.[Medline]
20. Subbaramaiah, K., Zakim, D., Weksler, B. B. & Dannenberg, A. J. (1997) Inhibition of cyclooxygenase: a novel approach to cancer prevention. Proc. Soc. Exp. Biol. Med. 216: 201–210.[Abstract]
21. Mariadason, J. M., Corner, G. A. & Augenlicht, L. H. (2000) 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.
22. Yang, W. C., Velcich, A., Mariadason, J., Nicholas, C., Corner, G., Houston, M., Edelmann, W., Kucherlapati, R., Holt, P. & Augenlicht, L. H. (2001) p21WAF1/cip1 is an important determinant of intestinal cell response to sulindac in vitro and in vivo. Cancer Res. 61: 6297–6302.
23. El-Deiry, W. S., Tokino, T., Waldman, T., Oliner, J. D., Velculescu, V. E., Burrell, M., Hill, D. E., Healy, E., Rees, J. L., Hamilton, S. R., Kinzler, K. W. & Vogelstein, B. (1995) Topological control of p21 waf1/cip1 expression in normal and neoplastic tissues. Cancer Res. 55: 2910–2919.
24. Shiff, S. J., Qiao, L., Tsai, L.-L. & Rigas, B. (1995) Sulindac sulfide, an aspirin-like compound, inhibits proliferation, causes cell cycle quiescence, and induces apoptosis in HT-29 colon adenocarcinoma cells. J. Clin. Invest. 96: 491–503.
25. Deng, C., Zhang, P., Harper, J. W., Elledge, S. J. & Leder, P. (1995) Mice lacking p21 cip1/waf1 undergo normal development, but are defective in G1 checkpoint control. Cell 82: 675–684.[Medline]
26. Yang, W. C., Mathew, J., Velcich, A., Edelmann, W., Kucherlapati, R., Lipkin, M., Yang, K. & Augenlicht, L. H. (2001) Targeted inactivation of the p21 WAF1/cip1 gene enhances Apc initiated tumor formation and the tumor promoting activity of a Western-style high risk diet by altering cell maturation in the intestinal mucosa. Cancer Res. 61: 565–569.
27. Mariadason, J. M., Rickard, K. L., Barkla, D. H., Augenlicht, L. H. & Gibson, P. R. (2000) Divergent phenotypic patterns and commitment to apoptosis of Caco-2 cells during spontaneous and butyrate-induced differentiation. J. Cell. Physiol. 183: 347–354.[Medline]
28. Mariadason, J. M., Bordonaro, M., Aslam, F., Shi, L., Kuraguchi, M., Velcich, A. & Augenlicht, L. H. (2001) Down-regulation of B-catenin-TCF signaling is linked to colonic epithelial cell differentiation. Cancer Res. 61: 3465–3471.
29. Augenlicht, L. H., Anthony, G. M., Chruch, T. L., Edelmann, W., Kucherlapati, R., Yang, K. Y., Lipkin, M. & Heerdt, B. G. (1999) Short chain fatty acid metabolism, apoptosis and Apc initiated tumorigenesis in the mouse gastrointestinal mucosa. Cancer Res. 59: 6005–6009.
30. Mariadason, J. M., Arango, D., Corner, G. A., Aranes, M. J., Hotchkiss, K. A., Yang, W. C. & Augenlicht, L. H. (2002) A gene expression profile that defines colon cell maturation in vitro. Cancer Res. 62: 4791–4804.
31. Muller, H., Bracken, A. P., Vernell, R., Moroni, M. C., Christians, F., Grassilli, E., Prosperini, E., Vigo, E., Oliner, J. D. & Helin, K. (2001) E2Fs regulate the expression of genes involved in differentiation, development, proliferation, and apoptosis. Genes Dev. 15: 267–285.
32. Pretlow, T. P., Siddiki, B., Augenlicht, L. H., Pretlow, T. G. & Kim, Y. S. (1999) Aberrant crypt foci (ACF) - earliest recognized players or innocent bystanders in colon carcinogenesis. In: Colorectal Cancer: Molecular Mechanisms, Premalignant State, and its Prevention (Schmiegel, W., ed.) Kluwer Academic Publishers, Lancaster, England.
33. Velcich, A. & Augenlicht, L. H. (1993) Regulated expression of an intestinal mucin gene in HT29 colonic carcinoma cells. J. Biol. Chem. 268: 13956–13961.
34. Velcich, A., Palumbo, L., Jarry, A., Laboisse, C., Rachevskis, J. & Augenlicht, L. (1995) Patterns of expression of lineage specific markers during the in vitro differentiation of HT29 colon carcinoma cells. Cell Growth Differ. 6: 749–757.[Abstract]
35. Velcich, A., Palumbo, L., Selleri, L., Evans, G. & Augenlicht, L. (1997) Organization and regulatory aspects of the human intestinal mucin gene (MUC2) locus. J. Biol. Chem. 272: 7968–7976.
36. Velcich, A., Yang, W. C., Heyer, J., Fragale, A., Nicholas, C., Viani, S., Kucherlapati, R., Lipkin, M., Yang, K. & Augenlicht, L. (2002) Colorectal cancer in mice genetically deficient in the mucin Muc2. Science 295: 1726–1729.
37. Augeron, C. & Laboisse, C. L. (1984) Emergence of permanently differentiated cell clones in a human colonic cancer cell line in culture after treatment with sodium butyrate. Cancer Res. 44: 3961–3969.
38. Arango, D., Corner, G. & Augenlicht, L. H. (2000) Low-level amplification of c-myc in colon cancer cell lines can result in a significant growth advantage. Proceedings of the American Association for Cancer Research Meeting, San Francisco, CA.
39. Arango, D., Corner, G. A., Wadler, S., Catalano, P. J. & Augenlicht, L. H. (2001) c-myc/p53 interaction determines sensitivity of human colon carcinoma cells to 5-fluorouracil in vitro and in vivo. Cancer Res. 61: 4910–4915.
40. Augenlicht, L. H., Wadler, S., Corner, G., Richards, C., Ryan, L., Multani, A. S., Pathak, S., Benson, A., Haller, D. & Heerdt, B. G. (1997) Low-level c-myc amplification in human colonic carcinoma cell lines and tumors: a frequent, p53-independent mutation associated with improved outcome in a randomized multi-institutional trial. Cancer Res. 57: 1769–1775.
41. Kinzler, K. W. & Vogelstein, B. (1996) Lessons from hereditary colorectal cancer. Cell 87: 159–170.[Medline]
42. van Es, J. H., Giles, R. H. & Clevers, H. C. (2001) The many faces of the tumor suppressor gene APC. Exp. Cell Res. 264: 126–134.[Medline]
43. Lipkin, M., Blattner, W. E., Fraumeni, J. F., Lynch, H. T., Deschner, E. & Winawer, S. (1983) Tritiated thymidine (0p,0h)labeling distribution as a marker for hereditary predisposition to colon cancer. Cancer Res. 43: 1899–1904.
44. Bordonaro, M., Mariadason, J. M., Aslam, F., Heerdt, B. G. & Augenlicht, L. H. (1999) Butyrate induced cell cycle arrest and apoptotic cascade in colonic carcinoma cells: modulation of the B-catenin-Tcf pathway, and concordance with effects of sulindac and trichostatin, but not curcumin. Cell Growth Differ. 10: 713–720.
45. Femino, A. M., Fay, F. S., Fogarty, K. & Singer, R. H. (1998) Visualization of single RNA transcripts in situ. Science 280: 585–590.
46. Levsky, J. M., Shenoy, S. M., Pezo, R. C. & Singer, R. H. (2002) Single cell gene expression profiling. Science 297: 836–840.
47. Wilson, A. J., Velcich, A., Kurland, A., Shenoy, S. M., Pezo, R. C., Levsky, J. M., Singer, R. H. & Augenlicht, L. H. (2002) Novel detection and differential utilization of a c-myc transcriptional block in colon cancer chemoprevention. Cancer Res. 62: 6006–6010.
48. Heruth, D. P., Zirnstein, G. W., Bradley, J. F. & Rothberg, P. G. (1993) Sodium butyrate causes an increase in the block to transcriptional elongation in the c-myc gene in SW837 rectal carcinoma cells. J. Biol. Chem. 268: 20466–20472.
49. Jain, M., Arvanitis, C., Chu, K., Dewey, W., Leonhardt, E., Trinh, M., Sundberg, C. D., Bishop, J. M. & Felsher, D. W. (2002) Sustained loss of a neoplastic phenotype by brief inactivation of MYC. Science 297: 102–104.
50. Newmark, H. L., Lipkin, M. & Maheshwari, N. (1990) Colonic hyperplasia and hyperproliferation induced by a nutritional stress diet with four components of western-style diet. J. Natl. Cancer Inst. 82: 491–496.
51. Newmark, H. L., Lipkin, M. & Maheshwari, N. (1991) Colonic hyperproliferation induced in rats and mice by nutritional-stress diets containing four components of a human western-style diet (series 2). Am. J. Clin. Nutr. 54: 209S–214S.
52. Newmark, H. L., Yang, K., Lipkin, M., Kopelovich, L., Liu, Y., Fan, K. & Shinozaki, H. (2001) A western-style diet induces benign and malignant neoplasms in the colon of normal C57Bl//6 mice. Carcinogenesis 22: 1871–1875.
This article has been cited by other articles:
|
|
 |
|
  K. Schror, P. Mehta, and J. L. Mehta Cardiovascular Risk of Selective Cyclooxygenase-2 Inhibitors Journal of Cardiovascular Pharmacology and Therapeutics, April 1, 2005; 10(2): 95 - 101. [Abstract] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
