| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Molecular Oncology, Markers, Clinical Correlates |
Simmons Cancer Center [N. S. W., R. B. G., S. S., D. T.], and Departments of Medicine [R. B. G., U. V., E. F., C. B.], Pathology [T. G.], and Surgery [C. S., J. F.], University of Texas Southwestern Medical Center, Dallas, Texas 75390
 |
ABSTRACT |
|---|
 Top ABSTRACT IntroductionMaterials and MethodsResultsDiscussionREFERENCES |
|---|
Experimental Design: cDNA microarray analysis was used to detect differences in gene expression between normal tissue and colon tumors and polyps isolated from 20 patients. To identify genes that are important in regulating the growth properties of colorectal cancer, RNA interference (RNAi) was used to disrupt expression of several of the overexpressed genes in a colon tumor cell line, HCT116, which showed similar patterns of gene expression as many of the patient tumors.
Results: Expression changes of 
2-fold in approximately one-third of the patients were consistently observed for 2632 of a total of 9592 genes (574 up-regulated genes and 2058 down-regulated genes). Subsequent analysis of 13 genes by quantitative real-time PCR confirmed the reliability of this analysis. RNAi-mediated disruption of the expression of one of these genes, survivin, a potent inhibitor of apoptosis, severely reduced tumor growth both in vitro and in an in vivo xenograft model.
Conclusions: The combined use of microarray analysis and RNAi provides an excellent system to define the role of specific genes that are up-regulated in cancer lead to the increased in vitro and in vivo growth of colon tumors.
|
Introduction |
|---|
|
TopABSTRACTIntroductionMaterials and MethodsResultsDiscussionREFERENCES |
|---|
135,000 patients diagnosed each year (1)
. At least 15% of CRCs occur in dominantly inherited patterns (2)
, with the two best-defined familial forms being FAP and hereditary nonpolyposis CRC. Insight into the genetic changes that give rise to these two familial forms of cancer has led to a better understanding of the changes that lead to sporadic colon cancer. CRC typically develops over decades and involves multiple genetic events. This has led to the development of a multistep model of colorectal tumorigenesis (3)
. It is generally believed that one of initiating steps in colorectal carcinogenesis is mutation in the APC tumor suppressor gene (4
, 5)
. APC mutations, which generally lead to a truncated protein or loss of one allele, are detected in about 75–80% of sporadic cancers and are the cause of FAP (2
, 3)
. APC is known to bind to a cell signaling/transcription factor molecule, ß-catenin. Activation of the Wnt pathway normally signals the association of ß-catenin with members of the T-cell factor/lymphocyte-enhancer factor family and translocation to the nucleus (6)
. This complex can activate the transcription of a variety of target genes including c-myc (7)
and cyclin D1 (8
, 9)
. Loss of APC function leads to an abnormal accumulation of ß-catenin and dysregulated Wnt signaling (10)
. There is strong evidence to suggest that a distinct pathway associated with MSI can activate another type of CRC. Germ-line defects in one of several mismatch repair genes are associated with the dominantly inherited syndrome hereditary nonpolyposis CRC (2) . About 15% of sporadic tumors also show high levels of MSI. Although MSI tumors have mutations in APC, they tend to be less frequent and of a different nature than tumors that are microsatellite stable (11) . For example, frameshift mutations in APC are more common in MSI tumors, especially those with high-level instability, than in MSS tumors. Whether initiated by APC or MSI, both types of colorectal tumors require additional "hits" for a carcinoma to develop. Although the exact nature of these additional mutations seems to vary between tumors initiated by the two different mechanisms, they include additional defects in K-ras, p53, and DCC (2) . Finally, a number of additional mutations in genes involved in cell adhesion, such as E-cadherin (12) , and in tissue degradation, such as urokinase plasminogen activators and matrix metalloproteinases (13) , are likely important in metastasis.
Despite this information, precise knowledge of how these genetic alterations lead to the development and progression of colorectal carcinomas remains to be resolved. It is likely that a number of downstream targets of these mutated genes are actually responsible for the proliferative advantage and survival of the cells that give rise to colorectal tumors. Thus, there is a clear need for identification of these downstream targets to develop better strategies for management of colorectal neoplasms.
Traditional methods of identifying novel targets that are involved in colon cancer progression are based on studies of individual genes. The development of oligonucleotide or cDNA microarrays has made such traditional methods obsolete, as they permit the expression levels of tens of thousands of genes to be monitored simultaneously and rapidly (14 , 15) . Microarray data analysis can be broadly divided into two tasks: grouping of genes to discover broad patterns of biological behavior and filtering of genes to identify specific genes of interest (16) . Most studies of malignancy have used microarrays for the former task to better characterize expression patterns in the neoplasm or for prediction of clinical outcome. For example, microarrays have been used to identify distinct types of diffuse large B-cell lymphomas (17) , to confirm and extend the morphological classification of human lung tumors (18) , to predict the prognosis of breast cancer patients and to determine which will benefit from adjuvant therapy (19) , to better classify and distinguish acute myeloid leukemia and acute lymphoblastic leukemia (20) , and to better understand the changes in gene expression in the progression from normal tissue through colorectal adenoma and adenocarcinoma (21 , 22) . Microarrays have also been used to identify genes and expression profiles correlated with colorectal metastasis in three paired cell line models of colorectal tumor metastasis (23) . Only a few studies have made novel target identification their goal because of the difficulty in rapidly identifying the function and/or importance of individual genes. For example, microarray analysis of human melanoma cell lines with low or high metastatic potential revealed that metastatic melanomas overexpress a number of genes, including thymosin ß4, fibronectin, and RhoC. Introduction of RhoC was demonstrated to enhance the metastatic behavior of poorly metastatic human melanoma cell lines (24) .
One technique that has been used recently to rapidly and efficiently identify the function of individual genes is RNAi (25, 26, 27) . This process is evolutionarily conserved in plants, worms, and flies and has been demonstrated recently to function in mammalian cells (28) . The use of RNAi in mammalian cells involves the transfection of an annealed 21-mer of sense and antisense RNA oligonucleotides (siRNAs) corresponding to a portion of a gene of interest. These RNAs then bind specifically to the cellular RNA and activate a process that leads to degradation of the mRNA and a subsequent 80–90% decrease in the levels of the corresponding protein. Thus, RNAi provides an important technique to specifically down-regulate the expression of cellular genes.
To identify relevant target genes that are important in the pathogenesis of CRC, we used microarrays to identify genes differentially expressed in malignant versus normal samples isolated from individual patients with surgically resected colon polyps and tumors. We additionally investigated whether RNAi-mediated reduction in the levels of genes identified by the microarray analysis, in an established colon cancer cell line, might result in decreased proliferation both in culture and when implanted in nude mice. This cell line, like the patient tumors, was shown to overexpress the selected genes as compared with normal colon tissue. These studies allowed us to rapidly identify one new target, which when disrupted, significantly affected tumor growth in vivo and in vitro. Furthermore, our results suggested that the combination of microarrays and RNAi is a powerful tool for the identification of genes important in the pathogenesis of CRC and is, thus, an efficient means of identifying novel therapeutic targets.
|
Materials and Methods |
|---|
|
TopABSTRACTIntroductionMaterials and MethodsResultsDiscussionREFERENCES |
|---|
Cell Lines.
The colon cancer adenocarcinoma cell line, HCT116 (29)
, was obtained from the American Type Culture Collection (Manassas, VA). These cells were propagated in DMEM (Life Technologies, Inc., Rockville, MD), supplemented with 10% fetal bovine serum (Life Technologies, Inc.), 2 mM L-glutamine, 50 units/ml of penicillin, and 50 µg/ml of streptomycin (all from Life Technologies, Inc.). Total RNA was extracted in the same manner as from tissue samples.
cDNA Arrays.
Microarrays were constructed from a clone set purchased from Research Genetics (Huntsville, AL). A total of 9592 individual clones from Research Genetics Human Sequence Verified Set, plates 1–100, were represented on the array. Clones were grown overnight in 96-well microtiter plates; plasmid DNA was isolated using the Qiagen Real Prep 96 kit; clone inserts were amplified using vector specific primers (Universal Forward 5'-CTGCAAGGCGATTAAGTTGGGTAAC-3' and Universal Reverse 5'-GTGAGCGGATAACAATTTCACACAGGAAACAGC-3'); and the products were precipitated and then resuspended in a 7% solution of DMSO. Clones were printed on lysine-coated plates using a custom-built arraying robot (MAGNA spotter). Printed slides were cross-linked by UV irradiation and postprocessed by conventional methods (15)
.
Hybridization and Analysis.
Twenty-five µg of RNA from matched tumor/normal sample were reverse transcribed into cDNA using Superscript II (Invitrogen, Carlsbad, CA) with oligodeoxythymidylic acid primers (Invitrogen), and cy5- (Tumor) and cy3- (Normal) labeled dCTP (Amersham, Piscataway, NJ). After purification through a microcon 30 spin column, the cDNAs were combined and hybridized to the cDNA array in a solution containing 10 µg of Cot-1 DNA (Invitrogen), 32 µg of tRNA (Sigma, St. Louis, MO), 10 µg of PolyA RNA (Sigma), 4x SSC, and 0.25% SDS. After an overnight hybridization at 62°C, slides were washed, scanned using a GenePix 4000B Scanner (Axon Instruments, Inc., Union City, CA), and analyzed by GenePix Pro software (Axon Instruments, Inc.). Data were normalized using the software program, MarC-V, developed at University of Texas Southwestern Medical Center (30)
to control for uneven incorporation of dyes. In brief, a lower-bound thresholding calculation is first performed such that no background subtracted intensity value is less than a threshold value. This threshold is calculated for both the Cy3 and Cy5 channels based on many internally replicated negative controls, which consist of DMSO solvent alone. Once the threshold has been applied, ratios are computed for each element and log transformed (base 10). Then, a ratio normalization step is performed. The assumption in this calculation is that for a given array of genes, there will be a subset of expression ratios that will not vary far from 1 (or 0 in log space). Each log ratio is normalized by multiplying each denominator by a normalization coefficient defined as 10 raised to the mean log ratio of the entire array. Finally, a confidence score based on spot intensity is calculated for each gene. Microarray data from the 20 patients were maintained in a Microsoft Excel spreadsheet. Before additional analysis, the data were filtered to exclude genes with low intensities as determined by the confidence score. Essentially, genes of which the background-subtracted value did not exceed at least 4 SDs above the mean blank in one channel and 2 SDs above the mean blank in the other channel were excluded (confidence score of 36% in MarC-V). The data were then sorted to obtain genes differentially expressed by 2-fold in at least 6 of the 20 patients. The sequence of selected genes was verified before continued analysis using a vector-specific primer (Universal Reverse).
Quantitative Real-Time PCR.
For validation of array results by quantitative PCR, cDNA was prepared from the HCT116 cell line, and each tumor and normal pair using a combination of oligodeoxythymidylic acid, and random primers and Superscript II (all from Invitrogen). Oligos were designed around intron-exon boundaries for each gene using Primer Express software (Applied Biosystems, Foster City, CA). Each PCR was carried out in triplicate in a 20-µl volume using Sybr Green Mastermix (Applied Biosystems) for 15 min at 95°C for initial denaturing, followed by 40 cycles of 95°C for 30 s and 60°C for 1 min in the ABI Prism 7700 Sequence Detection System. Each primer set was first tested to determine optimal concentrations, and products were run out on a 3% agarose gel to confirm the appropriate size. Subsequently, the ABI Dissociation Curves software was used following a brief thermal protocol (95°C 15 s and 60°C 20 s, followed by a slow ramp to 95°C) to control for multiple species in each PCR amplification. cDNA prepared from Universal RNA (Stratagene, La Jolla, CA) was used to construct a standard curve for each gene. Values for each gene were normalized to expression levels of 18S, and then a ratio comparing expression in tumor versus normal was calculated. The sequences of the primers used for RT-PCR were as follows: 18S forward, 5'-AGGAATTGACGGAAGGGCAC-3', reverse, 5'-GGACATCTAAGGGCATCACA-3; HMGIY forward, 5'-GAAGGAGCCCAGCGAAGTG-3', reverse, 5'-TTCTCCAGTTTTTTGGGTCTGC-3'; c-myc forward, 5'-CAGCTGCTTAGACGCTGGATT-3', reverse, 5'-GTAGAAATACGGCTGCACCGA; CTP synthetase forward, 5'-TGCAGTTGGCAGTGGTTGA-3', reverse, 5'-TGTCTACGACCACGGGATGA; survivin forward, 5'-TCCGGTTGCGCTTTCCT-3', reverse, 5'-TCTTCTTATTGTTGGTTTCCTTTGC-3'; B-myb forward, 5'-AGAGGGATAGCAAGTGCAAGGT-3', reverse, 5'-TGTACTGGCATTGCTGGTCAGT-3'; ATDC forward, 5'-CAAGGACGACCTGCTCAATGT-3', reverse, 5'-CGATGGTCACCACCGTTCTC-3'; EB1 forward, 5'-GGCTCCTTCCCTTGTTGCTC-3', reverse, 5'-CGTCTCCGTTGCCCACAC-3'; BRCA2 forward, 5'-GCGCGGTTTTTGTCAGCTTA-3', reverse, 5'-TGGTCCTAAATCTGCTTTGTTGC-3'; MAPK3 forward, 5'-CTTCCCTGGCAAGCACTACC-3', reverse, 5'-GTTTCGGGCTTCATGTTGA-3'; carbonic anhydrase II forward, 5'-AAACAAAGGGCAAGAGTGCTG-3', reverse, 5'-AGTGAGCCTGGGTAGGTCCA-3'; biliary glycoprotein forward, 5'-GAACCAAAGCGACCCCATC-3', reverse, 5'-CCAATCACAATGCCAGCAAT-3'; TRAIL forward, 5'-CGTGTACTTTACCAACGAGCTGA-3', reverse, 5'-ACGGAGTTGCCACTTGACTTG-3'; and ADAMTS1 forward, 5'-AGTGCCTACATGATTACATCATTTCTG-3', reverse, 5'-AGGGAGATCGCCTGGGAG-3'.
RNA Oligonucleotides.
siRNA oligonucleotides with two thymidine residues (dTdT) at the 3' end of the sequence were designed for c-myc (sense, 5'-AACAGAAAUGUCCUGAGCAAU-3'), survivin (sense, 5'-AAGCAUUCGUCCGGUUGCGCU-3''), and HTLV-1 tax (sense, 5'-GAUGGACGCGUUAUCGGCU-3'). These oligonucleotides along with their corresponding antisense oligonucleotides were synthesized (Dhamacon Research Inc., Lafayette, CO), dissolved in 2'-deprotection buffer supplied by the manufacturer, combined, annealed at 60°C for 45 min then ambient temperature for 30 min, precipitated, and finally resuspended in 1x annealing buffer [30 mM HEPES-KOH (pH 7.9), 100 mM potassium acetate, and 2 mM magnesium acetate] at 20 µM.
Transfection of RNA Oligonucleotides.
Approximately 105 cells were plated per well in six-well plates in DMEM supplemented with 10% FBS, glutamine, and Pen/Strep. The following day, siRNAs were transfected using Oligofectamine (Invitrogen) to result in a final RNA concentration of 100 nM per well. The cells were harvested at different time points and lysed, and Western blot analysis was performed.
Western Blot Analysis.
Cells were lysed [40 mM Tris-HCl (pH 8.0), 500 mM sodium chloride, 0.1% NP40, 6 mM EDTA, 6 mM EGTA, 5 mM sodium fluoride, 1 mM sodium orthovanadate, and 5 mM ß-glycerophosphate], and Western blot analysis was performed following conventional protocols. The antibodies and dilutions used included anti-c-myc (Santa Cruz Biotechnology, Santa Cruz, CA) at 1:400, antisurvivin (Santa Cruz) at 1:250, and anti-ß-Tubulin (Sigma) at 1:1000. After extensive washing, the membranes were incubated with antimouse or antirabbit IgG-horseradish peroxidase-conjugated antibody (Amersham) at a 1:1000 dilution and developed using enhanced chemiluminescence (Amersham).
Cellular Proliferation Assays.
At 24 h after RNA transfection, HCT116 cells transfected with the indicated siRNA oligonucleotides were harvested and replated in 96-well plates in quadruplicate. The cells were then incubated with 200 µl of tissue culture medium overnight and pulsed with 1 µCi/well of [3H]thymidine (Perkin-Elmer, Boston, MA) for 16 h the next day. 3H-labeled DNA was counted on a Beckman (Fullerton, CA) LS 6000 liquid scintillation ß counter.
Cell Cycle Analysis.
At 48, 72, and 96 h after RNA transfection, HCT116 cells were labeled with 15 µl/well of a 1 mM solution of BrdUrd for 45 min. The cells were then harvested, permeabilized, and stained with a FITC-labeled antibody to BrdUrd followed by 7AAD according to instructions provided with the BrdUrd Flow kit (BD Biosciences, San Jose, CA). Cells were analyzed on a Becton Dickinson (San Jose, CA) FACScan instrument using CellQuest software (Becton Dickinson). Propidium iodide (5 µg/ml) staining was performed on cells harvested at 48, 72, and 96 h, and permeabilized with 70% ethanol overnight at -20°C. The cells were analyzed as for the BrdUrd/7AAD stain.
Soft Agar Colony Assay.
At 24 h after RNA transfection, the cells were mixed with tissue culture medium containing 0.6% agar to result in a final agar concentration of 0.4%. Then 1 ml of this cell suspension was immediately plated in six-well plates coated with 1 ml/well of 0.6% agar in tissue culture medium. The colonies were counted in triplicate wells 12 days after plating, and the number of colonies per 104 cells was calculated.
Murine Xenograft Model.
Institutional guidelines and an Institutional Animal Care and Research Advisory Committee approved protocol were followed for mouse studies. Four to 6-week-old male nudenu/nu mice were obtained from Taconic (Germantown, NY) and housed in clean specific pathogen-free rooms in groups of 4. At 24 h after RNA transfection, the cells were harvested, washed twice in ice-cold serum-free DMEM, and counted. The cells were resuspended in the same medium at 107 cells/ml, and mice were injected with 2.5 x 106 cells s.c. into the left or right flank. The tumors were measured in three axes from day 7 onwards, and the tumor volume was calculated from these measurements. As per institutional requirements, the mice were euthanized once tumors grew to >2 cm in diameter or developed necrosis.
Statistical Analysis.
SAS (SAS Institute, Inc., Cary, NC) was used to analyze the data. A one-way ANOVA procedure followed by Dunnett’s t test was used to compare the statistical significance of each of the experimental siRNA treatments versus the control tax siRNA treatment where appropriate.
|
Results |
|---|
|
TopABSTRACTIntroductionMaterials and MethodsResultsDiscussionREFERENCES |
|---|
. Array data were normalized to correct for differences in the incorporation of label through the use of the MarC-V algorithm developed at University of Texas Southwestern Medical Center for this purpose (30)
and described in more detail in "Materials and Methods." In brief, a lower-bound thresholding calculation was performed such that no background subtracted intensity value was less than a threshold value. Once the threshold was applied, ratios were computed for each element and log transformed (base 10). Then, a ratio normalization step was performed that controlled for unequal incorporation of dyes. Finally, a confidence score based on spot intensity was calculated for each gene. Microarray data from the 20 patients were maintained in a Microsoft Excel spreadsheet. Because uncertainty of the array measurements will decrease as the intensity of each spot increases, the data were first filtered to exclude genes with low intensities as determined by the confidence score. Essentially genes of which the background-subtracted value on average among the 20 patients did not exceed at least 4 SDs above the mean blank in one channel and 2 SDs above the mean blank in the other channel were excluded. This reduced the total number of genes from which reliable expression data could be obtained to 6087.
|
2-fold up- or down-regulated in approximately one-third (6 of 20) of the patients. By these criteria, there were 574 commonly up-regulated genes and 2058 commonly down-regulated genes. The genes found to be up- or down-regulated in the largest number of patients are listed in Tables 2
and 3
.
|
|
|
presents the results of the PCR analysis for these genes. The number of patients showing 2-fold up- or down-regulation for each gene is presented. The results of RT-PCR showed changes in gene expression consistent (changes of 2-fold either up or down were considered up- or down-regulated, respectively) with microarray data in 166 of the 255 samples (65%). The lack of total concordance was attributed to variation in the array data, because these experiments were only carried out a single time, whereas the RT-PCR data were obtained in triplicate. Nevertheless, for all 13 of the genes, at least one-third or more of the patients were again shown to over- or underexpress each gene.
|
The data in Table 4
show that all 13 of the dysregulated genes validated in the patients showed a similar pattern of altered expression in HCT116 as compared with normal colon tissue. The known oncogene, c-myc, and the inhibitor of apoptosis protein, survivin, were chosen for additional analysis using RNAi. The c-myc gene is a known downstream target of ß-catenin, a protein of which the abnormal accumulation has been shown to be involved in colorectal carcinogenesis (7)
. Survivin was identified by hybridization screening of a human P1 genomic library with the cDNA of effector cell protease receptor-1 (34)
and is known to be overexpressed in multiple types of tumors (34, 35, 36)
. In several studies, loss of survivin function results in apoptosis in the affected cells (37, 38, 39)
. The pattern of expression of c-myc and survivin in tumor versus normal tissues for the 20 patients examined in this study as well as HCT116 are presented in Fig. 1
. Eleven of the patients overexpress (2-fold) c-myc (range, 2.0–9.0-fold), and 15 overexpress survivin (range, 2.0–7.5-fold).
|
. It is clear from these results that siRNA directed against c-myc and survivin reduces the levels of these proteins without affecting the levels of a control protein, ß-tubulin. Furthermore, transfection of siRNA oligonucleotides directed against both genes was as effective in reducing protein expression, as were experiments targeting each gene separately.
|
.
|
. This is consistent with the known role of c-myc in promoting entry to the cell cycle (41)
. siRNA against survivin resulted in a more substantial decrease in the number of cells entering S phase (Fig. 4)
. This is likely attributable in part to the large increase in cells showing hyperdiploid DNA content in the presence of survivin siRNA. Survivin is known to localize to the centrosomes and is thought to help regulate centrosome duplication during mitosis. The loss of survivin expression has been shown previously to result in abnormal centrosome duplication, and the formation of multipolar spindles and polyploidy (39)
.
|
. Interestingly, HCT116 cells treated with siRNA against both c-myc and survivin showed a lower frequency of apoptotic cells. These data have been additionally confirmed using a fluorometric caspase-3 enzyme assay that demonstrated significant levels of caspase activity only in survivin-treated cells (data not shown).
|
. The mean ± SE from three separate experiments is plotted here. siRNA directed against survivin, but not c-myc or c-myc and survivin, led to a significant decrease in colony formation as compared with siRNA directed against tax.
|
at the 3-week point. There was an 7-fold decrease in the average size of the HCT116 tumors derived from cells that were transfected with siRNA against survivin as compared with HCT116 cells that were transfected with siRNA directed against tax. Reductions in the size of HCT116 tumors derived from cells transfected with siRNA against c-myc or c-myc and survivin together resulted in more modest decreases in size (26% and 39% reduction, respectively). Only the reduction in tumor growth seen with survivin siRNA was significant at the P = 0.05 levels as measured by a one-way ANOVA followed by a Dunnet’s t test. These results indicate that siRNA-mediated reductions in survivin can result in prolonged decreases in tumor growth after implantation of HCT116 cells in nude mice. |
Discussion |
|---|
|
TopABSTRACTIntroductionMaterials and MethodsResultsDiscussionREFERENCES |
|---|
A number of studies have already used microarrays or another large-scale expression analysis tool, serial analysis of gene expression, to examine CRC. These studies have demonstrated the presence of a number of transcripts differentially expressed between colon tumor and normal samples (21 , 46 , 47) . In addition, several studies have compared gene expression during the progression of normal, to adenoma, to carcinoma, and finally to metastatic disease, and have shown that these stages are molecularly distinct, with unique gene expression patterns at each point (22 , 23 , 48 , 49) . Finally, one group in Japan has used microarrays to identify novel downstream targets of ß-catenin in transfected cell lines and analyzed the effect of altering expression of these genes on growth of colon tumors (50 , 51) . Although these previous studies were among the first to molecularly characterize CRC and to identify specific genes of which the expression was dysregulated in colon cancer, most did not expand on these analyses to identify those genes of which the altered expression played a specific role in the pathogenesis of the disease.
In our studies, to validate the significance of the array data and to identify possible targets involved in the pathogenesis of colon cancer, 13 genes were chosen for additional analysis. Although several of these genes have been studied individually for their role in controlling cellular growth, their association with colon cancer based on an unbiased screening approach has not been demonstrated previously. Thus, this work validates the association of several of the genes with colon cancer. Among these genes were a number of transcription factors, including the HMGIY, v-myc myelocytomatosis viral oncogene homologue (c-myc), and v-myb myeloblastosis viral oncogene homologue-like 2 (B-myb). The oncogenic and growth-promoting activities of c-myc are well described. c-myc not only promotes entry into the cell cycle, but it also promotes cell growth (40, 41, 42, 43, 44, 45) . Previous studies have shown that levels of c-myc are elevated in colorectal tumors (52) , and it is a known target of ß-catenin, a gene of which the abnormal accumulation is strongly linked to the development of many types of tumors, including CRC (7) . B-myb is also important in cell cycle control and is known to activate both CDC2 and cyclin D1 (53) . Recent microarray, serial analysis of gene expression, and comparative genome hybridization analyses have demonstrated elevated levels of B-myb in a number of malignancies, including ovarian, lung, breast, and cancers of the gastroesophageal junction (54, 55, 56, 57) ; however, to our knowledge, this is the first report of elevated levels of B-myb in CRC. HMGIY is the target of 6p21.3 rearrangements in a number of benign mesenchymal tumors and was found recently to be up-regulated in colon tumors using microarray analysis (21 , 58) .
Several additional proteins with a role in transcriptional regulation and/or DNA damage repair were also found elevated, including BRCA2 and ATDC. BRCA2 is involved in DNA damage repair, and inherited mutations in this gene predispose to breast, ovarian, and other cancers (59) . Elevated levels of this gene have not been noted previously in CRC. The ATDC gene was originally identified by its ability to complement the radiosensitivity defect of an AT fibroblast cell line (60) . Although individuals with AT show immunological, neurological, and developmental defects, and an increased risk of cancer (61) , this is the one of the first observations of an elevation of this gene in tumors.
Three of the overexpressed genes have a variety of functions, which may also have relevance in the pathogenesis of CRC. CTP synthetase is the rate-limiting enzyme of cytosine triphosphate biosynthesis (62) . Elevated levels of this enzyme have been noted previously in neoplastic versus normal tissue (63) . The microtubule-associated protein, EB1, has been shown to bind to the carboxy terminal region of the APC gene (64) . APC is a tumor suppressor gene, the mutation of which is one of the earliest events in colorectal carcinogenesis (4 , 5) . EB1 colocalizes along with APC to microtubules and may play a role in connecting APC to cellular division, coordinating normal growth and differentiation processes in the colon (65) . EB1 has not been shown previously to be elevated in colon tumors. Survivin is another protein shown previously to interact with microtubules, specifically the centrosome (38) . It is involved in proper duplication of the centrosomes during cell division and also plays a role in inhibiting apoptosis during the normal cell cycle (37, 38, 39) . It is known to be elevated in a wide variety of tumors, and is not expressed to any significant extent in normal tissues (34 , 36) . Interestingly, there is evidence that APC via APC/ß-catenin/T-cell factor-4 signaling down-regulates survivin expression (66) .
The genes that we identified as being under-expressed in colon tumor versus normal tissue and validated by quantitative PCR show a variety of functions that may also be related to colon tumor pathogenesis. The ADAMTS1 is a novel protein with metalloprotease, disintegrin, and thrombospondin domains. It has been shown previously to inhibit endothelial cell proliferation and to have antiangiogenic activity (67) . TRAIL is a type II membrane protein that is known to induce apoptosis (68) . Biliary glycoprotein is a cell adhesion molecule of the immunoglobulin family that behaves as a tumor inhibitor protein in colon and prostate cancers (69 , 70) . It has been shown previously to be down-regulated in colon cancer (71) . Carbonic anhydrase II reversibly hydrates CO2, and it is thought to function in transport and metabolic processes. Several studies have reported that loss of this protein consistently accompanies the progression to malignant transformation (72) . Finally, mitogen-activated protein kinase 3 is a member of a family of tyrosyl-phosphorylated and MAPKs that participate in cell cycle control (73) . It is somewhat surprising to find it down-regulated in colon tumor samples, but results from the NCI60 Microarray Project find a similar down-regulation of this kinase in several colon cancer cell lines (74) .
Although a number of these genes could potentially be key players in the development of CRC, their actual role in promoting the growth of colon tumors remains to be proven. To identify relevant target genes that are important in the pathogenesis of CRC, we additionally investigated whether reducing the levels of these genes in an established colon cancer cell line might alter its proliferation either in culture or when implanted in nude mice. Such results would be consistent with a possible role for these genes in the pathogenesis of colon cancer. For these studies, the colon cancer cell line HCT116 was chosen. RNAi was used to significantly reduce the levels of both c-myc and survivin proteins in this cell line, either separately or together. Loss of either or both proteins led to a reduction in the proliferation of these cells and distinct cell cycle defects. SiRNA directed against c-myc resulted in a slightly reduced number of cells entering S phase, whereas siRNA directed against survivin and c-myc/survivin resulted in a large increase in multiploid cells. These observations are consistent with a role for c-myc in promoting entry to the cell cycle and survivin in controlling proper cytokinesis. In addition, siRNA directed against survivin resulted in an increase in apoptotic cells. These results are consistent with the previously established role of survivin as an antiapoptotic gene. Survivin is thought to colocalize to the centrosome along with caspase-3 and caspase-9 during mitosis. Phosphorylation of survivin by p34cdc2 is necessary for the colocalization of survivin and caspase-9, and subsequent inhibition of apoptosis (39 , 75) .
Despite the data showing decreased proliferation after treatment with siRNA against c-myc and/or survivin, and despite data in the literature that has shown an association of c-myc with colon cancer (52) , experiments examining the number of colonies formed in soft agar after RNAi treatment for c-myc, survivin, or both, revealed that only survivin-depleted cells showed a significant reduction in tumorigenesis. This was mirrored by studies in vivo in which HCT116 cells transfected with siRNA oligonucleotides against c-myc, survivin, or both were injected s.c. into nude mice. These results are quite consistent with in vivo studies examining survivin and c-myc expression in patients with CRC. Expression of survivin was shown to correlate with decreased apoptosis, and increased proliferation and angiogenesis in the tumors (35) . Survivin immunoreactivity was increased significantly in the transition of colon adenoma with low dysplasia to either adenomas with high dysplasia or carcinomas, indicating that survivin may play a role in colorectal tumorigenesis. Finally, expression of survivin in colorectal carcinomas was associated with decreased survival rates (76) . On the other hand, a previous study reported elevated levels of c-myc RNA in primary adenocarcinomas versus normal colonic mucosa but failed to find any correlation between c-myc overexpression and recurrence of disease or patient survival (77) . Thus, with relative ease, our results using RNAi were able to confirm the findings of much more labor-intensive efforts in patients at showing a correlation between overexpression of a gene and disease progression.
It was anticipated that the simultaneous targeting of two overexpressed genes with different cellular functions would result in a greater reduction in tumorigenicity than did disruption of either gene alone. Although disruption of both c-myc and survivin together did result in a greater decrease in proliferation as measured by [3H]thymidine incorporation than did disruption of either gene alone, disruption of both genes did not reduce the number of colonies formed in soft agar nor did it decrease the size of tumors in vivo to a greater extent than did disruption of either gene alone. The reduced number of apoptotic cells seen with the combination of c-myc and survivin versus survivin alone may explain this. That is, the cells may undergo less apoptosis in the absence of both genes than in the absence of survivin alone. This finding is consistent with recent observations suggesting that c-myc plays an important role in sensitizing cells to a variety of apoptotic triggers (78) . Thus, it is important to consider possible interactions when using combination therapies that disrupt the functions of multiple distinct genes.
One intriguing aspect of this work is that it suggests that RNAi itself may prove to be a novel therapeutic tool. The effectiveness of RNAi in vivo, even 3 weeks after initial treatment of the cells, suggests its effects are long lasting. However, pretreatment of the tumor is a very artificial situation. Conditions will have to be developed in which the siRNA oligonucleotides are stable in vivo before using this technique for true in vivo therapy. The development of stable plasmids expressing siRNA oligonucleotides (79 , 80) may be a step in the right direction.
In summary, we have demonstrated that the use of RNAi when coupled with microarray analysis provides an excellent system to define the role of specific genes that are dysregulated in cancer on both the in vitro and in vivo growth of the tumor. Although both c-myc and survivin were overexpressed in the colon tumors, and although both have known roles in cell growth, the use of RNAi demonstrated that only disruption of survivin affected tumorigenesis significantly. Because its expression is restricted primarily to tumor tissue, survivin is an attractive therapeutic target for the treatment of colon and other cancers. These results validate the use of our system for examining unknown genes shown to be overexpressed in cancer.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 Supported by The Harold Simmons Cancer Center, The National Women’s Cancer Research Alliance, The Grant Dove Foundation, The Raymond Nasher Cancer Research Program, The Helena and Alden Wagner Cancer Fund, and NIH Grant CA74128. 
2 To whom requests for reprints should be addressed, at Simmons Cancer Center, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390-8594. Phone: (214) 648-4990; Fax: (214) 648-4152; E-mail: noelle.williams{at}utsouthwestern.edu, or carlos.becerra{at}utsouthwestern.edu
3 The abbreviations used are: CRC, colorectal cancer; FAP, familial adenomatous polyposis; RNAi, RNA interference; APC, adenomatous polyposis coli; RT-PCR, reverse transcription-PCR; MSI, microsatellite instability; MSS, microsatellite stability; siRNA, small interfering RNA; AT, ataxia telangiectasia; HMGIY, high mobility group protein isoforms I and Y; ATDC, ataxia-telangiectasia group D-associated protein; BRCA2, breast cancer 2, early onset; EST, expressed sequence tag; TRAIL, tumor necrosis factor-related apoptosis-inducing ligand; ADAMTS1, a disintegrin-like and metalloproteinase with thrombospondin type 1 motif, 1; BrdUrd, bromodeoxyuridine; MAPK, mitogen-activated protein kinase; FACS, fluorescence-activated cell sorter.
Received 8/ 2/02; accepted 11/13/02.
|
REFERENCES |
|---|
|
TopABSTRACTIntroductionMaterials and MethodsResultsDiscussionREFERENCES |
|---|
