Cancer Research Audrey Hepburn Protein Translation and Cancer
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Irby, R. B.
Right arrow Articles by Lee, N. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Irby, R. B.
Right arrow Articles by Lee, N. H.
[Cancer Research 65, 1814-1821, March 1, 2005]
© 2005 American Association for Cancer Research

Cell and Tumor Biology

Iterative Microarray and RNA Interference–Based Interrogation of the Src-Induced Invasive Phenotype

Rosalyn B. Irby1,2, Renae L. Malek3, Greg Bloom1, Jennifer Tsai3, Noah Letwin3,4, Bryan C. Frank3, Kathleen Verratti3, Timothy J. Yeatman1 and Norman H. Lee3,4

1 Department of Surgery, H. Lee Moffit Cancer Center & Research Institute, College of Medicine, University of South Florida, Tampa, Florida; 2 Penn State Cancer Institute, Penn State College of Medicine, Hershey, Pennsylvania; 3 Department of Functional Genomics, Institute for Genomic Research, Rockville, Maryland; and 4 Department of Pharmacology, George Washington University Medical Center, Washington, District of Columbia

Requests for reprints: Norman H. Lee, Department of Functional Genomics, The Institute for Genomic Research, 9712 Medical Center Drive, Rockville, MD 20850. Phone: 301-795-7585; Fax: 301-838-0208; E-mail: nhlee{at}tigr.org and Timothy J. Yeatman, Department of Surgery, H. Lee Moffit Cancer Center, 12902 Magnolia Drive, Tampa, FL 33612. E-mail: yeatman{at}moffit.usf.edu.


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Src kinase has long been recognized as a factor in the progression of colorectal cancer and seems to play a specific role in the development of the metastatic phenotype. In spite of numerous studies conducted to elucidate the exact role of Src in cancer progression, downstream targets of Src remain poorly understood. Gene expression profiling has permitted the identification of large sets of genes that may be functionally interrelated but it is often unclear as to which molecular pathways they belong. Here we have developed an iterative approach to experimentally reconstruct a network of gene activity regulated by Src and contributing to the invasive phenotype. Our strategy uses a combination of phenotypic anchoring of gene expression profiles and loss-of-function screening by way of RNA-mediated interference. Using a panel of human colon cancer cell lines exhibiting differential Src-specific activity and invasivity, we identify the first two levels of gene transcription responsible for the invasive phenotype, where first-tier genes are controlled by Src activity and the second-tier genes are under the influence of the first tier. Specifically, perturbation of first-tier gene activity by either pharmacologic inhibition of Src or RNA-mediated interference–directed knockdown leads to a loss of invasivity and decline of second-tier gene activity. The targeting of first-tier genes may be bypassed altogether because knockdown of second-tier genes led to a similar loss of invasive potential. In this manner, numerous members of a "transcriptional cascade" pathway for metastatic activity have been identified and functionally validated.

Key Words: Src • invasion • microarrays • gene expression • RNAi


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
A number of oncogenes are known to exhibit increased expression and activity in human colon cancer. Src kinase is well known as an oncogene that is overexpressed in these cancers and has been the focus of a number of studies in recent years (1–4). A member of the Src family of nonreceptor tyrosine kinases, Src functions in the signal transduction cascade and seems to play a role in the progression of a number of human cancers. Despite intense investigation into the functions of Src in human cancer progression, identification of the downstream targets of Src remains poorly understood (5). Src is activated by a number of receptor tyrosine kinases, such as epidermal growth factor receptor (6, 7) and insulin-like growth factor receptor (8). Activated Src stimulates the MAP kinase pathway (9) and signal transduction and activator of transcription factors (10), resulting in altered gene expression. Src also directly influences cellular activities, such as adhesion, through focal adhesion kinase and cadherins (11, 12).

Microarrays have become a popular technique to study the comparative expression of a large number of genes in a single experiment (13). Recently, gene expression patterns in a panel of Src-transformed rodent cell lines were compared with those from staged human colon tumors known to contain activated Src (14). The analysis revealed a Src-specific "transformation signature" containing a number of genes encoding components of cell cycle regulation, transcription factors, and cytoskeletal-associated proteins. Nevertheless, as is the general case for microarray studies, there has been a lack of functional validation coupled with an inability to map differentially regulated genes into a coherent network. In the present study, we have examined a series of well-studied human colon cancer cell lines that make up a part of the National Cancer Institute-60 panel (www.genome.wi.mit.edu/MPR/NCI60/NCI60.html). Herein, we describe an approach of successive iterations of phenotypic anchoring (i.e., phenotyping, expression profiling, and pattern matching) together with targeted gene disruption by RNA-mediated interference RNAi; refs. (15, 16) to identify a sequence of transcriptional events downstream of Src that are important for the invasive characteristics of colon cancer cells, thus allowing us to associate genes with function.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
Cell Culture and Small Interfering RNA Transfections. Human colon cancer cells, HT29, SW620, HCT116 (obtained from American Type Culture Collection), and KM12C (obtained from I. Fidler, Department of Cancer Biology, M.D. Anderson Cancer Center, Houston, TX) were cultured in 100 mm dishes containing RPMI medium with 10% fetal bovine serum at 37°C and 5% CO2. Cells were cultured for 24 hours to 50% confluence before transfection with the small interfering RNA (siRNA) duplex using LipofectAMINE (Invitrogen, Carlsbad, CA), following manufacturer's instructions. Briefly, the siRNA duplex mixed in 50 µL OptiMEM and LipofectAMINE mixed in 50 µL OptiMEM were simultaneously incubated at room temperature for 5 minutes. Subsequently, the two mixtures were combined and incubated at room temperature for 20 minutes. The siRNA:LipofectAMINE complex was added to the cells and left in the plate for the duration of the experiment. Final concentration of siRNA duplex was 100 pmol per milliliter of culture medium.

Small Interfering RNA. Design of siRNA duplexes was accomplished by using the software siRNA Target Finder (Ambion, Austin, TX; http://www.ambion.com/techlib/misc/siRNA_finder.html) and chemically synthesized (Dharmacon, Lafayette, CO). siRNA duplex RBPi-604 (sense: GAUGAAG ACAAGCGCCGUCdTdT, antisense: GACGGCGCUUGUCUUCAUCdTdT) targets the retinoblastoma binding protein 1 (RBP1) mRNA (Genbank accession number NM_002892). siRNA duplex CCRi-436 (GGUGGUCAACAGCAUGUACdTdT, GUTCAUGCUGUUGACCACCdTdT) targets the chemokine receptor 9 (CCR9) mRNA (NM_006641). siRNA duplex ADAMi-565 (GAGGCUGAGAACUCAGCUCdTdT, GAGCUGAGUUCUCAGCCUCdTdT) targets the disintegrin and metalloproteinase domain 21 (ADAM21) mRNA (NM_003813). siRNA duplex CD53i-229 (CAACUUCGGAGUGCUCUUCdTdT, GAAGAGCACUCCGAAGUUGdTdT) targets the CD53 antigen mRNA (NM_000560). siRNA duplex SASi-652 (GCAUUCAGACGAAGCCCUGdTdT, CAGGGCUUCGUCUGAAUGCdTdT) targets the sarcoma amplified sequence (SAS) mRNA (NM_005981). siRNA duplex Sp100i-707 (GAUGAUUCGCUAGGAAGCCdTdT, GGCUUCCUAGCGAAUCAUCdTdT) targets the nuclear antigen Sp100 mRNA (NM_003113). siRNA duplex PRDM2i-1853 (GACUGCUCAGAGGUAACACdTdT, GUGUUACCUCUGAGCAGUCdTdT) targets the PR domain containing protein 2 with zinc finger domain (PRDM2) mRNA (NM_012231). siRNA duplexes GLUTi-774 (CUGGCCAAAUUCAGUCCCGdTdT, CGGGACUGAAUUUGGCCAGdTdT). GLUTi-1356 (GAGUACUGAGCCCUGAAGCdTdT, GCUUCAGGGCUCAGUACUCdTdT) targets the glutaminase mRNA (AF223943). siRNA duplexes RPL32i-291 (CGUCAAGGAGCUGGAAGUGdTdT, CACUUCCAGCUCCUUGACGdTdT) and RPL32i-378 (GCCAUCGUGGAAAGAGCUGdTdT, CAGCUCUUUCCACGAUGGCdTdT) target the ribosomal protein L32 mRNA (NM_000994). siRNA duplexes KERi-660 (GGAGCAUCAGGAGGAAGUCdTdT, GACUUCCUCCUGAUGCUCCdTdT) and KERi-1084 (CGCCAGAACAACGAAUACCdTdT, GGUAUUCGUUGUUCUGGCGdTdT) target the cytokeratin 20 mRNA (BC031559). siRNA duplexes DEADi-1261 (GACAUGACGCCACUGAAGCdTdT, GCUUCAGUGGCGUCAUGUCdTdT) and DEADi-2005 (GAUGCCAGAGUGGGUCCUCdTdT, GAGGACCCACUCUGGCAUCdTdT) target the DEAD/H box polypeptide 32 mRNA (BC002473). siRNA duplexes TFDPi-680 (CGAGUCAGCUUAUGACCAGdTdT, CUGGUCAUAAGCUGACUCGdTdT) and TFDPi-1196 (GGCUCUGGAGCCAUACGUGdTdT, CACGUAUGGCUCCAGAGCCdTdT) target the transcription factor Dp1 mRNA (BC011685). Negative control green fluorescent protein siRNA duplex GFPi-122 is from Caplen et al. (GCAAGCUGACCCUGAAGUUCAU, GAACUUCAGGGUCAGCUUGCCG; ref. 17). Negative control siRNA duplex CONT-1 (GUAGGCAAGCAUCUGAGCUdTdT, AGCUCAGAUGCUUGCCUACdTdT) is a sequence-scrambled version of CCR9i-436. Negative control siRNA duplex CONT-2 (CCGACUGGGUACCAAGUAGdTdT, CUACUUGGUACCCAGUCGGdTdT) is a sequence-scrambled version of ADAMi-565. Negative control siRNA duplex CONT-3 (CGAUCAUCCCGCAGGAAGAdTdT, UCUUCCUGCGGGAUGAUCGdTdT) is a sequence-scrambled version of DEADi-1261.

Affymetrix U133A GeneChip. The Human Genome U133A GeneChip (Affymetrix, Santa Clara, CA) contains 22,283 probe sets representing over 18,000 named genes based on RefSeq database sequences. For each cell line, total RNA was prepared by TRIzol (Invitrogen) extraction, quantified, and validated for integrity by gel electrophoresis. Target synthesis, hybridization, and posthybridization staining were done as previously described (14), using standard protocols as recommended by the manufacturer (Affymetrix). Stained chips were scanned on a GeneArray Scanner (Affymetrix), and data files were processed using the Microarray Suite 5.0 software to calculate signal intensities, provide probe set detection calls, scale hybridization experiments, and calculate gene expression change.

Spotted DNA Microarrays, Target Labeling, and Hybridization. Spotted DNA microarray fabrication has been described previously (14, 18). Arrays were spotted with the PCR amplicons of 32,448 human cDNA clones representing ~27,487 unique genes based on the TIGR Gene Indices (19). Gene annotation can be accessed at http://www.tigr.org/tigr-scripts/magic/r1.pl. Preparation of labeled cDNA target and hybridization were done as previously described (20). Briefly, total RNA was isolated from three to six independent cultures of human colon cancer cell lines using RNeasy Mini kit as per the manufacturer's protocol (Qiagen, Valencia, CA). Total RNA (15 µg) was used to synthesize fluorescent cDNA target by reverse transcription with random primers and aminoallyl dUTP followed by chemical coupling with the Cy5 or Cy3 ester. Cy5-labeled cDNA from an individual test cell line was mixed with Cy3-labeled cDNA derived from the Universal Human Reference RNA (Stratagene, La Jolla, CA), and cohybridized onto the array. A reverse dye labeling strategy was included to account for potential dye-gene labeling bias (20).

Image Scanning, Statistics, Cluster Analysis, and Pattern Matching. Image scanning, fluorescence intensity measurements, background subtraction, data normalization with locally weighted linear regression, experimental noise determination, and cluster analysis were done as described previously (14, 18, 20). Fluorescence intensities were measured at 10 µm per pixel resolution (GenePix Pro 4000B, Axon Instruments, Inc., Foster City, CA). Expression ratios were calculated as the background subtracted ratio of median intensities (635 nm emission/532 nm emission) from the two channels. The following parameters identified well-measured features: feature spot diameter > 50 µm; saturated pixels < 50%; sum of background subtracted median intensities for each wavelength > 1,000 units (scale of 0-65,535). Differentially regulated genes that were statistically significant were identified by F test with a 1% false discovery rate using the ArrayStat software (Imaging Research, Inc., St. Catharines, Ontario, Canada, www.imagingresearch.com). For cluster analysis of significantly regulated genes, we used TIGR MultiExperiment Viewer, available at http:/www.tigr.org/softlab/. Before clustering, logarithms (base 2) were calculated on the measured median background subtracted fluorescence ratio of each gene. To identify candidate invasion-causing genes, pattern matching was used (21). In this technique, a hypothetical expression profile or "template" is created on the basis of phenotyping information (i.e., phenotypic anchoring) and used to search for genes with an analogous expression pattern (i.e., pattern matching). For example, a positively correlated template will identify any gene that is highly expressed in highly invasive cell lines and weakly expressed in the weakly invasive lines. Candidate genes were selected from matches exhibiting a correlation coefficient >0.85 and P < 0.01; RNAi in conjunction with loss-of-function assays were done to validate gene function.

Matrigel Assay. Human colon cancer cells were grown to 80% confluence, trypsinized, and counted. 5 x 104 (for pharmacologic inhibitor studies) or 1.5 x 104 (for siRNA transfection studies; lower cell density required for optimization of transfection) cells were seeded into the top well of the Matrigel inserts in serum-free medium. The lower well was filled with 800 µL medium with 10% serum. Six to ten hours following seeding, the cells in the upper wells were transfected with the appropriate siRNA duplex as detailed above. Cells were allowed to grow and invade for 96 hours. The cells were scraped from the upper side of the inserts with a cotton swab and the cells on the lower side were stained with crystal violet and counted under a microscope. All cells were counted. The experiment was repeated at least thrice in quadruplicate.

Homotypic Adhesion Assay. Cells were grown to confluence in 12-well tissue culture dishes. Cells were trypsinized in trypsin containing either 1 mmol/L EDTA (TE) or 1 mmol/L calcium (TC) for 30 minutes at 37°C. Cells were pipetted up and down five times gently and counted in a coulter counter (Beckman Instruments, Fullerton, CA). The results were presented as TC/TE. The assays were done thrice in triplicate.

Western Blotting and Src Kinase Assay. Fifty micrograms of total cellular protein from cell lysates were subjected to Western blotting with an anti-Src GD11 monoclonal antibody (Upstate, Charlottesville, VA) as previously described (22). For Src kinase assays (22), Src was immunoprecipitated from cell lysates (50 µg protein) with GD11 and resuspended in kinase reaction buffer containing enolase and [{gamma}-32P]ATP.

Quantitative Real-Time Reverse Transcription-PCR. To validate microarray data and gene knockdowns with RNAi, quantitative real-time reverse transcription-PCR (RT-PCR) was done as described previously (14). Total RNA from cell lines were reversed transcribed using random primers as per manufacturer's protocol. The resulting cDNA was diluted and used as template for RT-PCR. PCR primers were selected for specificity by National Center for Biotechnology Information's basic local alignment search tool of the human genome and amplicon specificity was verified by first derivative melting curve analysis using software provided by Applied Biosystems (Foster City, CA). Analysis of gene expression was done on an ABI Prism 7700 Sequence Detection System using SYBR green. The expression of the housekeeping genes Sp1 transcriptional activation cofactor (Genbank accession number BC005250) and protein inhibitor of activated signal transduction and activator of transcription protein PIASy (AK022481) was used for normalization. These genes did not exhibit differential expression in our microarray assays. RT-PCR primer pair sequences for each gene are as follows: Sp1 transcriptional activation cofactor, AACGGCTTGAAACAGCTGAG, AGGCAAATCATCAGGCAAAG; protein inhibitor of activated signal transduction and activator of transcription protein PIASy, GAGAAGAAGCCCACCTGGA, ACACTCGCTCAGGATCTTCG; RBP1, GAGGATGCAATGCCTCTGAT, TGCTACATCCTCAATGCTGG; CCR9, AAAGGGGACACAGAAGCACT, AGGCTCAAGAGCAGGGTAGA; disintegrin and ADAM21, AAGACGCCTGTTGTCTGTTG, TGGCATGAACTTGCAGTCTT; CD53 antigen, CGGAATCATCACCATCTGTG, TCATAGCCCTATGGTCTGGC; SAS, CATGATCATCCTTGGTTTGG, TCTGTCTGTTTGCTTCGGTT; nuclear antigen Sp100, GCTAAGTATACGCTGCGGTG, TTTCTTGTGCTTGGTGGTTC; and PRDM2, GCTTTTCCCTTCTGCTGTTG, CAGCCAGTTTCCCTTCTCTG.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
Phenotypic Anchoring of Colon Cancer Cell Lines. Colon cancer cell lines could be segregated phenotypically based on c-Src protein kinase activity as measured by in vitro phosphorylation assays with the exogenous substrate enolase (Fig. 1A). Kinase activity was directly related to the amount of expressed c-Src protein (Fig. 1A). Consistent with these findings, high c-Src-specific activity in the SW620 and HT29 cell lines was associated with a high potential to invade Matrigel (Fig. 1B) indicative of high metastatic potential (23). On the other hand, low c-Src activity in KM12C and HCT116 cells was linked with a low potential to invade Matrigel. The role of c-Src in invasiveness was confirmed by the tyrosine kinase inhibitor PD180970, which potently impaired c-Src activity in SW620 and HT29 cells and correspondingly decreased invasive activity (Fig. 1A and B). Of interest, SW620 cells exhibited residual invasive activity although c-Src activity was practically abolished by inhibitor treatment. This suggests the presence of Src-dependent and Src-independent mechanisms for invasion. Also, inhibitor treatment did not completely eliminate kinase activity in HT29 cells due likely to the higher expression of c-Src in this cell line.



View larger version (27K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. A, Src protein levels and kinase activity. Src protein levels were assessed by Western blotting of colon cancer cell lysates. A representative fluorogram is shown. Kinase activity was assessed by subjecting exogenous enolase to phosphorylation by Src immunoprecipitated from cell lysates. A representative autoradiogram is shown. Data are from three independent experiments. B, Src kinase activity correlates with invasive potential assessed by Matrigel assay. Columns, mean of three to five independent experiments; bars, SE. Highly invasive SW620 and HT29 cells were treated (+I) with the tyrosine kinase inhibitor PD180970. *, Significantly different from corresponding PD180970-treated cells (P < 0.05) by ANOVA and post hoc Tukey test.

 
Next, we expression profiled the panel of colon cancer cell lines to identify candidate genes involved in invasive activity. Our approach is based on extracting gene expression patterns that statistically associate with either highly or weakly transformed phenotypes (14). Using the Affymetrix U133A GeneChip containing ~18,000 named genes, we interrogated RNA derived from the KM12C, HCT116, SW620 (without and with pharmacologic inhibitor PD180970), and HT29 (without and with PD180970) cell lines. Gene expression patterns that are significantly correlated with the phenotyping data (P < 0.01) were computationally derived via a pattern matching technique (21), resulting in the identification of two relevant clusters (Fig. 2). One cluster of 28 genes (herein called the "high invasive cluster") was overexpressed in the highly invasive SW620 and HT29 cell lines with high c-Src activity and was underexpressed in the weakly invasive KM12C and HCT116 cells harboring low c-Src activity (Fig. 2, left). Hence, we hypothesize that a gene(s) belonging to the high invasive cluster likely facilitates invasive activity. The second cluster of genes was down-regulated in highly invasive cells and up-regulated in the weakly invasive cells, and likely represent genes involved in suppressing invasiveness ("low invasive cluster"; Fig. 2, right). Of interest was the finding that treatment of the highly invasive cell lines with PD180970 resulted in expression patterns now resembling those found in the weakly invasive cells (Fig. 2). It should be noted that this phenomenon of expression pattern reversal was specific to the high and low invasive clusters and not seen in other clusters (data not shown). These observations, coupled with the effects of PD180970 on c-Src activity and Matrigel assays (Fig. 1), lend further support of the notion that genes in the "invasive clusters" participate in metastatic activity.



View larger version (57K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. Cluster images showing two classes of gene expression profiles associated with invasive colon cancer cell lines. Scaled Affymetrix U133A GeneChip data were log2 transformed, filtered for min-max difference ≥ 1, median centered, and subjected to pattern matching and hierarchical clustering. Genes were identified as being up-regulated (left) or down-regulated (right) in the highly invasive cell lines HT29 and SW620 relative to the weakly invasive cell lines KM12C and HCT116. Red, highly expressed genes; green, weakly expressed genes; black, intermediate expression. The expanded top dendrogram (left, bottom) shows the relationship between the cell lines. HT29 and SW620 cells were treated (+I) with the tyrosine kinase inhibitor PD180970. Note the relationship of gene expression patterns where inhibitor-treated highly invasive cells now resemble weakly invasive cells.

 
RNAi of Candidate Invasive Genes. To validate the role of the up-regulated genes (i.e., high invasive cluster) in facilitating invasiveness, we perturbed HT29 cells with siRNAs directed against seven genes, namely the disintegrin and ADAM21, CCR9, nuclear antigen Sp100, PRDM2, RBP1, and the tetraspanins CD53 and SAS. The protein products of these genes are known mediators of cell signaling and are predicted to affect downstream gene expression, hence the reason for their selection. Of the seven genes targeted, knockdown of transcripts for ADAM21, CCR9, or CD53 led to a significant impairment in the ability of this cell line to invade Matrigel (Fig. 3A). As expected, transfection of HT29 cells with negative control siRNA GFPi-122 (targets GFP), CONT-1 (scrambled CCR9i-436), or CONT-2 (scrambled PRDM2i-1853) had no discernable effect on invasiveness (Fig. 3A). Successful gene knockdown was validated by quantitative RT-PCR where mRNA levels for each of the seven targeted genes were decreased by 51% to 80% with the majority of the reductions in the 72% to 80% range (Fig. 3C). Hence, the inability of some siRNAs to impair HT29 invasiveness was not due to transfection inefficiencies or a failure of the RNA-induced silencing complex to target mRNAs for degradation (15).



View larger version (21K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3. A, effect of specific gene knockdowns on invasive potential in colon cancer cell line HT29. Cells were transfected with siRNAs targeting the indicated mRNA species and allowed to grow and invade Matrigel over a 96-hour period. Green fluorescent protein siRNA (GFPi-122) served as a negative control as well as scrambled siRNA sequences of CCR9i-436 (CONT-1) and ADAMi-565 (CONT-2). Columns, mean of 5-10 independent experiments; bars, SE. B, effect of specific gene knockdowns on homotypic cell adhesion. HT29 cells were transfected with siRNAs for 24 hours before performing cell dissociation studies. GFPi-122, CONT-1, and CONT-2 served as negative controls. Columns, mean of three independent experiments; bars, SE. Note that ADAM21, CCR9, and CD53, implicated in cell invasiveness, had no effect in cell adhesion, suggesting that different transcriptional circuitries are involved. C, effectiveness of gene knockdowns was validated by quantitative RT-PCR. *, Significantly different from negative control treated cells (P < 0.05) by ANOVA and post hoc Tukey test. GFP, CONT-1, and CONT-2 treatments were not significantly different from each other (P > 0.05).

 
RNAi of Invasive Genes Does Not Affect Homotypic Cell Adhesion. Having identified the genes for ADAM21, CCR9, and CD53 as participating in the invasion process of HT29 cells, we next tested whether these same genes would affect another aspect of the cell transformation process, specifically homotypic cell adhesion (22). Homotypic cell adhesion of HT29 cells treated with the siRNA for ADAM21, CCR9, or CD53 was not significantly different from the adhesive properties of cells treated with any of the negative control siRNAs (GFPi-122, scrambled CCR9i-436, or scrambled PRDM2i-1853; Fig. 3B). Hence, genes critically involved in invasiveness do not seem to be important in homotypic cell adhesion. As might be expected, targeted gene knockdown of Sp100, PRDM2, RBP1, or SAS, ineffective in ameliorating invasiveness, was also without effect on HT29 homotypic cell adhesion (Fig. 3B). To identify genes involved in cell to cell interactions, one would simply design an "adhesion template" to phenotypically anchor gene expression profiles to homotypic cell adhesion data.

Identification and RNAi of Second-Tier Genes Implicated in Invasiveness. To identify second-tier genes (i.e., genes downstream of ADAM21, CCR9, and CD53) involved in tumor cell invasiveness, we expression profiled HT29 cells that were perturbed with the same siRNAs used in the Matrigel phenotyping assay. To interrogate a greater number of genes, we used a custom human cDNA microarray containing ~32,000 elements. Among the seven siRNA treatment groups (in addition to three negative control groups: GPFi-122 siRNA and two scrambled siRNA sequences), an F test with a 1% false discovery rate identified 1,911 genes across the seven treatment groups that were significantly differentially regulated compared with the negative control groups. We once again utilized pattern matching (21) to identify a common set of genes that were phenotypically anchored to weak invasivity in response to siRNA treatment (i.e., genes that were only down-regulated by ADAM21, CCR9, and CD53 siRNA treatments; Fig. 4A). Our hypothesis was that the gene products of ADAM21, CCR9, and CD53 cooperatively up-regulate a core set of downstream invasiveness genes (i.e., second-tier genes). Hence, RNAi knockdown of any one of these three genes would lead to a decreased expression of this core set and subsequent loss of invasiveness. When we compared the profiles of ADAM21, CCR9, and CD53 knockdowns to the GPF negative control, a total of 76 genes were identified by pattern matching as second-tier candidates (Fig. 4B). The number of candidate genes declined to 10 when we further restricted the second-tier genes to match a pattern of low expression in the ADAM21, CCR9, and CD53 knockdowns but high relative expression in the GFP negative control and knockdowns for Sp100, PRDM2, RBP1, and SAS (Fig. 4A and C). To recall, knockdown of Sp100, PRDM2, RBP1, and SAS had no effect on HT29 cell invasiveness and hence we would predict that the downstream genes affected by these knockdowns do not play a critical role in invasiveness. From a practical point of view, this method of pattern restriction allows us to prioritize which genes to subsequently target.



View larger version (18K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4. Gene expression profiling of HT29 cells treated with siRNAs targeting first-tier genes: ADAM21, CD53, CCR9, Sp100, PRDM2, RBP1, and SAS. GFPi-122 siRNA served as a negative control. A, pattern matching (PM) identifies correlated (P < 0.01) genes specifically down-regulated by siRNA treatment. The phenotypically anchored, anti-invasion template selects for second-tier genes exhibiting decreased expression following siRNA targeting of ADAM21, CD53, and CCR9. Down-regulation of second-tier genes is hypothesized to facilitate an anti-invasion phenotype. B, clustergram of genes down-regulated in cells treated with siRNAs targeting ADAM21, CD53, and CCR9 relative to siRNA-negative control GFPi-122. C, clustergram of genes down-regulated in cells treated with siRNAs targeting ADAM21, CD53, and CCR9 but not to siRNAs targeting Sp100, PRDM2, RBP1, SAS, and GFP. Note that Sp100, PRDM2, RBP1, and SAS siRNA treatments did not affect invasiveness and hence are hypothesized not to effect the expression of critical genes facilitating invasiveness. The 10 second-tier genes are hypothetical protein FLJ90022(Genbank accession W92257), brain-specific protein (H02328), glutaminase (W72090), ribosomal protein L32 (R43544), cytokeratin 20 (BC031559), DEAD/H box polypeptide 32 helicase (BC002473), transcription factor Dp-1 (BC011685), and three expressed sequence tags (H02328, H97875, H94236).

 
Of the 10 identified second-tier genes, we targeted glutaminase, ribosomal protein L32, cytokeratin 20, DEAD/H box polypeptide, and transcription factor Dp-1 for a second round of RNAi knockdown and Matrigel assays (Fig. 5). Knockdown of each of these genes led to a decrease in HT29 colon cancer cell invasiveness, supporting our contention of a gene transcriptional circuitry that is involved in at least one aspect of tumorigenesis, namely cellular invasiveness. For each gene, two independent siRNAs were designed to target different parts of the mRNA molecule (Fig. 5). This was done to ensure that "off-target" effects were not responsible for the suppression of invasiveness because it is unlikely that two independent siRNAs against the same transcript will target a common off-target transcript leading to the same phenotype (24).



View larger version (24K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5. Effect of specific second-tier gene knockdowns on invasive potential. Genes were identified for the next round of knockdown experiments based on the analysis summarized in Fig. 4. HT29 cells were transfected with the indicated second-tier siRNA species and were allowed to grow and invade Matrigel over a 96-hour period. GFPi-122 siRNA served as a negative control as well as scrambled siRNA sequences of ADAMi-565 (CONT-2) and DEADi-1261 (CONT-3). Columns, mean of three independent experiments; bars, SE. *, Significantly different from negative control treated cells (P < 0.05) by ANOVA and post hoc Tukey test. GFPi-122, CONT-2, and CONT-3 siRNA treatments were not significantly different from each other (P > 0.05).

 

   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
As an early member of a signal transduction cascade, Src exerts pleotropic effects on cell behavior through a number of pathways (1). As a result, the expression of a number of genes is altered and some of these alterations are expected to modify the cancer phenotype. We attempted not only to identify those genes whose expression is altered through increased Src activity but also those genes whose altered expression participated in the metastatic phenotype of the cell. Hence, it was critical that a panel of cancer cell lines could be statistically segregated, at the very least, into high and low metastatic phenotypes. Phenotypic segregation of cell line panels has been successfully applied to other metrics of cancer, such as anchorage-independent cell growth and foci formation (14, 18). Having separated cell lines according to their invasivity, pattern matching was imposed to identify candidate causal genes. Whereas this strategy helps narrow the list of implicated genes, a drawback of microarray studies has been the lack of functional validation. With this in mind, RNAi technology was included to knockdown the expression of tiered candidate genes in cancer cells that were subsequently screened for a loss of function.

We have identified a subset of c-Src up-regulated first-tier genes (ADAM21, CCR9, and CD53) imparting invasive properties to cancer cells. However, it should be noted that not all genes identified in the first round of experimentation seem to participate in invasion, as knockdown of the transcripts for Sp100, PRDM2, RBP1, and SAS did not reduce invasiveness. Notwithstanding, we can begin constructing a model of the "invasive cancer cell" along with its transcriptional circuitry based on the known cellular roles of the RNAi-validated first-tier genes and/or their gene family members (Fig. 6). It is important to note that our validated genes have not previously been linked experimentally to invasion. CCR9 is a member of the G-protein-coupled receptor family capable of coupling to arrestin via the ß-{gamma} subunits of Gi and activating extracellular signal-regulated kinase1/2, resulting presumably in direct effects on the cytoskeletal architecture and secondary effects on gene expression to facilitate chemotaxis/migration (25–27). CD53 belongs to a family of transmembrane proteins, collectively known as tetraspanins, that facilitate either promigratory or antimigratory activity (again via both cytoskeketal effects and secondary gene expression changes) depending on which tetraspanin family member is complexed to which integrin heterodimer (28–30). The surface proteins from the ADAM family are known facilitators of cell migration/invasiveness by proteolytic cleavage of extracellular matrix components (31). Moreover, ADAM family members can elicit changes in gene expression that have been attributed to the transactivation of receptor tyrosine kinases, such as the epidermal growth factor receptor through a process known as sheddase activity (32). The capacity of all three first-tier genes to affect gene expression independently supports the notion of a transcriptional cascade system that sustains tumorigenesis. Hence, it is our contention that interference of ADAM21, CCR9, or CD53 expression impairs the transcription of downstream targets (second-tier genes), thereby impeding cancer cell invasiveness as predicted by our model (Fig. 6).



View larger version (32K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6. Modeling Src-induced transcriptional regulation of genes associated with invasion. Activation of Src induces a transcriptional program that involves multiple tiers of genes. High Src activity is associated with invasivity and overexpression of first-tier genes CCR9, ADAM21, and CD53. Knockdown of these genes by RNAi or pharmacologic inhibition impairs cell invasiveness, in part, by presumably affecting downstream transcription of second-tier genes. Cellular roles of first-tier genes in the invasion process are predictive based on known functional annotation (i.e., gene ontology) of the genes themselves or their family members. RNAi and loss of function validate the involvement of second-tier genes cytokeratin 20, ribosomal protein L32, glutaminase, DEAD/H, and transcription factor Dp1 in the invasion process. Our approach is amenable to the continued mapping of subsequent tiers of transcription thereby providing a complex circuit of the cell. Circuits can be mapped to different aspects of cancer, such as resistance to apoptosis, invasiveness, and homotypic cell adhesion.

 
Of the second-tier genes identified by a subsequent round of microarrays and pattern matching, cytokeratin 20 has historically served as a diagnostic marker for colorectal carcinoma (33, 34) , whereas high expression of ribosomal protein L32, glutaminase, and DEAD/H box polypeptides has been associated with various cancers and/or metastatic lesions (35–38). To the best of our knowledge, we now provide evidence for the first time that these genes likely play a significant role in invasiveness as defined by loss of function following gene knockdown. Another second-tier gene of interest is Dp1 (for dimerization partner 1), which forms a heterodimer with the E2F family of transcription factors to enhance both DNA binding affinity and transactivation function of the latter (for review, see ref. 39). Target genes include those involved in DNA synthesis, cell cycle progression, and proteolysis (39, 40). A recent study indicates a trend toward increased E2F-1 expression during metastatic progression of colorectal carcinoma (41). Moreover, ectopic overexpression of E2F-1 has been shown to increase invasive activity in squamous carcinoma cell lines (42). Our results show that knockdown of E2F's binding partner, Dp1, impairs invasive activity. Noteworthy, gene expression profiling of clinical samples from patients with colon cancer reveals that the Dp1 transcript (along with CD53, ribosomal protein L32, and cytokeratin 20) is significantly up-regulated (P < 0.05) in metastatic Duke's C and D stage lesions compared with nonmetastatic lesions (data not shown). Based on these findings, future knockdown studies will target both the second-tier gene Dp1 and members of the E2F family, allowing us to navigate even further downstream the transcriptional cascade circuit (i.e., third-tier genes).

We expect the segregation of cell lines into a high or low metric classification scheme to be dependent on the phenotype assayed. Hence, the lists of candidate causal genes resulting from correlation analyses will likely vary across phenotypes. Indeed, knockdown of ADAM21, CCR9, or CD53, leading to impairment of colon cancer cell invasiveness, did not impede homotypic cell adhesion loses. The latter characteristic is seen in transformed cells possessing a defect in the cadherin/catenin pathway (22). For this reason, we believe that separate (or partially overlapping) transcriptional circuits exist for different aspects of cancer progression. Hence, at least one transcriptional circuit seems to promote cell invasiveness in colon cancer cells, whereas another likely facilitates loses in cellular adhesion.

Using an iterative strategy encompassing techniques, such as microarrays, RNAi, and loss-of-function screening, we now have in place a systematic approach to map tiered gene networks involved in cancer progression. In the present study, we have investigated the first two tiers of the metastatic process. Presumably, this transcriptional map will become more complex as other features of cancer progression are studied (e.g., resistance to apoptosis, anchorage-independent cell growth, and resistance to hypoxia). A great deal of attention has been given to developing molecular markers and/or identifying gene products that can serve as therapeutic targets of cancer. It is conceivable that our tiered network can be used to facilitate the identification of ideal targets, such as those located at critical branch points in the network controlling multiple aspects of cancer progression.


   Acknowledgments
 
Grant support: This study supported in part by a National Heart Lung and Blood Institute Grant (HL59781-01; N.H. Lee) and National Cancer Institute Grants (CA101355-01, CA098522-01, CA85429-04, CA85052-04; T.J. Yeatman).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received 10/ 6/04. Revised 12/ 3/04. Accepted 12/23/04.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Yeatman TJ. A renaissance for SRC. Nat Rev Cancer 2004;4:470–80.[CrossRef][Medline]
  2. Bjorge JD, O'Connor TJ, Fujita DJ. Activation of human pp60c-src. Biochem Cell Biol 1996;74:477–84.[Medline]
  3. Bolen J, Veillette A, Schwartz A, DeSeau V, Rosen N. Activation of pp60c-src protein kinase activity in human colon carcinoma. Proc Natl Acad Sci U S A 1987;84:2251–5.[Abstract/Free Full Text]
  4. Cartwright C, Kamps M, Meisler A, Pipas J, Eckhart W. pp60c-src activation in human colon carcinoma. J Clin Invest 1989;83:2025–33.
  5. Biscardi J, Tice D, Parsons S. c-Src, receptor tyrosine kinases, and human cancer. Adv Cancer Res 1999;76:61–119.[Medline]
  6. Maa M, Leu T, McCarley D, Schatzman R, Parsons S. Potentiation of epidermal growth factor receptor-mediated oncogenesis by c-Src: implications for the etiology of multiple human cancers. Proc Natl Acad Sci U S A 1995;92:6981–5.[Abstract/Free Full Text]
  7. Mao W, Irby R, Copolla D, et al. Activation of c-Src by receptor tyrosine kinases in human colon cancer cells with high metastatic potential. Oncogene 1997;15:3083–90.[CrossRef][Medline]
  8. Sekharam M, Nasir A, Kaiser HE, Coppola D. Insulin-like growth factor 1 receptor activates c-SRC and modifies transformation and motility of colon cancer in vitro. Anticancer Res 2003;23:1517–24.[Medline]
  9. Fincham V, Frame M, Haefner B, Unlu M, Wyke A, Wyke J. Functions of the v-Src protein tyrosine kinase. Cell Biol Int 1994;18:337–44.[CrossRef][Medline]
  10. Turkson J, Bowman T, Garcia R, Caldenhoven E, De Groot RP, Jove R. Stat3 activation by Src induces specific gene regulation and is required for cell transformation. Mol Cell Biol 1998;18:2545–52.[Abstract/Free Full Text]
  11. Brunton VG, Ozanne BW, Paraskeva C, Frame MC. A role for epidermal growth factor receptor, c-Src and focal adhesion kinase in an in vitro model for the progression of colon cancer. Oncogene 1997;14:283–93.[CrossRef][Medline]
  12. Calalb MB, Zhang X, Polte TR, Hanks SK. Focal adhesion kinase tyrosine-861 is a major site of phosphorylation by Src. Biochem Biophys Res Commun 1996;228:662–8.[CrossRef][Medline]
  13. Miller LD, Long PM, Wong L, Mukherjee S, McShane LM, Liu ET. Optimal gene expression analysis by microarrays. Cancer Cell 2002;2:353–61.[CrossRef][Medline]
  14. Malek RL, Irby RB, Guo QM, et al. Identification of Src transformation fingerprint in human colon cancer. Oncogene 2002;21:7256–65.[CrossRef][Medline]
  15. Hannon GJ. RNA interference. Nature 2002;418:244–51.[CrossRef][Medline]
  16. McManus MT, Sharp PA. Gene silencing in mammals by small interfering RNAs. Nature Rev 2002;3:737–47.
  17. Caplen NJ, Parrish S, Imani F, Fire A, Morgan RA. Specific inhibition of gene expression by small double-stranded RNAs in invertebrate and vertebrate sytems. Proc Natl Acad Sci U S A 2001;98:9742–7.[Abstract/Free Full Text]
  18. Teramoto H, Malek RL, Behbahani B, Castellone MD, Lee NH, Gutkind JS. Identification of H-Ras, RhoA, Rac1 and Cdc42 responsive genes. Oncogene 2003;22:2689–97.[CrossRef][Medline]
  19. Merrick JM, Osman A, Tsai J, Quackenbush J, LoVerde PT, Lee NH. The Schistosoma mansoni gene index: gene discovery and biology by reconstruction and analysis of expressed gene sequences. J Parasitol 2003;89:261–9.[CrossRef][Medline]
  20. Yang IV, Chen E, Hasseman JP, et al. Within the fold: assessing differential expression measures and reproducibility in microarray assays. Genome Biol 2002;3:research0062.
  21. Pavlidis P, Noble WS. Analysis of strain and regional variation in gene expression in mouse brain. Genome Biol 2001;2:RESEARCH0042.
  22. Irby RB, Yeatman TJ. Increased Src activity disrupts cadherin/catenin-mediated homotypic adhesion in human colon cancer and transformed rodent cells. Cancer Res 2002;62:2669–74.[Abstract/Free Full Text]
  23. Irby RB, Mao W, Coppola D, et al. Activating SRC mutation in a subset of advanced human colon cancers. Nat Genet 1999;21:187–90.[CrossRef][Medline]
  24. Berns K, Hijmans EM, Mullenders J, et al. A large-scale RNAi screen in human cells identifies new components of the p53 pathway. Nature 2004;428:431–7.[CrossRef][Medline]
  25. Perry SJ, Lefkowitz RJ. Arresting developments in heptahelical receptor signaling and regulation. Trends Cell Biol 2002;12:130–8.[CrossRef][Medline]
  26. Vlahakis SR, Villasis-Keever A, Gomez T, Vanegas M, Vlahakis N, Paya CV. G protein-coupled chemokine receptors induce both survival and apoptotic signaling pathways. J Immunol 2002;169:5546–54.[Abstract/Free Full Text]
  27. Youn BS, Kim CH, Smith FO, Broxmeyer HE. Role of the CC chemokine receptor 9/TECK interaction in apoptosis. Apoptosis 2002;7:271–6.[CrossRef][Medline]
  28. Berditchevski F. Complexes of tetraspanins with integrins: more than meets the eye. J Cell Sci 2001;114:4143–51.[Abstract/Free Full Text]
  29. Sugiura T, Berditchevski F. Function of {alpha}3ß1-tetraspanin protein complexes in tumor cell invasion. Evidence for the role of the complexes in production of matrix metalloproteinase 2 (MMP-2). J Cell Biol 1999;146:1375–89.[Abstract/Free Full Text]
  30. Shaw LM, Turner CE, Mercurio AM. The {alpha}6Aß1 and {alpha}6Bß1 integrin variants signal differences in the tyrosine phosphorylation of paxillin and other proteins. J Biol Chem 1995;270:23648–52.[Abstract/Free Full Text]
  31. Bauvois B. Transmembrane proteases in cell growth and invasion: new contributors to angiogenesis? Oncogene 2004;23:317–29.[CrossRef][Medline]
  32. Prenzel N, Zwick E, Daub H, et al. EGF receptor transactivation by G-protein-coupled receptors requires metalloproteinase cleavage of proHB-EGF. Nature 1999;402:884–8.[Medline]
  33. Moll R. Cytokeratins in the histological diagnosis of malignant tumors. Int J Biol Markers 1994;9:63–9.[Medline]
  34. Tot T. Identifying colorectal metastases in liver biopsies: the novel CDX2 antibody is less specific than the cytokeratin 20 + /7-phenotype. Med Sci Monit 2004;10:BR139–43.[Medline]
  35. Causevic M, Hislop RG, Kernohan NM, et al. Overexpression and poly-ubiquitylation of the DEAD-box RNA helicase p68 in colorectal tumours. Oncogene 2001;20:7734–43.[CrossRef][Medline]
  36. Medina MA. Glutamine and cancer. J Nutr 2001;131:2539–42S.
  37. Karan D, Kelly DL, Rizzino A, Lin MF, Batra SK. Expression profile of differentially-regulated genes during progression of androgen-independent growth in human prostate cancer cells. Carcinogenesis 2002;23:967–75.[Abstract/Free Full Text]
  38. Zacharias DP, Lima MM, Souza AL Jr, et al. Human cutaneous melanoma expresses a significant phosphate-dependent glutaminase activity: a comparison with the surrounding skin of the same patient. Cell Biochem Funct 2003;21:81–4.[Medline]
  39. Hitchens MR, Robbins PD. The role of the transcription factor DP in apoptosis. Apoptosis 2003;8:461–8.[CrossRef][Medline]
  40. Stanelle J, Stiewe T, Theseling CC, Peter M, Putzer BM. Gene expression changes in response to E2F1 activation. Nucleic Acids Res 2002;30:1859–67.[Abstract/Free Full Text]
  41. Banerjee D, Gorlick R, Liefshitz A, et al. Levels of E2F-1 expression are higher in lung metastasis of colon cancer as compared with hepatic metastasis and correlate with levels of thymidylate synthase. Cancer Res 2000;60:2365–7.[Abstract/Free Full Text]
  42. Zhang Y, Turkson J, Carter-Su C, et al. Activation of Stat3 in v-Src-transformed fibroblasts requires cooperation of Jak1 kinase activity. J Biol Chem 2000;275:24935–44.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 Cancer Res.Home page
O. Gautschi, C. G. Tepper, P. R. Purnell, Y. Izumiya, C. P. Evans, T. P. Green, P. Y. Desprez, P. N. Lara, D. R. Gandara, P. C. Mack, et al.
Regulation of Id1 Expression by Src: Implications for Targeting of the Bone Morphogenetic Protein Pathway in Cancer
Cancer Res., April 1, 2008; 68(7): 2250 - 2258.
[Abstract] [Full Text] [PDF]

Home page
 HypertensionHome page
N. H. Lee, B. J. Haas, N. E. Letwin, B. C. Frank, T. V. Luu, Q. Sun, C. D. House, S. Yerga-Woolwine, P. Farms, E. Manickavasagam, et al.
Cross-Talk of Expression Quantitative Trait Loci Within 2 Interacting Blood Pressure Quantitative Trait Loci
Hypertension, December 1, 2007; 50(6): 1126 - 1133.
[Abstract] [Full Text] [PDF]

Home page
 Cancer Res.Home page
S. Balasenthil, A. E. Gururaj, A. H. Talukder, R. Bagheri-Yarmand, T. Arrington, B. J. Haas, J. C. Braisted, I. Kim, N. H. Lee, and R. Kumar
Identification of Pax5 as a Target of MTA1 in B-Cell Lymphomas
Cancer Res., August 1, 2007; 67(15): 7132 - 7138.
[Abstract] [Full Text] [PDF]

Home page
 J. Pharmacol. Exp. Ther.Home page
G. Crudden, R. E. Chitti, and R. J. Craven
Hpr6 (Heme-1 Domain Protein) Regulates the Susceptibility of Cancer Cells to Chemotherapeutic Drugs
J. Pharmacol. Exp. Ther., January 1, 2006; 316(1): 448 - 455.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Irby, R. B.
Right arrow Articles by Lee, N. H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Irby, R. B.
Right arrow Articles by Lee, N. H.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online