- Research article
- Open Access
- Open Peer Review
Gene expression relationship between prostate cancer cells of Gleason 3, 4 and normal epithelial cells as revealed by cell type-specific transcriptomes
© Pascal et al; licensee BioMed Central Ltd. 2009
- Received: 12 March 2009
- Accepted: 18 December 2009
- Published: 18 December 2009
Prostate cancer cells in primary tumors have been typed CD10-/CD13-/CD24hi/CD26+/CD38lo/CD44-/CD104-. This CD phenotype suggests a lineage relationship between cancer cells and luminal cells. The Gleason grade of tumors is a descriptive of tumor glandular differentiation. Higher Gleason scores are associated with treatment failure.
CD26+ cancer cells were isolated from Gleason 3+3 (G3) and Gleason 4+4 (G4) tumors by cell sorting, and their gene expression or transcriptome was determined by Affymetrix DNA array analysis. Dataset analysis was used to determine gene expression similarities and differences between G3 and G4 as well as to prostate cancer cell lines and histologically normal prostate luminal cells.
The G3 and G4 transcriptomes were compared to those of prostatic cell types of non-cancer, which included luminal, basal, stromal fibromuscular, and endothelial. A principal components analysis of the various transcriptome datasets indicated a closer relationship between luminal and G3 than luminal and G4. Dataset comparison also showed that the cancer transcriptomes differed substantially from those of prostate cancer cell lines.
Genes differentially expressed in cancer are potential biomarkers for cancer detection, and those differentially expressed between G3 and G4 are potential biomarkers for disease stratification given that G4 cancer is associated with poor outcomes. Differentially expressed genes likely contribute to the prostate cancer phenotype and constitute the signatures of these particular cancer cell types.
- Prostate Cancer
- Prostate Cancer Cell
- Prostate Cancer Cell Line
- Luminal Cell
- Prostatic Intraepithelial Neoplasia
The Gleason grade is a pathology characterization of prostate tumors. Tumor glands can appear from well-differentiated to poorly differentiated. The degree of differentiation is scaled from 3 to 5 (grades 1 and 2 are no longer in wide use), with 3 to indicate tumors with glandular differentiation, 4 to indicate tumors with aglandular differentiation, and 5 to indicate no differentiation with cancer cells not organized into recognizable structures. Each tumor is assigned a score of two Gleason grades. Gleason 3+3 is associated with favorable outcomes and patient survival while tumors containing 4 or 5 components have a poorer prognosis. A molecular correlate of the Gleason grade was reported where grade 3 could be distinguished from grades 4 and 5 by gene expression, whereas grades 4 and 5 were indistinguishable by gene expression . Another study, however, found a different set of genes associated with degree of tumor differentiation . Reis et al.  found that a higher fraction (6.6%) of intronic ncRNAs were significantly correlated to the degree of prostate tumor differentiation (Gleason score) when compared to unannotated genomic regions (1%) or from exons of known genes (2%). The difference in gene signatures reported by different laboratories has yet to be explained, but is likely due to a combination of external factors as well as actual biological difference. These studies have attempted to define a set of gene expression profiles as determined by microarray analysis that could supplement the traditional diagnostics of pathology and clinical parameters to predict disease course . For example, a 5-gene model was proposed that could segregate recurrent from non-recurrent cancer . Although gene signatures have been shown to associate with Gleason grade , none so far have been able to discriminate locally invasive from non-invasive tumors.
We have previously shown that prostate tumors contained multiple Gleason grades as well as multiple cancer cell types distinguishable by their cluster designation (CD) phenotypes [7, 8]. CD antigens are cell surface molecules, and each cell type has a unique complement of these molecules. This allowed us to isolate, for example, CD26+ luminal cells, CD104+ basal cells, CD49a+ stromal fibromuscular cells, and CD31+ endothelial cells from prostate tissue samples by magnetic cell sorting (MACS) for microarray analysis using the Affymetrix GeneChip to determine cell type-specific gene expression or transcriptome . These cell type transcriptomes are a powerful tool to discover lineage relationship between cell types. We applied the same cell sorting methodology here to determine the transcriptome of CD26+ prostate cancer cells of Gleason 3+3 and Gleason 4+4 tumor specimens. These specimens, unlike those of 3+4 or 4+3, ensured that cells representative of only grade 3 or 4 would be obtained. The transcriptomes were then compared with each other to identify differentially expressed genes, as well as with those of the normal cell types. Furthermore, dataset comparisons between the transcriptomes of primary tumor cells and prostate cancer cell lines, which were established from metastatic lesions and showed unique CD expression [10, 11], demonstrated that they were not representative of cells in primary tumors.
Prostate tumor tissue and cell sorting
Anonymized tissue samples were obtained at radical prostatectomy under approval by the University of Washington Institutional Review Board. Pertinent pathology information was requested from a research coordinator who was authorized to access the patient database in the Department of Urology. The resected glands were cut into 3-mm thick transverse blocks, and frozen sections from representative right and left areas of the apex, middle and base regions were stained to locate the tumor foci. With the stained sections as guide, tumor tissue samples were excised from the corresponding regions and placed in sterile tubes. Processing of tissue pieces for cell sorting was as described [9, 12]. The collected samples were minced and digested by collagenase type I (Invitrogen, Carlsbad, CA) in serum-free RPMI1640 media supplemented with 10-8 M dihydrotestosterone on a magnetic stirrer at room temperature overnight. The resultant cell suspension was filtered with 70-μm Falcon cell strainer to remove debris, diluted with equal volume of Hanks balanced salt solution (HBSS), and aspirated with 18-gauge needle. Cells were pelleted by centrifugation and the digestion media was analyzed by Western blotting for TIMP1 protein  to gauge purity of the cancer specimens. The single cell preparation was partitioned into stromal and epithelial fractions on discontinuous Percoll density gradients (Amersham Pharmacia, Piscataway, NJ). Cells banding at the epithelial density were collected and used for sorting with CD26 antibody. The tumor specimens of 05-179 and 08-032 weighed 0.6 g and 6 g respectively, and were above the minimal weight requirement for cell sorting. The pathology characteristics of tumors from which cancer cells were obtained were as follows - 05-179: Gleason 3+3, T2cN0Mx, 4.7 cc tumor volume; and, 08-032: Gleason 4+4, T3b, 27 cc tumor volume.
MACS cell isolation and FACS analysis of sorted cells
Gradient-purified epithelial cells were labeled with phycoerythrin (PE)-conjugated anti-CD26 (1:20; M-A261, BD PharMingen, San Diego, CA) and sorted by MACS (Miltenyi Biotec, Auburn, CA) as described previously for the normal cell types . Aliquots of positive and negative cell fractions were analyzed by fluorescence activated cell sorting (FACS, Becton Dickinson, Mountain View, CA) to monitor sort efficiency as reported previously; only >85% CD26+ fractions were used for microarray analysis . The sorted cancer cells were pelleted by centrifugation and total RNA was extracted using RNaqueous (Ambion, Austin, TX).
Prostate cancer cell lines
Prostate cancer cell lines LNCaP, C4-2, CL1 (and its subclones CL1.1 and CL1.31), DU145 and PC3 were cultured and harvested for array analysis as described in our previous report on differential gene expression between LNCaP and C4-2 .
Serial 5-μm sections were prepared from frozen blocks, fixed in cold acetone, and processed for immunohistochemistry using a three-step indirect avidin-biotin-peroxidase procedure. The primary antibodies used were the CD antibodies described in our previous reports [7, 16]. Anti-Dlx1 (clone AB5724, Chemicon/Millipore, Billerica, MA), was used at 1:500. Sections were counterstained with hematoxylin. The immunostained sections were imaged with Olympus BX41 microscope (Olympus, Melville, NY) equipped with MircoFire digital camera (Optronics, Goleta, CA). Images were processed with Photoshop CS (Adobe Systems, San Jose, CA).
Western blot analysis
Cell-free tissue digestion media preparations (g tissue/ml media) diluted in equal volume of HBSS were resolved by gradient polyacrylamide gel electrophoresis and transferred to membrane filter for probing by AGR2 antibody (1:500; 1C3, Abnova, Taipei, Taiwan). The 17-kDa AGR2 protein was visualized using ECL (GE Healthcare, Piscataway, NJ). Prostate specific antigen (PSA) antibody (1:1,000; A67-B/E3, Santa Cruz Biotechnology, Santa Cruz, CA) was used as sample loading control.
DNA array analysis
Quality and concentration of RNA were determined using Agilent 2100 Bioanalyzer and RNA Nano or Pico Labchip (Agilent Technologies, Santa Clara, CA) where appropriate. Only RNA samples of sufficient concentration and showed no degradation as evidenced by distinct 18S and 28S ribosomal bands were used for microarray experiments. The Human Genome U133 Plus 2.0 GeneChips (Affymetrix, Santa Clara, CA) were used for expression profiling. The U133 array contains probesets representing 54,675 genes, splice variants, and ESTs. The GeneChips were prepared, hybridized, and scanned according to the protocols provided by Affymetrix (P/N 702232 Rev. 2) [9, 15]. In brief, 200 ng RNA was reverse transcribed with poly (dT)/T7 promoter primer, and the cDNA was made double-stranded. In vitro transcription was performed with biotinylated ribonucleotides, and the biotin-labeled cRNA was hybridized to the GeneChips. The chips were washed and stained with streptavidin-PE using FS-450 fluidics station (Affymetrix). Data was collected with Affymetrix GeneChip Scanner 3000. A total of three CD26+ prostate cancer array datasets were obtained from two patients; with 05-179 analyzed twice and 08-032 analyzed once. Two replicates were run for each prostate cancer cell line: CL1, CL1.1, CL1.31, DU145 and PC3. Datasets previously obtained from five non-sorted whole-tissue tumor specimens and their matched non-cancer specimens were used for comparison . All transcriptome datasets are deposited in our public UESC database http://scgap.systemsbiology.net/, which makes them available for data query .
Computational analysis of datasets
For differential gene expression, datasets were analyzed by HTself, a self-self based statistical method for low replication microarrays  using a strategy very similar to that described by Pascal et al. . To apply this method, all possible combination of pair-wise comparisons among experiments were taken to create sets of ratios. Gene expression level was defined as the normalized and summarized intensities of each GeneChip probeset, and was presented as its logarithmic value: X = log2(Normalized intensity). This step was carried out using the standard robust multi-array average (RMA) method , implemented in the analysis pipeline SBEAMS . The strength of differential expression between any pair of experiments was estimated by M i = log2(ratio) = X i -X a , where a represented one particular cell type and i represented each given cell type in the set: B (basal), L (luminal), S (stromal), E (endothelial), G3 (Gleason 3) or G4 (Gleason 4). To estimate reliability, the HTself method used virtual self-self experiments (i.e., G3 cancer vs. G3 cancer) to derive intensity-dependent cutoffs. Accordingly, a probeset was considered significantly differentially expressed if at least 80% of its log-ratio combinations were outside the 99.9% credibility intensity-dependent cutoff. Moreover, an average greater than 8-fold difference in expression level was chosen. The computational analysis results were verified by dataset query of known differentially expressed genes. In principal components analysis (PCA) of the various transcriptome datasets, a gene expression subspace was obtained that brought out the principal sources of variability among transcriptomes of the different prostate cell types. The rotation matrix was obtained by using averages of X G3 , X G4 , X L , X B , X S , and X E , and these were plotted as projections on the principal components space. The various transcriptomes were then projected into this PCA-generated subspace, which could be rotated freely to visualize the spatial separation of the individual nodes. Pathway analysis of selected genes was done with KEGG .
Gene expression of cancer cells of a Gleason 3+3 tumor
For the single specimen of 05-179 cancer, statistical analysis of gene expression with traditional methods was not applicable. Instead, HTself was used to analyze for differential gene expression between the CD26+ G3 cancer transcriptome and the previously determined transcriptome of the normal counterpart, CD26+ luminal . Cluster analysis comparing the transcriptome datasets of CD26+ G3 cancer cells and the previously determined principal cell types of the prostate  was summarized (Figure 1B). The gene expression profile of CD26+ cancer was closer to that of luminal than basal, in agreement with previous CD phenotype analysis  and cell type-specific expression data of cytokeratin subtypes .
The array data was reported as RMA-normalized Affymetrix signal intensities implemented in SBEAMS or as a composite value M i = log2(ratio) = X i -X a , where a represented cancer and i represented normal. For dataset query, we enabled simple searches to display results in downloadable formats as previously described . Query results of signal intensity values were graphically shown on a grayscale. Gene expression levels determined by microarray for CD10/MME, CD13/ANPEP, CD24, CD26/DPP4, and CD38 were concordant with immunohistochemistry, i.e., CD26+/CD13-/CD10-; increased CD24 and decreased CD38 compared to luminal cells (Figure 1C). The CD26+ cancer cells also had elevated expression of prostate cancer-associated genes AMACR (α-methylacyl-CoA racemase), HPN (hepsin) and PCA3 (prostate cancer antigen 3). These results indicated minimal contamination of CD26+ luminal cells from non-cancer tissue that might be present in the tumor specimen. In the same display, the expression profiles of these CD in the various prostate cancer cell lines were in agreement with their CD expression previously determined by flow cytometry [10, 11]. For example, LNCaP and C4-2 are CD10+; C4-2 differs from LNCaP in being CD26+; PC3 is CD13+, DU145 is CD24+. Overall, the gene expression and CD phenotype of the CD26+ cancer cells differed substantially from those of the cancer cell lines indicating that it is unlikely that any of the cell lines, which were established from metastatic lesions, could have derived from the CD26+/CD10-/CD13- cancer cell type found in a majority of tumors.
Differential gene expression between CD26+ G3 cancer cells and CD26+ luminal cells analyzed by HTself identified 121 genes with increased cancer expression and 86 genes with decreased cancer expression by at least 8-fold relative to luminal cells (see Additional File 1 for full list, and Additional File 2 for self-self experiment and HTself cutoff for differential expression). Although they were up-regulated, AMACR and HPN were not among those with the highest fold-increase in expression in this set. BRE, which encodes a death receptor-associated anti-apoptotic protein with an inhibitory function in the mitochondrial pathway, was the most up-regulated, followed by FLJ242562, R55784, MMP12, and KCNH8. IL24, which was highly up-regulated in this dataset, was recently found to be up-regulated in tumor-associated stromal cells as well . The many ESTs (highlighted in tan, Additional File 1), e.g., KIAA0960, also contributed to the signature of this prostate cancer cell type. Several Wnt pathway genes (highlighted in yellow, Additional File 1) were identified. Down-regulated genes by at least 8-fold relative to luminal cells included peptidases [ENPEP (glutamyl aminopeptidase), CPM (carboxypeptidase M), in addition to ANPEP/CD13 and MME/CD10], TNFRSF19 (involved in apoptosis), ADD3 (adducin, important for cell-cell contact in epithelial tissues), GEDP (the gene differentially expressed in the prostate), and MSMB/PSP94. MSMB, KLK3/PSA and ACPP (prostatic acid phosphatase) constitute the well-known three most abundant prostatic secretory products. Unlike KLK3 and ACPP, luminal expression of MSMB appears not to be regulated by stromal influence , and is known to be down-regulated in cancer cells .
Gene expression of cancer cells of a Gleason 4+4 tumor
The 08-032 tumor was scored Gleason 4+4, margin positive with seminal vesicle involvement. The Gleason scoring of 08-032 indicated that virtually all the cancer cells represented the Gleason grade 4 (G4) type while those of 05-179 the G3 type (no frozen tumor tissue block was available for 08-032 for CD immunohistochemistry). Verification of the "purity" of this rather large (6 g) tumor specimen, i.e., absence of non-cancer, was determined by Western blot analysis of tissue collagenase digestion media for TIMP1 protein (data published in ref. ). TIMP1 protein is produced by luminal cells and not cancer cells . Specimen 08-032 showed minimal reactivity for TIMP1, indicating a relatively pure sample with little contamination by non-cancer. Differential gene expression was analyzed between G3 and luminal cells, and between G4 and G3 cells.
Comparison of cancer cell and cancer tissue transcriptomes
Prostate tumors used to generate CP tissue expression datasets.
tumor volume (cc)
Wnt signaling in prostate cancer
FZD8, WIF1 and SOX7 were identified by HTself as differentially expressed between cancer and luminal cells, and in whole tissue CP compared to NP (see Figure 3 and Additional File 1). In the cancer cells as well as cancer cell lines (data not shown), there was increased expression of Wnt receptor FZD8; decreased expression of Wnt inhibitor WIF1, which functions to sequester Wnt molecules from receptor binding, and that of the transcription factor SOX7. Stromal expression of WNT2 and SFRP4 suggests Wnt signaling between stromal and epithelial cells in the prostate. SFRP4 was also recently found to up-regulated in tumor-associated stromal cells . FZD8, WIF1, SOX7 identified in our analysis appeared to contribute to an elevated expression of genes controlled by the Wnt molecule (increase in receptors and decrease in inhibitors). Uncontrolled Wnt signaling had been reported in many types of cancer, including prostate . Here, we identified three key members of this signaling pathway that were abnormally expressed in the CD26+ cancer cells.
Transcription factors with increased expression in cancer
Prostate cancer biomarkers
Genes encoding secreted or extracellular proteins with increased expression in prostate cancer may serve as potential biomarkers for diagnosis or prognosis. Several genes that encode these proteins were upregulated in the CD26+ cancer cells: AGR2, BCMP11, CEACAM5/CD66e, CRISP3, KCNG3, KCNH8, LOX and NEO1 (Figure 4B). These proteins could be measured in biospecimens by using, for example, quantitative proteomics methods. Except for AGR2 (anterior gradient 2, signal intensity at 4168 vs. 77 in luminal), their signal intensity indicated that their abundance was in the moderate class (e.g., 268 for BCMP11; 217 for CRISP3). BCMP11 (breast cancer membrane protein, also known as AGR3) was originally discovered by proteomics in breast cancer cell lines T-47D and MCF-7 . BCMP11 was specific to cancer and not found in normal ductal epithelial cells. The related AGR2 was previously shown to be androgen-inducible and overexpressed in prostate cancer [32, 33]. A ~19-kDa AGR2 protein was detected in cancer tissue preparation by mass spectrometry of 2-D gel electrophoresis spot . Western blot analysis of CP vs. NP media confirmed overexpression of AGR2 protein in cancer (Figure 4C). Note that the lower level of AGR2 protein in 08-032 CP was concordant with decreased expression of AGR2 in this G4 cancer transcriptome (see below). LOX and NEO1 also appeared to be expressed by NP stromal cells, which would make them less suitable as cancer biomarkers.
Genes potentially associated with node metastasis
The cancer cell transcriptomes were compared to the cancer cell lines for markers potentially associated with advanced disease. These cells have been in culture for many years and have undergone changes as a result; however, the expression of certain genes related to prostate cancer metastasis could still be maintained. For example, for lymph node metastasis, potentially important genes could be identified through a comparison of the CD26+ cancer and LNCaP transcriptomes. Node metastases are CD10+, whereas the CD26+ cancer cells are CD10-, so it is possible that LNCaP is a representative of the CD10+ cancer cell type found in ~30% of prostate tumors and nearly all positive nodes . The two datasets were analyzed to identify genes differentially expressed by ≥8-fold. All genes known to be involved in CELL PROLIFERATION as annotated in GeneOntology were filtered out. There are other genes whose expression appeared to be linked to cell culturing , but because they belong to different functional categories they were not filtered out. Genes with increased expression in LNCaP (and its lineage-related C4-2, which has extensive shared gene expression ) were examined in CD26+ cancer. Figure 4D shows the expression of cancer/testis melanoma antigens MAGE and GAGE. Expression of GAGE-7 in LNCaP was reported in the literature . Kufer et al.  reported that MAGE expression in circulating epithelial cells was detected significantly more in patients carrying a higher risk of disease recurrence than those with a lower risk. Patients with metastasis had higher MAGE expression compared to those with localized disease. There was also concordance in expression between primary tumors and bone marrow. In line with these genes, a gene called preferentially expressed antigen in melanoma (PRAME) was also scored in LNCaP and C4-2. Other candidates included DDC, MANEA, ZWINT, while ERG and SERPINA3 were examples of genes scored with decreased expression in LNCaP compared to CD26+ cancer. Genes with the least fold difference in expression, not unexpectedly, were those encoding ribosomal proteins. For pathway visualization, the LNCaP>CD26+ cancer analysis result was uploaded to KEGG. Two notable networks for the data were ANDROGEN AND ESTROGEN METABOLISM and FATTY ACID METABOLISM. It is known that androgens and lipid metabolism are coupled in LNCaP cells leading to accumulation of neutral lipids .
Comparison between Gleason 3 and Gleason 4 cancer cells
The expression of epithelial genes such as ACPP, AZGP1 (zinc α2-glycoprotein), KLK2/hK2, KLK3, and MSMB showed a trend toward down-regulation relative to luminal cells (Figure 5C). With regard to the epithelial cytokeratins (KRT), relative expression levels of luminal cell KRT8 and KRT18 were increased in G3 and G4, whereas those of basal and intermediate cell KRT5, KRT14, KRT17 were significantly decreased. KRT19, which is expressed by both basal and luminal epithelial cells and is thought to be a marker of the transit-amplifying population , was slightly increased in G3, but decreased in G4 with respect to luminal cells.
The LNCaP genes as potential markers of metastatic potential were also queried between G3 and G4 (Figure 5D). Note the decrease in ERG (2.3-fold) and SERPINA3 (53.5-fold) in G4 cells as that found in the G3/LNCaP comparison. Of the others, PRAME showed a 13.6-fold increase in G4. No expression of MAGE or GAGE was seen.
For pathway visualization, the G3 vs. G4 cancer analysis result was uploaded to KEGG. Notable networks for the data included EXTRACELLULAR MATRIX-RECEPTOR INTERACTION and CYTOKINE-CYTOKINE RECEPTOR INTERACTION. One of the most differentially up-regulated genes in G4 cells was CCL3 (chemokine C-C motif ligand, Additional File 3). CCL3 encodes the α subunit of MIP1 (macrophage inflammatory protein-1), which is active in recruitment and activation of polymorphonuclear leukocytes in acute inflammation. The β subunit gene, CCL4, was also differentially expressed in G4. CCL4 was recently reported to be associated with prostate cancer recurrence by Blum et al. .
Comparison with other cancer datasets
Transcriptome analysis of CD26+ prostate cancer cells demonstrated that it was possible to isolate and characterize gene expression in cancer cells from solid tumors weighing 0.5 g or more utilizing our cell sorting methodology. The extensive tissue processing also did not appear to alter substantially the gene expression profiles of the cancer cells or the tumor-associated stromal cells reported previously . Because each tumor could be unique in certain aspects we chose to determine differentially expressed genes by HTself data analysis rather than treating them as biological replicates of each other. The CD26+ cancer datasets obtained were representative of Gleason grades 3 and 4. Grade 3 tumors show a glandular morphology whereas grade 4 tumors are typically less glandular. Sorting for grade 5 is problematic as Gleason 5+5 tumors are rarely seen. According to many pathologists, cells of Gleason grade 3 are those that are first recognizable as cancerous. Compared to the normal counterpart luminal cells, the expression of a variety of genes was found to be different. This could be due to the altered expression of several transcription factors. Some of the genes identified are involved in cell division and response to apoptotic signaling. Genes of the Wnt pathway, which is important in epithelial-stromal cell interaction, appeared to be dysregulated to produce increased Wnt signaling in the cancer cells. Targeting this particular pathway may prove effective in prostate cancer.
The cancer cell types profiled in this study were the CD10- luminal-like types, although one populated a Gleason 3+3 tumor and the other a Gleason 4+4 tumor. Their respective transcriptomes represent the signatures of two distinct prostate cancer cell types. As greater numbers of cancer cell types from different cases are analyzed, it will be interesting to determine if dissimilarities in gene expression increase with higher Gleason grades. This could be visualized using PCA. The number of possible prostate cancer cell types is currently unknown, and tumor behavior could be dependent on its composition of various cancer cell types. For example, the CD10+ type is associated with lymph node metastasis and poor outcomes [8, 45]. Isolation of this cell type is hampered by the fact that luminal cells are also CD10+ and CD10 staining of biopsy is currently not a standard practice. Other cell types like CD44+  are even more rarely encountered.
The identification of candidate biomarkers for cancer detection and disease stratification in this study is of practical utility. Their identity would provide for a targeted approach to analyze for their presence in body fluids such as urine and blood. Genes that encode secreted proteins are particularly useful as most, if not all, clinical tests are based on immunodetection of these types of molecules. Since cancer cells are non-functional, it was not expected that many such proteins would be found. As can be inferred from the array signal intensity levels, expression of the major functional prostatic secretory proteins ACPP, AZGP1, KLK2, KLK3, and MSMB was down-regulated in the cancer cells compared to the luminal cells. Of the candidates, AGR2 appeared most promising. AGR2 is a relatively small protein and is produced apparently at a high level based on the array signals (>4000). AGR2 was also readily detectable in the tissue digestion media of cancer specimens by Western blot analysis. AGR2 expression may be decreased in higher Gleason tumors. Thus, AGR2 could potentially serve as a target biomarker for early disease detection. In contrast, the secreted CCL3/CCL4 could serve as a biomarker of tumors containing a G4 cancer cell type. Care must be taken to ensure that these potential biomarker proteins are not produced significantly by other cell types. Expression of these candidates will be examined in greater detail by tissue microarray analysis. A multimarker panel could conceivably be generated to incorporate all informative genes to diagnose prostate cancer.
In summary, we have determined the transcriptomes of CD26+ primary prostate cancer cells isolated from Gleason grade 3 and grade 4. Grade 4 is usually associated with poor clinical outcomes. Therefore, genes differentially expressed between these two grades have prognostic potential. The method we have described to isolate primary prostate cancer cells for transcriptome analysis should be applicable to other solid tumors.
We thank Adam van Mason for specimen collection; Dr. Marc Dall'Era for comments. This research was funded in part by grants CA111244, CA98699, CA85859 and DK63630 from the National Institutes of Health. RZNV was partially supported by P50-GMO-76547 when he was at the Institute for Systems Biology.
- True L, Coleman I, Hawley S, Huang CY, Gifford D, Coleman R, Beer TM, Gelmann E, Datta M, Mostaghel E, Knudsen B, Lange P, Vessella R, Lin D, Hood L, Nelson PS: A molecular correlate to the Gleason grading system for prostate adenocarcinoma. Proc Natl Acad Sci USA. 2006, 103 (29): 10991-10996. 10.1073/pnas.0603678103.View ArticlePubMedPubMed CentralGoogle Scholar
- Glinsky GV, Glinskii AB, Stephenson AJ, Hoffman RM, Gerald WL: Gene expression profiling predicts clinical outcome of prostate cancer. J Clin Invest. 2004, 113 (6): 913-923.View ArticlePubMedPubMed CentralGoogle Scholar
- Reis EM, Nakaya HI, Louro R, Canavez FC, Flatschart ÁV, Almeida GT, Egidio CM, Paquola AC, Machado AA, Festa F, Yamamoto D, Alvarenga R, da Silva CC, Brito GC, Simon SD, Moreira-Filho CA, Leite KR, Camara-Lopes LH, Campos FS, Gimba E, Vignal GM, El-Dorry H, Sogayar MC, Barcinski MA, da Silva AM, Verjovski-Almeida S: Antisense intronic non-coding RNA levels correlate to the degree of tumor differentiation in prostate cancer. Oncogene. 2004, 23 (39): 6684-6692. 10.1038/sj.onc.1207880.View ArticlePubMedGoogle Scholar
- Bettuzzi S, Scaltriti M, Caporali A, Brausi M, D'Arca D, Astancolle S, Davalli P, Corti A: Successful prediction of prostate cancer recurrence by gene profiling in combination with clinical data: a 5-year follow-up study. Cancer Res. 2003, 63 (13): 3469-3472.PubMedGoogle Scholar
- Febbo PG, Sellers WR: Use of expression analysis to predict outcome after radical prostatectomy. J Urol. 2003, 170 (6 Pt 2): S11-19. 10.1097/01.ju.0000095567.03807.a1.View ArticlePubMedGoogle Scholar
- Singh D, Febbo PG, Ross K, Jackson DG, Manola J, Ladd C, Tamayo P, Renshaw AA, D'Amico AV, Richie JP, Lander ES, Loda M, Kantoff PW, Golub TR, Sellers WR: Gene expression correlates of clinical prostate cancer behavior. Cancer Cell. 2002, 1 (2): 203-209. 10.1016/S1535-6108(02)00030-2.View ArticlePubMedGoogle Scholar
- Liu AY, Roudier MP, True LD: Heterogeneity in primary and metastatic prostate cancer as defined by cell surface CD profile. Am J Pathol. 2004, 165 (5): 1543-1556.View ArticlePubMedPubMed CentralGoogle Scholar
- Dall'Era MA, True LD, Siegel AF, Porter MP, Sherertz TM, Liu AY: Differential expression of CD10 in prostate cancer and its clinical implication. BMC Urol. 2007, 7: 3-10.1186/1471-2490-7-3.View ArticlePubMedPubMed CentralGoogle Scholar
- Oudes AJ, Campbell DS, Sorensen CM, Walashek LS, True LD, Liu AY: Transcriptomes of human prostate cells. BMC Genomics. 2006, 7: 92-10.1186/1471-2164-7-92.View ArticlePubMedPubMed CentralGoogle Scholar
- Liu AY: Differential expression of cell surface molecules in prostate cancer cells. Cancer Res. 2000, 60 (13): 3429-3434.PubMedGoogle Scholar
- Freedland SJ, Seligson DB, Liu AY, Pantuck AJ, Paik SH, Horvath S, Wieder JA, Zisman A, Nguyen D, Tso CL, Palotie AV, Belldegrun AS: Loss of CD10 (neutral endopeptidase) is a frequent and early event in human prostate cancer. Prostate. 2003, 55 (1): 71-80. 10.1002/pros.10202.View ArticlePubMedGoogle Scholar
- Liu AY, True LD, LaTray L, Nelson PS, Ellis WJ, Vessella RL, Lange PH, Hood L, Engh van den G: Cell-cell interaction in prostate gene regulation and cytodifferentiation. Proc Natl Acad Sci USA. 1997, 94 (20): 10705-10710. 10.1073/pnas.94.20.10705.View ArticlePubMedPubMed CentralGoogle Scholar
- Liu AY, Zhang H, Sorensen CM, Diamond DL: Analysis of prostate cancer by proteomics using tissue specimens. J Urol. 2005, 173 (1): 73-78.View ArticlePubMedGoogle Scholar
- Pascal LE, Goo YA, Vêncio RZN, Page LS, Chambers AA, Liebeskind ES, Takayama TK, True LD, Liu AY: Gene expression down-regulation in CD90+ prostate tumor-associated stromal cells involves potential organ-specific genes. BMC Cancer. 2009, 9 (1): 317-10.1186/1471-2407-9-317.View ArticlePubMedPubMed CentralGoogle Scholar
- Oudes AJ, Roach JC, Walashek LS, Eichner LJ, True LD, Vessella RL, Liu AY: Application of Affymetrix array and Massively Parallel Signature Sequencing for identification of genes involved in prostate cancer progression. BMC Cancer. 2005, 5: 86-10.1186/1471-2407-5-86.View ArticlePubMedPubMed CentralGoogle Scholar
- Liu AY, True LD: Characterization of prostate cell types by CD cell surface molecules. Am J Pathol. 2002, 160 (1): 37-43.View ArticlePubMedPubMed CentralGoogle Scholar
- Pascal LE, Deutsch EW, Campbell DS, Korb M, True LD, Liu AY: The urologic epithelial stem cell database (UESC) - a web tool for cell type-specific gene expression and immunohistochemistry images of the prostate and bladder. BMC Urol. 2007, 7: 19-10.1186/1471-2490-7-19.View ArticlePubMedPubMed CentralGoogle Scholar
- Vêncio RZ, Koide T: HTself: self-self based statistical test for low replication microarray studies. DNA Res. 2005, 12 (3): 211-214. 10.1093/dnares/dsi007.View ArticlePubMedGoogle Scholar
- Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, Speed TP: Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003, 4 (2): 249-264. 10.1093/biostatistics/4.2.249.View ArticlePubMedGoogle Scholar
- Marzolf B, Deutsch EW, Moss P, Campbell D, Johnson MH, Galitski T: SBEAMS-Microarray: database software supporting genomic expression analyses for systems biology. BMC Bioinformatics. 2006, 7: 286-10.1186/1471-2105-7-286.View ArticlePubMedPubMed CentralGoogle Scholar
- Kanehisa M, Goto S: KEGG: kyoto encyclopedia of genes and genomes. Nucl Acids Res. 2000, 28 (1): 27-30. 10.1093/nar/28.1.27.View ArticlePubMedPubMed CentralGoogle Scholar
- Okada H, Tsubura A, Okamura A, Senzaki H, Naka Y, Komatz Y, Morii S: Keratin profiles in normal/hyperplastic prostates and prostate carcinoma. Virchows Arch A Pathol Anat Histopathol. 1992, 421 (2): 157-161. 10.1007/BF01607049.View ArticlePubMedGoogle Scholar
- Chan PS, Chan LW, Xuan JW, Chin JL, Choi HL, Chan FL: In situ hybridization study of PSP94 (prostatic secretory protein of 94 amino acids) expression in human prostates. Prostate. 1999, 41 (2): 99-109. 10.1002/(SICI)1097-0045(19991001)41:2<99::AID-PROS4>3.0.CO;2-C.View ArticlePubMedGoogle Scholar
- Goo YA, Goodlett DR, Pascal LE, Worthington KD, Vessella RL, True LD, Liu AY: Stromal mesenchyme cell genes of the human prostate and bladder. BMC Urol. 2005, 5: 17-10.1186/1471-2490-5-17.View ArticlePubMedPubMed CentralGoogle Scholar
- Terry S, Yang X, Chen MW, Vacherot F, Buttyan R: Multifaceted interaction between the androgen and Wnt signaling pathways and the implication for prostate cancer. J Cell Biochem. 2006, 99 (2): 402-410. 10.1002/jcb.20983.View ArticlePubMedGoogle Scholar
- Pelzer AE, Bektic J, Haag P, Berger AP, Pycha A, Schäfer G, Rogatsch H, Horninger W, Bartsch G, Klocker H: The expression of transcription factor activating transcription factor 3 in the human prostate and its regulation by androgen in prostate cancer. J Urol. 2006, 175 (4): 1517-1522. 10.1016/S0022-5347(05)00651-8.View ArticlePubMedGoogle Scholar
- Brubaker KD, Vessella RL, Brown LG, Corey E: Prostate cancer expression of runt-domain transcription factor Runx2, a key regulator of osteoblast differentiation and function. Prostate. 2003, 56 (1): 13-22. 10.1002/pros.10233.View ArticlePubMedGoogle Scholar
- Rostad K, Mannelqvist M, Halvorsen OJ, Øyan AM, Bø TH, Stordrange L, Olsen S, Haukaas SA, Lin B, Hood L, Jonassen I, Akslen LA, Kalland KH: ERG upregulation and related ETS transcription factors in prostate cancer. Int J Oncol. 2007, 30 (1): 19-32.PubMedGoogle Scholar
- Tomlins SA, Rhodes DR, Perner S, Dhanasekaran SM, Mehra R, Sun XW, Varambally S, Cao X, Tchinda J, Kuefer R, Lee C, Montie JE, Shah RB, Pienta KJ, Rubin MA, Chinnaiyan AM: Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Science. 2005, 310 (5748): 644-648. 10.1126/science.1117679.View ArticlePubMedGoogle Scholar
- Cantile M, Kisslinger A, Cindolo L, Schiavo G, D'Antò V, Franco R, Altieri V, Gallo A, Villacci A, Tramontano D, Cillo C: cAMP induced modifications of HOX D gene expression in prostate cells allow the identification of a chromosomal area involved in vivo with neuroendocrine differentiation of human advanced prostate cancers. J Cell Physiol. 2005, 205 (2): 202-210. 10.1002/jcp.20384.View ArticlePubMedGoogle Scholar
- Adam PJ, Boyd R, Tyson KL, Fletcher GC, Stamps A, Hudson L, Poyser HR, Redpath N, Griffiths M, Steers G, Harris AL, Patel S, Berry J, Loader JA, Townsend RR, Daviet L, Legrain P, Parekh R, Terrett JA: Comprehensive proteomic analysis of breast cancer cell membranes reveals unique proteins with potential roles in clinical cancer. J Biol Chem. 2003, 278 (8): 6482-6489. 10.1074/jbc.M210184200.View ArticlePubMedGoogle Scholar
- Zhang JS, Gong A, Cheville JC, Smith DI, Young CY: AGR2, an androgen-inducible secretory protein overexpressed in prostate cancer. Genes Chromosomes Cancer. 2005, 43 (3): 249-259. 10.1002/gcc.20188.View ArticlePubMedGoogle Scholar
- Zhang Y, Forootan SS, Liu D, Barraclough R, Foster CS, Rudland PS, Ke Y: Increased expression of anterior gradient-2 is significantly associated with poor survival of prostate cancer patients. Prostate Cancer Prostatic Dis. 2007, 10 (3): 293-300. 10.1038/sj.pcan.4500960.View ArticlePubMedGoogle Scholar
- Kovalev LI, Shishkin SS, Khasigov PZ, Dzeranov NK, Kazachenko AV, Toropygin I, Mamykina SV: [Identification of AGR2 protein, a novel potential cancer marker, using proteomics technologies]. Prikl Biokhim Mikrobiol. 2006, 42 (4): 480-484.PubMedGoogle Scholar
- Liu AY, LaTray L, Engh van den G: Changes in cell surface molecules associated with in vitro culture of prostatic stromal cells. Prostate. 2000, 44 (4): 303-312. 10.1002/1097-0045(20000901)44:4<303::AID-PROS7>3.0.CO;2-J.View ArticlePubMedGoogle Scholar
- Chen ME, Lin SH, Chung LW, Sikes RA: Isolation and characterization of PAGE-1 and GAGE-7. New genes expressed in the LNCaP prostate cancer progression model that share homology with melanoma-associated antigens. J Biol Chem. 1998, 273 (28): 17618-17625. 10.1074/jbc.273.28.17618.View ArticlePubMedGoogle Scholar
- Kufer P, Zippelius A, Lutterbüse R, Mecklenburg I, Enzmann T, Montag A, Weckermann D, Passlick B, Prang N, Reichardt P, Dugas M, Köllermann MW, Pantel K, Riethmüller G: Heterogeneous expression of MAGE-A genes in occult disseminated tumor cells: a novel multimarker reverse transcription-polymerase chain reaction for diagnosis of micrometastatic disease. Cancer Res. 2002, 62 (1): 251-261.PubMedGoogle Scholar
- Swinnen JV, Esquenet M, Goossens K, Heyns W, Verhoeven G: Androgens stimulate fatty acid synthase in the human prostate cancer cell line LNCaP. Cancer Res. 1997, 57 (6): 1086-1090.PubMedGoogle Scholar
- Inokuchi J, Lau A, Tyson DR, Ornstein DK: Loss of annexin A1 disrupts normal prostate glandular structure by inducing autocrine IL-6 signaling. Carcinogenesis. 2009, 30 (7): 1082-1088. 10.1093/carcin/bgp078.View ArticlePubMedPubMed CentralGoogle Scholar
- Xiao W, Zhang Q, Jiang F, Pins M, Kozlowski JM, Wang Z: Suppression of prostate tumor growth by U19, a novel testosterone-regulated apoptosis inducer. Cancer Res. 2003, 63 (15): 4698-4704.PubMedGoogle Scholar
- Bjartell A, Johansson R, Björk T, Gadaleanu V, Lundwall Å, Lilja H, Kjeldsen L, Udby L: Immunohistochemical detection of cysteine-rich secretory protein 3 in tissue and in serum from men with cancer or benign enlargement of the prostate gland. Prostate. 2006, 66 (6): 591-603. 10.1002/pros.20342.View ArticlePubMedGoogle Scholar
- Hudson DL, Guy AT, Fry P, O'Hare MJ, Watt FM, Masters JR: Epithelial cell differentiation pathways in the human prostate: identification of intermediate phenotypes by keratin expression. J Histochem Cytochem. 2001, 49 (2): 271-278.View ArticlePubMedGoogle Scholar
- Blum DL, Koyama T, M'Koma AE, Iturregui JM, Martinez-Ferrer M, Uwamariya C, Smith JA, Clark PE, Bhowmick NA: Chemokine markers predict biochemical recurrence of prostate cancer following prostatectomy. Clin Cancer Res. 2008, 14 (23): 7790-7797. 10.1158/1078-0432.CCR-08-1716.View ArticlePubMedPubMed CentralGoogle Scholar
- Tomlins SA, Mehra R, Rhodes DR, Cao X, Wang L, Dhanasekaran SM, Kalyana-Sundaram S, Wei JT, Rubin MA, Pienta KJ, Shah RB, Chinnaiyan AM: Integrative molecular concept modeling of prostate cancer progression. Nat Genet. 2007, 39 (1): 41-51. 10.1038/ng1935.View ArticlePubMedGoogle Scholar
- Fleischmann A, Schlomm T, Huland H, Köllermann J, Simon P, Mirlacher M, Salomon G, Chun FH, Steuber T, Simon R, Sauter G, Graefen M, Erbersdobler A: Distinct subcellular expression patterns of neutral endopeptidase (CD10) in prostate cancer predict diverging clinical courses in surgically treated patients. Clin Cancer Res. 2008, 14 (23): 7838-7842. 10.1158/1078-0432.CCR-08-1432.View ArticlePubMedGoogle Scholar
- Paradis V, Eschwège P, Loric S, Dumas F, Ba N, Benoît G, Jardin A, Bedossa P: De novo expression of CD44 in prostate carcinoma is correlated with systemic dissemination of prostate cancer. J Clin Pathol. 1998, 51 (11): 798-802. 10.1136/jcp.51.11.798.View ArticlePubMedPubMed CentralGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://0-www.biomedcentral.com.brum.beds.ac.uk/1471-2407/9/452/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.