| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Imaging, Diagnosis, Prognosis |
Authors' Affiliations: 1 Vascular Biology Program, Department of Surgery, Children's Hospital; 2 Department of Pathology, Brigham and Women's Hospital; 3 Departments of Orthopedic Surgery and Biostatistics, Children's Hospital, Harvard Medical School; 4 Department of Urology, Massachusetts General Hospital, Boston, Massachusetts; and 5 Department of Urology, University Hospital, Ulm, Germany
Requests for reprints: Bruce R. Zetter, Vascular Biology Program, Department of Surgery, Karp Family Research Laboratories, 11.125, Children's Hospital, 300 Longwood Avenue, Boston, MA 02115. Phone: 617-919-2320; Fax: 617-730-0268; E-mail: bruce.zetter{at}childrens.harvard.edu.
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
Experimental Design: We investigated collagen XXIII expression in prostate cancer tissue and did a retrospective analysis of association with prostate-specific antigen (PSA)–defined disease recurrence. The presence of collagen XXIII in prostate cancer patient urine was also assessed before and after prostatectomy.
Results: Collagen XXIII protein was detected at very low levels in benign prostate tissue and was significantly increased in prostate cancer. Distant metastases exhibited significantly higher collagen XXIII levels compared with either localized prostate cancer or regional (lymph node) metastases. Patients with high collagen XXIII levels had a 2.8-fold higher risk of PSA failure with median time to failure of 8.1 months, compared with low collagen XXIII patients with a median time to failure of 5 years. Multivariate Cox regression showed that the presence of collagen XXIII was significantly associated with time to PSA recurrence, independent of other clinical variables. Collagen XXIII was also detected in prostate cancer patient urine, with reduced levels after prostatectomy, indicating potential as a noninvasive fluid biomarker.
Conclusions: We present the first report demonstrating increased collagen XXIII expression in prostate cancer tissue. We show that collagen XXIII level is a significant independent predictor of PSA-defined disease recurrence, suggesting a potential role as a molecular biomarker of prostate cancer progression and metastasis.
The progression of prostate cancer to a metastatic state involves multiple cellular and genetic changes that include increased expression of metastasis related genes and decreased expression of metastasis suppressor genes. A profile of the molecules expressed by a cancer cell during its transformation from a benign to a metastatic phenotype would allow us a better understanding of the process of tumor progression, provide molecular markers that could be used to predict disease outcome, and potentially provide new targets for therapeutic intervention. In the last few years, cDNA microarray analysis of normal prostate, prostate cancer, and metastasis has revealed many of the underlying molecular changes that occur during prostate cancer progression. Among those, some have now been shown to correlate with prostate cancer progression, including enhancer of zeste homologue 2 (5), 
-methylacyl-CoA racemase (6), and Hepsin (7, 8).
Using mRNA differential display analysis as a search tool for biomarkers of metastasis, we had identified a novel collagen, type XXIII, as being elevated in highly metastatic rat prostate carcinoma cells relative to levels in poorly metastatic sublines (9). Collagen XXIII is a member of the emerging family of transmembrane collagens. The protein consists of a short amino-terminal cytoplasmic domain, a transmembrane region, and three collagenous domains flanked by short noncollagenous regions. In this report, we investigated the expression of collagen XXIII in prostate cancer tissue, and present evidence that collagen XXIII is associated with metastasis and PSA-defined disease recurrence. Analysis of prostate cancer patient urine samples before and after prostatectomy further revealed that collagen XXIII may have potential as a biomarker in human fluids.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Antigen affinity purification was done over the collagen XXIII carboxyl-terminal GST fusion protein, cross-linked to glutathione Sepharose using dimethyl pimelimidate (Pierce Biotechnology), after passage over a GST peptide column to remove GST reactive antibodies. Fractions were eluted using 0.15 mol/L NaCl, 0.2 mol/L glycine HCl (pH 2.0), immediately neutralized with 2 mol/L Tris-HCl (pH 8.5), and protein-containing fractions pooled as purified IgG. For indirect immunofluorescent blocking studies, PC-3 cells plated on fibronectin-coated coverslips were transfected with pcDNA3 human collagen XXIII, fixed 48 h posttransfection, and stained with 6.1B antiserum preincubated with GST protein or GST-collagen XXIII protein. Staining was detected with Texas red–conjugated anti-rabbit IgG (Amersham Biosciences).
Cell culture, transfection, and lysate preparation. DU145, PC-3, and 293 cells were obtained from the American Type Culture Collection. DU145 and 293 were maintained in recommended medium at 37°C, 95% relative humidity and 5% CO2. PC-3 was maintained in RPMI containing 10% fetal bovine serum and penicillin-streptomycin-glutamine (Invitrogen). HaCaT cells were kindly provided by Dr. Diane Bielenberg (Children's Hospital, Boston, MA). In overexpression studies, 293 or PC-3 cells were transfected using Fugene 6.0 (Roche Diagnostics). Whole-cell lysates were prepared in radioimmunoprecipitation assay buffer in the presence of Complete protease inhibitor cocktail (Roche Diagnostics). Protein concentration was measured by bicinchoninic acid assay (Pierce Biotechnology).
Cloning and plasmid construction. Human collagen XXIII was cloned from K562 leukemia Marathon cDNA using GC2 Advantage Polymerase (BD Biosciences Clontech) and primers 5'EndF2, CGGAGAGTCCACGCGCGACGAACG and HstopEcoR1, GGAATTCTCACTTATGCCAGCAGCCAGGCA. Human collagen XIII was cloned from HaCaT cell RNA isolated using an RNeasy kit (Qiagen). RNA was reverse transcribed using Superscript III (Invitrogen) and amplified using primers colXIII5'fwd CCCCGCAGTTTGGATAGAGCCTT and COLXIII3'rev GCATCACTTGTTCCAGCAGCCTTG. PCR products were inserted into pCR2.1-TOPO (Invitrogen) and subcloned into pcDNA3 (Invitrogen). Identity was confirmed by DNA sequencing.
For expression of cell-associated collagen XXIII, site-directed mutagenesis was done to prevent furin cleavage, and loss of protein from the cell. Collagen XXIII protein (accession AAO18361) was mutated at the furin cleavage site 105KIRTAR110, changing R110 to A110 to prevent furin-mediated shedding from the cell surface. Primers h23furmutFWD GATCCGGACTGCTGCGGAAGCTCCATCCG and H23furmutREV CGGATGGAGCTTCCGCAGCAGTCCGGATC were used to introduce this mutation into pcDNA3-human collagen XXIII plasmid using the QuikChange mutagenesis kit (Stratagene). Cloning of pcDNA3-rat collagen XXIIImyc has been previously described (9).
Western blot analysis of cell lysates. Cell lysates were separated using SDS-PAGE and transferred onto Immobilon-P polyvinylidene difluoride (Millipore) membrane. For peptide competition experiments, whole-cell lysate from PC-3 cells transfected with pcDNA3-human collagen XXIIIR110A was separated using a one-well gel for equal protein loading across the gel. Following transfer, the membrane was cut into strips, probed with individual antiserum preparations, and reassembled. Membranes were developed using Enhanced Chemiluminescence Plus chemiluminescence kit (Amersham), and stripped using Reblot (Chemicon) for additional antibody probing. Collagen XIII antibody was kindly provided by Drs. Anne Latvanlehto and Taina Pihlajaniemi (University of Oulu, Oulu, Finland). Mouse monoclonal actin antibody was obtained from Chemicon.
In situ hybridization. Antisense and sense collagen XXIII RNA probes were prepared using collagen XXIII cDNA inserted into pcDNA3 (Invitrogen) and synthesized using a digoxigenin RNA labeling kit (Roche). Collagen XXIII was amplified from AT6.1 cell cDNA using GF 5'-CCAGATCATAGTGGAGCCAGGGCC-3'and GR 5'-CCAACCTGTACAGATCCGTTTGGA-3' primers. Formalin-fixed, paraffin-embedded sections of human prostate carcinoma were dewaxed, rehydrated, and digested with proteinase K (50 µg/mL) in 100 mmol/L Tris, 50 mmol/L EDTA buffer (pH 8.0) for 8 min at 37°C. Hybridization was done in an automated instrument (Ventana Medical Systems) for 60 min at 42°C with 10 pmol/L digoxigenin-labeled riboprobe in 100 µL of hybridization buffer [50% deionized formamide, 4x SSC, 10% dextran sulfate, 1% SDS, and denatured herring sperm DNA (400 µg/mL)] per section under a liquid coverslip. The highest stringency of posthybridization washes was at 45°C for 15 min in 0.1x SSC. Bound digoxigenin-labeled probe was detected by antidigoxigenin alkaline phosphatase conjugate and visualized by nitroblue tetrazolium and 5-bromo-4-chloro-3-indolylphosphate color reaction. Sections were counterstained with nuclear fast red.
Tissue microarray patient population. Patients diagnosed with clinically localized and locally advanced prostate cancer had undergone radical prostatectomy and pelvic lymph node dissection in curative intent at the University of Ulm Hospital (Ulm, Germany) as standardized in that institution between the years 1986 and 2002, and were included in the outcome study with prior internal review board approval. We have recently described this cohort in detail (10). Serum PSA levels were taken every 6 months during follow-up in the first 2 years and at least yearly afterward. Assay types varied over the years of this study and did not conform to a specific protocol.
Immunohistochemical analysis and tissue microarray image analysis. Tissue sections were deparaffinized in xylene and rehydrated in graded alcohols. Sections were treated with 3% hydrogen peroxide (Sigma-Aldrich) in methanol for 15 min to block endogenous peroxidase, and antigen retrieval was done by microwave in Antigen Unmasking Solution (Vector Laboratories). Immunohistochemistry was done using an avidin-biotin-peroxidase complex method, with extensive TBS washes between steps. Slides were blocked using normal goat serum followed by the avidin/biotin blocking kit, before incubating with collagen XXIII 6.1B affinity-purified antibody overnight at 4°C. Slides were then incubated with biotinylated goat anti-rabbit IgG, followed by a preformed avidin-biotin complex. Specifically bound antibodies were visualized by using peroxidase substrate, 3,3'-diaminobenzidine tetrahydrochloride (reagents from Vector Laboratories). Sections were counterstained with Mayer's hematoxylin. Negative controls included rabbit IgG or preimmune serum. Tissue microarray slide images were captured and analyzed using the ACIS II system (Chromavision) system as previously described (6). Primary prostate cancer tissue microarrays within the retrospective analysis cohort were immunostained in the same experimental run to enable accurate slide-to-slide comparison of staining and analysis of staining cutoff. Given the heterogeneity of prostate tissue samples, study pathologists used a computer-based selection tool to highlight areas within each 0.6-mm core for analysis. Cores with only stroma or nondiagnostic areas were excluded from further analysis.
Statistical analysis. Collagen XXIII staining data are normalized (z-scored) within each tissue microarray before data combination for analysis (6). Analysis of levels of collagen XXIII across different tissue types was done using the two-sample Student's t test with Levene's test to assess equality of variances. The Cox proportional hazards regression model with right-censored data was used to evaluate univariate associations of collagen XXIII and other clinical and pathologic variables with time to prostate specific antigen (PSA)–specific disease recurrence for patients (n=76) with a minimum of 5-year follow-up (11). Evidence of nonproportionality was checked using a log-minus-log survival plot, which indicated a constant difference over time between groups, supporting the assumption of proportionality (12). PSA-specific disease recurrence was used as a marker of disease progression. PSA recurrence was defined as an increase in postoperative serum PSA level >0.4 ng/mL in two consecutive measurements with the earlier date defined as date of failure. Gleason scores were reduced into well-differentiated (grades 5 and 6), moderately differentiated (grade 7), and poorly differentiated (grades 8 and 10) categories. Tumor stage was grouped into organ (prostate) confined (T2) and extraprostatic (T3) disease, due to low frequencies of individual tumor stage subcategories. This patient cohort did not contain T1 or T4 tumor stages. To adjust for confounding among covariates as well as unequal lengths of follow-up, multivariate Cox regression analysis was applied using a forward stepwise selection procedure and entering all variables analyzed by univariate analysis as candidates. Hazard ratios and 95% confidence intervals (95% CI) were calculated for significant independent predictors of PSA failure with significance of the variables determined by the Wald test (12). The predictive value of collagen XXIII for time to PSA-specific disease survival was also examined using the Kaplan-Meier product-limit method with survival curves compared with the log-rank (Mantel-Cox) test (13). Data analysis was done using the SPSS software package (version 14.0, SPSS, Inc.). Two-tailed values of P < 0.05 were considered statistically significant for all analyses.
Processing of patient urine samples. Patient urine samples were collected under approved institutional review board protocols issued by Massachusetts General Hospital and Children's Hospital, Boston. Frozen samples were thawed on ice; diluted in 1 volume of 10 mmol/L ammonium bicarbonate buffer (pH 8) supplemented with protease inhibitors, 1 mmol/L phenylmethylsulfonyl fluoride, 5 mmol/L phenanthroline, and 5 mmol/L benzamidine (Sigma-Aldrich), and centrifuged at 3,000 x g to remove insoluble material. Samples were concentrated by centrifugation using Centricon Plus-20, 5,000 MWCO devices (Millipore) and aliquots were lyophilized. Protein concentration was determined using the Bio-Rad protein assay (Bio-Rad Laboratories). Additional aliquots were resuspended in Laemmli buffer for SDS-PAGE. Control protein consisted of cell lysate from PC-3 cells transfected with human collagen XXIIIR110A mixed with molecular weight marker. Urine blots were probed using goat polyclonal PSA antibody (C-19; Santa Cruz Biotechnology) and mouse monoclonal Tamm Horsfall glycoprotein antibody (Cedarlane Laboratories). Collagen XXIII levels in prostate cancer patient urine blots were determined using enhanced chemiluminescence image capture using a ProEXPRESS imaging system (Perkin-Elmer) for semiquantitative Western blot analysis. Collagen XXIII level was normalized to the same positive control sample run on all gels.
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Collagen XXIII protein expression is up-regulated in human prostate cancer and in prostate cancer metastases. To study the expression of collagen XXIII in human prostate cancer, immunohistochemical staining was done on prostate tissue microarrays containing benign prostate tissue, prostate cancer, and prostate cancer metastases, using affinity-purified collagen XXIII antibody. Normal prostate tissue was largely negative for collagen XXIII staining, whereas prostate adenocarcinomas showed strong collagen XXIII staining in the tumor epithelia (Fig. 1A ). Tissue staining was quantified using the Chromavision Automated Cellular Imaging System II. Individual tissue cores were measured in terms of collagen XXIII–positive stained area (brown chromagen area), intensity, blue area (nuclear hematoxylin stain), and percentage staining (percentage brown area/blue area). Collagen XXIII was expressed at significantly higher levels in primary prostate tumors than in benign prostate epithelia, as assessed by mean stained brown area, P = 0.004 (two-sample t test, equal variances not assumed), intensity (P < 0.001), or percentage staining (P = 0.002; Fig. 1B; Table 1 ).
|
|
To determine the cell types responsible for collagen XXIII synthesis, we did in situ hybridization. Although we observed collagen XXIII protein in prostate tumor tissue epithelia, we had shown previously that collagen XXIII can be cleaved from the cell surface (9). We therefore could not discount the possibility that in vivo, collagen XXIII synthesis may occur in other cell types and localize at prostate epithelia. Prostate carcinoma tissue sections were hybridized with a sense control or an antisense collagen XXIII riboprobe (Fig. 2A and B ). Collagen XXIII mRNA expression was observed in prostate tumor epithelial cells but not observed in adjacent stromal tissue.
|
Additional high-density prostate cancer tissue microarrays with clinical follow-up data were analyzed using immunohistochemical staining for collagen XXIII, encompassing over 800 prevalidated, pathologist-reviewed tissue cores. Disease recurrence after prostatectomy was defined as two consecutive increases of PSA level >0.4 ng/mL. Stained brown area, intensity, and percentage staining were normalized for array comparison, and mean and maximum values were calculated from multiple tissue cores of individual patients. Patients who had undergone preoperative hormone ablation therapy and those with <5 years follow-up were excluded from analyses, resulting in a 76-patient subset whose demographics are listed in Table 2 . Mean follow-up time for non-PSA recurrence patients was 6.5 years, with a maximum of 8.3 years.
|
|
1.2) was found to be a significant predictor of PSA recurrence (hazard ratio, 2.5; 95% CI, 1.3-4.8; Table 3).
Recurrence-free survival curves were determined using Kaplan-Meier analysis (Fig. 3
). Patients with high collagen XXIII (1.2) had a higher risk of PSA failure, compared with patients with low collagen XXIII (<1.2). Analysis of the area under the curves indicated a highly significant difference in time to PSA failure (log-rank test, P = 0.003). High collagen XXIII patients had a median time to PSA failure of 8.1 months (95% CI, 0-2.2 years) compared with low collagen XXIII patients who had a median time to failure of 5.0 years (95% CI, 2.0-8.0 years). The cumulative percentage of patients with PSA recurrence at 1, 3, and 5 years is indicated in Table 4
, for low and high collagen XXIII. At 5 years, high collagen XXIII patients had only a 20% PSA-specific survival rate, whereas low collagen XXIII patients had a 51% PSA-specific survival rate.
|
|
Collagen XXIII can be detected in prostate cancer patient urine. Collagen XXIII can be shed from the cell surface by furin protease activity (9), suggesting that collagen XXIII might be present in biological fluids and thus have potential as a noninvasive screening tool. The ability to predict likelihood of disease recurrence using a patient fluid biomarker may provide further justification for the treatment strategy, whether active surveillance (deferred treatment) protocols for patients at low risk or radical prostatectomy combined with adjuvant therapies for patients at high risk of recurrence.
To determine whether collagen XXIII could be detected in human prostate cancer patient urine, samples were collected after digital rectal examination, both before and after prostatectomy. Western blot analysis was done on 20 µg of protein from patient-matched samples. Figure 4A shows that collagen XXIII can be detected in the urine of prostate cancer patients. Following prostatectomy, matched patient samples were found to have dramatically lower urinary collagen XXIII levels, indicating that collagen XXIII urine protein originated from the prostate. As expected, reduced PSA levels were also observed. The kidney-produced Tamm-Horsfall glycoprotein was used as a control for urinary protein quantity. Analysis of matched samples from 13 patients revealed that the level of collagen XXIII was significantly decreased after prostatectomy, P < 0.05 (paired t test), as shown in Fig. 4B. Seven of 13 patients showed a decrease in collagen XXIII. Interestingly, 1 of 13 patients has had an early PSA-defined disease recurrence (<3 years), and that patient showed no decrease in collagen XXIII after prostatectomy. Further studies will be required to determine whether there is any association of urinary collagen XXIII level and disease recurrence.
|
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
A number of other promising biomarkers have been identified in recent years and some exhibit different patterns of expression during prostate cancer progression, compared with collagen XXIII. For example, E-cadherin decreases in prostate cancer (17), whereas -methylacyl-CoA racemase (18) increases in prostate cancer but decreases in more aggressive disease. Decreased nuclear p27 has also been shown to correlate with disease progression (19). A biomarker profile may be more powerful than any individual marker alone for insight into the underlying biology of the primary tumor, and, indeed, an enhancer of zeste homologue 2/E-cadherin combination increased the predictive power of either marker alone (14). Additional studies will be required to determine whether the power of collagen XXIII in disease prediction can be improved by combination with additional biomarkers.
The adoption of molecular markers into clinical practice is likely to be most beneficial in cases of organ-confined disease, when clinical criteria classify the tumor at low risk of progression. Although radical prostatectomy is often selected, a favorable biomarker profile in biopsy tissue of low collagen XXIII, negative enhancer of zeste homologue 2/E-cadherin, and high -methylacyl-CoA racemase may give further support to offer active surveillance (deferred treatment) protocols to more elderly patients or those with confounding disease states. Conversely, a poor molecular biomarker profile might suggest a more aggressive treatment approach.
A potential advantage of collagen XXIII over many of the reported biomarkers is that it is cleaved from the cell surface by the ubiquitous protease, furin. The shift in protein size observed between the full-length transmembrane control protein and collagen XXIII in urine is consistent with furin cleavage (9). Furin level and activity are up-regulated in a number of cancers (20–22). As furin activity is likely to play a role in the shedding of collagen XXIII into prostate cancer patient urine (and perhaps serum), it will be interesting to determine the expression of furin protease in prostate cancer and whether this influences the level of collagen XXIII found in patient fluids. Collagen XXIII is not incorporated into various epithelial basement membranes (23), which suggests that shedding may occur at the luminal cell surface, resulting in high levels in the prostatic fluid in prostate cancer. Despite a dramatic decrease in collagen XXIII in patient urine after prostatectomy, some protein remained, suggesting either that residual disease remained or that collagen XXIII is reaching urine from another tissue source. We have amplified collagen XXIII cDNA from human heart, retina, and brain (9); however, it is unclear whether these levels of RNA expression will be relevant at the protein level. Recent evidence indicates that expression in mouse tissues is very low and distribution is restricted (23). Although the use of collagen XXIII as a urine biomarker will require an extensive prospective analysis with large sample numbers, we believe our data provides preliminary evidence for its utility. Our data show, at least at the tissue level, that collagen XXIII is independent of the only current fluid biomarker, PSA, as a prognostic indicator. Larger-scale urine studies will be needed to show conclusively whether urinary levels of collagen XXIII have the same predictive accuracy as the tissue levels.
The functional roles of either the integral membrane or the cleaved form of collagen XXIII are currently unknown. The subfamily of transmembrane collagens, also called membrane-associated collagens with interrupted triple helices (24), also includes collagens XIII, XVII, and collagen XXV (25). A role in pathologic diseases has been shown for collagens XVII, XIII, and XXV. Collagen XVII is mutated in epidermolysis bullosa, is a serum antigen in a number of skin blistering conditions (reviewed in ref. 26), and is up-regulated in oral and squamous basal cell carcinomas (27, 28). Collagen XIII is a serum antigen in the autoimmune disease, Graves' ophthalmopathy (29), and was recently reported to be up-regulated in mesenchymal tumors and in the stroma of epithelial tumors (30), whereas collagen XXV is deposited in Alzheimer's plaques (31). Our data now indicate a role for collagen XXIII in prostate cancer.
We observed extremely high collagen XXIII expression levels in the majority of distant prostate cancer metastases. There are two major explanations for this phenomenon. First, collagen XXIII expression in positive tumor cells could be further up-regulated by factors specific to the secondary site. Second, collagen XXIII expression could result in a selective advantage for cells to metastasize (32). Tumor metastasis is a multistage event involving local invasion and remodeling of extracellular matrix, intravasation, extravasation, and establishment at the secondary site (33). Considering the association of collagen XXIII expression in the primary tumor with later disease progression (as measured by PSA failure), this hypothesis will merit further investigation.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
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.
Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).
J. Banyard and L. Bao share first authorship.
Current address for L. Bao: Astra Zeneca, Waltham, MA 02451.
Received 8/29/06; revised 1/12/07; accepted 2/12/07.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
-methylacyl CoA racemase expression in localized prostate cancer is associated with an increased rate of biochemical recurrence and cancer-specific death. Cancer Epidemiol Biomarkers Prev 2005;14:1424–32.-Methylacyl coenzyme A racemase as a tissue biomarker for prostate cancer. JAMA 2002;287:1662–70.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | Cell Growth & Differentiation |
