SEARCH

SEARCH BY CITATION

Keywords:

  • SOX5;
  • prostate cancer;
  • SMAL-PCR DNA fingerprinting;
  • LCM;
  • chromosome 12p12.1

Abstract

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Identification of genomic alterations associated with the progression of prostate cancer may facilitate the better understanding of the development of this highly variable disease. Matched normal, premalignant high-grade prostatic intraepithelial neoplasia and invasive prostate carcinoma cells were procured by laser capture microdissection (LCM) from human radical prostatectomy specimens. From these cells, comparative DNA fingerprints were generated by a modified PCR-based technique called scanning of microdissected archival lesion (SMAL)-PCR. Recurrent polymorphic fingerprint fragments were used in tagging altered chromosomal regions. Altered regions were found at cytobands 1p31.3, 1q44, 2p23.1, 3p26.3, 3q22.3, 4q22.3, 4q35.2, 5q23.2, 8q22.3, 8q24.13, 9q21.3, 9q22.32, 10q11.21, 11p13, 12p12.1, 13q12.1, 16q12.2 and 18q21.31. Candidate genes in the surrounding area that may possibly harbor mutations that change normal prostatic cells to progress into their tumor stages were proposed. Of these fragments, a 420 bp alteration, absent in all 26 normal samples screened, was observed in 2 tumors. This fragment was cloned, sequenced and localized to chromosome 12p12.1. Within this region, candidate gene sex determining region Y-box 5 (SOX5) was proposed. Further studies of SOX5 in cell lines, xenografts and human prostate specimens, at both the RNA and protein levels, found overexpression of the gene in tumors. This overexpression was then subsequently found by fluorescent in situ hybridization to be caused by amplification of the region. In conclusion, our results suggest LCM coupled with SMAL-PCR DNA fingerprinting is a useful method for the screening and identification of chromosomal regions and genes associated with cancer development. Further, overexpression of SOX5 is associated with prostate tumor progression and early development of distant metastasis. © 2008 Wiley-Liss, Inc.

Prostate cancer is the most common malignancy occurring among men in North America and is the second leading cause of all cancer deaths. It has been estimated that in 2008, ∼ 28,660 deaths would occur in the United States as a result of this disease.1 One factor contributing to the high incidence and mortality rate, despite extensive research, is our limited knowledge of the molecular mechanisms that drive the progression of a normal cell to its malignant phenotype.2, 3 This highlights the need for a better understanding of the molecular mechanism of prostate cancer progression, with the goal of developing novel targets for the diagnosis and treatment of the disease.

It is believed that prostate cancer, similar to other epithelial cancers, develops from normal epithelial (luminal) cells which progress through distinct histopathological stages, including the precursor stage prostatic intraepithelial neoplasia (PIN), invasive carcinoma of ranging Gleason grades and metastatic disease.2, 3 This progression is thought to be caused by the stepwise accumulation of mutations in genes critical to functions such as cell cycle control, cell growth, differentiation, apoptosis, cell adhesion and other functions.

In the past, molecular studies of the disease pose major challenges since prostatic lesions are usually small, often containing only very few cells embedded in a mass of heterogeneous tissue. Most previous studies had analyzed bulk cancer tissues that were contaminated by a large proportion of noncancerous cells including stromal, microvasculature, and inflammatory cells, as well as cells representing different histological stages; thus such strategies do not yield precise molecular profiles of the respective lesions. Fortunately, with the advent of novel technological platforms and sensitive molecular techniques in recent years, the isolation of pure cell populations and the analysis of a small number of cells have become feasible.

By combining laser capture microdissection (LCM)4, 5 and a modified PCR-based DNA fingerprinting technique, scanning of microdissected archival lesions (SMALs)-PCR6, 7 that has been optimized for the analysis of low amount of DNA from archival, formalin-fixed, paraffin-embedded samples; we successfully generated comparative genomic DNA fingerprint profiles from matched normal, high-grade PIN (hPIN) and invasive carcinoma cells from 34 human prostatectomy specimens. Our studies identified several frequently altered chromosomal regions. Mapping these regions within a 2 Mbp range was undertaken and candidate genes were identified. Here, we describe the identification of a region of alteration (chromosome 12p12.1) that is recurrently observed in prostate tumors. Within this region, a candidate gene sex determining region Y (SRY)-box5 (SOX5) was proposed. Further studies of SOX5 found altered expression of the gene in tumors, suggesting that SOX5 may play a role in prostate cancer progression. Subsequent studies by fluorescent in situ hybridization (FISH) found SOX5 overexpression to be a result of amplification of the region. It is worthy to note that using this combined LCM and SMAL-PCR approach; our group previously identified LIM domain only 2 (LMO2) on another frequently altered chromosomal region 11p13 and found it to play a pivotal role in the progression of prostate cancer.8

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Clinical samples for SMAL-PCR DNA fingerprinting studies

A total of 224 prostate samples obtained from 89 patients were selected from the Vancouver, Richmond or Kelowna General Hospital (Vancouver, Richmond, Kelowna, BC, Canada). Excluded were patients undergoing hormone ablation therapy. The study was approved by the appropriate local ethics committees. Of the 224 samples, 160 were archived, formalin-fixed paraffin-embedded and 64 were fresh samples which were frozen as previously described.9 Histological diagnoses were made and tumors graded according to the Gleason method by pathologists T. Bainbridge or D. Webber.

LCM and sample preparation

One section from each of the samples was stained with hematoxylin and eosin (H&E) and examined by pathologists Bainbridge or Webber to mark areas that contained morphologically distinct normal, hPIN and invasive carcinoma cells. Emphasis was placed that the normal cells were distant to the hPIN or cancer cells. Formalin-fixed, paraffin-embedded serial sections (5 μm) were deparaffinized in xylene, rehydrated, stained with 0.05% Toluidine Blue O (Sigma Aldrich Co., St Louis, MO) in 100 mM citrate buffer (pH 5.5) and dehydrated in 100% ethanol and xylene. Serial sections from fresh frozen samples were placed on HistoBond glass slides (Paul Marienfeld GmbH & Co. KG, Lauda-Koenigshofen, Germany), fixed in 70% ethanol prior to the staining and dehydration procedure. Microdissection was performed using the LCM PixCell II (MDS Analytical Technologies, Sunnyvale, CA) as previously described.9 Microdissected cells on each LCM cap were lysed in a total of 50 μL lysis buffer (10 mM Tris, pH 8.0, 1 mM EDTA, 1% Tween 20 and 0.1% Proteinase K) at 42°C overnight. Lysates were extracted with phenol-chloroform-isoamyl alcohol (25:24:1) in Phase Lock Gel™ Eppendorfs (light, 0.5 mL, Invitrogen Corp., Carlsbad, CA) and DNA precipitated by adding 3 volumes of 100 mM NaOAc (pH 4.6), 95% ethanol.

Determination of quantity and quality of DNA from microdissected cells

The quantity and quality of extracted DNA was evaluated by multiplex PCR as previously described.10 DNA quantity was determined by comparing the signal intensity, as determined by densitometric analysis with the EagleEye II Still Video System (Stratagene, La Jolla, CA), of DNA from microdissected cells to those obtained for the genomic DNA standards (10, 5, 2.5, 1.25, 0.6 or 0.3 ng/μL) and DNA quality was determined by comparing the differential intensity of the 2 PCR amplified fragments from the experimental DNA.

SMAL-PCR DNA fingerprinting

SMAL-PCR primers were selected on the basis of the number of fragments generated and their distribution on a 6% nondenaturing polyacrylamide gel.10 Oligonucleotide primers (5′-AATCGGGCTA-3′) and (5′-GAAACGGGTC-3′) were chosen for this study and used to generate SMAL-PCR fingerprints of DNA from 34 pairs of normal, hPIN and cancer samples as previously described.11 Human genomic DNA and a reaction with no DNA template were used as positive and negative controls, respectively. Gains or losses of signals were identified by pairwise comparison of SMAL-PCR fingerprints. Signal alterations that occurred in multiple tumors were chosen for further analysis.

Cloning, sequencing and mapping of polymorphic fragments

DNA of polymorphic fragments were excised from gels, cloned and sequenced as previously described.11 Multiple clones from each band were sequenced to confirm that the alteration was the same in different cases. Sequences were matched to the March 2006 version of human genome assembly sequence database using the BLAT UCSC Genome Bioinformatics (http://genome.ucsc.edu). Matched results were interpreted as confirmed chromosome locations for each clone. Genes within a 2 Mbp radius were considered as candidate genes involved in the genesis of prostate cancer.

Cell lines and xenograft model

Prostate carcinoma cell lines PC3, LNCaP and DU145 were obtained from the American Type Culture Collection (Rockville, MD).12 Immortalized, normal human prostate epithelial cell line HPr-1 was previously established in our laboratory.13 Immortalized, normal human prostate epithelial cell line NPTx was provided as a gift by Drs. Robert Bright and Susan Topalian of NCI, NIH, Bethesda, MD.14 AI CWR22 is an androgen-independent recurrent tumor derived from AD CWR22 following castration of the host.15 The CWR22R xenograft was kindly provided as a gift by Dr. Franky Chan at the Chinese University of Hong Kong, Hong Kong.

RNA isolation, cDNA synthesis and RT-PCR

Total RNA was isolated with TRIZOL Reagent (Invitrogen, Carlsbad, CA) according to manufacturer's instructions. Total RNA was used for cDNA synthesis and RT products were used for RT-PCR with the following primer sequences: SOX5 forward 5′-TGCCTGGTGGATGGCAAAAAGC-3′ and reverse 5′-TGCTAGACACGCTTGAGTGC-3′; Actin forward 5′-GTTGCTATCCAGGCTGTGCT-3′ and reverse 5′-AGCACTGTGTTGGCGTACAG-3′.

Quantitative real time PCR of normal and prostate tumor clinical samples

Ten fresh frozen tissue specimens were collected from 5 patients (paired normal and tumor) from the Kelowna General Hospital (Kelowna, BC, Canada) or the Queen Mary Hospital (Hong Kong). cDNA was subjected to quantitative real time PCR (qPCR) with a SYBR Green PCR Kit (Applied Biosystems, Foster City, CA). SOX5 primer sequences were the same as those used for RT-PCR. 18S was used as an internal control (18S forward 5′-CTCTTAGCTGAGTGTCCCGC-3′ and reverse 5′-CTGATCGTCTTCGAACCTCC-3′). Quantitation was performed using the ABI PRISM 7700 Sequence Detector (Applied Biosystems, Foster City, CA). Calculations were performed as previously described.8

Clinical samples for tissue microarray studies

A total of 152 formalin-fixed, paraffin-embedded prostate tissue specimens were selected from the Department of Pathology, Queen Mary Hospital, The University of Hong Kong. Included were 10 normal, 41 BPH, 16 hPIN and 85 prostate carcinoma (PC) specimens. Excluded were tissue samples from patients who had undergone prior hormone deprivation or radiation therapy. Histological diagnoses were reviewed by pathologist K.W. Chan. Based on their combined Gleason score (GS), the PC were divided into low (n = 22) and high (n = 63) grade subgroups, defined as cases with a combined GS < or ≥ 7, respectively. Archival dates of the specimens fell between 1995 and 2004. Clinical information of the patients is summarized in Table I.

Table I. Patient Clinical and Pathologic Features (Tissue Microarray Studies)
 No. of cases%Median (range)
  1. Pre-PSA, preoperative prostate-specific antigen.

Histological group152  
Normal106.6 
Benign prostatic hyerplasia4127 
High grade PIN1610.5 
Prostate cancer (combined GS < 7)2214.5 
Prostate cancer (combined GS ≥ 7)6341.4 
Age (years)  76 (35–94)
Pre-PSA level (ng/mL)64 52.5 (0–2392)
Low (PSA < 10 ng/mL)914.1 
High (PSA ≥ 10 ng/mL)5585.9 
Site of metastasis detected46  
Bone4189.1 
Lymph node36.5 
Other24.3 

Tissue microarray development and immunohistochemistry

All 152 prostate specimens were used to construct a tissue microarray (TMA), as previously described.8 Each specimen was sampled in triplicates to account for tumor heterogeneity (n = 456). TMA sections were pretreated in 0.01M sodium citrate buffer (pH 6) for 15 min in a microwave oven. Endogenous peroxidase was blocked with 0.6% hydrogen peroxidase, followed by incubation with PBS containing 10% normal blocking protein (DAKO, Carpinteria, CA). Specimens were incubated with polyclonal rabbit anti-human SOX5 (Abcam Inc., Cambridge, MA). A standard anti-rabbit secondary antibody (DAKO Envision Plus System-HRP, DAKO, Carpinteria, CA) was applied. The reaction was then developed by Liquid DAB+ Substrate-Chromogen System (DAKO, Carpinteria, CA). All sections were counterstained with hematoxylin.

Evaluation of immunohistochemical staining

Stained sections were reviewed by Chan KW and Chan YP, who had no knowledge of the patient data. All available scores of each sample were examined and the score with the highest staining intensity was selected for further quantification. Foci were assessed for the percentage of cells staining positive both in the nucleus or cytoplasm; and an overall staining intensity score (negative, weak, moderate, strong and very strong) was given.

Statistical analysis

Statistical analyses were done using SPSS v11.5 software. Differences in SOX5 expression among different clinico-pathological stages were analyzed by Chi-square test (χ2 test). Kaplan-Meier analysis was used to estimate the median time required for development of distant metastasis. Log Rank and Breslow tests were used to compare survival functions of patients with different SOX5 expression levels. Paired sample t-test was used to compare SOX5 expression between nontumor and tumor groups in the qPCR studies. Statistical significance was declared if p < 0.05.

Fluorescent in situ hybridization

Amplification of SOX5 was examined by FISH using the original full formalin-fixed, paraffin-embedded tissue blocks from 12 patients, selected from the TMA mentioned above (2 normal/BPH, 2 PC with low SOX5 expression and 8 PC with high SOX5 expression). Dual-color FISH was undertaken using 4 BAC clones (RP11-444N1, RP11-437F6, RP11-653A22 and RP11-229F13) labeled in Spectrum Orange (Vysis, Downer's Grove, IL), covering the SOX5 gene at 12p12.1; and a reference BAC clone localized to the centromere of chromosome 12, labeled in Spectrum Green (Vysis). BACs were obtained from the BACPAC Resource Center (BPRC) at the Children's Hospital Oakland Research Institute (Oakland, CA). FISH reactions were performed as described previously with minor modifications.16, 17 Briefly, deparaffinized sections were treated with proteinase K (400 μg/mL) between 37 and 55°C for 2 to 4 hr, followed by denaturation in 70% formamide, 2× SSC at 75°C for 8 min. Fifty nanograms of each probe were mixed in a 20 μL hybridization mixture (containing 55% formamide, 2× SSC and 2 μg human Cot1 DNA), denatured at 75°C for 6 min and then hybridized to the denatured section at 37°C overnight. After hybridization, sections were washed 3 times in 50% formamide and 2× SSC at 45°C for 3 min each. Stained sections were counterstained with 4′-6-diamidino-2-phenylindole in an antifade solution and were examined under a Zeiss Axiophot microscope equipped with a triple-band pass filter. Amplification was defined as the presence of 2 times as many gene signals when compared with the reference probe in the same chromosome.

Results

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Assessment of prostate tissue samples for SMAL-PCR DNA fingerprinting studies

A total of 224 prostate samples originating from 89 patients were collected for SMAL-PCR DNA fingerprinting studies. Since we wished to compare DNA fingerprints from matched normal, hPIN and/or invasive PC cells, 31 samples were not included in this study because they contained so few or no hPIN or carcinoma cells. It is known that the quality of tissue samples varies with the collection, preservation and storage conditions and while the frozen samples were handled according to a standardized protocol9 to guarantee proper quality for DNA, RNA and proteins, a standardized protocol for the formalin-fixed, paraffin-embedded samples does not exist. Therefore, to judge if the extracted DNA would yield useful DNA fingerprints, a stringent multiplex PCR test was used to evaluate the quantity and quality of DNA extracted from LCM procured cells, prior to subjecting the samples for downstream analyses. DNA from fresh samples yielded ∼ 60 ng DNA/5,000 cells, amplified well in our PCR process, and consistently produced 50 or more SMAL DNA fingerprint bands in size ranging from 100 to 1,000 bp. Based on the quality of DNA extracted from frozen specimens, we set the same criteria for selection of DNA samples extracted from formalin fixed tissue specimens; to ensure that similar fingerprint profile were readily produced from both frozen and formalin fixed samples. Only 1/3 of the DNA samples from formalin fixed tissue samples exhibited the same characteristics than those of frozen samples, whereas the other 2/3 of these DNA samples either yielded very low or no amounts of DNA or did not support amplification and would not yield useful fingerprints (data not shown). These DNA samples were also rejected for further analysis. Therefore, of the total samples collected for the study, only 84 DNA samples originating from 34 patients yielded useful DNA fingerprints for analysis.

SMAL-PCR DNA fingerprints of matched microdissected normal, hPIN and prostate cancer cells reveal multiple chromosomal mutational sites

To identify chromosomal sites that harbor putative genetic alterations that are relevant to the molecular events in the formation of prostate cancer, a comparative analysis was undertaken of fingerprints that were generated from DNA extracted from matched normal, hPIN and carcinoma cells, using a modified PCR-based DNA fingerprint technique which we named SMAL-PCR. These cells were obtained in pure form from sections by LCM to avoid contamination with other cell types such as stromal cells, inflammatory cells, etc. An example of LCM microdissection is shown in Figure 1. Figure 1a shows an H&E stained reference slide with an area of hPIN cells which, as shown in Figure 1d, have been captured onto a LCM cap from a Toluidine Blue O stained serial section; images of this section before and after microdissection are shown for comparison reasons in Figures 1b and 1c.

thumbnail image

Figure 1. H&E stained reference section (a) was used as a guide for LCM microdissection (bd) of high grade PIN from Toluidine Blue O stained sections of formalin-fixed paraffin-embedded prostate archival biopsy. (e, f) Recurrent polymorphisms in DNA fingerprints of matched prostate normal (N), hPIN (P) and carcinoma (C) cells from 5 of 34 prostate cancer patients screened. (e) Red arrows indicate polymorphic fragments at 203 bp in hPIN of Patient 11, at 156 bp (a double band) in hPIN and at 320 bp (see also f) as well as at 495 bp in cancer cells of Patient 6 and at 184 bp and 635 bp in hPIN of Patient 7. (f) Example of a 320 bp recurrent polymorphism observed in prostate carcinoma cells from Patients 4, 6 and 16. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Examples of DNA fingerprints obtained from matched prostate normal (N), hPIN (P) and carcinoma (C) cells microdissected from formalin-fixed paraffin-embedded tissue sections are shown in Figures 1e and 1f. Fingerprints were performed in triplicates to ensure reproducibility. Red arrows indicate polymorphic fragments which are defined as additional or missing DNA fragments when compared with the fingerprints of their matched normal cells, such as the additional 203 bp polymorphic fragment found in hPIN of Patient 11, the hPIN fragment of 156 bp as well as 2 fragments of 320 and 495 bp in cancer cells of Patient 6; or the 2 hPIN fragments of 184 and 635 bp in Patient 7 (Fig. 1e). All the polymorphic fragments shown in these patients recurred in other patients. DNA from these recurrent bands were subsequently cloned, sequenced and used as tags to identify chromosomal regions presumed to harbor mutations likely to play a role in the formation of prostate cancer. For example, a 320 bp polymorphism recurred in prostate cancer Patients 4, 6 and 16 (red arrows Fig. 1f) and was localized to chromosome 3p26.3.

The findings of our fingerprint studies are summarized in Table II where data of identified recurring alterations are sorted based on the chromosomal locations (cytoband) of the cloned fragments. Listed in the table are the patient involved, frequency of the alteration detected, size of each of the recurrent polymorphic fingerprint bands, the chromosomal locations (cytoband) of the cloned fragments and whether hPIN and/or PC cells contained (+) these tagged chromosomal regions; NP indicates that hPIN or cancer cells were not present in the tissue specimens and was thus unavailable for comparative analysis.

Table II. Chromosomal Alterations Detected by SMAL-PCR Fingerprinting in hPIN and Prostate Carcinomas
  1. NP, hPIN or cancer cell not present in tissue specimens; hPIN, high grade prostatic intraepithelial neoplasia; PC, prostate carcinoma.

inline image

Some sites, including 1p31.1, 1q44 and 9q22.32, were consistently found in both matched hPIN and carcinoma foci in all the patients carrying the polymorphism (highlighted in light gray in Table II). In addition to sites that were detected in both matched hPIN and carcinoma foci of a patient, some chromosomal sites were consistently detected only in the hPIN foci but not in their corresponding invasive lesions (2p23.1, 8q22.3 and 18q21.31), suggesting that these regions may likely harbor putative genes that play important roles in the development and early progression of prostate cancer to more invasive stages. Other polymorphisms were consistently detected in carcinomas only but not in their corresponding normal epithelium or premalignant hPIN (3p26.3, 4q35.2, 8q24.13, 11p13 and 12p12.1), suggesting that these altered regions may likely harbor putative genes that play important roles in prostate cancer progression especially in the later stages of tumor invasion.

In summary, a total of 18 chromosomal sites presumed to contain genetic alterations were identified. These regions mapped to chromosomes 1p31.3, 1q44, 2p23.1, 3p26.3, 3q22.3, 4q22.3, 4q35.2, 5q23.2, 8q22.3, 8q24.13, 9q21.3, 9q22.32, 10q11.21, 11p13, 12p12.1, 13q12.1, 16q12.2 and 18q21.31. Genes that may be considered prostate cancer genes and map within a 2 Mbp range were reviewed.

Identification of candidate gene SOX5 in a frequently altered chromosome region 12p12.1

Of the 18 fragments identified by SMAL-PCR DNA fingerprinting, a 420 bp alteration, absent in all 26 normal samples screened, was observed in 2 of these tumors. However, it should be stressed that these numbers do not reflect the frequency of amplification because SMAL-PCR is dependent on the presence of primer sites that are affected by polymorphisms among individuals. The alteration appeared as additional bands in the tumor fingerprints that were absent in the patient-matched normal DNA fingerprints. This fragment was cloned from both cases and shown to be identical in sequence. BLASTN analysis of these sequences against the GenBank databases localized it to human bacterial artificial chromosome (BAC) RP11-16A24 (AC024225), which maps to 12p12.1 by sequence alignment against the March 2006 release of the human genome assembly sequence database. Within this region, several candidate genes [i.e., KRAS, RASSF8, LRMP and BHLHB3] (Fig. 2) were proposed, of which includes SRY-box5 (SOX5).18–29 Based on the argument that chromosome 12p12.1 is a frequently altered chromosomal site in prostate cancer and the fact that SOX5 has recently been shown to play a role in several cancer types, including those of testicular, skin, brain, lymph node and nasopharyngeal19–23; the authors decided that further work for this study would be focused on the SOX5 gene. However, it should be stressed that studies on other candidate genes in other identified frequently altered chromosomal regions are also currently underway in our laboratory.

thumbnail image

Figure 2. List of known genes found on chromosome 12p12.1.

Download figure to PowerPoint

SOX5 is preferentially expressed in the more advanced, androgen-independent prostate cancer cell lines and xenografts

We began by examining the expression of SOX5 in a panel of prostate cell lines and xenografts by semiquantitative RT-PCR analysis. As a model for nonaggressive prostate cancer cells (androgen-dependent), we chose LNCaP cells and AD CWR22 xenografts, and as a model for highly aggressive and metastatic cells (androgen-independent), we chose PC3 and DU145 cells, which were originally isolated from bone and brain metastasis, respectively, and AI CWR22 xenografts. Prostate cell lines HPr-1 and NPTx were also incorporated as immortalized, normal controls.

Positive SOX5 expression was observed in DU145, PC3 and AI CWR22 xenografts, contrasting with the weaker or absent signal observed in the less invasive AD CWR22 xenografts and LNCaP or immortalized, normal HPr-1 and NPTx. Note that although PC3 only displayed a weak SOX5 band, expression was still evidently higher than the immortalized, normal HPr-1, NPTx or the androgen-dependent LNCaP cells. Taken together, this suggest that SOX5 maybe up-regulated, and potentially required, during prostate cancer progression from a normal, androgen-dependent to an androgen-independent state of the disease (Fig. 3).

thumbnail image

Figure 3. RT-PCR analysis of SOX5 cDNA levels reverse-transcribed from RNA extracted from immortalized, normal prostate epithelial cell lines NPTx, HPr-1; prostate cancer cell lines LNCaP, DU145, PC3; and androgen-dependent (AD) and androgen-independent (AI) cells derived from CWR22 prostate xenografts. Actin was amplified as an internal control.

Download figure to PowerPoint

SOX5 is preferentially overexpressed in human prostate cancer specimens, as detected by both immunohistochemistry staining and qPCR analyses

Immunohistochemistry (IHC) for SOX5 was performed on a human prostate TMA constructed from 10 normal, 41 BPH, 16 hPIN specimens and 85 invasive PCs. The PC group was further classified into low and high grades, based on their combined GSs (Table I).

SOX5 expression was detected in all histological groups although expression was significantly stronger in PC specimens. For this part of the study, prostate specimens were classified into “nontumor” and “tumor” subgroups, with the former group comprising of normal, BPH and hPIN specimens and the latter group comprising of invasive PC samples, respectively. The χ2 test revealed a significant correlation in SOX5 expression between the nontumor and tumor subgroups (p = 0.002); where evidently more samples within the tumor subgroup displayed higher SOX5 expression levels (47 of 85; 55.3%) than the nontumor group (19 of 67; 28.4%) (Figs. 4a and 4b).

thumbnail image

Figure 4. (a) Immunohistochemical staining of SOX5 protein in the human prostate tissue microarray where representative nontumor (normal, BPH or hPIN specimens) (a, b) and tumor (invasive PC specimens) (c, d) cells are shown. Panels a and b show nontumor prostate cells with low overall and higher nuclear SOX5 expression, respectively; where Panels c and d show prostate tumor cells with high overall and even more intense nuclear SOX5 expression, respectively. The pictures are taken at ×400 magnification. (B) Bar-chart summarizing the overall immunohistochemical staining of SOX5 proteins in nontumor vs. tumor specimens. Numbers represent percentages of cases. (c) Kaplan-Meier analysis showing the relationship between SOX5 expression and the time to development of distant metastasis in patients with prostate cancer of high Gleason score (GS ≥ 7; n = 63). Patients with high SOX5 expression (dashed line) develop metastasis in a significantly shorter time (median = 3 months) than those showing low SOX5 expression (solid line; median = 37 months). (d) Relative SOX5 mRNA expression levels were determined by qPCR from RNA preparations obtained from 5 paired normal and prostate carcinoma specimens. The level of SOX5 in each prostate tumor specimen was higher than that in its corresponding normal prostate tissue (p = 0.047). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Within the invasive PC group, there was also a significant correlation in SOX5 expression and the time leading to distant metastasis. For this part of the study, patients showing negative to moderate SOX5 staining intensity levels were classified as “low” SOX5 expression whereas patients with strong to very strong levels of SOX5 staining intensity levels were classified as “high” SOX5 expression. The Kaplan-Meier and Breslow tests found that among prostate cancer patients with high GS (≥7), the time leading to the development of distant metastasis is significantly different between patients with low and high SOX5 expression (p = 0.016, Log Rank test; p = 0.005, Breslow test). Since the Breslow test emphasizes on events at earlier time points than when compared with the Log Rank test, the lower p value indicates that such difference is more significant, specifically in the earlier time period. Patients with high SOX5 expression tend to develop distant metastasis within a shorter period (estimated median = 3 months), when compared with patients with low SOX5 expression (estimated median = 37 months). For the PC patients with low GS, no evident correlation could be detected between the 2 SOX5 expression levels (p = 0.89, Log Rank test) (Table III, Fig. 4c). For this part of the study, only 46 of 85 PC specimens had follow up metastatic information available for analysis (Table III). SOX5 expression levels were not found to correlate with PSA levels nor mean age of the patients involved in the study (data not shown).

Table III. Breakdown of SOX5 Expression in the PC Stages of the TMA Studies and the Number of Corresponding Metastatic Events for the Kaplan-Meier Analysis
Gleason gradeSOX5 expressionNo. of casesEvents (metastasis)Censored cases
No. of cases%
  1. GS, Gleason score.

Low (combined GS < 7)Low72571.4
High1551066.7
High (combined GS ≥ 7)Low31151651.6
High3224825
Total 854639 

SOX5 mRNA expression level in 5 paired frozen normal and tumor specimens were also compared by qPCR. SOX5 mRNA was found to be significantly upregulated in prostate tumor compared with the normal specimen (p = 0.047, t-test) (Fig. 4d). Taken together, the results suggest that SOX5 may play a positive role in the neoplastic transformation and metastatic development of prostate cancer.

SOX5 overexpression is a result of gene amplification at 12p12.1

Past studies in the form of alteration at chromosome 12p12.1 has been controversial in prostate cancer. Both amplification and deletion in the region has previously been suggested.30–33 To further delineate the specific form of chromosomal alteration pinpointed to by SMAL-PCR DNA fingerprinting analysis at 12p12.1, leading to overexpression of SOX5, FISH analysis was carried out on whole-mount formalin-fixed paraffin-embedded sections from 12 patient cases showing either no, low or high SOX5 expression, as detected by IHC analyses. SOX5 amplification was not detected in the normal or BPH tissue specimens while amplification was evident in 1 of 2 prostate tumor cases showing low SOX5 expression; and 5 of 8 cases showing high SOX5 expression (Fig. 5). Of the other 3 cases in which amplification was not detected, 2 cases were not informative as they did not display proper FISH signal.

thumbnail image

Figure 5. Detection of amplification of SOX5 by dual-color FISH on formalin-fixed paraffin-embedded prostate clinical tissue specimens. The ideogram of chromosome 12 (top) shows the localization of SOX5 and BAC clones overlapping the gene on 12p12.1. Bottom, examples of nuclei hybridized with the 4 BAC clones to SOX5 (red) and the control reference probe to the centromere of chromosome 12 (green). Examples of amplification of SOX5 with increased copy number of 12p12.1 signals compared with reference signals from 3 representative patients are shown. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

A major challenge to human prostate cancer research has been the characterization of the molecular events associated with prostate cancer progression. Progress in achieving this goal has been hindered by the small size of the prostatic lesions and the heterogeneous cellular environment in which lesions are usually embedded. Prostate tumors are histologically diverse, containing a large proportion of stromal cells, and expression profiles from bulk tissues could be substantially affected by contamination with surrounding noncancerous cells and by the proportions of various cell types. For example, genes that are highly expressed in stromal cells, but not in epithelial cells, may be selected as downregulated elements in microarray analyses of bulk tissues, because stromal components are less abundant in cancer tissues than in normal prostate. In addition, molecular studies in the past have been greatly hampered by the poor quality and the limited availability of biological material most of which is reserved for diagnostic purposes. Even when pathologic tissue specimens are available, formalin, the most widely used fixative in histopathology practice, induces extensive cross-linking between nucleic acids and proteins thus making the DNA molecules rigid and susceptible to mechanical shearing, leading to degradation of the DNA, which in turn severely limits the use of these precious samples for molecular studies.34, 35 In accordance with past studies, indeed only a fraction of the archived tissue samples collected for our study here was useful for subsequent PCR DNA fingerprint analysis. As a result of these constraints, to date, only a limited number of studies have examined microdissected, matched pure cell population representing normal, hPIN and different stages of carcinoma. And of those that have been published, investigators have employed different techniques like comparative genomic hybridization,36 loss of heterozygosity,37 cDNA microarray,38–40 2D-PAGE41 or SELDI profiling analyses9, 42; all of which required great amount of starting DNA, RNA or protein material, of which is not always available due to the small cell number of the prostatic lesions.

We report here the first study to investigate matched pure cell populations representing normal epithelium, hPIN and invasive PC lesions using LCM technology and a modified genome-wide screening method of SMAL-PCR DNA fingerprinting. LCM allows for the rapid procurement of specific cell populations from the heterogeneous environment of the tissue, making it possible to obtain precise molecular profiles of the respective lesions.4, 5 SMAL-PCR DNA fingerprinting, which is essentially a modified technique of the RAPD-PCR (randomly amplified polymorphic DNA-PCR) approach,6, 7 allows for an unbiased, simultaneous genome-wide scanning of multiple loci, optimized for minute quantities of starting DNA. Using 1 primer pair alone, the technique is able to reveal ∼ 50 DNA fingerprint fragments per reaction with only 2 ng of DNA. With this platform, we were able to identify polymorphic fingerprint fragments that were used as tags to mark chromosomal regions. Some fragments were detected only in 1 particular patient whereas others recurred in multiple patients. Although we cannot rule out that sites that were only detected once do contain alterations implicated in prostate cancer progression, a lot of work will have to be done to ensure that they are not simply a PCR artifact or each represents just 1 single mutational event in that particular individual. Thus in view of this, in this study, we chose to concentrate on the recurrent changes that are more likely to represent mutational sites playing a role in prostate cancer formation. A number of chromosomal sites detected in our study were consistent with chromosomal instability regions that have previously been implicated to be frequently altered in prostate cancer, such as chromosome regions 4q22.3, 8q22.3, 8q24.13, 12p12.1 and 18q21.31.43 We attribute the observation that only a partial list of the alterations we found was consistent to those previously published to the fact that the SMAL technique is only able to reveal a fraction of the mutations when just 1 set of primer pair is used, as in this study. When increasingly more primer pairs are used to scan different loci of the same samples in the future, it is likely that regions of chromosomal instability will begin to cluster at specific regions and an increased number of altered chromosomal regions will also be revealed. In addition, we also identified new chromosomal mutational sites (1p31.3, 1q44, 2p23.1, 3p26.3, 3q22.3, 4q35.2, 5q23.2, 9q21.3, 9q22.32, 10q11.21, 11p13, 13q12.1 and 16q12.2) that have never been reported to be associated with this disease before.

Prostate cancer progression from normal epithelial (luminal) cell through low and high grade PIN and to invasive carcinoma is thought to be caused by stepwise alterations in the genome. The number of aberrations typically is small in premalignant lesions and substantially greater in more advanced lesions, supporting a role for acquisition of chromosomal aberrations in tumor progression. If this theory holds true, we should observe in our results that alterations found in PIN formed a subset of those found in carcinoma. Although some of the patients screened in our study do show this accumulation pattern, as in Patients 3 and 8, others do not fit this hypothesis. We attribute this variability to clonal heterogeneity/multifocality. Clinically resected prostate cancers often have multiple PIN and tumor lesions. Conceptually, disease multiclonality/multifocality are thought to occur in several ways involving (i) clonal expansion of a tumor from a single progeny cell followed by subsequent lateral spread; (ii) a field cancerization effect that primes the tissue to cancer at multiple sites simultaneously; or (iii) simply random independent genetic events. Therefore, if the PIN and tumor samples were to be microdissected from different areas of 1 individual prostate specimen, it is possible that clonal heterogeneity and multifocality may result in alterations found in PIN and not in the carcinoma.

A common criticism of SMAL-PCR fingerprinting is its inability to distinguish the type of alteration that produced the clone. In a SMAL fingerprint, alterations are either classified as a gain or loss of band in tumor DNA vs. the normal reference DNA. However, the mechanism underlying the change is unknown. For example, a gain can be derived due to a chromosomal amplification where a specific locus increases in copy number. But on the other hand, a gain can also be derived by a chromosomal deletion where the deletion would bring 2 SMAL sequences close enough together for PCR amplification. Similarly, a loss can not only be characterized as a deletion of a chromosomal region eliminating primer-binding sites, but can also be derived when an insertion separates 2 primer binding sites far enough to prevent PCR amplification. In addition, other genes outside the deletion may be the real cause as their expression may be regulated by mechanisms like hyper/hypomethylation. This disadvantage of SMAL-PCR DNA fingerprinting, however, can be solved by follow-up studies like delineation of the form of alteration by techniques like FISH; and expression analysis of genes in the region of the clone.

Subsequent analyses of candidate chromosomal regions identified by SMAL-PCR DNA fingerprinting found candidate gene SOX5 in the recurrently altered chromosome region 12p12.1. SOX5 is a member of the SOX [Sry-related high-mobility group (HMG) box] family of HMG DNA-binding domain transcription factors and has long been recognized to play a key role in the regulation of embryonic development and in the determination of cell fate.44–50 Early studies have found SOX5 to play a role in mouse chondrogenic differentiation whereas recent studies have found SOX5 to be involved in testicular seminomas, melanomas, nasopharyngeal carcinomas, gliomas and lymphomas.19–23 Specially, SOX5 maps to a region of amplification on the short arm of chromosome 12 in testicular seminomas while it is also found to be preferentially expressed and responsive in glioma patients. However, to date, the gene has not been suggested to be involved in any other cancer types. This is the first report to establish that overexpression of SOX5 is associated with prostate cancer progression. SOX5 mRNA expression was found to be overexpressed in the more aggressive androgen-independent, than when compared with the androgen-dependent or immortalized normal cell lines and xenografts. SOX5 expression also correlated positively with prostate tumor progression and the earlier development of distant metastasis, as shown in our human prostate cancer specimens. Succeeding studies by FISH found SOX5 overexpression to be a result of amplification of the region. Further understanding of the mechanistic pathway by which SOX5 regulates prostate cancer progression will be the subject of our further investigation. In addition, more studies on other candidate genes in other identified frequently altered chromosomal regions may also aid in the identification of novel molecular targets for the prevention and treatment of prostate cancer, as well as for the better understanding of the molecular mechanisms involved in prostate carcinogenesis.

Experiments in our current study have been largely dependent on 2 sets of clinical specimens, collected from Canada and Hong Kong. It should be stressed that the 2 sets of clinical samples collected were not used for direct comparison between one another. In contrast, samples collected in Canada were used in the SMAL-PCR DNA fingerprinting studies whereas the samples collected in Hong Kong were used to examine, by IHC, SOX5 expression found using the DNA fingerprinting method. In this regard, the second set of clinical specimens collected in Hong Kong can be viewed as a verification of data collected from the first set of clinical specimens collected in Canada. The fact that alteration of SOX5 on chromosome 12p12.1 can be found across the 2 sets of samples collected in different places around the world, representing different race, ethnicities, socio-economic status, etc., lends further support to the importance of this gene in prostate cancer on a global perspective.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The authors thank Ms. Pat Allard, Dr. Samina Noorali and Ms. Mihoko Whalen for their expert technical assistance.

References

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References