Skip Navigation


Carcinogenesis Advance Access originally published online on October 17, 2007
Carcinogenesis 2008 29(1):120-128; doi:10.1093/carcin/bgm226
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
29/1/120    most recent
bgm226v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Wang, X.
Right arrow Articles by Klein, R. D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Wang, X.
Right arrow Articles by Klein, R. D.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2007. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

The resistance to the tumor suppressive effects of COX inhibitors and COX-2 gene disruption in TRAMP mice is associated with the loss of COX expression in prostate tissue

Xingya Wang, Jennifer K.L. Colby1, Peiying Yang2, Susan M. Fischer1,*, Robert A. Newman2 and Russell D. Klein{dagger}

Department of Human Nutrition and Molecular Carcinogenesis and Chemoprevention Program, The Ohio State University Comprehensive Cancer Center, The Ohio State University, 325 Campbell Hall, 1787 Neil Avenue, Columbus, OH 43210, USA
1 Science Park—Research Division, The University of Texas M.D. Anderson Cancer Center, PO Box 389, Park Road 1C, Smithville, TX 78957, USA
2 Department of Experimental Therapeutics, The University of Texas M.D. Anderson Cancer Center, 8000 El Rio, Houston, TX 77054, USA

* To whom correspondence should be addressed. Tel: +1 512 237 9482; Fax: +1 512 237 9566;Email: smfischer{at}mdanderson.org


   Abstract
Top
 Abstract
Introduction
 Materials and methods
 Results
 Discussion
 Funding
 References
 
Over-expression of cyclooxygenase-2 (COX-2) and prostaglandin E2 has been demonstrated to play a significant role in the tumorigenesis of colon, lung, breast, bladder and skin. However, inconsistent and controversial reports on the expression and activity of COX-2 in prostate cancer raised the question of whether COX-2 plays a pivotal role in prostate carcinogenesis. To address this question, we examined the effects of COX-2 inhibition on prostate tumorigenesis in the transgenic adenocarcinoma mouse prostate (TRAMP) model. Three-week-old TRAMP mice were fed control, celecoxib- or indomethacin-supplemented diets for 27 weeks. A TRAMP/COX-2 knockout mouse model was also generated to determine the effects of the loss of the COX-2 gene on prostate tumorigenesis in TRAMP mice. These studies demonstrated that neither non-steroidal anti-inflammatory drugs (NSAIDs) nor genetic disruption of COX-2 was inhibitory in terms of tumor and metastases incidence, lobe weight or types of pathological lesions. A careful analysis of wild-type and TRAMP tissues was undertaken for the expression of cyclooxygenase-1 (COX-1) and COX-2 using immunoblotting, quantitative real time polymerase chain reaction (qRT-PCR) and immunohistochemistry approaches in TRAMP dorsal prostate tissue from 10- and 16-week-old, as well as tumor from 30-week-old mice. We found that the expression of COX-1 and COX-2 dramatically decreased during TRAMP carcinogenesis. Using the probasin promoter, a COX-2 over-expressing mouse model was also generated but failed to show any pathology in any of the prostate lobes. Collectively, our results suggest that COX-2 may not play a tumorigenic role during prostate carcinogenesis in the TRAMP model.

Abbreviations: AP, anterior prostate; COX-1, cyclooxygenase-1; COX-2, cyclooxygenase-2; DP, dorsal prostate; KO, knockout; LN, lymph node; LP, lateral prostate; mRNA, messenger RNA; NSAID, non-steroidal anti-inflammatory drug; PC, prostate cancer; PD, poorly differentiated carcinoma; PGE2, prostaglandin E2; PIN, prostatic intra-epithelial neoplasia; TRAMP, transgenic adenocarcinoma mouse prostate; VP, ventral prostate; WD, well-differentiated carcinoma; WT, wild-type


   Introduction
Top
 Abstract
 Introduction
Materials and methods
 Results
 Discussion
 Funding
 References
 
The rate-limiting step for the synthesis of prostaglandins is the metabolism of arachidonic acid by cyclooxygenase-1 and cyclooxygenase-2 (COX-1 and COX-2). COX-1 is generally considered to be constitutively expressed in nearly all tissues and has functions in maintaining tissue homeostasis, whereas COX-2 is inducible by a variety of factors (1). Over-expression of COX-2 in several cancers, including colon, breast, lung, bladder, skin and pancreas, has been reported by numerous studies (1). Prostaglandins have been reported to induce cell proliferation, enhance angiogenesis and promote invasion and metastasis of these cancers (2,3). Evidence from epidemiological, clinical and preclinical animal studies, including genetically modified mouse models of over-expression or disruption of the COX-2 gene, demonstrates that COX-2 plays a significant role in the development of above cancers (1,3,4). Non-steroidal anti-inflammatory drugs (NSAIDs) that inhibit COX activity have received much attention as potential chemopreventive agents for these cancers.

Studies on the role of COX-2 in prostate carcinogenesis, however, are inconsistent and controversial. Retrospective case–control and prospective cohort epidemiologic analyses examining the risk of prostate cancer (PC) relative to NSAID use have resulted in mixed findings. Whereas results from some studies indicate no protective effect of NSAIDs (5–9), there are several reports of a significant reduction in PC risk attributable to frequent NSAID use (10–13). Initial reports from several research groups indicated that COX-2 was over-expressed in human PCs (14–19); however, there are a number of discrepancies within these studies. These include differences in the cell types found to express COX-2 within prostate tissue, expression in normal or benign tissues and the relationship of level of expression to grade or stage of PC (1,20). More recent studies have found that COX-2 is only minimally expressed in most PCs and that elevated expression is only observed in areas of chronic inflammation or proliferative inflammatory atrophy (21,22).

There have been relatively few animal studies reported that have tested the level of COX-2 expression and the effect of NSAIDs on prostate carcinogenesis. The transgenic adenocarcinoma mouse prostate (TRAMP) model for PC is now well established and demonstrates several characteristics that make it advantageous for chemoprevention studies (23,24). The pathobiology of the TRAMP model has recently been extensively reviewed (23). Two groups of investigators recently reported that the COX-2-selective inhibitor celecoxib decreased prostate tumorigenesis in the TRAMP transgenic model (25,26). Celecoxib, however, has been shown to have multiple non-COX-2-mediated activities that could account for its effect on tumorigenesis in this model (27,28). Until now, investigations into the role of COX-2 in PC were largely hindered by the lack of prostate-specific COX-2 over-expression and knockout (KO) mouse models. The potential for non-COX-mediated effects of NSAIDs, combined with the inconsistent observations regarding COX-2 over-expression in human PC, led us to question whether COX-2 inhibition plays a role in the prevention of tumorigenesis in the TRAMP model. To address this question, we compared the effect of a non-selective COX inhibitor (indomethacin), a COX-2-selective inhibitor (celecoxib) and genetic disruption of COX-2 gene expression on prostate carcinogenesis in the TRAMP model. Interestingly, we found that the expression and activity of both COX-1 and COX-2 are dramatically decreased during TRAMP carcinogenesis and that inhibition of COX activity either by NSAID treatment or by disruption of the COX-2 gene is largely ineffective in preventing PC in the TRAMP model. We recently generated a transgenic mouse model (Pb.COX-2) with the COX-2 gene under the control of the probasin promoter, which allows for COX-2 over-expression only in prostate tissue. In support of our findings in the TRAMP model, over-expression of COX-2 in mouse prostate tissue had no pro-tumorigenic effects on prostates of Pb.COX-2 mice.

To our knowledge, we are the first laboratory to examine a direct role of COX-2 in prostate carcinogenesis using genetically modified mouse models that either over-express or lack expression of COX-2 in the prostate. Our results suggest that COX-2 does not play a tumorigenic role in prostate carcinogenesis in the TRAMP model. The discrepancy of current knowledge on COX-2 in human prostate carcinogenesis needs to be further elucidated in future studies.


   Materials and methods
Top
 Abstract
 Introduction
 Materials and methods
Results
 Discussion
 Funding
 References
 
Materials
Prostaglandin E2 (PGE2), polyclonal anti-COX-2 (murine), polyclonal anti-COX-1 (murine) and COX-1 (murine)-blocking peptide were purchased from Cayman Chemical (Ann Arbor, MI). Polyclonal anti-rabbit PGHS-2 was purchased from Oxford Biomedical Research (Oxford, MI). Biotin-conjugated mouse anti-COX-2 monoclonal antibody was from BD (Franklin Lakes, NJ). Goat polyclonal anti-COX-2 (c-20) and goat polyclonal anti-COX-1 (M-20) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). TRIzol reagent was purchased from Invitrogen (Carlsbad, CA). RNeasy Mini RNA extraction kit was purchased from Qiagen (Valencia, CA). The iScript cDNA Synthesis Kit was purchased from Bio-Rad (Hercules, CA). Taqman® Gene Expression Assay kits and Taqman® Universal PCR Mastermix were purchased from Applied Biosystems (Foster City, CA).

Genetically engineered mice
C57BL/6J TRAMP mice were originally obtained as a kind gift from Dr Norman Greenberg (24) and were maintained as a hemizygote colony. TRAMP (C57 x FVB) F1 mice were generated by crossing C57BL/6J TRAMP hemizygotes with FVB/n mice. Male COX-2 KO mice (B6;129P2-Ptgs2tm1Smi) were obtained from the Taconic (Hudson, NY) Emerging Models Program by kind permission of Dr Robert Langenbach (29). Male COX-2(–/–) mice were crossed with female C57BL/6J mice to generate COX-2(+/–) mice. COX-2(+/–) mice from this cross were then mated to generate male COX-2(–/–) and COX-2(+/+) mice. Male COX-2(–/–) mice were then crossed with female hemizygous C57BL/6J TRAMP mice to generate female mice heterozygous for COX-2 and hemizygous for the TRAMP transgene (C+/–:T+/–). To produce mice for the tumor study, female C+/–:T+/– mice were crossed with either male COX-2 –/– or COX-2 +/+ mice in order to generate hemizygous TRAMP males wild-type (WT) (C+/+:T+/–) or deficient (C–/–:T+/–) for COX-2. The Pb.COX-2 transgenic mice were on the FVB background, which over-express COX-2 under AAR2 probasin promoter (pAAR2Pb) control. Genotyping was done by PCR on DNA isolated from tail snips according to the established protocols for the TRAMP transgene (24) and the Ptgs2tm1Smi mutation (29). All mice were maintained in HEPA-filtered cages according to the Institutional Laboratory Animal Care and Use Committee-approved protocols at Science Park—Research Division, MD Anderson Cancer Center, University of Texas and at the Ohio State University.

TRAMP diet study
Diets were formulated by Research Diets (New Brunswick, NJ) and were based on the AIN-76A semi-purified rodent diet. Celecoxib at either 500 or 1500 p.p.m. and indomethacin at 4 p.p.m. were added to the diets prior to pelleting. The selected doses of celecoxib and indomethacin were well tolerated by TRAMP mice as reported in the previous study (25). TRAMP mice at 3 weeks of age were randomly distributed by weight into groups (n = 30 per group). The TRAMP mice were fed either AIN-76A control diet or COX inhibitor-supplemented AIN-76A diets until the animals were 30 weeks old. Food intakes were recorded every other day. Body weights were measured weekly.

Tumor studies
All mice were weighed and palpated for tumors weekly starting at 12 weeks of age. Mice were necropsied when palpable tumors reached 2 cm in diameter or mice reached 30 weeks of age. Mice that died prior to 30 weeks from causes unrelated to prostate tumor growth were censored from the analyses. At necropsy, the dorsal (DP), lateral (LP) (27), ventral (VP) and anterior (AP) lobes of the prostate were microdissected into individual lobes whenever possible. Weights of all prostatic lobes, regional lymph nodes (LNs) and tumors were recorded. The techniques and protocols for histological dissection of the prostate lobes were in concordance with the 2001 Bar Harbor Pathology Workshop (30). Mice were examined for any grossly observable metastatic tumors in LN, liver and lung. All samples were fixed in 10% buffered formalin and then paraffin embedded for subsequent examination.

Histopathology
The histopathology of embedded tissues from all mice was assessed by scoring 4 µm thick sections stained with hematoxylin and eosin. All sections were coded, blinded and randomized prior to independent analysis by R.D.K. and X.W. Prostatic lobes, including DP, LP, VP and AP, were assessed according to the published criteria that have been well established for the TRAMP model (23). Only lobes that were fully dissectible were included in the analysis. All glands in a given section were classified as normal, prostatic intra-epithelial neoplasia (PIN), well-differentiated carcinoma (WD), moderately differentiated carcinoma or poorly differentiated carcinoma (PD). The phylloide-like lesions described by Kaplan-Lefko et al. (23) and an additional classification of atrophy as described by Shappell et al. (30) were also included in the analysis. For each pair of lobes, the percent of the tissue that was normal, PIN, WD, moderately differentiated carcinoma, PD, phylloide-like lesions and atrophy was determined. These data were used to score each pair of prostatic lobes for the percentage of each pathological grade present and these scores were averaged for all mice in a group. LN, lungs and liver were assessed for metastases by scoring hematoxylin and eosin sections for the presence or absence of apparent metastatic lesions.

Immunohistochemistry
Immunohistochemical staining was done as described (3). Primary polyclonal antibodies against COX-1 and COX-2 (Cayman Chemical) were used at a dilution of 1:500. All immunohistochemical stainings were performed in an OptiMax Automated Cell Staining System (BioGenex, San Ramon, CA). Vas deferens tissue from WT mice was used as positive control for COX-1 and COX-2, whereas vas deferens from a COX-2 KO mouse served as a negative control for COX-2. A COX-1-blocking peptide was used as control for the specificity of COX-1 staining.

Immunoblotting
Immunoblot analysis was carried out using the standard procedures. Four commercial sources of the primary antibodies against COX-2 were analyzed at the following dilutions: 1:500 (Santa Cruz Biotechnology), 1:750 (Cayman Chemical), 1:250 (BD) and 1:500 (Oxford Biomedical Research). Two sources of the COX-1 antibodies were analyzed at dilutions of 1:500 for the Santa Cruz Biotechnology antibody and 1:1000 for the Cayman Chemical antibody.

RNA extraction and real-time PCR (qRT-PCR)
Total RNA was isolated by TRIzol reagent and further purified by RNeasy Mini kit as described by Qiagen. Both the quantity and quality of total RNA were analyzed by the Agilent Bioanalyzer 2100 system. Total RNA was reverse transcribed with an iScript cDNA Synthesis Kit. qRT-PCR was performed with Taqman gene expression primers and ABI mastermix according to the manufacturer’s instructions on an iCycler IQ qRT-PCR Detection System (Bio-Rad).

Quantitation of PGE2
PGE2 was measured on DP and PD using a liquid chromatography/mass spectrometry/mass spectrometry system as described by Kempen et al. (31). The result was expressed as nanograms of PGE2 per milligram protein.

Statistical analysis
Data are presented as means ± SDs or means ± 95% confidence intervals. Inter-group comparisons of PC incidence were made using Fisher’s exact test (two sided). Survival curves were compared using Kaplan–Meier analysis, and the difference between the groups was estimated using the log-ranked test. Comparisons among multiple groups were performed by one-way analysis of variance. A P value of <0.05 was considered statistically significant. All analyses were performed using SPSS 13.0 software (Chicago, IL).


   Results
Top
 Abstract
 Introduction
 Materials and methods
 Results
Discussion
 Funding
 References
 
Cyclooxygenase expression and activity in TRAMPs
Both COX-1 and COX-2 proteins are expressed in DP tissue from WT mice as determined by commercial antibodies from Santa Cruz Biotechnology (Figure 1A). Only DP tissues were used for the determination of COX expression and subsequent PGE2 measurement, since DP had the most severe pathological lesions compared with other lobes (Figure 3C and 4D). Interestingly, the expression of both COX proteins was dramatically reduced in TRAMP DP tissue as early as 10 weeks of age. COX protein expression remained at a reduced level at 16 weeks of age and was further reduced in tissue from PD. The commercial antibody against COX-2 from Cayman Chemical detected the expression of COX-2 in both the WT and TRAMP tissue; however, there was high background binding. The COX-2 antibody from BD only detected COX-2 expression in the positive control, whereas the COX-2 antibody from Oxford Biomedical Research failed to produce any signal. The Cayman Chemical polyclonal COX-1 antibody worked as well as the COX-1 antibody from Santa Cruz Biotechnology. To further confirm the results obtained by immunoblot analysis, we performed real-time PCR to quantify relative expression levels of COX messenger RNA (mRNA) in WT and TRAMP tissues (Figure 1B). The relative expression levels of COX mRNA in DP and PD tissues correlated extremely well with protein levels. TRAMP DP tissue expressed significantly less COX mRNA at 10 and 16 weeks of age and expression was further decreased in PD tissue.


Figure 1
View larger version (60K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1. Cyclooxygenase enzymes and PGE2 levels in DP tissue are reduced in TRAMP mice compared with WT mice. (A) Western blot analysis of COX-1 and COX-2 protein levels in DP tissue from WT (10 weeks), TRAMP DP (10 and 16 weeks) and from histologically confirmed PD tissue. Mouse vas deferens tissue was used as positive control for COX (+). Tissue lysates from two different mice were analyzed for each tissue type and images are representative of two independent experiments. (B) Quantitative RT–PCR of COX-1 and COX-2 mRNA levels in WT (10 weeks) and TRAMP DP tissues (10 and 16 weeks) and PD. Each bar represents the mean COX-1 or COX-2 mRNA expression (error bars are 95% confidence interval) of five animals relative to a common calibrator sample generated by pooling the DP, VP and LP of a single WT mouse. 18s RNA levels were used to normalize the expression of individual samples. Data points with an asterisk are significantly different from other data points in the same panel (P < 0.05). (C) COX-2 immunohistochemical staining of serial sections from TRAMP DP tissue with PIN, moderately differentiated carcinoma (MD) and PD. Vas deferens (Vas-D) from WT mouse served as positive control, whereas vas deferens from a COX-2 KO mouse served as negative control. (D) Steady-state endogenous PGE2 levels in DP tissue from 10-week-old WT and TRAMP mice. Tissues were analyzed for PGE2 by liquid chromatography/mass spectrometry/mass spectrometry; bars represent the mean of values obtained from at least five mice (error bars are standard deviation). Bars with an asterisk are significantly different from other bars in the same panel (P < 0.05).

 
Since most studies reporting COX-2 over-expression were determined by immunohistochemistry methods (14–16,18,19), we also undertook a careful analysis of COX expression in WT and TRAMP tissues using validated methods. The COX-2 immunohistochemical staining protocol was optimized to consistently produce strong staining of positive control tissue (vas deferens) from WT mouse or prostate lobes from the Pb.COX-2 mouse, while not staining negative control vas deferens from COX-2 KO mouse (Figure 1C). We did not, however, detect COX-2 protein expression in prostate tissues from WT or TRAMP mice at any stage of pathology (Figure 1C). These results suggest that COX-2 antibody could only detect COX-2 staining in the positive control tissues but was not sensitive enough to pick up the relatively low expression in prostate tissues. We also found that commercially available COX-2 antibodies produce strong non-specific epithelial and stromal staining in both the TRAMP and COX-2 null prostate tissues if immunohistochemical conditions are not carefully controlled (data not shown). Strong COX-1 protein expression was initially observed in both WT and TRAMP tissues under conditions lacking a negative control (data not shown). However, use of the COX-1-blocking peptide demonstrated that COX-1 staining in prostate tissue is non-specific. In correlation with the expression of COX enzymes, PGE2 levels were significantly lower in DP from 10-week-old TRAMP mice compared with WT littermates (Figure 1D). PGE2 levels remained significantly lower in TRAMP PDs than in DP from WT mice (data not shown). Collectively, the above results suggest that both the expression and the activity of COX enzymes are significantly decreased in the TRAMP tissue compared with the WT littermates.

Histopathological evaluation of TRAMP mice
We characterized the histopathology of TRAMP mice by collecting prostate tissues at different ages (6–35 weeks). The results correlated well with the pathological changes described previously (23). Briefly, histological PIN and early carcinoma were present in 100% of TRAMPs at 8 weeks of age, which progresses to PD with distant site metastasis by 16–32 weeks of age. Metastatic sites included the regional LNs, liver and lung. The WD and moderately differentiated carcinoma were observed at all ages between 8 and 35 weeks old. We scored prostatic glandular ducts and metastases in TRAMP tissues for the tumor studies according to the histopathologic criteria described previously (23). The representative hematoxylin and eosin-stained sections of DP, LP, VP and AP tissues from TRAMP mice and sections of TRAMP lung and liver tissues with micrometastases are shown (Figure 2).


Figure 2
View larger version (163K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 2. Histopathologic criteria used to score prostatic glandular ducts and metastases in tissues from TRAMP tumor studies. Representative hematoxylin and eosin-stained sections of DP (A), LP (B), VP (C) and AP (D) tissues from TRAMP mice as well as sections of TRAMP lung (E) and liver (F) tissues with micrometastases are shown. For sections from each lobe glandular ducts with histological features of normal ducts (N), PIN, WD and moderately differentiated carcinoma (MD) are indicated if present. Micrometastases in the lung and the liver are indicated by arrows. Length of the black bar present in each panel is 100 µm.

 
Effects of celecoxib and indomethacin on the incidence of PDs and metastases in TRAMP mice
As summarized in Table I, feeding TRAMP mice NSAID diets did not reduce the incidence of tumor and metastases in LN, liver and lung compared with TRAMP mice fed the control diet. By 30 weeks, ~60% of TRAMP mice developed palpable tumors (PD) in control- (15 of 26), 500 p.p.m. celecoxib- (18 of 28) and 1500 p.p.m. celecoxib- (15 of 25) fed groups. Although there was a slight reduction in the incidence of PDs (12 of 26) and LN metastases (13 of 26) in mice fed 4 p.p.m. indomethacin, this difference was not significant statistically. There were no significant differences in the size and weight of the PDs between groups (data not shown). Tumor-free survival analysis revealed that NSAID treatment did not prolong the time to development of palpable tumor of TRAMP mice (Figure 3A). There were no significant observed differences in body weights or food intakes for animals that received either control or experimental diets (data not shown).


Figure 3
View larger version (42K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 3. Treatment of TRAMP mice with cyclooxygenase inhibitors does not significantly alter prostate carcinogenesis. Three-week-old TRAMP mice were treated with an AIN-76A semi-purified diet (control, C) or the control diet supplemented with 500 p.p.m. celecoxib (500 Cele), 1500 p.p.m. celecoxib (1500 Cele) or 4 p.p.m. indomethacin (Indo) for 27 weeks. (A) Kaplan–Meier survival analysis for the incidence of palpable tumors. Mice that did not complete the entire 27-week treatment period due to causes other than a palpable tumor are indicated as being censored. (B) Wet weights of AP (n = 13–16), DP (n = 12–16), LP (n = 8–12) and VP (n = 10–13) lobes dissected from mice treated with the indicated diets. Only lobes that were fully dissectible were included in the analysis. Bars represent means, error bars are standard deviation and means within a panel with different letters are significantly different from each other (P < 0.05). (C) Incidence of pathologic lesions in the glandular ducts of prostatic lobes from TRAMP mice fed the indicated diets. The percent of normal (Norm) ducts and ducts containing PIN, WD, moderately differentiated carcinoma (MD), phylloide-like lesions (Phyl), atrophic lesions (Atrophy) and PD was determined for each lobe. Only clearly identifiable lobes were included in the analysis; AP (n = 17–24), DP (n = 19–28), LP (n = 14–17) and VP (n = 1417). Bars represent means and error bars are standard deviation. (D) PGE2 levels in DP tissue from 10-week-old TRAMP mice fed an AIN-76A semi-purified diet (control) or 500 p.p.m. celecoxib, 1500 p.p.m. celecoxib or 4 p.p.m. indomethacin for 2 weeks. Tissues were analyzed for PGE2 by liquid chromatography/mass spectrometry/mass spectrometry; bars represent the mean of values obtained from at least five mice (error bars are standard deviation). Bars with different letters are significantly different from each other (P < 0.05).

 


View this table:
[in this window]
[in a new window]

  		Table I. Incidence of PD and metastases(Mets) in TRAMP mice fed diets containing cyclooxygenase inhibitors

 
Effects of NSAIDs on prostate histopathology and PGE2 levels in TRAMP mice
We further analyzed the pathologic lesions in the glandular ducts of each individual pair of prostatic lobes from the control and experimental mice. We first measured the wet weight of each individual lobe by using lobes that were only fully dissectible upon necropsy, e.g. DP (n = 12–16), LP (n = 8–12), VP (n = 10–13) and AP (n = 13–16). Although the average wet weights of AP, LP and DP were lower in the celecoxib-fed groups than in the control and indomethacin-treated groups, the differences were not statistically significant (Figure 3B). However, there was a significant reduction in wet weight of VP from mice fed 1500 p.p.m. of celecoxib (Figure 3B).

The pathologic lesions in the glandular ducts of individual prostatic lobes were assessed using the published criteria (23). Only clearly identifiable (under the microscope) lobes were included in the analysis. Therefore, we analyzed a total of 19–26 DP, 14–17 LP, 14–17 VP and 17–24 AP. As shown in Figure 3C, treatment of TRAMP mice with both NSAIDs did not significantly alter the distribution of pathologic lesions in the glandular ducts of prostatic lobes. Interestingly, in correlation with the effects of 1500 p.p.m. celecoxib on the wet weight of VP, celecoxib at 1500 p.p.m. slightly, but not significantly, increased the percentage of PIN and decreased the percentage of WD in VP compared with the TRAMP mice fed control diet. However, these effects could be due to the non-COX effects of celecoxib.

To verify the effectiveness of our feeding regimen on TRAMP tissue, 10-week-old TRAMP mice (n = 5 per group) were fed above control or experimental diets for 2 weeks. We found that PGE2 concentration in the TRAMP DP tissue was significantly reduced by 1500 p.p.m. celecoxib and indomethacin compared with the DP of control mice; indomethacin had the greatest significant inhibitory effect (Figure 3D). The above data suggest that NSAIDs effectively reduced the activity of the COX enzymes in TRAMP mice, but had very minimal, if any, effect on the inhibition of prostate tumorigenesis in this animal model.

Effects of genetic disruption of COX-2 on prostate tumorigenesis in TRAMP mice
We next examined whether disruption of the COX-2 gene in the TRAMP mice could inhibit prostate tumorigenesis by crossing TRAMP transgenic mice with COX-2 KO mice. Figure 4A shows that the endogenous PGE2 level was significantly lower in the DP tissue of COX-2 KO mice compared with the COX-2 WT mice. PGE2 concentrations were further reduced in the PD of TRAMP mice with either COX-2 WT or COX-2 KO (Figure 4A). Similar to the treatment with celecoxib and indomethacin, there were no significant effects of the loss of COX-2 gene expression on the incidence of PDs, LN or liver metastases in TRAMP mice by 30 weeks of age (Table II). We observed a significant increase of the percentage of lung metastases in TRAMP x COX-2 KO mice compared with the control littermates, which might be due to the intrinsic variation in the TRAMP model. There were no significant differences in the size and weight of the PDs between groups (data not shown). Tumor-free survival analysis revealed that loss of the COX-2 gene slightly, but not significantly, prolonged the time to development of palpable tumor in the TRAMP x COX-2 KO mice compared with the COX-2-positive littermates (Figure 4B). The lower overall incidence of PDs in these mice (as compared with the NSAID study) is probably due to background strain difference as these mice were not crossed with FVB mice.


Figure 4
View larger version (36K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 4. Genetic disruption of COX-2 expression slightly reduces prostate carcinogenesis in the lateral and VP lobes of TRAMP mice. TRAMP mice either WT (n = 33) or homozygous mutant (KO, n = 35) for the COX-2 gene were followed until 30 weeks of age at which time all surviving animals were necropsied. (A) Steady-state endogenous PGE2 levels in DP tissue and PDs. Tissues were analyzed by liquid chromatography/mass spectrometry/mass spectrometry and bars represent the mean of values obtained from at least five mice (error bars are standard deviation). Bars with different letters are significantly different from each other (P < 0.05). (B) Kaplan–Meier survival analysis for the incidence of palpable tumors. Mice that did not complete the entire 27-week experimental period due to causes other than a palpable tumor are indicated as being censored. (C) Wet weights of AP (n = 23 WT, 19 KO), DP (n = 23 WT, 22 KO), LP (n = 18 WT, 16 KO) and VP (n = 18 WT, 18 KO) lobes dissected from mice with the indicated genotypes. Only lobes that were fully dissectible were included in the analysis. Bars represent means, error bars are standard deviations and KO means within a lobe with an asterisk are significantly different from WT (P < 0.05). (D) Incidence of pathologic lesions in the glandular ducts of prostatic lobes from TRAMP mice with the indicated genotypes. The percent of normal (Norm) ducts and ducts containing PIN, WD, moderately differentiated carcinoma (MD), phylloide-like lesions (Phyl) and PD was determined for each lobe. Only clearly identifiable lobes were included in the analysis; the number of lobes assessed for each genotype was the same as that for the wet weight analysis. Bars represent means and error bars are standard deviation, and bars in KO with an asterisk are significantly different from WT for specific type of lesion (P < 0.05).

 


View this table:
[in this window]
[in a new window]

  		Table II. Incidence of PD and metastases (Mets) in TRAMP mice either WT or homozygous mutant (KO) for the COX-2 gene

 
As shown in Figure 4C, disruption of the COX-2 gene in the TRAMP mice only decreased the wet weight of LP in a significant manner, but not the weights of other lobes. We further analyzed the effects of loss of COX-2 on the pathologic lesions in the glandular ducts of individual prostatic lobes. We found that genetic disruption of COX-2 expression only slightly reduced prostate tumorigenesis in the LP and VP lobes of TRAMP mice (Figure 4D). Taken together, these data suggest that COX-2 deficiency was not sufficient to inhibit tumorigenesis in the TRAMP model.

Effects of COX-2 over-expression on prostate tumorigenesis
To provide further evidence for a role for COX-2 in prostate tumorigenesis, we generated transgenic mice with COX-2 under the control of a probasin promoter (Pb.COX-2), allowing prostate-specific expression of COX-2. Analysis using immunohistochemistry and immunoblotting demonstrated that COX-2 is predominantly expressed in VP, followed by LP and DP (data not shown). The synthesis of PGE2 was also significantly increased in the three lobes from the Pb.COX-2 mice than from the WT littermates. PGE2 levels in DP, LP and VP were 31, 37 and 28 ng/mg protein in the Pb.COX-2 mice, respectively, whereas the concentrations of PGE2 were 15, 2 and 1 ng/mg protein in DP, LP and VP in the WT littermates, respectively. Consistent with what we observed from the NSAID study, over-expression of COX-2 in the Pb.COX-2 mice failed to induce any of the histopathology seen in the TRAMP model (data not shown).


   Discussion
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
Funding
 References
 
There is substantial evidence for the up-regulation of COX-2 in a number of cancers and its pro-tumorigenic activity in these cancers (1). COX-2-selective inhibitors have been intensely studied for their cancer chemopreventive potential. However, available data regarding the expression of COX-2 in PC and its role in prostate carcinogenesis have been inconclusive. Considering the lack of definitive evidence showing a benefit of COX-2 inhibition in PC cells indicates a need for a more critical examination of the role for COX-2 in prostate carcinogenesis (32).

In this study, we addressed this question using both pharmacologic and genetic approaches. We first determined that neither celecoxib nor indomethacin was able to inhibit the development and progression of PC in TRAMP mice. Our results contradict two earlier studies in TRAMP mice in which celecoxib reduced the incidence of PIN, decreased tumor incidence and blocked metastasis (25,26). There appears to be little agreement regarding experimental design, however, between our study and those studies. These differences include selection of background strain of the TRAMP mice, number of animals used per group and the end points to be assessed. Background strain has been reported to affect the prostate tumor phenotype in the TRAMP model (33). In agreement with our findings, Zeng et al. (34) found that the selective COX-2 inhibitor nimesulide failed to inhibit prostate tumorigenesis in probasin/SV40 T antigen transgenic rats, a model in which carcinogenesis is genetically initiated in a same manner as in the TRAMP model. Recently, COX-2-selective inhibitors have been reported to have functions that are independent of COX-2 activity that could account for their effects on prostate tumorigenesis (27,32,35). Taken together, experiments using pharmacologic approaches do not appear to provide strong evidence that COX-2 plays a pro-tumorigenic role in prostate carcinogenesis.

We next used genetic approaches to definitively determine the role of COX-2 in prostate carcinogenesis. In support of the results from the NSAID study, deletion of COX-2 in the TRAMP mice had essentially no effect on prostate tumorigenesis. Similarly, we did not observe any pathological lesions in the prostates of the COX-2 over-expression mice. In contrast, genetic studies using either transgenic or KO technology have firmly established a link between COX-2 and tumorigenesis in cancers of colon, mammary, skin and bladder (1,3). To our knowledge, we are the first laboratory to examine the role of COX-2 in prostate carcinogenesis using genetic approaches. Our results indicate that COX-2 may not play a critical role in prostate carcinogenesis. However, given the recent evidence that COX-2 may be over-expressed only in areas of inflammation in PC tissues (21,22), a role for COX-2 in either the initiation or promotion of this disease by eliciting an inflammatory response cannot be ruled out.

Since our results are in contradiction to many previously published studies, we performed a careful analysis of the expression and activity of the COXs in the WT and TRAMP tissues. Unexpectedly, both the expression and activity of COX-1 and COX-2 were substantially and progressively reduced as the PC progressed in the TRAMPs compared with the WT littermates. To show that this phenomenon was not unique to the TRAMP model, we examined the expression of COX mRNAs in the AP and DP of 10-week-old prostate-specific phosphatase and tensin homolog deleted on chromosome 10 KO mice (36). Similar to what we found in the TRAMP mice, COX mRNA levels were significantly reduced in both the AP and DP of conditional phosphatase and tensin homolog deleted on chromosome 10 KO mice compared with the WT littermates (data not shown). Interestingly, a substantial amount of in vitro studies reported that the expression of COX-2 was either low or non-detectable in PC cells (22,32,35,37,38), but high in normal prostate epithelial cells (38). Up-regulation of COX-2 results in increased synthesis of PGE2, which is the key mediator of the COX-2 pathway that activates downstream signaling and thus contributes to tumorigenesis by inducing cell proliferation, angiogenesis, invasion and metastasis (3). The exact role of PGE2 in prostate carcinogenesis, however, is currently unknown. Although several studies have assumed that PGE2 levels are increased in PC tissues and in vitro-cultured PC cells, in most studies PGE2 levels were not measured. We found that PGE2 levels were significantly reduced in TRAMP tissue compared with the WT littermates. Taken together, the loss of COX expression and its activity during prostate carcinogenesis concurs well with the results from our tumor studies and further suggests that COX-2 is unlikely to play a critical role in prostate carcinogenesis.

A review of the literature reveals possible dissimilarities between studies on COX-2 expression in PC, including differences in the cell types found to express COX-2, expression levels in benign versus malignant tissue and the relationship of level of expression to grade or stage of PC (1,20). Most studies used immunohistochemistry methods to determined COX-2 expression that had neither positive nor negative controls (16,18,19,25). To address this issue, we carefully optimized immunohistochemistry protocols to consistently produce strong staining of positive control tissue, while not staining negative control tissue. To address the same issue, Zha et al. (22) conducted a very comprehensive and well-controlled analysis of COX-2 expression in human PC tissues. Tissues from a total of 144 PC patients were analyzed, far more than in any similar study. In contrast to the previous results, the authors found only scattered expression of COX-2 in <1% of cells and no difference between cancer and normal cells as determined by immunohistochemistry. The authors found an increase in COX-2 expression only in areas of inflammation, which was in agreement with Wang et al. (21) who reported that COX-2 up-regulation is only associated with chronic inflammation in benign prostate hyperplasia. Zha et al. (21) further performed qRT-PCR analysis on a subset of frozen tissues. Interestingly, in agreement with our findings, the authors found that the mean mRNA levels of COX-2 were considerably reduced in malignant prostate as compared with the normal tissue, suggesting a down-regulation of COX-2 in PC. Similarly, by using well-controlled immunohistochemistry and in situ hybridization analysis, Shappell et al. (17) found that COX-2 expression was more often reduced in prostate tumor compared with the benign tissue, and this reduction was significant for Gleason score 5 and 6 tumors. However, the authors did not examine the level of COX-2 expression in normal prostate tissue. Although Gupta et al. (25) found a progressive increase in COX-2 expression in the TRAMP tissue from 8–24 weeks old, COX-2 expression was lost after 24 weeks of age. Thus, there are a substantial number of reports on human and murine tissues that show COX-2 expression diminishes during progression of the lesion.

Our study also suggests that the expression of COX-1 was progressively lost during prostate tumorigenesis in the TRAMP model. Given the fact that indomethacin failed to inhibit prostate tumorigenesis in TRAMP mice, COX-1 may also not play a role in the development and progression of this disease. Contrary to the situation in most epithelial tissues where COX-2 is not expressed under normal conditions, COX-2 is highly expressed in normal prostate (39). The constitutive expression of COX-2 is probably to be important in maintaining normal homeostasis and function of the prostate (38). Although the mechanisms underlying the marked decrease of COX mRNA and protein levels during prostate tumorigenesis have yet to be determined, there are a number of possible mechanisms, including negative feedback regulation due to decreased substrate availability due to competition with lipoxygenases for arachidonic acid, inhibition of COX-2 transcription due to competition for binding on the COX-2 promoter between WT p53 and TATA-binding proteins (40) and methylation of the CpG island upstream of the COX-2 gene (41). Regardless of the mechanism, it is not probable that loss of COX expression contributes to the tumorigenesis of PC, however, further studies are needed to address this question and the basis for this molecular change.

In conclusion, we have demonstrated that neither pharmacological inhibition nor genetic modification of COX-2 had an effect on prostate carcinogenesis. The lack of a pro-tumorigenic role for COX-1 and COX-2 in prostate carcinogenesis is further supported by our observation that both the expression and activity of COX-1 and COX-2 are progressively lost during prostate carcinogenesis. This is the first study to definitively establish the lack of a positive link between COX-2 and the progression of PC. Given the epidemiological evidence that NSAID use may be protective against PC (10–13) and the evidence that COX-2 may be associated with inflammation processes in PC, a role for COX-2 in the initiation of PC cannot yet be ruled out. Further work will be required, however, to determine the significance of the loss of COX genes in prostate carcinogenesis, and the role of COX-2, if there is any, in the early stage of PC development.


   Funding
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Funding
References
 
National Cancer Institute (CA091865 [GenBank] to R.D.K. and CA107588 [GenBank] to S.M.F.)


   Footnotes
 
{dagger} In memorium (1962–2006) of Russell D.Klein, Ph.D. died 1 December 2006 after a year-long battle with leukemia. He was a promising young scientist, an exemplary mentor and a fine human being. Back


   Acknowledgments
 
We thank Dr Mark L.Failla for his critical suggestions during the preparation of the manuscript, Dr Steve K.Clinton for providing the immunohistochemical facility and for critical suggestions, Robert Rengel for assistance in mouse breeding, Kim Carter for technical assistance with the immunohistochemical stainings.

Conflict of Interest Statement: None declared.


   References
Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 Funding
 References
 

  1. Zha S, et al. Cyclooxygenases in cancer: progress and perspective. Cancer Lett. (2004) 215:1–20.[CrossRef][Web of Science][Medline]
  2. Cao Y, et al. Fatty acid CoA ligase 4 is up-regulated in colon adenocarcinoma. Cancer Res. (2001) 61:8429–8434.[Abstract/Free Full Text]
  3. Klein RD, et al. Transitional cell hyperplasia and carcinomas in urinary bladders of transgenic mice with keratin 5 promoter-driven cyclooxygenase-2 overexpression. Cancer Res. (2005) 65:1808–1813.[Abstract/Free Full Text]
  4. Kismet K, et al. Celecoxib: a potent cyclooxygenase-2 inhibitor in cancer prevention. Cancer Detect. Prev. (2004) 28:127–142.[CrossRef][Web of Science][Medline]
  5. Irani J, et al. Effect of nonsteroidal anti-inflammatory agents and finasteride on prostate cancer risk. J. Urol. (2002) 168:1985–1988.[CrossRef][Web of Science][Medline]
  6. Langman MJ, et al. Effect of anti-inflammatory drugs on overall risk of common cancer: case-control study in general practice research database. BMJ (2000) 320:1642–1646.[Abstract/Free Full Text]
  7. Leitzmann MF, et al. Aspirin use in relation to risk of prostate cancer. Cancer Epidemiol. Biomarkers Prev. (2002) 11:1108–1111.[Abstract/Free Full Text]
  8. Norrish AE, et al. Non-steroidal anti-inflammatory drugs and prostate cancer progression. Int. J. Cancer (1998) 77:511–515.[CrossRef][Web of Science][Medline]
  9. Paganini-Hill A, et al. Aspirin use and chronic diseases: a cohort study of the elderly. BMJ (1989) 299:1247–1250.[Abstract/Free Full Text]
  10. Garcia Rodriguez LA, et al. Inverse association between nonsteroidal anti-inflammatory drugs and prostate cancer. Cancer Epidemiol. Biomarkers Prev. (2004) 13:649–653.[Abstract/Free Full Text]
  11. Habel LA, et al. Daily aspirin use and prostate cancer risk in a large, multiracial cohort in the US. Cancer Causes Control (2002) 13:427–434.[CrossRef][Web of Science][Medline]
  12. Nelson JE, et al. Inverse association of prostate cancer and non-steroidal anti-inflammatory drugs (NSAIDs): results of a case-control study. Oncol. Rep. (2000) 7:169–170.[Web of Science][Medline]
  13. Roberts RO, et al. A population-based study of daily nonsteroidal anti-inflammatory drug use and prostate cancer. Mayo Clin. Proc. (2002) 77:219–225.[Abstract/Free Full Text]
  14. Kirschenbaum A, et al. Expression of cyclooxygenase-1 and cyclooxygenase-2 in the human prostate. Urology (2000) 56:671–676.[CrossRef][Web of Science][Medline]
  15. Lee LM, et al. Expression of cyclooxygenase-2 in prostate adenocarcinoma and benign prostatic hyperplasia. Anticancer Res. (2001) 21:1291–1294.[Web of Science][Medline]
  16. Madaan S, et al. Cytoplasmic induction and over-expression of cyclooxygenase-2 in human prostate cancer: implications for prevention and treatment. BJU Int. (2000) 86:736–741.[CrossRef][Web of Science][Medline]
  17. Shappell SB, et al. Alterations in lipoxygenase and cyclooxygenase-2 catalytic activity and mRNA expression in prostate carcinoma. Neoplasia (2001) 3:287–303.[CrossRef][Web of Science][Medline]
  18. Uotila P, et al. Increased expression of cyclooxygenase-2 and nitric oxide synthase-2 in human prostate cancer. Urol. Res. (2001) 29:23–28.[Medline]
  19. Yoshimura R, et al. Expression of cyclooxygenase-2 in prostate carcinoma. Cancer (2000) 89:589–596.[CrossRef][Web of Science][Medline]
  20. Hussain T, et al. Cyclooxygenase-2 and prostate carcinogenesis. Cancer Lett. (2003) 191:125–135.[CrossRef][Web of Science][Medline]
  21. Wang W, et al. Cyclooxygenase-2 expression correlates with local chronic inflammation and tumor neovascularization in human prostate cancer. Clin. Cancer Res. (2005) 11:3250–3256.[Abstract/Free Full Text]
  22. Zha S, et al. Cyclooxygenase-2 is up-regulated in proliferative inflammatory atrophy of the prostate, but not in prostate carcinoma. Cancer Res. (2001) 61:8617–8623.[Abstract/Free Full Text]
  23. Kaplan-Lefko PJ, et al. Pathobiology of autochthonous prostate cancer in a pre-clinical transgenic mouse model. Prostate (2003) 55:219–237.[CrossRef][Web of Science][Medline]
  24. Greenberg NM, et al. Prostate cancer in a transgenic mouse. Proc. Natl Acad. Sci. USA (1995) 92:3439–3443.[Abstract/Free Full Text]
  25. Gupta S, et al. Suppression of prostate carcinogenesis by dietary supplementation of celecoxib in transgenic adenocarcinoma of the mouse prostate model. Cancer Res. (2004) 64:3334–3343.[Abstract/Free Full Text]
  26. Narayanan BA, et al. Regression of mouse prostatic intraepithelial neoplasia by nonsteroidal anti-inflammatory drugs in the transgenic adenocarcinoma mouse prostate model. Clin. Cancer Res. (2004) 10:7727–7737.[Abstract/Free Full Text]
  27. Kulp SK, et al. 3-Phosphoinositide-dependent protein kinase-1/Akt signaling represents a major cyclooxygenase-2-independent target for celecoxib in prostate cancer cells. Cancer Res. (2004) 64:1444–1451.[Abstract/Free Full Text]
  28. Williams CS, et al. Celecoxib prevents tumor growth in vivo without toxicity to normal gut: lack of correlation between in vitro and in vivo models. Cancer Res. (2000) 60:6045–6051.[Abstract/Free Full Text]
  29. Morham SG, et al. Prostaglandin synthase 2 gene disruption causes severe renal pathology in the mouse. Cell (1995) 83:473–482.[CrossRef][Web of Science][Medline]
  30. Shappell SB, et al. Prostate pathology of genetically engineered mice: definitions and classification. The consensus report from the Bar Harbor meeting of the Mouse Models of Human Cancer Consortium Prostate Pathology Committee. Cancer Res. (2004) 64:2270–2305.[Abstract/Free Full Text]
  31. Kempen EC, et al. Simultaneous quantification of arachidonic acid metabolites in cultured tumor cells using high-performance liquid chromatography/electrospray ionization tandem mass spectrometry. Anal. Biochem. (2001) 297:183–190.[CrossRef][Web of Science][Medline]
  32. Wagner M, et al. Resistance of prostate cancer cell lines to COX-2 inhibitor treatment. Biochem. Biophys. Res. Commun. (2005) 332:800–807.[CrossRef][Web of Science][Medline]
  33. Klein RD. The use of genetically engineered mouse models of prostate cancer for nutrition and cancer chemoprevention research. Mutat. Res. (2005) 576:111–119.[Web of Science][Medline]
  34. Zeng Y, et al. Inhibition of prostate carcinogenesis in probasin/SV40 T antigen transgenic rats by raloxifene, an antiestrogen with anti-androgen action, but not nimesulide, a selective cyclooxygenase-2 inhibitor. Carcinogenesis (2005) 26:1109–1116.[Abstract/Free Full Text]
  35. Patel MI, et al. Celecoxib inhibits prostate cancer growth: evidence of a cyclooxygenase-2-independent mechanism. Clin. Cancer Res. (2005) 11:1999–2007.[Abstract/Free Full Text]
  36. Wang S, et al. Prostate-specific deletion of the murine Pten tumor suppressor gene leads to metastatic prostate cancer. Cancer Cell (2003) 4:209–221.[CrossRef][Web of Science][Medline]
  37. Lim JT, et al. Sulindac derivatives inhibit growth and induce apoptosis in human prostate cancer cell lines. Biochem. Pharmacol. (1999) 58:1097–1107.[CrossRef][Web of Science][Medline]
  38. Subbarayan V, et al. Differential expression of cyclooxygenase-2 and its regulation by tumor necrosis factor-alpha in normal and malignant prostate cells. Cancer Res. (2001) 61:2720–2726.[Abstract/Free Full Text]
  39. O'Neill GP, et al. Expression of mRNA for cyclooxygenase-1 and cyclooxygenase-2 in human tissues. FEBS Lett. (1993) 330:156–160.[Web of Science][Medline]
  40. Subbaramaiah K, et al. Inhibition of cyclooxygenase-2 gene expression by p53. J. Biol. Chem. (1999) 274:10911–10915.[Abstract/Free Full Text]
  41. Yegnasubramanian S, et al. Hypermethylation of CpG islands in primary and metastatic human prostate cancer. Cancer Res. (2004) 64:1975–1986.[Abstract/Free Full Text]
Received July 5, 2007; revised September 13, 2007; accepted October 2, 2007.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?



This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
29/1/120    most recent
bgm226v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Wang, X.
Right arrow Articles by Klein, R. D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Wang, X.
Right arrow Articles by Klein, R. D.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?