Skip Navigation


Carcinogenesis Advance Access originally published online on October 27, 2006
Carcinogenesis 2007 28(4):848-857; doi:10.1093/carcin/bgl204
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
28/4/848    most recent
bgl204v1
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 arrow Search for citing articles in:
ISI Web of Science (3)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Tijhuis, M. J.
Right arrow Articles by Kampman, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Tijhuis, M. J.
Right arrow Articles by Kampman, E.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2006. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

Glutathione S-transferase phenotypes in relation to genetic variation and fruit and vegetable consumption in an endoscopy-based population

Mariken J. Tijhuis1, Marleen H.P.W. Visker1, Jac M.M.J.G. Aarts2, Wilbert H.M. Peters3, Hennie M.J. Roelofs3, Liesbeth Op den Camp2, Ivonne M.C.M. Rietjens2, Anne-Marie J.F. Boerboom2, Fokko M. Nagengast3, Frans J. Kok1 and Ellen Kampman1,*

1 Division of Human Nutrition, Wageningen University Wageningen, The Netherlands
2 Division of Toxicology, Wageningen University Wageningen, The Netherlands
3 Department of Gastroenterology, Radboud University Nijmegen Medical Centre Nijmegen, The Netherlands

*To whom correspondence should be addressed at: Division of Human Nutrition, Wageningen University, PO Box 8129, 6700 EV Wageningen, The Netherlands. Tel: +31 317 48 38 67; Fax: +31 317 48 27 82; E-mail: Ellen.Kampman{at}wur.nl.


   Abstract
Top
 Abstract
Introduction
 Methods
 Statistical analyses
 Results
 Discussion
 References
 
High glutathione S-transferase (GST) activity may contribute to colorectal cancer prevention. Functional polymorphisms are known in the GSTM1, GSTT1, GSTA1 and GSTP1 genes. The influence of these GST polymorphisms and recent fruit and vegetable consumption on GST levels and activity has not been investigated simultaneously in a human population. Also, it is not clear if blood GST activity reflects rectal GST activity. Therefore, we determined GST polymorphisms in 94 patients scheduled for sigmoidoscopy. Rectal GST isoenzyme levels (GSTM1, GSTM2, GSTT1, GSTA and GSTP1) were measured by quantitative western blotting, and rectal and white blood cell total GST activities were measured spectrophotometrically using 1-chloro-2,4-dinitrobenzene (CDNB) as a substrate. Vegetable and fruit consumption was assessed by dietary record. As expected, the GSTM1 and GSTT1 deletion polymorphisms, and the GSTA1 g.-69C->T polymorphism significantly affected the respective isoenzyme levels. Also, rectal GST isoenzyme levels differed between those with and without recent consumption of Alliaceae, Cucurbitaceae, Apiaceae and citrus fruit. Rectal GST activity, however, was not clearly influenced by fruit and vegetable consumption. It was most significantly determined by the GSTP1 c.313A->G polymorphism; compared with the 313AA genotypes, the 313AG and 313GG genotypes showed 36 and 67 nmol/min/mg protein (P < 0.001) lower GST activity, respectively. The correlation between rectal and white blood cell GST activities was low (r = 0.40, P < 0.001), and the relevance of the various genetic and dietary factors appeared to differ between the two tissues. In conclusion, this study indicates that the GST enzyme system is influenced by both GST polymorphisms and consumption of fruits and vegetables. The latter appeared more important for individual rectal GST isoenzyme levels than for total GST activity, which could affect detoxification of isoenzyme-specific substrates. The study results do no support the use of white blood cell GST activity as a surrogate measure for rectal GST activity.

Abbreviations: CRC, colorectal cancer; GST, glutathione S-transferase; DNP-SG, 2,4-dinitrophenyl-S-glutathione.


   Introduction
Top
 Abstract
 Introduction
Methods
 Statistical analyses
 Results
 Discussion
 References
 
The gastro-intestinal tract is constantly exposed to exogenous compounds with genotoxic potential. This genotoxicity can be inherent to the compound itself or the result of endogenous bioactivation. The human body is equipped with a defense system, among which are phase II biotransformation enzymes, which alter the toxic compounds and facilitate their excretion. Phase II biotransformation enzymes have been classified into several families, based on the type of reaction they catalyze, one of the most important being the glutathione S-transferases (GSTs, EC 2.5.1.18 [EC] ) (1). These enzymes catalyze the conjugation of reduced glutathione to a wide range of electrophilic substrates, including ultimate carcinogens (2,3). In the distal part of the gastro-intestinal tract, the colon and rectum, occurrence of neoplasia is high (4). This may in part be due to its high cell turnover (5), but also be due to low basal GST expression (6,7). Support for an inverse association between GST capacity and tumor incidence is, however, mostly indirect. Other tissues, e.g. breast and lung, also show high tumor risk with relatively low GST expression and the reverse, low tumor risk with high GST expression, applies to the small intestine and liver (8). In observational studies, lower cancer incidence has been linked to diets high in fruits and vegetables (9). An explanation for this observation may be that fruit and vegetable components, e.g. from Brassica vegetables and citrus fruits, induce detoxification enzymes such as GSTs (10). In a small number of human dietary intervention studies GSTs have been reported to be inducible, in various tissues; in the colorectal area (11,12), in plasma or peripheral lymphocytes (13–16) and in saliva (16,17). Because GST levels are relatively low in the colorectal area as compared with other organs, upregulation of GST enzymes may have considerable protective impact in the colon and rectum (6,7). The induction process involves activation of certain signal transduction pathways by fruit and vegetable components acting through transcription factor binding sites, which are present in the promoters of GSTs (18).

Interestingly, interindividual polymorphic variation exists in the GST genes (3). Depending on exposure to (pro)carcinogens, individuals with different genetic GST variants have been reported to have moderately different cancer risks (3), adding more support for a role of GSTs in human cancer susceptibility. In the coding sequence of GST genes, genetic polymorphisms can affect e.g. the catalytic activity of the enzymes (19). In the regulatory sequence, genetic variation can result in altered binding of transcription factors and altered mRNA levels (20), translating into changes in GST isoenzyme levels (21). In light of the effects that fruit and vegetable components may have on the regulatory region, polymorphisms in this area could be important for nutritional strategies aiming to upregulate GST enzyme capacity.

The aim of the present human observational study was to comprehensively assess GST protein phenotypes (i.e. GST isoenzyme levels and total GST activity) in the rectum in relation to genetic variation (i.e. GSTM1, GSTT1, GSTA1 and GSTP1 polymorphisms) and recent consumption of fruits and vegetables. Moreover, white blood cell GST activities were measured to investigate if these factors show similar effects on blood GST phenotype, and thus if blood can be used as a surrogate tissue.


   Methods
Top
 Abstract
 Introduction
 Methods
Statistical analyses
 Results
 Discussion
 References
 
Population
Participants were recruited in two outpatient endoscopy clinics in the Netherlands from patients scheduled for a sigmoidoscopy, between January 2003 and June 2004. Eligibility criteria were: age between 18 and 75 years, Caucasian, no chronic inflammatory bowel disease (past or present), no inflammation in the distal colon at the time of endoscopy, no sporadic colorectal cancer (CRC; past or present), and no bowel resection. Of all eligible patients, 235 (64%) were invited to participate in the study. Invitation was dependent on whether the time between identification of the patient and the start date of the food record was sufficient for the consent procedure. Of all invited patients, 105 agreed to participate (45%). Main reasons for not participating were the stress caused by the prospect of the endoscopy in general and fear for the biopsy specifically. Of all participants, 11 were excluded at endoscopy or after inspection of tissue pathology (PA) results because inclusion criteria were not met. This resulted in a final study population of 94 individuals. The study was approved by the Medical Review Boards of both hospitals. All participants gave their written informed consent.

Medical, dietary and lifestyle information
The indication for endoscopy was recorded from the endoscopy request form. Information on the macroscopic and (if available) microscopic result of the endoscopy was recorded from the endoscopy report and (if available) PA report.

Participants kept a 3-day dietary record, the third day ending at the time of endoscopy. Type and amount of food consumed were recorded in a structured open entry format. All records were checked for quality and completeness by the same, trained dietician of the Division of Human Nutrition of Wageningen University; ~50% of dietary records required follow-up by telephone, which was successful for the majority within 3 days after endoscopy. Processing into food quantities and coding was done according to the most recent standard manual on food portions and household measures and the Dutch Food composition Table (22,23). Conversion into amounts of nutrients was done using the VBS Food Calculation System (24). Fruits were subdivided in citrus (fruit and juice) and non-citrus fruits, vegetables in botanically defined subtypes: Alliaceae (e.g. garlic, leek), Apiaceae (e.g. celery, carrot), Brassicaceae (e.g. cauliflower, broccoli), Compositae (e.g. endive, lettuce), Cucurbitaceae (e.g. zucchini, cucumber), Solanaceae (e.g. bell pepper, tomato; potato not included), and a restgroup.

General lifestyle information was collected through a semi-structured questionnaire containing questions about age, sex, weight, height, smoking habits, medication, disease and family history of cancer.

Specimen collection and preparation
Biopsies
Flexible sigmoidoscopy was performed with the patient in left lateral decubitus position. Biopsies (~20 mg each) were taken from normal rectal mucosa at a distance of 5–15 cm from the anal verge and were snap-frozen in liquid nitrogen.

Blood. Blood (3 x 9 ml) was drawn shortly after endoscopy by venipuncture in Vacuette EDTA K3 Tubes (Greiner Bio-One, Alphen a/d Rijn, the Netherlands).

Leukocytes and lymphocytes. Within 5 min after blood draw, duplicate portions of 1.4 ml whole blood were mixed with 12.6 ml Puregene RBC lysis solution (Gentra systems, BIOzym group, Landgraaf, The Netherlands) and kept on ice for 10–30 min. Samples were then centrifuged at 4°C and 2500 g for 10 min. Leukocyte pellets were not collected of the first 19 participants. Duplicate portions of 4 ml of whole blood were used for isolation of lymphocytes on Histopaque-1077 (Sigma-Aldrich, Zwijndrecht, The Netherlands) according to the instruction of the manufacturer. Leukocyte and lymphocyte pellets were resuspended in 1 ml phosphate-buffered saline (PBS, pH 7.4, Invitrogen, Breda, The Netherlands), centrifuged for 5 min at 10 000 g and stored at –80°C. Leukocyte DNA. One tube of whole blood was centrifuged at 1100 g for 10 min. After removal of plasma, the buffy coat layer was remixed with the blood remnant for later DNA extraction and stored at –80°C.

Laboratory analyses
Genotyping
DNA was extracted from buffy coat cells (QIAamp 96 DNA blood kit, Qiagen Benelux B.V., Venlo, The Netherlands), and samples were stored with negative controls at 4°C. The GSTM1 and GSTT1 deletion polymorphisms were determined simultaneously by allele-specific multiplex PCR (25), in which a ß-globin gene fragment was co-amplified as internal positive control. The GSTA1 g.–69C->T polymorphism was determined by PCR–RFLP according to Coles et al. (26). Some modifications to the Coles' protocol were made; annealing temperature was set at 61°C and Eam1104I (Fermentas GmbH, St Leon-Rot, Germany) was used as restriction enzyme. The GSTP1 c.313A->G polymorphism was assessed by PCR–RFLP according to Harries et al. (27). GSTA1 g.–69C->T and GSTP1 c.313A->G genotypes were in Hardy–Weinberg equilibrium (HWE; {chi}2 = 2.37, P-value = 0.12 and {chi}2 = 0.07, P-value =0.79, respectively).

The 5' regulating region of the GSTP1 gene was screened for new polymorphisms by DNA sequencing of the –434 to +296 region of the GSTP1 gene (relative to the translation initiation site of GenBank accession no. AY324387 [GenBank] ; experimental details available on request). No new polymorphisms were identified. We confirmed the GSTP1 g.217G->A (rs1079719 on the NCBI SNP website), g.227G->A (rs1871041) and g.272C->G (rs4147581) polymorphisms and genotyped them by pyrosequencing (the g.223 G-insertion polymorphism was also confirmed, but genotyping was unsuccessful). For genotyping, PCR was performed with AccuPrime GC-rich DNA polymerase (Invitrogen). The resulting 279 bp amplicon was used for two reverse directed pyrosequencing analyses; the first to analyze the 233–209 region of GSTP1 in order to genotype the g.217G->A and g.227G->A polymorphisms and the second to analyze the 275–266 region of GSTP1 in order to genotype the g.272C->G polymorphism. The GSTP1 g.217G->A and g.227G->A genotypes were in HWE {chi}2 =0.15, P = 0.70 and {chi}2 =1.74, P = 0.19, respectively), whereas the GSTP1 g.272C->G polymorphism was not ({chi}2 =5.58, P = 0.018). All polymorphisms were genotyped in duplicate; reproducibility was 100%.

Protein assays
Two rectal biopsies were homogenized on ice using a frozen (–20°C) pestle, and suspended in 100 µl of 20 mM Tris–HCl (pH 7.5). Leukocyte and lymphocyte pellets were resuspended in 100 µl 20 mM Tris–HCl and cells were lysed by sonification (Sonorex RK100 ultrasonic bath, Bandelin electronic, Berlin, Germany) on ice during 10 min. The resulting rectal tissue and white blood cell lysates were centrifugated at 16 000 g and supernatants were alliquoted and refrozen at –80°C, until further measurement.

Total protein was measured by the BCA protein assay reagent kit (Pierce Rockford, IL, USA) using BSA as a standard, following the manufacturer's instructions.

Levels of rectal GST M1, M2, T1, A and P1 were determined by western blotting using monoclonal antibodies (11) and subsequent densitometric analyses of immunoblots. Known amounts of purified GSTs were run in parallel with the samples and served as standards. For quantification of GSTM2 protein, M2 bands were calculated relative to the M1 standard. The detection limit of the immunoblot assays was 20 ng GST protein/mg total protein. GST isoenzyme levels were was normalized to total protein content and expressed as ng GST protein/mg total protein.

Total GST enzyme activity was measured spectrophotometrically using 1-chloro-2,4-dinitrobenzene (CDNB, Sigma-Aldrich) as a substrate according to the method of Habig et al. (28), but adapted for microplate reader (SpectraMax 340, Molecular Devices Corporation) (29). Measurements were performed at 340 nm and 37°C, for 3 min, in triplicate. Data were analyzed using SOFTmaxPRO software (version 2.2.1, Molecular Devices Corporation). The average coefficient of variation in GST activity was 7.3% for rectal samples, 12.4% for leukocytes and 11.2% for lymphocytes. GST enzyme activity was expressed as nmol 2,4-dinitrophenyl-S-glutathione (DNP-SG) produced/min/mg total protein.

Sample storage
White blood cell pellets were stored intact at –80°C for 6.3 ± 4.5 months after blood sampling until preparation and refreezing. GST activity was then measured within 1 month. Rectal biopsies were stored intact at –80°C for 7.5 ± 4.5 months after tissue sampling until homogenation and refreezing; GST activity was then measured within 1 month and GST isoenzyme levels ~3 months later.


   Statistical analyses
Top
 Abstract
 Introduction
 Methods
 Statistical analyses
Results
 Discussion
 References
 
From two subjects no rectal tissue was obtained and for another two subjects there was no dietary information available. One extreme outlier in lymphocyte GST activity (523 nmol/min/mg protein) was excluded from lymphocyte-analyses. For two subjects with the GSTM1 null genotype, a GSTM1 protein value was measured. Since some cross-reactivity with other GSTM proteins may have occurred here, these two GSTM1 values were set to zero.

Vegetable and fruit consumption was dichotomized into ‘did’ or ‘did not’ consume on 1 of the 2 days before endoscopy. Linear regression models were used to evaluate factors confounding the associations between genotype and phenotype, or between fruit and vegetable consumption and phenotype (i.e. >10% change in ß-estimate), or statistically significantly contributing to phenotype (i.e. P < 0.05). GST genotype–phenotype associations were evaluated for age, sex, sample storage time and family history of CRC, and models were adjusted for age, sex and sample storage time. Sample storage time was calculated as the time between specimen collection and measurement performance, and contributed statistically significantly to phenotype in most models. Although age and sex did not contribute significantly, they were included because they are important general population parameters. Fruits and vegetables–GST phenotype associations were evaluated for age, sex, sample storage time, outpatient clinic, family history of CRC,season, smoking, coffee consumption, alcohol consumption, presence of diverticula, hemorrhoids and adenomas, and models were adjusted for age, sex, sample storage time and smoking. Smoking was defined as smoking on one of the two days before endoscopy.

The difference in GST phenotype outcome and its 95% confidence interval (CI) was reported for the variant genotypes as compared with the most common homozygous genotype variant, and for the ‘did-consume’ groups as compared with the non-consumers. For the reference groups (the most common homozygous genotype variant and the non-consumers), least-squares adjusted means were calculated, using mixed models.

Pearson correlation coefficients were calculated to evaluate the strength of the association between rectal and white blood cell GST activities.

Haplotypes for GSTP1 polymorphisms were estimated using the Hplus program version 2.5, available online (Fred Hutchinson Cancer Research Center. Hplus. http://qge.fhcrc.org/hplus). The GSTP1 g.217G->A, g.227G->A and g.272C->G variant nucleotides were not linked and were therefore investigated separately. Inclusion of the GSTP1 c.313A->G polymorphism (corresponding to genomic position 1377) in haplotype estimation resulted in six haplotypes, the two most common were: GGG-A (44.2%) and AGC-G (30.5%). All statistical analyses were performed using SAS software, version 9.1 (SAS institute, Cary, NC).


   Results
Top
 Abstract
 Introduction
 Methods
 Statistical analyses
 Results
Discussion
 References
 
Table I shows the distribution of general characteristics and consumption of fruit and vegetable subtypes between groups of lower and higher rectal GST activity within the study population. Rectal GST activity was lower with higher age, longer duration of ex-smoking, and the presence of hemorrhoids and diverticula. In the higher rectal GST activity group, there were more Alliaceae consumers.


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

  		Table I General study population characteristics by rectal GST activity

 
In Table II the GST genotypes and phenotypes are presented for the rectal GST activity groups and for the total population. Rectal GST activity was higher with presence of the GSTP1 272G- and 313A-alleles. All GST isoenzyme levels were higher with higher total rectal GST activity, GSTM2 and GSTP1 most pronounced. GSTP1 was the most abundant rectal GST enzyme measured, GSTA the least abundant. Interindividual variation in GST M1, M2, T1, A and P1 isoenzyme levels was about 10, 15, 7, 17 and 13-fold, respectively. Total GST activities in leukocytes and lymphocytes were higher with higher total rectal GST activity, though not significantly in lymphocytes. The interindividual variation in the GST activity measures was ~4-fold. The correlations between rectal and white blood cell GST activities were 0.41 between rectal and leukocyte, 0.35 between rectal and lymphocyte, and 0.41 between leukocyte and lymphocyte GST activity, P < 0.001 for all coefficients.


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

  		Table II GST genotype and phenotype characteristics of the total study population and by rectal GST activity

 
Rectal GST isoenzyme levels in relation to genetic variation
GST polymorphisms were associated with the levels of their isoenzymes (data not in table); presence or absence of the GSTM1 gene resulted in GSTM1 levels of 2150 ± 1148 and 0 ng/mg protein and for the GSTT1 gene corresponding GSTT1 levels were 4646 ± 2466 and 0 ng/mg protein. GSTA level differed significantly between GSTA1 g.–69C->T genotypes (Figure 1); The CC genotypes had the highest (762 ± 413 ng/mg, median value 677), the CT genotypes an intermediate (378 ± 264 ng/mg, median value 368) and the TT genotypes no detectable GSTA level (P < 0.001). The GSTP1 level appeared to be higher among individuals with a GSTP1 313 variant G-allele: the 313GA and 313GG genotypes had 772 (P = 0.17) and 1440 (P = 0.088) ng/mg protein higher GSTP1 levels compared with the AA genotype, respectively. The GSTP1 217GA genotype also showed a borderline significantly higher GSTP1 level (1026 ng/mg, P = 0.072) compared with the 217 GG genotype. There was no association between GSTP1 level and the GSTP1 227 or 272 variants.


Figure 1
View larger version (13K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1 Rectal GSTA level by GSTA1 g.–69C->T genotype. Boxes and bars represent the interquartile range and the median value, respectively; Whiskers represent distance to smallest and largest sample value, excluding outliers; open circles represent values of 1.5–3 box lengths.

 
Rectal GST isoenzyme levels in relation to fruit and vegetable consumption
Citrus fruit and juice consumption combined was positively associated with GSTM1 level (Table III). Total vegetable consumption and consumption of Apiaceae were inversely associated with GSTM2 level. Consumption of Alliaceae was associated with higher GSTT1 level. There was no significant association between fruit or vegetable consumption and GSTA level. However, when the GSTA1 g.–69C->T genotype was added to the GSTA-fruit and vegetable models, GSTA level was significantly higher among consumers of Solanaceae: +152 ng/mg protein (P = 0.024), and Cucurbitaceae: +152 ng/mg protein (P = 0.035), compared with non-consumers (data not shown in table). Consumption of Alliaceae and Cucurbitaceae was positively associated with GSTP1 level (Table III). This higher GSTP1 level was most pronounced for the GSTP1 272CC genotype: +3758 ng/mg protein (P = 0.0039) with Alliaceae consumption and +2340 ng/mg protein (P = 0.074) with Cucurbitaceae consumption; similar results were observed for the 227GA genotype, but this did not reach statistical significance. There was no such difference when stratifying for the GSTP1 g.217G->A or c.313A->G genotypes.


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

  		Table III Rectal GST isoenzyme levels*(ng/mg protein) by recent fruit and vegetable consumption (yes or no)

 
Rectal and white blood cell total GST activity in relation to genetic variation
Rectal GST activity
The polymorphisms in GSTM1, GSTT1 and GSTA1 did not affect rectal GST activity significantly (Table IV). Rectal GST activity did differ significantly between GSTP1 c.313A->G genotypes; the 313AG and GG genotypes had 36 and 67 nmol/min/mg protein lower rectal GST activities than the AA genotype, respectively. The GSTP1 272GG genotypes showed significantly higher rectal GST activity, likely because all 272GG individuals were GSTP1 313AA. The GSTP1 g.217G->A and g.227G->A polymorphisms did not affect rectal GST activity significantly.


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

  		Table IV Rectal and white blood cell GST activity* (in nmol DNP-SG/min/mg protein) by GST genotype

 
White blood cell GST activity
In leukocytes, the GSTM1 deletion polymorphism was associated with a 27 nmol/min/mg protein lower GST activity (Table IV). As in rectum, the GSTP1 313G-variant showed lower activity in leukocytes, but this was not statistically significant. The GSTP1 272CG genotype was significantly inversely associated with leukocyte activity, which was in contrast to its effect on rectal GST activity. In lymphocytes, the GSTP1 c.313A->G polymorphism showed a similar trend as in rectum: the 313AG and 313GG genotypes had 17 and 35 nmol/min/mg protein lower GST activity than the 313AA genotype, respectively. The GSTP1 217GA genotypes had significantly lower and the GSTP1 227GA genotypes had significantly higher lymphocyte GST activities than their homozygous wild-types (GSTP1 217GG and 227GG, respectively).

Rectal and white blood cell total GST activity in relation to fruit and vegetable consumption
Rectal GST activity
There was no statistically significant association between fruit and vegetable consumption on the two days prior to sigmoidoscopy and rectal GST activity (Table V). However, when stratifying for the GSTM1 deletion polymorphism, there was a significant difference in rectal GST activity between those who did and did not report fruit consumption of any kind among GSTM1 null individuals (+43 nmol/min/mg protein, P = 0.019, data not in table), which was not observed among GSTM1 positive individuals (–1.5, P = 0.95). Also, when adding the GSTP1 c.313A->G polymorphism to the rectal GST activity and fruit and vegetable models, there was a significant difference between consumers and non-consumers of Alliaceae with respect to total rectal GST activity (+24 nmol/min/mg protein, P = 0.031).


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

  		Table V Rectal and white blood cell total GST activity* (in nmol DNP-SG/min/mg protein) by recent fruit and vegetable consumption (yes or no)

 
White blood cell GST activity
Consumption of citrus juice was slightly inversely associated with leukocyte GST activity, and consumption of Alliaceae and Cucurbitaceae positively (Table V). Lymphocyte GST activity appeared to be slightly positively associated with citrus fruit consumption.

Similar to observations in rectal biopsies, lymphocyte GST activity among GSTM1 null individuals who reported fruit consumption was higher compared with those who did not (+18, nmol/min/mg protein P = 0.068), and this was not observed among GSTM1 positive individuals (–16 nmol/min/mg protein, P = 0.28); In leukocytes, GST activity among GSTM1 null and positive individuals with fruit consumption was +17 (P = 0.13) and –5.1 (P = 0.76), respectively. The difference in rectal GST activity, observed with Alliaceae consumption when the GSTP1 c.313A->G polymorphism was added to the statistical model, could also be seen in lymphocytes: +13 nmol/min/mg protein (P = 0.045) and became slightly more pronounced in leukocytes: +17 nmol/min/mg protein (P = 0.076).


   Discussion
Top
 Abstract
 Introduction
 Methods
 Statistical analyses
 Results
 Discussion
References
 
The results from this human observational study confirm that both genetic variation and recent consumption of fruits and vegetables influence the GST enzyme system. The GSTM1, GSTT1 and GSTA1 g.–69C->T polymorphisms clearly affected their respective isoenzyme levels. Consumption of specific fruits and vegetables was also associated with differences in specific rectal GST isoenzyme levels; GSTM1 level was higher among consumers of citrus, GSTM2 level was lower among total vegetable and Apiaceae consumers, GSTT1 level was higher among consumers of Alliaceae, and GSTP1 level was higher among consumers of Alliaceae and Cucurbitaceae. The consequences for the rectal GST activity are not obvious; activity was most affected by the GSTP1 c.313A->G polymorphism, whereas fruit and vegetable consumption in itself did not seem influential. GST activity in white blood cells did appear to be affected by fruit and vegetable consumption to some extent, and next to GSTP1 polymorphisms also by the GSTM1 deletion genotype.

We will first discuss the polymorphisms, and then fruit and vegetable consumption, in relation to GST phenotype.

In our population, the GSTA1 g. –69C->T polymorphism affected rectal GSTA enzyme level, which is consistent with knowledge about the functionality of the polymorphism (30). GSTA level as measured by us, consists of GSTA1 and GSTA2 protein. As we observed GSTA levels of zero (among the GSTA1 –69TT-genotypes), the expression of the GSTA2 subunit was apparently below detection in these subjects. Possibly, the reversed co-expression between GSTA1 and GSTA2, as reported for liver (25), does not operate in the colorectal area. The GSTP1 c.313A->G polymorphism did not have a clear relationship with its enzyme level, but clearly affected rectal total GST activity. It leads to an amino acid change, with consequences for substrate binding and thermal stability, i.e. lower GSTP1 activity (19,31). The abundance of GSTP1 enzyme in the rectum, in combination with the lower activity of the variant GSTP1 protein, explains the great influence of the GSTP1 c.313A->G polymorphism on total GST activity in the rectum. The effect of the other GSTP1 polymorphisms may be related to the GSTP1 c.313A->G polymorphism, as the 313A-allele occurred most frequently with the 272G-allele, and the 313G-allele with the 217A-allele. However, a larger population should be genotyped for more definitive answers concerning their linkage.

Consumption of different types of fruits and vegetables appeared to influence rectal individual GST isoenzyme levels, while it did not seem to influence rectal total GST activity. In the literature, several GST protein phenotypes and fruit and vegetable subtypes have been investigated, in different tissues. Most studies have focussed on Brassicaceae (11,13,14,16,17,32–34): higher levels of GSTA (11,13–15) and GSTP1 (11), and higher total GST activity (17) and GSTM activity (15) have been reported. We observed higher GSTM1 levels in rectum among Brassicaceae consumers (~30%, not significant), but a significant positive effect of Brassicaceae consumption on GST isoenzyme levels or GST activity was not seen. Lampe et al. (15) also studied Alliaceae and Apiaceae vegetables; After Alliaceae supplementation, they observed higher GSTM activity. In our study, individuals who consumed Alliaceae had significantly higher GSTT1 and GSTP1 level. The latter is in contrast with Wark et al. (35) who, also from observational data, noted downregulation of rectal GSTP1 with higher Alliaceae consumption. After Apiaceae supplementation, Lampe et al. (15) observed lower serum GSTA levels. They have also reported decreased CYP1A2 activity with Apiaceae consumption (36). In our study, rectal GSTM2 appeared to be downregulated by Apiaceae. There are indications from cell studies that phytochemicals from Apiaceae and Brassicaceae downregulate certain enzymes (37,38). Also, we observed lower GSTP1 levels with total vegetable consumption, to a level of ~85% of those who did not consume any vegetables on one of the two days before endoscopy; This effect appeared to be due to Apiaceae and Brassicaceae vegetables. A similar decrease in GSTP1 level was noted in lymphocytes by Persson et al. after a mixed vegetable diet, containing Apiaceae and Brassicaceae (39). Overall, inconsistencies concerning the effect of fruit and vegetable consumption on GST phenotype remain. Differences in study results may be related to study design (intervention versus observational), time frame and method of food consumption measurement (food frequency questionnaire versus food record), coding of consumption (yes/no versus high/low or continuous), type of tissue that was sampled, or laboratory methods. In general, studies are small, and only few have taken genetic variation into account.

Our a priori calculation of power and sample size required for comparison of phenotype by genotype groups was based on data by Coles et al. (26) and Siegel et al. (40). We estimated that a total group of 100 subjects and differing genotype frequencies would yield a power of >75% to detect relevant protein differences between at least two of the three genotype groups. To uncover the interplay between several different exposures and genes, however, larger studies are needed. As sampling of rectal material is invasive and impractical, blood GST activity could be a useful surrogate measure. We therefore compared white blood cell and rectal GST activity, but the correlation between the two was low. Despite earlier reports (33,41), blood may not be a good surrogate tissue. Possibly, taking genotype into account improves its value.

The study population consisted of individuals undergoing sigmoidoscopy for diagnostic reasons. Possible disadvantages are presence of colorectal abnormalities, bowel preparation, and lower food consumption. Colorectal abnormalities (i.e. diverticula, hemorrhoids and adenomas) were minor however, and though their distribution over the GST activity median-based groups was different, they did not influence statistical models. Bowel preparation for endoscopy may have affected phenotype, by itself or by lowering absorption and delivery of (non)nutrients to the rectal crypts. If it did, subjects at least were affected similarly, as preparations were similar. Subjects were not restricted in their food consumption, but energy intake was relatively low (7930 ± 2136 kJ/day), as was vegetable intake, which may have resulted in relatively low amounts of inducer. The purpose of our food record was to estimate actual and not habitual consumption, because induction of GST enzymes occurs rapidly, in a matter of hours to days after consumption of the inducer (17,34,42).

Assessment of polymorphisms was reproducible, and distributions of the GSTM1, GSTT1, GSTA1 g.–69C->T and GSTP1 c.313A->G polymorphisms were similar to previous findings (43,44). The distribution of the other GSTP1 polymorphism are not easily compared with that of the multi-ethnic population on the NCBI SNP website (www.ncbi.nlm.nih.gov/SNP, accessed September 19, 2006).

Interindividual GST expression was highly variable in our study, which is in line with other reports (3). Between studies, reported GST isoenzyme levels (7,35,45) also vary, as do GST activities (11,12,33,35,41,46,47). This may be related to the method of total protein measurement, by which GST measures are normalized; these are known to recover different amounts of protein (48,49). It may also be related to differences in population characteristics, such as genetic differences, age, sex, smoking or dietary habits.

A general point of attention is the substrate used to measure GST activity, which was CDNB, as in most studies. The different GST enzymes contribute differently to its metabolism: the order of highest to lowest specific activity is GSTM2, GSTM1, GSTP1, GSTA, whereas GSTT1 has no activity for CDNB (1). Total GST activity is an important measure for generic carcinogenic substrates. But, if not all GST isoenzymes or only those with lower specific activity are affected by a polymorphism or fruit and vegetable consumption, an effect may be difficult to visualize. It may, however, still be relevant, because numerous substrates are specifically metabolized by individual GST isoenzymes (e.g. N-acetoxy PhIP by GSTA1) (2,3).

The results of the present study contribute to a better understanding of both possibilities and limitations for food-mediated health effects. When investigating GST phenotype, genotype needs to be taken into account. White blood cell total GST activity may be unsuitable as a surrogate for rectal total GST activity. Thus, large and invasive studies are needed to evaluate the true effects of fruits and vegetables on rectal GSTs in human populations.


   Acknowledgments
 
The authors wish to thank Petra Vissink and Els Siebelink for the dietary assessment; Lucy Okma, Annie van Schaik and René te Morsche for lab-related support; Jan Harryvan for GSTM1/T1 genotyping; The endoscopy staff of the Radboud Univeristy Nijmegen Medical Center and the Canisius-Wilhelmina Hospital Nijmegen, in particular Drs Pieter Friederich and Adriaan Tan, for their support in recruitment; and all study subjects for their kind participation. This work was supported by the Netherlands Organization for Health Research and Development (ZonMW), grant number: 21000054, and the Dutch Digestive Diseases Foundation (MLDS), grant number WS 00-31.

Conflict of Interest Statement: None declared.


   References
Top
 Abstract
 Introduction
 Methods
 Statistical analyses
 Results
 Discussion
 References
 

  1. Hayes J.D. and Pulford D.J. (1995) The glutathione S-transferase supergene family: regulation of GST and the contribution of the isoenzymes to cancer chemoprotection and drug resistance. Crit. Rev. Biochem. Mol. Biol. 30:445–600.[ISI][Medline]
  2. Hayes J.D., Flanagan J.U., Jowsey I.R. (2005) Glutathione transferases. Annu. Rev. Pharmacol. Toxicol. 45:51–88.
  3. Coles B.F. and Kadlubar F.F. (2003) Detoxification of electrophilic compounds by glutathione S-transferase catalysis: determinants of individual response to chemical carcinogens and chemotherapeutic drugs? Biofactors 17:115–130.[ISI][Medline]
  4. Ferlay J., Bray F., Pisani P., Parkin D.M. (2004) GLOBOCAN 2002: cancer incidence, mortality and prevalence worldwide. IARC CancerBase No. 5. version 2.0(IARCPress, Lyon).
  5. Deschner E.E. (1987) Cell turnover and colon tumor development. Prev. Med. 16:580–585.[CrossRef][ISI][Medline]
  6. Peters W.H.M., Kock L., Nagengast F.M., Kremers P.G. (1991) Biotransformation enzymes in human intestine: critical low levels in the colon? Gut 32:408–412.[Abstract/Free Full Text]
  7. Coles B.F., Chen G., Kadlubar F.F., Radominska-Pandya A. (2002) Interindividual variation and organ-specific patterns of glutathione S-transferase alpha, mu, and pi expression in gastrointestinal tract mucosa of normal individuals. Arch. Biochem. Biophys. 403:270–276.[CrossRef][ISI][Medline]
  8. Grubben M.J.A.L., Nagengast F.M., Katan M.B., Peters W.H.M. (2001) The glutathione biotransformation system and colorectal cancer risk in humans. Scand. J. Gastroenterol. Suppl. 234:68–76.
  9. Riboli E. and Norat T. (2003) Epidemiologic evidence of the protective effect of fruit and vegetables on cancer risk. Am. J. Clin. Nutr. 78:559S–569S.[Abstract/Free Full Text]
  10. Park E.J. and Pezzuto J.M. (2002) Botanicals in cancer chemoprevention. Cancer Metastasis Rev. 21:231–255.[CrossRef][ISI][Medline]
  11. Nijhoff W.A., Grubben M.J.A.L., Nagengast F.M., Jansen J.B., Verhagen H., van Poppel G., Peters W.H.M. (1995) Effects of consumption of Brussels sprouts on intestinal and lymphocytic glutathione S-transferases in humans. Carcinogenesis 16:2125–2128.[Abstract/Free Full Text]
  12. O'Dwyer P.J., Szarka C.E., Yao K.S., et al. (1996) Modulation of gene expression in subjects at risk for colorectal cancer by the chemopreventive dithiolethione oltipraz. J. Clin. Invest. 98:1210–1217.[ISI][Medline]
  13. Bogaards J.J., Verhagen H., Willems M.I., van Poppel G., Van Bladeren P.J. (1994) Consumption of Brussels sprouts results in elevated alpha-class glutathione S-transferase levels in human blood plasma. Carcinogenesis 15:1073–1075.[Abstract/Free Full Text]
  14. Nijhoff W.A., Mulder T.P.J., Verhagen H., van Poppel G., Peters W.H.M. (1995) Effects of consumption of brussels sprouts on plasma and urinary glutathione S-transferase class-alpha and -pi in humans. Carcinogenesis 16:955–957.[Abstract/Free Full Text]
  15. Lampe J.W., Chen C., Li S., Prunty J., Grate M.T., Meehan D.E., Barale K.V., Dightman D.A., Feng Z., Potter J.D. (2000) Modulation of human glutathione S-transferases by botanically defined vegetable diets. Cancer Epidemiol. Biomarkers Prev. 9:787–793.[Abstract/Free Full Text]
  16. Steinkellner H. (2000) In Hietsch G., Sreerama L., Haidinger G., Gsur A., Kundi M., Knasmuller S (Eds.). Induction of glutathione-S-transferases in humans by vegetable diets. Royal Society of Chemistry, special publication vol. 255:193–198.
  17. Sreerama L., Hedge M.W., Sladek N.E. (1995) Identification of a class 3 aldehyde dehydrogenase in human saliva and increased levels of this enzyme, glutathione S-transferases, and DT-diaphorase in the saliva of subjects who continually ingest large quantities of coffee or broccoli. Clin. Cancer Res. 1:1153–1163.[Abstract]
  18. Chen C. and Kong A.N. (2004) Dietary chemopreventive compounds and ARE/EpRE signaling. Free Radic. Biol. Med. 36:1505–1516.[CrossRef][ISI][Medline]
  19. Ali-Osman F., Akande O., Antoun G., Mao J.X., Buolamwini J. (1997) Molecular cloning, characterization, and expression in Escherichia coli of full-length cDNAs of three human glutathione S-transferase Pi gene variants. Evidence for differential catalytic activity of the encoded proteins. J. Biol. Chem. 272:10004–10012.[Abstract/Free Full Text]
  20. Guy C.A., Hoogendoorn B., Smith S.K., Coleman S., O'Donovan M.C., Buckland P.R. (2004) Promoter polymorphisms in glutathione-S-transferase genes affect transcription. Pharmacogenetics 14:45–51.[CrossRef][ISI][Medline]
  21. Finley J.W. (2003) The antioxidant responsive element (ARE) may explain the protective effects of cruciferous vegetables on cancer. Nutr. Rev. 61:250–254.[CrossRef][ISI][Medline]
  22. Donders-Engelen M., van der Heijden L.J.M., Hulshof K.F.A.M. (2003) Measures, weights, and Code Numbers 2003. Zeist, Wageningen(TNO-Nutrition and Food Research and Wageningen University, The Netherlands) (In Dutch).
  23. NEVO S. (2001) Nederlands Voedingsstoffenbestand [Dutch Food Composition Tables]. Voedingscentrum, The Hague.
  24. Scholte B. (2004) VBS Food Calculation System. BaS Nutrition Software.
  25. Arand M., Muhlbauer R., Hengstler J., Jager E., Fuchs J., Winkler L., Oesch F. (1996) A multiplex polymerase chain reaction protocol for the simultaneous analysis of the glutathione S-transferase GSTM1 and GSTT1 polymorphisms. Anal. Biochem. 236:184–186.[CrossRef][ISI][Medline]
  26. Coles B.F., Morel F., Rauch C., Huber W.W., Yang M., Teitel C.H., Green B., Lang N.P., Kadlubar F.F. (2001) Effect of polymorphism in the human glutathione S-transferase A1 promoter on hepatic GSTA1 and GSTA2 expression. Pharmacogenetics 11:663–669.[CrossRef][ISI][Medline]
  27. Harries L.W., Stubbins M.J., Forman D., Howard G.C., Wolf C.R. (1997) Identification of genetic polymorphisms at the glutathione S-transferase Pi locus and association with susceptibility to bladder, testicular and prostate cancer. Carcinogenesis 18:641–644.[Abstract/Free Full Text]
  28. Habig W.H. and Jakoby W.B. (1981) Assays for differentiation of glutathione S-transferases. Methods. Enzymol. 77:398–405.[Medline]
  29. van Zanden J.J., Ben Hamman O., van Iersel M.L., Boeren S., Cnubben N.H., Lo Bello M., Vervoort J., van Bladeren P.J., Rietjens I.M. (2003) Inhibition of human glutathione S-transferase P1-1 by the flavonoid quercetin. Chem. Biol. Interact. 145:139–148.[CrossRef][ISI][Medline]
  30. Morel F., Rauch C., Coles B., Le Ferrec E., Guillouzo A. (2002) The human glutathione transferase alpha locus: genomic organization of the gene cluster and functional characterization of the genetic polymorphism in the hGSTA1 promoter. Pharmacogenetics 12:277–286.[CrossRef][ISI][Medline]
  31. Johansson A.S., Stenberg G., Widersten M., Mannervik B. (1998) Structure-activity relationships and thermal stability of human glutathione transferase P1-1 governed by the H-site residue 105. J. Mol. Biol. 278:687–698.[CrossRef][ISI][Medline]
  32. Lampe J.W. and Peterson S. (2002) Brassica, biotransformation and cancer risk: genetic polymorphisms alter the preventive effects of cruciferous vegetables. J. Nutr. 132:2991–2994.[Abstract/Free Full Text]
  33. Clapper M.L., Szarka C.E., Pfeiffer G.R., Graham T.A., Balshem A.M., Litwin S., Goosenberg E.B., Frucht H., Engstrom P.F. (1997) Preclinical and clinical evaluation of broccoli supplements as inducers of glutathione S-transferase activity. Clin. Cancer Res. 3:25–30.[Abstract]
  34. Pantuck E.J., Pantuck C.B., Anderson K.E., Wattenberg L.W., Conney A.H., Kappas A. (1984) Effect of brussels sprouts and cabbage on drug conjugation. Clin. Pharmacol. Ther. 35:161–169.[ISI][Medline]
  35. Wark P.A., Grubben M.J.A.L., Peters W.H.M., Nagengast F.M., Kampman E., Kok F.J., van't Veer P. (2004) Habitual consumption of fruits and vegetables: associations with human rectal glutathione S-transferase. Carcinogenesis 25:2135–2142.[Abstract/Free Full Text]
  36. Lampe J.W., King I.B., Li S., Grate M.T., Barale K.V., Chen C., Feng Z., Potter J.D. (2000) Brassica vegetables increase and apiaceous vegetables decrease cytochrome P450 1A2 activity in humans: changes in caffeine metabolite ratios in response to controlled vegetable diets. Carcinogenesis 21:1157–1162.[Abstract/Free Full Text]
  37. Peterson S., Lampe J.W., Bammler T.K., Gross-Steinmeyer K., Eaton D.L. (2006) Apiaceous vegetable constituents inhibit human cytochrome P-450 1A2 (hCYP1A2) activity and hCYP1A2-mediated mutagenicity of aflatoxin B1. Food Chem. Toxicol. 44:1474–1484.[CrossRef][ISI][Medline]
  38. Cauchi S., Han W., Kumar S.V., Spivack S.D. (2006) Haplotype-environment interactions that regulate the human glutathione S-transferase P1 promoter. Cancer Res. 66:6439–6448.[Abstract/Free Full Text]
  39. Persson I., He L., Fang C., Normen L., Rylander R. (2000) Influence of vegetables on the expression of GSTP1 in humans—a pilot intervention study (Sweden). Cancer Causes Control 11:359–361.[CrossRef][ISI][Medline]
  40. Siegel D., McGuinness S.M., Winski S.L., Ross D. (1999) Genotype–phenotype relationships in studies of a polymorphism in NAD(P)H: quinone oxidoreductase 1. Pharmacogenetics 9:113–121.[ISI][Medline]
  41. Szarka C.E., Pfeiffer G.R., Hum S.T., Everley L.C., Balshem A.M., Moore D.F., Litwin S., Goosenberg E.B., Frucht H., Engstrom P.F. (1995) Glutathione S-transferase activity and glutathione S-transferase mu expression in subjects with risk for colorectal cancer. Cancer Res. 55:2789–2793.[Abstract/Free Full Text]
  42. Hoensch H.P. and Hartmann F. (1981) The intestinal enzymatic biotransformation system: potential role in protection from colon cancer. Hepatogastroenterology 28:221–228.[Medline]
  43. Tijhuis M.J., Wark P.A., Aarts J.M., Visker M.H., Nagengast F.M., Kok F.J., Kampman E. (2005) GSTP1 and GSTA1 polymorphisms interact with cruciferous vegetable intake in colorectal adenoma risk. Cancer Epidemiol. Biomarkers Prev. 14:2943–2951.[Abstract/Free Full Text]
  44. Garte S., Gaspari L., Alexandrie A.K., et al. (2001) Metabolic gene polymorphism frequencies in control populations. Cancer Epidemiol. Biomarkers Prev. 10:1239–1248.[Medline]
  45. Hengstler J.G., Bottger T., Tanner B., Dietrich B., Henrich M., Knapstein P.G., Junginger T., Oesch F. (1998) Resistance factors in colon cancer tissue and the adjacent normal colon tissue: glutathione S-transferases alpha and pi, glutathione and aldehyde dehydrogenase. Cancer Lett. 128:105–112.[CrossRef][ISI][Medline]
  46. Clapper M.L., Hoffman S.J., Carp N., Watts P., Seestaller L.M., Weese J.L., Tew K.D. (1991) Contribution of patient history to the glutathione S-transferase activity of human lung, breast and colon tissue. Carcinogenesis 12:1957–1961.[Abstract/Free Full Text]
  47. Heckbert S.R., Weiss N.S., Hornung S.K., Eaton D.L., Motulsky A.G. (1992) Glutathione S-transferase and epoxide hydrolase activity in human leukocytes in relation to risk of lung cancer and other smoking-related cancers. J. Natl Cancer Inst. 84:414–422.[Abstract/Free Full Text]
  48. Smith P.K., Krohn R.I., Hermanson G.T., Mallia A.K., Gartner F.H., Provenzano M.D., Fujimoto E.K., Goeke N.M., Olson B.J., Klenk D.C. (1985) Measurement of protein using bicinchoninic acid. Anal. Biochem. 150:76–85.[CrossRef][ISI][Medline]
  49. Lleu P.L. and Rebel G. (1991) Interference of Good's buffers and other biological buffers with protein determination. Anal. Biochem. 192:215–218.[CrossRef][ISI][Medline]
Received May 2, 2006; revised September 23, 2006; accepted October 16, 2006.


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


This article has been cited by other articles:


Home page
 CarcinogenesisHome page
A. Rossini, D.C.M. Rapozo, S.C. Soares Lima, D.P. Guimaraes, M.A. Ferreira, R. Teixeira, C.D.P. Kruel, S.G.S. Barros, N.A. Andreollo, R. Acatauassu, et al.
Polymorphisms of GSTP1 and GSTT1, but not of CYP2A6, CYP2E1 or GSTM1, modify the risk for esophageal cancer in a western population
Carcinogenesis, December 1, 2007; 28(12): 2537 - 2542.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
28/4/848    most recent
bgl204v1
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 arrow Search for citing articles in:
ISI Web of Science (3)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Tijhuis, M. J.
Right arrow Articles by Kampman, E.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Tijhuis, M. J.
Right arrow Articles by Kampman, E.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?