Download - Alteration of glutathione S-transferase levels in Barrett's metaplasia compared to normal oesophageal epithelium

Alteration of glutathione S-transferase levels in Barrett’smetaplasia compared to normal oesophageal epitheliumSarah C. Cobbe, Gillian C. Scobie, Elizabeth Pohler, John D. Hayes,Neil M. Kernohan and John F. Dillon

Objective Oesophageal cancer associated with the

premalignant condition Barrett’s oesophagus has

increased in incidence over the last few years. Phase II

detoxifying enzymes, including glutathione S-transferases

(GSTs) protect the mucosa from carcinogens, which can

cause oxidative damage to cells. Therefore, a reduction in

these anti-oxidant enzymes can increase the risk of

carcinogenesis. The aim of this study was to compare the

extent of GST expression in normal oesophageal tissue,

Barrett’s oesophagus and oesophageal adenocarcinoma.

Design Antibodies raised against GST alpha, GST mu,

GST pi and microsomal GST were used to identify

expression of these proteins in tissue sections.

Method Paraffin-embedded sections were stained using

standard immunohistochemical techniques to

demonstrate the pattern of expression of GST proteins in

biopsy specimens. Twelve sections of Barrett’s metaplasia

and an equal number of specimens from normal

oesophageal tissue were examined, together with sections

from adenocarcinoma and normal gastric mucosa.

Results Expression of the GST enzymes appeared to be

reduced in Barrett’s tissue compared to normal

oesophageal tissue. Nuclear staining featured in some of

the normal tissue sections, but not in Barrett’s tissue.

Conclusion The reduction in GST expression suggested in

Barrett’s tissue is an interesting finding, as it is possible

that reduced expression of these detoxifying enzymes may

contribute to the risk of development of adenocarcinoma

in Barrett’s mucosa. Eur J Gastroenterol Hepatol 15:41–47

& 2003 Lippincott Williams & Wilkins

European Journal of Gastroenterology & Hepatology 2003, 15:41–47

Keywords: oesophagus, glutathione S-transferase, cancer, Barrett’soesophagus

Department of Molecular and Cellular Pathology, Biomedical Research Centre,University of Dundee, Dundee, UK.

Correspondence and reprint requests to Dr John Dillon, Ward 5 & 6, NinewellsHospital and Medical School, Dundee DD1 9SY, UK.Tel: +44 (0)1382 660 111; fax: +44 (0)1382 425 504;e-mail: [email protected]

Received 6 February 2002 Revised 2 August 2002Accepted 19 August 2002

IntroductionThroughout the world, the incidence of oesophageal

carcinoma has dramatically escalated over the past few

decades. In Scotland, the incidence of carcinoma has

increased by 2% per year in women and by 3% per year

in men over a 30-year period [1,2]. Epidemiological

data from the USA, Scandinavia, France, Switzerland,

Australia and New Zealand demonstrate that this is an

international trend [3–11].

The increase in incidence of oesophageal cancer has

coincided with an alteration in the most common

histological type of tumour found and the usual location

of these tumours within the oesophagus [12,13]. The

prevalence of oesophageal adenocarcinoma has now

surpassed that of squamous carcinoma in Western

Europe and the USA. In particular, cases of adenocarci-

noma associated with the premalignant condition

Barrett’s oesophagus have increased. The factors re-

sponsible for this alteration in geographical distribution

and increase in incidence are as yet unidentified.

Genetic susceptibility plays a role; however, this factor

alone cannot sufficiently explain the extent of the

increase in oesophageal tumours [14]. The relatively

short time period during which this increase has oc-

curred suggests the aetiology is likely to be environ-

mental.

Cells are protected against exposure to harmful environ-

mental xenobiotics by phase II drug detoxification

reactions. Such enzyme reactions increase the polarity

of xenobiotics, reducing their toxicity and promoting

subsequent excretion in the urine or bile. The suscept-

ibility of cells to oxidative damage by toxic chemicals is

increased if these phase II enzymes are not expressed

[15], and damage to key genetic material is a vital step

in the carcinogenic process.

The glutathione S-transferase (GST) isoenzymes are

phase II detoxifying enzymes, which were initially

described in 1961. They are dimeric proteins that

catalyse the conjugation of electrophilic chemicals with

Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Original article 41

0954-691X & 2003 Lippincott Williams & Wilkins

reduced glutathione [16,17]. Since their discovery, sub-

stantial interest in the role of these enzymes in cancer

prevention has arisen. Two GST superfamilies exist

that comprise either cytosolic or membrane-bound

proteins. Evidence suggests that cytosolic GST in-

creases bodily resistance to carcinogens [18], products

of oxidative stress, environmental pollutants and certain

anti-cancer drugs [19]. In humans, these cytosolic

isoenzymes have been allocated into eight classes: GST

alpha (GSTA), mu (GSTM), pi (GSTP), sigma, theta

(GSTT), kappa, zet and omega [20–23].

Low levels of GST correlate with increased sensitivity

of cells to damage by noxious substances, including

carcinogens [24]. This is demonstrated by the fact that

people lacking the GSTM1 gene (50% of the popu-

lation), and those lacking GSTT1 (16% of the popu-

lation), have an increased susceptibility to tumours of

the bladder, colon, skin and lung, especially if they are

heavy smokers [25,26]. Certain tumour types character-

istically over- or under-express GST [27–31]. GSTP1 is

notably increased in tumours of the colon, lung and

stomach and this rise often occurs during the pre-

neoplastic stage [30]. Therefore, GST levels may be

considered as surrogate markers for certain pre-neo-

plastic lesions; however, due to the ubiquitous distribu-

tion and variation in levels of GST, it is only useful in

clear cut cases such as the one described here.

In this study, we investigated the pattern of GST

expression in normal oesophageal tissue, Barrett’s tissue

and adenocarcinoma of the oesophagus to detect

whether these enzymes may be involved in the patho-

genic process. We selected a panel of four GST anti-

bodies: three for the cytosolic classes alpha, mu and pi,

and the fourth for a microsomal transferase. These

particular GSTs were selected because their expression

in other tumour types has already been investigated, as

mentioned above. Therefore, we felt it would be of

value to establish their behaviour within the oesopha-

gus. Tissue sections were stained using standard im-

munohistochemical techniques.

Materials and methodsMaterials

Chemicals, unless otherwise stated, were obtained from

AnalaR BDH chemicals (Poole, UK) or National Diag-

nostics (Atlanta, Georgia, USA). The primary antibodies

used were raised against GSTA1/2, GSTP1, GSTM1

and microsomal GST (MGST-1) [32]. Paraffin blocks

containing tissue samples were selected from the

pathology archive, Ninewells Hospital and Medical

School, Dundee. Section types used were from normal

oesophagus (n ¼ 8), Barrett’s oesophagus (n ¼ 11), oeso-

phageal adenocarcinoma (n ¼ 8), normal gastric tissue

(n ¼ 8), gastric carcinomas (n ¼ 2) and normal duode-

num (n ¼ 1).

Immunohistochemistry

Chosen tissue sections embedded in paraffin wax were

sliced at a thickness of 4 �m using a microtome (Leica

RM 2135, Leica, Nussloch, Germany). The sections

were resolubilized to allow the antibody access to the

antigens, and endogenous peroxidases were blocked for

10 min. Heat and citrate buffer were used to reduce

molecular bonding within the tissue.

The primary antibodies used were rabbit polyclonals

raised against the different GST subclasses and diluted

in phosphate-buffered saline (PBS) (Table 1). Normal

goat serum was applied prior to the antibody to reduce

non-specific binding.

The secondary antibody was anti-rabbit immuno-

globulin G (IgG) (Biogenex, San Ramon, California,

USA), which recognizes the rabbit primary polyclonal.

The secondary antibody system involved the formation

of a complex between a biotin molecule conjugated to

the secondary antibody and streptavidin, both diluted

1 : 25 in PBS.

To stain the sections, 3,39diaminobenzidine (DAB)

tetra hydrochloride (Sigma, Poole, Dorset, UK) (0.5 �g/

ml in PBS) was applied with 30% v/v hydrogen

peroxide (H2O2), followed by copper sulphate and

haematoxylin. The slides were then dehydrated,

mounted with distrene 80 dibutyl phthalate xylene and

examined microscopically.

Western blot analysis was used to demonstrate that the

GSTs were expressed in gastrointestinal tissue, and to

confirm the specificity of the immune cross-reactivity.

Rat liver and human stomach were probed by each

antibody, and bands correlating with the molecular

weight of GST were identified (Fig. 1). A second

isoform was identified when probing for GSTA1/2, and

degradation products were found with GSTM1 and

MGST-1. However, this should not have affected the

immunohistochemical techniques used in this study

because they were used only to define the cellular

localization of proteins. The disruptive nature of the

western blotting technique, during which cell lysis

occurs, can often result in degradation products.

ResultsNormal oesophagus

Positive superficial staining and weak basal staining was


Table 1 Dilution of glutathione S-transferase antibodies

Antibody Dilution

GSTA1/2 1:500GSTM1 1:100GSTP1 1:50MGST-1 1:100

42 European Journal of Gastroenterology & Hepatology 2003, Vol 15 No 1

observed with all the GST antibodies in the normal

oesophagus (Figs 2,3). The pattern of staining was

analogous for GSTP1 and GSTA1, suggesting that they

operate in similar areas of the oesophageal mucosa.

Nuclear staining was detected in a few of the normal

oesophageal sections for these two enzymes. However,

no particular pattern was identified in the distribution

and number of cells, with nuclear staining appearing to

be random (Fig. 2).

The GSTM1 gene is only present in 50% of the

general population [25]. Among the normal oesophageal

sections stained (n ¼ 7), one section was negative for

GSTM1 staining and one was equivocal. The other

normal oesophageal sections stained with GSTM1

showed weakly positive cytosolic staining which was

more superficially pronounced. Some nuclear staining

was witnessed, though not as consistently as with

GSTA1 and GSTP1.

Intense staining for MGST-1 was observed in the

superficial layers of the mucosa, as with the other

GSTs, in normal oesophageal tissue. No nuclear stain-

ing was identified.


Fig. 1

Western blot demonstrating the presence of the glutathione S-transferases (GSTs) GSTP1, GSTA1, GSTM1 and MGST-1 in human gastric bodyand rat liver control. The normal molecular weight of GST is 29.8 kDa.

Fig. 2

Normal oesophagus stained with GSTA1 (310 objective). Cytoplasmicstaining is observed near the surface of the epithelium. Scatteredpositive nuclei are also present.

Fig. 3

Normal oesophagus stained with GSTM1 (310 objective).Cytoplasmic staining is evident throughout the epithelium, althoughnuclear immunoreactivity is not represented in this field.

GSTs in Barrett’s oesophagus Cobbe et al. 43

Barrett’s oesophagus

In general, staining with all the GSTs in Barrett’s

oesophagus was very weak or negative (Fig. 4). Faint

staining was, however, present in the very superficial

layers of some of the slides. The superficial differen-

tiated cells stained weakly in the cytoplasm, but to a

lesser degree than in normal oesophagus. Staining was

weaker proportional to the depth of the gland layer.

Intestinal metaplasia showed almost negative staining

in the crypts (Fig. 5). There was no nuclear staining

present in any of the Barrett’s tissue sections as

opposed to normal oesophagus.

Staining was decreased or mostly negative for both

GSTA1 and GSTP1, particularly in the Barrett’s intest-

inal metaplasia. This was also observed, though not so

convincingly, in the gastric metaplasia.

Of the 11 Barrett’s sections examined, six appeared to

express GSTM1, four were negative and one section

was equivocal. Staining, as with GSTA1 and GSTP1,

was more pronounced in the basal layers. Interestingly,

in one section, GSTM1 was expressed in the normal

squamous tissue on the slide; however, this was lost on

the Barrett’s tissue (Fig. 6).

The pattern of staining in Barrett’s oesophagus with

MGST-1 did not noticeably differ from that in normal

oesophageal tissue.

Oesophageal adenocarcinoma

With all the GST antibodies, staining of the tumour

cells was found to be heterogeneous, and we could

ascertain no consistent pattern (Fig. 7).

Normal gastric cells

In general, the gastric foveolar cells stained weakly,

especially with GSTP1. The specialized cells were

positive, as was noted with all the antibodies in the

selected panel.

Gastric carcinoma

With the intestinal-type carcinoma, all the GSTs stained

positively (Fig. 8). However, the antral diffuse-type

carcinoma stained noticeably less than the intestinal-


Fig. 5

Barrett’s oesophagus stained with GSTP1 (320 objective). Intestinal-type epithelium with Barrett’s mucosa shows very low levels ofimmunoreactivity in the crypt epithelium, although weak cytoplasmicstaining persists in the gastric foveolar-type epithelium.

Fig. 4

Barrett’s oesophagus stained with GSTP1 (310 objective). Themetaplastic epithelium shows a markedly reduced level of staining withGSTP1 in comparison to the squamous epithelium. No nuclear stainingis present.

Fig. 6

Normal oesophageal mucosa positively staining for GSTM1, withnegative glandular epithelium in the same field.


type carcinoma with all of the antibodies tested. Some

areas were found to be completely negative (Fig. 9).

Normal duodenum

A section of normal duodenum was stained to see

whether the pattern of staining in the intestinal meta-

plasia resembled that of the normal intestinal mucosa.

Strongly positive staining was found with all the GSTs,

in the specialized cells as expected but also in the

epithelium.

DiscussionThe purpose of this study was to investigate whether

the GST family are involved in the tumourigenic

process from normal oesophagus to Barrett’s oesopha-

gus, and finally to adenocarcinoma. The general trend

witnessed suggests that metaplastic tissue may have

lower levels of GST compared to normal oesophagus.

Since the GSTs are protective detoxifying enzymes,

this decrease in expression could facilitate the expres-

sion of adenocarcinoma on a background of Barrett’s

metaplasia.

The normal tissue sections studied were extracted from

regions surrounding oesophageal lesions, which were

likely to have been exposed to noxious compounds.

Therefore, the strongly positive superficial staining ob-

served may be a protective mechanism induced by the

chemical stress to which oesophageal cells are often

directly exposed.

The GSTP1 antibody demonstrated the most convin-

cing reduction in staining in Barrett’s tissue. GSTP1 is

a widely distributed member of the cytosolic GST

family, and increased levels of this enzyme have been

implicated as a tumour marker in other cancers, such as

colon, skin and lung [30]. However, in this study, it

appears that a lack of GSTP1 in premalignant tissue

may allow progression to malignancy.

A number of sections of normal and Barrett’s oesopha-

gus did not stain for GSTM1, which would be expected

since only 50% of the population have the gene to

encode this protein [25]. This finding supports our

confidence in the specificity of the antibodies. Interest-

ingly, in one slide containing normal and Barrett’s

tissue adjacent to one another, staining for the mu

antibody was present in the former but was lost in the


Fig. 7

A poorly differentiated intestinal-type adenocarcinoma of theoesophagus (320 objective). Patchy staining is seen with GSTP1.

Fig. 8

Intestinal-type gastric carcinoma stained with GSTA1 (320 objective).Although stained with a different antibody to that used in Figure 6, thisimage is representative and demonstrates that intestinal-typecarcinomas express higher levels of GST enzymes than the diffusesubtypes of gastric adenocarcinoma.

Fig. 9

Diffuse-type gastric carcinoma stained with GSTM1 (320 objective).Signet ring cells show reduced levels of staining with the GST incomparison to intestinal-type adenocarcinoma.


latter. The loss of expression of GSTM1 suggests that

this may be a factor in the metaplastic process, though

this is speculative as it was only observed on one slide.

Normal gastric and duodenal sections were studied to

detect any similarities in GST expression between

these tissues and Barrett’s gastric and intestinal meta-

plasia. Gastric tissue and gastric metaplasia showed

similar weak staining; however, normal duodenal sec-

tions were markedly positive, unlike the pale staining

of the intestinal metaplasia. This result is in accordance

with the fact that intestinal-type metaplasia is more

prone to undergoing malignant change.

The expression of GSTs in oesophageal adenocarcioma

is complex, with variable changes identified for each

subclass. Furthermore, within individual tumours, het-

erogeneity of staining with each individual antibody

was noted. A larger study is necessary to reconcile these

changes in protein expression with the development of

oesophageal adenocarcinoma. This was initially thought

to contradict the theory that decreased levels of these

detoxifying enzymes allowed the progression of the

malignant process. However, it is well known that

tumours often express proteins in a dysregulated and

heterogeneous manner, which may explain the increase

in GST in the neoplastic stage.

Previous literature generally validates the pattern of

staining we identified in normal and Barrett’s tissue.

Therefore, despite being too small to provide definite

conclusions on the subject, this study highlights possi-

ble mechanisms that could explain the pathological

process occurring in Barrett’s metaplasia. There has

been a limited number of studies of phase II enzyme

levels in Barrett’s oesophagus, thus the present study is

in a relatively new field of research. Investigations

performed by van Lieshout et al. in the Netherlands in

1999 demonstrated low levels of GST activity in human

Barrett’s epithelia as opposed to normal oesophagus

[33]. They studied Barrett’s tissue samples retrieved

from human oesophageal endoscopies and compared

them to normal tissue. Their study is comparable with

the present one as both have utilized human samples

from endoscopic biopsy and used sensitive immuno-

logical techniques to identify the induction of enzymes.

van Lieshout et al. identified a 35% reduction in GST

enzyme activity in Barrett’s tissue compared to the

normal oesophagus. They also studied the isoenzymes

present by densometrical analysis following western

blotting and reported GSTP and GSTM to be de-

creased by 30% and 24% respectively in Barrett’s

tissue. These results suggest a decrease similar to the

one that we witnessed using immunohistochemistry.

However, van Lieshout et al. claimed that GSTA was

considerably raised, by 625%, in Barrett’s tissue com-

pared to normal tissue. Conversely, we found that, like

GSTP1, GSTA1/2 was decreased in Barrett’s tissue.

This disparity could be explained by differences in the

antibodies used, rather than the actual quantity of

protein detected.

Compton et al. also reported a similar decrease in

GSTP expression in Barrett’s tissue using immuno-

histochemistry [34]. In addition, with northern blot

analysis they observed reduced GSTP1 mRNA levels

in Barrett’s tissue compared to either normal oesopha-

gus or adenocarcinoma samples. This is relevant to our

study as we also observed a decrease in Barrett’s tissue,

coupled with a further increase in adenocarcinoma

tissue.

However, in a further study, van Lieshout et al. used

immunohistochemistry to investigate the expression of

GSTA1 and GSTP1 in normal oesophagus, Barrett’s

oesophagus and adenocarcinoma [35]. In contrast to

their previous study, they described an increase in

GSTP1 expression between normal and Barrett’s oeso-

phagus as opposed to a decrease. They also claimed

that GSTA1 was not present in normal oesophagus, but

was markedly increased in Barrett’s, then reduced again

in adenocarcinoma. Differences in the technique and

antibodies used probably account for these discrepan-

cies, which further experimentation is required to

resolve.

Nuclear staining with GSTs has been previously

reported in the oesophagus and in other tissues such as

the uterus and skin [35–37]. The nuclear staining we

observed in normal oesophagus is possibly a defensive

reaction to the threat of carcinogens or reflux of acid or

bile in the oesophagus. Potential reasons for GSTs to

enter the nucleus are to protect the DNA or to act as

transcription factors by triggering gene expression.

Before this hypothesis can be proved, a more substan-

tial panel of normal tissue samples extracted from the

area around a lesion should be compared with normal

tissue from volunteers without oesophageal pathology.

The fact that nuclear staining was witnessed in a small

number of normal sections suggests that the transmis-

sion of the GST into the nucleus is part of a stress

response, and its presence depends on which challenges

the cells were exposed to prior to being collected. It is

possible that prolonged and multiple episodes of acid

stress may desensitize the phase II response, though

this theory requires investigation.

One of the drawbacks of using immunohistochemistry

is that it is a semi-quantitative technique, therefore

levels of staining give only an impression of the amount

of protein present as opposed to an exact figure. The

positive staining seen in the immunohistochemical

procedure was authenticated by negative control slides

without visible brown staining. This proved that the



level of background staining seen in the slides was

minimal; however, the problem with using polyclonal

antibodies is that they can undergo nonspecific binding.

The use of pre-immune sera from the rabbit in which

the antibodies were raised would have helped distin-

guish further nonspecific staining, but this was not

available in our study. Thus, the results witnessed may

not be entirely specific, although the technique is

highly sensitive. Visual observations are not quantita-

tive and can give no absolute evidence. However,

overall impressions can be drawn from the alterations in

patterns and obvious changes in staining intensity.

In the future, further research is required, using a

substantially larger number of sections for immuno-

histochemistry. However, immunohistochemistry has

definite limitations, and therefore more sophisticated

techniques using fresh tissue samples would yield

results of higher quality. This would also produce a

more accurate reflection of the tumourigenic process invivo.

AcknowledgementsS.C.C. is very grateful to all the staff of the Gastro-

enterology Laboratory in Ninewells Hospital for allow-

ing her to work there, and for all their useful advice on

experimental techniques.

References1 McKinney A, Sharp L, Macfarlane GJ, Muir CS. Oesophageal and gastric

cancer in Scotland 1960–90. Br J Cancer 1995; 71:411–415.2 Powell J, McConkey CC. The rising trend in oesophageal adenocarcinoma

and gastric cardia. Eur J Cancer Prev 1992; 1:265–269.3 Blot WJ, Devesa SS, Kneller RW, Fraumeni JF Jr. Rising incidence of

adenocarcinoma of the esophagus and gastric cardia. JAMA 1991;265:1287–1289.

4 Moller H. Incidence of cancer of oesophagus, cardia and stomach inDenmark. Eur J Cancer Prev 1992; 1:159–164.

5 Hansson LE, Sparen P, Nyren O. Increasing incidence of both majorhistological types of esophageal carcinomas among men in Sweden. Int JCancer 1993; 54:402–407.

6 Hansen S, Wiig JN, Giercksky KE, Tretli S. Esophageal and gastriccarcinoma in Norway 1958–1992: incidence time trend variability accord-ing to morphological subtypes and organ subsites. Int J Cancer 1997;71:340–344.

7 Tuyns AJ. Oesophageal cancer in France and Switzerland: recent timetrends. Eur J Cancer Prev 1992; 1:275–278.

8 Liabeuf A, Faivre J. Time trends in oesophageal cancer incidence in Coted’Or (France), 1976–1993. Eur J Cancer Prev 1997; 6:24–30.

9 Levi F, Randimbison L, La Vecchia C, Erler G, Te VC. Multiple primarycancers to indicate associations between smoking and cancer incidence:Vaud and Neuchatel, Switzerland, 1974–1994. Int J Cancer 1998;76:913–914.

10 Thomas RJ, Lade S, Giles GG, Thursfield V. Incidence trends in oesopha-geal and proximal gastric carcinoma in Victoria. Aust NZ J Surg 1996;66:271–275.

11 Armstrong RW, Borman B. Trends in incidence rates of adenocarcinomaof the oesophagus and gastric cardia in New Zealand, 1978–1992. Int JEpidemiol 1996; 25:941–947.

12 Devesa SS, Blot WJ, Fraumeni JK Jr. Changing patterns in the incidenceof esophageal and gastric carcinoma in the United States. Cancer 1998;83:2049–2053.

13 Daly JM, Karnell LH, Menck HR. National Cancer Data Base report onesophageal carcinoma. Cancer 1996; 78:1820–1828.

14 Pera M, Trastek VF, Carpenter HA, Fernandez PL, Cardesa A, Mohr U,Pairolero PC. Barrett’s disease: pathophysiology of metaplasia and adeno-carcinoma. Ann Thorac Surg 1993; 56:1191–1197.

15 Gonzalez FJ, Gelboin HV. Role of human cytochromes P450 in themetabolic activation of chemical carcinogens and toxins. Drug Metab Rev1994; 26:165–183.

16 Mannervik B, Danielson UH. Glutathione transferases – structure andcatalytic activity. CRC Crit Rev Biochem 1988; 23:283–337.

17 Booth J, Boyland E, Sims P. An enzyme from rat liver catalyzing conjugateswith glutathione. Biochem J 1961; 79:516–521.

18 Mannervik B, Awasthi YC, Board PG, Hayes JD, Di Ilio C, Ketterer B, et al.Nomenclature for human glutathione transferases [letter]. Biochem J 1992;282:305–306.

19 Hayes JD, Pulford DJ. The glutathione S-transferase supergene family:regulation of GST and the contribution of the isoenzymes to cancerchemoprotection and drug resistance. Crit Rev Biochem Mol Biol 1995;30:445–600.

20 Board PG, Webb GC. Isolation of a cDNA clone and localization of humanglutathione S-transferase 2 genes to chromosome band 6p12. Proc NatlAcad Sci USA 1987; 84:2377–2381.

21 Pearson WR, Vorachek WR, Xu SJ, Berger R, Hart I, Vannais D, PattersonD. Identification of class-mu glutathione transferase genes GSTM1-GSTM5 on human chromosome 1p13. Am J Hum Genet 1993; 53:220–233.

22 Board P, Webb GC, Coggan M. Isolation of a cDNA clone and localizationof the human glutathione S-transferase 3 genes to chromosome bands11q13 and 12q13-14. Ann Hum Genet 1989; 53:205–213.

23 Pemble S, Schroeder KR, Spencer SR, Meyer DJ, Hallier E, Bolt HM, et al.Human glutathione S-transferase theta (GSTT1): cDNA cloning and thecharacterization of a genetic polymorphism. Biochem J 1994; 300:271–276.

24 Gonzalez FJ, Idle JR. Pharmacogenetic phenotyping and genotyping.Present status and future potential. Clin Pharmacokinet 1994; 26:59–70.

25 Seidegard J, Pero RW, Markowitz MM, Roush G, Miller DG, Beattie EJ.Isoenzyme(s) of glutathione transferase (class Mu) as a marker for thesusceptibility to lung cancer: a follow up study. Carcinogenesis 1990;11:33–36.

26 Salagovic J, Kalina I, Stubna J, Habalova V, Hrivnak M, Valansky L, KohutA, Biros E. Genetic polymorphism of glutathione S-transferases M1 and T1as a risk factor in lung and bladder cancers. Neoplasma 1998; 45:312–317.

27 Farber E, Parker S, Gruenstein M. The resistance of putative premalignantliver cell populations, hyperplastic nodules, to the acute cytotoxic effects ofsome hepatocarcinogens. Cancer Res 1976; 36:3879–3887.

28 Kitahara A, Satoh K, Nishimura K, Ishikawa T, Ruike K, Tsuda H, Ito N.Changes in molecular forms of rat hepatic glutathione S-transferase duringchemical hepatocarcinogenesis. Cancer Res 1984; 44:2698–2703.

29 Sato K, Tsuchida S, Tamai K. Anti-cancer drug resistance and glutathioneS-transferases. Gan To Kagaku Ryoho 1989; 16:592–598.

30 Howie AF, Forrester LM, Glancey MJ, Schlager JJ, Powis G, Beckett GJ,et al. Glutathione S-transferase and glutathione peroxidase expression innormal and tumour human tissues. Carcinogenesis 1990; 11:451–458.

31 Murray GI, Shaw D, Weaver RJ, McKay JA, Ewen SW, Melvin WT, BurkeMD. Cytochrome P450 expression in oesophageal cancer. Gut 1994;35:599–603.

32 Hayes JD, Mantle TJ. Use of immuno-blot techniques to discriminatebetween the glutathione S-transferase Yf, Yk, Ya, Yn/Yb and Yc subunitsand to study their distribution in extrahepatic tissues. Evidence for threeimmunochemically distinct groups of transferase in the rat. Biochem J1986; 233:779–788.

33 van Lieshout EM, Tiemessen DM, Witteman BJ, Jansen JB, Peters WH.Low glutathione and glutathione S-transferase levels in Barrett’s esopha-gus as compared to normal esophageal epithelium. Jpn J Cancer Res1999; 90:81–85.

34 Compton KR, Orringer MB, Beer DG. Induction of glutathione S-transfer-ase-pi in Barrett’s metaplasia and Barrett’s adenocarcinoma cell lines. MolCarcinog 1999; 24:128–136.

35 van Lieshout EM, van Haelst UJ, Wobbes T, Peters WH. Immunohisto-chemical localization of glutathione S-transferase alpha and pi in humanesophageal squamous epithelium, Barrett’s epithelium and carcinoma. JpnJ Cancer Res 1999; 90:530–535.

36 Martens JE, Smedts F, ter Harmsel B, Helmerhorst TJ, Ramaekers FC.Glutathione S-transferase pi is expressed in (pre) neoplastic lesions of thehuman uterine cervix irrespective of their degree of severity. AnticancerRes 1997; 17:4305–4309.

37 Moral A, Palou J, Lafuente A, Molina R, Piulachs J, Castel T, Trias M.Immunohistochemical study of alpha, mu and pi class glutathione S-transferase expression in malignant melanoma. MMM Group. Multidisciplin-ary Malignant Melanoma Group. Br J Dermatol 1997; 136:345.

