In Vitro Binding of GADD45A to RNA:DNA Hybrids.

R-loops are three-stranded RNA:DNA hybrid structures that frequently form during transcription. While R-loop misregulation is associated with genome instability, cells also harness RNA-DNA hybrids in scheduled, "regulatory," R-loops to control gene expression. One regulatory role involves epigenetic gene regulation by the R-loop reader Growth Arrest and DNA Damage 45A (GADD45A). This small stress related protein promotes DNA demethylation by recruiting TET dioxygenase and Thymine DNA glycosylase to specific genomic loci. GADD45A requires adapters for its genomic localization. One such class of adapters are R-loops formed at certain CpG island promoters to which GADD45A binds directly, targets the demethylation machinery, and confers an open chromatin state. Here, we describe protocols for carrying out in vitro binding assays with GADD45A to RNA:DNA hybrids to biochemically study its direct binding to R-loops, specifically GADD45A pulldown and EMSA (electrophoretic mobility shift) assays.

Gadd45α modulates aversive learning through post-transcriptional regulation of memory-related mRNAs.

Learning is essential for survival and is controlled by complex molecular mechanisms including regulation of newly synthesized mRNAs that are required to modify synaptic functions. Despite the well-known role of RNA-binding proteins (RBPs) in mRNA functionality, their detailed regulation during memory consolidation is poorly understood. This study focuses on the brain function of the RBP Gadd45α (growth arrest and DNA damage-inducible protein 45 alpha, encoded by the Gadd45a gene). Here, we find that hippocampal memory and long-term potentiation are strongly impaired in Gadd45a-deficient mice, a phenotype accompanied by reduced levels of memory-related mRNAs. The majority of the Gadd45α-regulated transcripts show unusually long 3' untranslated regions (3'UTRs) that are destabilized in Gadd45a-deficient mice via a transcription-independent mechanism, leading to reduced levels of the corresponding proteins in synaptosomes. Moreover, Gadd45α can bind specifically to these memory-related mRNAs. Our study reveals a new function for extended 3'UTRs in memory consolidation and identifies Gadd45α as a novel regulator of mRNA stability.

GADD45A binds R-loops and recruits TET1 to CpG island promoters.

R-loops are DNA-RNA hybrids enriched at CpG islands (CGIs) that can regulate chromatin states1-8. How R-loops are recognized and interpreted by specific epigenetic readers is unknown. Here we show that GADD45A (growth arrest and DNA damage protein 45A) binds directly to R-loops and mediates local DNA demethylation by recruiting TET1 (ten-eleven translocation 1). Studying the tumor suppressor TCF21 (ref. 9), we find that antisense long noncoding (lncRNA) TARID (TCF21 antisense RNA inducing promoter demethylation) forms an R-loop at the TCF21 promoter. Binding of GADD45A to the R-loop triggers local DNA demethylation and TCF21 expression. TARID transcription, R-loop formation, DNA demethylation, and TCF21 expression proceed sequentially during the cell cycle. Oxidized DNA demethylation intermediates are enriched at genomic R-loops and their levels increase upon RNase H1 depletion. Genomic profiling in embryonic stem cells identifies thousands of R-loop-dependent TET1 binding sites at CGIs. We propose that GADD45A is an epigenetic R-loop reader that recruits the demethylation machinery to promoter CGIs.

Neil DNA glycosylases promote substrate turnover by Tdg during DNA demethylation.

DNA 5-methylcytosine is a dynamic epigenetic mark with important roles in development and disease. In the Tet-Tdg demethylation pathway, methylated cytosine is iteratively oxidized by Tet dioxygenases, and unmodified cytosine is restored via thymine DNA glycosylase (Tdg). Here we show that human NEIL1 and NEIL2 DNA glycosylases coordinate abasic-site processing during TET-TDG DNA demethylation. NEIL1 and NEIL2 cooperate with TDG during base excision: TDG occupies the abasic site and is displaced by NEILs, which further process the baseless sugar, thereby stimulating TDG-substrate turnover. In early Xenopus embryos, Neil2 cooperates with Tdg in removing oxidized methylcytosines and specifying neural-crest development together with Tet3. Thus, Neils function as AP lyases in the coordinated AP-site handover during oxidative DNA demethylation.

Long noncoding RNA TARID directs demethylation and activation of the tumor suppressor TCF21 via GADD45A.

DNA methylation is a dynamic and reversible process that governs gene expression during development and disease. Several examples of active DNA demethylation have been documented, involving genome-wide and gene-specific DNA demethylation. How demethylating enzymes are targeted to specific genomic loci remains largely unknown. We show that an antisense lncRNA, termed TARID (for TCF21 antisense RNA inducing demethylation), activates TCF21 expression by inducing promoter demethylation. TARID interacts with both the TCF21 promoter and GADD45A (growth arrest and DNA-damage-inducible, alpha), a regulator of DNA demethylation. GADD45A in turn recruits thymine-DNA glycosylase for base excision repair-mediated demethylation involving oxidation of 5-methylcytosine to 5-hydroxymethylcytosine in the TCF21 promoter by ten-eleven translocation methylcytosine dioxygenase proteins. The results reveal a function of lncRNAs, serving as a genomic address label for GADD45A-mediated demethylation of specific target genes.

Bulky dna adducts in cord blood, maternal fruit-and-vegetable consumption, and birth weight in a European mother-child study (NewGeneris).

BACKGROUND: Tobacco-smoke, airborne, and dietary exposures to polycyclic aromatic hydrocarbons (PAHs) have been associated with reduced prenatal growth. Evidence from biomarker-based studies of low-exposed populations is limited. Bulky DNA adducts in cord blood reflect the prenatal effective dose to several genotoxic agents including PAHs. OBJECTIVES: We estimated the association between bulky DNA adduct levels and birth weight in a multicenter study and examined modification of this association by maternal intake of fruits and vegetables during pregnancy. METHODS: Pregnant women from Denmark, England, Greece, Norway, and Spain were recruited in 2006-2010. Adduct levels were measured by the 32P-postlabeling technique in white blood cells from 229 mothers and 612 newborns. Maternal diet was examined through questionnaires. RESULTS: Adduct levels in maternal and cord blood samples were similar and positively correlated (median, 12.1 vs. 11.4 adducts in 108 nucleotides; Spearman rank correlation coefficient = 0.66, p < 0.001). Cord blood adduct levels were negatively associated with birth weight, with an estimated difference in mean birth weight of -129 g (95% CI: -233, -25 g) for infants in the highest versus lowest tertile of adducts. The negative association with birth weight was limited to births in Norway, Denmark, and England, the countries with the lowest adduct levels, and was more pronounced in births to mothers with low intake of fruits and vegetables (-248 g; 95% CI: -405, -92 g) compared with those with high intake (-58 g; 95% CI: -206, 90 g). CONCLUSIONS: Maternal exposure to genotoxic agents that induce the formation of bulky DNA adducts may affect intrauterine growth. Maternal fruit and vegetable consumption may be protective.

Epigenetic deregulation of TCF21 inhibits metastasis suppressor KISS1 in metastatic melanoma.

Metastatic melanoma is a fatal disease due to the lack of successful therapies and biomarkers for early detection and its incidence has been increasing. Genetic studies have defined recurrent chromosomal aberrations, suggesting the location of either tumor suppressor genes or oncogenes. Transcription factor 21 (TCF21) belongs to the class A of the basic helix-loop-helix family with reported functions in early lung and kidney development as well as tumor suppressor function in the malignancies of the lung and head and neck. In this study, we combined quantitative DNA methylation analysis in patient biopsies and in their derived cell lines to demonstrate that TCF21 expression is downregulated in metastatic melanoma by promoter hypermethylation and TCF21 promoter DNA methylation is correlated with decreased survival in metastatic skin melanoma patients. In addition, the chromosomal location of TCF21 on 6q23-q24 coincides with the location of a postulated metastasis suppressor in melanoma. Functionally, TCF21 binds the promoter of the melanoma metastasis-suppressing gene, KiSS1, and enhances its gene expression through interaction with E12, a TCF3 isoform and with TCF12. Loss of TCF21 expression results in loss of KISS1 expression through loss of direct interaction of TCF21 at the KISS1 promoter. Finally, overexpression of TCF21 inhibits motility of C8161 melanoma cells. These data suggest that epigenetic downregulation of TCF21 is functionally involved in melanoma progression and that it may serve as a biomarker for aggressive tumor behavior.

Biomarkers for hazard identification in humans.

BACKGROUND: Oxidative stress enhances lipid peroxidation (LPO), which both are implicated in the promotion and progression stages of carcinogenesis, in particular under conditions of chronic inflammation and infections. Exocyclic etheno-DNA adducts, which are formed by LPO-products such as 4-hydroxy -2-nonenal, are strongly pro-mutagenic DNA lesions. METHODS: The development of ultra-sensitive detection methods for etheno-adducts in human tissues, white blood cells( WBC) and urine has provided evidence that these adducts are elevated in affected organs of cancer-prone patients, probably acting as a driving force to malignancy. RESULTS: Two recent studies that yielded some new insights into disease causation are briefly reviewed: DNA-damage in WBC of mother-newborn child pairs, and lipid peroxidation derived DNA damage in patients with cancer-prone liver diseases. Our results indicate that biomonitoring of etheno-DNA adducts in humans are promising tools (i) to better understand disease aetiopathogenesis, allowing hazard identification(ii) to monitor disease progression and (iii) to verify the efficacy of chemopreventive and therapeutic interventions. Such clinical trials are warranted.

Quantifying etheno-DNA adducts in human tissues, white blood cells, and urine by ultrasensitive (32)P-postlabeling and immunohistochemistry.

Exocyclic etheno-DNA adducts are formed by the reaction of lipid peroxidation products, such as 4-hydroxy-2-nonenal (HNE) with DNA bases to yield 1,N (6)-etheno-2'-deoxyadenosine (εdA), 3,-N (4)-etheno-2'-deoxycytidine (εdC), and etheno-2'-deoxyguanosine. These adducts act as a driving force for many human malignancies and are elevated in the organs of cancer-prone patients suffering from chronic inflammation and infections. Here, we describe the ultrasensitive and specific techniques for the detection of εdA and εdC in tissue and white blood cell (WBC) DNA. This approach is based on -combined immunopurification by monoclonal antibodies and (32)P-postlabeling analysis. The detection limit is about five adducts per 10(10) parent nucleotides, requiring 5-10 µg of DNA. In addition, we describe techniques for immunohistochemical detection of εdA and εdC in tissue biopsies, and the approaches for the -analysis of εdA and εdC excreted in urine. The utility of these detection methods for human studies is based on: (1) high sensitivity and specificity, (2) low amounts of DNA required, (3) capability to detect "background" levels of etheno-DNA adducts in biopsies, WBC, and urine samples of healthy subjects, and (4) reliable monitoring of the disease-related increase of these substances in patients.The described methods are useful in diagnosis and monitoring of chronic degenerative diseases, including cancer, atherosclerosis, and neurodegenerative disorders.

Accumulation of lipid peroxidation-derived DNA lesions in iron-overloaded thalassemic mouse livers: comparison with levels in the lymphocytes of thalassemia patients.

In thalassemia patients, iron overload can stimulate lipid peroxidation (LPO), thereby generating miscoding DNA adducts. Adducted DNA was measured in the lymphocytes of beta-Thal/Hb E patients and healthy controls and in the organs of thalassemic mice. epsilondA, epsilondC and M(1)dG residues were quantified by (32)P-postlabeling-TLC/HPLC. M(1)dG levels in lymphocyte DNA from patients were 4 times as high as in controls, while the increase in epsilondA and epsilondC was not significant. Adducted DNA accumulated in the liver of thalassemic mice having >2.7 mg Fe/g tissue dry weight; DNA adducts and iron were highly correlated. epsilondA was not specifically generated in certain mouse liver cell types as revealed by immunohistochemical staining. We found elevated LPO-induced DNA damage in the liver of thalassemic mouse and in lymphocytes, implicating that massive DNA damage occurs in the liver of thalassemia patients. We conclude that promutagenic LPO-derived DNA lesions are involved in the onset of hepatocellular carcinoma in these patients.

Activator protein 2 alpha (AP2alpha) suppresses 42 kDa C/CAAT enhancer binding protein alpha (p42(C/EBPalpha)) in head and neck squamous cell carcinoma.

The tumor suppressor C/CAAT enhancer binding protein alpha (C/EBPalpha) is a transcription factor involved in cell cycle control and cellular differentiation. A recent study showed that C/EBPalpha is frequently downregulated in head and neck squamous cell carcinoma (HNSCC) by DNA methylation in an upstream regulatory region. Here, we investigated how DNA methylation in the upstream regulatory region disrupts the transcriptional regulation of C/EBPalpha in HNSCC. The results reveal that aberrant methylation correlates with methyl binding domain protein binding and repressive histone modifications. This methylated region contains previously uninvestigated AP2alpha binding sites. AP2alpha suppresses C/EBPalpha promoter activity and protein expression. Interestingly, silencing AP2alpha by shRNA increases the antiproliferative isoform of C/EBPalpha (p42(C/EBPalpha)). Furthermore, growth analysis revealed that these 2 isoforms yield very different proliferative properties in HNSCC.

Frequently methylated tumor suppressor genes in head and neck squamous cell carcinoma.

Head and neck squamous cell carcinoma (HNSCC) is a very aggressive cancer. In advanced stages, the patient has poor chances of receiving effective treatment, and survival rates are low. To facilitate timely diagnosis and improve treatment, elucidation of early detection markers is crucial. DNA methylation markers are particularly advantageous because DNA methylation is an early event in tumorigenesis, and the epigenetic modification, 5-methylcytosine, is a stable mark. A genome-wide screen using Restriction Landmark Genomic Scanning found a set of genes that are most commonly methylated in head and neck cancers. Five candidate genes: septin 9 (SEPT9), sodium-coupled monocarboxylate transporter 1 (SLC5A8), functional smad-suppressing element on chromosome 18 (FUSSEL18), early B-cell factor 3 (EBF3), and iroquois homeobox 1 (IRX1) were methylated in 27% to 67% of the HNSCC patient samples tested. Furthermore, approximately 50% of the methylated tumor samples shared methylation between two of the five genes (most commonly between EBF3 and IRX1), and 15% shared methylation between three of the five genes. Expression analysis revealed candidate gene down-regulation in 25% to 93% of the HNSCC samples, and 5-aza-2'-deoxycytidine treatment was able to restore expression in at least 2 of 5 HNSCC cell lines for all of the genes tested. Overexpression of the three most frequently down-regulated candidates, SLC5A8, IRX1, and EBF3, validated their tumor suppressor potential by growth curve analysis and colony formation assay. Interestingly, all of the candidates identified may be involved in the transforming growth factor beta signaling pathway, which is often disrupted in HNSCC.

Polyunsaturated fatty acids modulate NOX 4 anion superoxide production in human fibroblasts.

The strong ROS (reactive oxygen species) production, part of an antioxidant response of human fibroblasts triggered by DHA (docosahexaenoic acid; C(22:6,n-3), served as a model for deciphering the relative contribution of NOX (NADPH oxidase) to ROS production, as the role of this enzymatic system remains controversial. Using hydroxyethidium fluorescence for fibroblast ROS production, RT (reverse transcriptase)-PCR for NOX 4 mRNA quantification and mRNA silencing, we show that ROS production evolves in parallel with the catalytic activity of NOX and is suppressed by siNOX 4 (small interference oligonucleotide RNA directed against NOX 4) silencing. Apocynin and plumbagin, specific inhibitors of NOX, prevent ROS production in this cellular model and confirm the role of NOX 4 for this production. Furthermore, we show that, in cell lysates, NOX 4 activity can be modulated by PUFAs (polyunsaturated fatty acids) at the micromolar level in the presence of calcium: NOX 4 activity is increased by arachidonic acid (C20:4,n-6) (approximately 175% of the control), and conjugated linoleic acid (C18:2 [9Z,11E]) is a potent inhibitor (50% of the control). Unexpectedly, intracellular superoxide dismutase does not participate in the modulation of this ROS production and the opposite effects of some PUFAs, described in our experiments, could suggest another way of regulating NOX activity.

Conjugated linoleic acid, unlike other unsaturated fatty acids, strongly induces glutathione synthesis without any lipoperoxidation.

Enhancement of the redox status of cells is a cytoprotective strategy against oxidative damage. We recently showed that DHA upregulates glutathione (GSH) content via an induction of its related enzymes gamma-glutamylcysteine ligase and glutathione reductase. In the present study, we investigated the effects of eight other fatty acids on the redox status and lipid peroxidation of human fibroblasts. After 48 h, only arachidonic acid and conjugated linoleic acid (CLA) enhanced GSH content through an induction of gamma-glutamylcysteine ligase. CLA was more potent than arachidonic acid in inducing GSH synthesis. For all the fatty acids tested, lipoperoxidation, estimated by cell malondialdehyde measurement, did not differ from that of controls at 48 h but dramatically increased at 7 d, except for CLA. Lipoperoxidation is associated at 7 d with a high level of reactive oxygen species and with increased haemoxygenase-1 and cyclooxygenase-2 mRNA expression. As demonstrated by a tert-butylhydroperoxide cytotoxicity test, the GSH synthesis obtained with arachidonic acid is not sufficient to protect the cells, whereas this protective effect was obvious with CLA at 48 h as well as at 7 d. The present results show that CLA is the only PUFA able to induce GSH synthesis without any change in oxidative balance, whereas an upregulation of cyclooxygenase-2 by other PUFA is concomitant with an overproduction of malondialdehyde and reactive oxygen species. The particular hairpin conformation obtained for CLA by molecular modelling could account for this specific biological effect.

Dietary beta-carotene inhibits mammary carcinogenesis in rats depending on dietary alpha-linolenic acid content.

To investigate whether dietary alpha-linolenic acid (ALA) content alters the effect of beta-carotene on mammary carcinogenesis, we conducted a chemically induced mammary tumorigenesis experiment in rats randomly assigned to four nutritional groups (15 rats per group) varying in beta-carotene supplementation and ALA content. Two oil formula-enriched diets (15 %) were used: one with 6 g ALA/kg diet in an essential fatty acids (EFA) ratio of linoleic acid:ALA of 5:1 w/w (EFA 5 diet), the other with 24 g ALA/kg diet in an EFA ratio of 1:1 w/w (EFA 1 diet), both designed with a similar linoleic acid content. beta-Carotene was either added (10 mg/kg diet per d) or not added to these diets. beta-Carotene supplementation led to decreased tumour incidence and tumour growth when added to the EFA 5 diet, whereas it had no effect when added to the EFA 1 diet. The decreased tumour growth did not result from an involvement of lipoperoxidation (tumour malondialdehyde content being similar between the groups) or from an inhibition of tumour cell proliferation (as there was an unchanged S phase fraction in the tumours). We concluded that an adequate content of ALA in the diet is required to allow a protective effect of beta-carotene in mammary carcinogenesis. Whether such an interaction between ALA and beta-carotene influences the risk of breast cancer in women needs to be investigated.

Docosahexaenoic acid enhances the antioxidant response of human fibroblasts by upregulating gamma-glutamyl-cysteinyl ligase and glutathione reductase.

The chemopreventive effects of dietary n-3 PUFA in various pathologies has so far remained controversial, and we were interested in studying their potential influence on cell redox status. DHA (22 : 6n-3), a typical highly unsaturated n-3 PUFA, was used at 30 micromol/l in a model of human fibroblast cell culture. A dose-response effect, roughly linear, was checked for DHA between 0 and 60 micromol/l, and was accompanied by a large increase in intracellular GSH content. A time course study of this effect shows that, after a short fall, as soon as 4 h after the beginning of the experiment, the large increase in the GSH content was associated with elevated catalytic activities of gamma-glutamyl-cysteinyl ligase, glutathione reductase and glutathione S-transferase. This coordinated response is characteristic of an antioxidant response and was confirmed by the induction of expression of mRNA for gamma-glutamyl-cysteinyl ligase, glutathione reductase and haem-oxygenase. This large increase in the GSH content contributes to decreasing the reactive oxygen species level, as assessed by the decreased accumulation of dichlorofluorescein inside cells. To our knowledge, this is the first report on a specific and potent effect of DHA for decreasing the oxidative stress of human fibroblasts.

Kinetic measurement by LC/MS of gamma-glutamylcysteine ligase activity.

gamma-Glutamylcysteine ligase (GCL) combines cysteine and glutamate through its gamma carboxyl moiety as the first step for glutathione (GSH) synthesis and is considered to be the rate-limiting enzyme in this pathway. The enzyme is a heterodimer, with a heavy catalytic and a light regulatory subunit, which plays a critical role in the anti-oxidant response. Besides the original method of Seelig designed for the measurement of a purified enzyme, few endpoint methods, often unrefined, are available for measuring it in complex biological samples. We describe a new, fast and reliable kinetic LC/MS method which enabled us to optimize its detection. l-2-Aminobutyrate is used instead of cysteine (to avoid glutathione synthetase interference) as triggering substrate with saturating concentrations of glutamate and ATP; the gamma glutamylaminobutyrate formed is measured at m/z=233 at regular time intervals. Reaction rate is maximum because ATP is held constant by enzymatic recycling of ADP by pyruvate kinase and phosphoenolpyruvate. The repeatability of the method is good, with CV% of 6.5 and 4% for catalytic activities at, respectively 0.9 and 34 U/l. The affinities of rat and human enzymes for glutamate and aminobutyrate are in good agreement with previous published data. However, unlike the rat enzyme, human GCL is not sensitive to reduced glutathione and displays a more basic optimum pH.

Alpha-tocopherol suppresses mammary tumor sensitivity to anthracyclines in fish oil-fed rats.

Polyunsaturated fatty acids (PUFAs) have been reported to enhance the efficacy of chemotherapeutic agents that produce reactive oxygen species such as anthracyclines. We previously reported in a human breast cancer cell line that the increased cytotoxic activity of anthracyclines by several PUFAs was abolished by antioxidants and enhanced by pro-oxidants, suggesting that lipid peroxidation was involved in this effect. To determine the relevance of this observation in vivo, we examined the effect of the oxidative status of the diet on the activity of epirubicin against N-methylnitrosourea-induced mammary tumors in Sprague-Dawley rats. Three groups of rats were fed a basal diet enriched with dietary n-3 PUFA (sardine oil, 15%) alone (control group), with addition of an antioxidant (alpha-tocopherol, 100 UI/kg diet), or with addition of an oxidant system (dehydroascorbate/naphthoquinone). When the first mammary tumor reached 1 cm2, epirubicin was administrated weekly for 3 wk, and subsequent change in tumor size was documented over time. Two weeks after the end of epirubicin injections, tumor size was increased by 34% in the control group. In the pro-oxidant group, tumor size was decreased by 50%. In contrast, tumor size was increased by 188% in the antioxidant group. Thus, addition of pro-oxidants in a fish oil-enriched diet increased the sensitization of mammary tumors to chemotherapy, whereas addition of alpha-tocopherol suppressed tumor response in vivo, indicating that interaction between components of the diet has to be carefully controlled during chemotherapy.

Plasma lipid hydroperoxides measurement by an automated xylenol orange method.

Lipid hydroperoxides (LH) appear to be good candidates as initial biomarkers of oxidative stress. We describe an automated method to quantify it, based on a known principle: oxidation of Fe II to Fe III by lipid hydroperoxides, under acidic conditions, followed by complexation of Fe III by xylenol orange. This method requires only a 10-microl sample volume of heparinized plasma or serum. It has been carried out automatically, with two reagents, in a two-end-point mode with bichromatic detection at 570 and 700 nm. The within-run precision, measured on a low- and a high-level plasma, was 5.0+/-0.3 and 14.0+/-0.6 microM (n=25 for each series). The between-run precision (one run for 18 days), evaluated on two commercial controls, was 5.6+/-0.5 microM (CV=8.9%) and 7.9+/-0.5 microM (CV=6.3%). The recovery of known amounts of tert-butylhydroperoxide (1 and 2 microM) added to human plasma was 98%. The specificity was demonstrated by the excellent correlation of the values of 42 samples measured either directly, with a simple dilution, or after gel permeation chromatography. The reference interval determined on 21 subjects was 4.9+/-1.7 microM. This was in the upper range of previously published values but our recovery and chromatographic experiments strongly suggest that former methods have underestimated the true content of LH in human plasma.

Fast liquid chromatography-mass spectrometry glutathione measurement in whole blood: micromolar GSSG is a sample preparation artifact.

We describe a new, fast (6 min) and reliable method to measure reduced or oxidized glutathione (GSH) or (GSSG) in whole blood. The method is based on a LC/MS measurement in positive electrospray ionization mode after a chromatographic separation on a specific column which does not need any counter-ion in the mobile phase, improving the sensitivity of detection. A 50 microl sample of whole blood is sufficient for analysis. We demonstrate that the lack of an alkylating agent during the sample preparation brings out an underestimation of GSH and an artefactual production of GSSG, corresponding to 2-3% of GSH. The simultaneous use of N-ethyl-maleimide and a strong deproteinising acid prevents these two drawbacks. This efficient and new method of preparation and analysis lets us show that, unexpectedly, GSH is stable in whole blood for some hours and that deproteinised samples can be stored without GSH loss for at least three weeks at -20 or -80 degrees C. The reference interval, measured on 22 volunteers, on blood samples collected either with heparin or with EDTA, is 1310 +/- 118 microM for GSH and 0.62 microM for GSSG. The within-run precision of this method, with gamma glutamyl-glutamic acid as an internal standard, evaluated in three successive series (n = 30), lies between 2.1 and 4.8% for a GSH level at 580 or 1150 microM. The one step sample preparation we propose seems well suited for GSH routine measurements in hospital laboratories and avoids any underestimation of GSH, a now well accepted biomarker of oxidative stress.