Mice with genetic gamma-glutamyl transpeptidase deficiency exhibit glutathionuria, severe growth failure, reduced life spans, and infertility.

A mouse mutant with glutathionuria was discovered by screening for amino acidurias in the progeny of ethylnitrosourea-mutagenized mice. Total glutathione concentration was increased in both blood and urine but decreased in liver homogenates from affected mice. Glutathionuric mice exhibited lethargy, severe growth failure, shortened life spans and infertility. gamma-Glutamyl transpeptidase activity was deficient in kidney homogenates of glutathionuric mice. The glutathionuric phenotype in these mice is inherited as an autosomal recessive trait. This mouse mutant will be a useful animal model for the study of gamma-glutamyl transpeptidase physiology and glutathione metabolism.

GSH is the most abundant cellular thiol and functions as the principal reducing reagent in all cell types (1). A partial listing of the antioxidative functions of GSH include: protection against mitochondrial damage, protection against oxygen toxicity in the lung, protection against lipid peroxidation, detoxification of electrophilic compounds through conjugation, preservation of proper sulfide bonds in proteins, a postulated function in anticarcinogenesis, and a role in the immune system (2). GSH metabolism also provides a source of cysteine for cells (3). ␥-Glutamyl transpeptidase (␥-GT; EC 2.3.2.2) 1 catalyzes the initial step in the degradation of GSH (4). ␥-GT is a key step in the ␥-glutamyl cycle (5), a series of degradative and synthetic reactions that mediate cellular GSH metabolism. Several reviews of ␥-GT physiology and function have been published (4 -6), but despite intensive investigation, the exact role ␥-GT plays in GSH metabolism or its putative contribution to renal amino acid transport have not been definitively determined. Bound to secretory endothelial cell membranes in several organs but predominantly in proximal renal tubule cells, ␥-GT participates in the transmembrane transport of GSH and in interorgan GSH exchange ( Fig. 1) (7). Meister (8) proposed that ␥-GT also contributes to amino acid transport in the proximal renal tubule through transpeptidation of GSH and subsequent tubule cell uptake of ␥-glutamyl amino acids. In vivo model systems that have lost ␥-GT activity are an exquisitely powerful tool for the study of ␥-GT function and its relationship to GSH metabolism. Administration of chemical inhibitors of ␥-GT to animals results in both glutathionuria and glutathionemia (9), but chemically treated animal models are limited by several drawbacks including the temporary nature of inhibition and the difficulty of long-term continuous inhibitor administration. Also, the degree and specificity of enzyme inhibition in various tissues (particularly the brain) of these chemically treated animals is unknown. Genetic ␥-GT deficiency has been described in only five humans (6), and the effects of various different disease states or environmental influences upon ␥-GT deficient individuals cannot be adequately evaluated given the rarity of the disorder. A genetic animal model of total ␥-GT deficiency overcomes the limitations of previously existing experimental systems and provides a useful tool for the study of ␥-GT function and GSH physiology.
N-Ethyl-N-nitrosourea (ENU) is the most potent point mutagen known (10), yielding mutation rates up to 300 ϫ 10 Ϫ5 in mice, depending upon the specific locus tested. ENU mutagenesis has been used by several laboratories to generate mutant mouse strains that model specific human genetic diseases (11). Since the report of phenylketonuria secondary to phenylalanine hydroxylase deficiency in ENU-generated mouse mutants (12), our laboratory has focused upon developing mouse models of other human inborn errors of metabolism. To this end, we screen the progeny of ENU-treated mice for metabolic abnormalities using a variety of urine biochemical analyses. Using this protocol, we have previously isolated a mouse strain with recessively inherited sarcosine dehydrogenase deficiency (13). In this report, we describe a genetic mouse model of autosomal recessively inherited ␥-GT deficiency developed by random mutagenesis using ENU. These mice exhibit glutathionuria/emia, severe growth failure, shortened life spans, and inability to reproduce. This strain, designated GGT enu1 , provides a useful experimental system for the study of ␥-GT physiology and GSH metabolism.

EXPERIMENTAL PROCEDURES
All reagents were purchased from Sigma unless otherwise specified. Mutagenesis and Metabolic Screening Protocol-Male C57Bl/6J mice were treated with 50 mg/kg body weight ENU weekly for a total of three doses according to the method of King et al. (14). Breeding and metabolic screening of potentially mutant progeny was carried out as described previously (13). Urine samples were collected on filter paper from all G 3 progeny after weaning (21-28 days of age) following an overnight fast. Each urine sample was evaluated using a battery of qualitative chemical analyses including sodium cyanide-sodium nitroprusside reagent for the detection of disulfides and free sulfhydryls (15). Urine amino acids were examined with one-dimensional paper chromatography using butanol:acetic acid:water (12:3:5) solvent and detection with ninhydrin according to the method of Smith (16). Quantitative urine and plasma amino acid analysis and quantitative urinary organic acid analysis were performed in any mouse that had an abnormal result by amino acid paper chromatography. Quantitative urine and plasma amino acids were analyzed on a Beckman System 6300 automatic amino acid analyzer according to the methods of Slocum and Cummings (17). Quantitative urine organic acids were measured by trimethylsilyl derivatization followed by gas chromatography-mass spectrometry according to the method of Hoffmann et al. (18). Duplicate one-dimensional paper chromatograms were examined specifically for the presence of sulfur-containing amino acids using platinic iodide reagent (15) in any mouse that had an abnormal cyanide-nitroprusside screening test. Urine from the mouse mutant described in this report and the identity of the sulfur-containing compounds therein were further examined by two-dimensional high voltage thin layer chromatography with ninhydrin detection (15) in the laboratory of Dr. Vivian Shih, Massachusetts General Hospital, Boston, MA.
Tissue Glutathione Concentration-Total glutathione concentration in urine, plasma, and tissues of control and glutathionuric (GGT enu1 ) mice was determined enzymatically using the method of Tietze (19). GSH reacts with the sulfhydryl reagent 5,5Ј-dithiobis-(2-nitrobenzoic acid) (DTNB or Ellman's reagent) to produce 2-nitro-5-thiobenzoic acid and its glutathionyl adduct. Yeast glutathione reductase added to the reaction mixture reduces GSSG in the presence of NADPH to form GSH which is then free to react with DTNB. The rate of 2-nitro-5-thiobenzoic acid production is measured by monitoring absorbance at 412 nm and is proportional to the total GSH concentration in the solution. For the assay, all reagents are prepared in 125 mM sodium phosphate, pH 7.5, 6.3 mM sodium EDTA buffer. 700 l of 0.3 mM NADPH, 100 l of 6 mM DTNB, and 200 l of sample are combined in a 1-ml cuvette and allowed to equilibrate at 30°C in a water bath for 5 min. The reaction was initiated by adding 10 l of 50 unit/ml yeast glutathione reductase, and the increase in absorbance at 412 nm was recorded for 6 min at 30°C. Total glutathione concentration in unknowns was calculated from a standard curve of known GSSG concentrations varying from 0 to 0.02 mM but are reported as millimoles/liter GSH. The absorbance at 412 nm does not appreciably change over 6 min without the addition of glutathione reductase.
Urine samples for total glutathione measurement were collected from urine-soaked filter paper by NH 4 OH elution. Urine glutathione concentration was measured without further sample modification and was corrected for urine creatinine concentration. Blood was obtained by direct cardiac puncture and anticoagulated with 25 l of 0.1 M EDTA. Plasma, collected by centrifugation, was deproteinized with 0.1 volume 35% w/v sulfosalicylic acid. Following centrifugation, total glutathione concentration was measured in the supernatant. Solid tissue samples were weighed and homogenized in 10 volumes of 10% w/v trichloroacetic acid with five up and down strokes of a Pro2000 tissue homogenizer (Pro Scientific, Inc.). Following centrifugation at 3000 ϫ g, total glutathione concentration was measured in the supernatant and corrected for wet weight of the tissue sample.
Kidney ␥-GT Activity-␥-GT activities in whole kidney homogenates of control (C57Bl/6J) and GGT enu1 mice were determined by the method of Orlowski and Meister (20). 10% w/v kidney homogenates were prepared in 0.1 M Tris-HCl, pH 9.0, with five strokes of a Pro2000 electric homogenizer while the samples remained on ice. The samples were kept on ice or stored at Ϫ20°C until ␥-GT activity was measured. In this assay, the ␥-glutamyl moiety of the artificial substrate, ␥-glutamyl-pnitroanilide, is transferred to glycylglycine by ␥-GT and the optically active product, p-nitroaniline, is generated. 1.8 ml of 11.11 mM glycineglycine ϩ 2.78 mM ␥-glutamyl-p-nitroanilide in 0.1 M Tris-HCl, pH 9.0, is added to 10 -500 g of sample protein in a 200-l volume. The reaction is incubated at 37°C for 10 min and then stopped with the addition of 50 l of 50% w/v trichloroacetic acid followed by 2 ml of 2 M acetic acid. The absorbance of the sample at 410 nm is measured versus a duplicate reaction to which trichloroacetic acid and acetic acid had been added immediately to stop the reaction at time 0. The amount of p-nitroaniline produced in the reaction was calculated using the molar absorptivity of p-nitroaniline at 410 nm (⑀ ϭ 8.8 mM Ϫ1 cm Ϫ1 ) and ␥-GT activity was expressed in terms of nanomoles of p-nitroaniline produced/min/mg of protein. Protein concentrations were determined by a modification of a bicinchoninic acid method (Pierce) (21).
Immunoblotting-␥-GT was partially purified from wild type Harlan Sprague Dawley rat, wild type C57Bl/6J mouse, and GGT enu1 mouse kidney homogenates according to the method of Hughey and Curthoys ␥-GT on the tubule epithelial membrane transfers the ␥-glutamyl moiety of GSH in the glomerular filtrate to an acceptor such as an amino acid (AA) or water. The products of this reaction, cysteinylglycine and the ␥-glutamyl-acceptor adduct are taken up by the epithelial cell and degraded further enzymatically. The ultimate products of these reactions, namely glycine, cysteine, and glutamate, may be utilized to resynthesize GSH. GSH is then secreted into either the peritubular plasma or back into the tubule lumen. (22). Briefly, kidney microsomal fractions were isolated by differential centrifugation and resuspended in 0.1 M Tris buffer, pH 9.0, 1% Triton X-100. Alkaline phosphatase (Sigma diagnostic kit) and ␥-GT activities were measured on the microsomal fractions from each animal. Protein electrophoresis was performed on SDS-10% polyacrylamide gels (23), and proteins were transferred by electrophoretic elution (24) to polyvinylidene difluoride membranes (Immobilon-P, Millipore). ␥-GT protein was identified using polyclonal rabbit antisera (courtesy of Dr. Henry Pitot) raised against purified rat ␥-GT. As a control, an integral renal tubule membrane protein, the ␤1 subunit of Na,K-ATPase, was detected on a duplicate immunoblot using polyclonal rabbit sera raised against amino acids 152-340 of the rat Na,K-ATPase ␤1 subunit as deduced from the cDNA sequence (Upstate Biotechnology, Lake Placid, NY). On both blots, the primary antibodies were localized with alkaline phosphatase-conjugated goat anti-rabbit IgG antibody (Sigma), and alkaline phosphatase activity was detected using nitro blue tetrazolium/bromochloroindolyl phosphate substrate (Pierce).

Initial Detection of Mice with Glutathionuria-Male
C57Bl/6J inbred mice were treated weekly with ENU by intraperitoneal injection for a total of three doses as described previously (14). After the mice had regained fertility, each was bred to C57Bl/6J females to generate five to ten G 1 males. G 1 (n ϭ 31) males were bred with females to generate G 2 offspring and then each of the G 1 males were back-crossed with two or three of their G 2 daughters to generate a total of 553 potentially homozygous mutant G 3 mice. Urine samples from two male mice in a single sibship gave a positive test result with sodium cyanide-sodium nitroprusside reagent, indicating the presence of a disulfide or free sulfhydryl-containing compound in the urine. Qualitative urine amino acid analysis by one-dimensional paper chromatography and ninhydrin detection demonstrated a large red-purple spot with R F 0.13 and a smaller pink spot with R F 0.31 which are not present on the amino acid chromatograms of control mouse urine. The larger spot reacted with both the cyanide-nitroprusside and iodoplatinate reagents indicating the presence of sulfur in the compound. This large spot had the same R F and color on ninhydrinstaining as GSSG purchased from Sigma. The smaller pink spot corresponded to GSH, which was present at much lower concentrations than GSSG in the mutant mouse urine. Oxidized and reduced glutathione were also identified by twodimensional high voltage electrophoresis of mutant mouse urine by Drs. Kimiyo Mogami and Vivian Shih, Massachusetts General Hospital, Boston, MA.
Quantitative urine amino acid analysis using the Beckman amino acid analyzer (high pressure liquid chromatography) showed three abnormal peaks on the chromatogram (Fig. 2). A single, sharp peak with a retention time of 12.0 min co-chromatographed with both GSH and ␥-glutamylcysteine. A broad peak at 18 -27 min corresponded to GSSG. The third peak at 37 min was broad and smaller than the first two peaks and has been tentatively identified as bis-␥-glutamylcystine (␥-glutamylcysteine disulfide), a substance which has also been detected in the urine of humans with ␥-GT deficiency (25). The method of urine collection used in our screening protocol likely promoted oxidation of GSH to GSSG (and of ␥-glutamylcysteine to bis-␥-glutamylcystine), yielding the large amount of GSSG compared with GSH in most urine samples from the affected mice. Pretreatment of mutant mouse urine with dithiothreitol resulted in complete disappearance of the broad peaks associated with GSSG and bis-␥-glutamylcystine and accentuation of the peak corresponding to GSH and ␥-glutamylcysteine (data not shown).
Phenotype of the Affected Mice-At birth, glutathionuric mice were indistinguishable from their unaffected litter mates. The total number of offspring in litters which contained glutathionuric pups was not significantly different from the size of control C57Bl/6J litters (5-12 pups/litter) in our colony. By approximately 14 days of age, glutathionuric pups appeared smaller than their litter mates, and at weaning (3-4 weeks of age), the body weight of affected mice was significantly lower (Fig. 3). Other than their size, the mice at weaning appeared physically healthy and morphologically normal (Fig. 4).
Severe growth failure persisted after weaning (Fig. 3). As the glutathionuric mice aged, their fur became rough and dull, but there was no change in coat color. The mice became less active, but were easily agitated. Many of the glutathionuric mice developed a severe thoracolumbar kyphosis. One severely kyphotic mouse developed paralysis of the hindlimbs, and two males suffered maceration of the penis due to priapism. Approximately 10% of the mice exhibit unilateral cataracts which are physically apparent by 2-3 months of age. A single mouse had bilateral cataracts. Unfortunately, neither male nor female glutathionuric mice have produced offspring. The line of glutathionuric mice has been maintained with considerable difficulty only by breeding unaffected siblings of glutathionuric mice which potentially are heterozygous for the glutathionuric trait.
Gross anatomic examination of several mice revealed no abnormalities except for small size of all organs in proportion to the small body size overall. In one mouse sacrificed at age 14 months, a large cystic mass completely replaced the right kidney. The right ureter, urinary bladder, and the left kidney and ureter appeared grossly normal. In another mouse sacrificed at age 15 months, a solitary solid tumor was found in the liver. Aortic perfusion with formalin-saline was performed on two male glutathionuric mice. Microscopic examination of most internal organs with hematoxylin-eosin staining was completely normal except for oligospermia apparent in the testes. Nissl staining of coronal sections from the lens of a single mouse confirmed the subcapsular location of a clinically appar-ent unilateral cataract. Histology of the contralateral lens in that mouse was normal. The brain and spinal cord of two mice were sectioned and examined with hematoxylin-eosin or Nissl staining. No anatomic abnormalities of the nervous system were detected. The hematocrit of four different glutathionuric mice was similar to that of control mice, ranging from 50 -55%. The microscopic appearance of Wright-stained peripheral blood smears from glutathionuric mice was normal compared with blood smears from control C57Bl/6J mice. Neither male nor female glutathionuric mice have successfully produced offspring. The mean age of glutathionuric mice (n ϭ 60) at death was 242 Ϯ 15 days (ϮS.E.) with a range of 70 -512 days, while the phenotypically normal siblings of glutathionuric mice (n ϭ 88) lived an average of 339 Ϯ 22 days (p Ͻ 0.001) with a range of 74 -929 days.
Quantitative Amino Acid Analysis-In addition to the large amount of glutathione detected, quantitative urine amino acid analysis revealed slight elevations of many amino acids including threonine, glycine, cystine, isoleucine, leucine, ornithine, and lysine compared with control mouse urine samples (Table  I). This may indicate a mild generalized defect of amino acid reabsorption in the renal tubules of glutathionuric mice. Additionally, taurine, a sulfur-containing amino acid that is usually the most abundant amino acid in control mouse urine, was consistently low in urine from glutathionuric mice. Plasma cystathionine levels were slightly lower in mutant mice (3.1 Ϯ 3.8 mol/liter) than in controls (8.5 Ϯ 1.8 mol/liter), but no other significant differences in plasma amino acid concentrations were detected. Specifically, the plasma levels of other sulfur-containing amino acids including methionine and nonprotein-bound cystine were normal in glutathionuric mice. Quantitative amino acid analysis was performed on trichloroacetic acid extracts of liver, kidney, and brain of glutathionuric and control mice. No significant differences in any amino acids including free cystine were detected (data not shown). Tissue Glutathione Concentrations-Using the method of Tietze (19), glutathione concentration was measured in urine and plasma, and trichloroacetic acid extracts of liver, kidney, and brain from glutathionuric and control mice (Table II). As expected, urine glutathione excretion was very elevated in the mutant mice. Plasma glutathione concentration was also elevated in the mutant mice. Total glutathione was also elevated in trichloroacetic acid extracts of kidney and brain from glutathionuric mice. However, total glutathione in liver extract from glutathionuric mice was significantly decreased compared with control mice.
Kidney ␥-GT Activities-␥-GT-specific activities were measured in Tris-HCl extracts of whole kidney from glutathionuric and control mice (Table III). Kidney ␥-GT activity of glutathionuric mice was nearly 100-fold lower than ␥-GT activity measured in control C57Bl/6J mice. Under reaction conditions similar to those used to measure ␥-GT activity in kidney extracts from control mice (25-50 g of protein added to the reaction and 10 min of incubation), ␥-GT activity was frequently undetectable in kidney extracts from glutathionuric mice. Increasing the amount of glutathionuric mouse kidney extract in the reaction to 500 g of protein allowed the measurement of a slight amount of ␥-GT activity. Incubation of the reaction with glutathionuric mouse kidney extract at 37°C for up to 4 h did not result in appreciably higher p-nitroaniline production (data not shown). A 1:1 mixture of kidney extracts from a glutathionuric mouse and from a control mouse had measured ␥-GT activity equaling 50% of specific ␥-GT activity measured in control extracts alone, indicating that kidney extract from glutathionuric mice did not inhibit the ␥-GT reaction when active enzyme was added. Kidney ␥-GT activity was also measured in four mice that had produced glutathionuric offspring and were therefore carriers of the glutathionuric trait. Mean kidney ␥-GT activity in the carriers was 72% of that in control mice.
Immunoblotting-Kidney microsomal fractions from the Harlan Sprague Dawley rat, C57Bl/6J mouse, and GGT enu1 mouse were analyzed for the presence of ␥-GT protein by Western blot analysis using polyclonal rabbit sera raised against purified rat ␥-GT protein. The specific activity of ␥-GT in each kidney microsomal fraction was 557 Ϯ 10.1 nmol/min/mg of protein in wild type C57Bl/6J mouse, 64.6 Ϯ 5.9 nmol/min/mg in GGT enu1 mouse, and 2070 Ϯ 365 nmol/min/mg in the Harlan Sprague Dawley rat. The specific activity of alkaline phosphatase, another renal tubule membrane-associated enzyme, in these microsomal fractions was 788 Ϯ 39.1 nmol/min/mg of protein in the wild type C57Bl/6J mouse (ratio of ␥-GT to alkaline phosphatase activity ϭ 0.71), 839 Ϯ 35.2 nmol/min/mg in the GGT enu1 mouse (ratio ϭ 0.071), and 703 Ϯ 16.6 nmol/ min/mg in the Harlan Sprague Dawley rat (ratio ϭ 2.9). Aliquots containing approximately 10 g of total protein from each microsomal fraction were loaded onto a 10% SDS-polyacrylamide gel. Immunoblotting revealed two densely stained bands with apparent molecular masses of approximately 55 and 35 kDa in the rat kidney microsomal fraction (Fig. 5). This result agrees with the immunoblot of purified rat ␥-GT obtained by Coloma and Pitot (26). Two additional less intensely stained bands with apparent molecular masses of 150 and 25 kDa were also detected in the rat. The 150-kDa band might be the ␥-GT precursor protein which contains both the large and small subunits; the 25-kDa band could possibly be a deglycosylated form of the small subunit (26). Alternatively, these bands could be artefacts. Two polypeptides are detected in wild type mouse kidney microsomal protein with apparent molecular masses of approximately 55 and 25 kDa, but the smaller polypeptide in control mouse stained less intensely than the larger subunit. This could be caused by inefficient detection of mouse ␥-GT protein when using sera raised against rat ␥-GT protein, or the smaller subunit in control mouse could have been partially inadvertently degraded during the isolation procedure. Only a small amount of a polypeptide with 150 kDa apparent molecular mass is visible in the GGT enu1 lane. Neither mature ␥-GT subunit was detected in GGT enu1 kidney in the trial depicted in Fig. 5, but faint traces of each subunit (apparent molecular masses 55 and 35 kDa) were occasionally detected in GGT enu1 kidney on repeated immunoblots. This result is consistent with the small amount of residual ␥-GT activity measured in GGT enu1 kidney homogenate.
Inheritance of Glutathionuria-Autosomal recessive inheritance of glutathionuria in the mice was suspected from pedigree inspection. No vertical transmission of glutathionuria was seen. Ninety-nine glutathionuric mice were detected out of 323 total offspring (30.6%) from 46 matings. Of these, 44 were male   and 55 were female. In an autosomal recessive model, 81 glutathionuric mice would have been expected out of 323 total offspring (25%). This difference between the expected and actual number of affected mice is statistically significant ( 2 ϭ 5.33, p Ͻ 0.05, one degree of freedom) but becomes insignificant if one assumes that the glutathionuric phenotype of only three mice had been assigned incorrectly.
Because glutathionuric mice did not breed, the mutant strain has been maintained using two breeding schemes. First, unaffected female siblings of glutathionuric mice are bred to their carrier fathers. Second, unaffected male siblings of glutathionuric mice are bred to wild type C57Bl/6J females. The female offspring of this mating are backcrossed to their father and the next generation of offspring are then screened for glutathionuria. Using the latter breeding method, the glutathionuric trait (that is, the carrier state for glutathionuria) has been transmitted through 12 generations without any alteration in the glutathionuric phenotype of homozygous mice.

DISCUSSION
The glutathionuric mouse described here is an animal model of genetic ␥-GT deficiency. Glutathionuria was detected by paper chromatography in progeny of mice treated with ENU, a potent chemical point mutagen, and confirmed by high voltage two dimensional electrophoresis, quantitative amino acid analysis and specific spectrophotometric measurement of GSH concentration. Plasma glutathione levels were also elevated in the glutathionuric mice. Given that glutathionuria in association with ␥-GT deficiency has been described in humans (27,28) and in rats treated with chemical inhibitors of ␥-GT (9), ␥-GT deficiency is the most likely cause of glutathionuria in the mutant mice. ␥-GT deficiency in kidney homogenates from the glutathionuric mice was confirmed using a spectrophotometric assay of ␥-GT activity. Western blot analysis of Triton X-100 treated kidney microsomes from glutathionuric mice revealed only a trace of ␥-GT cross reactive material compared with control mouse kidney microsomes. Glutathionuria is inherited in an autosomal recessive manner in this mouse strain. We conclude that this mouse strain, which we designate GGT enu1 , is a model for genetic ␥-GT deficiency.
The progeny of ENU-treated mice may harbor random point mutations at multiple genetic loci. We have not ruled out the possibility that mutations at more than one genetic locus are required for the GGT enu1 phenotype, but the autosomal recessive inheritance pattern suggests that this phenotype is caused by mutation at a single genetic locus. This claim is further supported by the fact that glutathionuric offspring have been produced by twelfth generation progeny of the original ENUtreated mouse. Mice in the 12th generation, because of out-breeding to wild type C57Bl/6J mice, share only 10.6% genetic identity (2 of 20 chromosomes) with first generation mice. In humans, a family of at least four ␥-GT genes exists (29), mapping to chromosome 22q (30). At least two of the human ␥-GT genes are transcribed as the human kidney cDNA differs from the placental and liver cDNAs (31). In rats (32) and mice (33), several different ␥-GT mRNAs with different 5Ј ends but identical coding regions are transcribed from a single gene. In rat kidney, translation of ␥-GT mRNA results in a precursor polypeptide which is subsequently cleaved into two glycoprotein subunits of M r ϳ29,000 and ϳ49,500 (34). Only traces of ␥-GT protein were detected by immunoblotting of kidney microsomal fraction from glutathionuric GGT enu1 mice. These data suggest that glutathionuric GGT enu1 mice are homozygous for a mutation (probably at the ␥-GT locus) which significantly interferes with either the production or stability of active ␥-GT protein.
Recently, the technique of homologous recombination in embryonic stem cells has been used to disrupt the murine ␥-GT gene (35). Mice that are homozygous for the ␥-GT deletion (GGT ml /GGT m1 ) have no detectable ␥-GT activity in kidney, pancreas, small intestine, or seminal vesicles. Plasma glutathione levels in GGT m1 /GGT m1 mice (175 M) are very similar to our results in GGT enu1 mice (213 M), but urinary glutathione excretion is six to seven times greater in GGT m1 /GGT m1 mice (15378 M versus 2261 M in GGT enu1 mice). The growth pattern of GGT m1 /GGT m1 mice mirrors that of GGT enu1 mice, but the life span of knock-out mice (10% survival at 25 weeks of age) is shorter than in GGT enu1 mice (50% survival at 25 weeks of age). Other physical differences include the ubiquitous development of bilateral cataracts and change in coat color from black to agouti in GGT m1 /GGT m1 mice. All these differences may be explained by the presence of slight residual ␥-GT activity in GGT enu1 mice compared with complete ␥-GT deficiency in GGT m1 /GGT m1 mice. The phenotypic similarities between the two mouse strains further supports our claim that GGT enu1 mice are a model for genetic ␥-GT deficiency.
Urine GSH excretion from the GGT enu1 mice was almost 600 times greater than that from control mice (Table II). The mechanism by which ␥-GT deficiency causes glutathionuria has been previously proposed. Two-thirds of the glutathione in plasma is extracted by the kidney (36). It is both filtered through the glomerulus and secreted by the tubule cells into the proximal renal tubule (37). Renal proximal tubule cells exhibit the highest ␥-GT activity of any tissue, and this activity is mainly localized to the cell surface facing the tubule lumen (38). ␥-GT catalyzes the initial step in the degradation of renal tubular GSH by cleaving the ␥-glutamyl bond and transferring the ␥-glutamyl moiety to a broad repertoire of acceptors including many amino acids, dipeptides, water or GSH itself (Fig. 1). The products of the ␥-GT reaction, namely cysteinylglycine and the ␥-glutamyl-acceptor complex, are transported into the renal tubule cells and processed enzymatically to produce the amino acids cysteine, glycine, and glutamate. Resynthesis of GSH from these three amino acids completes the proposed ␥-glutamyl cycle (5). Without ␥-GT activity in the brush border membrane, GSH cannot be recovered from glomerular filtrate and is lost in urine. Glutathionuria in GGT enu1 mice confirms the previous experience in humans with genetic ␥-GT deficiency and in rodents with chemically induced ␥-GT deficiency, demonstrating that ␥-GT activity is required for reabsorption of GSH from the urine.
Plasma GSH concentration in the GGT enu1 mice was increased 5-fold as compared with control mice (Table II). Chemical inhibition of ␥-GT activity in mice and rats also increases plasma GSH levels (9,39). In a ␥-GT-deficient person, plasma GSH levels were increased to 4.2-7.3 mol/liter above control values of 1-1.6 mol/liter (27). A second patient also exhibited an elevated GSH plasma level of 19.4 mol/liter (28). Plasma GSH levels are approximately 10-fold higher in ␥-GT-deficient mice than in ␥-GT-deficient humans, possibly suggesting greater residual ␥-GT activity in affected humans, but plasma GSH levels in control mice are also substantially higher than in human controls (ϳ25 versus ϳ1 mol/liter, respectively). Species-specific variation in normal GSH physiology among humans and mice most likely accounts for the species-specific differences in GSH levels in normal and ␥-GT-deficient individuals. The redundancy and tissue-specific expression of human ␥-GT genes could be responsible for the phenotypic differences between ␥-GT-deficient mice and humans.
GSH levels were altered in other tissues of GGT enu1 mice. Total liver GSH content in GGT enu1 mice was decreased 66% compared with control animals. In mice treated with the ␥-GT inhibitor D-␥-glutamyl-(o-carboxy)phenylhydrazide, total liver GSH content was decreased by only 17%, probably because of incomplete or temporary ␥-GT inhibition (9). An 80% decrease in liver GSH was measured in GGT m1 /GGT m1 mice (35). One possible explanation for decreased liver GSH content in ␥-GT deficiency might be that failure of GSH recovery from urine leads to systemic deficiency of GSH precursors and limited liver GSH synthesis. Of the three amino acids which constitute GSH (glutamate, glycine, and cysteine), only the supply of cysteine is likely to be a limiting factor in GSH synthesis. An alternative explanation for liver GSH deficiency is impaired recovery of GSH from bile due to ␥-GT deficiency affecting biliary epithelium.
It has been proposed that renal brush border ␥-GT catalyzes transpeptidation with luminal amino acids, particularly cystine, and that uptake of ␥-glutamyl-amino acids from the tubule lumen is an important pathway for amino acid transport in the kidney (8). In GGT enu1 mice, urinary threonine, glycine, cystine, isoleucine, leucine, ornithine, and lysine concentrations were slightly increased as compared with controls (Table  I). Cystine is an excellent ␥-glutamyl acceptor (40), and a decrease in recovery of cystine from the renal tubule due to ␥-GT deficiency could explain increased urinary cystine excretion in GGT enu1 mice. Glycine and lysine are also reasonable ␥-glutamyl acceptors, but other amino acids which are active ␥-glutamyl acceptors, most notably glutamine, are not elevated in urine from GGT enu1 mice. Amino aciduria can be a manifestation of generalized proximal renal tubule dysfunction (renal Fanconi syndrome), but histologic examination of kidneys from GGT enu1 mice revealed no abnormalities, and there is no evidence of renal losses of other metabolites such as glucose. These data support but do not prove the theory that GSH and ␥-GT play a nonexclusive role in the reabsorption of amino acids from the renal tubule. GGT enu1 mice will be a valuable experimental model for the study of ␥-GT and its relationship to amino acid transport in the kidney.
If ␥-GT deficiency does impair amino acid reabsorption in the renal tubule, disturbed amino acid flux could be responsible for the poor linear growth and weight gain of GGT enu1 mice. Systemic cysteine deficiency due to chronic urinary loss of GSH could also contribute to poor growth. Levels of free cystine and other amino acids are normal in plasma and tissue homogenates from GGT enu1 mice, but these data may not accurately reflect the status of intracellular amino acid pools or interorgan amino acid flux. Decreased urinary excretion of taurine, an amino acid derived from cysteine, in GGT enu1 mice may reflect a relative deficiency of cysteine or other sulfur-containing amino acids. GGT enu1 mice excreted 40 -45 mol of GSH/day based upon an average daily creatinine excretion of 0.6 mg/ creatinine/day in normal mice (41). If a normal mouse consumes 5 g of mouse chow (Teklad Mouse Breeder Diet 8626) per day, the average daily intake of cysteine equals 130 mol of cysteine per day. This cysteine intake may not be adequate to replace urinary GSH losses and provide for normal cysteine requirements. Plasma cystine deficiency was detected in GGT m1 /GGT m1 mice; the growth velocity of GGT m1 /GGT m1 mice was almost completely corrected when the mice were given N-acetylcysteine supplements as a source of extra cysteine (35). Studies to determine the effect of cysteine supplementation on the growth of the GGT enu1 mice are in progress. No significant problems with appetite, linear growth, or weight gain were reported in any of the glutathionuric humans.
GGT enu1 mice exhibit neurologic abnormalities including decreased general activity but with agitation and tremor when stimulated. Histological examination at the light microscopic level of their brains did not reveal any abnormalities. Impaired GSH and amino acid transport in the brain could disrupt neurotransmitter or other pathways. Three unrelated humans with ␥-GT deficiency had mild mental retardation and one of them had severe behavioral problems (27,28,42). Recently, two female siblings from Australia have been reported with ␥-GT deficiency (43). The older sister was discovered at 6 weeks of age to have glutathionuria through a total population screening program examining urine amino acids by paper chromatography. Her health, growth, and development have been normal except for an absence seizure disorder which developed at 10 years of age. Her younger sister has ␥-GT deficiency but no seizure disorder. Her mild mental retardation has been attributed to Prader-Willi syndrome associated with an interstitial deletion of chromosome 15q11-13. Although the causal relationship between the abnormal neurologic signs and ␥-GT deficiency remains uncertain in humans, our findings in the GGT enu1 mice indicate that ␥-GT is required for normal neurologic function.
In summary, GGT enu1 mice provide convincing evidence that ␥-GT function is required for normal growth and neurologic function. Further studies to elucidate the mechanism by which ␥-GT deficiency leads to the physical and neurologic abnormalities in GGT enu1 mice will enable the metabolic and cellular roles of ␥-GT to be better defined. The GGT enu1 mouse also provides a valuable experimental platform for the study of ␥-GT and GSH metabolism in response to infection, environmental toxins, carcinogenic agents, and other disease processes.