γ-Glutamyl Leukotrienase, a γ-Glutamyl Transpeptidase Gene Family Member, Is Expressed Primarily in Spleen*

We have recently identified a mouse enzyme termed γ-glutamyl leukotrienase (GGL) that converts leukotriene C4 (LTC4) to leukotriene D4(LTD4). It also cleaves some other glutathione (GSH) conjugates, but not GSH itself (Carter, B. Z., Wiseman, A. L., Orkiszewski, R., Ballard, K. D., Ou, C.-N., and Lieberman, M. W. (1997) J. Biol. Chem. 272, 12305–12310). We have now cloned a full-length mouse cDNA coding for GGL activity and the corresponding gene. GGL and γ-glutamyl transpeptidase constitute a small gene family. The two cDNAs share a 57% nucleotide identity and 41% predicted amino acid sequence identity. Their corresponding genes have a similar intron-exon organization and are located 3 kilobases apart. A search of Genbank and reverse transcription-polymerase chain reaction analysis failed to identify additional family members. Mapping of the GGL transcription start site revealed that the GGL promoter is TATA-less but contains an initiator, a control element for transcription initiation. Northern blots for GGL expression were negative. As judged by ribonuclease protection,in situ hybridization, and measurement of enzyme activity, spleen had the highest level of GGL expression. GGL is also expressed in thymic lymphocytes, bronchiolar epithelial cells, pulmonary interstitial cells, renal proximal convoluted tubular cells, and crypt cells of the small intestine as well as in cerebral, cerebellar, and brain stem neurons but not in glial cells. GGL is widely distributed in mice, suggesting an important role for this enzyme.

Glutathione conjugates play a central role in normal physiology and in responses to injury (1)(2)(3)(4)(5)(6). Among the more important GSH derivatives are conjugates of eicosanoids, xenobiotics, and carcinogens. At least three types of eicosanoid-GSH conjugates have been identified, including derivatives of leukotriene A 4 , prostaglandins, and hepoxilins. The LTA 4 -GSH conjugate (LTC 4 ) 1 and its cleavage products, LTD 4 and LTE 4 , are powerful mediators of bronchoconstriction, vasoconstriction, mucus formation, and edema. As a result, they are important mediators of asthma, coronary artery spasm, and nephropa-thies (7)(8)(9)(10)(11)(12)(13)(14)(15). Prostaglandins play diverse biological roles and are involved in the development of the inflammatory response, inhibition of platelet aggregation, and regulation of immune responses (6). Prostaglandins are inactivated and cleared by conjugation to GSH (3,6). Hepoxilin A 3 forms a GSH conjugate, hepoxilin A 3 -C, that is known to be a potent regulator of hippocampal neurons (4). In addition, several neurotransmitters, including serotonin and dopamine, form GSH conjugates, suggesting an additional role for this pathway in the central nervous system (16,17). GSH conjugation along with glucuronide formation is the major pathway by which toxins, drugs, and carcinogens such as CH 3 Hg, acetaminophen, and aflatoxin are detoxified and excreted (5, 18 -21).
Until recently, it was thought that the sequential cleavage of GSH conjugates and their derivative cysteinyl-glycine conjugates were catalyzed by two enzymes. GGT was known to cleave a ␥-glutamyl group from GSH conjugates, and a dipeptidase, often identified as membrane-bound dipeptidase, was believed to cleave the Cys-Gly bond of conjugates to yield Cys derivatives (2). We have recently developed mice deficient in GGT and have shown that tissues from these mice will not cleave GSH, but they retain the ability to metabolize LTC 4 to LTD 4 , its Cys-Gly derivative (22,23). We have partially purified this activity and termed it ␥-glutamyl leukotrienase (GGL) to reflect this fact (23); however, the enzyme will also cleave S-decyl GSH, other alkyl GSH derivatives, and S-nitrosyl GSH (23,24). A human cDNA coding for an enzyme termed GGT-rel shares at least some properties with GGL (25). While we were preparing this manuscript for publication, a report identifying the rat homologue of human GGT-rel appeared (26).
In other experiments, we have produced membrane-bound dipeptidase-deficient mice and found that tissues from these mice are capable of metabolizing LTD 4 to LTE 4 , the Cys derivative (27). Thus, like the cleavage of GSH conjugates, more than one enzyme is capable of catalyzing the second step of this pathway.
A new picture is emerging in which a variety of enzymes participate in the biotransformation of GSH conjugates. At present, it is unknown how many enzymes are involved in each step and what role each plays in the regulation of these important biological molecules. To begin to examine this pathway in more detail, we have cloned a full-length mouse GGL cDNA and the corresponding gene and examined its expression in mouse tissues.

EXPERIMENTAL PROCEDURES
Materials-GGT-deficient mice were developed in our laboratory (22). All restriction enzymes were purchased from Boehringer Mannheim, and [␣- 32 4 , and other chemicals were purchased from Sigma unless otherwise indicated.
RNA Isolation-Total RNAs were isolated from various GGT-deficient mouse tissues following the acid guanidinium thiocyanate proce-* This work was supported by National Institutes of Health Grant ES-08668. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF077765 and AF080593.
To obtain the GGL cDNA sequence further downstream of cDNA I, spleen poly(A) ϩ RNA from GGT-deficient mice was reverse transcribed using a lock-docking cDNA synthesis primer obtained from the Marathon cDNA amplification kit (CLONTECH). 3Ј-RACE fragments were obtained by PCR amplification of the cDNA using primers AP1 and GGL1 (Fig. 1). The products were then cloned into pT7Blue(R)T-Vector and sequenced. The clone containing the polyadenylation signal was selected and further amplified using primers GGL2 and GGL3 ( Fig. 1) to obtain GGL cDNA II. To obtain the GGL cDNA sequence further upstream of cDNA I, spleen poly(A) ϩ RNA from GGT-deficient mice was reverse transcribed using primer GGL4 (Fig. 1). 5Ј-RACE fragments were amplified using AP2 from the Marathon cDNA amplification kit and GGL4. The products were then cloned and sequenced as above. The clone with the longest 5Ј end was chosen and further amplified by PCR using primers GGL5 (Fig. 1) and GGL4 to obtain GGL cDNA III.
To obtain a full-length GGL cDNA, GGL cDNA II and cDNA III were fused and amplified by PCR using primers GGL5 and GGL3 (Marathon cDNA amplification kit as described by the manufacturer). A full-length GGL cDNA was cloned into pT7Blue(R)T-Vector, and six clones were sequenced. The sequence of GGL cDNA was chosen by comparing six GGL cDNA sequences (derived from PCR cloning) and the exon sequence of the gene (isolated from genomic library, see below).
All sequencing was carried out by an automated Applied Biosystems 373 DNA sequencer using a Dye Terminator Cycle Sequencing Ready Reaction DNA sequencing kit (ABI-Perkin Elmer). Comparison and alignment of GGL cDNA with other cDNAs and homology analyses of DNA and amino acid sequences were accomplished with the IntelliGenetics (IG) Suite (IntelliGenetics Inc., Palo Alto, CA).
In Vitro Transfection of Murine NIH/3T3 Cells with Full-length GGL cDNA-Murine NIH/3T3 cells were maintained in Dullbecco's modified Eagle's medium with 10% fetal bovine serum. Full-length GGL cDNA was released from pT7Blue(R)T-Vector and subcloned into a pCMV vector (CLONTECH). The construct was cotransfected into NIH/3T3 cells with pGK-neo using the calcium phosphate precipitation method, and the stable transfectants were selected in the presence of geneticin as described (29). LTC 4 to LTD 4 converting activity (GGL activity) and GSH cleavage were assayed by high performance liquid chromatography methods (23). The GGL enzyme activities for organs presented in this paper (Table I) are higher than those presented earlier (23) for which we used material frozen at Ϫ80°C. All assays reported in this paper are carried out on homogenates of fresh organs. ␥-Glutamyl-pnitroanilide cleaving activity was assayed as described previously (22).
GGL Gene Structure Identification-GGL genomic clones were isolated from a mouse FixII genomic library (Stratagene) screened with GGL cDNA containing the full-length coding region as described previously (30,31). The intron-exon organization of the gene was determined by the same procedures as used for the mouse GGT gene (30) and mouse GSH synthetase gene (31).
Primer Extension Analysis-26 g of mouse spleen poly(A) ϩ RNA was used for primer extension analysis. The primer was GGL6 (5ЈCT-GGACAGAAAGGATGACGGACAGCCTTACAGATAGGTGG3Ј).
The reaction was carried out essentially as described previously (32).
Detection of GGL RNA Expression in Mouse Tissues-Expression of GGL RNA in various mouse tissues was detected by Northern blot (31) and ribonuclease protection assay (Ambion). To obtain the probe for ribonuclease protection, a GGL cDNA fragment stretching from base Ϫ28 to base 173 ( Fig. 1) was cloned into the pT7Blue(R)T plasmid. The plasmid was linearized with BamHI and transcribed by T7 RNA po-lymerase in the presence of [␣-32 P]UTP using an in vitro transcription kit (Stratagene). 100 g of RNA from tissues of GGT-deficient mice (spleen, small intestine, kidney, liver, lung, and brain) and 2.0 ϫ 10 5 cpm of GGL probe were used for each assay following the manufacturer's instruction. Hybridization was carried out at 42°C for 18 h followed by RNase digestion at 37°C for 1 h with a 100-fold dilution of RNase R solution in the kit.
In Situ Hybridization-A GGL cDNA fragment stretching from base Ϫ28 to base 254 ( Fig. 1) was cloned into the pT7Blue(R)T plasmid in both orientations. To generate antisense and sense probes, the plasmids were then linearized by BamHI and transcribed by T7 RNA polymerase in the presence of digoxigenin-labeled UTP (Boehringer Mannheim). Sections of paraffin-embedded organs from wild-type and GGT-deficient mice were incubated on slides with 0.001% proteinase K for 20 min at 37°C. After acetylation with 0.25% acetic anhydride for 5 min at room temperature (to block positive charges on tissue induced by protease digestion), the sections were dehydrated and prehybridized for 1 h at 45°C in a hybridization buffer consisting of 50% deionized formamide, 2 ϫ standard SSC solution, 20 mM Tris(hydroxymethyl)aminomethane, pH 8.0, 1 ϫ Denhardt's solution, 1 mM ethylenediaminetetraacetic acid, 100 mM dithiothreitol, and 0.5 mg/ml yeast tRNA. The sections were then incubated with an antisense riboprobe (or a sense probe as a negative control) at 45°C overnight followed by successive incubations in 2 ϫ SSC for 1 h, 1 ϫ SSC for 30 min at room temperature, 0.5 ϫ SSC with RNase A at 65°C for 30 min, and 0.5 ϫ SSC for 1 h at room temperature. The tissues were incubated with a polyclonal sheep antidigoxigenin antibody labeled with alkaline phosphatase. No differences were seen in patterns of in situ hybridization in GGT-deficient and wild-type mice. In all cases, hybridizations with the sense strand were negative.
Exposure of Mice to Oxygen-GGT-deficient mice (22) and their wildtype littermates were exposed to 80% oxygen continuously for 120 h as described previously with minor modifications (33). They were sacrificed by metofane inhalation, then cervical dislocation. RNA was isolated from lungs of these mice for Northern blotting (31). Lung homogenates from oxygen-exposed and air-breathing control GGT-deficient mice were used to assay GGL activity (23).

RESULTS
Isolation and Analysis of Mouse GGL cDNA-As an initial step in examining the pathway(s) involved in the biotransformation of GSH conjugates, we cloned a full-length GGL cDNA by an RT-PCR strategy using poly(A) ϩ RNA from the spleen of GGT-deficient mice and primers for GGT-rel, a human gene product that shares some substrate specificity and properties with GGL (22,23,25). Although a search for genes related to human GGT-rel in the mouse by Northern blotting was negative (25), we were able to detect LTC 4 to LTD 4 conversion (GGL activity) in GGT-deficient mice (23). We reasoned that because GGL and GGT-rel shared some functions, it was likely that GGL would be at least in part homologous with GGT-rel. We isolated a 2.3-kb GGL cDNA containing an open reading frame of 1719 nucleotides, a 5Ј UTR of 314 nucleotides, and a 3Ј UTR of 284 nucleotides followed by the polyadenylation site, which is 18 nucleotides upstream of the poly(A) tail. The GGL cDNA and its deduced amino acid sequences are shown in Fig. 1. The cDNA shares an overall 68% nucleotide sequence identity with human GGT-rel cDNA (25) and an 89% identity with the recently identified rat homologue (26). The predicted amino acid sequence shows 77% identity with that of human GGT-rel (Fig.  2) and 88% identity with the rat sequence (26). The predicted heavy chain of mouse GGL contains 388 amino acids compared with 387 amino acids for human GGT-rel, whereas the predicted light chain contains 185 amino acids compared with 199 amino acids for human GGT-rel (Figs. 1 and 2 and Ref. 25). Closer inspection of the two cDNAs revealed that a stretch of 48 nucleotides in the region coding for the light chain of GGT-rel (corresponding to amino acids 446 -461 of human GGT-rel in Fig. 2) was missing in GGL, which would account for the somewhat smaller predicted size of the GGL protein (573 amino acids) compared with the human GGT-rel protein (586 amino acids). The deletion of 48 nucleotides compared with human GGT-rel sequence also occurred in the rat GGT-rel cDNA sequence, which predicts 572 amino acids (26). From the putative structure of the rat protein, the histidine at position 371 of mouse GGL is deleted in rat GGT-rel ( Fig. 2 and Ref. 26). GGL has six putative N-glycosylation sites, all located in the putative heavy chain (Fig. 1).
We also found that GGL and mouse GGT share an overall 57% cDNA sequence identity and 41% predicted amino acid sequence identity (Fig. 2). The predicted amino acid identity of mouse GGL and GGT sequence is highest in three regions confirm that the cDNA we cloned from GGT-deficient mouse spleen codes for GGL activity, we transfected the cloned fulllength cDNA under the regulation of the cytomegalovirus promoter into murine NIH/3T3 cells. Normally, these cells do not express either GGL or GGT. We isolated three clones with enzyme activity, thus demonstrating that this cDNA codes for GGL (Fig. 3). The specific activities of the three clones for conversion of LTC 4 to LTD 4 were 1.93, 3.48, and 4.10 mol/g of protein/h, respectively. In addition to LTD 4 , we detected LTE 4 in our reactions (Fig. 3). The appearance of this leukotriene results from the cleavage of LTD 4 to LTE 4 by dipeptidase(s) in NIH/3T3 cells (23,34). We were unable to demonstrate cleavage of GSH or ␥-glutamyl-p-nitroanilide (a substrate used to assay GGT) by transfected 3T3 cells (data not shown); this result confirms our earlier findings with partially purified enzyme preparations from GGT-deficient mice (23).
Genomic Structure of Mouse GGL-A Southern blot of mouse DNA digested with EcoRI, HindIII, both EcoRI and HindIII, and BamHI and probed with a 0.37-kb GGL cDNA fragment yielded single bands (data not shown), indicating that GGL is a single copy gene. Subsequently, from 5 ϫ 10 5 phages of a mouse genomic FixII library, we isolated seven individual clones that hybridized with the cDNA probe. These were mapped with restriction endonucleases and subcloned. Intron-exon boundaries were identified by sequencing the subclones that hybridized to the cDNA probe and comparing their genomic sequences with the GGL cDNA sequence. As a result, we found that GGL has 12 exons spanning approximately 26 kb (Fig. 4). The average length of exons is 190 bp, whereas that of introns is about 2.2 kb. All the intron-exon boundaries follow the GT/AG rule of splice site consensus (35). Interestingly, the intron-exon organization of GGL is almost identical to that of GGT with all the splice sites of GGL being consistent with those of GGT except for intron 9, which is spliced 6 bp (2 amino acids) upstream of the site corresponding to that of GGT (Fig. 4).
Analysis of GGL RNA-To determine if GGL and GGT are similar in their 5Ј UTR, we examined the 5Ј UTR of mouse GGL cDNA by the RACE technique. We examined 46 clones containing the 5Ј UTR: 30 from kidney, 6 from liver, and 10 from spleen, and found only one 5Ј UTR that is part of first coding exon of GGL (Fig. 1). In contrast, the mouse GGT gene contains at least nine 5Ј-noncoding exons that generate seven types of RNA with different 5Ј UTRs (32, 36 -38). These findings suggest that regulation of mouse GGL transcription is different from that of mouse GGT (39).
We also found a shorter GGL RNA in spleen, small intestine, kidney, liver, and lung. Subsequent sequence analysis showed that its cDNA lacks exon 10 (data not shown), a deletion that should produce a shift in the reading frame and termination at amino acid 456. This shorter RNA could only be detected after two rounds of PCR; therefore, at most, it represents a very small portion of total GGL RNA and probably lacks physiological significance.
Sequence of the 5Ј-Flanking Region and Determination of Transcription Start Site-The DNA sequence containing part of GGL 5Ј UTR and its 5Ј flanking region is presented in Fig.  5A. The transcription start site of GGL was determined by primer extension and ribonuclease protection analyses. For primer extension, mouse spleen poly(A) ϩ RNA was used with yeast tRNA as a negative control. The primer GGL6 for reverse transcription is located within the 5Ј UTR of the GGL gene (Fig. 5A, bases ϩ111 to ϩ150 corresponding to bases Ϫ180 to Ϫ219 in Fig. 1). A product, 150 nucleotides in length, was found (Fig. 5B), indicating the 5Ј UTR of GGL consists of 329 nucleotides (150 ϩ 179). The result was confirmed by ribonuclease protection assay (Fig. 5C). The probe for ribonuclease protection assay was transcribed from a subclone containing GGL DNA sequence from bases Ϫ110 to ϩ150 (Fig. 5A). The conditions for the reaction were the same as for the ribonuclease protection assay for detecting GGL RNA expression (see "Experimental Procedures") except that a 50-fold dilution of RNase R solution was used. The protected fragment (Fig. 5C) seems slightly longer than the expected 150 nucleotides. This appar- ent extra length is the result of comparing the mobility of RNA with DNA markers. The transcription start site is marked as ϩ1 (Fig. 5A). The start site and its surrounding sequence (bold in Fig. 5A) match the PyPyAϩ1NT/APyPy consensus sequence for an initiator (40).
A TATA box was not found in GGL promoter region. However, an A/T-rich sequence is located 14 -26 bp upstream of the transcription start site (Fig. 5A). In addition, AP2 and CCAAT boxes and three tandem copies of SP1 binding sites are present in the putative GGL promoter region (Fig. 5A).
GGL and GGT Are Members of the Same Gene Family-Because of the similarity between GGL and GGT, we reasoned that one might be derived from the other by gene duplication and divergence, and they might be tightly linked. Therefore, we examined our mouse genomic GGT clones (30) and found that one of them (gtA) contained the 5Ј end of GGL about 3 kb 3Ј of GGT.
We used two strategies to search for additional GGT/GGL family members. First, we used the peptide sequences of GGT and GGL to search Genbank ("BLAST P" version 2.0.4, National Center for Biotechnology Information) but found no significant sequence homology to these two proteins. We also tried an approach based on RT-PCR and degenerate primers (41). Briefly, three pairs of degenerate oligonucleotides corresponding to three conserved regions of the amino acid sequences (amino acids 68 -73, 154 -159, and 404 -409 of GGL; see Fig. 2) were used to amplify cDNAs reverse transcribed from RNAs of liver and kidney of GGT-deficient mice. The PCR products were subsequently cloned and sequenced, but again, we failed to identify any additional family members.
Detection of GGL RNA Expression in Mouse Tissues-To determine the level of GGL expression in mouse tissues, we used Northern blot analysis and ribonuclease protection assays. Northern blot analysis failed to detect any GGL signal from spleen, small intestine, kidney, liver, and lung, suggesting that GGL is expressed at low levels or not at all in these organs (data not shown); These findings are in agreement with data on human and rat GGT-rel (25,26). When more sensitive ribonuclease protection assays were performed, a protected fragment of the expected size was obtained from most organs (Fig. 6). Spleen expressed the highest level of GGL RNA, whereas small intestine, kidney, lung, and brain expressed lower levels. Additional ribonuclease protection experiments, semi-quantitative RT-PCR (results not shown), and enzyme assays all indicated that spleen expresses the highest level of GGL (Table I and Ref. 23).
Localization of GGL Expression by in Situ Hybridization-To study the distribution of GGL more specifically, we used in situ hybridization to examine GGL expression in tissues of GGTdeficient mice (Fig. 7, Table I). Enzyme activity data as another measurement of GGL expression are also presented in Table I.
Almost all lymphocytes present in the spleen were positive by in situ hybridization (Fig. 7A). Thymus was positive in the medulla and paracortical region (Table I). Although we did not determine the phenotype of these lymphocytes, the locations of GGL-positive cells and their abundance in these organs suggest that both T and B cells express GGL. Interestingly, thymus expressed much lower GGL enzyme activity than spleen (Table I), although both tissues stained strongly by in situ hybridization. These findings suggest that GGL expression may be regulated post-transcriptionally in different tissues.
In the lung, all columnar bronchiolar epithelial cells and numerous interstitial cells in the alveolar septae were positive ( Fig. 7B and Table I), but it was not possible to determine, unambiguously, the lineage of these interstitial cells. Chondrocytes in bronchial cartilage were also positive. Some proximal pulmonary arteries in both GGT-deficient and wild-type mice contain cardiac myocytes (a normal finding in mice), which stained positively for GGL. Although a great number of cells were stained positively by in situ hybridization and a signal was detected by ribonuclease protection assay, lung expressed very low GGL activity (Table I).
In situ hybridization on kidney was positive in proximal and distal tubules as well as in some collecting ducts (Fig. 7C).
Other structures, such as glomeruli and blood vessels, were negative. The crypt cells of the small intestine (Fig. 7D) were positive, whereas villous cells showed only focally positive stain. The myocardium was positive (Table I). In the liver, a few isolated periportal hepatocytes and lymphocytes in the periportal region were positive, whereas other hepatocytes and other cells were negative. It is likely that the small percentage of GGL-positive cells in liver is the explanation for our inability FIG. 6. Ribonuclease protection assay of GGL expression in mouse tissues. RNA (100 g) from GGT-deficient mouse tissues and 2 ϫ 10 5 cpm of GGL probe were used. Hybridization was carried out at 42°C for 18 h. t, tRNA as a negative control; Sp, spleen; SI, small intestine; K, kidney; Li, liver; Lu, lung; Br, brain. MW, molecular weight marker (pBR322 DNA-MspI digest). I, undigested probe; II, protected fragment. (See "Experimental Procedures" for additional detail).  Glial cells, white matter, and vessels are negative. All sections were also hybridized to digoxigenin-labeled sense probe; no reaction product was seen (all sections were photographed at 70ϫ magnification and were not counterstained). Tissues were from GGT-deficient mice. to detect expression by ribonuclease protection (Fig. 6) and low level of GGL activity by enzyme assay (Table I).
In the brain, all cortical neurons, including those of the hippocampus and basal ganglia, and brain stem neurons were positive (Table I). The signal was confined to neuronal bodies. In cerebellum, Purkinje cells were strongly positive, whereas granular neurons were weakly positive. In contrast, white matter and glial cells were negative (Fig. 7, E and F). Endothelial cells were also negative.
Expression of GGL in Mouse Lung after Oxygen Exposure-Increased levels of rat GGT-rel RNA have been reported in lung after chronic exposure to isobutyl nitrite and in tumors induced by this agent (26). To examine the effects of other agents on the induction of GGL expression, we exposed wild-type and GGTdeficient mice to 80% oxygen for 120 h and performed Northern analysis of RNA from lungs of these mice and air breathing controls. We were unable to detect any GGL RNA in either O 2 -exposed or control mice. Failure to find elevated levels of GGL RNA after oxygen exposure is not the result of generalized tissue damage since both wild-type and GGT-deficient mice are responsive to oxygen injury as judged by elevated levels of ␥-glutamyl cysteine synthetase RNA in exposed mice. 2 We also examined GGL enzyme activity in GGT-deficient mice exposed to O 2 and their air-breathing controls. No apparent difference of GGL activity in lung was found between O 2 -exposed GGTdeficient mice and their air-breathing controls (results not shown). DISCUSSION We have cloned a mouse cDNA (Fig. 1) that encodes GGL activity (Fig. 3). Its high sequence identity with human and rat GGT-rel ( Fig. 2 and Ref. 26) suggests that GGT-rel is the human/rat homologue of GGL (23,25,26).
As judged by experiments with 3T3 cells carrying a fulllength GGL cDNA, mouse GGL cleaves GSH conjugates (LTC 4 ) ( Fig. 3) but not GSH itself (or ␥-glutamyl-p-nitroanilide) (see also Refs. 23 and 24). We emphasize this point because the function of GGL seems to be the cleavage of these conjugated GSH molecules and not the cleavage of GSH itself (25,26).
Our data also demonstrate that the mouse GGL gene and the mouse GGT gene are related by nucleotide sequence and are located within a few kilobases of one another (Fig. 4). Thus, GGL and GGT constitute a small gene family in which one gene probably evolved from the other by gene duplication. It may be that GGL is descended from GGT as a need arose for more selective cleavage of conjugated glutathiones in higher organisms. A search for other family members using Genbank and PCR strategies did not yield additional candidates. At present, we cannot rule out the possibility that other family members exist since their sequences may be too divergent to be identified by our methods.
Although we have not investigated the regulation of GGL expression in detail, we have used the RACE technique to look for a variety of 5Ј ends as evidence of different promoters or genes. The fact that all the GGL cDNAs we have identified to date in kidney (30 clones), liver (6 clones), and spleen (10 clones) have the same 5Ј UTR suggests that GGL is expressed from one or a limited number of promoters. In contrast, the selective expression of GGT in different tissues in the mouse is governed in part by at least seven independent promoters (39). Hence, the two genes have adopted different regulatory strategies in the mouse.
GGL is expressed in spleen at high levels relative to other mouse organs (Figs. 6 and 7, Table I). We do not know what role, if any, GGL plays in immune function, but it is worth noting that relative to other organs, GGT expression in mouse spleen is relatively low (23). In fact, about 90% of the cleavage of GSH conjugates (as measured by LTC 4 cleavage) is carried out by GGL in this organ (23).
Under normal circumstances, GGL is expressed at very low levels in most organs. The expression of GGL in bronchiolar epithelium and the interstitial cells of the lung is consistent with the role leukotrienes are believed to play in normal lung function and asthma; however, exposure to high levels of O 2 for 5 days did not result in detectable increases in steady-state levels of GGL RNA or in increases in GGL enzyme activity in lung. In contrast, other investigators have found elevated levels of rat GGT-rel RNA in lungs from rats exposed for 2 years to isobutyl nitrite (26). GGL is expressed in the proximal tubules of the kidney, which is the site of highest GGT expression (Table I and Refs. 23 and 39). GSH cleavage is an important function of GGT in kidney; as mentioned above, GGL does not cleave GSH. In addition, GGT is able to cleave all known substrates for GGL (23,24). These observations suggest that GGL may have an as yet unrecognized function in kidney. In the brain, GGL expression is confined to neurons. Virtually all cerebral, cerebellar, and brain stem neurons express GGL as judged by in situ hybridization, whereas glial cells, vascular elements, and endothelial cells are GGL-negative (Fig. 7, Table  I). At present, the meaning of this observation remains obscure.
In summary, GGL is a member of the GGT gene family and cleaves a subset of molecules cleaved by GGT. Highest levels of expression are found in spleen. The wide distribution of GGL in many organs suggests an important function for this enzyme.