N-terminal Extension of Canine Glutamine Synthetase Created by Splicing Alters Its Enzymatic Property*

It was found that an extra exon exists in the first intron of glutamine synthetase gene, generated by means of alternative splicing (Shin, D., Park, S., and Park, C. (2003) Biochem. J. 374, 175–184). Inclusion of this exon decreased the translation of glutamine synthetase (GS) in human, dog, and mouse. When translated in vitro with the canine GS transcript containing the exon, we obtained two different species of GS enzymes. Besides the known 45-kDa protein, the extended form of GS was identified with additional 40 amino acids on its N-terminal end. An upstream ATG in the extra exon served as a translation initiator for the long form of GS. When the long transcript was translated in vivo in animal cells, only the long GS was expressed. On the other hand, the long GS is less predominant relative to the short one in canine tissues including brain and liver. Subcellular fractionation of canine brain revealed that the long GS is present in all cellular compartments as is the short one, which is consistent with fluorescence mi-croscopy data obtained with green fluorescent protein fused to GS. The short (SGS) and long (LGS) forms of canine GS were purified in Escherichia coli and shown to have similar K m values for L -glutamate and hydroxyl- amine. However, the K m values for ATP were slightly altered, 1.3 and 1.9 m M for the short and long GSs, re- spectively. The K i s for L -methionine- S -sulfoximine

Glutamine synthetase (GS 1 ; EC 6.3.1.2; L-glutamate ammonia ligase) is an enzyme that catalyzes the ATP-dependent conversion of glutamate and ammonia into glutamine, and thus plays a critical role in eliminating the excitotoxic glutamate in animal brains (1). The expression of eukaryotic GS is regulated at transcriptional and post-transcriptional levels and becomes unstable with oxidation mediated by metal or free radical peptides, generated from a fragmentation of ␤-amyloid, and by growth hormone or glutamine (2).
The mammalian GS protein has been reported to form an eight-subunit oligomer (3) with unknown three-dimensional structure. The bindings of chloride and manganese/magnesium ions to allosteric and activator sites of GS, respectively, cause changes in GS conformation, without affecting its oligomerization (4). As an essential trace metal in vivo, the manganese ion is mostly (ϳ80%) bound to GS in astrocytes (5), although its concentration is variable in other mammalian tissues. Magnesium is also bound to mammalian GS in vivo (6). L-methionine-S-sulfoximine (MSOX), a structural analogue of L-glutamate, is converted to MSOX-phosphate, mimicking the tetrahedral intermediate formed between an enzyme-bound ␥-glutamyl phosphate and ammonia at the active site of GS (7). This reaction product becomes an irreversible and noncovalent inhibitor of the enzyme. MSOX was originally isolated from nitrogen chloride-treated zein and characterized as a toxin causing induction of convulsions, hysteria, and epileptic fits in a number of animals (8). The K i values of MSOX are 105, 1, 161, and 100 M for GS proteins of sheep, Escherichia coli, pea, and spinach, respectively (8). Despite its crucial role in many neurological diseases (9 -13), information on mammalian GS is rather scarce, and inconsistencies are found in various reports on its mass and subunit arrangements (8).
In vertebrate, compartment-specific GS isozymes are produced from a single gene and targeted to either mitochondria or cytosol in a tissue-specific manner (14). The N terminus of GS is variable in size and sequence and serves as a subcellular targeting signal (15). Besides the known 44-kDa protein, the GS-like protein of 54 kDa was reported from human brain. This enzyme is mostly found in crude mitochondrial fractions and possesses higher hydroxylamine-L-glutamine transferase activity than the 44-kDa GS (16,17).
Modulation of the translational efficiency of mammalian GSs by 5Ј transcript leader region has been suggested as a mechanism for regulating its expression (18). This phenomenon was studied in detail with the canine GS, a 45-kDa enzyme with 373 amino acids, which is translated from the gene of ϳ10 kb in length, being organized into seven exons and six introns. An extra exon is found in the first intron of glutamine synthetase gene, which is subjected to an alternative splicing. The long transcript with extra 5Ј-UTR is translated less efficiently than the short one. This translational regulation is partially due to an abortive initiation at the upstream ATG located in the extra exon (18). It is known that the 5Ј transcript leader regions in most vertebrate mRNAs are less than 100 nucleotides long (19). However, two thirds of the oncogenes and other genes associated with growth and differentiation have several hundred nucleotides of transcript leader region. An ATG within this leader sequence is suggested to be involved in the regulation of downstream translation either by generating a peptide (20) or by promoting an alternative in-frame translation (21).
During the analysis of translation of GS gene in canine brain, we found that the long transcript with the extra 5Ј-UTR is exclusively translated into the 49-kDa of GS protein. This protein was characterized to have different enzymatic properties, presumably due to a conformational alteration.

EXPERIMENTAL PROCEDURES
Reagents and Primers-All reagents used in the present study were purchased from Sigma, unless otherwise stated. PCR primers for subcloning were synthesized by Genotech co. (Taejon, Korea). PCR products were separated by agarose gel electrophoresis, purified using QIAquick Gel Extraction kits (Qiagen). All plasmid constructs were sequenced using the Big Dye termination kits for the ABI-Prism 3100 DNA sequencer (PerkinElmer Life Sciences).
Animal Tissues-Tissues from the frontal lobe, heart, skeletal muscle, kidney, and liver of dog (Sapsari, a Korean breed, 22) were obtained from an adult animal, with an approval by the University Animal Care and Use Committee at KAIST (Taejon, Korea). The mouse brain was obtained from an adult animal of the FVB strain, which was purchased from Daehan Breeding Center Co. Ltd (Seoul, Korea) and housed under controlled temperature and lighting (22°C with 12-hour light-dark cycle) with free access to food and water, according to the National Institutes of Health guideline for the care and use of laboratory animals. The brain tissue of human was a mixture of samples from the parietal lobe, cortex, and white matter, kindly provided by Chong-Jai Kim (Seoul National University, College of Medicine, Seoul, Korea), with an informed consent of 16-year-old male who died of primitive neuroectodermal tumor in the chest wall and with an approval by the Ethical Committee of Seoul National University Hospital (Seoul, Korea).
In Vitro Transcription and Translation-The short and long transcripts of canine GS, covering the exon 1 to exon 7, were reverse transcribed with oligo(dT) 30 , amplified with primers cGSBamHI-f (5Ј-GGGGGATCCAGAGCCGAGAATGGGAGCGG-3Ј; AF544242, nt 1-20) and cGSXhoI-r1 (5Ј-GGGCTCGAGTTAGTTTTTGTACTGGAAG-GG-3Ј; AF544242, nt 1369 -1349), and subcloned into the BamHI and XhoI sites of pcDNA3.1(ϩ) (Invitrogen), yielding pSGS and pLGS, respectively. The linearized templates of 5 g by digestion with XhoI restriction enzyme were used for in vitro transcription using the mMESSAGEmMACHINE TM kit (Ambion) to obtain capped transcripts. The integrity of transcripts was ascertained by visualizing on denaturing polyacrylamide gel electrophoresis. In vitro translations were carried out in rabbit reticulocyte lysates (Promega). Equimolar amounts of capped in vitro transcripts were used as templates for protein synthesis in a 50 l reaction mixture containing 35 l of reticulocyte lysate, 20 Ci of [ 35 S]methionine (Amersham Biosciences), amino acid mixture (without methionine), and 40 units of RNasin (Promega). After 90 min incubation at 30°C, the synthesized protein products were analyzed by electrophoresis on 10% SDS-polyacrylamide gel and visualized by autoradiography.
Western Blotting Analysis-Protein samples were heat denatured in the presence of 2-mercaptoethanol and SDS and separated electrophoretically on a 10% SDS-polyacrylamide gel under denaturing conditions. The proteins were then transferred electrophoretically onto a nitrocellulose membrane (Schleicher & Schuell), and reacted sequentially with a mouse anti-glutamine synthetase monoclonal antibody (CHEMICON international) and an horseradish peroxidase-conjugated mouse anti-mouse IgG secondary antibody (Amersham Biosciences). The GS protein bands were visualized with ECL TM Western blotting detection reagents (Amersham Biosciences) on an x-ray film.
Transient Expressions in Mammalian Cells-293T (human kidney) and SK-N-SH (human brain neuroblastoma) cells were grown to ϳ15% confluence as monolayers in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum on 3.5 cm dishes for 24 h prior to transfection with either 2 g of pSGS or pLGS, using LipofectAMINE plus reagent (Invitrogen) according to the manufacturer-supplied protocol. After 12-hour of transfection, complete medium was removed from the cells, and the induction medium containing Dulbecco's modified Eagle's medium (plus 10% fetal bovine serum) without L-glutamine was added (23). After 24 h of media change, the cells were washed in phosphate-buffered saline once, harvested in 10 mM imidazole-HCl/2 mM EDTA/5 mM 2-mercaptoethanol (pH 7.2), and sonicated, from which soluble supernatants were collected for Western blotting.
Preparation of Subcellular Fractions-Subcellular fractionation of brain tissue was performed as described previously (24). The canine brain of 1.2 grams was placed in 5 volumes of ice-cold "buffered sucrose" (320 mM sucrose/4 mM HEPES-KOH (pH 7.3)/1 mM MgCl 2 /0.5 mM CaCl 2 ) containing 2 mM dithiothreitol and the mixture solution of protease inhibitors, and the material was kept at 4°C throughout the preparation. The tissue was homogenized in a glass-Teflon homogenizer using 12 up-and-down strokes at 900 rpm, to which the same volume of buffered sucrose was added. The homogenate was centrifuged for 10 min at 900 ϫ g av . The resulting pellet (P1) was stored, whereas supernatant (S1) was collected and centrifuged for 15 min at 12,000 ϫ g av . The supernatant (S2) was stored, and the pellet was washed in 5 ml of buffered sucrose and re-centrifuged for 15 min at 13,000 ϫ g av to obtain a pellet (P2). The S2 was centrifuged for 2 h at 250,000 ϫ g av to yield a pellet (P3) and a cytosol fraction (S3). The P1 contained nuclei, and P2 with mitochondria/lysosomes/peroxisomes, P3 with plasma membrane/ endoplasmic reticulum, and S3 with cytosol.
Purification of the Recombinant GS Enzymes-The short and long coding sequences of canine GS were amplified with primers cGSNdeI-f1 (5Ј-GGGCATATGGCCACCTCTGCGAG-3Ј; AF544242, nt 248 -264)/ cGSNdeI-f2 (5Ј-GGGCATATGGACCAGCGGGAG-3Ј; AF544242, nt 128 -142) and cGSXhoI-r2 (5Ј-GGGCTCGAGGTTTTTGTACTGGAAG-GG-3Ј; AF544242, nt 1366 -1349), and subcloned into the NdeI and XhoI sites of pET21b (Novagen) to obtain the pET-SGS and pET-LGS plasmids. E. coli strain BL21(DE3) was transformed with pET-SGS or -LGS to produce recombinant GSs. The GS protein was induced with a treatment of 0.5 mM isopropyl-1-thio-␤-D-galactopyranoside to cells of 0.3-0.4 O.D. at A 600 . The cells were then allowed to grow 4 h more at 30°C and harvested by centrifugation with JA14 rotor at 4°C and 8000 rpm for 15 min. The pelleted cells were resuspended in 5 mM imidazole/ 0.5 M NaCl/20 mM Tris-Cl (pH 7.9), sonicated on ice, and centrifuged 25,000 rpm in SW41Ti rotor for 30 min at 4°C. The supernatant was filtered through a 0.45 M pre-sized membrane, and loaded onto a nickel-nitrilotriacetic acid agarose column (Qiagen) that had been equilibrated with the same buffer. The column was washed with 15 volumes of the previous buffer and 6 volumes of 60 mM imidazole/0.5 M NaCl/20 mM Tris-Cl (pH 7.9). The GS protein was eluted in 1 M imidazole/0.5 M NaCl/20 mM Tris-Cl (pH 7.9), collected with the fractions of 1 ml each, and pooled for the active fractions. Purity was assessed by 10% SDS-PAGE as described by Sambrook and Russell (25), and protein concentration was determined with a dye-binding assay (Bio-Rad) using bovine serum albumin as a standard. The purified protein was dialyzed against 10 mM imidazole-HCl/2 mM EDTA/5 mM 2-mercaptoethanol (pH 7.2) and stored at 4°C, as reported by Listrom et al. (26).
Enzyme Assay-For measuring GS activity, the synthetase reaction was carried out using a colorimetric method as described by Meister (27). The mixture consisted of 100 mM imidazole-HCl (pH 7.2), 50 mM sodium L-glutamate (pH 7.2), 25 mM 2-mercaptoethanol, 10 mM ATP (pH 7.2), 20 mM MgCl 2 , and 125 mM of freshly titrated hydroxylamine (pH 7.2). The reaction mixture of 500 l was pre-incubated for 2 min at 37°C, and the reaction was initiated by the addition of 0.1 nM of enzyme. The reaction was quenched after 15 min with 750 l of 0.37 M FeCl 3 /0.67 M HCl/0.2 M trichloroacetic acid, and the absorbance was read at 535 nm against a blank lacking the enzyme. The controls containing purified GS without ATP or L-glutamate showed no absorbance. All absorbance measured were within the range of the ␥-glutamylhydroxamate standard curve. A unit of GS activity was defined as the amount of enzyme catalyzing the synthesis of 1 mol of product in 1 min at 37°C under the conditions given above. The kinetic velocity was expressed as units per nM of protein.
Fluorescence Spectroscopy-1 M of GS protein in 10 mM HEPES-KOH (pH 7.2) was mixed with KCl, MnSO 4 , or MgSO 4 at the indicated final concentration and incubated at 25°C for 10 min. Intrinsic fluorescence was monitored with the 8100 Series 2 spectrofluorometer (SLM-AMINCO) equipped with a temperature-controlled cuvette compartment maintained at 25°C. The tryptophan fluorescences after being excited at 295 nm were monitored in the range of 310 -400 nm.
Analytical Gel Filtration Chromatography-The GS protein was passed through 0.2-m nylon filter (Whatman) and loaded onto a Superdex 200 column (bed dimensions, 10 ϫ 300 mm; Amersham Biosciences) equilibrated with 10 mM HEPES-KOH (pH 7.2) at 25°C. The protein peaks were detected with a Ä KTAprime Chromatography system and PrimeView software (Amersham Biosciences) with an absorbance at 280 nm. The apparent molecular masses were estimated using intrapolation in the plots of V e (elution volume) Ϫ V o (void volume) versus log [molecular weight] from standard proteins (Amersham Biosciences) of chymotrypsinogen A (19.9 kDa), albumin (64.7 kDa), catalase (219 kDa), and thyroglobulin (699 kDa). The void volume (6.2 ml) was determined using blue dextran 2000 (1 mg/ml).

Novel Translation Initiation from the Alternative Transcript of
Canine Glutamine Synthetase Gene-It was found previously that alternatively spliced GS transcript acquiring extra 5Ј-UTR decreased the normal translation (18). In an effort to obtain further evidences on this observation, we carried out an in vitro translation of transcripts made from the GS plasmids in rabbit reticulocyte lysate. Although the short transcript generated a 45 kDa protein (SGS), the long one with extra 5Ј-UTR gave rise to a 49-kDa protein (LGS) with a small amount of SGS (Fig. 1A). This indicates that translation occurs from an upstream ATG (uATG at Ϫ120) located in the extra exon, resulting in an in-frame 40 amino acids extension of the N-terminal SGS (Fig. 2).
Because the rabbit reticulocyte, a highly specialized cell, is known to have ill-balanced translation system (28), expressions of the GSs were analyzed transiently in 293T cells (Fig. 1B). The 293T cells transfected with the plasmid containing cDNA for SGS (pSGS) produced a band of expected size (45 kDa) on Western blot, whereas the LGS plasmid (pLGS) generated only the 49-kDa protein band, which is different from the result of in vitro translation (Fig. 1A). When both pSGS and pLGS were transfected, equal amounts of proteins were expressed. The same patterns of GS expressions were obtained for cDNAs of SGS and LGS in SK-N-SH cells (data not shown), suggesting that the primary role of long GS transcript is to produce the LGS protein in vivo.
The Long Form of GS Is Found in Canine Brain and Liver Tissues-To examine the presence of LGS in the canine tissue, soluble extracts from the frontal lobe, heart, skeletal muscle, kidney, and liver tissues were separated on SDS-PAGE and blotted with monoclonal antibody raised against the mouse GS. The canine GSs are highly expressed in brain, moderately in liver, and barely in other tissues (Fig. 3A), with lower amounts of the long from. In human and mouse brains, only the 45-kDa GS bands were detected (Fig. 3B), presumably due to an absence of an in-frame uATG.
Similar to canine GS expression in animal cells, the expressions of the human and mouse GSs were analyzed transiently FIG. 1. Translations of canine GS transcripts in vitro and in vivo. A, the two canine GS transcripts synthesized in vitro were translated in a rabbit reticulocyte lysate, which were analyzed by SDS-PAGE and autoradiography. Unlike the short transcript (Short), the long one (Long) has extra 5Ј-UTR corresponding to exon 1Ј. Equal amounts of capped transcripts (0.5/0.5 pmol; 1/1 pmol per 50 l reaction) for short and long were used. Translation from the short transcript generated a 45-kDa product, whereas the long transcript produced an additional 49-kDa band, each designated here as SGS and LGS, respectively. B, both canine GS transcripts were cloned into the expression vector, yielding pSGS for short and pLGS for long ones. They were transfected into 293T cells and analyzed by Western blotting with monoclonal antibody for GS. In the negative control (Mock), no protein was detected (2). The plasmid harboring short transcript (pSGS) produces a 45 kDa product, as in the case of in vitro translation, whereas the pLGS produces only 49 kDa protein without the short form. When both plasmids were transfected with equal molarity, similar amounts of two proteins were detected (pSGSϩLGS).

FIG. 2.
The sequence of extra exon and its corresponding peptide. The exon 1Ј contained in the long transcript is shaded with its encoded amino acids shown below the codon sequences. The translation of SGS is initiated from the start codon (ϩ1), whereas the one for LGS is located at Ϫ120. in 293T cells. The 293T cells transfected with the plasmid containing cDNA for the short transcript of human GS produced the band of expected size (45 kDa), so did the long one with less amount of protein (data not shown). The short transcript of mouse GS expresses the 45-kDa protein, whereas the long one produces none (not shown), presumably due to translational repression as described previously (18).
Intracelluar localization of GS protein was analyzed by subcellular fractionation of the canine brain, which were blotted with GS antibody. Absolute amounts of both enzymes were varied in different fractions, i.e. cytosol (S3), endoplasmic reticulum/plama membrane (P3), nuclei (P1), and other particulate fraction (Fig. 4A, P2). However, their ratios are about the same, suggesting that the extra 40 amino acids are not associated with specific targeting. This pattern of SGS/LGS distribution is consistent with the result of subcellular localization of the GS proteins fused to GFP, produced from pEGFP-SGS and pEGFP-LGS. When two constructs were transfected into Madin-Darby canine kidney cells, the GS-GFP fusion proteins were expressed evenly in all subcellular regions (Fig. 4B).
The Glutamine Synthetase Isoforms Differ in Their Enzymatic Properties-The two different canine GS genes were cloned into an inducible pET21b vector, from which the enzymes were purified to homogeneity. The colorimetric synthetase reaction assay with L-glutamate, hydroxylamine, and ATP as substrates revealed that K m values of SGS and LGS are 1.1 Ϯ 0.12 and 1.3 Ϯ 0.14 mM for L-glutamate, 1.6 Ϯ 0.41 and 1.7 Ϯ 0.31 mM for hydroxylamine, and 1.3 Ϯ 0.24 and 1.9 Ϯ 0.23 mM for ATP, respectively (Table I). The values for L-glutamate and hydroxylamine are similar in both enzymes, but the affinity for ATP is higher in SGS than in LGS. It is known that MSOX, with the presence of ATP, is phosphorylated by GS, leading to an irreversible and non-covalent inhibition of the enzyme (27). When MSOX was titrated in the presence of 50 mM L-glutamate, the K i values were 0.067 and 0.124 mM for SGS and LGS, respectively (Table I), indicating that the affinity for MSOX is lower in LGS than in SGS. The affinity changes for ATP and MSOX suggest that the binding sites for both FIG. 3. Detection of GS proteins in mammalian tissues. A, results of Western blotting analysis for five different canine tissues. Soluble extract of each 10 g of tissue was separated by SDS-PAGE, transferred onto nitrocellulose membrane, and detected by GS monoclonal antibody. Canine GS protein is highly expressed in frontal lobe and moderately in liver. SGS is more abundant than LGS. B, equal amounts of mouse, human, and canine brain tissues were analyzed by SDS-PAGE and Western blot.
LGS was detected only in canine brain tissue, not in mouse and human.   (4) demonstrated that bindings of various effectors, such as chloride, manganese, and magnesium, to allosteric or activator site of the bovine brain GS cause a conformational change as monitored by fluorescence quenching, without affecting the oli-gomerization state. There are seven tryptophans and fifteen tyrosines in the short GS protein, whereas the long one contains an extra tryptophan in the N-terminal peptide (Fig. 2). To assess whether the extended peptide of LGS affects the conformation of GS, we monitored fluorescence quenching by the presence of KCl, MnSO 4 , or MgSO 4 , each treated alone or combined together. The fluorescence changes are shown here as the ratio of fluorescence (F) for the sample to that of SGS enzyme itself (Fig. 5, F o ). When excited at 295 nm for tryptophan, addition of MnSO 4 to SGS resulted in fluorescence quenching of about 15%, whereas MgSO 4 increased fluorescence of 30% (Fig. 5). In contrast, LGS shows 34% quenching with MnSO 4 , and only 8% increase with MgSO 4 . Nevertheless, the patterns of fluorescence changes are similar. However, the tryptophan fluorescence of SGS increased ϳ133% by an addition of 10 mM chloride, whereas no change was observed in LGS. Further addition of MnSO 4 and MgSO 4 to the chloridetreated SGS resulted in 85% quenching relative to the untreated. For LGS, quenching was 17% with MnSO 4 , whereas MgSO 4 increased fluorescence of about 5-6%. The notable differences in fluorescence between SGS and LGS were observed upon chloride binding, with enormous increase in SGS. Based on the fact that chloride ion binds to the allosteric sites of GS, such fluorescence changes may reflect conformational differences between the two enzymes.
On the other hand, the basal levels of tryptophan fluorescences are different between SGS and LGS such that the emission of LGS at 338 nm is 1.5-fold higher than that of SGS. The difference could be due to either the presence of extra tryptophan residue in LGS or the change in microenvironment of other tryptophans, or both. To exclude the possibility that the LGS is caused by its oligomerization, we examined oligomeric states of the enzymes by gel-filtration chromatography. As shown in Fig. 6, the degrees of multimerization in both proteins are similar, in which the majority exists as octamer of 70 -73%. This suggests that the fluorescence changes in the GS proteins are not associated with their quaternary structures. DISCUSSION We characterized two isoforms of GS proteins, 45 and 49 kDa, translated from one functional gene. The long form was generated by in-frame translation from the upstream ATG located in the extra exon of long GS transcript. The extra peptide did not reveal any sequence similarity in the data base search. Because the sequence analysis predicted O-linked glycosylation site for N-acetylglucosamine at ϩ40 position (threonine) of LGS, we treated LGS with ␤-acetyl hexosaminidase or with sodium hydroxide for ␤-elimination (29). However, the result was negative (data not shown), although we cannot completely rule out the possibility of other modifications including phosphorylation.
In a previous study (18), it was shown that the ratio of the short to long transcripts of canine GS was similar in various brain tissues. However, relative ratio of the proteins was different, with far more SGS (Fig. 3A), implying that there is a difference in stability or translation between the transcripts of LGS and SGS. The known regulators of GS expression at post-transcriptional level include glutamine, growth hormone, and oxidations mediated by metal or free radical peptides generated by a fragmentation of ␤-amyloid (2).
Most eukaryotic mRNAs contain a short recognition sequence that facilitates initial binding of mRNA to the small subunit of the ribosome (19). The sequence information of the long UTR (Fig. 2) as well as the analyses of in vitro (Fig. 1A) and in vivo (Fig. 1B) translations indicated that the synthesis of LGS depends on the uATG codon located at Ϫ120. Efficiency of translation from the ATG codon in eukaryotes relies on an optimal translation initiation context (particularly A Ϫ3 and/or G ϩ4 ). Unlike the regular ATG at ϩ1, the translation start site of LGS ( Ϫ120 ATG) has a G at ϩ4, but not A at the Ϫ3 position, presumably serving as less efficient translation initiators. Nevertheless, LGS seems to be expressed from the long transcript at higher concentration in vitro than SGS. Interestingly, the long GS transcript produces only LGS in vivo (Fig. 1), although the SGS contains an ATG codon with conserved Kozak sequence (both A Ϫ3 and G ϩ4 ). This might be due to a readthrough translation of LGS from uATG without stopping at the SGS initiation codon (ϩ1 position in Fig. 2).
In the case of human GS, two uATGs exist at Ϫ326 and Ϫ317 nucleotides upstream of the authentic ATG (18). However, a stop signal is found at Ϫ248 position, which is 78 and 69 nucleotides downstream of the uATGs. This stop signal may allow the re-initiation of SGS translation in the long transcript of human GS. In the mouse GS exon, one uATG is located at Ϫ83 nucleotides upstream of ATG, which is in out-of-frame from the coding sequence. The translation from uATG may encounter a stop codon appeared after the initiation ATG, and thus we speculated that the abortive translation from uATG inhibits the authentic SGS translation by skipping the normal initiation.
The heterogeneity of GS, i.e. existence of several forms inside one taxon, is well known and widespread (14). Most of subtypes of GS are compartmentalized in different subcellular organelles (15). The human genome contains at least five GS homologs (30). There are at least two enzymes possessing GS activities in the human brain, one oligomeric enzyme consisting of 44 kDa subunits (GS) and the other with 54 kDa subunits (GS-like protein) that is enriched in mitochondrial fraction (16). One may speculate that the addition of N-terminal amino acids affects functional activities or intracellular location of the GS protein. When we examine subcellular localizations of SGS and LGS in mammalian cell or brain tissue, we were unable to find any differences in compartmental localization between LGS and SGS in canine brain (Fig. 4).
We compared enzymatic properties of the two enzymes, in which the affinities for L-glutamate and hydroxylamine are similar, although those for ATP and MSOX were different (Table I). This may indicate that there is a difference in enzyme conformation, at least in the substrate binding sites for ATP and MSOX. To further investigate this, we examined chloride and divalent metal ion bindings to allosteric and activator sites, respectively, which induce the conformational changes of GS (4). The most prominent differences in fluorescence change were found in the chloride binding, in which a considerable increase was observed for the short protein compared with the long one, suggesting a conformational difference between the two proteins.
Because the fluorescence change could also be due to a nonspecific binding of chloride on the substrate site if there is a fluorophore located in the substrate site of GS, we examined the fluorescence changes upon bindings of ADP, L-glutamate, and MSOX. The marginal difference in fluorescence was detected below the level of 5%, indicating that there is no tryptophan residue in the substrate site that significantly alters its fluorescence upon substrate binding (data not shown). On the other hand, to assess whether the fluorescence change is due to the oligomerization of enzyme, we detected fluorescence in the sample with only monomers, obtained by treating 1% CHAPS (zwitterionic detergent). Here, the fluorescence change was less than 10% of the values obtained for purified protein with mostly oligomeric forms, indicating that the tryptophan fluorescence is not probing the oligomerization (data not shown).
We attempted to detect any structural difference due to an extension of N-terminal amino acids in LGS using circular dichroism spectroscopy. However, an analyzable spectra were not obtained because of the high propensity of GS oligomerization that disturbs circular dichroism measurement. At any rate, we already presented other enzymatic and spectroscopic data indicating the conformational alteration by an extension of N-terminal amino acids in LGS, although physiological relevance of this structural variation still remains to be elucidated.