Isolation and Characterization of Glutamine Synthetase from Chicken Neural Retina

Abstract A procedure is described for the isolation of glutamine synthetase (EC 6.3.1.2) from chicken neural retina, and the enzyme has been purified to homogeneity. Its amino acid composition is given and the results of electron microscopic examinations are described. Electrophoresis in polyacrylamide gels containing either sodium dodecyl sulfate or urea showed that the enzyme consists of homogeneous subunits with a molecular weight of 42,000 ± 2,000. The molecular weight of the native enzyme was found by sedimentation equilibrium analysis to be 392,000. The sedimentation coefficient of the enzyme (15.8 S) and the Stokes radius (59 A) were determined and their combination gave a molecular weight of 386,000. Calculations from these data would indicate the presence of nine subunits in the native enzyme; however, other considerations favor an octameric structure with a molecular weight of the order of 340,000. The possible reasons for the apparent discrepancy between this value and those obtained from the sedimentation data are discussed. Electron microscopy showed the native enzyme molecules as rectangular particles with dimensions of about 90 x 90 x 125 A; occasionally the individual molecules were seen associated in chain-like aggregates.

From theDepartment of Biology, The University of Chicago, and MeArgonne Cancer Research Hospital,7 Chicago, Illinois 60637 SUMMARY A procedure is described for the isolation of glutamine synthetase (EC 6.3.1.2) from chicken neural retina, and the enzyme has been purified to homogeneity.
Its amino acid composition is given and the results of electron microscopic examinations are described. Electrophoresis in polyacrylamide gels containing either sodium dodecyl sulfate or urea showed that the enzyme consists of homogeneous subunits with a molecular weight of 42,000 f 2,000. The molecular weight of the native enzyme was found by sedimentation equilibrium analysis to be 392,000. The sedimentation coefficient of the enzyme (15.8 S) and the Stokes radius (59 A) were determined and their combination gave a molecular weight of 386,000.
Calculations from these data would indicate the presence of nine subunits in the native enzyme; however, other considerations favor an octameric structure with a molecular weight of the order of 340,000. The possible reasons for the apparent discrepancy between this value and those obtained from the sedimentation data are discussed.
Electron microscopy showed the native enzyme molecules as rectangular particles with dimensions of about 90 x 90 x 125 A; occasionally the individual molecules were seen associated in chain-like aggregates.
Stadtman's group has purified and studied glutamine synthetase from Escherichiu coli (6) and Bacillus subtilis (7). More recently, glutamine synthetase from pea and rat liver (8) have also been examined.
Although there are certain remarkable similarities between the two bacterial enzyme * Supported by a postdoctoral stipend from Training Grant TOl-HD 00297.
$ Supported by National Institutes of Health Grant 5-ROlHE-13505. 4 Supported by National Institutes of Health Grant HD 01253. y Operated by the University of Chicago for the United States Atomic Energy Commission. preparations and between the sheep brain, rat liver, and pea enzymes, none of these are physically or chemically identical.
It has long been known that in the neural retina of the chicken the level of glutamine synthetase is very high, and that its specific activity there is considerably above that in other tissues, including the brain (9). The discovery of the hormonal induction of this enzyme in the embryonic neural retina of the chick in vivo and in vitro by ll-P-hydroxycorticosteroids (10) has focused considerable interest on this system as a particularly favorable model for analysis of control mechanisms in embryonic differentiation (11). The induction of glutamine synthetase in the retina represents de novo synthesis and accumulation of the enzyme (12-14); it is correlated with the functional differentiation and maturation of the retina, and considerable information has been forthcoming concerning macromolecular events and gene expression in this system (9)(10)(11)(12)(13)(14).
A detailed knowledge of the properties of the enzyme is essential for further analysis of this enzyme induction, as well as for comparative studies of glutamine synthetases from various microbial and eukaryotic systems. We describe here the isolation of glutamine synthetase from chicken neural retina tissue and some physicochemical properties of the enzyme.

MATERIALS AND METHODS
M&&&--Eyes from freshly slaughtered fowl were used as the source of retina tissue. The retina was removed from the eyes, thoroughly washed with cold Tyrode's physiological salt solution and lyophilized.
About 2 g of lyophilized retina could be obtained from 200 heads. The sources of the chemicaIs were as follows: DEAE-cellulose, 0.93 meq per g (Sigma) ; hydroxylapatite for chromatography (Bio-Rad) ; catalase (Nutritional Biochemicals) ; and ferritin from horse spleen (Mann).
Assays-Glutamine synthetase activity was assayed by the hydroxamate reaction, as described before (11). However, for the determination of enzyme activity in the hydroxylapatite column fractions, this method could not be used because of interference due to high phosphate concentration.
For this, 50 ~1 of each fraction was diluted to 0.35 ml with 0.1 M acetate buffer, pH 5, prior to reaction with hydroxylamine.
Catalase was assayed by following the decrease in absorbance at 240 nm of a 3-ml reaction mixture containing 30 pmoles of potassium phosphate buffer, pH 7.2, 18 pmoles of hydrogen peroxide, and 25 ~1 of the sample (15). Activities were determined in terms of increase in absorbance at 240 nm/20 s/25 J of enzyme fraction. Disc Gel Eleclrophoresis-SDS-polyacrylamide gel electrophoresis was performed as described by Weber and Osborn (17). For electrophoresis of the native, undissociated enzyme, 4% gels were used in 0.005 M phosphate buffer, pH 7.8. Electrophoresis in urea was performed after exposing the enzyme to 4 M urea for 3 hours at 25". The 4 y0 of polyacrylamide gels used for this purpose also contained 4 M urea in 0.005 M phosphate buffer, pH 7.8. A current of 8 to 10 ma per tube was generally applied for 3 to 4 hours and after the run the gels were stained with Coomassie blue and destained electrophoretically. Zonal Centrijugation-Sedimentation coefficients were measured by the method of Martin and Ames (15). Samples containing approximately 0.3 mg of each protein in a total volume of 0.1 ml were layered over 4.5 ml of 5 to 20% sucrose gradients in 0.05 M Tris-HCl buffer, pH 7.5, or in 0.01 M phosphate buffer, pH 7.1 (the latter was used when protein determinations by the Lowry procedure were desired).
Centrifugation time was 18 hours at 35,000 rpm in Spinco SW 39 rotor at 4". After piercing the tubes, the contents were fractionated and assayed for glutamine synthetase and marker protein peaks.
Gel Filtration-The gel filtration experiments employed a column of Sephadex G-200 (Pharmacia).
In general, the procedure of Siegel and Monty (18) was followed.
The buffer consisted of 0.04 M sodium phosphate, pH 8.0. After the column had been fully equilibrated with buffer, 1.35 ml of a solution containing 10 mg of serum albumin, 5 mg of catalase, 10 mg of ferritin, and 1.5 mg of retinal glutamine synthetase in 0.04 M phosphate buffer, pH 8, with 10% glycerol was applied to the column, and 1.4.ml fractions were collected.
The peak positions in the eluate were determined and the parameter Kd defined by the following equation (18) was calculated.
where V, is the elution volume at the peak, V'O is the void volume of the column, V, is the volume of the gel components, and Vt is the total volume of the gel bed.
Analytical Ultracentrijugation-A Spinco model E ultracentrifuge equipped with Raleigh interference optics was used. Sedimentation equilibrium measurements were made at three different concentrations of glutamine synthetase in the Yphantis six-channel centerpiece (19). Fringe displacements were measured with a two dimensional Nikon microcomparator.
Sedimentation velocity measurements of the enzyme activity were made with a separation cell following the procedure of Yphantis and Waugh (20).
Electron lllicroscopy-Preparations of purified glutamine synthetase in 0.1 M phosphate buffer, pH 6.5, were applied to carboncoated, 400 mesh copper grids, rinsed with 0.1 M potassium chloride, and negatively stained with one of the following solutions: 2oj, uranyl acetate, unbuffered; 1 y0 sodium phosphotungstate, pH 7; or 1% ammonium molybdate, pH 7. In some experiments, negative staining was applied following fixation of specimens in 2% glutaraldehyde, pH 7. Comparable results were obtained with all three stains with or without prior fixation; 1 The abbreviation used is: SDS, sodium dodecyl sulfate. since image contrast was best with uranyl acetate, this was employed in the present work, without prior fixation.
For measurements of particle dimensions of glutamine synthetase, the enzyme preparations were negatively co-stained with one of two internal standards, tobacco mosaic virus2 or tropomyosin tactoids (21). Both of these macromolecules have well documented dimensions which have been established by x-ray diffractometry and electron microscopy (21, 22). The outside diameter of tobacco mosaic virus is 180 A; the tropomyosin tactoids possess a transverse periodicity of 396 f 6 A. Electron microscopy was performed with an AEI EM 6B instrument operating at 60 kv, with a 50 pm objective aperture and utilizing original magnifications of 20,000 or 40,000 diameters.
Amino Acid Composition-Amino acid analyses were carried out after hydrolyzing the enzyme for 20 and 40 hours in a sealed hydrolysis tube in 6 N HCI. Portions of the hydrolysate were analyzed with a Beckman amino acid analyzer model 12OC.

Isolation of Retinal Glutamine Synthetase
The enzyme preparations were maintained between O-4" throughout this procedure. All buffers contained 0.005 M mercaptoethanol.
A small sample from each step was assayed for enzyme activity and protein content.
A summary of the isolation procedure is shown in Table I.
Step 1. Preparation of Crude Extract-Dry lyophilized adult chick retina (400 mg) was sonicated in 20 ml of 0.005 M potassium phosphate buffer, pH 7.8, and spun at 100,000 x g for 30 min. The supernatant, called the crude extract, was diluted 2.fold and used for ammonium sulfate fractionation in Step 2.
Step 2. Ammonium Sulfate Fractionation-To 40 ml of the crude extract, 25 ml of cold (4") saturated ammonium sulfate was added dropwise with continuous stirring.
Following an additional stirring for 15 min the mixture was centrifuged at 15,000 X g for 15 min. The supernatant was decanted and brought to 50% saturation with respect to ammonium sulfate by adding another 15 ml of saturated ammonium sulfate. The misture mined by chromatography on a colutin of Sephadex G-ZOO with was centrifuged after 15 min and the precipitate was dissolved bovine serum albumin, catalaxe, and ferrit'in as standards. From in 3 to 4 ml of 0.005 M phosphate buffer and was dialyzed over-the plot of Ka uerSus the Stokes radius, a, for the marker proteins night against the same buffer. (Fig. l), the Stokes radius for the enzyme was found to be 59 A.
Step 3. Chromatograph,y on IIydroxylapatite Column-The hilolecular weight of the enzyme was then calculated to be 386,000 dialyzed preparation from Step 2 was applied to a hydroxylapa-with the relationship M = 67rqNaS/l -VP, where q is the tite column (15 X 300 mm; bed height 12 cm). Potassium viscosity, N is Avogsdro's number, and p is the density of t,he phosphate buffer (0.1 M; pH 7) was passed through the column solvent. at about 50 ml per hour with a peristaltic pump; 3-ml fractions were collected, and the absorbance of the fractions was read at Polyacrylamide Gel Electrophoresis and Subunit Composition 280 nm; when the absorbance of the eluent dropped to the level The electrophoretic patterns of the native undissociated enof the buffer (after about 150 ml of buffer), elution was begun zyme and of the monomeric form obtained after treatment with with a gradient of 0.15 to 0.35 M phosphate buffer, pH 7.2. The urea or SDS are shown in Fig. 2. The native enzyme displayed enzyme emerged from the column between 0.2 to 0.3 M buffer a major fast moving band, and in addition a minor slow moving concentration.
All fractions with enzyme activity were pooled, band which represented 5 to 10% of the total protein.
Dissociaand the combined eluent was concentrated to about 4 to 5 ml by tion of the same preparations into monomeric form by treat,ment passing through Amicon PM-30 membrane filters (43-mm diamewith urea or with SDS, produced only a single band (Fig. 2). ter) in a 52.ml Amicon filtration cell under 50 to 80 p.s.i. nitrogen Since the electrophoretic mobilities of proteins in gels depend in pressure. These filters were tested before and were found to the presence of urea on charge, and in the presence of SDS on retain the enzyme almost quantitatively.
The concentrate molecular weight, the finding of only one band under both t,hese from the filtration cell was dialyzed overnight against 0.005 ~4 conditions strongly suggests t.hat the subunits of retinal glutamine phosphate buffer, pH 7.8, to remove the salts and then used in synthetase are identical.
These resu1t.s also indicate tha.t the Step 4. minor band observed in the electrophoretic pattern of the native Step 4. Chromatography on DEAE-cellulose Column-The enzyme is unlikely to be due to an impurity in the preparat,ions solution from Step 3 was applied to a DEAE-cellulose column and may represent an aggregated form of the enzyme. Further (15 X 300 mm; bed height, 15 cm) previously equilibrated with information on this point, was sought in the following experi-0.005 M phosphate buffer, pH 7.8. About 150 ml of buffer was merits. passed through the column at a flow rate of approsimately 50 Preparations of the purified enzyme were chromatographed on ml per hour. Under these conditions, glutamine synthetase remained tightly bound to DEAE-cellulose while some other pro: Sephadex G-200 column or sedimented in 5 to 20% sucrose gradients. When the enzyme activity was assayed in the column teins were washed out. The enzyme was then eluted with 0.1 M eluent or in the gradient fractions, only a single peak of activity phosphate buffer of pH 6.5. Three-milliliter fractions were colwas observed; however, when the fractions from the sucrose lected. The enzyme activity appeared within first 5 to 10 fracgradient were assayed for protein, a small protein peak sedimenttions. The active fractions were pooled and concentrated with ing faster than the bulk of the native enzyme was detected ( Fig.  PM 30 filters. The enzyme solution could be stored in ice in 3). We assume that this small peak corresponds to the minor 0.1 M phosphate buffer, pH 6.5, for at least a month without any band on the electrophoretic pattern of the native enzyme The significant loss in activity. exact nature of this component is not clear; its slower mobility Molecular Weight of Native Enzyme relative to the major band and its apparently identical subunit composition suggest that it represents an aggregated form of the The molecular weight of the native enzyme was determined by native enzyme; this conclusion is supported by the presence of the sedimentation equilibrium method of Yphantis (19). Ex-small aggregates of the native enzyme in electron micrographs periments were performed at three different concentrations (see below). (0.735, 0.55, and 0.75 mg per ml) in 0.1 M phosphate buffer, The molecular weight of the subunit was determined to be pH 6.5.
42,000 f 2,000 from SDS-polyacrylamide gel electrophoresis, Plots of log ((Y -Y,) + 3) versus r2 were made and the slopes were determined by least squares fit, using those values which were clearly in the linear region, and neglecting the fringe displacements where they were densely packed. The mean molecular weight calculated from the three slopes (1.465, 1.477, and 1.539) was 392,000.
An independent estimate of the molecular weight of the native enzyme was obtained by combining the sedimentation coefficient and the Stokes radius according to the procedure of Siegel and Monty (18). This method, although less accurate than the eedimentation equilibrium method described earlier, allows measure- with bovine serum albumin, ovalbumin, and chymotrypsinogen as standards (Fig. 4). Although a comparison of this value with the molecular weight of the native enzyme (392,000) would indicate the presence of nine subunits, an octameric structure is suggested by the electron micrographs of purified enzyme preparations (see below) which show a box-like structure similar to that of the octameric sheep brain glutamine synthetase. The morphology of the retinal glutamine synthetase molecules was examined following negative staining, as described under "Materials and Methods." Fig. 5 demonstrates the excellent over-all homogeneity of the purified preparation of the enzyme. The only particles noticeable are those assumed to be the enzyme, and no other macromolecular structures were observed on the grid. The individual molecules most frequently appear as compact rectangular structures with an "H"-like shape. In certain molecules, substructure can be detected within the lateral arm of each "H"; however, the present data are insufficient to propose a definitive three-dimensional, low resolution model of the molecule. Occasionally, the individual enzyme molecules were seen associated into aggregates in the form of chains (Fig. 5, inset). Both side-to-side and end-to-end aggregates were observed.
Particle dimensions were estimated from plates containing as an internal standard either tobacco mosaic virus or tropomyosin (Fig. 6) ; similar estimates were derived in both cases. The cylindrical tubules of tobacco mosaic virus are approximately twice the diameter of the enzyme particles. We have measured the length and width of 100 enzyme molecules in the "H" projection. The minor axis measured 90 Z!Z 7 A (mean f S.D.), the major axis 127 f 10 A. These electron microscopic results are compatible with a three-dimensional rectangular structure, 90 x 90 x 125 A. The size of the subunits or monomers of glutamine synthetase cannot be measured in our micrographs with adequate precision, but if the molecule contains eight subunits, one at each corner of a rectangular structure, then the monomers are probably no larger than 45 x 45 x 65 A.

Amino Acid Composition and. Related Studies
The amino acid composition of the chicken retina glutamine synthetase (Table II) showed close similarities with that of glutamine synthetase from sheep brain or bacteria; however, the differences in the content of glutamic acid, histidine, and leucine show that these enzymes are not identical. The partial specific volume of the retinal enzyme was calculated from the amino acid composition (23) to be 0.725 at 20'. This value was used in the determination molecular weight of the native enzyme. Attempts to determine the NHS-terminal residue of retinal enzyme in a Beckman amino acid sequencer with the use of the phenyl iso- bration bar = 1030 A; magnification X 220,660. The inset illustrates an area in which the molecules have aggrethiocyanate reaction were unsuccessful since no reaction occurred, presumably due to a blocked NH*-terminal amino acid.

Other Properties
The effect of increasing concentrations of urea on the activity of retinal glutamine synthetase is shown in Fig. 7. The enzyme is almost completely and irreversibly inactivated by concentrations of urea higher than 3 M, which, as in the case of sheep brain enzyme (4) presumably dissociates the enzymes into monomeric, catalytically inactive subunits.
The absorption spectrum of retinal glutamine synthetase shows a peak at 280 nm and a trough at 250 nm with a &80:A260 ratio of 1.6 suggesting that, unlike the E. coli enzyme (24), the retina enzyme does not contain covalently bonded AMP. DISCUSSION Information concerning the characteristics of retinal glutamine synthetase is of interest for several reasons. First, by adding to what is already known about glutamine synthetase from mammalian brain and liver, and bacteria (8) it extends our knowledge about this group of proteins; second, as explained in the introduction, the importance of the hormonal induction of glutamine synthetase in the embryonic retina as a model system for studying control mechanisms in differentiation makes further knowledge of the characteristics of the retinal enzyme essential (see also Reference 25). As a step in this direction, this paper describes the purification of glutamine synthetase from the chicken neural retina, its molecular weight, subunit composition, and other physicochemical properties.
The procedure employed here for the isolation of the enzyme is a modification of that used for its purification from sheep brain (3). The acid precipitation step at pH 4.3 has been replaced, in the case of the retinal enzyme, by ammonium sulfate fractionation, since this resulted in a relatively greater increase of specific activity. The specific enzyme activity in the 100,000 x g supernatant from the sonicate of the retina is considerably higher than that in the brain or liver; therefore, a IO-fold purification sufficed for obtaining the enzyme in a homogeneous form. The yield of the purified enzyme was approximately 1 mg from 400 mg of lyophilized retina tissue; to obtain this amount of tissue, approximately 100 adult chicken eyes had to be dissected; this imposes a certain limitation on routine large scale preparation of highly purified retinal glutamine synthetase.
Previous studies have shown that bacterial glutamine synthetase is dodecameric with a subunit molecular weight of 50,000 (6, 7) ; the enzyme from sheep brain, rat liver, and pea seeds is apparently octameric with subunit molecular weights estimated to be 49,000, 44,000, and 45,000, respectively (8). In the present study, the molecular weight of the subunit of the retinal enzyme was determined by SDS-polyacrylamide gel electrophoresis to be 42,000 + 2,000; this value is in the general range for those reported for glutamine synthetase from other eukaryote cells; furthermore, it is consistent with previous observations that in the retina, the enzyme is synthesized by polysomes estimated to comprise 12 to 14 ribosomes (12). Following dissociation of the purified enzyme into monomeric form by urea or by SDS, analysis in polyacrylamide gels yielded a single protein component; this indicates strongly that the subunits are identical, and is in a.greement with the findings of subunit identity in sheep brain enzyme (3). The evidence for subunit homogeneity in retinal glutamine synthetase is important in considerations of genetic controls of this enzyme synthesis and regulation; however, it should be pointed out that the identity of the subunits in the retinal glutamine' synthetase could not be as yet confirmed by NHz-terminal analysis, since the NHZterminal end apprars to be blocked.
In contrast, the sheep brain enzyme has been reported to contain a free NHz-terminal arginine (3).
The amino acid compositions of glutamine synthetase from chicken retina, sheep brain, and bacteria (Table II) display  considerable similarities, which raises the possibility of their evolutionary derirat.ion from a common gene. Previous studies have demonstrated immunologiral similarities hctwrcn glutamine synthctase from sheep brain and from neural retina of chick embryos (14). The high aspartate content of the retinal enzyme explains the preferential incorporation of this amino acid into glutamine synthetase in retina cells which are actively synthesizing the enzyme due to hormonal induction (12).
ThP discrepant results of t,he various determinations of molecular weight of retinal glutamine synthetase should be examined.
The value obtained by sedimentation equilibrium was 392,000, and by sedimentation velocity and gel permeation was 386,000.
However, accepting the subunit molecular weight to be 42,000 and the octameric form of the enzyme, the calculated molecular weight is 336,000. These disparate results may be due to inaccuracies inherent in all three methods of mcasurements used. Thus, measurements of subunit molecular weight by gel electrophoresis have an error of about 5%, a value that could account for some of the difference between t,he expected and the observed molecular weights of the octamer.
In the case of sedimentation equilibrium measurements, the presence in solution of aggregates or stacks of the enzyme octamers, such as seen in the electron micrographs (Fig. 5, inset) could have artifactually raised the value. It can be calculated, for instance, that the presence of 16% of an octamer doublet (calculated mol wt 672,000) would suffice to raise the weight average molecular weight from the expected 336,000 t,o the observed value of 392,000. There is, at present, no way to determine whether such aggregates persist even at very high dilutions. Aggregation and stacking of the native enzyme was also found in glutamine synthetase isolated from bacteria3 (26). Thus, in the face of the electron microscopic and other physicochemical data available to us at present, it seems reasonable to conclude that the molecular weight of the chicken retina glutamine synthetase is of the order of 340,000, and that this enzyme consists of 8 apparently identical subunits.