Preliminary crystal structure of Acinetobacter glutaminasificans glutaminase-asparaginase

THEJOURNAL OF BIOLOGICAL CHEMISTRY

Vol. 263, No. 1, Issue of Januaxy 5, pp. 150-156,1988 Printed in U.S.A.

Preliminary Crystal Structure of Acinetobacter glutaminasificans Glutaminase-Asparaginase* (Received for publication, June 18, 1987)

Herman L. Amman$, Irene T. WeberQT,Alexander WlodawerQll, Robert W. HarrisonQll, Gary L. GillilandQ,Kenan C. Murphy$, LennartSjolinII ,and Joseph Roberts** From the $Department of Chemistry and Biochemistry and Center for Advanced Research in Biotechnology, University of Maryland, College Park, Maryland 20742, the $Center for Chemical Physics, National Bureau of Standards, Gaithersburg, Maryland 20899, the 11 Department of Inorganic Chemistry, Chalmers University of Technology, Goteborg, Sweden, and the **College of Pharmacy, University of South Carolina, Columbia, South Carolina 29208

The preliminary structure of aglutaminase-asparaginase from Acinetobacterglutaminasificans is reported. The structure was determined at 3.0-A resolution with a combination of phase information from multiple isomorphous replacement at 4-5-A resolution and phase improvement and extension by two density modification techniques. The electron density map was fitted by a polypeptide chain that was initially polyalanine. This was subsequently replaced by a polypeptide with an amino acid sequence in agreement with the sizes and shapes of the side chain electron densities. The crystallographic R factor is 0 . 3 0 0 following re; strained least squares refinement with data to 2.9-A resolution. The A. glutaminasificans glutaminase-asparaginase subunit folds into two domains: the aminoterminal domain contains a five-stranded /3 sheet surrounded byfive a helices, whilethe carboxyl-terminal domain contains three a helices and less regular structure. The connectivity is not fully determined at present, due in part to the lack of a complete amino acid sequence. The A. glutaminasificans glutaminase-asparaginase structure has been used successfully to determine the relative orientations of the molecules in crystals of Pseudomonas 7A glutaminase-asparaginase, in crystals of Vibrio succinogenes asparaginase, and in a new crystalform ofEscherichia coli asparaginase (space group 1222, one subunit per asymmetric unit).

The parenteral administration of L-asparaginase’ (L-asparagine amidohydrolase, EC 3.5.1.1) leads to the regression of certain lymphomas and leukemias in experimental animals *This workwas supported at the University of Maryland by National Science Foundation Grant PCM-79-07501 and National Institutes of Health Grant CA 33741, and at theUniversity of South Carolina by National Institutes of Health Grant CA 40446. Certain commercial equipment, instruments, or materials are identified in this paper in order to adequately specify the experimental procedure. Such identification does not imply recommendation or endorsement by the National Bureau of Standards. 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 solelyto indicate this fact. ll Current address: National Cancer Institute, Crystallography Laboratory, BRI-FCRF, P. 0. Box B, Frederick, MD 21701. ’Most asparaginases have limited glutaminase activity. We use the term “asparaginase” for those amidohydrolases in which the glutaminase activity is about 5% or less of the asparaginase activity and “glutaminase-asparaginase” to describe those enzymes with more equal reactivity toward asparagine and glutamine.

and humans (Oettgen e t al., 1967; Kidd, 1970; Broome, 1981; Prager et al., 1982). The Escherichia coli enzyme has been used in the treatment of acute lymphoblastic leukemia for about 20 years (Oettgen et al., 1967; Hrushesky et al., 1976), and other amidohydrolases have been used in clinical trials. Asparaginase-sensitive tumor cells generally show a diminished capacity to synthesize L-asparagine because of relatively low levels of the enzyme, L-asparagine synthetase, which results in the requirement for an exogenous supply of the amino acid. It is unclear, however, exactly how L-asparaginase causes death of sensitive tumor cells, and the biochemical basis for L-asparaginase sensitivity has not been firmly established (Keefer et al., 1985). Glutamine, like asparagine, is not an essential component in the human diet. The activity of glutaminase-asparaginases against asparaginase-resistant cells (Schmid and Roberts, 1974) andinasparaginase-resistantpatients(Spiersand Wade, 1976) has been demonstrated. Recently, a combination of a glutaminase-asparaginase from Pseudomonas 7A, 6-diazo5-oxonorleucine, and acivicin has been shown to be active against human mammary and colon tumors in vitro and in tumors growing in nude mice (Roberts, 1983). The simultaneous elimination of both amino acids might confer a therapeutic advantage. The bacterial amidohydrolases are tetramers of identical subunits with molecular masses of 120,000-147,000 daltons; the individual subunits are not catalytically active. There is substantial evidence from solution (Murthy and Knox, 1976) and x-ray crystallographic experiments (Epp et al., 1971; Lee et al., 1975; Wlodaweret al., 1977; Yonei et al., 1977; Ammon et al., 1983, 1985) that the tetramerconsists of a tetrahedral arrangement of subunits with overall 222 symmetry. Although several of these enzymes have been crystallized and preliminary crystallographic publications have appeared, there have been noreports of high resolution structures. The preliminary investigations include: Erwinia carotovora asparaginase (North et al., 1969); E. coli ATCC 9637 asparaginase (Born and Bauer, 1970); Proteus vulgaris asparaginase (Lee et al., 1975);E. coli HAP asparaginase (Itai et al., 1976; Yoneiet al., 1977); Acinetobacter glutaminasificans glutaminase-asparaginase (Wlodawer et al., 1975, 1977); Pseudomonas 7A glutaminase-asparaginase (Wlodawer et al., 1977; Ammon et al., 1983); and Vibrio succinogenes asparaginase (Ammon et a$ 1985). The structure of the E. coli HAP asparaginase at 5.8 A was brieflyreported in a meeting abstract (Mitsuiet al., 1978), but neither a detailed description of the chain nor atomic coordinates have become available. The x-ray crystal structures of four bacterial amidohydrolases are under investigation in our laboratories. Selected

150

Preliminary Crystal Structure of a Glutaminase-Asparaginase TABLEI

TABLE I1 Data collection summary forthe A. glutaminasificans glutaminaseasparaginase native and three derivatives

Crystal and substrate activity data for fouramidohydrolases The activity data are from Wriston and Yellin (1973), Howard and Camenter (1972).and D. Kafkewitz (1985).Dersonal communication). Source

A. glutaminasificans glutaminase-asparaginase Pseudomonas 7A glutaminase-asparaginase

V. succinogenes asparaginase

E. coli asparaginase

%Iative

Space group and unitcell

Unique subunits'

I222 a = 96.56 b = 112.51 A c = 71.17 P212121 a = 118.0 b = 131.2 4 c = 85.1 A P2212, a = 71.3 b = 85.84 c = 114.0 A I222 a = 62.06 b = 73.69 4 c = 140.93 A

1

100/77

4

100/51

2

0.4/100

4

4 4

4

activity' L-Gln/L-Asn

Native"

SMerb

PCMB'

4.7 Resolution 4.0 (A) 5.3 2.9 Unique x-ray data 1689 3068 1454 9413 1 mMsatd satd Derivative soaking conditions 3 weeks 3 weeks Native versus deriv0.1370.1040.212 ative residual'

PtenClb

9 weeks

Two crystals used. sodium mersalyl; PCMB, p-chloromercuribenzoate; PtenC1, platinum ethylenediamine dichloride. R = 2 I Fp - Fpn I /CFp, where FP and FpH are the native and scaled derivative F values. a

1

4/100

"Number of crystallographically unique subunits in the crystal asymmetric unit.

crystal and substrate activity data for the enzymes are given in Table I. Preliminary crystallographic data for the Acinetobacterglutaminasificans glutaminase-asparaginase, Pseudomonas 7A glutaminase-asparaginase, and Vibrio succinogenes asparaginase have been published (Wlodawer e t al., 1975, 1977; Ammon et al., 1983,1985). The crystals of E. coli asparaginase reported here have not been observed previously. In thispaper, a solution for the structuEe of A. glutaminasificans glutaminase-asparaginase at 2.9-A resolution is reported. A complete amino acid sequence for A . glutaminasificans glutaminase-asparaginase was not available* during the interpretation of the electron density map, and, therefore, the structure presented here must be treatedas preliminary. However, the success in correlating this A. glutaminasificans glutaminase-asparaginase structure with the diffraction data for other amidohydrolases through the use of rotation and translation function techniques (i.e. A . glutaminasificans glutaminase-asparaginase versus Pseudomonas 7A glutaminaseasparaginase and A. glutaminasificans glutaminase-asparaginase versus V. succinogenes asparaginase; Ammon e t al., 1983, 1985) indicated the usefulness of presenting the model at its current stage. The results from an A . glutaminasificans glutaminase-asparaginase versus E. coli asparaginase rotation function calculation also are reported here. EXPERIMENTALPROCEDURES

E. coli Asparaginase Crystallization and Data Collection The crystallization of E. coli asparaginase from acetate buffer with n-propyl alcohol as the precipitant has been reported (Itai et al., 1976). When we tried this, however, the result was poorly formed needles which redissolved several days after forming. The conditions that gave stable rod-shaped crystals were 5-8-pl hanging drops of 510 mg/ml protein in0.01 M acetate buffer, pH 5.0, over 2-ml reservoirs of acetate buffer, 12-20% n-propyl alcohol, and 12-20% 2-methyl2,4-pentanediol, a t room temperature. No crystals were obtained a t cold room temperatures (4 "C). The space group, unit cell, and x-ray diffraction data to 3.5-A resolution were obtained from one crystal of E. coli asparaginase with a Nicolet imaging proportional counter, an electronic area detector (Howard et al., 1987). The x-ray source used to generate CuK, The only published sequence for an amidohydrolase is that of an

E. coli asparaginase (Maita et al., 1974).

151

* SMer,

radiation was an Elliot GX-21 rotating anode, operating at 70mA and 40 kV with a 0.3 X 3.0-mm focal spot and a 0.3-mm collimator. Monochromatization was provided by a Huber graphite monochromator. The detector was mounted on a Supper oscillation camera controlled by a Cadmus 9000 microcomputer. This instrument was made available to us by the Genex Corp., Gaithersburg, MD. All data were collected at well controlled room temperature (16-18 "C). The area detector chamber was mounted 14 cm from the crystal, and the carriage angle was set at 10" during data collection. The determination of the crystal orientationand unit cell parameters and the integration of reflection intensities were performed with the XENGEN program system (Howard et al., 1987). The x-ray diffraction data included 14,318 observations of 2,773 reflections extending to 3.5-A resolution; R,, on intensitywas 0.056; and 2,170 of the unique data had significant intensity with Z > 1.5 u ( 0 . This crystal form is orthorhombic, space group 1222, with unit cell parameters a = 62.06, b = 73.69, and c = 140.93 A. The space group and cell parameters were confirmed subsequently by precession photography. One 34,080dalton subunit (Maita et al., 1974) per crystal asymmetric unit corresponds to a V,,, (Matthews, 1968)of 2.26 A3/dalton, which is similar to thatobserved for other amidohydrolases. Procedures Used to Determine the Structure of A . glutaminasificans Glutaminase-Asparaginase Crystal Growth-Crystals of A. glutaminasificans glutaminase-asparaginase were grownas described previously (Wlodawer et al., 1977) by the hanging-drop vapor diffusion method (Wlodawer et al., 1975). The protein was purified by the method of Roberts et al. (1972) and stored as a lyophilized powder at -20 "C. The powder (about 50% protein) was dissolved in 0.01 M phosphate buffer, pH 7.2, to give a protein concentration of 12-14 mg/ml. After extensive microdialysis against 0.01 M phosphate, the protein concentration was in the 1012 mg/ml range. Polyethylene glycol ( M , 400)was added tothe dialysis solution to a concentration of 10-15% (v/v). 8-9-p1 droplets were deposited on either siliconized glass or plastic coverslips, which were then placed over 2 ml of 0.01 M phosphate buffer (pH 7.2) plus 20-25% polyethylene glycol 400solutions in the wells of Linbro tissue culture plates. In some experiments, crystals grew as large as 0.5 x 0.5 X 1.0 mm within a few weeks. Most often, however, only small crystals were obtained, and for the past few years we have been unable to grow any large crystals. The lack of large diffraction-quality crystals has been a limiting factor in this investigation of the structure of A. glutaminasificans glutaminase-asparaginase. Preparation of Heavy Atom Derivatives-Heavy atom derivatives were prepared by soaking A. glutaminasificans glutaminase-asparaginase crystals in solutions of compounds containing the heavy atoms in 0.01 M phosphate, pH 7.2, plus 25% polyethylene glycol400. Successful derivatives were made with sodium mersalyl, sodium p chloromercuribenzoate, and platinum ethylenediamine dichloride. The conditions are summarized in Table 11. Data Collection-A. glutaminasificans glutaminase-asparaginase crystals areorthorhombic, space oup 1222,with unit cell parameters a = 96.56, b = 112.51, c = 71.17 $X-ray intensity data for the native Weighted R factor: R, = - G~,(Z)f)/u~]*/~(Zij/ui,)z)l~*; where G, = g,+ A,s, + B,s? s = sinO/h, and g, A,and B are scaling parameters.

152

Preliminary Crystal Structure of a Glutaminase-Asparaginase

and three derivatives were measured on a Picker FACS-I diffractometer equipped with a sealed copper x-ray tubeand anincident beam graphite monochromator. The procedures, which have been described in detail previously (Ammon et al., 1983), involve: a 9-step w scan over the reflection center; determination of a general background surface as a function of 20, x, and 4; +scan absorption correction; radiation damage correction; and integrated intensity calculation by the fit of a Gaussian profile to the background-corrected step-scan data. Friedel pairs were measured for all crystals at k20. Table I1 summarizes the x-ray data collected for A. glutaminasificans glutaminase-asparaginase. Structure Solution and Heavy Atom Refinement-Each of the three derivative data sets was scaled independently to thenative data with an overall scale factor and a term to compensate for the fall-off of intensities as a function of resolution. The Harker sections of the difference Patterson maps calculated with the coefficients (FPHF p ) 2 were readily interpretable and revealed a major site and one or two minor sites for each derivative. One of the sites was common to all three derivatives. This was the most highly occupied site in sodium mersalyl and p-chloromercuribenzoate and thesecond highest site in platinum ethylenediamine dichloride (Table 111). The heavy atom coordinates, occupancies, and isotropic temperature factors for each derivative were refined separately, and therelative hands of the data were determined with difference electron density maps. Subsequently, the three-dimensional data were subjected to alternating cycles of least squares refinement and phase determination using a modification of the lack-of-closure procedure of Dickerson et al. (1968) and the program PROTEIN of Steigemann j1974). The refinement provided phases for 1340 reflections $0 5.3 A based on three derivatives, for another 256 reflections to 4.7 A with two contributing derivatives, and for 1129 additional reflections to 4.0-A resolution based on the p-chloromercuribenzoate derivative alone. Anomalous data were not utilized at this stage of phase determination. Maps calculated with MIR4data to5.3-A resolution and MIR/SIR data to4.0 A clearly showed the molecular boundary. The protein chain, however, could not be followed for any length, and subsequently, the phases were improved and the resolution was extended by means of two image enhancement/density modification procedures. Phase Extension and Improvement-The first of the methods was the image enhancement/density modification procedure of Wang (1985). In this technique, the initial map and a value for the fraction of the unit cell volume occupied by solvent are used to compute a mask to delineate the protein and solvent regions. The map is filtered with this mask by setting the densities outside of the protein area to a constant value, phases are calculated by inversion of the modified map, and the procedure is repeated several times. Usually the mask is recalculated 2-3 times, and the overall procedure is repeated to convergence. Initially, only the platinum ethylenediamine dichloride derivative data were used, since this derivative gave the best refinement statistics. Four cycles of the Wang procedure resulted in a change in the figure of merit from 0.50 to 0.69 and a decrease in the R factor for map inversion from 0.463 to 0.274. Because the anomalous data had not been used directly for phasing and theFriedel pairs had not been merged, the choice of the hand of the structure was tested by changing the hand of the heavy atom positions and repeating the process described above. When the opposite hand was assumed, the figure of merit at the start and at the endwas 0.01 lower, while the R factor at convergence was higher (0.281). These results suggested that the initial choice of the hand for the heavy atom siteswas correct. Further phase refinement of the platinum ethylenediamine dichloride data was discontinued in favor of utilizing all three derivatives in phase modification and extension. Separate phase probabilities were calculated for each of the three derivatives, with all data available for each, and these probability curves were combined with the program MERGE (part of Wang's program package). The initial map was calculated with the MIR phases to 4.0 A (2725 paired reflections); map inversion gave a figure of merit of 0.54 and an R factor of 0.443. Four cycles of refinement were performed with the first mask, followed bycalculation of another mask and four more cycles of map inversion/phase calculation, and then by calculation of another mask and eight more cycles of refinement. The following steps were taken at this stage to extend the phase information: 1) add 300 reflections at 4-A resolution for which no MIR data were available, followed by 4 cycles of refinement; 2)

'The abbreviations used are: MIR, multiple isomorphous replacement; SIR, single isomorphous replacement.

TABLE 111 Relative occupanciesand fractional heavy atom coordinates for the three heavy atom derivatives of A. glutaminasificans glutaminaseasparaginase Mean fieure of merit = 0.75 for ranee 20.0-4.0-A resolution. Derivative

occu-

x

DRnCV

Y

Z

Ram

Sodium mersalyl

1.00 0.3367 0.3648 0.3918 0.56 0.1788 0.2139 0.3014 54.5% 0.09 0.0039 0.0167 0.2324

Platinum ethylenediamine dichloride

0.63 0.0109 0.3243 0.2347 0.40 0.3451 0.3532 0.3877 42.4% 0.26 0.3020 0.4736 0.1253

p-Chloromercuribenzoate

0.65 0.08

0.3328 0.3645 0.3940 0.1176 0.0931 0.1195 45.6%

add 1146 data in the 4-3.6 A range, 6 cycles; 3) add 117l~reflections at 3.6-3.3 A, 7 cycles; 4) add 2137 reflections at 3.3-3.0 A, 9 cycles. The figure of merit for the final set of phases was 0.74, the R factor for map inversion was 0.215, the phase shift from the initial set was 71.4", the shift per cycle was 1.1", and the number of reflections phased without reference to heavy atom data was 4784, almost twothirds of the total. A map calculated with these phases (hereafter referred to as theW map) was used to fit the model described below (see Figs. 1 and 2). The other technique used to improve and extend the phases originally derived from the MIR data was developedby one of the authors? With this technique, the distribution of the density values in the map is forced to be Gaussian. This is related to a class of image processing algorithms referred to ashistogram equalization and histogram specification (Gonzalez and Wintz, 1977). If it is assumed that the map density values will cluster around some average value and thatmany reflections are used to calculate the map, then the central limit theorem of statistics states that the ideal distribution of density values should be close to Gaussian. A histogram is calculated for the distribution of density values in the initial map, and this distribution is remapped onto a Gaussian distribution with the same mean and standarddeviation. New phases are obtained by inverting this map. The MIR map was passed through this procedure for 15 cycles. The initial resolution was 4.2 A, and the final resolution was 3.2 A. Most of the new reflections were added in the first 10 cycles in increments corresponding to 0.1-8, resolution. The figure of merit increased from 0.78 to 0.96 while the map inversion R factor after modification was loweredfrom 0.188 to 0.028. The totalphase shift was 64.7".The resulting map will be referred to as the H map (see Figs. 1 and 2). RESULTS

The Structureof A. glutaminasificans Glutaminase-Asparaginase Interpretation of A. glutaminasificans Glz$aminase-Asparagine*Electron Density Maps-Both the 3-A W map and the

3.2-A H map were contouredandplottedontransparent acetate sheets. Several CY helices and a @ sheet were visible in both maps. Dots representing C, atoms were placed in the H map in continuous stretchesof recognizable secondary structure. The longest continuous stretch contained two helices, each 16 residues long, which is reminiscent of theotwo long a helices observed by Mitsui e t al. (1978) in the5.8-A resolution map of E. coli asparaginase. A total of 172 residues was fitted, which represents five CY helices and 16 other stretches, including the five strands of a p sheet. The positions of these C,. atoms were digitized (program DIGITby R. W. H.) by placing the contoured map sections on the data tablet of an Evans and Sutherland PS-330 graphics terminal and touching the sheets with the graphics pen at the C,, positions. This proceR. W. Harrison, J. Appl. Crystallogr., submitted for publication.

Preliminary Crystal Structureof a Glutaminase-Asparaginase

153

FIG. 1. The residues of the a4 helix in A. glutaminasificans glutaminase-asparaginase from 97 to 106 are shown with theW map density inA and theH map density inB. Both electron densitymaps were contoured at a level of 1.0 u.

dure was simple and fast. The two-electron density maps andmetry-related molecules touched. Here the interpretationwas the C,, coordinates were examined on a PS 330 computer aided by the I222 symmetry since, in some cases, the strand being fit lay next to the same strand from a symmetry-related graphics system with the program, FRODO (Jones, 1978). The H map and W map are compared in Fig. 1, which molecule, and a tracing was chosen that led the strand back was being built. Three additional helices shows a n a helix, and in Fig. 2, which shows three strands of into the subunit that the sheet. Although the H map and W map were found to were built at the proteinsurface into density that was present be very similar in the interior of the molecule, there were in the H map, but only partially visible in the W map. 300 alanine residues were fitted into the continuous density. appreciable differences in certain regions of the protein surThese were in 15 separate strands because of the lack of face. The W map occasionally masked out protein density that was visible in the H map, while more noise was present connecting density, ranging in size from 3 to 45 residues. It in the H map in solvent areas. It was important to examine was then decided t o make thepolypeptide chain as continuous both maps while building the polypeptide chain; the interpre- as possible by using distance criteria: two strandsowere connected if there were only two ends within about5-A distance. tation relied heavily on the H map at the molecular surface. Polyalanine was built into the starting C, positions and ex- This involved switching the direction of strands in several tended into continuous density where possible. There were instances. The separated strands were joined in such a way several gaps in the density, especially at the proteinsurface. as to avoid circular connections and knots. An essentially The connectivity was ambiguous in several places wheresym- arbitrary choice was made between pairs of connections in

Preliminary Crystal Structureof a Glutaminase-Asparaginase

154

B

FIG. 2. Three strands from the @ sheet in the amino-terminal domain of A. gluturninasificuns glutaminase-asparaginase (residues 26-30, 87-90, and 116-120) in electron density contoured at a level of 1.0 u from A , the W map, and B, the H map.

one case where four ends lay close together. The A. glutaminasificans glutaminase-asparaginase subunit was found to fold into two domains. One domain consisted of a B sheet surrounded by a helices, and reference was made to known cu-p domain structures to aid in connecting strands. The density for this domain was easier to interpret than for the second domain of the protein. This procedure of building logical connections resulted in two polyalanine chains, residues 1-274 and 300-359, for a total of334 amino acids. It should be noted that theE. coli asparaginase with 321 residues (Maita et al., 1974) and A. glutaminasificans glutaminaseasparaginase have similar molecular weights, and thus there probably are fewer amino acids in the correct A. glutaminasificans glutaminase-asparaginase structure than are present in this model. The structure which consisted of two polyalanine chains was refined with the reciprocal space least squares program PROLSQ (Hendrickson, 1985). The crystallographic R factor was lowered from 0.4t4 to 0.343 for 7023 structure factors in the shell from 7 to 2.9 A, while the root mean square deviations

of the bopded distances from ideality decreased from 0.071 8, to 0.028 A. Because a complete A. glutaminasificans glutaminase-asparaginase sequence was not available at the time, an attempt was made to identify the various amino acids on the basis of size and shape of the side chain density, with a bias toward polar side chains on the molecular surface and hydrophobic side chains for interior residues. This model, with its guessed sequence, was further refined after refitting. The final R factor was 0.300 for the 2.9-A data, andothe deviation of bonded distances from ideality was 0.023 A. The model resulting from this refinement is described below; it has also been correlated with the x-ray data from three other amidohydrolases. Description of the A. glutaminasificans Glutaminase-Asparaginase Model Structure-The present model for the A. glutaminasificans glutaminase-asparaginase structure has331 amino acids in two chains, numbered as residues 1-273 and 300-357, as shown in Fig. 3. The A . glutaminasificans glutaminase-asparaginase subunit folds into two domains, and there are three long (Y helices lying on the surface between

Preliminary Crystal Structure of a Glutaminase-Asparaginase

155

the domains. Residues 1-130 and 255-273 are in the amino- norleucine and A. glutaminasificans glutaminase-asparagiterminal domain, andresidues 131-254 and 300-357 form the nase, or Pseudomonas 7A glutaminase-asparaginase, revealed carboxyl-terminal domain. TableIV lists the secondary struc- that radioactivity was associated with Thr-12 of A. glutaminasificans glutaminase-asparaginaseandthe homologous turalelements in thisinterpretation.Theamino-terminal domain consistsof a five-stranded /3 sheet surrounded by five Thr-20 of Pseudomonas 7A glutaminase-asparaginase (Holcenberg et al., 1978). Thr-12 of A. glutaminasificans glutamina helices. The carboxyl-terminal domain was more difficult to interpret, and the structure either more is irregular or else ase-asparaginase has been tentatively assigned to the helix a1 at the start of the amino-terminaldomain. If this assignhas several incorrect connections. This domain contains three a helices as well as four strands that may represent a short /3 ment is correct, the 6-diazo-5-oxonorleucine radiolabel does sheet. Theamino-terminaldomain showsa five-stranded not bindbetween the two domains of the A . glutaminasificans twisted /3 sheet with four parallel and one antiparallel strand glutaminase-asparaginase subunit.The structurecorrespondin coli asparaginase, residues (Fig. 3), although the strand direction is not definitive. Several ing to the putative active site E. of the surface connectionsbetween /3 strands are still ambig- 117-119 ( E . coli numbering), has not been definitely identiuous. This domain structure with a five-strand /3 sheet and fied. external (Y helices is a commonfeatureinotherproteins (Richardson, 1981). Studies of Pseudomonas 7A Glutaminase-Asparaginase and of Predictions of the secondary structure of the E. coli aspaa New Crystal Formof E. coli Asparaginase raginase were made with the method of Chou and Fasman A. glutaminasificansGlutaminase-AsparaginaseVersus E. (1978) and the published amino acid sequence (Maita et al., 1974). The method predicted 10 a-helices, three of which were coli Asparaginase Rotation Function Investigation-The relarelatively long (12-15 residues). When the firstresidue of the tive orientation of the A. glutaminasificans glutaminase-asknown E. coli asparaginase sequence was aligned with the paraginase and E. coli asparaginase subunits has been estabamino terminus of the A. glutaminasificans glutaminase-as- lishedwith Pattersonrotationfunctioncalculationsperparaginase structure, the positionsof five of the predicted E. formed initially with the Crowther (1972) fast rotation funccoli asparaginase helices were close to helices (012, a4,a6,a7, tion method. The calculations used the following data and and a8) in the preliminary tracing of the A. glutaminasificans glutaminase-asparaginase structure. The last four of these TABLE IV helices corresponded approximately in length as well as in Secondary structureof A . glutaminasificans glutaminaseposition. The a1 helix, which is located close to the amino asDarwinme subunit terminus, was not predicted, although thesizes of the amino Amino-terminal Carboxyl-terminal acid side chains correlate well with the amino-terminal residomain domain dues in the published fragment of the A . glutaminasificans 5-12 f f l 152-165 a5 glutaminase-asparaginase sequence (Holcenberg et al., 1978). 35-43 a2 320-328 a7 Active Site of A. glutaminasificans Glutaminase-Asparagia3 334-344 f f 8 68-76 we-Several active site residues have been targeted in ex96-112 a4 periments with the irreversible inhibitors, 5-diazo-5-oxonor256-266 a6 valine and 6-diazo-5-oxonorleucine, which resemble aspara15-21 Pl 22-31 gine and glutamine, respectively. 14C-Labeled 5-diazo-5-oxo02 45-50 P3 norvaline has been shown to react with a serine or threonine 84-92 04 in residues 117-119 of the E. coli asparaginase (Peterson et 115-123 05 al., 1977), and similar experiments with [14C]6-diazo-5-oxo~

,

ASP350

FIG. 3. A stereo view of the C-atoms of the A. glutaminasificans glutaminase-asparaginase subunit, numberedevery 50 residues. The preliminary model wastraced as two polypeptide chains which werenumbered as residues 1-274 and 300-357. The amino acid types have been estimated from the shapes of the electron densities (the amino acid sequence is currently being determined).

Preliminary Crystal Structure

156

Glutaminase-Asparaginase of a

parameters: pseudo-norxpalized structure factors (Eh) to a? E,,' minimum of 0.8,30-A radius of integration, and 10-4.7-A data. A cross-rotation function map computed with the native A . glutaminasificans glutaminase-asparaginase versus E. coli asparaginase data sets contained a single maximum, 9.0 u above the map average, at a = p = y = 90" ( K = 180", J. = 135", 4 90"). A map obtained with F, values derived from the 2.9-A A. glutaminasificans glutaminase-asparaginase model versus E. coli asparaginase Fo values contained two maxima, both 6.0 u above the map average, at a, (3,y of (go", go", 90") and (go", 75', 90"). Refinement of these solutions with the rigid bodyprogram TRAREF (Huber andSchneider, 1985) gave K , J., and 4 values of (179.1", 136.6", 90.0") and (178.6", 128.1", 90.0"). These solutions indicate that the A. glutaminasificans glutaminase-asparaginase and E. coli asparaginase subunits can be brought into correspondence by an about 180" rotation about an axis that is close to thediagonal of the bc face of the unit cell. The relative orientation is one that maintains the subunit-subunit interactionsin A. glutaminasificans glutaminase-asparaginase and E. coli asparaginase, which further emphasizes their structural similarities. The Structure of Pseudomonas 7A Glutaminase-AsparagiWe-The Pseudomonas 7A glutaminase-asparaginase crystal asymmetric unit contains a complete tetramer with approximate 222 noncrystallographic symmetry. The preliminary Pseudomonas 7A glutaminase-asparaginase self-rotation and A. glutaminasificans glutaminase-asparaginase versus Pseudomonas 7A glutaminase-asparaginase cross-rotation function calculations, as well as A. glutaminasificans glutaminase-asparaginase versusPseudomonas 7A glutaminaseasparaginase translation function studies, have been reported previously (Ammon et al., 1983). A tetramer built from the A. glutuminasificans glutaminase-asparaginase model has been oriented and positioned in the Pseudomonas 7A glutaminaseasparaginase unit cell. A Pseudomonas 7A glutaminase-asparaginase electron density map obtained after density modification and averaging (Bricogne, 1976) over the noncrystallographic tetramer, at 3.0-A resolution, clearly shows a helices and a (3 sheet. Details of the Pseudomonas 7A glutaminaseasparaginase structure will be reported elsewhere.

=.

CONCLUSIONS

The structure of A. glutaminu$ficans glutaminase-asparaginase has been determined to 3-A resolution by a combination of standard MIR phasing and two different density modification techniques. Thestructure has been fitted with a guessed amino acid sequence and has been refined to an R factor of 0.300. The A. glutaminasificans glutaminase-asparaginase subunit contains eight a helices and a (3 sheet. Each subunit folds into two domains, although the connectivity must be treated aspreliminary until thecomplete amino acid sequence is available. The amino-terminal domain folds into a five-stranded parallel /3 sheet surrounded by five a helices. The A. glutaminasificans glutaminase-asparaginase model has been successfullycorrelated with the diffraction data for other amidohydrolases through the use of molecular replacement. The atomic coordinates of the preliminary A. glutaminasificans glutaminase-asparaginase structure have been used in rotation/translation function studies of the three other amidohydrolases, E. coli asparaginase, V. succinogenes asparaginase, and Pseudomonas 7A glutaminase-asparaginase. The rotation function solutions with the A. glutaminasificans glu-

taminase-asparaginase Fo values and model F, values are essentially identical. Pseudomonas 7A glutaminase-asparaginase crystals have a full tetramer in the asymmetric unit; an interpretable electron density map has been obtained starting from an A. glutaminasificans glutaminase-asparaginase tetramer that had been oriented and positioned in the Pseudomonas 7A glutaminase-asparaginase unit cell. Once the complete amino acid sequence is available for A. glutaminasificans glutaminase-asparaginase, it should be possibleto resolve any ambiguities in the connectivity of the polypeptide chain. Acknowledgments-We thank Dr. John Holcenberg for the supply of protein in the initial stages of this work and for fruitful discussions; the Genex Corp., Gaithersburg, MD, for access to their x-ray data collection facility; and Dr. Andrew Howard for assistance in establishing the crystallographic parameters for E. coli asparaginase from the area detector data. REFERENCES Ammon, H. L., Murphy, K. C., Wlodawer, A., Holcenberg, J. S., and Roberts, J. (1983) Acta CrystauOgr. Sect. B: Strut. Sci. B 3 9 , 250-257 Ammon. H.L.. Mumhv. K. C.. Chandrasekhar.. K... and Wlodawer.. A. (1985) J. . Mol. hid. 1 8 4 , 1?9:181 ' Born, L., and Bauer, K. (1970) Naturwissenschnften 67,545 Bricogne, G. (1976) Acta Crystallogr. Sect. A: Found.Crystallogr. A32,832-847 Broome, J. D. (1981) Cancer Treat. Rep. 66,111-114 Chou, P. Y., and Fasman, G. D. (1978) Annu. Rev. Biochem 47,251-276 Crowther, R. A. (1972) in The Molecular Replucement Method (Rossmann, M. G., ed) pp. 173-178, Gordon and Breach Science Publishers, Inc., New York Dickerson, R. E., Weinzierl, J. E., and Palmer, R. A. (1968) Acta Czystallogr. Sect. B: Struct. Sci. B24,997-1003 Epp, O., Steigemann, W., Formanek, H., and Huber, R. (1971) Eur. J. Biochem. 20,432-437 Gonzalez, R. C., and Wintz, P. T. (1977) Digital Image Processing, pp. l l b 1 3 6 , Addison-Wesley Publishing Co., Reading, MA Hendrickson, W. A. (1985) Methods Enzymol. 116,252-270 Holcenberg, J. S., Ericsson, L., and Roberta, J. (1978) Biochemistry 1 7 , 411dl 7

H&rd, J. B., and Carpenter, F. H. (1972) J. Biol. Chem. 247, 1020-1030 Howard, A. J., Gilliland, G. L., Finzel, B. C., Poulos, T. L., Ohlendorf, D. H., and Salemme, F. R. (1987) J. Appl. Crystallogr., in press Hrushesky, W. J., Slavik, M., and Muggia, F. (1976) Proc. Am. SOC. Clin. Annu. Meet. Oncol. 1 7 , 311-321 Huber, R., and Schneider, M. (1985) J. A p 1 Crystdlogr. 18,165-169 Itai, A., Yonei, M., Mitsui, Y.,and Iitaka, 4.'(1976) J . Mol. Biol. 105,321-325 Jones, T.A. (1978) J. Appl. Crystallogr. 11,268-272 Keefer, J. F., Moraga, D. A., and Schuster, S. M. (1985) Biochem. P h u r m o l .

-BA-,

KKO-KGK " I

"Y

Kidd, J. (1970) in Experimental and Clinical Effects of L-Asparuginme, (Gundman, E., and Oettgen, H. F., eds) pp. 1-14, Springer-Verlag, Heidelberg Lee, B., Yang, H. J., Henry, G. M., Seymour, J. P., and Chibata, I. (1975) J. Bwl. Chem. 260,6228-6231 Maita, T., Morokuma, K., and Matauda, G. (1974) J. Bwchem. (Tokyo) 76, 1351-1354 Matthews, B. W. (1968) J. Mol. Biol. 33,491-497 Mitsui, Y., Satow, Y., Watanabe, Y., Hirono, S., Yonei, M., Urata, Y., Torii, K., and Iitaka, Y. (1978) Acta CrystauOgr. Sect. A: Abstructs Eleventh Internotional Congress A34, S60 Murthy, N. s.,and Knox, J. R. (1976) J. Mol. Biol. 106,567-575 North, A. C. T., Wade, H. E., and Cammack, K. A. (1969) Nature 2 2 4 , 594595 Oettgen, H. F.,Old, L. J.. Boyse, E. A., Campbell, H. A., Philips, F. S., Clarkson, B. D., Tallal, L.. Leeper, R. D., Schwartz, M. K., and Kim, J. H. (1967) Cancer Res. 27,2619-2631 Peterson. R. G.. Richards. F. F.. and Handschumacher. R. E. (1977) . . J . Bid. Chem. '262.2072-2076~' Prager, M. D., Baechtel, F. S., and Heifetz, A. (1982) in Bacteria and Cancer, Proceedinas of an International Cohquium (Jeljaszewicz, J., Pulverer, G., and Roszgowski, W., eds) pp. 291-306;AcademicPress, London Richardson, J. (1981) Adv. Protein Chem. 34,167-339 Roberts, J. (1983) Division of Microbial and Biochemical Technology,American Chemical Society Annual Meetln Wash. D. C., Paper MCBT-55 Roberts. J.. Holcenberg. -. J. S., - and f h o w y , W. C. (1972) J. Bid. Chem. 2 4 7 , 84-90 Schmid, F. A., and Roberts, J. (1974) Cancer Chemother. Rep. 68,829-840 Spiers, A. S. D., and Wade, H. E.(1976) Br. Me$. J. l ? , 1317;1319 Steigemann (1974) Ph.D. thesis, Technische Unlversltat, Munchen Wan B C (1985) MethodsEnzymol. 115,90-112 WI&&r, A,, Hodgson, K. 0.. and Bensch, K. (1975) J. Mol. Biol. 99, 295~

~~~~~~~~

~~

133

Wk$wer, A., Roberts, J., and Holcenberg,J. S. (1977) J . Mol. Biol. 112,515ill3

Wriston, J. C., and Yellin, T.0.(1973) Adu. Enzymol. Relot. Areas Mol. Biol. 39,185-248 Yonei, M., Mitsui, Y., and Iitaka, Y. (1977) J. Mol. Biol. 1 1 0 , 179-186