Communication Vol. 263, No. 7 Issue of March 5, pp. 3643-3646,1988 THE JOURNAL OF BIOLOGICAL CHEMISTRY

0 1988 by The American Society for Bibchemistry and Molecular Biology, Inc. Printed in U.S.A.

Crystal Structure of the Binary Complex of Ribulose- 1,5- bisphosphate Carboxylase and Its Product, 3-Phospho-~-glycerate*

(Received for publication, September 16, 1988) Tomas Lundqvist and Gunter Schneider From the Swedish University of Agricultural Sciences, Department of Molecular Biology, Uppsala Biomedical Center, P. 0. Box 590, ,9751 24 Uppsala, Sweden

The crystal structure of the binary complex of non- activated ribulose- 1,S-bisphosphate carboxylaseloxy- genase from Rhodospirillum rubrum and its produc$ 3-phospho-~-glycerate has been determined to 2.9-A resolution. This structure determination confirms the proposed location of the active site (Schneider, G., Lindqvist, Y., Branden, C.-I., and Lorimer, G. (1986) EMBO J. 5, 3409-3415) at the carboxyl end of the &strands of the a/B-barrel in the carboxyl-terminal domain. One molecule of 3-phosphoglycerate is bound per active site. All oxygen atoms of 3-phosphoglycerate form hydrogen bonds to groups of the enzyme. The phosphate group interacts with the sidechains of resi- dues Arg-288, His-321, and Ser-368, which are con- served between enzymes from different species as well as with the main chain nitrogens from residues Thr- 322 and Gly-323. These amino acid residues constitute one of the two phosphate binding sites of the active site. The carboxyl group interacts with the side chains of His-287, Lys-191, and Asn-111. Implications of the activation process for the binding of 3-phosphoglyc- erate are discussed.

Ribulose bisphosphate carboxylase/oxygenase (ribulose-P2 carboxylase,’ EC 4.1.1.39) has attracted a lot of interest due to its central role in the carbon metabolism of plants and photosynthetic microorganisms (for a review see Andrews and Lorimer, 1987). The enzyme catalyses the primary steps in both the photosynthetic carbon dioxide fixation and in photorespiration. The carboxylation reaction involves addi- tion of carbon dioxide to ribulose bisphosphate, yielding two molecules of 3-P-glycerate. The competing oxygenation re- action, during which molecular oxygen is added to the sub- strate ribulose bisphosphate, yields one molecule of 3-P- glycerate and one molecule of phosphoglycolate, which sub- sequently enters the glycolate pathway. In this pathway, energy is dissipated as heat and photorespiration thus consid- erably reduces the efficiency of photosynthesis and hence plant productivity.

The enzyme from higher plants and most photosynthetic

* This work was supported by grants from the Swedish Natural Science Research Council and the E. I. du Pont Company. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduer- tisernent” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The abbreviations used are: ribulose-Pz carboxylase, ribulose-1,5- bisphosphate carboxylase; 3-P-glycerate, 3-phospho-D-glycerate.

microorganisms comprises eight large (56 kDa) and eight small subunits (14 kDa). In contrast to the hexadecameric plant enzyme, ribulose-P2 carboxylase from Rhodospirillum rubrum is a dimer of only large subunits (Schloss et al., 1979). The three-dimensional structure of the nonactivated form of ribulose-P2 carboxylase from R. rubrum has been described (Schneider et al., 1986a). The subunit consists of two domains, an amino-terminal mixed @-sheet with helices on both sides and a carboxyl-terminal a/@-barrel domain. The active site has been located at the carboxyl-terminal end of the @-strands of the a/@-barrel in the carboxyl-terminal domain (Schneider et al., 1986a). The subsequent structure determination of two higher plant type ribulose-Pz carboxylases from spinach’ and tobacco (Chapman et al., 1987, 1988) has shown that the structure of the large subunits of the L ~ S S type ribulose-PZ carboxylases is highly similar to the structure of the subunit of the R. rubrum enzyme.

The location of the active site in the structure of ribulose- P2 carboxylase from R. rubrum was determined from the position of some conserved amino acid residues, which are essential for catalysis. In the following communication, we present direct crystallographic evidence for the localization of the active site in ribulose-P2 carboxylase. Furthermore, the structure determination of the binary complex of nonacti- vated ribulose-P2 carboxylase with its product, 3-P-glycerate, allows an assignment of function for some of the conserved amino acid side chains at the active site of the enzyme.

MATERIALS AND METHODS

The carboxylase used in this structure determination is a recom- binant protein (Sommerville and Sommerville, 1984) containing 24 additional amino acids from @-galactosidase at the amino terminus. The crystallization procedure for the native, nonactivated enzyme has been described elsewhere (Schneider et al., 1986b). The crystals are monoclinic, spacegroup P21 with cell dimensions a = 65.5 A, b = 70.6 A, c = 104.1 A, and @ = 92.1”. The binary complex of the enzyme with 3-P-glycerate was prepared by soaking crystals of the native enzyme in crystallization buffer containing 30 mM 3-P-glycerate for 2 days.

All intensity data were collected on a computer controlled STOE single counter four-circle diffractometer at 4 “C in a cold room. The procedures used for data collection have been described in detail elsewhere (Eklund et al., 1976). Using six crystals, a complete dataset to 2.9 A resolution was collected, comprising 21,500 unique reflec- tions. The observed structure factors amplitudes were scaled against the corresponding amplitudes from the native crystals. Electron den- sity maps were computed with coefficients (2/F,/ - IFc/) and (IFo/ - /Fc/), where F, denotes the observed structure factors and F, the calculated structure factors from the refined native model, which at present has a crystallographic R-factor of 19.7% at 2.3 A res~lution.~

The model built from these electron density maps was then refined by restrained parameter least squares minimization using the fast Fourier transform version of the program Prolsq (Konnert and Hen- drickson, 1980) contained in the CCP4 program suite, interspersed with manual model building using the program Frodo (Jones, 1985). The 3-P-glycerate complex has at present a crystallographic R-value of 20.3% (10-2.9 A). The model has reasonably good stereochemistry, as indicated by the root mean square-deviation for bond lengths of 0.018 A.

Anderson, I., Knight, S., Schneider, G., Lindqvist, Y., Lundqvist, T., Branden, C.-I., and Lorimer, G . (1988) Nature, in press.

a T. Lundqvist and G. Schneider, unpublished results.

3643

3644 Binding of Product to Ribulose Bisphsphate Carboxyluse

RESULTS AND DISCUSSION

In the monoclinic crystal form, the whole dimer is in the crystal asymmetric unit. The two subunits are thus crystal- lographically independent, and the structure determination therefore gives independent results for both subunits. The obtained results in this study are, within the error limits of our electron density maps, identical, and we describe the results as valid for both subunits.

The initig difference Fourier electron density map, calcu- lated at 2.9 A resolution, showed one major peak per subunit with a height of five times the standard deviation of the electron density map. A 3-P-glycerate molecule could be fitted into the electron density with the highest difference electron density being the position of the phosphate group. The crystallographic R-value of the initial model was 28%. After 30 cycles of least squares refinement using data between 10 and 2.9 A interspersed with manual adjustment this value had dropped to 20.3%, keeping the geometry of the model close to ideal values. The electron density maps based on this model showed a quality comparable to those for the native structure at the corresponding resolution. For final confirmation, a difference Fourier electron density map, from which the contribution of the 3-P- glycerate had been excluded from the structure factor calculation was calculated. Fig. 1 shows the difference

electron density for the bound product, 3-P-glycerate, in this electron density map.

The active site is located in the interface between the carboxyl end of the a/O-barrel domain of one subunit and the amino-terminal domain of the second subunit (Fig. 2). Resi- dues from both domains are therefore likely to participate in the catalytic mechanism. Residues from both domains also participate in the binding of 3-P-glycerate (Fig. 3).

The phosphate binding site is made up of residues from loops 5, 6, and 7 at the carboxyl end of the a/p-barrel. Side chains from residues Arg-288, His-321, and Ser-368 as well as main chain nitrogens from Thr-322 and Gly-323 are all in proper hydrogen bonding distance to the phosphate group (Figs. 3 and 4). This phosphate binding site is also the binding site of Pt(CN)q-, one of the heavy metal derivatives used for phase determination in the initial structyre determination (Schneider et al., 1986a). This anion is known to bind to general anion binding sites in proteins, eg . a phosphate bind- ing site in liver alcohol dehydrogenase (Eklund et ul., 1976). A preliminary comparison with the structure of the large subunit from the ribulose-Pz carboxylase from spinach showed that this phosphate binding site is, at present resolu- tion, identical to the binding site for the Pz-phosphate of the transition state analog 2-carboxy-~-arabinitol 1,5-bisphos- phate.2

All oxygen atoms of the product are involved in hydrogen bonds to polar or charged amino acid side chains of the protein (Fig. 4). The 0-2 oxygen interacts with the side chain of Ser-

L

FIG. 1. Difference electron density for 3-phosphoglycerate, as observed in the binary complex of ribuloae+l,6-b~hos- phate carboxylase and 3-phosphoglycerate. The map shown here was calculated with coeffkienta (IFo/ - /FJ), where F, are the observed structure factor amplitudes for the binary complex and F, FIG. 2. The active site of ribulose-ps cfuboxyla~e at the are the calculated structure factors from the rehed model without interface between subunits. The a//?-banel domain from one any contributions of the 3-phosphoglycerate molecule to the structure subunit is shown in yellow, and parts of the amino-terminal domain factor calculation. The contour level shown here is two times the from the second subunit are shown in red (C-ar positions only). The standard deviation of the electron density map. product, bound at the active site, ia shown in blue.

Binding of Product to Ribulose Bisphosphate Carboxylase 3645

FIG. 3. Stereo picture of the ac- tive site of ribulose-P2 carboxylase with bound product, 3-phosphoglyc- erate.

FIG. 4. Schematic diagram showing the polar interactions of the 3-phosphoglycerate molecule bound at the active site of ribul-Pa carboxyl^.

368 and the hydroxyl group of 3-P-glycerate forms a hydrogen bond to the sidechain of Asn-111 from the second subunit. The carboxyl group of the 3-P-glycerate is oriented toward His-287, Lys-191, and Asn-111. All those amino acid side chains, interacting with the product, are conserved in the ribulose-P2 carboxylases from different species.

During activation, the e-amino group of Lys-191 becomes carbamylated and is then part of a magnesium binding site (Lorimer, 1981). Activation thus drastically changes the prop- erties of this site from a positive anion binding site to a negative metal binding site. In the nonactivated enzyme, the carboxyl group of 3-P-glycerate interacts with the positively charged lysine side chain. It has been show by epr studies of the complex of activated enzyme with Cu(I1) replacing the Mg(II), and 3-P-glycerate, that oxygen atoms from the prod- uct are part of the coordination sphere of the metal ion. The

structure of the binary complex of enzyme and product places the carboxyl group of the product in the close vicinity of the putative metal binding site and thus supports the results from epr spectroscopy (Styring and Briindkn, 1985).

Many phosphorylated sugar compounds are known to affect the state of activation. Most of these compounds, eg. 3-P- glycerate, are known to enhance activation and are therefore called positive effectors. Attempts have been made to develop a unifying model to account for the many diverse experimental observations (Badger and Lorimer, 1981). From the present model we can suggest a mode of action for these positive effectors. The metal binding site with the carbamate is located at the bottom of the active site. Any phosphorylated sugar compound, which binds with the phosphate group in the phosphate binding site described here and is a ligand of the metal ion will decrease the solvent accessibility of the metal ion and the carbamate and thus minimize the dissociation rate of the ternary complex, keeping the enzyme in the active form.

A preliminary comparison of the structure of nonactivated ribulose-Pz carboxylase from R. rubrum with the structure of the activated spinach enzyme shows that the overall confor- mation of the subunit and the dimer interactions are very similar. The ability of the decarbamylated enzyme to bind the transition state analogue and other effectors implies that the enzyme still can provide a very efficient binding of sugar phosphates. The non-carbamylated enzyme is also catalyti- cally competent, performing the decarboxylation of the six carbon intermediate 2-carboxy-3-keto-arabinitol1,5-bisphos- phate (Pierce et al., 1986).

REFERENCES Andrews, T. J., and Lorimer, G. H. (1987) in The Biochemistry of

Plants (Hatch, M. D., ed) Vol. 10, pp. 131-218, Academic Press, Orlando, FL

Badger, M. R., and Lorimer, G. H. (1981) Biochemistry 20, 2219- 2225

Chapman, M. S., Se Won Suh, Curmi, P. M. G., Cascio, D., Smith, W. W., and Eisenberg, D. (1987) Nature 329,354-356

Chapman, M. S., Se Won Suh, Curmi, P. M. G., Cascio, D., Smith, W. W., and Eisenberg, D. (1988) Science 241,71-74

Eklund, H., Nordstrom, B., Zeppezauer, E., Soderlung, G., Ohlsson, I., Boive, T., Soderberg, B.-O., Tapia, O., Briindbn, C.-I., and

3646 Binding of Product to Ribulose Bisphosphate Carboxylase Akeson, A. (1976) J . Mol. Biol. 102, 27-59 D., and Hartman, F. C. (1979) J . Bacterid. 137,490-501

Jones, T. A. (1985) Methods Enzyrnol. 115, 157-170 Schneider, G., Lindqvist, Y., Branden, C.-I., and Lorimer, G. (1986a) Konnert, J. H., and Hendrickson, W. A. (1980) Acta Crystallogr. A36, EMBO J. 5 , 3409-3415

Lorimer, G. (1981) Biochemistry 20, 1236-1240 187, 141-143 Pierce, J., Andrews, T. J., and Lorimer, G. H. (1986) J . Biol. Chem. Somerville, C. R., and Somerville, S. (1984) Mol. Gen. Genet. 193,

Schloss, J. V., Phores, E. F., Long, M. W., Norton, I. L., Stringer, C. Styring, S., and Brandin, R. (1985) Biochemistry 24, 6011-6019

344-350 Schneider, G., Branden, C.-I., and Lorimer, G. (1986b) J. Mol. Biol.

261, 10248-10256 214-219