Functional analysis of the putative catalytic bases His-321 and Ser-368 of Rhodospirillum rubrum ribulose bisphosphate carboxylase/oxygenase by site-directed mutagenesis.

Numerous candidates have been suggested according to chemical and structural criteria for the active site base of ribulose bisphosphate carboxylase/oxygenase that catalyzes substrate enolization. We evaluate the functional significance of two such candidates, His-321 and Ser-368 of the Rhodospirillum rubrum enzyme, by site-directed mutagenesis. Position 321 mutants retain 3-12% of wild-type rates of both overall carboxylation and the initial enolization, with little effect on Km for CO2 or ribulose bisphosphate. Position 368 mutants exhibit approximately 1% of wild-type carboxylation but 4-9% of enolization, also accompanied by little effect on Km values. The modest catalytic facilitations elicited by these residues are incompatible with either acting as the crucial base. The enhanced efficiency of the position 368 mutants in enolization versus carboxylation clearly indicates that Ser-368 effects catalysis preferentially beyond the point of proton abstraction. Both sets of mutants bind the reaction intermediate analogue, 2-carboxy-D-arabinitol bisphosphate, stoichiometrically. Ligand exchange from complexes with position 321 mutants is increased relative to wild type, whereas complexes with position 368 mutants are more exchange-inert. Therefore, His-321 may assist stabilization of the transition state mimicked by the analogue.

terminal aci-acid to form the second molecule of PGA (derived from COz and C1 and C2 of ribulose-Pz). The activated enediol(ate) intermediate may also react with molecular oxygen, which leads to oxygenolytic cleavage of substrate and diversion of cellular carbon into energetically wasteful photorespiration (5, 6). In contrast to the limited understanding of protein determinants that distinguish between the two gaseous substrates (7,8), the involvement of several acidbase groups at the active site has been invoked on sound chemical grounds to fulfill the minimal requirements of the overall reaction pathway (3). Diverse approaches have identified such groups and have provided insights into some of their precise functions (for reviews, see Refs. 9 and 10 and the citations therein). However, the identity of the base responsible for abstracting the C3 proton of the substrate to form the enediol(ate) intermediate common to both the carboxylation and oxygenation pathways remains an unresolved issue.
Despite a wealth of chemical, mutagenic, and kinetic evidence supporting Lys-166' of the Rhodospirillum rubrum ribulose-Pn carboxylase (and, by analogy, Lys-175 of the spinach enzyme) as the base that mediates the initial enolization step (11)(12)(13), crystallographic studies of both the R. rubrum and spinach enzymes place this conserved residue too far from the C3 proton of substrate for such a role (14,15). Although no other amino acid side chain has, in fact, been found that fulfills this structural criterion in any of the crystallographic structures of the carboxylase yet determined (a limitation due, in part, to the ill-defined orientation of bound ligands (14, E ) ) , a number of other candidates have been offered (14-16). These include His-321 and Ser-368, which are evolutionarily conserved amino acids between the simpler dimeric (Lz) form of bacterial carboxylases and the more complex hexadecameric (L8Ss) form of plant, algal, and cyanobacterial carboxylases. In the x-ray structure of the spinach ribulose-Pz carboxylase complexed with the reaction intermediate analogue carboxyarabinitol-Pz, His-327 (His-321 of the R. rubrum enzyme) is a ligand for one of the phosphate groups, and Ser-379 (Ser-368 of the R. rubrum enzyme) forms a hydrogen bond with the C3 hydroxyl group of the bound analogue ( Fig.  1) (14). The closest approach of an imidazole 2itrogen of His-327 to C3 of carboxyarabinitol-Pp is -5.1 A, whereas the hydroxyl group of Ser-379 at -3.7 A is the closest of any side chain with the potential to mediate proton transfer. These two residues also engage each other as part of an extensive hydrogen-bonding network conserved in LsSs forms of the carboxylase that ultimately leads to a buried glutamyl residue. However, the absence of several members of this hydrogenbonding chain in the R. rubrum enzyme indicates that the entire network is not crucial to carboxylase activity. Recently, Unless otherwise specified, amino acid numbers refer to the position in the R. rubrum enzyme.
Active site residues of ribulose-Pz carboxylase in the immediate vicinity of bound carboxyarabinitol-Pz as determined by x-ray crystallography. The bound inhibitor is shown in the trans conformation with respect to the hydroxyls a t C2 and C3; the current resolution does not permit a distinction between cis and trans (14). Residue numbers refer to the R. rubrum enzyme; B denotes residues from the adjacent subunit. (Adapted from the schematic of the spinach enzyme given in Ref. 16.) the x-ray structure of a catalytically incompetent complex of activated R. rubrum carboxylase with ribulose-P2 has been reported, showing essentially the same location of these residues in the active site relative to bound phosphorylated ligand (15). In the present study, we address directly by site-directed mutagenesis the putative role of His-321 or Ser-368 of the R. rubrum ribulose-P2 carboxylase as the catalytic base responsible for initiating catalysis.
Expression and Mutagenesis-Cloning of the R. rubrum rbc gene into the expression vectors pFL200 and pFL245 was described previously (20). pFL245, derived from pFL200, contains an optimized ribosome binding site that increases the expression efficiency by approximately 40%. Site-directed mutagenesis was effected by the method of Kunkel (21) using dU-substituted template derived from either pFL2OO (for the production of His-321 replaced mutants) or pFL245 (for the production of Ser-368 replaced mutants). All constructs were expressed in Escherichia coli strain MV1190 grown in 2X-YT media (22) supplemented with 1% glycerol and 50 mg/ml ampicillin. The cultures were harvested 2.5 h after induction with 0.1 mM isopropyl /3-D-thiogalactopyranoside.
Protein and Activity Assays-Carboxylase activity was assayed by a modified radiometric filter disk assay (25). Assays were routinely performed a t ambient temperature in a buffer (pH 8.0) containing 50 mM Bicine, 10 mM MgC12, 1 mM EDTA, 25 mM NaH14C03 (-8 mCi/ mmol), and either 400 or 1000 pM ribulose-Pz. Enzymes were preincubated at room temperature for 15-30 min in the same buffer lacking ribulose-P2 to ensure full activation (spontaneous carbamylation of Lys-191 and coordination of active site M e (26)). The reactions were initiated by addition of ribulose-Pz. Aliquots were periodically applied to trifluoroacetic acid-soaked filter disks, which then were dried and counted by liquid scintillation (Eco-Lite liquid scintillant, ICN). The K,,, values for ribulose-P2 were determined by varying its concentration from 18 to 320 p~ a t a fixed bicarbonate concentration of 25 mM. Bicarbonate concentration was varied from 10 to 100 mM (-5 mCi/mmol) for the determinations of its K, values while holding the ribulose-P2 concentration constant a t 1 mM. In these latter experiments, bovine erythrocyte carbonic anhydrase (Sigma) was added at 0.1 mg/ml just prior to initiation of reaction; the ionic strength was maintained a t a constant level by suitable additions of NaC1.
The enolization partial reaction was assayed by monitoring the enzyme-catalyzed detritiation of [3-"H]ribulose-Ps under conditions used above for carboxylase assays (18,27). Despite a lack of rate dependence upon bicarbonate concentration by the mutant enzymes studied here (data not shown), 66 mM NaHCOs (unlabeled) was used to ensure full activation (i.e. carbamylation) throughout the duration of the assays. Reactions were initiated by the addition of [3-:'H] ribulose-P2 (2 mM, -0.2 mCi/mmol); periodically, 15-pl aliquots were quenched by dilution with 100 p1 of freshly prepared 100 mM NaBH4. Quenched samples were evaporated to dryness and subjected to scintillation counting.
Stabilities of the quaternary complexes of the enzymes with carboxyarabinitol-Pz were determined by a gel filtration procedure (11, with I4C-labeled inhibitor (100 p~) for 1 h. Exchange of bound ligand 28). The complexes were formed by incubation of enzyme (10-15 p~) was carried out by adding unlabeled ligand (1 mM) to these incubations and periodically subjecting portions to gel filtration.
Protein concentrations were determined by the use of Bradford's (29) reagent obtained from Bio-Rad. Pure ribulose-P2 carboxylase isolated from R. rubrum was used as the standard.
Enzyme Purifications-All buffers used during purifications contained 10% glycerol. Cell-free extracts were prepared by passing cell suspensions (1 g of cell paste/2 ml of pH 8.0 activation buffer (50 mM Bicine, 10 mM MgCL, 1 mM EDTA, and 66 mM NaHC03) that contained 1 p M leupeptin and 1 mM phenylmethylsulfonyl fluoride) three times through a French pressure cell (Aminco) a t 1000 psi. The mutant enzymes were purified by a modification of a fast protein liquid chromatography protocol that utilizes two consecutive MonoQ (Pharmacia LKB Biotechnology Inc.) anion exchange chromatographic steps (30). Carboxylase-containing fractions were located by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate (Pharmacia PhastSystem) or by assays for catalytic activity when practical. T o process a larger amount of cell paste (5-10 g) than The single-letter code is used to describe mutant proteins. The first letter denotes the amino acid present in the wild-type enzyme at the numbered position. The final letter denotes the amino acid present at the corresponding position in the mutant protein.

Active Site Residues
of Ribulose-P2 Carboxylasc previously (1-3 g), an isocratic ion exchange separation (1 ml of DEAR Fast Flow Sepharose (Pharmacia) per 1 ml ofcell-free extract) was carried out prior to chromatography on MonoQ. The extract was diluted 1:l with additional extraction buffer and applied to the DEAE-Sepharose column, which was washed with two bed volumes of activation buffer followed by two bed volumes of activation buffer containing 20 mM NaCI. The carhoxvlase was then eluted with the same buffer containing 150 mM NaCl and concentrated to <5 ml by ultrafiltration (Amicon Centriprep-30). Prior to application to the first MonoQ column, the concentrate was diluted to 1 0 ml with a low ionic strenkqh, pH 8.0 buffer (5 mM Ricine, 5 mM MgCI,, 0.5 mM EDTA, and 25 mM NaHC0:d and filtered (Millex-HA 0.45 pm filter, Millipore). The first MonoQ column (HR 10/10) was eluted with a NaCl gradient and the second (HI3 10/10) with a Tris-citrate gradient, as described previously (30). The pooled fractions of purified carboxylase obtained from the second MonoQ column were concentrated (Amicon Centricon-30) and dialyzed against activation buffer containing 20'0 glycerol. The dialyzed enzyme solutions were frozen in liquid N, and stored at -80 "C. Wild-t,ype carhoxvlase was isolated from R. rubrum ( 3 1 ).

Expression, Purification, and Structural Integrity of Mutant
Carboxylases-The levels of mutant carboxylases in cell-free extracts of E. coli transformed by pFL200-or pFL245-derived constructs were observed to be 5-10% of the total soluble protein. These levels are comparable to those of wild-type carboxylase produced from these vectors (20). The mutant enzymes were purified to 80-9096 homogeneity as determined by both nondenaturing and denaturing electrophoresis ( Fig.  2; data not shown for position 368 mutants). All of the mutant enzymes comigrate with wild-type enzyme during electrophoresis under denaturing conditions, indicating that full-length gene products are produced. Mobilities identical to that of wild-type carboxylase during nondenaturing electrophoresis ( Fig. 2R; data not shown for position 368 mutants) indicate t h a t all of the mutant carboxylases form stable dimers like t h e wild-type enzyme.
Kinetic Parameters (Table I) 21,500; lysozyme, M , 14,400. Phast(;el (I'harmacia) 12.5"; and 10-15% media were used for denaturing and nondenaturing electrophoresis, respectively, utilizing the I'hnrmacia I'hastSvstem apparatus with "SDS" or "native" hul'fer strips obtained from the supplier. perturbed (<lO-fold increased), indicating that the side chain at position 321 has only minor effect upon the productive binding of substrate. The comparative kvn,/Km, values illustrate t h a t a 50-100-fold reduction of catalytic efficiency in the carboxylation reaction accompanies substitution of His-321. Among this series of mutant proteins, the rather conservative Asn and Gln substitutions exhibit the higher levels of activity.
The extremely low activities of the mutant proteins with Arg and Lys at position 321 (-O.OOFir;) approach the lower limits for quantification of carboxylase activity. The rates of ribulose-P,-dependent '"CO, fixation catalyzed by these mutants are inhibited by low concentrations (1.5-fold excess over active site) of carhoxyarabinitol-P~ (data not shown), demonstrating that the incorporation of radioactivity reflects true carboxylase activity. Precise K , values of H321K and H321H for ribulose-P2 could not be determined because of the high concentration of these proteins 010-20 p~ active site) required to reliably measure carboxylation rates. However, a K,,, -50 p~ was estimated in each case. The inherently low activities of the H321K and H321R mutant enzymes apparently reflect conformational disruption (see below) in addition to the direct consequences of altering a key active site side chain.
The mutant enzymes with substitutions for Ser-368 exhibit lower, yet significant., levels of carboxylase activity than the enzymes with substitutions for His-321. The K , for ribulose-P, with S368A is essentially unchanged from the wild-type value, so removal of the seryl hydroxyl group has no apparent consequence on the binding of ribulose-P,. Replacement of the hydroxyl with a sulfhydryl increases the observed K,, for ribulose-P2 by 10-fold, perhaps reflecting the larger van der Waals radius. The comparable catal-ytic deficiencies of the S368C and S368A mutant enzymes do not support a nucleophilic role for Ser-368 in catalysis.
T h e K,, for bicarbonate is unchanged between wild type and both the H321N and S368A mutants, demonstrating that neither residue is necessary for interactions with suhstrate co,.
Under the routine conditions of the "C fixation assay (25 mM NaHCO,,), no significant decrease in the endpoint of "C fixation (reflective of ril)ulose-P, exhaustion) relative to that with wild-type enzyme is observed in any of the mutant carboxylases (data not shown), implying that the carboxylase:oxygenase activity ratios are not drastically reduced. However, these assays would not have revealed elevated carbox-y1ase:oxygenase ratios.
Catalysis of Enediol Formation-The putative involvement of His-321 and Ser-368 in the enolization partial reaction was directly examined by assaying the efficiencies of the mutant carboxylases in catalyzing detritiation of [3-"H]ribulose-P~. Like any partial step of an enzyme-catalyzed reaction, the rate of deprotonation of ribulose-P, must he at least as great as t h a t of the overall k,.,,,. However, deprotonation is partially rate limiting in the overall reaction, appearing -2-fold slower than kcnt when measured with trace-labeled [:l-,"H]rihulose-T', (27,32). Therefore, if t h e deficiency of a mutant carboxylase is not due to altering the proton acceptor, the apparent rate of enolization could be enhanced relative to overall carboxylation, whereas replacement of the proton acceptor with a functionally inert side chain could result in the initital proton abstraction step becoming solely rate limiting with a consequential increase in kinetic isotope effect and reduction in the ratio of enolization and carboxylation rates.
The relative enolization rates for the His-321 and Ser-368 mutant carboxylases at saturating levels of [B-'H]ribulose-T', are shown in   catalyze this partial reaction at the same rate relative to carboxylation as is observed with wild-type enzyme. Therefore, barring an increase in the kinetic isotope effect of C3 proton abstraction accompanying the amino acid substitutions, these mutants are indeed impaired in the initial proton abstraction step, but not preferentially impaired. In contrast, the Ser-368 replaced carboxylases catalyze the enolization reaction &fold more rapidly than overall carboxylation (Table  11), relative to the wild-type enzyme. Thus, the primary role of the hydroxymethyl side chain of Ser-368 does not appear to be that of the base which accepts the C3 proton of ribulose-P P .
Formation and Stability of Reaction-Intermediate Analogue Complexes-The catalytic competence of ribulose-P, carboxylase requires prior derivitization of an active site lysyl residue (Lys-191) by nonsubstrate Cop to form a carbamate that provides one ligand for the essential MS' (26). Because only the activated form of the enzyme binds the reaction-intermediate analogue carboxyarabinitol-Pa with great tenacity (19,28), formation of this quaternary complex ( E . CO,. M$+. carboxyarabinitol-Pp) provides a convenient diagnostic for the ability of mutant carboxylases to undergo proper activation and to properly bind phosphorylated ligands. Position 321 substituted carboxylases (except for H321K and H321R) bind [2-'4C]carboxyarabinitol-P2 stoichiometrically, in analogy with wild-type enzyme. However, the rates of ligand exchange following challenge with a 10-fold excess of unlabeled inhibitor are increased significantly relative to the exchange rate for wild-type carboxylase (Fig. 3, Table I). Even Virtually no binding of carboxyarabinitol-P, to H321R or H321K could be detected, indicating that complexes are either not formed or are insufficiently stable to permit isolation during gel filtration (-1 h). The fact that these mutants catalyze the carboxylation of ribulose-P,, albeit at extremely low levels (see above), proves that they do bind ribulose-P,, and by inference, carboxyarabinitol-P2. Additionally, because small molar excesses of carboxyarabinitol-P, inhibit the ac-

Active Site Residues
of Ribulose-P2 Carboxylase tivities of these mutants, carboxyarabinitol-P2 appears to bind in competition with substrate. Nonetheless, the inability to isolate a quaternary complex with these mutants argues that Lys or Arg at position 321 structurally perturbs the active site to a much greater extent than do the other replacements. The interaction of carboxyarabinitol-P, with ribulose-Pz carboxylase has been characterized as a two-step process involving initial rapid, reversible binding that induces a slow conformational change to form the final exchange-resistant complex (19,33). Conceivably, the introduction and/or improper alignment of positively charged side chains into the polyvalent, polar region of the active site (14) may preclude or alter the conformational change which normally severely restricts solvent accessibility to this site. Such a disruption could contribute to the catalytic deficiencies of these mutant^.^ In contrast to the His-321 replaced enzymes, the position 368 mutant carboxylases form complexes with carboxyarabinitol-Pz (Figure 3, inset; Table I) that exhibit similar (tIl2 -2 days for S368C) or greater ( t l / , -5.5 days for S368A) stabilities than wild type.

DISCUSSION
This study represents a continuation of our efforts to define the roles of active site residues of ribulose-P2 carboxylase and to uncover the identity of the elusive base that initiates the catalytic pathway by abstraction of the C3 proton of ribulose-P,. Previous chemical and mutagenesis studies revealed Lys-166 as a residue that embodies salient features anticipated for the base involved in proton abstraction. These include enhanced nucleophilicity and an unusually low pK, of 7.9 (34), which matches an inflection observed in the pH dependence of the deuterium isotope effect with [3-2H]ribulose-P2 as substrate (32). Furthermore, the inflection was insensitive to the solvent dielectric constant, compatible with an amine serving as the base. Even more compelling, K166G lacks detectable carboxylase and enolization activities (12) (which defines the contribution of Lys-166 to rate enhancement as >lo5), yet is able to catalyze the turnover of the 6-carbon carboxyketone reaction intermediate to PGA (13).
A direct catalytic role of the lysyl e-amino group was verified by the partial restoration of activity effected by selective aminoethylation of the K166C mutant protein (ll), a covalent modification that generates a side chain differing from a lysyl side chain only in the replacement of the y-methylene group with a sulfur atom. The incomplete restoration of activity (-20%) observed in these experiments may reflect the lower pK, of the aminoethylcysteinyl side chain relative to that of lysyl (9.5 uersus 10.5 in model compounds) (35), thereby supporting a direct correlation between catalytic potential and the pK, of the amino group at position 166.
Despite the diversity of experimental observations that implicate Lys-166 as the base, crystallographic studies argue to the contrary, because the t-amino group in question appears to be too distant (-6.5 A) from the position of the C3 proton of ribulose-P, as deduced from the structure of the spinach enzyme complexed with carboxyarabinitol-P2 (14,16) and the structure of the R. rubrum enzyme complexed with ribulose-P, (15). However, these structures may not accurately repre-The inability of positively charged residues to functionally replace His-321 could indicate that its side chain is neutral in the native enzyme. While this may indeed be the case, such an interpretation must be considered speculative in the absence of direct measurements. T h e charged functionalities of the substitutions (Arg or Lys) may merely extend further into the active-site cavity to globally disrupt the active site, perhaps by approaching Arg-288, the primary residue involved in neutralizing the negative charge of the bound phosphate group. sent the conformational form of the enzyme that predominates at inception of catalytic turnover. Carboxyarabinitol-P, mimics the gem-diol catalytic intermediate which occurs in the reaction pathway beyond the point of CO, addition to the enedioUate). Binding of this inhibitor is biphasic, entailing a conformational change in the second slow stage to reach the final stable complex. Therefore, the positions of active site residues may differ somewhat in this quaternary complex relative to the state of the enzyme that binds ribulose-P, and is poised for C3 proton abstraction. Although the complex of the R. rubrum enzyme with ribulose-P, would appear to provide a more relevant structure for addressing this question, the crystals utilized were catalytically incompetent because of lattice interactions that apparently preclude side-chain movements necessary for catalysis. Consequently, structures representing additional reaction coordinates are needed for rigorous assessments, derived solely from distance measurements, of catalytic function of side chains.
Because differences in interpretation between chemical and structural studies are not yet reconciled, we have undertaken a systematic evaluation by site-directed mutagenesis of other putative bases for mediating enolization of ribulose-P, that have been offered by crystallography; these include His-321 and Ser-368 (14-16). If such an approach is to be credible, the expected properties of a mutant protein that lacks the crucial base must be contemplated. In this regard, consideration of the application of site-directed mutagenesis to other enzymes that catalyze C-H bond cleavages proves instructive. For example, in the case of triosephosphate isomerase, which also catalyzes an enolization reaction, proton transfer is effected by Glu-165 (36-39). Replacement of this residue by Aia or Gly reduces Kcat by >lo6. In actuality, this factor may underestimate the contribution of a single base, because the residual activity of E165A and E165G is still dependent on base catalysis through recruitment of H 2 0 (40). Similarly, replacement of Lys-258 of aspartate aminotransferase, the tamino group of which abstracts the C,-proton of the amino acid substrate complexed with enzyme as an aldimine, decreases the rate of catalysis by lo6-10' (41). Another example is provided by the key base of A5-3-ketosteroid isomerase (Asp-38), which enhances the rate of substrate enolization by lo6 (42). This value, too, may be low because of evidence for recruitment of an alternate, inefficient base such as H 2 0 (43).
Thus, a rate enhancement of at least lo6 provided by a general base which abstracts a proton from a carbon atom appears commonplace among diverse enzymes. Although our results demonstrate stringency for both His-321 and Ser-368 for maximal catalytic efficiency of ribulose-P2 carboxylase, the rather modest catalytic facilitation (10-100-fold) provided by these residues, in contrast to the literature examples cited, appears incompatible with either side chain directly mediating the initial proton abstraction step.
Mutant enzymes with substitutions for His-321 that lack the side chain functionality (H321A), retain hydrogen-bonding potential (H321S), or display isosterism as well as competence in hydrogen-bonding (H321N, H321Q) maintain -5-10% of the wild-type activities in both the carboxylation and enolization reactions. The replacements of Ser-368, designed to remove functionality (S368A) or to increase nucleophilicity while maintaining hydrogen-bonding potential (S368C), allow substantial enolization activities (4-10%) that exceed the corresponding levels a t which these mutant enzymes catalyze overall carboxylation. Whether the 10% relative kc,, in enolization retained by these two classes of mutants should be viewed as a significant residual activity or a drastic loss of activity is a matter of interpretation; we prefer the former Active Site Residues of Ribulose-P2 Carboxylase 24739 based on comparisons presented in the preceding paragraph. One consequence of the preponderance of interacting charged and polar residues at the active site may be that the potential for functionally conservative replacements simply does not exist among the nineteen choices available. A seemingly modest alteration (e.g. S368C) could disrupt the delicately balanced microscopic charge and polarity at the active site with concomitant catalytic consequences. Both His-321 and Ser-368 interact directly with the reaction intermediate analogue and with other proximal side chains. The localized structural perturbation wrought by any amino acid substitution might then be propagated so as to impair catalysis indirectly, even though the gross conformational integrity of these mutants does not appear compromised in that they are dimeric, undergo carbamylation (activation), interact with phosphorylated ligands, and retain some catalytic activity.
If neither His-321 nor Ser-368 serves the role of initial proton acceptor, what are their possible functions? The similar impairments in both enolization and carboxylation activities that accompany substitution of His-321 appear to exclude a preferential influence of this residue on any single step of catalysis and are compatible with conformational perturbations as indirect causes of catalytic deficiencies. Alternatively, the decreased stabilities of the complexes of position 321 mutant proteins and carboxyarabinitol-P2 could be viewed as evidence that His-321 assists in stabilizing a transition state mimicked by the reaction-intermediate analogue.
Such an interpretation is consistent with the three-dimensional structure that shows proximity of the histidyl side chain and the C5 phosphate of the bound analogue (see Fig.  1). Credence to this interpretation is furthered by the observation that the K,,, for ribulose-P2 is insignificantly altered by replacement of His-321, thereby prompting the view that the histidyl residue contributes to the preferential binding of a transition state relative to substrate.
The 5-fold enhanced enolization activities relative to the residual carboxylation rates of position 368 mutants clearly eliminate Ser-368 as a viable candidate for the primary base. Furthermore, if Ser-368 were to mediate enolization, removal of the @-hydroxyl (S368A) should dramatically reduce the rate of enolization, whereas the introduction of a better nucleophile (e.g. the sulfhydryl of S368C) might be less detrimental. Instead, the S368A mutant is even more efficient than s368C in promoting enolization. The increased stability of the carboxyarabinitol-P2 complex of both position 368 mutants appear t o rule out stabilization of the mimicked transition state by Ser-368; however, absence of the seryl hydroxyl could allow an alternate binding conformation that leads to a dead-end complex. In the crystallographic structure of the spinach carboxylase with bound carboxyarabinitol-P2, the relative orientation of the hydroxyls at C2 and C3 (and therefore the exact placement of the C3 hydroxyl group) cannot be defined from the electron density map (14). The denoted interaction of Ser-368 with carboxyarabinitol-P,, which resembles the gem-diol intermediate (a hydrated ketone) but lacks one of the hydroxyls at C3, can only occur if the analogue binds in the conformation with trans-hydroxyls (Fig. 1). Depending on which gem-diol hydroxyl group of the true reaction intermediate interacts with Ser-368, a potential role in assisting addition of Hz0 to enediol(ate) could be envisioned.
If Lys-166 is eliminated by crystallographic studies (which may be premature) and if His-321 and Ser-368 are eliminated by the observations described herein, the only structurally plausible amino acid side chains that remain as prospects for the base that enolizes ribulose-P2 are His-287 and the carbamate of Lys-191. Position 191 mutants totally lack detectable carboxylation and enolization activities. Restoration of partial catalytic competence to inactive K191C was achieved by interaction with aminoethanesulfonate, thereby invoking a catalytic requirement for the carbamate nitrogen but not defining that role (44). The high acidity of carbamate nitrogens does not appear suitable for a role in proton transfer, but rigorous exclusion is not yet warranted by direct experimental observations. In preliminary experiments (45); the catalytic role of His-287 has been explored by mutagenesis.
Our results indicate that substitutions at this position reduce activity levels by about lo4, while allowing stable binding of carboxyarabinitol-P2. Future, detailed kinetic and structural characterization of position 287 mutant proteins should clarify the potential of His-287 as the key base. A final possibility for C3 proton abstraction, based on the x-ray structure of the inactive carboxylase-ribulose-P, complex, entails intramolecular transfer to the C1 phosphate group of substrate (15). This proposal stems from the proximity of the phosphate group to the C3 proton in one of the models deduced from this relatively low resolution area of the structure. While such mechanisms have been documented for other enzymes (46, 47), confirmation or refutation of substrate-assisted carboxylase activity will require further investigation.
Aknowledgments-We express sincere gratitude to Drs. C.-I. Bran-d6n and G. Schneider (Swedish University of Agricultural Sciences, Uppsala) for providing crystallographic coordinates for the spinach carboxylase quaternary complex and the R. rubrum carboxylase binary complex with PGA. Dr. T. S. Soper is acknowledged for discussions and assistance in the initial stages of this work.