Examination of the Function of Active Site Lysine 329 of Ribulose-bisphosphate Carboxylase/Oxygenase as Revealed by the Proton Exchange Reaction*

Diverse approaches that include site-directed muta- genesis have indicated a catalytic role of Lys-329 of ribulosebisphosphate carboxylase/oxygenase from Rhodoepirillum rubrum. To determine whether Lys-329 is required for the initial enolization of ribulose bisphosphate or for some subsequent step in the overall reaction pathway, the competence of position 329 mu- tant proteins (devoid of carboxylase activity) in catalyzing exchange of solvent protons with the C-3 proton of substrate has now been examined. Irrespective of the amino acid substitution for Lys-329, the mutant protein retains 2-6% of the wild-type activity in the proton exchange reaction. The complete stability of ribulose bisphosphate during the enolization catalyzed by mutant protein suggests that the major effect of Lys-329 is to facilitate the addition of gaseous sub- strates (COz or 0 2 ) to the enediol intermediate. The exchange reaction requires Mg2+, is COa-dependent, and is inhibited by the transition-state analogue 2-carboxyarabinitol1,B-bisphosphate. A mutant protein in which Lys-191, the site for carbamylation by COz in an obligatory activation step, is replaced by a cys- teinyl residue totally lacks proton exchange activity. Barely detectable exchange activity (-0.2% of wild-type) is displayed by the Lys-166 + Cys mutant pro- tein, consistent with the previously implicated role of Lys-166 in the deprotonation of ribulose the assay solution 45 min prior to initiation: PEP carboxylase (0.5 units), malate dehydrogenase (5 units), NADH (3 mM), and PEP (3 mM). Based on the monitoring of NADH oxidation (20) in a separate control, which was identical to the exchange reaction assay solution including 3 mM ribulose-P2 but lacking any form of ribulose-PZ carboxylase, the consumption of endogenous C02/HC0, (300-500 p ~ ) was complete within 90 s. Additional 500 p~ NaHC03 subse-quently added to the same solution was also rapidly consumed, demonstrating the efficiency and excess capacity of the COZ sink.

Both activities are dependent on a unique activation process in which the €-amino group of Lys-191' condenses with CO, yielding a carbamate that is stabilized by catalytically essential Mg2+ (4, 5). The carbamate may be detected and quantified as a component of an exchange-inert quaternary complex formed when the transition-state analogue carboxyarabinitol-Pp binds to the activated enzyme (6).
Subsequent to carbamylation of the enzyme, the initial catalytic step is abstraction of the C-3 proton of ribulose-P2 to generate the corresponding 2,3-enediol. Carboxylase or oxygenase activity then reflects whether C 0 2 or O2 reacts with the enediol (7). In the carboxylation pathway, reaction of CO2 with the 2,3-enediol gives rise to 2-carboxy-3-keto-~-arabini-to1 1,5-bisphosphate (7). This six-carbon reaction intermediate undergoes hydration and carbon-carbon scission to liberate one molecule of PGA derived from C-3, C-4, and C-5 of ribulose-P,. The other product of the scission reaction, a carbanion derived from COZ and C-1 and C-2 of ribulose-Pz, must then undergo inversion and protonation to form the second molecule of PGA (for a more detailed description of the overall reaction pathway, see Ref. 3). Among these partial reactions, the enolization of ribulose-Ps and the formation of PGA from the six-carbon intermediate may be assayed independently of the overall carboxylase reaction (7,8). Enolization is monitored by the exchange of solvent protons with the C-3 proton of ribulose-P,; with six-carbon intermediate, labeled with 14C in the carboxyl group, its conversion to PGA is followed as an increase in acid-stable radioactivity.
The dissection of partial reactions, if catalyzed by sitedirected mutant proteins devoid of overall carboxylation activity, provides an avenue for ascribing the involvement of active-site residues to discrete steps. For example, K166G3 (despite lacking detectable carboxylase or oxygenase activity) undergoes carbamylation and catalyzes the hydrolysis of the six-carbon intermediate; it does not, however, catalyze the enolization reaction (9). These observations provided direct support for the earlier suggestion (10) that the e-amino group of active-site Lys-166 serves as a base required for the enolization of ribulose-Pn.
Residue numbers refer to the R. rubrum carboxylase unless noted otherwise.
Mutant proteins are designated with one-letter abbreviations of amino acids. The first residue named is that found in the wild-type enzyme; the number identifies its location in the polypeptide chain.
Lys-329 is another active site residue of ribulose-P2 carboxylase that has also been probed by site-directed mutagenesis (11). Irrespective of the nature of the amino acid replacement, each of the mutant proteins lacked enzymic activity. The ability of these mutant proteins to bind phosphorylated ligands suggested a role of Lys-329 in catalysis rather than in substrate binding. Conversion of Lys-329 to aminoethylcysteine (net replacement of the y-methylene with sulfur) by aminoethylation of K329C yielded an active enzyme displaying a 4-fold reduced kcat but normal K, values relative to those of wild-type enzyme (12). Thus, the consequences of introducing amino acid substitutions for Lys-329 did not appear to reflect indirect conformational effects but rather the removal of a group involved in catalysis.
If the assignment of Lys-166 to the initial enolization is correct, the participation of Lys-329 likely occurs at some subsequent step in the reaction pathway. Position 329 mutant proteins might then retain the ability to catalyze the enolization of ribulose-P2. Confirmation of this postulate is provided by the present study, in which characteristics and requirements for enolization are also described.

EXPERIMENTAL PROCEDURES
Materials-Commonly used chemicals and reagents were purchased at the highest level of purity readily available. Other commercial materials and vendors were as follows: Bicine, PEP, PEP carboxylase (maize), and malate dehydrogenase (bovine heart) from Sigma; tritiated water (5 Ci/ml) from ICN Radiochemicals. Ribulose-P2 and carboxyarabinitol-Pz were prepared according to published procedures (13, 14). [3-3H]Ribulose-P2 (specific activity = 11,000 cpm/ nmol) was kindly provided by Dr. George H. Lorimer of DuPont or prepared by incubation of ribulose-P2 in tritiated water in the presence of K329C, which catalyzes exchange between the C-3 proton and solvent protons, and reisolation of the labeled ribulose-P2. Details of this procedure will be published elsewhere.
Wild-type ribulose-Pn carboxylase from Rhodospirillum rubrum and its site-directed mutants cloned in Escherichia coli were purified as reported earlier (15,121 and stored at -70 "C as concentrated stocks (10-70 mg/ml) in a pH 8.0 buffer consisting of 50 mM Bicine, 66 mM NaHC03, 10 mM MgC12, 1 mM EDTA, 10 mM 2-mercaptoethanol, and 20% (v/v) glycerol. As needed, aliquots of the stocks were dialyzed at room temperature against storage buffer lacking glycerol. Protein concentrations were based on absorbancies at 280 nm and an 2 % of 12.0 (15). The wild-type enzyme had a specific activity of 4-5 units/ mg in the direct "CO, fixation assay (16,17).
Exchange of Solvent Tritium with the C-3 Proton of Ribulose-P2-The assay procedure was that described by Saver and Knowles (18). Reaction mixtures (final volumes = 200 pl) at pH 8.0 and 23°C contained 20 mCi of tritiated water, 3 mM ribulose-P2, 50 mM Bicine, 10 mM MgCl,, 66 mM NaHC03, 1 mM EDTA, and 5-300 pg of wildtype or mutant carboxylase (the amount depending on the activity level), which was added last. In those cases in which the exchange dependence of M%+ or HCO; was examined, the Bicine buffer lacked the relevant component. Following the addition of 1 p1 of octanol to minimize foaming, the assay solutions were brought to 1 M NaBH4 to terminate the reactions. Thirty min later, glacial acetic acid (7 drops) was added to the quenched reaction mixtures to decompose excess borohydride. Samples were then diluted to 1.5 ml with water and lyophilized to dryness; lyophilization from 1.5 ml of water was repeated several times. Residues were dissolved in 10 ml of water and applied to a 0.8 X 12-cm DEAE-cellulose (Whatman DE52) column equilibrated with 0.05 M NH,HC03 (pH 8.0). Reduced ribulose-P2 and PGA were resolved by elution of the column with a 50-ml linear gradient of 0.05-0.2 M NH4HC03. The collected 1-ml fractions were routinely assayed for radioactivity. If the specific radioactivity of a phosphate ester was required, 300-p1 aliquots of peak fractions were digested at 37°C for 2 h with 60 units of E. coli alkaline phosphatase (Bethesda Research Laboratories). The released inorganic phosphate was quantified with an acid/molybdate reagent (19).
Exchange of Tritium from [3-3H/Ribulose-P2 with Solvent Protom-Conditions and constituents of the assay (8) were essentially as described in the preceding paragraph. The final concentration of ribulose-P, in the 200-pl reaction mixtures ranged from 0.4 to 3 mM; in all cases, 0.08 mM of the total was provided by [3-3H]ribulose-P2 (specific activity -11,000 cpm/nmol). All assays were conducted at the same ionic strength by appropriate adjustments with NaC1. Reactions were initiated with either ribulose-P2 or enzyme (2-5 pg of wild-type or 20-50 pg of mutant). Periodically, aliquots (20 pl) were quenched by diluting into 100 pl of 75 mM NaBH4 contained in scintillation vials; the excess borohydride was decomposed 5 min later with 500 p1 of 2 M acetic acid. Samples were dried in an oven at 110°C for 20 min; residues were dissolved in 500 pl of Hz0 and, after the addition of 10 ml of scintillant (ACS, Amersham Corp.), counted. In those instances in which the exchange reaction was to be examined in the absence of COZ, the following components were introduced into the assay solution 45 min prior to initiation: PEP carboxylase (0.5 units), malate dehydrogenase (5 units), NADH (3 mM), and PEP (3 mM). Based on the monitoring of NADH oxidation (20) in a separate control, which was identical to the exchange reaction assay solution including 3 mM ribulose-P2 but lacking any form of ribulose-PZ carboxylase, the consumption of endogenous C02/HC0, (300-500 p~) was complete within 90 s. Additional 500 p~ NaHC03 subsequently added to the same solution was also rapidly consumed, demonstrating the efficiency and excess capacity of the COZ sink.

RESULTS
General Features of Exchange of Solvent Tritium with the C-3 Proton of Ribulose-P2 as Catalyzed by K329C"When ribulose-P2 is incubated with activated wild-type enzyme in the presence of tritiated water, HCO,, and M e , the reisolated substrate (prior to its complete conversion to PGA) contains tritium at C-3, reflecting protonation of the enediol intermediate (18). This observation is reproduced by the data illustrated in Fig. 1A. Before the addition of enzyme, chromatography on DEAE-cellulose of a solution of ribulose-P2, subsequent to borohydride reduction for stabilization as the corresponding pentitol-P2, does not reveal any radioactivity emerging at the positions for phosphate esters. (The labeled water that remains after lyophilization is eluted during the extensive washing prior to initiation of the gradient and is not shown). Ten min after addition of the wild-type enzyme, the processed reaction mixture contains labeled PGA and labeled pentitol-P2. The carboxylation of ribulose-P2 is completed within 45 min, and a single peak of labeled PGA is observed. The specific radioactivity of this PGA is about onesixth that of solvent protons, in agreement with the reported isotope effect for the terminal proton transfer step (la).
As illustrated in Fig. lB, K329C also catalyzes exchange of solvent protons into ribulose-Pp. Note even a trace of PGA can be detected, consistent with the earlier claim, based on conventional assays with 14C02, that position 329 mutant proteins lack carboxylase activity (11). Omission of bicarbonate from the reaction mixture results in only modest reduction of tritium incorporation. The exchange reaction requires M$+, as in the case of wild-type enzyme (la), and is inhibited by the transition-state analogue carboxyarabinitol-P2. The enhanced labeling of ribulose-P2 (Fig. 1B) compared to that of PGA at completion of the reaction depicted in Fig. 1A suggests that ribulose-Pz fully equilibrates with solvent due to repetitive deprotonation/reprotonation.
Even if the exchange reaction as catalyzed by K329C is allowed to proceed for 6 h (the data in Fig. 1B represent a 1h incubation), the ribulose-Pz concentration, as measured with wild-type carboxylase, remains unchanged.
By analogy with wild-type carboxylase, the incorporation of solvent tritium into ribulose-Pz during incubation with K329C presumably entails exchange of the C-3 proton. The experiment shown in Fig. 1C confirms this expectation. After incubation of ribulose-P2 with mutant protein in tritiated water for 3 h (other conditions as described for Fig. 1, A and  B ) , wild-type enzyme is introduced. Subsequent chromatography reveals PGA as the only radioactive product. Hence, all of the radioactivity associated with the pentitol-Pz peak in  Fig. 1C. Furthermore, the peak area for PGA in Fig. 1C matches that of the end-point sample in Fig. 1A rather than the much larger peak area for pentitol-Pz in Fig. 1B. This pattern is entirely consistent with the fact that the C-3 hydrogen (or titrium) is lost to solvent during the catalytic turnover of ribulose-Pz to PGA. Thus, the degree of labeling of PGA when the carboxylase reaction is carried out in tritiated water must be the same whether [3-3H]ribulose-Pz or unlabeled ribulose-Pz is used.
The time course of the exchange reaction catalyzed by K329C, under the same conditions as those for the experiments summarized by Fig. I, is shown in Fig. 2. The specific radioactivity of ribulose-Pz reaches a maximum of -720 cpm/ nmol, which is -53% that of the solvent (-1360 cpm/nmol 3HzO). Thus, complete equilibration is reached with an indication of a small equilibrium isotope effect. Enolization of [3-3H]Ribulose-Pz-To determine the rates of enolization more conveniently, the loss of tritium from [3-3H]ribulose-P2 to solvent was monitored. As expected from the observed transfer of solvent tritium into substrate, the position 329 mutant proteins also display enolization activity when measured in the reverse direction (Fig. 3A). Little, if any, activity can be detected with K191C (Lys-191 is the site of carbamylation (4, 5 ) ) or with K166C (Lys-166 has been suggested previously to be required in the enolization reaction (10)). Whether or not K329C is preincubated with 66 mM HCO; and 10 mM M$+ to allow carbamylation (activation), subsequent exchange assays at these ligand concentrations reveal similar rates. Presumably, the activation process (carbamylation) is rapid compared to enzyme catalyzed deprotonation of ribulose-Pz so that preactivation is not necessary. Table I summarizes the relative rates of enolization observed with numerous mutant carboxylases and the wild-type enzyme. Irrespective of the substitution for Lys-329, the resultant mutant proteins catalyze loss of substrate tritium at 2-6% of the wild-type rate. A similar exchange activity is observed with E48Q, which retains slight overall carboxylase activity (21,22); Glu-48 is an active site residue of unknown function. The K,,, for ribulose-Pz in the exchange reaction catalyzed by wild-type enzyme is -11 p~, a value not dissimilar to the one determined for overall carboxylation. With K329C, the K,,, for ribulose-Pz in the exchange reaction is increased to -180 p~ (data not shown). Even when the mutant proteins are assayed at 1-1.5 mg/ml in contrast to the wild-type enzyme at 10-40 pg/ml, the threshold for detection of enolization activity is -0.05% of wild-type. Given this limitation, K191C is devoid of activity, but the position 166 mutant proteins catalyze exchange at barely detectable rates. This conclusion has been confirmed by the demonstration of transfer of solvent tritium into substrate with the protocol described in Fig. 1, but longer reaction times (6 h) and higher protein concentrations (1.5 mg/ml) were used (data not shown). The extremely low level of activity precludes reliable rate measurements.
The apparent lack of appreciable COZ dependence of transfer of solvent tritium into substrate (Fig. 1B) prompted a closer examination of enolization with [3-3H]ribulose-Pz. If K329C is preincubated with 66 mM HCO; and 10 mM M P to promote complete carbamylation, the exchange rate is stimulated moderately as the bicarbonate concentration in the assay solutions is increased (Fig. 3B). Without the addition of exogenous bicarbonate to the assay solution, which in the assay solution (0); K329C (150 pg/ml) without added bicarbonate but with the phosphoenolpyruvate carboxylase system in the assay solution (reaction initiated with ribulose-Pz) (A); control lacking any carboxylase (0). C, wild-type (10 pg/ml) (0); wild-type (10 pg/ml) without added bicarbonate in the assay solution (A); wildtype (10 pg/ml) without added bicarbonate but with the PEP carboxylase system in the assay solution (0); wild-type (10 pg/ml) without added bicarbonate but with the PEP carboxylase system in the assay solution (reaction initiated with ribulose-P2) (A); control lacking any carboxylase (0). The insets in B and C show the specific activity nevertheless results in a final concentration of 2 mM (1.6 mM carryover from the stock protein and 0.4 mM endogenous from dissolved air), the exchange rate is about one-half of maximum. This rate is unaltered by depleting the assay solution of free CO2/HC03 by inclusion of PEP carboxylase, PEP, NADH, and malate dehydrogenase, which collectively act as a sink for Cot. In these experiments, designed to determine the influence of C02 on the enolization rate, the reactions  (16,17) and were in agreement with published values (11,12,21,26). The exchange activities were determined by monitoring the loss of tritium from [3-3H] were initiated with preactivated mutant carboxylase. Even in the absence of free CO,/HCO,, the carbamylated form of the enzyme should predominate due to its stabilization by ribulose-Pz (23). By contrast, if the enolization reaction is initiated with ribulose-Pz after incubation of the enzyme in an assay solution that lacks exogenous bicarbonate and contains the PEP carboxylase system so that the non-carbamylated form of the enzyme predominates, the rate of deprotonation is decreased to only 5% of its maximal value (Fig. 3B). A very similar pattern is observed with wild-type enzyme (Fig. 3C).
Based on standard spectrophotometric assays (15, ZO), ribulose-Pz carboxylase is not significantly inhibited by PEP nor is PEP carboxylase significantly inhibited by ribulose-P2 under the conditions described in the legend to Fig. 3. Depletion of CO,/HCO; is complete within 90 s. Thus, the effects of the PEP carboxylase system on the deprotonation reaction as catalyzed by K329C and wild-type enzyme can be ascribed to depletion of COZ/HCO; and hence carbamylated protein.

DISCUSSION
The evaluation of position 329 mutants of the carboxylase as catalysts for the enolization of ribulose-Pz is instructive in several ways. First, the retention of this activity by proteins that are devoid of carboxylase activity demonstrates that enolization can proceed independently of the other partial reactions in the overall carboxylase pathway. Comparisons of the enolization activity possessed by different mutant carboxylases containing single amino acid substitutions for various active site residues should then reveal which side chains are necessary for the initial proton abstraction step. Second, the demonstration that position 329 mutant proteins are indeed enzymes, albeit not carboxylases, which act on ribulose-Pz provides evidence of an active-site topology similar to that of the wild-type enzyme. Apparently, the catalytic deficiency of these mutant proteins is due to the absence of a catalytic 6amino group rather than to indirect conformational changes. Third, the absence of dephosphorylation, isomerization, and/ or epimerization of ribulose-Pz during proton exchange indicates that the enediol never dissociates from the enzyme. Future rapid quench experiments should provide the equilibrium constant between enzyme-bound ribulose-Pz and the corresponding enediol. Fourth, lack of addition of either gaseous substrate to the enediol generated by the position 329 mutant suggest that with wild-type enzyme neither carbox-ylation nor oxygenation is spontaneous but require direct participation by amino acid side chains.
To elaborate on the last assertion in the preceding paragraph, the carboxylase-catalyzed reaction proceeds by a Theorell-Chance type of kinetic mechanism: ordered addition and enolization of ribulose-Pz followed by bimolecular reaction of the enediol with gaseous substrate (1,2). Evidence in favor of this ordered, sequential reaction includes NMR, direct binding, isotope trapping, and kinetic analyses indicating that the gaseous substrates do not bind to free enzyme or to enzyme. ribulose-P2. Earlier speculation (24) that the universal oxygenase activity of ribulose-P2 carboxylases reflects an unavoidable consequence of the inherent reactivity of the enediol is certainly compatible with the elucidated kinetic mechanism. Carboxylation or oxygenation of the enediol could then be viewed as nonenzymic. However, the enolization of ribulose-Pz, as catalyzed by K329C, without concomitant carboxylation or oxygenation (oxygen was never excluded from the incubations), argues that the enzyme plays an active role in facilitating the addition of gaseous substrate to enediol. One possibility is that Lys-329 is needed to enhance the reactivity of the enediol through polarization and development of the nucleophilic center at C-2.
Initial characterizations (11)  In our preliminary report (25), we asserted that the noncarbamylated forms of both wild-type and position 329 mutant proteins exhibit inherent activity in the proton exchange reaction. In retrospect, this conclusion was unjustifiably dogmatic due to our failure to adequately deplete assay solutions of COz. By the use of PEP carboxylase to render assay solutions virtually COZ-free,4 the observed rates of proton exchange with both wild-type and mutant enzymes can be reduced to 5% of their respective maximal values achieved at high bicarbonate concentrations. Whether residual exchange activities are inherent to the non-carbamylated enzyme or reflective of the difficulties of eliminating all COz is unclear. However, the complete absence of exchange activity associated with K191C (Lys-191 is the site of carbamylation), even though this protein binds stoichiometric levels of CO, and displays high affinity for carboxyarabinitol-Pz (261, favors an obligatory role of carbamylation in enolization. Clearly, Lys-329 is not required for the initial deprotonation of ribulose-P2. Irrespective of the nature of the amino acid substitution at position 329, which include glycyl or alanyl residues, the resulting mutant proteins display apparent exchange activities 2-6% as great as that for the wild-type enzyme. The absolute rate of deprotonation of C-3 of ribulose- Pz by the wild-type carboxylase as measured by NMR (1) is 2-3-fold greater than kcat for overall carboxylation; compared to this corrected value, the observed exchange activities of the mutant proteins are reduced to 0.7-3% of wild-type. However, these relative rates are uncorrected for the tritium isotope effect, which may approach the intrinsic kinetic tritium isotope effect as the only chemistry entails cleavage of a C-H (C-T) bond. The intrinsic tritium isotope effect for the R.
rubrum carboxylase has been estimated as >2.6 (27) and 10-16 (2). If the higher value is correct and can be applied to the exchange reaction catalyzed by the mutant proteins, their true activity in this partial reaction is about 30% of wild-type enzyme.
Lys-329 must participate in catalysis at some step subsequent to the enolization of ribulose-Pz. As noted above, the apparent lack of addition of CO, or O2 to the enediol of ribulose-Pz generated by the position 329 mutants suggests that Lys-329 in the wild-type enzyme facilitates this step. Failure of ribulose-P, to be cleaved to products, despite its enolization, does not, however, provide conclusive proof that COS or Oz does not react with enediol. The position 329 mutants could be competent catalyzing reversible addition of gaseous substrates to enediol but incompetent in the conversion of the resultant intermediates to products. However, two observations are inconsistent with this possibility. In the overall carboxylase reaction, the six-carbon, carboxylated intermediate partitions almost exclusively in the forward direction to yield PGA, rather than regenerating enediol via decarboxylation (7). Also, if the deficiency of the position 329 mutant proteins were in the processing of six-carbon, carboxylated intermediate, HCO;(COz) should inhibit the proton exchange reaction. In contrast, bicarbonate actually stimulates the proton exchange reaction. Thus, a more plausible argument can be made for the participation of Lys-329 in the reaction of enediol with CO, or Oz rather than in some later step.
Although we have not thoroughly examined the exchange reaction as catalyzed E48Q, its level of activity is similar to that of the position 329 mutant proteins. Relative to wildtype enzyme, E48Q is only -0.05% as active in overall Carboxylase activity, but >1% (uncorrected for the tritium isotope effect) as active in enolization activity. Thus, as in the case of Lys-329, the major influence of Glu-48 appears to be at some step subsequent to the initial deprotonation of ribulose-Pa.
The recently published (28) 2.3-A structure of activated spinach ribulose-Pz carboxylase, complexed with carboxyarabinitol-Pz, reveals an intersubunit ionic bond between the active site residues Glu-60 and Lys-177 (Glu-48 and Lys-168 of the R. rubrum carboxylase). Therefore, disruption of subunit-subunit association upon replacing Lys-166 or Lys-168 of the R. rubrum carboxylase with an aspartyl or glutamyl residue, respectively, is understandable (29, 30). LYS-334 (which corresponds to Lys-329 of the R. rubrum enzyme) is within ionic bonding distance of the C-2 carboxylate of the bound transition-state analogue, a location compatible with the present postulate that the €-amino group of Lys-329 might enhance the reactivity of the enediol of ribulose-Pz. However, Anderson et al. (28) include Lys-329 as a candidate for the base that enolizes ribulose-Pz; we believe that this possibility is unlikely due to the substantial enolization activity displayed by the position 329 mutant proteins. Furthermore, this lysyl residue has a pKa of 9.0 (lo), whereas the base (irrespective of its identity) that abstracts the C-3 proton from ribulose-Pz has a pK, of 7.5 (2). This latter pK, approximates that of Lys-of the spinach enzyme); this fact, in conjunction with a wealth of chemical and mutagenesis data, which include lack of significant enolization activity by position 166 mutant proteins, provides rather compelling arguments for Lys-166 as a base that accelerates proton abstraction in the first step of the overall reaction. Some doubt is cast on this conclusion by the crystal structure, which shows the lysyi 175 e-amino group closer to the C-1 oxygen than to the C-3 proton of ribulose-Pz. The authors state, however, that the structure of the enzyme-carboxyarabinitol-Pz complex "simulates a later stage in the reaction sequence" rather than the initial enolization. This qualification should be taken quite seriously because of the major conformational differences between the deactivated (non-carbamylated) protein, the activated (carbamylated) protein, and the activated protein complexed with carboxyarabinitol-P,. For example, in the 2.9-A structure of non-carbamylated R. rubrum enzyme (31), Lys-329 is far removed from Lys-191 (the carbamylation site) and probably would not even be classified as an active site residue (precise distances cannot be stated because side chains were not fitted to electron densities and a short segment of a-carbon backbone including Lys-329 was ill-defined'). Also, whereas both the enzyme. carboxyarabinitol-P, structure (28) and the non-carbamylated tobacco enzyme structure (32) confirm an intersubunit location of the active site comprised of side chains from adjacent subunits, as first discovered by hybridization of site-directed mutant proteins (22), the earlier structure of non-carbamylated R. rubrum enzyme prompted a somewhat different conclusion. The low kat for the conversion of the six carbon, carboxylated reaction intermediate to PGA in comparison to the overall kc, for carboxylation of ribulose-Pz has also been interpreted to reflect conformational differences between carbamylated enzyme and carbamylated enzyme complexed with carboxyarabinitol-Pz (7). Given the multiple conformational states of ribulose-P, carboxylase, the translation of structural, chemical, and kinetic observations into plausible mechanistic inferences remains especially challenging.