The Relationship between Side Reactions and Slow Inhibition of Ribulose-bisphosphate Carboxylase Revealed by a Loop 6 Mutant of the Tobacco Enzyme*

The first directed mutant of a higher plant ribulose-bisphosphate carboxylase/oxygenase (Rubisco), constructed by chloroplast transformation, is catalytically impaired but still able to support the plant's photosynthesis and growth (Whitney, S. M., von Caemmerer, S., Hudson, G. S., and Andrews, T. J. (1999) Plant Physiol. 121, 579–588). This mutant enzyme has a Leu to Val substitution at residue 335 in the flexible loop 6 of the large subunit, which closes over the substrate during catalysis. Its active site was intact, as judged by its barely impaired competency in the initial enolization step of the reaction sequence, and its ability to bind tightly the intermediate analog, 2′-carboxy-d-arabinitol-1,5-bisphosphate. Prompted by observations that the mutant enzyme displayed much less slow inhibition during catalysis in vitro than the wild type, its tendency to catalyze side reactions and its response to the slow inhibitor d-xylulose-1,5-bisphosphate were studied. The lessening in slow inhibition was not caused by reduced production of inhibitory side products. Except for pyruvate production, these reactions were strongly enhanced by the mutation, as was the ability to catalyze the carboxylation of d-xylulose-1,5-bisphosphate. Rather, reduced inhibition was the result of lessened sensitivity to these inhibitors. The slow isomerization phase that characterizes inhibition of the wild-type enzyme by d-xylulose-1,5-bisphosphate was completely eliminated by the mutation, and the mutant was more adept than the wild type in catalyzing the benzylic acid-type rearrangement of d-glycero-2,3-pentodiulose-1,5-bisphosphate (produced by oxidation of the substrate, d-ribulose-1,5-bisphosphate). These observations are consistent with increased flexibility of loop 6 induced by the mutation, and they reveal the underlying mechanisms by which the side reactions cause slow inhibition.

ribulose-P 2 in photosynthetic CO 2 fixation and photorespiration (1)(2)(3)(4)(5). Despite excellent crystal structures of various liganded and unliganded Rubiscos from various sources (6) and intensive mutagenic study of algal (3) and bacterial (7) Rubiscos, progress toward understanding the catalytic mechanism has been hampered by an inability to express the dominant higher plant form of the enzyme in heterologous hosts. This precluded application of the power of directed mutagenesis to the higher plant enzyme until Whitney et al. (8) constructed the first such mutant by chloroplast transformation in the natural host, tobacco. Because it was desirable to maintain photosynthetic viability, a mutation (a Leu to Val substitution at residue 335 of the plastid-encoded large subunit) was chosen with the aim of changing the kinetic parameters (particularly the CO 2 /O 2 specificity (9)) as much as possible without seriously disabling catalytic performance at elevated CO 2 concentration. The substitution successfully reduced the CO 2 /O 2 specificity 4-fold, and the Michaelis constants for CO 2 , O 2 , and ribulose-P 2 2-to 10-fold, while retaining sufficient substrate-saturated activity to allow growth under CO 2 enrichment. The changes were readily apparent in the photosynthetic characteristics of the leaves of the mutant plant (8).
Leu-335 is located near the apex of loop 6 of the Rubisco large subunit, which closes over the active site after ribulose-P 2 binds (6,10). The adjacent residue, Lys-334, is catalytically essential, and it makes contacts with the P1 phosphate and the C2 carboxylate formed by addition of CO 2 (10), which stabilize the transition state for carboxylation (11). The Leu-335 residue makes van der Waals contacts with the P2 phosphate (attached to C5 of the substrate) (10). Thus the effects of the Val-335 substitution were interpreted in terms of its alteration of the position and orientation of the critical ⑀N of Lys-334 (8).
Here we show that the perturbation of loop 6 caused by the Val-335 substitution affects catalysis more pervasively than can be explained by a simple misorientation of the adjacent Lys-334. The alteration exacerbates the inhibitor-producing side reactions, but the changed structure and dynamics of the loop permit the inhibitors to escape from, or be rearranged by, the active site. As a result, fallover inhibition during catalysis is suppressed or eliminated, depending on conditions, revealing the underlying causes of the phenomenon and suggesting how it may have evolved.

EXPERIMENTAL PROCEDURES
Materials-Wild-type tobacco (Nicotiana tabacum L. cv Petit Havana [N,N]) was grown in a naturally illuminated glasshouse. The plastome mutant that produces Rubisco with a Leu to Val mutation at residue 335 of the large subunit (8) was grown in an artificially illuminated growth cabinet in an atmosphere containing 0.3% (v/v) CO 2 . Wild-type and L335V Rubiscos were purified either by crystallization as described previously (29) or by polyethylene glycol precipitation and anion-exchange chromatography (Mono Q, Amersham Biosciences) using a procedure similar to that described for spinach Rubisco (18). Tobacco Rubisco concentrations were estimated from the absorbance at 280 nm, assuming that the protein concentration in mg⅐ml Ϫ1 is given by A 280 ϫ 0.7 (30). 1-3 H-and 3-2 H-labeled and unlabeled ribulose-P 2 were synthesized as described (25,31). Unlabeled and carboxy-14 C-labeled carboxy-pentitol-P 2 were synthesized, and carboxyarabinitol-P 2 was purified from these preparations as described (32). Xylulose-P 2 was synthesized from glycolaldehyde-2-phosphate, prepared by periodate oxidation of D-ribose-5-phosphate (33) and dihydroxyacetone phosphate using rabbit muscle aldolase (34), and purified by the same anion-exchange chromatographic procedure used for ribulose-P 2 (25). All bisphosphate preparations were desalted by using Sephadex G-10 chromatography (25).
Measurement of Activity-Ribulose-P 2 carboxylase activity was determined at 25°C using a modification of a spectrophotometric assay (35). Rubisco was activated at 50°C for 10 min in 100 mM EPPS-NaOH buffer solution, pH 8.0, containing 20 mM MgCl 2 , 1 mM EDTA, and 20 mM NaHCO 3 . Each assay solution (2 ml) contained 100 mM EPPS-NaOH, pH 8.0, 20 mM MgCl 2 , 0.5 mM ribulose-P 2 , 0.5 mM dithiothreitol, 0.2 mM NADH, 2 mM ATP, 10 mM phosphocreatine, 20 mM NaHCO 3 , 0.1 mg⅐ml Ϫ1 carbonic anhydrase, 25 units⅐ml Ϫ1 creatine phosphokinase, 25 units⅐ml Ϫ1 3-phosphoglycerate kinase, 20 units⅐ml Ϫ1 glyceraldehyde phosphate dehydrogenase, 55 units⅐ml Ϫ1 triose phosphate isomerase, and 20 units⅐ml Ϫ1 glycerol phosphate dehydrogenase. The reaction was initiated by the addition of activated Rubisco (3-12 nM active sites), and A 340 was measured at 5-s intervals for 2000 s. When anaerobic conditions were required, the buffer was pre-sparged with N 2 before addition of 20 mM NaHCO 3 , and the assays were carried out in septum-capped cuvettes with the headspace flushed with N 2 . Characteristically of higher plant Rubisco (19), activity declined exponentially from an initial rate (v i ) to reach a final, steady-state rate (v f ). Data for product accumulation versus time were fitted to the following equation (19), where k obs is the observed first-order rate constant. Residual activity versus time was plotted using the parameter estimates obtained and the differentiated form of Equation 1, in Equation 2.
Xylulose-P 2 carboxylase activity was measured using the same method, but with 25-50 M xylulose-P 2 instead of ribulose-P 2 , and using 0.5-2.0 M active sites. Controls lacking Rubisco or ATP showed no activity.
Enolization of Ribulose-P 2 -The enolization rate was measured by SCHEME 1. Side reactions catalyzed by Rubisco. The rates for the wild-type (Leu-335) tobacco Rubisco (where known) are shown as fractions of the substratesaturated rate of carboxylation. The analogous partitionings for the Val-335 mutant Rubisco are in parenthesis. Reactions where the partitioning (relative to carboxylation) is enhanced in the mutant are indicated by block arrows; the one where it is reduced is shown with a dashed arrow.
NMR spectrometry by observing the rate at which the deuteron of [3-2 H]ribulose-P 2 was exchanged with a proton from solution. Assays were carried out at 25°C in buffer containing 50 mM Tris-HCl, pH 8.0, 10 mM MgCl 2 , 1 mM EDTA, 10% (v/v) 2 H 2 O, and 2.9 mM [3-2 H]ribulose-P 2 . Reactions were initiated by the addition of Rubisco (to 3 M sites) that had been activated as described above. Spectra were collected at 2-min intervals using a Varian INOVA 600-MHz NMR spectrometer. The water resonance was suppressed by the W5 WATERGATE pulse (36). The increasing intensity of the resonance at 4.37 ppm due to the 3-H proton with time was fitted to an exponential equation, and the initial rate was corrected for the 10% 2 H 2 O present in the solvent. No exchange occurred in controls lacking enzyme.
Reaction Products-Assays were carried out at 25°C in buffer containing 100 mM EPPS-NaOH, pH 8.0, 20 mM MgCl 2 , 1 mM EDTA, and 1 M [1-3 H]ribulose-P 2 . Assays under carboxylating conditions used buffer that had been sparged with N 2 before the addition of 20 mM NaHCO 3 ; assays under predominantly oxygenating conditions used buffer that had been sparged with O 2 and had a final concentration of NaHCO 3 (carried over with the preactivated enzyme solution) of 10 M. Assays were initiated with preactivated enzyme (described earlier) to a final concentration of 3.7 M active sites and terminated at 10 min after complete consumption of the ribulose-P 2 . To measure the total reaction products, the reaction was terminated by adding SDS to a final concentration of 2% (w/v). Denatured protein was removed using a Millipore Ultrafree-MC filter unit (M r 10,000 cutoff) and the filtrate was loaded onto a Amersham Biosciences Mono Q 5/5 column equilibrated with 10 mM EPPS/10 mM boric acid/NaOH, pH 8.0, and the reaction products were separated using a NaCl gradient. The radioactively labeled compounds were detected by online scintillation counting (Ultima-Flo, Packard). The contributions of ribulose-P 2 and xylulose-P 2 to the unresolved pentulose bisphosphate peak were measured by collecting the peak, treating with alkaline phosphatase at pH 9.3, and carrying out chromatographic analysis of the pentuloses at 85°C on a water-equilibrated Bio-Rad HPX-87C column. To measure products that remained bound to the enzyme, the reactions were terminated by ultrafiltration (as described above) before denaturation. The retentate was washed with four 400-l aliquots of buffer before elution of the bound products with a 10 mM EPPS/10 mM boric acid/NaOH, pH 8.0, 2% (w/v) SDS and chromatographic analysis as described above. Radioactivity in the wash fractions was measured by scintillation counting.
Products that remained bound after complete consumption of oxidized ribulose-P 2 were analyzed using a similar procedure. [1-3 H]Ribulose-P 2 (50 M) was oxidized for 3 h with 1 mM CuSO 4 as described (25). Oxidized [1-3 H]ribulose-P 2 (8 M) was reacted at room temperature for 10 min with 10 M preactivated Rubisco active sites in 50 mM EPPS-NaOH buffer, pH 8.0, 15 mM MgCl 2 , 1 mM EDTA, 10 mM NaHCO 3 before separation of Rubisco-bound radioactivity from unbound by gel filtration on a 1-ϫ 20-cm column of Sephadex G-50 fine (Amersham Biosciences) equilibrated with the same buffer solution. The protein peak was collected, and the radioactivity it contained was measured and released by adding SDS to 2% (w/v). Denatured protein was removed by ultrafiltration, and an aliquot of the filtrate was chromatographed as described above. In some further experiments, 5 mM H 2 O 2 was included in the solution during consumption of the oxidized ribulose-P 2 by Rubisco; in others, addition of the H 2 O 2 was delayed until 5 min after mixing the oxidized ribulose-P 2 with Rubisco. Rubisco-bound radioactivity was separated by gel filtration and measured. The rate of pyruvate production during catalysis was measured spectrophotometrically as described (37) and compared with the rate of P-glycerate production measured with the standard assay described above.
Assays Under CO 2 and O 2 Limitation-Continuous spectrophotometric assays were conducted in N 2 -sparged, septum-capped cuvettes with a minimal initial CO 2 /HCO 3 Ϫ concentration (50 -100 M), limited to that carried over with the preactivated enzyme preparation (activated as described above), and in the presence of excess ribulose-P 2 (1 mM), so that ample ribulose-P 2 remained after CO 2 /HCO 3 Ϫ exhaustion. Deoxypentodiulose-P formation was measured from the absorbance of its quinoxaline adduct with o-phenylenediamine as described by Morell et al. (18) with a final Rubisco concentration of 10 M active sites used to initiate the assay. Xylulose-P 2 formation was measured using a method based on that described by Edmondson et al. (22) in a buffer solution containing 100 mM EPPS-NaOH, pH 8.0, 20 mM MgCl 2 , 1 mM EDTA, 1 mg⅐ml Ϫ1 rabbit muscle aldolase, 0.15 mg⅐ml Ϫ1 rabbit muscle glycerol phosphate dehydrogenase, 0.2 mM NADH, and 1 mM ribulose-P 2 . Reactions were initiated by addition of preactivated Rubisco (final concentration, 4 M active sites), and the rate of decrease in A 340 was measured. Both assays were conducted at 25°C for 10 min after initiation, and rates of absorbance change in controls lacking enzyme were subtracted.
Inhibition by Xylulose-P 2 -Rapid equilibrium inhibition by xylulose-P 2 (competitive with respect to ribulose-P 2 ) was measured as described (38) by adding preactivated enzyme (final concentration, 12 nM sites) to initiate spectrophotometric assays containing mixtures of ribulose-P 2 (2.5-320 M) and xylulose-P 2 (0 -40 M). Initial activities for each substrate/inhibitor combination were measured, and the data for all combinations were fitted simultaneously to the hyperbolic equation for competitive inhibition to estimate the K m (ribulose-P 2 ) and K i (xylulose-P 2 ).
Slow inhibition of initially fully carbamylated Rubisco was monitored using a method based on that of Zhu and Jensen (23). Rubisco (0.05-1 M sites) was fully carbamylated by incubation with 100 mM EPPS-NaOH buffer, pH 8.0, 20 mM MgCl 2 , 20 mM NaHCO 3 , 1 mM EDTA. Then 0.4 -500 M xylulose-P 2 was added, and, at intervals thereafter, assays were initiated by adding a 200-l sample of this mixture to a 2-ml spectrophotometric assay mixture as described above. Initial activities were measured, the data were analyzed, and the various parameters were estimated as described below.
Inhibition by Carboxyarabinitol-P 2 -Binding of carboxyarabinitol-P 2 to Rubisco was measured using the method of Pierce et al. (32), adapted for use with the spectrophotometric assay. Rubisco (10 -50 nM sites) was fully carbamylated by incubation with 100 mM EPPS-NaOH buffer, pH 8.0, 20 mM MgCl 2 , 20 mM NaHCO 3 , 1 mM EDTA, and the other components required for the spectrophotometric assay (see above). Then 0.1-2.0 M carboxyarabinitol-P 2 was added and the mixture was incubated at 25°C for various periods before assays were initiated by adding ribulose-P 2 to 0.5 mM. The initial activities measured decayed exponentially as a function of the duration of incubation with carboxyarabinitol-P 2 . The data were analyzed, and the various parameters were estimated as described below. Release of carboxypentitol-P 2 was measured by observing the exchange of enzyme-bound [ 14 C]carboxypentitol-P 2 with unbound, unlabeled carboxypentitol-P 2 using a gel-filtration procedure to separate bound and unbound label (32).
Analysis of Slow Binding Inhibition-Both xylulose-P 2 and carboxyarabinitol-P 2 are slow binding inhibitors of Rubisco. This has been modeled previously as a two-step process (32): an initial rapid equilibrium interaction with the active site that competes with the substrate, ribulose-P 2 , followed by a slow isomerization of the initial complex to a much tighter complex that is likely to be associated with closure of flexible loops of the protein over the ligand within the active site (39).
where E represents enzyme, I, the inhibitor, EI the complex in rapid equilibrium with E and I, EI*, the slowly or non-exchanging complex, and k 1 to k 4 are the rate constants of the individual forward and reverse reactions. Assuming that k 3 and k 4 are much smaller than k 1 and k 2 and that [I] Ͼ Ͼ [E], the rate equation for this process is, and v is the Rubisco activity at time t, v i is the Rubisco activity at time zero, E a (ϭ [E] ϩ [EI]) and E t are the concentrations of non-tightly complexed and total Rubisco active sites, respectively, I is the inhibitor concentration, and K i ϭ k 2 /k 1 . When k 4 ϭ 0 (i.e. the tight complex is non-exchangeable and inhibition proceeds to completion, as it does with carboxyarabinitol-P 2 ), these relationships simplify to those used previously (32); however, if k 4 is finite, inhibition proceeds until an equilibrium is reached at which a portion of the enzyme remains unsequestered in the tight complex and active. This active fraction is given by, where v f is the final activity after preincubation to equilibrium. Because of the saturation of the rate of inhibitor binding inherent in the model (Equation 3), a residual active fraction persists even at saturating concentrations of the slow binding inhibitor. This fraction is given by  (40) for slow isomerization binding. Rubisco was incubated for various periods with various concentrations of carboxyarabinitol-P 2 or xylulose-P 2 , and its activity was measured immediately after mixing with ribulose-P 2 as described above. To estimate K i , k 3 , and k 4 , these data were fitted simultaneously to Equations 4 and 5, using multiple curve-fitting software (OriginLab, Northampton, MA).
Carbamylation of Lys-201 in the active site of Rubisco and subsequent binding of a divalent metal ion are prerequisites for catalysis (41). This introduces a complication into the model embodied in Equation 3, because some slow binding inhibitors, such as carboxyarabinitol-P 2 (32), require the carbamylated, metal-complexed active site for tight binding, whereas others such as xylulose-P 2 (23) bind more tightly to the uncarbamylated, metal-free site. This may cause k 1 or k 2 , and thus K i , to depend on the concentrations of CO 2 and divalent metal, but, provided that these concentrations are kept constant and the binding and release of CO 2 and divalent metal are rapid compared with inhibitor release, the above relationships remain applicable.
Activation of Initially Uncarbamylated, Metal-free Rubisco-The standard spectrophotometric assay (see above) was used to measure the rates at which uncarbamylated metal-free Rubisco became active after simultaneous exposure to Mg 2ϩ ions, CO 2 , and ribulose-P 2 . The inactive complexes were formed by incubating 2.2-7.5 M uncarbamylated Rubisco sites in 100 mM EPPS-NaOH buffer, pH 8.0, 1 mM EDTA with or without 80 M xylulose-P 2 for 30 min at 25°C. To initiate activation, these solutions were added to complete assay mixtures containing 20 mM MgCl 2 , 20 mM NaHCO 3 , and 0.5 mM ribulose-P 2 as described above (final Rubisco concentration, 150 nM sites). When the enzyme was preincubated with xylulose-P 2 complex, the xylulose-P 2 concentration carried over into the final assay solution was 400 nM.

RESULTS
The Val-335 Mutation Does Not Interfere with Carboxyarabinitol-P 2 Binding-The Val-335 mutation did not impair the ability of the active site to bind carboxyarabinitol-P 2 stoichiometrically (8). We could detect little exchange of bound [ 14 C]carboxypentitol-P 2 with unbound unlabeled carboxypentitol-P 2 with either the Val-335 mutant Rubisco or the Leu-335 wild type during 9 days at room temperature (data not shown). However, the kinetics of binding was slightly affected. Carboxyarabinitol-P 2 binds in a two-step process, involving a rapid equilibrium phase followed by slow isomerization to form the non-exchangeable complex (32). Although the affinity of carboxyarabinitol-P 2 in the former process was modestly strengthened by the mutation, the maximal rate of the latter isomerization at saturating carboxyarabinitol-P 2 was slightly reduced in the mutant (Table I, see "Experimental Procedures" for details). These data are qualitatively similar to those observed for spinach Rubisco (32,42).
Enolization of Ribulose-P 2 Is Only Slightly Impaired by the Val-335 Mutation-The Val-335 mutant Rubisco enolized [3-2 H]ribulose-P 2 at two-thirds of the rate of the wild-type enzyme (0.6 s Ϫ1 , cf. 0.9 s Ϫ1 , data not shown). This impairment is barely significant compared with the 6-fold reduction in the k cat value for carboxylation that the mutation causes (8).
The Val-335 Mutation Reduces Inactivation during Catalysis-As is always observed with higher plant Rubiscos (19,24,25,43), the activity of the wild-type Leu-335 Rubisco from tobacco declined progressively after addition of ribulose-P 2 . However, this fallover was much less marked with the Val-335 mutant enzyme ( Fig. 1 and Table II). Under aerobic conditions (Fig. 1A), the mutant enzyme lost activity very slowly (k obs one-seventh of that of the wild type), losing only 14% of its initial activity in 500 s, compared with a 43% loss for the wild type. Moreover, the decline of the mutant appeared to be qualitatively different from that of the wild type. Consistent with previous reports (19,25), the data for the wild type, when fitted to Equation 1, estimate a final steady-state rate (v f ) of 32% of the initial rate (v i ). Similar fitting of the mutant's data project to a v f near zero, and this projection was confirmed in longer assays (up to 8000 s, not shown).
Under anaerobic conditions (Fig. 1B), shown previously to partially suppress fallover by limiting oxidation of ribulose-P 2 (25), the decline in activity of the wild-type enzyme was considerably reduced in extent (v f was 74% of v i ) but occurred at a similar rate (Table II). By contrast, exclusion of O 2 abolished fallover by the Val-335 mutant completely.
Xylulose-P 2 Carboxylase Activity Is Increased by the Val-335 Mutation-Consistent with previous observations (44,45), wild-type tobacco Rubisco catalyzed the carboxylation of xylulose-P 2 at a very slow rate (one six-thousandth of the rate with ribulose-P 2 , Fig. 1C and Table II). The kinetics of fallover with xylulose-P 2 as substrate was quite similar to those with ribulose-P 2 . The Val-335 mutant catalyzed xylulose-P 2 carboxylation faster than the wild type both in absolute terms (1.6-fold) and especially in terms of its ratio with respect to ribulose-P 2 carboxylation (10-fold) (Table II). Furthermore, it did so with no signs of fallover. Indeed, an increase in xylulose-P 2 carboxylase activity during xylulose-P 2 carboxylation was observed consistently with the mutant enzyme (Fig. 1C).
The Val-335 Mutation Alters the Amounts of Side Reaction Products-The side-reaction product profile of the wild-type and mutant Rubiscos was observed by chromatographic analysis of the products of conversion of [1-3 H]ribulose-P 2 by a stoichiometric excess of Rubisco active sites under conditions, which promoted carboxylation or oxygenation (Figs. 2 and 3). In all cases, the conversion was virtually complete; only traces of ribulose-P 2 (Ͻ0.5%) remained at the end of the reaction. Small amounts of some early-eluting non-phosphorylated compounds, which included pyruvate and products of trace phosphatase contamination, were poorly resolved by this procedure and were disregarded. Otherwise, all of the products of known side reactions were well resolved, except for the pentulose bisphosphates, which were resolved after dephosphorylation.
Under carboxylating conditions, the Leu-335 wild-type enzyme produced labeled P-glycerate nearly exclusively, and none of it remained bound to the enzyme (Fig. 2, A and B). The Val-335 mutant enzyme produced mostly P-glycerate, along with a trace of carboxytetritol-P 2 (0.3%), some of which remained bound (Fig. 2, C and D). When the assay conditions  promoted oxygenase activity, Leu-335 produced mostly P-glycolate and P-glycerate (76 and 23%, respectively). Because the 3 H label was attached to C1 of the ribulose-P 2 , the P-glycerate molecule produced by oxygenation was not labeled. Again, these products were fully released. Some labeled xylulose-P 2 was also produced (0.3%), most of it remaining bound to the wild-type enzyme (Fig. 3, A and B). The resistance of this trace of bound xylulose-P 2 to consumption by carboxylation implies that it must be bound to inactive, uncarbamylated enzyme likely to be abundant in these assays at low CO 2 concentration. The increased oxygenation, relative to carboxylation, catalyzed by Val-335 Rubisco (8) was reflected in the ratio of [ 3 H]Pglycolate (91%) to [ 3 H]P-glycerate (7%) produced and released. The mutant enzyme also produced several side products: pentodiulose-P 2 (0.4%), xylulose-P 2 (0.4%), and carboxytetritol-P 2 (0.8%), the latter remaining substantially bound to the enzyme (Fig. 3, C and D). With the mutant, the xylulose-P 2 produced did not remain bound.
Other experiments (not shown) revealed that pentodiulose-P 2 was not fully retained on the enzyme during the long period (ϳ2 h) required for the centrifugal ultrafiltration and washing procedure used to separate bound products in these experiments (Figs. 2 and 3). This caused variable retention of labeled pentodiulose-P 2 . Either this species is slowly released itself (consistent with the reported K d of 0.13 M (25)) or it might be slowly converted to monophosphate products on the active site before release. Therefore, the amounts of bound pentodiulose-P 2 detected in those experiments must be considered to be minimum estimates. Carboxytetritol-P 2 appeared to be better retained.
The rate of production of pyruvate, a by-product of the carboxylation reaction resulting from ␤-elimination of the acicarbanion intermediate (Scheme 1) was measured spectrophotometrically under carboxylating conditions. In agreement with previous observations with other Rubiscos (37), the Leu-335 wild-type produced pyruvate at 0.69% of the rate at which it produced pairs of P-glycerate molecules. However, the analogous ratio for the Val-335 mutant was only one-fifth as great (0.13%).
Binding and Processing of Ribulose-P 2 Oxidation Products-As reported previously (25), two of the side-reaction products seen with the Leu-335 mutant enzyme, pentodiulose-P 2 and its product of benzylic acid-type rearrangement, carboxytetritol-P 2 (Scheme 1), can be produced non-enzymatically by oxidation of ribulose-P 2 in the presence of Cu 2ϩ ions. Chromatographic resolution of the oxidation products (Fig. 4A) revealed the presence of both of these compounds, together with a variety of monophosphorylated and non-phosphorylated compounds arising from further degradation of the unstable pentodiulose-P 2 (25). This mixture of oxidation/degradation products, including ϳ40% residual unoxidized ribulose-P 2 , was reacted with Leu-335 and Val-335 Rubiscos to completion under carboxylating conditions. Unbound compounds were then removed by gel filtration, as quickly as possible (within 20 min) to maximize retention of bound pentodiulose-P 2 . The unbound compounds were almost entirely monophosphate species (data not shown). With both Rubiscos, the fraction of radioactivity that remained bound correlated approximately with the total pentodiulose-P 2 plus carboxytetritol-P 2 in the oxidized mixture (Table III). Further information about the identity of the bound species was sought by adding H 2 O 2 , which is known to cleave Rubisco-bound pentodiulose-P 2 and release the cleaved products, P-glycolate and P-glycerate (25). With both wild-type and mutant enzymes, adding H 2 O 2 simultaneously with the oxidized ribulose-P 2 preparation eliminated over 75% of the bound radioactivity. When the H 2 O 2 was added to the reaction mixture 5 min after the oxidized ribulose-P 2 preparation, when conversion to products had been completed, the result was similar with the wild-type Rubisco but different with the Val-335 mutant, which retained 80% of the radioactivity bound in the absence of H 2 O 2 (Table III). Apparently, at least two compounds derived from oxidized ribulose-P 2 must remain bound to Rubisco in these experiments. One is released from the enzyme by H 2 O 2 , and the other is not, and the former is substantially converted to the latter by the mutant enzyme, but not by the wild type. This conclusion was confirmed and the two compounds were identified as pentodiulose-P 2 and carboxytetritol-P 2 by anion-exchange chromatography of the radioactivity bound in the absence of H 2 O 2 and released by treatment with SDS. Whereas the wild-type enzyme bound and retained pentodiulose-P 2 and carboxytetritol-P 2 approximately in the same proportions as they existed in the oxidized ribulose-P 2 preparation (Fig. 4B), the proportions were reversed with the mutant, which had catalyzed the benzylic acid-type rearrangement of most of the pentodiulose-P 2 present in the oxidized ribulose-P 2 to carboxytetritol-P 2 (Fig. 4C). Obviously, the mutation must alter the structure of the active site in a way that facilitates this rearrangement.  (Table II). Using these estimates, curves for the percentage of the initial activity remaining (right ordinate) as a function of time were plotted using Equation 2. Solid lines, Leu-335; dotted lines, Val-335.

The Val-335 Mutation Promotes Side Reactions
Originating from the Enediol Intermediate-Although the rates of "misfire" reactions originating from the enediol intermediate (Scheme 1 (12,13,18,22,23)) are small compared with the rate of CO 2 addition to this intermediate, they can be measured in the absence of CO 2 and O 2 where consumption of ribulose-P 2 by carboxylation and oxygenation is prevented and the fraction of the enzyme with enediol bound is maximized. Although CO 2 is required to activate Rubisco by carbamylation, this requirement can be fulfilled by preincubation in a HCO 3 Ϫ -containing solution. When a small aliquot of preactivated enzyme is sub-sequently added to the assay mixture, the final HCO 3 Ϫ is small and rapidly exhausted by carboxylation. Decarbamylation of higher plant Rubisco in CO 2 -free medium occurs very slowly while ribulose-P 2 persists (20). This device readily allows spectrophotometric measurement (see "Experimental Procedures") of both ␤ elimination of the enediol to produce deoxypentodiulose-P and misprotonation to produce xylulose-P 2 . Both assays showed an initial lag (presumably caused by the coupling systems and consumption of the last traces of CO 2 /HCO 3 Ϫ ) followed

FIG. 2. Chromatography of reaction products under carboxylating conditions. Leu-335 (A and B) and Val-335 (C and D) Rubiscos consumed [1-3 H]ribulose-P 2 by carboxylation and the total (A and C)
and enzyme-bound (B and D) products were chromatographed on an anion-exchange column with a NaCl gradient (indicated in B). See "Experimental Procedures" for details. The numbers in parentheses show the radioactivity recovered in each compound as a percentage of the total radioactivity eluting after 10 min in the total-product chromatogram. by a near linear period. When expressed as percentages of the carboxylation k cat , both activities were substantially increased by the Leu-335 mutation (10-fold for deoxypentodiulose-P and 4-fold for xylulose-P 2 , Table IV).
Inhibition by Xylulose-P 2 -Fully CO 2 /Mg 2ϩ -activated wildtype Rubisco was slowly inhibited by exposure to xylulose-P 2 (Fig. 5A). Similar slow inhibition was seen with spinach Rubisco by Zhu and Jensen (23) who attributed it to the formation of a decarbamylated enzyme-xylulose-P 2 complex. We analyzed xylulose-P 2 inhibition in terms of the model for slow binding inhibition (Equation 3) (40) used previously to model inhibition of Rubisco by carboxyarabinitol-P 2 (32,42). The rate constants for slow binding and release, k 3 and k 4 , were obtained by fitting data for the rate of inhibition at varying xylulose-P 2 concentrations to Equations 4 and 5. The results obtained show that, although the K i values for the rapid-equilibrium steps are similar for xylulose-P 2 and carboxyarabinitol-P 2 binding, there are striking differences in the rate constants for the subsequent isomerization step (Table I). First, the isomerization step for xylulose-P 2 binding is very slow (approximately one-hundredth of the rate of the analogous step for carboxyarabinitol-P 2 binding); second, xylulose-P 2 is re-

FIG. 4. Chromatography of enzyme-bound reaction products after complete consumption of oxidized ribulose-P 2 under carboxylating conditions. [1-3 H]Ribulose-P 2 was oxidized with CuSO 4 (chromatogram of the oxidized products shown in panel A) and reacted
to completion with Leu-335 and Val-335 Rubiscos at saturating CO 2 , and the Rubisco-bound products were isolated by gel filtration. These constituted 8.0% (Leu-335) and 6.4% (Val-335) of the radioactivity in the starting oxidized ribulose-P 2 preparation (Table III). Chromatography of the bound compounds released with SDS is shown in B (Leu-335) and C (Val-335).

TABLE III Effect of H 2 O 2 on binding of ribulose-P 2 oxidation products to Rubisco
Experimental details are given under "Experimental Procedures" and the legend of Fig. 4. H 2 O 2 was absent (none) or added (5 mM final concentration) to Rubisco simultaneously with the oxidized [1-3 H]ribulose-P 2 preparation (0 time) or 5 min afterwards when the conversion to initial products had been completed. Rubisco-bound radioactivity was separated rapidly by gel filtration and measured.  Table II), and the solid fitted lines were plotted using these estimates. No inhibition was seen for Val-335 Rubisco, even at 5.0 M xylulose-P 2 (dashed line). B, preactivated Leu-335 (q) and Val-335 (E) Rubiscos were exposed to the stated concentrations of xylulose-P 2 for 60 min before assay. The residual activity (v f ) is plotted as a fraction of the activity seen in the control lacking xylulose-P 2 (v i ). The solid line is drawn according to Equations 5 and 6 using the parameter estimates obtained from the experiment shown in panel A. No inhibition was seen for Val-335 Rubisco at xylulose-P 2 concentrations ranging up to 300 M (dashed line). leased from its tight complex at least 5000-fold faster than is carboxyarabinitol-P 2 ; and third, although the Val-335 mutation perturbs the kinetics of carboxyarabinitol-P 2 only modestly, it completely abolishes the slow isomerization phase of xylulose-P 2 binding so that no slow inhibition was seen, even after prolonged incubation at xylulose-P 2 concentrations 20fold greater than its rapid equilibrium K i (Fig. 5).
As predicted for slow isomerization binding where release from the tight complex occurs at an appreciable rate relative to isomerization (Equation 3), inhibition of the wild-type, Leu-335 Rubisco by xylulose-P 2 plateaued at less than complete inhibition. A fraction of activity was retained (25% in this case), even after prolonged incubation to equilibrium at saturating xylulose-P 2 concentrations (Fig. 5B). Although the data for v f /v i versus xylulose-P 2 concentration may be fitted to Equations 5 and 6, the value of the parameter estimates obtained by this method is limited by strong interdependency between k 3 and k 4 . Nevertheless, a curve drawn using the estimates of these parameters obtained from the data of Fig. 5A fits these data approximately (Fig. 5B).
The data reported in Fig. 5 are the initial activities seen immediately after 10-fold dilution of xylulose-P 2 -treated ECM preparations into the assay solution containing saturating ribulose-P 2 . Under these assay conditions, reactivation slowly occurred. In all cases, this recovery proceeded with a k obs of ϳ7 ϫ 10 Ϫ4 s Ϫ1 (data not shown), which is similar to the k 4 value estimated from the rate of xylulose-P 2 inhibition (Table  I), as expected.
The action of xylulose-P 2 as a rapid equilibrium inhibitor, competitive with respect to ribulose-P 2 , was assessed by measuring the activity of preactivated Rubisco immediately after addition to assay mixtures containing various ribulose-P 2 and xylulose-P 2 concentrations. Both wild-type and mutant enzymes showed classic, competitive inhibition. The K i for xylulose-P 2 was 4.8 M for the Leu-335 form, which is consistent with the value reported for spinach Rubisco (38), and agrees approximately with the estimate obtained from the slow binding experiment (Fig. 5A). For the Val-335 form, the estimate increased to 20.7 M (Table I).
Activation of Uncarbamylated Rubisco in the Presence of Xylulose-P 2 -When uncarbamylated, metal-free Rubisco (E) is added to an assay mixture containing saturating concentrations of CO 2 , Mg 2ϩ , and ribulose-P 2 , activity accelerates as the essential Mg 2ϩ -stabilized carbamate (ECM) forms in the active site. Although the rate of ECM formation is slowed by the presence of ribulose-P 2 , which binds tightly to E (46), activity nevertheless increases from an initial zero to a final steady state where it matches that seen in controls where full CO 2 / Mg 2ϩ activation (ECM formation) was induced before addition of ribulose-P 2 . The observed rate constant for this activation was increased slightly by the Val-335 mutation (Fig. 6), perhaps indicating that ribulose-P 2 binds a little less tightly to the E form of the mutant than the wild type.
The tendency of ligands, such as xylulose-P 2 , to bind to the uncarbamylated E form of Rubisco may be gauged by observing the effect of preincubation of E with the ligand on the rate of activation in the above assay. The substantially reduced rate of activation caused by pre-exposure of the Leu-335 wild type to xylulose-P 2 (Fig. 6A) indicates that xylulose-P 2 binds to E and is released from it much more slowly than the rate of carbamylation under the same conditions without preincubation with xylulose-P 2 . Similarly slow release from the E-xylulose-P 2 complex was reported for spinach Rubisco (23,44). The k obs for xylulose-P 2 release from the wild-type tobacco enzyme estimated by this means (Fig. 6A) is similar to the release parameter, k 4 , estimated from the rate of xylulose-P 2 inhibition of the ECM form (Table I). However, in the experiment shown in Fig.  6A, we can be sure that the starting EI* species is uncarbamylated. Therefore, this observation lends support to the view that the tight complex with xylulose-P 2 is uncarbamylated, regardless of the carbamylation status of the starting, uninhibited enzyme (23,44). In marked contrast, the effect of preexposure to xylulose-P 2 on subsequent activation of the Val-335 mutant was barely perceptible (Fig. 6B).

DISCUSSION
The Leu-335 3 Val Substitution Does Not Grossly Disturb the Active Site of Rubisco-Consistent with previous assessments (8), the Val-335 mutant of tobacco Rubisco is properly assembled and fully soluble. It crystallizes easily and can be purified in quantity by the same procedures used for the wildtype protein. Its apparently unimpaired tight binding of the reaction intermediate analog, carboxyarabinitol-P 2 , and its only slightly impaired capacity for enolization of the substrate indicate that perturbation of the active by the mutation must be quite subtle.
Slow Inhibition during Catalysis Is Less Severe and Different in Character with Val-335 Rubisco-Consistent with the view that fallover inhibition of Rubisco during catalysis is caused by accumulation of slow binding inhibitors at the active site (21,22), assay time-course data (Fig. 1) fit well to Equation 1, which can be derived from the two-step, slow isomerization model (40) of slow inhibition (see "Experimental Procedures"). However, interpretation of the meanings of the kinetic parameters estimated in this way is clouded by observations that at least two FIG. 6. Activation of uncarbamylated, metal-free Rubisco. Uncarbamylated metal-free complexes of Leu-335 (A) and Val-335 (B) Rubiscos, preincubated with (q, dashed line) or without (⅜, solid line) 80 M xylulose-P 2 , or with saturating Mg 2ϩ and CO 2 (OE, dotted line), were added at zero time to spectrophotometric assay mixtures with saturating Mg 2ϩ , CO 2 , and ribulose-P 2 concentrations as described under "Experimental Procedures." Under these assay conditions, the concentration of xylulose-P 2 carried over from the preincubation, when present, was less than one-tenth of its K i and ribulose-P 2 was present at more than 25 times its K m . Therefore, rapid equilibrium inhibition by xylulose-P 2 during the assay may be neglected, and activity should increase exponentially until it matches the initial activity measured in the fully CO 2 /Mg 2ϩ -activated, inhibitor-free controls (v con ). Data for product accumulation versus time were fitted to Equation 1 with v i ϭ 0, v f ϭ v con , and k obs was estimated. different inhibitors are responsible and their concentrations vary during the assay because they are largely generated during the course of the assay itself, either by Rubisco-catalyzed misprotonation of the enediolate intermediate (22) (Scheme 1) or by non-enzymatic oxidation of ribulose-P 2 (25).
In the presence of atmospheric concentrations of O 2 , the Val-335 mutation greatly reduced the severity of fallover inhibition and altered its character in the sense that the inhibition proceeded to completion (i.e. v f ϳ 0) rather than coming to a steady state where a finite fraction of the activity remained, as with the wild type (Fig. 1A). This suggests that the residual inhibition seen with the Val-335 mutant is caused by an inhibitor whose rate of release (k 4 , Equation 3) from the tight complex (EI*) is close to zero, unlike wild-type fallover where the finite steady-state v f indicates that k 4 must be appreciable compared with k 3 .
Exclusion of O 2 reduced the extent of fallover inhibition of the wild-type enzyme (Fig. 1B), consistent with previous observations that non-enzymatic oxidation of ribulose-P 2 to pentodiulose-P 2 occurring during the assay was one of the main causes of the fallover phenomenon (25). Total abolition of fallover inhibition of the Val-335 mutant enzyme under the same conditions (Fig. 1B) is consistent with pentodiulose-P 2 , or a product derived from it, being the sole cause of the residual inhibition seen with the mutant under aerobic conditions.
All Abortive Side Reactions, Except Pyruvate Production, Are Exacerbated by the Val-335 Mutation-Obviously the reduced fallover inhibition seen with the mutant is not a result of reduced production of inhibitory by-products; production of all of these was increased by the mutation, except pyruvate. Another active site mutation, Thr-65 3 Val in Synechococcus PCC6301 Rubisco, also partitioned less product toward pyruvate (12). Apparently, loosening the active site or reducing the catalytic flux can, in some circumstances, aid the stereospecific protonation that completes the carboxylation sequence (Scheme 1).
The increased tendency to catalytic misfire induced by the mutation was measured most quantitatively with the byproducts arising from the enediolate intermediate, xylulose-P 2 and deoxypentodiulose-P, because spectrophotometric assays can be devised (Table IV). The 4-and 10-fold increases in the partitioning of product toward these compounds attest to the impairment of the ability of the mutant enzyme to control proton access to the Re face of C3 of the enediolate and to maintain the planarity of the O1, C1, C2, and C3 atoms that is so essential for suppressing ␤ elimination of the P1 phosphate (2). The monophosphate deoxypentodiulose-P is not likely to be a strong inhibitor of Rubisco, but the decreased fallover inhibition, despite enhanced partitioning toward xylulose-P 2 , suggests that the interaction of this bisphosphate with the active site must be considerably loosened by the mutation.
Further information about misfire products was obtained by chromatographic analysis of total and enzyme-bound products after complete conversion of labeled ribulose-P 2 (Figs. 2 and 3).
Here the gradient of the anion-exchange chromatography was optimized to focus attention on the bisphosphate and carboxylated-monophosphate products. The experiment under predominantly oxygenating conditions (Fig. 3) was the most informative, because the low CO 2 concentration maximized the fraction of active sites with enediol bound, and thus the side reactions that stem from this intermediate. Furthermore, the high O 2 / CO 2 ratio promotes flux through the oxygenation pathway, allowing detection of the pentodiulose-P 2 and carboxytetritol-P 2 products that derive from the peroxyketone intermediate (Scheme 1). Xylulose-P 2 was the only bisphosphate by-product arising from the Leu-335 wild type (Fig. 3A), much of it remaining bound to the enzyme (Fig. 3B). The Val-335 mutant also produced pentodiulose-P 2 and carboxytetritol-P 2 (Fig. 3C) with only the latter remaining bound (Fig. 3D). This supports the view that xylulose-P 2 is not bound tightly by the mutant enzyme. It also demonstrates that the mutant does not fully suppress H 2 O 2 elimination from the peroxyketone intermediate but is able to promote the benzylic acid-type rearrangement of the elimination product, pentodiulose-P 2 , to carboxytetritol-P 2 (Scheme 1), which remains tightly bound at the active site. Enhanced production of pentodiulose-P 2 and carboxytetritol-P 2 was observed previously with mutants of R. rubrum Rubisco (14,15).
The Val-335 Mutation Promotes Rearrangement of Pentodiulose-P 2 to Carboxytetritol-P 2 within the Active Site-Exposure to Cu-oxidized ribulose-P 2 , which contains pentodiulose-P 2 and carboxytetritol-P 2 (25), provided a more direct demonstration of the ability of the active site to bind pentodiulose-P 2 and, in the case of the mutant, to rearrange it to carboxytetritol-P 2 . These experiments benefited from the recognition that, unlike carboxytetritol-P 2 , pentodiulose-P 2 is not stable when bound to Rubisco and, therefore, rapid isolation of the complex by gel filtration is required for optimal detection. These experiments also exploited the ability of added H 2 O 2 to return pentodiulose-P 2 to the catalytic pathway for oxygenation, leading to its release from the enzyme as P-glycolate and P-glycerate (25). The results (Table III and Fig. 4) demonstrate clearly that both wild-type and mutant enzymes bound both pentodiulose-P 2 and carboxytetritol-P 2 and that the mutant converted the former to the latter with much greater facility than the wild type. Similar facilitation of this benzylic acid-type rearrangement was observed previously with the Lys-329 3 Ala (analogous to residue 334 in the higher plant enzyme) mutant of R. rubrum Rubisco (15).
Slow Binding Inhibition of Rubisco-The data (Figs. 5 and 6 and Table I) are consistent with xylulose-P 2 being a slow binding inhibitor following the classic two-step, slow isomerization model (Equation 3) (40). Its inhibition is analogous to the more thoroughly studied inhibition by 2Ј-carboxyarabinitol-P 2 and 4Ј-carboxyarabinitol-P 2 (32,42) but differs in that it is two orders of magnitude slower and that it has an appreciable release rate, so that it comes to an equilibrium where 25% of the activity remains uninhibited even at saturating xylulose-P 2 concentrations. Whether the slow rate of inhibition is related to the preference of xylulose-P 2 for binding to the uncarbamylated E form of the active site (23,44) remains to be determined. This could be assessed by comparison with another slow binding inhibitor, such as 2Ј-carboxyarabinitol-1-phosphate, which is known to prefer the carbamylated site (ECM) (47) but binds less tightly than carboxyarabinitol-P 2 (48). Measurement of the rate of slow inhibition by xylulose-P 2 commencing with E rather than ECM (as in Fig. 5) also might shed light on this question.
Saturation of inhibition by xylulose-P 2 at less than complete inhibition was observed previously (23,38), and this phenomenon is shared by one or more inhibitors remaining in Rubisco assays after complete consumption of ribulose-P 2 (21), 2Ј-carboxyarabinitol-1-phosphate (48,49), and pentodiulose-P 2 (25). Although this might suggest anti-cooperativity (i.e. binding of the ligand to one or more sites on a Rubisco octamer makes binding to subsequent sites less favored) (21), it is more economically explained as being simply the expected consequence of the model for slow inhibition embodied in Equation 3, where a rate of inhibition that becomes saturated at high inhibitor concentrations is coupled with an appreciable rate of release (see "Experimental Procedures").
Alterations to Fallover Inhibition Caused by the Val-335 Mutation-Two different slow binding inhibitors have been definitely implicated in fallover inhibition: xylulose-P 2 produced by misprotonation of the enediolate (22) and pentodiulose-P 2 produced by non-enzymatic oxidation of ribulose-P 2 during its synthesis and storage and also under the conditions of the Rubisco assay (25). Another pentulose bisphosphate misprotonation product, 3-keto-D-arabinitol-1,5-bisphosphate, has been hypothesized to contribute to fallover inhibition (18,22,24,50). However, its identification as its arabinitol product after dephosphorylation and borohydride reduction is compromised by observations that pentodiulose-P 2 can generate predominantly the same product after similar treatment (14), presumably because of stereochemical bias during reduction. Therefore, we do not consider that the contribution of 3-keto-D-arabinitol-1,5bisphosphate to fallover inhibition has been established securely. The observation, that production of putative 3-keto-Darabinitol-1,5-bisphosphate depended on the presence of O 2 (50), heightens suspicion that the species detected was really the ribulose-P 2 oxidation product, pentodiulose-P 2 . If so, xylulose-P 2 and pentodiulose-P 2 may be the only significant causes of fallover inhibition.
One of the central observations of this study is the total abolition of the slow isomerization phase of xylulose-P 2 inhibition by the Val-335 mutation. Rapid equilibrium inhibition by this species still persists in weakened form (Table I), but no sign of the second step of the two-step binding mechanism (Equation 3) was evident. Apparently, the minimal perturbation of loop 6 in the mutant (effectively, removal of one methylene carbon) was sufficient to loosen its interactions with the active site and allow the rapid release of the ligand. The binding of other slow binding bisphosphates, such as pentodiulose-P 2 , might be similarly affected, but the instability of the latter compound precludes detailed study.
When slow inhibition by xylulose-P 2 (and perhaps 3-keto-Darabinitol-1,5-bisphosphate, if it exists) was prevented by the Val-335 mutation and pentodiulose-P 2 production was blocked by anoxia, fallover was abolished completely (Fig. 1B), a situation never encountered previously with any higher plant Rubisco. Four pieces of circumstantial evidence collectively mount a strong case that carboxytetritol-P 2 is the inhibitor responsible for the reduced degree of fallover inhibition retained by the mutant under aerobic conditions (Fig. 1A). First, unlike the wild-type enzyme, the mutant catalyzes the production of significant amounts of pentodiulose-P 2 , the precursor of carboxytetritol-P 2 , when O 2 is present (Figs. 2 and 3). This substantially augments the basal amounts of this dicarbonyl compound produced by non-enzymatic oxidation during the course of the assay, the lack of which causes the moderate alleviation of fallover inhibition of the wild-type enzyme in anoxia (Fig. 1, A and B) (25). Second, the mutation greatly facilitates the benzylic acid-type rearrangement of pentodiulose-P 2 to carboxytetritol-P 2 ( Fig. 4 and Table II). Third, bound carboxytetritol-P 2 must be released from the active site imperceptibly slowly, despite the loosening of loop 6 induced by the mutation. This is shown by the recovery of enzyme-bound carboxytetritol-P 2 even after four-times repeated washing by centrifugal ultrafiltration over a 2-h period, a procedure that led to a loss of most of the bound pentodiulose-P 2 (Figs. 2 and 3). Fourth, consistent with this, the residual fallover inhibition displayed by the mutant projects to complete inhibition, suggesting that it is caused by an inhibitor whose release rate (k 4 , Equation 3) is close to zero.
The Val-335 mutation therefore illustrates the relative contributions of the two main causes of fallover inhibition by eliminating one (by facilitating rapid release of any pentulose bisphosphates produced within the active site by misprotonation) and exacerbating the other when O 2 is present (by permitting rearrangement of pentodiulose-P 2 to carboxytetritol-P 2 , which binds even more tightly). Reduced fallover inhibition of spinach Rubisco observed after chemical modification of some Arg residues with phenylglyoxal (26) might reflect similar facilitation of pentulose bisphosphate release from active sites indirectly perturbed by modification of remote Arg residues.
Alterations to Xylulose-P 2 Carboxylation-The increased capacity to catalyze xylulose-P 2 carboxylation induced by the mutation (Fig. 1C and Table II) presumably is the consequence of looser packing of loop 6 against the substrate. By some means, this apparently allows an increase in the rate of abstraction of the C3 proton from xylulose-P 2 , which, if xylulose-P 2 binds exactly analogously to ribulose-P 2 , would not be positioned appropriately for abstraction by carbamylated Lys-201 (1, 2). The fallover-like decline in xylulose-P 2 carboxylase activity of the wild-type enzyme during assay (Fig. 1C) parallels the time course of xylulose-P 2 inhibition of ribulose-P 2 carboxylase activity measured on subsequent exposure to ribulose-P 2 (Fig. 5A). This presumably reflects the progressive decarbamylation of the active site induced by xylulose-P 2 (23,44). Nor is the lack of fallover-type inhibition with the mutant enzyme's xylulose-P 2 carboxylase activity ( Fig. 1C) surprising, given the mutant's lack of the slow isomerization, second step of xylulose-P 2 binding likely to be required for such decarbamylation. What is surprising is the progressive increase in the apparent xylulose-P 2 carboxylase activity of the Val-335 mutant that occurs during the assay (Fig. 1C). This observation might be consistent with a finite amount of 3-epimerization of xylulose-P 2 (correct reprotonation of the enediol intermediate to produce ribulose-P 2 , Scheme 1) occurring even at saturating CO 2 . Any traces of ribulose-P 2 so produced and released from the enzyme initially would compete very poorly with the vastly higher xylulose-P 2 concentrations present but the competition would become more equal as ribulose-P 2 accumulated, leading to an increasing rate of product formation. In the case of the Val-335 mutant lacking xylulose-P 2 -induced decarbamylation, this process was apparently more than sufficient to offset any residual inhibition caused by production of pentodiulose-P 2 and carboxytetritol-P 2 in the active site. This inhibition might be expected to be slight, because, although pentodiulose-P 2 should be produced in the same fraction of catalytic turnovers under aerobic conditions regardless of whether xylulose-P 2 or ribulose-P 2 was the substrate, turnover with xylulose-P 2 was 600fold slower (Table II).
The rates of carboxylation of xylulose-P 2 were much slower than the rates of xylulose-P 2 synthesis from ribulose-P 2 under CO 2 /O 2 -free conditions (5% for Leu-335, 14% for Val-335; Tables II and IV). However, under the CO 2 -saturated conditions of the usual assays, the steady-state concentrations of the enzyme-enediol complex will be lower and therefore xylulose-P 2 production should be reduced. Indeed, fallover inhibition was observed to be lessened at CO 2 saturation (19). Under these conditions, carboxylation of xylulose-P 2 may make a significant contribution to the total rate of release of xylulose-P 2 .
What Is the Adaptive Advantage of Fallover Inhibition?-The insights into the mechanism of fallover inhibition provided by the Val-335 mutant prompt questions about its adaptive significance. Slow isomerization inhibition of Rubisco, according to Equation 3, whether directed at the free E form or the carbamylated ECM form, provides an elegant device for regulating Rubisco activity that has an inbuilt limit that prevents total inhibition, always permitting a basal level of activity. Coupled with the Rubisco activase system for ATP-powered release of the inhibitor, itself regulated by energy charge and redox conditions (28), it becomes a sophisticated and effective mechanism for regulating Rubisco activity according to the prevailing light intensity, temperature, and requirement for photosynthate. However, the many examples of algal and bacterial Rubiscos that lack fallover inhibition (27,43,51,52) suggest additional significance. Like the Val-335 mutant, fallover-free Rubiscos generally have lower CO 2 /O 2 specificity and lower catalytic effectiveness (k cat /K m for carboxylation), than the higher plant enzyme (53). They also have differences in the sequence of loop 6 and its environs. Perhaps the attainment of greater carboxylase specificity and efficacy, by evolving better complementarity between the active site and the transition state for CO 2 addition, has come at the cost of making the closure of loop 6 over the substrate so precise and so tight that close analogs of the substrate or its enediolized version are unable to escape without external assistance. In co-opting the motor protein, Rubisco activase, as the release agent, and further co-opting the release mechanism for regulatory purposes, natural selection has made an elegant virtue out of an ugly necessity.