Structural characterization and the determination of negative cooperativity in the tight binding of 2-carboxyarabinitol bisphosphate to higher plant ribulose bisphosphate carboxylase.

When CO2/Mg2+-activated spinach leaf ribulose-1,5-bisphosphate carboxylase (EC 4.1.1.39) is incubated with the transition-state analog 2-carboxyarabinitol 1,5-bisphosphate, an essentially irreversible complex is formed. The extreme stability of this quaternary complex has allowed the use of native analytical isoelectric focusing, anion-exchange chromatography, and nondenaturing polyacrylamide gel electrophoresis to probe the mechanism of the binding process and the effects of ligand tight-binding on the structure of the protein molecule. Changes in the chromatographic and electrophoretic properties of the enzyme upon tight binding of the inhibitor reveal that the ligand induces a conformational reorganization which extends to the surface of the protein molecule and, at saturation, results in a 16% decrease in apparent molecular weight. Analysis of ligand binding by isoelectric focusing shows that (i) incubating the protein with a stoichiometric molar concentration of ligand (site basis) results in an apparently charge homogeneous enzyme population with an isoelectric point of 4.9, and (ii) substoichiometric levels of ligand produce differential effects on each of the charge microheterogeneous native enzyme forms. These stoichiometry-dependent changes in electrofocusing band patterns were employed as a probe of cooperativity in the ligand tight-binding process. The tight-binding reaction was shown to be negatively cooperative.

When C02/Mg2+-activated spinach leaf ribulose-1,5bisphosphate carboxylase (EC 4.1.1.39) is incubated with the transition-state analog 2-carboxyarabinitol 1,5-bisphosphate9 an essentially irreversible complex is formed. The extreme stability of this quaternary complex has allowed the use of native analytical isoelectric focusing, anion-exchange chromatography, and nondenaturing polyacrylamide gel electrophoresis to probe the mechanism of the binding process and the effects of ligand tight-binding on the structure of the protein molecule. Changes in the chromatographic and electrophoretic properties of the enzyme upon tight binding of the inhibitor reveal that the ligand induces a conformational reorganization which extends to the surface of the protein molecule and, at saturation, results in a 16% decrease in apparent molecular weight. Analysis of ligand binding by isoelectric focusing shows that (i) incubating the protein with a stoichiometric molar concentration of ligand (site basis) results in an apparently charge homogeneous enzyme population with an isoelectric point of 4.9, and (ii) substoichiometric levels of ligand produce differential effects on each of the charge microheterogeneous native enzyme forms. These stoichiometry-dependent changes in electrofocusing band patterns were employed as a probe of cooperativity in the ligand tightbinding process. The tight-binding reaction was shown to be negatively cooperative.
2-Carboxyarabinitol 1,5-bisphosphate is a potent inhibitor of higher plant ribulose-1,5-bisphosphate carboxylase/oxygenase(Rubisc0') (EC4.1.1.39),thelarge, heteromultimericenzyme that catalyzes the carboxylation and oxygenation of Ru-Po. The interaction of 2-carboxyarabinitol-P2 with C0,/Mg2+activated Rubisco is characterized by a slow process which results in very tight binding (KO 5 10 PM) (1). It is postulated * This work was supported in part by United States Department of Energy Contract DE-AC02-81ER10902 and is published as Paper No. 7429, Journal Series, Nebraska Agricultural Experiment Station. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. To whom all correspondence should be addressed. that tight binding occurs as a result of structural changes in or around the catalytic site which are promoted by similarities between 2-carboxyarabinitol-P2 and 2-carboxy-3-keto-arabinitol 1,5-bisphosphate (2), the 6-carbon transition-state intermediate of the carboxylase reaction (3). Studies on the mechanism of 2-carboxyarabinitol-P2 binding to the active spinach enzyme have demonstrated that this interaction takes place in at least two steps (1).
Step 1: Rubisco + CABP -Rubisco-CABP (active) fast Step 2: Rubisco-CABP -Rubisco*CABP slow According to this mechanism, 2-carboxyarabinitol-P2 (CABP) equilibrates rapidly with C02/Mg2+-activated Rubisco, forming a relatively loose complex (Rubisco-CABP, KI = 0.4 p~) , which may then undergo a slow, irreversible isomerization to the inactive, exchange-inert quaternary complex of enzyme. CO,. Mg2+. CABP (Rubisco*CABP). Support for an ordered binding mechanism is found in the observation of initially competitive kinetics between Ru-P2 and 2-carboxyarabinitol-P2 and the ensuing irreversible, noncompetitive inhibition of enzymic activity (1). Furthermore, the decreased rate of irreversible inhibition of catalysis observed in the presence of Ru-P2 (2) and the essentially irreversible trapping of activator CO, and Me2+ by 2-carboxyarabinitol-P2 (4) suggest that this phosphorylated ligand binds within the catalytic center. While the mechanism of the slow, irreversible inhibition is not clearly understood, qualitative similarities in the UV difference spectra of enzyme-CABP and non-CO,/M$+-activated enzyme-Ru-P, complexes are consistent with a mechanism in which the initial binding of the inhibitor promotes conformational changes similar to those occurring during catalysis and which, owing to the catalytic incompetence of the ligand, subsequently form an inactive, exchange-inert complex (2).
The extent to which 2-carboxyarabinitol-P2 binding distorts the native Rubisco molecule is as yet unanswered, as is the question of whether homotropic effects occur upon subsequent ligand binding. Previous analyses have shown that the C02/M$+-activated carboxylase does not display sigmoidal kinetics with respect to Ru-P, and hence is not positively cooperative (5). However, given the complex activation and catalytic mechanisms of Rubisco (6), as well as the sensitivity of the enzyme and gaseous substrate to variations in ionic strength, kinetic analyses, alone, are not likely to elucidate the allosteric effects of ligand binding. As such, determination of the equilibrium binding distribution of a slow, tight-binding inhibitor within a population of enzyme molecules seems a more fruitful approach for determining the presence or absence of cooperative effects with respect to ligand binding.
In the research described herein, a variety of techniques were employed to analyze the gross structural effects of tight binding of 2-carboxyarabinitol-P* on the COJMg2+-activated spinach carboxylase. These findings were then used to demonstrate that the tight binding of this phosphorylated ligand to CO,/M$+-activated spinach leaf Rubisco displays negative cooperativity. The proof is provided by experiments in which the relative preference of 2-carboxyarabinitol-P2 for binding sites on unligated and partially 2-carboxyarabinitol-P2-saturated Rubisco was determined by native isoelectric focusing (IEF).
Spinach Rubisco Purification and Assay-Homogeneous Rubisco, as determined by denaturing and nondenaturing polyacrylamide gel electrophoresis, was isolated from market spinach leaves according to the method of Brown et al. (10). The protein was stored a t 4 "C as an ammonium sulfate precipitate. Prior to use, the enzyme was dissolved in 25 mM Bicine/NaOH, p H 8.0, and desalted by two passes through centrifugal columns made from 1-ml plastic pipet tips packed with Bio-Rad P-6DG desalting gel (centrifugation time, 2 min a t 1840 X g). Protein concentrations (mg/ml) were determined by multiplying A:hmmrn by 0.61 (11). Calculations of active-site concentrations were based on the approximate native M , of 560,000 (the primary structure indicates a precise M , of 533,256 for the spinach leaf holoenzyme) and 8 sites/holoenzyme. Carboxylase activity was measured following at least 15 min activation with 10 mM MgC1,/20 mM NaH"C03 (0.1 Ci/mol) in an assay medium containing 25 mM Bicine/NaOH, pH 8.0,0.2 mg/ml bovine serum albumin, and 0.1 mM Nap EDTA. Specific activity was typically 2.3 pmol/min/mg protein for a 1-min assay a t 0.5 mM Ru-P2 and 30 "C. The 0.5-ml assays were stopped by adding 100 p1 of 6 M CH&OOH, taken to dryness at 75 "C, and acid-stable IrC dpm were determined by liquid scintillation spectroscopy.
Horizontal Slab Gel Isoelectric Focusing-Enzyme (approximately 3 mg/ml in 25 mM Bicine, pH 8.0) was activated for at least 30 min at 30 "C by adding 0.1 volume, each, of NaHCO, and MgClz (200 and 100 mM, respectively). Aliquots (200 pl) of activated enzyme (7 nmol of active sites) were added to tubes containing varying amounts of 2carboxyarabinitol-Pp in 50 pl of Bicine plus 20 mM NaHCO,, 10 mM MgC12, p H 8.0, and these were incubated for 40 min a t 30 "C. The samples were then gel-filtered by two passes through centrifugal columns, and protein concentration was determined as previously described. The desalted Rubisco samples were immediately applied near the cathodic electrode of a 5% polyacrylamide horizontal slab gel containing 1.6% LKB, pH 5-8, and 0.4% LKB, pH 3.5-10, carrier ampholytes. All samples were applied to the 0.5-mm thick gel in duplicate such that the 205 X 125-mm gel could ultimately be divided into two halves containing identical complements of protein. Electrofocusing was performed for 3 h a t 20 watts (constant power) and 4 "C on a Bio-Rad flatbed electrofocusing apparatus using 1 M NaOH and 1 M H3P04 as the cathodic and anodic electrode solutions, respectively. The pH gradient was measured with a surface pH electrode, and the gel was refocused for an additional 15 min to sharpen the bands. The gel was then cut in half and the two mirror images were immediately fixed in 45% CH3CH20H/10% CH,COOH (v/v). Protein was visualized by staining one-half of the gel in 45% CH,CH,OH/ 10% CH3COOH containing 0.1% (w/v) Coomassie Brilliant Blue R-250 and destaining in 10% ethanol/lO% acetic acid. The remainder of the fixed gel was prepared for fluorography following the procedures outlined by New England Nuclear. The fixed gel was treated with an enhancing agent (EN3HANCE, New England Nuclear), dried, and placed firmly in contact with Kodak X-Omat AR film. The film was exposed a t -80 "C for approximately 3 weeks and then developed. Densitometric analyses of the fluorographs and Polaroid negatives of the stained gels were performed using a Schoeffel SD-3000 spectrodensitometer.
Anion-exchange Chromatography-Enzyme (3-4 mg/ml) was activated as above in 25 mM Tris-HC1, pH 8.0, containing 20 mM NaHC03 and 10 mM MgC12. After a 1-h incubation with [I4C]2-carboxyarabinitol-P2 (20 Ci/mol) at several 2-carboxyarabinitol-P~/holoenzyme molar ratios, samples containing 10-15 mg of Rubisco were applied to a 0.7 X 15-cm column of DE53 anion-exchange resin (Whatman) equilibrated with 150 mM NaC1, 25 mM Tris-HCl, pH 8.2, and eluted with a 150-ml linear, 150-350 mM, NaCl gradient a t 25 "C and a flow rate of 3 ml/min. Samples of eluant protein to be used for IEF analysis were processed as described above except that in the desalting step centrifugation time was reduced to 15 s in order to concentrate the protein. IEF was performed according to the protocol outlined above. In order to eliminate phase separation, aliquots (100 pl) for the determination of binding stoichiometry were dried at 90 "C in the presence of 100 pI of 6 M CH3COOH prior to liquid scintillation counting.
Electrophoretic Analysis-Activated spinach Rubisco, Z-carboxyarabinitol-P2-treated as described above, was subjected to nondenaturing polyacrylamide vertical slab gel electrophoresis. Slabs (1.2-mm thick) of 4.5, 5,6, or 7% separating gel and 3% spacer gel concentrations were prepared using the buffer system described by Laemmli (12), but without sodium dodecyl sulfate, and run at 4 "C until the tracking dye was within approximately 1 cm of the bottom. The position of the dye front was marked, and the gels were stained and destained as previously described. The migration of the bromphenol blue dye and protein bands from the bottom of the 3% spacer gel was measured immediately, and the results were expressed as the ratio of protein to dye migration (RM). Since RM varies exponentially with gel concentration, a plot of log R M uersus separating gel concentration was used to analyze changes in apparent molecular weight (13).
Differential 2-Carboxyarabinitol-P2-binding Experiments-Phase 1 (see Fig. 1): spinach Rubisco (2.6 mg/ml; 4.6 p~ holoenzyme) was activated at 30 "C for a t least 20 min with 20 mM NaHC03 and 10 mM MgCl, in 25 mM Bicine/NaOH, pH 8.0. Aliquots were then incubated for 30 min a t 30 "C wit,h 2-carboxyarabinitol-P, or ["C]2carboxyarabinitol-P, (20 Ci/mol) a t 2-carboxyarabinitol-P, to holoenzyme molar ratios of 0, 2, 4, and 6. Phase 2 (see Fig. 1): aliquots of unligated Rubisco from the same stock of activated enzyme used in Phase 1 were added to the Phase 1 enzyme so as to double either the amount of protein (mg/ml) ( Fig. 1, Protocol A ) or the number of free binding sites (Fig. 1, Protocol B). Nonradioactive 2-carboxyarabinitol-P,, at a concentration sufficient to bring the second aliquot of enzyme to the same partial saturation as the first, was then added. Following another 30-min incubation at 30 "C and centrifugal desalting, samples from both phases were subjected to native IEF on 5% polyacrylamide horizontal slab gels. In order to reduce systematic error in experiments involving both unlabeled and [14C]2-carboxyarabinitol-P,, the concentrations of these two reagents were matched empirically by determining the aliquot size of each required to produce an identical change in the IEF pattern of native spinach Rubisco (see Fig. 2).
Calculation of Ligand Distribution for Differential Binding Experiments-Binding distributions of any noncooperative system with an initially uniform number of binding sites/protein molecule can be calculated by a binomial distribution (14)(15)(16). A generalization of this was derived to apply to arbitrary initial distributions and two-phase irreversible binding. In Equation 1, F,,, is the fraction of protein with m moles of ligand ( L ) bound, n is the total number of of binding sites/enzyme molecule, and k , is the intrinsic probability of tight binding for enzyme species with m ligands irreversibly bound. Simultaneous integration was performed using a fourth-order Runge-Kutta procedure. Agreement to six significant figures with the binomial distribution for an 8-site noncooperative case was obtained (starting with Fo = 1) using 100 steps. gel illustrates the native and 2-carboxyarabinitol-P2-treated spinach carboxylase IEF patterns (Fig. 2). The untreated enzyme routinely focuses as several bands within a diffuse region of protein (11) between pH 5.5 and 5.8, while the 2carboxyarabinitol-P2-saturated enzyme (8 2-carboxyarabinitol-P2/holoenzyme) focuses as a single tight band at a PI of 4.9. These distinct focusing characteristics of the native and ligand-saturated enzymes are independent of protein load and essentially invariant with enzyme preparation or IEF system (including LKB pH 3.5-10, 5-8, or 4-6 gradients; polyacrylamide tube or vertical slab gels or a Bio-Lyte granulated bed). Fractionation, extraction, and assay of focused spinach Rubisco reveal that all portions of the native pattern possess Ru-P2-dependent carboxylase activity. Ixxbation of the activated enzyme with 2-carboxyarabinitol-P2 at ligand to holoenzyme molar ratios of between 1 and 8 results in an essentially linear decrease in enzymic activity (data not shown).
The effect of binding substoichiometric quantities of 2carboxyarabinitol-P2 on the native IEF pattern reveals a complex interaction between ligand tight-binding and the physical properties of the native enzyme forms (Fig. 2). Upon addition of 1 2-carboxyarabinitol-P2/holoenzyme, the diffuse native carboxylase pattern begins to tighten into a series of discrete bands. Concomitant with this change, some protein migrates to a more acidic region of the pH gradient, although the majority of the sample still remains within the boundaries of the native pattern. Two prominent bands, designated N and 8, appear at isoelectric points of 5.45 and 5.30, respectively. N remains a visually distinct band at 2-carboxyarabinitol-P, levels of between 1 and 5/holoenzyme, while 8 is detectable at up to 6 2-carboxyarabinitol-P2/holoenzyme. N is not sufficiently resolved to allow densitometric determination of its relative protein content; scans of p, however, show that it contains 14-19% of the total stained protein at between 1 and 5 2-carboxyarabinitol-P2/holoenzyme. A new series of closely spaced bands accumulates between pH 5.3 and 4.9 as 2-carboxyarabinitol-P2 is increased from 1 to 6/holoenzyme, while protein in the original native region is virtually depleted at a 2-carboxyarabinitol-P2 to holoenzyme ratio of 5 to 1. Between 6 and 8 2-carboxyarabinitol-P2/holoenzyme, the protein pattern tightens markedly, ultimately giving a single tight band at a 2-carboxyarabinitol-P2 to enzyme ratio of 8 to 1. No effect is observed on the PI 4.9 band when the protein is treated with a molar excess of 2-carboxyarabinitol-P2 indicat-ing that, as expected, no more than eight ligands bind to the holoenzyme.
Fluorography of E~z~~~-[ '~C ] C A B P Complexes-In order to examine the relationship between binding stoichiometry, isoelectric point and banding pattern, fluorography and protein staining were performed on identical halves of gels containing ['4C]2-carboxyarabinitol-P2-treated Rubisco. The fluorographs and stained protein patterns on the gels are qualitatively similar a t all ratios of ligand to holoenzyme examined (Fig. 3). Comparison of the untreated protein pattern with the fluorographs at 1 and 2 2-carboxyarabinitol-P2/ holoenzyme reveals that 2-carboxyarabinitol-P2 is bound to bands which have PI values identical to those of the unligated protein. When densitometric tracings of the stained and fluorographic patterns are compared at each of the 2-carboxyarabinitol-P2 stoichiometries examined, no marked differences are observed in the amount of ligand bound per unit of stainable protein as a function of PI. However, limitations in the sensitivity and resolution of fluorography and densitometry preclude a more quantitative analysis of the 2-carboxyarabinitol-P2/protein stoichiometry in each of the bands.
Resolution of Enzyme-CABP Complexes by Anion-exchange Chromatography-In light of the magnitude of change observed in the charge microheterogeneity of native Rubisco upon tight binding of 2-carboxyarabinitol-PZ (Figs. 2 and 3), anion-exchange chromatography was employed in an attempt to separate the treated enzyme on the basis of binding stoichiometry, thus permitting quantitation of the binding distribution. Ion-exchange chromatography has been previously used to resolve chemically modified proteins differing by as little as two net surface charges (17). Similarly, this technique is very sensitive to the distribution of charge on the surface of proteins (18).
The bottom panel of Fig. 4 shows the protein desorption A number of other gradient and buffer systems were also employed, but this system provided the best resolution of the enzyme-CABP complexes.
Measurement of I4C ligand bound to the eluted protein (Fig.  4, top panel) shows a constant stoichiometry of 7.7 2-carboxyarabinitol-Pp/holoenzyme across the elution peak of the 2carboxyarabinitol-Pz-saturated enzyme, while the profiles of enzyme incubated at 1, 2, and 4 2-carboxyarabinitol-P2/holoenzyme show the majority of the eluted protein to have a stoichiometry of within +1 of that present during preincubation. When a mixture of native enzyme and that treated with eight ['4C]2-carboxyarabinitol-Pz/holoenzyme is chromatographed, a single, broad " , , , elution peak results, suggesting that protein-protein interactions may be affecting the resolution of the system. However, determination of 14C radioactivity across the peak indicates that only the front half of the protein profile is radioactive. When eluant samples of enzyme treated at various substoichiometric levels of 2-carboxyara-binitol-P, were electrofocused, those from the leading edge of the elution profile possessed more acidic PI values, as well as slightly higher 2-carboxyarabinitol-P, stoichiometries (see Fig. 4), than those at the trailing end of the peak (data not shown). These observations substantiate the indications from fluorography (Fig. 3) that no large deviations relative to the mean 2-carboxyarabinitol-P2 stoichiometry occur within the enzyme population at each molar ratio of 2-carboxyarabinitol-Pz to holoenzyme examined. In addition, the resolution obtained by anion-exchange chromatography serves as a further verification of the various IEF patterns observed. However, the IEF pattern of non-2-carboxyarabinitol-Pptreated enzyme was not resolved by anion-exchange chromatography, while all eluant samples from chromatographed, 2carboxyarabinitol-Pz-saturated enzyme focused at a PI of 4.9.
Nondenaturing Polyacrylamide Gel Electrophoresis of Enzyme-CABP Complexes-The polyacrylamide gel matrix participates directly in the electrophoretic separation by means of a sieving effect which retards macromolecules. The extent to which sieving slows protein migration is a function of protein shape, while the electrical force on a macromolecule depends only on its net charge. Consequently, as reported by Hedrick and Smith (13), the effect of separating gel concentration on electrophoretic mobility provides information about the size and charge relationships between proteins. When applied to a single protein, this analysis allows a distinction to be made between the effects of ligand binding on the net charge and/or shape of the macromolecule.
Activated spinach Rubisco pretreated with 0, 2, 4, 6, or 10 2-carboxyarabinitol-P2/holoenzyme was electrophoresed on nondenaturing vertical slab gels of 4.5,5,6, and 7% separating gel monomer concentrations. The effect of ligand binding at these various molar ratios on electrophoretic mobility is illustrated in Fig. 5A. The stained gels at each of the different polyacrylamide concentrations demonstrate that each addition of 2-carboxyarabinitol-P, increases protein mobility. A plot of log protein mobility relative to the dye front (R") uersus polyacrylamide gel concentration at the five Z-carboxyarabinitol-Pz stoichiometries examined yields five nonparallel lines which, when extrapolated, intersect at a 2% separating gel concentration (Fig. 5B). At this hypothetical gel concentration, the free protein and the CABP-enzyme complexes would migrate identically since sieving and charge effects would cancel. Since only sieving effects vary with gel concentration, the finding of nonparallel lines extrapolating to a common point near 0% gel concentration strongly suggests that the various enzyme-CABP complexes differ primarily in hydrodynamic shape (13). That is, under the electrophoretic conditions employed, the various protein samples have essentially identical net charge, but differ in molecular size. The magnitude of these conformational differences is highlighted in Table I, where the effect of 2-carboxyarabinitol-P,-binding on the apparent molecular weight of native Rubisco is shown. Since the binding of eight 2-carboxyarabinitol-P2 and 8 CO,/MgZf molecules adds a mass of less than 1% of the -560-kDa Rubisco molecule, the observed differences in hydrodynamic molecular weight must reflect ligandinduced alterations in the shape of the enzyme molecule. The decrease in apparent molecular weight of about 90,000 for the 2-carboxyarabinitol-Pp-saturated enzyme (Table I) documents the magnitude of this structural reorganization, while the finding of changes in electrophoretic mobility with each change in 2-carboxyarabinitol-P, stoichiometry (Fig. 5 ) shows that sequential ligand binding results in sequential changes in protein structure. In addition, it can be seen that the decrease observed in apparent molecular weight is not linear

TABLE I Effect of 2-carboxyarabinitol-Pz tight-binding on the apparent molecular weight of spinach Rubisco
The molar ratio of 2-carboxyarabinitol-Pz (CABP) to holoenzyme was that present during the 40-min preincubation with activated enzyme (see Fig. 5). Apparent masses were calculated using the linear slope-molecular weight relationship given by Hedrick 1). As indicated, the average binding stoichiometry was made constant at the end of each experimental phase (see Fig. 1). The numbers in each column represent the final mean binding stoichiometry of protein from each phase at the end of the simulated two-phase experiment. The range of stoichiometries about each mean is given by the appropriate binomial distribution since no site-site interaction occurs. The numbers in parentheses represent the per cent of total protein from each phase at the end of the experiment. CABP, 2-carboxyarabinitol-P2. with respect to 2-carboxyarabinitol-P2 stoichiometry (Table   I). Since each 2-carboxyarabinitol-P2 molecule contributes the same overall formal charge and mass to the enzyme, this nonlinearity is suggestive of subunit interactions which vary as a function of the number of ligands bound. Differential Binding Experiments-In view of the above findings, a protocol was devised to investigate possible allosteric effects of the 2-carboxyarabinitol-P2-induced conformational changes. An outline of the experimental design is presented in Fig. 1. Two variations of the basic experimental protocol (A, B ) were employed. In Protocol A equal quantities of protein (i.e. total sites), with and without tightly bound 2carboxyarabinitol-P2, competed for the binding of additional 2-carboxyarabinitol-P2 during Phase 2, while in the second (Protocol B ) equal numbers of unligated sites (i.e. sites without ligand irreversibly bound) competed for additional 2-carboxyarabinitol-P2. In both designs, the amount of 2-carboxyarabinitol-P2 added in Phase 2 was sufficient to bring 2-carboxyarabinitol-P2 partial saturation back to the same level obtained at the end of Phase I . It should be noted that the choice of protocol affects the sensitivity of IEF since with Protocol A protein from Phase 1 comprises 50% of the final protein irregardless of 2-carboxyarabinitol-P2 stoichiometry, while the fraction of protein from Phase 1 in the final mixture varies with 2-carboxyarabinitol-P2 stoichiometry in Protocol B (see Table 11). Although Protocol A has the obvious advantage of maintaining a constant protein ratio in each binding phase, it is also subject to random errors as 2-carboxyarabinitol-P2 stoichiometry is increased. Consequently, the number of binding sites and the quantity of 2-carboxyarabinitol-P2 (relative to sites from Phase 1) increase as a function of 2carboxyarabinitol-Pz stoichiometry. For example, if the protein in Phase 1 has 4 2-carboxyarabinitol-P,/holoenzyme and an equimolar concentration of activated Rubisco is added during Phase 2, the unligated enzyme will have twice as many free catalytic sites in solution as the partially bound enzyme or 67% of the total unoccupied sites. In contrast, Protocol B is less subject to such errors since the amount of both enzyme and 2-carboxyarabinitol-P, added in Phase 2 decreases with increasing 2-carboxyarabinitol-PP stoichiometry.
Since both an irreversible and a reversible process are involved in 2-carboxyarabinitol-P2 binding (l), the validity of experiments in which ligand is added in more than one aliquot must be established by showing that, under the conditions employed (i.e. protein and ligand concentrations), the final distribution of ligand within the enzyme population is independent of the path by which it is added. This was verified experimentally by bringing activated Rubisco (4.6 PM, as described under "Experimental Procedures") to a final 2carboxyarabinitol-P2/holoenzyme stoichiometry of 1.5, 3, or 6 in either one or several successive steps. After each aliquot, the solution was allowed to incubate for 30 min a t 30 "C prior to the addition of more 2-carboxyarabinitol-P,. IEF analysis and carboxylase activity measurements showed that this interval was sufficient for complete binding. Both the final activity and native IEF pattern were found to be independent of the number of aliquots by which ligand was added.
Molar ratios of 2, 4, and 6 2-carboxyarabinitol-P2/holoenzyme were chosen for the differential binding experiments so that distinct components of the IEF profile could be monitored a t each 2-carboxyarabinitol-P2 stoichiometry examined (see Fig. 2). Calculations of the final distribution of ligand among the protein molecules from Phase 1 and Phase 2 (Fig.  1) for an 8-site noncooperative binding process (Table 11) allow direct comparison of the actual results from the differential binding experiments to those predicted for a noncooperative process.
The results of an experiment using Protocol A (protein basis) are presented in Fig. 6. The IEF patterns obtained in Phase 1 a t ligand ratios of 2,4, and 6 2-carboxyarabinitol-P2/ holoenzyme (lanes b, d, and f, respectively) are in agreement with those in Fig. 2. The absence of 2-carboxyarabinitol-Pz exchange between Phase 1 and Phase 2 protein is evidenced by the appearance of native Rubisco in the pH 5.5-5.8 region of the gel when buffer instead of 2-carboxyarabinitol-Pz was added in Phase 2 (data not shown). By comparison of the protein patterns obtained after each phase at 2, 4, and 6 2carboxyarabinitol-P,/holoenzyme (lanes b/c, d/e, and f/g, respectively), it is evident that 2-carboxyarabinitol-PZ preferentially binds to sites on protein with low levels of ligand bound. The ligand distribution calculation predicts that for an experiment using Protocol A with a noncooperative enzyme at 2 2-carboxyarabinitol-Pz/holoenzyme, protein added in Phase 2 should have a final average stoichiometry of 1.1 2carboxyarabinitol-P,/holoenzyme (Table 11). At this stoichiometry, all of the major native enzyme bands (PI 5.5-5.8; lane a) should be readily visible. However, these bands are clearly not evident even though Phase 2 protein makes up 50% of the total protein loaded onto the gel (Table 11). Similar results were obtained for the experiments at molar ratios of 4 and 6 2-carboxyarabinito1-P2/ho10enzyme (Fig. 6). The minor banding differences between Phase 1 and Phase 2 proteins (lunes b/c, d/e, and f/g of Fig. 6) can be accounted for by variations in the addition of enzyme and ligand between the two phases. Fig. 7 illustrates the results of a differential 2-carboxyarabinitol-P2-binding experiment performed on a free site basis (Protocol B of Fig. 1). At 2 2-carboxyarabinitol-P,/holoenzyme, the gel patterns of equal amounts of protein after Phase 1 and Phase 2 (lanes u and b) are identical, while the calculations presented in Table I1 predict that protein added in Phase 1 and 2 should have final average 2-carboxyarabinitol-PZ stoichiometries of 2.8 and 1.0, respectively, for a noncooperative system. Since the IEF profiles of equal amounts of protein at 1 and 3 2-carboxyarabinitol-P2/holoenzyme differ markedly (see Fig. 2), these findings clearly demonstrate that at these levels of partial saturation there are changes in the intrinsic (site) rate for the 2-carboxyarabinitol-P, tight-binding process. At 4 2-carboxyarabinitol-P,/holoenzyme (lunes c and d), protein added in Phase 2 should have a final average 2-carboxyarabinitol-PP stoichiometry of 2.0 for a noncooperative system (Table 11). Direct comparison of the more acidic end of the IEF pattern after Phase 2 (lune d ) with protein at 2 2-carboxyarabinitol-Pz/holoenzyme (lune a ) shows the stoichiometry to be greater than predicted. At 6 2-carboxyarabinitol-P,/holoenzyme (lunes e and f ) , this trend appears to continue, although the errors associated with determination of protein (site) and ligand concentrations, as well as the small fraction of total protein added in Phase 2 (Table 11), preclude detailed analysis at this 2-carboxyarabinitol-P,/ holoenzyme level.
In order to examine the effects of competition for 2-carboxyarabinitol-Pz between unligated and partially saturated Rubisco on the distribution of the partially ligated enzyme species after Phase 2, differential binding experiments were performed using ['4C]2-carboxyarabinitol-P2 in the first phase (see Fig. 1). A single polyacrylamide gel was cut in half after electrofocusing and employed for both protein staining and fluorography. Fig. 8 shows the protein staining (lunes a-f) and fluorography (lanes g-1) patterns from an experiment performed using Protocol B (free site basis) at 2 and 4 2carboxyarabinitol-P,/holoenzyme. Lanes a-f'show the typical protein patterns obtained with this protocol (see Fig. 7). From the correspopding fluorograph, it can be seen that some migration of protein from the first phase does occur during the second phase. This is particularly evident in the more basic end of the pattern (e. g. lunes h, i and j , k ) . However, there is little net shift of the fluorographic patterns to more acidic PI values, indicating that only minor changes occur in binding stoichiometry. These findings are consistent with the presence of negative cooperativity, since the binding of unlabeled 2-carboxyarabinitol-P, during the second phase is apparently restricted to Rubisco molecules which have low ligand stoichiometries at the end of Phase 1.

2-Carboxyarabinitol-P2 binding to
higher plant Rubisco has been investigated previously by kinetic and spectral analyses (1, 2, 4, 19). These studies have shown that the activated enzyme and this phosphorylated ligand interact by a two-step mechanism, which presumably involves a change in protein conformation (2,19,20). The extent and effects of this proposed structural reorganization have not been elaborated, however. The experiments reported herein extend these earlier studies by demonstrating that the decrease in carboxylase specific activity is a linear function of the amount of 2carboxyarabinitol-P, bound per holoenzyme and that tight binding is correlated with specific, reproducible changes in the native IEF pattern, as well as in the chromatographic and electrophoretic behavior of the protein. Specifically, the effects of 2-carboxyarabinitol-Pz on the hydrodynamic shape, charge, and surface properties of the native protein demonstrate the occurrence of a significant, ligand-induced confor- exceed that predicted by a binomial distribution. In particular, the native IEF gels show little or no protein in the pH 4.9 region, where all protein migrates when it is 2-carboxyarabinitol-P2-saturated, prior to the addition of 5-6 2-carboxyarabinitol-Ps/holoenzyme. The binomial distribution model of ligand binding for an %site enzyme predicts that only 10% of the protein should have eight ligands bound when the mean 2-carboxyarabinitol-P2 stoichiometry is 6/holoenzyme. Anion-exchange chromatography resolves only a limited range of 2-carboxyarabinitol-P2 stoichiometries, and gel electrophoresis, while clearly able to separate enzymes with differences of 2-4 in 2-carboxyarabinitol-P2 stoichiometry, shows fairly tight bands. If the probability of a tight-binding event varied markedly in any subset of the native enzyme population, it would he reflected in the final binding distribution. Thus, each of the native forms of the enzyme appear to have similar rates of ligand tight-binding at each level of 2-carboxyarabinitol-P, saturation examined. The possibility of positive cooperativity in the 2-carboxyarabinitol-P2 tight-binding process is also precluded by these experiments, since this, too, would result in a broad (ie. nonbinomial) distribution of 2-carboxyarabinitol-P, stoichiometries.
Gross Conformational Effects of 2-Carboxyarabinitol-P, Tight-binding-Although the tightening of the native IEF titration pattern upon 2-carboxyarabinitol-P2 saturation is highly suggestive of a ligand-induced conformational change, the possible effects of ligand charge-addition and protein buffering on the native heterogeneity cannot be excluded as the cause of this phenomenon. However, these possibilities are not obviously compatible with two aspects of the IEF titration patterns. Regarding the addition of ligand charge, if the charge of this phosphorylated 6-carbon molecule was the sole determinant of the observed PI shifts, enzyme treated with subst,oichiometric quantities of 2-carboxyarabinitol-P2 would be expected to have a more uniform distribution of protein among the isoelectric bands than the native enzyme. This follows directly from the thermodynamics of binding since the random addition of charge to a population of molecules distributed among several discrete charge states must result in a new distribution in which the occupancy of any state approaches one over the number of molecules which can reach that state by charge addition. Thus, the intense protein bands observed at intermediate ligand concentrations, e.g. 3, 4, and 5 2-carboxyarabinitol-P2/holoenzyme, are irreconcilable with ligand charge addition effects alone. Another possible explanation for the native protein's microheterogeneity, protein buffering effects, appears equally unlikely. Given the diffuse, multicomponent 0.3 pH-unit range in which the native enzyme focuses, it is not apparent how a change in PI on going from the unbound to the fully ligated state (which yields a single, visually tight band covering less than 0.05 pH-unit) could mask the native surface charge heterogeneity by simple protein buffering, unless it is assumed that the native charge microheterogeneity is a product of amino acid residues at the surface of the protein which all have pK values in the 5.5-6.0 region; this is an unlikely assumption. Consequently, changes in the buffering capacity of multiple, reproducible protein forms does not seem adequate to explain band tightening.
The less sensitive, but complementary techniques of anionexchange chromatography and native gel electrophoresis yield more specific insight into the extent and nature of the conformational changes which occur upon 2-carboxyarabinitol-P, tight binding. A variety of published observations suggest that the active sites of Rubisco are deeply buried within the enzyme molecule (see Discussion in Ref. 20). Thus, the finding that enzyme with 2-carboxyarabinitol-Pz bound, and hence possessing a higher formal negative charge, elutes at a lower ionic strength than the untreated enzyme (Fig. 4) suggests that a structural change occurs in Rubisco upon tight binding of 2-carboxyarabinitol-P2. This alteration is not limited to the catalytic region of the enzyme, but affects the anion-exchange accessible surface groups of the protein. Native electrophoretic analysis of the untreated enzyme and the enzyme-CABP complexes illustrates the magnitude of the conformational change brought about by tight binding of 2-carboxyarabinitol-P2; the binding of stoichiometric amounts of ligand results in a 16% decrease in the apparent molecular weight of the carboxylase molecule.
Differential Effects of 2-Carboxyarabinitol-P, on the Isoelectric Forms of the Native Enzyme-A number of observations from the native IEF protein and fluorography data suggest differential effects of 2-carboxyarabinitol-P2 on the microheterogeneous charge forms of the native enzyme. First, upon treatment with substoichiometric amounts of 2-carboxyarabinitol-P2 the diffuse native pattern tightens into a series of discrete bands. Obviously, for this to occur, enzyme forms which, in the unbound state, focus at slightly different PI values must upon addition of 2-carboxyarabinitol-P2 focus a t the same PI. This is best exemplified a t 2-carboxyarabinitol-P2 stoichiometries of 1 or 2/holoenzyme where the distribution function limits differences in 2-carboxyarabinitol-P2 stoichiometry within the native enzyme population. Here, enzyme molecules, which in the unligated state focus at different PI values, must, upon binding the same amount of 2-carboxyarabinitol-P2, focus at the same PI. In addition, some bands of the native population do not change in PI, yet fluorography reveals that they have bound 2-carboxyarabinitol-P2. Second, the intensities of bands 01 and p at 1-5 2-carboxyarabinitol-P2/holoenzyme are inconsistent with their genesis from distinct fractions of the native population without invoking negative cooperativity in the tight-binding process. In the case of p, applying the binomial distribution to the protein band intensity as a function of the 2-carboxyarabinitol-P2 molar ratio leads to the conclusion that 50-70% of the total protein must at some point focus in the PI 5.30 region. Since no bands within the native pattern contain this much protein, either differential conformational changes or negative cooperativity with respect to 2-carboxyarabinitol-P2 tight binding must be invoked. Implications of the Binding Mechanism-In order to address these possibilities experimentally, we have examined the implications of 2-carboxyarabinitol-P2 binding to the partially ligated protein. Although there are several questions to be considered, perhaps the most basic of these concerns the elucidation of the mechanism of cooperativity in the tightbinding process. In the experimental analysis of this mechanism, we have compared the structural products of a single ligand-binding event in a homogeneous carboxylase solution with those generated in a solution containing the partially ligated product of this first reaction, an addition of unligated, activated Rubisco, and a second addition of ligand. Such a comparison permits us to examine the binding distribution of 2-carboxyarabinitol-P2 in a population of enzyme molecules in which a subset of that population has a given number of catalytic sites already occupied by the tight-binding ligand. Since the Rubisco-CABP complexes of both reactions are stable and amenable to analysis by native polyacrylamide isoelectric focusing, specific information concerning their ligation state and distribution is provided.
Three possible binding distribution alternatives were envisaged in considering the interaction of 2-carboxyarabinitol-P2 in a solution composed of partially saturated, i.e. 2, 4, or 6 2carboxyarabinitol-P2/holoenzyme, and unligated Rubisco. One of these corresponds to the ligand binding solely to the partially saturated holoenzyme. A second relates to the binding of 2-carboxyarabinitol-P2 only to unligated enzyme. The third relates to a nonpreferential binding situation in which the ligand binds to both partially ligated and unligated Rubisco. In the investigation of these three possible alternatives, native IEF, coupled with the knowledge that different binding states of the holoenzyme display diagnostic "fingerprint" profiles upon such analysis (see Fig. 2), provides a unique experimental system. It is also noteworthy that when unligated enzyme is placed in solution with partially ligated Rubisco, a composite IEF profile is observed. Consequently, if the first distribution model, i.e. ligand binding only to the partially ligated enzyme, was correct, the IEF gels should reveal a change in that portion of the profile contributed by the partially bound Rubisco; the pattern of the unligated holoenzyme would remain unchanged. Similarly, in order for the second alternative, i.e. binding of CABP solely to the unligated enzyme, to be valid, the banding pattern of the partially bound holoenzyme would remain constant while the banding contribution of the native enzyme would no longer be evident in the region of the gel in which it is normally observed. If the third model, i.e. nonpreferential binding to both enzyme species, was correct, a complete rearrangement of the banding profile would be expected with little or no resemblance to the original banding patterns of the two contributing enzyme forms.
The experimental results (Figs. 6-8) demonstrate clearly that the added ligand preferentially binds to the unligated enzyme in a solution of native and partially ligated Rubisco, or, in other words, to the least saturated of two partially complexed CABP-Rubisco species in a solution consisting of two subsaturated enzyme populations. This was evident even when there were fewer total unligated Rubisco molecules present than ligated enzymes (Fig. 7 ) . Multiple (>2) additions of 2-carboxyarabinitol-P2 to attain the appropriate ligand concentration also did not alter these conclusions. Consequently, these findings are in accord with the interpretation that this tight binding transition-state analog binds preferentially to the unligated or least ligated form of the enzyme.
The analyses presented in this paper have important implications for cooperativity in spinach Rubisco since there is ample precedent in the hemoglobin literature (21,22) to suggest that the quaternary structure, not the state of ligation per se, controls the ligand affinity and chemical kinetics of an enzyme. More specifically, this literature suggests that with each addition of ligand to a multimeric enzyme, changes in subunit-subunit interactions take place (23); for Rubisco and the purposes of our analysis, these effects are manifested as surface-charge perturbations which are visualized by native IEF. More important, however, these alterations may be responsible for modulating the kinetic properties of the enzyme (24). Consequently, the collective binding distribution results presented herein imply that activated spinach Rubisco is negatively cooperative.