Investigation of the action patterns of pectinmethylesterase isoforms through kinetic analyses and NMR spectroscopy. Implications In cell wall expansion.

Well characterized pectin samples were incubated with cell wall-bound and -solubilized pure isoforms of pectinmethylesterase from mung bean hypocotyls (Vigna radiata). Both enzyme activity and average product structure were determined at intervals along the deesterification pathway at pH 5.6 and 7.6. The latter analyses were performed by 13C NMR spectroscopy, and the degree of esterification was probed by both 13C NMR and potentiometric measurements. A dichotomy was observed in the behavior of the alpha and gamma isoforms when compared with that of the beta isoenzyme. Ideal blockwise deesterification mechanisms reproduced the experimental average structures (methylester distribution) throughout the course of the reaction. In the case of the alpha and gamma isoforms, a single chain mechanism associated with a free carboxyl group at the second nearest neighbor position could be postulated at pH 5.6, whereas some multiple attack character was required to reproduce the data at pH 7.6. Several mechanisms that differed from the preceding ones were compatible with the data for the beta isoform at the two pH values. Both the nature of the polysaccharides produced in these reactions and the role of pectinmethylesterase in the cell wall-stiffening process along the growth gradient are discussed.

Pectins that represent around 30% of the primary plant cell walls play a key role in plant physiology as well as in plant pathology. The general structure of pectic polymers consists of homogalacturonan linear chains (smooth regions) interspersed with highly branched galacturonic chains (hairy regions). Some of the galacturonic residues linked by ␣-1,4 glycosidic bonds are methyl-esterified at the carboxyl group. The degree of methylesterification (DE) 1 varies greatly depending on the plant organ and the degree of differentiation of the cells. Young, plastic cell walls are generally characterized by a high content of highly methylated pectin that decreases in parallel with the loss of extensibility of the walls, whereas the amount of acidic residues increases (1)(2)(3). Moreover, the balance between high and low methylated pectins varies inside the wall of a single cell (4) generating microdomains. Not only the number but also the distribution of free, unesterified galacturonate carboxyl groups within the galacturonan regions will control the gellforming capacity of the pectin and thereby the porosity and the extensibility of the apoplasm (5,6). It is commonly accepted that the polygalacturonic backbone is polymerized in the cis-Golgi cisternae, methylesterified in the medial-Golgi, substituted with side chains in the trans-Golgi, and exported to the cell walls as a highly methylated polygalacturonan (7). At a later stage, it is deesterified in muro by cell wall pectinmethylesterases (PMEs). Many proteins exhibiting PME activity have been purified, and their biochemical features such as molecular weight, optimal pH, pI, and substrate specificity have been established (8). In some cases, the corresponding genes have also been cloned and sequenced (9 -11), but so far three-dimensional models based on either crystallographic data or NMR analyses are lacking.
A limited number of investigations of the action pattern of PME have been reported (12)(13)(14)(15)(16). Partially deesterified polymers have been submitted to hydrolysis by endogalacturonases that are known to stop their attack in the vicinity of a methylated galacturonic unit, and the fragments released were subsequently characterized (13). In another approach, the activity coefficients ␣ Ca 2ϩ (␣ Ca 2ϩ is a complex function of the linear charge density of the macromolecule) of the calcium pectinates were determined after enzymatic deesterification (14). From these investigations, it was proposed that alkaline PME from higher plants (tomato, orange, alfalfa) catalyze deesterification of pectin linearly along the chain of the molecule, giving rise to blocks of free carboxyl groups, whereas acidic microbial PMEs cause a random cleavage of esterified carboxyl groups (14). In the case of the plant PMEs, studies have also pointed to the importance of a free carboxyl group in the vicinity of the active site (13,14,17). A major drawback to the afore-going methods is that they provide only indirect information on the distribution of the carboxyl units after PME action and do not lead to a precise description of the reaction mechanism.
In quantitative studies of the action pattern of tomato PME, Grasdalen et al. have established the methoxyl distribution at various stages (DE) of the reaction with 1 H NMR spectroscopy (15,16). The three mechanisms, which are generally evoked in theoretical transformations of polysaccharides (18), were considered (see Scheme I): (a) multiple chain mechanism, where the enzyme-substrate complex dissociates after each reaction, resulting in conversion (deesterification) of just one residue for each attack; (b) single chain mechanism, where an enzymic attack is followed by a contiguous conversion of all residues in an E-block starting with the E-residue in the attacked UE-diad (where U and E are unesterified and esterified residues, respectively); and (c) multiple attack mechanism, where the enzyme deesterifies a limited average number of residues for every active enzyme-substrate encounter. Their results were best interpreted with the multiple attack mechanism, assuming that the enzyme attacked alternating UE sequences and deesterified linearly preferentially toward the reducing end. The degree of polymerization (DP n ) of the pectin samples used in this work was fairly low (n ϳ 15 residues), and end group effects were necessarily important that might partially explain why the enzymatic deesterification deviated from an idealized block-producing mechanism. However, the major pitfall of these studies resides in the enzyme preparation as the tomato PME was not purified and contained several isoforms. Indeed, in most cases cell walls contain several PMEs differing in their biochemical properties and encoded by different genes (11). Moreover, the behavior of PME in solution sometimes differs from that exhibited by cell wall-bound enzymes (8), and it would be hazardous to extrapolate physiological explanations of the functions of PME in vivo using such an approach. Comparison of PME activity for both cell wall-bound and solubilized states is a necessary precaution if one wishes to understand the action mode in vivo.
In the present study, partially depolymerized pectin samples were incubated with either purified PME isoforms or cell wallbound enzymes to elucidate the PME action patterns. Under suitable conditions, substrates with an average DP n of ϳ100 can be obtained by thermally induced depolymerization, and it was demonstrated with various complementary techniques such as size exclusion chromatography, potentiometry, viscosimetry, and 13 C NMR that these samples retain the characteristics of the parent polymer (DE and methoxyl distribution) while adopting the random-coil conformation (as opposed to the aggregated forms of the native pectins), which is mandatory for solutionstate NMR structural studies (19). From the behavior of solubilized and cell wall-bound PMEs, we ultimately aimed to shed light on the role of the different PME isoforms in the cell wallstiffening process, which occurs along the mung bean hypocotyl.

MATERIALS AND METHODS
Pectin Samples-The pectin sample (apple pectin classic AU 201 (DE 74.4%)) was graciously supplied by Herbstreith & Fox KG (Neuenbü rg, Germany). The thermal depolymerization protocol and the characteristics of the resulting polymers have been described previously (18). As regards the samples used in the present work, 13 C NMR end group analysis indicated a DP n of 100, whereas the one estimated (20) from the intrinsic viscosity ([] ϭ 0.5) was roughly 50. The totally esterified pectin was a gift from Dr. Baron.
Enzyme Isolation and Assay-Cell walls were isolated from the upper 2.5 cm of hypocotyl tissues of 3-day-old seedlings of mung bean (Vigna radiata L. Wilzeck) according to a procedure previously described (1). PMEs bound to the cell wall fragments were solubilized with 1 M NaCl, and the different isoforms were recovered according to Bordenave and Goldberg (21,22). Assay for PME activity estimation (total volume 6 ml) contained 30 mg of thermally depolymerized polymers. PME activity was measured titrimetrically by following the release of protons in the presence of 150 mM NaCl. The protons were titrated with 10 mM NaOH under nitrogen, the pH being maintained at 5.6 or 7.6 with an automatic titrator (TTT 80, Radiometer). The reaction rate, as eq of H ϩ released/min/mg of protein, was estimated from the time necessary to reduce the DE by 1.67% i.e. the time necessary to release 2.5 eq of H ϩ from the pectin sample. The reaction was stopped by lowering the pH to 3 with 0.05 N HCl when less than around 0.05 eq of H ϩ were released/ min in the incubation medium. All samples were stored at Ϫ20°C before NMR analysis.
NMR Spectra-In the case of a random distribution of methoxyl groups, thermally depolymerized pectins with a wide range of DE (35-75%) can be analyzed at room temperature, as the major fraction of the polymers adopts the random-coil conformation and is detected by NMR spectroscopy (19). However, because of a change in the distribution of the methoxyl groups upon deesterification with PME and/or the salt concentration, the pectins incubated with PME were much more prone to aggregation than the starting material. To optimize the signalto-noise ratio in the 13 C spectra, the samples were desalted on an ion-exchange resin (DOWEX, H ϩ form) and then dialyzed at 4°C to obtain a less acidic pH (4 to 6 depending on the DE). Finally, EDTA (0.5 eq) was added to remove divalent cations. Totally decoupled 13 C spectra were recorded at 60°C on aqueous solutions of the pectins (ϳ2% w/v) at 100.6 MHz.
Deesterification Simulations-Starting samples containing all possible distributions of the methoxyl groups for a 10-residue fragment of given DE (i.e. 210, 120, 45, 10 fragments for DE values of 60, 70, 80, and 90%, respectively) were first generated as a linear two-dimensional array. A polydisperse population (between 100 to 4000 chains) of polymers of given DP n (n ϭ 25, 50, 100, or 200) and DE (73%) with a random sequence were then built from these 10-residue fragments using a random number generator. The DE and the triad populations (EEE, EEU, UEE, UEU, EUE, UUE, EUU, and UUU, where the underscore designates the residue that is either being considered in the simulation algorithm or detected in NMR spectra) and the homopolymer E-block and U-block populations (for block-lengths of 1-100 residues) of the resulting two-dimensional arrays were evaluated and compared with the corresponding data of the pectins used for assaying enzyme activity. The enzymatic reactions were simulated as follows (see Scheme I). Random generation of the point of attack (the point of attack XE2X or PA is defined by the integers, which correspond to the chain and residue positions in the two-dimensional array), evaluation of constraint criteria (i.e. PA must be a methoxylated residue; for certain mechanisms, 1-3 neighboring residues were required to be E or U units, and all positions between 1-10 residues away from PA were considered for these constrained units), and sequential (single chain mechanism) or single residue (multiple chain) deesterification toward a given end of the macromolecule until a deesterified unit or the end of the chain was encountered (or a given number of residues, k, in the case of the multiple attack mechanism). For suitable decrements in the DE (5%), the fractional triad populations (EEE, EEU, UEE, UEU, EUE, UUE, EUU, and UUU) and the homopolymer E-block and U-block populations of the resulting sample of polydisperse polymers were evaluated. The average E-and U-block lengths, N E and N U , which included all the E or U residues except the isolated ones (UEU or EUE), were as follows: N E ϭ ((DE-UEU)/EEU) and N U ϭ (((1 Ϫ DE) Ϫ EUE)/UUE). The following R-factor has been used to evaluate the quality of the fits with the experimental data: , where the summation extends over all the DE values for which data were collected. Replacing the sums by the corresponding squared experimental average deviations afforded the experimental R-factor. The standard deviation in the triad populations, estimated from simulations initiated with different seeds for the random number generator was Ͻ Ͻ1%.
Potentiometry-The degree of esterification was potentiometrically estimated as previously reported (1). Briefly, samples were dialyzed and weighed, and a portion was set aside for colorimetric estimation of the galacturonic acid content according to Blumenkranz and Asboe-Hansen (23). After transformation into the H-form with Amberlite IR SCHEME I. E, U, and X are esterified, unesterified, and either esterified or unesterified residues, respectively; the esterified residue (underlines) to the left of 2 is the point of attack.
120H, the weight and the uronic acid content were again determined to assure that all of the starting pectin was recovered from the ionexchange resin. The sample was then neutralized under nitrogen flux with sodium hydroxide. After saponification (4 h, 4°C) the new negative charges were estimated as above. The degree of esterification was calculated as the number of new charges over the total amount of galacturonic acid.

RESULTS
Relationships between PME Activity and DE-To visualize the effects of the degree of methylesterification of the substrate on the reaction rate of the deesterification process, plots of activity versus DE have been given in Fig. 1 for the long term incubations, which were performed either at pH 5.6 (E) or pH 7.6 (q) with bound (A and C) or solubilized (B, D, E, and F) PMEs. None of the enzymatic fractions was able to generate totally deesterified pectins, as the reaction stopped before the DE had dropped to values that varied between 20 and 50% depending on the enzyme used in the assay.
PMEs bound to their natural matrix (Fig. 1A) or solubilized with NaCl (Fig. 1B) reacted similarly to the decrease in DE. In both cases, whatever the DE, the reaction rate was higher at pH 7.6, but the differences between the reaction rates estimated at pH 7.6 and 5.6 were larger when the enzymes were still bound to the cell walls. Moreover, at pH 7.6, two successive phases were observed depending on the DE; during the first phase (from DE 70% to DE 40 -50%), the reaction rate lowered very slightly with DE, whereas below this level it decreases much more rapidly. At pH 5.6 this decrease was observed as soon as the DE dropped below 60%. This lowering of the reaction rate was not because of a paucity of methylester bonds because even at a DE of 50% the concentration of methyl groups in the assay was still 10-fold higher than the K m .
A dichotomy appears in the behavior of the three purified enzymes. Two of them, PME ␣ and PME ␥ , reacted in a similar way to decreasing DE. At pH 7.6, plots of the reaction rate versus DE presented a plateau for high DE values (74 -45%) followed by an abrupt regular drop to zero at lower DE values (24 and 32% for ␣ and ␥, respectively, Fig. 1, D and E). At pH 5.6, the decrease in activity started at DE values of about 65-70%, and it was sharper for PME ␣ than for PME ␥ . At high DE, contrary to cell wall fragments or isoform mixtures, the maximal activity was higher at pH 5.6 than at pH 7.6, a situation that persisted over a larger range of DE values in the case of PME ␥ . These data explain why PME ␣ and PME ␥ have been reported to display different pH sensitivities when incubated with citrus pectin whose DE is around 65%; with such substrates PME ␥ exhibits an acidic optimal pH, whereas PME ␣ activity is barely modified between pH 5.6 and 7.6 (21). The behavior of PME ␤ differs markedly from that of the other two isoforms (Fig. 1F); at pH 7.6, for high DE values a plateau in activity is not detected, and for all DE values, activities are higher than at pH 5.6. This tendency was observed both when the enzyme was bound (Fig. 1C) and when it was solubilized (Fig. 1F), the only difference being an attenuation in the effect of the pH after solubilization of the enzyme. Finally, when a totally esterified pectin sample was incubated with PME ␤ , enzymic activity was negligable.
Triad Populations and Block Lengths of Pectin Incubated with PME Estimated from 13  The pH of the incubation medium was maintained either at pH 5.6 (E) or at pH 7.6 (q). In all cases the initial reaction rate (i.e. the rate at DE 70 -75%) at pH 5.6 has been normalized to 100. A, crude cell wall fragments containing all PME isoforms; B, PMEs solubilized from the cell walls; C, cell wall fragments treated successively with 0.2, 0.4, and 0.5 M NaCl, which contain only isoform PME ␤ ; D, E, and F, purified isoforms PME ␣ , PME ␥ , and PME ␤ , respectively. pectins can be estimated from the corresponding 1 H (15, 16) or 13 C (24) NMR spectra (Fig. 2). The 100.6 MHz 13 C NMR triad population data for different incubation periods with the various enzyme preparations have been collected in Table I along with the corresponding DE values. These latter data were in good agreement with the DE independently established by potentiometry, indicating that the random-coil species detected with the latter technique was representative of the polymer as a whole. The discrepancies in the minimum DE values obtained either during the kinetic activity measurements (Fig. 1) or in the course of NMR analyses (Table I) stems from a regain in enzyme activity upon raising the pH from 3 to 4 -6 during the preparation of the NMR samples (see "Materials and Methods"). For example, upon lowering the pH of a solution containing PME ␣ to 3 and storing at 5°C for 2 weeks, the enzyme recovers 5 and 16% of its normal activity when the pH is raised to 5.6 and 7.6, respectively.
From inspection of the triad populations, it can be seen that the results obtained by incubation with either PME ␣ or PME ␥ at a given pH were analogous (see DE 23% at pH 5.6 for PME ␣ and PME ␥ ). As regards the esterified residues (XEX) at pH 5.6, the EEE and UEU populations were the major and minor ones, respectively, at DE values as low as 23%. In contrast, at pH 7.6, the EEE and EEU ϩ UEE populations were similar, and the UEU population became significant below DE values of 50%. Subsequently, all the data for PME ␣ and PME ␥ were simultaneously fitted to the same theoretical plots of triad populations versus DE to increase the number of experimental points. In the case of incubations with PME ␤ , for all DE values the EEE and EEU ϩ UEE populations were similar, whereas the UEU population was a minor one for DE values above 30%. As a result, data concerning incubations with PME ␤ were considered separately.
The NMR data for pectin incubated with the preparation containing a mixture of all the solubilized enzymes (␣, ␤, and ␥) displayed a similar pattern to that observed for the ␣ or ␥ isoforms. As regards the cell wall-bound enzymes, the triad populations of pectins that had been treated with the crude preparation were analogous to those obtained with the ␣ and ␥ isoforms. In contrast, the results observed with pectins which had been incubated with cell wall fragments that had been extracted with NaCl solutions (here only, the ␤ isoform remained bound to the cell wall fragments) tended to be similar to the data recorded for the pure ␤ isoform.

Comparison of Experimental and Simulated Populations as a Function DE for Various Reaction Mechanisms-
The theoretical polymers generated for simulation of various reaction mechanisms all displayed both a Bernoullian distribution of triad populations and average E-block lengths, N E , identical to those of the partially depolymerized pectic samples. From simulations that were conducted in both directions (from left-to-right and from right-to-left), it was ascertained that the randomnumber generator had not introduced a skewed distribution of the methoxyl groups. 1000 chains were necessary to minimize end group effects for simulations with the experimental DP n (i.e. 100,000 residues). When only one constraint was associated with the deesterification process (i.e. a free carboxyl group in a specified position with respect to the point of attack) the reaction continued down to the requested DE value of 4%. However, the reaction stopped at much higher DE values (10 -30%) when the number of constraints increased and/or DP n decreased. It should be emphasized that end groups effects strongly affect the UEU population for calculations involving shorter chains. The fits for the more pertinent simulations have been collected in Table II.
The multiple chain mechanism (MCM) did not reproduce the experimental triad populations irrespective of the number or type of constraints, and the corresponding R f values, which are given in Table II, were very high for all isoforms (ϳ5-10 times greater than the experimental precision). This was to be expected as the range of N E values for a Bernoullian distribution (ϳ2-3 for DE values between 20 to 50%) is much lower than the experimental ones (e.g. 4.0 -5.2 for incubations with the ␣ and ␥ isoforms).
The single chain and multiple attack mechanisms (SCM and MAM) were next systematically evaluated with 0 -3 constraints. In the case of the latter mechanism, multiple attack leading to the deesterification of 2-5 (degree of multiple attack, k in the scheme) residues/active encounter was considered. Simulations with more than one constrained or specified residue led to monotonously decreasing reaction profiles for the EEE and EEU ϩ UEE triad populations, which did not reproduce the experimental data and gave very poor agreement for UEU. However, in the case of simulations with one constraint, reasonable fits to the experimental data were observed for some of the SCM and MAM simulations. Generally speaking, the UEU versus DE curve varied significantly when the neighboring constrained U residue was located between 1 and 3 residues away from the point of attack, but when the constrained residue was placed at an even greater distance to the point of attack, subsequent variations in the UEU versus DE profiles became negligable.
In the case of the incubations with the ␣ and ␥ isoforms at pH 5.6, the best fit was obtained for all the esterified triad populations with the SCM reaction as long as the constrained U residue was separated from the point of attack by one unit (XUXE2), and plots of the corresponding theoretical triad populations (solid lines (E and छ) for DP values of 200 and 25, respectively) have been given in Fig. 3A (experimental data at pH 5.6 (Ⅺ)). An R f value of 0.0123 was calculated for the fit between the experimental and theoretical data, and it was analogous to the experimental precision (R f expt 0.0200). Ap-

FIG. 2. Esterified carboxyl carbon region of the 100 MHz 13 C NMR spectrum of a depolymerized pectin sample (DP 100 and DE 37.5%) that had been incubated with PMEinf>␤ at pH 7.6.
Assignments for the triads with an esterified central residue have been indicated, and the entire carboxyl region is given in the inset.
praisal of Table II suggests that the SCM mechanism involving attack of UE diads (XUE2) also gives a good fit to the experimental data (R f 1.42). However, when one inspects the R f values for the individual triad populations (Table II, values in italics), it can be seen that this mechanism is far from reproducing the experimental populations for UEU (0.0113 versus 0.0052 for R f and R f expt , respectively). As these latter populations are low, their contribution to the R f values is often masked by that of the much larger EEE and the EEU ϩ UEE triad populations. Comparison of the UEU population versus DE for long (DP 200 (Fig. 3A, solid lines (E)) and short (DP 25 (Fig. 3A, solid lines (छ)) chains shows two effects; for the shorter chains this population is smaller at low DE, and the reaction stops at higher DE than for the longer chains. At pH 7.6, the best results for incubations with the ␣ and ␥ isoforms were obtained for MAM involving the deesterification of 2-5 residues with the free carboxyl unit in the same position (XUXE2) (Fig. 3A, experimental and theoretical data, f and  dotted lines, respectively).
In the case of the ␤ isoform, many of the SCM and MAM simulations (without constraints, XU(X) n E2 and .. XE(X) n E2, where 10 Ն n Ն 2) gave the same reasonable fits to the experimental data. SCM and MAM mechanisms without constraints or with only an esterified residue in the vicinity of the point of attack could be rejected, as the ␤ isoform was unable to deesterify totally esterified pectin. The smallest R f value was obtained when the free carboxyl group was between 3 to 10 residues away from the point of attack (XUXXE2), but other mechanisms reproduced all the individual triad populations within the experimental precision. The SCM (XUXE2) and MAM (XUXE2, k Ն 2) mechanisms, which gave the best fit to the experimental NMR data for PME ␣ and PME ␥ , were notable exceptions. Here, inspite of the low R f , certain individual triad populations gave very poor fits. A plot of the SCM triad profiles  for the most favorable mechanism for the ␤ isoform (solid lines; XUXXE2) has been given in Fig. 3B along with the corresponding experimental data at pH 5.6 and 7.6 (Ⅺ and f, respectively). Finally, the plots of the average length of the E-and U-blocks (N E and N U ) versus the DE have been given in Fig. 4A (q , ␣ and ␥; E, ␤) for the mechanisms which best fit the experimental triad populations (see Figs. 3, A and B, and Table II). Similar profiles were obtained for all the satisfactory mechanisms of a given isoform irrespective of the pH. As can be seen in Fig. 4A, in the case of the ␣ and ␥ isoforms, the average E-block length  Fig. 3: q, ␣ and ␥ isoforms; E, ␤ isoform. B, plots of the evolution of the UEU and UEEU (left) and EUE and EUUE (right) populations with DE for the simulations of ideal blockwise deesterification mechanisms in Fig. 3: q and f, ␣ and ␥ isoforms; ␤, E and Ⅺ. remains constant (4.8) as the DE decreases from 74% to about 15% and then increases when the DE drops to below 10%. In contrast, for the ␤ isoform, the average E-block length decreases slowly with the DE% and then increases sharply when the DE drops to below 10%. The experimental data confirm this behavior for incubations with the ␣ and ␥ isoforms on the one hand and the ␤ isoform on the other. In parallel, differences in the theoretical populations of short E-blocks (i.e. UEU and UEEU) for the preferred mechanisms for the ␣ and ␥ isoforms (on the left, q and f) and the ␤ isoform (on the left, E and Ⅺ) are also observed (Fig.  4B). The former mechanisms (␣/␥) lead to much stronger populations for UEU than UEEU, whereas the latter one (␤) produces more similar populations for these short blocks. Inspection of the average U-block lengths and the populations for the short Ublocks as a function of DE in Figs. 4, A and B showed that fairly analogous distributions were produced by all the isoforms. DISCUSSION The data reported above shed new light on the reactions in the apoplasm, which modify the structural features of the pectic polymers. Previous investigations on pectin from mung bean hypocotyl (1) have shown that the DE decreases along the growth axis from 58.4 to 44.5%, whereas the PME activity solubilized from the cell walls doubles. This suggested that these enzymes were involved in the accumulation of galacturonan in mature stiff cell walls. It is well known that the pH of the apoplasm is acidic (pH 5.5-6.0), and under these conditions, the V max of the different PME isoforms is markedly different, because studies with citrus pectin (DE 65%) have led to activities of 0.4, 0.08, and 0.8 eq min Ϫ1 g Ϫ1 for PME ␣ , PME ␤ , and PME ␥ respectively (25). These data show that PME ␣ and PME ␥ activities are undoubtedly those that contribute the most to the decrease in DE in vivo. Fig. 1 indicates that, at acidic pH, the activity of these two isoforms drops abruptly for DE values below 60%. The balance between methyl and free carboxyl groups can then be considered as a key factor in the control of PME-mediated deesterification. This feedback control will prevent the generation of totally deesterified polymers but allows the formation of long U blocks along the polymer chain. The presence of these blocks in mature cell walls favors the formation of junction zones via calcium ions and will then contribute to the cell wall stiffening that develops along the hypocotyl.
An explanation for the dichotomy in the action patterns of the PME isoforms from V. radiata has been obtained by simulation of the triad populations established from 13 C NMR. In the case of the pure ␣ and ␥ isoforms, a single chain mechanism requiring a free carboxyl group at the second nearest neighbor position (XUXE2X) gave the best fit at pH 5.6, whereas at pH 7.6 the reaction displayed multiple attack character involving the deesterification of only between 2 and 5 residues. Thus, the pH strongly effects the efficiency of the deesterification process. A change in reaction mechanism with a variation in pH has been reported for porcine pancreatic ␣-amylase (26). In contrast, successful simulation of the action pattern of the ␤ included a specific interaction (E or U) at least three residues away from the active site (e.g. XUXXE2X). It should be emphasized that the methoxyl distribution in the polysaccharide produced by PME-mediated deesterification depends on the conditions (constrained residues) governing the reaction and not on the type of process (MCM, SCM, or MAM). It would appear that the distance between crucial complementary polar groups and the catalytic site varies slightly in going from the ␣/␥ isoforms to the ␤ isoform, but three-dimensional structures for all the isoforms will be required to fully interpret these results. At present, only the complete amino acid sequence of PME ␥ has been determined (25).
Interestingly, the lowest DE values attained experimentally for the pure isoforms (for example, 13.5-23% at pH 5.6) are significantly higher than those predicted by the simulations (Ͻ5.5 to 2.5% for chains with DP ϳ50 -100), which implies that the reaction stops because of factors that are not accounted for in the simulation algorithm (i.e. reasons other than a lack of substrate with a suitable methoxyl distribution). One possible explanation is that the free carboxyl group on the pectin near the active site may become protonated, thus weakening the polar interaction. However, van der Waals interactions between some of the methoxyl groups and hydrophobic amino acids may make an important contribution to the binding energy and partially explain why, when the methoxyl group population or DE decreases below 45-55%, the enzymatic activity drops very abruptly. In the case of protein-protein interactions, the stabilization from a hydrophobic contact has been estimated to be about half of that produced by complementary polar groups (27), whereas in the case of protein-carbohydrate interactions (28), the favorable effects of electrostatic and van der Waals contacts have been considered to be similar. From the afore-going discussion, it can be postulated that a combination of polar and nonpolar interactions probably contributes to pectin-PME binding. The methoxyl group distribution in pectins incubated with the ␣ and ␥ or ␤ PME isoforms from V. radiata are significantly different, particularly with respect to the short E-blocks. The block-producing character of the deesterification process leads to highly aggregated polymers, and in the future, such enzymes may find industrial applications as they should be able to confer tailor-made rheological properties to commercial pectins.