The Crystal Structure of Pectate Lyase Pel9A from Erwinia chrysanthemi*

The “family 9 polysaccharide lyase” pectate lyase L (Pel9A) from Erwinia chrysanthemi comprises a 10-coil parallel β-helix domain with distinct structural features including an asparagine ladder and aromatic stack at novel positions within the superhelical structure. Pel9A has a single high affinity calcium-binding site strikingly similar to the “primary” calcium-binding site described previously for the family Pel1A pectate lyases, and there is strong evidence for a common second calcium ion that binds between enzyme and substrate in the “Michaelis” complex. Although the primary calcium ion binds substrate in subsite -1, it is the second calcium ion, whose binding site is formed by the coming together of enzyme and substrate, that facilitates abstraction of the C5 proton from the sacharride in subsite +1. The role of the second calcium is to withdraw electrons from the C6 carboxylate of the substrate, thereby acidifying the C5 proton facilitating its abstraction and resulting in an E1cb-like anti-β-elimination mechanism. The active site geometries and mechanism of Pel1A and Pel9A are closely similar, but the catalytic base is a lysine in the Pel9A enzymes as opposed to an arginine in the Pel1A enzymes.

Polysaccharide lyases (EC 4.2.2.-) are carbon-oxygen lyases that exploit ␤-elimination chemistry to cleave C5 uronic acid containing pyranoside substrates such as polygalacturonates, alginates, hyaluronan, and chondroitin. In contrast to the 90 sequence-derived families of glycoside hydrolases, polysaccharide lyases have been grouped into just 12 families (1). 1 Threedimensional structures have been reported for enzymes from polysaccharide lyase families (PL) 2 families 1, 3, 5, 6, 8, and 10, and of these, PL-1 (3)(4)(5), PL-3 (6), and PL-6 (7) consist of "right-handed parallel ␤-helix" domains (called "parallel ␤-helix" domain throughout the rest of this article). The parallel ␤-helix domain is superhelical and comprises three short ␤-strands, typically three to four residue, per coil of the superhelix; these coils stack coil-on-coil to form three parallel ␤-sheets that extend the whole length of the domain, and these ␤-sheets are referred to as PB1, PB2, and PB3. Schematic diagrams of the three parallel ␤-helix domains, including that of Pel9A reported in this report, are shown in Fig. 1. The turns between the short ␤-strands are known as T1, T2, and T3, and these turns connect PB1 to PB2, PB2 to PB3, and PB3 to PB1 of the next coil of the superhelix, respectively (see Fig. 2a).
Glycoside hydrolases (EC 3.2.1.-) hydrolyze the carbon-oxygen bond between two pyranoside residues or between a pyranoside and another moiety. The three-dimensional structures of the solved glycoside hydrolases show more structural diversity, although the ␤/␣ barrel occurs frequently, often with catalytic carboxylates at the ends of ␤-strands 4 and 7 (8,9). So far, only one glycoside hydrolase family has been shown to have parallel ␤-helix architecture, family GH-28 (10), a family that includes the enzyme polygalacturonase (11), although three other families have been predicted as parallel ␤-helices (12). Carbohydrate esterases can be grouped into 13 sequence-based families, and the parallel ␤-helix fold occurs again in carbohydrate esterase family CE-8, a family of pectin methylesterases (13). The parallel ␤-helix fold of each of these families has its distinct structural features (14).
Pectate lyases cleave polymeric ␣-1,4-linked galacturonic acids (GalA) of the pectate component of the plant cell wall by a ␤-elimination mechanism that generates a 4,5-unsaturated oligogalacturonate product. All pectate lyases, with the exception of PelW (15), require calcium for in vitro activity and presumably utilize the abundant calcium in the plant cell wall for activity in vivo. The role of calcium has been controversial, but it is now clear that it binds to both enzymes and substrates and can mediate enzyme substrate interactions by binding between enzyme and substrate (16). Some pectate lyases have been shown to have a preference for partially methylated polymers of polygalacturonic acid, and the non-calcium-dependent pectin lyases that cleave fully methylated substrates also belong to family PL-1 (17).
Erwinia chrysanthemi is a pathogenic enterobacterium that causes soft-rot diseases in a variety of crops. The bacterium enters the plant through wounds and secretes polygalacturonate-active enzymes that break down the complex polysaccharide pectin. Pectin is a key component of the plant cell wall, and these enzymes enable the further invasion of the bacterium. E. chrysanthemi is known to produce six PL-1 (PelA, -B, -C, -D, -E, and -Z), a PL-2 (PelW), a PL-3 (PelI), and two PL-9 (PelL and PelX) pectate lyases. PelL is an endo-acting, calcium-dependent pectate lyase with relatively low activity against polygalacturonate (ϳ100-fold less than the major PL-1 pectate lyase PelB) and with a pH optimum around 8.0 (18,19).
Here we report the 1.6-Å resolution crystal structure of the endo-acting pectate lyase L (EC 4.2.2.2) from E. chrysanthemi 3937 (called Pel9A throughout this article as it is a paradigm for the polysaccharide lyase family 9 enzymes). The fold of Pel9A was correctly predicted as part of the Critical Assessment of Techniques for Protein Structure Prediction competition (20), but the modelers had severe difficulties producing a molecular model for Pel9A because of the repetitive nature of the fold, the short ␤-strands, and the many possibilities for excursions between the ␤-strands. When the "primary" calcium-binding sites of the previously described Pel1A and the Pel9A structure described here are superimposed, the active site geometry is seen to be closely similar, but lysine in Pel9A replaces the catalytic arginine of the Pel1A enzymes. The K273A mutant of Pel9A is inactive but correctly folded, supporting the assignment of lysine 273 as the catalytic base.

MATERIALS AND METHODS
Production of Native Pectate Lyase L-The pectate lyase L gene from E. chrysanthemi (PelL) was overproduced in Escherichia coli BL21 cells and purified from the periplasm by ion-exchange and gel filtration chromatography (21). The purified protein gave a single band on Coomassie Blue-stained SDS-PAGE gels. The protein was concentrated to 20 -30 mg/ml for crystallization, and the largest crystals grew from 16 to 17% 1,6-hexanediol, 12-13% polyethylene glycol 3350, 100 mM sodium chloride with 400 mM citrate, pH 5.6. The crystals grew as plates up to 1.0 mm long, 0.4 mm across, and 0.2 mm thick. The crystals are orthorhombic, space group P2 1 2 1 2 1 , with a ϭ 55.33 Å, b ϭ 57.83 Å, c ϭ 114.93 Å, and with a single molecule in the asymmetric unit.
Preparation of Heavy Atom Derivatives and Data Collection-Data were collected from crystals cooled to 100 K in the nitrogen gas stream of an Oxford Cryosystems cryocooler using synchrotron radiation and image plate or CCD detector (Table I). No additional cryoprotectant was needed for successful flash cooling for native or heavy atom soaked crystals. Heavy atom derivatives were prepared by soaking Pel9A crystals in the heavy atoms dissolved in 20% polyethylene glycol 3350, 20% hexane 2,5-diol, 400 mM sodium acetate with 100 mM sodium chloride. Heavy atom data were collected from crystals soaked in platinum orange (chloro(2,2Ј:6Ј,2Љ-terpyridine) platinum(II) chloride), chloroplatinate (long and short soaks), lead acetate, and sodium hexachloroiridite. The concentrations and soak times used are given in Table I as is a summary of the crystallographic statistics. Data were reduced using DENZO and SCALEPACK (22), and subsequent calculations used CNS (23) and the CCP4 program suite (24).
Structure Solution-Inspection of the platinum orange difference Patterson revealed a single platinum-binding site. The correct hand of the solution was that which gave the greatest peak heights on crossphased difference Fourier maps. The heavy atom positions from the difference Patterson and the difference Fourier maps were refined using MLPHARE (24), and protein phases were calculated to a 2.3-Å resolution. DM (26) was then used to improve the protein phases, and the difference Fourier maps were recalculated, reviewed, and re-refined. DM was then run at 1.6 Å to give protein phases that were used in WARP (27). WARP successfully built 85% of the protein model. Refinement was completed using CNS and REFMAC (28) combined with model building using O (29). PROCHECK was used to judge the stereochemical quality of the final structure (30). Sequence alignments used T-Coffee (31). Crystallographic statistics are given in Table I. The presence of a calcium ion at the active site was confirmed by inspection of an anomalous difference map.
Construction and Analysis of Mutant Proteins-The single mutations K273A and K273R were introduced in PelL (Pel9A) by site-directed mutagenesis with the QuikChange kit (Stratagene). The primers K273A (5Ј-ggcaacgggttcGCCctaggaggaaacc-3Ј) and K273A-rc, the reverse complementary to K273A, were used to create K273A, and the primers K273R (5Ј-ggcaacgggttcCGCctaggaggaaacc-3Ј) and K273R-rc, the reverse complementary to K273R, were used to generate K273R 99.1% Ramachandran plot (generously allowed) 0.6% Overall G-factor 0.03 a R-factor and R free are given by Α ͉F obs Ϫ F calc ͉/Α F obs , where F obs and F calc represent, respectively, the observed and calculated structure factors. The R-factor was calculated using the working reflection set (95% of the data), and the R free was calculated using the free reflection set (5%) not included at any stage of refinement. b r.m.s.d., root mean square deviation.
where I j is the intensity measurement for reflection j, ͗I͘ is the mean intensity for multiply recorded reflections and ͚ h is the sum over the unique reflections. c R iso , ͚͉F ph Ϫ F p ͉/͚͉F p ͉, the mean relative isomorphous difference between the native protein (F p ) and the derivative (F ph ) structure factors. d Phasing power, ͗F h ͘/͗E͘, where ͗F h ͘ is the root mean square heavy atom structure factor and E is the residual lack of closure error. e These parameters are shown as means for the acentric and centric reflections.
where F h is the calculated heavy atom structure factor.
(codon 273 is in uppercase letters). The two amino acid substitutions led to the creation of a unique AvrII restriction site. After the nucleotide sequences were verified, the stability and enzymatic activity of mutant proteins were tested. The plasmid pT7L1.6 carrying the pelL gene under control of the ⌽10 promoter and the derivative plasmids bearing the mutant pelL genes were introduced into E. coli BL21 (DE3) cells, and the pectate lyase was overproduced and extracted from the cells as described previously (21). The quantity of mutant proteins in whole cell lysates and periplasmic extracts was compared with that of wild type PelL by immunoblotting with PelL antibodies. The pectate lyase activity in periplasmic extracts was measured spectrophotometrically at 230 nm. The quantities of K273A,K273R mutant proteins produced were comparable with wild type PelL, indicating that these mutations do not affect the protein production and stability, but the mutants were inactive, supporting the assignment of lysine 273 as an important catalytic residue.

RESULTS AND DISCUSSION
The Architecture of Pel9A-The three-dimensional structure of Pel9A from E. chrysanthemi was solved at a resolution of 1.6 Å (see "Materials and Methods" and Tables I and II). The mature polypeptide chain (residues 26 -425) of Pel9A folds to form a right-handed parallel ␤-helix (residues 95-356) with distinct structural features and with 10 turns of superhelical architecture (Fig. 1). The parallel ␤-helix domains of Pel1A and Pel3A are shorter than Pel9A with only eight complete superhelical turns but, like polygalacturonase, Pel9A has 10 turns ( Fig. 1). When compared with Pel1A, Pel9A has considerably shorter and better-ordered loops. All three enzymes shown in Fig. 1 have an N-terminal ␣-helix that caps the hydrophobic core of the parallel ␤-helix.
The ␤-helix of Pel9A is both preceded and followed by less regular regions, and there is a single disulphide bridge between cysteines 28 and 114 that links the third residue of the Nterminal extension to the beginning of the second ␤-strand of PB2. The N-terminal extension forms the first strand of PB2 (residues 40 -44) and an extravagant but well ordered loop (residues 47-62) before forming the ␣-helix (residues 62-70) that contributes Phe-63 and Met-67 to the capping of the hydrophobic core at the N-terminal end of the ␤-helix. There then follow 10 turns of superhelical architecture, starting with the second strand of PB2 (residues 74 -77). At the end of the ␤-helix domain, the polypeptide breaks away abruptly from the final ␤-strand of PB3 to form the C-terminal extension (residues 357-425) that comprises an extended irregular hairpin that packs against PB3 and further irregular structure, part of which is stabilized by a sodium ion in the crystal structure. Trp-357 of the C-terminal extension terminates the aromatic stack on PB1 (see below) and plugs the hydrophobic core at the C-terminal end of the ␤-helix.
In comparison with the other right-handed ␤-helix structures solved, Pel9A has a very regular arrangement of side chains and main chain in the T3 turns (between PB3 of coil n and PB1 of coil n ϩ 1). This regularity is conferred by the remarkably well ordered like-on-like stacks in this region, which includes an eight-residue glycine stack, a six-residue asparagine ladder at T3 (asparagines 174, 198, 231, 255, 293, and 317), and a four-residue alanine stack (alanines 195, 228, 252, and 290) (Fig. 2, a and b). In Pel9A, there are also additional short asparagine ladders within the T2 turn (Asn-312, Asn-338, and Asn-355) and within the T1 turn (Asn-164, Asn-188, and Asn-221). The regularity in the T3 region extends to the main chain conformation that forms hydrogen bonds up and down the length of the parallel ␤-helix. It is the T2 turn and PB3 that possess the asparagine ladder and the aromatic stack characteristic of the Pel1A enzymes (Fig. 2, c and d). The regular aromatic stack of Pel1A (Fig. 2d) is replaced by the alanine stack described above in Pel9A, and for many turns, the less regular aromatic stack on PB1 fills the parallel ␤-helix (phenylalanines 211, 236, 272, and 298 and tyrosine 321; Fig.  2b). Although an aromatic residue is shown on PB2 of Pel9A (Fig. 2a), this is not involved in extensive stacks. There is also an external aromatic stack on the surface of PB2 in Pel9A involving His-172, Tyr-196, Trp-229, Phe-253, Phe-291, and Tyr-315. The distinctive features of Pel9A, the aromatic stack on PB1, and the asparagine ladder within the T3 turn are interesting and show how different sequences can pack the core of the ␤-helix domain. The success of the parallel ␤-helix in polygalacturonate-active enzymes is presumably because the binding surface on the external surface of PB1 and adjacent loops is well suited to binding polygalacturonates.
Calcium Binding to Pel9A and Pel1A Enzymes-A single calcium ion is seen bound to Pel9A (Fig. 3a) despite no addition of calcium during purification and crystallization, which implies that the site has at least M affinity and that the enzyme has scavenged the calcium present in trace quantities from the solutions. The calcium-binding site is formed by four aspartates on the external surface of PB1. This calcium-binding site is strikingly similar both in its position on the surface of PB1 and in the nature of the calcium ligands to the first calcium site identified both in the Pel1A structures (4,32) and in the pH 9.5 Pel3A structure (6). Superimposition of the calcium ion of Pel9A with the equivalent calcium in the PelC substrate complex reveals that Asp-131, Glu-166, and Asp-170 of PelC are equivalent to Asp-209, Asp-233, and Asp-237, respectively, of Pel9A. These superimposed carboxylates correspond to the three calcium-binding aspartates seen in Bacillus subtilis pectate lyase (BsPel). If the common calcium-binding residues are superimposed, then the ␤-helices are brought into close structural alignment, revealing that, when compared with the Pel1A enzymes, the two additional coils of Pel9A are N-terminal to the calcium-binding site such that in Pel9A, the calcium-bind- ing aspartates are on coils 5 and 6, whereas in PelC, they are on coils 3 and 4 (Fig. 3). PelC and Pel9A superimpose with a root mean square deviation of 1.8 Å for 171 equivalent ␣-carbon atoms.
Data collected from native Pel9A crystals soaked in substrates and calcium did not reveal bound ligand or additional calcium-binding sites (data not shown), but the Pb sites used to calculate the protein phases do indicate the presence of a second calcium-binding site close to the active center. In the PelC "Michaelis" complex formed using the inactive mutant R218K and Ca 2ϩ /GalA 4 (32), substrate occupies the Ϫ1 to ϩ3 subsites, and the mutated Arg-218 is the base and abstracts the C5 proton from the galacturonic residue at the ϩ1 subsite (Fig. 3). The first calcium binds the carboxylate oxygen of the galacturonic residue at the Ϫ1 subsite. A second calcium is bound by Glu-166 (OE1 and OE2), by the ring oxygen (O5) and one carboxylate oxygen of the galacturonic residue at subsite ϩ1, and by a second carboxylate oxygen from the galacturonic residue at subsite Ϫ1. A third calcium binds the second carboxylate oxygen of the galacturonic residue at the ϩ1 subsite and the ring oxygen of the residue at the ϩ2 subsite. The fourth and fifth calcium are close to the putative base and bind the ring and carboxylate oxygens of the galacturonic residue at the ϩ3 subsite.
Clearly, the first calcium site is conserved between the Pel1A and Pel9A enzymes, and the aspartate contributing to the second site is conserved (Fig. 3), suggesting that in the Pel9A substrate complex, a second calcium would bridge between Asp-233 and substrate carboxylates at the ϩ1 and Ϫ1 subsites. This suggestion is supported by the binding of Pb to Pel9A; one Pb-binding site is close to the position at which calcium binds to Pel9A, and a second Pb binds in the position anticipated for the FIG. 3. The calcium-binding sites of Pel9A and Pel1A. a, anomalous difference Fourier map (chicken wire mesh) revealing the presence of a metal bound to Pel9A by aspartates 209, 233, and 237 (Asp-234 is also a calcium-ligand, not shown). The putative catalytic base, lysine 273, is also shown. b, calcium-binding site of Pel1A (BsPel) showing calcium bound by aspartates 184, 223, and 227. The catalytic base, arginine 279, is also shown. c, substrate binding to Pel1A (PelC coordinates kindly provided by Prof. Fran Jurnak). Galacturonates occupying subsites Ϫ1 (left) and ϩ1 (right) are shown together with four calcium-binding sites (ligands for sites three and four are not shown). d, substrate modeled in to the active center of Pel9A. The primary calcium-binding site is strikingly similar to that of Pel1A (BsPel). The second calcium site is anticipated to bind the complex at a similar position (as suggested by the Pb2 site), and the third and fourth sites will be different if they exist in Pel9A. Phenylalanine 239 in Pel9A forms a platform for the galacturonate in subsite Ϫ1. This figure was prepared using BOBSCRIPT (2). second calcium in the Pel9A-substrate complex (Fig. 3d). The aspartates contributing to the third and fourth calcium-binding sites in PelC are not present in Pel9A. However, Asp-299 in Pel9A may form a calcium-binding site together with substrate carboxylates, and this calcium would be in a similar position to the fourth calcium in PelC.
There are no arginines in the substrate-binding cleft of Pel9A, and so substrate carboxylate interactions with arginines involving both carboxylate oxygens and hydrogens of NH1 and NH 2 nitrogens of the arginine, as seen for the galacturonic residue at subsite ϩ2 and Arg-245 in PelC, cannot occur in Pel9A. However, when compared with PelC, Pel9A has two additional carboxylates, Asp-299 and Glu-180, in the substratebinding cleft, and these could form calcium-mediated interactions with substrate carboxylate oxygens. Asp-299 (coil 8) is close to the putative base, and Glu-180 (coil 4) is on the leaving group side of the cleft.
Pel9A has low activity on polygalacturonate (19) but is reported to be effective in the maceration of plant tissue (18). Polygalacturonate is a model substrate, and Pel9A may have more activity against its true plant cell wall substrate. Alternatively, Pel9A may have low activity as it is constitutively expressed, suggesting its role as a reconnaissance molecule. Pel9A produces GalA 4 as opposed to the GalA 2 produced by the other lyases (19). The low levels of GalA 4 produced may alert the bacteria to the presence of plant pectin without alerting the plant to the impending attack resulting from the induction of the other lyases.
Catalytic Mechanism-Pel1A and Pel9A enzymes cleave ␣-1,4 bonds of polygalacturonate by anti-␤-elimination characterized by base-catalyzed abstraction of the proton from C5 and elimination of the substituent at O4 to give an unsaturated product. ␤-elimination chemistry is reviewed in Anderson (33). In the family Pel1A enzymes, it has been argued that an arginine is the Brønsted base that abstracts the proton from C5 (16). Arginine is a surprising base for an enzyme active at pH 8.0, but calcium ions close to the arginine in the Michaelis complex could depress its pK a (34); as discussed below, the putative base is lysine in Pel9A as opposed to arginine. There is no obvious enzyme or substrate Brønsted acid to protonate the glycosidic oxygen in these enzymes, and protonation is presumed to occur via solvent water. There are, however, substrate hydroxyls that could stabilize the departing anion. Hydrogens from the O3 hydroxyl of the galacturonate in subsite ϩ1 and from the O2 hydroxyl of the galacturonate in subsite Ϫ1 could stabilize the oxyanion of the leaving group (Figs. 3  and 4). The argument for arginine as base is strongly supported by the recent Pel10A structure that has convergent active site geometry with the Pel1A enzymes, including an active site arginine, but a non-related architecture (35). Perhaps the most remarkable difference between the active sites of Pel9A and these other polysaccharide lyases is that in Pel9A, there is no arginine equivalent to the putative base in Pel9A (Arg-218 in PelC). If arginine is substituted for lysine in the coordinates of the inactive PelC complex, then the end of Lys-273 in Pel9A is close to the end of the Arg-218 in PelC. Because lysine is shorter than arginine for the ends of the side chains to arrive in the same location relative to bound substrate, the lysine is situated on the coil adjacent to the calcium-binding site, whereas the arginine is on the next but one coil (Fig. 3). When Lys-273 was mutated to alanine, no activity could be detected, a result consistent with lysine acting as the base. When Lys-273 was mutated to arginine, the activity was less than 5% that of the native enzyme, suggesting that arginine is capable of catalyzing the reaction but is not in an optimal position for catalysis. Lys-296 was also a candidate for the base in Pel9A, but this lysine is not conserved across the family Pel9A enzymes, unlike lysine 273 and the calcium-binding aspartates. Presumably, the PelC mutant Arg-218 Lys is inactive, allowing the Michaelis complex to be formed because lysine is a poorer base and not optimally positioned with respect to the C5 proton.
The formation of the Michaelis complex in the Pel10A structure (35) was a result of the mutation Asp-389 3 Asn. This mutation probably destroyed a calcium-binding site important for catalysis (equivalent to calcium-binding site three in PelC). The withdrawal of electrons from the carboxylate oxygens of the substrate by the interaction with calcium ions or basic residues increases the acidity of the C5 proton (4,36,37) and therefore facilitates bond cleavage. This interaction is missing in the Pel10A complex.
The two-dimensional mechanistic landscape of the ␤-elimination reaction catalyzed by PelL can be best considered in terms of the More O'Ferrall diagram (Fig. 4a), in which the courses of the three pathways, E1, E2, and E1cb, are determined by the order of bond cleavage with the cleaved C␣-H bond along the x axis and the C␤-X bond along the y axis (33,38). In the E1cb reaction, abstraction of the ␣-proton results in a carbanion intermediate followed by subsequent elimination of the C␤ substituent. The reverse order of C␤-X cleavage before C␣-H generates a carbocation intermediate, and this E1 reaction is considered unlikely in enzyme-catalyzed reactions FIG. 4. a, More O'Ferrall diagram for the ␤-elimination of galactoconfigured uronic acids (33,34,38) where the bond order for the H-C␣ cleavage is plotted along the x axis, and C␤-O bond cleavage is plotted along the y axis. b, putative E1cb/asynchronous E2 reaction mechanism for Pel9A enzymes in which the lysine abstracts the proton, resulting in leaving group elimination. The galacturonate shown is in subsite ϩ1, with the galacturonates labeled R 1 and R 2 in subsites Ϫ1 and ϩ2, respectively. Factors important in catalysis by Pel9A are acidification of the C5 proton by the interaction of the substrate carboxylate with a calcium ion and with the NH 2 group of Asn-268 and the resonant nature of the intermediate. The calcium shown is identified on the basis of the second Pb site in Pel9A (Fig. 3d). Asn-268 also stabilizes ϩ1 substrate carboxylate in Pel9A (PelL) but is not absolutely conserved in all family PL-9 enzymes. A third calcium may also be involved in stabilizing the substrate carboxylate. Leaving group departure may be facilitated by the hydroxyls of galacturonates in ϩ1 (O 3 , shown) and Ϫ1 (O 2 , not shown, see Fig. 3d). (33). The E2 pathway describes concerted but not necessarily synchronous proton abstraction and leaving group elimination such as would occur if the lifetime of the carbanion is less than the vibrational frequency of the C␤-X bond (39).
The presence of a putative catalytic base (lysine 273) in Pel9A, but no clear catalytic acid, suggests that the favored reaction pathway will be E1cb or toward E1cb rather than E1 on the More O'Ferrall potential energy diagram. The calcium ion interaction with the substrate carboxylate at C5 will increase the acidity of the C5 proton, and this, too, favors cleavage of the C␣-H bond, again favoring an E1cb mechanism (Fig. 4b). Substrate hydroxyls may facilitate leaving group departure.
There are three possible reasons why Pel9A is less active on polygalacturonates and oligogalacturonates than the Pel1A enzymes. Firstly, the lysine of Pel9A is a less potent base than arginine. Secondly, although there is strong evidence that the second calcium is bound in a similar way to that seen in PelC, the third calcium that is important in increasing the acidity of the C5 proton may be absent or at least not in the same position as in PelC. Thirdly, polygalacturonate may be a poor substrate for Pel9A.
Comparison of Pel9A and Pel1A Lyases with Polygalacturonase-Polygalacturonase has 10 turns of ␤-helix, as does Pel9A, and when polygalacturonase is superimposed coil-forcoil on Pel9A, then two of the putative active site residues of polygalacturonase Asp-202 and Asp-224 are superimposed on two of the calcium-binding residues Asp-209 and Asp-234 of Pel9A. When compared with Pel1A enzymes, not only are the aspartates in the same position in polygalacturonase, but also the arginine, the putative base in the Pel1A enzymes (32). Not only is the chemistry different in polygalacturonase, but also, a different bond is broken. In the hydrolase, it is the C1-O bond that is hydrolyzed, and in the lyase, it is the substituent at C4 that is eliminated, breaking the C4 -O bond. It is fascinating to see a common topology and the related active site geometries in these enzymes giving rise to different catalytic residues (Pel9A when compared with Pel1A) and different active site chemistries (lyases as compared with hydrolase).