INTRODUCTION

Pectate lyases (Pels)¹ are major virulence factors of plant pathogenic Erwinia sp., the causal agents of soft-rotting diseases on a broad range of plants (1). These bacteria produce an array of Pel isozymes that depolymerize cell wall polygalacturonides in the presence of Ca²⁺ ions and accordingly destroy the integrity of plant tissues, a process classically called ``maceration'' (2, 3). In addition, pectate lyases have been shown to activate defense systems in plants, presumably by releasing oligogalacturonides from the plant cell wall that function as defense elicitors (4). Pectate lyase proteins are secreted extracellularly by Erwinia sp. via the general secretory pathway (5), involving elements of the Sec system as well as specialized extracellular secretion proteins encoded by out genes (6). Although secretion to the periplasm is determined by stereotypic leader peptides, unknown structural features of the Pels predicate their entrance into the Erwinia Out secretion system. Because Escherichia coli cells lack the Out system, Pel proteins are not appreciably secreted extracellularly and instead accumulate in the periplasmic space.

Many of the genes encoding Pels have been cloned and characterized (2, 7). The overexpression of various pel genes in E. coli has also afforded large amounts of Pel proteins that can be easily purified from the periplasmic space. Recently, the three-dimensional structures of PelC (8, 9) and PelE (10) from Erwinia chrysanthemi EC16 were determined. These proteins have an unusual structure, the ``parallel  helix,'' which is generated by coiling a  strand into a large, right-handed helix with an unusual stacking of asparagines on consecutive turns of the parallel  helix core (11). Although PelC and PelE share only 22% amino acid identity, their core structures are very similar, with differences only in the size and conformation of the four loop regions that protrude from and cover the parallel  helix core (10, 12). The parallel  helix structure has also been found in a pectate lyase from Bacillus subtilis (13), the Salmonella typhimurium phage P22 tailspike protein (14), E. coli UDP-N-acetylglucosamine acyltransferase (15), and Bordetella pertussis P.69 pertactin (16). A topologically related structure, termed a -roll, has been observed in an alkaline protease from Pseudomonas aeruginosa (17). Like Pels, several of these proteins are involved in pathogenesis and appear to function at cell surfaces.

Extracellularly secreted pectate lyases share regions of sequence similarity with fungal pectin lyases (18, 19, 20) and certain plant pollen- and style-specific proteins (21, 22, 23, 24). Structure-based multiple alignment of these Pel-related proteins revealed the presence of potentially catalytic, invariant amino acid residues, not only around the Ca²⁺ binding site but also in the midregion of the core structure of the parallel  helix, designated as the vWiDH region (25, 26) (see Fig. 1). These observations suggested that all of the Pel-related proteins might have pectinolytic activity. This prediction was recently supported when a pollen-specific Pel-homologue from Japanese cedar was shown to exhibit in vitro pectinolytic activity (27).

We have initiated a project to understand amino acids important in pectinolytic activity, maceration of plant tissue, elicitor activity, and extracellular secretion, as well as for folding and stability of the parallel  helix. In this paper we report the construction of several oligonucleotide site-directed PelC mutant proteins and their characterization.

MATERIALS AND METHODS

Strains of E. coli employed and the plasmids utilized and constructed are summarized in Table I. E. coli HMS174(DE3) (28) was used with the expression vector pRSET5A, containing a T7 promoter upstream of a multiple cloning site (29). pRSET405Pst, which contains the entire pelC gene downstream of a T7 promoter, was constructed by inserting a ~1.4-kilobase pair XbaI-HindIII fragment from pPEL405 (30) into the same sites of pRSET5A and deleting the polylinker PstI site (Table I). For expression, the mutant and wild-type T7 plasmids were generally transformed into E. coli HMS174(DE3) (Table I). E. coli HB101 was also used for the expression of several pelC mutants cloned in the expression vector pINK1 (31). The same strain was also used for expression of the pelE gene (pPEL748, Table I).

Table I. Bacterial strains, plasmids, and oligonucleotide primers used Designation Relevant characteristicsa Source of reference Escherichia coli  DH5 F-lacZ M15 endA hsdR17 supE44 thi-1 gyrA relA1 Bethesda Research Laboratories  HB101 F-hsdS20(hsdR hsdM) recA13 ara-14 proA2 lac1 galK2 rpsL20(Strr) xyl-5 mtl-1 supE44 (32)  HMS174(DE3) F-recA r-K12m+K12m+K12Rifr (T7 lysogen)b (28) Plasmids  pUC119 Cloning vector, Apr (42)  pRSET5A Expression vector carrying the T7 promoter, Apr (29)  pINK1 Expression vector, Apr (31)  pPEL405 Clone carrying the pelC gene in pUC119 (30)  pRSET405Pst 1.4-kb XbaI-HindIII fragment from pPEL405 inserted into pRSET5A. A PstI site in the fragment was removed by partial PstI digestion, S1 digestion and religation This study  pPEL410 pelC expression construct (30)  pPEL748 pelE expression construct (30) Oligonucleotide primers mismatched primersc  L-10E 5-GAGCGCCAGCAGTCCAG-3 (antisense) This study  Y7D 5-GTAGCGGCGCCACCGGTA-3 (antisense) This study  D28R 5-GATATTGACGATCTGCATCG-3 (antisense) This study  C72S 5-CACTGGCCGATATTGGCG-3 (antisense) This study  M118N 5-CGATACGGTTCTGTACCACC-3 (antisense) This study  I120F 5-CAGGTAGCCACGCATGTTC-3 (antisense) This study  D129N 5-GTCGCCTTTAGCGCCG-3 (antisense) This study  D131E 5-GCGGATCATGCCATCTT-3 (antisense) This study  D131N 5-CTAAAGATGGCATGATCC-3 (sense) This study  W142H 5-CGTTATGGTCAACGACATTCGGCG-3 (antisense) This study  D144N 5-CAATTCGTTATGAACCCAGAC-3 (antisense) This study  H145Q 5-CAATTCGTTGTCAACCCAG-3 (antisense) This study  C155S 5-GTGCCGTCCTCATGGTTG-3 (antisense) This study  E166D 5-ACGGCGGAAAAGGTGGTG-3 (antisense) This study  E166Q 5-ACGGCGGAAAAGGTGGTG-3 (antisense) This study  D170N 5-CCTTGATAACGGCGG-3 (antisense) This study  K172H 5-GTTGATATAGGCGCGTCGAAC-3 (sense) This study  K172L 5-GTTGATATAGGCGCGTCGAAC-3 (sense) This study  K172R 5-GACGCGCCGATATCAACG-3 (antisense) This study  K190A 5-CAGACCCACTTTCACACCGTG-3 (antisense) This study  T206A 5-GTTATGGTGATAAATATTGC-3 (antisense) This study  N210S 5-CGTTGTAGTAATGGTG-3 (antisense) This study  R218A 5-CAACGGCAGGGCGTTAACG-3 (antisense) This study  R218E 5-CAACGGCAGGGCGTTAACG-3 (antisense) This study  R218K 5-CAACGGCAGGGCGTTAACGTC-3 (antisense) This study  R218L 5-CGGCAGGGCGTTAACG-3 (antisense) This study  R223A 5-CTGCCGTTGCAAGGTGGT-3 (sense) This study  H228Q 5-GTTGTAAGCAACCAGACC-3 (antisense) This study  G280N 5-GATGTTGTTTTTCAGCAC-3 (antisense) This study  C329S 5-TGTCCTTCACTTGTGC-3 (antisense) This study  K342E 5-GTGGCCAGATTACCCAC-3 (antisense) This study  C352S 5-GAGATTATTTGGCTGTGC-3 (antisense) This study Matched primers  F(M13/pUC) 5-GTAAAACGACGGCCAGT-3 (sense) This study  F200 5-CTGGCTATGAAATCACTC-3 (sense) This study  F400 5-CACGTCTGGATGCCAACG-3 (sense) This study  F600 5-CAACTTCGGCATCTGGAT-3 (sense) This study  F800 5-GTCGAACACCGTCACCGT-3 (sense) This study  F1000 5-CACCAACATCACCGGTTC-3 (sense) This study  F1200 5-GACACCAAGCCTTATGTG-3 (sense) This study  R(M13/pUC) 5-AACAGCTATGACCATG-3 (antisense) This study a  Str, streptomycin; Rif, rifampicin; Ap, ampicillin; r, resistance. b  Lysogen carrying the lacI gene, the lacUV5 promoter, the beginning of the lacZ gene, and the gene for T7 RNA polymerase. c  Bases underlined correspond to substituted amino acid codons, and bold letters represent mismatched bases with the wild-type pelC gene.

DNA manipulations were generally done according to the methods of Maniatis et al. (32). Site-directed oligonucleotide mutagenesis of the pelC gene was performed by PCR using essentially an overlapping extension method (33, 34). The synthesized primers for generating pelC mutants were reverse or forward and contained 18-24 nucleotides with 1-3 mismatched base pairs, depending on the desired amino acid change (Table I). Six primers were also designed to give overlapping fragments for the second round of PCR, and also for sequencing ~200-base pair intervals of the final clones. All mutagenesis primers were obtained from Cruachem Inc., Sterling, VA and were used in combination with M13 forward or reverse sequencing primers. pPEL405 was used as the template DNA in the first round of PCR, and Vent^TM DNA polymerase (New England Biolabs) was employed because of its high fidelity. In first round PCR, 30 cycles of amplification were carried out: 30 s at 94 °C for denaturation, 30 s at 55 °C for annealing, and 60 s at 72 °C for extension in each cycle. The resulting fragments were diluted and mixed to make full-length fragments and reamplified, but at 50 °C for annealing. These second round fragments were then treated with proteinase K followed by phenol-chloroform extraction (35), digested with XbaI and HindIII, and then cloned into pRSET5A. E. coli DH5 transformants were screened for the proper mutation by extraction of plasmid DNA and sequencing. Following selection of desired pelC mutants, the genes were sequenced in entirety to confirm fidelity.

E. coli strain HMS174(DE3) cells harboring pRSET5A constructs were grown to stationary phase at 28 °C with shaking in 15 ml of Luria broth (LB) supplemented with 50 µg ml¹ ampicillin. These cells were then added to 250 ml of the same medium and, when the culture had attained an A₆₀₀ of 0.6-0.8 (after approximately 4 h at 28 °C), isopropyl-1-thio--D-galactopyranoside (IPTG) was added to 0.6 mM. The cells were incubated for an additional 4 h at 28 °C with shaking and then recovered by centrifugation at 8,000 × g.

In the case of strain HB101 carrying pelC wild-type (pPEL410, Table I) or mutant constructs in pINK1, cells were grown for ~48 h at 22 °C in LB medium supplemented with 1 mM IPTG at culture initiation. For production of PelE, HB101 cells carrying pPEL748 were grown for ~20 h at 28 °C on LB medium supplemented with ampicillin but no IPTG. Cells were harvested by centrifugation at 8,000 × g. For both the HB101 and HMS174(DE3) expression systems, periplasmic fractions containing PelE, PelC, or mutant PelC proteins were prepared according to the method of Witholt et al. (36). The cells from a 250-ml culture were suspended in 25 ml of 0.2 M Tris-HCl, pH 8.0, and centrifuged at 8,000 × g for 10 min. The supernatant was discarded and the cells suspended in 4 ml of 0.2 M Tris-HCl, pH 8.0. Five ml of 1 M sucrose in the Tris-HCl buffer and 100 µl of 50 mM EDTA were added, mixed well, and allowed to stand for 5 min. Then 100 µl of freshly prepared lysozyme solution (6 mg/ml) was added, followed by 10 ml of distilled water and thorough mixing. The mixture was allowed to stand for 20 min at 22 °C. The resulting spheroplasts were pelleted by centrifugation at 15,000 × g for 15 min, and the supernatant periplasmic fraction was retained.

PelE, PelC, and PelC mutant proteins were purified using cation exchange chromatography, essentially as described previously (31). Periplasmic fractions were dialyzed against 5 mM Tris-HCl, pH 8.0, for 36-48 h at 4 °C (tubing was from Spectrum Medical Industries, Los Angeles, CA, M_r 6000-8000 cut-off). The dialyzed preparations were centrifuged at 15,000 × g for 5 min.

The clear supernatants were adjusted to pH 8.0 if necessary and applied to a 1.4 × 15-cm column of CM Bio-Gel (Bio-Rad) in 5 mM Tris-HCl, pH 8.0, at room temperature. After application of samples and washing with the starting Tris-HCl buffer, Pel proteins were eluted batchwise with the same buffer containing 0.25 M sodium chloride. Fractions of 2.8 ml were collected and analyzed for absorbance at 280 nm and pectinolytic activity. Peak pectate lyase fractions were pooled, dialyzed against 5 mM Tris-HCl, pH 8.0, for 24 h, concentrated using a Centricon 10 device (Amicon Inc., Beverly, MA), and stored at 20 °C. In some experiments, the proteins were dialyzed against distilled water. Little loss of activity was observed through all treatments.

Production of wild-type and mutant proteins by bacterial cells and the purity of proteins were determined by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (37) (12% gels, 0.75 mm thick) followed by Coomassie Brilliant Blue R250 staining and Western blot analysis (34). Following electrophoretic transfer to nitrocellulose membranes, pre-PelC, mature PelC, and mutant proteins were detected using a rabbit polyclonal PelC antiserum at 1:5000 dilution.

The activity of PelE and various PelC proteins was determined by monitoring the absorbance increase at 232 nm of a 1-ml reaction mixture containing sodium polypectate at 22 °C. To start the reaction, 5 µl of appropriately diluted protein were mixed with 870 µl of 50 mM Bis-Tris-propane (BTP), pH 9.5, containing the optimal CaCl₂ concentration (0.5 mM for wild-type PelC) and 125 µl of 1% sodium polypectate. For preparation of stock solutions of 1% sodium polypectate, polygalacturonic acid (85-90% purity, Sigma) was dissolved in deionized water, boiled for 5 min, and centrifuged at 10,000 × g for 10 min to remove sedimenting materials. 50 mM BTP, pH 9.5, containing the optimal CaCl₂ concentration was prepared just prior to assays. The assay for wild-type PelE was identical to that for PelC, with the exception that the pH was 9.0. One unit of pectate lyase activity was defined as described previously (7) (1 µmol of unsaturated product formed/min, which equaled 1.73 absorbance units min¹), and the specific activity was expressed as units mg¹ protein. For each protein, the specific activity reported was optimized at optimal pH and CaCl₂ concentration values. Protein concentration was determined by the method of Bradford (38).

Plant tissue maceration assays were done using mesocarp tissue of cucumber fruit, essentially according to the method of Mussell and Morre (39). Cylinders were cut from the mesocarp tissue with a cork borer (0.5 mm diameter), sliced into 5-mm lengths, and immediately immersed in distilled water. After standing for ~1 h, five cucumber cylinders were gently blotted onto filter paper to remove excess water, weighed to the nearest milligram, and placed into 5 ml of 50 mM BTP, pH 7.5, 0.5 mM CaCl₂ containing 0-10 µg ml¹ of PelC or mutant proteins. Concentrations of the mutant proteins were adjusted depending on their specific activity. After incubation for 1 h at 37 °C with shaking at 80 rpm, the cylinders were vortexed at maximum speed for 20 s and then poured onto a concave wire mesh filter (~0.8 mm mesh size) to remove loosened cells. The resultant cylinders were then blotted gently with filter paper and weighed to the nearest mg. Maceration was expressed as the weight loss of each sample during incubation and was calculated per 500 mg of starting tissue. Three replications were used for the wild-type and mutant proteins, and the data reported were normalized to the activity of wild-type PelC protein.

Activity of PelC and mutant proteins to elicit glyceollin production in soybean cotyledons was determined by slight modification of previous methods (40, 41). Eight cotyledons were freshly harvested from 8-10-day-old cv. Harosoy 63 plants when the primary leaves had just begun to expand. Each cotyledon was wounded on the undersurface, and 200 µl of protein solution in water containing 300 µg ml¹ streptomycin and 10 µg ml¹ rifampicin were placed onto the cut surface. After 24 h of incubation at room temperature under fluorescent lights, 50 µl of fluid were recovered from the cut surfaces of each of five representative cotyledons and the pooled samples were added to 20 ml of distilled water. The absorbance of this solution at 285 nm was determined and is proportional to the concentration of glyceollin (40). At least three replications were carried out for each enzyme per experiment.

RESULTS

Screening of the PelC PCR mutant constructs by DNA sequencing at the mutation sites showed that the desired mutations were obtained from approximately 50% of the constructs analyzed, as theoretically expected. Undesired mutations occurred randomly with a probability of ~1/4000 bases, and such genes were generally discarded. Following confirmation of complete sequences, genes encoding mutant PelC proteins were cloned into pRSET5A or pINK1 (Table I) and transformed into E. coli HMS174(DE3) or HB101, respectively, for expression. Fig. 1 shows the location of mutated amino acids on the PelC structure that were used in this study.

pelC and the various mutated genes in pRSET5A generally directed the production of large amounts of protein in E. coli cells, much of which was secreted into the periplasmic space (Fig. 2). Most of the PelC mutant proteins were purified satisfactorily using CM Bio-Gel chromatography with elution by 0.25 M NaCl in 5 mM Tris-HCl, pH 8.0. However, two mutants, R218A and R218L, did not bind well to the column at pH 8.0, presumably as a consequence of their lower pI values. Dialyzed periplasmic fractions containing these mutant proteins were therefore subjected to repeated CM Bio-Gel purification with 10 mM Tris-HCl at pH 7.0 and elution with 0.25 M NaCl in the same buffer. Approximately 3 mg of purified protein were obtained from a 250-ml bacterial culture (~0.5 g wet weight cells). As expected, differences in molecular weight among the various mutant proteins were not detected by SDS-gel electrophoresis. In the T7, but not the pINK1 expression system, approximately half of the translated products were not cleaved and remained in the membrane fraction of spheroplasts in the form of unprocessed preproteins, detected as higher weight bands on SDS gels (data not shown). They were assumed to be an artifact of the high level T7 expression system, because they were not normally observed using the pINK1 system.

Unlike wild-type PelC and most mutants, several PelC mutant proteins (L-10E, Y7D, D28R, C72S, M118N, D144N, H145Q, N210S, H228Q, and C329S) were produced at high level in the T7 system, but appeared to remain associated with the membrane fraction as preproteins or mature proteins, based on their sizes on SDS-gel electrophoresis with detection by Coomassie Blue staining or Western blotting (data not shown). No pectinolytic activity in the standard spectrophotometric assay could be detected for these mutants and the preproteins and mature proteins were both readily digested by trypsin (data not shown). With D28R, C72S, and D144N, small amounts of protein were present in the periplasmic fraction as detected by Western blotting, but were insufficient for isolation and purification. The remaining mutants gave only traces of proteins in the periplasmic fraction as deduced by Western blotting. The double cysteine mutant, C329S/C352S, gave greatly reduced protein production, even using the T7 expression system.

When D28R, C72S, and D144N were recloned into pINK1 and expressed in E. coli HB101, the proteins were present in the periplasmic fraction, albeit at reduced levels relative to wild-type PelC, since only ~3 mg/liter culture were recovered by cation exchange chromatography. All three of these mutant proteins were contaminated with 10-30% of a ~57-kDa protein that did not react with PelC antiserum on Western blots (data not shown). In these cases, it was necessary to run successive CM Bio-gel columns until the 57-kDa protein was removed. Recloning of the mutant genes encoding L-10E, H145Q, H228Q, and C329S into pINK1 resulted in greatly reduced production by strain HB101 relative to wild-type PelC.

The specific pectate lyase activities of wild-type PelC and all isolable mutant proteins are shown in Table II, along with PelE as a control. All mutations located in the vWiDH and surrounding regions of PelC that yielded isolable proteins (D28R, W142H, D144N, T206A, and K342E) retained high pectinolytic specific activity. Two additional oligonucleotide mutants (I120F and G280N) and one spontaneous mutant (T275I) also showed specific activities 80% or more of the wild-type level (Table II). The double mutant, C72S/C155S, gave relatively low activity (14% of the wild type, Table II), but C72S gave even lower activity (~6% of the wild type). Several mutants (R218A, R218E, R218K, R218L, R223A, K190A, K172H, K172L, K172R, D129N, D131E, D131N, D170N, E166D, and E166Q) were constructed as a consequence of their proximity to the active site region deduced by x-ray diffraction studies (Fig. 1). With the exception of D129N, E166D, and K172R, these proteins showed greatly reduced specific activities. They will be discussed elsewhere.²

Table II. Maximum specific activity, plant tissue maceration, and elicitor activity of wild-type PelC, PelC mutant proteins, and wild-type PelE PelC mutant proteins Maximum specific activitya Plant tissue macerationb Elicitor activityc Wild-type PelC 2900 420 0.978  D28R 2200 330 0.511  C72S 180 25 0.351 C72S/C155S 400 160 0.416  I120F 2300 350 1.186  D129N 1400 210 0.497  D131E <2 <3 0.201  D131N <3 <4 0.108  W142H 2800 370 0.854  D144N 1800 280 0.826  E166D 2700 410 0.779  E166Q <1 <3 0.136  D170N <8 <7 0.194  K172H 80 80 0.972  K172L 120 60 0.988  K172R 1300 290 0.896  K190A <1 <3 0.089  T206A 2400 340 0.744  R218A <1 <2 0.291  R218E <2 <7 0.322  R218K <15 <8 0.112  R218L <5 <5 0.129  R223A <18 <20 0.391  T275I 2300 350 0.797  G280N 2600 390 0.733  K342E 2100 300 0.518 Wild-type PelE 1700 4300 0.876 a  Maximum specific activities of PelC, PelC mutant enzymes, and PelE are expressed as units mg1 of purified protein and determined spectrophotometrically at 22 °C at the optimal pH and CaCl2 concentration for each enzyme. The maximum specific activity of each protein was calculated from three replications with triplicates. b  Plant tissue maceration activity is expressed as mg weight loss per 500 mg of cucumber mesocarp tissue by 1 µg ml1 of PelC, PelC mutant proteins, or PelE per 1 h at 37 °C. Plant tissue maceration activity of each protein was calculated from three replications with triplicates. Weight loss of buffer control was subtracted. c  Elicitor activity is expressed as the A285 nm of wound droplets taken from excised soybean cotyledons treated with 1 µg ml1 of PelC, PelC mutant proteins or PelE in distilled water. Cotyledons were incubated at 22 °C under fluorescent lights for 24 h. Absorbance at 285 nm of the control (without enzymes) was 0.111 and was not subtracted from experimental values. Elicitor activity of each protein was calculated from three replications with triplicates.

PelC and isolated mutant proteins were assayed for maceration activity (Table II), and this was compared with their in vitro pectinolytic specific activity (Fig. 3). These results showed that tissue maceration activity was closely correlated with pectinolytic specific activity, but three different K172 mutants and C72S/C155S showed higher maceration activity than expected, based on their pectinolytic specific activity (Fig. 3). Confirming previous results (30), PelE yielded ~10-fold greater maceration activity than PelC, based on in vitro activity (Table II).

Pectic enzymes behave as elicitors of plant defense reactions (3), presumably by generating oligouronides from the plant cell wall that are the actual elicitors. As shown in Fig. 4, wild-type PelC elicited accumulation of the soybean phytoalexin, glyceollin, at a minimal concentration of about 0.1 µg/ml. Two mutant proteins (D129N and D170N) showed lesser ability to elicit glyceollin (Fig. 4), consistent with their decreased specific pectinolytic activities (Table II). E166D, which showed similar pectinolytic activity to wild-type PelC (Table II), also elicited with similar efficiency (Fig. 4). However, when a larger collection of mutant proteins were compared, a relatively poor correlation of specific pectinolytic activity and elicitor activity was observed (Fig. 5; Table II). For instance, K172L, K172H, and K172R gave high elicitor activity relative to their in vitro pectinolytic activity. These same mutants also were more active than expected in the tissue maceration assay (Fig. 3). Several other mutants (R223A, D144N, and I120F) yielded higher elicitor activity than predicted from their specific pectinolytic activity. Only D28R gave elicitor activity significantly below that expected from its in vitro activity (Fig. 5). Surprisingly, PelE showed the same level of elicitor activity in the soybean assay as did PelC (Table II), despite the fact that PelE macerated cucumber mesocarp tissue ~10 times more efficiently. These results suggest that elicitor activity is not solely dependent on pectinolytic activity, as measured in vitro.

DISCUSSION

There is a considerable body of information on the organization, sequence and regulation of pel genes that, combined with the three-dimensional structures of three Pel proteins, has provided significant understanding of these important virulence factors in Erwinia sp. An analysis of the structure-based multiple sequence alignment for the entire Pel superfamily suggested that the active site resides in the Ca²⁺-binding region (Fig. 1) (25, 26). Supporting this idea, pectinolytic activity was virtually abolished in mutants D131E, D131N, E166Q, D170N, K190A, R218A, R218E, R218K, R218L, and R223A. All of these amino acids cluster around the Ca²⁺-binding site, with the first three coordinating directly to the cation. The next question was whether the loss of pectinolytic activity was due to loss of Ca²⁺ binding, impairment of a catalytic residue, a decrease in substrate affinity or an abnormality in protein folding. Because all of the mutants above readily form crystals isomorphous to wild-type PelC, an abnormality in protein folding is unlikely. The question of Ca²⁺ binding and catalytic impairment will be addressed in detail elsewhere,² and the effects of various mutations on substrate binding are currently being studied. Interestingly, the data in Table II eliminate the possibility that the vWiDH region is involved in pectinolytic cleavage observed under in vitro conditions, because W142H and D144N as well as mutations T206A and K3442E in spatially adjacent amino acids exhibited enzymatic specific activities near the wild-type level. The question of function for the vWiDH region is illuminated by the fact that several mutations in this region led to proteins that remained associated with the bacterial membrane fraction and were poorly exported to the periplasm. While additional experiments are needed, these results raise the possibility that the vWiDH region is involved in secretion or protein folding in the periplasm.

In addition to pectinolytic function, the PelC mutants were also screened for their ability to macerate plant tissue under in vivo conditions and to elicit the formation of phytoalexins in soybean cotyledons. Generally, there was a good correlation between in vitro pectinolytic activity, which utilized PGA as the substrate, and maceration efficiency of the various PelC mutants (Fig. 3). However, four proteins (K172H, K172L, K172R, and C72S/C155S) showed significantly higher maceration activity relative to their pectinolytic activity. The relatively high maceration activity of these mutants is interesting, because it suggests that important features of PelC other than its ability to cleave PGA per se are involved in plant tissue maceration. In this regard, it should be noted that PelC itself is about 10-fold less efficient in maceration than PelE from Erwinia chrysanthemi (Table II; Ref. 30). It is therefore possible that the PelC K172 mutants have been modified to approach the maceration efficiency of PelE and that additional mutations could be made which would further increase the maceration activity of PelC.

The tendency for some PelC mutants to give greater maceration activity than predicted from their in vitro activity was more pronounced when the same proteins were assayed for their ability to elicit formation of the phytoalexin, glyceollin, in soybean cotyledons. Thus, K172L, K172H, and K172R gave elicitor activity that was not significantly different from wild-type PelC, despite their much lower in vitro pectinolytic activities (Table II, Fig. 5). One mutant, I120F, gave elicitor activity that was greater than wild-type PelC (Fig. 5), but its maceration activity was proportional to its in vitro pectinolytic activity (Fig. 3). Although some of the mutants exhibited elicitor activity that was less than wild-type PelC relative to their in vitro pectinolytic activity, only D28R was significantly lower. It is interesting that, despite its 10-fold greater maceration efficiency, PelE exhibited the same elicitor activity as PelC. These results all indicate that factors other than ability to cleave PGA are involved in the interactions of PelC and PelE with plant cells.

Another outcome of the research was the development of a more effective overexpression system for producing large quantities of Pel proteins. For all previous studies of E. chrysanthemi PelC and PelE, including determination of the three-dimensional structures, the pINK1 expression system was used (31). A more efficient T7 expression vector, pRSET5A, was employed for the production of PelC and most mutant proteins in this study. Expression directed by pRSET5A in E. coli HMS174(DE3) generally provided large amounts of Pel proteins in the periplasmic fraction for isolation and characterization. Approximately 6 mg of PelC or the mutant proteins could be purified from 1 g of fresh bacteria following 4 h of induction at 28 °C. However, a portion of the translated products appeared to remain associated with the membrane fraction as the preprotein when the T7 system was used. This may be a consequence of PelC synthesis levels exceeding the capacity of the general secretory pathway.

Certain amino acid changes in PelC resulted in reduced protein production and/or production of protein that was associated with the bacterial membrane fraction and not significantly exported into the periplasm. One such mutant was L-10E, in which a glutamate residue was inserted into the hydrophobic region of the PelC leader peptide sequence. Western blots confirmed that, as expected, the resulting preprotein accumulated in the E. coli membrane fraction with little or no cleavage to give the processed, mature protein. Several other PelC mutants (Y7D, D28R, C72S, M118N, D144N, H145Q, N210S, H228Q, and C329S) also were not produced as soluble periplasmic proteins in the T7 system, but remained associated with the bacterial inner membrane. In contrast to L-10E, Western blots established that a significant portion of these mutant PelC proteins ran on SDS gels at the size of the mature, processed PelC protein (data not shown). Thus, the signal peptide appeared to have been cleaved, but the mutant proteins failed to fold or release properly, and instead remained associated with the inner membrane. Some of these mutants were sensitive to trypsin digestion (e.g. M118N and H145Q; data not shown), suggesting that they were in an unfolded or semi-folded state. When these same mutant proteins were produced in the pINK1 expression system, only trace amounts of protein were detected by Western blots in either the membrane or periplasmic fractions. This may be interpreted as indicative of proteolysis of improperly folded protein products in the slower expressing pINK1 system. In contrast, expression of D28R, C72S, and D144N in the pINK1 system at 22 °C led to low but recoverable amounts of processed, periplasmic proteins from E. coli strain HB101. These periplasmic preparations exhibited the peculiarity of having significant contamination with a 57-kDa protein that was difficult to remove except by repeated cation exchange chromatography. Further experiments are under way to determine if the 57-kDa protein is specifically associated with the mutant PelC proteins.