The Cleavable N-terminal Domain of Plant Endopolygalacturonases from Clade B May Be Involved in a Regulated Secretion Mechanism*

Polygalacturonases represent the most abundant carbohydrate hydrolase family in the Arabidopsis thalianagenome, and they are thought to be involved in nearly all of the developmental processes requiring cell wall modifications during the life cycle of the plant. By phylogenetic analysis, plant polygalacturonases fall into at least three groups, one of which is distinguished from the others by the presence of an additional N-terminal domain. We have used RDPG1, the polygalacturonase involved in pod dehiscence in oilseed rape (Brassica napus), as a model to investigate the function of this domain. We have confirmed that this domain is absent in the mature protein by determination of the N-terminal sequence of mature RDPG1 purified from oilseed rape pod. We have furthermore investigated the accumulation and subcellular localization of the precursor containing the N-terminal domain and of the mature protein throughout the development and maturation of the pod. Using recombinant expression in Pichia pastoris, we have produced the RDPG1 precursor, and we present evidence that the N-terminal domain of plant polygalacturonases is not involved in folding or inactivation of the precursor but may play a role in the intracellular transport of this protein family via a novel regulated secretion pathway.

Polygalacturonases (PGs) 1 (EC 3.2.1.15) are one of the most important classes of enzymes associated with pectin degradation and cell wall rearrangements. Pectins are a major component of the cell wall of dicotyledonous plants, where they make up about 30% of the primary wall and most of the middle lamella (1). Diverse roles have been proposed for pectins, including the regulation of porosity and hydration and mechan-ical properties of the cell wall, cell expansion, and cell-cell adhesion. All of these functions imply rearrangement or disassembly of the pectin network, and pectin metabolism is thus critical to many developmental processes. The presence of PGs has indeed been detected in all developmental processes requiring cell wall modifications. The role of PGs in the late phases of fruit ripening, in organ abscission, in pod dehiscence, and in pollen maturation is well documented (2)(3)(4)(5)(6)(7)(8), but PGs have also been associated with wounding, interactions with parasites, rhizobium-legume symbiosis establishment, seed germination, vascular tissue differentiation, and pathogen defense response (9 -12). Recently, expression of PGs has also been evidenced at low levels and ubiquitously in vegetative growing tissues, where PGs are thought to be involved in growth and development (13).
PG genes are highly frequent in plant genomes, and their deduced amino acid sequences are rather divergent. In Arabidopsis thaliana, they represent the largest family of polysaccharide hydrolases with 52 genes distributed on the five chromosomes (14), and this is likely to be the case in other plant species as well. This abundance indicates a tight differential regulation of the spatial and temporal expression of PGs as is suggested by the sequence diversity found in the untranslated 5Ј regions of the PG genes by Northern analysis. The sequence divergence observed in the coding regions may be associated with different properties of each of the PG enzymes such as pH optimum, substrate preference, and mode of hydrolysis, which would affect physiological functions and cell wall breakdown.
All known plant PGs are secretory proteins, having a signal peptide. Until recently plant PGs were classified in three clades, A, B, and C, based on amino acid sequence comparison (15) (Fig. 1A). However, this classification is not absolute, and it may not be sufficient any longer, considering the growing number of PG sequences in the data bases. A recent phylogenetic analysis including 43 plant PG sequences (13) suggests that plant PGs fall into five classes, three of which corresponding to the previously defined clades A, B, and C. The newly revealed clades D and E include A. thaliana genes only at present and are more closely related to clade B than to clade C. The expression patterns remain to be investigated in clades D and E, whereas clades A, B, and C contain more sequences and are better characterized. Clade C comprises solely PGs that are expressed in pollen and that are thought to be exo-enzymes, because only exo-PG activity has been detected in pollen (7). Clades A and B comprise endo-PGs and probably also exo-PGs (12). Clade A includes PGs expressed in fruit and abscission zones, and clade B includes PGs expressed in fruit, dehiscence zones, and germinating seeds. Thus, it is not possible to distin-guish between clades A and B based on mode of action or tissue specificity. However, one feature that sets clade B, and possibly clade E, apart from the other clades is the presence of an additional 40 -50-amino acid-long N-terminal domain immediately following the signal peptide (3,5,13,15).
Propeptides, which are most often located at the N terminus of the mature protein domain, are a common feature of many protein precursors. However, their function has been extensively studied only in protease and peptide hormone precursors, in which the removal of the propeptide is associated with activation of the enzyme or hormone moiety (16). In addition to their role in zymogen activation, propeptides may carry sorting signals for the correct targeting of their cognate proteins to their final subcellular location (17). More recently, propeptides were found necessary for the correct folding of bacterial and yeast secretory proteases in vivo as well as in vitro in denaturation-renaturation experiments (18). By analogy, it has commonly been assumed that the predicted propeptide of PGs from clade B is involved in keeping the enzyme inactive until its activity is required or in directing the protein to a specific location within the cell wall, but these hypotheses have never been tested (3,13,19). Thus, elucidating the function of this N-terminal domain may help to understand the physiological relevance behind the distinction between clades A and B. The rdpg1 gene, encoding the oilseed rape (Brassica napus) pod dehiscence related PG and belonging to clade B was isolated in our laboratory (6) as well as by Jenkins et al. (4). RDPG1 provides a useful model to study clade B enzymes, including the function of their N-terminal domain or predicted propeptide.
We have therefore revisited and confirmed the presence of a cleavable N-terminal domain, which is absent in the mature protein for PGs from clade B using RDPG1 as a model, and we have furthermore investigated the site and time of cleavage of this N-terminal domain from the precursor protein in the plant. Recombinant expression of the RDPG1 precursor in Pichia pastoris provided us with a unique source of the precursor and enabled us to address the function of the N-terminal domain in folding and in activation of RDPG1. Finally, we present data pointing at a function for the N-terminal domain of PGs from clade B in the intracellular transport of its enzyme moiety via a novel and regulated secretion pathway. The elusive and transitory nature of most precursors makes their study difficult, and to our knowledge, this is the first time that the function of the propeptide-like domain of a protein different from a hormone or a protease has been studied.

EXPERIMENTAL PROCEDURES
Plant Material-Culture of oilseed rape cultivars Topas and Fido and flower tagging to determine the days after anthesis (DAA) at each harvest were carried out as described by Petersen et al. (6). Whole plants were sprayed with 2-methyl-4-chlorophenoxyacetic acid (4-CPA) as described by Chauvaux et al. (20).
Production of Polyclonal Antibodies to Pro-RDPG1 and to Its Nterminal Domain or Presumed Propeptide-Pure pro-RDPG1 recombinantly expressed in P. pastoris was used as antigen to produce antiserum OA01 to pro-RDPG1. The presumed propeptide of RDPG1 was expressed in E. coli as a fusion protein with His 6 -dihydrofolate reductase (N-terminally to the presumed propeptide). For this, the nucleotides 156 -284 region corresponding to the presumed propeptide of RDPG1 (L24-S66) was amplified by PCR from the full-length cDNA clone RDPG1 encoding pro-RDPG1 (6) using primers (5Ј-GGATCCTT-GAGTAGCAACGTAGAT-3Ј and 5Ј-CTCGAGTCATGATTCGGTAGT-GGGTCT-3Ј) designed to introduce a BamHI and a XhoI restriction sites (underlined in the sequences) at the 5Ј-and 3Ј-ends, respectively, and a stop codon (in italics in the sequence). The PCR fragment was then cloned into pQE-40 (Qiagen), and the resulting plasmid was checked by sequencing and transformed into Escherichia coli SG13009[pREP4]. Expression of the recombinant protein was induced with 2 mM isopropyl-␤-thiogalactopyranoside. Cell harvest, lysis, and purification of the recombinant protein under denaturing conditions using a nickel-nitrilotriacetic acid agarose column were performed according to the manufacturer's directions (Qiagen). The purified His 6dihydrofolate reductase-presumed propeptide fusion protein was dialyzed against a 50 mM potassium phosphate buffer, pH 7.5, containing 100 mM KCl, 2 M urea, and glutathione (2 mM GSH and 0.2 mM GSSG). Rabbits were immunized both intramuscularly in the hind leg and subcutaneously in the neck region at 4-week intervals with 70 g of antigen emulsified with RIBI adjuvant R-730 (RIBI ImmunoChem Research Inc.). Preimmune serum was collected before the first injection, and antisera were collected 12 days after each of four injections.
Immunopurification of RDPG1 from Oilseed Rape Dehiscence Zones (DZ)-An immunoaffinity column was prepared by coupling 5.5 mg of pro-RDPG1/RDPG1 specific IgG to 1 ml of Affi-Gel Hz Hydrazyde affinity support according to the manufacturer's instructions (Bio-Rad). Coupling yield was 64%. Pro-RDPG1/RDPG1 specific IgG was purified from antiserum OA01 using a protein A-Sepharose CL-4B column (Amersham Pharmacia Biotech). DZ also comprising the replum and the septum were separated from the remaining pod wall and seeds of mature pods and frozen into liquid nitrogen. DZ (5 g) were ground in liquid nitrogen, and a protein extract was obtained by further grinding with sand and 20 ml of 10 mM MOPS, pH 7.0, supplemented with a protease inhibitor mixture (Complete TM , Roche Molecular Biochemicals) and centrifugation (12,000 ϫ g for 20 min). The resulting salt-free supernatant contained the readily soluble protein. The pellet was resuspended in 20 ml of water and centrifuged as described above. The supernatant was discarded and the pellet was resuspended in 10 ml of 1 mM MOPS, 1 M LiCl, pH 7.0, and stirred for 30 min on ice. The sample was centrifuged as described above. The resulting pellet was discarded, whereas the resulting LiCl supernatant represented proteins that are strongly bound to the cell wall through ionic interactions. For immunopurification of RDPG1, the salt-free supernatant or the 2-fold diluted LiCl supernatant was mixed with 1 ml of immunoaffinity matrix and incubated overnight at 4°C, under rotation, for binding of RDPG1 to the antibody. The gel was then poured into a disposable column and allowed to settle. Washing was carried out with 10 ml of phosphatebuffered saline. After a pre-elution wash with 10 ml of 10 mM sodium phosphate, pH 7.0, RDPG1 was eluted with 100 mM glycine, pH 2.0, and collected in 1-ml fractions into tubes containing 50 l of 1 M sodium phosphate, pH 8.0.
Recombinant Expression of Pro-RDPG1 and ⌬Pro-RDPG1 in P. pastoris-The nucleotides 156 -1388 and 285-1388 regions of the cDNA clone RDPG1 (6), corresponding to the presumed proenzyme (L24-P433) and mature (S67-P433) forms of RDPG1, respectively, were amplified by PCR using the forward primers 5Ј-AAAGAATTCCTTTGAGTAG-CAACGTAGAT-3Ј and 5Ј-AACTGCAGCATCAACTGTTAGTGTTTC-3Ј and the reverse primer 5Ј-GCTCTAGATCATTAAGGGCATTTAGG-3Ј. These primers had been designed to introduce an EcoRI or PstI restriction site at the 5Ј-end of the PCR product and a XbaI site at its 3Ј-end (underlined in the sequences), as well as two stop codons at the 3Ј-end (in italics in the sequence). The EcoRI/XbaI-or PstI/XbaI-digested PCR product, corresponding to nucleotides 156 -1388 or 285-1388 regions, respectively, was cloned into the P. pastoris expression vector pPICZ␣B (Invitrogen). The RDPG1 cDNA fragments were inserted in frame at the 3Ј-end of the Saccharomyces cerevisiae ␣-factor secretion signal to ensure secretion of the recombinant proteins and a stop codon had been included before the myc epitope His 6 tag. The resulting recombinant plasmids proRDPG1pPICZ␣B and ⌬pro-RDPG1pPICZ␣B were transformed into E. coli XL1-Blue, checked by sequencing, linearized with SacI, and subsequently transformed in the P. pastoris X-33 strain by electroporation according to the manufacturer (Invitrogen). Transformants were selected on plates with YPDS medium and Zeocin (100 g/ml). For each construct, approximately 50 transformants were screened for high secretion of the recombinant protein in 10-ml cultures under expression inducing conditions for up to 96 h. Aliquots of the culture medium (150 l) were taken out every 24 h and analyzed by immuno-dot-blot. The transformants with the highest level of secreted recombinant protein (pro45 and A11 for pro-RDPG1 and ⌬pro-RDPG1, respectively) were chosen for large-scale expression.
Purification of Pro-RDPG1 and ⌬Pro-RDPG1 Recombinantly Expressed in P. pastoris-For large scale expression of pro-RDPG1 and ⌬pro-RDPG1, the P. pastoris transformant A11 or pro45 was cultivated in BMGY/BMMY according to the manufacturer's instructions (Invitrogen). After 72 h of culture in BMMY, cells were harvested by centrifugation (3000 g for 10 min) and discarded. The supernatant (2 liters) was filtered through Whatman folded filters and a protease inhibitor mixture (Complete TM , Roche Molecular Biochemicals) was added. All of the following steps were performed at 4°C. The supernatant was ultrafiltrated to a final volume of 100 ml using a Minitan unit equipped with a high flux Biomax polysulfone membrane with a cut-off of 10 kDa (both from Millipore). The resulting crude extract was applied to a SP-Sepharose Fast Flow XK50/100 column (Amersham Pharmacia Biotech) equilibrated with 100 mM sodium acetate, pH 5.5, containing 0.2 M NaCl. After a washing step using the same buffer, pro-RDPG1 or RDPG1 was eluted with 100 mM sodium acetate, pH 5.5, containing 0.45 M NaCl. The fractions containing enzyme activity were pooled; dialyzed twice for 2 h each against 25 mM sodium acetate, pH 5.5, containing 50 mM NaCl; and applied to a Mono-Q HR5/5 and a Mono-S HR5/5 column (Amersham Pharmacia Biotech) connected in this order and equilibrated with 50 mM sodium acetate, pH 5.5, containing 70 mM NaCl. The enzyme passed through the Mono-Q column and bound to the Mono-S column, from which it was subsequently eluted using a 70 -1000 mM NaCl gradient in 20 ml or a 70 -500 mM NaCl in 60 ml for RDPG1 and pro-RDPG1, respectively. For pro-RDPG1, the Mono-Q column was omitted. The pure enzymes were then stored in 10% glycerol at -18°C. The concentration of solutions of pure pro-RDPG1 and ⌬pro-RDPG1 was estimated spectrophotometrically using extinction coefficients (E 1% ) of 0.793 and 0.851, respectively (21).
Determination of Endopolygalacturonase Activity-Endopolygalacturonase activity was routinely assayed viscosimetrically in an Ostwald microviscosimeter (Schott-Gerä te) at 35°C in 2 ml. The reaction was initiated by the addition of enzyme in 1 ml of sodium acetate 200 mM, pH 5.0, to 1 ml of a 1% pectin solution (citrus pectin, 31% ME, Hercules) in the same buffer also containing 10 mM EDTA and 0.01% sodium azide. Viscosity was measured at time intervals, and 1 activity unit was defined as in Ref. 22.
Denaturation-Renaturation Experiments-Denaturation-renaturation was achieved in two different ways. 8 M guanidium chloride, 50 mM sodium acetate, pH 5.0, was added to 100 l of pro-RDPG1 (90 g) or 50 l of ⌬pro-RDPG1 (160 g) to a final guanidinium chloride concentration of 6.5 M. After incubation for 30 min at room temperature, the samples were renatured by buffer exchange on a Nap-5 column (Amersham Pharmacia Biotech) equilibrated with 50 mM sodium acetate, pH 5.0. Guanidinium chloride was omitted in control samples. Alternatively, 20 l of pro-RDPG1 and ⌬pro-RDPG1 (20 and 65 g, respectively) were freezedried and resolubilized in the same volume of 50 mM sodium acetate, pH 5.0 (control), or the same buffer containing 6.5 M guanidinium chloride, or the same buffer containing 6.5 M guanidinium chloride and 10 mM DTT. After incubation for 1 h at room temperature the samples were renatured by a rapid 60-fold dilution with 50 mM sodium acetate, pH 5.0. In all cases, the samples were then placed on ice, and endopolygalacturonase activity was measured at various time intervals. Emission fluorescence spectra were recorded on a thermostated PerkinElmer Life Sciences LS50B luminescence spectrometer using an excitation wavelength of 275 nm.
Electron Microscopy and Immunolocalization of Pro-RDPG1, RDPG1, and Pectin-Slices of the pod wall tissue, 1 mm thick, which contained the dehiscence zone were fixed and embedded as in Ref. 6. Nonspecific binding sites in the sections were saturated with 6% bovine serum albumin/0.1% fish gelatin/5% normal goat serum/0.05% glycine in phosphate-buffered saline, pH 7.6, for 30 min. The sections were then incubated with the RDPG1 antiserum (1:1000) or the anti-pectin antibody JIM 5 (23) (1:100) in phosphate-buffered saline/1% bovine serum albumin overnight at 4°C. Following rinsing the sections were incubated for 2 h at room temperature with the secondary antibody (10 nm of gold-gold anti-rabbit, 1:10, British Biocel International). The sections were washed, dried onto Formvar film (Agar Aid Bishop's), and counterstained with lead citrate and examined using a transmission electron microscope (model H-7000, Hitachi) at an accelerating voltage of 75,000 V. Determination of the distribution of labeling, as consolidated in Table I, relied on examination of four sections from each of three pods for each treatment.

Immunopurification of RDPG1 from Mature Pod DZ and
Determination of Its N-terminal Sequence-Native RDPG1 was purified to homogeneity from mature pod dehiscence zones (approximately 50 DAA) using immunoaffinity chromatography on an RDPG1-antiserum column in order to determine its precise N terminus and thereby verify the existence of a cleavable, propeptide-like, N-terminal domain. Purification of RDPG1 to homogeneity, as assessed by SDS-PAGE, could thus be achieved in a single step. From 5 g of oilseed rape DZ (fresh weight), 0.25 mg of pure RDPG1 was obtained. SDS-PAGE revealed the presence of two bands in the immunopurified RDPG1: a major band of 49 kDa and a minor band of 54 kDa. Western blots probed with an antiserum raised against the N-terminal domain or against the whole RDPG1 showed that the major band corresponded to mature RDPG1, whereas the minor band contained pro-RDPG1 ( Fig. 2A). N-terminal sequencing of the protein in the major band identified Glu-65 as the exact N terminus of the mature RDPG1. This confirms that RDPG1 is synthesized as a precursor containing a cleavable, propeptide-like N-terminal domain and furthermore reveals the precise length of this N-terminal domain and the exact N terminus of the mature RDPG1. Thus the RDPG1 precursor is composed of a predicted 23-amino acid-long signal peptide (Met-1-Ala-23) (24), a 41-amino acid-long cleavable N-terminal domain (Leu-24 -Thr-64), and a 367-amino acid-long catalytic domain starting at Glu-65. Furthermore, N-terminal sequencing of the purified protein revealed microheterogeneity with three minor forms starting at Ser-58, Asp-60, and Ser-66, respectively (Fig. 1B).

Investigation of a Putative Role of the Cleavable N-terminal Domain of RDPG1 in Protein Folding in Vivo by Comparison of
Recombinant Expression of Pro-RDPG1 and ⌬Pro-RDPG1 in P. pastoris-In order to investigate whether the N-terminal domain of RDPG1 is necessary for the correct folding of its cognate protein in vivo, we compared the expression of pro-RDPG1 and ⌬pro-RDPG1 (the N-terminal domain truncated form of pro-RDPG1, corresponding to the mature RDPG1) recombinantly in P. pastoris. For this, a truncated cDNA lacking the signal peptide coding sequence (bases 1-155) or lacking the signal peptide and the N-terminal domain coding sequence (bases 1-284) was cloned into the P. pastoris vector pPICZ␣ in frame with a DNA sequence encoding a yeast signal peptide to direct the recombinant protein to the endoplasmic reticulum and through the secretory pathway. P. pastoris was transformed with the resulting plasmids (Fig. 1, C and D). Both constructs were able to direct secretion of comparable levels of pro-RDPG1 and ⌬pro-RDPG1, respectively, into the culture medium, indicating that cleavable N-terminal domain is not required for correct folding of RDPG1 in P. pastoris, as unfolded proteins are not allowed to be secreted but are degraded inside the cell. However, P. pastoris could not posttranslationally process the precursor protein into the mature endopolygalacturonase, as shown by SDS-PAGE (Fig. 2B) and by N-terminal sequencing (Fig. 1C). This indicates that the protease(s) responsible for this maturation process in the plant is absent in P. pastoris. The normally transitory nature of precursors is a great obstacle to their isolation and study and the inability of P. pastoris to process pro-RDPG1 into RDPG1 provided us with a unique source of the pro-RDPG1 precursor for further investigations of the function of its cleavable N-terminal domain.
Purification of Recombinant Pro-RDPG1 and ⌬Pro-RDPG1 and Investigation of a Putative Role of Cleavable N-terminal Domain in Enzyme Activation-In order to compare their specific activities, recombinant pro-RDPG1 and ⌬pro-RDPG1 were purified from the P. pastoris culture broth before their activity was measured. The purification to homogeneity of pro-RDPG1 and ⌬pro-RDPG1 was achieved with ultrafiltration followed by two ion exchange chromatography steps (Table II). For ⌬pro-RDPG1, the last step was a combined anion and cation exchange chromatographic separation, in which the enzyme passed through the anion exchanger and bound to the cation exchanger. From 2 liters of culture broth, 8 -10 mg of 500 -600fold purified pro-RDPG1 and 3-4 mg of 600-800-fold purified ⌬pro-RDPG1 were obtained. In the last separation step ⌬pro-RDPG1 eluted in a single peak, corresponding to a single band in SDS-PAGE. In contrast, pro-RDPG1 eluted in four peaks. SDS-PAGE revealed that each peak contained a single protein, but of varying molecular masses, and immunoblotting confirmed that all four proteins originated from pro-RDPG1, because they were all recognized by an antiserum to RDPG1. N-terminal sequencing of each isoform showed that this heterogeneity was due to random, partial proteolysis within the cleavable N-terminal domain part of pro-RDPG1, resulting in isoforms starting at Ser-26 (full-length pro-RDPG1 and most abundant form), Asp-42, Lys-57, and Ser-58, respectively (Fig.  1C). Full-length pro-RDPG1 was used for all further analyses. Specific activities of pure pro-RDPG1 and ⌬pro-RDPG1, determined using various pectic substrates, were very similar, showing that the precursor is fully active. Moreover, the specific activities of pro-RDPG1 and ⌬pro-RDPG1 were comparable to that of the native, mature enzyme immunopurified from oilseed rape pod DZ, indicating that the enzymatic properties of the recombinant proteins obtained from P. pastoris were similar, if not identical, to those of the native protein, thus confirming that these recombinant proteins are good models to study pro-RDPG1 and RDPG1. Specific activities using our standard pectin substrate were 410, 447.5, and 372 units/mg for pro-RDPG1, ⌬pro-RDPG1, and native RDPG1, respectively. Thus, the cleavable N-terminal domain of RDPG1 does not play a role in enzyme activation, in contrast to nearly all propeptides in enzyme precursors in which this has been investigated.

Investigation of a Putative Role of the Cleavable N-terminal Domain of RDPG1 in Protein Folding Using in Vitro Denaturation-Renaturation Experiments-
The denaturation and renaturation of pro-RDPG1 and ⌬pro-RDPG1 was investigated by incubation of the proteins with 6.5 M guanidinium chloride followed either by buffer exchange to a negligible guanidium chloride concentration or by rapid dilution to 0.1 M guanidinium chloride. Reduction of the disulfide bridges was done by addition of 10 mM DTT in the denaturation buffer. In order to confirm that pro-RDPG1 and ⌬pro-RDPG1 had been denatured upon incubation with guanidinium, with or without DTT, their intrinsic fluorescence spectra were recorded. For both proteins, a reduction in E max and a shift in max toward higher wavelengths were observed in the presence of guanidinium, thus confirming that the proteins were denatured. The -shift was even higher in samples containing both guanidinium and DTT, suggesting, as expected, that the proteins are more extensively unfolded in the presence of DTT than in guanidinium alone (Table III). Recovered enzymatic activity was used as a measure of the extent of renaturation. In all cases, the maximum level of recovered activity was achieved after 15 min of renaturation and was stable for at least 2 h (data not shown). In experiments without disulfide bridge reduction, pro-RDPG1 and ⌬pro-RDPG1 recovered up to 92.6 and 74.6% of their initial activity, respectively (Fig. 3). This confirms that the cleavable N-terminal domain of RDPG1 is not required for folding. For both proteins, buffer exchange on a Nap-5 column yielded a higher renaturation level than rapid dilution, probably because the chromatographic support provides a refolding scaffold to the protein. In all cases, the levels of recovered activity were somewhat higher for pro-RDPG1 than for ⌬pro-RDPG1, but this marginal difference most likely reflects that these proteins have different requirements for optimal renaturing conditions, and the conditions used here may have been more favorable for pro-RDPG1 than for ⌬pro-RDPG1. The level of renaturation after denaturation in the presence of DTT was considerably lower than in the absence of DTT for both pro-RDPG1 and ⌬pro-RDPG1 (Fig. 3). This suggests that at least some disulfide bridges have been eliminated by the DTT treatment and that their correct reformation is essential for activity. Formation of disulfide bridges is known to be a rate-limiting step in folding, and correct reoxidation of up to all 12 cysteine residues of pro-RDPG1 and ⌬pro-RDPG1 may thus require a longer renaturation time than in our experimental conditions, as well as the presence of a catalyst, such as protein disulfide isomerase (25).   Investigation of the Oligomerization State of Pro-RDPG1 and ⌬Pro-RDPG1-We investigated the oligomerization state of pro-RDPG1 and ⌬pro-RDPG1 in order to find out whether the cleavage of the N-terminal domain mediates oligomerization of the mature RDPG1, as it is the case for the propeptide of some caspases (26,27). For this, we performed size exclusion chromatography of pro-RDPG1 and ⌬pro-RDPG1, both as a mixture and separately. The elution volumes of both proteins (14.0 ml and 14.7 ml for pro-RDPG1 and ⌬pro-RDPG1, respectively) corresponded to the expected elution volume of the monomers, and no difference in the elution volumes was observed when the proteins were chromatographed as a mixture. This shows that both pro-RDPG1 and ⌬pro-RDPG1 are monomeric and excludes the possibility that the cleavable N-terminal domain of RDPG1 plays a role in oligomerization.
Expression and Subcellular Localization of RDPG1 in Oilseed Rape Pod DZ-In order to investigate whether the cleavable N-terminal domain of RDPG1 could be involved in intracellular protein transport, we examined the fate of the RDPG1 precursor containing the N-terminal domain (pro-RDPG1) and of the mature enzyme (RDPG1) in the oilseed rape pod DZ throughout the development and maturation of the pod. Starting with anthesis, the development of the pod in oilseed rape can be segmented into three phases. The first phase (0 -20 DAA) is characterized by elongation of the pod to its full size and differentiation of the DZ, starting at around 10 DAA. During the second stage (20 -50 DAA) rigidification of the cell wall of edge cells and replum cells occurs and seed growth takes place. At the end of this phase, degradation of the pectin in the middle lamella in the DZ is initiated. In the final stage (50 -70 DAA), senescence takes place, and extensive degradation of the pectic lamella in the DZ leads to cell separation and dehiscence (6,28).
The accumulation of pro-RDPG1 and of RDPG1 was investigated on immunoblots of protein extracts from DZ at various ages between 17 and 52 DAA (Fig. 4). Pro-RDPG1 could not be detected in any extract (data not shown), whereas the mature RDPG1 could be detected in the DZ, where it appeared between 17 and 24 DAA and accumulated onwards, resulting in high levels in mature and senescent DZ. The subcellular localization of RDPG1 was further investigated by electron microscopy in ultrathin sections of DZ of transversally cut pods at various ages between 35 and 52 DAA using the antiserum to the whole RDPG1 protein and a gold-conjugated secondary antibody ( Fig.  5 and Table 1). In 35-DAA DZ, labeling was detected strictly intracellularly and was absent from the cell wall (Fig. 5B). The amount of intracellular RDPG1 then declined from 47 DAA and onwards. Concomitantly, RDPG1 labeling appeared in the cell wall, in the pectin-rich middle lamella, where it accumulated throughout senescence (Fig. 5C). The intracellular RDPG1 labeling appeared to be restricted to defined areas, reminiscent of subcellular organelles, possibly the endoplasmic reticulum, thus suggesting that RDPG1 is stored retained inside the cell for a period of time prior to its secretion into the wall. No labeling was observed using the N-terminal domain antiserum in the DZ of green or senescent pods, confirming that pro-RDPG1 does not accumulate in the DZ.
Chauvaux et al. (20) have observed that there is a decrease in auxin content specifically in the DZ of oilseed rape pods prior to degradation of pectin in the middle lamella of the cell wall leading to dehiscence. Furthermore they showed that treatment of the pods with the auxin analogue 4-CPA results in a 10-day delay of cell separation in the DZ. These results indicate that a low level of auxin in the DZ is necessary for dehiscence to take place. In order to investigate whether the release of RDPG1 to the cell wall required a low level of auxin in the DZ, we sprayed oilseed rape plants with 4-CPA. Subcellular localization of RDPG1 was analyzed by electron microscopy as above. Upon spraying with 4-CPA at 39 DAA, the appearance of RDPG1 in the cell wall was delayed from 47 to 52 DAA. At the same time the intracellular accumulation of RDPG1 was significantly higher than in nonsprayed, control plants, indi- cating that secretion of RDPG1 to the cell wall was blocked by the presence of 4-CPA and RDPG1 thus accumulated intracellularly (Fig. 5, C and D). Hence the ratio of intracellular RDPG1 to cell wall RDPG1 was altered in 47-49-DAA DZ in 4-CPA sprayed plants as compared with nonsprayed, control plants. Concomitantly with the delay of the appearance of RDPG1 in the cell wall, a delay in pectin degradation was observed as evidenced by the immunolabeling using the JIM5 monoclonal antibody (23), specific for pectic homogalacturonan with low degree of methyl esterification, the substrate of RDPG1. Subsequently, a delay in cell separation was also observed (data not shown), establishing a clear dependence of dehiscence on the release of RDPG1 to the cell wall. DISCUSSION The postulate that the additional 40 -50-amino acid-long N-terminal domain of PGs from clade B is a propeptide was based on one example only, that of the tomato fruit PG. Comparison of the protein sequence deduced from full-length cDNA clones with the N-terminal sequence of the PG purified from tomato fruit revealed that the first 47 amino acids following the signal peptide were absent in the mature protein. This was further confirmed in in vitro translation experiments in the presence of microsomes (19,29). In the present study, we have established that this is also true for the oilseed rape pod polygalacturonase RDPG1, another member of this protein subfamily. In identifying the additional N-terminal domain of another PG from clade B as a cleavable domain, which is absent in the mature protein, we have confirmed experimentally the universal presence of a cleavable N-terminal domain in all PGs from clade B. The cleavable N-terminal domains of PGs from clade B do not share significant sequence homology but have a high content of charged amino acids. Moreover, the cleavable N-terminal domain of RDPG1 exhibits susceptibility to random proteolysis, as shown by the occurrence of isoforms with different N termini, all residing within this domain, upon recombinant expression of pro-RDPG1 in P. pastoris. This indicates that the cleavable N-terminal domain is not fully and compactly folded. All of these are general structural features of propeptides (30,31). The microheterogeneity found at the N terminus of the native RDPG1, purified from pod DZ, could be an indication that the processing of pro-RDPG1 into RDPG1 is a multistep process involving several proteases, e.g. an endoprotease and one or several aminopeptidases. Moreover, one or more proteases required for the cleavage of pro-RDPG1 are absent in the yeast P. pastoris, suggesting that this processing is unique to plants.
Our study reveals that the cleavable N-terminal domain of PGs from clade B is not involved in keeping its cognate enzyme inactive. This result is important, because it reveals that the often encountered hypothesis that the cleavable N-terminal domain of PGs from clade B is involved in preventing premature activation of the enzyme is wrong (3,13,19). We also show that the cleavable N-terminal domain of PGs from clade B is not required for folding or dimerization. Our results on folding from in vivo experiments and from in vitro denaturation-renaturation experiments correlate very well and support each other. However, this does not exclude the possibility that the N-terminal domain might have an influence on the kinetics of the folding of PGs from clade B in vivo. In conclusion, although the cleavable N-terminal domain of PGs from clade B presents the structural characteristics of a propeptide, it does not have the classical functions of a propeptide, in keeping its cognate enzyme as an inactive zymogen and/or in assisting protein folding. Indeed, in virtually every case in which this has been investigated, zymogen propeptides have been shown to keep their cognate enzyme inactive. Therefore, the cleavable N-terminal domain of PGs from clade B cannot be regarded as a true propeptide.
Our time course study of the accumulation of RDPG1 in the pod DZ using immunoblots and gold immunolocalization shows that the protein accumulates over a long period of time in the pod DZ in accordance with the rdpg1 promoter activity analysis (32). The precursor form is absent from the bulk of RDPG1 in oilseed rape pod DZ. Only following a high enrichment of protein extracts in pro-RDPG1 and RDPG1 by immunoaffinity chromatography could pro-RDPG1 be detected in the oilseed rape DZ (Fig. 2A). This shows that pro-RDPG1 is a transitory form and that the N-terminal domain is cleaved early after biosynthesis of pro-RDPG1. Strikingly, our electron micrograph results furthermore indicate that RDPG1 is temporarily stored inside the cell for about 3 weeks after its biosynthesis and before its release to the cell wall. These data are corroborated by the observation that the auxin analogue 4-CPA retarded or even prohibited the release of RDPG1 to the cell wall whereas it did not seem to affect its biosynthesis. Because RDPG1 is responsible for the degradation of pectin in the middle lamella, leading to dehiscence (4, 6, 32), the abundant JIM5 immunogold labeling of the pectin in the middle lamella upon spraying with 4-CPA, provides an additional proof of the delay in secretion of RDPG1 upon spraying with 4-CPA, thus confirming the results of our RDPG1 immunogold labeling. Altogether, this suggests that the biosynthesis of RDPG1 and its secretion to the cell wall do not occur simultaneously but are two distinct processes, exhibiting different hormonal dependence, and that the secretion of RDPG1 to the cell wall does not occur via the default pathway but is a regulated event.
The presence of PG mRNAs in planta long before pectin undergoes enzymatic breakdown has been observed in oilseed rape pod dehiscence zone (this study), in tomato, and in Charentais melon (33)(34)(35). This raises the question of how premature action of PGs on their substrates is prevented and various hypotheses have previously been proposed. In tomato fruit, it has been proposed that PG could associate with the ␤-subunit, a small protein that could restrict PG activity by binding either to the enzyme or to its substrate (36). Recently Brummel et al. (37) have obtained data indicating that the tomato fruit expansin Exp1 may control access of PG to its substrate via pectin relaxation. In addition to such mechanisms regulating substrate accessibility within the cell wall, the sequestration of PG in a subcellular compartment until the triggering of its secretion to the wall could provide another efficient mechanism to prevent it from degrading pectin prematurely.
In plant cells, secretion to the cell wall is the default pathway for all proteins entering the endoplasmic reticulum. Transport of secretory proteins to another destination, e.g. the vacuole, requires a sorting signal within the protein sequence. This signal can be located on a cleavable N-or C-terminal peptide or within the mature protein (38 -41). Therefore, our present data lead us to propose that RDPG1 is transported to the cell wall through a novel regulated secretion pathway. More precisely, we propose that RDPG1 is sorted away from the default secretion pathway and is stored intracellularly until its secretion to the cell wall is triggered by a signal that is yet to be uncovered and that may be regulated by the level of plant hormones and by the developmental stage of the tissue. Such a secretion mechanism could allow a fast and highly dynamic cell wall response, with the enzyme being physically isolated from its substrate until needed but instantly provided when needed. We further propose that the sorting signal for this novel pathway resides within the cleavable N-terminal domain. Sequence comparison of the cleavable N-terminal domain of 14 PGs from clade B has identified the presence of a four-amino acid consensus sequence on 11 of the 14 sequences. 2 This consensus sequence, (D/E)X(G/P)(Y/F), where X is not Gly, therefore represents a good candidate for a sorting signal. If the sorting signal lies within the cleavable N-terminal domain, it is reasonable to imagine that this novel secretion pathway is valid for all PGs from clade B. In turn, the physiological basis for the distinction between clade B and the other clades could be their secretion mechanism, and thus PGs from clade B could represent a class of PGs involved in rapid cell wall changes.
The expression in transgenic plants of a reporter protein (e.g. GUS or GFP) alone or fused with the cleavable N-terminal domain of RDPG1 and driven by the RDPG1 promoter and signal peptide and comparison of the fate of the reporter protein expressed with or without the cleavable N-terminal domain will help to challenge the hypothesis that the cleavable N-terminal domain of PGs from clade B is involved in the sorting of these proteins toward a novel regulated secretion pathway.