A Pectate Lyase-Coding Gene Abundantly Expressed during Early Stages of Infection Is Required for Full Virulence in Alternaria brassicicola

Alternaria brassicicola causes black spot disease of Brassica species. The functional importance of pectin digestion enzymes and unidentified phytotoxins in fungal pathogenesis has been suspected but not verified in A. brassicicola. The fungal transcription factor AbPf2 is essential for pathogenicity and induces 106 genes during early pathogenesis, including the pectate lyase-coding gene, PL1332. The aim of this study was to test the importance and roles of PL1332 in pathogenesis. We generated deletion strains of the PL1332 gene, produced heterologous PL1332 proteins, and evaluated their association with virulence. Deletion strains of the PL1332 gene were approximately 30% less virulent than wild-type A. brassicicola, without showing differences in colony expansion on solid media and mycelial growth in nutrient-rich liquid media or minimal media with pectins as a major carbon source. Heterologous PL1332 expressed as fusion proteins digested polygalacturons in vitro. When the fusion proteins were injected into the apoplast between leaf veins of host plants the tissues turned dark brown and soft, resembling necrotic leaf tissue. The PL1332 gene was the first example identified as a general toxin-coding gene and virulence factor among the 106 genes regulated by the transcription factor, AbPf2. It was also the first gene to have its functions investigated among the 19 pectate lyase genes and several hundred putative cell-wall degrading enzymes in A. brassicicola. These results further support the importance of the AbPf2 gene as a key pathogenesis regulator and possible target for agrochemical development.


Introduction
Alternaria brassicicola is a destructive plant pathogen and causes black spot disease on almost all plant species in the Brassicaceae [1][2][3]. Disease symptoms appear mainly on the leaves and stems of host plants, including Brassica oleracea (vegetables), B. rapa (vegetables, oilseeds, and forages), B. juncea (vegetables and seed mustard), the vegetable oil-producing species B. napus (oilseeds) [4], and the model plant Arabidopsis thaliana [5]. This disease is of worldwide economic importance [1][2][3]6,7] and can result in 20 to 50% yield reductions in crops such as canola and rape [7].
Alternaria brassicicola is a necrotrophic plant pathogen and its disease symptoms include the necrosis of host tissues, occasionally surrounded by yellow halos. The pathogenesis mechanisms employed by necrotrophic fungi are simplistically described as being comprised of two steps. The first step is the killing of host cells or inducing programmed cell death with toxins [8][9][10][11][12][13][14]. The next step is deconstruction of the dead tissue and assimilating it into the fungal biomass using various carbohydrate-active enzymes (CAZys) commonly known as cell walldegrading enzymes (CWDEs). It has been suspected that toxins and CAZys play important roles in pathogenesis [15], however, we are still searching for genes whose loss-of-function mutation causes a reduction in virulence.
For a successful parasitic lifestyle, efficacious invasion and subsequent colonization are crucial and the number of genes in each family involved in this process are speculated to be increased. Pectin-digesting enzymes are prominent examples. There are 19 pectate lyases and 7 pectin esterases in A. brassicicola, twice as many as in their homologs in other dothideomycete fungi [33]. Pectin-digesting enzymes are speculated to be involved in the invasion and colonization of host tissues by depolymerizing pectins in the middle lamella and plant cell walls, making them important virulence factors. Six pectate lyase genes (AB05514.1, AB00904.1, AB10322, AB06838.1, AB03608, AB10575.1) are induced by AbVf19 during the late stages of infection, after establishment and colonization, when plant tissues are necrotic [34]. Loss-offunction mutations of the most abundantly expressed gene (AB10322.1) or other pectate lyase genes, however, do not result in a reduction in virulence [15]. This suggests that the lost function of individual pectin digestion enzymes is either replaced or complemented by unknown enzymes. Alternatively, the major function of AB10322.1 is in something other than pathogenesis. Functional redundancy among CAZys, and functional specialization of individual genes within each family have been proposed previously to explain similar observations in Cochliobolus carbonum [35].
Recently, we identified two pectate lyase genes, PL1332 (AB01332.1) and PL4813 (AB04813.1), which are exponentially induced as early as 4 hours after fungal contact with the surface of its host and lasting up to 24 hours postinoculation [36]. These two genes are regulated by the transcription factor AbPf2, which is involved in the early stage of pathogenesis. Deletion strains of the AbPf2 gene are nonpathogenic, but its other phenotypes are the same as wild-type A. brassicicola in saprophytic growth, both in the presence and absence of stress-inducing chemicals [36]. In this study, we tested a hypothesis that the PL1332 gene encoding a pectin digestion enzyme is an important virulence factor. The results of this study provide another reason to further investigate the functions of other genes regulated by AbPf2 and to consider this transcription factor a good target for efficient management of diseases caused by A. brassicicola.

Expression patterns of pectate lyase genes during plant infection and saprophytic growth
Two putative pectate lyase-coding genes, PL1332 and PL4813, regulated by the transcription factor AbPf2, were dramatically induced soon after conidia were inoculated on leaves of host plants [36]. Expression levels of these genes and six other pectate lyase genes were further quantified and compared with transcripts of a gene encoding elongation factor 1-α (Ef1-α) (Fig 1). The expression levels of Ef1-α were more consistent than all other genes encoding housekeeping proteins [36]. The expression levels of all eight pectate lyase genes were less than 3% of the transcripts of Ef1-α at 4 hours postinoculation (hpi) (Fig 1A), but transcript levels of PL1332 and PL4831 were dramatically increased afterwards and reached levels comparable to Ef1-α by 12 hpi (Fig 1B). Subsequently, their expression levels decreased to less than 2% by 48 hpi when colonization was established. The expression levels of these genes remained low during saprophytic growth on both dead host tissue and axenic media (Fig 1D-1G). Notably, the presence of pectin as a major carbon source did not induce their expression (Fig 1G). The other six pectate lyase-coding genes (AB05514.1, AB00904.1, AB10322, AB06838.1, AB03608, AB10575.1) are putatively regulated by the AbVf19 transcription factor [34]. Although their expression was induced by AbVf19 during the late stage of infection, the magnitude of induction varied from less than 5% to over 200% compared to the expression levels of Ef1-α. Particularly, AB10322 and AB06838 among the six pectate lyase-coding genes were expressed at their highest levels during the late stages of infection. All eight of the pectate lyase-coding genes were expressed at low levels during saprophytic growth in a liquid culture medium and none were induced when pectin was a major carbon source in the medium.

Sequence similarity between two genes encoding pectate lyases
In addition to their similar expression pattern, the PL1332 and PL4813 genes shared three short blocks of similar sequences within a 1 kb sequence upstream from the start codon (Fig 2A) and an identical motif of a putative promoter [36]. The length of its genomic DNA was 835 nucleotides and contained one putative intron and two exons. We determined their cDNA sequence and defined the coding region for both genes (Fig 2A). There were three exons and two introns in each gene. The length of the coding regions and their nucleotide sequences were similar (Table 1 and S1 Fig). Their introns were also similar in their location, length, and sequence. Similarity in their expression profiles, gene structures, and gene sequences suggested that their functions were similar. For this reason, we decided to study one of the two genes instead of both.

Replacement of PL1332 with a HygB cassette
We designed a construct to create deletion strains of the PL1332 gene by replacing the coding region with a Hygromycin B transferase (HygB) gene cassette (Fig 2B). Southern hybridization with three probes against the genomic DNA extracted from eight transformants confirmed that the PL1332 gene was absent in all eight transformants ( Fig 2C). The PL1332 coding region was replaced by a single copy of the HygB resistance cassette in seven strains and by multiple copies in one of the gene-deletion strains (Δpl1332-7). In contrast to replacement of the PL1332 gene with a HygB cassette, the PL4813 gene was left intact in all strains.

Reduction in virulence of the Δpl1332 strains
We performed virulence assays using two strains, Δpl1332-1 and Δpl1332-2, to further characterize virulence attributes associated with PL1332. Both deletion strains produced lesions  approximately 30% smaller in diameter (p<0.001) than the wild type in detached-leaf assays ( Table 2 and Fig 3A). We performed similar experiments using the Δpl1332-1 strain on leaves still attached to the plant. The size of lesions produced by Δpl1332-1 strain was similarly reduced compared to the wild type on the leaves attached to whole plants ( Fig 3B). We also compared lesions produced by the wild-type, Δpl1332-1 strain, and a strain complemented with the wild-type allele. Lesions produced by the complemented strain were similar to those produced by the wild type ( Table 2). Results of these virulence assays provided evidence that loss of the PL1332 gene caused the reduction in virulence and that PL1332 was important for full virulence.

No differences in vegetative growth
We evaluated the importance of PL1332 on the growth of colonies on solid media and on mycelium production in liquid media that were either rich in nutrients, or contained only essential minerals supplemented with pectin as a major carbon source. On solid, nutrient-rich PDA, the average colony diameter was similar for the deletion strain and wild-type A. brassicicola (Table 3). Colony size was also similar for the Δpl1332-1 strain and wild type on minimal mineral agar supplemented with pectin. We inoculated two different liquid media with fungal mycelium from either the Δpl1332-1 strain or wild-type A. brassicicola, and then measured their dry weights four days later. Dry weight of the mycelium was similar for the mutant strain and the wild type in nutrient-rich GYEB as well as in a minimal medium supplemented with citrus pectin or glucose as a major carbon source (Table 4). Both strains and the wild type grew poorly in the minimal media supplemented with citrus pectin or glucose compared to growth in nutrient-rich media. This suggested that pectin digestion enzymes were not induced by citrus pectin and that PL1332 was not important in the use of citrus pectin.

Enzyme activities of PL1332 expressed in Escherichia coli
We failed to measure knockout effects of the PL1332 gene on the enzyme activity of pectate lyases secreted in the culture medium because the PL1332 gene was expressed at extremely low levels in the liquid medium, with or without pectin (Fig 1, GYEB and pectin). Further, it was not possible to measure enzyme activity in the inoculum collected from the infection sites when the PL1332 gene was highly induced because the fungal biomass was extremely small at 4 to 24 hours postinoculation. To verify its enzyme activity, we expressed the PL1332 protein by cloning the gene in a heterologous protein expression system ( Fig 4A). PL1332 was expressed as a fusion protein by linking it to maltose binding protein (MBP) (Fig 4B). Two amino acids at the N-terminus were deleted during the cloning of PL1332 cDNA in the proper reading frame, following the MBP-coding region. After IPTG-was induced, all transformants abundantly expressed the~68 KDa proteins expected from the fusion of MBP and PL1332 (MBP-PL1332). The fusion proteins were soluble and stayed in the cytosol of E. coli cells. However, the proteins were partially degraded after purification, unlike the intact MBP proteins (Fig 4B, compare lane 7 and lane 10). Treatment of the fusion protein with Factor Xa to remove the MBP domain caused complete degradation of the protein during overnight incubation at 4°C (data not shown). Thus, we were not able to perform enzyme assays using PL1332 after removing the MBP binding domain. Instead we performed enzymatic assays using the fusion proteins that were partially degraded. Enzyme activity was measured by the extent of enzymatic digestion of polygalacturonic acid to oligogalacturonic acid using a titrimetric stop-reaction method.
Purified MBP-PL1332 fusion proteins did not show enzyme activity under any of the test conditions (S2 Fig). We speculated that the fusion proteins were either degraded during protein purification and subsequent enzyme-assay conditions, or the enzyme required unknown co- Table 3. Growth of wild-type and the Δpl1332-1 strain of Alternaria brassicicola on PDA or on water agar with 1% (w/v) pectin.

Chemical
Colony diameter (mm) Wild type Δpl1332-1 Colony diameter indicates the average colony diameter ± standard deviation.
doi:10.1371/journal.pone.0127140.t003 factors for its activity. To circumvent possible problems caused by protein degradation, the absence of unknown cofactors in the reaction mixture, or both, we measured enzyme activity in the soluble fraction of whole lysates of E. coli that expressed MBP-PL1332 fusion proteins. Soluble lysate of E. coli expressing MBP was used as a control and it showed weak enzyme activity, as expected (Fig 5 MBP-PL1332). The soluble lysate of E. coli expressing MBP-PL1332 fusion proteins, however, showed significantly stronger enzyme activity (p < 0.01) than the control. We performed similar experiments using PL1332 proteins fused to glutathione-S-transferase (GST). The GST-PL1332 fusion proteins in soluble bacterial lysate also showed stronger enzyme activity than GST in soluble bacterial lysate (Fig 5 GST-PL1332).

Induced necrosis of host tissue
Toxins play important roles in necrotrophic parasitism in other fungi and we have been searching for similar toxins in A. brassicicola. We considered pectate lyases in general as toxin candidates because pectins are important components of the architecture of plant tissue. We tested whether PL1332 protein was toxic to host plants using MBP-PL1332 fusion proteins in a soluble fraction of bacterial lysate. Maltose binding protein in soluble bacterial lysate, protein wash buffer, or sterilized deionized water were used as controls. When each solution was injected between the leaf veins of host plants, local tissues around the injection sites immediately appeared waterlogged, but the symptom disappeared in about 3 hours (Fig 6).

Discussion
Pectins are structural heteropolysaccharides and key components of primary and secondary cell walls of flowering plants, and they are important for the protection of plants from abiotic stresses and biotic invasion [37][38][39]. Genes encoding pectinolytic enzymes are important virulence factors and their deletion or disruption causes a reduction in virulence of several phytopathogenic fungi, such as Aspergillus flavus, Botrytis cinerea, and Claviceps purpurea [40][41][42]. The genome of A. brassicicola contains about twice the number of genes encoding pectin-digestion enzymes as other dothideomycete fungi [33]. These enzymes are probably important for its pathogenic lifestyle, but evidence has been lacking until now. Previously, the disruption of a pectate lyase gene that was abundantly expressed during plant infection caused little or no reduction in the virulence of A. brassicicola [15]. Further, disruption of four other putative pectate lyase genes in this necrotroph did not change its virulence (Cho, unpublished data). Identification of individual pectate lyase genes associated with pathogenesis had been challenging until pectate lyase gene PL1332 was identified on the molecular level as an important virulence factor. This study clarifies why previous approaches were unsuccessful in identifying pectate lyase genes important in pathogenesis. There are 19 pectate lyase genes and 7 pectin esterase genes in A. brassicicola [33]. The pectate lyase-coding gene, AB10322, expressed at high levels by the fungal mycelium in necrotic plant tissue during host infection, was considered important for pathogenesis and selected as the best candidate for a virulence factor. In retrospect, though AB10322 was abundantly expressed during the late stages of infection, it was poorly expressed during the early stages of infection. Furthermore, four additional genes were also highly expressed during the late stage of infection (Fig 1). Pectate lyases encoded by AB10322 and the other four genes probably play important roles in deconstructing pectins and unlocking sugars for use as basic structural components of the fungal biomass. These available sugars would be important for subsequent colony expansion of the necrotroph and appearance of the typical disease symptom: macerated plant tissue. Thus, a gene-disruption strain was expected to result in a slower expansion of disease symptoms compared to the wild type, in contrast to no changes by the disruption of AB10322. A major reason for the unchanged virulence of the ab10322 strain was probably functional redundancy among the five pectate lyase-coding genes expressed at moderate to high levels during the late stage of infection (Fig 1). Functional redundancy was previously proposed for another pathogenic fungus, Cochliobolus carbonum [35].
These experimental results offer answers to lingering questions on the role of toxins in the pathogenesis of A. brassicicola. PL1332 was a strong toxin and caused necrosis in the host plants tested (Fig 6). This protein, however, was smaller in molecular weight than the previously reported 35 KDa AB-toxin [29,30]. It was also different from host-specific toxins that are secondary metabolites produced by several pathotypes of A. alternata [17,21,22,24,25]. These secondary metabolites are toxic to selected host plants and essential for pathogenicity. In comparison, PL1332 was toxic to all three host plants tested and gene-deletion strains were still pathogenic, suggesting that it is a general toxin rather than a host-specific toxin.
Pectins are major components of plant cell walls and the middle lamellae that help bind cells together. Therefore, the digestion of pectins will cause tissue collapse, cell membrane rupture, and subsequent tissue necrosis. It is also possible that oligopectins, or pectin derivatives digested by the enzyme, triggered host defense reactions [43] and programmed cell death. In these cases, enzyme activity was necessary to cause the necrosis. Alternatively, the enzyme activity of PL1332 is not required for necrosis, nor is the xylanase activity of Xyn11A secreted by B. cinerea [12]. Thirty amino acids in the xylanase Xyn11A are sufficient for toxicity, even without enzyme activity of xylanase. The latter idea was supported in our study when purified fusion proteins showed no detectable levels of enzyme activity (S2 Fig), but produced toxic effects (S3 Fig). This observation suggests that pectate-lyase enzyme activity is not necessary for toxicity. Further study is needed to clarify whether the toxic effects of PL1332 on host tissues resulted from or were independent of the enzyme activity. If the toxic effect is not caused by pectate lyase activity, introduction of the Δpl1332 with a defective PL1332 enzyme, or expression cassettes for short oligopepetide coded by PL1332, would restore full virulence and defective PL1332 proteins or short oligopeptides would still be toxic to host plants.
Regardless of the molecular mechanism, PL1332 is an important general toxin and it affects more than one species of host plant. It is the first protein molecularly substantiated in A. brassicicola that shows toxicity to host tissue. Unlike the five pectate lyase genes abundantly expressed during the late stages of infection, the PL1332 and PL4813 genes were highly expressed during early infection. These results suggest that these genes specifically interact with host plants early in the infection process, before or during penetration, rather than later during the conversion of sugars and colony expansion. This study raises several interesting questions, including the roles of the PL4813 gene on virulence, functional redundancy between the PL1332 and PL4813 genes, possible synergistic effects of the mutation of both genes, and the importance of the five late-stage genes in virulence.
This study was initiated based on discovery of the key-pathogenesis regulator, AbPf2 [36]. Its transcription was induced during the early stages of host infection, followed by the induction of 106 putative downstream genes, including PL1332 [36]. Data generated in this study provide the first evidence that these downstream genes might be important in pathogenesis. It is of note that this transcription factor also regulates six genes that encode small secretion proteins. They may act as effectors and be important in the interaction between host plants and pathogenic fungi [44][45][46] or fungus-like oomycete pathogens [47,48]. It would be practical and fruitful to explore the roles of these six putative effector genes that are explosively induced during the brief early infection stage. They may increase our understanding of the biological aspect of pathogenesis, especially the mode of secretion of these effector proteins and host-pathogen interactions. Our study results also provide a reason why the deletion or disruption of the AbPf2 gene causes a loss of pathogenicity in A. brassicicola. Inhibition of AbPf2 would provide full protection to plants from A. brassicicola infection, while inhibition of pectae lyase activities would provide partial protection. Therefore, inhibition of AbPf2 would be a better target than individual downstream genes, including PL1332. It is feasible to screen natural or synthetic compounds that inhibit the functions of AbPf2 and ultimately pathogenesis.

Maintenance of fungal strains and Southern hybridization
Growth and maintenance of Alternaria brassicicola Schweinitz & Wiltshire (ATCC96836), pathogenicity assays using deletion strains or wild-type A. brassicicola, and its transformation and nucleic acid isolation were performed as described previously [49]. Each strain created during this study was purified by two rounds of single-spore isolation to obtain a uniform genetic background. Loss-of-function mutation was verified by Southern hybridization using three probes, respectively representing the 5' upstream region, PL1332 coding region, or HygB gene cassette. Southern hybridizations were performed as described previously [15] following manufacturer's protocol (Roche Diagnostics, Mannheim, Germany) with appropriate modifications. Fungal DNA extracted from each transformant or wild-type A. brassicicola was digested with the endonuclease, HindIII. All three probes were synthesized with a PCR DIG Probe Synthesis Kit according to the manufacturer's manual (Roche Diagnostics, Mannheim, Germany). A gene-specific probe was generated with the primer set 1332ProbeF (CCCTCAACATCCCAGCTAGA) and 1332ProbeR (TGTTAATGGCGACAAGGTCA). The probe for the 5' flanking region was produced with 1332-DP1 (CGCACCCGTAAGAAGAAGAA) and 1332-DP2' (TTCAAAGTGGCAGAGCA CAC), and the HygB-specific probe was produced with HygIn84 (CTTGGCTGGAGCTAGTGGAG) and HygIn1343 (ATTTGTGTACGCCCGACAGT). Gene-deletion strains were maintained as glycerol stock in separate tubes with one tube used for each assay. The sequence data were deposited in the NCBI GenBank (KR024320-KR024323).

Pathogenicity assays
Either whole plants or detached leaves harvested from 5-to 8-week-old Brassica oleracea (green cabbage) were inoculated with 1-2 x 10 3 conidia in 10 μl of water. After infection, the plant materials were maintained in mini humidity chambers and the development of disease symptoms observed for 7 days. Pathogenicity assays were conducted multiple times and the disease symptoms recorded 5 dpi with a digital camera.

Creation of gene deletion-strains for PL1332
We made Δpl1332 deletion strains by replacing the 915 base pairs (bp) spanning the partial promoter (70 bp), whole protein-coding region (835 bp), and partial sequence of 3' untranslated region (bp nt) with a HygB resistance cassette (Fig 2B). The replacement construct was produced by two rounds of PCR as described previously [36]. Initially, a 978-bp-long 5' flanking region of the PL1332 gene, 1436-bp-long HygB cassette, and 954-bp-long 3' flanking region were amplified with three sets of primers; 1332-DP1 and 1332-DP2 (ATCAGTTAACGTCGACCTCGTTCAAAGT GGCAGAGCACAC); 1332-DP3 (GTGTGCTCTGCCACTTTGAACGAGGTCGACGTTAACTGAT and1332-DP4 (ATTGTGCTTTCCGTGGAGTCCGTCGACGTTAACTGGTTCC); 1332-DP5 (GGAACCAGTTAACGTCGACGGACTCCACGGAAAGCACAAT) and 1332-DP6 (AACTTTTCGGC AAAATCTCG). Subsequently, PCR products were mixed and used as template DNA to create the final construct by amplifying DNA with 1332-DP1 and 1332-DP6. Sequence similarity between PL1332 and PL4813 at the 5' flanking region was undetectable except for three short blocks marked in Fig 2A. In addition, there was no sequence similarity at the 3' flanking region. In short, the construct was designed to replace only the PL1332 coding region without affecting the PL4813 coding region. The final construct was transformed into the protoplast of wild-type A. brassicicola as described previously [15,36].

Complementation of Δabpf2-1 strain
The ΔPl1332-1 strain was complemented with the wild-type PL1332 allele and its native promoter as described previously [50].
Growth assays in the presence of major carbon sources Colony growth assays were performed on either potato dextrose agar (PDA) plates or water agar plates. Water agar (2% w/v) plates contained 0.5% (NH 4  . Each flask was inoculated with 4-6 x 10 5 conidia of either Δpl1332-2 or wildtype A. brassicicola and incubated in the dark at 25°C with continuous agitation at 100 rpm. The flasks were shaken vigorously by hand several times during the first eight hours to prevent conidia from aggregating and sticking to the wall of the flask. Mycelia were harvested at 4 days postinoculation, washed with distilled water, dried at 70°C overnight, and their dry weights measured.

Determination of cDNA sequence
Total RNA was extracted from a mixture of plant leaves and wild-type A. brassicicola at 12 hpi and used to produce cDNA as previously described [36]. Open reading frames of PL1332 and PL4813 were amplified from the cDNA with primer sets PL1332F (TTCACTGCCTTGACCAT TACCG) and PL1332 R (CATTGTGCTTTCCGTGGAGT); PL4813F (GGCCAGACTCTGAACAT TCC) and PL4813seqR (TTGCATTGCATTCTTTCTCG), respectively. The nucleotide sequence of the PCR products of each gene was determined with the primer sets used for the PCR amplification. Their cDNA sequence was then compared with a known genomic sequence to determine the structure of each gene.

Quantitative real-time PCR
Expression of the eight pectate lyase genes in wild-type A. brassicicola was measured by quantitative RT-PCR. We collected mixed samples of fungal and leaf tissue from inoculated B. oleracea at times that generally represented the five stages of pathogenesis: conidial attachment to the host plant and initiation of germination (4 hpi), penetration (12 hpi), colonization (48 hpi), saprophytic growth on necrotic host tissues (72 hpi), and saprophytic growth and conidiation (120 hpi). Tissues were frozen in liquid nitrogen as soon as they were collected. RNA extraction, cDNA synthesis, and qRT-PCR were performed as previously described [15,34]. Standard curves were produced with purified amplified DNA products of 10 pg/μl, 1 pg/μl, 100 fg/μl, 10 fg/μl, and 1 fg/μl starting concentrations. A baseline subtracted curve fit was used to generate standard curve data. Absolute amounts of transcripts were calculated using correlation coefficient formulae generated from the standard curve in each run with a length correction of 700-800 bp actual transcripts compared to 100-150 bp amplicons. Relative amounts of the transcripts of eight pectate lyase genes were calculated as (transcripts PL / transcripts of Ef1-α) x 100. The elongation factor 1-α (Ef1-α) was used as a housekeeping gene to normalize transcript amounts of pectate lyase genes because it was the most consistently expressed under all conditions tested based on previous gene expression profile studies during the parasitic and saprophytic growth of wild-type A. brassicicola [34,36,50,51].

Expression of PL1332 in Escherichia coli
An open reading frame of PL1332 was amplified from the cDNA with primers PL1332F_BamHI (AAggatccTTCACTGCCTTGACCATTACCG) and PL1332R3_HindIII (CCaagcttCATTGTGCTTTCCGTGGAGT), digested with BamHI and HindIII and cloned in a pMAL-c2x plasmid (NEB, Ipswich, MA). The plasmid was transformed into E. coli and selected transformants in the presence of ampicillin. Plasmids purified from 12 selected colonies of transformants were purified and their enzyme digestion patterns examined. Further, the nucleotide sequence of plasmids isolated from three colonies was determined using the M13 forward primer (GTAAAACGACGGCCAGT) to verify the presence of PL1332 genes and the intact continuous reading frame from the MBP. PL1332 protein produced from this plasmid was translated as a fusion protein from the start codon of maltose binding protein. PL1332 expression by the three transformants was tested and one was selected to produce the enzyme following the protocol in Current Protocols in Molecular Biology (1994), with slight modification. A single colony was transferred from an LB (Luria-Bertani) agar plate to 10 ml of LB broth with ampicillin and incubated overnight at 30°C, and then 1 ml of the cultured inoculum was transferred into 100 ml of LB broth medium. To induce expression of the PL1332 protein, 0.3 mM IPTG was added to the 100-ml culture and incubated for an additional 4-5 hours at 30°C with continuous agitation at 200 rpm. The bacterial cells were harvested and stored at -80°C until use. The stored cells were thawed and resuspended in 30 ml of bind and wash buffer (20 mM Tris-HCl, pH7.4, 200 mM NaCl, 1 mM EDTA, 10 mM β-mercaptoethanol) with one tablet of protease inhibitor (Roche, Basel, Switzeland), 300 μl of phosphatase inhibitor and 1% 4-Nitrophenyl phosphaste di(tri) salt. The cells were disrupted for about 3 minutes in ice-water with 30-second intervals of pulse and pause and 35 amplitudes of an ultrasonic liquid processor (Misonix model: S-4000-010, Newtown, CT). Proteins were further purified with MBP-binding agarose resin following manufacturer's protocol with slight modification (Elpis Biotech, Daejeon, Korea). A total of 30 ml of supernatant was combined with the prewashed amylose resin and incubated overnight at 4°C for binding. The protein was further washed and eluted with 10 mM maltose in 500 μl bind and wash buffer. We also used a supernatant of whole bacterial lysates for the enzymatic assays instead of purified MBP-PL1332. In addition, the PL1332 cDNA was cloned in a pGEX-6P1 vector transformed into E. coli, which produced PL1332-Glutathione S-Transferase (GST) fusion protein. The whole lysate of the E. coli that produced the PL1332-GST fusion protein was also used for further biological analyses. In these experiments, the supernatant of the whole lysates of the E. coli transformed with an empty vector was used as a negative control.

Enzymatic assays
The enzyme activity of pectate lyase was measured by a titrimetric stop reaction method, following a previously described protocol with appropriate modifications [52]. A solution of 5% (w/v) polygalacturonic acid (Cat# P3889, Sigma-Aldrich, St. Louis, MO) at pH 4.0 was mixed with either PL1332 fusion proteins or control proteins to a total volume of 5 ml and incubated at 25°C for 5 minutes. Then 5 ml of 100 mM I 2 solution and 1 ml of 106 mg/ml Na 2 CO 3 were added to the reaction mixture and incubated in the dark for 20 minutes. The mixture was then acidified by adding 2 ml of 2.0 N H 2 SO 4 . The free iodine was titrated with continuous stirring against 100 mM Na 2 S 2 O 3 using 1.0% (w/v) starch as an indicator. We calculated relative amounts of the titrant that measures free iodines that were not covalently bound to oligogalacturonic acids. Enzyme units were calculated using the formula, Units/ml = [(milliliters of titrant for blank-milliliters of titrant for test) x dilution factor x 100] / (0.1 x 5 x 2), and Units/μg protein = (Units/ml enzyme)/(μg protein/ ml enzyme). To visualize the relative amounts of free iodine at the end of the pectate lyase reactions, we added equal amounts of Na 2 S 2 O 3 to each reaction to sequester the same amount of free iodine. Finally, we added starch to visualize residual iodine.

Test of necrosis-inducing activity of PL1332 protein
The necrosis-inducing activity of PL1332 was examined with the PL1332 fusion proteins. The proteins were resolved in protein-elution buffer (10 mM maltose, 20 mM Tris-HCl, pH 7.4, 200 mM NaCl, 1 mM EDTA, 10 mM β-mercaptoethanol) or 10 mM Tris-HCl buffer and injected into young leaves of B. oleracea, B. juncea, or B. campestris var. chinensis using syringes with 26G x 13 mm needles. The intercellular space became waterlogged with protein solution and the soaked tissue remained visible for several hours. After infiltration, the leaves were maintained in mini humidity chambers and the development of necrotic tissue was observed for up to one week. Infiltration experiments were conducted more than three times and the progress of tissue damage was recorded daily with a digital camera.