Functional characterization of unique enzymes in Xanthomonas euvesicatoria related to degradation of arabinofurano-oligosaccharides on hydroxyproline-rich glycoproteins

In this study, we clarified the functions of three uncharacterized enzymes, XCV2724, XCV2728, and XCV2729, in Xanthomonas euvesicatoria, the causal agent of bacterial spot of tomato and pepper. The genes corresponding to the three enzymes are homologs of hypBA1, hypBA2, and hypAA from Bifidobacterium longum and are unique to Xanthomonas spp. among plant pathogenic bacteria. Functional characterization of the recombinant enzymes expressed using microbial systems revealed that they degrade the arabinofurano-oligosaccharides present on hydroxyproline (Hyp)-rich glycoproteins (HRGPs) such as extensin and solanaceous lectins in plant cell walls. These enzymes work coordinately to degrade the oligosaccharides. First, XeHypAA (XCV2728), belonging to the glycoside hydrolase (GH) 43 family, releases L-arabinose from L-arabinofuranose (Araf)-α1,3-Araf-ß1,2-Araf-ß1,2-Araf-ß-Hyp (Ara4-Hyp), cleaving its α1,3 bond; second, XeHypBA2 (XCV2729), belonging to the GH121 family, releases the disaccharide Araf-ß1,2-Araf from Araf-ß1,2-Araf-ß1,2-Araf-ß-Hyp (Ara3-Hyp); finally, XeHypBA1 (XCV2724), belonging to GH family 127, releases L-arabinose from Araf-ß-Hyp (Ara-Hyp). In summary, the main oligosaccharide structure of Ara4-Hyp on the HRGPs is degraded to Ara3-Hyp, then to Ara-Hyp, and finally to Ara monosaccharides by the action of these three enzymes. HRGPs containing oligosaccharide substrates have been reported to contribute to plant defense, and interestingly, the promoter region of the operon (xehypBA2 and xehypAA) contains the plant-inducible promoter box for binding the regulator protein HrpX involved in pathogenicity. We then analyzed the expression level of the operon gene in hrp-inducing medium and in plants and constructed gene-deletion mutants. However, although the operon was evidently upregulated by HrpX, three single-gene deletion mutants (ΔxehypBA1, ΔxehypBA2, ΔxehypAA) and even a triple-gene deletion mutant (ΔxehypBA1-BA2-AA) remained pathogenic, and had no effect on nonhost resistance, either, indicating that these three enzymes are not involved in either pathogenicity or nonhost resistance reactions. This is the first report of enzymes in plant pathogenic bacteria that catalyze the degradation of Hyp-linked-L-arabinofuranosides in plant cell walls.

Fujita et al. [1,2] searched the Pfam database for homologs of the three novel genes encoding HypBA1, HypBA2 and HypAA from B. longum, and interestingly, homologs were found only in the genomes of Xanthomonas spp. among plant pathogenic bacteria. Our interest was piqued because the enzymes from B. longum work on sugar chains of HRGPs that contribute to plant defense [1,2]. In addition, xcv2729 from X. euvesicatoria, corresponding to the homolog of hypBA2, is expressed inductively by HrpG and HrpX [17], two upstream regulators of hrp genes encoding the type III secretion system (a membrane-embedded nanomachine) that is essential for pathogenicity in Xanthomonas spp. [18,19]. Therefore, to investigate whether these enzymes from Xanthomonas spp. have the same functions as those of B. longum and are involved in pathogenicity in this study, we cloned the homologous genes and characterized the recombinant enzymes from X. euvesicatoria (formerly X. campestris pv. vesicatoria), which causes bacterial leaf spot [20,21]. This bacterium infects solanaceous plants, such as tomato and pepper, that contain both extensin and solanaceous lectins. This is the first report of enzymes that catalyze the degradation of Hyp-linked-L-arabinofuranosides in plant pathogenic bacteria. We also discuss why X. euvesicatoria may have these three unique enzymes.

Bacterial strains, growth conditions and plasmids
The bacterial strains and plasmids used in this study are listed in Table 1. X. euvesicatoria UPB139 corresponds to strain 85-10 in the KEGG database (T00288) [20][21][22] and was grown at 28˚C using complex nutrient-yeast-glycerol medium (NYG) [23] or hrp-inducing medium (XVM2) that provides an environment similar to the plant extracellular space [24]. Escherichia

Construction of protein expression vectors
The genomic DNA of X. euvesicatoria UPB139 was extracted using a NucleoSpin Microbial DNA kit (Takara Bio, Otsu, Japan) and used for further PCR amplification. The primers used in this work are all shown in S1 Table. A fragment of xcv2724, encoding amino acids (aa) 44 to 791, was amplified with primer set PEX1/PEX2 to eliminate the N-terminal signal peptide (Nsp). The amplified fragment was cloned into pET-23b digested with NdeI and XhoI using an In-Fusion HD Cloning kit (Takara Bio), yielding pET23-xcv2724. A fragment of xcv2728, encoding aa 25 to 528, was amplified with primer set PEX5/PEX6 without N-sp. The amplified fragment was cloned into the linearized pBIC3 using the Brevibacillus in vivo cloning (BIC) method [28,29]; briefly, a mixture of the PCR product and the linearized plasmid is directly transferred into B. choshinensis competent cells in which the insert and the plasmid are spontaneously combined via homologous recombination. Thus, the expression vector of xcv2728 was constructed without using E. coli, resulting in pBIC3-xcv2728. A fragment of xcv2729,

Expression and purification of recombinant protein
For the expression of XeHypBA1, vector pET23-xcv2724 was transferred into E. coli BL21 (λDE3) cells, then a single colony was added to 50 mL LB containing 50 mg/mL kanamycin and grown to an OD 600 of 0.6 at 37˚C. The culture was then induced with 1 mM isopropyl ß-

Enzymatic assays
A 10-μl reaction mixture for thin-layer chromatography (TLC) analysis of dansylated substrates contained 50 mM sodium acetate buffer (pH 4.5), 50 μM substrate, and 1 μL of the expressed recombinant enzyme. After 12 h at 30˚C, the reaction mixtures were spotted on a Silica Gel 60 aluminum plate (Merck, Darmstadt, Germany) and developed with a 3:1:1 solvent (v/v/v) of 1-butanol/acetic acid/water and finally visualized with UV light. For orcinol-stained TLC analysis, the 100-μL reaction mixture contained 50 mM sodium acetate buffer (pH 4.5), 35 μM substrate, and 2 μL of an expressed enzyme, and the reaction was conducted at 30˚C for 12 h. Spotted silica gels were developed with a 2:1:1 solvent (v/v/v) of ethyl acetate/acetic acid/ water. Sugars were visualized by spraying an orcinol-sulfate reagent onto the silica gel plate [31]. For high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD) analysis, oligosaccharides in the 100-μL reaction mixture were analyzed with a CarboPac PA-1 column. The column was eluted at a flow rate of 1.0 mL/min with the following gradient: 0-5 min, 100% eluent A (0.1 M NaOH); 5-30 min, 0-100% eluent B (0.5 M sodium acetate and 0.1 M NaOH); and 30-35 min, 100% eluent B.

RNA expression analysis
Total RNA from X. euvesicatoria UPB139 was extracted using the Nucleospin RNA II (MACHERY-NAGEL, Duren, Germany) according to the manufacturer's instructions, except that the phenol-chloroform-isoamyl alcohol mixture was used only when inoculated plants were macerated in the first step. To confirm the absence of any genomic DNA contamination, extracted RNAs were directly subjected to PCR. DNA-free RNA was then converted to cDNA using the ReverTra Ace (Toyobo, Osaka, Japan) with random hexamers. Quantitative reversetranscription PCR (qRT-PCR) was conducted in a LightCycler Nano System (Roche Diagnostics, Rotkreuz, Switzerland) with TB Green Premix Ex Taq II (Takara Bio). Cycling conditions were initial denaturation for 2 min at 95˚C; 45 cycles of 95˚C for 15 s, 55˚C for 15 s, and 72˚C for 15 s. Relative levels of gene expression were calculated using the 2 -ΔΔCt method [32]. For each amplification run, the calculated threshold cycle for each gene amplification was normalized against that of the reference gene 16S rRNA. Three technical replicates were performed each time. Primers used in this analysis are shown in S1 Table. Generation of gene-deletion mutants X. euvesicatoria mutants were constructed via double homologous recombination using the suicide vector pK18mobSacB [27], which harbors the sacB gene as a counterselection marker. All primers used in this experiment are shown in S1 Table. Single-gene deletion mutants of xehypBA1, xehypAA, and xehypBA2 were generated as below. For the xehypBA1 deletion, the two primer pairs, DIS1/DIS2 containing the EcoRI site and DIS3/DIS4 containing the XbaI site were used to amplify the 492-bp upstream region and the 590-bp internal region, respectively. The upstream region was first cloned into the EcoRI site of pK18mobSacB via In-Fusion HD Cloning kit (Takara Bio), while the internal region was cloned into the XbaI site in the same manner, yielding pK18-xcv2724. Similarly, for the xehypAA deletion, primer pairs DIS13/DIS14 and DIS15/DIS16 were used to amplify the 584-bp and 510-bp fragments and cloned into pK18mobSacB, yielding pK18-xcv2728. For the xehypBA2 deletion, primer pairs, DIS7/DIS8 and DIS9/DIS10 were used to amplify the 517-bp and 591-bp fragments and cloned into pK18mobSacB, yielding pK18-xcv2729. For the xehypAA-xehypBA2 double deletion, primer pairs DIS7/DIS8 and DIS15/DIS16, described above were used to construct vector pK18-xcv2728-2729. For the xehypBA1-xehypAA-xehypBA2 triple deletion, the double deletion mutant was used as a recipient and the xehypBA1 gene was deleted in the manner described above. The hrpX-deletion mutant of X. euvesicatoria was created using primer pairs DIS19/ DIS20 and DIS21/DIS22 to amplify the 500-bp and 419-bp fragments, which were then cloned into pK18mobSacB, yielding pK18-hrpX.
The deletion vectors were inserted into E. coli S17-1 (λpir) cells and then introduced into X. euvesicatoria by biparental conjugation. Crystal violet (0.3% w/v) was used to select a kanamycin-resistant isolate X. euvesicatoria, discriminating it from a kanamycin-resistant E. coli for the first screening. A marker-exchanged mutant was obtained by double homologous recombination using sucrose selection (SacB) as reported previously [33]. Deletion mutants were confirmed by PCR and sequencing.

Inoculation tests
Micro-Tom plants [34] were grown from seeds in a plant growth chamber at 28˚C with 16 h light/8 h dark. Plants 4-5 weeks old were dipped into a suspension of bacteria (OD 600 of 0.1) containing 0.02% (v/v) surfactant Silwet L-77 and 10 mM MgCl 2 . Plants were then covered with plastic bags for 48 h to maintain a moist environment, grown for 10 more days in the chamber, and checked for symptoms. Leaves of other 4-5-week-old plants were infiltrated with a 100-fold dilution of a bacterial suspension grown to an OD 600 of 0.1.
For assessing nonhost responses, tobacco plants (Nicotiana tabacum) were grown in a greenhouse up to the 4-5 leaf stage and transferred to a growth chamber at 25˚C under a 16-h light/8-dark 3 days before inoculation. Leaves were infiltrated with a bacterial suspension (OD 600 of 0.4) in 10 mM MgCl 2 using a syringe. Inoculated plants were then grown in the chamber for another 3 days and checked for symptoms.

Expression and purification of XeHypBA1, XeHypBA2, and XeHypAA
XeHypBA1 consisted of 791 aa. The recombinant protein without the signal peptide (43 aa) was expressed in E. coli BL21 (λDE3) and obtained as a soluble protein. SDS-PAGE showed that the purified recombinant XeHypBA1 protein migrated as a single band with an estimated molecular mass of 82.8 kDa (Fig 2A).
XeHypBA2 consisted of 1452 aa. The recombinant protein without the signal peptide (42 aa) was expressed in E. coli BL21 (λDE3) as a fusion protein, in which the protein of interest was fused with the trigger factor chaperone because a soluble protein was not obtained without a trigger that induces protein solubility. SDS-PAGE showed one band with a slightly lower molecular mass than the expected size 200 kDa (XeHypBA2, 152 kDa; and trigger factor, 48 kDa; Fig 2B).
XeHypAA consisted of 528 aa, and the recombinant protein without the signal peptide (24 aa) was expressed in B. choshinensis SP3 as a secretory protein because we could not obtain a soluble protein in E. coli even with the trigger factor. The expression level of XeHypAA in B. choshinensis was low, and the purified protein was not visible in an SDS-PAGE gel stained with Coomassie brilliant blue. Thus, we used Western blotting with His-tag antibodies to confirm expression of the target protein (Fig 2C). Western blotting showed a His-tagged protein with an estimated molecular mass of 55.3 kDa (Fig 2C).

Substrate specificity of XeHypBA1, XeHypBA2, and XeHypAA
XeHypBA1 preferred Ara-Hyp as a substrate and liberated L-arabinose (Fig 3A and 3B) and was only slightly active on ß-Ara 2 (Fig 3C), but did not use Ara 4 -Hyp and Ara 3 -Hyp at all. Ara 2 -Hyp was completely degraded in its dansylated form, but only partially when unmodified (Fig 3A and 3B). These activities differ from those of HypBA1 from B. longum, which degrade ß-Ara 2 , Ara 3 -Hyp, Ara 2 -Hyp, and Ara-Hyp. These results indicate that XeHypBA1 mainly acts on Ara-Hyp.
We also examined the synergistic effects of the three enzymes on the degradation of Hyplinked arabino-oligosaccharides. Ara 4 -Hyp was completely degraded by combined activity of XeHypAA, XeHypBA2, and XeHypBA1 (Fig 4).
The substrate specificities of each enzyme are summarized in Table 2, and the coordinated degradation of the oligosaccharides is shown in Fig 5.

Expression analysis of xehypBA2-AA operon
The stop codon of xehypBA2 and the start codon of xehypAA are overlapped, typical of an operon gene. Thus, to confirm whether the two genes form an operon, we conducted RT-PCR using primers EXG1 and EXG2 (S1 Table), designed to amplify a 346-bp fragment containing the junction region of the two genes. Expected fragments were obtained from cDNAs derived from the bacteria in XVM2 (hrp-inducing medium) (Fig 6A), indicating that xehypBA2 and  Characterization of arabinofuranosidases from Xanthomonas euvesicatoria xehypAA are transcribed into a single mRNA. No PCR product was obtained from extracted RNAs, demonstrating the absence of genomic DNA contamination in the RNAs (Fig 6A).
In the promoter region of the xehypBA2-AA operon, a conserved cis-regulatory element, PIP box with the consensus sequence TTCGCN 15 -TTCGC [17,35] was found. Thus, to Characterization of arabinofuranosidases from Xanthomonas euvesicatoria investigate whether the operon was regulated by HrpX, we analyzed the expression of the operon gene over time using qRT-PCR; expression in XVM2 reached a maximum at 10 h and was significantly higher than in NYG (complete medium) (Fig 6B). Next, we compared the expression level of the wild type with that of the hrpX-deletion mutant in XVM2 after a 10-h incubation. The ΔhrpX mutant was generated as shown in S1 Fig. The expression level of the gene in the wild type was significantly higher than in the ΔhrpX mutant (Fig 6C), indicating that the gene was upregulated by HrpX. We also analyzed the expression of the gene in infected plants (Micro-Tom). The expression was much higher than in NYG at all times tested (15 h-5 days), and was highest at 30 h after inoculation (Fig 6D). Obviously, the expression of the operon gene was induced in infected plants.

Pathogenicity of mutants
Because the xehypBA2-AA operon was regulated by HrpX, we constructed a gene-deletion mutant of the operon gene, as well as that of xehypBA1 to investigate whether these genes are involved in pathogenicity. Single-gene deletion mutants of each gene (ΔxehypBA1, ΔxehypBA2, ΔxehypAA) and a triple-gene deletion mutant of the three genes (ΔxehypBA1-BA2-AA) were created through biparental mating. The triple-gene deletion mutant was constructed from the ΔxehypBA1 mutant. The obtained mutants were confirmed by the reduced size of the PCR products (Fig 7). Micro-Tom plants were inoculated with one of the various mutants using the dipping method (see Materials and methods), and plants were examined for symptoms 12 days later. Single-gene deletion mutants, and even the triple-gene deletion mutant, induced the same symptoms on leaves as on the wild type (Fig 8). The double-gene deletion mutant (Δxe-hypBA2-AA) also maintained pathogenicity (data not shown). The hrpX-deletion mutant (ΔhrpX) did not cause any symptoms (Fig 8). These results indicate that the three genes are not involved in the pathogenicity of X. euvesicatoria.
The influence of the respective gene-deletion mutants on nonhost responses was tested by inoculating N. tabacum plants. All the mutants except for ΔhrpX induced chlorotic reactions in the same manner as the wild type (Fig 9), indicating that the three genes have no effect on nonhost resistance.

Discussion
Because the genes in B. longum that encode the enzymes to degrade arabino-oligosaccharides on HRGPs are found only in Xanthomonas spp. among plant pathogenic bacteria [1,2], we cloned and characterized the homologous genes xehypBA1, xehypBA2, and xehypAA derived from X. euvesicatoria that infects tomato and pepper to better understand the role of these enzymes in pathogenicity of the bacterium.
The gene xehypBA1 is the homolog of hypBA1 from B. longum. HypBA1, assigned to the new GH family 127, is a novel ß-L-arabinofuranosidase that liberates L-arabinose from ß-Ara 2 , Ara-Hyp, Ara 2 -Hyp, and Ara 3 -Hyp [1]. However, XeHypBA1 liberated L-arabinose specifically from Ara-Hyp and slightly from ß-Ara 2 and Ara 2 -Hyp; the enzyme likely works better on Ara 2 -Hyp-DNS than Ara 2 -Hyp (unmodified form) because the conformation of Ara 2 -Hyp-DNS may provide a better fit with the enzyme than that of Ara 2 -Hyp (Fig 3A, 3B and 3C), but the cause is unclear. HypBA1 from B. longum is an intracellular protein; B. longum takes up ß-Ara 2 and degrades the disaccharide in the cells. XeHpyBA1, on the other hand, is an extracellular protein and can work directly on the arabino-oligosaccharides on HRGPs (data not shown). These results indicate that XeHpyBA1 is quite different from HypBA1. The gene xehypBA2 is the homolog of hypBA2 from B. longum. Because the recombinant protein of XeHypBA2 was not expressed in the pET system as a soluble protein, we used the pCold TF system that expresses a fusion protein with the trigger factor chaperon as a soluble protein. In the SDS-PAGE analysis, the fusion protein migrated at a slightly lower molecular mass than expected (Fig 2B), probably because the fusion protein was not completely unfolded by the SDS even after extensive heating due to the complexity of the protein structure. HypBA2 from B. longum, assigned to the new HG family 121, is a novel ß-L-arabinobiosidase that has strict substrate specificity for Ara 3 -Hyp [2]. It is an extracellular protein with a membraneanchoring region (MAR) at its C-terminal region, suggesting that the enzyme may be on the bacterial surface and that it liberates ß-Ara 2 from HRGPs adjacent to the cells. XeHypBA2 also liberates ß-Ara 2 only from Ara 3 -Hyp, not from Ara 2 -Hyp and Ara 4 -Hyp, so it also is highly specific for the structure of Ara 3 -Hyp (Fig 3D and 3E). It is also an extracellular protein, but it has no MAR, indicating that the secreted enzyme freely degrades Ara 3 -Hyp on HRGPs.
The gene xehypAA is the homolog of hypAA from B. longum. The recombinant protein of XeHypAA was not expressed either in the pET system or even in the pCold TF system. We then used the Brevibacillus expression system, which is well-suited for heterologous protein expression. HypAA from B. longum, containing a GH43 domain, is an α1,3-specific α-L-arabinofuranosidase, which recognizes the Araf-α1,3-Araf structure of Ara 4 -Hyp [unpublished data]. XeHypAA also specifically degraded Ara 4 -Hyp to Ara 3 -Hyp and L-arabinose, recognizing the α1,3 bond of arabinose that is only on Ara 4 -Hyp (Fig 3F). HypAA is a secretory enzyme with MAR at its C-terminal region like HypBA2, but XeHypAA has no MAR. XeHypAA also may freely degrade Ara 4 -Hyp on HRGPs.
When we examined the synergistic effects among various combinations of the three enzymes on the degradation of Ara 4 -Hyp (Fig 4), they coordinately degraded Ara 4 -Hyp ( Fig  5). Ara 4 -Hyp and Ara 3 -Hyp are the major Hyp-linked L-arabinofuranosides in dicotyledons. In particular, Ara 4 -Hyp accounts for 33-75% of the total Hyp residues in plant cell walls [16]. Because XeHypBA2 cannot directly hydrolyze Ara 4 -Hyp, XeHypAA is required for further degradation. Thus, the operon construction of the two genes, xehypBA2 and xehypAA, is quite reasonable.
With regard to the involvement of HRGPs of extensin and lectins in plant defense [3][4][5][6][7][8][9], Brown et al. [36] reported that, in the pepper-X. campestris interaction, restriction of bacterial colony development was linked to the formation of an amorphous papillae-like matrix containing HRGPs around bacterial cells. Because the enzymes from X. euvesicatoria can remove the arabinofurano-oligosaccharides on HRGPs, we expected that deglycosylated HRGPs would be unstable and degradable by proteases, leading to a reduction in plant resistance. As a matter of fact, in the promoter region of the operon gene (xehypBA2-AA), there is a PIP box, which binds with the global-regulator HrpX to regulate pathogenicity-related genes. In the qRT-PCR for the operon gene using a hrpX-deletion mutant, the gene was evidently regulated by HrpX (Fig 6C). Koebnik et al. [17] also reported that expression of xcv2729 (xehypBA2) was dependent on HrpG and HrpX. Furthermore, the gene was upregulated in inoculated plants (Fig 6D). In the qRT-PCR of xehypBA1, surprisingly, the gene was also upregulated in infected plants, even though there were no PIP box-like sequences in the promoter region (data not shown). These data strongly suggest that the genes are involved in pathogenicity. Thus, we created in-frame deletion mutants for each enzyme gene. However, in contrast to expectations, the single-gene deletion mutants and even the triple-gene mutant remained pathogenic (Fig  8). We also investigated whether the mutants exert an influence on the resistance reactions of a nonhost plant because of the possibility that the monosaccharides (Ara) or disaccharides (ß-Ara 2 ) freed by the enzymes might act as an elicitor. However, the inoculated tobacco plants (N. tabacum) developed chlorosis (not fast cell death) in the areas surrounded by veins (Fig 9), typical resistance reactions of N. tabacum against X. euvesicatoria [37]. These results indicate that the enzymes are not involved in either pathogenicity or nonhost resistance reactions.
B. longum uses these enzymes to degrade the arabinofurano-oligosaccharides on HRGPs that reach the intestine and thus frees L-arabinose for use as their carbohydrate source [1]. X. euvesicatoria cannot utilize L-arabinose as a carbon source [38], and as described above, the enzymes are not involved in pathogenicity. So why does X. euvesicatoria have these enzymes? Besides their presence in X. euvesicatoria, homologs of hypBA2 and hypAA are conserved in only two species of Bifidobacterium (B. longum and B. pseudocatenulatum), but not in other intestinal bacteria such as Bacteroides, Salmonella, Clostridium, and Escherichia [2]. The homologs are also found in some actinomycetes such as Streptosporangium roseum, Actinosynnema mirum, and Micromonospora aurantiaca [2]. On the other hand, in Xanthomonas spp., homologs are conserved in 12 of 13 species deposited in public databases but not in other plant pathogenic bacteria such as Pseudomonas, Erwinia, Pectobacterium, and Burkholderia. In sum, the genes are distributed among some species in several genera of bacteria, but are well conserved in the species of Xanthomonas. Therefore, these genes might have been transferred horizontally from Xanthomonas to bifidobacteria or actinomycetes. If Xanthomonas spp. had acquired these genes originally, what role did they originally play? Initially, these enzymes may have contributed to pathogenicity, but plants may have overcome it, and in response the bacteria evolved other pathogenicity factors to cause disease. In other words, these enzymes may now be useless for Xanthomonas spp. We also deleted the homologous genes from X. campestris pv. campestris that infects cruciferous plants, and the deletion did not alter its pathogenicity (data not shown), just as we found for X. euvesicatoria. Recently, the full genome sequence of X. phaseoli that infects common bean, Phaseolus vulgaris has become available, and the homolog of xehypBA2 in X. phaseoli was annotated as a pseudogene (XppCFBP6546P_19765) [39]. A putative peptidase gene located adjacent to xehypAA from X. euvesicatoria is predicted to encode a prolyl oligopeptidase to hydrolyze proline-containing peptides (Fig 7A). The enzyme may be able to degrade HRGPs rich in hydroxyproline. However, a frameshift mutation has been caused by a single nucleotide deletion, creating a stop codon in the internal region, indicating that the gene has also become a pseudogene. These facts indicate that the genes encoding enzymes related to degradation of arabino-oligosaccharides on HRGPs may be in the process of disappearing in Xanthomonas spp.

Author Contributions
Data curation: Masayuki Nakamura.