Deletion of TFP-related genes differentially affects natural competence and movement of X. fastidiosa

Individual deletion mutant strains were generated by site-directed mutagenesis via natural transformation to determine the functional roles of 38 TFP-related genes (S1 Table) [40]. Genes were chosen according to genome annotation and searches based on homology to counterparts encoded by Pseudomonas aeruginosa [9, 54]. To the best of our knowledge, these genes represent the entire set of TFP genes encoded by X. fastidiosa strain TemeculaL. Assessment of natural competence was conducted using the pAX1-Cm plasmid, which recombines into the neutral site 1 of the X. fastidiosa genome and inserts a chloramphenicol (Cm) resistance cassette [39, 50], whereas twitching motility was determined by measuring colony fringe widths on solid agar plates [25]. Although the knockout of genes studied here did not always modulate natural competence and twitching motility in the same manner, the correlation between these two traits among mutant strains was positive (R² = 0.56, P<0.001; S3 Table), indicating that they were usually increased or reduced in a similar manner upon deletion of TFP genes.

A set of ten genes were considered core TFP components as deletion of such genes abrogated both natural competence and movement of X. fastidiosa (Figs 1 and S1–S3). They include the ATPases pilB and pilT, the prepilin peptidase pilD, the TFP assembly platform pilC, the whole alignment subcomplex pilMNOP, the secretin pilQ and the regulatory gene pilZ. Deletion of the sensor gene pilS of the PilRS two-component regulatory system (Figs 1 and S1–S3) significantly increased both natural competence and twitching motility, while deletion of the regulator pilR showed that this gene is essential for twitching motility but not for transformation. The minor pilin fimT3 was the only gene found to be essential exclusively for natural competence (Figs 1 and S1). Curiously, this Δfimt3 strain showed significantly increased twitching motility than wild-type (WT) cells (Figs 1 and S2 and S3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Natural transformation and twitching motility phenotypes of X. fastidiosa mutant strains used in this study.

Quantification of natural transformation was performed by enumerating total viable and recombinant culturable cells and results are expressed as the ratio of recipient cells transformed (recombination frequency; values shown in the x-axis of the chart). Twitching motility was determined by spotting cells on agar plates and measuring the movement fringe width after 4 days of growth at 28°C (values shown in the y-axis of the chart). Results of mutant strains for paralogous genes are shown using the same color for text and symbol. The non-motile and non-recombinant mutants are shown as a black dot. The WT is represented as a dark blue triangle and all other mutants are represented as black triangles. Data represent means and standard errors. Different letters in parenthesis indicate significant difference in fringe width as analyzed by ANOVA followed by Tukey’s HSD multiple comparisons of means (P<0.05; n = three to 14 independent replicates with eight to 48 internal replicates each). * and ** indicate significant difference (P<0.05 and P<0.005, respectively) of recombination frequency through natural competence in comparison to the WT as determined using Student’s t-test (n = three to 21 independent replicates with two internal replicates each). The detection limit for recombination frequency was 10⁻⁷ and for twitching motility was 10 μm. Mutants below the detection limit are shown in the non-motile and non-recombinant group.

https://doi.org/10.1371/journal.ppat.1011154.g001

Seven sets of paralogs exist in the X. fastidiosa TFP cluster (PilA1-3, PilE1-2, PilV1-2, PilW1-2, PilX1-2, PilY1-1-3 and FimT1-3). Intriguingly, the knockout of paralogs presented varied (mostly opposing) phenotypes suggesting duplication and neofunctionalization of the genes. The minor pilins pilE1V1W1X1, which are encoded in the same operon (S2 Table), were essential for twitching motility but not for transformation (Figs 1 and S2 and S3), although these mutants did show significantly lower recombination frequencies than WT (Figs 1 and S1). In comparison to the WT, ΔpilE2 presented significant higher twitching motility and no changes in natural transformation, ΔpilV2 had significant higher natural transformation with no altered movement, ΔpilW2 showed significant higher natural transformation and significant lower movement, and ΔpilX2 possessed both significant lower natural transformation and twitching motility (Figs 1 and S1–S3). Knockout of pilA1 significantly increased both natural competence and twitching motility, whereas deletion of pilA2 and pilA3 significantly reduced twitching motility and did not alter natural transformation (Figs 1 and S1–S3). A strain with deletion of both pilA1 and pilA2 abolished twitching motility (Figs 1 and S2–S3), as described elsewhere [40]. Additionally, deletion of the minor pilin fimT1 significantly decreased both natural competence and movement, while knockout of fimT2 significantly increased natural competence and did not change twitching motility, and ΔfimT3 lost natural competence and presented significantly higher movement than the WT (Figs 1 and S1–S3). As for pilY1-1, pilY1-2 and pilY1-3, which are tip adhesins, their deletion significantly reduced twitching motility. However, ΔpilY1-1 and ΔpilY1-2 presented significant lower natural competence whilst ΔpilY1-3 had significant higher natural competence than the WT (Figs 1 and S1–S3).

Lastly, deletion of the pilotin pilF and of the retraction ATPase pilU significantly reduced natural competence and movement. Knockout of the chemotaxis genes pilGHIJL and chpBC significantly reduced twitching. However, ΔpilG, ΔpilH and ΔpilL showed increased natural competence, while ΔpilJ and ΔchpC had lower natural competence, and ΔpilI and ΔchpB did not change this phenotype in comparison to the WT (Figs 1 and S1–S3).

TFP-related genes differentially modulate other virulence traits of X. fastidiosa

We evaluated phenotypes such as growth, biofilm formation, cell aggregation (measured by settling rate), and virulence in planta of the individual TFP mutants. Growth rates were measured by change in OD_600nm over time (S4 and S5 Figs). Since this growth analysis considered only turbidity that may be affected by cell attachment (biofilm) to surfaces and cell-to-cell aggregation, the number of culturable X. fastidiosa cells (measured as CFU/ml) was also evaluated. Only the non-recombinant and non-motile ΔpilC and ΔpilP mutants showed significant changes in the number of cells in comparison to WT, with both having significantly higher populations (S6 Fig). Biofilm formation is a critical aspect of virulence in X. fastidiosa. Individual deletion of pilB, pilD, pilH, pilI, pilM, pilN, pilO, pilQ, pilR, pilT, pilX1 and fimT1 significantly increased biofilm formation, whereas ΔpilA3 and ΔfimT3 formed significant less biofilm than WT (S7 Fig). Planktonic growth was significantly increased for strains with individual deletion of pilA1, pilC, pilE1, pilF, pilM, pilN, pilO, pilP, pilQ, pilR, pilV1, pilW1, pilW2, pilX1, pilA1pilA2, pilY1-1, fimT3 and chpC, while individual knockout of pilD, pilT and pilY1-3 significantly decreased planktonic growth (S8 Fig). Moreover, only ΔpilD, ΔpilT and ΔpilX2 showed significant higher settling rates, with other mutants not differing significantly from the WT (S9 Fig). Biofilm formation and settling rate, as well as planktonic growth and growth rate were positively correlated (S3 Table). Meanwhile, biofilm formation and twitching motility/growth rate, settling rate and planktonic growth/growth rate, and growth rate and total viable X. fastidiosa had significant negative correlations (S3 Table). No other phenotype presented a significant correlation to recombination frequency besides twitching motility.

Virulence of a subset of the X. fastidiosa mutant strains was assessed by inoculating Nicotiana tabacum L. cv. Petite Havana SR1 plants and scoring disease incidence and severity. Mutants chosen for analysis had different twitching motility phenotypes, such as higher movement (ΔpilA1 and ΔpilS), lower movement (ΔpilA2) and non-motility (ΔpilA1pilA2, ΔpilQ and ΔpilR). Additionally, mutants in the DNA-binding fimT3 (see below) and the paralogs fimT1 and fimT2 were tested and no significant change in virulence was found compared to WT. WT-inoculated plants presented the highest disease severity over the time course of assays (Fig 2A), with all plants inoculated with the mutant strains having significantly lower AUDPC (Area Under the Disease Progress Curve) values, except plants inoculated with ΔpilA2 (Fig 2B). All inoculated plants reached 100% disease incidence at the end of assays, except plants inoculated with ΔpilA1, which had 89% disease incidence (±11%, standard error). However, only WT-inoculated plants consistently presented severe symptoms of leaf scorch, while plants inoculated with the mutant strains mostly presented mild symptoms (Fig 2D). In addition, when evaluating the population of X. fastidiosa in planta by qPCR, only the ΔpilA1pilA2 double mutant showed a defect in colonizing the top leaf of the plant host, and the ΔpilQ and ΔpilR mutants presented significant higher population in the basal leaf of plant hosts (Fig 2C). Together, results demonstrate that the deletion of analyzed genes significantly reduced symptoms development in infected tobacco plants (except for deletion of pilA2), which was not explained by bacterial distribution, since most strains could effectively colonize the entirety of plants (except ΔpilA1pilA2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Deletion of pil genes negatively affects virulence of X. fastidiosa.

a Disease severity progression over time in inoculated tobacco plants. X. fastidiosa WT and mutant strains were inoculated into Nicotiana tabacum L. cv. Petite Havana SR1 plants (PBS mock inoculation used as control). Mutant strains chosen for inoculation presented different twitching motility phenotypes, including higher twitching motility (ΔpilA1 and ΔpilS), lower twitching motility (ΔpilA2) and non-motility (ΔpilA1pilA2, ΔpilQ and ΔpilR). Leaf scorch symptoms were recorded for measurements of disease incidence and severity once a week during nine weeks after appearance of the first disease symptoms. At the final time point of evaluation, disease incidence in all inoculated plants reached 100%, except for plants inoculated with ΔpilA1, which reached 88.88% disease incidence (±11.11%, standard error). On the other hand, disease severity reached 96% in the WT, 35% in ΔpilA1, 86% in ΔpilA2, 57% in ΔpilA1pilA2, 73% in ΔpilQ, 72% in ΔpilR and 56% in ΔpilS. Data represent means and standard errors from two independent experiments (n = seven to ten plants in each independent experiment). b Mean AUDPC per treatment group. AUDPC was calculated using data from disease severity over nine weeks after first disease symptom appearance. WT is highlighted in blue, and the dashed blue line indicates the mean value of AUDPC for WT-inoculated plants. AUDPC was significantly lower for plants inoculated with all mutant strains in comparison to WT-inoculated plants, except for ΔpilA2. Data represent means and standard errors. Different letters on top of bars indicate significant difference as analyzed by ANOVA followed by Tukey’s HSD multiple comparisons of means (P<0.05; n = two independent experiments with seven to ten plants each). c X. fastidiosa in planta population determined by qPCR at the last time point of evaluation. The population of WT and mutant strains throughout inoculated plants was calculated using petioles of basal and top leaves. ΔpilA1pilA2 cells showed a defect in colonizing the top leaf of plant hosts, while ΔpilQ and ΔpilR showed a significant higher population in the basal leaf of infected plants. This shows that mutant strains still move inside the xylem of plant hosts, but absence of deleted genes impairs full symptom development. Data represent means and standard errors. Different letters on top of bars indicate significant difference as analyzed by ANOVA followed by Tukey’s HSD multiple comparisons of means (P<0.05; n = two independent experiments with three leaf replicates each). d Figure panel with representative pictures of leaf scorch symptoms in WT- and mutant strains-inoculated plants (top and basal leaves), as well as control plants (PBS mock inoculation). Only WT-inoculated plants consistently presented severe leaf scorch symptoms. Similar events were captured in two independent experiments.

https://doi.org/10.1371/journal.ppat.1011154.g002

The minor pilin FimT3 is the DNA receptor of the X. fastidiosa type IV pilus

The observation of the minor pilin FimT3 being the only TFP component that is essential exclusively for natural transformation (Figs 1 and S1–S3) led us to hypothesize that this component is the DNA receptor of the TFP of X. fastidiosa. We first analyzed the piliation phenotype of ΔfimT3 in comparison to WT by transmission electron microscopy (TEM). Both presented long TFP as well as short type I pili, with no apparent changes in the piliation of X. fastidiosa cells upon deletion of fimT3 (Fig 3A). Next, the ability of these cells to take up DNA from the extracellular environment was evaluated by exposing them to Cy-3-labeled pAX1-Cm plasmid and observing under a fluorescence microscope after a DNase I treatment to remove extracellular DNA. Upon exposure to the fluorescently labeled DNA fragment, WT cells imported DNA allowing for protection from exogenous DNase I, but no fluorescent DNA foci were present in ΔfimT3 cells (Fig 3B). DNA foci were mostly observed at one pole of the cells, which coincides to the described phenotype of X. fastidiosa that forms TFP at only one of its poles [52]. The DNA-uptake ability of the mutant strains of other minor pilins that decreased the natural transformation ability of X. fastidiosa upon deletion was also evaluated for comparison. These included ΔpilE1, ΔpilV1, ΔpilW1, ΔpilX1, ΔpilX2, ΔfimT1 and ΔfimT2 (Figs 1 and S1). The ΔfimT2 mutant strain was included because this gene is a paralog of fimT3, even though its deletion increased the natural competence of X. fastidiosa (Figs 1 and S1). All analyzed mutant strains maintained the DNA-uptake (competence) phenotype (S10 Fig), but at different levels (S11 Fig). WT cells presented the highest percentage of cells with DNA foci, with ΔpilE1 having the lowest proportion of cells with DNA acquisition (S11 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. FimT3 DNA binding activity.

a Transmission electron microscopy micrographs of pilus formation by X. fastidiosa WT and ΔfimT3 cells. Arrows are pointing to TFP. Similar events were captured in two independent experiments. Images were captured at 31,500× magnification. Scale bars, 500 nm. b Uptake of Cy-3-labeled pAX1-Cm plasmid into a DNase I resistant state by X. fastidiosa cells observed using a fluorescence microscope. The images correspond to the bright field (left), Cy-3 channel (center) and merged images (right). In merged images, arrows are pointing to fluorescent DNA foci at the cell poles. Similar events were captured in three independent experiments. Images were captured at 100× magnification. Scale bars, 1.5 μm. c DNA-binding ability of purified FimT1s, FimT2s and FimT3s assessed by agarose EMSA. Similar events were captured in three independent experiments. d Surface electrostatics representation of FimT3 merged with its ribbon representation. The arginine amino acid residues (R160 and R162) that were mutated for functional studies are highlighted in the figure. The electrostatic potential (eV) color scale is shown in the figure. e DNA-binding ability of purified FimT3s and FimT3s-R160AR162A assessed by agarose EMSA. Similar events were captured in three independent experiments. f Titration of the DNA binding activity of FimT3s and FimT3s-R160AR162A by native acrylamide EMSA. Similar events were captured in two independent experiments. g Densitometry analysis of the DNA binding activity of FimT3s and FimT3s-R160AR162A in acrylamide EMSA. The fluorescent intensity of the shifted DNA bands of Cy-3-labeled Km resistance cassette presented in f was measured using ImageJ. The fluorescent intensity of shifted bands when treated only with 1, 2, 5, and 10 μM of each protein was measured, since they presented higher shifts in electrophoretic mobility within this protein concentration range. h Area under the fluorescence curve. The area under the fluorescence curves in g was calculated to quantify the DNA-binding affinity of FimT3s and FimT3s-R160AR162A. Statistical significance was determined using Student’s t-test (* indicates P<0.05 in comparison to the wild-type protein; n = two independent replicates). Multiple bands of the pAX1-Cm plasmid observed in c and e are due to the multiple forms of the plasmid after purification, with each form displaying different sizes in agarose gels.

https://doi.org/10.1371/journal.ppat.1011154.g003

To determine the DNA-binding ability of FimT3, and compare it to its paralogs FimT1 and FimT2, the soluble portion of these proteins (named FimT3s, FimT1s and FimT2s, respectively), which lacked the conserved N-terminal hydrophobic α-helices of pilins [27], were amplified and cloned for expression. These proteins were then purified and assayed for DNA binding by agarose electrophoretic mobility shift assay (EMSA). Results showed FimT3s was the only protein capable of binding DNA, since no shift was observed with FimT1s and FimT2s when using as much as 10 μM of each protein (Fig 3C). Moreover, FimT3s was able to bind to different DNA sequences, independent of homology to the X. fastidiosa TemeculaL genome, indicating that binding is not sequence-specific (S12 Fig). Collectively, fimT3 being the only gene essential exclusively for natural competence, the ability of FimT3 to bind to DNA and the loss of DNA acquisition from the extracellular environment by cells upon its deletion, demonstrate that this minor pilin is the DNA receptor of the X. fastidiosa TFP.

To further characterize FimT3, we searched for specific amino acid residues involved in DNA-binding. The binding of DNA to TFP via a minor pilin was demonstrated in ComP of Neisseria, ComZ of Thermus thermophilus, and VC0858 of Vibrio cholerae [12, 24, 61]. Amino acids alignments revealed that FimT3 and ComP share only 28% identity (64% similarity) over 20% of query sequence, while FimT3 and ComZ have no significant identity. Alignment of FimT3 with VC0858 of Vibrio cholerae showed 35% identity and 62% similarity over 18% of query sequence. In spite of the low overall homology the sequence alignments showed that an arginine residue at position 162 of FimT3 aligned with a lysine of ComP (K108) that was previously demonstrated to be essential for its DNA-binding ability [12] (S13A Fig). Additionally, an arginine residue at position 160 of FimT3 aligned with an arginine of VC0858 (R168) that was also important for the DNA-binding activity of the V. cholerae pilus [24] (S13B Fig). These two arginine residues of FimT3 (R160 and R162) were predicted in silico to have the highest probabilities of binding DNA among the residues of this protein (S4 Table). Moreover, alignment of FimT3 with a FimT ortholog that is essential for natural competence of Acinetobacter baylyi [46] (28% identity and 46% similarity over 91% of query sequence) showed alignment of R162 of FimT3 to an arginine of FimT from A. baylyi. Likewise, FimT3 shares 27% identity (48% similarity) over 80% of query sequence with a recently described DNA-binding FimT ortholog of Legionella pneumophila, in which the aligned R160 and R162 have been demonstrated as important to bind DNA [8] (S13C Fig).

Analysis of a structural model produced by the Phyre2 algorithm, showed that R160 and R162 of FimT3 are located within an electropositive surface and could form a pocket to bind DNA (Fig 3D). Thus, to assess the role of these arginine residues in the DNA-binding ability of FimT3, we generated a FimT3s variant by site-directed mutagenesis in which both arginine residues were replaced by alanine residues (FimT3s-R160AR162A). Agarose EMSAs revealed that WT-FimT3s had a higher affinity for DNA than FimT3s-R160AR162A based on the amount of protein needed to shift 50% of the input DNA (Fig 3E). Similar results were obtained when using the more sensitive acrylamide EMSA and the kanamycin (Km) resistance cassette sequence as target DNA (previously determined to be bound by FimT3s in S12 Fig). An abrupt shift was detected when using concentrations of FimT3s ranging between 1 and 5 μM, whereas a much more subtle and gradual shift was observed for samples treated with FimT3s-R160AR162A within the same concentration range (Fig 3F). We then performed a densitometry analysis of the shifted bands treated with concentrations ranging between 1 and 10 μM of each protein (Fig 3G) and calculated the area under the curve to quantify the DNA-binding affinity of FimT3s and FimT3s-R160AR162A. While the DNA-binding affinity of FimT3s-R160AR162A was lower than that of FimT3s (Fig 3H), the mutations did not completely abolish DNA-binding ability suggesting other residues are contributing to this activity. Nonetheless, these results show that R160 and R162 are important for DNA-binding activity of FimT3.

FimT3 is widely conserved within X. fastidiosa strains and, among plant pathogens, is encoded by other members of the Xanthomonadaceae family

Previously, DNA-binding homologues of FimT3 have been found in many representative species of γ-Proteobacteria, including the Xanthomonadaceae family to which X. fastidiosa belongs [8]. We performed searches using the Conditional Reciprocal Best BLAST (CRB-BLAST) [4], as well as phylogenetic analyses, to further explore the distribution of the fimT paralogs within this family. All three paralogs of fimT are encoded by X. fastidiosa, X. taiwanensis, many clade II Xanthomonas spp. [34] and few Pseudoxanthomonas spp. (Fig 4). Curiously, early branching clade I Xanthomonas spp. only encode fimT1 and fimT3 (Fig 4). Specifically, fimT3 was present in 1,416 out of 3,001 genomes of the Xanthomonadaceae (47%), the majority of them belonging to species within the genus Xanthomonas. In the fimT3 clade, X. fastidiosa strains grouped together, including X. taiwanensis, with a similar trend being observed for Xanthomonas spp. strains (Fig 4). However, clade I and clade II Xanthomonas spp. and other Xanthomonadaceae species clustered mostly separated (Fig 4). The alignment of the different FimT3 sequences encoded by X. fastidiosa strains showed a maximum of six nonconservative substitutions within strains belonging to the subspecies pauca, which encode FimT3 that are 13 amino acids shorter than other strains (S14A Fig). Alignment of representative FimT3 sequences among strains belonging to the Xanthomonadaceae family revealed that the arginine residues at positions 160 and 162 of FimT3 from X. fastidiosa strain TemeculaL were conserved in most orthologous sequences of this protein (S14B Fig). Additionally, R160 was always preceded by a glycine residue, constituting a highly conserved GRxR motif (S14B Fig), as seen elsewhere [8]. Although amino acid identities ranged from 95.60% to 100% within X. fastidiosa, and from 15.5% to 80% compared to other species, only 41 out of 1,416 FimT3 sequences encoded by the Xanthomonadaceae do not carry the GRxR motif (S5 Table). Overall, results indicate that FimT3 is encoded by representative plant pathogens of the Xanthomonadaceae and that R160 and R162 are highly conserved within FimT3 sequences.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Phylogenetic tree based on nucleotide sequences of the three fimT paralogs from Xanthomonadaceae bacterial strains with whole-genome sequences available.

The phylogenetic tree was built using the Maximum-likelihood method and visualized using FigTree. Branches with bootstrap values below 70% were collapsed, except where indicated. This was performed to keep conciseness of the figure while indicating the different clusters for fimT1, fimT2 and fimT3. X. fastidiosa is highlighted in blue within the tree. When known, clade I Xanthomonas spp. are highlighted in orange, and clade II Xanthomonas spp. are highlighted in purple. Bacterial species encoding all three fimT paralogs are indicated in bold, while those encoding two paralogs are shown in italics and those encoding only one paralog are underlined within the tree. The order of appearance of bacterial species in collapsed branches follows the original order of appearance within the tree. For representation purposes, branches are highlighted by color according to each fimT paralog. Brown: fimT1; Dark blue: fimT2; Red: fimT3. *** Indicates clusters within the fimT3 clade which contain few FimT3 sequences with no GRxR motif (see S5 Table and S14 Fig). Abbreviations are the following. Coralloluteibacterium: Cst–C. stylophorae. Luteimonas: Lab–L. abyssi; Lae–L. aestuarii; Laq–L. aquatica; Lar–L. arsenica; Lch–L. chenhongjianii; Lcu–L. cucumeris; Lde–L. deserti; Lfr–L. fraxinea; Lgi–L. gilva; Lgr–L. granuli; Lhu–L. huabeiensis; Llu–L. lumbrici; Lma–L. marina; Lme–L. mephitis; Lpd–L. padinae; Lpn–L. panaciterrae; Lsp–L. sp.; Lte–L. terrae; Lti–L. terricola; Lwe–L. wenzhouensis; Lyi–L. yindakuii. Lysobacter: Lys–L. sp. Pseudoxanthomonas: Pbe–P. beigongshangi; Pbr–P. broegbernensis; Pco–P. composti; Pda–P. daejeonensis; Pdo–P. dokdonensis; Pge–P. gei; Phe–P. helianthi; Pin–P. indica; Pja–P. japonensis; Pji–P. jiangjuensis; Pkl–P. kalamensis; Pko–P. kaohsiungensis; Pkr–P. koreensis; Pme–P. mexicana; Psc–P. sacheonensis; Psn–P. sangjuensis; Psp–P. sp.; Psx–P. spadix; Psu–P. suwonensis; Ptw–P. taiwanensis; Pwi–P. winnipegensis; Pwu–P. wuyuanensis; Pye–P. yeongjuensis. Rehaibacterium: Rte–R. terrae. Silanimonas: Sle–S. lenta. Stenotrophomonas: Sac–S. acidaminiphila; Sbe–S. bentonitica; Sch–S. chelatiphaga; Scy–S. cyclobalanopsidis; Sda–S. daejeonensis; Sge–S. geniculata; Sgi–S. ginsengisoli; Shu–S. humi; Sin–S. indicatrix; Skr–S. koreensis; Sla–S. lactitubi; Sma–S. maltophilia; Sne–S. nematodicola; Sni–S. nitritireducens; Spn–S. panacihumi; Spv–S. pavanii; Spi–S. pictorum; Srh–S. rhizophila; Sse–S. sepilia; Ssp–S. sp; Ste–S. terrae; Stu–S. tumulicola. Vulcaniibacterium: Vge–V. gelatinicum; Vte–V. tengchongense; Vth–V. thermophilum. Xanthomonas: Xab–X. albilineans; Xaf–X. alfalfae; Xar–X. arboricola; Xax–X. axonopodis; Xbr–X. bromi; Xcm–X. campestris; Xcn–X. cannabis; Xcs–X. cassavae; Xci–X. cissicola; Xct–X. citri; Xco–X. codiaei; Xcu–X. curcubitae; Xdy–X. dyei; Xer–X. euroxanthea; Xeu–X. euvesicatoria; Xfl–X. floridensis; Xfr–X. fragariae; Xho–X. hortorum; Xha–X. hyacinthi; Xhd–X. hydrangea; Xml–X. maliensis; Xms–X. massiliensis; Xme–X. melonis; Xna–X. nasturtii; Xor–X. oryzae; Xpe–X. perforans; Xpi–X. pisi; Xpo–X. populi; Xpr–X. prunicola; Xsa–X. sacchari; Xso–X. sontii; Xsp–X. sp.; Xth–X. theicola; Xtr–X. translucens; Xva–X. vasicola; Xve–X. vesicatoria. Xylella: Xtw–X. taiwanensis.

https://doi.org/10.1371/journal.ppat.1011154.g004

Bacterial strains, plasmids, and culture conditions

All strains and plasmids used in this study are listed in S6 Table. All X. fastidiosa mutant strains used here are derivatives of the X. fastidiosa subsp. fastidiosa strain TemeculaL [59], and were originated by site-directed mutagenesis of each gene of interest using natural competence as described elsewhere [40]. Primers used in this process are listed in S7 Table. X. fastidiosa was recovered from -80°C glycerol stocks and routinely cultured for seven days at 28°C on periwinkle wilt (PW) agar plates [19], modified by removing phenol red and using bovine serum albumin (1.8 g/l) (BSA; Gibco Life Sciences Technology), and sub-cultured onto fresh PW agar plates for another seven days at 28°C before use. X. fastidiosa mutant strains were grown similarly but using PW plates amended with Km. All assays were performed using the subcultured X. fastidiosa strains. Cells were suspended and cultured in PD3 broth [19] to perform phenotypic assays, while phosphate-buffered saline (PBS) was used to suspend cells in liquid for in planta assays. Bacterial suspensions were performed by thorough pipetting followed by agitation using a vortex to foster greater homogenization and minimize the effects of cell-cell adhesion. Whenever needed, the antibiotics Km and Cm were used at concentrations of 50 and 10 μg/ml, respectively. When used together with Cm, the Km concentration was reduced to 30 μg/ml. Luria-Bertani (LB) medium (BD Difco) was used to culture E. coli cells. When needed, Km, Cm and ampicillin (Amp) were added to LB at concentrations of 50, 35 and 100 μg/ml, respectively.

Phenotypic analyses of X. fastidiosa WT and mutant strains

Analyses of natural competence among strains [39], twitching motility [38], growth curve and growth rate [39], biofilm formation and planktonic growth [17, 39], settling rate [39], piliation observation by TEM [40], and virulence in planta using tobacco as model system [28] were performed as described elsewhere. All experiments had at least three biological replicates, unless otherwise stated.

DNA uptake assays

DNA uptake assays were performed as similarly described for the analysis of natural competence among X. fastidiosa strains (see SI Materials and Methods in S1 Text). Briefly, recipient cells were suspended in PD3 broth to OD_600nm = 0.6, spotted onto PD3 agar plates and grown for three days at 28°C. Then, 1 μg of fluorescently labeled pAX1-Cm plasmid (10-μl volume), labeled using the Label IT Nucleic Acid Labeling kit, Cy3 (Mirus Bio LLC), was added on top of cells, air-dried, and incubated at 28°C for another 24 h. After, cells were harvested in 150 μl of PD3 broth, and a 50 μl aliquot was treated with 10 units of DNase I (New England Biolabs) for 10 minutes at 37°C to degrade the remaining extracellular DNA, and cells were observed using a 100× oil immersion objective in a Nikon Eclipse Ti inverted microscope (Nikon). Image acquisition was performed using a Nikon DS-Q1 digital camera (Nikon) controlled by the NIS-Elements software version 3.0 (Nikon), which was also used to create merged fluorescent images. To detect Cy3, an excitation wavelength of 590 nm was used (tetramethyl rhodamine isothiocyanate; TRITC filter). The proportion of cells that acquired extracellular DNA was calculated as the percentage of cells with fluorescent DNA foci to cells without fluorescent DNA foci. Experiments were performed at least three times independently.

Protein cloning, expression, and purification

Cloning, expression, and purification of FimT1s, FimT2s and FimT3s were performed similarly as described elsewhere using the pHIS-Parallel1 plasmid (S6 Table) [71, 78]. Amino acid exchanges in FimT3s were performed by designing a pair of primers (S7 Table) that exchanged the arginine residues at positions 160 and 162 by alanine residues. This pair of primers was designed using the QuickChange Primer Design tool (Agilent Technologies, Inc.; https://www.agilent.com/store/primerDesignProgram.jsp). For amino acid exchanges, the whole pHIS-Parallel1-fimT3s construct was amplified via Pfu DNA polymerase (G-Biosciences) using a standard protocol from the manufacturer and this pair of primers in a S1000 thermal cycler (Bio-Rad). Expression and purification of FimT3s-R160AR162A were performed as described above (see SI Materials and Methods in S1 Text) [78]. PCR products were purified using the Gel/PCR DNA Fragments Extraction kit (IBI Scientific). Correct cloning and amino acid exchanges in all constructs were confirmed by Sanger sequencing (Sequetech Corporation) using the T7 promoter forward primer (S7 Table).

DNA binding assays

Agarose EMSAs were mainly performed to assess the DNA-binding ability of purified FimT1s, FimT2s and FimT3s. Briefly, 200 ng of DNA (usually pAX1-Cm plasmid) were incubated for 30 minutes at 28°C with increasing concentrations of purified proteins in 20 μL EMSA reaction buffer (50 mM Tris-HCl, pH 7.5; 50 mM NaCl; 200 mM KCl; 5 mM MgCl₂; 5 mM EDTA, pH 8.0; 5 mM DTT; 0.25 mg/ml BSA). Lysis buffer containing 250 μM imidazole, in which proteins were suspended, was included as blank control. After the incubation period, DNA was separated by electrophoresis on a 0.8% agarose gel containing GelRed nucleic acid gel stain (Biotium) in Tris-acetate-EDTA buffer (120 V for 40 minutes). Experiments were performed at least three times independently. On the other hand, native acrylamide EMSAs were used to perform titration of the DNA binding activity of FimT3s and FimT3s-R160AR162A. In summary, 60 ng of Cy-3 labeled Km resistance cassette were incubated for 30 minutes at 28°C with increasing concentrations of purified proteins in 20 μL EMSA reaction buffer and separated by electrophoresis on a 3.5% native acrylamide gel in Tris-borate-EDTA buffer (40 V for 4 hours). Native acrylamide gels were pre-run at 40 V for 30 minutes before being used. DNA samples were directly visualized using the ImageQuant LAS 4010 Imaging System (GE Healthcare), since the used DNA was fluorescently labeled with Cy3. Experiments were performed two times independently. Densitometry analysis of the shifted bands of Cy-3 labeled DNA in native acrylamide gels was performed by measuring the fluorescence intensity of these bands when treated with 1, 2, 5, and 10 μM of each protein using ImageJ [67]. Then, these values were used to calculate the area under the fluorescence curves, as performed to obtain the AUDPC (see SI Materials and Methods in S1 Text) [72], to quantify the DNA-binding affinity of FimT3s and FimT3s-R160AR162A by determining the progress of shifted bands in relation to each protein concentration.

Bioinformatic analyses

FimT3 modeling was performed using the Phyre2 web portal [42] and visualized via PyMOL version 2.4.0 (Schrödinger, LLC). The surface electrostatics of FimT3 was predicted using the APBS Electrostatics plugin from PyMOL. Sequence alignments of FimT3 with ComP, VC0858 and FimT from A. baylyi and L. pneumophila were performed by retrieving the respective sequences from NCBI, aligning through the T-Coffee Multiple Sequence Alignment Server [21] and visualizing using the BoxShade webserver (https://embnet.vital-it.ch/software/BOX_form.html). For FimT3 analyses within the Xanthomonadaceae family, DNA sequences of different FimT homologs from Xylella, Xanthomonas, Lysobacter and Stenotrophomonas were used as query for screening genomes of Xanthomonadaceae using CRB-BLAST [4] with default settings. Nucleotide sequences for the CRB-BLAST hits were then retrieved and aligned using MAFFT [41], followed by RAxML version 8.0.24 [73] to build a phylogenetic tree, which was visualized using Fig Tree version 1.4.4 (http://tree.bio.ed.ac.uk/). FimT3-encoding sequences were retrieved from the phylogenetic tree using the TREE2FASTA Perl script [63], translated into amino acid sequences and aligned using Clustal Omega (Clustal 12.1) to determine percentage of identical amino acids [48]. The visualization of the alignment of representative FimT3 sequences among X. fastidiosa strains and representative bacterial species belonging to the Xanthomonadaceae family was performed using the BoxShade webserver. The FimT3 GRxR motif sequence logo was generated using the WebLogo webserver [16].

Data analysis

Data from natural competence assays (recombination frequency) and the area under the fluorescence curve were compared by two-tailed Student’s t-test. Data from twitching motility, growth rate, viable X. fastidiosa CFU/ml obtained during natural competence assays, biofilm formation, planktonic growth, settling rate, AUDPC, X. fastidiosa population in planta and percentage of cells acquiring DNA from the extracellular environment were individually analyzed by one-way analysis of variance (ANOVA) followed by Tukey’s HSD multiple comparisons of means in R 4.0.0 under the package multcomp [31]. Correlation among analyzed phenotypes was determined by Pearson’s correlation using the SigmaPlot software version 11.0 (Systat Software Inc.).