|Home | About | Journals | Submit | Contact Us | Français|
Complete sequencing of the Xylella fastidiosa genome revealed characteristics that have not been described previously for a phytopathogen. One characteristic of this genome was the abundance of genes encoding proteins with adhesion functions related to biofilm formation, an essential step for colonization of a plant host or an insect vector. We examined four of the proteins belonging to this class encoded by genes in the genome of X. fastidiosa: the PilA2 and PilC fimbrial proteins, which are components of the type IV pili, and XadA1 and XadA2, which are afimbrial adhesins. Polyclonal antibodies were raised against these four proteins, and their behavior during biofilm development was assessed by Western blotting and immunofluorescence assays. In addition, immunogold electron microscopy was used to detect these proteins in bacteria present in xylem vessels of three different hosts (citrus, periwinkle, and hibiscus). We verified that these proteins are present in X. fastidiosa biofilms but have differential regulation since the amounts varied temporally during biofilm formation, as well as spatially within the biofilms. The proteins were also detected in bacteria colonizing the xylem vessels of infected plants.
Aggregative growth is a common feature in the microbial world, and its discovery radically changed our concept of microbial growth dynamics. A cellular aggregate adhering to a surface is known as a biofilm. It has important characteristics, such as greater resistance to antimicrobial compounds (34, 54), increased capacity of the cells to take up nutrients from the environment (59), and higher detoxification efficiency resulting from an increase in expression of genes encoding efflux pumps (43). These characteristics give the biofilm cells a great adaptive advantage.
Biofilm growth also confers advantages to plant pathogens by promoting virulence and protection against plant defense responses. Bacteria can colonize different niches in the plant, from aerial surfaces to roots and the vascular system, and biofilm formation can play a role at all of these sites of colonization. In the vessels, biofilms are very important since the cells need to survive in a competitive habitat where plant defense compounds are produced in response to infection (7).
Biofilm development is divided into at least the following five phases: (i) reversible attachment, (ii) irreversible attachment, (iii) beginning of maturation, (iv) mature biofilm, and (v) dispersion (13, 50). In Xylella fastidiosa strain 9a5c, the maturation phase occurs between days 15 and 20 in vitro, while dispersion occurs between days 25 and 30, as observed by our analysis of biofilm formation using different methods, including scanning electron microscopy and quantification of exopolysaccharides, biomass, and the total protein (unpublished data). The establishment and development of biofilms of plant-colonizing bacteria share several features with the establishment and development of biofilms of human bacterial pathogens, such as regulation by quorum sensing, nutrient starvation regulation, and phase variation. Motility is also an important factor not only for the initiation and development of the biofilm but also for dispersion (50). Attachment is mediated by surface-associated structures, which include both polysaccharides and proteins classified as fimbrial and afimbrial adhesins, depending on the structure to which they contribute. Fimbrial adhesins form filamentous structures, while afimbrial adhesins produce projections on the outer membrane (23).
X. fastidiosa, a Gram-negative phytopathogen that grows as a biofilms in both plant xylem vessels and the cybarium of insect vectors, is a major threat to plant production around the world. In Brazil, it has a major economic impact on citriculture since it causes citrus variegated chlorosis disease (CVC) (35, 39, 42). The biofilm formed by X. fastidiosa blocks the xylem vessels of susceptible citrus plants, impairing water flow. This blockage leads to a drastic reduction in fruit size (32) and, consequently, severe economic losses resulting from reduced plant productivity (4).
Due to the economic damage caused by CVC, there has been a major effort to generate more information about its biology. This led to sequencing of the genome of the pathogen. The X. fastidiosa genome harbors a wide variety of genes encoding adhesins (53). Bacterial cell surface adhesins are important in the initial phases of adherence to surfaces, as well as in bacterium-bacterium interactions and microcolony development (15). Insight into X. fastidiosa has also come from genome analysis of a strain of X. fastidiosa which causes Pierce's disease of grape (58). Studies of this strain showed that the cellular aggregation process involves type I and type IV fimbrial adhesins. The two types of fimbriae present different adhesion forces that help bacteria adhere to a substrate (10, 30). Adhesion proteins have also been demonstrated to mediate adherence to carbohydrates of leafhopper foregut surfaces (27). In addition, both fimbrial and afimbrial adhesins are important for plant pathogenicity (38, 41). However, the expression of these proteins during X. fastidiosa biofilm formation either in vitro or in planta is still poorly understood. For X. fastidiosa strains causing CVC, nothing is known about the role of these proteins in pathogenicity or biofilm formation, although some adhesion-encoding genes, such as pilA2, pilC, xadA1, and xadA2, were found to be upregulated either in virulent strains of X. fastidiosa or during biofilm formation (12, 14). These results suggest that the biofilm mode of growth is important for successful colonization of the citrus host by X. fastidiosa strains that cause CVC. In this work we focused on the temporal expression of the PilA2 and PilC fimbrial proteins and XadA1 and XadA2, which are afimbrial adhesins, during in vitro development of X. fastidiosa CVC biofilms. We demonstrated that the temporal and spatial patterns of expression of the fimbrial and afimbrial adhesins are very different during biofilm development in vitro. Moreover, we also verified that these adhesins are present in X. fastidiosa cells in symptomatic plants of three different hosts (citrus, periwinkle, and hibiscus).
X. fastidiosa subsp. pauca strain 9a5c (51), for which a genome sequence is available and which was obtained from INRA (Institut National de La Recherche Agronomique, Bordeaux, France), was used in all studies. Bacteria were extracted from petioles of symptomatic plants that had been ground in phosphate-buffered saline (PBS), and the suspensions were spread on periwinkle wilt (PW) medium (8). The first colonies were observed between days 10 and 15 after inoculation, and cells were inoculated into 50 ml of PW medium and grown at 130 rpm and 28°C. The cultures were transferred weekly to fresh medium (1 week corresponded to one passage) and used to obtain both biofilm and planktonic cells. Biofilms were recovered from passages 1 (7 days, 14 generations) to 8 (56 days, 112 generations). Planktonic cells which did not form a biofilm were collected after the passage 18 (126 days, 252 generations) in PW medium and were used after 10 days of growth. For X. fastidiosa strain 9a5c, the doubling time was 12 h.
In order to amplify highly antigenic regions of the target proteins for antibody production, primers were designed for flanking coding sequences having hydrophilic and antigenic regions (Lasergene 99; DNASTAR). More than one pair of primers were designed for different antigenic regions of the fimbrial proteins PilA2 (Tfp pilus assembly protein, major pilin FimA/PilA [type IV fimbrillin]) (Bioinformatics Laboratory open reading frame identification number [LBI ORF ID] XF2539) and PilC (type IV fimbrial assembly protein PilC) (LBI ORF ID XF2538) and the afimbrial proteins XadA2 (surface protein adhesin YadA-like) (LBI ORF ID XF1529) and XadA1 (surface protein adhesin YadA-like) (LBI ORF ID XF1516). These proteins were originally annotated FimA, PilC, Hsf, and UspA1 but later were reannotated as indicated above; the current accepted annotation is available at www.xylella.lncc.br. EcoRI and HindIII sites were placed in the 5′ and 3′ regions of each primer for directional cloning in pET28a (Novagen). The Primer Select software from Lasergene99 (DNASTAR) was used to design the primers. The primer sequences are shown in Table Table11.
Each 25-μl PCR mixture used to obtain amplicons contained 100 ng of total 9a5c DNA, 50 ng of each primer (forward and reverse), 2.5 mM deoxynucleoside triphosphates (dNTP), 1.25 μl of 50 mM MgSO4, 2.5 μl of 10× buffer (Invitrogen), 1 U of Taq Platinum high-fidelity polymerase (Invitrogen), and Milli-Q H2O. The amplification program was one cycle at 94°C for 3 min, followed by 35 cycles of 55°C for 1 min, 72°C for 1.5 min, and 94°C for 1 min and then 10 min at 72°C for further extension. The amplified fragments were analyzed in a 1% agarose gel stained with ethidium bromide and were visualized under UV light. The fragments were excised from the gel and purified using a GFX PCR DNA and gel band purification kit (Amersham Biosciences). The amplicons were cloned in the pGEM-T vector (Promega). Escherichia coli DH5α competent cells were transformed with the constructs, and the transformants were grown in LB containing 100 μg/ml of ampicillin.
The recombinant plasmids were sequenced with the two primers used for amplification of the target genes and primers for the pGEM-T vector (T7 and SP6 promoter primers). For sequencing the xadA1-1 and xadA1-2 fragments (1.5 and 2.0 kb, respectively), internal primers were also used in order to obtain the complete nucleotide sequences. The reaction mixtures were prepared by following the Applied Biosystems instructions for a BigDye Terminator cycle sequencing Ready Reaction DNA sequencing kit (v 3.0), and the resulting DNAs were run in an ABI 3730 automatic sequencer (Applied Biosystems). The quality of a cloned DNA fragment was confirmed by performing similarity searches against the X. fastidiosa database using BlastN and BlastX (http://www.ncbi.nlm.nih.gov/sutils/genom_table.cgi) tools. Plasmids containing sequences that exhibited 100% identity with the expected nucleotide sequences were cleaved with EcoRI and HindIII (Invitrogen) for directional cloning in pET28a using T4 DNA ligase (Invitrogen). pET28a was designed for expression of the recombinant protein fused to a six-His tag in both the N- and C-terminal regions of the protein. E. coli Rosetta competent cells were transformed with the recombinant plasmids, and selection was performed in LB plates containing 100 μg/ml of kanamycin. PCR colony screening was used to exclude false positives.
The target proteins were induced using 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) in a 100-ml culture incubated for 2 h at 37°C. For the proteins that appeared in the insoluble phase, a lower temperature (25°C) was used for induction. The cells were collected by centrifugation at 4,000 × g for 20 min at 8°C, suspended in 1 ml of buffer 1 (50 mM Tris, 300 mM NaCl; pH 7.5) containing 1 mg/ml lysozyme and 1 mM phenylmethylsulfonyl fluoride (PMSF), and lysed by sonication on ice with an ultrasonic cell disruptor (Unique). The cell debris and supernatant were separated by centrifugation at 12,000 rpm for 15 min at 8°C, and the expression pattern for each sample was monitored by SDS-PAGE performed using a 12% or 15% separation gel under reducing conditions as described by Laemmli (29); the gel was stained with Coomassie brilliant blue R-250 and destained with methanol-acetic acid.
Soluble proteins were purified using an immobilized metal affinity chromatography (IMAC) column packed with 1.0 ml of nickel-nitrilotriacetic acid (Ni-NTA) resin. Bound proteins were eluted with 4 ml of 200 mM imidazole in 50 mM Tris (pH 7.5), 300 mM NaCl. Fractions were used to estimate the total protein concentration (Bradford assay) and were analyzed by SDS-PAGE.
Even at low induction temperatures PilC was found only in the insoluble fraction of the extract. Therefore, this protein was cut out of the gel for immunization.
After obtaining the target proteins, we confirmed their identity by mass spectrometry analysis. For this, samples were excised from an SDS-PAGE gel and digested as described by Fiorani Celedon et al. (16), using trypsin (Invitrogen), before sequence analysis. The peptide mixtures were identified by online liquid chromatography (LC) using a Cap-LC coupled to a quantitative time of flight (Q-TOF) Ultima API mass spectrometer (Waters). Five microliters of sample was loaded onto a nanoease trapping column (0.18 by 23.5 mm; Waters) for preconcentration, which was followed by separation of peptides in a nanoease LC column (Symmetry 300 C18 3.5 μm; 75 by 100 mm; Waters). Peptides were eluted using a 60-min linear gradient of solvent B (95% [vol/vol] acetonitrile [ACN] and 0.1% [vol/vol] formic acid in water) at a flow rate of 250 nl min−1. Solvent A consisted of 5% (vol/vol) ACN and 0.1% (vol/vol) formic acid in water. The whole analysis was performed using the positive-ion mode with a needle voltage of 3 kV. The mass range used was m/z 300 to 2,000, and tandem mass spectrometry (MS/MS) spectra were acquired for the most intense peaks (≥15 counts). Multiple charged precursor ions were selected for fragmentation and peptide sequencing using automated data-dependent acquisition (DDA) MassLynx software (Waters), switching from the MS mode to the MS/MS mode and then returning to the MS mode. The resulting fragmented spectra were processed using the ProteinLynx v4.0 software (Waters). MASCOT MS/MS Ion Search (Matrix Science) was used to blast the sequences against the Swiss-Prot and NCBI nr databases. Combined MS-MS/MS searches were conducted with a parent ion mass tolerance of 50 ppm, an MS/MS mass tolerance of 0.1 Da, carbamidomethylation of cysteine (fixed modification), and methionine oxidation (variable modification). According to MASCOT probability analysis, only significant (P < 0.05) hits were accepted.
Antibodies against the target proteins were obtained using New Zealand White rabbits. The purified proteins, as well as PilC (excised from an SDS-PAGE gel), were individually mixed with Freund's complete adjuvant (Sigma) and injected into individual rabbits. Proteins mixed with Freund's incomplete adjuvant were injected two more times, at 10 and 20 days after the first injection. The concentration of inoculated proteins was 150 μg. The quality of the antibodies was tested by performing a direct enzyme-linked immunosorbent assay (ELISA), using 3 μg of the target proteins as antigens and PBS as the negative control. The antibody dilutions varied from 1:10,000 to 1:256,000. The antibodies produced positive results even at a dilution of 1:250,000, and no cross-reaction was observed with any of the other proteins used for antibody production.
Three replicate experiments were carried out for extraction of biofilm proteins. Cells grown in PW medium until the passage 8 were transferred to flasks containing 50 ml of medium and grown under the conditions described above. The biofilms that attached to the glass at the medium-air interface were collected after 3, 5, 10, 15, 20, and 30 days of growth, which corresponded to the different biofilm developmental phases (13), and total protein was extracted. To do this, the medium was discarded, and 1 ml of wash buffer (50 mM Tris [pH 8.0], 25 mM NaCl, 5 mM EDTA [pH 8.0], 2% Triton X-100) was used to collect biofilm cells. Lysozyme and PMSF (1 mM each) were added to the extract, and the samples were kept on ice for 20 min. Cells were lysed by sonication, which was followed by centrifugation at 10,000 × g for 10 min. The supernatant and the pellet were used for protein analysis. For planktonic conditions, the cells were obtained after passage 18, when no biofilm was observed at the medium-air interface, and were collected by centrifugation at 6,000 × g for 5 min. Total protein was extracted by treating the cells as described above for the biofilms. For all of the experiments, the amount of total protein obtained from each biofilm phase was normalized before the gel was loaded, using a bovine serum albumin (BSA) standard curve for comparison, for which R2 was 0.956.
X. fastidiosa proteins from different biofilm developmental phases and planktonic growth were quantified using the procedure described by Lowry et al. (31). For Western blot assays, for each of the growth conditions 6 μg of proteins was separated by 9% or 12% SDS-PAGE, depending on the protein size. Individual Western blot experiments were performed for each of the antibodies tested, and preimmune rabbit antiserum was used to verify that there was a possible nonspecific cross-reaction. After electrophoresis, gels were incubated in Tris-glycine buffer (48 mM Tris, 39 mM glycine, 0.04% SDS, 20% methanol) for 1 h, and the proteins were transferred to a nitrocellulose membrane (Hybond-C Ultra; Amersham) according to the specifications of the manufacturer (Pharmacia) for the Multiphor II Novablot electrophoretic transfer unit. Membranes were treated by following Amersham's Hybond-C nitrocellulose membrane protocol, using 1× TBS (150 mM NaCl, 10 mM Tris-HCl; pH 8.0), a 16-h blocking period, and incubation with antibodies at a 1:10,000 dilution for 6 h. The membranes were incubated with a 1:10,000 dilution of goat anti-rabbit alkaline phosphatase conjugate (Sigma) for 1 h at 37°C and washed three times with 1× TBS, 0.1% Tween 20. For detection of the proteins, the membranes were incubated in a freshly prepared substrate solution (0.1 M Tris-HCl [pH 9.5], 0.1 M NaCl, 5 mM MgCl2, 2 mg 5-bromo-4-chloro-3-indolylphosphate, 4 mg Nitro Blue Tetrazolium). When the desired staining was observed, the reaction was stopped by washing the membranes in distilled water. At least three membranes were tested in each biological experiment.
Signal intensity was quantified using Image J 1.42q (http://rsb.info.nih.gov/ij) software. The values were used for statistical analyses (P ≤ 0.05, t test).
For the immunofluorescence analysis, biofilms were obtained using cover glasses placed in the bottom of 24-well Nunclon Delta SI Multidishes (Nunc) containing 1 ml of liquid PW medium in each well. One plate was used for each time point, and 100 μl of a preinoculum containing X. fastidiosa 9a5c (optical density, 0.1) was added to four different wells. The culture plates were maintained statically at 28°C and were analyzed after 3, 5, 10, 15, 20, and 30 days of growth. The PW medium was gently removed from the wells with a pipette, and the biofilm adhering to cover glasses was washed as described by Roper et al. (49) but without fixation of the cells since the aim of this work was to evaluate just the adhering cells. The antibodies and the preimmune sera were used individually in a dilution of 1:1,000. For localization of proteins in the biofilms that formed, we used 1 ml of goat anti-rabbit rhodamine-conjugated IgG (Santa Cruz Biotechnlogy) at a 1:10,000 dilution in PBS. The biofilm cells were stained with Syto 9 diluted 1:600 in autoclaved Milli-Q H2O. Images were obtained using an oil immersion lens (numerical aperture, 1.0) with an Olympus UIS2 fluorescence microscope; for Syto 9, a filter that permits excitation at wavelengths between 460 and 490 nm and emission at wavelengths between 500 and 520 nm was used, and for rhodamine, a filter that allows excitation at wavelengths between 510 and 550 nm and emission at wavelengths between 570 and 590 nm was used. To verify that a possible nonspecific fluorescent signal was present, two negative controls were prepared from biofilms after 30 days of growth, one without secondary rhodamine-conjugated antibodies and the other without incubation with primary antibody.
Small fragments of petioles from citrus (sweet orange; Citrus sinensis cv. Pera), periwinkle (Catharantus roseus), and hibiscus (Hibiscus schizopetalus) plants infected by X. fastidiosa were fixed in a modified Karnovksy solution (2% paraformaldehyde and 2.5% glutaraldehyde in 0.05 M cacodylate buffer [pH 7.2] with 0.001 M CaCl2) for 1 to 2 h. The plants were naturally infected with X. fastidiosa, whose presence was confirmed by previous PCR assays. The citrus plants exhibited characteristic symptoms of CVC, while the periwinkle plants exhibited general chlorosis and stunting and the hibiscus plants had leaf scald symptoms (28). Fixed tissues were dehydrated using an ethanol series (30%, 50%, 70%, and 90% ethanol for about 10 min each) and then transferred to chilled 100% ethanol. Dehydrated tissues were infiltrated for 6 h with a mixture of 100% ethanol and LR White resin (1:1) at 4°C and overnight with pure LR White at 4°C. Infiltrated tissues were then transferred to size 3 gelatin capsules filled with pure LR White and allowed to polymerize at 60°C for 2 days. Embedded tissues were cut into thin sections with a Leica UC 6 ultramicrotome equipped with a Diatome diamond knife, and the sections were mounted in 100-mesh nickel grids covered with Formvar films. Immunolocalization assays with antibodies against the target proteins were performed using the protocol of Vandenbosch (57). As controls for the experiment, images of noninfected xylem vessels of infected plants, the walls of infected xylem vessels, and parenchyma were evaluated and compared. The background was reduced by adjusting the antibody dilution to 1:250. About 40 immunogold tests for each plant with at least 80 sections per test were done, which generated a minimum of 20 images for each protein in each plant species in the experiment. Transmission electron microscopic examination was done using a Zeiss EM 900 electron microscope at 80 kV, and the images were registered digitally with an F-view charge-coupled-device (CCD) monochrome camera (Soft Imaging System).
Clones containing recombinant plasmids produced sufficient amounts of the proteins of interest for further purification. The PilA2, XadA2, and XadA1 recombinant proteins were in the soluble fraction of the protein extracts. In contrast, PilC was in only the insoluble fraction, even when low induction temperatures were used. The fractions containing PilA2, XadA2, and XadA1 were purified by IMAC, and it was possible to recover large amounts of purified proteins after elution with imidazole. For PilC, we excised the band corresponding to the target protein directly from a polyacrylamide gel. Induction and purification were monitored by SDS-PAGE (data not shown). The identities of the purified proteins were confirmed by LC-MS/MS (data not shown). The proteins were inoculated into rabbits for antibody production.
The amounts of the fimbrial and afimbrial proteins in biofilms were verified by Western blotting using cells obtained after 3, 5, 10, 15, 20, and 30 days of growth (Fig. (Fig.1),1), which corresponded to the initial adhesion of the cells to the substrate (3 and 5 days), microcolony formation (10 days), early development of the biofilm architecture (15 days), mature biofilm (20 days), and the biofilm after some apparent dispersion had occurred (30 days) (unpublished data). Analysis of PilA2, PilC, and XadA2 revealed that these proteins had apparent molecular masses of approximately 15 kDa, 55 kDa, and 200 kDa, respectively. These masses are in accordance with the values expected for these proteins (http://www.lbi.ic.unicamp.br/xf/). For XadA1, Western blotting revealed a band at approximately 70 kDa (Fig. (Fig.1D);1D); the predicted molecular mass of XadA1 from X. fastidiosa 9a5c is 98 kDa. Since the possibility of contamination was excluded by the mass spectrometry analysis (Table (Table2),2), there was apparently cleavage of the protein or its genome annotation is wrong.
The temporal patterns of abundance of the two fimbrial proteins (PilA2 and PilC) were similar, and statistically significant differences between the initial and later phases of biofilm formation were observed (Fig. 1A and B). The patterns of abundance for the afimbrial proteins were different from each other in Western blot analyses. Very little XadA2 protein was observed at the beginning of biofilm formation. By 10 days, the amount of this protein had increased, but a statistically significant increase compared with the amount in young biofilm cells was observed only after 30 days of growth (Fig. (Fig.1C).1C). XadA1 was produced rather constitutively, and there was no statistically significant change in the amount of the XadA1 protein at any biofilm phase (Fig. (Fig.1D1D).
To verify that these adhesins are associated with biofilm growth, we compared their abundance with that in planktonic cells. Under our experimental conditions, none of the adhesins evaluated was detected in planktonic X. fastidiosa cells (data not shown).
In order to localize the adhesion proteins during X. fastidiosa biofilm formation in vitro, we immunolabeled the proteins and visualized them by using fluorescence microscopy. The PilA2 and PilC fimbrial proteins were apparent throughout a biofilm by day 3. However, by days 20 and 30, these proteins seemed to be more concentrated in some parts of the biofilm than in other parts (Fig. (Fig.22 and and3).3). The XadA2 afimbrial protein was not observed in small cell aggregates on day 3 of growth but was seen by day 5, when cellular aggregates were a little larger. By day 10, when the biofilm began to exhibit a multilayer organization, this protein was quite evident, suggesting that it may have a role in cell-cell adhesion and three-dimensional structuring (Fig. (Fig.4).4). The biofilm distribution of XadA1 was quite different from that of the other proteins (Fig. (Fig.5).5). At day 10, when biofilm maturation had started, this protein was present primarily in gaps in the biofilm structure. At days 15 and 20, when the biofilm was thickest, the localization of this protein was somewhat different; it was located primarily in discrete spots, forming an “island-like” pattern in the biofilm. No other protein exhibited this pattern. The XadA1 protein remained localized in “island-like” inclusions on day 30, even though the spots were less numerous than they were in earlier phases of growth and the labeling seemed not to be as tightly linked to the biofilm (Fig. (Fig.55).
To verify the localization of the adhesion bacterial proteins in vivo, we analyzed X. fastidiosa cells present in sections of xylem vessels of infected periwinkle, hibiscus, and citrus leaves exhibiting disease symptoms. PilA2 is one of the rod-forming units of the type IV pili, and immunogold labeling demonstrated that it was present in cell membranes and outside the cells in all of the hosts tested (Fig. (Fig.6A,6A, B, and C). The PilC protein, which is involved in pilus assembly, was also detected in the three hosts tested, but it was mainly in the bacterial cell membrane (Fig. (Fig.6D,6D, E, and F). In contrast, the afimbrial proteins detected were outside the cells. XadA2 was outside the cells but close to the outer membrane of cells in the three plant hosts (Fig. (Fig.7A,7A, B, and C). The high level of background labeling for XadA1 in periwinkle did not allow analysis of this protein in this plant, but there was little labeling of host materials in hibiscus and citrus, which allowed XadA1 to be visualized in these species. Similar to the findings for XadA2, most of this protein was observed to be close to the outer membrane, but some labeling was also observed at more distant sites (Fig. (Fig.7D,7D, E, and F).
Hierarchically ordered gene expression circuits are characteristic of biofilms (40). In the present work we demonstrated that adhesion proteins are expressed temporally throughout biofilm development in an orderly pattern. X. fastidiosa biofilm development can be divided into five different phases, including initial attachment, microcolony formation, beginning of maturation, mature biofilm, and dispersion (13). Growth of biofilms is a well-known behavior in bacteria, and in X. fastidiosa it is necessary for development of symptoms due to occlusion of the xylem vessels (1, 41). A study of the Pierce's disease X. fastidiosa Temecula strain suggested that its colonization of vessels is related to its adhesion capacity and twitching motility, which are mediated by adhesion proteins (9).
Type IV pili are important for adhesion and twitching motility in bacteria (2, 11, 18, 30, 36, 44, 46). As expected, the temporal patterns of variations in the amounts of PilA2 and PilC were somewhat similar during X. fastidiosa biofilm formation since these proteins are involved in type IV pilus assembly. Larger amounts were observed at the beginning of development and the dispersion phase, when cells might be expected to be more motile. Thus, type IV pili must have different roles during biofilm development, and adhesion to the surface and twitching motility are associated with biofilm spread early in development. The observation that the PilA2 and PilC proteins were present in spots in the biofilms is intriguing and suggests that labeled cells may be released from a biofilm not only as individual cells but also as clusters. Twitching may take place very early in biofilm development and may be necessary for the formation of macrocolonies. It is known that twitching motility allows existing microcolonies to join and form macrocolonies that develop into a mature biofilm (56).
Immunogold electron microscopy of the fimbrial proteins in periwinkle, hibiscus, and citrus plants revealed that they were located in the same subcellular compartment of the X. fastidiosa biofilms. PilA2 was located in the cell membrane, as well as outside the cell, while PilC was preferentially close to the membrane. This observation is in agreement with the expected localization for anchoring of immature pilus subunits and assembly of these subunits (24, 25). The small amount of PilA2 observed was not unexpected since assembly of the rod could block the epitopes of this protein, so that only the epitopes of proteins that are in the initial steps of assembly of the pili are available for interaction with the antibody (24).
The afimbrial protein XadA1 was detected in all phases of biofilm development, and XadA2 was detected mainly in later phases of biofilm development. Western blot analysis of XadA1 revealed a protein that was smaller than expected. This protein is very similar to UspA1 of Moraxella catarrhalis, for which a reduction in size was reported previously; this reduction in size was also dependent on the temperature to which the protein was subjected (21, 37). The discrepancy between the expected and observed sizes of XadA1 could be due to the same factors that apparently affect UspA1 processing. The detection of XadA1 in all phases of biofilm formation suggests that it may have a role in the initial adhesion to surfaces, as well as in cellular aggregation. UspA1 of M. catarrhalis is important for adhesion to other cells and thus for biofilm formation (45). It is noteworthy that XadA1 mutants of X. fastidiosa Temecula were reported to be defective in adhesion as single cells to glass surfaces but that cell-cell aggregates could still form. Thus, XadA1 is apparently involved in the initial adhesion of Temecula cells (15). In the mature biofilm phase, XadA1 is distributed in an “island-like” pattern, suggesting that it may have a role in biofilm development by altering cell-cell adhesion and aggregation. However, this hypothesis needs to be investigated further.
The other afimbrial protein evaluated in this work, XadA2, was observed mainly only after 10 days, indicating that it is likely not involved in adherence to the substrate but instead is involved in cell-cell adhesion associated with higher cell density. A protein similar to XadA2 from Haemophilus influenzae is known to have specific domains (Hia) that contribute to adherence to mammalian cells in vitro (6) and direct attachment to specific proteins (19, 55). These domains are present in the X. fastidiosa protein (Fig. (Fig.8),8), and they could mediate protein-protein interactions. XadA1 and XadA2 belong to the family of trimeric autotransporters, and thus they may have adhesive activity that mediates bacterial interactions either with host cells, which have their own extracellular matrix proteins, or with other unknown proteins (6, 20, 48). Even though the X. fastidiosa XadA1 protein resembles UspA1 and YadA, two of the most-studied proteins of the family, it contains more Hep-Hag repetitive regions (Fig. (Fig.8).8). XadA2 exhibits similarity to Hsf of H. influenzae, but in X. fastidiosa there are many more domains that are related to adhesive functions, such as the Hep-Hag and hemagglutinin domains (Pfam domains PF05658 and PF05662) (Fig. (Fig.8).8). Additionally, analysis of XadA2 revealed the presence of motifs that are unique to proteins from X. fastidiosa (Pfam domain PF06669) (3) (Fig. (Fig.8).8). These structural characteristics suggest that XadA1 and XadA2 may have distinct functions in biofilm formation and host interactions.
These afimbrial adhesins may complement the role of other afimbrial adhesins, such as the hemagglutinins. In X. fastidiosa strain Temecula, the hxfA gene encodes one of the hemagglutinins. A mutant with a mutation in this gene showed increased virulence and monolayer biofilm formation in vivo, suggesting that the mutation allowed faster colonization of the xylem vessels, resulting in more severe symptoms (17). This observation suggests that in X. fastidiosa Hxf may be related to aggregation and therefore biofilm growth may be a way to avoid rapid colonization of the host, which is undesirable for the pathogen since it could ultimately lead to the death of the plant host.
XadA1 and XadA2 immunogold labeling was detected close to the cell membrane in a pattern similar to that reported for YadA of Yersinia enterocolitica, in which aggregation by the protein head domain located outside the cell directed similar localization (22). However, some XadA1 was also detected far from the cells, and together with immunofluorescence observation of apparent protein aggregates far from cells, this suggests that this protein is secreted into the environment. This hypothesis and the possible roles of such secreted proteins need to be investigated further.
Interestingly, the expression of most of the genes encoding the proteins studied in this work during biofilm formation in vitro (12) does not correlate well with the accumulation of proteins at a given time that was detected by Western blot analysis. Expression of xadA2, for instance, was detected mainly on days 5 and 10 during biofilm formation in vitro, yet we observed that the proteins were present mainly after day 10. This lag between gene expression and protein appearance is intriguing. It seems likely that different patterns of transcription, translation, and protein degradation occur during biofilm development. Translation could be affected by small regulatory RNAs mediated by gacA. Activity of small RNAs related to gacA has been demonstrated for Pseudomonas aeruginosa and also for X. fastidiosa (26, 47, 52). In the latter case, regulation via gacA directly affects xadA1 and xadA2 expression and thus cell-cell adhesion and biofilm formation (52). These small RNAs could posttranscriptionally regulate the expression of a variety of proteins, including afimbrial adhesins, thus explaining the delay in the production of these adhesins during biofilm formation. The regulation of these genes could also be mediated through a diffusible signal factor (DSF)-dependent signaling system. Mutations in rpfF, which is responsible for the synthesis of DSF in X. fastidiosa, impairs biofilm formation in an insect vector (42) and leads to greater virulence in plants due to an increased ability to move in the xylem vessels (5, 42).
Both fimbrial and afimbrial proteins are involved in biofilm formation in X. fastidiosa in vitro. The expression of these adhesins seems to be substantially different at different phases of biofilm development, suggesting that there are different contributions to various biofilm processes. The variation in the spatial distribution pattern of these proteins in the biofilm also suggests that they contribute in different ways to biofilm functions. The detection of these proteins in xylem vessels of infected plants indicates that the biofilms in plants may resemble those produced in vitro and that these structures likely contribute to the infection process.
We thank Steven Lindow for critical revision of the manuscript and Renato Salaroli for transmission electron microscopy technical support.
This work was supported by research grants from Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) (grants 04/14576-2, 06/52681-8 and INCT-Citros 08/57909-2, 573848/08-4). M.A.T., A.P.D.S., E.W.K., C.A.L., M.A.M., and A.A.D.S. were recipients of research fellowships from CNPq.
Published ahead of print on 14 May 2010.