Phosphoproteomic Analysis Reveals the Effects of PilF Phosphorylation on Type IV Pilus and Biofilm Formation in Thermus thermophilus HB27*

Thermus thermophilus HB27 is an extremely thermophilic eubacteria with a high frequency of natural competence. This organism is therefore often used as a thermophilic model to investigate the molecular basis of type IV pili–mediated functions, such as the uptake of free DNA, adhesion, twitching motility, and biofilm formation, in hot environments. In this study, the phosphoproteome of T. thermophilus HB27 was analyzed via a shotgun approach and high-accuracy mass spectrometry. Ninety-three unique phosphopeptides, including 67 in vivo phosphorylated sites on 53 phosphoproteins, were identified. The distribution of Ser/Thr/Tyr phosphorylation sites was 57%/36%/7%. The phosphoproteins were mostly involved in central metabolic pathways and protein/cell envelope biosynthesis. According to this analysis, the ATPase motor PilF, a type IV pili–related component, was first found to be phosphorylated on Thr-368 and Ser-372. Through the point mutation of PilF, mimic phosphorylated mutants T368D and S372E resulted in nonpiliated and nontwitching phenotypes, whereas nonphosphorylated mutants T368V and S372A displayed piliation and twitching motility. In addition, mimic phosphorylated mutants showed elevated biofilm-forming abilities with a higher initial attachment rate, caused by increasing exopolysaccharide production. In summary, the phosphorylation of PilF might regulate the pili and biofilm formation associated with exopolysaccharide production.

Thermus thermophilus HB27 is a Gram-negative, rodshaped, and extremely thermophilic eubacterium isolated from a geothermal area (1). This organism grows at temper-atures up to 85°C and has an optimal growth temperature of 70°C. The thermostable enzymes obtained from members of the genus Thermus are of considerable interest because of their potential in research, biotechnological, and industrial applications (2,3). In addition, T. thermophilus HB27 is a suitable laboratory model for genetic manipulation, as it is easily cultured under laboratory conditions and has a natural transformation system that is much more efficient than those of other Thermus spp. (4). Intriguingly, thermophiles are also found in biofilms, enclosed within a matrix consisting of extracellular polymeric substances, in various natural and artificial thermal environments (5,6). Bacteria form biofilms in order to adapt and survive in harsh environments (7,8). Over the past few decades, biofilm formation has been a major focus of microbial research and, as such, has been studied in relationship to bacterial pathogenesis, immunology, biofouling, microbial technology, and industrial applications (7, 9 -12).
Members of the genus Thermus, like many other thermophiles, have evolved two main mechanisms for thermoadaption. One is biofilm formation, which confers protection against environmental stresses such as high temperature and the presence of antibiotics (8). In previous studies, a novel exopolysaccharide, TA-1, was isolated from a T. aquaticus YT-1 biofilm, and both its primary structure and its immunological activity were determined (13). In addition, we showed that the overexpression of uridine diphosphate (UDP)-galactose-4Ј-epimerase (GalE), which catalyzes the reversible interconversion of UDP-galactose and UDP-glucose, in T. thermophilus HB27 increases biofilm production because of the enzyme's involvement in an important step of exopolysaccharide (EPS) 1 biosynthesis (14). The other mechanism that enables Thermus to thrive in extreme habitats is natural trans-formation (i.e. the ability to take up free DNA). In hot environments, natural transformation allows the horizontal exchange of genetic information between extremophiles, including of genes that promote thermoadaptation (15)(16)(17). Recent studies showed that the type IV pili (T4P) on the cell surface of T. thermophilus HB27 not only are required for natural transformation (18,19), but also mediate adhesion and twitching motility (20). Also, together with the degree of EPS production, the presence of T4P on the bacterial cell surface contributes to the regulation of biofilm formation (21). However, despite extensive research on the physiological, biochemical, and genetic traits of thermophiles, the mechanisms underlying these functions and their role in thermal adaptation have not been fully elucidated (16,(22)(23)(24).
Advances in the field of phosphoproteomics have come from high-resolution mass spectrometry and prokaryotic genome sequencing, which have confirmed the phosphorylation of many bacterial proteins on serine/threonine and tyrosine residues (25,26). In surveys of phosphorylation-related functions, bacterial serine, threonine, and tyrosine phosphoproteins have been shown to regulate many physiological and adaptation processes, such as central carbon catabolism, the heat shock response, osmolarity, starvation, EPS synthesis, virulence, and sporulation (25)(26)(27). These observations have been followed by more detailed, species-specific phosphoproteomics investigations, including in Bacillus subtilis (28), Escherichia coli (29), Lactococcus lactis (30), Halobacterium salinarum (31), Klebsiella pneumonia (32), Pseudomonas spp. (33), Rhodopseudomonas palustris (34), and T. thermophilus HB8 (35). In this study, the role played by the global phosphorylation network of the thermophile T. thermophilus HB27 in the physiological processes that mediate the stress responses and thermotolerance of this bacterium was examined. Specifically, we used strong cation exchange (SCX) chromatography and titanium dioxide (TiO 2 ) (28 -30) enrichment to characterize the phosphoproteomic map of T. thermophilus HB27. Genetic manipulation of this strain indicated that phosphorylation of the PilF protein, which contains an ATP-binding motif (TTC1622/pilF) and drives T4P formation, is involved in both EPS production and piliation, thereby influencing the biofilm formation during thermophilic adaptation.

MATERIALS AND METHODS
Bacterial Strains and Growth Conditions-Wild-type PilF and its mutant versions were expressed in the host strain T. thermophilus HB27, grown under aerobic conditions at 70°C in Thermus modified (TM) medium (4). E. coli DH5␣ and BL21 (DE3) competent cells (Novagen, Madison, WI) were used as hosts for genetic manipulations of plasmids and for the overexpression of proteins, respectively. E. coli strains were grown in Luria-Bertani (LB) medium (36) at 37°C. When needed, kanamycin (30 g/ml) and/or ampicillin (100 g/ml) was added to TM or LB plates for plasmid selection.
Protein Extraction, Digestion, and Phosphopeptide Enrichment-T. thermophilus HB27 was grown aerobically to mid-exponential phase (optical density at 600 nm [OD 600] ϭ 0.8). The cells were harvested, and the resulting pellets were washed twice with PBS and then resuspended in fresh lysis buffer containing 50 mM Tris-HCl (pH 7.5), PhosSTOP phosphatase inhibitor mixture tablets (Roche), 6 M urea, and 2 M thiourea. Cell extracts, obtained using a French press, were subsequently centrifuged at 25,000g for 30 min at 4°C. The protein concentration was measured using the Bradford protein assay (Bio-Rad). Samples containing 25 mg of protein were disulfide-reduced with 1 mM dithiothreitol for 1 h at 37°C, alkylated with 5 mM iodoacetamide for 1 h in the dark, and trypsin digested (1:50 w/w) at 37°C overnight. The tryptic peptides were acidified with 5% trifluoroacetic acid (TFA) to pH 2.7. SCX and TiO 2 chromatography were used to fractionate and enrich the phosphopeptides (37,38). Briefly, the tryptic peptides were loaded onto a 1-ml Resource S column (GE Healthcare) in solvent A (5 mM KH 2 PO 4 , 30% acetonitrile (ACN), adjusted to pH 2.7 with TFA) at a flow rate of 1 ml/min and the flow-through was collected. The bound peptides were eluted in a gradient of 0%-30% solvent B (5 mM KH 2 PO 4 , 30% ACN, 350 mM KCl, adjusted to pH 2.7 with TFA) over 30 min. For phosphopeptide enrichment, TiO 2 beads (GL Sciences, Tokyo, Japan) were pre-incubated with the loading buffer (1 M glycolic acid in 80% ACN and 5% TFA). Each SCX fraction was then mixed and incubated with 2 mg TiO 2 beads for 30 min at room temperature. The beads were washed with loading buffer and wash buffer (80% ACN and 5% TFA), and the bound phosphopeptides were eluted with 1% NH 4 OH in 40% ACN, pH Ͼ 10.5. The eluates were desalted and dried for LC-MS/MS analysis.
Liquid Chromatography-Mass Spectrometric Analysis-Nano-HPLC-MS/MS analysis was performed on a nanoAcquity system (Waters, Milford, MA) connected to an LTQ-Orbitrap XL hybrid mass spectrometer (Thermo Electron, Bremen, Germany) equipped with a PicoView nanospray interface (New Objective, Woburn, MA). Peptide mixtures were diluted with 1% formic acid, loaded onto a C18 BEH column (75-m inner diameter, 25-cm length; Waters, Milford, MA), and separated using a linear gradient of 5%-40% ACN in 0.1% formic acid run over 90 min at a flow rate of 300 nl/min at a column temperature of 35°C. The mass spectrometer was operated in datadependent mode. A survey scan of the MS spectra (m/z 300 -1800) was acquired in the Orbitrap, with a resolution of 60,000 at m/z 400 after the accumulation of 10 6 ions. The most intense ions (up to 10) were further sequenced via collision-induced dissociation fragmentation (normalized collision energy 35%) in the LTQ after the accumulation of 7000 ions. An activation q ϭ 0.25 for 30 ms and a multistage activation at 97.98, 48.99, 32.66, and 24.49 Thomson (Th) were applied in the MS 2 acquisitions. The dynamic exclusion list was restricted to a maximum of 500 entries with a maximum retention period of 90 s. The Orbitrap measurements were performed with the lock mass option enabled to improve mass accuracy.
Data Processing and Validation-The raw data were processed using MaxQuant software, version 1.1.1.25 (39), to generate the peak lists. High-resolution profile MS/MS data were deconvoluted before the 10 most abundant peaks per 100 Th were extracted. Using the Andromeda search engine (40), we analyzed the derived peak lists against a composite target-decoy protein database constructed from the NCBI T. thermophilus HB27 protein database of a total of 2210 sequences (accession numbers: AE017221 and AE017222) downloaded from the NCBI Reference Sequence database (www.ncbi. nlm.nih.gov/protein) on March 30, 2009 (15). The database also included commonly observed contaminants of 101 protein sequences (version 1.0 of the common repository of adventitious proteins downloaded from the Global Proteome Machine) and 5880 Saccharomyces cerevisiae protein sequences downloaded from NCBI Reference Sequence database on March 30, 2009, that also serve as possible contaminants during cell culture. A search parameter of cysteine carbamidomethylation was set as the fixed modification; methionine oxidation, protein N-terminal acetylation, and phosphorylation on serine, threonine, and tyrosine residues were set as variable modifica-tions. The digestive enzyme was trypsin, and the maximum number of missed cleavages was two. The minimum peptide length was six amino acids. The Andromeda search results were further processed by MaxQuant. In order to achieve reliable identifications, the maximum false discovery rate of peptides and proteins was set at 0.01, which allowed the posterior error probability of peptide-spectrum matches to be less than 0.05.
Bioinformatics Analysis-The biological function and cellular component of each identified phosphoprotein were grouped using the Blast2GO tool (41). The Blast2GO suite is primarily based on the Gene Ontology annotation to perform functional annotation. Using Fisher's exact test, we obtained the Gene Ontology categories that were statistically overrepresented. All Gene Ontology terms of overrepresented proteins were selected based on p Ͻ 0.05. In addition, we used the prediction tool pSORTb (version 3.0) (42) to classify bacterial protein localization. BLASTP, a phylogenomic database (ProtClustDB) (43), protein family and domain databases (UniPort, InterPro, Pfam, and PROSITE), and the pathway database (KEGG) were used in parallel as complementary methods to further classify each identified protein. Information on Enzyme Code, the KEGG pathway, and motifs was also extracted and compiled into a dataset provided in supplemental Table  S1.
Construction of PilF Expression Plasmids and Site-directed Mutagenesis-Chromosomal DNA from T. thermophilus HB27 was isolated according to a standard phenol extraction protocol (44). Plasmid DNA was purified using the Bioman plasmid purification kit (Bioman, Taipei, Taiwan). Phusion High-Fidelity DNA Polymerase (Finnzymes, Espoo, Finland) was used for all PCR amplifications. Based on its DNA sequence, the pilF (GeneID: 2774631) gene was PCR-amplified using the primers listed in supplemental Table S2. NdeI and HindIII restriction enzyme sites (MBI Fermentas, Vilnius, Lithuania) were added to the N and C termini of pilF, respectively. The PCR product of pilF was inserted into a linearized pGEM7Z vector (Promega, Madison, WI) for nucleotide sequence determination (Genomics BioSci & Tech, Taipei, Taiwan). Subsequently, pilF was subcloned into the pET28a expression vector (Novagen) at the corresponding NdeI and HindIII restriction sites to generate plasmid pET28a-pilF.
Mutations at the identified phosphorylation sites of PilF were generated using the mutagenic primers in overlap extension PCR (45), in which two rounds of PCR were performed to generate mutations at the desired positions of a gene. pET28a-pilF served as the template in the first round, with two flanking primers and two internal primers (each containing the mutated nucleotides) used to generate the Nand C-terminal PCR products in separate reactions. The primers were designed to create an overlap region in the two resulting PCR fragments. In the second round, the purified PCR products from the first round served as templates. The two flanking primers were used to amplify the full-length mutant PilF gene. Finally, the mutant pilF constructs were ligated with pET28a as described above to create mutant versions of pET28a-pilF.
To aerobically overexpress wild-type and mutant PilFs in T. thermophilus HB27, the inducible nitrate reductase promoter (Pnar) region in plasmid pMKE2 (46) (Biotools B&M, Madrid, Spain) was replaced with the endogenous pilF promoter (PpilF). A 157-bp PpilF promoter containing the flanking region of the XbaI-NcoI sites was PCR-amplified, and the resulting DNA fragment was cloned into the corresponding sites of plasmid pMKE2 to replace the Pnar promoter, yielding pMKE2-PpilF. The expression vector pMEK2, allowing overexpression of His-tagged enzymes in T. thermophilus, was developed based on pET28b(ϩ). To generate pMKE2-Pro-pilF, the wild-type and mutant pilF genes, derived from pET28a-pilF and its mutant versions, were directly subcloned into pMKE2-PpilF containing the same multiple cloning sites (NdeI and HindIII).

Overexpression of Recombinant PilFs in T. thermophilus HB27-
Successful recombinant pMKE2-Pro-pilF and its mutant derivatives were transformed into T. thermophilus HB27 using a modified version of the protocol described by Koyama et al. (4). Briefly, T. thermophilus HB27 was grown overnight at 70°C in 10 ml of TM medium. Overnight cultures were diluted 1:10 in fresh medium and grown for an additional 6 h at 70°C. One-milliliter aliquots of these cells were transformed with the required amount of plasmids. The transformants were selected 2 h later on 2% agar plates containing 30 g kanamycin/ml and then incubated for 2 to 4 days at 70°C. To verify the overproduction of PilF and its mutant forms in T. thermophilus HB27, each transformant colony was inoculated in 5 ml of TM medium containing 30 g kanamycin/ml and then grown overnight with shaking at 200 rpm for plasmid mini-preparations and anti-His-tag antibody detection.
Expression and Purification of Recombinant PilFs in E. coli-Recombinant wild-type and mutated forms of PilF were obtained using the respective plasmids to transform E. coli BL21 (DE3). E. coli cells containing the distinct plasmids were grown overnight at 37°C in 100 ml of LB medium containing 30 g kanamycin/ml. Overnight cultures were transferred into 1000 ml of fresh medium and inoculated until A 600 ϭ 0.6 was reached. The cultures were induced with isopropyl-␤-D-thiogalactopyranoside (final concentration of 0.5 mM) for 4 h and then harvested via centrifugation at 6000g for 15 min at 4°C, resuspended in cold lysis buffer (50 mM Tris-HCl pH 7.5, 0.25 M NaCl, 5 mM imidazole), and disrupted by sonication on ice. The E. coli proteins were precipitated by heating the cell lysates at 65°C for 20 min and then separating the soluble proteins from the debris in a 10-min centrifugation step at 8000g. The clear supernatant was filtered through 0.45-m pore size filters (Millipore, Bedford, MA) and incubated with resins containing nickel ions (Ni 2ϩ ). Recombinant PilF and its site-specific mutants carrying the hexahistidine tag were purified via Ni-affinity chromatography (Novagen) and eluted in 250 mM imidazole. Fractions containing the desired protein were identified by means of SDS-PAGE on a 12% gel and concentrated via ultrafiltration through a pore membrane with a 30-kDa cutoff (Amicon Ultra-15, Millipore).
Transmission Electron Microscopy of T4P-Thermus cells were grown on TM agar plates for 2 days and then directly resuspended on a copper grid (300 mesh) containing a drop of PBS buffer. After three washes in drops of PBS buffer, the grids were stained with 2% uranyl acetate for 3 min (the excess stain was drawn off with filter paper) and then dried in vacuo overnight. The cells were then observed using a 120-kV transmission electron microscope (JEOL JEM 1400) equipped with a Gatan UltraScan 4000 CCD camera. Images were taken at a magnification of 25,000ϫ with a defocus of ϳ1.5 m.
Biofilm Formation Assays-Static biofilms of Thermus spp. that developed on microtiter plates or glass slides were characterized as described previously (14). The inoculated broth cultures were grown to stationary phase in 5 ml of TM medium at 70°C and then used to prepare serial dilutions in fresh medium down to A 600 ϭ 0.006. Aliquots of the diluted cells were dispensed into each of the wells of a 96-well plate. A set of eight wells filled only with 300 l TM medium served as the control in each plate. The plates were incubated at 70°C for 24 h without shaking, after which 0.1% crystal violet was added to each well for 10 min at room temperature. Unbound crystal violet stain was removed and the wells were washed gently three times with distilled water. The bound crystal violet in each well was solubilized via the addition of 300 l of 95% ethanol. A microtiter plate reader (MRX Revelation, Dynex Technologies, Denkendorf, Germany) was used to read the optical density at 595 nm. For analyses of biofilm architecture, glass slides were placed into 12-well microtiter plates containing diluted cells at A 600 ϭ 0.06. The plates were incubated at 70°C for 3 h. The slides were then gently taken out, rinsed once in PBS buffer (pH 7.5) to remove nonadherent planktonic cells, air-dried overnight, and processed for scanning electron microscopy (described in the supplementary "Materials and Methods" section).
Quantitative Measurement of EPSs-The sugar content of the biofilms was estimated via the phenol-sulfuric acid method (47). T. thermophilus HB27 strains were grown on TM agar plates at 70°C for 2 days. The bacterial culture on each agar plate was scraped into 2 ml of PBS buffer, separating the soluble EPS dissolved in the PBS buffer by means of centrifugation at 10,000g for 10 min. The EPS concentration was determined using a phenol-sulfuric acid assay, with dextrans as the standard. The total sugar content was normalized to the bacterial concentration. The EPS preparation was then dialyzed overnight with water for subsequent use in GC-MS analyses.

RESULTS
Phosphoproteome of Exponentially Growing T. thermophilus HB27-To identify phosphorylated proteins in extremely thermophilic eubacteria, a phosphoproteomic analysis of T. thermophilus HB27 was carried out using the gel-free shotgun LC-MS/MS approach (28 -31). Extracts of trypsin-digested proteins from exponentially growing T. thermophilus HB27 were fractionated via SCX. The resulting 15 SCX peptide fractions, including the flow-through, were subjected to phosphopeptide enrichment using TiO 2 and then analyzed five times on a high-accuracy LTQ-Orbitrap MS. Ninety-three unique phosphopeptides were identified from 53 proteins, with an estimated false-positive rate of less than 1% (Table I). Among the identified phosphopeptides, the locations of 67 nonredundant phosphorylation sites were determined with the highest level of confidence (class I sites: 99% peptide identification confidence (48)): 38 (56.7%) on serine, 24 (35.8%) on threonine, and 5 (7.4%) on tyrosine. The relative Ser/Thr/Tyr phosphorylation ratio was similar to that of other bacterial phosphoproteomes (28,29,31,33,35). Annotated MS/MS spectra of all identified phosphopeptides and a detailed list of the identified phosphoproteins are provided in supplemental Fig. S1 and supplemental Table S1, respectively.
Classification of the Identified Phosphoproteins-The identified phosphoproteins in T. thermophilus HB27 were categorized according to their cellular localization and biological function (Fig. 1), as determined using the databases noted in "Materials and Methods." Accordingly, as shown in Fig. 1A, 37 phosphoproteins could be assigned to the cytoplasm, 4 to the ribosome, 2 to the cytoplasmic membrane, and 1 to the nucleoid. Of the 53 phosphoproteins identified in this study, 44 (86.8%) could be functionally annotated, and 7 remained hypothetical proteins and were therefore placed within the "Unknown" category. From the biological process perspective, the 46 phosphoproteins were mostly involved in metabolic pathways, including amino acid (22.6%), carbohydrate (18.9%), and nucleotide (13.2%) metabolism (Fig. 1B). These results are consistent with those of other studies, and the functional distribution was similar to that determined for other bacteria, in which proteins involved in housekeeping path-ways and central metabolisms are significantly overrepresented (28 -30, 32, 33, 49).
Phosphorylation of Thr 368 and Ser 372 Residues of the ATPbinding Motif-containing Protein PilF-In many species of bacteria, biofilm formation and natural competence have been shown to depend on the retractable protein appendages of T4P, located on the cell surface (19,24,60,61). One of the phosphoproteins identified in this analysis was PilF, which contains an ATP-binding motif (TTC1622/pilF). PilF is essential for correct T4P synthesis and, through its ATPase activity, drives the natural transformation system, as shown in several species of bacteria (59). This protein is phosphorylated at two sites of phosphorylation, Thr-368 and Ser-372, located on the two individual phosphopeptides RGGGRLEDpTLVQSGK and RGGGRLEDTLVQpSGK (Table I). The MS/MS spectra of these peptides, indicating the exact sites of their phosphorylation, are shown in Fig. 2. In a previously described sequence alignment of ATPase families within type II and type IV secretion systems (59), the double phosphorylation sites of PilF were localized to the third conserved GSPII (general secretion pathway) motif at the N terminus of the protein. This domain is crucial for normal protein functioning in type II and IV secretion systems (62,63).
Effects of Site-directed Mutagenesis of PilFs on Colony Morphology and Piliation-In order to elucidate the role of Thr-368 and Ser-372 phosphorylation in PilF, plasmids expressing wild-type or site-specific mutagenized PilFs (T368V, T368D, S372A, S372E, and the double mutants T368V/S372A and T368D/S372E) under the control of the constitutively active pilF promoter were generated and individually transformed into T. thermophilus HB27. A transformant containing the unmodified vector (pMEK2) was used as the control strain. All of the transformants were examined for twitching motility, Sites of phosphorylation with 99% peptide identification confidence assigned by MaxQuant software are indicated in bold: pS, phosphoserine; pT, phosphothreonine; pY, phosphotyrosine. Ambiguous sites are indicated in bold with brackets as p͓S͔, p͓T͔, and p͓Y͔. Detailed information on the detected phosphopeptides can be found in supplemental Table S1. discernible in T. thermophilus HB27 by the formation of large, flat, spreading colonies on agar surfaces (20). Interestingly, the results showed that the colonies formed by the control strain and by mutants expressing a nonphosphorylatable form of PilF (i.e. T368V, S372A, and T368V/S372A) were large, flat, and spreading, whereas the colonies of mutants with mimic phosphorylated forms of PilF (T368D, S372E, and T368D/ S372E) and the strain overexpressing wild-type PilF (PilF wt) were smaller (Fig. 3A). Moreover, both PilF wt and the mutant strains T368D, S372E, and T368D/S372E, carrying phosphorylation mimic PilFs, lost their twitching motility and exhibited pili-formation-deficient phenotypes (Fig. 3C). When the strains were grown in liquid medium, the nonphosphorylated mutant strains (S372A and T368V/S372A) had a lower gravity sedimentation rate, indicating that the mutants form pili and have higher twitching motility (Fig. 3B).
To further examine the stoichiometry of phosphorylation in the PilF wt strain, the relative quantities of the dephosphorylated phosphopeptides (RGGGRLEDTLVQSGK and GGGRLEDTLVQSGK) in phosphorylated and phosphatasetreated PilF were measured via mass spectrometry, and the areas of the respective peaks were compared. The results showed 63% phosphorylation site occupancy in cells grown to mid-exponential phase (supplemental Table S5).
PilF ATPase activity powers the mechanical force necessary for T4P pilus extension; conversely, following a reduction in ATPase activity, T4P function is interrupted (64). We found that both the PilF wt strain and the serial mutant PilF strains retained weak ATPase activity, as measured by ␥-phosphate release from ATP hydrolysis (supplemental Fig. S2). Also, the secondary and quaternary structures of these strains were similar with respect to their circular dichroism spectra and the results of analytical ultracentrifugation analyses, respectively (supplemental Figs. S3 and S4 and supplemental Table S6). Thus, all of the recombinant proteins formed hexameric complexes, and there was no disruption of their ATPase activity. These results implied that seryl/threonyl phosphorylation of PilF is essential to pili formation and pili-mediated twitching motility but does not determine or alter the conformation, quaternary assembly, and ATPase activity of the protein.
Effects of Site-directed Mutagenesis of PilFs on Biofilm Formation and EPS Production-Apart from twitching motil- ity, T4P is known to involve in the formation of complex colonial structures in biofilms (60). In a previous study, we demonstrated the ability of T. thermophilus HB27 to form biofilms (14). Thus, in the present work we further investigated the effect of PilF phosphorylation on biofilm-formation ability, using crystal violet staining and scanning electron microscopy. The results showed significant reductions (p Ͻ 0.05) in the biofilm-formation ability of strains expressing nonphosphorylatable forms of PilF (T368V, S372A, and T368V/S372A) relative to the PilF wt strain or the strains with phosphorylation simulated PilFs (T368D, S372E, and T368D/S372E), in which biofilm formation was normal (Fig. 4A). Additionally, strains T368V, S372A, and T368V/S372A had a lower initial attachment rate during biofilm formation (supplemental Fig. S5), and their cell surfaces had a granular appearance (supplemental Fig. S6), unlike the thick and smooth matrix that normally coats the bacterium and that was also seen in the transformants with normal biofilm production. These differences in biofilm formation and the nature of the cell surface led us to further explore EPS production and the maintenance of biofilm structure (7,8). In an EPS quantitative assay, the PilF mutant strains T368V, S372A, and T368V/S372A produced less EPS than strains S372E, T368D, and S372E/T368D with the mimic phosphorylated PilF, strain PilF wt, or vector control (Fig. 4B). The relative quantities of EPS measured via GC-MS were similar to the amounts determined using the phenolsulfuric acid method (supplemental Table S7). In terms of sugar composition, the glycosyl residues of the EPS produced by all of the examined strains consisted of 75% galactose, 15% mannose, and 7% glucose. These results indicated a relationship between PilF phosphorylation and EPS production in regulating biofilm formation in thermophilic bacteria. DISCUSSION To date, global phosphoproteome analyses from archaea (31) to eubacteria (28 -30, 32, 33) have identified quite a few phosphoproteins. Despite the presence of the same phosphosites, there is very little overlap between these phosphoproteins and thus between the respective phosphoproteomes (49). In a comparison of phosphoproteomes within the same species (i.e. T. thermophilus strains HB27 (this study) and HB8 (35)), ϳ45% of the phosphoproteins were found to be identical (supplemental Table S8), and, with the exception of valyl-tRNA synthetase (TTC0805) and chaperonin GroEL (TTC1714), the phosphorylation sites of these proteins were the same. The remaining 55% of the phosphoproteins of these two closely related thermophiles were distinct, perhaps accounting for strain-specific differences in regulatory mechanisms and biological functions.
Protein Phosphorylation Regulates T4P-mediated Functions in T. thermophilus HB27-In this study, the phospho- protein PilF (TTC1622) was shown to have two phosphorylation sites, Thr-368 and Ser-372, located on distinct monophosphopeptides (Table I and Fig. 2). In many species of bacteria, PilF is essential to T4P-mediated natural transformation, which in T. thermophilus HB27 is required for DNA uptake from thermal environments and has allowed thermoadaption (17)(18)(19). Furthermore, PilF forms hexameric complexes at the bacterial inner membrane and is the only ATPase driving T4P-mediated functions (59). This study is the first to describe PilF phosphorylation and thus to suggest a mechanism for its regulation with respect to T4P.
Phosphorylated PilF Interferes with Pilus Formation and Retards Motility in T. thermophilus HB27-In a previous study, it was found that long pilus structures are still formed but competence is defective in T. thermophilus with a deleted pilF gene (65). We found that overexpression of wild-type PilF resulted in a nonpiliated phenotype (Fig. 3C, PilF wt). This inhibition of pilus formation was likely due to the higher levels of phosphorylated PilF, as the phosphorylation stoichiometry of strain PilF wt was 63% higher than that of the control (vector) strain (supplemental Table S5). Interestingly, natural transformation ability was nonetheless maintained in strain PilF wt, as evidenced by transformation frequencies of 10 Ϫ4 to 10 Ϫ5 transformants per viable count. This result is consistent with that of the previous report (20).
In terms of the role of PilF phosphorylation, an interesting point is that the phosphorylation-simulated mutants including single or double acidic residue substituted PilFs (T368D, S372E, and T368D/S372E) all displayed nonpiliated and nontwitching phenotypes (Fig. 3). Conversely, in these nonphosphorylated mutants (T368V, S372A, and T368V/S372A), pili were formed and, at the same time, twitching motilities were retained (Fig. 3). In other words, the nonphosphorylation of PilF might possess a functional pilus apparatus because of its twitching motility. Likewise, in our results, natural competence was independent of piliation and preserved in all of the transformants at frequencies similar to those of the wild-type control (ϳ10 Ϫ4 to 10 Ϫ5 transformants per viable count) (20). Additionally, neither ATPase activity nor hexameric complex formation was lost in any of the mutated strains, including PilF wt (supplemental Figs. S2-S4). Taken together, these findings suggest that PilF functions regulated by serine/threonine phosphorylation are required for T4P formation and T4Pdriven motility but not for natural transformation and ATPase activity. Accordingly, the two independent phosphorylation sites, Thr-368 and Ser-372, in PilF may individually have the same regulatory ability in the switch from the assembly to the retraction of T4P.
Phosphorylated PilF Increases EPS Production and Thus Enhances Biofilm Formation-T4P is a multifunctional protein, involved not only in natural competence but also in adherence, movement, biofilm formation, and virulence in a wide variety of bacterial species, including Myxococcus xanthus, P. aeruginosa, Neisseria spp., E. coli, and Vibrio cholerae (60,66,67). In T. thermophilus HB27, T4P was shown to be important in natural transformation (17)(18)(19), adhesion, and twitching motility (20). In this study, we examined not only the regulation of these T4P-mediated functions by PilF phosphorylation but also the role of T4P-mediated motility in promoting the initiation of biofilm formation. Our results suggested that the retarded motility exhibited by mutants carrying the mimic phosphorylated PilF proteins (T368D, S372E, and T368D/S372E) enhanced biofilm formation (Fig. 4A). This observation is in contrast to reports on P. aeruginosa (68) and M. xanthus (69) in which bacterial movement by twitching motility and social motility, respectively, was shown to facilitate surface attachment, microcolony formation, and therefore biofilm maturation. Surface-attached cells were quantified by solubilizing the dye in ethanol and determining the absorbance at 600 nm. Data are the means of three independent experiments, with each sample tested in triplicate. The asterisks (*) indicate statistical significance (p Ͻ 0.05). B, the total EPS content extracted from T. thermophilus HB27 strains expressing wild-type or mutant PilF proteins normalized to each of their bacterial concentrations, as determined in a phenol-sulfuric acid assay. Error bar represents the standard deviations of three independent measurements. Asterisks indicate statistical significance (p Ͻ 0.05).
In our mutants, it is tempting to consider that phosphorylated PilFs enhance biofilm formation by interfering with pili formation, which limits motility and thereby increases initial adhesion (supplemental Fig. S5).
In M. xanthus, the membrane accumulation of pilin subunits (PilA) was shown to alter EPS production and social motility (70), whereas in our T. thermophilus HB27 mutants the expression of mimic phosphorylated or phosphorylatable PilFs influenced the degree of EPS production and thus of biolfim formation (Fig. 4B). According to Cava et al. (71), the composition of polysaccharides detected in T. thermophilus cell wall is 74% reducing sugars. In addition, previous studies indicated that in T. thermophilus HB27, the overexpression of GalE, which participates in the Leloir pathway for galactose metabolism, leads to an increase in biofilm formation (14). In this study, these mutant strains also incorporated a higher proportion (ϳ75%) of galactose into the EPS on their cell wall (supplemental Table S7). These findings are consistent with the results of our study, in which galactose was the major component extracted from EPS and its relative amount was positively related to effects on the ability to form biofilm. In terms of the relationship between bacterial phosphorylation and EPS biosynthesis, current evidence suggests that either deletion of specific bacterial tyrosine kinases or mutation of the phosphorylation sites on EPS biosynthetic proteins results in the loss of EPS production, especially in pathogens (32,72,73). Consequently, in T. thermophilus HB27, phosphorylated PilF might influence sugar incorporation into the cell wall and thereby biofilm production capacity.
T4P-driven motility is typically powered by the two-protein systems ATPase PilB/PilF (pilin polymerase) and PilT/PilU (pilin depolymerase), controlling pilus fiber extension and retraction, respectively (64,67,74). This is the case in mesophilic bacteria such as P. aeruginosa (64), Neisseria spp. (75), and M. xanthus (74), whereas in T. thermophilus HB27 there is only one T4P-related ATPase, PilF (17,59). Although we did not carry out structure-function analyses of T. thermophilus PilF, this protein is unusual because although its primary sequence is closely related to PilB/PilF, it is functionally closer to PilT/PilU (59). How Thermus cells coordinate pilus fiber retraction and extension using only one ATPase and the mechanism by which protein phosphorylation coordinates these functions in thermophilic bacteria are challenging questions to be addressed in future studies.
The phosphoproteins identified in our global phosphoproteome analysis of T. thermophilus HB27 were shown to be involved in a diverse range of cellular processes. One such phosphoprotein, PilF, powers the multiple functions of T4P in this bacterium, including the acquisition of DNA for transformation, in addition to playing an essential role in thermoadaptation. PilF phosphorylation was shown to be related to the sugar composition of the EPSs, which are secreted into the cell wall to assist in cell attachment preceding biofilm formation. Our results provide insights into the mechanisms under-lying the well-regulated balance in thermophiles between phosphorylation-regulated T4P and EPS production.
Acknowledgments-LTQ-Orbitrap data were acquired at the Academia Sinica Common Mass Spectrometry Facilities located at the Institute of Biological Chemistry. Additional technical assistance was provided by the NRPGM Core Facilities for Proteomics and Glycomics, located at the Institute of Biological Chemistry, Academia Sinica. We thank the Cryo-EM Common Facility of the Scientific Instrument Center, Academia Sinica, for transmission electron microscopy services and Tzu-Chi University (Hualien, Taiwan) for carrying out scanning electron microscopy.