ABSTRACT
The structure of pili from the archaeon Methanococcus maripaludis is unlike that of any bacterial pili. However, genetic analysis of the genes involved in the formation of these pili has been lacking until this study. Pili were isolated from a nonflagellated (ΔflaK) mutant and shown by sodium dodecyl sulfate-polyacrylamide gel electrophoresis to consist primarily of subunits with an apparent molecular mass of 17 kDa. In-frame deletions were created in three genes, MMP0233, MMP0236, and MMP0237, which encode proteins with bacterial type IV pilin-like signal peptides previously identified by in silico methodology as likely candidates for pilus structural proteins. Deletion of MMP0236 or MMP0237 resulted in mutant cells completely devoid of pili on the cell surface, while deletion of the third pilin-like gene, MMP0233, resulted in cells greatly reduced in the number of pili on the surface. Complementation with the deleted gene in each case returned the cells to a piliated state. Surprisingly, mass spectrometry analysis of purified pili identified the major structural pilin as another type IV pilin-like protein, MMP1685, whose gene is located outside the first pilus locus. This protein was found to be glycosylated with an N-linked branched pentasaccharide glycan. Deletion and complementation analysis confirmed that MMP1685 is required for piliation.
Members of the archaeal domain are widespread in nature, inhabiting some of the most extreme environmental niches (e.g., hydrothermal vents, soda lakes, and hot springs) found on Earth, as well as a large variety of habitats considered “nonextreme” (e.g., the ocean and the soil) (14). Archaea have been shown to possess a variety of unusual surface structures, some of which appear to have no bacterial counterpart (3, 47). Furthermore, there is evidence from studies of numerous archaea showing that many diverse surface structures are composed of proteins made with type IV pilin-like signal peptides and processed by a type IV prepilin peptidase homologue. Thus far, flagella, certain pili, Iho670 fibers, and the proposed bindosome all share this feature (3, 46, 47).
The archaeal flagellum is the best-studied cell surface organelle in the domain Archaea; it is a ubiquitous swimming apparatus among the motile archaeal members (11, 31, 48). Though superficially resembling the bacterial flagellum, the archaeal flagellum more closely resembles the bacterial type IV pilus system while also possessing certain archaeon-specific traits (48). The archaeal flagellar filament is architecturally similar to the bacterial type IV pilus (18, 19, 63, 64) and is composed of multiple flagellin subunits, except in the case of Sulfolobus, where a single flagellin gene is found (59). Prior to incorporation into the growing flagellar filament, the flagellins undergo two known posttranslational modification steps: (i) N-linked glycosylation (15, 16, 57, 76) and (ii) cleavage of the N-terminal signal peptide by the preflagellin peptidase (4, 9, 10). Though direct proof is lacking, circumstantial evidence—including the lack of a hollow central channel like that found in bacterial flagella (18, 63, 64)—suggests that the archaeal flagellum is assembled by incorporating new flagellin subunits at the base (31, 32).
Very recently, a number of studies of a variety of archaeal genera have presented the first data on pili—structures that have been observed in electron micrographs of different archaea for many years (22, 33, 38, 55, 73). It was shown in Methanothermobacter thermoautotrophicus that the pili are composed of a 16-kDa protein encoded by the mth60 gene (60). This study presented the first evidence of a role for archaeal pili in adhesion. In both Methanococcus and Sulfolobus, there is clear evidence of a type IV pilus-like locus, with two or three pilin-like genes encoding proteins with type IV pilin-like (class III) signal peptides (51) and processed by a prepilin peptidase. In Sulfolobus solfataricus, this enzyme, PibD, is the same one that processes the flagellins, pilins, and sugar binding proteins with class III signal peptides (3, 78). Evidence has been presented that in Haloferax volcanii, the PibD equivalent also processes flagellins and pilin-like proteins (65). However, in M. maripaludis, there is a dedicated second peptidase, EppA, for pilin processing that is separate from the flagellin-processing enzyme FlaK (58). In S. solfataricus, transcription of the pilus locus is strongly upregulated, with the number of pili on the surface of the cells greatly enhanced, upon treatment of cells with UV light (26-28). Recently, it was shown that deletion of a typical type IV pilus-associated ATPase gene within the pilus locus of this organism leads to nonpiliated cells (28). The pili have been shown to lead to cellular aggregation and to be necessary for surface adherence (28, 78).
In the case of Methanococcus, where the flagella greatly outnumber the pili on the cell surface (38), study of pili became much easier once genetic techniques (45) allowed the generation of flagellumless mutants that left pili as the sole surface appendages. This was recently made possible through the generation of various in-frame deletions in genes of the fla operon (17), as well as in flaK (essential for preflagellin processing and subsequently flagellar synthesis) in Methanococcus maripaludis (49). The ΔflaK mutant is nonmotile and nonflagellated; observations by electron microscopy revealed that this strain possesses only pili as surface appendages. When M. maripaludis pili were isolated and examined by electron cryomicroscopy, they were found to have a structure unlike that of any other known bacterial pili, with two subunit packing arrangements found to coexist within the same filament (72). In silico analysis and signal peptide processing assays were used to identify a locus in M. maripaludis that contained several genes encoding proteins with predicted and, in some cases, demonstrated class III signal peptides that were suggested to be likely candidates for pilus structural proteins (58). In addition, a number of other proteins were identified from the FlaFind PERL program analysis of the genome as likely possessing class III signal peptides. Here, to complement our earlier structural data (72) and the in silico data (58), we present the first genetic analysis of the locus proposed to be involved in the biogenesis of these pili. We report that the pili are composed mainly of glycoprotein subunits with apparent molecular masses of 17 and 15 kDa as determined by migration on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and demonstrate by deletion analysis and complementation in trans that two of the three predicted pilin-like genes (58) are essential for piliation. However, the gene encoding the major structural pilin was, surprisingly, not found within this genetic locus but was identified by mass spectrometry (MS) analysis of purified pilin protein as the product of the MMP1685 gene, which is located at a distinct locus within the genome separate from the other pilin-like genes.
MATERIALS AND METHODS
Microbial strains and growth conditions.M. maripaludis Δhpt (Mm900) (45) and ΔflaK (49) mutants and all of the subsequent mutants derived from them were routinely grown in Balch medium III (8) as previously described (36). Methanoculleus marisnigri was grown in Balch medium III statically at room temperature. For the in-frame deletion mutagenesis experiments with M. maripaludis Mm900, cells were grown in McCas medium supplemented with 8-azahypoxanthine (240 μg/ml) or neomycin (1 mg/ml) as required at various steps in the transformation procedure outlined by Moore and Leigh (45). In complementation experiments, the plasmids were maintained in M. maripaludis by puromycin (2.5 μg/ml) selection. Escherichia coli strain DH5α (Novagen) was used for intermediate cloning steps. Cells were grown at 37°C in Luria-Bertani medium supplemented with ampicillin (100 μg/ml) when necessary.
In-frame deletions of MMP0237, MMP0236, MMP0233, and MMP1685 in M. maripaludis.An MMP0237 in-frame deletion strain was created according to the methodology of Moore and Leigh (45). Briefly, primers (Table 1) were designed to PCR amplify the approximately 1-kb upstream and downstream flanking sequences from the M. maripaludis genomic DNA separately. The reverse primer from the upstream amplification and the forward primer from the downstream amplification had AscI sites incorporated so that the PCR products could be digested with AscI and ligated together. Another round of PCR with the forward primer from the upstream amplification and the reverse primer from the downstream amplification allowed the amplification of a 2,532-bp fragment containing an in-frame deletion within the MMP0237 gene, leaving 57 amino acids of the original protein's 151 amino acids after accounting for the insertion of the AscI site. This product was cloned into pCRPrtNeo (45) via the flanking BamHI sites, creating pKJ660. Sequencing of the cloned fragment was performed to ensure that the subsequent deletion would be in frame. The Mm900 ΔflaKΔ0237 mutant was made by transforming the Mm900 ΔflaK mutant with pKJ660 as described by Tumbula et al. (66). All of the steps were performed anaerobically. Briefly, 5 ml of freshly grown Mm900 ΔflaK mutant cells was washed and resuspended in transformation buffer (50 mM Tris-HCl, 0.35 M sucrose, 0.38 M NaCl, 1 mM MgCl2, 0.00001% resazurin, pH 7.5). The resuspended culture was mixed with 5 μg of DNA in the presence of polyethylene glycol (18%, wt/vol). The transformation mixture was grown overnight without selection and then subcultured into 10 ml medium with neomycin (for selection for vector integration) and incubated at 37°C while shaking for 48 h. The culture was then used to inoculate McCas medium without neomycin to allow a second recombination event to remove the vector. At this stage, a recombination event that removes the plasmid insertion can occur either on the same side of the deletion as the first event, returning the chromosome to its wild-type sequence, or on the opposite side of the first event, creating a deletion.
Primers and plasmids used in this study
The culture was plated on McCas agar containing 8-azahypoxanthine, which would be lethal to any cells that retained the vector-borne hpt gene. The plates were incubated at 37°C in an anaerobic canister (CO2-H2 pressurized with 3 ml 25% sodium sulfide). Single colonies were picked after 5 days and screened by PCR and Southern blot analysis to identify deletion mutants.
An in-frame deletion of MMP0236 was created by the same methodology, leaving 37 of the original protein's 132 amino acids after accounting for the insertion of the AscI site. For MMP0233, after insertion of the AscI restriction site, 42 of the original protein's 129 amino acids remained. For MMP1685, after insertion of the AscI site, 29 of the original protein's 74 amino acids were left. Details of the primers and plasmids used for these strain creations are summarized in Table 1.
PCR screening of deletion mutants.PCR primers were designed to provide a quick PCR screen of the transformants for gene deletion (Table 1). For MMP0237, primers were designed to give 703- versus 413-bp products using wild-type versus mutant cells as the template, respectively. For MMP0236, primers were designed to give 477- versus 185-bp products using wild-type versus mutant cells as the template, respectively. For MMP0233, primers were designed to give 699- versus 433-bp products using wild-type versus mutant cells as the template, respectively. For MMP1685, primers were designed to give 692- versus 547-bp products using wild-type versus mutant cells as the template, respectively. Overnight cultures of transformants grown from single colonies on 8-azahypoxanthine plates were washed twice with 2% NaCl and used directly in a 2% NaCl suspension as templates for PCR. The PCR was performed with Taq DNA polymerase with the following protocol: 95°C for 5 min; 29 cycles of 94°C for 45 s, 45°C for 45 s, and 72°C for 1 min; and a final cycle with an extension time of 10 min at 72°C. The PCR products were electrophoresed on 0.8% agarose gels for analysis alongside a 100-bp ladder marker (New England Biolabs, Ipswich, MA). For MMP0233, MMP0236, and MMP0237, the gene deletions were confirmed by Southern blotting experiments (below). However, since the MMP1685 gene was so small, additional PCRs were done to show that the gene was deleted. Forward and reverse primers were designed for a portion of the MMP1685 gene that was deleted in the mutant. This forward primer was used with the initial screening reverse primer, while the reverse primer was used with the initial screening forward primer, using either wild-type or MMP1685 deletion mutant strain cells as the template.
Southern blot analysis of the deletion mutants.To confirm the MMP0237 gene deletion in the genome of the M. maripaludis ΔflaKΔ0237 mutant strain, PCR primers were used to amplify an ∼350-bp fragment across the deletion region using M. maripaludis genomic DNA as the template. The amplified product was subsequently used to generate a digoxigenin (DIG)-labeled probe using a DNA labeling kit (Boehringer Mannheim, Mannheim, Germany). The probe as designed would detect, in a Southern blot, a 2.2-kb fragment that would result from EcoRV-digested wild-type genomic DNA, whereas with the gene deletion, a 1.8-kb fragment would be bound. The same probe was used to probe the MMP0236 deletion; it yields a 1.9-kb fragment in the M. maripaludis ΔflaKΔ0236 deletion mutant.
For the M. maripaludis ΔflaKΔ0233 mutant, the MMP0233 screening primers were used to generate the DIG-labeled probe. DIG-labeling was performed as described above; the probe was designed to detect a 2.3-kb fragment from AseI-digested wild-type genomic DNA versus a 2.0-kb fragment in the deletion mutant.
Southern blotting was carried out as previously described (61). For all three deletion screens, the same hybridization temperature of 50°C was used.
Cloning of M. maripaludis pilin-like genes for complementation into the corresponding deletion strains.The expression vector pWLG40 (41) was used for cloning of MMP0237, MMP0236, MMP0233, and MMP1685 for complementation in the M. maripaludis ΔflaKΔ0237, ΔflaKΔ0236, ΔflaKΔ0233, and ΔflaKΔ1685 mutant strains, respectively. The genes of interest were amplified by PCR with forward and reverse primers that had NsiI and MluI restriction sites incorporated, respectively. PCRs were performed using Vent DNA polymerase under the following conditions: 95°C for 5 min; 29 cycles of 94°C for 45 s, 50°C for 45 s, and 72°C for 1 min; and a final cycle with an extension time of 10 min at 72°C. The amplified products were purified using a PCR purification column (Qiagen, Chatsworth, CA) and cloned into pWLG40 under the control of the constitutive hmv promoter via the NsiI and MluI restriction sites, generating pKJ661 (for MMP0237), pKJ666 (for MMP0236), pKJ690 (for MMP0233), and pKJ880 (for MMP1685). These plasmids were transformed into the appropriate deletion strains for complementation testing by using the same transformation methodology as outlined above. The primers and plasmids used for all complementations are listed in Table 1.
Purification of pilus filaments from the M. maripaludis ΔflaK mutant.Pili were purified from the M. maripaludis ΔflaK mutant, a nonflagellated but piliated mutant of strain S2, using the protocol described previously (72). Protein samples were stained with the PAS reagent to detect glycoproteins (23). A whole-cell lysate of M. marisnigri, which has a glycoprotein S layer of 138 kDa known to be stained by PAS (77), was included on the same gel as an internal control.
SDS-PAGE and in-gel proteolytic digestion.Protein samples were subjected to SDS-PAGE as previously described (40). Silver staining (2) and Coomassie staining (24) were carried out as previously described. Gel bands containing approximately 1 to 2 μg of pilin protein each were excised and digested either with trypsin or with endoproteinase AspN. Tryptic digestion was carried out in the presence of Rapigest in accordance with the manufacturer's recommended procedure (Waters, Milford, MA). The protein bands for AspN in-gel digestion were first reduced and alkylated in accordance with standard protocols. The gel bands were vacuum dried, and 250 ng of AspN (Roche) in 20 μl of 50 mM sodium phosphate, pH 8.0, was added to them. Sufficient buffer was added to the vials to cover the bands, and digestion was carried out overnight at 37°C.
MS analysis of intact pilin protein.The intact purified pilin protein was analyzed using the nonaqueous electrospray ionization (ESI)-MS method as previously described (39). Briefly, the protein solution/suspension was evaporated to complete dryness on a Savant centrifugal evaporator and resuspended in concentrated formic acid (10 μl). The protein was solubilized by the addition of hexafluoroisopropanol (90 μl). All mass spectra were acquired on a Q-TOF2 hybrid quadrupole time-of-flight mass spectrometer (Waters). The nonaqueous pilin solutions were infused into the nanoelectrospray interface at 1 μl/min, and spectra were recorded from m/z 800 to m/z 2,500 (one acquisition per second). Protein molecular weight profiles were derived from the spectra using MaxEnt (Waters). Top-down tandem MS (MS/MS) analysis was performed on the protein multiply charged ions as first described by Schirm et al. (56) and used recently by us to profile the flagellar glycan modifications on Clostridium botulinum (67). The resulting MS/MS spectra were examined by hand for amino acid sequence information, which was then searched using the BLAST algorithm (6) against the M. maripaludis genome to identify the gene of origin.
MS analysis of pilin digests.Tryptic and AspN in-gel digests were analyzed by nano-liquid chromatography (nLC)-MS/MS using a NanoAquity UPLC system (Waters) coupled to a Q-TOF Ultima hybrid quadrupole time-of-flight mass spectrometer (Waters). The digests were injected onto a Symmetry C18 trap (180 μm by 20 mm, 5-μm; Waters) and separated on a BEH130C18 column (100 μm [inside diameter] by 10 cm, 1.7-μm; Waters) using the following gradient conditions: 5% to 45% acetonitrile (ACN) in 0.2% formic acid in 35 min and 45% to 95% ACN in 5 min. The mass spectrometer was set to automatically acquire the MS/MS spectra of the three most abundant multiply charged ions in each survey scan. MS/MS spectra were searched against the NCBI nonredundant database, as well as a custom database containing putative protein sequences derived from the M. maripaludis S2 genome using MASCOT 2.0.1 (Matrix Science, United Kingdom). Glycopeptide MS/MS spectra were interpreted by hand.
The AspN pilin digest was also examined by nano-ESI (nESI)-MS/MS on an LTQ XL linear ion trap mass spectrometer (ThermoFisher Scientific). The pilin AspN digest was desalted using a C18 ZipTip (Millipore) prior to analysis. The ZipTip was first prepared by washing it in 100% ACN and then in 0.5% acetic acid (AcOH). The digest solution was adjusted to ∼0.5% AcOH and aspirated a number of times through the ZipTip. The ZipTip was rinsed thoroughly with 0.5% AcOH, and the peptides/glycopeptides were eluted in 5 μl of 40% ACN-0.5% AcOH. The eluates were loaded into a PicoTip emitter needle (New Objective) installed in the LTQ instrument. A voltage of 2 to 2.5 kV was used to generate a stable electrospray. Glycopeptide MS/MS spectra were generated by collisional activated dissociation. Collision offsets were adjusted for each ion to achieve optimal fragmentation. The resulting MS/MS spectra were interpreted by hand.
Electron microscopy.M. maripaludis was grown overnight, and 1 ml of cells was briefly washed and resuspended in phosphate-buffered saline. Purified pili were recovered from KBr gradients by centrifugation (111,000 × g, 90 min), and the pellets were resuspended in distilled H2O. Samples were negatively stained with 2% phosphotungstic acid and supported on carbon-Formvar-coated copper grids. They were examined in a Hitachi 7000 electron microscope operating at an accelerating voltage of 75 kV.
RESULTS
The recent generation of a nonflagellated M. maripaludis mutant via in-frame deletion of the preflagellin peptidase gene flaK resulted in cells that still possessed thin appendages termed pili. Compared to the surfaces of wild-type cells that are dominated by flagella (68), electron microscopic examination of the ΔflaK cell surface revealed that several pili, located peritrichously on the cell surface, were the sole cell surface appendage type (49).
This generation of a flagellumless mutant with only pili as surface appendages made it possible to attempt the purification of the much less abundant pilus structures without potential contamination with flagella. Following a procedure developed originally to isolate flagella, pili were purified from the ΔflaK mutant (Fig. 1A). Unlike the archaeal flagellum, no anchoring structure or hook was apparent in the pilus samples. The pilus filament has a diameter of approximately 6 nm—much smaller than the flagellar filament, which is 12 nm in width. When the purified pili were subjected to SDS-PAGE analysis and stained with Coomassie blue, one major band with an apparent molecular mass of ∼17 kDa and a minor band of about 15 kDa were detected (Fig. 1B). The protein had unusual Coomassie staining properties—the protein stained poorly despite a large amount of loaded protein—a fuzzy halo band was often observed. This was similar to earlier reports on the unusual staining properties of many bacterial flagellin proteins (2). The sample was tested for possible glycosylation by PAS staining. However, the pilin bands did not stain with the PAS reagent under conditions where the glycosylated S layer of Methanoculleus marisnigri stained positively (results not shown). Attempts at N-terminal sequencing of the separated bands were unsuccessful, suggesting a blocked N terminus.
Electron microscopy and SDS-PAGE analysis of purified M. maripaludis pilus samples. (A) Purified M. maripaludis pilus filaments. The sample was negatively stained with 2% phosphotungstic acid (pH 7.0). Bar, 100 nm. (B) Purified M. maripaludis pilus filaments were separated by SDS-PAGE and subjected to Coomassie blue staining. Molecular masses (lane M contained molecular size markers) are indicated on the left. The arrows on the right point out the 17- and 15-kDa bands in lane 1 subsequently analyzed by mass spectroscopy.
Genome sequence analysis in search of type IV pilin-like signal peptides led to the identification of a number of type IV pilin-like genes, MMP0233 (encoding a protein with a predicted molecular mass of 14.4 kDa, with the signal peptide), MMP0236 (encoding a protein with a predicted molecular mass of 14.2 kDa, with the signal peptide), and MMP0237 (encoding a protein with a predicted molecular mass of 17.3 kDa, with the signal peptide), in a single gene cluster (58). In the cases of MMP0233 and MMP0237, it was also previously demonstrated that both proteins were processed by the type IV prepilin peptidase homologue EppA found in the same gene cluster, further supporting their assignment as type IV pilins (58). All three ORFs possess a characteristic type IV pilin-like signal peptide sequence, with [RK]G↓Q[VI] preceding a hydrophobic stretch (Fig. 2A), within the first 30 amino acids in the annotated protein sequence. In addition, both MMP0233 and MMP0236 contain glutamic acid at the +5 position, a highly conserved amino acid in bacterial type IV pilins. Interestingly, MMP0237 contains aspartic acid at this position.
Posttranslational modifications of M. maripaludis pilins. (A) Alignment of the N termini of the M. maripaludis putative pilins showing the cleavage sites for processing by the prepilin peptidase homolog (indicated by the arrow). (B) Presence of potential N-linked glycosylation sites (bold type) in M. maripaludis pilins.
In an attempt to link these candidate genes to the pilus structure, MMP0233, MMP0236, and MMP0237 were separately targeted for in-frame deletion in an M. maripaludis strain in which flaK had already been deleted, by the recently developed markerless mutant methodology (45). The deletions were confirmed by two independent methods, i.e., by using PCR primers designed to give size-differentiated products in wild-type versus mutant cells carrying deletion-containing versions of the gene, respectively (data not shown), and by Southern blotting using a DIG-labeled probe predicted to hybridize to size-shifted fragments in digested DNA of the wild type versus that of the mutant (data not shown). In both instances, the behavior of the mutants was as predicted.
Upon careful examination of the cell surface by electron microscopy, the ΔflaKΔ0237 and ΔflaKΔ0236 mutants with changes in both flaK and a putative pilin gene were both found to be completely nonpiliated (Fig. 3 A, C, and D), in addition to lacking flagella because of the flaK deletion. To provide further proof that MMP0236 and MMP0237 are directly involved in the formation of the pilus structure, complementation studies were performed. The MMP0236 or MMP0237 gene was cloned into pWLG40 under the control of the constitutive hmv promoter. The cloning strategy was to express the protein in its native form, without additional tags, such as a C-terminal His tag, in order to minimize the potential for interference in the interactions required of the subunits for assembly of the pilus shaft. Examination of the complemented strains by electron microscopy (Fig. 3B, C, E, and F) demonstrated that piliation was clearly restored, although the number of pili observed in both complemented strains was typically less than that observed on the ΔflaK mutant strain.
Electron microscopy of M. maripaludis pilin gene mutants and complemented strains. (A) M. maripaludis ΔflaKΔ0237 mutant showing absence of pili. (B) M. maripaludis ΔflaKΔ0237(pWLG40::0237) mutant showing presence of pili. (C) Enlargement of boxed area in panel B showing pili. (D) M. maripaludis ΔflaKΔ0236 mutant showing absence of pili. (E) M. maripaludis ΔflaKΔ0236(pWLG40::0236) mutant showing presence of pili. (F) Enlargement of boxed area in panel E showing pili. (G) M. maripaludis ΔflaKΔ0233 mutant showing presence of a few pili. (H) Enlargement of boxed area in panel G showing a couple of pili. (I) M. maripaludis ΔflaKΔ0233(pWLG40::0233) mutant showing presence of pili. (J) Enlargement of boxed area in panel I showing pili. Arrows point to pili. All samples were negatively stained with 2% phosphotungstic acid (pH 7.0). Bars, 200 nm.
In contrast to the MMP0236 and MMP0237 deletion results, deletion of MMP0233 did not completely eliminate piliation. The resulting ΔflaKΔ0233 mutant cells were generally found to possess one or two pili (Fig. 3G and H). Interestingly, when these cells were complemented with the constitutively expressed MMP0233 gene, restoration of piliation to a similar level to that seen in ΔflaK mutant cells was observed and it was not uncommon, in some complementation experiments, to observe cells that had piliation well in excess of that observed on typical ΔflaK mutant cells (Fig. 3I and J).
Since all three pilin-like genes were involved in piliation, attempts were made to identify the major structural protein(s), which proved challenging because of the nature of the pilin. Initially, we attempted to identify the protein(s) in the purified pilus preparation by in-gel tryptic digestion and analysis of peptides by nLC-MS/MS. These efforts were generally unsuccessful, although a single tryptic peptide corresponding to T54-68, SNIEILSVSGEGSYK, from MMP0236 was identified in one analysis (data not shown). Reports in the literature of the hydrophobic nature of some Gram-positive pilin proteins led us to explore alternative MS methods to characterize this recalcitrant protein. Nonaqueous ESI-MS methods facilitated the determination of the intact mass of the major pilin protein (Fig. 4A). The pilin sample was dissolved in a formic acid-hexafluoroisopropanol solution and infused into the electrospray interface of the Q-TOF2 mass spectrometer, yielding the spectrum presented in Fig. 4A. This mass spectrum was deconvoluted to obtain a molecular mass profile for the protein centered around 9,700 Da (Fig. 4A, inset). The complexity of the deconvoluted mass profile suggests extensive posttranslational modification.
Nonaqueous ESI-MS analysis of pilin protein. (A) Mass spectrum of the intact protein sample acquired on the Q-TOF2 mass spectrometer. The molecular mass profile derived from this mass spectrum is presented in the inset. (B) Top-down MS/MS spectrum of the pentuply charged (5+) protein ion at m/z 1,938.2. The amino acid sequence derived from the fragment ion series was searched using the BLAST algorithm against all M. maripaludis protein sequences in the NCBI nonredundant database and matched to hypothetical protein MMP1685. (C) Amino acid sequence of MMP1685 showing the four potential sites of N linkage (underlined). In addition, based on the masses of the fragment ions observed in the top-down MS/MS spectrum (panel B), it was determined that the N-terminal amino acid is pyroglutamic acid.
Top-down MS/MS analysis was performed on the most abundant ions in the protein mass spectrum, all of which produced a spectrum virtually identical to that presented in Fig. 4B. Manual interpretation of the fragment ion series observed in the lower half of the top-down MS/MS analysis yielded a 12-residue amino acid sequence which, when searched using the BLAST algorithm against all M. maripaludis protein sequences, was matched to a sequence from the N-terminal region of the putative protein MMP1685 (Fig. 4C). This protein is 74 amino acids long and includes an N-terminal 12-amino-acid type IV pilin-like signal peptide predicted by FlaFind (Fig. 2A) (58). The amino acid at the +1 position is Q, as found in the MMP0233, MMP0236, and MMP0237 proteins, and that at the +5 position is E, as was found in both the MMP0233 and MMP0236 proteins. MS/MS analysis of the peptide confirmed that the site of cleavage of the signal peptide was position 12 (Q at the +1 position) and identified the mature pilin N-terminal residue as a pyroglutamic acid residue (Fig. 4C), which explains the previous failure to obtain N-terminal sequence information. Furthermore, this protein has only a single lysine residue and no arginine residues, which explains the failure of attempts at identification using trypsin digestion and MS analysis.
The mature MMP1685 protein has a predicted molecular mass of 6,398 Da, which is considerably lower than that observed by mass analysis of the intact purified pilin (9,728 Da). However, it does possess four consensus sites for N-glycosylation and the intact mass analysis suggested considerable variation in modification. To investigate the apparent discrepancy between the predicted and observed molecular masses further, we examined peptides from an in-gel AspN digest of both putative pilin protein bands observed on SDS-PAGE by nLC-MS/MS on the Q-TOF Ultima mass spectrometer. The mass spectrum of the major chromatographic peak is presented in Fig. 5A and appears to be composed of a series of related glycopeptides. The MS/MS spectrum of the major doubly charged ion at m/z 1,082.41 (Fig. 5B) contains a number of b and y fragment ions, confirming that this glycopeptide originated from the C terminus of MMP1685. Furthermore, this spectrum is dominated by the same glycan fragment ions previously observed in the MS/MS spectra of N-linked glycopeptides derived from M. maripaludis flagellin (37), suggesting that this peptide is modified with a similar glycan. Interestingly, the pilin glycan possesses an additional branched hexose not found on the flagellin glycan. nESI-MS/MS of the same glycopeptide ion on the LTQ XL mass spectrometer generated less extensive fragmentation of the glycan (Fig. 5C) and suggests that the hexose is appended to the GalNAc residue linking the glycan to the asparagine side chain.
nLC-MS/MS analysis of an AspN in-gel digest of pilin protein. (A) Mass spectrum of glycopeptides observed in the pilin digest. The bulk of the glycopeptide heterogeneity observed is likely an artifact due to the loss of labile sugar moieties caused by front-end collision-induced dissociation. (B) MS/MS spectrum acquired on the Q-TOF Ultima mass spectrometer for the doubly protonated glycopeptide ion at m/z 1,082.4. The spectrum is dominated by fragmentation of the glycan moiety, which appears to be very similar to that on the flagellin proteins from this species. Peptide fragment ions were also observed in this MS/MS spectrum and were matched to the C-terminal peptide of the hypothetical protein MMP1685. (C) MS/MS spectrum acquired on the LTQ linear ion trap for the same glycopeptide ion at m/z 1,082.4. The observed glycan fragmentation patterns indicate that the hexose is attached to HexNAc as a side branch. The likely identities of the glycan components are presented in the inset in panel A. Sug, methyl(5S)-2-acetamido-2,4-dideoxy-α-l-erythro-hexos-5-ulo-1,5-pyranose.
If MMP1685 is the main structural protein of the pili, then it must be essential for piliation. Consequently, MMP1685 was targeted for deletion in a ΔflaK mutant background and mutants carrying a deletion in this gene were found (data not shown). Screening by several PCR-based methods indicated that a clean deletion mutant was obtained. The mutant with MMP1685 deleted was shown by electron microscopy to no longer produce pili at the cell surface (Fig. 6 A and B). Pilus production was restored to the ΔflaKΔ1685 mutant strain upon complementation with a plasmid-borne version of the MMP1685 gene (Fig. 6C and D).
Electron microscopic examination of M. maripaludis ΔflaKΔ1685 mutant cells carrying the vector control (A and B) and carrying the complementation construct (C and D). Panels B and D are enlargements of the boxed areas depicted in panels A and C, respectively. Samples were negatively stained with 2% phosphotungstic acid (pH 7.0). Bars, 200 nm.
DISCUSSION
While pili were observed on various archaeal species by electron microscopy as early as the 1970s, information on the function and genetics of these surface appendages remains scant. Numerous classes of bacterial pili exist (e.g., type I and IV pili, sex pili, curli, and the sortase-dependent pili of certain Gram-positive bacteria), where functions, structure, and mode of assembly can differ (7, 12, 20, 29, 43, 54). With the study of archaeal pili in its infancy, information regarding any of these features is much more limited. However, several research groups working on such diverse archaeal genera as Methanococcus (58, 72), Sulfolobus (28, 78), and Methanothermobacter (60) have begun to focus on these extracellular appendages in order to answer fundamental questions regarding these structures.
In M. maripaludis, a single locus was identified that contained three putative type IV pilin-like genes (MMP0233, MMP0236, and MMP0237) encoding proteins with typical class 3 signal peptides that are cleaved by a type IV prepilin peptidase-like enzyme, leaving mature proteins with a hydrophobic N terminus (58). It was expected that the major structural pilin(s) represented by the 17- and 15-kDa bands identified by SDS-PAGE analysis of purified pilus samples would be one of these proteins. Indeed, the initial prediction of MMP0233, MMP0236, and MMP0237 being involved in the formation of the pili (58) was confirmed when in-frame deletions of two of the putative pilin genes (MMP0236 and MMP0237) led to completely nonpiliated cells (Fig. 3) and deletion of MMP0233 led to poorly piliated cells. Piliation was restored to the cell surface of these mutants when the deleted pilin genes were complemented in trans. It was assumed that MMP0236, MMP0237, or both gene products would be the major structural pilin since both have predicted molecular masses close to the size of the major identified subunits (MMP0236 is 14.2 kDa, and MMP0237 is 17.3 kDa). Unexpectedly, however, while we did identify a single MMP0236 peptide in the purified pilus sample, the major pilin monomer was shown by MS analysis to be another, glycosylated, protein with a predicted class III signal peptide, MMP1685. The MMP1685 gene is located outside the locus containing MMP0233, MMP0236, and MMP0237. The pilin protein was shown to have an intact mass of approximately 9,700 Da and displayed unique properties which made it difficult to analyze by traditional MS methods and which may explain its aberrant migration on SDS-PAGE (Fig. 1B). Based on the intact protein's mass (9,700 Da) compared to that predicted from the mature pilin protein sequence (6,400 Da), it appears that this very small protein is glycosylated at three of the four putative N-linked sequons (Fig. 2B) with a glycan with a mass of 1,196 Da.
Type IV pilins in bacteria can be classified as type IVa and type IVb, with the two groups distinguished by several characteristics (62). Type IVa pilins (as found in Pseudomonas aeruginosa, Neisseria meningitidis, and Neisseria gonorrhoeae) have very short signal peptides of only 5 or 6 amino acids, with the mature pilins being about 150 amino acids long. The type IVb pilins (represented by the TCP pilins of Vibrio cholerae and PilS of Salmonella enterica), on the other hand, have longer signal peptides (15 to 30 amino acids) and a longer mature pilin length of about 190 amino acids. A unique subset of the type IVb pilins is represented by the Flp pilins of Aggregatibacter actinomycetemcomitans, which have signal peptides of 10 to 26 amino acids and a very short mature pilin length of only 50 to 80 amino acids. The three pilin-like proteins MMP0233, MMP0236, and MMP0237, are all somewhat smaller than the typical type IV pilins found in bacteria, at about 120 to 140 amino acids (without the signal peptide), while mature MMP1685 is only 62 amino acids long. In its size and signal peptide length, MMP1685 resembles Flp pilins, although MMP1685 lacks the Flp motif at the N terminus that includes adjacent glutamate and tyrosine residues (34). In three of the four pilins (MMP0233, MMP0236, and MMP1685), there is a +5 glutamic acid that is also almost universally found in bacterial type IV pilins, while MMP0237 has aspartic acid at this spot. We are not aware of any bacterial type IV pilins that have a +5 aspartic acid; studies that have targeted the +5 glutamic acid of bacterial type IV pilins for site-directed mutagenesis have not reported mutants harboring a change from glutamic acid to aspartic acid (1, 42, 50). Consequently, whether this difference from the other pilus systems might be significant for pilus assembly in this organism is unknown. The M. maripaludis pilins do, however, lack the two disulfide-bridged cysteines in the C-terminal segment that are also highly conserved among type IV pilins, although the newly identified type IV pilin of Clostridium perfringens is also missing these (69), as are the short Flp pilins (34). EppA (MMP0232) has been shown to be the signal peptidase that processes the prepilins MMP0233 and MMP0237 (58). A key feature that seems to be required for EppA processing of the substrates is the presence of a conserved glutamine at position +1. This feature is also found in MMP1685 (Fig. 2), indicating that this major pilin is also likely processed by EppA.
Analysis of the separated subunits of the purified pili indicated the presence of an N-linked glycan at multiple positions of the small MMP1685 protein. We have shown that the flagellins from M. maripaludis are modified with an unbranched tetrasaccharide attached at multiple N linkage sites on each of three flagellin protein subunits (37). Interestingly, the glycan on MMP1685 is identical, with the exception of an additional hexose added as a branch to the N-acetylgalactosamine linking the glycan to the asparagine side chains. It has been assumed that all surface proteins modified with an N-linked glycan would receive the same glycan, after its biosynthesis by a general glycan assembly pathway, through the action of the oligosaccharyltransferase AglB. In Methanococcus voltae, for example, the same glycan has been found on both flagellins and S-layer proteins (70). It appears that substitutions to this glycan can be added, as seen in this present study, although the process by which this occurs is currently unknown. It remains to be determined if this glycan addition is a pilin-specific process or whether this is an unexpected consequence of rendering cells unable to assemble flagella due to the initial flaK mutation. We will determine the structure of the S-layer glycan from this mutant in future studies to determine if this is a pilin-specific process. While glycoprotein staining of the major pilin subunits was negative, this was also true of the M. maripaludis flagellin glycoproteins (35, 37). Despite their small size, MMP0233, MMP0236, and MMP0237 all contain multiple N-linked sequon targets (N-X-S/T) (Fig. 2B) which could be modified with the same glycan found on MMP1685. As yet, only a single peptide of MMP0236 has been detected in pilus samples and this peptide did not contain an N-linked sequon. The presence of pilins that contain an N-linked glycan, as shown here for MMP1685, is novel; in bacteria, pilin glycosylation, when reported, has been of the O linkage type, typically to serine or threonine residues (52, 53, 71).
Of the four pilin-like genes deleted, only the deletion of MMP0233 did not totally abolish piliation. Its deletion resulted in a minimal level of piliation, while complementation resulted in cells that often exhibited increased piliation, sometimes much greater than that of the parent flaK mutant. MMP0233 may normally be present in cells in relatively small amounts. Expression of MMP0233 from the strong hmv promoter on the complementation vector may result in artificially elevated protein levels, leading to the higher pilus number upon complementation. Antibodies against MMP0233 are not available to monitor cellular protein abundance under normal and complementation conditions. The plasmid itself is reported to be a low-copy-number vector present at about three copies per genome (75). Clearly, the role of MMP0233 is not essential for piliation to occur, although its presence has a positive impact, which seems, then, to rule it out as a major structural pilin. It is common for type IV pilus gene clusters to encode multiple proteins with prepilin signal peptides, with one of them being a major structural pilin and several minor pilins which may have minor structural roles (5). In at least some systems, such as Neisseria gonorrhoeae, mutants lacking any of the pilin-like proteins have greatly reduced levels of piliation (74), as observed for MMP0233 mutants in this study. One might speculate that the role of MMP0233 is as a minor pilin involved in the initiation of pilus formation, an explanation that would be consistent with the large number of pili sometimes observed on complemented cells that might be expressing higher-than-normal levels of the protein. MMP0236 and MMP0237 could then be minor components needed for the pilus structure, with MMP1685 playing the role of the major structural component.
It is unknown how many genes might be required for the formation of archaeal pili and whether the locus containing MMP0233, MMP0236, and MMP0237 contains other genes necessary for pilus assembly, regulation, and function. One gene that is obviously lacking in this locus is an ATPase needed for assembly of the pili. Most type IV pili extend and retract, resulting in twitching motility (a flagellum-independent bacterial movement over solid surfaces), with the incorporation and removal of the subunits at the base of the structure usually being mediated by two separate ATPases (13, 25). No identifiable pilin-like ATPase gene is present in the M. maripaludis locus, although such an ATPase has been found associated with pilin genes in Sulfolobus and indeed has already been shown to be critical for pilus formation in that organism (28). MMP1685 appears to be cotranscribed with, at most, one other gene, one that encodes a possible metallophosphoesterase.
In bacteria, type IV pili can play various biological roles, depending on the environmental niche of the organism. The involvement of many type IV pili in twitching motility is well documented (44), but they can have additional roles, such as in cell-to-cell attachment and biofilm formation (30), transformation of DNA, and phage attachment (20, 21). If archaeal pili represent a true counterpart of bacterial type IV pili, they could be involved in some, or all, of these functions. Roles in aggregation and possible mating, as well as surface adherence, have already been suggested for the type IV-like pili in Sulfolobus (28, 78), but no role for pili in M. maripaludis is known.
ACKNOWLEDGMENTS
This work was funded by the National Research Council of Canada (S.M.L., J.F.K.) and by a Discovery Grant from the Natural Sciences and Engineering Research Council of Canada (to K.F.J.).
We thank Shino Mizuno and Yoshika Nosaka for technical support.
FOOTNOTES
- Received 13 July 2010.
- Accepted 30 October 2010.
- Accepted manuscript posted online 12 November 2010.
- Copyright © 2011, American Society for Microbiology
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵