Identification of a Putative Acetyltransferase Gene, MMP0350, Which Affects Proper Assembly of both Flagella and Pili in the Archaeon Methanococcus maripaludis

Abstract

Glycosylation is a posttranslational modification utilized in all three domains of life. Compared to eukaryotic and bacterial systems, knowledge of the archaeal processes involved in glycosylation is limited. Recently, Methanococcus voltae flagellin proteins were found to have an N-linked trisaccharide necessary for proper flagellum assembly. Current analysis by mass spectrometry of Methanococcus maripaludis flagellin proteins also indicated the attachment of an N-glycan containing acetylated sugars. To identify genes involved in sugar biosynthesis in M. maripaludis, a putative acetyltransferase was targeted for in-frame deletion. Deletion of this gene (MMP0350) resulted in a flagellin molecular mass shift to a size comparable to that expected for underglycosylated or completely nonglycoslyated flagellins, as determined by immunoblotting. Assembled flagellar filaments were not observed by electron microscopy. Interestingly, the deletion also resulted in defective pilus anchoring. Mutant cells with a deletion of MMP0350 had very few, if any, pili attached to the cell surface compared to a nonflagellated but piliated strain. However, pili were obtained from culture supernatants of this strain, indicating that the defect was not in pilus assembly but in stable attachment to the cell surface. Complementation of MMP0350 on a plasmid restored pilus attachment, but it was unable to restore flagellation, likely because the mutant ceased to make detectable flagellin. These findings represent the first report of a biosynthetic gene involved in flagellin glycosylation in archaea. Also, it is the first gene to be associated with pili, linking flagellum and pilus structure and assembly through posttranslational modifications.

Although originally believed to be limited to eukaryotes, the addition of carbohydrate to protein is a posttranslational modification now known to be utilized in prokaryotes as well. While biochemical and genetic studies conducted with organisms such as Campylobacter jejuni (43, 50, 51) have provided a well-characterized model for N- and O-linked glycosylation pathways in bacteria, similar investigations of the archaeal equivalents have only recently been reported (35, 59).

The number of confirmed and putative glycoproteins reported in archaea is significant and is based on evidence that ranges from glycoprotein staining, unusual sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) migration, glycosylation inhibition, and lectin binding to completely characterized glycan structures and linkages (reviewed in reference 14). The first prokaryotic glycoprotein to be described in detail was the surface (S)-layer protein from the halophilic archaeon Halobacterium salinarum (32, 33). It was found to contain both N- and O-linked saccharides. Since then, evidence of glycosylation in the S-layer proteins of other organisms such as Haloarcula japonica (56), Haloferax volcanii (48), Methanococcus voltae (55), Methanothermus spp. (8, 23), Staphylothermus marinus (40), and Sulfolobus spp. (17) has established these proteins as the most widely recognized glycoproteins in archaea.

In addition to S-layer proteins, the structural proteins (flagellins) that make up the archaeal flagellum have also been reported to be glycosylated. Although similar from a functional perspective, the archaeal flagellum differs significantly from its bacterial counterpart in structure and assembly (reviewed in reference 37). Bacterial flagellin proteins can be glycosylated, but thus far the linkage of the glycan is limited to O linkage (30), not including the sheath proteins found to be associated with the periplasmic flagella of spirochetes (15, 27). The bacterial flagellar filament is assembled by passing structural proteins through a central channel in the flagellum via a specialized type III secretion system, thus elongating the structure at the distal end (31). The archaeal flagellum, on the other hand, does not have a central channel large enough for this assembly pathway to occur (12, 54). Thus, components are apparently incorporated at the base of the growing structure, as in many pilus systems and as first postulated by Jarrell et al. (19). Closer resemblances of archaeal flagella to type IV pilus systems rather than to bacterial flagella have been noted by several groups (18, 39, 54). Putative glycosylation has been reported in the flagellins of many archaeal species mainly based on glycan-detecting stains, deglycosylation, and glycan inhibition studies (14). Definitive evidence for flagellin glycosylation, however, has been found in H. salinarum (47, 58) and M. voltae (55), in which the structures of the attached glycans have been determined. A trisaccharide [β-ManpNAcA6Thr-(1-4)-β-Glc-pNAc3NAcA-(1-3)-β-GlcpNAc] was shown by mass spectrometry to be attached to all four of the flagellin proteins of M. voltae. In contrast to bacterial glycosylated flagellins, the structure in M. voltae is N linked, as is also true of the glycan of H. salinarum. In M. voltae and H. salinarum, the same N-linked saccharide attached to the flagellins was also found on the S-layer proteins, suggesting a common glycosylation process for the two proteins.

In archaea, little insight into the mechanism of the biosynthesis, assembly, and attachment of O-linked glycans is available and only recently have some steps in the pathway of N-linked glycosylation been elucidated (59). In eukaryotes and bacteria, N-linked glycosylation is localized to the endoplasmic reticulum or cytoplasmic membranes, respectively (reviewed in reference 50). Nucleotide-activated monosaccharides are sequentially added to a membrane-embedded lipid carrier by using a number of glycosyltransferases. After assembly, the glycan is transported across the membrane via a flippase before being transferred to an asparagine residue located within an Asn-X-Ser/Thr consensus sequence (where X is any amino acid except proline) by using an oligosaccharyltransferase (STT3 homolog). The similarity of this process between eukaryotes and bacteria suggests that N-linked glycosylation in archaea may also occur in a comparable manner. Work conducted on flagellin and S-layer proteins of M. voltae has identified two enzymes involved in the assembly and attachment of the N-linked glycan (10, 44). A glycosyltransferase, AglA, was necessary for the addition of the terminal sugar of the N-linked trisaccharide, while AglB, homologous to the STT3 oligosaccharyltransferase, was needed for transfer of the completed glycan to the protein. Subsequent inactivation of these genes also demonstrated that at least a portion of the glycan was necessary for proper flagellum assembly and function (10). Recently, another gene (Mv1751, aglH) was identified as the glycosyltransferase responsible for the attachment of the first (linking) sugar of the glycan (N-acetylglucosamine) in M. voltae due to its sequence similarity to alg7 in eukaryotes and its ability to complement a conditional lethal alg7 mutant in yeast (44). Studies conducted with H. volcanii have also identified an archaeal STT3 homolog (AglB) and two glycosyltransferases (AglD, AglE) involved in S-layer glycosylation (1-3). As for the enzymes involved in the actual biosynthesis of the sugars that compose the N- or O-linked glycans, there has been little research reported. Only recently have Namboori and Graham (35) biochemically identified numerous enzymes involved in acetamido sugar biosynthesis in Methanococcus, specifically, enzymes involved in the biosynthesis of N-acetylglucosamine and derivatives found in the glycans attached to flagellins and S-layer proteins, as well as methanogenic cofactor B.

In addition to the presence of flagella on archaeal cell surfaces, the occurrence of thinner filaments has long been observed in many archaea (13, 21), including M. voltae (25) and Methanococcus maripaludis (36). The study of the structure, assembly, and genetics of these filaments has only recently gotten under way (36, 49). The requirement of posttranslational modifications for the assembly of these structures remains unknown.

In this study, for the first time, a deletion of a gene involved in the biosynthesis, rather than the assembly, of the sugars of the glycan N linked to flagellins is reported. Besides affecting flagellation, surprisingly, this gene also plays a role in anchoring the pilus structures to the cell surface, indicating a link between flagellum and pilus posttranslational modifications.

MATERIALS AND METHODS

Microbial strains and growth conditions.

M. maripaludis strain Mm900 was grown at 30°C in McCas medium (34) under an atmosphere of CO₂-H₂ (20:80). Neomycin (final concentration of 1 mg/ml), puromycin (2.5 μg/ml), and 8-azahypoxanthine (240 μg/ml) were added for selection when required. For complementation, nitrogen-free minimal medium (7) was used and supplemented with sterile, anaerobic NH₄Cl (10 mM) or l-alanine (10 mM) as a nitrogen source when necessary. Escherichia coli DH5α, grown at 37°C in Luria-Bertani medium, was used for plasmid cloning, and ampicillin (100 μg/ml) was used for selection when necessary. E. coli strain BL21(DE3)/pLysS was used for the overexpression of FlaB1.

Construction of the deletion plasmid.

The primers and plasmids used in this study are listed in Table 1. Construction of the in-frame deletion of MMP0350 was performed as described previously (11), with primers 157_2kb_up and 157_Start_Del to amplify the upstream fragment and 157_Stop_Del and 157_2kb_down to amplify the downstream fragment relative to the targeted gene. The pCRPrtNeo plasmid harboring the MMP0350 in-frame deletion fragment was designated pKJ645.

TABLE 1.

Primers and plasmids used in this study

Primer or plasmid	Sequencea (5′-3′) or features	Restriction site, source, or reference
Primers
157_2kb_up	CGGATCCTCTGGTATATATTCATATGG	BamHI
157_Start_Del	TGGCGCGCCTTTCCACATGTGCGGTAGGA	AscI
157_Stop_Del	TGGCGCGCCGATTAATATGATACCTATTG	AscI
157_2kb_down	CGGATCCGGTTATGAACCGGAATTGGA	BamHI
157_Del_For	TTTAAAACGTCACTTTCTTCG
157_Del_Rev	CAATCATTCCGCTTTTTAAC
157_For_Nsi	CCAATGCATGGGTAGTTATCAGGCACATCC	NsiI
157_Rev_Xba	GCTCTAGATTAGTGATGGTGGTGATGATGATCCCCCAATAAATCATGATC	XbaI
B1_For	GGAATTCCATATGAAATAACAGAATTCATG	NdeI
B1_Rev	CCGCTCGAGTTGAAGTTCTACAAGTG	XhoI
FlaE_For	GGAATTCCATATGGCATTATCATCAATATTACTCG	NdeI
FlaE_Rev	CCGCTCGAGAATCCCATAGTATTGCTCTGCG	XhoI
Plasmids
pCRPrtNeo	hmv promoter-hpt fusion + Neor cassette in pCR2.1TOPO; Ampr Kanr	27
pKJ645	pCRPrtNeo with in-frame deletion of MMP0350	This study
pHW40	nif promoter-lacZ fusion + Purr cassette; Ampr	J. Leigh
pKJ656	pHW40 with C-terminally His-tagged MMP0350	This study
pET23a	Expression vector; Ampr	Novagen
pKJ557	pET23a with C-terminally His-tagged flaE	This study
pKJ584	pET23a with C-terminally His-tagged flaB1	This study

^aThe restriction sites indicated are underlined.

Construction of complementation plasmid.

The MMP0350 gene was amplified by PCR with primers 157_For_Nsi, incorporating an NsiI restriction site, and 157_Rev_Xba, which integrated a C-terminal histidine tag and an XbaI restriction site. The gene was ligated into NsiI/XbaI-digested pHW40 (obtained from J. Leigh) and transformed into E. coli. The pHW40 plasmid containing the MMP0350 gene was designated pKJ656. MMP0350 was under the control of the nif promoter, and an intermediate level of expression was obtained by supplementing nitrogen-free medium with l-alanine, while repression of expression was obtained by the addition of NH₄Cl (28, 29).

Generation of M. maripaludis mutants.

Markerless in-frame deletions were constructed with a double-recombination event as demonstrated previously (34). Potential deletion mutants were screened by PCR with primers 157_Del_For and 157_Del_Rev. With these primers, amplification from mutant cells generated a 396-bp fragment, as opposed to a 933-bp fragment in wild-type cells. Proper in-frame deletion was confirmed by sequencing the 396-bp PCR product, as well as by Southern blot analysis.

Southern blot analysis.

Chromosomal DNA was extracted from cultures of wild-type and mutant M. maripaludis as previously described (16) and digested with the restriction enzyme DraI. Southern blotting conditions were as previously described (52). A digoxigenin-labeled probe was generated by PCR amplification across the in-frame deletion with primers 157_Del_For and 157_Del_Rev and digoxigenin-dUTP (Roche Molecular Biochemicals).

Overexpression of FlaB1 and FlaE.

A C-terminally histidine-tagged version of the M. maripaludis flaB1 gene was amplified with primers B1_For and B1_Rev, which incorporated NdeI and XhoI restriction sites, respectively. The gene was ligated into the multiple cloning site of expression vector pET23a, creating plasmid pKJ584. This plasmid was transformed into E. coli strain BL21(DE3)/pLysS, and expression of FlaB1 was induced at log-phase growth by the addition 0.4 mM isopropyl-β-d-thiogalactopyranoside, followed by a further 2-h incubation.

Similarly, a His-tagged version of FlaE was generated with primers FlaE_For and FlaE_Rev and the PCR product was cloned into pET23a to create pKJ557. Following transformation into E. coli and subsequent induction of protein synthesis, the His-tagged FlaE protein was purified via Ni affinity chromatography and with an antigen for antibody generation in chickens. Antibodies were produced by RCH Antibodies, Sydenham, Ontario, Canada.

Immunoblotting.

Whole-cell lysates or a crude membrane fraction of M. maripaludis were electrophoresed through a 15% acrylamide gel (26) and then transferred to an Immobilon-P membrane (Millipore, Bedford, MA) as previously described (53). Immunoblots were developed with chicken polyclonal M. voltae anti-FlaB2 (shown to cross-react with M. maripaludis flagellins) or anti-M. maripaludis FlaE as the primary antibody and horseradish peroxidase-conjugated rabbit anti-chicken immunoglobulin Y (Jackson ImmunoResearch Laboratories, West Grove, PA) as the secondary antibody. Blots were developed with a chemiluminescence kit (Roche Molecular Biochemicals) according to the manufacturer's instructions.

Electron microscopy.

M. maripaludis whole-cell preparations or centrifuged cell supernatants were negatively stained with 2% phosphotungstic acid (pH 7.0) and transferred to Formvar-coated gold grids. Whole cells were washed in 50 mM MgSO₄ prior to adsorption onto the grids. Cell supernatants were prepared by pelleting 3 ml of cells by centrifugation (3,000 × g, 15 min, Sorvall RC2-B) and removing the top 1 ml of supernatant. The material in the supernatant was pelleted (90,000 × g, 1 h, 4°C, Beckman TL-100) and washed in water. Following another centrifugation (90,000 × g, 1 h, 4°C, Beckman TL-100), pelleted material was resuspended in 3 μl of water and transferred to the grid. Grids were viewed on a Hitachi H-700 electron microscope operating at 75 kV.

RESULTS

Generation of an in-frame deletion of MMP0350.

It has not been previously shown that M. maripaludis utilizes the N-linked glycosylation system. However, studies conducted with a related species, M. voltae, have shown that a trisaccharide composed of an acetylated mannuronic acid with attached threonine, a diacetylated glucuronic acid, and an N-acetylglucosamine residue [β-ManpNAcA6Thr-(1-4)-β-Glc-pNAc3NAcA-(1-3)-β-GlcpNAc] is N linked to the flagellin and S-layer proteins (55). The flagellin proteins of M. maripaludis migrated on SDS-PAGE at a molecular weight larger than that predicted by the amino acid sequence, which suggested that a glycan may be present in this organism as well. All three of the M. maripaludis flagellins contain the conserved Asn-X-Ser/Thr sequon required for the attachment of the N-linked glycan, and preliminary mass spectroscopy data indicate that all three of the flagellin proteins do, in fact, have an attached N-linked glycan. The total glycan mass detected on M. maripaludis flagellins is significantly larger than that of the 779-Da glycan of M. voltae. While the complete structure remains to be elucidated, it is known that the first two sugars in the chain are acetylated (an N-acetylated hexosamine and a diacetylated glucuronic acid; S. Logan, D. VanDyke, and K. F. Jarrell, unpublished data). To identify genes involved in the biosynthesis of these sugars, the M. maripaludis genome was scanned for genes annotated as acetyltransferases. The gene MMP0350 was found to possess a hexapeptide repeat motif characteristic of a number of transferases. The strongest similarities found in BLAST searches conducted with the MMP0350 sequence were to known acetyltransferases, and thus, this gene was considered a reasonable candidate to carry out an acetylation step in the biosynthesis of M. maripaludis glycan precursors. To test this possibility, MMP0350 was targeted for in-frame deletion. The vector pCRPrtNeo, carrying a 2-kb fragment containing an in-frame deletion of MMP0350 (pKJ645), was transformed into M. maripaludis strain Mm900. Colonies were eventually picked from plates containing 8-azahypoxanthine and screened by PCR to identify strains harboring the desired deletion. Mutants were identified as strains yielding a PCR product of 396 bp, compared to the wild-type size of 933 bp (Fig. 1A). Deletion of the targeted gene was further confirmed by Southern blot analysis with a probe hybridizing to a 1,194-bp DraI fragment from the genomic DNA of the deletion mutant, compared to a 1,744-bp DraI fragment from the genomic DNA of wild-type cells (Fig. 1B).

FIG. 1.

(A) Discrimination between M. maripaludis Mm900 (wt) and the MMP0350 deletion mutant (Δ) by PCR with primers 157 Del_For and 157 Del_Rev. M, marker representing a 100-bp DNA ladder (New England BioLabs). (B) Discrimination between M. maripaludis Mm900 (wt) and the MMP0350 deletion mutant (Δ) by Southern blot assay of DNA digested with DraI.

MMP0350 deletion decreases flagellin molecular weight.

To determine if the loss of MMP0350 had an effect on flagellin proteins, whole-cell lysates of the MMP0350 deletion strain and the Mm900 wild-type strain were subjected to immunoblotting with antiflagellin antisera (Fig. 2). Compared to wild-type flagellins, there was a significant decrease in flagellin molecular mass in ΔMMP0350 mutant cells. Since the antisera used in the immunoblot assays cross-reacts with all of the M. maripaludis flagellins, analysis of the Western blot data indicates that, as in M. voltae, interference at any step in the N-linked glycosylation pathway results in the same defect in glycosylation in all flagellins. To get an estimate of the degree of underglycosylation of the flagellins in the mutant, a His-tagged version of one of the major flagellins, FlaB1, was expressed in E. coli. E. coli does not glycosylate methanococcal flagellins, and the cross-reacting band detected by immunoblotting with antiflagellin antisera represents completely nonglycosylated flagellin proteins. This protein, however, is not identical to mature nonglycosylated methanococcal FlaB1, as the E. coli-produced version retained the 12-amino-acid signal peptide that M. maripaludis flagellins are initially synthesized with (E. coli lacks the specific peptidase required to remove the unusual signal peptide), as well as the His tag at the C terminus. These additional amino acids cause E. coli-generated FlaB1 to migrate as a slightly larger protein than would be expected of native, unglycosylated FlaB1 flagellin. Figure 2 shows that the MMP0350 deletion caused the flagellin proteins from this mutant to migrate slightly farther than E. coli-overexpressed FlaB1, demonstrating that the mutant flagellin proteins are likely lacking most, if not all, of the N-linked glycan.

FIG. 2.

Western blot assay of whole-cell lysates of M. maripaludis Mm900 cells (wt), MMP0350 deletion-containing cells (Δ), and E. coli cells expressing FlaB1 flagellin proteins (Ec). The blot was developed with antibodies raised against M. voltae FlaB2, which cross-react against M. maripaludis flagellins. The values on the left are molecular masses of marker proteins in kilodaltons.

MMP0350 deletion affects flagellar filament assembly.

When wild-type and MMP0350 deletion-containing cells were examined for motility by phase-contrast microscopy, it was observed that wild-type cells were weakly motile, consistent with the formal description of the species (22), while mutant cells were completely nonmotile. To observe the effect that MMP0350 deletion had on flagellar filament structure, wild-type and ΔMMP0350 and ΔflaK mutant cells were examined by electron microscopy (Fig. 3). FlaK is the preflagellin peptidase found in M. maripaludis and is required for flagellin processing prior to their assembly into flagellar filaments (6). Cells lacking this enzyme are nonflagellated (5). Therefore, an M. maripaludis mutant carrying an in-frame deletion of the flaK gene (S. Y. M. Ng, D. J. VanDyke, B. Chaban, and K. F. Jarrell, unpublished data) was used as a nonflagellated control cell. Observation by electron microscopy showed that, compared to wild-type cells, in which the presence of multiple flagella was apparent (Fig. 3A), the MMP0350 mutant did not show any indication of flagellar filaments either attached to the cell or free on the grid (Fig. 3C). This was comparable to flaK deletion-containing cells which, as expected, also did not possess flagella (Fig. 3B).

FIG. 3.

Electron micrographs of (A) Mm900 cells, (B) ΔflaK mutant cells, (C) ΔMMP0350 mutant cells, and (D) pili isolated from supernatant of ΔMMP0350 mutant cells. Arrows indicate pilus structures. Samples were negatively stained with 2% phosphotungstic acid (pH 7.0). Scale bars = 250 nm.

Deletion of MMP0350 also affects pilus anchoring.

Electron microscopy examination of ΔMMP0350 and ΔflaK mutant cells to view the status of flagellar filaments also led to some notable observations about the cell surface structures believed to be pili. Wild-type M. maripaludis cells were found to possess both flagellar filaments and thinner pilus structures (Fig. 3A). The flaK mutant, which has lost the ability to assemble flagellar filaments, still possessed pili on the cell surface (Fig. 3B). The MMP0350 mutant, however, lacked both flagella and pili (Fig. 3C). These observations were made on cultures following overnight incubation with shaking at 110 rpm. To determine if the ΔMMP0350 mutant cells could not make pili or could not attach them properly to the cells, cultures of ΔMMP0350 and ΔflaK mutant cells were grown overnight with shaking and the culture supernatants were examined by electron microscopy. The supernatant from ΔMMP0350 mutant cells grown under shaking conditions contained apparently fully assembled pilus structures (Fig. 3D). There were no flagellar filaments detected, however, indicating that the absence of flagella in these cells was not due to a similar loss of the structure into the medium. No pilus structures were detectable in the supernatants of ΔflaK mutant cells (not shown), verifying that the pili remained attached to the surface of these cells under these growth conditions. These data indicate that although the MMP0350 mutant cannot assemble flagellar filaments, it can still assemble pili. However, these pili differ from wild-type pilus structures as they are poorly attached to the cell surface. When ΔMMP0350 mutant cells were grown statically, they were observed to retain some pili on their surface (data not shown).

Mutant complementation restores pili.

A C-terminally histidine-tagged version of MMP0350 was expressed from a plasmid under the control of the regulatable nif promoter in ΔMMP0350 mutant cells in an effort to restore function. In nitrogen-free medium supplemented with l-alanine (where intermediate levels of transcription from the nif promoter occur [28]), it was found that the cells regained the ability to maintain pili on the cell surface under shaking conditions (Fig. 4A). Indeed, the complementation under these conditions sometimes resulted in more pili per cell than observed on the flaK mutant (compare Fig. 4A and C). This same restoration of pili was not observed when cells were grown in nitrogen-free medium supplemented with NH₄Cl (Fig. 4B), conditions under which transcription from the nif promoter is repressed. Interestingly, complementation of MMP0350 was not successful in restoring flagellation. When intermediate expression of MMP0350 occurred, flagellar filaments were not detected by electron microscopy. However, further examination indicated that, over the course of the complementation experiment, ΔMMP0350 mutant cells stopped making amounts of any of the three flagellin proteins that could be detected by immunoblotting (Fig. 5A). Western blot assays developed with antisera against FlaE, the gene for which is cotranscribed with those for the flagellin proteins, demonstrated that, besides the three flagellins, the MMP0350 mutant did not synthesize detectable levels of FlaE either, suggesting that transcription of the entire flagellar operon was shut down (Fig. 5B). These results were also found in the complementation strain grown in nitrogen-free medium supplemented with alanine (data not shown). Interestingly, this effect was not observed in the case of the nonflagellated flaK mutant, as both flagellin and FlaE were detected by immunoblotting. Due to their lack of synthesis, the ability of the complemented MMP0350 gene to restore the flagellins to the wild-type molecular mass could not be assessed. However, it seems likely that the inability of the complemented cells to assemble flagella is due to the mutant cells' inability to synthesize sufficient flagellin protein rather than to the inability of the reintroduced MMP0350 gene product to compensate for the MMP0350 gene deletion.

FIG. 4.

Electron micrographs of ΔMMP0350 mutant cells containing pKJ656 under conditions where (A) there is intermediate expression of MMP0350 (growth in the presence of l-alanine as the sole nitrogen source) or (B) expression of MMP0350 is repressed (growth in the presence of NH₄Cl as the sole nitrogen source). (C) ΔflaK mutant cells showing the wild-type level of piliation without interference of flagella are shown for comparison. Samples were negatively stained with 2% phosphotungstic acid (pH 7.0). Scale bars = 250 nm. Arrowheads and arrows indicate pilus structures.

FIG. 5.

Western blot analysis of membrane fractions of Mm900 cells (lanes 1 and 4), ΔflaK mutant cells (lanes 2 and 5), and ΔMMP0350 mutant cells (lanes 3 and 6) developed with (A) anti-M. voltae flagellin FlaB2 or (B) anti-M. maripaludis FlaE. The values on the left of each panel are molecular masses in kilodaltons.

DISCUSSION

In this study, a gene involved in sugar biosynthesis has been identified that affects M. maripaludis flagellin glycosylation, flagellum assembly, and pilus anchoring. This is the first gene found to affect pili in any archaeon. This is also the first report of a deletion of a gene involved in the biosynthesis of the monosaccharide components of the glycan, as the only reported genes affecting N-linked glycosylation thus far have been glycosyltransferases and oligosaccharyltransferases responsible for glycan assembly and transfer to the protein (1, 3, 10, 44).

The related species M. voltae has flagellin proteins with an attached N-linked glycan that is necessary for proper flagellum assembly (10). The structure of the M. voltae N-glycan has been elucidated, and it has been shown to have a mass of 779 Da (55). Preliminary mass spectrometry analysis of M. maripaludis flagellins has indicated that a larger glycan (Logan et al., unpublished) is attached. The first two sugars are known to be acetylated, and therefore, it is predicted that there must be acetyltransferases responsible for the addition of these groups to the monosaccharides. MMP0350 shows sequence similarity to known acetyltransferases and could be involved in attaching the acetyl groups to one or more of the monosaccharides.

Flagellins in the MMP0350 mutant migrate as though they are lacking most, if not all, of the normally attached glycan, compared to the unmodified preflagellin control that was expressed in E. coli. With respect to the proposed N-linked assembly pathway in archaea (10), it is possible that incorrect synthesis of one or both of the first two N-acetylated sugars of the glycan could result in improper recognition by enzymes involved in the pathway and, therefore, cessation of glycan assembly at an early step. We speculate, given the molecular mass shift observed in the flagellins of the MMP0350 mutant, that this gene product is most likely involved in the acetylation of the second sugar, a diacetylated glucuronic acid. This would result in either no glycan or, at most, a glycan consisting of only a single sugar, an N-acetylhexosamine, being attached to the protein. It has recently been shown in C. jejuni that mutation of the acetyltransferase PglD, required for biosynthesis of bacillosamine, results in mutants with proteins lacking the N-linked glycan (24). It is apparent by electron microscopy observation that the glycan in M. maripaludis is indeed serving a functional role when it comes to the assembly of flagellar filaments. The MMP0350 mutant has underglycosylated flagellins which, as in M. voltae (10), cannot be assembled into flagellar filaments.

The MMP0350 deletion also resulted in a lack of pili present on cells that were agitated during incubation. These pili were found as intact structures in the medium of shaken cultures. This suggests that the pili can be made but are improperly anchored to the cell surface. The pilus structures observed in archaea have never been characterized, and indeed, the occurrence of several types of pili, as found within the domain Bacteria, is a definite possibility. The data suggest that a glycosylation step is involved in the anchoring of the pili to the cell in M. maripaludis. It is not known whether these structures contain glycosylated proteins, although pilins in other systems such as the type IV pilins of Pseudomonas aeruginosa 1244 (9) and Neisseria meningitidis (46) are known to be O glycosylated. A number of mutations have been made in N. meningitidis (20, 41, 42) that have affected pilin glycosylation, as detected by shifts in SDS-PAGE migration. However, the interference with glycosylation was not shown to affect pilus morphology or cell adherence. In P. aeruginosa 1244, a pilO glycosyltransferase mutant deficient in pilin glycosylation was also found to assemble normal-appearing pili, although it did exhibit reduced twitching motility (45). The finding that in M. maripaludis a deletion affecting the N-linked glycosylation state of the flagellin proteins also causes a change in the pilus phenotype provides evidence that glycosylation plays an important role at some step in pilus assembly and/or attachment. Since the acetyltransferase gene deletion could affect early steps in the biosynthesis of sugars that could ultimately enter either the N- or O-linked glycosylation pathway, it is not known whether the glycosylation defect in the case of pili is N linked, as in the flagellin proteins. Observation of complete pilus filaments in the medium of shaken cultures implies that, unlike in flagella, glycosylation is required not for filament assembly but rather for proper anchoring to the cell. This may be the role of a minor pilin structural component which requires glycosylation in order to function.

The shedding of assembled pilus filaments into the supernatant has been previously observed in several different bacterial pilus systems. In E. coli Pap pili, deletion of a minor pilin-like protein, PapH, caused a similar shedding, indicating its role of anchoring the structure to the outer membrane (4). In P. aeruginosa type IV pili, a mutation in the ATPase pilU gene caused a hyperpiliated phenotype with fragile pili that were often observed as clumps separated from the cell (57). A hyperpiliated phenotype, however, is clearly not observed with the MMP0350 mutant in M. maripaludis. Improper attachment of pili to the cell wall has been demonstrated in gram-positive organisms. Nobbs et al. (38) demonstrated that pilus-like structures in Streptococcus agalactiae were much more weakly anchored when either one of two sortase enzymes was deleted. Sortases are transpeptidases that cleave and allow bonds between the protein and cell wall. M. maripaludis, however, lacks a classic bacterium-like cell wall and has only an S-layer located external to the cytoplasmic membrane. It is plausible that the glycosylation of a specific anchoring pilin or cell membrane protein is needed for proper anchoring of pili to the cell and that interference with glycan assembly in the MMP0350 mutant leads to this protein being unable to carry out this necessary function.

Immediately following the isolation of a mutant containing the MMP0350 in-frame deletion, lower-molecular-weight flagellins were detected by immunoblotting. However, long-term cultivation of the mutant resulted in a lack of detectable flagellins and FlaE, another flagellar operon protein. While this may be explained by the cells somehow sensing that the flagellins are not being incorporated into filaments and therefore shutting off the synthesis of the structural proteins, other mutants unable to assemble flagella, such as the flaK mutant, continue to make flagellins at least to a level detectable by immunoblotting techniques. Interestingly, we have observed this shutdown of the flagellar operon by other mutations that also affect glycosylation (unpublished data). Complementation of a MMP0350 deletion strain with MMP0350 expressed from a plasmid was thus unable to restore flagellation, although restoration of the pilus attachment was observed.

Collectively, we have presented the first genetic data on a biosynthetic gene involved in flagellin glycosylation in archaea. It is also the first gene knockout to affect pili in this domain, linking flagellum and pilus structure and assembly through posttranslational modifications. These data further expand on the importance of glycosylation in archaeal surface structures, which now include S-layers, flagella, and pili.