Identification of Genes Involved in the Biosynthesis of the Third and Fourth Sugars of the Methanococcus maripaludis Archaellin N-Linked Tetrasaccharide

Abstract

N-glycosylation is a protein posttranslational modification found in all three domains of life. Many surface proteins in Archaea, including S-layer proteins, pilins, and archaellins (archaeal flagellins) are known to contain N-linked glycans. In Methanococcus maripaludis, the archaellins are modified at multiple sites with an N-linked tetrasaccharide with the structure Sug-1,4-β-ManNAc3NAmA6Thr-1,4-β-GlcNAc3NAcA-1,3-β-GalNAc, where Sug is the unique sugar (5S)-2-acetamido-2,4-dideoxy-5-O-methyl-α-l-erythro-hexos-5-ulo-1,5-pyranose. In this study, four genes—mmp1084, mmp1085, mmp1086, and mmp1087—were targeted to determine their potential involvement of the biosynthesis of the sugar components in the N-glycan, based on bioinformatics analysis and proximity to a number of genes which have been previously demonstrated to be involved in the N-glycosylation pathway. The genes mmp1084 to mmp1087 were shown to be cotranscribed, and in-frame deletions of each gene as well as a Δmmp1086Δmmp1087 double mutant were successfully generated. All mutants were archaellated and motile. Mass spectrometry examination of purified archaella revealed that in Δmmp1084 mutant cells, the threonine linked to the third sugar of the glycan was missing, indicating a putative threonine transferase function of MMP1084. Similar analysis of the archaella of the Δmmp1085 mutant cells demonstrated that the glycan lacked the methyl group at the C-5 position of the terminal sugar, indicating that MMP1085 is a methyltransferase involved in the biosynthesis of this unique sugar. Deletion of the remaining two genes, mmp1086 and mmp1087, either singularly or together, had no effect on the structure of the archaellin N-glycan. Because of their demonstrated involvement in the N-glycosylation pathway, we designated mmp1084 as aglU and mmp1085 as aglV.

INTRODUCTION

N-glycosylation is a prevalent protein posttranslational modification that is not limited to eukaryotes as originally thought but is also found in both prokaryotic domains, the Bacteria and Archaea (1–6). Unlike the bacterial N-glycosylation system, which is thought to be restricted to only a few bacterial subgroupings, such as Deltaproteobacteria and Epsilonproteobacteria (6), N-glycosylation is much more prevalent in the Archaea. A gene encoding an oligosaccharyltransferase, required for the transfer of the assembled glycan from the lipid carrier to the target protein, has been detected in all but two of sequenced archaeal genomes, representing all five archaeal phyla (7, 8). In addition, the N-glycosylation pathway in Archaea differs from the other domains in the diversity of N-glycan structures reported, variations in the lipid carriers and, at least in the halophiles, the versatility in the pathways used to assemble similar glycans in different species (2, 4, 9–11).

Although N-glycosylation is widely distributed in Archaea, studies combining both genetic and structural information about N-glycosylation pathways are focused mainly on a few model species (2, 9, 12), such as the mesophilic methanogens Methanococcus voltae and Methanococcus maripaludis (13–17), the thermoacidophile Sulfolobus acidocaldarius (9, 18–20), and the halophiles Haloferax volcanii and Haloarcula marismortui (2, 11, 21–26). As seen for many archaeal traits, the archaeal N-glycosylation pathway appears as a partial amalgam of the pathway found in eukaryotes and bacteria (1, 2, 27). In general, the archaeal N-glycan precursor is first synthesized onto a lipid carrier (mono- or pyrophosphorylated dolichol) by sequential addition of sugar monomers by glycosyltransferases using sugar-nucleotide donors on the cytoplasmic side of the cytoplasmic membrane. The lipid-linked oligosaccharide is then thought to be flipped across the cytoplasmic membrane by a flippase and transferred en bloc by the oligosaccharyltransferase AglB onto Asn residues within selected Asn-Xaa-Ser/Thr motifs in the target protein. To date, the identification of the putative flippases has remained elusive, except that in H. volcanii an involvement of AglR has been demonstrated in the flipping of one of the charged dolichol carriers (26).

Reporter proteins used for the genetic and structural analysis of archaeal N-glycosylation pathways are usually the S-layer protein (13, 18, 23) and archaellin (formerly archaeal flagellin [28]) (13, 16, 29), although other proteins have also been discovered to be modified with N-glycans, including pilins (15), cytochrome b_558/556 of Sulfolobus acidocaldarius (30), and even archaeal viral proteins (31).

In M. maripaludis S2, archaella are composed of three archaellin proteins: the major archaellins FlaB1 and FlaB2 form the filament, while the minor archaellin FlaB3 is responsible for the hook region (32). All three archaellins are modified by an N-linked tetrasaccharide attached at multiple sites: three sites in FlaB1, four sites in FlaB2 and two sites in FlaB3 (16). The structure of this N-glycan was determined to be Sug-1,4-β-ManNAc3NAmA6Thr-1,4-β-GlcNAc3NAcA-1,3-β-GalNAc, where Sug is [(5S)-2-acetamido-2,4-dideoxy-5-O-methyl-α-l-erythro-hexos-5-ulo-1,5-pyranose), found exclusively in this species (16). Interestingly, the N-glycan attached to the major pilin in M. maripaludis is a slightly modified pentasaccharide containing an extra hexose attached as a branch to the first sugar, N-acetyl-galactosamine (15).

In M. maripaludis S2, the oligosaccharyltransferase and glycosyltransferases responsible for the addition of the second, third, and fourth sugar residues of the N-glycan have been identified as AglB (MMP1424), AglO (MMP1079), AglA (MMP1080), and AlgL (MMP1088), respectively (17). The assembly and function of archaella were significantly affected in mutant cells carrying in-frame deletions of these genes, which all resulted in a truncated N-glycan. Archaellins modified with N-glycans lacking the last one (algL mutant) or two (aglA mutant) sugar residues are still assembled into archaella on the cell surface, although the motility of mutant cells is impaired compared to wild-type cells, with motility directly correlated to the length of the N-glycans. Further reduction in glycan size to a single sugar, as in an aglO mutant, or blocking the transfer of any N-glycan to archaellins, as in an aglB mutant, led to nonmotile, completely nonarchaellated cells. These results indicated that a minimum disaccharide N-glycan attachment to archaellins is essential for archaella assembly (17). Similarly, the addition of N-glycan onto archaellins has also been shown to be necessary for stable archaella formation and motility function in S. acidocaldarius (19) and H. volcanii (29). Interestingly, the addition of N-glycan is not required for pilins to assemble into pili in M. maripaludis, since aglB mutant cells still assemble a wild-type number of pili, even though the major structural pilin is normally a glycoprotein (17).

Besides the genes involved in the assembly of the glycan onto the lipid carrier, a number of genes involved in the biosynthesis pathways for the individual sugars of the glycan have been identified or characterized in either in vivo or in vitro studies (17, 33). Most recently, the products of the mmp1081-mmp1082-mmp1083 (aglXYZ) gene cluster were shown to be responsible for the acetamidino group modification of the third sugar (34). Several of these genes are clustered in the genome, including ones encoding enzymes involved in sugar biosynthesis (mmp1076, mmp1077, mmp1090, aglX, aglY, and aglZ) (33, 34; Y. Ding, I. Brockhausen, and K. Jarrell, unpublished data) and glycosyltransferases (aglO, aglA, and aglL [17]) (see Fig. 1). Here, we identify a four-gene operon, mmp1084 to mmp1087, located between aglZ and aglL and demonstrate an involvement of two of these genes in the modification of sugars in the N-linked glycan. We present evidence that MMP1084, which belongs to the glutamine amidotransferase class II (Gn_AT_II) superfamily, is involved in the transfer of the threonine residue linked to the third sugar of the N-glycan, whereas MMP1085 is responsible for the addition of the methyl group linked to C-5 on the fourth sugar, consistent with its predicted function as an S-adenosylmethionine (SAM)-dependent methyltransferase. Deletion of the remaining two genes, mmp1086 and mmp1087, either singularly or together, had no effect on the structure of the final attached N-glycan.

Fig 1.

Genomic area surrounding the mmp1084-to-mmp1087 region, where a number of genes involved in the N-glycosylation pathway are located. mmp1077 encodes a phosphomutase converting d-glucosamine-6-phosphate to d-glucosamine-1-phosphate. MMP1076 is a GlcN-1-P uridylyltransferase/acetyltransferase that catalyzes the acetylation of the 2-amino group of GlcN-1-P and the transfer of GlcNAc-1-P to UTP to form UDP-GlcNAc (33). aglO, aglA, and aglL encode the glycosyltransferases for the second, third and fourth sugars, respectively (17) The aglXYZ gene cluster is responsible for the acetamidino group of the third sugar (34). mmp1090 encodes UDP-Gal/GalNAc 4-epimerase (Ding et al., unpublished).

MATERIALS AND METHODS

Strains and growth conditions.

M. maripaludis S2 Δhpt (Mm900) (35) and subsequent mutants derived from Mm900 were grown anaerobically in Balch medium III (36) under an atmosphere of CO₂-H₂ (20:80) at 37°C with shaking. In the in-frame deletion mutagenesis experiments, Mm900 derived transformants were grown in McCas medium with 1 mg of neomycin/ml or 240 μg of 8-aza-hypoxanthine/ml as required at the various steps (35). In complementation experiments, M. maripaludis mutant strains harboring the appropriate complementation plasmid were grown in nitrogen-free medium supplemented with either 10 mM l-alanine or 10 mM NH₄Cl as the sole nitrogen source (37) in the presence of 2.5 μg of puromycin/ml for plasmid selection. Escherichia coli TOP10 cells used for cloning steps were grown in Luria-Bertani medium at 37°C with 100 μg of ampicillin/ml for selection when required.

RT-PCR.

To determine whether the genes mmp1084 to mmp1087 formed a cotranscribed unit, reverse transcriptase PCR (RT-PCR) was performed using primers (listed in Table 1) designed to amplify across the intergenic regions of neighboring genes. RNA template was extracted from wild-type cells using an RNeasy Minikit (Qiagen, Inc., Mississauga, Ontario, Canada) with optional DNase digestion (Qiagen) according to the manufacturer's protocol. cDNA was amplified using a One-Step RT-PCR kit (Qiagen) in accordance with the supplied protocol. In addition, PCR amplifications were performed with the same primer combinations using genomic DNA to verify amplicon size and specificity of primer pairs or the purified RNA not subjected to reverse transcription as a control to rule out possible contamination of the RNA sample with genomic DNA.

Table 1.

Primers used in this study

Technique and primer	Sequence (5′–3′)a	Restriction site
RT-PCR
1083–1084_F	GTTTTGGGTGCTGGAGAACTCG
1083–1084_R	AATTTGGCGAATACTACGAAGC
1084–1085_F	TGTATCTTCGGGATTAAATCGC
1084–1085_R	TATAACTGCACCAAATGTCTGG
1085–1086_F	ATAAGTGCTCAATTACATCAAGG
1085–1086_R	AATCTTGGAAATTGTAATGAGG
1086–1087_F	TTGGATCCAACAAAGATTCTCC	BamHI*
1086–1087_R	ATCTCACACTTTCATACCTACG
1087–1088_F	TTTCACGAGTAACTTTCCTTCG
1087–1088_R	AATCTTGGAAAACATTACTTGC
In-frame deletions/plasmid construction
1084del-up_F	ACCAAGGATCCGTAAATAACAATGTAAATGGATG	BamHI
1084del-up_R	AAAGGCGCGCCCCTATATTTCCGGCATAAAATC	AscI
1084del-down_F	AAAGGCGCGCCATGGATACCTATTATGACGCTAC	AscI
1084del-down_R	TTAAGGATCCCAAGTGCTGTTTCGAAGGATAAC	BamHI
1085del-up_F	AAAGGATCCAACACTATGAATTTTATAGAACG	BamHI
1085del-up_R	AAAGGCGCGCCATGCATGCAATAGATAATCATG	AscI
1085del-down_F	AAAGGCGCGCCCCACATATGCGTAATGTTTTA	AscI
1085del-down_R	AAAGGATCCGTGCTATAATATATCTTGAATCC	BamHI
1086del-up_F	ATTCTAGATCTAGAGTAGAACTAACTAATGATTGTAACTATCAGTG	XbaI
1086del-up_R	GCATGGCGCGCCGGTGTGTACAATATTACATTGTG	AscI
1086del-down_F	GCATGGCGCGCCCCATATTATGCTTTACTTCAATGTAGGC	AscI
1086del-down_R	ATTCTAGATCTAGATCGTGTATTTTATCGAGCCCATGAGC	XbaI
1087del-up_F	ATTCTAGATCTAGAACCACAACCTAAAGCTTCGAACATCACTGTG	XbaI
1087del-up_R	GCATGGCGCGCCCAATCATTAGTTAGTTCTACTATAGC	AscI
1087del-down_F	GCATGGCGCGCCGTGAAGCTCCATCGCCATTGTTC	AscI
1087del-down_R	ATTCTAGATCTAGAAATTCACCGCTAACATCCATGTGTG	XbaI
In-frame deletions/screening
1084seq_F	ATTTTCAGGATATTCTGTTAAAGG
1084seq_R	ACGATGCAATTAATGTGGGAGC
1085seq_F	AGTAAATAACAATGTAAATGG
1085seq_R	ATATTTTCTGTATCTTCGGG
1086seq_F	TACCTACGGCAAAATGTGAA
1086seq_R	AATTCACCGCTAACATCCAT
1087seq_F	CTCTTGTCGGGATAACATGG
1087seq_R	GTGTGGACAATACTTACATTGTG
Complementation
1084comp_F	GGCCATGCATGCCGGAAATATACCTTGAA	NsiI
1084comp_R	CCGGACGCGTTTATTTATATTTTATATCATAATATTTTTCAATGTATTC	MluI
1085comp_F	GGCCATGCATGGATGTTAGCGGTGAATTTATGAAAACTAATACCTGGTTAAACTATAAACATGATTATCTATTGCACGCAT	NsiI
1085comp_R	CCGGACGCGTTCACCGATTTGAATTTATCTCC	MluI

^aRestriction sites added to primers are underlined. *, Natural BamHI site in this primer. The abolished NsiI site after a silent T-to-C mutation is indicated in boldface.

In-frame gene deletion plasmid construction.

Plasmids used for in-frame deletions of mmp1084, mmp1085, mmp1086, and mmp1087 were constructed as previously described (35), using the primers listed in Table 1. Confirmation of the in-frame nature of each deletion was obtained by DNA sequencing. Table 2 lists the plasmids used in the present study.

Table 2.

Strains and plasmids used in this study

Strain or plasmid	Description and/or genotypea	Source or reference
Strains
E. coli TOP10	F− mcrA Δ(mrr-hsdRMS-mcrBC) ϕ80lacZΔM15 Δ lacX74 nupG recA1 araD139 Δ(ara-leu)7697 galE15 galK16 rpsL(Strr) endA1 λ−	Invitrogen
M. maripaludis S2
Mm900	Δhpt	35
Δmmp1084	Mm900Δmmp1084	This study
Δmmp1085	Mm900Δmmp1085	This study
Δmmp1086	Mm900Δmmp1086	This study
Δmmp1087	Mm900Δmmp1087	This study
Δmmp1086Δmmp1087	Mm900Δmmp1086 Δmmp1087	This study
Plasmids
pCRPrtNeo	hmv promoter-hpt fusion plus Neor cassette in pCR2.1Topo; Ampr	35
pKJ795	pCRPrtNeo with in-frame deletion of mmp1084	This study
pKJ715	pCRPrtNeo with in-frame deletion of mmp1085	This study
pKJ882	pCRPrtNeo with in-frame deletion of mmp1086	This study
pKJ783	pCRPrtNeo with in-frame deletion of mmp1087	This study
pHW40	nif promoter-lacZ fusion plus Purr cassette; Ampr	John Leigh
pKJ959	pHW40 with mmp1084 complement	This study
pKJ955	pCRII-TOPO with mmp1085	This study
pKJ958	pHW40 with mmp1085 complement	This study

^aNeo^r, neomycin resistance; Pur^r, puromycin resistance; Amp^r, ampicillin resistance; Str^r, streptomycin resistance.

Generation of M. maripaludis mutants carrying in-frame deletions in targeted genes.

Mutants carrying markerless in-frame deletions of mmp1084, mmp1085, mmp1086, and mmp1087 were generated as described previously (35). Single transformant colonies picked from McCas-Noble agar containing 8-azahypoxanthine were screened by PCR using primers (listed in Table 1) that would amplify across the target gene to identify deletion mutants. Mutants identified in this screen were restreaked, and single colonies were again screened by PCR to ensure purity. To create the mutant deleted for both mmp1086 and mmp1087, mmp1086 was deleted from the mmp1087 deletion mutant. Table 2 lists the strains generated in this study.

Complementation studies.

mmp1084 and mmp1085 were amplified using the complementation primers listed in Table 1, digested with NsiI and MluI, and ligated into the complementation vector pHW40, resulting in the inserted gene being under the control of the inducible nif promoter (37). To facilitate cloning, an internal NsiI site in mmp1085 was removed by a silent T-to-C change at nucleotide position 72 using the long forward primer 1085comp_F (Table 1). The fidelity of the cloned genes in all of the complementation plasmids (listed in Table 2) was confirmed by DNA sequencing. The complementation vectors were transformed into the corresponding deletion mutants under the selection of 2.5 μg of puromycin/ml, as described previously (37, 38). Complemented mutants were grown in the presence of puromycin in nitrogen-free medium supplemented with either l-alanine (nif promoter induced) or NH₄Cl (nif promoter repressed) (32, 37).

Western blot analysis.

M. maripaludis whole-cell lysates were subjected to SDS–15% PAGE (39) and then transferred to an Immobilon-P membrane (Millipore, Bedford, MA) (40). Major archaellin FlaB2 was detected with chicken anti-FlaB2 specific antibodies (34). Horseradish peroxidase-conjugated rabbit anti-chicken immunoglobulin Y (Jackson ImmunoResearch Laboratories, West Grove, PA) was used as secondary antibody. Blots were developed with an enhanced chemiluminescence kit (Millipore) according to the manufacturer's instructions.

Archaella isolation.

Archaella from the various single gene deletion strains, as well as the Δmmp1086 Δmmp1087 double mutant, were isolated as previously described (41).

Semisolid swarm motility plate assay.

Overnight wild-type cells and various mutants were pelleted anaerobically and resuspended in Balch III medium to a final optical density at 600 nm of 3.0. Then, 5 μl of each resuspension was inoculated into semisolid swarm plates (Balch III medium containing 0.25% agar), followed by incubation at 37°C for 5 days (17).

Mass spectrometry analysis.

Purified archaella (50 μg) from the single mutants were digested overnight with trypsin (Promega, Madison, WI) at a ratio of 30:1 (protein-enzyme [vol/vol]) in 50 mM ammonium bicarbonate at 37°C. The archaella isolated from the Δmmp1086 Δmmp1087 double mutant was first resolved by SDS–15% PAGE and then stained with collioidal Coomassie blue. The archaellin band was excised and subjected to an overnight in-gel tryptic digestion. All of the archaellin digests were analyzed by nano-liquid chromatography-tandem mass spectrometry (Nano-LC-MS/MS) using a NanoAquity UPLC system (Waters, Milford, MA) coupled to a QTOF Ultima hybrid quadrupole time-of-flight mass spectrometer (Waters). The digests were injected onto an Acclaim PepMap100 C₁₈ μ-precolumn (5 mm by 300 μm [inner diameter]; Dionex/Thermo Scientific, Sunnyvale, CA) and resolved on a 1.7-μm BEH130 C₁₈ column (100 μm by 100 mm [inner diameter]; Waters) using the following gradient conditions: 5 to 45% acetonitrile (ACN) in 0.1% formic acid for 35 min and 45 to 95% ACN for 5 min. The flow rate was 450 nl/min. MS/MS spectra were acquired on doubly, triply, and quadruply charged ions and searched against the NCBInr database using the Mascot search engine (Matrix Science, Ltd., London, United Kingdom). The spectral data sets were searched for glycopeptide MS/MS spectra, which were then interpreted by hand.

Electron microscopy.

M. maripaludis S2 wild-type and deletion mutants were grown overnight. A 1-ml portion of the cells was briefly washed and resuspended in phosphate-buffered saline. Samples were negatively stained with 2% phosphotungstic acid and supported on carbon-Formvar-coated copper grids. The samples were then examined in a Hitachi 7000 electron microscope operating at an accelerating voltage of 75 kV.

RESULTS

The genomic location of mmp1084 to mmp1087 between genes already shown to be involved in the assembly or biosynthesis of the archaellin N-linked glycan (Fig. 1) led us to examine these genes for a possible role in the N-linked glycosylation pathway.

Bioinformatic analysis of mmp1084 to mmp1087.

As a first step in obtaining information about the putative functions of MMP1084 to MMP1087, bioinformatics-based analyses were performed with these protein sequences. When MMP1084 was used as a query in a BLAST search (i.e., the Basic Local Alignment Search Tool [http://blast.ncbi.nlm.nih.gov/]) (42), the results showed significant similarities to numerous sequences, most of which are asparagine synthases from a variety of different species. BLAST analysis also indicated that MMP1084 possessed a Gn_AT_II superfamily domain that is found at the N terminus of various enzymes, including asparagine synthase. MMP1085 analysis revealed a conserved domain with specific hits to SAM-dependent methyltransferases. MMP1086 and MMP1087 share 56% identity, and both have an N-terminal radical-SAM domain with a FeS/SAM binding site, including the conserved CxxxCxxC motif, which coordinates the iron-sulfur cluster. Radical SAM enzymes are widespread and are known to be involved in such diverse processes as metabolism, DNA repair, and the biosynthesis of vitamins, coenzymes, and antibiotics. The two proteins also contain a C-terminal DUF4008 domain. DUF4008, domain of unknown function 4008, is a functionally uncharacterized domain found in bacteria and archaea. Similar conserved domains were found using another protein functional analysis server, InterProScan (http://www.ebi.ac.uk/Tools/pfa/iprscan/ [43]). The potential cellular location of the four proteins was examined using PRED-SIGNAL (http://bioinformatics.biol.uoa.gr/PRED-SIGNAL/) (44), TMHMM (http://www.cbs.dtu.dk/services/TMHMM/), and PSortb (http://www.psort.org/) (45). No signal peptides or transmembrane domains were found, and the location of the proteins was predicted by PSortb (set to archaeal proteins) to be cytoplasmic, except for MMP1086 for which a location could not be predicted. Interestingly, homologues of mmp1084 to mmp1087 are not found in the four other sequenced M. maripaludis strains (C5, C6, C7, and X1).

RT-PCR analysis of the mmp1084-to-mmp1087 gene cluster.

mmp1084 to mmp1087 are all oriented in the same direction and are separated by short intergenic regions (5, 37, and 45 bp, respectively), suggesting that they likely form a single transcriptional unit. To test this, primers were designed to amplify across the intergenic regions between adjacent two genes from mmp1083 to mmp1088, as shown in Fig. 2A. In the RNA samples extracted from Mm900 cells subjected to reverse transcription, amplification products of the predicted size linking mmp1084 and mmp1085 (563 bp), mmp1085 and mmp1086 (550 bp), and mmp1086 and mmp1087 (484 bp) were obtained, whereas, as expected, no products were amplified between mmp1083 and mmp1084 (expected size, 584 bp) or between mmp1087 and mmp1088 (expected size, 510 bp), which are genes transcribed in the opposite direction (Fig. 2B). PCRs using the same primers and the total RNA that had not undergone the reverse transcriptase reaction did not yield any amplification products, demonstrating that the RNA was free of contaminating DNA. Standard PCRs using Mm900 genomic DNA and the same primers were also conducted as positive controls to confirm the size of the amplicons and the specificity of the primers.

Fig 2.

(A) mmp1084-mmp1087 operon. The gene region surrounding mmp1084 to mmp1087, which was targeted for RT-PCR to determine potential cotranscription. Black lines below the genes represent the predicted amplicons obtained from RT-PCR. (B) RT-PCR experiment indicating cotranscription of mmp1084, mmp1085, mmp1086, and mmp1087. Standard PCRs using Mm900 genomic DNA (DNA lanes) and the respective RT primers that amplify the intergenic regions were performed for amplicon size confirmation. RT-PCR (RT lanes) was run using total RNA extracted from Mm900 cells with the same RT primers. The RT lanes that have bands at the same size as the DNA lanes indicate the cotranscription of the according genes. Standard PCRs were also performed using total RNA that did not undergo reverse transcription as a template (RNA lanes) to rule out possible DNA contamination of the RNA sample.

Generation of Δmmp1084, Δmmp1085, Δmmp1086, and Δmmp1087 single mutants and the Δmmp1086 Δmmp1087 double mutant.

To identify the possible involvement of mmp1084 to mmp1087 in the N-glycosylation system in M. maripaludis S2, in-frame deletions of each gene were generated and confirmed by PCR of transformant cells using primers that amplified across the targeted gene (Fig. 3). In each case, Mm900 genomic DNA (wild type) was also used as a template with the same primer pair as the corresponding deletion mutant to confirm the specificity of the primers and to provide a comparison wild-type size of the target gene. In the PCR amplification using cells of each deletion mutant as a template, the size of the PCR product was smaller than that from wild-type genomic DNA by the length predicted from the size of the deletion made in the target gene. In each case, sequencing of the PCR product confirmed that each deletion made was in frame. In the case of the double mutant Δmmp1086 Δmmp1087, each gene deletion was confirmed individually by PCR and sequencing. As shown in Fig. 3B, the amplicons obtained with the Δmmp1086 Δmmp1087 deletion strain as a template using the mmp1086 and mmp1087 screening primers were smaller than those from the wild-type genomic DNA by the expected lengths, illustrating that the Δmmp1086 Δmmp1087 double mutant was successfully generated.

Fig 3.

Confirmation of in-frame deletions of Δmmp1084, Δmmp1085, Δmmp1086, and Δmmp1087 single mutants and the Δmmp1086 Δmmp1087 double mutant by PCR. (A) Washed whole cells of each deletion strain, as well as wild-type (WT) cells, were used as a template for the PCR confirmation of each in-frame deletion, using corresponding sequencing primers as marked in the figure. (B) For the Δmmp1086 Δmmp1087 double mutant, the in-frame deletions of mmp1086 and mmp1087 were determined individually. Washed whole cells of the Δmmp1086 Δmmp1087 mutant were used as a template with sequencing primers for both mmp1086 and mmp1087 mutants. In all cases, amplicons of the expected size from both the wild type and deletion mutants were obtained. M, 1-kb DNA ladder.

Western blot analysis of the archaellins from the mutants.

To examine whether the deletion of any of the mmp1084 to mmp1087 genes had any detectable effect on the glycan N-linked to the M. maripaludis archaellins, the migration of the archaellin reporter glycoprotein FlaB2 in each deletion strain was initially examined by Western blots probed with antibody specific for this archaellin (34). It has been shown previously that even small truncations to the glycan have effects on archaellin migration that are detectable by Western blots (17, 34). As shown in Fig. 4, FlaB2 migrated as a slightly smaller protein in both Δmmp1084 and Δmmp1085 mutants compared to the wild-type cells, suggesting a possible effect on the glycan. In contrast, no change in the migration of FlaB2 in either Δmmp1086 or Δmmp1087 mutants compared to wild-type cells was evident. However, the 56% identity between MMP1086 and MMP1087 suggests that they might compensate for each other. To exclude the possibility that in either the Δmmp1086 or the Δmmp1087 single mutant, the remaining gene product complements the function of the in-frame deletion, FlaB2 from the Δmmp1086 Δmmp1087 double mutant was also analyzed by Western blot analysis, as shown in Fig. 4. The apparent molecular mass of FlaB2 from the Δmmp1086 Δmmp1087 double mutant was the same as that from wild-type cells, indicating that at least under the culture conditions used neither mmp1086 nor mmp1087 appeared to be involved in the N-glycosylation pathway. Since the roles of aglY and aglZ in N-glycosylation were not evident when cells are grown in the presence of ammonia (34), the size of FlaB2 in the Δmmp1086 Δmmp1087 double mutant was also examined under ammonia-limiting conditions by growing cells in nitrogen-free medium supplemented with alanine. No reduction in the size of FlaB2 in the double mutant was observed in Western blots under these conditions either (data not shown).

Fig 4.

Western blot of whole-cell lysates of the wild type, Δmmp1084, Δmmp1085, Δmmp1086, and Δmmp1087 single mutants, and the Δmmp1086 Δmmp1087 double mutant probed with anti-FlaB2 antibodies. The decrease in apparent molecular mass of FlaB2 in the Δmmp1084 and Δmmp1085 mutants was the initial evidence of a truncated N-glycan in these mutants. No change in the migration of FlaB2 was evident in the Δmmp1086 mutant, the Δmmp1087 mutant, or the Δmmp1086 Δmmp1087 double mutant.

In-frame deletions of mmp1084 to mmp1087 do not interfere with archaellation.

In order to determine whether proteins encoded by mmp1084 to mmp1087 are involved in archaella assembly, the Δmmp1084, Δmmp1085, Δmmp1086, and Δmmp1087 deletion mutants, as well as the Δmmp1086 Δmmp1087 double mutant, were examined by electron microscopy. As shown in Fig. 5, numerous archaella were observed on the surfaces of all of the mutant cells, as well as the wild-type cell, suggesting that deletion of these genes had no gross effect on archaella assembly, number, or cellular location.

Fig 5.

Electron microscopy of wild-type and mutant cells obtained in the present study. Numerous archaella are observed attached to the cell surface of the Δmmp1084, Δmmp1085, Δmmp1086, and Δmmp1087 single mutants and the Δmmp1086 Δmmp1087 double mutant, as well as the wild-type cell. Scale bars, 500 nm.

Semisolid swarm motility plate assay.

Examination of wild-type cells and the various mutants by light microscopy demonstrated that all strains were motile. However, since the motility of even wild-type cells is weak (46), motility was also examined using semisolid agar. Here, a motility similar to that of wild-type cells was observed for the Δmmp1084, Δmmp1085, Δmmp1086, Δmmp1087, and Δmmp1086 Δmmp1087 deletion mutants (data not shown).

Mass spectrometry analysis of archaellins.

Mass spectroscopy of isolated archaella confirmed the Western blot results that showed effects on glycosylation in the Δmmp1084 and Δmmp1085 deletions but not in Δmmp1086, Δmmp1087, or Δmmp1086 Δmmp1087 deletion strains (Fig. 6). Tryptic glycopeptides from the Δmmp1084 archaellin were predominantly modified with the tetrameric glycan designated 203Da-258Da-257Da-217Da (Fig. 6b) corresponding to the loss of the threonine moiety from the third sugar in the wild-type glycan (Fig. 6a). A minor trimeric glycan species, 203Da-258Da-257Da, was also observed in the Δmmp1084 mutant strain. Archaellins from the Δmmp1085 mutant were modified with a different tetrameric glycan 203Da-258Da-358Da-203Da, indicating that the methyl group on the fourth sugar was absent (Fig. 6c). Archaella from the Δmmp1086, Δmmp1087, and Δmmp1086 Δmmp1087 deletion strains were modified with the wild-type glycan, 203Da-258Da-358Da-217Da (Fig. 6d).

Fig 6.

Nano-LC-MS/MS analysis of the FlaB2 tryptic glycopeptide, T⁵³⁻⁸¹, from the wild-type and mutant strains investigated in the present study. The FlaB2 tryptic peptide T⁵³⁻⁸¹ contains one site of N-glycosylation (DSTEQVASGLQIMGISGYQAGTANANITK). The major carbohydrate oxonium ions are identified in the MS/MS spectra using symbols to indicate the sugar residues present as follows: ■, GalNAc; ●, GlcNAc3NAcA; ▲, ManNAc3NAmA6Thr; solid pentagon, ManNAc3NAmA; ⧫, (5S)-2- acetamido-2,4-dideoxy-5-O-methyl-α-l-erythro-hexos-5-ulo-1,5-pyranose (217-Da sugar); and ∗, previous sugar lacking a methyl group (203-Da sugar). The b and y ions arising from fragmentation of the peptide bonds are also shown. (a) The structure of tetrameric wild-type glycan (MS/MS spectrum acquired in an earlier study) has been described previously (16). (b) The glycopeptide from the Δmmp1084 mutant strain was modified with a tetrasaccharide lacking the threonine modification that is linked to the third sugar residue of the wild-type glycan. (c) The same tryptic peptide from the Δmmp1085 mutant was also modified with a tetrasaccharide but one whose terminal sugar is 14 Da less than that in the wild-type glycan. (d) The equivalent glycopeptide from the Δmmp1086 Δmmp1087 double mutant was modified with wild-type glycan.

Complementation of mmp1084 and mmp1085.

To confirm that deletions of mmp1084 and mmp1085 are the only contributors to the defects observed by mass spectrometry, Δmmp1084 and Δmmp1085 mutants were complemented with plasmids bearing the pertinent wild-type copy of the deleted gene under an inducible nif promoter, which is induced under ammonia-limiting conditions (as when l-alanine, for example, is used as the sole nitrogen source) while repressed in the presence of ammonia (37, 47). When cells were grown with alanine as the sole nitrogen source the apparent molecular mass of FlaB2 was restored to wild-type size in both mmp1084 and mmp1085 complementations (Fig. 7), indicating that the molecular mass loss in FlaB2 in each mutant was caused by the in-frame deletion of that single gene. When ammonia was used as the nitrogen source, the apparent mass of FlaB2 in both complementations was also restored. This effect has been previously observed in the complementation of certain enzymes involved in the N-linked glycosylation process, likely due to a low basal level of expression from the nif promoter even in the presence of ammonia (17). This low level of expression may be sufficient to provide enough product to still complement the deletion.

Fig 7.

Western blot analysis of complementation experiments with Δmmp1084 and Δmmp1085 mutants. A Western blot of whole-cell lysates from Δmmp1084 and Δmmp1085 mutants harboring respective complementation vectors cultured in nitrogen-free medium supplemented by either alanine (Ala, promoter on) or ammonia (NH₄⁺, promoter off) was probed with anti-FlaB2 antibodies. Wild-type (WT) and deletion mutants were used as controls to show the apparent molecular masses of FlaB2 attached to wild-type and truncated N-glycans.

DISCUSSION

Using archaellins and S-layer proteins as the reporter proteins, significant progress has been made in the study of N-glycosylation pathways in several model archaeal organisms, such as H. volcanii, S. acidocaldarius, M. voltae, and M. maripaludis (2, 9, 48). In these systems, numerous agl genes in the N-glycosylation pathways have been identified encoding the oligosaccharyltransferase aglB, glycosyltransferases and enzymes involved in the biosynthesis of component sugars in the N-glycan (2, 9, 12, 14, 19, 20, 22–26, 33, 34, 49–53). The viability of an aglB deletion mutant shows that the N-glycosylation of proteins is not essential for M. voltae (14), M. maripaludis (17), or H. volcanii (22). However, disruption of the system leads to defects in archaellation in both methanogens (14, 17) and H. volcanii (29) and the stability of the S layer in the halophile, which results in growth defects at elevated NaCl concentrations (23). In contrast to the situation in these euryarchaeotes, an aglB deletion could not be created in the crenarchaeote S. acidocaldarius, indicating an essential role for this protein modification in this organism (9). Interference with N-glycosylation to the point where truncated glycans are still attached in S. acidocaldarius leads to both nonarchaellation and growth defects at higher salinities (9, 19, 20).

In M. maripaludis, genes encoding the oligosaccharyltransferase and several glycosyltransferases were identified (17), and it was shown that interference in the N-glycosylation pathway resulted in adverse effects on archaella formation and function but not on pilus assembly (14, 17). Recently, proteins involved in the biosynthesis of the glycan component sugars have been identified (33, 34, 54, 55). We expand here on these findings and identify two genes whose products are involved in the synthesis of the final two sugars of the glycan: mmp1084 is involved in the threonine modification of the third sugar, and mmp1085 encodes the methyltransferase required for the biosynthesis of the terminal sugar.

In M. maripaludis S2, an in-frame deletion of mmp1084 resulted in a truncated N-glycan which was shown by mass spectrometry analyses to be missing only the threonine modification of the third sugar. Possible roles for MMP1084 include involvement in either the biosynthesis of the threonine or the transfer of this amino acid onto the lipid-linked tetrasaccharide. In silico analysis shows that MMP1084 belongs to the Gn_AT_II superfamily and has sequence similarity to the putative asparagine synthase AsnB in many sequenced genomes. However, in the genome of M. maripaludis S2, mmp0918 has been annotated as asnB, and it is located with adjacent genes whose products are annotated to be involved in amino acid metabolism, suggesting that mmp0918 is more likely to encode the true asparagine synthase. Genes homologous to mmp0918 are found in a conserved genomic island in numerous members of the Methanococcales, including in the sequenced genomes of four other M. maripaludis strains (C5, C6, C7, and X1), indicating that mmp0918 is a housekeeping gene. On the other hand, homologues of mmp1084 are rare and are not found in other sequenced M. maripaludis strains, although they can be identified in M. voltae strains PS and A3. In keeping with the assigned function of MMP1084, the structure of the N-linked glycan attached to M. voltae PS archaellins and S layer protein includes a threonine attached to the C-6 of a mannuronic acid as the third sugar (13). The lack of an mmp1084 homolog in other sequenced M. maripaludis strains suggest that these strains do not have a glycan carrying a threonine-modified mannuronic acid. In addition, the growth of Δmmp1084 cells is not impaired in nitrogen-free medium supplemented with NH₄Cl as the sole nitrogen source, which means the de novo synthesis pathway of amino acids is not affected in the mutant. This would seem to rule out a role for MMP1084 in threonine biosynthesis.

Further analysis of MMP1084 supports a role for this protein in the transfer of threonine to the third sugar. MMP1084 has 22% identity and 39% similarity with the product of orf3 in the K40 gene cluster in E. coli O8:K40, a rare example of an identified gene that is responsible for transferring an amino acid, serine, onto the carboxyl group at the C-6 position of a hexuronic acid (56). Like MMP1084, ORF3 also possesses a conserved domain found in members of the Gn_AT_II superfamily. In E. coli O8:K40, the K40 CPS repeat unit is a trisaccharide (6GlcNAc-4-β-GlcA6Ser-4-α-GlcNAc) in which the serine is linked to the second sugar, and deletion of orf3 does not block the addition of the last sugar onto the CPS repeat unit. Similarly, in M. maripaludis, in-frame deletion of mmp1084 leads to a truncated N-glycan that is only missing the threonine linked to the third sugar residue but retains the full tetrasaccharide backbone. These results suggest that the amino acid modification may be the last step in the glycan synthesis pathway, occurring only after the fourth sugar is added to the growing glycan. This would be consistent with previous observations on the glycan of an aglL (encoding the fourth glycosyltransferase) mutant in which not only is the last sugar missing but also the threonine of the third sugar (17). Sequence analysis of MMP1084 does not show any evidence of a signal peptide or transmembrane domain, and it is predicted to be a cytoplasmic protein. It may function in the cytoplasm after the action of the glycosyltransferase for the last sugar but before the flipping of the lipid-linked N-glycan precursor to the external face of the cytoplasmic membrane. However, it is also possible that MMP1084 could be membrane associated if it interacts with integral membrane proteins, perhaps in a complex devoted to glycan formation.

A mutant carrying an in-frame deletion of mmp1085 had a truncated N-glycan lacking only the methyl group O-linked to the C5 hydroxyl group on the unique fourth sugar residue. This is consistent with the identification of a conserved domain in MMP1085 associated with SAM-dependent methyltransferases and with BLAST search results that show significant sequence similarity of MMP1085 to SAM-dependent methyltransferases. O-methylation is a relatively common glycan modification and has been studied in another archaeal N-glycan attached to both S-layer protein and archaellins in H. volcanii. In H. volcanii, the C-4 position in the hexuronic acid fourth sugar residue in the pentasaccharide N-glycan is methylated through the action of AglP, also identified as a SAM-dependent methyltransferase (49). Several bacterial O-antigenic polysaccharides also contain a variety of different sugars with O-methylation modifications at various positions (C-2, C-3, C-4, or C-6) catalyzed by SAM-dependent methyltransferases (57–60). However, little sequence similarity is shared between the bacterial methyltransferases and the archaeal AglP and MMP1085. Thus, mmp1085 is the first gene identified to be involved in the biosynthesis pathway of the unique terminal sugar, (5S)-2-acetamido-2,4-dideoxy-5-O-methyl-α-l-erythro-hexos-5-ulo-1,5-pyranose. Since deletion of mmp1085 resulted in the loss of the methyl group of the terminal sugar, we assume that the methyl group is added onto the second hydroxyl group on the C-5 after the formation of the dialdose.

The products of the last two genes in the operon, mmp1086 and mmp1087, did not have any demonstrable effect on the N-linked glycan. There was no detectable difference in the size of FlaB2 in Δmmp1086 and Δmmp1087 mutants compared to the wild-type version in Western blots, and mass spectrometry analysis of the purified archaella showed that a wild-type glycan was attached. Since MMP1086 and MMP1087 share 56% amino acid sequence identity and both contain an N-terminal radical-SAM domain and a C-terminal DUF4008 domain, it was considered that a loss of one of the two proteins might be compensated by the other. However, Western blot and mass spectrometry analyses of the Δmmp1086 Δmmp1087 double mutant also showed wild-type patterns. These data indicate that neither mmp1086 nor mmp1087 are involved in the N-glycosylation system under normal growth conditions. However, since the roles in N-glycosylation of the neighboring genes aglY and aglZ were not evident under normal growth conditions but only when the cells were ammonia limited (34), the Δmmp1086 Δmmp1087 mutant was also examined by Western blotting after growth under these conditions. However, again, only a wild-type size of FlaB2 was observed. These findings do not rule out the possibility that mmp1086 and mmp1087 might be involved in the N-glycosylation pathway under different growth conditions.

Recent analysis of sequenced archaeal genomes indicated that in many cases, especially among haloarchaea, there is a clustering of genes putatively involved in N-glycosylation in the immediate vicinity of aglB (8). However, this is not true of most methanogens, including M. voltae and M. maripaludis or the crenarchaeon S. acidocaldarius (8). Nonetheless, gene clusters involved in N-glycosylation not linked to aglB also exist. In M. maripaludis, a group of genes involved in the assembly or biosynthesis of the archaellin N-glycan are located in the genomic region from mmp1076 to mmp1094 (Fig. 1). Included in this region are the three glycosyltransferase genes aglO (mmp1079), aglA (mmp1080), and algL (mmp1088) (17), mmp1076 (which encodes GlcN-1-P uridylyltransferase/acetyltransferase), mmp1077 (which encodes a phosphomutase that converts α-d-glucosamine-6-phosphate to α-d-glucosamine-1-phosphate) (33), mmp1090 (which encodes a UDP-Gal/GalNAc 4-epimerase [Ding et al., unpublished]), aglXYZ (mmp1081-mmp1083; responsible for the acetamidino modification of the third sugar) (34) and, in the present study, mmp1084 and mmp1085. In addition to this genetic locus, another cluster of genes (mmp0350 to mmp0359) has been identified and shown by deletion analysis and mass spectrometry to include genes that encode enzymes involved in biosynthesis of the second and third sugars of the glycan (54; S. Siu, K. F. Jarrell, S. M. Logan, and J. Kelly, unpublished data). Interestingly, in a related mesophilic methanococcus, M. voltae A3, homologues of both gene clusters are all located in a single large gene locus (from Mvol_0230 to Mvol_0247) (Fig. 8).

Fig 8.

Gene locus from Mvol_0225 to Mvol_0247 in M. voltae A3. Many homologues of genes involved in the N-glycosylation pathway in M. maripaludis could be found in this gene island, including mmp1084 to mmp1087, aglA, aglL, aglXYZ, mmp1076, and mmp1090. The mmp numbers or agl gene designations under the M. voltae genes indicate the respective homologues in M. maripaludis S2. No homologues of Mvol_0233 or Mvol_0244 to Mvol_0246 could be found in the M. maripaludis S2 genome.

In the present study, we examined four genes found in a single operon—mmp1084, mmp1085, mmp1086, and mmp1087—for their potential role in the N-linked glycosylation process in M. maripaludis S2. We were able to assign roles to two gene products. MMP1084 is the putative threonine transferase for the third sugar which functions after AglL has added the fourth sugar to the lipid carrier, and MMP1085 is the methyltransferase which transfers the methyl group to the second hydroxyl group on C-5 position in the terminal sugar. No role for MMP1086 or MMP1087 in N-glycan biosynthesis could be identified. Because of their demonstrated involvement in the N-glycosylation pathway, we designate mmp1084 as aglU and mmp1085 as aglV, in keeping with the nomenclature scheme for genes involved in N-linked glycosylation in Archaea first proposed by Chaban et al. (14) (Fig. 9).

Fig 9.

Summary of the function of AglU and AglV in the N-glycosylation pathway in M. maripaludis. MMP1084 is designated as AglU, and MMP1085 as AglV. AglU is the putative threonine transferase for the threonine modification of the third sugar. AglV is the putative methyltransferase which transfers the methyl group to the second hydroxyl group on C-5 position in the terminal sugar. The previously identified oligosaccharyltransferase AglB and glycosyltransferases AglA, AglO and AglL are also marked in this model (17).