Identification of genes involved in the biosynthesisand attachment of Methanococcus voltae N-linkedglycans: insight into N-linked glycosylation pathwaysin Archaea

Bonnie Chaban,1 Sebastien Voisin,2 John Kelly,2

Susan M. Logan2 and Ken F. Jarrell1*1Department of Microbiology and Immunology, Queen’sUniversity, Kingston, Ontario, K7L 3N6, Canada.2Institute for Biological Sciences, National ResearchCouncil, Ottawa, Ontario, K1A 0R6, Canada.

Summary

N-linked glycosylation is recognized as an importantpost-translational modification across all threedomains of life. However, the understanding of thegenetic pathways for the assembly and attachment ofN-linked glycans in eukaryotic and bacterial systemsfar outweighs the knowledge of comparable pro-cesses in Archaea. The recent characterization ofa novel trisaccharide [b-ManpNAcA6Thr-(1-4)-b-GlcpNAc3NAcA-(1-3)-b-GlcpNAc] N-linked to aspar-agine residues in Methanococcus voltae flagellin andS-layer proteins affords new opportunities to investi-gate N-linked glycosylation pathways in Archaea.In this contribution, the insertional inactivation ofseveral candidate genes within the M. voltae genomeand their resulting effects on flagellin and S-layerglycosylation are reported. Two of the candidategenes were shown to have effects on flagellin andS-layer protein molecular mass and N-linked glycanstructure. Further examination revealed inactivationof either of these two genes also had effects onflagella assembly. These genes, designated agl(archaeal glycosylation) genes, include a glycosyltransferase (aglA) involved in the attachment of theterminal sugar to the glycan and an STT3 oligosac-charyl transferase homologue (aglB) involved in thetransfer of the complete glycan to the flagellin andS-layer proteins. These findings document the firstexperimental evidence for genes involved in any gly-cosylation process within the domain Archaea.

Introduction

It is well known that the addition of carbohydrate to proteinis an important post-translational modification required forthe proper folding and function of many eukaryotic pro-teins (Gemmill and Trimble, 1999; Robbins, 1999). Theold paradigm that glycosylation was limited to eukaryoteshas been recognized as false and, in recent years, studiesof organisms like Campylobacter jejuni have greatlyimproved our knowledge of how bacteria carry out bothN-linked and O-linked glycosylation (Schmidt et al., 2003;Szymanski et al., 2003; Szymanski and Wren, 2005).Indeed, studies in C. jejuni have established an excellentmodel for an N-linked glycosylation pathway in bacteria,with the activities of the characterized pgl (protein glyco-sylation) gene cluster assembling and transferring aknown heptasaccharide from a membrane-anchoredundecaprenylpyrophosphate-linked donor to an aspar-agine residue in proteins at the classic Asn-X-Ser/Thrmotif (Szymanski et al., 2003; Glover et al., 2005a).However, this same breadth of knowledge is not availablefor archaeal glycosylation systems. Evidence for proteinglycosylation in Archaea ranges from detailed structuralcharacterization of specific glycans and their linkage sitesto indirect characterization of glycosylation using lectinbinding, glycan staining and aberrant migration of proteinsin SDS-PAGE (reviewed by Eichler and Adams, 2005).

As the surface (S)-layer protein (or cell surface glyco-protein, CSG) of Halobacterium salinarum was recog-nized and characterized as the first prokaryoticglycoprotein (Mescher et al., 1974; Mescher andStrominger, 1976), this organism remains one of the bestcharacterized in terms of archaeal glycosyation. EachH. salinarum S-layer protein was found to be modified byone N-linked sulphated repeating unit saccharide, 10N-linked sulphated oligosaccharides and 20 O-linkedneutral oligosaccharides (Mescher and Strominger, 1976;Lechner and Sumper, 1987; Sumper, 1987). The flagellinproteins were also modified with the same N-linked sul-phated oligosaccharides. Biochemical analysis demon-strated that sulphated precursors of these saccharideswere linked to dolichyl monophosphate (Lechner et al.,

Accepted 8 May, 2006. *For correspondence. E-mail jarrellk@post.queensu.ca; Tel. (+1) 613 533 2456; Fax (+1) 613 533 6796.

Molecular Microbiology (2006) 61(1), 259–268 doi:10.1111/j.1365-2958.2006.05226.xFirst published online 1 June 2006

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd

1985). This is similar to the dolichol used by the eucaryalprotein N-glycosylation system, but distinct from the unde-caprenol used by bacterial systems (Lechner et al., 1985;Eichler and Adams, 2005). However, despite the impres-sive collection of biochemical data, nothing has yet beenreported in H. salinarum, or any other Archaea, regardingthe genes that comprise the pathways leading to thebiosynthesis, assembly or attachment of archaealglycans.

Another archaeal species that is known to possessglycosilated flagellin and S-layer proteins is Methanococ-cus voltae. Flagella of M. voltae are composed of twomajor flagellins (FlaB1 and FlaB2) and two minor flagellins(FlaA and FlaB3) (Bardy et al., 2002). Recent analysis ofM. voltae flagellins determined that the proteins weremodified with a novel trisaccharide, b-ManpNAcA6Thr-(1-4)-b-GlcpNAc3NAcA-(1-3)-b-GlcpNAc, N-linked to Asn(Voisin et al., 2005). All of the peptides containing N-linkedsequons from the four flagellin proteins were character-ized by nano-liquid chromatography-electrospray tandemmass spectrometry (nanoLC-MS/MS) (five on FlaB1, sixon FlaB2 and two on FlaB3, one of two on FlaA; thesecond sequon-containing peptide was not included in the60% coverage of this minor flagellin) and were shown tocarry the 779 Da trisaccharide in N-linkage (Voisin et al.,2005). In addition, a peptide containing an N-linkedsequence motif or sequon (Asn-X-Ser/Thr) from theS-layer protein was also detected by MS/MS and wasfound to be modified in an identical fashion. This indicatesthat like H. salinarum, M. voltae S-layer and flagellin pro-teins share a common glycan and, presumably, acommon N-linked glycosylation pathway.

The N-linked glycosyation pathway in bacteria hasrecently been reviewed (Szymanski and Wren, 2005) andshown to be similar in general mechanism to thatdescribed for eukaryotes. In both cases, glycosyl trans-ferases link nucleotide-activated sugars to the appropriatelipid carrier in the membrane. This occurs on the cytoplas-mic side of the cell membrane in bacteria or the cytoplas-mic face of the endoplasmic reticulum in eukaryotes.Once assembled, the glycan-lipid is translocated (orflipped) across the membrane by a flippase. Eukaryotic

glycans are further modified at this stage, while in thebacterial system, no evidence has been presented foradditional modification of the glycans. Finally, an oligosac-charide transferase (the OTase complex in eukaryotes;STT3 protein in bacteria) transfers the glycan en bloc tothe target protein at the Asn residue in the Asn-X-Ser/Thrsequon (Szymanski and Wren, 2005). At the present time,the same sequence of events described for bacteria aretheorized to occur in Archaea (Sumper, 1987; Eichler andAdams, 2005).

In this study, the first demonstration of specific genesinvolved in the assembly and attachment of an archaealglycan are reported. Using M. voltae as a model organ-ism, several genes with characteristics of glycan biosyn-thetic or protein glycosylation genes from bacteria wereinactivated. Two of these genes were shown to play acritical role in the biosynthesis of glycosilated flagellinsand S-layer proteins.

Results

Generation of M. voltae mutants

Three distinct genetic loci, originally identified by BLAST

analysis to contain genes with homology to known bac-terial genes involved in either protein glycosylation orglycan biosynthesis, were targeted. The three loci arepresented in Fig. 1 and the genes which were inacti-vated are listed in Table 1. Of the 10 genes targeted,nine knock-outs were successfully made (Mv1751 beingthe exception) and confirmed via Southern blot analysis(Mv151 and Mv1749 shown in Fig. 2; remainder notshown). For both Mv151 and Mv1749, Southern blotdata revealed tandem insertion of the inactivationplasmid into the target gene region (Fig. 2B). This resulthas been previously documented to occur with thismethodology in M. voltae (Berghöfer and Klein, 1995;Thomas et al., 2001).

As seen in Fig. 1, most of the genes targeted in thisstudy are transcribed in the same direction as genes lyingimmediately adjacent. Intergenic regions are often short,ranging from 11 bp to 325 bp, creating the potential for

Fig. 1. Schematic of M. voltae gene regionsof interest. Inactivation of black colouredgenes had noticeable effects on flagellin andS-layer molecular weights; grey genes had noeffect and the striped gene is still beingtargeted.

1102 1103 1104 1105 1106 11081107 1832

152 155 156 157 301 302 303 304

1745 17471746 1748 17531751

73 1809 250 251 252

Region 1

Region 2

Region 3

1102 1103 1104 1105 1106 11081107 183 1512

152 155 156 157 301 302 303 304152 155 156 157 301 302 303 304

1745 17471746 174 17498 17531751

73 1809 250 251 25273 1809 250 251 252

Region 1

Region 2

Region 3

260 B. Chaban et al.

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Molecular Microbiology, 61, 259–268

multigene transcripts. As the insertional inactivationmethods used in this study are predicted to have polareffects downstream of target genes (Jarrell et al., 1996),when a novel phenotype was obtained, additional knock-outs were constructed in genes located immediatelydownstream to confirm that the observed phenotype wasdue to the gene inactivated and not a result of polareffects on downstream genes (Table 1).

Effects of flagellin glycosylation

To determine if insertional inactivation resulted in analtered flagellin phenotype, whole cell lysates of eachM. voltae mutant were examined by immunoblot utilizingM. voltae anti-FlaB2 antibody. Of the nine mutantstested, two (Mv151 and Mv1749) had detectable shifts inflagellin molecular weight (Fig. 3). To determine moreprecisely the nature of the change in flagellin molecularweight, two forms of M. voltae flagellins were used ascontrols. M. voltae flagellins, like all archaeal flagellins,are first synthesized as preproteins. These preflagellinsare processed into mature proteins via the cleavage of a12-amino-acid leader sequence by a dedicated pre-flagellin peptidase (Correia and Jarrell, 2000). For a fullyglycosilated size marker, wild-type M. voltae flagellinswere used. Wild-type M. voltae FlaB1/B2 flagellins aredetected by immunoblot at 31–33 kDa (given the highsequence similarity between FlaB1 and FlaB2, poly-clonal antibodies raised against FlaB2 cross-react withFlaB1, Bayley and Jarrell, 1999). This size representsfully modified flagellin [possessing five (FlaB1) and six(FlaB2) glycans/protein] with the 12-amino-acid signalpeptide sequence removed (Fig. 3, lanes 1 and 7). For anon-glycosilated size marker, M. voltae FlaB2 flagellinoverexpressed in Escherichia coli was used (Bayley andJarrell, 1999). Utilizing an established preflagellin pepti-dase assay (Correia and Jarrell, 2000), M. voltae FlaB2flagellins can be generated with (26.5 kDa) and without(25 kDa) their signal peptide sequence. Both the pre-flagellin and mature flagellin forms are seen in Fig. 3,lane 4. As E. coli cannot glycosilate the archaeal flagel-lins, these forms of FlaB2 represent unglycosilatedflagellins with and without the signal peptide.

For Mv1749, an apparent molecular weight shift ofFlaB1/B2 was seen (Fig. 3). Unfortunately, this mutantwas extremely unstable and upon transfer of the originalculture (seen in Southern blot, Fig. 2A), subsequentgrowth contained a mixed population of mutant and wild-type cells (lane 5, Fig. 3) and a complete reversal towild-type flagellin molecular weight was seen within only afew subcultures (lane 6). Despite this, initial immunoblotsof Mv1749 reveal a FlaB1/B2 band that is consistent withan unglycosilated, but N-terminally processed flagellinprotein (lower band, lane 5).Ta

ble

1.G

enes

from

M.v

olta

eta

rget

edfo

rin

activ

atio

n.

Gen

ede

sign

atio

nP

utat

ive

nam

eaH

ypot

hesi

zed

func

tion

ingl

ycos

ylat

ion

No.

ofT

MD

bD

emon

stra

ted

effe

cton

glyc

osyl

atio

n

Mv1

51G

lyco

sylt

rans

fera

se(a

glA

)G

lyco

sylt

rans

fera

se1

Yes

Mv1

52A

spar

agin

esy

nthe

tase

[glu

tam

ine-

hydr

olys

ing]

Non

ec0

No

Mv1

55G

lyco

sylt

rans

fera

seG

lyco

sylt

rans

fera

se0

No

Mv1

56U

DP

-glu

cose

4-ep

imer

ase

Bio

synt

hesi

s1

No

Mv2

51H

ypot

hetic

alpr

otei

nB

iosy

nthe

sis

3N

oM

v110

8P

olys

acch

arid

ebi

osyn

thes

ispr

otei

nF

lippa

se13

No

Mv1

748

Hyp

othe

tical

prot

ein

Non

ec0

No

Mv1

749

Olig

osac

char

yltr

ansf

eras

eS

TT

3su

buni

t(a

glB

)O

ligos

acch

aryl

tran

sfer

ase

13Ye

sM

v175

1N

-ace

tylg

luco

sam

ine

phos

phat

etr

ansf

eras

eG

lyco

sylt

rans

fera

se7

Unk

now

nM

v180

9M

ethy

ltra

nsfe

rase

Bio

synt

hesi

s0

No

a.P

utat

ive

nam

ew

asas

sign

edby

BLA

ST

com

paris

onw

ithco

mpl

eted

arch

aeal

geno

mes

.b

.N

umbe

rof

pred

icte

dtr

ansm

embr

ane

dom

ains

,ba

sed

onT

Mpr

ed(H

ofm

ann

and

Sto

ffel,

1993

).c.

Not

hypo

thes

ized

toha

vean

yro

lein

glyc

osyl

atio

n.T

hese

gene

sw

ere

targ

eted

beca

use

they

wer

edi

rect

lydo

wns

trea

mof

age

new

itha

dete

ctab

legl

ycos

ylat

ion

effe

ct.

Archaeal glycosylation genes 261

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Molecular Microbiology, 61, 259–268

In the case of Mv151, the FlaB1/B2 proteins migratedwith an apparent molecular weight that was intermediatebetween mature glycosilated and unglycosilated flagellin(lane 2, Fig. 3). This mutant was also unstable and amixed population of mutant/wild-type cells was apparentafter limited subculturing of the initial mutant culture (lane3) and several additional subcultures resulted in flagellinand chromosomal phenotypes that had completelyreverted to wild type.

For each of these two mutants, the revertants were stillpuromycin resistant, although Southern blotting resultsindicated that the genes of interest in each case no longercontained the inserted plasmid.

For the remaining seven mutants generated, noobvious effects on flagellin apparent molecular weightswere seen (data not shown). While these findings do notcompletely rule out a role for these genes in the N-linkedglycosylation process, it appears that their inactivationdoes not have a gross effect on glycan assembly. Corre-spondingly, these mutants were stable to repeated sub-culturing, in contrast to the instability of Mv151 andMv1749. The lone exception was Mv1108. Mv1108 wasidentified in BLAST searches as a homologue of polysac-charide biosynthesis proteins and heteropolysacchariderepeat unit export proteins (flippases) responsible fortransfer of a lipid-linked glycan across the cell membraneto the periplasm (Table 1). The pure mutant culture ofMv1108 showed no effect on flagellin apparent molecularweight. However, as with Mv151, 1749, the mutant inMv1108 was unstable and rapidly reverted back to a wild-type chromosomal genotype upon subculture, which mayindicate that such mutants are compromised in an impor-tant cell function.

Effect on flagella structure

To evaluate the overall effect of each respective mutationon flagella function, cells were visually examined underphase contrast microscopy for motility. Wild-type

M wt Δ M wt ΔA

151 1749

B3.5 kb

151 wt

151 Del 4.8 kb 3.4 kb 2.1 kb

1749 wt

1749 Del

1.9 kb

3.5 kb 5.5 kb 3.9 kb

M wt Δ M wt ΔA

151 1749

M wt Δ M wt ΔA

151 1749

B3.5 kb

151 wt

151 Del 4.8 kb 3.4 kb 2.1 kb

1749 wt

1749 Del

1.9 kb

3.5 kb 5.5 kb 3.9 kb

Fig. 2. Confirmation of gene knock-outs for M. voltae mutants Mv151 (aglA) and Mv1749 (aglB).A. Southern blot analysis. l-HindIII marker represents 23, 9, 6.5, 4.3, 2.3 and 2 kb.B. Theoretically expected Southern blot band sizes. Markers with, diamond tops indicate restriction sites used (Hinf1 for Mv151 and HindIII forMv174a), black boxes represent internal fragments of the gene used as Dig-labelled probes and dotted lines represent the pac/pUC vector.Note that for both mutants, tandem insertion of the inactivation plasmid was seen.

1 2 3 4 5 6 7

Fig. 3. Western blot of M. voltae flagellins FlaB1/2. Markers denote35 and 25 kDa. Lanes 1 and 7, wild-type M. voltae; lanes 2 and 3,Mv151 (aglA) as mutant and mixed mutant/wild-type reversioncultures respectively; lane 4, M. voltae FlaB2 expressed in E. coli,with and without signal peptide (see text for details); lanes 5 and 6,Mv1749 (aglB) as mutant/wild-type reversion mixed culture andcompleted reverted culture respectively.

262 B. Chaban et al.

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Molecular Microbiology, 61, 259–268

M. voltae cells were motile, while Mv151 cells appearedweakly motile and Mv1749 cells appeared non-motile. Allother M. voltae mutants generated in this study appearedas motile as wild-type cells.

Mv151 and Mv1749 cells were further examined bytransmission electron microscopy for flagella structureand number (Fig. 4). While wild-type cells possessedmany flagella (Fig. 4A), Mv151 cells had only a few fla-gella, many of which appeared short/broken (Fig. 4B),and Mv1749 cells appeared non-flagellated (Fig. 4C).Mv152 and Mv1748 (mutants in genes located immedi-ately downstream of Mv151 and Mv1749) were alsoexamined and had flagella indistinguishable from wildtype (data not shown). This again confirmed that theeffects on glycosylation observed in mutants Mv151 andMv1749 were due to disruption of those genes and not topolar effects on downstream genes. In addition, as it hasbeen previously shown that only N-terminally processedflagellins can be assembled into flagella filaments (Bardyand Jarrell, 2003), the molecular weight shift observed inMv151 FlaB2 (Fig. 3) is not due to an unprocessed signalpeptide.

Characterization of Mv151 glycans

Following the observation by electron microscopy thatMv151 cells produced limited amounts of flagella, thefilaments were sheared off, purified and examined bymass spectroscopy. The glycosylation profiles fromMv151 tryptic peptides were compared with thoseobtained for flagellin peptides from the parent strain. Asexpected, nanoLC-MS/MS of tryptic peptides containingN-linked sequons from wild-type flagellin producedspectra containing oxonium ions typical for the trisaccha-ride which had been previously characterized (Voisinet al., 2005). In contrast, the MS/MS spectra of trypticpeptides containing N-linked sequons from Mv151 mutantflagellin displayed a unique series of fragment ions indi-cating the loss of the terminal modified mannuronic acid(m/z 319). The spectra for the tryptic peptide

100YNESSYK106 from wild type and Mv151 flagellin areshown in Fig. 5 and clearly demonstrates that the glycanmade by Mv151 cells lacks the oxonium ion correspond-ing to the terminal monosaccharide (m/z 319). It should benoted that detailed analysis of the Mv151 flagellin trypticpeptides revealed minor amounts (approximately 8% ofall glycopeptide spectra examined) of peptides whichcarried a full-length glycan. The fully modified glycopep-tides from the Mv151 strain were presumably a conse-quence of cells within the population reverting back to wildtype and producing limited amounts of flagellin glycosi-lated with a complete glycan structure. A careful reexami-nation of the wild-type strain data revealed that more than95% of examined glycopeptides were fully modified: gen-erally, there was only one glycopeptide lacking the com-plete trisaccharide per wild-type nanoLC-MS/MS analysis.

Effect on S-layer protein

Because it had been previously determined that theS-layer protein in M. voltae contained the same trisaccha-ride modification as the flagellins (Voisin et al., 2005), weexamined the apparent molecular weight of the S-layerproteins from all M. voltae mutants by SDS-PAGE. As wasthe case with the flagellins, only Mv151 and Mv1749 hadany detectable change in their S-layer molecular weights(Fig. 6), and shifts back to wild-type protein molecularweight (76 kDa; Koval and Jarrell, 1987) correspondedwith mutant reversion.

Discussion

In this study, genes involved in the assembly and attach-ment of a novel archaeal trisaccharide N-linked toM. voltae flagellins and S-layer proteins have beenidentified. Several genes in the M. voltae genome weretargeted for inactivation and examined for a role inN-linked glycosylation of flagellin and S-layer protein. Forthis inaugural study, the techniques utilized focused ongross alterations to flagellin and S-layer protein apparent

A B CA B C

Fig. 4. Electron micrographs of (A) wild-type M. voltae; (B) Mv151 mutant (aglA); and (C) Mv1749 mutant (aglB). Scale bar equals 1 mm.

Archaeal glycosylation genes 263

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Molecular Microbiology, 61, 259–268

molecular weight. This does not rule out the possibilitythat the remaining genes that were targeted contributed tominor changes that were undetectable by the methodsemployed. On the basis of a change in flagellin andS-layer protein molecular weight detectable by SDS-PAGE, structural analysis of flagellin protein by MS/MSand examination of cells by electron microscopy, twogenes have been identified which are clearly involved inthe glycosylation process.

The first gene demonstrated to have an effect in theN-linked glycan of M. voltae was Mv151, renamed aglA.The protein encoded by this gene was shown to beinvolved either in the biosynthetic pathway or in theattachment of the terminal sugar to the flagellin andS-layer glycan. As the AglA protein has a glycosyl trans-ferase group 1 domain (pfam00534) and is predicted tohave one transmembrane domain close to the C-terminus(TMpred, Hofmann and Stoffel, 1993), the likely functionof the protein is in assembly of the glycan on a lipid-linkedcarrier at the cytoplasmic face of the membrane throughthe addition of the terminal monosaccharide. Because of

the unusual nature of the M. voltae terminal sugar (amodified mannuronic acid with the amino acid threoninecovalently attached at position 6), it is currently unclearwhether AglA recognizes mannuronic acid or the mannu-ronic acid with the attached threonine. This will depend, inpart, on whether the threonine is added during the bio-synthesis of the terminal residue or whether it is added tothe complete glycan independently.

Although the aglA– mutant only affected the terminalmonosaccharide of the glycan, the resulting effect on fla-gella assembly was pronounced. The amount of FlaB2protein detectable in aglA– strains was significantly lesscompared with wild type (data not shown), correspondingwith the severely reduced flagella number seen by elec-tron microscopy. This implies that the glycans on flagellinproteins are involved in the flagella assembly process.The nature of this involvement, whether it is in recognition,structure, or some other role remains to be investigated.

The second gene demonstrated to have an effect wasMv1749, renamed aglB. AglB is a homologue of theeukaryotic (STT3p) and bacterial (PglB) oligosaccharyltransferases. The STT3p protein has been demonstratedto be an essential component in the nine subunit eukary-otic OTase complex that transfers pre-assembled glycansto target proteins via N-linkages (Zufferey et al., 1995).The most conserved region of this protein, WWDYG, hasbeen characterized as the active site of the OTase (Yanand Lennarz, 2002). The STT3p homologue in bacteria,PglB, has also been shown to transfer a glycan to Asn inthe common N-linked sequon. In C. jejuni, PglB was theonly protein necessary to accomplish this (Szymanski

A B

100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 14000

%

100

0

%

100

Fig. 5. nanoLC-MS/MS spectra of glycosilated tryptic peptide 100YNESSYK106 from the flagellin FlaB2: (A) wild-type M. voltae or (B) Mv151mutant. In (A) (wild-type M. voltae), the spectrum is dominated in the high m/z range by the sequential loss of the carbohydrate moieties(indicated in the spectrum by the open arrows) and in the low m/z range by the carbohydrate oxonium ions (m/z 319.1 and 259.1 plus theirdehydration products). The glycan composition of the major fragment ions is depicted schematically within the brackets above each ion. Thestraight line represents the tryptic peptide and each additional symbol represents the monosaccharide constituents of the glycan [�, m/z 204(203 Da); �, m/z 259 (258 Da); �, m/z 319 (318 Da)]. In (B) (Mv151 mutant), in the high m/z range, only ions corresponding to either theunmodified tryptic peptide or tryptic peptide glycosilated with a single 203 Da monosaccharide is apparent. In the low m/z range, the absenceof the m/z 319 carbohydrate oxonium ion is readily apparent while the m/z 259 oxonium ion and its dehydration product are still present.

1 2 3 4 5 6 71 2 3 4 5 6 7

Fig. 6. Coomassie blue stain of SDS-PAGE separated S-layerproteins. Lanes 1, 4 and 7, wild-type M. voltae (76 kDa); lanes 2and 3, Mv151 (aglA) as mutant and mixed mutant/wild-typereversion cultures respectively; lanes 5 and 6, Mv1749 (aglB) asmutant and mutant/wild-type reversion cultures respectively.Proteins were separated on a 12.5% acrylamide gel.

264 B. Chaban et al.

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Molecular Microbiology, 61, 259–268

et al., 2003) and the same catalytic WWDYG motif wasfound. The archaeal STT3 homologue, AglB, appears tofunction in the same way as its bacterial counterpart, withthis predicted membrane-bound protein containing ahomologous putative catalytic region of WWDNG locatedon the exterior side of the cytoplasmic membrane(TMpred, Hofmann and Stoffel, 1993). Some variability inthe fourth amino acid of the STT3 catalytic domain hasbeen observed before (Szymanski and Wren, 2005), as inthe sequence WWDWG from the bacterial genusDesulfovibrio. The otherwise highly conserved nature ofthe STT3p/PglB/AglB protein across all three domains oflife underlies its importance in N-linked glycosylation.

Another property of the STT3 subunit is that it transfersglycans en bloc to target proteins. AglB has demonstratedthe ability to attach incomplete glycans as seen by thedisaccharides present on flagellins and S-layer proteins inM. voltae aglA mutants. This type of relaxed specificityhas been observed in Saccharomyces cerevisiae alg(asparagine-linked glycosylation) mutants deficient inassembly steps of the lipid-linked oligosaccharide, whereincomplete oligosaccharides were still transferred to Asnresidues of secretory proteins by the OTase complex,albeit with reduced efficiency (Turco et al., 1977; Munozet al., 1994). More recently, relaxed substrate specificitywas also demonstrated for PglB from Campylobacter,where undecaprenyl pyrophosphate-linked saccharidesof various lengths were transferred to a peptide substrate(Glover et al., 2005b).

While none of the other seven genes which were inac-tivated in this study produced an obvious change to flagel-lin and S-layer glycosylation profiles, the high reversionrate observed for Mv1108 mutants and the inability toobtain mutants in the Mv1751 gene suggest that thesegenes could be involved in process(es) which affect cellfitness or are essential to cell viability respectively. Theprotein encoded by Mv1751 appears to be a glycosyltransferase which has homology to the eukaryoticenzyme N-acetylglucosamine-1-phosphate transferase(Alg7/GPT). Alg7/GPT catalyses the linkage of UDP-

activated N-acetylglucosamine to dolichol phosphate,which is the first step of N-linked glycan synthesis ineukaryotes (Eckert et al., 1998). As the linking sugar ofthe M. voltae glycan is also N-acetylglucosamine (Voisinet al., 2005), and Archaea appear to use dolichyl mono-phosphate as their anchoring lipid carrier (Lechner et al.,1985), the possibility of Mv1751 carrying out a homolo-gous reaction to Alg7/GPT is plausible (Fig. 7). However,the knock-out of this gene has been unsuccessful to dateand experimental confirmation of this role is still inprogress. As for a possible role of Mv1108, BLAST analysisof this gene indicates that it is similar to heteropolysac-charide export proteins. Mv1108 may therefore beinvolved in some export process that is important to thecell, leading to its rapid reversion to maintain cell fitness.

Focusing on the assembly and attachment pathway, ourproposed model for M. voltae N-linked glycosylation isconsistent with current theories (Eichler and Adams,2005) and is presented in Fig. 7. In this current study, weclearly demonstrate that Mv151 (aglA) and Mv1749 (aglB)are involved in the glycosylation process. This study rep-resents a significant initial step in understanding post-translational modifications in archaeal proteins, as it is thefirst time any gene has been experimentally shown to beinvolved in glycosylation within the domain Archaea.

Findings in this study also illuminate many challengesthat will have to be faced in the continued study ofM. voltae N-glycosylation. Analysis of the genomes ofM. voltae and its close relatives Methanococcus maripalu-dis (Hendrickson et al., 2004) and Methanocaldococcusjannaschii (Bult et al., 1996) reveals a plethora of genesannotated as involved in various aspects of sugar/glycansynthesis and assembly. Many appear to have similarity tobacterial genes involved in lipopolysaccharide, capsuleand cell wall synthesis. However, methanococci do notmake these various structures, leaving this rather long listof genes open to a possible revised role in archaealN-linked glycosylation. In addition, mutations that disruptthe formation of the N-linked glycan not only affect properflagella assembly, but also have an effect on the S-layer

Fig. 7. Proposed model for flagellin andS-layer M. voltae N-linked glycan assemblyand attachment. Steps 1–3 diagram theassembly of the trisaccharide via glycosyltransferases (AglA for step 3) onto a lipidcarrier at the cytoplasmic face of thecytoplasmic membrane. Step 4 represents thetranslocation of the glycan to the exterior ofthe membrane via a flippase enzyme. Finally,step 5 shows the attachment of the completeglycan to the target protein via an STT3oligosaccharyl transferase (AglB).1

2

3 - AglA

4N-X-S/T

flagellin

Cytoplasmic

S-layer

5 - AglB

membrane

Archaeal glycosylation genes 265

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Molecular Microbiology, 61, 259–268

protein. Without the benefit of additional cell wall struc-tures, like peptidoglycan, most archaeal cells (includingM. voltae) have only the S-layer proteins external to thecytoplasmic membrane to maintain the cell’s structuralintegrity. S-layer proteins interact to create a crystalline-like lattice surrounding the cell. Mutations that ultimatelyalter the structure of the S-layer protein could affect thisprotein interaction and disrupt the structural integrity ofthe cell. As a result, such mutants could suffer serious cellstability problems. It is this effect on the S-layer that likelyexplains why the agl mutants are so unstable. This wouldbe unlike effects on S-layers in bacteria where the cellshave structural integrity provided by the peptidoglycanlayer. In these cases, S-layers are unnecessary and,indeed, are often ‘lost’ by cells grown under laboratoryconditions. Despite future challenges, the results from thisstudy establish a solid foundation upon which to build anunderstanding of N-linked glycosylation pathways inArchaea.

Experimental procedures

Microbial strains and growth conditions

Methanococcus voltae strain PS was grown in Balch medium3 at 37°C, under an atmosphere of CO2/H2 (20:80) andsupplemented with 7.5 mg ml-1 puromycin (MP Biomedicals,Solon, OH) when necessary. E. coli DH5a was used for allin vitro cloning and was grown in Luria–Bertani medium at37°C and supplemented with 100 mg ml-1 ampicillin whennecessary.

Construction of knock-out plasmids

All sequence data referred to in this manuscript can be foundin GenBank/DDBJ/EMBL databases under the AccessionNumbers DQ372941–DQ372943. With the exception of theplasmid for Mv152, knock-out plasmids were constructed asinsertional activation plasmids as described by Thomas et al.(2001). Briefly, an internal portion of the target gene wasamplified via PCR, using primers that incorporated terminalEcoRI restriction sites. These fragments were then clonedinto the corresponding EcoRI site within the pac/pUC vector(Thomas et al., 2001), transformed into E. coli, and screened.Fidelity of the cloned fragment was confirmed by sequencing.For Mv152, the vector pPAC60 (Thomas et al., 2002) wasused as the backbone plasmid for knock-out creation vialinear gene replacement utilizing a puromycin resistance(purR) cassette. Briefly, 1 kb of sequence upstream anddownstream of Mv152 (800 bp outside the coding region,200 bp inside the coding region) was amplified via PCR, withthe addition of restriction enzyme sites, and ligated intopPAC60. The upstream 1 kb fragment was ligated into themultiple cloning site (mcs), at HindIII and HaeII, before thepurR cassette, while the downstream 1 kb fragment wasligated into the mcs at KpnI and SalI, after the purR cassette.This created a construct with 2 kb of M. voltae sequence, withthe internal portion of Mv152 being replaced by the purR

cassette. Fragment orientation and sequence was confirmedafter transformation into E. coli and screening. This constructwas linearized by restriction digest (at HindIII) beforetransformation.

Generation of M. voltae mutants

A liposome delivery method first described for Methanosa-rcina acetivorans (Metcalf et al., 1997) was used with minormodifications (Thomas et al., 2001) for all transformations ofM. voltae. Single colonies were picked anaerobically andtransferred into Balch medium 3 containing puromycin forsubsequent analyses.

Southern hybridization analyses

Chromosomal DNA was isolated from wild-type and trans-formant cultures of M. voltae as previously described(Gernhardt et al., 1990), digested with restriction enzymes,electrophoresed and transferred to nylon membranes (RocheMolecular Biochemicals) by the capillary transfer method.Digoxigenin (DIG)-labelled probes were generated by PCRamplification of internal gene regions (identical to gene frag-ments used for cloning) using DIG-dUTP (Roche MolecularBiochemicals) as recommended by the manufacturer. South-ern hybridizations were carried out at 55°C and stringencywashes and developing were done as described by Thomaset al. (2001).

SDS-PAGE and immunoblotting

Whole cell preparations of M. voltae were subjected to SDS-PAGE as described by Laemmli (1970). Gels were eitherstained with Coomassie brilliant blue G250 (Faguy et al.,1996) or transferred to Immobilon-P transfer membranes(Millipore, Bedford, MA) as described by Towbin et al. (1979).Immunoblots were developed using chicken polyclonal anti-FlaB2 antibodies (shown to react with the major flagellinsFlaB1 and FlaB2; Bayley and Jarrell, 1999) as the primaryantibody and a horseradish peroxidase-conjugated rabbitanti-chicken IgY (Jackson Immunoresearch Laboratories,West Grove, PA) as the secondary antibody. Blots weredeveloped with a chemiluminescence kit (Roche MolecularBiochemicals) according to the manufacturer’s instructions.

Preflagellin peptidase assay

A preflagellin peptidase assay previously described (Bayleyand Jarrell, 1999; Correia and Jarrell, 2000) was used togenerate two versions of unglycosilated M. voltae FlaB2flagellin. E. coli membranes containing overexpressed pre-flagellin FlaB2 were used as substrate with M. voltae mem-branes as the source of the preflagellin peptidase, FlaK. Theunprocessed and processed forms of FlaB2 (with or withoutthe signal peptide) could be distinguished by immunoblot.

Flagella extraction

Flagella were sheared from intact M. voltae cells asdescribed by Kalmokoff et al. (1988).

266 B. Chaban et al.

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Molecular Microbiology, 61, 259–268

nanoLC-MS/MS analysis

Flagella samples were analysed by nanoLC-MS/MS asdescribed by Voisin et al. (2005). Briefly, flagella extractswere trypsin digested and analysed using a Q-TOF 2 hybridquadrupole time-of-flight mass spectrometer coupled to aCapLC capillary HPLC system (Waters, Milford, MA). Themass spectrometer was set to automatically acquire MS/MSspectra on doubly, triply and quadruply charged ions. AllMS/MS spectra were examined manually for the presence ofunusual modifications.

Electron microscopy

Intact M. voltae cells were fixed with 4% glutaraldehyde priorto absorption onto formvar-coated gold grids. Negative stain-ing was done with 2% phosphotungstic acid (pH 7.0). Gridswere viewed on a Hitachi H-700 electron microscope operat-ing at 75 kV.

Acknowledgements

We gratefully thank J. Leigh and W.B. Whitman for access tothe draft M. voltae genome. B.C. was supported by a NaturalSciences and Engineering Research Council (NSERC) Post-Graduate Scholarship. This work was supported by an oper-ating grant from NSERC (to KFJ) and by NRC’s Genomicsand Health Initiative (SML, SV).

References

Bardy, S.L., and Jarrell, K.F. (2003) Cleavage of preflagellinsby an aspartic acid signal peptidase is essential for flagel-lation in the archaeon Methanococcus voltae. Mol Micro-biol 50: 1339–1347.

Bardy, S.L., Mori, T., Komoriya, K., Aizawa, S., and Jarrell,K.F. (2002) Identification and localization of flagellins FlaAand FlaB3 within flagella of Methanococcus voltae. J Bac-teriol 184: 5223–5233.

Bayley, D.P., and Jarrell, K.F. (1999) Overexpression ofMethanococcus voltae flagellin subunits in Escherichia coliand Pseudomonas aeruginosa: a source of archaealpreflagellin. J Bacteriol 181: 4146–4153.

Berghöfer, Y., and Klein, A. (1995) Insertional mutations inthe hydrogenase vhc and frc operons encoding selenium-free hydrogenases in Methanococcus voltae. Appl EnvironMicrobiol 61: 1770–1775.

Bult, C.J., White, O., Olsen, G.J., Zhou, L., Fleischmann,R.D., Sutton, G.G., et al. (1996) Complete genomesequence of the methanogenic archaeon, Methanococcusjannaschii. Science 273: 1058–1073.

Correia, J.D., and Jarrell, K.F. (2000) Posttranslational pro-cessing of Methanococcus voltae preflagellin by preflagel-lin peptidases of M. voltae and other methanogens. JBacteriol 182: 855–858.

Eckert, V., Blank, M., Mazhari-Tabrizi, R., Mumberg, D.,Funk, M., and Schwarz, R.T. (1998) Cloning and functionalexpression of the human GlcNAc-1-P transferase, theenzyme for the committed step of the dolichol cycle, by

heterologous complementation in Saccharomycescerevisiae. Glycobiology 8: 77–85.

Eichler, J., and Adams, M.W.W. (2005) Posttranslationalprotein modification in Archaea. Microbiol Mol Biol Rev 69:393–425.

Faguy, D.M., Bayley, D.P., Kostyukova, A.S., Thomas, N.A.,and Jarrell, K.F. (1996) Isolation and characterization offlagella and flagellin proteins from the thermoacidophilicarchaea Thermoplasma volcanium and Sulfolobusshibatae. J Bacteriol 178: 902–905.

Gemmill, T.R., and Trimble, R.B. (1999) Overview of N- andO-linked oligosaccharide structures found in various yeastspecies. Biochim Biophys Acta 1426: 227–237.

Gernhardt, P., Possot, O., Foglino, M., Sibold, L., and Klein,A. (1990) Construction of an integration vector for use inthe archaebacterium Methanococcus voltae and expres-sion of a eubacterial resistance gene. Mol General Genet221: 273–279.

Glover, K.J., Weerapana, E., and Imperiali, B. (2005a) In vitroassembly of the undecaprenylpyrophosphate-linked hep-tasaccharide for prokaryotic N-linked glycosylation. ProcNatl Acad Sci USA 102: 14255–14259.

Glover, K.J., Weerapana, E., Numao, S., and Imperiali, B.(2005b) Chemoenzymatic synthesis of glycopeptides withPgIB, a bacterial oligosaccharyl transferase from Campy-lobacter jejuni. Chem Biol 12: 1311–1315.

Hendrickson, E.L., Kaul, R., Zhou, Y., Bovee, D., Chapman,P., Chung, J., et al. (2004) Complete genome sequence ofthe genetically tractable hydrogenotrophic methanogenMethanococcus maripaludis. J Bacteriol 186: 6956–6969.

Hofmann, K., and Stoffel, W. (1993) TMbase – A database ofmembrane spanning proteins segments. Biol ChemHoppe-Seyler 374: 166.

Jarrell, K.F., Bayley, D.P., Florian, V., and Klein, A. (1996)Isolation and characterization of insertional mutations inflagellin genes in the archaeon Methanococcus voltae. MolMicrobiol 20: 657–666.

Kalmokoff, M.L., Jarrell, K.F., and Koval, S.F. (1988) Isolationof flagella from the archaebacterium Methanococcusvoltae by phase separation with Triton X-114. J Bacteriol170: 1752–1758.

Koval, S.F., and Jarrell, K.F. (1987) Ultrastructure and bio-chemistry of the cell wall of Methanococcus voltae. J Bac-teriol 169: 1298–1306.

Laemmli, U.K. (1970) Cleavage of structural proteins duringthe assembly of the head of bacteriophage T4. Nature 227:680–685.

Lechner, J., and Sumper, M. (1987) The primary structure ofa procaryotic glycoprotein. Cloning and sequencing of thecell surface glycoprotein gene of halobacteria. J Biol Chem262: 9724–9729.

Lechner, J., Wieland, F., and Sumper, M. (1985) Biosynthe-sis of sulfated saccharides N-glycosidically linked to theprotein via glucose. J Biol Chem 260: 860–866.

Mescher, M.F., and Strominger, J.L. (1976) Purification andcharacterization of a prokaryotic glucoprotein from the cellenvelope of Halobacterium salinarium. J Biol Chem 251:2005–2014.

Mescher, M.F., Strominger, J.L., and Watson, S.W. (1974)Protein and carbohydrate composition of the cell envelopeof Halobacterium salinarium. J Bacteriol 120: 945–954.

Archaeal glycosylation genes 267

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Molecular Microbiology, 61, 259–268

Metcalf, W.W., Zhang, J.K., Apolinario, E., Sowers, K.R., andWolfe, R.S. (1997) A genetic system for Archaea of thegenus Methanosarcina: liposome-mediated transformationand construction of shuttle vectors. Proc Natl Acad SciUSA 94: 2626–2631.

Munoz, M.D., Hernandez, L.M., Basco, R., Andaluz, E., andLarriba, G. (1994) Glycosylation of yeast exoglucanasesequons in alg mutants deficient in the glucosylation stepsof the lipid-linked oligosaccharide. Presence of glucotrioseunit in Dol-PP-GlcNAc2Man9Glc3 influences both glycosy-lation efficiency and selection of N-linked sites. BiochimBiophys Acta 1201: 361–366.

Robbins, P.W. (1999) Yeast glycosylation – foundations andbuilding blocks. Biochim Biophys Acta 1426: 225–226.

Schmidt, M.A., Riley, L.W., and Benz, I. (2003) Sweet newworld: glycoproteins in bacterial pathogens. Trends Micro-biol 11: 554–561.

Sumper, M. (1987) Halobacterial glycoprotein biosynthesis.Biochim Biophys Acta 906: 69–79.

Szymanski, C.M., and Wren, B.W. (2005) Protein glycosyla-tion in bacterial mucosal pathogens. Nat Rev Microbiol 3:225–237.

Szymanski, C.M., Logan, S.M., Linton, D., and Wren, B.W.(2003) Campylobacter – a tale of two protein glycosylationsystems. Trends Microbiol 11: 233–238.

Thomas, N.A., Pawson, C.T., and Jarrell, K.F. (2001) Inser-tional inactivation of the flaH gene in the archaeon Metha-

nococcus voltae results in non-flagellated cells. Mol GenetGenomics 265: 596–603.

Thomas, N.A., Mueller, S., Klein, A., and Jarrell, K.F. (2002)Mutants in flaI and flaJ of the archaeon Methanococcusvoltae are deficient in flagellum assembly. Mol Microbiol46: 879–887.

Towbin, H., Staehelin, T., and Gordon, J. (1979) Electro-phoretic transfer of proteins from polyacrylamide gels tonitrocellulose sheets: procedure and some applications.Proc Natl Acad Sci USA 76: 4350–4354.

Turco, S.J., Stetson, B., and Robbins, P.W. (1977) Compara-tive rates of transfer of lipid-linked oligosaccharides toendogenous glycoprotein acceptors in vitro. Proc Natl AcadSci USA 74: 4411–4414.

Voisin, S., Houliston, R.S., Kelly, J., Brisson, J.R., Watson,D., Bardy, S.L., et al. (2005) Identification and character-ization of the unique N-linked glycan common to the flagel-lins and S-layer glycoprotein of Methanococcus voltae. JBiol Chem 280: 16586–16593.

Yan, Q., and Lennarz, W.J. (2002) Studies on the function ofoligosaccharyl transferase subunits. Stt3p is directlyinvolved in the glycosylation process. J Biol Chem 277:47692–47700.

Zufferey, R., Knauer, R., Burda, P., Stagljar, I., te Heesen, S.,Lehle, L., and Aebi, M. (1995) STT3, a highly conservedprotein required for yeast oligosaccharyl transferase activ-ity in vivo. EMBO J 14: 4949–4960.

268 B. Chaban et al.

© 2006 The AuthorsJournal compilation © 2006 Blackwell Publishing Ltd, Molecular Microbiology, 61, 259–268