Cj1123c (PglD), a multifaceted acetyltransferase from … › 6d9d › 5b420dd56c520f0d1bfe40d… · Cj1123c (PglD), a multifaceted acetyltransferase from Campylobacter jejuni Melinda

Cj1123c (PglD), a multifaceted acetyltransferasefrom Campylobacter jejuni

Melinda Demendi and Carole Creuzenet

Abstract: Campylobacter jejuni produces both N- and O-glycosylated proteins. Because protein glycosylation contributesto bacterial virulence, a thorough characterization of the enzymes involved in protein glycosylation is warranted to assesstheir potential use as therapeutic targets and as glyco-engineering tools. We performed a detailed biochemical analysis ofthe molecular determinants of the substrate and acyl-donor specificities of Cj1123c (also known as PglD), an acetyltrans-ferase of the HexAT superfamily involved in N-glycosylation of proteins. We show that Cj1123c has acetyl-CoA-depend-ent N-acetyltransferase activity not only on the UDP-4-amino-4,6-dideoxy-GlcNAc intermediate of the N-glycosylationpathway but also on the UDP-4-amino-4,6-dideoxy-AltNAc intermediate of the O-glycosylation pathway, implying func-tional redundancy between both pathways. We further demonstrate that, despite its somewhat relaxed substrate specificityfor N-acetylation, Cj1123c cannot acetylate aminoglycosides, indicating a preference for sugar-nucleotide substrates. In ad-dition, we show that Cj1123c can O-acetylate UDP-GlcNAc and that Cj1123c is very versatile in terms of acyl-CoA do-nors as it can use propionyl- and butyryl-CoA instead of acetyl-CoA. Finally, using structural information available forCj1123c and related enzymes, we identify three residues (H125, G143, and G173) involved in catalysis and (or) acyl-donorspecificity, opening up possibilities of tailoring the specificity of Cj1123c for the synthesis of novel sugars.

Key words: Campylobacter jejuni, acetyltransferase, protein glycosylation, Cj1123c.

Resume : Campylobacter jejuni produit des proteines N- et O-glycosylees. Puisque la glycosylation contribue a la viru-lence des bacteries, une caracterisation rigoureuse des enzymes impliquees dans la glycosylation des proteines est justifieepour evaluer leur utilisation potentielle comme cibles therapeutiques ou comme outils en ingenierie des sucres. Nous avonsrealise une analyse biochimique detaillee des determinants moleculaires de la specificite des substrats et des donneursd’acyles de Cj1123c (connue aussi sous l’appellation PglD), une acyltransferase de la suprafamille HexAT impliquee dansla N-glycosylation des proteines. Nous montrons que Cj1123c possede une activite acyltransferase dependante de l’acetyl-CoA, non seulement sur l’intermediaire UDP-4-amino-4,6-dideoxy-GlcNAc de la voie de N-glycosylation mais aussi surl’intermediaire UDP-4-amino-4,6-dideoxy-AltNAc de la voie de O-glycosylation, ce qui implique une redondance fonction-nelle entre les deux voies. Nous demontrons de plus que, malgre sa specificite assez relachee pour la N-acetylation,Cj1123c ne peut acetyler les aminoglycosides, ce qui indique sa preference pour des substrats de type sucre-nucleotide. Deplus, nous montrons que Cj1123c peut O-acetyler l’UDP-GlcNAc et est aussi versatile en termes de donneurs d’acyl-CoAdu fait qu’elle peut utiliser le propionyl- et le butyryl-CoA a la place de l’acetyl-CoA. Finalement, en utilisant l’infor-mation disponible sur Cj1123c et les enzymes apparentees, nous identifions trois residus (H125, G143 et G173) impliquesdans la catalyse et (ou) dans la specificite du donneur d’acyles, ouvrant des possibilites de modifier la specificite deCj1123c pour synthetiser de nouveaux sucres.

Mots-cles : Campylobacter jejuni, acetyltransferase, glycosylation des proteines, Cj1123c.

[Traduit par la Redaction]

Introduction

Campylobacter jejuni has become the prototype for thestudy of bacterial protein glycosylation, being thus far theonly identified bacterium that has both O- and N-linked pro-tein-glycosylation pathways. The O-glycosylation pathway isdedicated to the glycosylation of flagellins via pseudaminicacid and its derivatives (Creuzenet 2004; Obhi and Creuze-net 2005; Thibault et al. 2001) and is very similar to thatobserved in Helicobacter pylori (Creuzenet et al. 2000b;

Schoenhofen et al. 2006c). In contrast, the N-glycosylationpathway targets multiple proteins that are modified by a di-acetamidobacillosamine-containing heptasaccharide (Szy-manski et al. 1999; Young et al. 2002). N-glycosylation ofbacterial proteins appears to be unique to C. jejuni, with theexception of the flagellin N-glycosylation pathway found inarchaebacteria (Voisin et al. 2005). Both protein glycosyla-tion pathways utilize the same common sugar-nucleotideprecursor, UDP-GlcNAc, and comprise the same three enzy-matic steps of C6 dehydration, C4 amination, and 4-N-acety-

Received 20 November 2008. Revision received 10 January 2009. Accepted 15 January 2009. Published on the NRC Research PressWeb site at bcb.nrc.ca on 25 April 2009.

M. Demendi and C. Creuzenet.1 Department of Microbiology and Immunology, Infectious Diseases Research Group, University ofWestern Ontario, London, ON N6A 5C1, Canada.

1Corresponding author (e-mail: [email protected]).

469

Biochem. Cell Biol. 87: 469–483 (2009) doi:10.1139/O09-002 Published by NRC Research Press

lation. Differences in the stereochemistry of the C6 dehydra-tases result in the formation of two distinct products by eachpathway: UDP-diacetamidobacillosamine for N-glycosyla-tion and UDP-4-acetamido-4,6-dideoxy-AltNAc, which isfurther modified into CMP-pseudaminic acid, for O-glycosy-lation (Creuzenet 2004; Goon et al. 2003; Obhi and Creuze-net 2005; Olivier et al. 2006; Schoenhofen et al. 2006a,2006b; Vijayakumar et al. 2006).

The enzymes involved in each of these pathways have po-tential as therapeutic targets because each pathway contrib-utes to virulence and is unique to pathogens ( Goon et al.2003; Szymanski et al. 1999, 2002; Vijayakumar et al.2006). These enzymes also have potential as engineeringtools for the production of carbohydrate antigens for vacci-nation purposes because they allow the synthesis of uniquenucleotide-activated sugars predominantly incorporated incomplex polysaccharides (glycoproteins or glycolipids)present on the surface of pathogens. The chemical synthesisof such polysaccharides is difficult and costly, and enzy-matic synthesis represents a promising alternative (Gloveret al. 2005a, 2005b). Acetyltransferases are particularly im-portant because acetylation can modulate the antigenicity ofpolysaccharides, affect the efficiency of opsonic killing ofbacteria (Bhasin et al. 1998; Cerca et al. 2007; Gudlavalletiet al. 2007), and affect bacterial surface properties essentialfor virulence (Bystricky and Szu 1994). While O-acetylatedpolysaccharides are common (Bhasin et al. 1998; Deszo etal. 2005; Kooistra et al. 2001; Lewis et al. 2007; MacLeanet al. 2006; Yildirim et al. 2005), N-acetylated surface poly-saccharides are more scarce beyond the frequent 2-N-acety-lation of sugars that arises directly from UDP-GlcNAc. Thefact that N-acetylated polysaccharides are often associatedwith pathogens (Allen and Maskell 1996; Di Fabio et al.1992; Knirel et al. 2006; Rocchetta et al. 1999) furtherunderscores that N-acetylation might play a role in virulenceand that N-acetyltransferases might also have value as thera-peutic targets. Detailed biochemical studies are warranted tofully harness the potential of such N-acetyltransferases via abetter understanding of the molecular basis for substrate and,when applicable, donor specificity.

The sequence of Cj1123c, the N-acetyltransferase of theUDP-diacetamidobacillosamine biosynthesis pathway, con-tains several incomplete hexapeptide repeats that trigger theformation of a characteristic left-handed b-helical structurefound in members of the hexapeptide acyltransferase(HexAT) superfamily (Beaman et al. 1997; Raetz and Ro-derick 1995; Vaara 1992). Structural data available forCj1123c since 2006 (Protein Data Bank No. 2npo ) or pub-lished recently (Olivier and Imperiali 2008; Rangarajan etal. 2008) confirmed this sequence-based structural predictionand further demonstrated the existence of the characteristictrimer that is necessary for the formation of the donor- andsubstrate-binding site in other acyltransferases of the HexATfamily.

HexAT enzymes are fairly versatile with regard to theirsubstrates (simple sugars versus sugar-nucleotides that canbe acylated or not or other complex structures), acyl donors(small groups derived from coenzyme A (CoA) donors ver-sus bulky fatty acids derived from acyl carrier proteins), andreaction chemistry (O- vs. N-acetylation) despite the conser-vation of structural features (Anderson and Raetz 1987; An-

derson et al. 1985; Brown et al. 1999; Buetow et al. 2007;Coleman and Raetz 1988; Gehring et al. 1996; Kelly et al.1993; Mengin-Lecreulx and van Heijenoort 1994; Olsen etal. 2007; Rende-Fournier et al. 1993; Rund et al. 1999; Su-gantino and Roderick 2002; Wenzel et al. 2005; Williamsand Raetz 2007).

Using the structural information available for Cj1123c andother HexATs as well as using direct biochemical analyses,we investigated the molecular determinants of the substrateand donor specificities of Cj1123c. We show that Cj1123chas N-acetyltransferase activity on the UDP-4-amino-4,6-dideoxy-GlcNAc (UDP-amino-dGlcNAc) intermediate ofthe N-glycosylation pathway, and we further demonstratethat Cj1123c is a much more versatile enzyme than antici-pated. We show not only that Cj1123c has relaxed substratespecificity for N-acetylation but also that it can act as an O-acetyltransferase and is able to use various acyl-CoA donors.Finally, we identify residues involved in catalysis and acyl-donor specificity, opening up possibilities of tailoring thespecificity of Cj1123c for the synthesis of novel sugars.

Materials and methods

Cloning of histidine-tagged Cj1123c in the pETexpression system

The open reading frame of cj1123c was amplified by PCRfrom the chromosomal DNA of C. jejuni ATCC 700819 us-ing primers P1 and P2, which contained NcoI and BamHIrestriction sites, respectively (Table 1). PCR was performedwith expand long-range template DNA polymerase (RocheDiagnostics) according to the manufacturer’s specificationswith annealing at 60 8C. The PCR fragment was ligatedinto a NcoI- and BamHI-cut pUC18 plasmid using standardprocedures, and the construct was transferred into calciumchloride competent Escherichia coli DH5a. Potential clonescarrying the cj1123c/pUC18 construct were selected on Lu-ria–Bertani (LB) agar containing 100 mg/mL ampicillin. Thecj1123c insert was then transferred from cj1123c/pUC18into the pET23 expression plasmid derivative (Newton andMangroo 1999) using the NcoI and BamHI restriction sitesand following the same procedures. The DNA sequences ofall constructs were confirmed by sequencing using the T7promoter primer. Sequencing was performed at the RobartsResearch Institute (London, Ontario).

The plasmids encoding the N-terminally histidine-taggeddehydratase Cj1293 and aminotransferases Cj1121c andCj1294 have been described previously (Creuzenet 2004;Obhi and Creuzenet 2005; Vijayakumar et al. 2006).

Mutagenesis of Cj1123cPoint mutations were generated in Cj1123c using the

QuikChange procedure following the manufacturer’s instruc-tions (Stratagene), except that the expand long-range tem-plate DNA polymerase was used and six cycles of PCRwere performed using each primer independently beforepooling the reactions and allowing the PCR to resume for19 more cycles as described previously (Demendi et al.2005). The cj1123c/pET23a construct described above wasused as a template and the primers used are listed in Table 1.Potential mutants were verified by DNA sequencing withthe T7 promoter primer.

470 Biochem. Cell Biol. Vol. 87, 2009

Published by NRC Research Press

Protein expression and purificationAll plasmids encoding the dehydratase, aminotransferases,

or acetyl-transferase were transformed into E. coliBL21(DE3)pLysS grown on LB containing 100 mg/mL am-picillin and 34 mg/mL chloramphenicol. The cells weregrown to an optical density of 0.6 (measured at 600 nm)and protein expression was induced with 0.1 mmol/L IPTGovernight at 30 8C. The cells were harvested and stored at –20 8C until needed.

For enzyme purifications, the cell pellets were resus-pended in 30 mL of binding buffer (50 mmol/L HEPES(pH 7.5), 50 mmol/L imidazole, 100 mmol/L NaCl) contain-ing 6 mg/mL lysozyme and incubated on ice for 10 min.The cells were lysed by three passages through a FrenchPress cell at 1500 psi. Cell debris and insoluble componentswere removed by ultracentrifugation for 45 min at 127 000gat 4 8C. The soluble enzymes were purified by nickel chela-tion as described previously (Creuzenet 2004; Vijayakumaret al. 2006). Cj1123c was further purified by cation-ex-change chromatography using a 1 mL MonoS column(Amersham) equilibrated in 50 mmol/L sodium phosphate(pH 6.0) containing 50 mmol/L NaCl. The protein waseluted from the column with a linear gradient of NaCl(50 mmol/L –1 mol/L in 30 column volumes). The fractionscontaining Cj1123c were pooled and dialyzed overnight(cutoff 12.5–14 kDa) into 50 mmol/L HEPES (pH 7.5).

The protein concentration was determined using BioRadprotein assay reagent. The proteins were analyzed by SDS–PAGE on 10% acrylamide gels with detection by silverstaining and Western blotting with anti-histidine tag anti-body (used at 1:3000 dilution) (Sigma) using standard pro-cedures. Detection was performed using a goat anti-mouseIgG antibody (at 1:2000) conjugated to AlexaFluor 680 andan Odyssey infrared imaging system (LI-COR Biosystems).

Mass spectrometry analysis of Cj1123cThe purified protein was analyzed by matrix-assisted laser

desorption–ionization (MALDI) mass spectrometry (MS) us-ing a 4700 Proteomics analyzer (Applied Biosystems, FosterCity, California) in the linear positive-ion mode to confirmits size. The MALDI matrix, a-cyano-4-hydroxycinnamicacid, was prepared as 5 mg/mL in 6 mmol/L ammoniumphosphate monobasic, 50% acetonitrile, and 0.1% trifluoro-acetic acid and mixed with the sample in a ratio of 1:1(v/v). Data acquisition and data processing were done usingthe 4000 Series Explorer and Data Explorer (both from

Applied Biosystems), respectively. The analysis was per-formed at the MALDI MS facility at the University ofWestern Ontario (London, Ontario).

Gel filtration analysisGel filtration analysis was carried out on a Superose 12

(GE Healthcare) column (10 mm diameter, 30 cm length)run at 0.5 mL/min in 50 mmol/L sodium phosphate and0.15 mol/L NaCl (pH 7.5) on an Akta purifier. The molecu-lar weight markers (Sigma, MW-GF-200) were bovine se-rum albumin (66 kDa), b-amylase (200 kDa), carbonicanhydrase (29 kDa), and cytochrome c (12.4 kDa). For eachrun, a 500 mL sample containing each of the markers dilutedat 0.1 mg/mL, with or without Cj1123c (wild-type or mu-tant) also at 0.1 mg/mL, was injected onto the column. Pro-tein elution was recorded by UV monitoring at 210 and280 nm. The relative proportions of trimers and monomerswere determined by integration of the area under each peak(from baseline to baseline) obtained after subtraction of thecontrol curve (markers only).

Synthesis and purification of UDP amino sugarsFor the synthesis of UDP-amino-dGlcNAc and UDP-4-

amino-4,6-dideoxy-AltNAc (UDP-amino-dAltNAc), the en-zymatic reactions contained 100 mg of dehydratase Cj1293,100 mg of aminotransferase (Cj1121c or Cj1294 as appropri-ate), 20 mmol/L UDP-GlcNAc (Calbiochem, >95% purity),40 mmol/L glutamate (pH 7), 10 mmol/L pyridoxal phos-phate, and 100 mmol/L Bis–Tris–propane buffer (pH 7.5) ina total volume of 1000 mL. The resulting UDP amino sugarswere purified by anion-exchange chromatography using a1 mL High Q Econopac column (BioRad) and a linear gra-dient of 50 mmol/L to 1 mol/L triethylammonium bicarbon-ate (pH 8.5) in 40 column volumes at 1 mL/min. Thepresence and purity of the expected UDP amino sugars inthe fractions were checked by capillary electrophoresis (CE)as described previously (Creuzenet et al. 2000a; 2000b) andthe appropriate fractions were pooled together and lyophi-lized. The substrates were also analyzed by MS (see below)to ensure identity and purity.

Cj1123c in vitro enzyme activity assaysTypical enzymatic reactions were carried out in 20 mL at

37 8C in 125 mmol/L Bis–Tris–propane buffer at pH 8, un-less specified otherwise. The amount of enzyme, substrate,CoA, and donors and the length of incubation time used for

Table 1. Name and sequences of primers used throughout this study.

Primer name Primer sequenceCj1123c P1 5’-AGGGTCCATGGCAAGAACTGAA AAAATTTATAT-3’Cj1123c P2 5’-GCGTCGGATCCTTACATCCTTTTTGCAGGTACTC-3’H125A forward 5’-CAAGCGTAATTGAGGCTGAATGTGTGATAGG-3’H125A reverse 5’-CCTATCACACATTCAGCCTCAATTACGCTTG-3’G143I forward 5’-GGGAGCTAAATGTGCGATTAATGTAAAAATTGG-3’G143I reverse 5’-CCAATTTTTACATTAATCGCACATTTAGCTCCC-3’G143V forward 5’-GAGCTAAATGTGCGGTTAATGTAAAAATTGG-3’G143V reverse 5’-CCAATTTTTACATTAACCGCACATTTAGCTC-3’G173M forward 5’-GCAGATGATAGTATTTTAGGTATGGGAGCAACTTTAGTTAAAAA-3’G173M reverse 5’-TTTTTAACTAAAGTTGCTCCCATACCTAAAATACTATCATCTGC-3’

Note: When applicable, the restrictions sites are underlined and the mutated nucleotides are in italics.

Demendi and Creuzenet 471


the reactions are indicated in the figure legends. The reac-tions were quenched by snap freezing in an ethanol – dryice bath and were subsequently analyzed by CE as describedpreviuosly (Creuzenet et al. 2000a, 2000b). The Cj1123c re-action products were purified by anion-exchange chromatog-raphy as described above for the UDP amino sugars. Theywere resuspended in water for MS analysis.

Cj1123c in vivo enzyme activity assays onaminoglycosides

To test the ability of Cj1123c to acetylate aminoglyco-sides in vivo, E. coli BL21(DE3)pLysS harbouring theCj1123c expression plasmid was grown overnight at 37 8Cin LB containing 100 mg/mL ampicillin and 34 mg/mLchloramphenicol, diluted 1:60 (v/v) in fresh media, andgrown until the optical density at 600 nm (OD600 nm) reached0.6. IPTG (0.1 mmol/L final) was added to induce Cj1123cexpression. After 3 h of induction, 200 mL aliquots of cellswere transferred to each well of a 100-well microtiter plateand supplemented with various concentrations of aminogly-cosides. Growth was monitored over time using a BioscreenC automated microbiology growth curve analysis system(MTX Lab Systems, Inc., Vienna, Virginia). Each batch ofcells was tested in triplicate for each condition examined.Three independent experiments were performed. The amino-glycosides tested were kanamycin, gentamycin, neomycin,and streptomycin at concentrations of 0.03 and 0.005 g/Lfor kanamycin representing 10- and 60-fold the minimuminhibitory concentration (MIC), 0.02 and 0.005 g/L for gen-tamycin representing 50- and 100-fold the MIC, 0.025 and0.005 g/L for neomycin representing 20- and 100- fold theMIC, and 0.0025 and 0.005 g/L for streptomycin represent-ing 10- and 50-fold the MIC.

Reactivity of the purified Cj1123c reaction products with2,4,6-trinitrobenzensulfonic acid (TNBS)

The purified Cj1123c reaction products (0.25 nmol) ob-tained by incubation of Cj1123c with the various substratesand CoA donors were incubated for 1 h at 50 8C with 5nmol of TNBS (Sigma) in 100 mmol/L HEPES (pH 7.5) ina total reaction volume of 20 mL. Similar reactions were car-ried out using UDP amino sugars and UDP-GlcNAc as pos-itive and negative controls. The samples were analyzeddirectly by CE after incubation.

MS analyses of the reaction productsMS was performed at the Don Rix MS facility of the Uni-

versity of Western Ontario on a Micromass Q-TOF micro-mass spectrometer equipped with a Z-spray sourceoperating in the negative-ion mode (40 V, 80 8C) as re-ported previously (Obhi and Creuzenet 2005; Vijayakumaret al. 2006).

Results

Overexpression and purification of Cj1123cCj1123c was overexpressed in the pET system (Studier et

al. 1990) as an N-terminally histidine-tagged protein. Theprotein migrated at ~29 kDa, which was larger than the ex-pected size of 22.2 kDa. However, this band was not presentin noninduced samples (data not shown), suggesting that this

was the overexpressed protein of interest that migratedanomalously. About 50% of the protein expressed was foundin the soluble fraction and could be readily purified to ho-mogeneity by nickel chelation followed by cation-exchangechromatography. The purified protein reacted readily withthe anti-histidine tag antibody (Fig. 1A, 1B). When analyzedby MALDI MS, the purified protein exhibited a mass of21.9 kDa (monoisotopic mass, data not shown), which isconsistent with the size expected for Cj1123c and withinthe precision limits for the MALDI MS instrumentation. Noother significant peaks were detected, thereby confirmingthe purity of the sample. Gel filtration analysis showed thatthe protein was present as ~85% trimers of 68 kDa and~15% monomers eluting at 22 kDa (Fig. 1C). The slight ex-tra mass observed on the trimer peak could indicate bindingof the cofactor in the trimer, although this is also within ac-ceptable experimental error of the gel filtration column usedin this mass range. The small peak that eluted at ~22 mLwas probably caused by buffer components rather than byCj1123c because this peak exhibited a high conductivitythat was not observed for the trimer and monomer ofCj1123c and because MS analysis of the Cj1123c prepara-tion indicated that no degradation of the protein had oc-curred. Altogether, the Western blotting data, MALDI–MSanalysis, and gel filtration data confirm the identity of thepurified protein as histidine-tagged Cj1123c, despite itsanomalous migration on SDS–PAGE gels. The purified en-zyme was stable for at least 3 months when stored at –20 8C in the presence of 25% glycerol.

Cj1123c has N-acetyltransferase activity on UDP-4-amino-4,6-dideoxy-GlcNAc

CE analysis demonstrated that Cj1123c is able to useUDP-amino-dGlcNAc as a substrate in the presence of ace-tyl-CoA (AcCoA) (Fig. 2A). The reaction was accompaniedby the release of CoA, consistent with the proposed activityof Cj1123c as an AcCoA-dependent acetyltransferase. Con-version of UDP-amino-dGlcNAc to its reaction product in-creased with the amount of enzyme used over a wide rangeof enzyme concentrations tested and in the presence of ex-cess substrate (Fig. 3A).

MS analysis of the purified reaction product revealed apeak at a mass to charge ratio (m/z) 631.1, which was 42mass units greater than that of the substrate (m/z 589.1)(Figs. 4A, 4B), as expected for the predicted N-acetylationreaction. The tandem MS (MS/MS) pattern of the peak atm/z 631.1 was consistent with acetylation on the glucosemoiety. Specifically, four peaks arising from the glucosering in the substrate (m/z 265, 283, 345, and 589) wereshifted by 42 mass units in the product (m/z 307, 325, 387,and 631), whereas all other peaks assigned to the UDP moi-ety were not affected by the enzymatic reaction.

Furthermore, whereas the substrate readily reacted withTNBS (Vijayakumar et al. 2006), a reagent specific for pri-mary amines, the purified Cj1123c reaction product did notreact with TNBS (data not shown), further confirming thatthe amino group was the target of the acetyl transfer, as ex-pected. Hence, these data demonstrate the AcCoA-mediated4N-acetyltransferase activity of Cj1123c on UDP-amino-dGlcNAc.



Cj1123c also has N-acetyltransferase activity on UDP-amino-dAltNAc

Based on mutagenesis studies carried out on Cj1293 andCj1121c (Goon et al. 2003; Vijayakumar et al. 2006), theN- and O-glycosylation pathways are expected to be fullysegregated and nonredundant. However, Cj1123c is 44%similar (15% identical) to Cj1313, the proposed UDP-amino-dAltNAc N-acetyltransferase of the O-glycosylationpathway (Schoenhofen et al. 2006a). Hence, we testedwhether, based on its high level of similarity to Cj1313,Cj1123c could also use the same substrate as Cj1313. CEanalysis demonstrated that Cj1123c is actually also able touse UDP-amino-dAltNAc as a substrate in an AcCoA-de-pendent manner (Fig. 2B). In contrast with the substrate(Obhi and Creuzenet 2005), the reaction product did not re-act with TNBS, indicating that the primary amine of thesubstrate had been modified (data not shown). MS analysisof the reaction product revealed a peak at m/z 631.3 consis-tent with N-acetylation resulting in the formation of UDP-4-acetamido-4,6-dideoxy-AltNAc (Figs. 4C and 4D). Becausethe mass of this product is identical to that of UDP-4-acet-amido-4,6-dideoxy-GlcNAc and because the MS/MS pat-terns of UDP-amino-dAltNAc and UDP-amino-dGlcNAcare identical (Obhi and Creuzenet 2005; Vijayakumar et al.2006), the MS/MS patterns of their acetylated products arealso expected to be indistinguishable. Hence, CE co-injec-

tion experiments were performed to demonstrate that the re-action products obtained after incubation of Cj1123c withUDP-amino-dGlcNAc or UDP-amino-dAltNAc are distinctmolecules. Indeed, when the two purified products were co-injected, two distinct peaks were obtained (Fig. 2C).

N-acetylation of UDP-amino-dAltNAc was not as effi-cient as that of UDP-amino-dGlcNAc because the substrateconversions obtained under the same conditions (37 8C,pH 8, 1.5 h incubation) were 24% (Fig. 2B) for the formerand 100% for the latter (Fig. 2A). This was further con-firmed by measuring the conversion of each substrate as afunction of enzyme concentration and time. A much longerincubation time and higher enzyme amount were necessaryto achieve the same level of substrate conversion using theUDP-amino-dAltNAc product (Fig. 3A).

Optimal parameters for activity of Cj1123cThe optimal temperature for activity of Cj1123c on UDP-

amino-dGlcNAc was 30 8C, with ~80% activity at 37 8C(Fig. 3B). Surprisingly, the optimal temperature was shiftedto 50 8C in the presence of UDP-amino-dAltNAc, with~70% activity at 37 8C. This suggests potential stabilizationof the enzyme against thermal denaturation upon binding tothe UDP-amino-dAltNAc substrate, which is turned over ata slower rate. The optimal pH was 8.5–9.5 for UDP-amino-dGlcNAc but was shifted to pH 8.0–8.5 for the UDP-amino-dAltNAc substrate (Fig. 3C).

Despite structure-based predictions, His-125 is importantbut not essential for catalysis

As mentioned above, the structural data available forCj1123c at the outset of these studies clearly definedCj1123c as a member of the HexAT family of acyltransfer-ases. In several of these HexATs, the trimeric organizationof the left-handed b-helix results in positioning of a histidineresidue in close proximity to the nitrogen or oxygen atom tobe acylated, resulting in proton abstraction that facilitatesthe interaction with the acyl donor and the direct SN2 attackon the donor thioester (Buetow et al. 2007; Olsen et al.2007; Williams and Raetz 2007; Wyckoff and Raetz 1999).This histidine residue acts as a catalytic base that renders theacceptor site of the substrate a better nucleophile. This histi-dine residue is His-363 in the E. coli GlmU that acetylatesGlcN-1-PO4 as the first step of UDP-GlcNAc synthesis(Brown et al. 1999; Gehring et al. 1996; Mengin-Lecreulxand van Heijenoort 1994; Olsen et al. 2007), His-125 in E.coli LpxA, the acyl carrier protein dependent acetyltransfer-ase LpxA that modifies UDP-GlcNAc with fatty acid for thesynthesis of lipid A (Anderson and Raetz 1987; Anderson etal. 1985; Coleman and Raetz 1988; Williams and Raetz2007), and His-247 in the acyl carrier protein dependentUDP-3-acyl-glucosamine acetyltransferase LpxD (Buetow etal. 2007; Kelly et al. 1993; Rund et al. 1999). A structure-based alignment of the sequence of Cj1123c with LpxA,LpxD, and GlmU allowed us to predict that His-125 mightserve as the catalytic base in Cj1123c because it alignedwell with the catalytic histidine residues of LpxA, LpxD,and GlmU (Fig. 5A). This prediction was also consistentwith the basic optimal pH of activity observed for Cj1123c(Fig. 3C). It was further confirmed by structural studies re-cently published (Olivier and Imperiali 2008; Rangarajan et

Fig. 1. Analysis of purified Cj1123c by SDS–PAGE and gel filtra-tion chromatography. (A and B) SDS–PAGE analysis showing thepurification of histidine-tagged Cj1123c after overexpression in Es-cherichia coli. (A) Silver staining; (B) anti-histidine tag Westernblotting. Lane 1, eluted proteins after nickel-affinity chromatogra-phy; lane 2, purified protein after nickel-affinity chromatographyfollowed by cation-exchange chromatography. (C) Gel filtrationchromatography showing the elution of Cj1123c wild-type (WT)and mutants as a trimer (T) or monomer (M). The elution times ofthe molecular weight markers at 200, 66, and 29 kDa are indicatedby triangles.



al. 2008). A H125A mutant was produced by site-directedmutagenesis. This mutant was still active when tested foracetylation of UDP-amino-dGlcNAc. As high as 60% con-version could be obtained with 500 ng of H125A after 2 hof reaction. A time course performed with three differentamounts of enzyme demonstrated that His-125 is neverthe-less important for catalysis, since 100% substrate conversionwas obtained with as little as 1 ng of wild-type enzymein £15 min, whereas barely 20% conversion was obtainedwith the same amount of mutant enzyme in 2 h (Figs. 6Aand 6B). The optimal pH of this mutant was only slightlyless basic than that of the wild-type enzyme (Fig. 3C), indi-cating that the assay conditions used were suitable for opti-mal activity and do not account for the observed reducedactivity. Hence, the decreased activity does confirm a roleof His-125 in catalysis, but the significant residual activitysuggests that His-125 is not absolutely essential.

Gel filtration analysis revealed that the H125A mutant isslightly impaired in its ability to form trimers, since aboutone third of the protein eluted as a monomer (Fig. 1C). Incontrast, for the wild-type protein, only ~15% of the protein

eluted as monomers. The reduced ability of the H125A mu-tant to trimerize could also in part explain the lower activityof the mutant because trimerization is necessary to assemblethe cofactor binding site.

Cj1123c cannot acetylate aminoglycosidesBased on the unexpected relaxed substrate specificity de-

scribed above and based on the fact that Cj1123c belongs tothe HexAT family that comprises numerous aminoglycosideacetyltransferases, we investigated whether Cj1123c coulduse kanamycin, gentamycin, neomycin, or streptomycin assubstrate. Because inactivation of aminoglycosides by N-acetylation is a common mechanism used by bacteria to re-sist exposure to such antibiotics (Wright 2007; Wright andLadak 1997), we first tested whether overexpression ofCj1123c could enhance the resistance of E. coli to theseaminoglycosides at concentrations ranging from 10- to 100-fold the minimum inhibitory concentration. This type of as-say has been used previously to assess the activity of otheracetyltransferases (Draker and Wright 2004). Expression ofCj1123c was induced prior to exposure to the antibiotic and

Fig. 2. Capillary electrophoresis analysis of the N-acetyltransferase activity of Cj1123c on (A) UDP-amino-dGlcNAc and (B) UDP-amino-dAltNAc. and (C) comparison of the reaction products obtained with each substrate. In Figs. 2A and 2B, a typical complete reaction of20 mL contained 1 mg of Cj1123c, 125 mmol/L substrate, and 250 mmol/L AcCoA in 125 mmol/L Bis–Tris–propane buffer (pH 8) (trace c).The reactions were incubated for 1.5 h at 37 8C. Modifications to this composition are no Cj1123c (trace a) or no AcCoA (trace b). In eachpanel, trace d represents a co-injection of the complete reaction shown in trace c with purified substrate. With each substrate (S), a newproduct (P1 or P2) distinct from the starting substrate was generated upon incubation with Cj1123c in an AcCoA donor (D) dependentmanner. Figure 2C demonstrates that the purified reaction products obtained with UDP-amino-dGlcNAc (trace a) and UDP-amino-dAltNAc(trace b) are different and yield two distinct peaks upon co-injection on the capillary electrophoresis (trace c). AU, arbitrary units.

Fig. 3. Dependence of the acetyltransferase activity of Cj1123c on (A) amount of enzyme present in the reaction, (B) reaction temperature,and (C) reaction pH. Solid circles, reactions of wild type with UDP-amino-dGlcNAc, 15 min incubation; solid squares, reactions of wild-type with UDP-amino-dAltNAc, 90 min incubation; open squares, reaction of WT with UDP-amino-dAltNAc, 15 min incubation; open cir-cles, reaction of H125A with UDP-amino-dGlcNAc, 90 min incubation. The reaction composition was as described in Fig. 2 except that inFig. 3A, the amount of enzyme was altered as indicated on the x-axis and in Fig. 3C, the buffer was replaced by 125 mmol/L Bis–Tris–propane buffer that had been adjusted to pH 6.5–8.5 or by 125 mmol/L CHES buffer that had been adjusted to pH 9.0–10.0. All reactionswere incubated at 37 8C.



Fig. 4. Mass spectrometry analysis of the reaction products of Cj1123c obtained after incubation with UDP-amino-dGlcNAc and UDP-amino-dAltNAc. (A) MS/MS spectrum of the substrate UDP-amino-dGlcNAc and (B) MS/MS spectrum of the reaction products obtainedupon incubation of UDP-amino-dGlcNAc with Cj1123c. Several peaks were shifted by 42 mass units compared with the substrate (Fig. 4A)as expected for the acetylation reaction. The dots indicate peaks arising from the UDP moiety. (C) MS spectrum of the substrate UDP-amino-dAltNAc and (D) MS spectrum of the product obtained upon reaction of Cj1123c with UDP-amino-dAltNAc. The 42 mass unit in-crease compared with the substrate (Fig. 4C) is also consistent with acetylation.

Fig. 5. Structure-based alignment of the sequences of Cj1123c, Escherichia coli LpxA and GmlU, and Chlamydia trachomatis LpxD andlocation of the residues targeted for site-directed mutagenesis in Cj1123c. (A) The structure-based alignment was performed using structuresavailable in the Protein Data Bank: 2npo, 1LXA, 2OI5, and 2IU8. The residues are grouped to highlight the structural arrangement as a left-handed b-helix comprising four residues per b-strand (straight line) and two residues per turn (arch). The catalytic histidine and cofactorbinding glycine and alanine residues are highlighted in bold in the shaded boxes. The LpxA glycine residue that acts as a hydrocarbon rulerand corresponding residue in Cj1123c are highlighted in a shaded box. (B) Location of the residues potentially involved in catalysis and incofactor binding in the crystal structure of Cj1123c. Each monomer of the trimeric unit is shown in a different color. This picture wasgenerated using structural data available in the Protein Data Bank for Cj1123c crystallized without AcCoA (2pno) using Pymol (DeLano2002), and coordinates for the CoA moiety were extracted from the recently released co-crystal structure (Protein Data Bank 2vhe).



induction was maintained throughout the experiment. Ex-pression of Cj1123c was ascertained by anti-His Westernblotting (data not shown). The overexpression of Cj1123creduced bacterial growth compared with uninduced culturesbut did not confer any detectable level of resistance to anyof the aminoglycosides tested, suggesting that none of themis a substrate for Cj1123c (data not shown). This was con-firmed by CE analysis whereby no CoA release was ob-served after incubation of purified Cj1123c with any of theaminoglycosides tested in the presence of AcCoA (data notshown). Altogether, these data indicate that, althoughCj1123c exhibits relaxed substrate specificity, its catalyticabilities appear to be confined to closely related sugar-nu-cleotide substrates and do not extend to aminoglycosides.

Cj1123c can O-acetylate UDP-GlcNAcBecause the prior data suggested that Cj1123c preferen-

tially utilizes sugar-nucleotides as substrates and becauseLpxA has been shown to act as both N- and O-acetyltrans-ferase (Sweet et al. 2004b), we investigated whetherCj1123c could potentially O-acetylate sugar-nucleotides.Whereas Cj1123c was unable to use UDP-Glc, UDP-Gal, orUDP-GalNAc as a substrate (data not shown), it could useUDP-GlcNAc and generated two products after overnight in-cubation (Fig. 7A, products Pa and Pb). O-acetylation ofUDP-GlcNAc was limited because only 0.04 nmol of Paand 0.01 nmol of Pb were produced per microgram of en-zyme in 12 h, whereas 2.5 and 0.6 nmol of N-acetylatedproducts were formed (also per microgram of enzyme) in1.5 h with UDP-amino-dGlcNAc and UDP-amino-dAltNAc,respectively. This suggested that O-acetylation of UDP-GlcNAc is probably not the physiological function ofCj1123c.

MS analysis of the partially purified reaction products (Pa

and Pb) revealed a product peak at m/z 648.1, which is 42mass units greater than UDP-GlcNAc and corresponds tomonoacetylated product(s). Further MS/MS analysis of theparent peak at m/z 648.1 revealed peaks at m/z 324.1, 342.1,and 404.0, which are also all 42 mass units greater than cor-responding peaks arising from the GlcNAc portion of UDP-GlcNAc (Fig. 7B). The presence of unaltered peaks corre-sponding to the UDP moiety confirmed that the nucleotideportion of the molecule was not modified, as expected. Nopeaks were observed at m/z 690 or 732, which would be ex-pected for di- and tri-acetylated products. Altogether, this in-dicates that Cj1123c can O-acetylate the GlcNAc moiety ofUDP-GlcNAc and that only one of the three free hydroxylspotentially available for acetylation (on C3, C4, and C6)was modified in each of the products formed, Pa and Pb.The predominance of product Pa over product Pb indicatesthat one of the three free hydroxyls is the preferred O-acety-lation site. The stereospecificity of Cj1123c for UDP-GlcNAc over the C4 epimer UDP-GalNAc and the fact thatN-acetylation occurs on the C4 nitrogen substituent of UDP-4-amino-4,6-dideoxy-AltNAc or -GlcNAc tend to suggestthat the C4 hydroxyl is the primary target of O-acetylation.Hence, product Pa likely represents a novel sugar: UDP-Glc2NAc4OAc. The precise location of the substituent couldnot be determined unambiguously because it was not possi-

Fig. 6. Time course of acetylation of UDP-amino-dGlcNAc bywild-type (WT) enzyme or by the H125A, G143I, G143V, andG173M mutants in the presence of AcCoA or propionyl-CoA(PropCoA). The reaction composition was the same as in Fig. 2 ex-cept that the amount of enzyme used varied as follows: asterisks,500 ng; diamonds, 10 ng; squares, 1 ng; circles, 100 pg; triangles,10 pg. The time course shows the lower rates of conversion cata-lyzed by the various mutants for acetylation of UDP-amino-dGlcNAc with AcCoA and lower rates of conversion catalyzed bythe WT enzyme with PropCoA than with AcCoA.

Fig. 7. Capillary electrophoresis and mass spectrometry analysis ofO-acetyltransferase activity of Cj1123c on UDP-GlcNAc. (A) Ca-pillary electrophoresis analysis of the reactions. The reaction com-position was the same as in Fig. 2 except that UDP-GlcNAc wasused as the substrate, the substrate concentration was 250 mmol/L,and 4 mg of enzyme was used. The reactions were incubated over-night at 37 8C. Trace a, no Cj1123c; trace b, complete reaction; S,substrate; Pa and Pb, reaction products; D, AcCoA donor. (B) MS/MS analysis of the peak at m/z 648.174 presenting the purified re-action products. The appearance of peaks at m/z 324.1, 342.1,404.0, and 648.1, which are 42 units greater than the correspondingpeaks in the original substrate, is consistent with mono-O-acetyla-tion on the sugar ring. The dots indicate peaks corresponding to theunaltered UDP moiety of the products.



ble to generate sufficient amounts of product for NMR anal-ysis.

Cj1123c can utilize donors other than AcCoAAs multiple potential CoA-based acyl donor species are

usually present in the cellular environment as a result oflipid metabolism, we investigated whether Cj1123c was spe-cific for AcCoA or not. Cj1123c was able to use propionyl-CoA and butyryl-CoA as donors for modification of UDP-amino-dGlcNAc because the release of CoA and the forma-tion of novel products whose migration was clearly differentfrom that of UDP-4-acetamido-4,6-dideoxy-GlcNAc wereobserved by CE analysis (Figs. 8A and 8B, compare withFig. 2A). Co-injections of the various purified productsdemonstrated that they are different compounds (Fig. 8C).

MS analysis of the purified reaction product obtained withpropionyl-CoA revealed the appearance of a peak at m/z645.1 (i.e., 14 mass units greater than what was observedfor the acetylated product and 56 mass units higher than thestarting substrate), which is consistent with the incorporationof the propionyl group in the molecule (Fig. 8D). The MS/MS fragmentation pattern of the parent peak (at m/z 645.1)was consistent with propionylation on the free amine func-tion of the sugar ring (Fig. 8D), with appearance of peaksat m/z 321, 339, 401, and 645 shifted by 56 mass units com-pared with the corresponding peaks observed in the substrate(Fig. 4A). Similarly, MS analysis of the purified reactionproduct obtained with butyryl-CoA as a donor confirmedbutyrylation of the substrate (Fig. 8E), with appearance ofpeaks at m/z 335, 353, 415, and 659 shifted by 70 mass unitscompared with the corresponding peaks observed in the sub-strate (Fig. 4A).

Lack of reactivity with TNBS further demonstrated modi-fication of the amine group of the substrate in the presenceof either cofactor (data not shown). These data indicate thatCj1123c can be used to produce propionylated and butyry-lated derivatives of UDP-amino-dGlcNAc.

Although 100% substrate conversion could be obtainedwith either donor, longer incubation times were necessaryto obtain the equivalent percentage of substrate conversionwith propionyl-CoA (Fig. 6C) and butyryl-CoA (data notshown) than with AcCoA. This suggests that the physiologi-cal donor for the acetyltransferase activity of Cj1123c isprobably AcCoA.

These data indicate somewhat relaxed specificity for theacyl group donor. However, there was still some level ofspecificity because no catalysis was observed with malonyl-CoA using UDP-amino-dGlcNAc as a substrate (data notshown). Also, the donor specificity was dependent on theidentity of the acceptor substrate, since no activity was ob-served using propionyl-, butyryl-, or malonyl-CoA withUDP-amino-dAltNAc as a substrate (data not shown).

Gly-173 does not affect acyl-CoA donor specificity inCj1123c but affects the rate of catalysis

Structural studies of LpxA indicated that the distal portionof the acyl chain lies along the main axis of the left-handedparallel b-helix (Williams and Raetz 2007) and that a gly-cine residue (Gly-173 in E. coli LpxA) could play a role indetermining the length of the acyl donor (Wyckoff et al.1998) by acting as a hydrocarbon ruler. A G173M mutation

prevented access of bulky acyl donors to the binding pocketand the reverse mutation in the Pseudomonas aeruginosaLpxA had the opposite effect (Wyckoff and Raetz 1999;Wyckoff et al. 1998). It is possible that the equivalent resi-due in Cj1123c could also influence the nature of the acyl-CoA donor. If so, introduction of a bulkier side chain mightfurther decrease the ability of Cj1123c to use bulky acyl do-nors such as propionyl- or butyryl-CoA. To test this hypoth-esis, a Cj1123c G173M mutant was constructed. Thismutant was active not only with the AcCoA donor but alsowith propionyl-CoA, allowing 100% conversion of the sub-strate in 2 h with either donor (Fig. 9). However, a timecourse indicated that its activity was significantly reducedcompared with the wild-type (Fig. 6D). These data suggestthat in Cj1123c, Gly-173 does not affect cofactor specificityamong the short-chain acyl donors tested but probably af-fects the rate at which the cofactor(s) can enter the bindingpocket.

Also, the curve obtained with 10 ng of enzyme plateaus,suggesting that decay of the enzyme occurredwithin <30 min of incubation at 37 8C. Gel filtration analy-sis excluded any significant defect in the ability of this mu-tant enzyme to oligomerize (Fig. 1C), thereby also excludingthat the decreased activity and decay of the mutant wascaused by protein unfolding, since this would not supporttrimer formation. The possibility that methionine oxidationcould also contribute to the observed effect cannot be totallyexcluded.

Gly-143 is important for catalysis and affects CoA-donorspecificity

As mentioned above, in LpxA, LpxD, and GlmU, thebackbone amide (NH) group of Gly-143, Gly-265, and Ala-380, respectively, has been proposed to be important for thecatalytic process by maintaining the thioester in the properorientation for nucleophilic attack as well as by functioningas an oxyanion hole during catalysis (Buetow et al. 2007;Olsen et al. 2007; Williams and Raetz 2007). Our structure-based alignment (Fig. 5A) suggests a similar role for thebackbone NH of Gly-143 of Cj1123c. To test this hypothe-sis, mutants G143I and G143V were generated by site-di-rected mutagenesis and tested for catalysis. The rationalefor choosing these mutations was that the presence of abulky side chain might prevent proper interaction of thebackbone amide with the thioester carbonyl and prevent cat-alysis. Against expectations, both mutants were still activeand the G143I mutant was more active than the G143V mu-tant (Fig. 9). A time course performed using variousamounts of enzyme showed that, although 100% conversioncould be reached with the G143I mutant after overnight in-cubation, its activity was also reduced compared with thewild-type (Fig. 6E) but was significantly higher than that ofG143V (Fig. 6F).

The more bulky side chains in the G143V and G143I mu-tants might also prevent entry of more bulky CoA donorsinto the binding pocket, such as propionyl- or butyryl-CoA.When tested with propionyl-CoA for acylation of UDP-amino-dGlcNAc, the G143V and G143I mutants bothshowed decreased catalysis, yielding only ~58% and 73%substrate conversion, respectively, after overnight incuba-



tion, whereas 100% conversion was obtained in 2 h with thewild-type enzyme under the same conditions (Fig. 9).

Gel filtration analysis revealed that the G143V mutantformed trimers in the same proportions as the wild-type andthat the G143I mutant was only very slightly impaired in itsability to form trimers, as ~20% of the protein eluted as amonomer (Fig. 1C). The change being very minimal com-pared with the wild-type (~15% monomers), it is not antici-pated to affect the catalytic efficiency significantly.

Overall, these data show that Gly-143 may affect the po-sition of the CoA donor via its backbone amide and that theabsence of a bulky side chain is advantageous for enzymeactivity. Introduction of a more bulky side chain enhancesthe specificity for AcCoA, which probably reflects sterichindrance in the catalytic pocket.

Discussion

Main activity of Cj1123c and catalytic residuesOur data demonstrating the N-acetyltransferase activity of

Cj1123c on UDP-amino-dGlcNAc are in full agreement withthose recently reported elsewhere (Olivier et al. 2006). Ouridentification of His-125 as the putative catalytic base andour selection of residues potentially involved in co-factor orsubstrate binding were based on the structure of the apo-en-zyme Cj1123c, the only structure available at the time, andon structural information available for LpxA/D and GlmUco-crystallized with their acyl donor, from which we sur-mised the position of the AcCoA and the catalytic pocket inCj1123c. Our selections are in agreement with two studiesreleased during the final stages of completion of this work(Olivier and Imperiali 2008; Rangarajan et al. 2008) andthat proposed a role for H125A as a catalytic residue. Whileour activity data on the H125A mutant also point to a roleof His-125 in catalysis, we show that His-125 is not essen-tial because significant levels of substrate conversion (60%)could be obtained. Our gel filtration data also suggest thatHis-125 might contribute to enzyme activity by stabilizing

Fig. 8. Capillary electrophoresis and mass spectrometry analysis of the N-acetyltransferase activity of Cj1123c on UDP-amino-dGlcNAc inthe presence of propionyl- or butyryl-CoA. For all panels, the reaction composition was as described in Fig. 2 except that AcCoA wasreplaced by (A) propionyl-CoA or (B) butyryl-CoA. In Figs. 8A and 8B Cj1123c was omitted from the reaction in trace a, whereas trace bshows the results of the complete reaction. S, substrate; D, donor; CoA, released CoA after reaction; Pn, reaction product. (C) Capillaryelectrophoresis co-injections of the purified reaction products obtained with AcCoA (P1), propionyl-CoA (P2), and butyryl-CoA (P3) de-monstrating that they are distinct compounds. Trace a, product P1; trace b, product P2; trace c, co-injection of P1 and P2; trace d, productP3; trace e, co-injection of P1 and P3; trace f, co-injection of P2 and P3. (D and E) MS/MS analysis of the propionylated and butyrylatedproducts. The dots indicate peaks corresponding to the unaltered UDP moiety of the products.

Fig. 9. Role of Gly-143 and Gly-173 in acyl-donor specificity ofCj1123c as measured by capillary electrophoresis using the G143V/I and G173M mutants. The reaction composition was the same asin Fig. 2 except that AcCoA was replaced by PropCoA when ap-propriate. Open bars, 2 h reaction; shaded bars, overnight reaction.This figure is a representative example of two independent experi-ments. WT, wild type.



the trimers, which would also explain the reduced activity ofthe H125A mutant.

Also, little difference was observed between the structuresof the apo-enzyme and of Cj1123c co-crystallized with CoAreleased recently (Olivier and Imperiali 2008; Rangarajan etal. 2008), thus reinforcing the validity of our predictionsconcerning the importance of G143 and G173 as co-factorbinding residues, although these predictions were based oncomparative analysis of apo-Cj1123c with LpxA and GlmU.The role for the NH group of Gly-143 in oxyanion-inter-mediate stabilization proposed recently (Rangarajan et al.2008) has not been tested experimentally in any study but isin accordance with our predictions and with the mutagenesisdata presented in this manuscript.

Among the characterized HexATs, Cj1123c utilizes a sub-strate that is most closely related to the substrate of P. aeru-ginosa WbpD, which is a putative CoA-dependent UDP-2-acetamido-3-amino-2,3-dideoxy-D-glucuronic acid 3-N-ace-tyltransferase (Wenzel et al. 2005). However, overall, thecomplement of residues important for catalysis in Cj1123cis different from that of WbpD. Based on LPS synthesiscomplementation and structural modeling studies, the N3 ofLys-136 and the side chain of Gln-60 were identified asimportant, although non-essential, residues for AcCoA bind-ing and catalysis, respectively, in WbpD (Wenzel et al.2005). No sequence- or structure-based match with Gln-60could be identified, but we determined that, instead, His-125 is an important catalytic residue in Cj1123c. Based onsequence alignments, Lys-136 of WbpD corresponds toLys-179 in Cj1123c. Although the side chain of this residuelies parallel to the distal portion (ADP portion) of AcCoA,its distance to AcCoA is too great for this residue to affectAcCoA binding in Cj1123c (Fig. 5B). Also, it lies very faraway from the acetyl group (Fig. 5B) and thus would notbe anticipated to affect acyl-donor specificity. In this study,we identified instead Gly-143 and Gly-173 as important res-idues for acyl-donor binding and specificity.

Versatility of Cj1123c in terms of substrate and acyl-donor specificity

Our data demonstrate novel aspects of the versatility ofCj1123c in terms of substrate and acyl-donor specificity. Abroad-spectrum substrate specificity is not unprecedentedfor acetyltransferases because it is often observed in amino-glycoside N-acetyltransferases (Hegde et al. 2001; Magal-haes and Blanchard 2005; Vetting et al. 2002; Wright andLadak 1997). One of them, Salmonella enterica AAC(6’)-Iy,has even been shown to acetylate basic histone proteins inaddition to a large spectrum of aminoglycosides (Vetting etal. 2004). However, most aminoglycoside acetyltransferasesare members of the GCN5-related N-acetyltransferase family(Vetting et al. 2002; Wolf et al. 1998; Wybenga-Groot et al.1999) and are structurally totally distinct from HexATs. Tothe best of our knowledge, apart from VatD, which inacti-vates streptogramin Group A antibiotics via O-acetylation(Rende-Fournier et al. 1993; Sugantino and Roderick 2002),and LpxA (Sweet et al. 2004b), no other example of aHexAT with multiple substrates has been described to date.In addition, together with the aminoglycoside 2’N-acetyl-transferase from Mycobacterium tuberculosis (Hegde et al.2001; Vetting et al. 2002) and LpxA (Sweet et al. 2004a,

2004b), Cj1123c is one of the few examples of acetyltrans-ferases that can act as both N- and O-acetyltransferase. Thefact that overexpression of Cj1123c in E. coli led to signifi-cantly reduced bacterial growth may reflect the ability ofCj1123c to O-acetylate UDP-GlcNAc. The large amounts ofCj1123c produced in that context might compensate for theslow character of the reaction such that O-acetylation wouldpartly deplete the stock of UDP-GlcNAc available for cellwall synthesis. No appropriate mutant of C. jejuni is cur-rently available to assess the role, if any, of the low level ofO-acetylation of UDP-GlcNAc that might occur because ofendogenous levels of Cj1123c expression in C. jejuni.

Our data also present the paradox that Cj1123c exhibitsrelaxed substrate specificity while being specific for the po-sition of the hydroxyl group targeted for O-acetylation. In-deed, only one of the three available hydroxyl groups wasacetylated at a time in UDP-GlcNAc, with preference forone location that we surmise is C4 hydroxyl. This is remi-niscent of other acetyltransferases, such as gentamycin ace-tyltransferase I, that acetylate only one of the four aminogroups present on the substrate (Williams and Northrop1978). This is yet another property shared between Cj1123cand aminoglycoside acetyltransferases.

The ability of Cj1123c to use various acyl donors is also acommon trait shared with CoA-dependent aminoglycosideacetyltransferases (Magalhaes and Blanchard 2005; Vettinget al. 2002; Wright and Ladak 1997). Similarly, LpxA fromseveral Bordetella species have relaxed acyl-chain specific-ity (Sweet et al. 2002), although in that case, the reaction isdependent on an acyl carrier protein donor.

Despite the many common functional features betweenCj1123c and aminoglycoside acetyltransferases regardingthe ability to acetylate several substrates, to proceed to N-or O-acetylation and to accommodate several acyl donors,Cj1123 was unable to acetylate the four aminoglycosidestested. Although generalization is not possible in the ab-sence of a larger-scale screen, these data could suggest thatthe UDP portion of the substrate, which is not present inaminoglycosides, might be important to ensure proper posi-tioning of the substrate within the catalytic site of Cj1123c.This would be very different from what is observed withGlmU, where acetylation of the sugar occurs before conden-sation with UDP (Mengin-Lecreulx and van Heijenoort1994). However, this would be consistent with recent mod-eling data showing Cj1123c in complex with a 4-amino-UDP sugar, which reveals that several residues interact withthe UDP moiety (Rangarajan et al. 2008).

As mentioned above, Cj1123c was recently co-crystal-lized with CoA (Rangarajan et al. 2008). Analysis of theco-crystals revealed that the carbonyl of Gly-173 interactswith N6 of the CoA adenine, acting similarly to the Gly-173 residue of LpxA as an outermost boundary of the acyl-donor binding pocket. However, one noticeable difference isthat in LpxA, Gly-173 interacts with the distal end of theacyl chain that gets transferred onto the acceptor amine,thereby having a role in acyl-donor specificity by exertingcontrol over the length of the acyl chain that fits in the bind-ing pocket. In contrast, in Cj1123c, Gly-173 interacts withthe portion of the AcCoA donor molecule that is ‘‘leftover’’ after transfer of the acetyl portion onto the acceptoramine, i.e., the distal portion of CoA that is identical no



matter which CoA donor is used. This explains why Gly-173does not play a role in donor specificity per se among theshort-chain acyl-donors tested, as observed with our G173Mmutant. Since introduction of the bulky methionine residuein the G173M mutant results in significantly reduced activ-ity, even though the residue is remote from the catalytic site,it is plausible that the binding of the cofactor or its entry inthe binding pocket may be rate limiting in this mutant.

When the amino sugar acceptor was modeled into thestructure of the Cj1123c/CoA complex, it appeared to bindto AcCoA in an L-shaped fashion instead of the linear con-formation observed in other HexATs (Rangarajan et al.2008). In this configuration, Gly-143 is in contact with thedonor carbonyl via its amide chain, which occurs similarlynotwithstanding the acyl-CoA donors (AcCoA or propionyl-CoA). However, of note is the fact that a small cavity isgenerated by this L-shaped configuration that is not presentin other HexATs and might allow for accommodation of thebulkier side chains of other acyl-CoA donors without per-turbing the fine positioning of the catalytic ternary complex.This could explain why Cj1123c can use propionyl-CoA andbutyryl-CoA fairly efficiently but that the ability to usethese donors is affected by introduction of a bulkier sidechain, as observed in our G143I and G143V mutants. It isalso plausible that conversion of Gly-143 to a residue witha bulkier side chain causes an alteration of the polypeptideconformation, which may cause the observed change in cat-alysis. Moreover, the carboxylic side chain of Glu-124,which protrudes in this cavity (Fig. 5B), would repel theacidic and more bulky malonyl-CoA, explaining why thiscannot be used as a donor by wild-type Cj1123c (data notshown).

We demonstrated that, like Cj1313, Cj1123c has N-acetyl-transferase activity on UDP-amino-dAltNAc. While Cj1123cbelongs to the HexAT family, Cj1313 belongs to the GCN5-related N-acetyltransferase family. Members of this familyoften function as dimers with one substrate/donor bindingsite per monomer, whereas HexATs function as trimers inwhich two monomers are necessary to form a substrate/do-nor binding and catalytic site. It is remarkable that thesetwo enzymes belonging to two structurally distinct acetyl-transferase families can use the same substrate and generatethe same reaction product. However, similarities in themechanism of action of enzymes of both families could ex-plain this. For example, direct nucleophilic attack by thesugar amine onto the thioester of the acyl donor can easilybe achieved in various structural contexts by involvement ofa catalytic base (histidine or other amino acids) that enhan-ces the nucleophilic character of the targeted amine. Themore surprising feature is perhaps that the precise andproper positioning of all partners of the necessary ternarycomplex could be achieved via very structurally differentbinding sites and nevertheless resulted in equally efficientlevels of catalysis.

Implications for engineering of novel sugarsThe relaxed substrate and acyl-donor specificities ob-

served for Cj1123c open up possibilities to synthesize novelsugars that might have applications as antigens for vaccina-tion against bacterial pathogens. Acetylation of sugars isknown to affect their antigenicity (DeShazer et al. 1998; De-

Shazer et al. 2001), and acetylated sugars are often found onbacterial pathogens (Allen and Maskell 1996; Di Fabio et al.1992; Knirel et al. 2006; Rocchetta et al. 1999). Being ableto synthesize enzymatically tailored sugars in their nucleo-tide-activated form will greatly facilitate the assembly ofcomplex carbohydrates that could be used for vaccinationpurposes in the absence of any pathogen or of any of theirpotentially toxic components, a field of glyco-engineeringthat is only beginning to be explored. It also provides noveltools to assess the specificity of glycosyltransferases in-volved in the assembly of complex carbohydrates. The C. je-juni genome encodes several other putative acyltransferasesof unknown function that are very similar to Cj1123c andpotentially have slightly different substrates and (or) donorspecificities. The biochemical characterization of Cj1123creported herein will allow better understanding of the molec-ular basis for substrate and donor specificity in this wide-spread family of enzymes.

AcknowledgementsThis work was funded through an operating grant from

the Canadian Institutes for Health Research to Dr. Creuzenet(MOP-62775). Dr. Creuzenet was the recipient of a Univer-sity Faculty Award from the Natural Sciences and Engineer-ing Research Council of Canada. We thank A. Yuen, I.Gryski, and E. Adiyaman for technical assistance at the out-set of this project. We thank A. Berghuis (McGill Univer-sity) for advice on structural representations using Pymol.We thank P. Hopf for assistance with preparing Fig. 1.

ReferencesAllen, A., and Maskell, D. 1996. The identification, cloning and

mutagenesis of a genetic locus required for lipopolysaccharidebiosynthesis in Bordetella pertussis. Mol. Microbiol. 19: 37–52.doi:10.1046/j.1365-2958.1996.354877.x. PMID:8821935.

Anderson, M.S., and Raetz, C.R. 1987. Biosynthesis of lipid A pre-cursors in Escherichia coli. A cytoplasmic acyltransferase thatconverts UDP-N-acetylglucosamine to UDP-3-O-(R-3-hydroxy-myristoyl)-N-acetylglucosamine. J. Biol. Chem. 262: 5159–5169. PMID:3549716.

Anderson, M.S., Bulawa, C.E., and Raetz, C.R. 1985. The bio-synthesis of gram-negative endotoxin. Formation of lipid A pre-cursors from UDP-GlcNAc in extracts of Escherichia coli. J.Biol. Chem. 260: 15536–15541. PMID:3905795.

Beaman, T.W., Binder, D.A., Blanchard, J.S., and Roderick, S.L.1997. Three-dimensional structure of tetrahydrodipicolinate N-succinyltransferase. Biochemistry, 36: 489–494. doi:10.1021/bi962522q. PMID:9012664.

Bhasin, N., Albus, A., Michon, F., Livolsi, P.J., Park, J.S., and Lee,J.C. 1998. Identification of a gene essential for O-acetylation ofthe Staphylococcus aureus type 5 capsular polysaccharide. Mol.Microbiol. 27: 9–21. doi:10.1046/j.1365-2958.1998.00646.x.PMID:9466251.

Brown, K., Pompeo, F., Dixon, S., Mengin-Lecreulx, D., Cambil-lau, C., and Bourne, Y. 1999. Crystal structure of the bifunc-tional N-acetylglucosamine 1-phosphate uridyltransferase fromEscherichia coli: a paradigm for the related pyrophosphorylasesuperfamily. EMBO J. 18: 4096–4107. doi:10.1093/emboj/18.15.4096. PMID:10428949.

Buetow, L., Smith, T.K., Dawson, A., Fyffe, S., and Hunter, W.N.2007. Structure and reactivity of LpxD, the N-acyltransferase of



lipid A biosynthesis. Proc. Natl. Acad. Sci. U.S.A. 104: 4321–4326. doi:10.1073/pnas.0606356104. PMID:17360522.

Bystricky, S., and Szu, S.C. 1994. O-acetylation affects the bindingproperties of the carboxyl groups on the Vi bacterial polysac-charide. Biophys. Chem. 51: 1–7. doi:10.1016/0301-4622(94)00002-6. PMID:8061223.

Cerca, N., Jefferson, K.K., Maira-Litran, T., Pier, D.B., Kelly-Quintos, C., Goldmann, D.A., et al. 2007. Molecular basis forpreferential protective efficacy of antibodies directed to thepoorly acetylated form of staphylococcal poly-N-acetyl-beta-(1–6)-glucosamine. Infect. Immun. 75: 3406–3413. doi:10.1128/IAI.00078-07. PMID:17470540.

Coleman, J., and Raetz, C.R. 1988. First committed step of lipid Abiosynthesis in Escherichia coli: sequence of the lpxA gene. J.Bacteriol. 170: 1268–1274. PMID:3277952.

Creuzenet, C. 2004. Characterization of CJ1293, a new UDP-GlcNAc C6 dehydratase from Campylobacter jejuni. FEBSLett. 559: 136–140. doi:10.1016/S0014-5793(04)00057-2.PMID:14960321.

Creuzenet, C., Belanger, M., Wakarchuk, W.W., and Lam, J.S.2000a. Expression, purification, and biochemical characteriza-tion of WbpP, a new UDP-GlcNAc C4 epimerase from Pseudo-monas aeruginosa serotype O6. J. Biol. Chem. 275: 19060–19067. doi:10.1074/jbc.M001171200. PMID:10747995.

Creuzenet, C., Schur, M.J., Li, J., Wakarchuk, W.W., and Lam, J.S.2000b. FlaA1, a new bifunctional UDP-GlcNAc C6 dehydratase/C4 reductase from Helicobacter pylori. J. Biol. Chem. 275:34873–34880. doi:10.1074/jbc.M006369200. PMID:10954725.

DeLano, W.L. 2002. The PyMol molecular graphics system. Avail-able from www.pymol.org. [accessed July 2008].

Demendi, M., Ishiyama, N., Lam, J.S., Berghuis, A.M., and Creu-zenet, C. 2005. Towards a better understanding of the substratespecificity of the UDP-N-acetylglucosamine C4 epimeraseWbpP. Biochem. J. 389: 173–180. doi:10.1042/BJ20050263.PMID:15752069.

DeShazer, D., Brett, P.J., and Woods, D.E. 1998. The type II O-antigenic polysaccharide moiety of Burkholderia pseudomalleilipopolysaccharide is required for serum resistance and viru-lence. Mol. Microbiol. 30: 1081–1100. doi:10.1046/j.1365-2958.1998.01139.x. PMID:9988483.

DeShazer, D., Waag, D.M., Fritz, D.L., and Woods, D.E. 2001. Iden-tification of a Burkholderia mallei polysaccharide gene cluster bysubtractive hybridization and demonstration that the encoded cap-sule is an essential virulence determinant. Microb. Pathog. 30:253–269. doi:10.1006/mpat.2000.0430. PMID:11373120.

Deszo, E.L., Steenbergen, S.M., Freedberg, D.I., and Vimr, E.R.2005. Escherichia coli K1 polysialic acid O-acetyltransferasegene, neuO, and the mechanism of capsule form variationinvolving a mobile contingency locus. Proc. Natl. Acad. Sci.U.S.A. 102: 5564–5569. doi:10.1073/pnas.0407428102. PMID:15809431.

Di Fabio, J.L., Caroff, M., Karibian, D., Richards, J.C., and Perry,M.B. 1992. Characterization of the common antigenic lipopoly-saccharide O-chains produced by Bordetella bronchiseptica andBordetella parapertussis. FEMS Microbiol. Lett. 76: 275–281.doi:10.1016/0378-1097(92)90348-R. PMID:1427018.

Draker, K.A., and Wright, G.D. 2004. Molecular mechanism of theenterococcal aminoglycoside 6’-N-acetyltransferase: role ofGNAT-conserved residues in the chemistry of antibiotic inacti-vation. Biochemistry, 43: 446–454. doi:10.1021/bi035667n.PMID:14717599.

Gehring, A.M., Lees, W.J., Mindiola, D.J., Walsh, C.T., and Brown,E.D. 1996. Acetyltransfer precedes uridylyltransfer in the forma-tion of UDP-N-acetylglucosamine in separable active sites of the

bifunctional GlmU protein of Escherichia coli. Biochemistry, 35:579–585. doi:10.1021/bi952275a. PMID:8555230.

Glover, K.J., Weerapana, E., and Imperiali, B. 2005a. In vitro as-sembly of the undecaprenylpyrophosphate-linked heptasacchar-ide for prokaryotic N-linked glycosylation. Proc. Natl. Acad.Sci. U.S.A. 102: 14255–14259. doi:10.1073/pnas.0507311102.PMID:16186480.

Glover, K.J., Weerapana, E., Numao, S., and Imperiali, B. 2005b.Chemoenzymatic synthesis of glycopeptides with PglB, a bacter-ial oligosaccharyl transferase from Campylobacter jejuni. Chem.Biol. 12: 1311–1315. doi:10.1016/j.chembiol.2005.10.004.PMID:16356848.

Goon, S., Kelly, J.F., Logan, S.M., Ewing, C.P., and Guerry, P.2003. Pseudaminic acid, the major modification on Campylobac-ter flagellin, is synthesized via the Cj1293 gene. Mol. Microbiol.50: 659–671. doi:10.1046/j.1365-2958.2003.03725.x. PMID:14617187.

Gudlavalleti, S.K., Lee, C.H., Norris, S.E., Paul-Satyaseela, M.,Vann, W.F., and Frasch, C.E. 2007. Comparison of Neisseriameningitidis serogroup W135 polysaccharide-tetanus toxoid con-jugate vaccines made by periodate activation of O-acetylated,non-O-acetylated and chemically de-O-acetylated polysacchar-ide. Vaccine, 25: 7972–7980. PMID:17936445.

Hegde, S.S., Javid-Majd, F., and Blanchard, J.S. 2001. Overexpres-sion and mechanistic analysis of chromosomally encodedaminoglycoside 2’-N-acetyltransferase (AAC(2’)-Ic) from Myco-bacterium tuberculosis. J. Biol. Chem. 276: 45876–45881.doi:10.1074/jbc.M108810200. PMID:11590162.

Kelly, T.M., Stachula, S.A., Raetz, C.R., and Anderson, M.S. 1993.The firA gene of Escherichia coli encodes UDP-3-O-(R-3-hy-droxymyristoyl)-glucosamine N-acyltransferase. The third stepof endotoxin biosynthesis. J. Biol. Chem. 268: 19866–19874.PMID:8366125.

Knirel, Y.A., Bystrova, O.V., Kocharova, N.A., Zahringer, U., andPier, G.B. 2006. Conserved and variable structural features inthe lipopolysaccharide of Pseudomonas aeruginosa. J. Endo-toxin Res. 12: 324–336. doi:10.1179/096805106X118906.PMID:17254386.

Kooistra, O., Luneberg, E., Lindner, B., Knirel, Y.A., Frosch, M.,and Zahringer, U. 2001. Complex O-acetylation in Legionellapneumophila serogroup 1 lipopolysaccharide. Evidence for twogenes involved in 8-O-acetylation of legionaminic acid.Biochemistry, 40: 7630–7640. doi:10.1021/bi002946r. PMID:11412117.

Lewis, A.L., Cao, H., Patel, S.K., Diaz, S., Ryan, W., Carlin, A.F.,et al. 2007. NeuA sialic acid O-acetylesterase activity modulatesO-acetylation of capsular polysaccharide in Group B Streptococ-cus. J. Biol. Chem. 282: 27562–27571. doi:10.1074/jbc.M700340200. PMID:17646166.

MacLean, L.L., Webb, A.C., and Perry, M.B. 2006. Structural elu-cidation of the O-antigenic polysaccharide from enterohemorrha-gic (EHEC) Escherichia coli O48:H21. Carbohydr. Res. 341:2543–2549. doi:10.1016/j.carres.2006.08.008. PMID:16949062.

Magalhaes, M.L., and Blanchard, J.S. 2005. The kinetic mechanismof AAC3-IV aminoglycoside acetyltransferase from Escherichiacoli. Biochemistry, 44: 16275–16283. doi:10.1021/bi051777d.PMID:16331988.

Mengin-Lecreulx, D., and van Heijenoort, J. 1994. Copurificationof glucosamine-1-phosphate acetyltransferase and N-acetylgluco-samine-1-phosphate uridyltransferase activities of Escherichiacoli: characterization of the glmU gene product as a bifunctionalenzyme catalyzing two subsequent steps in the pathway forUDP-N-acetylglucosamine synthesis. J. Bacteriol. 176: 5788–5795. PMID:8083170.



Newton, D.T., and Mangroo, D. 1999. Mapping the active site ofthe Haemophilus influenzae methionyl-tRNA formyltransferase:residues important for catalysis and tRNA binding. Biochem. J.339: 63–69. doi:10.1042/0264-6021:3390063. PMID:10085228.

Obhi, R.K., and Creuzenet, C. 2005. Biochemical characterizationof the Campylobacter jejuni Cj1294, a novel UDP-4-keto-6-deoxy-GlcNAc aminotransferase that generates UDP-4-amino-4,6-dideoxy-GalNAc. J. Biol. Chem. 280: 20902–20908. doi:10.1074/jbc.M413832200. PMID:15790564.

Olivier, N.B., and Imperiali, B. 2008. Crystal structure and cataly-tic mechanism of PglD from Campylobacter jejuni. J. Biol.Chem. 283: 27937-27946. doi:10.1074/jbc.M801207200.PMID:18667421.

Olivier, N.B., Chen, M.M., Behr, J.R., and Imperiali, B. 2006. Invitro biosynthesis of UDP-N,N’-diacetylbacillosamine by en-zymes of the Campylobacter jejuni general protein glycosylationsystem. Biochemistry, 45: 13659–13669. doi:10.1021/bi061456h. PMID:17087520.

Olsen, L.R., Vetting, M.W., and Roderick, S.L. 2007. Structure ofthe E. coli bifunctional GlmU acetyltransferase active site withsubstrates and products. Protein Sci. 16: 1230–1235. doi:10.1110/ps.072779707. PMID:17473010.

Raetz, C.R., and Roderick, S.L. 1995. A left-handed parallel betahelix in the structure of UDP-N-acetylglucosamine acyltransfer-ase. Science (Washington, D.C.), 270: 997–1000. doi:10.1126/science.270.5238.997. PMID:7481807.

Rangarajan, E.S., Ruane, K.M., Sulea, T., Watson, D.C., Proteau,A., Leclerc, S., et al. 2008. Structure and active site residues ofPglD, an N-acetyltransferase from the bacillosamine syntheticpathway required for N-glycan synthesis in Campylobacter je-juni. Biochemistry, 47: 1827–1836. doi:10.1021/bi702032r.PMID:18198901.

Rende-Fournier, R., Leclercq, R., Galimand, M., Duval, J., andCourvalin, P. 1993. Identification of the satA gene encoding astreptogramin A acetyltransferase in Enterococcus faeciumBM4145. Antimicrob. Agents Chemother. 37: 2119–2125.PMID:8257133.

Rocchetta, H.L., Burrows, L.L., and Lam, J.S. 1999. Genetics of O-antigen biosynthesis in Pseudomonas aeruginosa. Microbiol.Mol. Biol. Rev. 63: 523–553. PMID:10477307.

Rund, S., Lindner, B., Brade, H., and Holst, O. 1999. Structuralanalysis of the lipopolysaccharide from Chlamydia trachomatisserotype L2. J. Biol. Chem. 274: 16819–16824. doi:10.1074/jbc.274.24.16819. PMID:10358025.

Schoenhofen, I.C., McNally, D.J., Brisson, J.R., and Logan, S.M.2006a. Elucidation of the CMP-pseudaminic acid pathway inHelicobacter pylori: synthesis from UDP-N-acetylglucosamineby a single enzymatic reaction. Glycobiology, 16: 8C–14C.doi:10.1093/glycob/cwl010. PMID:16751642.

Schoenhofen, I.C., McNally, D.J., Vinogradov, E., Whitfield, D.,Young, N.M., Dick, S., et al. 2006b. Functional characterizationof dehydratase/aminotransferase pairs from Helicobacter andCampylobacter: enzymes distinguishing the pseudaminic acidand bacillosamine biosynthetic pathways. J. Biol. Chem. 281:723–732. doi:10.1074/jbc.M511021200. PMID:16286454.

Schoenhofen, I.C., Lunin, V.V., Julien, J.P., Li, Y., Ajamian, E.,Matte, A., et al. 2006c. Structural and functional characteriza-tion of PseC, an aminotransferase involved in the biosynthesisof pseudaminic acid, an essential flagellar modification in Heli-cobacter pylori. J. Biol. Chem. 281: 8907–8916. doi:10.1074/jbc.M512987200. PMID:16421095.

Studier, F.W., Rosenberg, A.H., Dunn, J.J., and Dubendorff, J.W.1990. Use of T7 RNA polymerase to direct expression of cloned

genes. Methods Enzymol. 185: 60–89. doi:10.1016/0076-6879(90)85008-C. PMID:2199796.

Sugantino, M., and Roderick, S.L. 2002. Crystal structure ofVat(D): an acetyltransferase that inactivates streptogramin groupA antibiotics. Biochemistry, 41: 2209–2216. doi:10.1021/bi011991b. PMID:11841212.

Sweet, C.R., Preston, A., Toland, E., Ramirez, S.M., Cotter, R.J.,Maskell, D.J., and Raetz, C.R. 2002. Relaxed acyl chain specifi-city of Bordetella UDP-N-acetylglucosamine acyltransferases. J.Biol. Chem. 277: 18281–18290. doi:10.1074/jbc.M201057200.PMID:11889134.

Sweet, C.R., Ribeiro, A.A., and Raetz, C.R. 2004a. Oxidation andtransamination of the 3’-position of UDP-N-acetylglucosamineby enzymes from Acidithiobacillus ferrooxidans. Role in the for-mation of lipid a molecules with four amide-linked acyl chains.J. Biol. Chem. 279: 25400–25410. doi:10.1074/jbc.M400596200.PMID:15044494.

Sweet, C.R., Williams, A.H., Karbarz, M.J., Werts, C., Kalb, S.R.,Cotter, R.J., and Raetz, C.R. 2004b. Enzymatic synthesis of lipidA molecules with four amide-linked acyl chains. LpxA acyl-transferases selective for an analog of UDP-N-acetylglucosaminein which an amine replaces the 3’-hydroxyl group. J. Biol.Chem. 279: 25411–25419. doi:10.1074/jbc.M400597200.PMID:15044493.

Szymanski, C.M., Yao, R., Ewing, C.P., Trust, T.J., and Guerry, P.1999. Evidence for a system of general protein glycosylation inCampylobacter jejuni. Mol. Microbiol. 32: 1022–1030. doi:10.1046/j.1365-2958.1999.01415.x. PMID:10361304.

Szymanski, C.M., Burr, D.H., and Guerry, P. 2002. Campylobacterprotein glycosylation affects host cell interactions. Infect. Im-mun. 70: 2242–2244. doi:10.1128/IAI.70.4.2242-2244.2002.PMID:11895996.

Thibault, P., Logan, S.M., Kelly, J.F., Brisson, J.R., Ewing, C.P.,Trust, T.J., and Guerry, P. 2001. Identification of the carbohy-drate moieties and glycosylation motifs in Campylobacter jejuniflagellin. J. Biol. Chem. 276: 34862–34870. doi:10.1074/jbc.M104529200. PMID:11461915.

Vaara, M. 1992. Eight bacterial proteins, including UDP-N-acetyl-glucosamine acyltransferase (LpxA) and three other transferasesof Escherichia coli, consist of a six-residue periodicity theme.FEMS Microbiol. Lett. 76: 249–254. doi:10.1111/j.1574-6968.1992.tb05471.x. PMID:1427014.

Vetting, M.W., Hegde, S.S., Javid-Majd, F., Blanchard, J.S., andRoderick, S.L. 2002. Aminoglycoside 2’-N-acetyltransferasefrom Mycobacterium tuberculosis in complex with coenzyme Aand aminoglycoside substrates. Nat. Struct. Biol. 9: 653–658.doi:10.1038/nsb830. PMID:12161746.

Vetting, M.W., Magnet, S., Nieves, E., Roderick, S.L., and Blan-chard, J.S. 2004. A bacterial acetyltransferase capable of regiose-lective N-acetylation of antibiotics and histones. Chem. Biol. 11:565–573. doi:10.1016/j.chembiol.2004.03.017. PMID:15123251.

Vijayakumar, S., Merkx-Jacques, A., Ratnayake, D., Gryski, I.,Obhi, R.K., Houle, S., et al. 2006. Cj1121c, a novel UDP-4-keto-6-deoxy-GlcNAc C4 aminotransferase essential for proteinglycosylation and virulence in Campylobacter jejuni. J. Biol.Chem. 281: 27733–27743. doi:10.1074/jbc.M511714200.PMID:16690622.

Voisin, S., Houliston, R.S., Kelly, J., Brisson, J.R., Watson, D.,Bardy, S.L., et al. 2005. Identification and characterization ofthe unique N-linked glycan common to the flagellins and S-layerglycoprotein of Methanococcus voltae. J. Biol. Chem. 280:16586–16593. doi:10.1074/jbc.M500329200. PMID:15723834.

Wenzel, C.Q., Daniels, C., Keates, R.A., Brewer, D., and Lam, J.S.2005. Evidence that WbpD is an N-acetyltransferase belonging



to the hexapeptide acyltransferase superfamily and an importantprotein for O-antigen biosynthesis in Pseudomonas aeruginosaPAO1. Mol. Microbiol. 57: 1288–1303. doi:10.1111/j.1365-2958.2004.04767.x. PMID:16102001.

Williams, A.H., and Raetz, C.R. 2007. Structural basis for the acylchain selectivity and mechanism of UDP-N-acetylglucosamineacyltransferase. Proc. Natl. Acad. Sci. U.S.A. 104: 13543–13550. doi:10.1073/pnas.0705833104. PMID:17698807.

Williams, J.W., and Northrop, D.B. 1978. Substrate specificity andstructure–activity relationships of gentamicin acetyltransferase I.The dependence of antibiotic resistance upon substrate Vmax/Km

values. J. Biol. Chem. 253: 5908–5914. PMID:681328.Wolf, E., Vassilev, A., Makino, Y., Sali, A., Nakatani, Y., and Bur-

ley, S.K. 1998. Crystal structure of a GCN5-related N-acetyl-transferase: Serratia marcescens aminoglycoside 3-N-acetyltransferase. Cell, 94: 439–449. doi:10.1016/S0092-8674(00)81585-8. PMID:9727487.

Wright, G.D. 2007. The antibiotic resistome: the nexus of chemicaland genetic diversity. Nat. Rev. Microbiol. 5: 175–186. doi:10.1038/nrmicro1614. PMID:17277795.

Wright, G.D., and Ladak, P. 1997. Overexpression and characteri-zation of the chromosomal aminoglycoside 6’-N-acetyltransfer-ase from Enterococcus faecium. Antimicrob. AgentsChemother. 41: 956–960. PMID:9145851.

Wybenga-Groot, L.E., Draker, K., Wright, G.D., and Berghuis,A.M. 1999. Crystal structure of an aminoglycoside 6’-N-acetyl-transferase: defining the GCN5-related N-acetyltransferasesuperfamily fold. Structure, 7: 497–507. doi:10.1016/S0969-2126(99)80066-5. PMID:10378269.

Wyckoff, T.J., and Raetz, C.R. 1999. The active site of Escherichiacoli UDP-N-acetylglucosamine acyltransferase. Chemical modi-fication and site-directed mutagenesis. J. Biol. Chem. 274:27047–27055. doi:10.1074/jbc.274.38.27047. PMID:10480918.

Wyckoff, T.J., Lin, S., Cotter, R.J., Dotson, G.D., and Raetz, C.R.1998. Hydrocarbon rulers in UDP-N-acetylglucosamine acyl-transferases. J. Biol. Chem. 273: 32369–32372. doi:10.1074/jbc.273.49.32369. PMID:9829962.

Yildirim, H.H., Li, J., Richards, J.C., Hood, D.W., Moxon, E.R.,and Schweda, E.K. 2005. Complex O-acetylation in non-type-able Haemophilus influenzae lipopolysaccharide: evidence for anovel site of O-acetylation. Carbohydr. Res. 340: 2598–2611.doi:10.1016/j.carres.2005.09.005. PMID:16199021.

Young, N.M., Brisson, J.R., Kelly, J., Watson, D.C., Tessier, L.,Lanthier, P.H., et al. 2002. Structure of the N-linked glycan pre-sent on multiple glycoproteins in the Gram-negative bacterium,Campylobacter jejuni. J. Biol. Chem. 277: 42530–42539.doi:10.1074/jbc.M206114200. PMID:12186869.



Documents

Cj1123c (PglD), a multifaceted acetyltransferase from … › 6d9d › 5b420dd56c520f0d1bfe40d… · Cj1123c (PglD), a multifaceted acetyltransferase from Campylobacter jejuni Melinda