Characterization of a Cluster of Three Glycosyltransferase Enzymes Essential for Moraxella catarrhalis Lipooligosaccharide Assembly

Moraxella catarrhalis isolates express lipooligosaccharide (LOS)molecules on their surface, which share epitopes similar tothat of the Neisseria and Haemophilus species. These commonLOS epitopes have been implicated in various steps of pathogenesisfor the different organisms. In this study, a cluster of threeLOS glycosyltransferase genes (lgt) were identified in M. catarrhalis7169, a strain that produces a serotype B LOS. Mutants in theseglycosyltransferase genes were constructed, and the resultingLOS phenotypes were consistent with varying degrees of truncationcompared to wild-type LOS. The LOS structures of each lgt mutantwere no longer detected by a monoclonal antibody (MAb 4G5) specificto a highly conserved terminal epitope nor by a monoclonal antibody(MAb 3F7) specific to the serotype B LOS side chain. Mass spectrometryof the LOS glycoforms assembled by two of these lgt mutantsindicated that lgt1 encodes an ${alpha}$ (1-2) glucosyltransferase andthe lgt2 encodes a ß(1-4) galactosyltransferase. However,these structural studies could not delineate the function forlgt3. Therefore, M. catarrhalis lgt3 was introduced into a definedß(1-4) glucosyltransferase Haemophilus ducreyi 35000glu–mutant in trans, and monoclonal antibody analysis confirmedthat Lgt3 complemented the LOS defect. These data suggest thatlgt3 encodes a glucosyltransferase involved in the additionof a ß(1-4)-linked glucose to the inner core. Furthermore,we conclude that this enzymatic step is essential for the assemblyof the complete LOS glycoform expressed by M. catarrhalis 7169.

INTRODUCTION

Top

Introduction

Moraxella catarrhalis is a gram-negative human respiratory pathogenthat causes 15 to 20% of acute otitis media in children (56).In addition, this bacterium is responsible for 10 to 35% oflower respiratory infections in adults with chronic obstructivepulmonary disease, the fourth leading cause of death in theUnited States (20). The emergence of M. catarrhalis as an importanthuman pathogen has occurred in the past 15 years due to theincreasing prevalence of ß-lactamase-positive strains,the high incidence of recurrent infections despite successfulantibiotic treatment, and the lack of an effective vaccine (10,39, 50, 56). Another factor that likely contributes to the persistenceof M. catarrhalis disease is the lack of understanding of thebasic bacterial factors and mechanisms that promote colonizationand survival in the host.

Although there have been a number of putative virulence factorsdescribed for M. catarrhalis, the actual role of these componentsin pathogenesis remains largely undefined (13, 19, 20, 31, 53).One of the most prominent surface components expressed in theouter membrane of this bacterium is the lipooligosaccharide(LOS) (12). The LOS of M. catarrhalis is similar to the lipopolysaccharide(LPS) of other gram-negative organisms, but it lacks a repeatingO antigen (21). Previous studies have suggested that LOS isimportant for the pathogenesis of other respiratory pathogens,such as Neisseria meningitidis and Haemophilus influenzae, aidingin adherence of the bacterium to the mucosal epithelium andin the destruction of the mucosal barrier (16, 30, 51, 52).Some studies have demonstrated that the LOS of M. catarrhalismay also aid in adherence of this bacterium to epithelial cells,while other studies have indicated that LOS may elicit antibodyproduction, suggesting that this major glycolipid has potentialas a vaccine candidate (2, 23, 24, 27).

Previous studies using polyclonal antisera and structural analyseshave identified three major LOS serotypes for M. catarrhalis(7-9, 21, 55). Clinical isolates were grouped into LOS serotypesA (60%), B (30%), and C (5%), with 5% of the strains unidentified(21, 26, 55). However, some cross-reactivity exists betweenserotypes A and C, and there has yet to be a correlation establishedbetween LOS serotype as it relates to colonization and infection(21, 43-45). Despite these data, the genes and gene productsresponsible for the biosynthesis and assembly of M. catarrhalisLOS and the role of this molecule in pathogenesis remain onlypartially defined. Previous studies have identified two genesinvolved in the biosynthesis of M. catarrhalis LOS: the UDP-glucose-4-epimerase(galE) gene in a serotype A strain (59) and the 2-keto-3-deoxyoctulosonicacid (KDO)-8-phosphate synthase (kdsA) gene in a serotype Bstrain (33).

In this study, we have identified and characterized a clusterof three glycosyltransferase genes involved in the biosynthesisof LOS from a serotype B strain, M. catarrhalis 7169. The genecluster was identified in the National Center for BiotechnologyInformation (NCBI) patented database of M. catarrhalis usingnucleotide sequence homology to glycosyltransferase genes fromother gram-negative respiratory pathogens. Construction of isogenicmutants for the first two genes in the cluster and analysisof corresponding LOS structures has determined that these lipooligosaccharideglycosyltransferase genes (lgt) encode an ${alpha}$ (1-2) glucosyltransferase(Lgt1) and a ß(1-4) galactosyltransferase (Lgt2).Deletion of the third putative glycosyltransferase (Lgt3) resultedin the production of a deeply truncated LOS molecule. Our studiessuggest that Lgt3 functions by the addition of a ß(1-4)-linkedglucose to the inner core of the LOS, and this step plays acritical role that allows for the continuation of the biosynthesisof the major oligosaccharide glycoform assembled by M. catarrhalis.This report describes the isolation and characterization ofglycosyltransferase genes involved in the biosynthesis of M.catarrhalis LOS, and these studies provide insight into thesteps involved in the assembly of this important glycolipid.

MATERIALS AND METHODS

Top

Materials and Methods

Strains, plasmids, and culture conditions. The bacterial strains and plasmids used in this study are describedin Table 1. M. catarrhalis 7169 was cultured on standard brainheart infusion (BHI) agar plates at 35.5°C in 5% CO₂. Escherichiacoli XL1-Blue was used as the host strain for plasmid DNA manipulations.E. coli was cultured in standard Luria-Bertani (LB) medium at35.5°C in 5% CO₂. Haemophilus ducreyi strains were culturedon chocolate agar plates or in BHI broth at 35.5°C in 5%CO₂, as previously described (5). Antibiotics were supplementedas necessary as 20-µg/ml kanamycin, 100-µg/ml ampicillin,or chloramphenicol (1 µg/ml).

View this table:
[in this window]
[in a new window]
TABLE 1. Bacterial strains and plasmids used in this study

Standard recombinant DNA techniques. Standard molecular biological techniques were used in this studyas previously described (49). Plasmids were obtained from E.coli using QIAprep spin miniprep kits (QIAGEN, Chatsworth, Calif.).Chromosomal DNA was obtained from M. catarrhalis using standardmethods, as previously described (48), and DNA was purifiedusing QIAGEN purification kits. PCR was performed using TaqDNA polymerase (Invitrogen, Carlsbad, Calif.), and all otherenzymes were obtained from New England Biolabs (Beverly, Mass.)and used according to company protocols. XL1-Blue E. coli wasmade electrocompetent, and plasmid DNA was introduced by electroporation,as previously described (49). DNA nucleotide sequences wereobtained for all constructs via automated DNA sequencing atthe RPCI Biopolymer Facility (Roswell Park Cancer Institute,Buffalo, N.Y.), and all sequences were analyzed using MacVector,version 7.2, software (Accelrys, San Diego, Calif.).

Construction of M. catarrhalis 7169 isogenic mutants. The PCR primers used to clone and sequence lgt1, lgt2, and lgt3from strain 7169 are listed in Table 2. The primers were designedbased on the sequence homologies between the M. catarrhalisgenome in the NCBI patent database and the reported sequenceof other known glycosyltransferase genes. Each individual genewas ligated into the TA cloning vector pGEM-T Easy (Promega,Madison, Wis.) to create plasmids pLGT-1KE, pLGT-2KE, and pLGT-3KE(Table 1), and isogenic mutants were constructed using an inversePCR strategy as previously described (14, 32, 33). Briefly,inverse PCR primers (Table 2) were designed to amplify a portionof the 5' end and 3' end of each gene using pLGT-1KE, pLGT-2KE,and pLGT-3KE as templates. This created an internal deletionwith engineered restriction sites (Table 2) in each individualgene that allowed for the nonpolar insertion of the aphA-3 cassette(37) (kanamycin resistance) amplified from pUC18K (Table 1)as has been described previously by this laboratory (14, 33).The mutant constructs pLGT-1kanB, pLGT-2kanM, and pLGT-3kanH(Table 1) were linearized and introduced into M. catarrhalis7169 by natural transformation using our standard techniques(14, 33). Chromosomal DNA was isolated from selected kanamycin-resistanttransformants and subjected to PCR and sequence analysis toconfirm that each inactivated gene had properly integrated intothe genome.

View this table:
[in this window]
[in a new window]
TABLE 2. Nucleotide sequence of oligonucleotide primers used for PCR-based cloning procedures in this study

Restoration of native LOS in 7169 lgt mutants through reversion to wild type. Primer sets 406-407, 508-509, and 698-699 (Table 2) were usedto amplify the native lgt1, lgt2, and lgt3 genes, respectively,from M. catarrhalis 7169 chromosomal DNA. The resulting PCRproducts were purified using the MinElute kit (QIAGEN) and usedto naturally transform 7169::lgt1KB, 7169::lgt2KM, and 7169::lgt3KH,respectively, by a method previously described (14). Potentialrevertant clones were tested for the loss of kanamycin resistancethrough replicate plating on both BHI and BHI plus kanamycin(20 µg/ml) agar plates. One clone for each lgt mutantrevertant that demonstrated the loss of kanamycin resistancewas chosen for further analyses (Table 1). Chromosomal DNA wasisolated from each revertant strain (Table 1) and subjectedto PCR and sequence analysis to confirm that each wild-typegene had properly integrated into the genome.

Preparation and analysis of M. catarrhalis LOS. LOS was extracted from M. catarrhalis 7169 and all derivativestrains using a modified proteinase K water-phenol method, aspreviously described (1, 25, 33). Briefly, M. catarrhalis washarvested from BHI plates, resuspended in a sodium dodecyl sulfate(SDS) buffer (0.06 M Tris base, 10 mM EDTA, 2% SDS), and heatedat 100°C for 5 min. After addition of proteinase K at 50µg/ml, the solution was incubated at 65°C for 1 hand then at 37°C overnight. The samples were sonicated for15 min, and the solution was precipitated and washed three timeswith 100% ethyl alcohol and 3 M sodium acetate (pH 5.2) overnightat –20°C to remove the SDS. After the final precipitation,the pellets were resuspended in 20 mM Tris and incubated with10 µg/ml of DNAseI and RNase for 2 h at 37°C. TheLOS was extracted (two times) with equal volumes of phenol,heated at 65°C, and then placed on ice for 1 h. The aqueouslayer was removed, and the LOS was precipitated and washed withethyl alcohol and sodium acetate overnight (three times). Afterthe final precipitation, the LOS molecules were resuspendedin 200 µl of distilled water and resolved by SDS-polyacrylamidegel electrophoresis on a 17.5% acrylamide gel using a bilayerstacker as previously described (25) and then visualized bysilver staining (54).

Western blot and colony lift analysis. Purified LOS isolated from M. catarrhalis were transferred to0.2-µm-pore-size nitrocellulose membranes using standardmethods (28). Western blots were blocked for 1 h at room temperaturewith 3% (wt/vol) nonfat milk-phosphate-buffered saline and incubatedovernight at 4°C with monoclonal antibodies (MAb) (34, 35).Colony lift assays were performed with H. ducreyi by a methodpreviously described (35). MAb 4G5 (an immunoglobulin G2a [IgG2a])and 3F7 (an IgM) were developed from a previous fusion (35).MAb 1B2-1B7 (an IgM) was obtained from the American Type CultureCollection (Manassas, Va.) and used as the probe in H. ducreyicolony lifts and Western blots. This antibody recognizes thesurface-exposed, terminal lactosamine moiety on the LOS of wild-typeH. ducreyi 35000 (11, 36, 58). Horseradish peroxidase-labeledprotein A was used to detect MAb 4G5 and horseradish peroxidase-labeledgoat anti-mouse IgM (Kirkegaard & Perry Laboratories, Inc.,Gaithersburg, Md.) was used to detect MAb 3F7 and MAb 1B2-1B7reactivity, and then the blots were developed as previouslydescribed (4, 34, 35).

Growth curve and outer membrane protein (OMP) profile analyses. Growth curves were performed for the M. catarrhalis strainsas previously described (4, 14). Static-phase bacteria wereharvested, and outer membranes were prepared and analyzed aspreviously described (4, 34, 35).

Mass spectrometric analysis of M. catarrhalis LOS. LOS from M. catarrhalis 7169 and the lgt mutants were analyzedby mass spectrometry (MS) to determine molecular masses, genericcompositions of the oligosaccharide, and the phosphorylationstates of the lipid A moiety. LOS was O-deacylated by treatmentwith anhydrous hydrazine for 35 min at 37°C (18). The conversionof LOS to di-N-acyl LOS makes it more water soluble and amenableto mass spectrometric analysis (15). The O-deacylated LOS wasdissolved in water and desalted by drop dialysis using a 0.025-µm-pore-sizenitrocellulose membrane (Millipore, Bedford, Mass.). Approximately25% of the sample was dried in a Speed-Vac (Savant Instruments,Inc., Holbrook, N.Y.), reconstituted in 1 µl of milli-Qwater, and desalted with cation-exchange resin beads (DOWEX,50x, NH₄⁺) (40). Prior to spotting onto the stainless steeltarget, the sample was mixed with a matrix-assisted laser desorptionionization (MALDI) matrix consisting of 1 µl of 160 mM2,5-dihydroxybenzoic acid, 87.5 mM 1-hydroxyisoquinoline solutionin 4:1 acetone-water and air dried on the target (38). TheseO-deacylated LOS samples were then analyzed by MALDI-MS usingan Applied Biosystems Voyager DESTR+ plus (Framingham, Mass.)with an N laser (337 nm) in negative-ion mode with linear optics.The delay time was 165 ns, and the grid voltage was 94% of thefull acceleration voltage (20 kV). Spectra were acquired, averaged,and mass calibrated with an external calibrator consisting ofan equimolar mixture of angiotensin I, ACTH 18-39, and ACTH7-38 (Bachem, Torrance, Calif.).

Complementation of Haemophilus ducreyi 35000glu–. pLS88 was used as an E. coli-H. ducreyi shuttle vector (Kan^r,Sm^r, Sul^r) for cloning and complementation in these studies(6, 11). PCR primers 836 and 837 (Table 2), engineered withSacII and PstI restriction sites, were used to amplify the M.catarrhalis lgt3 gene with 265 bp of 5' flanking DNA, whichwas then ligated to SacII/PstI-digested pLS88, to constructplasmid pLS3PR-E (Table 1). Similarly, PCR primers 838 and 839(Table 2) were used to amplify an internal portion of the M.catarrhalis lgt3 gene, which was then ligated to a SacII/PstI-digestedpLS88, to construct plasmid pLS3IN-1 (Table 1). These plasmidswere electroporated into H. ducreyi 35000glu– (Cat^r) aspreviously described (3, 11). Two kanamycin- and chloramphenicol-resistanttransformants were selected, termed 35000glu–(pLS3PR-E)and 35000glu–(pLS3IN-1) as described in Table 1, for furtheranalyses.

Nucleotide sequence accession numbers. All primers were originally designed using sequence 31 of M.catarrhalis patent number WO0078968 (patent located at NCBIdatabase under Incyte Genomics accession number AX067456). Thesequence of the cluster of genes reported in this paper weredeposited with GenBank and assigned the accession number AY789049.

RESULTS AND DISCUSSION

Top

Results and Discussion

Identification, cloning, and analysis of glycosyltransferase genes of M. catarrhalis 7169. An NCBI BLAST search, with nucleotide sequences of known glycosyltransferasegenes as queries, identified a cluster of three putative LOSgenes in sequence 31 of the M. catarrhalis patented genome (NCBIpatent number WO0078968). PCR primers, designed to each gene(Table 2), were used to amplify products from the chromosomeof M. catarrhalis 7169, an LOS serotype B strain (33). Threeamplicons corresponding to open reading frames (ORFs) of 984bp, 765 bp, and 1,605 bp were obtained, and their nucleotidesequences were determined. Figure 1 illustrates the arrangementof the cluster with predicted gene products of 328, 255, and535 amino acids for lgt1, lgt2, and lgt3, respectively. A databasesearch with the translated sequences revealed homology withthree glycosyltransferases from other gram-negative bacteria.Lgt1 has 26% similarity to the WbpZ of Pseudomonas aeruginosaand an uncharacterized glycosyltransferase from Methanosarcinamazei. WbpZ is a glycosyltransferase that adds the first rhamnoseresidue onto the A-band LPS molecule O-antigen region synthesizedby P. aeruginosa (47). Lgt2 has 55% similarity to the glycosyltransferasesLgtB of Neisseria meningitidis and LpsA of Mannheimia haemolytica.These genes encode a ß(1-4) galactosyltransferaseinvolved in the addition of galactose to the ${alpha}$ -chain of the N.meningitidis LOS molecule (41, 57, 60) and a glycosyltransferaseinvolved in LPS core biosynthesis for M. haemolytica (42), respectively.Lgt3 has 28% similarity to LgtF of Haemophilus ducreyi and WaaEof Klebsiella pneumoniae. These genes both encode ß(1-4)glucosyltransferases involved in the addition of glucose tothe LOS inner core (11, 46). Sequences upstream and downstreamof the lgt cluster did not reveal significant homology to otherLOS-associated genes.

View larger version (4K):
[in this window]
[in a new window]
FIG. 1. Organization of the lgt cluster in the M. catarrhalis 7169 genome. The ORFs lgt1, lgt2, and lgt3 are 984 bp, 765 bp, and 1,605 bp in size, respectively.

Construction of M. catarrhalis LOS mutants. To determine if the genes identified above code for specificglycosyltransferase enzymes, mutants were constructed by disruptingeach lgt as described in Materials and Methods. ChromosomalDNA from selected kanamycin-resistant transformants were analyzedby PCR and sequence analysis to confirm that proper mutagenesisoccurred in M. catarrhalis 7169, yielding the mutant strains7169::lgt1KB and 7169::lgt2KM (Table 1). The arrangement oflgt3 and lgt2 (Fig. 1) suggested that these genes were cotranscribed,and reverse transcription-PCR confirmed this hypothesis (datanot shown). Based on these data, a nonpolar mutation was constructedin lgt3 (as described in Materials and Methods), and the resultingtransformant was named 7169::lgt3KH (Table 1).

To demonstrate that the loss of each Lgt expression was dueto the internal deletion in each individual lgt as describedabove, the wild-type lgt1, lgt2, and lgt3 were reintroducedinto the chromosome of each respective lgt mutant by naturaltransformation. Homologous recombination was confirmed by PCRand sequencing, and the LOS of each revertant was characterizedcompared to the corresponding mutant LOS as described below.

Comparative growth and OMP analyses. There were no significant differences in growth rate or OMPcontent between the wild-type and the three lgt mutants.

Characterization of LOS profiles from lgt mutants and revertants. LOS isolated from wild-type 7169, the lgt mutants, and the lgtmutant revertants were analyzed by SDS-polyacrylamide gel electrophoresis(Fig. 2A). All three mutants synthesized truncated LOS moleculesthat migrated more rapidly than the wild-type 7169 glycoform(Fig. 2A, lane 1), indicating loss of carbohydrate moieties.The LOS synthesized by 7169::lgt2KM (Fig. 2A, lane 2) migratedslower than the LOS glycoforms from 7169::lgt1KB (Fig. 2A, lane4) and 7169::lgt3KH (Fig. 2A, lane 6), suggesting a loss ofonly one or two sugar moieties compared to the wild-type LOSglycoform. The LOS glycoform synthesized by 7169::lgt3KH (Fig.2A, lane 6) migrated the fastest relative to the wild-type LOSand the LOS from the other two mutants, suggesting a deepertruncation of the major glycoform. The LOS molecule synthesizedby each lgt mutant revertant migrated the same as the wild-typeLOS molecule (Fig. 2A, lanes 3, 5, and 7).

View larger version (30K):
[in this window]
[in a new window]
FIG. 2. Composite of a silver-stained SDS-17.5% acrylamide gel (A) and the corresponding Western blots with monoclonal antibodies 4G5 (B) and 3F7 (C). The gel contains LOS samples from M. catarrhalis wild-type 7169 (lanes 1), 7169::lgt2KM (lanes 2), 7169::lgt2KM(REV) (lanes 3), 7169::lgt1KB (lanes 4), 7169::lgt1KB(REV) (lanes 5), 7169::lgt3KH (lanes 6), and 7169::lgt3KH(REV) (lanes 7). PAGE, polyacrylamide gel electrophoresis.

Monoclonal antibody analyses of M. catarrhalis lgt mutants and revertants. Previous studies have shown that MAb 4G5 reacts to a conservedLOS epitope, containing the terminal portion of the main branch,expressed by all M. catarrhalis isolates tested (33, 59). Westernblot analysis demonstrated that MAb 3F7 reacts to an epitopethat is specific to M. catarrhalis strains expressing serotypeB LOS, suggesting that the side chain is important for antibodyreactivity (data not shown). These antibodies were used to characterizethe phenotypic changes that resulted from the LOS truncationsseen in each mutant and to confirm the restoration of the wild-typeLOS. Figures 2B and C demonstrate that all three mutants lostreactivity to both MAb 4G5 (Fig. 2B, lanes 2, 4, and 6) andMAb 3F7 (Fig. 2C, lanes 2, 4, and 6). The loss of these LOSepitopes further supports the conclusion that this gene clusteris involved in the assembly of both the full-length chain andthe serotype B-specific side chain. In addition, the LOS isolatedfrom the corresponding revertants regained reactivity to bothMAb 4G5 (Fig. 2B, lanes 3, 5, and 7) and MAb 3F7 (Fig. 2C, lanes3, 5, and 7), confirming that the internal deletion of eachgene alone was responsible for the resulting LOS phenotypes.

Structural analysis. Mutant and wild-type LOS were O-deacylated (O-LOS) and analyzedby negative-ion linear MALDI-time of flight (TOF) mass spectrometry.The mass spectrum of the strain 7169 wild-type O-LOS correspondswith the previously published structure of serotype B LOS (9,21) (Fig. 3A), yielding six deprotonated (M-H)^– molecularions. The ions with m/z of 2,633.2, 2,756.4, 2,795.3, 2,879.6,2,918.5, and 3,042.1 are consistent with an O-LOS containingeight to nine hexoses, two KDO residues, and up to two phosphoethanolamine(PEA) groups but lacking any N-acetylhexosamine within the oligosaccharide.The lack of N-acetylhexosamine further confirms the LOS as Bserotype, since both A- and C-serotype LOS contain a GlcNAc(7, 8, 21).

View larger version (53K):
[in this window]
[in a new window]
FIG. 3. Linear negative-ion MALDI-TOF spectra of O-LOS from M. catarrhalis strains. (A) Wild-type 7169; (B) lgt1 mutant-7169::lgt1KB; (C) lgt2 mutant-7169::lgt1KM; (D) lgt3 mutant-7169::lgt3KH. All ions are represented as average (M-H)^– masses. Intact O-LOS molecules show numerous molecular ions due to heterogeneity in hexose and phosphoethanolamine content. Molecular ions undergo in-source fragmentation to give lipid A and oligosaccharide prompt fragments as shown. *, ions occurring due to ß-elimination of phosphoric acid (H₃PO₄) from the O-LOS; OS, oligosaccharide; lipid A, O-deacylated lipid A; Hex, Hexose; Kdo, 2-keto-3-deoxyoctonate.

O-LOS isolated from the lgt1 mutant showed two (M-H)^–molecular ions with m/z of 2,107.8 and 1,984.8, correspondingto a truncated oligosaccharide consisting of four hexoses andtwo KDO moieties with zero to one PEA moieties, respectively(Fig. 3B). Also present in the mass spectrum is an ion witha m/z of 1,886.9, which results from the loss of phosphoricacid (H₃PO₄) from the ion with a four-hexose structure at am/z of 1,984.8. Composition analysis showed the presence ofglucose and KDO but no galactose in the LOS. Linkage analysisconfirmed the presence of a 3-, 4-, 6-substituted glucose andterminal glucose only, whereas for the wild type, 2- and 4-substitutedhexoses were also observed (data not shown).

This mutant is consistent with the loss of an ${alpha}$ (1-2) glucosyltransferase,which would leave an oligosaccharide containing two KDO, one3-substituted glucose, and three terminal glucose residues,thus:

{zjb0090546600e1}
MALDI-TOF-MSof the lgt2 mutant O-LOS showed the presence of (M-H)^–molecular ions smaller than the wild-type O-LOS but larger thanthe lgt1 mutant O-LOS (Fig. 3C). The ions with m/z of 2,308.9,2,432.2, and 2,555.5 correlate to an O-LOS containing six hexoses,two KDO moieties, and zero to two PEA moieties. Also presentwere ions with m/z of 2,334.5 and 2,210.7, which are thoughtto arise from the loss of H₃PO₄ from ions with m/z of 2,434.2and 2,308.9, respectively. Composition analysis of the O-LOSsuggested that the major component of the oligosaccharide wasglucose and that the mutant LOS lacks galactose. Linkage analysisof oligosaccharide derived from the LOS shows the presence ofa 3-, 4-, 6-substituted glucose and terminal glucose, as seenfor the lgt1 mutant; in addition, a partially methylated alditolacetate representing 2-substituted glucose was observed. Takentogether, the data for this lgt2 mutant are consistent witha structure containing 2-substituted glucose but lacking galactose:

{zjb0090546600e2}
This structure, truncated at the siteof galactose addition, would correlate with Lgt2 being a ß(1-4)galactosyltransferase.

Finally, MALDI-TOF MS analysis of O-LOS isolated from the lgt3mutant yielded ions with m/z of 2,308.9, 2,432.2, and 2,553.3which correspond to an oligosaccharide containing one hexose,two KDO moieties, and zero, one, or two PEA moieties, respectively(Fig. 3D). Also present were ions with m/z of 1,400.9 and 1,523.7corresponding to the loss of H₃PO₄ from the ions with m/z of1,498.4 and 1,621.6, respectively. This mutant produces themost deeply truncated LOS molecules of the mutants studied andcan only correspond to the structure:

wherelinkage between the glucose and KDO is likely to be ${alpha}$ (1-5), aspreviously shown by Edebrink et al. (9). It is unlikely thatthe Lgt3 enzyme is trifunctional; thus, it seems that if thisenzyme is a glycosyltransferase, it adds either 3-, 4-, or 6-linkedglucose to the growing LOS molecule and that this addition ofa single glucose molecule is required before further elongationof the OS can occur.

As previously mentioned, Lgt3 has homology to LgtF of H. ducreyi,a known ß(1-4) glucosyltransferase (11); thus, thefollowing studies were performed to test this hypothesis.

Complementation of the glucosyltransferase (lgtF)-deficient H. ducreyi 35000glu–. To determine whether the 7169 lgt3 functions as a ß(1-4)glucosyltransferase, complementation studies were performedusing the H. ducreyi 35000glu– mutant, previously developedin this laboratory (11). This isogenic mutant has a single defectin the ß(1-4) glucosyltransferase, LgtF, and assemblesa truncated LOS lacking the terminal, surface-exposed lactosamineepitope, expressed by the parent strain 35000 and detected byMAb 1B2-1B7 (36, 58). Two plasmids were constructed in the pLS88shuttle vector using standard cloning techniques (see Materialsand Methods). One plasmid, pLS3PR-E (Table 1), contained a 2.2-kbinsert consisting of the entire M. catarrhalis 7169 lgt3 ORFas well as its 5' flanking DNA of $~$ 265 bp. The other plasmid,pLS3IN-1 (Table 1), contained a 1.3-kb insert consisting ofan internal portion of the 7169 lgt3 gene without any 5' flankingDNA.

After electroporation of these constructs into H. ducreyi 35000glu–,kanamycin-chloramphenicol-resistant transformants were screenedfor the restoration of reactivity to MAb 1B2-1B7. Figure 4 isa composite showing colony lift assays probed with MAb 1B2-1B7,denoting the presence of the terminal lactosamine epitope onthe LOS of the wild-type 35000 (Fig. 4A) and the loss of thisepitope in 35000glu– (Fig. 4B) as expected (11, 36, 58).More importantly, these data demonstrate that MAb 1B2-1B7 reactsto the complemented 35000glu–(pLS3PR-E) (Fig. 4D), indicatingthat expression of lgt3 in trans corrected the defect, resultingin assembly of the full-length LOS chain. In addition, the mutantcomplemented with the internal portion of lgt3, 35000glu–(pLS3IN-1)did not restore the epitope recognized by MAb 1B2-1B7 (Fig.4C). These results indicate that the 7169 Lgt3 was expressedin the LgtF-deficient H. ducreyi mutant, resulting in complementationof the defect, providing evidence that the Lgt3 can functionas a ß(1-4) glucosyltransferase in M. catarrhalis7169.

View larger version (28K):
[in this window]
[in a new window]
FIG. 4. A composite of colony lift assays of H. ducreyi 35000 (A), 35000glu– (B), 35000glu–(pLS3IN-1) (C), and 35000glu–(pLS3PR-E) (D) probed with MAb 1B2-1B7.

While the data presented in these studies conclude that Lgt1and Lgt2 function as an ${alpha}$ (1-2) glucosyltransferase and a ß(1-4)galactosyltransferase, respectively, these data do not conclusivelydefine the function of Lgt3. In addition, our data also suggeststhat the Lgt1 may be able to add a ${alpha}$ (1-2)-linked glucose ontosimilar but not identical acceptor molecules in the two differentchains (Fig. 5). It is interesting that both Lgt1 and Lgt2 likelyfunction in the addition of carbohydrates to both the main branchand the serotype-specific branch of the LOS molecule; however,it is possible that the addition of only one sugar is occurringbefore another very similar enzyme is able to add the same sugarto the other branch.

View larger version (10K):
[in this window]
[in a new window]
FIG. 5. Biochemical structure of the wild-type 7169 LOS molecule and the positions of the glycosyltransferase enzymes Lgt1, Lgt2, and Lgt3 as they may function in LOS biosynthesis for this organism as determined in these studies.

Perhaps the most interesting results from our studies involvedthe analysis of the lgt3. Based on the homology data, we wereinitially surprised by the structural studies that confirmedthat the lgt3 mutant contained a single glucose attached tothe KDO. This was not the result of pleiotropic effects, asour molecular analysis of the mutant and the restoration ofthe wild-type phenotype demonstrated that the LOS defect wasthe result of disrupting lgt3 alone. However, these data didnot provide insight into a possible function for Lgt3. The LOSstructural data suggested three possible functions for Lgt3as either a ß(1-3), a ß(1-4), or ß(1-6)glucosyltransferase. The other possible explanation was thatLgt3 is a multifunctional enzyme; however, while this wouldbe a novel observation, more studies are needed to address thisissue.

The heterologous complementation studies, using the H. ducreyiß(1-4) glucosyltransferase mutant, provided importantinsight into a probable function for Lgt3. Using the M. catarrhalislgt3 to complement the isogenic 35000glu– mutant in transindicated that Lgt3 can function as a ß(1-4) glucosyltransferase.Furthermore, these data also show that the M. catarrhalis Lgt3is capable of adding a glucose onto a heptose substrate in theH. ducreyi 35000glu– background. This is a somewhat surprisingobservation, as there is no heptose in M. catarrhalis LOS andLgt3 would be predicted to add a glucose onto a glucose substratein this bacterium. Nevertheless, based on the previously publishedcharacterization of the H. ducreyi 35000glu– mutant, aß(1-4)-linked glucose had to be added to the heptoseto allow for the completion of the full-length H. ducreyi LOSchain and the restoration of the MAb 1B2-1B7 epitope (11, 36,58). In addition, this is the first study describing the expressionof a M. catarrhalis gene product in H. ducreyi, which is animportant observation, as currently there are no reliable vectorsthat allow for complementation in trans in M. catarrhalis.

These studies further suggested that the function of the Lgt3enzyme is essential for the continuation of both the main oligosaccharidechain and the serotype-specific side chain. Thus, in the absenceof the ß(1-4) linked glucose addition to the coreglucose, the biosynthesis and assembly of the main LOS chainare disrupted. Previous studies with N. meningitidis and H.influenzae have reported that the order of addition of specificsugar residues onto the growing LOS molecule is critical tothe function of enzymes that act further downstream in the biosynthesisof the full-length LOS molecule (22, 29). One study involvingrfaK of N. meningitidis concluded that the absence of this transferase,responsible for the addition of N-acetylglucosamine to HepII,causes a lack of ${alpha}$ -chain extension from HepI in the LOS rfaKmutant glycoform due to structural constraints imposed by theincomplete biosynthesis of the LOS inner core (29). Anotherstudy with glycosyltransferase enzymes of H. influenzae concludedthat lpsA, lic2A, lgtC, and lgtD were all required for the sequentialaddition of the glycoses to the terminal inner core heptoseto give the globotetraose structure of its LOS molecule (22).Based on our data and these previous studies, we hypothesizethat Lgt3 functions at a critical point in biosynthesis andassembly of M. catarrhalis LOS, and in the absence of this enzymaticactivity, the acceptor molecule will not be present for theproper function of glycosyltransferase enzymes that continuethe elongation of either the main branch or the side chain,preventing the completion of the principal glycoform. However,further biochemical analyses of Lgt3 with regard to specificsubstrates and acceptor molecules are needed to delineate theprecise action of this critical enzyme in M. catarrhalis LOSbiosynthesis. Together, these data provide insight into theassembly of M. catarrhalis LOS, and the specific mutants providevaluable tools that will be essential in determining the roleof this glycolipid in pathogenesis.

ACKNOWLEDGMENTS

This research was supported by Public Health Service Researchgrants AI46422, DC005837 (A.A.C.), and AI31254 (B.W.G.). K.J.E.is also supported as a graduate student fellow by NIH traininggrant AI07614. This work was supported by the Department ofEnergy-funded (DE-FGOZ-93ER-20097) Center for Plant and MicrobialComplex Carbohydrates.

We thank Melanie J. Filiatrault for the construction of theH. ducreyi 35000glu– (LgtF) mutant strain. We thank theCenter for Complex Carbohydrate Research at the University ofGeorgia, in particular Trina Abney and Parastoo Azadi, for thepreparation and analysis of the data involving the LOS isolatedfrom the M. catarrhalis 7169 and lgt mutants. We also thankAlan Lesse for assistance in sequence analysis.

FOOTNOTES

* Corresponding author. Mailing address: Department of Microbiology, University at Buffalo, Biomedical Research Bldg. Rm. 143, 3435 Main St., Buffalo, NY 14214. Phone: (716) 829-2673. Fax: (716) 829-3889. E-mail: aac{at}buffalo.edu.

REFERENCES

Top