Construction and Characterization of an Acapsular Mutant of Mannheimia haemolytica A1

Abstract

The nmaA and nmaB genes, which code for UDP-GlcNAc-2-epimerase and UDP-ManNAc-dehydrogenase, respectively, are involved in capsular polysaccharide biosynthesis in Mannheimia haemolytica A1. A chloramphenicol resistance (Cm^r) cassette cloned behind an M. haemolytica A1 promoter, plpcat, was created and used to interrupt nmaA and nmaB. A 1.3-kbp DNA fragment that encompasses part of nmaA and nmaB was replaced by the 1.0-kbp plpcat, resulting in a knockout mutant which is Cm^r and unable to synthesize N-acetylmannosamine (ManNAc) and N-acetylmannosaminuronic acid (ManNAcA). The DNA replacement was confirmed by Southern hybridization and PCR analyses of the nmaA and nmaB loci. Electron microscopy examination of the mutant showed the absence of capsular materials compared to the parent strain. The loss of NmaA and NmaB activity was confirmed by analysis of carbohydrate moieties using capillary electrophoresis. Serum sensitivity assays indicated that the acapsular mutant is as resistant as the encapsulated parent to complement-mediated killing by colostrum-deprived calf serum but is more sensitive to killing by immune bovine serum. Analysis of lipopolysaccharide prepared from the acapsular mutant and encapsulated parent confirmed that these strains have long O-polysaccharide chains, possibly conferring resistance to serum-mediated killing.

Mannheimia (Pasteurella) haemolytica A1 is the principal causative agent of bovine pneumonic pasteurellosis, or shipping fever (1, 39), resulting in significant economic loss to the cattle industry (50). Several M. haemolytica A1 virulence factors have been characterized; these include leukotoxin (3, 21, 27, 46), neuraminidase (45), lipopolysaccharide (LPS) (49), and transferrin binding proteins (35, 36). M. haemolytica A1 cells are also encapsulated. The capsular polysaccharide (CPS) of this organism has been implicated in mediating resistance to killing by serum (8), impairing phagocytosis by bovine neutrophils (8, 12) and alveolar macrophages (13), and facilitating microcolony formation in the pneumonic bovine lung (31, 32).

The M. haemolytica A1 CPS biosynthetic locus was recently analyzed (25). The 16-kbp DNA region contains 12 genes involved in the export and biosynthesis of the serotype-specific extracellular CPS. The genetic organization of the CPS biosynthetic cluster is similar to that established for group II capsules where the CPS biosynthesis genes are flanked by two operons encoding proteins involved in CPS export (51).

The M. haemolytica A1 CPS is composed of disaccharide repeats of N-acetylmannosaminuronic acid (ManNAcA) β-1,4 linked with N-acetylmannosamine (ManNAc) (2). We previously demonstrated, through complementation studies (25), that two CPS biosynthesis genes, nmaA and nmaB, are functional homologues of Escherichia coli K-12 enterobacterial common antigen biosynthesis genes wecB (rffE) and wecC (rffD), respectively. The wecB and wecC genes code for the enzymes UDP-GlcNAc-2-epimerase and UDP-ManNAc-dehydrogenase, respectively (24), and are required for the biosynthesis of enterobacterial common antigen in E. coli (29).

The construction of M. haemolytica mutants has been difficult for many researchers. Hence, without genetically defined mutants, the characterization of this organism's virulence factors has been limited. We have observed that M. haemolytica A1 poorly recognizes foreign promoters, hindering the use of regular antibiotic resistance cassettes as selective markers in the allelic exchange approach to isogenic mutant development. In this study, this problem was circumvented by creating a chloramphenicol resistance cassette that is expressed from an M. haemolytica promoter. Using this cassette, we then constructed a CPS mutant of M. haemolytica A1 using the allelic exchange approach described by Fedorova and Highlander (15). The mutation to the CPS biosynthesis genes nmaA and nmaB was characterized, and the biological significance of CPS in conferring serum resistance to M. haemolytica A1 was examined.

MATERIALS AND METHODS

Bacterial strains, plasmids, and culture conditions.

M. haemolytica A1 strain SH1217 was provided by Sarah Highlander (Baylor College of Medicine, Houston, Tex.). E. coli strain JN1, a bovine mastitis isolate, was provided by Carlton Gyles (University of Guelph, Guelph, Ontario, Canada). Pseudomonas aeruginosa strain PA01 was provided by Joseph Lam (University of Guelph). E. coli HB101 (6) was used for plasmid propagation and cloning. E. coli and P. aeruginosa strains were cultured in Luria-Bertani media supplemented with ampicillin (100 μg/ml), chloramphenicol (25 μg/ml), or spectinomycin (170 μg/ml) when required. M. haemolytica A1 strains were cultured in brain heart infusion (BHI) media supplemented with ampicillin (50 μg/ml), chloramphenicol (5 μg/ml), or novobiocin (5 to 10 μg/ml) when required. All cultures were grown at 37°C.

Enzymes and chemicals.

Restriction endonucleases, Taq polymerase, and DNA and protein molecular mass standards were purchased from Amersham Pharmacia Biotech (Baie d'Urfe, Quebec, Canada), Boehringer Mannheim (Laval, Quebec, Canada), Gibco BRL/Life Technologies (Burlington, Ontario, Canada), and Bio-Rad Laboratories (Mississauga, Ontario, Canada). Uridine-diphospho-N-acetyl-glucosamine (UDP-GlcNAc) and uridine-diphospho-N-acetyl-galactosamine were purchased from Sigma-Aldrich (Oakville, Ontario, Canada).

Growth curves.

The growth profiles of M. haemolytica strains was determined by inoculating 250 ml of BHI with 1% overnight culture and monitoring the optical density at 600 nm every 20 min, as well as spread plating appropriate dilutions to enumerate the corresponding number of CFU per milliliter.

DNA manipulation.

M. haemolytica cells were prepared and electroporated as described elsewhere (33). To improve the yield of plasmid DNA recovered from E. coli cells, a spectinomycin amplification step (42) was used to prepare DNA for subsequent electrotransformation of M. haemolytica cells.

A modified cat cassette, plpcat, was created to provide consistent expression of chloramphenicol acetyltransferase in M. haemolytica. PCR primers plp4/StyI (5′ GCAATCAGACCAAGGGCAACTAT 3′) and plp4/NdeI (5′ TCCAGTTTCATATGAAGTTTCCT 3′) were used to amplify a 300-bp region containing the promoter of a previously characterized M. haemolytica lipoprotein gene, plp4 (34), from M. haemolytica A1 genomic DNA. A 700-bp promoterless cat gene was amplified from plasmid pSL33 (26) using primers CAT/NdeI (5′ CTAGAAGAAGCTCATATGGAGAAA 3′) and CAT/MluI (5′ CGGGTCATTACGCGTCCAGAAA 3′). The PCR products were purified (Gibco/BRL PCR purification kit); digested with StyI and NdeI or NdeI and MluI, respectively; and cloned into StyI-, MluI-digested plasmid vector pPHCPX2.2 (25), to create pCPXCAT.

The knockout vector was created in two steps. First, PCR primers CAP/KpnI (5′ AGAATTTGGTACCTCGCCTGAA 3′) and CAP/XbaI (5′ GCAAATTTCTAGACATCCGCT 3′) were used to amplify a 3.9-kbp region containing nmaA and nmaB from M. haemolytica A1 genomic DNA. This fragment was purified, digested with KpnI and XbaI, and cloned into KpnI, XbaI-digested shuttle vector pNF2176, creating pLMCAP (Fig. 1A). pNF2176 is a shuttle vector that confers resistance to ampicillin and propagates in M. haemolytica and E. coli (16). Subsequently, a 1.3-kbp fragment, spanning parts of nmaA and nmaB, was excised from pLMCAP by SfuI and MunI digestion. Then, plpcat was PCR amplified from pCPXCAT using primers plpcat/SfuI (5′ AAGGGCAACTATTCGAAGCAA) and plpcat/MunI (5′ TGCTGTAATCAATTGTCGTTAA). The 1.0-kbp product was digested with SfuI and MunI and inserted into the SfuI, MunI sites of pLMCAP to construct the knockout vector pLMCAT (Cm^r Ap^r) (Fig. 1B).

FIG. 1.

Restriction map showing the M. haemolytica A1 nmaAB region that was used in the construction of mutagenesis plasmid pLMCAT. (A) Primers CAP/KpnI (K) and CAP/XbaI (X) were used to amplify a 3.9-kbp region of the M. haemolytica A1 chromosome containing genes nmaA and nmaB. This fragment was cloned into the XbaI, KpnI sites of plasmid pNF2176 to produce plasmid pLMCAP. (B) Plasmid pLMCAP was digested with restriction enzymes MunI and SfuI, removing a 1.3-kbp fragment that spans nmaA and nmaB. The chloramphenicol resistance cassette, plpcat (1.0 kbp), was inserted into the MunI, SfuI sites, creating the mutagenesis plasmid pLMCAT.

Antibiotic selection of mutant strains and plasmid curing.

Plasmid pLMCAT was introduced into M. haemolytica A1 SH1217 cells by electroporation. Transformants were plated onto BHI agar containing chloramphenicol, and 150 Cm^r isolates were replica plated onto BHI plates containing ampicillin. All isolates were Ap^r Cm^r. The isolates were pooled into 5 ml of BHI and propagated overnight without antibiotic selection. The overnight culture was diluted 1/100 into 5 ml of BHI containing novobiocin (10 μg/ml). On each of the next 2 days the culture was diluted 1/100 and grown in 3 ml of BHI containing novobiocin (7 μg/ml). On day 4, 0.1 ml of culture was plated onto BHI-chloramphenicol plates and incubated overnight. The next day 1,100 single colonies were picked and patched onto BHI-ampicillin plates and BHI-chloramphenicol plates to select for the integration of plpcat into the chromosome (conferring Cm^r) and loss of the plasmid vector (resulting in Ap^s).

PCR analysis of genomic DNA.

Genomic DNA was isolated from M. haemolytica cells using QIAGEN Genomic-tips (Mississauga, Ontario, Canada) according to the manufacturer's protocol. PCR amplification from genomic DNA was performed using PLATINUM Taq DNA polymerase (Gibco BRL) according to the manufacturer's suggested protocol. The PCR was performed with primers CAP/KpnI and CAP/XbaI as follows: thirty cycles of DNA denaturation at 94°C for 1 min, primer annealing at 60°C for 1.5 min, and extension at 72°C for 1.5 min.

Southern hybridization analysis of genomic DNA.

One microgram of genomic DNA was digested with HindIII for 3 h. The DNA was separated by electrophoresis on a 0.8% agarose gel and transferred to a nylon membrane as described elsewhere (30). The membrane was hybridized overnight under high-stringency conditions with one of the following probes: the 1.3-kbp MunI-SfuI fragment of pLMCAP or the 700-bp cat cassette, amplified using primers CAT/NdeI and CAT/MluI. The DNA probes were labeled with digoxigenin (DIG)-dUTP using the random primed labeling method according to manufacturer's protocol (Boehringer Mannheim). The Southern blot was washed under high-stringency conditions (0.5× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate], 1% [wt/vol] sodium dodecyl sulfate) twice for 15 min at 68°C. The DIG-labeled probe was detected using anti-DIG-ampicillin antibodies and disodium 3-{4-methoxyspiro[1,2-dioxethane-3,2′(5′-chloro)tricyclo(3.3.1.^3, 7)decan]-4-yl}phenyl phosphate (CSPD; Boehringer Mannheim) according to the manufacturer's protocol. The DIG-labeled DNA probe was stripped from the membrane by rinsing the blot in deionized water for 5 min at room temperature and washing the blot twice (twice for 15 min) in 0.2 M NaOH with 0.1% (wt/vol) sodium dodecyl sulfate at 37°C. The membrane was equilibrated in 2× SSC (pH 7.0) for 5 min at room temperature, air dried, and stored in a sealed bag.

Electron microscopy.

Bacterial cells were stained with cationized ferritin for electron microscopy as previously described (48). Briefly, fresh bacterial colonies from agar plates were suspended in 0.1 M HEPES buffer (pH 6.8), washed twice, and incubated for 30 min in 10-mg/ml ferritin solution (Sigma-Aldrich, Oakville, Ontario, Canada). The cells were washed twice with HEPES buffer and suspended in 2% glutaraldehyde for 1.5 h. Washed cells were suspended in 2% osmium tetroxide for 1.5 h, washed again, and enrobed in 2% Noble agar. The cells were suspended in 2% uranyl acetate for 1 h, washed again, and dehydrated with 100% ethanol. The cells were suspended in 50/50 L R White/Ethanol and then in pure L R White resin. Cells in L R White in a gelatin capsule were heated at 60°C and cut into ultrathin sections (75 nm), which were stained with uranyl acetate and lead citrate. The samples were viewed with a Leo 912AB electron microscope at 100 kV and analyzed with Soft Imaging System software, AnalySYS (Lakewood, Colo.).

NmaA-NmaB activity assay.

M. haemolytica strains were grown overnight in 50 ml of BHI, collected by centrifugation, and resuspended in 10 ml of 50 mM Tris-HCl (pH 7.9). The cells were sonicated on ice using a pulse setting, with 60% output, at 10-s intervals for 3 min and then centrifuged at 100,000 × g for 100 min at 4°C. The cellular supernatant was recovered and used the same day for the assay. The protein concentration of each of the crude cellular extracts was determined using the Bio-Rad Protein Assay Kit. The assay was adapted from the protocol previously described by Creuzenet et al. (10). For each assay, 1.7, 5.5, or 0.5 μg of total protein was used in a 35 μl-reaction volume containing 20 mM Tris-HCl (pH 7.9) and 1 mM UDP-GlcNAc (Sigma). When appropriate, 0.1 mM NAD⁺ and 0.1 mM NADP⁺ (Sigma) were added as cofactors. The reactions were incubated at 37°C for a range of times (from 2 min to 3 h) and were stopped by boiling for 5 min. The samples were frozen at −20°C until analysis by capillary electrophoresis (CE). Prior to CE analysis the samples were thawed, diluted by one-half with sterile water, and centrifuged at 6,000 × g for 5 min. The parameters for CE analysis were adapted from those previously described (10). CE analysis was performed using a P/ACE MDQ system (Beckman Instruments) with UV (254 nm) detection. The capillary of fused silica was pretreated with 0.2 M NaOH and H₂O rinses. Each sample was run for 20 min at 22 kV using 25 mM sodium tetraborate (pH 9.0) as the running buffer. Peak integration and calculation of peak area were done with the Beckman P/ACE Station software.

Serum sensitivity assay.

The serum sensitivity assay was modified from the procedure described by Slavic et al. (44). Overnight bacterial suspensions were subcultured in BHI and grown to mid-log phase. Cells were harvested from 1 ml of the subculture, washed once in buffered saline-gelatin (BSG) (8.5 g of NaCl, 0.3 g of KH₂PO₄, 0.6 g of NaHPO₄, and 10 ml 10% porcine gelatin combined in 990 ml of distilled H₂O; pH to 7.4), and suspended in BSG to a cell density of 10⁷ CFU/ml. Whole blood was obtained from a clinically healthy cow, or from a colostrum-deprived, immunologically naive Holstein calf, and serum was collected by centrifugation (800 × g, 15 min, 4°C). The anti-M. haemolytica A1 leukotoxin neutralizing antibody titer was determined for each serum as described previously (18). For each experiment, an aliquot of fresh serum was heat treated (56°C for 30 min) to inactivate both the alternative and classical complement pathways (HI serum), was left untreated (N serum), or was chemically treated with EGTA and MgCl₂ (10 mM each, final volume) to inactivate the classical complement pathway (CPI serum). The bacterial cells were incubated in 66% HI, N, or CPI serum at 37°C for 3 h with shaking (60 rpm). SH1217, LMCap1, and JN1 cells were sampled at 0, 1, 2, and 3 h; serially diluted; plated onto BHI, BHI-chloramphenicol (5 μg/ml), or Luria-Bertani plates, respectively; and incubated overnight at 37°C. Percentage viability was determined for each time point by dividing the mean number of CFU per milliliter by the initial count × 100.

LPS preparation and silver staining.

M. haemolytica A1 strains SH1217 and LMCap1, E. coli JN1, and P. aeruginosa PA01 cells were scraped from agar plates and washed in 0.5 ml of sterile saline. The cells were pelleted, suspended in 0.2 ml of Hitchcock and Brown lysis buffer (19), and boiled for 30 min. Proteinase K (Bioshop, Burlington, Ontario, Canada ) was added (final volume, 200 μg/ml), and the samples were incubated at 58°C for 2 h. The samples were electrophoresed in a 12.5% acrylamide gel. To visualize the LPS, the gel was then silver stained according to the procedure described by Fomsgaard et al. (17).

RESULTS

Isolation of mutagenic strain.

The knockout vector, LMCAT, was designed to disrupt the M. haemolytica A1 CPS biosynthesis genes nmaA and nmaB by allelic exchange. This plasmid was constructed by cloning a 3.9-kbp fragment, containing complete copies of nmaA and nmaB, into the M. haemolytica-E. coli shuttle vector pNF2176 (Ap^r) and the subsequent replacement of the 1.3-kbp MunI/SfuI fragment with the 1.0-kbp plpcat (Cm^r) (Fig. 1B). Six micrograms of plasmid pLMCAT was amplified, purified, and used for electroporation of M. haemolytica SH1217 cells. One hundred and fifty Cm^r isolates were recovered from BHI plates containing chloramphenicol (5 μg/ml) and replica plated onto BHI-ampicillin (50 μg/ml) and BHI-chloramphenicol (5 μg/ml). All the colonies were resistant to both ampicillin and chloramphenicol, indicating that they contained pLMCAT. Plasmid preparations performed on several isolates confirmed the presence of pLMCAT. The isolates were pooled and grown in subinhibitory concentrations of novobiocin. Novobiocin antagonizes the activity of the DNA gyrase B subunit and inhibits DNA replication. A subinhibitory concentration of novobiocin allows the cells to use the limited DNA gyrase activity for chromosomal replication and sacrificing plasmid DNA replication. In subsequent cultures, the subinhibitory concentration of novobiocin may become inhibitory as less plasmid DNA is maintained per cell, thus reducing the concentration of novobiocin. After novobiocin selection, 1,100 colonies were replica plated onto BHI-ampicillin (50 μg/ml) and BHI-chloramphenicol (5 μg/ml). After overnight incubation, 13 Cm^r but Ap^s colonies were recovered, suggesting that the plasmid may have been lost and plpcat may have integrated into the chromosome. Several isolates did not retain a Cm^r phenotype after passage for 3 days in BHI without chloramphenicol, while others retained plasmid DNA. Ultimately, six Cm^r Ap^s isolates were further characterized by PCR, and two were analyzed by Southern hybridization (see below).

PCR analysis of mutated locus.

PCR primers CAP/KpnI and CAP/XbaI were used to amplify the expected 3.9-kbp product from M. haemolytica SH1217 genomic DNA. Using genomic DNA prepared from the six Cm^r Ap^s isolates a PCR product of 3.6 kbp was amplified. In constructing the knockout vector, pLMCAT, a 1.3-kbp fragment was replaced with the 1.0-kbp plpcat, accounting for the 300-bp difference in these PCR products (Fig. 2).

FIG. 2.

PCR analysis of the nmaAB region of several putative mutants. The PCR was performed with CAP/KpnI and CAP/XbaI. The template DNA is from SH1217 (lane 1), and the following Cm^r Ap^s isolates were used: LMCap1, LMCap2, LMCap5, LMCap9, LMCap10, and LMCap11 (lanes 3 to 8, respectively). Lane 2 is a control, performed without a DNA template. The positions of the 3.9- and 3.6-kbp products are as indicated.

Southern hybridization analysis.

Southern hybridizations were performed to examine the integration of plpcat into the chromosome, replacing the 1.3-kbp MunI/SfuI fragment spanning nmaA and nmaB on the chromosome. The DIG-labeled cat cassette was used to probe HindIII-digested genomic DNA prepared from SH1217, as well as two Cm^r Ap^s isolates, LMCap1 and LMCap5. The cat probe hybridized with LMCap1 and LMCap5 DNA but not with SH1217 DNA (Fig. 3B). Specifically, the probe hybridized with a fragment of 1.8 kbp, as expected. The chromosome contains two HindIII sites, 2.1 kbp apart, that flank the MunI and SfuI sites. Taking into account the loss of 300 bp when the 1.3-kbp MunI/SfuI fragment was replaced with the 1.0-kbp cassette, the probe was expected to hybridize with a DNA fragment of approximately 1.8 kbp, as in Fig. 3B. A second Southern hybridization confirmed the loss of the 1.3-kbp MunI/SfuI fragment from the LMCap1 and LMCap5 chromosomes. The DIG-labeled MunI/SfuI fragment was used to probe the same membrane that was previously probed with the cat probe. The probe only hybridized with SH1217 DNA and not with DNA from LMCap1 or LMCap5 (Fig. 3C). This indicates that in LMCap1 and LMCap5 plpcat replaced the DNA between the MunI and SfuI sites, spanning nmaA and nmaB, conferring the phenotypic chloramphenicol resistance. Since there were no discernible differences between strains LMCap1 and LMCap5, only LMCap1 was used for the subsequent analyses.

FIG. 3.

(A) Agarose gel electrophoresis of HindIII-digested genomic DNA from SH1217 (lane 2), LMCap1 (lane 3), and LMCap5 (lane 4). Lane 1 contains HindIII-digested lambda DNA, and lane 5 contains a 1-kbp ladder. (B) Southern hybridization of gel from 3A, probed with a DIG-labeled cat gene. (C) Southern hybridization of gel from panel A, probed with DIG-labeled 1.3 kbp-MunI/SfuI DNA fragment. The molecular size standards, in kilobase pairs, are shown to the left of each panel.

Phenotypic analysis of LMCap1.

LMCap1 colonies grown on BHI were noted to be smaller and flatter and to have more-distinct borders than SH1217 colonies. H1217 colonies are also pearl-like, whereas LMCap1 colonies are more opaque in color. SH1217 and LMCap1 cultures have a similar growth profile, but LMCap1 cultures grow more quickly and reach mid-logarithmic phase faster than the SH1217 culture (data not shown). On the other hand, smaller colonies were observed when LMCap1 was complemented with pLMCAP. This may be due to a slight deleterious effect of having multiple copies of nmaA and nmaB genes in the cell.

Slide agglutination was performed using serial twofold dilution of sera from calves that had been vaccinated with a CPS-enriched formulation (unpublished results). The results showed that while SH1217 was agglutinated up to a log₂ dilution of 5, LMCap1 was only agglutinated up to a log₂ dilution of 2, suggesting a reduction in the amount of surface molecules.

Electron microscopy was used to examine ferritin-stained LMCap1 cells. The parental strain, SH1217, is covered with a thick layer that readily binds cationized ferritin. The CPS extends almost 200 nm from the bacterial surface without antibody stabilization and can also be seen to have sloughed off SH1217 cells (Fig. 4A). Conversely, CPS was absent from around LMCap1 cells, and the area surrounding the cells is clear of any CPS debris (Fig. 4B). Plasmid pLMCAP or shuttle vector pNF2176 was electroporated into LMCap1 cells. Both LMCap1(pNF2176) and LMCap1(pLMCAP) cells were ferritin stained and examined by electron microscopy. LMCap1(pNF2176) cells look identical to LMCap1: lacking a stained CPS coating and maintaining a clear area surrounding the cells (Fig. 4C). In contrast, LMCap1(pLMCAP) cells resemble the parental strain, SH1217, and express CPS, although it is slightly patchy and less uniformly distributed on the bacterial surface (Fig. 4D).

FIG. 4.

Electron micrographs of ferritin-stained thin sections of M. haemolytica cells SH1217 (A), LMCap1 (B), LMCap1(pNF2176) (C), and LMCap1(pLMCAP) (D). The bar represents 200 nm.

Confirming the loss of CPS biosynthesis.

Enzymatic assays were performed using crude cellular proteins isolated from LMCap1 cells to investigate if inactivation of nmaA and nmaB eliminated conversion of UDP-GlcNAc to UDP-ManNAc and UDP-ManNAcA. In the first assay, 1.7 μg of crude cellular protein isolated from SH1217 or LMCap1 was incubated with UDP-GlcNAc for 2, 3, or 4 h. The samples were then subjected to CE, along with a UDP-GlcNAc standard. The UDP-GlcNAc standard resolved after 11.8 min of electrophoresis (Fig. 5); this peak was also resolved from the SH1217 and LMCap1 samples. A second peak was resolved from the SH1217 and LMCap1 samples, at 12.6 min. It was determined that this peak represents UDP-GalNAc, using a UDP-GalNAc standard. A third peak was resolved from the SH1217 sample, at 11.4 min, and was absent from the LMCap1 sample and the UDP-GlcNAc control sample (Fig. 5). While it is suspected that the carbohydrate is UDP-ManNAc, the predicted product of NmaA activity on UDP-GlcNAc, a commercial standard was not available to confirm its identity.

FIG. 5.

CE analysis of NmaA and NmaB activity from SH1217 and LMCap1 crude protein samples. The reaction mixture contained 1.7 μg of total protein and 1 mM UDP-GlcNAc in a total reaction volume of 35 μl. The samples were incubated at 37°C for 2 h and electrophoresed. A UDP-GlcNAc sample is included as a control. The arrow indicates a carbohydrate species resolving at 11.4 min that is absent from the LMCap1 sample but present in the SH1217 sample. AU, arbitrary units.

The area under each peak was calculated for the SH1217 assay. As a percentage, the UDP-GalNAc peak represents 21.3% of the total detectable carbohydrates, the UDP-GlcNAc peak represents 73.2% of the total detectable carbohydrates, and the unknown peak represents 5.7%, after 2 h of incubation. Extending the incubation to 3 h did not alter the area under these three peaks, nor did increasing the amount of cellular protein used for the assay to 5.5 μg (data not shown).

Another assay was performed using 0.55 μg of protein isolated from SH1217 and LMCap1 cells, as well as from strains LMCap1(pNF2176) and LMCap1(pLMCAP). The complemented strain was analyzed to determine if synthesis of the carbohydrate resolving at 11.4 min (from the previous assay) could be restored in LMCap1 when nmaA and nmaB were introduced in trans. The cellular proteins were incubated with UDP-GlcNAc for 2 or 10 min. In the LMCap1(pLMCAP) sample, a peak resolved at 11.5 min that was absent from the LMCap1 and LMCap1(pNF2176) samples (Fig. 6). When the area under the peak was calculated, the unknown product was shown to have increased from 3.7 to 6.9% (of the total detectable carbohydrates) from 2 to 10 min of incubation. In this assay, NAD⁺ and NADP⁺ were added to the reaction to investigate if cofactors were required for the dehydrogenation step catalyzed by NmaB. In the SH1217 assay, very little product had been synthesized using 0.55 μg of protein and 2 min of incubation, but it was detectable when the sample was incubated for 10 min (Fig. 6).

FIG. 6.

CE analysis of NmaA and NmaB activity from SH1217, LMCap1, LMCap1(pNF2176), and LMCap1(pLMCAP) crude protein samples. The reaction mixture contained 0.5 μg of total protein and 1 mM UDP-GlcNAc in a total reaction volume of 35 μl. The samples were incubated at 37°C for 2 and 10 min and electrophoresed. A UDP-GlcNAc sample is included as a control. The arrow indicates a carbohydrate species resolving at 11.5 min that is absent from the LMCap1 and LMCap1(pNF2176) samples but present in the SH1217 sample and the complemented strain, LMCap(pLMCAP). AU, arbitrary units.

Serum sensitivity assays.

To investigate if the M. haemolytica A1 CPS confers resistance to complement-mediated killing, SH1217 and LMCap1 were examined for growth inhibition by 66% colostrum-deprived calf serum. A serum-sensitive control, E. coli JN1, was also examined to confirm the presence of complement activity in the serum. SH1217 was not killed in normal calf serum, and the average number of CFU per milliliter increased from 4.1 × 10⁶ to 3.2 × 10⁸ over 3 h of incubation. A similar trend was observed for LMCap1, increasing from an initial average of 3.5 × 10⁶ to 9.6 × 10⁷ CFU/ml over 3 h in normal calf serum. E. coli JN1 was effectively killed by the serum; the average number of CFU dropped from 1.2 × 10⁷ to 9.0 × 10³ per ml over 3 h. These values are presented as percent viability in Fig. 7. HI serum permitted growth of SH1217, LMCap1, and JN1 (data not shown).

FIG. 7.

Serum sensitivity assay of M. haemolytica A1 SH1217 (▪), acapsular mutant LMCap1 (▴), and E. coli control JN1 (⧫) in 66% colostrum-deprived normal calf serum. The data points for the M. haemolytica strains represent the means of three independent assays. Percent viability was calculated by dividing the mean number of CFU per milliliter, as determined for each time point, by the original number of CFU per milliliter × 100. The error bars represent the standard errors of the means.

It was also of interest to determine whether the M. haemolytica A1 CPS contributes to serum resistance in the presence of antibody. To do this, serum was collected from a healthy adult cow that had a high neutralizing antibody titer against M. haemolytica A1 leukotoxin (log₂, 9) as well as a log₂ agglutination titer of 5. Again, E. coli JN1 was included as a positive control for complement-mediated killing. Effective killing of SH1217 was observed in 66% normal serum within the first hour of incubation, with the average viable cell count dropping from 3.6 × 10⁶ to 1.5 × 10⁶ CFU/ml. After the first hour the cells were no longer killed, and the number of CFU per milliliter increased. Killing of LMCap1 was also observed in 66% normal serum within the first hour of incubation, dropping from an average of 3.8 × 10⁶ to 1.8 × 10⁵ CFU/ml. These values are presented as percent viability in Fig. 8A. LMCap1 was consistently killed to a greater degree than SH1217 within the first hour in the four independent assays performed. The normal serum killed JN1 (Fig. 8A). SH1217 and LMCap1 cells incubated in 66% CPI serum were not killed (Fig. 8B). The SH1217 cell count increased from an average of 4.7 × 10⁶ to 6.8 × 10⁷ CFU/ml over 3 h. Similarly, LMCap1 grew from an average of 3.8 × 10⁶ to 1.9 × 10⁸ CFU/ml over 3 h (Fig. 8B). CPI serum killed JN1 cells. All strains were resistant to HI serum and multiplied to 10 to 100 times the initial cells counts over the 3-h incubation (data not shown).

FIG. 8.

Serum sensitivity assay of M. haemolytica A1 SH1217 (▪), acapsular mutant LMCap1 (▴), and E. coli control JN1 (⧫) in 66% adult bovine serum (A) or 66% CPI serum (B). All data points represent the means of four independent assays. Percent viability was calculated by dividing the mean number of CFU per milliliter, as determined for each time point, by the original number of CFU per milliliter × 100. The error bars represent the standard errors of the means.

LPS profiles.

LPS was prepared from M. haemolytica A1 strains SH1217 and LMCap1, as well as from E. coli JN1, and analyzed by silver staining. Long O-polysaccharide side chains were confirmed on these strains by comparing to the well-characterized LPS profile of P. aeruginosa laboratory strain PAO1 (data not shown).

DISCUSSION

It has been observed that antibiotic resistance markers using E. coli promoters are poorly or inconsistently expressed in M. haemolytica, adding to the difficulties in constructing defined mutants of this organism. To circumvent this, a chloramphenicol resistance cassette, plpcat, was constructed by utilizing a previously characterized promoter region of an M. haemolytica lipoprotein gene, plp4 (34), upstream of the promoterless cat gene. Gene plp4 was shown to code for an OmpA-like lipoprotein (34), and its expression was not expected to be under the control of any specific regulatory mechanism. Indeed, consistent expression of chloramphenicol resistance was observed from plpcat. Using this marker, the mutagenic plasmid pLMCAT, which carries two disrupted CPS biosynthesis genes, nmaA and nmaB, was created. Plasmid pLMCAT was electroporated into M. haemolytica A1 SH1217, and plpcat replaced nmaAB on the chromosome by allelic exchange, producing LMCap1. These events were confirmed by PCR and Southern hybridization. This is the first demonstration of an M. haemolytica A1 strain with inactivated CPS biosynthesis genes.

LMCap1 broth culture was noted to have a slightly higher growth rate than SH1217 (data not shown). This is probably because these cells have less metabolic requirements due to the disrupted CPS biosynthesis operon, freeing more energy for cellular growth. A similar observation has been made for an A. pleuropneumoniae serotype 1 strain deficient in CPS biosynthesis (47). Electron microscopy analysis of LMCap1 indicates that this bacterium is lacking CPS. In addition to the absence of a CPS coat stained with ferritin, the inner membrane and periplasm appear to be intact. There is no evidence of accumulation of CPS precursors, which suggests that the biosynthesis of CPS had been eliminated. Conversely, electron micrographs of E. coli mutants with defective CPS transporters show “electron-lucent” spaces where polymer accumulates in the cytoplasm (4, 37, 52).

When the nmaAB genes were introduced into LMCap1 on plasmid pLMCAP, surface-expressed CPS was restored, as observed by electron microscopy (Fig. 4D). This suggests that the inactivation of nmaAB does not result in a polar effect on the four genes downstream of nmaAB or that these downstream genes are not involved in CPS biosynthesis. To date, these four gene products have not yet been characterized. Nevertheless, the six genes that encompass region 2 of the CPS biosynthesis cluster are believed to be in an operon because each open reading frame either overlaps with or begins within 1 to 15 bp from the stop codon of the previous open reading frame (25).

A disrupted CPS biosynthesis pathway in LMCap1 was confirmed by biochemical analysis. Total protein prepared from SH1217 catalyzed an in vitro conversion of UDP-GlcNAc to an unknown carbohydrate species that was absent when the cellular proteins from LMCap1 were used for the same assay (Fig. 5). Introduction of plasmid pLMCAP into LMCap1 restored the biosynthesis of the unknown carbohydrate species, confirming that NmaA and/or NmaB catalyzed the reaction. In E. coli (20) and Salmonella enterica serovar Typhimurium (24) WecB catalyzes the epimerization of UDP-GlcNAc to UDP-ManNAc, and then WecC oxidizes UDP-ManNAc to UDP-ManNAcA. Because NmaA and NmaB have been demonstrated to be functional homologues of WecB and WecC, respectively, by complementation studies (25) it is hypothesized that the unknown carbohydrate could be UDP-ManNAc or UDP-ManNAcA. Standards of these compounds were not commercially available, so the identity of the carbohydrate absent from the LMCap1 protein assay could not be determined. Kiser et al. (22) demonstrated epimerization of UDP-GlcNAc to UDP-ManNAc by a WecB homologue, Cap5P, from Staphylococcus aureus serotype 5. It is interesting that, in their study, increasing the amount of purified Cap5P or UDP-GlcNAc did not improve the efficiency of conversion to UDP-ManNAc. Similarly, in this study, increasing the amount of protein used in the assay did not improve the quantity of product synthesized from UDP-GlcNAc (data not shown). NmaA, which has 47% similarity with Cap5P, may be regulated in a manner similar to that suggested for the epimerase Cap5P. Cofactors NAD⁺ and NADP⁺ were included in one assay to investigate if these could influence the dehydrogenation step catalyzed by NmaB. No significant changes to the production of unknown product were observed (Fig. 6). The peak that resolved at 12.6 min from both the SH1217 and LMCap1 samples (Fig. 5) was determined, by using a standard, to be UDP-GalNAc. One of the components of the M. haemolytica A1 LPS O side chain is N-acetylgalactosamine (43). It is hypothesized that UDP-GlcNAc was epimerized to UDP-GalNAc from an LPS biosynthesis enzyme that was among the total cellular proteins used in the assay.

LMCap1 was found to be as resistant to complement-mediated killing as was the encapsulated parent, SH1217, in calf serum. This suggests that, in the absence of specific antibody, the M. haemolytica A1 CPS does not mediate resistance to complement. However, CPS did protect SH1217 from being killed to the same extent as LMCap1 in immune serum (Fig. 8A). One of the proposed functions of CPS is that it masks underlying surface antigens (41). It is possible that antibodies specific for various M. haemolytica A1 surface antigens were unable to bind the corresponding ligand in the presence of the CPS. Hence, LMCap1 was killed by immune serum to a greater extent due to increased accessibility of surface antigens. Treatment of the bovine serum with a calcium chelator rendered the classical pathway of the complement system inactive, leaving only the alternative pathway. The inability of complement in CPI serum to kill SH1217 or LMCap1 (Fig. 8B) supports the conclusion that only antibody-activated complement can cause bacteriolysis of these organisms in serum. This finding correlates with an earlier study by Chae et al. (8) that showed that experimentally decapsulated M. haemolytica A1 was more susceptible to killing by sera from vaccinated calves. Recently, several other acapsular mutants have been constructed from organisms belonging to the family Pasteurellaceae. Similar to these findings, acapsular mutants of Actinobacillus pleuropneumoniae serotype 1 (40) and Pasteurella multocida serogroup B:2 (5) demonstrated a serum-resistant phenotype. However, acapsular mutant strains of A. pleuropneumoniae serotype 5a (47) and P. multocida serogroup A (9) were reportedly serum sensitive. This suggests that the role CPS plays in mediating serum resistance can vary and that additional factors contribute to a serum-resistant phenotype.

M. haemolytica A1 has been shown to possess long O-polysaccharide side chains (14, 23). Analysis of LPS preparations from LMCap1 demonstrated that elimination of the CPS biosynthesis pathway did not affect synthesis of LPS (data not shown). Similar to CPS, the long O-polysaccharide side chains of LPS can sterically hinder the access of complement components to the bacterial membrane (38). The resistance of LMCap1 cells to complement-mediated killing in nonimmune serum could be due to long LPS O side chains that provided serum resistance. In E. coli there are several examples where defined acapsular mutants of smooth strains remained serum resistant. For example, in strains O6:K5, O18:K5, O75:K5, and O9:K30 it was shown that the primary barrier against complement-mediated serum killing was the O antigen side chain of LPS molecules (7, 11, 28).

In summary, an acapsular mutant strain of M. haemolytica A1, LMCap1, has been constructed. It was shown that disruption of CPS biosynthesis genes nmaA and nmaB resulted in the loss of surface-expressed CPS in LMCap1. In immune serum, LMCap1 was killed to a greater degree than the encapsulated parent, indicating that the M. haemolytica A1 CPS is mediating resistance to serum bacteriolysis through the classical complement pathway. Conversely, the mutant was as resistant as the parent to killing in colostrum-deprived calf serum. Hence, more research is needed to investigate what other factors, such as smooth LPS, contribute to M. haemolytica A1 serum resistance.