Sialic Acid Metabolism and Systemic Pasteurellosis
http://www.100md.com
感染与免疫杂志 2005年第3期
Laboratory of Sialobiology, Department of Pathobiology, University of Illinois at Urbana-Champaign, Urbana, Illinois
Pfizer Animal Health, Kalamazoo, Michigan
ABSTRACT
Pasteurella multocida subsp. multocida is a commensal and opportunistic pathogen of food animals, wildlife, and pets and a zoonotic cause of human infection arising from contacts with these animals. Here, an investigation of multiple serotype A strains demonstrated the occurrence of membrane sialyltransferase. Although P. multocida lacks the genes for the two earliest steps in de novo sialic acid synthesis, adding sialic acid to the growth medium resulted in uptake, activation, and subsequent transfer of sialic acid to a membrane acceptor resembling lipooligosaccharide. Two candidate-activating enzymes with homology to Escherichia coli cytidine 5'-monophospho-N-acetylneuraminate synthetase were overproduced as histidine-tagged polypeptides. The synthetase encoded by pm0187 was at least 37 times more active than the pm1710 gene product, suggesting pm0187 encodes the primary sialic acid cytidylyltransferase in P. multocida. A sialate aldolase (pm1715) mutant unable to initiate dissimilation of internalized sialic acid was not attenuated in the CD-1 mouse model of systemic pasteurellosis, indicating that the nutritional function of sialate catabolism is not required for systemic disease. In contrast, the attenuation of a sialate uptake-deficient mutant supports the essential role in pathogenesis of a sialylation mechanism that is dependent on an environmental (host) supply of sialic acid. The combined results provide the first direct evidence of sialylation by a precursor scavenging mechanism in pasteurellae and of a potential tripartite ATP-independent periplasmic sialate transporter in any species.
INTRODUCTION
Pasteurella multocida subsp. multocida (hereafter referred to as P. multocida) is an economically important opportunistic pathogen of livestock (hemorrhagic septicemia, shipping fever, and atrophic rhinitis), poultry (fowl cholera), wildlife (avian cholera), and laboratory rabbits (snuffles) (1). As a commensal of cats and dogs, P. multocida also is a significant cause of zoonotic abscesses arising mainly from bites or scratches, but it has been increasingly associated with pulmonary disease, sepsis, and meningitis in patients with underlying medical conditions that may compromise their immune status (4, 10, 13, 15, 16, 20, 21, 29, 30, 35, 38, 48). Although the so-called dermonecrotic toxin synthesized by some P. multocida strains produces atrophic rhinitis, the toxin is not essential for respiratory or systemic disease. The P. multocida virulence factors responsible for life-threatening invasive infections are poorly defined despite complete genomic DNA sequencing (36), microarray gene-expression profiling (7, 8), in vivo expression technology (25), and signature tagged mutagenesis (19, 23). Known or suspected virulence factors include adhesins, systems for iron acquisition, lipooligosaccharide (LOS), capsules, and neuraminidases (1). However, with the exception of capsule (11), the function in pathogenesis or relative importance of most other virulence factors is either unknown or uncertain. Evidence is provided here that the metabolism of sialic acids should be included as a potentially essential disease factor in systemic pasteurellosis.
We are interested in the metabolism of the sialic acids, a structurally diverse group of nine-carbon keto sugars (3, 9), and have hypothesized that the interplay between its metabolism and the host-microbe interaction is a unifying theme for understanding diseases caused by a wide range of invasive pathogens or commensals (51-53). For example, Escherichia coli K1 and certain meningococci synthesize sialic acid by a de novo pathway for assembly of sialic acid into outer membrane capsules or lipooligosaccharides (LOS) that are known to inhibit host innate immunity (44). Other pathogens, such as Neisseria gonorrhoeae and Haemophilus spp., do not synthesize sialic acid but scavenge host-derived (environmental) sialic acids for cell surface decoration (modification) involving endogenous sialyltransferases. All known sialyltransferases use the activated nucleotide sugar cytidine 5'-monophospho (CMP)-sialic acid as the obligate donor substrate for transfer of sialyl units to appropriate acceptor substrate molecules. In addition to de novo synthesis or scavenging of sialic acid, some pathogens may also catabolize environmental sialic acid for nutrition (carbon, nitrogen, and energy) or as a source of amino sugars for cell wall biosynthesis (40). How organisms that both decorate their surfaces with sialic acid and use it for nutrition regulate the metabolic decision between surface modification and degradation has not been investigated until recently (27, 41, 42, 54, 55).
Here, for the first time, sialyltransferase is demonstrated in P. multocida, and the role of sialometabolism in the host-pathogen interaction is investigated in an animal model of systemic pasteurellosis. In addition, the first evidence is provided for a novel sialate uptake system involving a putative tripartite ATP-independent periplasmic (TRAP) transporter of the type that functions in carboxylic acid transport in a wide range of bacterial species (18, 28). The combined results indicate that sialometabolism may be a common feature of the entire genus and begin to delineate the importance of sialylation to systemic pasteurellosis.
(A preliminary account of a portion of this study was presented by E. Vimr and C. Lichtensteiger at the 101st General Meeting of the American Society for Microbiology in Orlando, Fla., 2001 [abstr. B-446].)
MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions. Pm70 is a spontaneous capsule-negative, nalidixic acid-resistant turkey isolate kindly provided by V. Kapur (36). TF5 is a nalidixic acid-resistant type A bovine isolate with defined systemic virulence in the CD-1 mouse model (19). TF5E (PM1B-E11) and TF5P (PM3-C9) are TF5-derived strains bearing kanamycin-resistant disruptions of pm1711 and pm1709, respectively, isolated from CD-1 mice after signature-tagged mutagenesis of TF5 (19). The transcriptional orientations of the kanamycin cassettes in these mutants are given in Fig. 1. X-73 is a type A avian isolate kindly provided by M. Wolcott (National Wildlife Health Center, Madison, Wis.). Cincy is a type A strain isolated from a clinical case of human pneumonia involving pulmonary contusions; it was given to us by J. Solomkin (University of Cincinnati College of Medicine, Cincinnati, Ohio) for toxA typing by using our previously described method for PCR detection of toxigenic P. multocida (33). The results of this analysis indicated that the strain was nontoxigenic. Upon receipt, Cincy was passaged once before cryopreservation. The E. coli BL21(DE3) Star strain for histidine-tagged polypeptide overproduction was purchased from Invitrogen (Carlsbad, Calif.). EV5 is a neuA mutant lacking the CMP-sialic acid synthetase (activating enzyme) for polysialic acid capsule synthesis (54). S17-1 is a pir+ strain that supports replication of plasmids bearing pir-dependent replicons; it was used as the donor for mutagenesis of P. multocida as previously described (37).
Plasmid pGPKan is a kanamycin-resistant derivative of the suicide vector pGP704, kindly provided by M. Lee (University of Georgia, Athens). Plasmids pSX1000 and pSX1001 are ampicillin-resistant derivatives of pCR T7/CT TOPO TA (Invitrogen) overexpressing pm1710 or pm0187, respectively.
P. multocida strains were routinely propagated in brain heart infusion (BHI) or Haemophilus test medium (HTM) as previously described (53). Unless indicated otherwise, supplemented HTM contained 1 mg of Neu5Ac/ml, added from a 20-mg/ml filter-sterilized stock solution directly to HTM. E. coli strains were propagated in the Lennox formulation of lysogeny broth (5). When necessary, kanamycin, ampicillin, and nalidixic acid were included in media at 20, 100, and 20 μg/ml, respectively. Bacteria were propagated at 37°C with aeration by rotary mixing.
Mutant construction. Aldolase (pm1715-deficient) P. multocida mutants Pm70A and TF5A were constructed by first amplifying an 500-bp internal fragment of pm1715 using the gene-specific forward (5'-GGATGGTCTTTACGTTGGCGGCAGTAC-3') and reverse (5'-GATTTGACGAGCGCGCACACCATTGAC-3') oligonucleotide primers. The resulting amplicon was cloned into the pGEM T-Easy PCR cloning vector (Promega, Madison, Wis.), followed by excision of the fragment with EcoRI and cloning into the suicide vector pGPKan in E. coli strain S17-1. The plasmid was mated into Pm70 or TF5 with selection for kanamycin resistance and nalidixic acid counterselection (37). The gene disruptions resulting from plasmid cointegration were confirmed by Southern hybridization analysis with the internal gene fragment from pm1715 as the probe. Mutants were propagated in vitro in the presence of kanamycin to select for cointegrate maintenance.
CMP-sialic acid synthetase overproduction. The genes pm1710 and pm0187 encoding putative activating enzymes were cloned into pCR T7/CT TOPO TA cloning kit (version B) after PCR amplification of the complete open reading frames (ORFs) with N- and C-terminal primers to generate pSX1000 and pSX1001, respectively, by previously described methods (41). All PCR primers were synthesized by IDT (Coralville, Iowa). Polypeptides were overproduced by IPTG (isopropyl--D-thiogalactopyranoside) induction as C-terminal His6 fusions (42).
Biochemical assays. N-Acetylneuraminic acid (Neu5Ac; the most common sialic acid) was measured by colorimetric thiobarbituric acid assay (53). Sialyltransferase was detected by incubating membrane samples derived by sonic disruption in 20 mM Tris (pH 8.0), 5 mM magnesium acetate, and 2 mM dithiothreitol (TMD) buffer with CMP-[4-14C]Neu5Ac (50 mCi/mmol), followed by descending paper chromatography in solvent system I (7:3, ethanol-ammonium acetate [pH 7.5]) and liquid scintillation spectrometry of radioactivity remaining at the origin essentially as described previously (54, 55). Quantitative estimation of sialyltransferase activity is described in detail below. Protein was estimated by dye binding (Pierce Chemical Co., Chicago, Ill.) with bovine serum albumin as standard.
Sialidase was measured as relative fluorescence units with the fluorogenic substrate 2'-(4-methylumbelliferyl)-Neu5Ac (MuNeu5Ac) as previously described (50). Sialic acid uptake was measured by depletion assay using unlabeled or [4-14C]Neu5Ac (55 mCi/mmol) as previously described (53, 54). CMP-sialic acid synthetase was detected autoradiographically after incubation of soluble cell extracts with 10 mM CTP and 18 μM radiolabeled Neu5Ac, followed by chromatography in solvent system I or by thin-layer chromatography as previously described (45). Quantitative estimation of synthetase activity was accomplished by excising regions of the chromatograms containing CMP-Neu5Ac product, followed by liquid scintillation spectrometry. Sialate aldolase was detected autoradiographically after incubation of extracts with radiolabeled Neu5Ac and chromatography in solvent system II (n-propanol-1 M sodium acetate [pH 5.0]-water, 7:1:2), with [1-14C]ManNAc (50 mCi/mmol) as the standard (41). All radiochemicals were purchased from American Radiochemical Company (St. Louis, Mo.). Type V sialidase from Clostridium perfringens and all other chemicals were purchased from Sigma Chemical Company (St. Louis, Mo.).
Sialyltransferase assay. Quantitative comparisons of sialyltransferase activities in mammalian and avian P. multocida were carried out by a modification of the procedure described above. Single colonies from the indicated strains were inoculated into 25 ml of BHI and grown to stationary phase. Cultures of encapsulated strains were treated for 1 h at room temperature with 100 U of hyaluronidase to facilitate cell collection by centrifugation. All succeeding preparative steps were carried out at 0 to 4°C. Cell pellets were resuspended in 0.5 ml of TMD buffer and disrupted by sonication (Branson Cell Disrupter 185 with microtip): four cycles of 20 s each on ice at an output setting of 4 with 45 s of cooling between each cycle. Intact cells were removed by centrifugation at 4,000 x g for 4 min. Supernatants were diluted with TMD to a final volume of 3 ml, and membranes were collected by ultracentrifugation in a Beckman TL-1000 desktop ultracentrifuge with TLA100.3 rotor at 52,000 rpm for 15 min. The membrane pellets were resuspended in 1.5 ml of TMD and sonicated for two cycles as described above, followed by the removal of large fragments or any remaining unbroken cells by centrifugation as described above. Membranes were recollected by ultracentrifugation and finally resuspended in 0.05 ml of TMD for storage on ice prior to use. Assays were carried out in triplicate by incubating ca. 150 μg of membrane protein with radiolabeled CMP-Neu5Ac, followed by chromatography and liquid scintillation spectrometry as described above. The data were expressed a picomoles of Neu5Ac transferred to endogenous acceptor(s) ± the standard deviation per mg of protein in 15 min. Differences in specific activities were determined by analysis of variance (P < 0.0001), followed by comparisons of means by the Tukey method ( = 0.05).
Animal experiments. Outbred (5- to 7-week-old) CD-1 female mice were purchased from Charles River Laboratories (Wilmington, Mass.) and housed at the University of Illinois College of Veterinary Medicine animal care facility. Mice were injected intraperitoneally with ca. 700 CFU in 0.1 ml of phosphate-buffered saline containing equal numbers of wild-type (kanamycin-sensitive) and mutant (kanamycin-resistant) P. multocida. Actual input doses were determined by enumerating CFU after dilution plating and overnight incubation. Infected mice were euthanized 17 h postinoculation, and systemic (blood) CFU were differentially enumerated on selective medium. The data are expressed as the ratio of mutant to wild-type bacteria, or the competitive index (CI), which was normalized to the input doses as previously described (53). Control experiments in which mice were inoculated with just TF5A confirmed the stability of the mutation in this strain during the course of infection, indicating <0.1% loss of drug resistance. All animal experiments were approved by the University of Illinois animal care office after institutional review according to National Institutes of Health guidelines. CI differing from 1.0 was analyzed by a one sample, two-way t test using log transformation.
In vivo detection of sialyl acceptor. Wild-type P. multocida strain Pm70 or the Pm70A mutant were grown in 2 ml of BHI and exposed at an A600 of 0.4 for 1 h to 5 μg of Neu5Ac/ml containing a tracer amount of radiolabeled Neu5Ac. Bacteria were collected by centrifugation; residual liquid was removed by carefully wicking the tube walls around the pellets. Wild-type and mutant bacteria removed 40 and 70%, respectively, of the label from the culture medium during the 1-h incubation period. Bacterial pellets were resuspended in 9 μl of 10 mM Tris (pH 8.0) containing 5 μg of lysozyme/ml, followed by the addition of 1 μl of 0.5 M EDTA. Cells were ruptured by three freeze-thaw cycles, and 0.6 μl of 1 M MgCl2 was added prior to treatment with DNase I as previously described (49). Membrane and soluble samples were prepared by centrifugation. Half of each soluble fraction was analyzed by descending paper chromatography in solvent system II by autoradiography, as previously described (42). The membrane fraction was resuspended in 9 μl of water and fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in a 4 to 20% gradient gel. The gel was stained with Coomassie blue, fluoroenhanced, and subjected to autoradiography for 3 days prior to film development. Tricine gels (16.5%) were purchased from Bio-Rad.
RESULTS
Predicted functions of P. multocida strain Pm70 ORFs with presumed roles in sialometabolism. When a BLAST (2) comparison of known E. coli K1 sialometabolic gene products and the P. multocida strain Pm70 genomic DNA sequence (36) was carried out, orthologs of NeuB (sialate synthase) and NeuC (UDP-GlcNAc 2-epimerase) were not detected (Table 1). The absence of neuB- and neuC-like genes indicated that P. multocida lacks the two earliest steps in the pathway for de novo sialic acid synthesis (53). In contrast, P. multocida potentially encodes two CMP-Neu5Ac synthetases (pm0187 and pm1710), an ortholog of the 2,6-sialyltransferase (encoded by pm0188) from the marine organism Photobacterium damsela (57), and a potential 2,3-sialyltransferase that is homologous to one of the Haemophilus influenzae LOS sialyltransferases (8). The absence of genes for de novo sialate synthesis but the presence of ORFs potentially coding for activating enzyme(s) and sialyltransferases suggested that P. multocida may use a sialylation pathway similar to that described previously for H. influenzae (53). This precursor scavenging pathway involves the uptake of environmental sialic acid, followed by its activation by CMP-sialic acid synthetase (NeuA) and subsequent transfer of sialic acid to an appropriate membrane acceptor by sialyltransferase (Fig. 2). As shown in Table 1, P. multocida also potentially encodes a complete pathway for sialic acid catabolism, including two membrane sialidases with distinct substrate specificities for hydrolyzing host sialoglycoconjugates (37). The combined in silico analysis of Pm70 ORFs suggests that, with the exception of de novo Neu5Ac biosynthesis, P. multocida is genetically equipped to carry out a full range of sialometabolic functions including uptake, and either activation of internalized sialic acid for cell surface modification or cleavage to N-acetylmannosamine (ManNAc) and pyruvate for nutrition or source of amino sugars for cell wall biosynthesis (Fig. 2). Since P. multocida is an obligate commensal of mammals, birds, and reptiles, including dragons (38), all of which synthesize sialic acid, sialometabolism is likely to be a central aspect of this bacterium's unusually broad range of host interactions, including the potential to cause disease.
Detection of sialyltransferase. If P. multocida uses a sialylation mechanism similar to that of H. influenzae (53), the isolated membrane fraction should transfer sialic acid from exogenous CMP-Neu5Ac substrate to endogenous membrane acceptor(s). To test for this activity, we incubated the membrane fractions from sonically disrupted bacteria with radioactive CMP-Neu5Ac and measured the incorporation of labeled sialic acid into membranes by descending paper chromatography in solvent system I. Table 2 indicates that all of the P. multocida strains tested (including two type D strains, TF5E, and X-73 [results not shown]) had detectable transferase activity. The highest specific activity was consistently observed with mammalian isolates, a finding comparable in magnitude to the activities of E. coli and meningococcal polysialyltransferases (46). A control experiment with radiolabeled Neu5Ac instead of CMP-Neu5Ac as the substrate did not result in detectable sialyltransferase activity, which is consistent with the known substrate dependency of sialyltransferases for activated sialic acid.
To determine whether the variability in sialyltransferase specific activities resulted from endogenous sialidases (Fig. 2), hydrolysis of the fluorogenic sialoside MuNeu5Ac was measured at basic and acidic pH. As expected from the known acidic pH optima of P. multocida sialidases (37), activity against the sialoside was greater at pH 6.8 than at pH 8.0 (Table 2). However, there were no appreciable differences between sialidase activities of mammalian or the avian isolates tested. The greater activity of the aldolase-negative derivative of Pm70 (Pm70A) may reflect induction by trace sialic acid in the medium, as previously observed for the inducible nan system in E. coli (41, 42, 49).
To directly investigate the susceptibility of the sialylated acceptors to endogenous sialidases, radiolabeled Pm70 membranes were diluted 50-fold into assay buffer, pelleted by centrifugation, and resuspended to the same volume as for the initial radiolabeling. Samples assayed over a 3-h period showed no loss of radioactivity, indicating no effect of endogenous sialidase on sialic acid release. In contrast, 0.5 U of exogenous C. perfringens sialidase released over half of the radioactivity in 1 h, indicating sialyl residues were added by sialyltransferase in their expected terminal -glycoketosidic linkages to endogenous acceptor(s). Furthermore, proteinase K digestion did not release label, and Pm70 lacks the type A hyaluronate capsule, strongly suggesting that the endogenous acceptor is neither protein nor capsular polysaccharide. We concluded that sialyltransferase is a consistent phenotype of at least P. multocida subsp. multocida, which is responsible for most animal disease (1).
Identification of CMP-Neu5Ac synthetase. The absence of orthologs of neuB and neuC but the presence of sialyltransferase implies P. multocida scavenges environmental sialic acid and activates it intracellularly using one or both of the putative CMP-Neu5Ac synthetase candidates indicated in Table 1. One of these candidate synthetases (encoded by pm0187) is located adjacent to the putative 2,6-sialyltransferase encoded by pm0188, whereas the other (pm1710) is predicted to be the first gene of the P. multocida nan-like operon (Fig. 1). PCR analysis of 32 P. multocida type A or D isolates from our swine collection (33) resulted in the successful amplification of the expected DNA products from pm1710 or pm0187 in all strains tested (data not shown). The ubiquity of these candidate neuA orthologs in P. multocida is consistent with the sialyltransferase-positive phenotype documented in Table 2.
Campylobacter spp. and Legionella pneumophila have been shown to express multiple orthologs of neuA or neuB (32, 34), suggesting that both pm0187 and pm1710 could encode functional sialate activating enzymes despite the greater similarity of the pm1710 gene product to phospholipase than cytidylyltransferase (Table 2). To determine whether either of these ORFs codes for an active CMP-Neu5Ac synthetase, the respective genes were cloned into a vector designed to express gene products in frame with C-terminal His6 tags. As shown in Fig. 3A, pm0187 complemented an E. coli K1 neuA mutant for polysialic acid biosynthesis, whereas pm1710 did not (Fig. 3B), suggesting that only pm0187 encodes a functional synthetase. Similarly, while an extract of wild-type Pm70 had both detectable Neu5Ac aldolase and CMP-Neu5Ac synthetase activity (Fig. 4A, lane 3), only the His6-tagged pm0187 product produced CMP-Neu5Ac from CTP and radiolabeled Neu5Ac (compare lanes 2 to 4 to lanes 5 to 7 in Fig. 4B). The results of complementation and direct biochemical analyses indicate that the pm0187 is a functional NeuA ortholog.
To further investigate the potential cytidylyltransferase activity of pm1710, we compared its relative activity to that of the pm0187-encoded NeuA ortholog. As shown in Fig. 5, NeuA had, as expected, high synthetase activity that saturated between 56 and 113 μg of protein. In contrast, no activity of the extract containing the overproduced pm1710 gene product was detected until 159 μg of protein was tested, resulting in at least a 37-fold-lower specific activity than NeuA. These results suggest that if pm1710 encodes a cytidylyltransferase, its substrate is unlikely to be Neu5Ac. Further work is necessary to determine the function of the pm1710 gene product, but the present results indicate that NeuA encoded by pm0187 is the primary if not the sole Neu5Ac synthetase in P. multocida. The results in Table 1 suggest that pm1710 may encode a phospholipase or esterase, similar to the C-terminal domain of E. coli NeuA (31).
Scavenging environmental sialic acid. Although P. multocida lacks candidate genes for the two earliest steps in sialic acid synthesis (Table 1), the results described above demonstrate that it has the genetic information for Neu5Ac activation and sialyl transfer. In addition, an extract of wild-type P. multocida has a sialate aldolase activity that could initiate the catabolism of environmental sialic acid, as indicated by the product migrating with the same relative migration (Rf) value (0.83) as authentic ManNAc (Fig. 4A, lane 3). Therefore, if P. multocida uses a precursor scavenging mechanism for cell surface sialylation, there must be an uptake system for sialic acid that provides the substrate for both intracellular catabolic and activating enzymes (Fig. 2). The results presented in Table 1 suggested that pm0835 could encode an ortholog of NanT, whereas pm1708, together with pm1709, was a likely candidate for a TRAP sialate transporter.
Sialate uptake was demonstrated directly by a chromogenic depletion assay (53), indicating that wild-type strain TF5 quantitatively removed Neu5Ac during growth in supplemented HTM (Table 3). In contrast, growth of TF5P in supplemented HTM did not result in Neu5Ac depletion (Table 3), strongly suggesting that the polypeptide encoded by pm1709 is required for sialate uptake. This conclusion was confirmed by uptake of radiolabeled Neu5Ac during growth of the wild type but not TF5P in HTM containing 10 μM unlabeled and a tracer amount of labeled Neu5Ac. Wild-type TF5 took up nearly 50% of the label by 1 h, whereas the input counts per minute (cpm) remaining in the culture medium of TF5P remained constant when sampled at 1-, 2-, or 3-h intervals (99.2% ± 1.3% [standard deviation]). Interestingly, with wild-type bacteria the number of extracellular cpm actually increased after the initial 1-h labeling period, resulting in ca. 20% more radioactivity detected in the medium by 3 h compared to that present after the first hour. We interpret this increase in cpm as the excretion of fermentative end products, because Neu5Ac is clearly metabolized after uptake as shown by the nearly 70% growth stimulation of TF5 relative to TF5P (Table 3). Note that the uptake results imply P. multocida uses only one system for sialic acid transport, at least under the conditions of laboratory growth used for these experiments, suggesting that the putative NanT ortholog encoded by pm0835 is not involved in sialate uptake (Fig. 2).
Isolation and characterization of P. multocida sialate aldolase mutants. To confirm the presence of a functional sialate dissimilatory system in P. multocida, we constructed a gene disruption of the strain Pm70 nanA ortholog (pm1715) by homologous recombination with a pGPKan derivative harboring an internal (500 bp) fragment of the gene; the resulting mutant was designated Pm70A. ORF pm1715 is predicted to be the last gene of the P. multocida nan-like dissimilatory operon (Fig. 1C); thus, the gene disruption would have no effect other than loss of the proposed aldolase activity encoded by pm1715. An identical gene disruption was constructed in the bovine P. multocida isolate TF5 (18), which was designated TF5A. The loss of aldolase activity was detected by direct biochemical assay with radiolabeled Neu5Ac (Fig. 6, lane 6). Extracts prepared from either Pm70A or TF5A lacked detectable sialate aldolase activity (Fig. 6, lanes 1 and 5, respectively), whereas extracts from either wild-type parent (Fig. 6, lanes 2 and 3), as expected, produced a radiolabeled product with the same mobility as authentic ManNAc (Fig. 4, lane 7). An extract of TF5E bearing a nonpolar kanamycin resistance cassette (14) transcribed in the same direction as pm1711 expressed wild-type aldolase activity (Fig. 4, lane 4). Although we did not directly confirm that pm1711 encodes ManNAc-6-P epimerase, its homology to E. coli nanE (Table 1) and its proximity to sialate uptake and aldolase genes (Fig. 1) strongly supports the proposed epimerase function (Fig. 2). This assignment is consistent with detection of nanK and yhcH orthologs in P. multocida (Table 1) which, along with nanA and nanE, comprise the E. coli nan operon (Fig. 1A).
Endogenous sialyl acceptor. To identify the in vivo acceptor recognized by P. multocida sialyltransferase(s), we expected that a growing culture of Pm70A would efficiently transport and activate exogenous radiolabeled Neu5Ac, because there would be no competition for label from the catabolic system, as originally observed with E. coli nanA mutants (55). However, most of the label accumulated by Pm70A remained as free Neu5Ac (Fig. 7, lane 1), making it difficult to identify the endogenous membrane acceptor (Fig. 8A and B, lanes 5 and 7). In contrast, wild-type strain Pm70 produced a readily detectable CMP-Neu5Ac signal (Fig. 7, lane 2) that was correlated with a membrane-associated product with electrophoretic mobility near the dye front and apparent mass of <14 kDa (Fig. 8A and B, lanes 4 and 6). The radiolabeled product remaining in the wells is likely to be peptidoglycan, as suggested by past results with E. coli in which ManNAc dissimilation resulted in rapid labeling of the insoluble cell wall fraction (55). This suggestion is consistent with the low detectable amount of ManNAc in wild type (Fig. 7, lane 2) and with the absence of any detectable peptidoglycan in Pm70A (Fig. 8B, lanes 5 and 7), which cannot produce the ManNAc precursor. Note that, as expected, all of the soluble radiolabel that was in low-molecular-weight products passed through the gel in front of the dye marker, exiting into the anode buffer (Fig. 7B, lanes 2 and 3). When membranes that were used for the sialyltransferases assays presented in Table 2 were solubilized and fractionated by Tricine SDS-PAGE, only the broad band migrating at <14 kDa was detected after autoradiography (data not shown), a result which is consistent with the in vivo analysis shown in Fig. 8. We conclude that the sole macromolecular sialyl acceptor is a relatively low molecular weight product resembling LOS in its electrophoretic behavior. This conclusion is consistent with the biochemical results presented above, and previous results in the H. influenzae system (see reference 53 and references therein). Both the products of the in vivo and in vitro sialylation reactions will now be open to structural characterization.
Requirement for sialometabolism during systemic pasteurellosis. P. multocida strain Pm70 lacks the type A hyaluronate capsule that is essential for systemic virulence. Therefore, all animal experiments were carried out with encapsulated TF5 or its mutant derivatives. As summarized in Table 4, strain TF5 was previously found to be highly virulent in the CD-1 mouse model of systemic pasteurellosis (50% lethal dose [LD50] of <10 CFU), whereas TF5E and TF5P had >10,000 times higher LD50 values in this model (19). Because the data presented above strongly suggest that the pm1709 gene product is required for sialate transport, we concluded that sialometabolism is likely to be a necessary feature of systemic pasteurellosis. To correlate the CI with LD50 as an alternative measurement of virulence, we determined the CI for TF5E, indicating that a 500-fold decrease in relative fitness corresponds to a >10,000-fold increase in LD50 (Table 4). Because both the pm1711 and the pm1709 gene products are required for sialate catabolism, but only that of pm1709 is required for LOS sialylation, we sought to distinguish between the possible nutritional requirement for Neu5Ac in vivo and its alternative use for surface modification.
To distinguish between these possibilities, we carried out mixed infections with wild type and the TF5A aldolase-deficient mutant. As shown in Table 4, despite the relatively weak activation of Neu5Ac by Pm70A grown in HTM (Fig. 7), TF5A was as fit as the wild type for systemic propagation in the CD-1 mouse infection model. This result formally excludes a primary nutritional function of sialic acid catabolism in vivo, strengthening our hypothesis, as we previously described for the related H. influenzae system (53), that LOS sialylation is dependent on an external (environmental) source of sialic acid. Attenuation of TF5P is consistent with this hypothesis, since failure to transport exogenous sialic acid would result in the expression of unsialylated LOS. Independent analysis of H. influenzae sialylation has shown that very little sialic acid must be incorporated in vivo to provide a fitness advantage during otitis media (6). We suggest the same is likely to be true for P. multocida during systemic infection.
Although our combined results point to an essential role of sialylation in vivo, the attenuation of TF5E is difficult to reconcile with this hypothesis and suggests that, whereas Neu5Ac catabolism per se is not required for pathogenesis, as shown by the relatively wild-type fitness of the NanA mutant (Table 4), the ManNAc 6-phosphate expected to accumulate in a nanE mutant (41, 42) may be toxic during in vivo infection. Alternatively, a second mutation in TF5E, unrelated to the defect in pm1711, may account for the observed attenuation of this mutant. The single-infection control experiment described in Materials and Methods rules out instability of the pm1715 mutation as an explanation. Further biochemical analysis of LOS structure and genetic complementation will be necessary to determine the exact function of LOS sialylation in pasteurellosis and to understand the possible function of the pm1711 gene product in pathogenesis.
DISCUSSION
Since the discovery that H. influenzae sialylates its LOS by a precursor scavenging mechanism (52, 53), analogous pathways have been described in H. ducreyi (43) and H. somnus (26), suggesting that all haemophili may depend on an environmental source of sialic acid for LOS modification. Thus, bacteria that lack orthologs of neuB and neuC, which are needed for the two earliest steps in sialic acid biosynthesis (Fig. 2) (41, 42), but contain one or more copies of neuA and at least one sialyltransferase gene are likely to use precursor scavenging as a mechanism for surface sialylation. H. influenzae was subsequently shown to sialylate its LOS during infection in the chinchilla model of otitis media, and it was also shown that this surface modification was crucial to virulence (6). Although the exact function of sialylation during animal infection remains unclear, two recent studies have shown that sialic acid is required for normal H. influenzae biofilm formation, which may be a necessary process for colonization of mucosal surfaces (22, 47). In contrast, we have provided evidence that P. multocida sialylation, or at least sialate transport, may be necessary for systemic pasteurellosis, probably by protecting the sialylated bacteria from innate host defense mechanisms because mice do not normally carry P. multocida and would, therefore, not be expected to mount an antibody-dependent immune response under our conditions of experimental infection (1).
Perhaps in addition to a role of sialylation during systemic disease, the ability of obligate commensals and symbionts to modify their repertoire of surface carbohydrates arose as a mechanism for avoiding detection and clearance by the host's immune system. This idea would imply that sialylation may be required for persistence and that, if so, a therapeutic approach blocking surface modification could have broad clinical value by causing certain microbes to become sensitive to host clearance mechanisms. Evidence that surface carbohydrate modification plays a role in symbiosis comes from Bacteroides spp., in which the synthesis of eight structurally distinct capsular polysaccharides varies randomly in response to a site-specific recombinase for reversible inversion of the different capsule biosynthetic operon promoters (12). In contrast, our results point to the variation at the level of LOS sialylation being dependent on a supply of host sialic acid.
An elevated serum sialoglycoconjugate concentration has been shown to be a general marker of inflammation, suggesting that endogenous or microbial sialidases could increase the extracellular free sialic acid concentration in response to any ongoing inflammatory process (51). Thus, by increasing their cell surface sialylation in response to elevated environmental sialic acid, some microbes may be protected under conditions that normally activate innate defense mechanisms that would ordinarily lead to detection and clearance of unsialylated species (51). The P. multocida lifestyle appears to be unusually committed to sialometabolism, including two sialidases, two potential sialyltransferases, at least one activating enzyme, and a complete system for sialic acid dissimilation, including sialic acid transport (Fig. 2). The present results suggest that scavenging sialic acid is essential for at least systemic propagation. This robust sialometabolic system might also account in part for the wide P. multocida host range that includes diverse mammalian, avian, and reptilian species.
A functional precursor scavenging system minimally requires mechanisms for sialic acid uptake, activation, and some way to regulate the metabolic decision between sialate catabolism and surface modification. To date, all known sialyl acceptors in microbes that have been found to use a precursor scavenging mechanism are lipopolysaccharides in the outer membranes of gram-negative nasopharyngeal commensals. In the case of P. multocida, we have identified an activating enzyme (encoded by pm0187) and shown that a relatively low molecular weight membrane component resembling LOS is the endogenous acceptor. We have presented evidence here that at least the first gene (pm1709) of the potential operon defined by pm1007-1009 is required for sialate uptake. This uptake system is homologous to the TRAP transporters for carboxylic acids in diverse bacterial species, including H. influenzae (Fig. 1B). In contrast, the E. coli sialate transporter encoded by nanT (Fig. 1A) is a member of the major facilitator superfamily (MFS), indicating the existence of at least two distinct mechanisms of bacterial sialate uptake. Our results strongly suggest that P. multocida uses only the TRAP system for sialate uptake.
All known TRAP transporters include three linked genetic components, two of which encode integral membrane proteins that may be fused in some systems (e.g., HI0147 and pm1708 shown in Fig. 1B and C, respectively). The larger component codes for a predicted polypeptide with 12 membrane-spanning domains, resembling the secondary structures of most MFS permeases that couple solute uptake to one of the ion gradients across the cell membrane (28). The other membrane component is predicted to contain four membrane-spanning regions and is assumed to function in concert with the MFS-like permease (18, 28). The final components of all TRAP transporters are members of a family of periplasmic binding proteins that resemble the extracytoplasmic solute receptors of ATP-binding cassette-like uptake systems (28). Ligand binding to these extracytoplasmic solute receptors is thought to account for the high uptake affinities of multicomponent transporters (28). In P. multocida, we have provided evidence that the potential periplasmic component encoded by pm1709 is required for sialic acid uptake. Thus, P. multocida, H. influenzae, Fusobacterium nucleatum, Vibrio cholerae, and an unknown number of other pathogens or commensals are predicted to use TRAP transporters for sialic acid uptake (51). We have purified the pm1709 gene product as a His6 C-terminal fusion and shown that it binds Neu5Ac in the low micromolar range, confirming its predicted biochemical function as a sialate-binding protein (E. Vimr and S. Steenbergen, unpublished results).
TRAP transporters recognize a diverse range of mono- and dicarboxylic acids (28). Similarly, the sialic acids may be considered to be monocarboxylates, suggesting that the negative charge at carbon 1 may be necessary for recognition by the binding protein encoded by pm1709, which we propose to rename nanP (Table 1). Regardless of the exact binding mechanism, it should be possible to identify low-molecular-weight compounds that compete for sialic acid binding and thus block uptake. As in the case of the first influenza virus sialidase inhibitor (56), a compound with sufficient binding affinity and favorable on-off rates should prevent precursor scavenging, thereby blocking subsequent surface modification. Our results not only imply that such an inhibitor could be useful for treating systemic infection but also suggest potential value for preventing colonization or inhibiting the persistence of diverse bacterial species. Work directed toward rational design of sialate TRAP transporter inhibitors is in progress.
ACKNOWLEDGMENTS
We are grateful to J. Solomkin, V. Kapur, and M. Wolcott for providing bacterial strains. We also thank M. Lee for providing the suicide vector and for advice on mutant construction.
This study was supported in part by the Governor's Venture Technology Individual Grant Program and National Institutes of Health grant 2R01 AI42015 (E.V.).
Present address: Department of Molecular Biology and Microbiology, Tufts University, Boston, MA 02115.
Present address: DNA Indexing Laboratory, Springfield, IL 62707.
REFERENCES
1. Adlam, C., and J. M. Rutter. 1989. Pasteurella and pasteurellosis. Academic Press, Ltd., London, England.
2. Altschul, S. F., W. Gish, W. Miller, E. W. Myers, and D. J. Lipman. 1990. Basic local alignment search tool. J. Mol. Biol. 215:403-410.
3. Angata, T., and A. Varki. 2002. Chemical diversity in the sialic acids and related alpha-keto acids: an evolutionary perspective. Chem. Rev. 102:439-469.
4. Berge, A., A. Fagergren, and S. H. Stierenstedt. 2002. Pasteurella multocida septicemia in 2 Swedish patients. Scand. J. Infect. Dis. 34:138-139.
5. Bertani, G. 2004. Lysogeny at mid-twentieth century: P1, P2, and other experimental systems. J. Bacteriol. 186:595-600.
6. Bouchet, V., D. W. Hood, J. Li, J. R. Brisson, G. A. Randle, A. Martin, Z. Li, R. Goldstein, E. K. Schweda, S. I. Pelton, J. C. Richards, and E. R. Moxon. 2003. Host-derived sialic acid is incorporated in Haemophilus influenzae lipopolysaccharide and is a major virulence factor in experimental otitis media. Proc. Natl. Acad. Sci. USA 100:8898-8903.
7. Boyce, J. D., I. Wilkie, M. Harper, M. L. Paustian, V. Kapur, and B. Adler. 2002. Genomic scale analysis of Pasteurella multocida gene expression during growth within the natural chicken host. Infect. Immun. 70:6871-6879.
8. Boyce, J. D., I. Wilkie, M. Harper, M. L. Paustian, V. Kapur, and B. Adler. 2004. Genomic-scale analysis of Pasteurella multocida gene expression during growth within liver tissue of chickens with fowl cholera. Microbes Infect. 6:290-298.
9. Bravo, I. G., S. Garcia-Vallve, A. Romeu, and A. Reglero. 2004. Prokaryotic origin of cytidylyltransferases and alpha-ketoacid synthases. Trends Microbiol. 12:120-128.
10. Byrd, R. P., Jr., and T. M. Roy. 2003. Pasteurella multocida respiratory infection: an important cat-associated zoonosis. Arch. Intern. Med. 163:1239. Infect. 34:293-296.
11. Chung, J. Y., I. Wilkie, J. D. Boyce, K. M. Townsend, A. J. Frost, M. Ghoddusi, and B. Adler. 2001. Role of capsule in the pathogenesis of fowl cholera caused by Pasteurella multocida serogroup A. Infect. Immun. 69:2487-2492.
12. Comstock, L. E., and M. J. Coyne. 2003. Bacteroides thetaiotaomicron: a dynamic, niche-adapted human symbiont. Bioessays 25:926-929.
13. Cooke, F. J., A. Kodjo, E. J. Cluterbuck, and K. B. Bamford. 2004. A case of Pasteurella multocida peritoneal dialysis-associated peritonitis and review of the literature. Int. J. Infect. Dis. 8:171-174.
14. de Lorenzo, V., and K. N. Timmis. 1994. Analysis and construction of stable phenotypes in gram-negative bacteria with Tn5- and Tn10-derived minitransposons. Methods Enzymol. 235:386-405.
15. Donnio, P. Y., A. L. Lerestif-Gautier, and J. L. Avril. 2004. Characterization of Pasteurella spp. strains isolated from human infections. J. Comp. Pathol. 130:137-142.
16. Fayad, G., T. Modine, S. Mokhtari, L. Legout, C. Decoene, R. Azzaoui, S. Al-Ruzzeh, G. Naja, B. Larrue, F. Lecomte, E. Senneville, and H. Warembourg. 2003. Pasteurella multocida aortic valve endocarditis: case report and literature review. J. Heart Valve Dis. 12:261-263.
17. Fleischmann, R. D., M. D. Adams, O. White, R. A. Clayton, E. F. Kirkness, A. R. Kerlavage, C. J. Bult, J.-F. Tomb, B. A. Dougherty, J. M. Merrick, K. McKenney, G. Sutton, W. FitzHugh, C. Fields, J. D. Gocayne, J. Scott, R. Shirley, L.-I. Liu, A. Glodek, J. M. Kelley, J. F. Weidman, C. A. Phillips, T. Sprigs, and E. Hedblom, M. D. Cotton, T. R. Utterback, M. C. Hanna, D. T. Nguyen, D. M. Saudek, R. C. Brandon, L. D. Fine, J. A. Fritchman, J. L. Fuhrmann, N. S. M. Geoghagen, C. L. Gnehm, L. A. McDonald, K. V. Small, C. M. Frazer, H. O. Smith, and J. C. Venter. 1995. Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science 269:496-512.
18. Forward, J. A., M. C. Behrendt, N. R. Wyborn, R. Cross, and D. J. Kelly. 1997. TRAP transporters: a new family of periplasmic solute transport systems encoded by the dctPQM genes of Rhodobacter capsulatus and by homologs in diverse gram-negative bacteria. J. Bacteriol. 179:5482-5493.
19. Fuller, T. E., M. J. Kennedy, and D. E. Lowery. 2000. Identification of Pasteurella multocida virulence genes in a septicemic mouse model using signature-tagged mutagenesis. Microb. Pathog. 29:25-38.
20. Gowda, R. V., and R. Stout. 2002. Pasteurella multocida infection in a post-chemotherapy neutropenic host following cat exposure. Clin. Oncol. 14:497-498.
21. Green, B. T., K. M. Ramsey, and P. E. Nolan. 2002. Pasteurella multocida meningitis: case report and review of the last 11 y. Scand. J. Infect. Dis. 3:213-217.
22. Grenier, L. L., H. Watanabe, N. J. Phillips, J. Shao, A. Morgan, A. Zaleski, B. W. Gibson, and M. A. Apicella. 2004. Nontypeable Haemophilus influenzae strain 2019 produces a biofilm containing N-acetylneuraminic acid that may mimic O-linked glycans. Infect. Immun. 72:4249-4260.
23. Harper, M., J. D. Boyce, I. W. Wilkie, and B. Adler. 2003. Signature-tagged mutagenesis of Pasteurella multocida identifies mutants displaying differential virulence characteristics in mice and chickens. Infect. Immun. 71:5440-5446.
24. Hoyer, L. L., A. C. Hamilton, S. M. Steenbergen, and E. R. Vimr. 1992. Cloning, sequencing, and distribution of the Salmonella typhimurium LT2 sialidase gene, nanH, provide evidence for interspecies gene transfer. Mol. Microbiol. 6:873-884.
25. Hunt, M. L., D. J. Boucher, J. D. Boyce, and B. Adler. 2001. In vivo-expressed genes of Pasteurella multocida. Infect. Immun. 69:3004-3012.
26. Inzana, T. J., G. Glindemann, A. D. Cox, W. Wakarchuk, and M. D. Howard. 2002. Incorporation of N-acetylneuraminic acid into Haemophilus somnus lipooligosaccharide (LOS): enhancement of resistance to serum and reduction of LOS antibody binding. Infect. Immun. 70:4870-4879.
27. Kalivoda, K. A., S. M. Steenbergen, E. R. Vimr, and J. Plumbridge. 2003. Regulation of sialic acid catabolism by the DNA binding protein NanR in Escherichia coli. J. Bacteriol. 185:4806-4815.
28. Kelly, D. J., and G. H. Thomas. 2001. The tripartite ATP-independent periplasmic (TRAP) transporters of bacteria and archaea. FEMS Microbiol. Rev. 25:405-424.
29. Kouppari, G., A. Garoufi, G. Stamos, A. Zaphiropoulou, T. Harpathios, and N. J. Legakis. 1999. Pasteurella multocida septicemia and meningitis in an infant. Clin. Microbiol. Infect. 5:101-102.
30. Laupland, K. B., K. P. Rimmer, D. B. Gregson, and D. W. Megran. 2002. Spontaneous empyema and overwhelming septic shock due to Pasteurella multocida. Scand. J. Infect. Dis. 35:132-133.
31. Liu, G., J. Chunsheng, and C. Jin. 2004. CMP-N-acetylneuraminic acid synthetase from Escherichia coli K1 is a bifunctional enzyme. J. Biol. Chem. 279:17738-17749.
32. Luneberg, E., N. Zetzmann, D. Alber, Y. A. Knirel, O. Kooistra, U. Zahringer, and M. Frosch. 2000. Cloning and functional characterization of a 30 kb gene locus required for lipopolysaccharide biosynthesis in Legionella pneumophila. Int. J. Med. Microbiol. 290:37-49.
33. Lichtensteiger, C. A., S. M. Steenbergen, R. M. Lee, D. D. Polson, and E. R. Vimr. 1996. Direct PCR analysis for toxigenic Pasteurella multocida. J. Clin. Microbiol. 34:3035-3039.
34. Linton, D., A. V. Karlyshev, P. G. Hitchen, H. R. Morris, A. Dell, N. A. Gregson, and B. W. Wren. 2000. Multiple N-acetyl neuraminic acid synthetase (neuB) genes in Campylobacter jejuni: identification and characterization of the gene involved in sialylation of lipo-oligosaccharide. Mol. Microbiol. 35:1120-1134.
35. Liu, W., R. F. Chemaly, M. J. Tuohy, M. M. LaSalvia, and G. W. Procop. 2003. Pasteurella multocida urinary tract infection with molecular evidence of zoonotic transmission. Clin. Infect. Dis. 36:58-60.
36. May, B. J., Q. Zhang, L. L. Li, M. L. Paustian, T. S. Whittam, and V. Kapur. 2001. Complete genomic sequence of Pasteurella multocida, Pm70. Proc. Natl. Acad. Sci. USA 98:3460-3465.
37. Mizan, S., A. Henk, A. Stallings, M. Maier, and M. D. Lee. 2000. Cloning and characterization of sialidases with 2-6' and 2-3' sialyl lactose specificity from Pasteurella multocida. J. Bacteriol. 182:6874-6883.
38. Montgomery, J. M., D. Gillespie, P. Sastrawan, T. M. Fredeking, and G. L. Stewart. 2002. Aerobic salivary bacteria in wild and captive Komodo dragons. J. Wildl. Dis. 38:545-551.
39. Perrin, I., P. Blanc, T. Karam, and R. Carbajal. 2003. Meningitis and osteitis caused by Pasteurella multocida in a three-month-old infant. Arch. Pediatr. 10:439-441.
40. Plumbridge, J., and E. Vimr. 1999. Convergent pathways for utilization of the amino sugars N-acetylglucosamine, N-acetylmannosamine, and N-acetylneuraminic acid by Escherichia coli. J. Bacteriol. 181:47-54.
41. Ringenberg, M., C. Lichtensteiger, and E. Vimr. 2001. Redirection of sialic acid metabolism in genetically engineered Escherichia coli. Glycobiology 11:533-539.
42. Ringenberg, M. A., S. M. Steenbergen, and E. R. Vimr. 2003. The first committed step in the biosynthesis of sialic acid by Escherichia coli K1 does not involve a phosphorylated N-acetylmannosamine intermediate. Mol. Microbiol. 50:961-975.
43. Schilling, B., S. Goon, N. M. Samuels, S. P. Gaucher, J. A. Leary, C. R. Bertozzi, and B. W. Gibson. 2001. Biosynthesis of sialylated lipooligosaccharides in Haemophilus ducreyi is dependent on exogenous sialic acid and not mannosamine. Incorporation studies using N-acylmannosamine analogues, N-glycolyneuraminic acid, and 13C-labeled N-acetylneuraminic acid. Biochemistry 40:12666-12677.
44. Silver, R. P., and E. R. Vimr. 1990. Polysialic acid capsule of Escherichia coli K1, p. 39-60. In B. Iglewski and V. Miller (ed.), The bacteria: molecular basis of bacterial pathogenesis. Academic Press, Inc., New York, N.Y.
45. Steenbergen, S. M., and E. R. Vimr. 1990. Mechanism of polysialic acid chain elongation in Escherichia coli K1. Mol. Microbiol. 4:603-611.
46. Steenbergen, S. M., and E. R. Vimr. 2003. Functional relationships of the sialyltransferases involved in expression of the polysialic acid capsules of Escherichia coli K1 and K92 and Neisseria meningitidis groups B or C. J. Biol. Chem. 278:15349-15359.
47. Swords, W. E., M. L. Moore, L. Godzicki, G. Bukofzer, M. J. Mitten, and J. VonCannon. 2004. Sialylation of lipooligosaccharides promotes biofilm formation by nontypeable Haemophilus influenzae. Infect. Immun. 72:106-113.
48. Tseng, H. K., S. C. Su, C. P. Liu, and C. M. Lee. 2001. Pasteurella multocida bacteremia due to non-bite animal exposure in cirrhotic patients: report of two cases. J. Microbiol. Immunol. Infect. 34:293-296.
49. Vimr, E. R. 1992. Selective synthesis and labeling of the polysialic acid capsule in Escherichia coli K1 strains with mutations in nanA and neuB. J. Bacteriol. 174:6191-6197.
50. Vimr, E. R., L. Lawrisuk, J. Galen, and J. B. Kaper. 1988. Cloning and expression of V. cholerae neuraminidase gene nanH in Escherichia coli. J. Bacteriol. 170:1495-1504.
51. Vimr, E. R., K. A. Kalivoda, E. L. Deszo, and S. M. Steenbergen. 2004. Diversity of microbial sialic acid metabolism. Microbiol. Mol. Biol. Rev. 68:132-153.
52. Vimr, E., and C. Lichtensteiger. 2002. To sialylate or not to sialylate: that is the question. Trends Microbiol. 10:254-257.
53. Vimr, E., C. Lichtensteiger, and S. Steenbergen. 2000. Sialic acid metabolism's dual function in Haemophilus influenzae. Mol. Microbiol. 35:1113-1123.
54. Vimr, E. R., and F. A. Troy. 1985. Identification of an inducible catabolic system for sialic acids (nan) in Escherichia coli. J. Bacteriol. 164:845-853.
55. Vimr, E. R., and F. A. Troy. 1985. Regulation of sialic acid metabolism in Escherichia coli: role of N-acylneuraminate pyruvate-lyase. J. Bacteriol. 164:854-860.
56. Von Itzstein, M., W. Y. Wu, G. B. Kok, M. S. Pegg, J. C. Dyason, B. Jin, T. Van Phan, M. L. Smythe, H. F. White, S. W. Oliver, P. M. Colman, J. N. Varghese, D. M. Ryan, J. M. Woods, R. C. Bethell, V. J. Hotham, J. M. Cameron, and C. R. Penn. 1993. Rational design of potent sialidase-based inhibitors of influenza virus replication. Nature 363:418-423.
57. Yamamoto, T., M. Nakashizuka, and I. Terada. 1998. Cloning and expression of a marine bacterial -galactoside -2,6-sialyltransferase gene from Photobacterium damsela JT0160. J. Biochem. 123:94-100.(Susan M. Steenbergen, Car)
Pfizer Animal Health, Kalamazoo, Michigan
ABSTRACT
Pasteurella multocida subsp. multocida is a commensal and opportunistic pathogen of food animals, wildlife, and pets and a zoonotic cause of human infection arising from contacts with these animals. Here, an investigation of multiple serotype A strains demonstrated the occurrence of membrane sialyltransferase. Although P. multocida lacks the genes for the two earliest steps in de novo sialic acid synthesis, adding sialic acid to the growth medium resulted in uptake, activation, and subsequent transfer of sialic acid to a membrane acceptor resembling lipooligosaccharide. Two candidate-activating enzymes with homology to Escherichia coli cytidine 5'-monophospho-N-acetylneuraminate synthetase were overproduced as histidine-tagged polypeptides. The synthetase encoded by pm0187 was at least 37 times more active than the pm1710 gene product, suggesting pm0187 encodes the primary sialic acid cytidylyltransferase in P. multocida. A sialate aldolase (pm1715) mutant unable to initiate dissimilation of internalized sialic acid was not attenuated in the CD-1 mouse model of systemic pasteurellosis, indicating that the nutritional function of sialate catabolism is not required for systemic disease. In contrast, the attenuation of a sialate uptake-deficient mutant supports the essential role in pathogenesis of a sialylation mechanism that is dependent on an environmental (host) supply of sialic acid. The combined results provide the first direct evidence of sialylation by a precursor scavenging mechanism in pasteurellae and of a potential tripartite ATP-independent periplasmic sialate transporter in any species.
INTRODUCTION
Pasteurella multocida subsp. multocida (hereafter referred to as P. multocida) is an economically important opportunistic pathogen of livestock (hemorrhagic septicemia, shipping fever, and atrophic rhinitis), poultry (fowl cholera), wildlife (avian cholera), and laboratory rabbits (snuffles) (1). As a commensal of cats and dogs, P. multocida also is a significant cause of zoonotic abscesses arising mainly from bites or scratches, but it has been increasingly associated with pulmonary disease, sepsis, and meningitis in patients with underlying medical conditions that may compromise their immune status (4, 10, 13, 15, 16, 20, 21, 29, 30, 35, 38, 48). Although the so-called dermonecrotic toxin synthesized by some P. multocida strains produces atrophic rhinitis, the toxin is not essential for respiratory or systemic disease. The P. multocida virulence factors responsible for life-threatening invasive infections are poorly defined despite complete genomic DNA sequencing (36), microarray gene-expression profiling (7, 8), in vivo expression technology (25), and signature tagged mutagenesis (19, 23). Known or suspected virulence factors include adhesins, systems for iron acquisition, lipooligosaccharide (LOS), capsules, and neuraminidases (1). However, with the exception of capsule (11), the function in pathogenesis or relative importance of most other virulence factors is either unknown or uncertain. Evidence is provided here that the metabolism of sialic acids should be included as a potentially essential disease factor in systemic pasteurellosis.
We are interested in the metabolism of the sialic acids, a structurally diverse group of nine-carbon keto sugars (3, 9), and have hypothesized that the interplay between its metabolism and the host-microbe interaction is a unifying theme for understanding diseases caused by a wide range of invasive pathogens or commensals (51-53). For example, Escherichia coli K1 and certain meningococci synthesize sialic acid by a de novo pathway for assembly of sialic acid into outer membrane capsules or lipooligosaccharides (LOS) that are known to inhibit host innate immunity (44). Other pathogens, such as Neisseria gonorrhoeae and Haemophilus spp., do not synthesize sialic acid but scavenge host-derived (environmental) sialic acids for cell surface decoration (modification) involving endogenous sialyltransferases. All known sialyltransferases use the activated nucleotide sugar cytidine 5'-monophospho (CMP)-sialic acid as the obligate donor substrate for transfer of sialyl units to appropriate acceptor substrate molecules. In addition to de novo synthesis or scavenging of sialic acid, some pathogens may also catabolize environmental sialic acid for nutrition (carbon, nitrogen, and energy) or as a source of amino sugars for cell wall biosynthesis (40). How organisms that both decorate their surfaces with sialic acid and use it for nutrition regulate the metabolic decision between surface modification and degradation has not been investigated until recently (27, 41, 42, 54, 55).
Here, for the first time, sialyltransferase is demonstrated in P. multocida, and the role of sialometabolism in the host-pathogen interaction is investigated in an animal model of systemic pasteurellosis. In addition, the first evidence is provided for a novel sialate uptake system involving a putative tripartite ATP-independent periplasmic (TRAP) transporter of the type that functions in carboxylic acid transport in a wide range of bacterial species (18, 28). The combined results indicate that sialometabolism may be a common feature of the entire genus and begin to delineate the importance of sialylation to systemic pasteurellosis.
(A preliminary account of a portion of this study was presented by E. Vimr and C. Lichtensteiger at the 101st General Meeting of the American Society for Microbiology in Orlando, Fla., 2001 [abstr. B-446].)
MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions. Pm70 is a spontaneous capsule-negative, nalidixic acid-resistant turkey isolate kindly provided by V. Kapur (36). TF5 is a nalidixic acid-resistant type A bovine isolate with defined systemic virulence in the CD-1 mouse model (19). TF5E (PM1B-E11) and TF5P (PM3-C9) are TF5-derived strains bearing kanamycin-resistant disruptions of pm1711 and pm1709, respectively, isolated from CD-1 mice after signature-tagged mutagenesis of TF5 (19). The transcriptional orientations of the kanamycin cassettes in these mutants are given in Fig. 1. X-73 is a type A avian isolate kindly provided by M. Wolcott (National Wildlife Health Center, Madison, Wis.). Cincy is a type A strain isolated from a clinical case of human pneumonia involving pulmonary contusions; it was given to us by J. Solomkin (University of Cincinnati College of Medicine, Cincinnati, Ohio) for toxA typing by using our previously described method for PCR detection of toxigenic P. multocida (33). The results of this analysis indicated that the strain was nontoxigenic. Upon receipt, Cincy was passaged once before cryopreservation. The E. coli BL21(DE3) Star strain for histidine-tagged polypeptide overproduction was purchased from Invitrogen (Carlsbad, Calif.). EV5 is a neuA mutant lacking the CMP-sialic acid synthetase (activating enzyme) for polysialic acid capsule synthesis (54). S17-1 is a pir+ strain that supports replication of plasmids bearing pir-dependent replicons; it was used as the donor for mutagenesis of P. multocida as previously described (37).
Plasmid pGPKan is a kanamycin-resistant derivative of the suicide vector pGP704, kindly provided by M. Lee (University of Georgia, Athens). Plasmids pSX1000 and pSX1001 are ampicillin-resistant derivatives of pCR T7/CT TOPO TA (Invitrogen) overexpressing pm1710 or pm0187, respectively.
P. multocida strains were routinely propagated in brain heart infusion (BHI) or Haemophilus test medium (HTM) as previously described (53). Unless indicated otherwise, supplemented HTM contained 1 mg of Neu5Ac/ml, added from a 20-mg/ml filter-sterilized stock solution directly to HTM. E. coli strains were propagated in the Lennox formulation of lysogeny broth (5). When necessary, kanamycin, ampicillin, and nalidixic acid were included in media at 20, 100, and 20 μg/ml, respectively. Bacteria were propagated at 37°C with aeration by rotary mixing.
Mutant construction. Aldolase (pm1715-deficient) P. multocida mutants Pm70A and TF5A were constructed by first amplifying an 500-bp internal fragment of pm1715 using the gene-specific forward (5'-GGATGGTCTTTACGTTGGCGGCAGTAC-3') and reverse (5'-GATTTGACGAGCGCGCACACCATTGAC-3') oligonucleotide primers. The resulting amplicon was cloned into the pGEM T-Easy PCR cloning vector (Promega, Madison, Wis.), followed by excision of the fragment with EcoRI and cloning into the suicide vector pGPKan in E. coli strain S17-1. The plasmid was mated into Pm70 or TF5 with selection for kanamycin resistance and nalidixic acid counterselection (37). The gene disruptions resulting from plasmid cointegration were confirmed by Southern hybridization analysis with the internal gene fragment from pm1715 as the probe. Mutants were propagated in vitro in the presence of kanamycin to select for cointegrate maintenance.
CMP-sialic acid synthetase overproduction. The genes pm1710 and pm0187 encoding putative activating enzymes were cloned into pCR T7/CT TOPO TA cloning kit (version B) after PCR amplification of the complete open reading frames (ORFs) with N- and C-terminal primers to generate pSX1000 and pSX1001, respectively, by previously described methods (41). All PCR primers were synthesized by IDT (Coralville, Iowa). Polypeptides were overproduced by IPTG (isopropyl--D-thiogalactopyranoside) induction as C-terminal His6 fusions (42).
Biochemical assays. N-Acetylneuraminic acid (Neu5Ac; the most common sialic acid) was measured by colorimetric thiobarbituric acid assay (53). Sialyltransferase was detected by incubating membrane samples derived by sonic disruption in 20 mM Tris (pH 8.0), 5 mM magnesium acetate, and 2 mM dithiothreitol (TMD) buffer with CMP-[4-14C]Neu5Ac (50 mCi/mmol), followed by descending paper chromatography in solvent system I (7:3, ethanol-ammonium acetate [pH 7.5]) and liquid scintillation spectrometry of radioactivity remaining at the origin essentially as described previously (54, 55). Quantitative estimation of sialyltransferase activity is described in detail below. Protein was estimated by dye binding (Pierce Chemical Co., Chicago, Ill.) with bovine serum albumin as standard.
Sialidase was measured as relative fluorescence units with the fluorogenic substrate 2'-(4-methylumbelliferyl)-Neu5Ac (MuNeu5Ac) as previously described (50). Sialic acid uptake was measured by depletion assay using unlabeled or [4-14C]Neu5Ac (55 mCi/mmol) as previously described (53, 54). CMP-sialic acid synthetase was detected autoradiographically after incubation of soluble cell extracts with 10 mM CTP and 18 μM radiolabeled Neu5Ac, followed by chromatography in solvent system I or by thin-layer chromatography as previously described (45). Quantitative estimation of synthetase activity was accomplished by excising regions of the chromatograms containing CMP-Neu5Ac product, followed by liquid scintillation spectrometry. Sialate aldolase was detected autoradiographically after incubation of extracts with radiolabeled Neu5Ac and chromatography in solvent system II (n-propanol-1 M sodium acetate [pH 5.0]-water, 7:1:2), with [1-14C]ManNAc (50 mCi/mmol) as the standard (41). All radiochemicals were purchased from American Radiochemical Company (St. Louis, Mo.). Type V sialidase from Clostridium perfringens and all other chemicals were purchased from Sigma Chemical Company (St. Louis, Mo.).
Sialyltransferase assay. Quantitative comparisons of sialyltransferase activities in mammalian and avian P. multocida were carried out by a modification of the procedure described above. Single colonies from the indicated strains were inoculated into 25 ml of BHI and grown to stationary phase. Cultures of encapsulated strains were treated for 1 h at room temperature with 100 U of hyaluronidase to facilitate cell collection by centrifugation. All succeeding preparative steps were carried out at 0 to 4°C. Cell pellets were resuspended in 0.5 ml of TMD buffer and disrupted by sonication (Branson Cell Disrupter 185 with microtip): four cycles of 20 s each on ice at an output setting of 4 with 45 s of cooling between each cycle. Intact cells were removed by centrifugation at 4,000 x g for 4 min. Supernatants were diluted with TMD to a final volume of 3 ml, and membranes were collected by ultracentrifugation in a Beckman TL-1000 desktop ultracentrifuge with TLA100.3 rotor at 52,000 rpm for 15 min. The membrane pellets were resuspended in 1.5 ml of TMD and sonicated for two cycles as described above, followed by the removal of large fragments or any remaining unbroken cells by centrifugation as described above. Membranes were recollected by ultracentrifugation and finally resuspended in 0.05 ml of TMD for storage on ice prior to use. Assays were carried out in triplicate by incubating ca. 150 μg of membrane protein with radiolabeled CMP-Neu5Ac, followed by chromatography and liquid scintillation spectrometry as described above. The data were expressed a picomoles of Neu5Ac transferred to endogenous acceptor(s) ± the standard deviation per mg of protein in 15 min. Differences in specific activities were determined by analysis of variance (P < 0.0001), followed by comparisons of means by the Tukey method ( = 0.05).
Animal experiments. Outbred (5- to 7-week-old) CD-1 female mice were purchased from Charles River Laboratories (Wilmington, Mass.) and housed at the University of Illinois College of Veterinary Medicine animal care facility. Mice were injected intraperitoneally with ca. 700 CFU in 0.1 ml of phosphate-buffered saline containing equal numbers of wild-type (kanamycin-sensitive) and mutant (kanamycin-resistant) P. multocida. Actual input doses were determined by enumerating CFU after dilution plating and overnight incubation. Infected mice were euthanized 17 h postinoculation, and systemic (blood) CFU were differentially enumerated on selective medium. The data are expressed as the ratio of mutant to wild-type bacteria, or the competitive index (CI), which was normalized to the input doses as previously described (53). Control experiments in which mice were inoculated with just TF5A confirmed the stability of the mutation in this strain during the course of infection, indicating <0.1% loss of drug resistance. All animal experiments were approved by the University of Illinois animal care office after institutional review according to National Institutes of Health guidelines. CI differing from 1.0 was analyzed by a one sample, two-way t test using log transformation.
In vivo detection of sialyl acceptor. Wild-type P. multocida strain Pm70 or the Pm70A mutant were grown in 2 ml of BHI and exposed at an A600 of 0.4 for 1 h to 5 μg of Neu5Ac/ml containing a tracer amount of radiolabeled Neu5Ac. Bacteria were collected by centrifugation; residual liquid was removed by carefully wicking the tube walls around the pellets. Wild-type and mutant bacteria removed 40 and 70%, respectively, of the label from the culture medium during the 1-h incubation period. Bacterial pellets were resuspended in 9 μl of 10 mM Tris (pH 8.0) containing 5 μg of lysozyme/ml, followed by the addition of 1 μl of 0.5 M EDTA. Cells were ruptured by three freeze-thaw cycles, and 0.6 μl of 1 M MgCl2 was added prior to treatment with DNase I as previously described (49). Membrane and soluble samples were prepared by centrifugation. Half of each soluble fraction was analyzed by descending paper chromatography in solvent system II by autoradiography, as previously described (42). The membrane fraction was resuspended in 9 μl of water and fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in a 4 to 20% gradient gel. The gel was stained with Coomassie blue, fluoroenhanced, and subjected to autoradiography for 3 days prior to film development. Tricine gels (16.5%) were purchased from Bio-Rad.
RESULTS
Predicted functions of P. multocida strain Pm70 ORFs with presumed roles in sialometabolism. When a BLAST (2) comparison of known E. coli K1 sialometabolic gene products and the P. multocida strain Pm70 genomic DNA sequence (36) was carried out, orthologs of NeuB (sialate synthase) and NeuC (UDP-GlcNAc 2-epimerase) were not detected (Table 1). The absence of neuB- and neuC-like genes indicated that P. multocida lacks the two earliest steps in the pathway for de novo sialic acid synthesis (53). In contrast, P. multocida potentially encodes two CMP-Neu5Ac synthetases (pm0187 and pm1710), an ortholog of the 2,6-sialyltransferase (encoded by pm0188) from the marine organism Photobacterium damsela (57), and a potential 2,3-sialyltransferase that is homologous to one of the Haemophilus influenzae LOS sialyltransferases (8). The absence of genes for de novo sialate synthesis but the presence of ORFs potentially coding for activating enzyme(s) and sialyltransferases suggested that P. multocida may use a sialylation pathway similar to that described previously for H. influenzae (53). This precursor scavenging pathway involves the uptake of environmental sialic acid, followed by its activation by CMP-sialic acid synthetase (NeuA) and subsequent transfer of sialic acid to an appropriate membrane acceptor by sialyltransferase (Fig. 2). As shown in Table 1, P. multocida also potentially encodes a complete pathway for sialic acid catabolism, including two membrane sialidases with distinct substrate specificities for hydrolyzing host sialoglycoconjugates (37). The combined in silico analysis of Pm70 ORFs suggests that, with the exception of de novo Neu5Ac biosynthesis, P. multocida is genetically equipped to carry out a full range of sialometabolic functions including uptake, and either activation of internalized sialic acid for cell surface modification or cleavage to N-acetylmannosamine (ManNAc) and pyruvate for nutrition or source of amino sugars for cell wall biosynthesis (Fig. 2). Since P. multocida is an obligate commensal of mammals, birds, and reptiles, including dragons (38), all of which synthesize sialic acid, sialometabolism is likely to be a central aspect of this bacterium's unusually broad range of host interactions, including the potential to cause disease.
Detection of sialyltransferase. If P. multocida uses a sialylation mechanism similar to that of H. influenzae (53), the isolated membrane fraction should transfer sialic acid from exogenous CMP-Neu5Ac substrate to endogenous membrane acceptor(s). To test for this activity, we incubated the membrane fractions from sonically disrupted bacteria with radioactive CMP-Neu5Ac and measured the incorporation of labeled sialic acid into membranes by descending paper chromatography in solvent system I. Table 2 indicates that all of the P. multocida strains tested (including two type D strains, TF5E, and X-73 [results not shown]) had detectable transferase activity. The highest specific activity was consistently observed with mammalian isolates, a finding comparable in magnitude to the activities of E. coli and meningococcal polysialyltransferases (46). A control experiment with radiolabeled Neu5Ac instead of CMP-Neu5Ac as the substrate did not result in detectable sialyltransferase activity, which is consistent with the known substrate dependency of sialyltransferases for activated sialic acid.
To determine whether the variability in sialyltransferase specific activities resulted from endogenous sialidases (Fig. 2), hydrolysis of the fluorogenic sialoside MuNeu5Ac was measured at basic and acidic pH. As expected from the known acidic pH optima of P. multocida sialidases (37), activity against the sialoside was greater at pH 6.8 than at pH 8.0 (Table 2). However, there were no appreciable differences between sialidase activities of mammalian or the avian isolates tested. The greater activity of the aldolase-negative derivative of Pm70 (Pm70A) may reflect induction by trace sialic acid in the medium, as previously observed for the inducible nan system in E. coli (41, 42, 49).
To directly investigate the susceptibility of the sialylated acceptors to endogenous sialidases, radiolabeled Pm70 membranes were diluted 50-fold into assay buffer, pelleted by centrifugation, and resuspended to the same volume as for the initial radiolabeling. Samples assayed over a 3-h period showed no loss of radioactivity, indicating no effect of endogenous sialidase on sialic acid release. In contrast, 0.5 U of exogenous C. perfringens sialidase released over half of the radioactivity in 1 h, indicating sialyl residues were added by sialyltransferase in their expected terminal -glycoketosidic linkages to endogenous acceptor(s). Furthermore, proteinase K digestion did not release label, and Pm70 lacks the type A hyaluronate capsule, strongly suggesting that the endogenous acceptor is neither protein nor capsular polysaccharide. We concluded that sialyltransferase is a consistent phenotype of at least P. multocida subsp. multocida, which is responsible for most animal disease (1).
Identification of CMP-Neu5Ac synthetase. The absence of orthologs of neuB and neuC but the presence of sialyltransferase implies P. multocida scavenges environmental sialic acid and activates it intracellularly using one or both of the putative CMP-Neu5Ac synthetase candidates indicated in Table 1. One of these candidate synthetases (encoded by pm0187) is located adjacent to the putative 2,6-sialyltransferase encoded by pm0188, whereas the other (pm1710) is predicted to be the first gene of the P. multocida nan-like operon (Fig. 1). PCR analysis of 32 P. multocida type A or D isolates from our swine collection (33) resulted in the successful amplification of the expected DNA products from pm1710 or pm0187 in all strains tested (data not shown). The ubiquity of these candidate neuA orthologs in P. multocida is consistent with the sialyltransferase-positive phenotype documented in Table 2.
Campylobacter spp. and Legionella pneumophila have been shown to express multiple orthologs of neuA or neuB (32, 34), suggesting that both pm0187 and pm1710 could encode functional sialate activating enzymes despite the greater similarity of the pm1710 gene product to phospholipase than cytidylyltransferase (Table 2). To determine whether either of these ORFs codes for an active CMP-Neu5Ac synthetase, the respective genes were cloned into a vector designed to express gene products in frame with C-terminal His6 tags. As shown in Fig. 3A, pm0187 complemented an E. coli K1 neuA mutant for polysialic acid biosynthesis, whereas pm1710 did not (Fig. 3B), suggesting that only pm0187 encodes a functional synthetase. Similarly, while an extract of wild-type Pm70 had both detectable Neu5Ac aldolase and CMP-Neu5Ac synthetase activity (Fig. 4A, lane 3), only the His6-tagged pm0187 product produced CMP-Neu5Ac from CTP and radiolabeled Neu5Ac (compare lanes 2 to 4 to lanes 5 to 7 in Fig. 4B). The results of complementation and direct biochemical analyses indicate that the pm0187 is a functional NeuA ortholog.
To further investigate the potential cytidylyltransferase activity of pm1710, we compared its relative activity to that of the pm0187-encoded NeuA ortholog. As shown in Fig. 5, NeuA had, as expected, high synthetase activity that saturated between 56 and 113 μg of protein. In contrast, no activity of the extract containing the overproduced pm1710 gene product was detected until 159 μg of protein was tested, resulting in at least a 37-fold-lower specific activity than NeuA. These results suggest that if pm1710 encodes a cytidylyltransferase, its substrate is unlikely to be Neu5Ac. Further work is necessary to determine the function of the pm1710 gene product, but the present results indicate that NeuA encoded by pm0187 is the primary if not the sole Neu5Ac synthetase in P. multocida. The results in Table 1 suggest that pm1710 may encode a phospholipase or esterase, similar to the C-terminal domain of E. coli NeuA (31).
Scavenging environmental sialic acid. Although P. multocida lacks candidate genes for the two earliest steps in sialic acid synthesis (Table 1), the results described above demonstrate that it has the genetic information for Neu5Ac activation and sialyl transfer. In addition, an extract of wild-type P. multocida has a sialate aldolase activity that could initiate the catabolism of environmental sialic acid, as indicated by the product migrating with the same relative migration (Rf) value (0.83) as authentic ManNAc (Fig. 4A, lane 3). Therefore, if P. multocida uses a precursor scavenging mechanism for cell surface sialylation, there must be an uptake system for sialic acid that provides the substrate for both intracellular catabolic and activating enzymes (Fig. 2). The results presented in Table 1 suggested that pm0835 could encode an ortholog of NanT, whereas pm1708, together with pm1709, was a likely candidate for a TRAP sialate transporter.
Sialate uptake was demonstrated directly by a chromogenic depletion assay (53), indicating that wild-type strain TF5 quantitatively removed Neu5Ac during growth in supplemented HTM (Table 3). In contrast, growth of TF5P in supplemented HTM did not result in Neu5Ac depletion (Table 3), strongly suggesting that the polypeptide encoded by pm1709 is required for sialate uptake. This conclusion was confirmed by uptake of radiolabeled Neu5Ac during growth of the wild type but not TF5P in HTM containing 10 μM unlabeled and a tracer amount of labeled Neu5Ac. Wild-type TF5 took up nearly 50% of the label by 1 h, whereas the input counts per minute (cpm) remaining in the culture medium of TF5P remained constant when sampled at 1-, 2-, or 3-h intervals (99.2% ± 1.3% [standard deviation]). Interestingly, with wild-type bacteria the number of extracellular cpm actually increased after the initial 1-h labeling period, resulting in ca. 20% more radioactivity detected in the medium by 3 h compared to that present after the first hour. We interpret this increase in cpm as the excretion of fermentative end products, because Neu5Ac is clearly metabolized after uptake as shown by the nearly 70% growth stimulation of TF5 relative to TF5P (Table 3). Note that the uptake results imply P. multocida uses only one system for sialic acid transport, at least under the conditions of laboratory growth used for these experiments, suggesting that the putative NanT ortholog encoded by pm0835 is not involved in sialate uptake (Fig. 2).
Isolation and characterization of P. multocida sialate aldolase mutants. To confirm the presence of a functional sialate dissimilatory system in P. multocida, we constructed a gene disruption of the strain Pm70 nanA ortholog (pm1715) by homologous recombination with a pGPKan derivative harboring an internal (500 bp) fragment of the gene; the resulting mutant was designated Pm70A. ORF pm1715 is predicted to be the last gene of the P. multocida nan-like dissimilatory operon (Fig. 1C); thus, the gene disruption would have no effect other than loss of the proposed aldolase activity encoded by pm1715. An identical gene disruption was constructed in the bovine P. multocida isolate TF5 (18), which was designated TF5A. The loss of aldolase activity was detected by direct biochemical assay with radiolabeled Neu5Ac (Fig. 6, lane 6). Extracts prepared from either Pm70A or TF5A lacked detectable sialate aldolase activity (Fig. 6, lanes 1 and 5, respectively), whereas extracts from either wild-type parent (Fig. 6, lanes 2 and 3), as expected, produced a radiolabeled product with the same mobility as authentic ManNAc (Fig. 4, lane 7). An extract of TF5E bearing a nonpolar kanamycin resistance cassette (14) transcribed in the same direction as pm1711 expressed wild-type aldolase activity (Fig. 4, lane 4). Although we did not directly confirm that pm1711 encodes ManNAc-6-P epimerase, its homology to E. coli nanE (Table 1) and its proximity to sialate uptake and aldolase genes (Fig. 1) strongly supports the proposed epimerase function (Fig. 2). This assignment is consistent with detection of nanK and yhcH orthologs in P. multocida (Table 1) which, along with nanA and nanE, comprise the E. coli nan operon (Fig. 1A).
Endogenous sialyl acceptor. To identify the in vivo acceptor recognized by P. multocida sialyltransferase(s), we expected that a growing culture of Pm70A would efficiently transport and activate exogenous radiolabeled Neu5Ac, because there would be no competition for label from the catabolic system, as originally observed with E. coli nanA mutants (55). However, most of the label accumulated by Pm70A remained as free Neu5Ac (Fig. 7, lane 1), making it difficult to identify the endogenous membrane acceptor (Fig. 8A and B, lanes 5 and 7). In contrast, wild-type strain Pm70 produced a readily detectable CMP-Neu5Ac signal (Fig. 7, lane 2) that was correlated with a membrane-associated product with electrophoretic mobility near the dye front and apparent mass of <14 kDa (Fig. 8A and B, lanes 4 and 6). The radiolabeled product remaining in the wells is likely to be peptidoglycan, as suggested by past results with E. coli in which ManNAc dissimilation resulted in rapid labeling of the insoluble cell wall fraction (55). This suggestion is consistent with the low detectable amount of ManNAc in wild type (Fig. 7, lane 2) and with the absence of any detectable peptidoglycan in Pm70A (Fig. 8B, lanes 5 and 7), which cannot produce the ManNAc precursor. Note that, as expected, all of the soluble radiolabel that was in low-molecular-weight products passed through the gel in front of the dye marker, exiting into the anode buffer (Fig. 7B, lanes 2 and 3). When membranes that were used for the sialyltransferases assays presented in Table 2 were solubilized and fractionated by Tricine SDS-PAGE, only the broad band migrating at <14 kDa was detected after autoradiography (data not shown), a result which is consistent with the in vivo analysis shown in Fig. 8. We conclude that the sole macromolecular sialyl acceptor is a relatively low molecular weight product resembling LOS in its electrophoretic behavior. This conclusion is consistent with the biochemical results presented above, and previous results in the H. influenzae system (see reference 53 and references therein). Both the products of the in vivo and in vitro sialylation reactions will now be open to structural characterization.
Requirement for sialometabolism during systemic pasteurellosis. P. multocida strain Pm70 lacks the type A hyaluronate capsule that is essential for systemic virulence. Therefore, all animal experiments were carried out with encapsulated TF5 or its mutant derivatives. As summarized in Table 4, strain TF5 was previously found to be highly virulent in the CD-1 mouse model of systemic pasteurellosis (50% lethal dose [LD50] of <10 CFU), whereas TF5E and TF5P had >10,000 times higher LD50 values in this model (19). Because the data presented above strongly suggest that the pm1709 gene product is required for sialate transport, we concluded that sialometabolism is likely to be a necessary feature of systemic pasteurellosis. To correlate the CI with LD50 as an alternative measurement of virulence, we determined the CI for TF5E, indicating that a 500-fold decrease in relative fitness corresponds to a >10,000-fold increase in LD50 (Table 4). Because both the pm1711 and the pm1709 gene products are required for sialate catabolism, but only that of pm1709 is required for LOS sialylation, we sought to distinguish between the possible nutritional requirement for Neu5Ac in vivo and its alternative use for surface modification.
To distinguish between these possibilities, we carried out mixed infections with wild type and the TF5A aldolase-deficient mutant. As shown in Table 4, despite the relatively weak activation of Neu5Ac by Pm70A grown in HTM (Fig. 7), TF5A was as fit as the wild type for systemic propagation in the CD-1 mouse infection model. This result formally excludes a primary nutritional function of sialic acid catabolism in vivo, strengthening our hypothesis, as we previously described for the related H. influenzae system (53), that LOS sialylation is dependent on an external (environmental) source of sialic acid. Attenuation of TF5P is consistent with this hypothesis, since failure to transport exogenous sialic acid would result in the expression of unsialylated LOS. Independent analysis of H. influenzae sialylation has shown that very little sialic acid must be incorporated in vivo to provide a fitness advantage during otitis media (6). We suggest the same is likely to be true for P. multocida during systemic infection.
Although our combined results point to an essential role of sialylation in vivo, the attenuation of TF5E is difficult to reconcile with this hypothesis and suggests that, whereas Neu5Ac catabolism per se is not required for pathogenesis, as shown by the relatively wild-type fitness of the NanA mutant (Table 4), the ManNAc 6-phosphate expected to accumulate in a nanE mutant (41, 42) may be toxic during in vivo infection. Alternatively, a second mutation in TF5E, unrelated to the defect in pm1711, may account for the observed attenuation of this mutant. The single-infection control experiment described in Materials and Methods rules out instability of the pm1715 mutation as an explanation. Further biochemical analysis of LOS structure and genetic complementation will be necessary to determine the exact function of LOS sialylation in pasteurellosis and to understand the possible function of the pm1711 gene product in pathogenesis.
DISCUSSION
Since the discovery that H. influenzae sialylates its LOS by a precursor scavenging mechanism (52, 53), analogous pathways have been described in H. ducreyi (43) and H. somnus (26), suggesting that all haemophili may depend on an environmental source of sialic acid for LOS modification. Thus, bacteria that lack orthologs of neuB and neuC, which are needed for the two earliest steps in sialic acid biosynthesis (Fig. 2) (41, 42), but contain one or more copies of neuA and at least one sialyltransferase gene are likely to use precursor scavenging as a mechanism for surface sialylation. H. influenzae was subsequently shown to sialylate its LOS during infection in the chinchilla model of otitis media, and it was also shown that this surface modification was crucial to virulence (6). Although the exact function of sialylation during animal infection remains unclear, two recent studies have shown that sialic acid is required for normal H. influenzae biofilm formation, which may be a necessary process for colonization of mucosal surfaces (22, 47). In contrast, we have provided evidence that P. multocida sialylation, or at least sialate transport, may be necessary for systemic pasteurellosis, probably by protecting the sialylated bacteria from innate host defense mechanisms because mice do not normally carry P. multocida and would, therefore, not be expected to mount an antibody-dependent immune response under our conditions of experimental infection (1).
Perhaps in addition to a role of sialylation during systemic disease, the ability of obligate commensals and symbionts to modify their repertoire of surface carbohydrates arose as a mechanism for avoiding detection and clearance by the host's immune system. This idea would imply that sialylation may be required for persistence and that, if so, a therapeutic approach blocking surface modification could have broad clinical value by causing certain microbes to become sensitive to host clearance mechanisms. Evidence that surface carbohydrate modification plays a role in symbiosis comes from Bacteroides spp., in which the synthesis of eight structurally distinct capsular polysaccharides varies randomly in response to a site-specific recombinase for reversible inversion of the different capsule biosynthetic operon promoters (12). In contrast, our results point to the variation at the level of LOS sialylation being dependent on a supply of host sialic acid.
An elevated serum sialoglycoconjugate concentration has been shown to be a general marker of inflammation, suggesting that endogenous or microbial sialidases could increase the extracellular free sialic acid concentration in response to any ongoing inflammatory process (51). Thus, by increasing their cell surface sialylation in response to elevated environmental sialic acid, some microbes may be protected under conditions that normally activate innate defense mechanisms that would ordinarily lead to detection and clearance of unsialylated species (51). The P. multocida lifestyle appears to be unusually committed to sialometabolism, including two sialidases, two potential sialyltransferases, at least one activating enzyme, and a complete system for sialic acid dissimilation, including sialic acid transport (Fig. 2). The present results suggest that scavenging sialic acid is essential for at least systemic propagation. This robust sialometabolic system might also account in part for the wide P. multocida host range that includes diverse mammalian, avian, and reptilian species.
A functional precursor scavenging system minimally requires mechanisms for sialic acid uptake, activation, and some way to regulate the metabolic decision between sialate catabolism and surface modification. To date, all known sialyl acceptors in microbes that have been found to use a precursor scavenging mechanism are lipopolysaccharides in the outer membranes of gram-negative nasopharyngeal commensals. In the case of P. multocida, we have identified an activating enzyme (encoded by pm0187) and shown that a relatively low molecular weight membrane component resembling LOS is the endogenous acceptor. We have presented evidence here that at least the first gene (pm1709) of the potential operon defined by pm1007-1009 is required for sialate uptake. This uptake system is homologous to the TRAP transporters for carboxylic acids in diverse bacterial species, including H. influenzae (Fig. 1B). In contrast, the E. coli sialate transporter encoded by nanT (Fig. 1A) is a member of the major facilitator superfamily (MFS), indicating the existence of at least two distinct mechanisms of bacterial sialate uptake. Our results strongly suggest that P. multocida uses only the TRAP system for sialate uptake.
All known TRAP transporters include three linked genetic components, two of which encode integral membrane proteins that may be fused in some systems (e.g., HI0147 and pm1708 shown in Fig. 1B and C, respectively). The larger component codes for a predicted polypeptide with 12 membrane-spanning domains, resembling the secondary structures of most MFS permeases that couple solute uptake to one of the ion gradients across the cell membrane (28). The other membrane component is predicted to contain four membrane-spanning regions and is assumed to function in concert with the MFS-like permease (18, 28). The final components of all TRAP transporters are members of a family of periplasmic binding proteins that resemble the extracytoplasmic solute receptors of ATP-binding cassette-like uptake systems (28). Ligand binding to these extracytoplasmic solute receptors is thought to account for the high uptake affinities of multicomponent transporters (28). In P. multocida, we have provided evidence that the potential periplasmic component encoded by pm1709 is required for sialic acid uptake. Thus, P. multocida, H. influenzae, Fusobacterium nucleatum, Vibrio cholerae, and an unknown number of other pathogens or commensals are predicted to use TRAP transporters for sialic acid uptake (51). We have purified the pm1709 gene product as a His6 C-terminal fusion and shown that it binds Neu5Ac in the low micromolar range, confirming its predicted biochemical function as a sialate-binding protein (E. Vimr and S. Steenbergen, unpublished results).
TRAP transporters recognize a diverse range of mono- and dicarboxylic acids (28). Similarly, the sialic acids may be considered to be monocarboxylates, suggesting that the negative charge at carbon 1 may be necessary for recognition by the binding protein encoded by pm1709, which we propose to rename nanP (Table 1). Regardless of the exact binding mechanism, it should be possible to identify low-molecular-weight compounds that compete for sialic acid binding and thus block uptake. As in the case of the first influenza virus sialidase inhibitor (56), a compound with sufficient binding affinity and favorable on-off rates should prevent precursor scavenging, thereby blocking subsequent surface modification. Our results not only imply that such an inhibitor could be useful for treating systemic infection but also suggest potential value for preventing colonization or inhibiting the persistence of diverse bacterial species. Work directed toward rational design of sialate TRAP transporter inhibitors is in progress.
ACKNOWLEDGMENTS
We are grateful to J. Solomkin, V. Kapur, and M. Wolcott for providing bacterial strains. We also thank M. Lee for providing the suicide vector and for advice on mutant construction.
This study was supported in part by the Governor's Venture Technology Individual Grant Program and National Institutes of Health grant 2R01 AI42015 (E.V.).
Present address: Department of Molecular Biology and Microbiology, Tufts University, Boston, MA 02115.
Present address: DNA Indexing Laboratory, Springfield, IL 62707.
REFERENCES
1. Adlam, C., and J. M. Rutter. 1989. Pasteurella and pasteurellosis. Academic Press, Ltd., London, England.
2. Altschul, S. F., W. Gish, W. Miller, E. W. Myers, and D. J. Lipman. 1990. Basic local alignment search tool. J. Mol. Biol. 215:403-410.
3. Angata, T., and A. Varki. 2002. Chemical diversity in the sialic acids and related alpha-keto acids: an evolutionary perspective. Chem. Rev. 102:439-469.
4. Berge, A., A. Fagergren, and S. H. Stierenstedt. 2002. Pasteurella multocida septicemia in 2 Swedish patients. Scand. J. Infect. Dis. 34:138-139.
5. Bertani, G. 2004. Lysogeny at mid-twentieth century: P1, P2, and other experimental systems. J. Bacteriol. 186:595-600.
6. Bouchet, V., D. W. Hood, J. Li, J. R. Brisson, G. A. Randle, A. Martin, Z. Li, R. Goldstein, E. K. Schweda, S. I. Pelton, J. C. Richards, and E. R. Moxon. 2003. Host-derived sialic acid is incorporated in Haemophilus influenzae lipopolysaccharide and is a major virulence factor in experimental otitis media. Proc. Natl. Acad. Sci. USA 100:8898-8903.
7. Boyce, J. D., I. Wilkie, M. Harper, M. L. Paustian, V. Kapur, and B. Adler. 2002. Genomic scale analysis of Pasteurella multocida gene expression during growth within the natural chicken host. Infect. Immun. 70:6871-6879.
8. Boyce, J. D., I. Wilkie, M. Harper, M. L. Paustian, V. Kapur, and B. Adler. 2004. Genomic-scale analysis of Pasteurella multocida gene expression during growth within liver tissue of chickens with fowl cholera. Microbes Infect. 6:290-298.
9. Bravo, I. G., S. Garcia-Vallve, A. Romeu, and A. Reglero. 2004. Prokaryotic origin of cytidylyltransferases and alpha-ketoacid synthases. Trends Microbiol. 12:120-128.
10. Byrd, R. P., Jr., and T. M. Roy. 2003. Pasteurella multocida respiratory infection: an important cat-associated zoonosis. Arch. Intern. Med. 163:1239. Infect. 34:293-296.
11. Chung, J. Y., I. Wilkie, J. D. Boyce, K. M. Townsend, A. J. Frost, M. Ghoddusi, and B. Adler. 2001. Role of capsule in the pathogenesis of fowl cholera caused by Pasteurella multocida serogroup A. Infect. Immun. 69:2487-2492.
12. Comstock, L. E., and M. J. Coyne. 2003. Bacteroides thetaiotaomicron: a dynamic, niche-adapted human symbiont. Bioessays 25:926-929.
13. Cooke, F. J., A. Kodjo, E. J. Cluterbuck, and K. B. Bamford. 2004. A case of Pasteurella multocida peritoneal dialysis-associated peritonitis and review of the literature. Int. J. Infect. Dis. 8:171-174.
14. de Lorenzo, V., and K. N. Timmis. 1994. Analysis and construction of stable phenotypes in gram-negative bacteria with Tn5- and Tn10-derived minitransposons. Methods Enzymol. 235:386-405.
15. Donnio, P. Y., A. L. Lerestif-Gautier, and J. L. Avril. 2004. Characterization of Pasteurella spp. strains isolated from human infections. J. Comp. Pathol. 130:137-142.
16. Fayad, G., T. Modine, S. Mokhtari, L. Legout, C. Decoene, R. Azzaoui, S. Al-Ruzzeh, G. Naja, B. Larrue, F. Lecomte, E. Senneville, and H. Warembourg. 2003. Pasteurella multocida aortic valve endocarditis: case report and literature review. J. Heart Valve Dis. 12:261-263.
17. Fleischmann, R. D., M. D. Adams, O. White, R. A. Clayton, E. F. Kirkness, A. R. Kerlavage, C. J. Bult, J.-F. Tomb, B. A. Dougherty, J. M. Merrick, K. McKenney, G. Sutton, W. FitzHugh, C. Fields, J. D. Gocayne, J. Scott, R. Shirley, L.-I. Liu, A. Glodek, J. M. Kelley, J. F. Weidman, C. A. Phillips, T. Sprigs, and E. Hedblom, M. D. Cotton, T. R. Utterback, M. C. Hanna, D. T. Nguyen, D. M. Saudek, R. C. Brandon, L. D. Fine, J. A. Fritchman, J. L. Fuhrmann, N. S. M. Geoghagen, C. L. Gnehm, L. A. McDonald, K. V. Small, C. M. Frazer, H. O. Smith, and J. C. Venter. 1995. Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science 269:496-512.
18. Forward, J. A., M. C. Behrendt, N. R. Wyborn, R. Cross, and D. J. Kelly. 1997. TRAP transporters: a new family of periplasmic solute transport systems encoded by the dctPQM genes of Rhodobacter capsulatus and by homologs in diverse gram-negative bacteria. J. Bacteriol. 179:5482-5493.
19. Fuller, T. E., M. J. Kennedy, and D. E. Lowery. 2000. Identification of Pasteurella multocida virulence genes in a septicemic mouse model using signature-tagged mutagenesis. Microb. Pathog. 29:25-38.
20. Gowda, R. V., and R. Stout. 2002. Pasteurella multocida infection in a post-chemotherapy neutropenic host following cat exposure. Clin. Oncol. 14:497-498.
21. Green, B. T., K. M. Ramsey, and P. E. Nolan. 2002. Pasteurella multocida meningitis: case report and review of the last 11 y. Scand. J. Infect. Dis. 3:213-217.
22. Grenier, L. L., H. Watanabe, N. J. Phillips, J. Shao, A. Morgan, A. Zaleski, B. W. Gibson, and M. A. Apicella. 2004. Nontypeable Haemophilus influenzae strain 2019 produces a biofilm containing N-acetylneuraminic acid that may mimic O-linked glycans. Infect. Immun. 72:4249-4260.
23. Harper, M., J. D. Boyce, I. W. Wilkie, and B. Adler. 2003. Signature-tagged mutagenesis of Pasteurella multocida identifies mutants displaying differential virulence characteristics in mice and chickens. Infect. Immun. 71:5440-5446.
24. Hoyer, L. L., A. C. Hamilton, S. M. Steenbergen, and E. R. Vimr. 1992. Cloning, sequencing, and distribution of the Salmonella typhimurium LT2 sialidase gene, nanH, provide evidence for interspecies gene transfer. Mol. Microbiol. 6:873-884.
25. Hunt, M. L., D. J. Boucher, J. D. Boyce, and B. Adler. 2001. In vivo-expressed genes of Pasteurella multocida. Infect. Immun. 69:3004-3012.
26. Inzana, T. J., G. Glindemann, A. D. Cox, W. Wakarchuk, and M. D. Howard. 2002. Incorporation of N-acetylneuraminic acid into Haemophilus somnus lipooligosaccharide (LOS): enhancement of resistance to serum and reduction of LOS antibody binding. Infect. Immun. 70:4870-4879.
27. Kalivoda, K. A., S. M. Steenbergen, E. R. Vimr, and J. Plumbridge. 2003. Regulation of sialic acid catabolism by the DNA binding protein NanR in Escherichia coli. J. Bacteriol. 185:4806-4815.
28. Kelly, D. J., and G. H. Thomas. 2001. The tripartite ATP-independent periplasmic (TRAP) transporters of bacteria and archaea. FEMS Microbiol. Rev. 25:405-424.
29. Kouppari, G., A. Garoufi, G. Stamos, A. Zaphiropoulou, T. Harpathios, and N. J. Legakis. 1999. Pasteurella multocida septicemia and meningitis in an infant. Clin. Microbiol. Infect. 5:101-102.
30. Laupland, K. B., K. P. Rimmer, D. B. Gregson, and D. W. Megran. 2002. Spontaneous empyema and overwhelming septic shock due to Pasteurella multocida. Scand. J. Infect. Dis. 35:132-133.
31. Liu, G., J. Chunsheng, and C. Jin. 2004. CMP-N-acetylneuraminic acid synthetase from Escherichia coli K1 is a bifunctional enzyme. J. Biol. Chem. 279:17738-17749.
32. Luneberg, E., N. Zetzmann, D. Alber, Y. A. Knirel, O. Kooistra, U. Zahringer, and M. Frosch. 2000. Cloning and functional characterization of a 30 kb gene locus required for lipopolysaccharide biosynthesis in Legionella pneumophila. Int. J. Med. Microbiol. 290:37-49.
33. Lichtensteiger, C. A., S. M. Steenbergen, R. M. Lee, D. D. Polson, and E. R. Vimr. 1996. Direct PCR analysis for toxigenic Pasteurella multocida. J. Clin. Microbiol. 34:3035-3039.
34. Linton, D., A. V. Karlyshev, P. G. Hitchen, H. R. Morris, A. Dell, N. A. Gregson, and B. W. Wren. 2000. Multiple N-acetyl neuraminic acid synthetase (neuB) genes in Campylobacter jejuni: identification and characterization of the gene involved in sialylation of lipo-oligosaccharide. Mol. Microbiol. 35:1120-1134.
35. Liu, W., R. F. Chemaly, M. J. Tuohy, M. M. LaSalvia, and G. W. Procop. 2003. Pasteurella multocida urinary tract infection with molecular evidence of zoonotic transmission. Clin. Infect. Dis. 36:58-60.
36. May, B. J., Q. Zhang, L. L. Li, M. L. Paustian, T. S. Whittam, and V. Kapur. 2001. Complete genomic sequence of Pasteurella multocida, Pm70. Proc. Natl. Acad. Sci. USA 98:3460-3465.
37. Mizan, S., A. Henk, A. Stallings, M. Maier, and M. D. Lee. 2000. Cloning and characterization of sialidases with 2-6' and 2-3' sialyl lactose specificity from Pasteurella multocida. J. Bacteriol. 182:6874-6883.
38. Montgomery, J. M., D. Gillespie, P. Sastrawan, T. M. Fredeking, and G. L. Stewart. 2002. Aerobic salivary bacteria in wild and captive Komodo dragons. J. Wildl. Dis. 38:545-551.
39. Perrin, I., P. Blanc, T. Karam, and R. Carbajal. 2003. Meningitis and osteitis caused by Pasteurella multocida in a three-month-old infant. Arch. Pediatr. 10:439-441.
40. Plumbridge, J., and E. Vimr. 1999. Convergent pathways for utilization of the amino sugars N-acetylglucosamine, N-acetylmannosamine, and N-acetylneuraminic acid by Escherichia coli. J. Bacteriol. 181:47-54.
41. Ringenberg, M., C. Lichtensteiger, and E. Vimr. 2001. Redirection of sialic acid metabolism in genetically engineered Escherichia coli. Glycobiology 11:533-539.
42. Ringenberg, M. A., S. M. Steenbergen, and E. R. Vimr. 2003. The first committed step in the biosynthesis of sialic acid by Escherichia coli K1 does not involve a phosphorylated N-acetylmannosamine intermediate. Mol. Microbiol. 50:961-975.
43. Schilling, B., S. Goon, N. M. Samuels, S. P. Gaucher, J. A. Leary, C. R. Bertozzi, and B. W. Gibson. 2001. Biosynthesis of sialylated lipooligosaccharides in Haemophilus ducreyi is dependent on exogenous sialic acid and not mannosamine. Incorporation studies using N-acylmannosamine analogues, N-glycolyneuraminic acid, and 13C-labeled N-acetylneuraminic acid. Biochemistry 40:12666-12677.
44. Silver, R. P., and E. R. Vimr. 1990. Polysialic acid capsule of Escherichia coli K1, p. 39-60. In B. Iglewski and V. Miller (ed.), The bacteria: molecular basis of bacterial pathogenesis. Academic Press, Inc., New York, N.Y.
45. Steenbergen, S. M., and E. R. Vimr. 1990. Mechanism of polysialic acid chain elongation in Escherichia coli K1. Mol. Microbiol. 4:603-611.
46. Steenbergen, S. M., and E. R. Vimr. 2003. Functional relationships of the sialyltransferases involved in expression of the polysialic acid capsules of Escherichia coli K1 and K92 and Neisseria meningitidis groups B or C. J. Biol. Chem. 278:15349-15359.
47. Swords, W. E., M. L. Moore, L. Godzicki, G. Bukofzer, M. J. Mitten, and J. VonCannon. 2004. Sialylation of lipooligosaccharides promotes biofilm formation by nontypeable Haemophilus influenzae. Infect. Immun. 72:106-113.
48. Tseng, H. K., S. C. Su, C. P. Liu, and C. M. Lee. 2001. Pasteurella multocida bacteremia due to non-bite animal exposure in cirrhotic patients: report of two cases. J. Microbiol. Immunol. Infect. 34:293-296.
49. Vimr, E. R. 1992. Selective synthesis and labeling of the polysialic acid capsule in Escherichia coli K1 strains with mutations in nanA and neuB. J. Bacteriol. 174:6191-6197.
50. Vimr, E. R., L. Lawrisuk, J. Galen, and J. B. Kaper. 1988. Cloning and expression of V. cholerae neuraminidase gene nanH in Escherichia coli. J. Bacteriol. 170:1495-1504.
51. Vimr, E. R., K. A. Kalivoda, E. L. Deszo, and S. M. Steenbergen. 2004. Diversity of microbial sialic acid metabolism. Microbiol. Mol. Biol. Rev. 68:132-153.
52. Vimr, E., and C. Lichtensteiger. 2002. To sialylate or not to sialylate: that is the question. Trends Microbiol. 10:254-257.
53. Vimr, E., C. Lichtensteiger, and S. Steenbergen. 2000. Sialic acid metabolism's dual function in Haemophilus influenzae. Mol. Microbiol. 35:1113-1123.
54. Vimr, E. R., and F. A. Troy. 1985. Identification of an inducible catabolic system for sialic acids (nan) in Escherichia coli. J. Bacteriol. 164:845-853.
55. Vimr, E. R., and F. A. Troy. 1985. Regulation of sialic acid metabolism in Escherichia coli: role of N-acylneuraminate pyruvate-lyase. J. Bacteriol. 164:854-860.
56. Von Itzstein, M., W. Y. Wu, G. B. Kok, M. S. Pegg, J. C. Dyason, B. Jin, T. Van Phan, M. L. Smythe, H. F. White, S. W. Oliver, P. M. Colman, J. N. Varghese, D. M. Ryan, J. M. Woods, R. C. Bethell, V. J. Hotham, J. M. Cameron, and C. R. Penn. 1993. Rational design of potent sialidase-based inhibitors of influenza virus replication. Nature 363:418-423.
57. Yamamoto, T., M. Nakashizuka, and I. Terada. 1998. Cloning and expression of a marine bacterial -galactoside -2,6-sialyltransferase gene from Photobacterium damsela JT0160. J. Biochem. 123:94-100.(Susan M. Steenbergen, Car)