|Home | About | Journals | Submit | Contact Us | Français|
Haemophilus ducreyi, the causative agent of chancroid, produces a lipooligosaccharide (LOS) which terminates in N-acetyllactosamine. This glycoform can be further extended by the addition of a single sialic acid residue to the terminal galactose moiety. H. ducreyi does not synthesize sialic acid, which must be acquired from the host during infection or from the culture medium when the bacteria are grown in vitro. However, H. ducreyi does not have genes that are highly homologous to the genes encoding known bacterial sialic acid transporters. In this study, we identified the sialic acid transporter by screening strains in a library of random transposon mutants for those mutants that were unable to add sialic acid to N-acetyllactosamine-containing LOS. Mutants that reacted with the monoclonal antibody 3F11, which recognizes the terminal lactosamine structure, and lacked reactivity with the lectin Maackia amurensis agglutinin, which recognizes α2,3-linked sialic acid, were further characterized to demonstrate that they produced a N-acetyllactosamine-containing LOS by silver-stained sodium dodecyl sulfate-polyacrylamide gel electrophoresis and mass spectrometric analyses. The genes interrupted in these mutants were mapped to a four-gene cluster with similarity to genes encoding bacterial ABC transporters. Uptake assays using radiolabeled sialic acid confirmed that the mutants were unable to transport sialic acid. This study is the first report of bacteria using an ABC transporter for sialic acid uptake.
Haemophilus ducreyi is the causative agent of the sexually transmitted disease chancroid. This disease is most prevalent in developing countries (54); however, sporadic outbreaks occur in predominantly urban areas of the United States (62). Chancroid has been linked, as a cofactor, to the heterosexual transmission of the human immunodeficiency virus, especially in areas where both diseases are predominant (28, 54, 61, 62).
A number of putative virulence factors of H. ducreyi have been described which seem to play a role in pathogenicity. Two of these factors are toxins: a hemolytic toxin (3, 42, 71) and cytolethal distending toxin (10, 11, 13). The outer membrane proteins, DsrA and DltA, have been shown to promote resistance to killing by normal human serum (17, 34). The hemoglobin receptor HgbA (16, 56) and the Cu,Zn-superoxide dismutase (33, 41, 49) both seem to play a role in iron acquisition for H. ducreyi. Additionally, a number of proteins have been shown to play a role in adherence (9, 19, 34).
The lipooligosaccharide (LOS) produced by H. ducreyi is also a putative virulence factor. Previous studies have shown that LOS plays a role in adherence of bacteria to keratinocytes and human foreskin fibroblasts (2, 20). Structural studies have been performed on the LOS from a number of H. ducreyi strains (1, 6, 7, 20, 21, 39, 40, 51, 52). These studies have shown that one of the predominant glycoforms expressed by H. ducreyi terminates in N-acetyllactosamine. These same terminal sugars are also found on LOS structures expressed by Neisseria meningitidis, Neisseria gonorrhoeae, and Haemophilus influenzae (reviewed in reference 47). Previous findings have shown that some of the LOS structures from these bacteria mimic human antigens (35, 36; for reviews, see also references 26 and 63) and are involved in receptor-mediated interactions (24, 25, 58, 59).
The LOS glycoform terminating in N-acetyllactosamine can be further extended by the addition of a single sialic acid (N-acetyl-neuraminic acid [NeuAc]) to the terminal galactose residue. Studies have shown that sialic acid is an important virulence factor in N. gonorrhoeae and N. meningitidis, promoting resistance to bactericidal activity of normal human serum (5, 29, 43, 70). Additionally, sialylation of N. meningitidis LOS has been shown to increase resistance to phagocytosis by human dendritic cells (65). Previous studies from our laboratory have shown that the LOS of most H. ducreyi strains, including the prototype strain 35000HP, are highly sialylated (39), but the mechanism of sialic acid transport is unclear.
Until recently, the Escherichia coli NanT protein was the only known bacterial sialic acid transporter (38, 68). Sequence analysis and mutagenesis studies by our laboratory determined that there is no functional NanT homolog in H. ducreyi (46). Therefore, we performed random mutagenesis, followed by lectin and antibody screening, to identify mutants that lacked sialic acid in their LOS but retained the ability to produce an N-acetyllactosamine-containing LOS. We predicted that these mutants would have mutations in the sialyltransferase, the CMP--NeuAc synthetase, or the sialic acid transporter. The gene that was interrupted in each mutant was then determined. Several independent mutations in an ABC transporter gene cluster were identified, and these mutants were shown to be incapable of transporting sialic acid.
All strains used in this study are listed in Table Table1.1. H. ducreyi strains were cultured on chocolate agar plates at 35°C in a 5% CO2 atmosphere, as previously described (42). Chocolate agar was supplemented with 1 mM NeuAc (Calbiochem, La Jolla, CA) or kanamycin at 20 μg/ml when appropriate. E. coli strains were grown on Luria-Bertani (LB) plates or in LB broth and were supplemented with kanamycin when appropriate.
Several colonies from a 48-h chocolate agar plate of strain 35000HP were swabbed with a cotton swab onto a fresh chocolate agar plate. After incubation for 12 to 14 h at 35°C in a 5% CO2 atmosphere, competent cells were prepared as described by Palmer and coworkers (42). Electrocompetent cells were electroporated with 1 μl of EZ::TN<R6Kγori/KAN-2> transposome (Epicentre, Madison, WI), then suspended in 100 μl of brain heart infusion broth, and spread on a chocolate agar plate. After 6 h of incubation, cells were scraped and dilutions were plated on chocolate agar containing kanamycin at 20 μg/ml. After 48 h, approximately 2,500 clones were patched onto chocolate agar plates supplemented with kanamycin. After incubation overnight, cells were lifted onto nitrocellulose and then lysed by incubation in chloroform vapor for 5 min. Nonspecific protein binding sites were blocked by incubation of the filters in 1% skim milk in Tris-buffered saline (TBS) (0.05 M Tris, 0.15 M NaCl, pH 7.5). The filter was then incubated with a 1:200 dilution of Maackia amurensis agglutinin (MAA) lectin conjugated to horseradish peroxidase (EY Labs, San Mateo, CA) in TBS for 1 h. After three washes with TBS, the blot was developed with Bio-Rad (Hercules, CA) peroxidase detection reagents. Twenty-two clones that failed to react with MAA were then screened for reactivity with the murine monoclonal antibody 3F11 (8). Colonies were lifted and lysed as above, and then the filters were processed as described by Sun and coworkers (57).
LOS preparations for sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and mass spectrometric analyses were extracted from H. ducreyi cells grown for 2 days on chocolate agar plates. Cells were washed with PBS (pH 7.4) containing 0.5 mM MgCl2 and 0.15 mM CaCl2 and suspended in double-distilled water (ddH2O). LOS was extracted by the hot phenol method (27). One-tenth of the total LOS extracted from each sample was set aside for SDS-PAGE analysis. These samples were reconstituted in 10 μl of ddH2O, and 1 to 2 μl of these samples was diluted in 8 to 9 μl of Laemmli sample buffer containing 2% β-mercaptoethanol (Bio-Rad) for a total volume of 10 μl. Neuraminidase-treated LOS samples were reconstituted in 5 μl of 2× neuraminidase buffer (100 mM sodium acetate, 8 mM calcium chloride, pH 5.5) and incubated overnight at 37°C with 5 mU of neuraminidase isolated from Vibrio cholerae (5 μl of a 1-U/ml solution; Roche, Indianapolis, IN). These samples were diluted in Laemmli buffer as described above. All SDS-PAGE samples were boiled for 10 min, followed by a brief centrifugation, and 1 to 2 μl of each sample was loaded onto the gel. Samples were separated on a 15% SDS-PAGE gel (32). Silver staining was performed according to a protocol previously described by Tsai and Frasch (64).
Water soluble O-deacylated LOS (O-LOS) samples were prepared by treating the LOS extracted from one agar plate with 50 μl of anhydrous hydrazine, followed by acetone precipitation (44). Lyophilized O-LOS samples were desalted by reconstitution in ddH2O and performance of drop dialysis using 0.025-μm-pore-size nitrocellulose membranes (Millipore, Bedford, MA). Samples were then lyophilized and reconstituted in 5 to 10 μl of ddH2O. O-LOS samples were further desalted using Dowex 50X 100 to 200 mesh cation exchange beads (NH4+ form). Before samples were loaded onto the target, they were mixed 1:1 (vol/vol) with matrix (160 mM 2,5-dihydroxybenzoic acid, 87.5 mM 1-hydroxyisoquinoline solution in acetone-water; 4:1 [vol/vol]). For the analysis of O-LOS, matrix-assisted laser desorption ionization-mass spectrometry (MS) by using an Applied Biosystems (Framingham, MA) Voyager DE time-of-flight mass spectrometer was performed. Mass spectra were run in linear negative-ion mode with a nitrogen laser (337 nm) under delayed extraction conditions: 165-ns delay time with a grid voltage of 94% of full acceleration voltage (20 kV). Mass spectra were acquired, averaged (typically 100 laser shots), and externally calibrated with a standard peptide mixture consisting of angiotension I, and adrenocorticotropin fragments 1 to 7, 18 to 39, and 7 to 38 (Bachem, Torrance, CA). All masses measured under these conditions were average masses.
The sequences flanking the Tn5 element were first rescued and then sequenced. Genomic DNA was prepared from each clone using the Puregene reagents (Gentra Systems, Minneapolis, MN), digested with MfeI, self ligated, and transformed into E. coli EC100D pir+ (Epicentre). The DNA sequence flanking the mTn5 insertion was then determined in both directions by using dye terminator chemistries and oligonucleotide primers KAN-2 FP-1 and R6KAN-2 RP-1 (Epicentre). No clones were rescued from strain 35000HP-310. However, the sequence flanking the mutation in this strain was determined by a single-primer PCR methodology as described by Ducey and Dyer (15). The resulting sequences were used to query the H. ducreyi genome sequence using the BLAST algorithm (4) and Seqman from the DNASTAR suite of programs.
Several colonies from a 48-h chocolate agar plate incubation of each strain were swabbed with a cotton swab onto a fresh chocolate agar plate. After incubation for 12 to 14 h at 35°C in a 5% CO2 atmosphere, cells were suspended in RPMI 1640 without phenol red (ICN, Aurora, OH) supplemented with 1% (vol/vol) 200 mM glutamine (Invitrogen, Carlsbad, CA), 1% (vol/vol) 100 mM sodium pyruvate (Invitrogen), and 10% (vol/vol) 0.5 M HEPES, pH 7.5, then harvested by centrifugation at room temperature for 10 min at 3,200 × g. Cells were resuspended to an absorbance of 2.0 at 600 nm in the same medium. An aliquot of the cell suspension was equilibrated to 35°C for 3 min, and the reaction was started by the addition of a 1/10 volume of RPMI-based medium containing [3H]sialic acid (American Radiochemicals, St. Louis, MO), prepared such that the concentration of sialic acid (Sigma, St. Louis, MO) in the final reaction mixture was 2.5 μM. After the indicated period of time, the reaction mixture was rapidly filtered through a 0.22-μm-type GS filter (Millipore, Bedford, MA) and washed twice with 2 ml of RPMI-based medium. Filters were air dried and then counted in 10 ml of ScintiSafe 30% LSC-Cocktail (Fischer Scientfic, Hampton, NH) in a Beckman LS 6500 Scintillation Counter (Beckman Coulter, Fullerton, CA).
Mutagenesis was performed by introducing a mTn5 transposon randomly into the chromosome of H. ducreyi. Kanamycin-resistant clones were screened with the MAA lectin, which recognizes terminal α2,3-linked sialic acid. Clones that did not bind MAA were then screened with the monoclonal antibody 3F11, which recognizes terminal N-acetyllactosamine. Nine clones were identified that were 3F11 positive and MAA negative (Table (Table11).
LOS from seven of the mutants, 35000HP-305, -306, -310, -313, -319, -320, and -322 was isolated and compared by SDS-PAGE with LOS isolated from strains 35000HP and 35000HP-RSM203, a sialyltransferase (lst) mutant (6). The structure of 35000HP LOS has been determined, and nomenclature describing its various glycoforms has been developed; both are shown in Fig. Fig.1A1A (6). Silver-stained SDS-PAGE gels demonstrated that the previously identified sialylated glycoform (band A5a1) (39) was absent from LOS isolated from all seven mutants generated in this study, as well as the sialyltransferase mutant 35000HP-RSM203 (Fig. (Fig.1B).1B). The A5b1 and A5b2 bands, corresponding to the addition of GlcNAc or Gal-β1,4-GlcNAc, respectively, were very prevalent in all of the mutants and in the LOS from the sialyltransferase mutant 35000HP-RSM203. Figure Figure1C1C shows a representative SDS-PAGE gel of LOS isolated from the parent strain 35000HP, the lst mutant 35000HP-RSM203, and the mutant 35000HP-310 with or without neuraminidase treatment. In the neuraminidase-treated 35000HP LOS sample, the sialylated band (A5a1) was not visible. In addition, the di-N-acetyllactosamine band (A5b1) was only visible in the 35000HP LOS sample after neuraminidase treatment. Neuraminidase treatment had no visible affect on either the LOS from strain 35000HP-310 or the LOS from the sialyltransferase mutant 35000HP-RSM203. LOS from all of the mutants was treated with neuraminidase prior to separation by SDS-PAGE. This treatment had no visible affect on the LOS isolated from the mutants (data not shown). The LOS from the other two MAA negative and 3F11 positive mutants did not contain the A5b1 and A5b2 LOS-glycoforms (data not shown) and therefore were not further characterized in this study.
Mass spectra obtained from O-LOS isolated from the parent strain 35000HP and strain 35000HP-306 are shown in Fig. Fig.2.2. For clarity, only the full O-LOS region of the spectra is shown, since there were no differences observed in the lipid A region of the mutant compared to the parent strain. The LOS from 35000HP has been previously characterized, and nomenclature describing the various LOS glycoforms has been established (Fig. (Fig.1A)1A) (6). The A5* and A5 peaks, corresponding to LOSs that terminate in N-acetyllactosamine (with and without phosphoethanolamine, respectfully), are the most prevalent peaks in all of the spectra (Fig. 2A and B). However, the A5a1 and A5a1* peaks, previously identified as the A5 and A5* structures with the addition of one NeuAc, respectively, are only present in the 35000HP O-LOS.
Previous experiments from our laboratory demonstrated that supplementing the growth medium of H. ducreyi with additional sialic acid increases the abundance of the sialylated glycoforms present in H. ducreyi LOS (50). Figure Figure2B2B shows the O-LOS spectra from 35000HP and 35000HP-306 grown in medium supplemented with 1 mM sialic acid. The relative abundance of the A5a1* peak is increased in the 35000HP sample under these conditions of high sialic acid, corresponding to the previous findings (50). However, even when grown in the presence of 1 mM sialic acid, the sialylated peaks A5a1 and A5a1* are not detectable in the 35000HP-306 sample. Figures 2A and B both demonstrate that the A5b1, A5b1*, and A5b2 peaks, corresponding to the addition of GlcNAc (without and with phosphoethanolamine) and lactosamine to the A5 structure, respectively, are more readily detectable in the 35000HP-306 samples than in the 35000HP samples. These data correspond to previous findings (6) and indicate that the absence of the sialic acid on the terminal Gal of the A5 LOS-glycoform enables a larger percentage of the LOS to extend the oligosaccharide to a di-N-acetyllactosamine. These data are representative of the seven mutants characterized in this study.
Sequences flanking the transposon insertion in the seven mutants were sequenced. Surprisingly, none of these insertions mapped to the sialyltransferase gene, lst, or the CMP-NeuAc synthetase, neuA. Instead, these mutants all had Tn5 insertions in the genes designated Hd1669 to -1672 (Table (Table1;1; Fig. Fig.3).3). This gene cluster had previously been predicted to encode an ABC-transporter. Based on sequence analysis and phenotype, the Hd1669 to -1672 genes were designated satABCD (sat for sialic acid transport), respectively. The Hd1669 gene (satA) has two possible AUG start codons. When translated from the second AUG, the gene is predicted to encode for a protein with an Mr of 56,495. As translated, lysine is present at amino acid position 2 and ADA residues are present at positions 22 to 24. AXA is frequently present at a signal peptidase cleavage site. Together with the lysine at position 2, this sequence is typical of leader peptides, suggesting that Hd1669 contains a 24-amino-acid leader peptide. BLAST analysis indicates that this protein is a member of family 5 of the bacterial extracellular solute binding proteins (pfam00496). Family 5 includes the oligopeptide and dipeptide binding proteins OppA and DppA of E. coli, a number of other gram-negative periplasmic binding proteins (PBPs), and gram-positive homologs of these periplasmic proteins, which are lipoproteins. The Hd1670 and Hd1671 genes (satB and satC, respectively) are predicted to encode the integral membrane permease proteins with an Mr of 35,134 and 71,259, respectively. Hd1670 is a member of pfam00528, a family of binding protein-dependent transport system integral inner-membrane proteins. Hd1671 has two domains: the N-terminal domain, which shows high homology to other permease proteins, and the C-terminal domain, which has high homology to transport ATPases (National Center for Biotechnology Information conserved domain cd0267.1). The Hd1672 gene (satD) is predicted to encode a protein with an Mr of 29,593. This protein also contains a transport ATPase domain of the same class (National Center for Biotechnology Information conserved domain cd0267.1). The predicted protein sequence from SatD matched the consensus sequence of the critical conserved domains of ABC ATPases, and the sequence from SatC matched the consensus sequences except for the addition of one amino acid in the Walker A motif.
ABC transport systems with periplasmic binding proteins are considered uptake systems. This observation together with the lack of sialic acid in the LOS of the mutants is consistent with the hypothesis that the Hd1669 to -1672 genes encode the sialic acid transporter. However, none of the evidence presented unequivocally demonstrates that this gene cluster encodes the sialic acid transporter. Therefore, the mutants were directly tested for their ability to transport sialic acid. First, sialic acid transport by the parental strain was demonstrated by incubation of the parental strain with [3H]NeuAc, followed by filtration to remove sialic acid that was not cell associated. Figure Figure4A4A shows a time course experiment. These data show that the bacteria are able to transport [3H]NeuAc under these experimental conditions. Representative ABC transporter mutants were also tested. Figure Figure4B4B shows single 10-min time point data for cells of strains 35000HP-306 (satB mutant), 35000HP-313 (satD mutant), 35000HP-319 (satA mutant), 35000HP-322 (satC mutant), and 35000HP. These data demonstrated that all four mutants are deficient in the transport of [3H]NeuAc compared to the parental strain, 35000HP.
In this study, we have shown that H. ducreyi utilizes the products of the satABCD genes to encode an ABC transporter for sialic acid uptake. Sequence comparisons with other ABC transporter systems suggest that these genes encode the PBP SatA, the integral-membrane permease protein SatB, and the ATPase SatD. These comparisons also show that the N terminus of the SatC protein contains a permease domain, while the C terminus contains an ATPase domain. How might sialic acid be transported by this novel ABC transporter? A hypothetical model for sialic acid transport in H. ducreyi, based on the predicted protein functions, can be constructed as shown in Fig. Fig.5.5. This model is based on the maltose transporter in E. coli (12). The sialic acid transporter model predicts that sialic acid is first bound to the PBP SatA. Binding of the sialic acid may increase the affinity of SatA for SatB/SatC. Binding of the SatA-sialic acid complex to SatB/SatC may decrease SatA's affinity for sialic acid. This decrease may allow the release of sialic acid, the reorientation of SatB/SatC, and the exposure of sites needed for ATP hydrolysis. After ATP hydrolysis, sialic acid is transported to the cytoplasm, and the rest of the complex returns to its original conformation.
The sialic acid transporter identified in H. ducreyi is the first ABC transporter identified that transports sialic acid; however, sequence analysis suggests that it may be the first of many bacterial sialic acid ABC transporters yet to be identified and characterized (67). In addition to H. ducreyi, two other members of the Pasteurellaceae family, Actinobacillus pleuropneumoniae and Haemophilus somnus, may utilize an ABC transporter for sialic acid uptake. Both bacteria have proteins that are highly homologous to the H. ducreyi SatABCD proteins. Proteins predicted from unfinished genome sequences of A. pleuropneumoniae serovar 1 strain 4074 and H. somnus 129PT are 86% identical (>510 amino acids) and 75% identical (>454 amino acids), respectively, to SatA from H. ducreyi. The H. ducreyi SatB protein is 92% identical (>317 amino acids) to a predicted protein from A. pleuropneumoniae serovar 1 strain 4074 and 84% identical (>317 amino acids) to predicted proteins from H. somnus 129PT and H. somnus 2336. The H. ducreyi SatC protein is 85% identical (>648 amino acids) to a predicted protein from A. pleuropneumoniae serovar 1 strain 4074, 81% identical (>649 amino acids) to a predicted protein from H. somnus 2036 and 80% identical (>649 amino acids) to a predicted protein from H. somnus 2336. Similarly, the SatD protein is highly homologous to predicted proteins from A. pleuropneumoniae and H. somnus. These data strongly suggest that sialic acid is also transported by an ABC transporter in these members of the Pasteurellaceae.
The gram-positive bacteria Corynebacterium glutamicum, Corynebacterium diphtheriae, and Streptomyces avermitilis, also have homologs of the sat genes. The satA gene of H. ducreyi has a 40% G+C content, similar to the overall G+C content of the H. ducreyi genome (38.2% G+C). In contrast, these gram-positive organisms have a high G+C content both in their respective genomes and in their satA homologs, suggesting that the sat genes were not recently passed to or acquired from each other.
Mechanisms of sialic acid transport are diverse among bacteria, even within their subfamilies. The sialic transporter in E. coli, NanT, is a symporter of the major facilitator superfamily type of membrane transporters (38, 67). In this type of transport, protons or metal ions are coupled to solute uptake (67). The tripartite ATP-independent periplasmic-type of transporter has been proposed for sialic acid uptake in two members of the Pasteurellaceae family, Pasteurella multocida subsp. multocida and H. influenzae (55, 67). The tripartite ATP-independent periplasmic transporters are predicted to have a periplasmic binding component and one membrane-spanning protein (55, 67). The identification that H. ducreyi, also a member of the Pasteurellaceae family, transports sialic acid using an ABC transporter adds a third method of sialic transport. Thus, there are at least three different methods of sialic acid transport in bacteria. The finding that differences in the types of transport mechanisms used by these bacteria occur even within a family is intriguing and could represent the various needs and environments that these bacteria encounter, or alternatively, could be an example of convergent evolution.
Besides incorporation of sialic acid into LOS, lipopolysaccharide, and capsule, some bacteria are also able to utilize sialic acid as a carbon and nitrogen source (45, 68). Studies of E. coli have demonstrated that a number of genes are involved in the conversion of sialic acid to fructose-6-phosphate (45, 68). After transport, the first step in the sialic acid catabolic pathway is its conversion to N-acetylmannosamine (ManNAc) and pyruvate by the aldolase NanA (45, 68). ManNAc is then phosphorylated by the ATP-dependent kinase NanK (48); the product, ManNAc-6-P, is then converted to N-acetylglucosamine-6-phosphate (GlcNAc-6-P) by the epimerase NanE (48). Conversion of GlcNAc-6-P to glucosamine-6-phosphate (Glc-6-P) is performed by the deacetylase NagA (45). Lastly, NagB is involved in the conversion of Glc-6-P to fructose-6-phosphate (Fru-6-P) (45). Comparisons of the E. coli genes involved in the sialic acid catabolism pathway to the H. influenzae genome demonstrated that a number of potentially homologous genes were present (66). These genes include nanA, nagA, nagB, nanK, and nanE homologs. Mutagenesis performed on the putative H. influenzae aldolase-encoding gene, nanA, demonstrated that this mutant was unable to metabolize sialic acid and exhibited no detectable aldolase activity (66). Whole-cell enzyme-linked immunosorbent assay of H. influenzae nanA mutants, with the monoclonal antibody 3F11, demonstrated that the mutants had a dramatic increase in the level of sialylation of the LOS (66). These data suggest that in these organisms there exists a balance between incorporation of sialic acid into LOS and the use of sialic acid as a carbon source. Sequence analyses of H. ducreyi strain 35000HP, with the H. influenzae sialic acid catabolic pathway genes, indicate that a region 5′ of rfe contains putative nanA, nanE, nagA, and nagB homologs but does not seem to contain a nanK homolog. Due to the apparent absence of a homologous nanK in H. ducreyi, it is currently unclear if H. ducreyi possesses a functional sialic acid degradation pathway. Studies to discern whether this catabolic pathway is functional in H. ducreyi are currently under way.
Since the accumulation of high levels of sialic acid in the cytoplasm have been shown to be toxic in some bacteria (68), how might H. ducreyi control uptake and intracellular levels of sialic acid? In this and previous studies, H. ducreyi was grown on plates containing high levels of sialic acid (22, 50). The presence of the sialic acid did not seem to affect the growth of the bacterium. To more quantitatively test whether the presence of high levels of sialic acid affected the growth rate of H. ducreyi, growth curves of H. ducreyi under low (~0.5 μM basal level in medium) (50), medium (10 μM), and high (1 mM) sialic acid conditions were performed (46). There were no differences in the growth rates suggesting that H. ducreyi either regulates sialic acid uptake, degrades sialic acid, and uses it as a carbon source or that the bacteria are able to tolerate some intracellular accumulation of sialic acid. Both E. coli and H. influenzae seem to utilize their ability to degrade sialic acid as one mechanism of controlling the intracellular levels of sialic acid (66, 68, 69). As noted above, it is currently unclear if H. ducreyi is able to degrade sialic acid. Another possible mechanism for controlling sialic acid uptake may be through the PBPs. The PBPs from other bacteria have been shown to be important for controlling the amount and efficiency of solute uptake (12, 14, 30, 31, 37, 45). Therefore, it is possible that SatA plays an important role in controlling sialic acid uptake in H. ducreyi. One mechanism of control could be that the level of satA expression is affected by the intracellular levels of sialic acid. When the intracellular pool of sialic acid is high, expression of satA is downregulated; conversely, when the intracellular pool of sialic acid is low, expression of satA is upregulated. This type of regulation has been shown to be effective for controlling intracellular pools of iron in a wide range of bacteria (reviewed in reference 18). Another possible option for H. ducreyi to manage possible toxic levels of intracellular sialic acid is to have a relatively high tolerance for the accumulation of sialic acid. Under these conditions, the levels of SatA would not dramatically change under low or high sialic acid conditions; instead, satA would be expressed constitutively and sialic acid uptake would not be tightly regulated. If this option is correct, it suggests that H. ducreyi has a higher tolerance than E. coli for the accumulation of sialic acid in the cytoplasm. The identification of the components of the H. ducreyi sialic acid transporter should enable further investigation of the regulation of sialic acid uptake and utilization.
The role that sialylation of the LOS in H. ducreyi plays in virulence is currently unclear. However, similar to other bacteria (24, 58, 59), H. ducreyi LOS plays a role in adherence to human cells (2, 20). The exact role that LOS plays in facilitating this interaction has not yet been elucidated. Recent studies of H. influenzae demonstrated that sialylation of the LOS played a role in biofilm formation (23, 60). In addition, one of these studies demonstrated that these bacteria produce a biofilm containing α2,6-linked sialic acid (23). Sialylation of LOS in H. ducreyi could therefore play a role in adherence or possibly microcolony formation. Now that we have identified the sialic acid uptake system in H. ducreyi as a novel ABC transporter (SatABCD), thus completing the identification of the major components of the sialic acid biosynthesis pathway, it should be possible to design experiments that will determine the role of sialic acid in the biology of this human pathogen.
We acknowledge the outstanding technical assistance of Maria Hughes, Erin Tracy, and Huachun Zhong.
The DNA sequence was determined in the Core Facility of Columbus Children's Research Institute, which is supported in part by NIH grant K12 HD43372. This work was supported by NIH grant AI31254 (B.W.G.).
Editor: D. L. Burns