ABSTRACT
Porphyromonas gingivalis lipid A is heterogeneous with regard to the number, type, and placement of fatty acids. Analysis of lipid A by matrix-assisted laser desorption ionization-time of flight mass spectrometry reveals clusters of peaks differing by 14 mass units indicative of an altered distribution of the fatty acids generating different lipid A structures. To examine whether the transfer of hydroxy fatty acids with different chain lengths could account for the clustering of lipid A structures, P. gingivalis lpxA (lpxAPg ) and lpxDPg were cloned and expressed in Escherichia coli strains in which the homologous gene was mutated. Lipid A from strains expressing either of the P. gingivalis transferases was found to contain 16-carbon hydroxy fatty acids in addition to the normal E. coli 14-carbon hydroxy fatty acids, demonstrating that these acyltransferases display a relaxed acyl chain length specificity. Both LpxA and LpxD, from either E. coli or P. gingivalis, were also able to incorporate odd-chain fatty acids into lipid A when grown in the presence of 1% propionic acid. This indicates that E. coli lipid A acyltransferases do not have an absolute specificity for 14-carbon hydroxy fatty acids but can transfer fatty acids differing by one carbon unit if the fatty acid substrates are available. We conclude that the relaxed specificity of the P. gingivalis lipid A acyltransferases and the substrate availability account for the lipid A structural clusters that differ by 14 mass units observed in P. gingivalis lipopolysaccharide preparations.
Porphyromonas gingivalis is a gram-negative, asaccharolytic, anaerobic bacterium that is strongly associated with adult periodontitis (12, 18, 27). Its virulence factors include proteases (19, 34, 46), fimbria (2, 22, 54), hemagglutinins (17, 23, 35), and lipopolysaccharide (LPS) (3, 11, 13). LPS is a major component of the outer membrane of gram-negative bacteria. LPS contributes to the integrity of the bacterial cell envelope and is a strong modulator of the host innate immune system (3, 5, 9, 10, 13, 14, 16). Both positive and negative modulations of the host response by LPS are believed to play a role in the development of various infectious conditions (13, 15, 21, 24, 28, 32, 33, 44, 50). Most of this activity is known to be contained in the lipid portion of the LPS molecule known as lipid A (41, 51). Escherichia coli lipid A structure consists of a glucosamine disaccharide phosphorylated at the 1 and 4′ positions and is substituted with four β-hydroxy-myristates in amide linkage (2 and 2′ positions) and in ester linkage (3 and 3′ positions). The structure also contains a laurate in the secondary ester linkage on the 2′ β-hydroxy-myristate and a myristate in the secondary linkage on the 3′ β-hydroxy-myristate to generate the hexa-acyl lipid A canonical structure (51).
P. gingivalis LPS is significantly different from E. coli in that it consists of a heterogeneous mixture of lipid A structures that differ in the type and number of fatty acids and the number of phosphate groups (1, 26, 40). Compared to LPS from E. coli, the P. gingivalis lipid A structures are underacylated (four to five fatty acids) and underphosphorylated (a single phosphate at position 1) and that contain longer-chain (C15, C16, and C17) and branched (iso) fatty acids (26, 31). The major lipid A structures consist of di- and mono-phosphoryl penta-acylated lipid A and mono-phosphoryl tetra-acylated lipid A forms. Additional lipid A structural heterogeneity is found in the mono- and di-phosphoryl penta-acylated structures, as well as in the mono-phosphoryl tetra-acylated structure containing additional lipid A mass ions that differ by 14 mass units greater and less than m/z 1,770, 1,690, and 1,449, respectively. This type of heterogeneity results in “clusters” of matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) peaks that were originally observed by Kumada et al. (26) in P. gingivalis and more recently have been described for LPS obtained from Leptospira interrogans (36) and Stenotrophomonas maltophilia (53). In both studies, these clusters were suspected to represent fatty acid heterogeneity; however, a mechanism for the generation of these lipid A structural clusters has not been determined.
One mechanism that could result in “clusters” of lipid A structures that differ in mass by one methylene unit is the transfer of fatty acids of various chain lengths to the lipid A biosynthetic intermediates. LpxA and LpxD represent two lipid A biosynthetic acyltransferases that catalyze the transfer of ester-linked primary β-hydroxy-myristate to a monosaccharide precursor (UDP-glucosamine) at the 3 and 3′ positions and amide-linked β-hydroxy-myristate at the 2 and 2′ positions, respectively (37-39). An approach that has proved successful for determining the fatty acid chain length acyl specificity for lpxA and lpxD lipid A gene homologues obtained from several different bacterial species has been to complement E. coli mutants defective in either lpxA or lpxD and then determine the fatty acids found on lipid A (30, 45, 48, 49). In the present study, the expression of P. gingivalis lpxA and lpxD in E. coli as a means for assessing their contribution to P. gingivalis lipid A heterogeneity is described. We found that both lpxA and lpxD obtained from P. gingivalis, as well as the available fatty acid substrates, significantly contributed to lipid A fatty acid heterogeneity and provided a mechanism for the generation of “clusters” of lipid A structures found in P. gingivalis.
MATERIALS AND METHODS
Bacterial strains and medium. E. coli temperature-sensitive lpxD mutant strain RL25 was obtained from the E. coli Genetic Stock Center, New Haven, CT, and E. coli temperature-sensitive lpxA mutant strain SM101 was obtained from Christian Raetz, Duke University Medical Center, Durham, NC. E. coli temperature-sensitive strains were maintained at 30°C in Luria broth (LB) or plates made by adding agar to this medium (LB agar). Strains expressing wild-type E. coli or P. gingivalis lpxA or lpxD genes were maintained at 30°C in LB containing 100 μg of ampicillin/ml. For LPS isolations, strains were grown overnight at 42°C in the same medium containing 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside). To examine the ability of the E. coli strains to incorporate odd-chain fatty acids into their lipid A, bacteria were grown overnight in LB containing 1% propionic acid (Sigma). P. gingivalis strain 33277 was obtained from the American Type Culture Collection and cultured anaerobically in Trypticase soy broth supplemented with hemin and menadione as described previously (40).
Construction of strains expressing lpxAPg and lpxDPg .A search of the P. gingivalis W83 genome (http://www.tigr.org ) for sequences homologous to E. coli lpxA and lpxD identified a single P. gingivalis homolog of lpxA (PG0070) and a single homolog of lpxD (PG0072). Oligonucleotide primers were designed for PCR amplification of the homologues designated lpxAPg and lpxDPg . The sequences of the primers for amplification of lpxAPg were 5′-CGCATCGAATTCGTCAGAGACAAAAATCAGTCCGTTG-3′ (forward) and 5′-TTAAGCGGATCCTCACTCCATGGTTCCGCGGACAAT-3′ (reverse) and include the addition of EcoRI and HindIII restriction sites to the 5′ and 3′ primers, respectively. The amplified fragment was ligated into EcoRI/HindIII-digested pUC19 and transformed into E. coli TOP10F competent cells (Invitrogen, Carlsbad, CA). Clones were then selected by their ampicillin resistance. Similarly, lpxAEc , lpxDPg , and lpxDEc were amplified, ligated into EcoRI/BamHI-digested pUC18, and cloned into TOP10F cells by using the following primers to amplify the gene fragments and create EcoRI (5′) and BamHI (3′) restriction sites. The sequences of the primers used to amplify lpxDPg were 5′-CGCATCGAATTCGGAATTTACAGCCCAACAGATAGCT-3′ (forward) and 5′-TTAAGCGGATCCTCAGTGTTTGTTGTTTTTGCATAT-3′ (reverse). The sequences of the primers used to amplify lpxAEc were 5′-CGCATCGAATTCGATTGATAAATCCGCCTTTGTGCAT-3′ (forward) and 5′-TTAAGCGGATCCTTAACGAATCAGACCGCGCGTTGA-3′ (reverse), and the sequences used to amplify lpxDEc were 5′-CGCATCGAATTCGCCTTCAATTCGACTGGCTGATTTA-3′ (forward) and 5′-TTAAGCGGATCCTTAGTCTTGTTGATTAACCTTGCG-3′ (reverse). Recombinant plasmids found to contain inserts of the expected sizes were sequenced by BigDye termination reaction at the University of Washington DNA Sequence Facility. Plasmids confirmed to contain lpxAPg and lpxAEc were isolated and used to transform E. coli strain SM101, while plasmids confirmed to contain lpxDPg and lpxDEc were isolated and used to transform E. coli strain RL25.
Isolation and characterization of LPS.LPS was isolated by hot phenol-water extraction (55) and purified by nuclease digestions, protease digestions, and repeated ultracentrifugation (6). LPS preparations were determined to have <1% contaminating protein as determined by BCA reaction (Pierce, Rockford, IL). Fatty acid content was analyzed by gas chromatography of methyl ester derivatives as described previously (47). Lipid A was generated from purified LPS by hydrolysis at 100°C in acetate buffer (pH 4.5) in the presence of 1% sodium dodecyl sulfate (8). Negative-ion MALDI-TOF mass spectrometry of isolated lipid A was performed as described previously (15).
RESULTS
Strains of E. coli expressing P. gingivalis lpxA or lpxD incorporate 16 carbon hydroxy fatty acids into lipid A.It was hypothesized that the cluster of lipid A structures that differ by 14 atomic mass units (amu) found in P. gingivalis LPS preparations were due to a relaxed acyl chain length specificity found in the P. gingivalis acyltransferases lpxA and lpxD (Fig. 1). Therefore, conditional lethal E. coli strains in lpxA (SM101) and lpxD (RL25) that display a temperature-sensitive phenotype were transformed with a pUC19 plasmid containing either the E. coli lpxA gene (plpxAEc ) or the P. gingivalis homolog of lpxA (plpxAPg ). Similarly, E. coli RL25 was transformed with a pUC19 plasmid containing the E. coli lpxD gene (plpxDEc ) or the P. gingivalis homolog of lpxD (plpxDPg ) (see Table 1).
Characterization of P. gingivalis lipid A by negative ion mass spectrometry. Lipid A was cleaved and separated from the LPS as described by Caroff et al. (8). MALDI-TOF mass spectra were determined as previously described (15). P. gingivalis displays lipid A heterogeneity, in that LPS isolated from a single species contains multiple lipid A structures. Kumada et al. (26) elucidated the major structures of several of the major lipid A mass ions. In this figure, one major lipid A “cluster” centered around m/z 1,690 is shown. All values given are average mass rounded to the nearest whole number for singly charged deprotonated molecules. Note that peaks varying by 14 amu may be due to changes in the fatty acid composition.
Strains and constructs
Transformations were performed at the permissive temperature (30°C) and expression was induced at the restrictive temperature (42°C). Successful complementation, measured as the ability to restore a normal growth rate at 42°C of both the lpxA (SM101) and lpxD (RL25) conditional lethal strains, occurred in both the E. coli controls (transformed with plpxAEc and plpxDEc ) and the P. gingivalis homologues (transformed with plpxAPg and plpxDPg ).
LPS was then isolated from these strains and the lipid A structural composition was determined by MALDI-TOF mass spectroscopy and gas chromatographic analysis. As seen in Fig. 2, SM101 transformed with plpxAEc contained a lipid A structural composition that resembled wild-type E. coli in that one major lipid A structure was observed with the expected mass of 1,798 (Fig. 2A). In contrast, SM101 transformed with plpxAPg yielded this peak, and an additional peak at m/z 1,826, indicating the addition of a different long-chain fatty acid (Fig. 2B). Consistent with this, gas chromatographic analysis of the lipid A structures found in SM101 lpxAPg (Fig. 2B) demonstrated the presence of 3-hydroxy-hexadecanoic acid (C16OH) (Fig. 4A). This is consistent with the notion that in some lipid A molecules ester linked 3-hydroxy-tetradecanoic acid (C14OH) had been replaced by the longer 16 carbon fatty acid generating a lipid A with a mass of 1826 (see Fig. 5 for proposed structures).
Negative ion MALDI-TOF mass spectra of lipid A from the E. coli lpxA temperature-sensitive mutant (SM101) with a plasmid carrying either E. coli lpxA (plpxAEc ) (A) or P. gingivalis lpxA (plpxAPg ) (B). The bacteria were cultured at the restrictive temperature 42°C in the presence of 100 mM IPTG.
RL25 transformed with plpxDEc was similar to SM101 plpxAEc in that a single major lipid A structure at m/z 1,798 was observed (Fig. 3A). In contrast, lipid A MALDI-TOF analysis of LPS obtained from RL25 plpxDPg (Fig. 3B) demonstrated negligible wild-type lipid A (m/z 1,798) and revealed the presence of two lipid A structures at m/z 1,826 and at m/z 1,854. In addition, C16OH was found in a greater percentage of the total lipid A fatty acids compared to lpxA transformed cells (Fig. 4B). These data are consistent with the amide positions of the lipid A backbone being substituted with C16OH fatty acids at an incidence of between one and two per molecule. This accounts for the two peaks at m/z 1,826 and m/z 1,854, respectively (see Fig. 5 for proposed structures). In addition, these data indicate that the P. gingivalis lpxD may have a preference for longer-chain fatty acids compared to P. gingivalis lpxA. Since little, if any, wild-type lipid A structure was found, the majority of lipid A structures in this strain were modified.
Negative-ion MALDI-TOF mass spectra of lipid A from the E. coli lpxD temperature-sensitive mutant (RL25) with a plasmid carrying either E. coli lpxD (plpxDEc ) (A) or P. gingivalis lpxD (plpxDPg) (B). The bacteria were cultured at the restrictive temperature 42°C in the presence of 100 mM IPTG.
Fatty acid composition of LPS from E. coli transformed with P. gingivalis lpxA or lpxD mutant. (A) Strain SM101 expressing either E. coli lpxA (▪) or P. gingivalis lpxA (□); (B) strain RL25 expressing either E. coli lpxD (▪) or P. gingivalis lpxD (□). Fatty acid methyl esters were determined by gas chromatography and are expressed in the figure as a percentage of the total fatty acids identified. Determinations were done on bacteria grown on three separate occasions, and typical results are presented.
Proposed lipid A structures for chimeric lipid A generated when E. coli is complemented with plpxAPg or plpxDPg . The sites for the addition of lipid A fatty acids by lpxA and lpxD are indicated. Note that since lpxA and lpxD transfer fatty acids at the monosaccharide stage in lipid A biosynthesis, an asymmetric distribution of fatty acids with different chain lengths can occur. We did not determine whether the C16OH was added to the 2 or 2′ position or to the 3 or 3′ position.
Wild-type E. coli incorporates odd-chain fatty acids into lipid A when grown in the presence of propionic acid.Based upon the structure of P. gingivalis lipid A, the lpxA and lpxD homologues are expected to transfer the odd-chain fatty acids C15OH and C17OH, respectively, to the nascent lipid A (see Fig. 1). Since E. coli does not normally synthesize odd-chain fatty acids, these fatty acids would not normally be present in the fatty acid pools in E. coli. Therefore, to potentially examine a more natural fatty acid substrate pool for lpxAPg and lpxDPg , E. coli was grown in the presence of propionic acid to induce the formation of odd-chain fatty acids (20). Propionic acid has been shown to induce the synthesis of odd-chain fatty acids in E. coli through the formation of propyl CoA in the presence of excess propionic acid (20). This moiety is used for priming fatty acid synthesis rather than acetyl-coenzyme A, resulting in odd-chain fatty acids after elongation (29, 42).
Initially, a wild-type strain of E. coli (JM83) was grown in 1% propionic acid, and the lipid A structural content was determined by MALDI-TOF and gas chromatography. Control experiments demonstrated that this strain grown without propionic acid contained the expected single lipid A structure (bis-phosphorylated, hexa-acylated lipid A), as indicated by MALDI and gas chromatographic analysis (data not shown). LPS obtained from E. coli grown in propionic acid, however, contained a heterogeneous lipid A structural composition (Fig. 6). MALDI-TOF analysis revealed multiple lipid A structures that clustered around a mass of m/z 1,798 and differed by the equivalent of one methylene unit (14 mass units), suggesting that odd carbon fatty acids were being incorporated into the lipid A. Gas chromatographic analysis of LPS fatty acids in this strain (Fig. 9A) confirmed that in addition to the expected even-chain fatty acids, the lipid A also contained the odd-chain fatty acids tridecanoic acid (C13), pentadecanoic acid (C15), hydroxy tridecanoic acid (C13OH), and hydroxy pentadecanoic acid (C15OH).
Negative ion MALDI-TOF mass spectrum of lipid A from wild-type E. coli grown overnight in medium containing 1% propionate. Lipid A was cleaved and separated from the LPS as described by Caroff et al. (8). MALDI-TOF mass spectra were determined as previously described (15). Note that numerous peaks were observed that centered around m/z 1,798 and differed by 14 amu.
Propionic acid alters the fatty acid lipid A composition in E. coli SM101 and RL25 containing lpxA and lpxD.Next, SM101 plpxAEc , Sm101 plpxAPg , RL25 plpxDEc , and RL25 plpxDPg strains were cultured overnight in propionic acid, and the lipid A's were isolated and subjected to analysis by MALDI-TOF mass spectrometry and gas chromatographic analysis. Similar to wild-type E. coli, multiple lipid A structures were observed. Each lipid A structure differed by 14 mass units, indicating that odd-chain fatty acids were incorporated (Fig. 7 and 8). A second cluster of peaks was also observed in some preparations (centered around m/z 2,036) corresponding to hepta-acyl species containing palmitate. Similar hepta-acyl lipid A structures have been reported to be present in an lpxA temperature-sensitive mutant grown at the restrictive temperature and attributed to the action of the inducible acyltransferase CrcA (43).
Negative ion MALDI-TOF mass spectra of lipid A from mutant strains expressing E. coli or P. gingivalis acyltransferases and cultured in medium containing 1% propionic acid. (A) SM101 plpxAEc ; (B) SM101 plpxAPg . Lipid A was cleaved and separated from the LPS as described by Caroff et al. (8). MALDI-TOF mass spectra were determined as previously described (15). Note that numerous peaks were observed that differed by 14 amu.
Negative ion MALDI-TOF mass spectra of lipid A from mutant strains expressing E. coli or P. gingivalis acyltransferases and cultured in medium containing 1% propionic acid. (A) RL25 plpxDEc ; (B) RL25 plpxDPg . Lipid A was cleaved and separated from the LPS as described by Caroff et al. (8). MALDI-TOF mass spectra were determined as previously described (15). Note that numerous peaks were observed that differed by 14 amu.
Gas chromatographic analysis of these lipid A's (Fig. 9B and C) confirmed the presence of odd-chain-length fatty acids. In addition, more C15OH was incorporated into the lipid A when the lpxAPg strain was examined compared to the lpxAEc strain (Fig. 9B). Based upon the structure of the P. gingivalis lipid A, it is expected that C15OH, a natural substrate for lpxAPg , would be more efficiently transferred to lipid A. However, this was not observed for lpxDPg , where little or no C17OH fatty acid, the more natural substrate for lpxDPg , was found in the lipid A (Fig. 9C). The reason for this is not clear, but it likely represents less substrate availability for the longer-chain fatty acids, since it has been shown that the pool size for longer-odd-chain fatty acids are present in the least amount (20, 29).
Fatty acid composition of LPS from mutant strains expressing E. coli or P. gingivalis acyltransferases and cultured in medium containing 1% propionic acid. Fatty acid methyl esters were determined by gas chromatography. (A) Wild-type strain JM83. The data are expressed as a percentage of the total fatty acids identified. (B) Strain SM101 expressing lpxA from either E. coli (▪) or P. gingivalis (□). The data are expressed as a percentage of the total hydroxy fatty acids; (C) Strain RL25 expressing lpxD from either E. coli (▪) or P. gingivalis (□), the data are expressed as a percentage of total hydroxy fatty acids. Determinations were done on bacteria grown on three separate occasions, and typical results are presented.
In addition, it was found that when propionic acid was added to the growth medium, MALDI-TOF analysis revealed a lipid A structure with a mass of 1,798 for RL25 containing plpxDPg . This was not observed in the absence of propionic acid, where nearly all of the lipid A molecules contained either a single or a double substitution with C16OH (compare Fig. 8B to Fig. 3B). This suggests that when the fatty acid pool is altered, such as when odd- and even-chain fatty acids are present, both the number and the types of fatty acids can influence what will be transferred to lipid A.
DISCUSSION
The biosynthesis of lipid A is an ordered, tightly regulated process in gram-negative bacteria (37, 39). Although the structure and synthesis of the lipid A of E. coli has been studied most thoroughly, the lipid A of other unrelated bacteria are now also coming under scrutiny. Relative to E. coli lipid A, lipid A from other bacteria maintains the same basic structure. However, variations are found in the number, type, and placement of fatty acids and the number and substitutions of the phosphates. Lipid A synthesis in these bacteria proceeds through the same basic synthetic pathway as characterized in E. coli. Observed differences in the lipid A structure are explained either by differences in substrate specificity of the synthetic enzymes, resulting in the addition of different components, or by the presence of additional lipid A modifying enzymes that add or remove components auxiliary to normal synthesis (38, 49, 52). Studies of the substrate specificity of the lipid A acyltransferases LpxA and LpxD indicate that the length of the fatty acid chains present in lipid A is determined by the specificity of the acyltransferases for a particular chain length. These enzymes also contain a hydrocarbon ruler that “measures” fatty acid size (7, 30, 56-58).
The LpxA and LpxD acyltransferases of E. coli have exquisite specificity in catalyzing the incorporation of only one type of fatty acid (C14OH). In contrast, the lpxA homologue, found in Bordetella sp., has been found to have a relaxed fatty acid acyl chain specificity facilitating the transfer of fatty acids which differ by two methylene units (49). In the present study, the possibility that the P. gingivalis lpxA and lpxD homologues also demonstrated a relaxed fatty acid acyl chain length specificity which contributed to the clusters of lipid A structures that differ by one methylene unit (14 amu), which have been observed by us (11, 38) and others (25, 59), was examined. An alternate possibility that P. gingivalis may contain multiple copies of the lpxA or lpxD genes with differing fatty acid chain length specificities was not consistent with a search of the published P. gingivalis genome, which revealed only a single copy of each homologue. The ability of the P. gingivalis lpxA and lpxD homologues to facilitate the incorporation of both C14OH and C16OH fatty acids into E. coli lipid A demonstrated that they had a relaxed acyl chain specificity. Thus, we suggest that lpxA and lpxD contribute to the heterogeneity as a result of a relaxed acyl chain specificity that allowed them to transfer a small range of fatty acid chain lengths. Expression of P. gingivalis lpxA and lpxD in E. coli indicated that the P. gingivalis transferases exhibited a preference for longer-chain fatty acids over those transferred by the E. coli forms. In addition, the shorter-chain β-hydroxymyristates found in E. coli lipid A were also incorporated, suggesting that the P. gingivalis enzymes have a sufficiently relaxed acyl chain specificity to account for the chain length heterogeneity observed in P. gingivalis lipid A. These data are consistent with the notion that the additional lipid A mass ions found in P. gingivalis lipid A are the result of an altered distribution of the fatty acids generating different lipid A structures and provide a mechanism that explains the appearance of lipid A clusters observed in P. gingivalis mass spectra (see Fig. 1, for example).
Utilization of propionic acid as a substrate for fatty acid synthesis, where odd carbon fatty acids were synthesized in E. coli (20), revealed that the available fatty acid substrates may also contribute lipid A structural heterogeneity. For example, we demonstrate here that, in vivo, E. coli lpxA and lpxD can transfer fatty acids differing by a single carbon unit (i.e., 13 carbon and 15 carbon hydroxy fatty acids resulting in a lipid A with a high degree of fatty acid heterogeneity). The observation that wild-type E. coli (see Fig. 6) incorporated odd-chain fatty acids, containing wild-type levels of LpxA and LpxD lipid A biosynthetic enzymes, demonstrates that the availability of odd-chain fatty acid substrates can significantly affect the lipid A fatty acid composition. These results indicate that the apparent strict specificity of E. coli lipid A acyltransferases LpxA and LpxD is actually the sum of the relatively high specificity of the transferases combined with the lack of 13 and 15 carbon substrates, respectively, in the cellular fatty acid pools. Furthermore, comparison of the fatty acids incorporated into lipid A when LpxA and LpxD from E. coli and P. gingivalis were compared revealed that the average chain lengths incorporated by the P. gingivalis enzymes were larger since it was found that the P. gingivalis enzymes incorporated more C15OH and C16OH fatty acids (see Fig. 9). It is interesting that, due to the wide range of lipid A structures observed when E. coli was grown with 1% propionic acid, it is not clear whether P. gingivalis LpxA has a more relaxed chain specificity than the corresponding E. coli enzyme. For example, if P. gingivalis LpxA transfers an optimum fatty acid chain length of 15 carbons, which is consistent with the fatty acid data found in E. coli complemented with P. gingivalis lpxA grown either with or without propionate, then the P. gingivalis lpxA is similar to E. coli lpxA in that they both can transfer fatty acids that differ by one carbon unit (E. coli LpxA C14 ± 1C, yielding C13 and C15; P. gingivalis LpxA C15 ± 1C, yielding C14 and C16). Considering that both E. coli and P. gingivalis LpxA and LpxD can transfer odd-chain fatty acids to lipid A, in vitro assays to determine the precise fatty acid chain length optima will be required to reexamine the extent of relaxed acyl chain specificities and hence the “hydrocarbon ruler” motif for these enzymes.
The biological significance of P. gingivalis containing a heterogeneous mix of lipid A structural type that differs by minor changes in the fatty acid composition remains to be determined. This bacterium, and other gram-negative oral anaerobes (data not shown), may utilize this mix of lipid A structural types to confer optimal structural integrity to their outer membranes. However, the effects of different fatty acid chain lengths may also have significant effects on the innate host response. For example, we have previously demonstrated that E. coli containing the P. gingivalis htrB acyltransferase gene contains a lipid A structure that is altered by one fatty acid (C16 instead of a C12 is incorporated) and induces a significantly less potent interleukin-8 response (4). It is possible, therefore, that alteration in fatty acid substrate pools represents a novel mechanism by which bacteria may respond to local host microenvironments in order to evade or manipulate the innate host response.
FOOTNOTES
- Received 15 February 2008.
- Accepted 22 April 2008.
- Copyright © 2008 American Society for Microbiology
