Mutations in firA, encoding the second acyltransferase in lipopolysaccharide biosynthesis, affect multiple steps in lipopolysaccharide biosynthesis

The product of the firA (ssc) gene is essential for growth and for the integrity of the outer membrane of Escherichia coli and Salmonella typhimurium. Recently, Kelly and coworkers (T. M. Kelly, S. A. Stachula, C. R. H. Raetz, and M. S. Anderson, J. Biol. Chem., 268:19866-19874, 1993) identified firA as the gene encoding UDP-3-O-(R-3-hydroxymyristoyl)-glucosamine N-acyltransferase, the third step in lipid A biosynthesis. We studied the effects of six different mutations in firA on lipopolysaccharide synthesis. All of the firA mutants of both E. coli and S. typhimurium examined had a decreased lipopolysaccharide synthesis rate. E. coli and S. typhimurium strains defective in firA produced a lipid A that contains a seventh fatty acid, a hexadecanoic acid, when grown at the nonpermissive temperature. Analysis of the enzymatic activity of other enzymes involved in lipid A biosynthesis revealed that the firA mutations pleiotropically affect lipopolysaccharide biosynthesis. In addition to that of UDP-3-O-(R-3-hydroxymyristoyl)-glucosamine N-acyltransferase, the enzymatic activity of the lipid A 4' kinase (the sixth step of lipid A biosynthesis) was decreased in strains with each of the firA mutations examined. However, overproduction of FirA was not accompanied by overexpression of the lipid A 4' kinase.

The firA gene of Escherichia coli and the analogous gene in Salmonella typhimurium have been shown to be essential for growth. The product of firA is important for maintaining resistance to hydrophobic antibiotics such as rifampin (19,27). The outer membrane of gram-negative bacteria functions as a permeability barrier to hydrophobic antibiotics (30). For this reason, mutations affecting the synthesis of the outer membrane cause the cells to become hypersensitive to hydrophobic antibiotics (30). Recently, Kelly et al. (24) have demonstrated that firA encodes UDP-3-0-(R-3-hydroxymyristoyl)glucosamine N-acyltransferase, the third committed step in lipid A biosynthesis. Studies with S. typhimurium indicate that the lipopolysaccharide produced in the firA mutant contains a seventh fatty acid, a hexadecanoic acid linked to the N-linked 3-hydroxymyristate on the reducing end sugar (18). A mutation in lpxB (lipid A disaccharide synthase; see Fig. 1) also causes accumulation of the lipopolysaccharide precursor 2,3diacyl-N-acetylglucosamine-1-phosphate (lipid X), some of which contains a hexadecanoic acid in the analogous position (termed lipid Y) (31). Lipid Y is not a substrate for the lipid A disaccharide synthase for further incorporation into lipopolysaccharide (23). 3-Deoxy-D-manno-octulosonate (KDO) cytidyltransferase activates KDO for the addition onto the lipid A precursor tetraacyl-disaccharide-1,4'-bis-phosphate (Fig. 1). Both E. coli and S. typhimurium grown in the presence of an inhibitor of this enzyme accumulate the lipid A precursor tetraacyl-disaccharide-1,4'-bis-phosphate (16). In S. typhimurium, much of this precursor tetraacyl-disaccharide-1,4'-bisphosphate contains a hexadecanoate in a position analogous to that found in the lipopolysaccharide of the firA mutant strain of S. typhimurium (36). However, E. coli treated with the same drug does not significantly accumulate the hexadecanoic acid-modified precursor of lipid A (16). In the present study, we identified the hexadecanoic acid-modified mature lipopolysaccharide in E. coli deficient in firA. The presence of heptaacyl lipid A in the lipopolysaccharide of firA-defective strains may indicate a block in lipid A biosynthesis or assembly. This block could cause accumulation of a precursor that can be modified with hexadecanoic acid and incorporated into the mature lipopolysaccharide. Alternatively, hexadecanoic acid acyltransferase activity may be increased by the stress of reduced lipopolysaccharide biosynthesis.
A defect in the lipid A 4' kinase is expected to cause accumulation of a precursor in the cytoplasm that could be modified by hexadecanoic acid acyltransferase. In the present study, we found that all mutations in firA, the gene for UDP-3-0-acyl-glucosamine acyltransferase, reduce the activity of the lipid A 4' kinase. However, overproduction of firA did not alter lipid A 4' kinase activity. Dicker and Seetharam (12) have proposed that FirA is important for RNA transcription. One mutation in the gene for an RNA polymerase subunit, rpoB, causes a synergistic increase in the temperature-sensitive phenotype of a firA mutation. In addition, FirA coprecipitates with RNA polymerase in immunoprecipitations of RNA polymerase from a strain overproducing FirA (12). Dicker and Seetharam (12) posited a catalytic requirement for FirA under certain conditions, as has been reported for w protein, a protein thought to be physically involved in the modification of polymerase during the stringent response (22). The very low level of FirA in vivo is consistent with such a catalytic role. An interaction of the FirA protein with RNA polymerase could explain the pleiotropic effects of mutations in firA.

MATERIALS AND METHODS
Bacteria, plasmids, and media. The bacterial strains used are described in Table 1. Strain JCR4 was created by integration of pGAH317 (21), with a 1.5-kbp kanamycin resistance  (35). Genes 1pxA, lpxB,firA, and kdtA, which are known to be involved in lipid A biosynthesis, are indicated. ACP, acyl carrier protein; GlcNAc, N-acetylglucosamine; GlcN, glucosamine. cassette inserted in a PstI site in hipA, into the chromosome of RL436 at 43°C (see Fig. 2). The temperature-sensitive mutation is in the upstream end of open reading frame 17 in RL436 (lOa). pGAH317 does not contain the downstream portion of open reading frame 17; therefore, growth at 43°C is possible only after recombination of the plasmid into the chromosome. P1 transduction (38) was used to move the hlpA::Kanr insertion into MF6R (dapD) with selection for the nearby dapD+ gene and kanamycin resistance. At this point, the strain has two copies of firA, one from the plasmid and one from the chromosome. High concentrations of ampicillin enrich for cells that have excised the plasmid from the chromosome, to allow high-copy-number replication of the plasmid carrying the ampicillin resistance gene, bla. Excision of the plasmid leaves a single copy of hlpA and firA in the chromosome, with a 50% chance of having the kanamycin resistance insertion in the chromosome. A P1 lysate was made on this strain with the kanamycin resistance cassette left in the chromosome. This mutation was transduced into MF6R/pMTfir with selection for dapD+ and kanamycin resistance. The resulting strain was checked for loss of ampicillin resistance. The recA56 mutation was introduced into the strain by Hfr mating with JC10240, selection for tetracycline resistance, and screening for UV sensitivity. The presence of a single copy of hipA-firA with the Kanr insertion in the chromosome of JCR4 was confirmed by Southern blotting (40).
The E. coli chromosomal portions of the plasmids used in this study are indicated in Fig. 2. pGAH317 has been described previously (21). pRsa contains an RsaI fragment encoding all of firA inserted into the SmaI site of pTZ18R (United States Biochemical Corporation). Plasmid pMTfir has the HindIll fragment containing firA from E. coli inserted into the HindIll site of vector pMT101 (17), which has a temperature-sensitive replicon. Plasmid pID2 (a kind gift from Ira Dicker) has been described previously (11). Plasmid pUCHS14-II contains the S. typhimurium firA gene and has been described previously (19).
Materials. Restriction enzymes were purchased from New England BioLabs, Beverly, Mass. Except as noted otherwise, chemicals were from Sigma, St. Louis, Mo. 32p; was purchased from ICN, Irvine, Calif. UDP-N-acetyl[3H]glucosamine and En3Hance were from New England Nuclear. Kodak SB film was used for autoradiography. Glass-backed silica gel 60 thin-layer chromatography plates were from E. Merck, Darmstadt, Germany.
Rate of lipopolysaccharide biosynthesis. The rate of lipopolysaccharide synthesis was measured as previously described (14). Cells were grown in L broth at 30°C and then switched to 43°C for 1 h (8 h for JCR4) prior to pulse labeling with 32P, for 5 min. Quantitation of 32p; incorporation into lipopolysaccharide and glycerophospholipids (14) was done with a Molecular Dynamics (Sunnyvale, Calif.) Phosphorimager after exposure to Kodak storage phosphor screens. Data analysis employed the software supplied with the Phosphorimager in the Louisiana State University Biotechnologies Core Laboratory.
Preparation of cell extracts and enzyme assays. Cell extracts for enzyme assays were prepared from stationary-phase cells grown at 30°C (43°C for 8 h for JCR4 and its control) in L broth. Cells were harvested by centrifugation at 5,000 x g for 10 min at 4°C. Cells were resuspended in 10 mM potassium phosphate (pH 7.0) and broken by passage through a French press at 18,000 lb/in2. Unbroken cells were removed by centrifugation at 5,000 x g for 5 min. Membranes were removed by centrifugation at 4°C for 60 min at 210,000 x g. The membrane fraction was suspended in Tris-HCl (pH 8.0) for use in the lipid A 4' kinase assay. Protein concentration was determined by the bicinchoninic acid (Pierce Chemical Co., Rockford, Ill.) method of Smith et al. (39). Protein concentrations were made equal prior to enzyme assays.
The early steps of lipid A biosynthesis were assayed as described by Anderson and Raetz (1) by using UDP-Nacetyl[3H]glucosamine or UDP-3-O-acyl[3H]glucosamine as the initial substrate. Products were separated by using silica gel thin-layer chromatography developed with chloroform-methanol-acetic acid-water (25:15:4:2, vol/vol). Products were identified by comigration with known standards.
The lipid A disaccharide synthase was assayed as described by Crowell and coworkers (9). Quantitation was done with a Molecular Dynamics Phosphorimager.
The lipid A 4' kinase was assayed as described by Ray and Raetz (37), with [32P]tetraacyl-disaccharide-monophosphate  (32). The lower phase was washed with 0.1 N HClmethanol (1:2) and evaporated to dryness under nitrogen. This n for 8 h at 43°C in L was suspended in chloroform-methanol (3:1). One hundredth rifugation at 4,000 x of a volume of 1 M ammonia was added to raise the pH. This sample was used directly for mass spectral analysis.
Electrospray mass spectra. Electrospray mass spectra were generated with a model 201 single-quadrapole mass spectrophotometer (Vestec Corporation, Houston, Tex.) that was dnaE A fitted with a 10-kV conversion dynode, had a 2,000 m/z mass range, and was located in the Louisiana State University Biotechnologies Core Laboratory. Sample mixtures were infused by syringe pump (Orion Research) at a flow rate of 6 ml/min, with spectra acquired as single 5-s scans. The source block was heated to 250°C, the spraying chamber was at 55°C, the repeller voltage was 18 V, and the nozzle voltage was set at region of the E. co/i 230 V. Calibration of the mass spectrophotometer was perycin resistance cassette formed prior to sample analysis, with 1-decanesulfonic acid, cent map of E. coli (2). sodium salt. regions of the chromo-Each spectrum is the average of more than 20 scans. High smid pMTfir is temper-values (noise) were ignored if they differed by greater than 5 standard deviations from the mean for that m/z.  (14), and radioactivity in each was determined. Values are expres of lipopolysaccharide to glycerophospholipids, normal genic host strain for each sample. firA66 is from strain is from JCR20,firA53 is from JCR53,firA111 is from St is the firA::Kanr mutation in strain JCR4. All strains 43°C for 1 h prior to labeling (except JCR4, which wa at 43°C prior to labeling).

RESULTS
Lipopolysaccharide production in firA mutai mutagenesis of the 4-min region of the E. coli produced a number of temperature-sensitive alle 26). The firA gene from temperature-sensitive s1 mented with a plasmid that produces only FirA M into strain MF6R to make isogenic strains (' measured the rates of lipopolysaccharide biosyn ous E. coli and S. typhimurium strains by measur incorporation of radiolabeled Pi into lipopolysa Figure 3 shows that cells with every mutant allele had a reduced rate of lipopolysaccharide bios pared with wild-type levels. The rate of lipor biosynthesis was normalized to the rate of glycer biosynthesis (14). We obtained similar results wi protein biosynthesis was used for normalizati shown). The most severe inhibition of lipor biosynthesis was in strain JCR4. JCR4 produced the chromosome because of an insertion in th upstream of the FirA coding sequence (Fig. 2) W' tial gene lpx4 in JCR4, a second promoter must exist between firA and 1pxA.
The above-described results confirm the finding of Kelly and coworkers (24) that firA is involved in lipopolysaccharide biosynthesis, but the results do not indicate whether firA is involved in multiple steps of lipid A biosynthesis.
strain produced a more slowly migrating lipopolysaccharide. A and glycero-The E. coli firA mutant strain also produced a lipopolysacchathe amount of ride with altered mobility. However, less than 50% of the E. ssed as the ratio coli lipopolysaccharide migrated at the slower rate in SDSlized to the iso-PAGE.
i SM66, firA200 The slower migration of the lipopolysaccharide of E. colifirA vIlll, and AfirA mutant strains may be due to an additional fatty acid on the were grown at lipid A. Alternatively, multiple other modifications of lipopolyts grown for 8 h saccharide have been identified, such as additional phosphates or phosphorylethanolamine (35). Electrospray mass spectrometry of lipid A preparations demonstrates the existence of heptaacyl lipid A in E. coli. Electrospray mass spectral analysis of the lipid A produced from E. coli firA' (JCR4/pGAH317) and the firA mutant nts. Localized (JCR4) indicates production of heptaacyl lipid A by JCR4 (Fig.   chromosome  5). The lipid A for mass spectral analysis was isolated by using les offirA (14, mild conditions (1% acetic acid), which are less likely to trains compleremove the acyl-oxy-acyl fatty acids than are the harsher ras transduced hydrolysis conditions (0.2 N HCl) used previously (14,18). This Table 1). We method of isolation of lipid A does hydrolyze some of the ithesis in varianomeric phosphate (Fig. 5). Both the firA+ and firA mutant ing the rate of strains contain hexaacyl lipid A with a -1 (1798) or -2 (899) iccharide (14). charge (Fig. 5). In addition, both samples have the hexaacyl offirA tested lipid A that has lost one of the phosphates. Along with these ,ynthesis com-bands, the firA mutant lipid A also contains molecules with a )olysaccharide molecular mass corresponding to the addition of a hexaophospholipid decanoate. This heptaacyl form is found with a -2 (1018) hen the rate of charge and a -1 charge with loss of a phosphate (1957). We ion (data not did not detect pentaacyl forms of lipid A in these E. coli strains.
polysaccharide These results prove that E. coli firA mutant strains produce no FirA from a heptaacylated lipid A. The identification of a lipid A molete firA operon cule lacking a phosphate is most likely an artifact of the JCR4 contains isolation procedure. Therefore, these results do not indicate or sitive replicon rule out the presence of an underphosphorylated lipid A, as 3°C only when would be expected from lowered production of the 4' kinase. normal FirA The above-described results and those of Helander and tent the essen-coworkers (18) indicate that a mutation infirA causes produc- tion of a heptaacyl lipopolysaccharide. Other mutations in lipid A biosynthesis that result in accumulation of lipid A precursors on the cytoplasmic side of the inner membrane accumulate modified precursors with the extra fatty acid hexadecanoic acid (31,36). However, the hexadecanoic acid-containing precursor is not a substrate for the disaccharide synthase, the fifth step of lipid A biosynthesis. For this reason, lipid A precursors modified with a hexadecanoic acid prior to the disaccharide synthase (step 5 in Fig. 1) would not be expected to be incorporated into the lipopolysaccharide. Alternatively, the hexadecanoyl acyltransferase may be activated under conditions of stress. A specific assay of the disaccharide synthase indicated that the activity of this enzyme was unaltered in all of the firA alleles tested. The lipid A 4' kinase is defective in allfirA mutant strains. The lipid A 4' kinase adds a phosphate to the 4' position of tetraacyl-disaccharide-1-phosphate. In E. coli, this step is required for addition of the first core sugar, KDO (4). We determined the lipid A 4' kinase activities in strains with different alleles of firA ( Fig. 6 and 7). Strains with every temperature-sensitive allele of firA (firA66, firA200, firA53, firAll], AfirA, and ssc) had defective 4' kinase activity; however, strains overexpressingfirA (JMfir/pGAH317, JMfir/pID2, and SH7622/pUCSH14-II) did not overexpress this activity. It should be noted that the firA66 mutation reduced the level of UDP-3-O-acylglucosamine acyltransferase to less than 1% of wild-type levels (data not shown). The activity of the lipid A 4' kinase may be lowered because the firA mutant strains produce a substance that inhibits the lipid A 4' kinase. If the decrease in lipid A 4' kinase activity is due to an inhibitor being produced in firA mutant strains, the activity of the 4' kinase of a wild-type strain would be expected to be reduced if mixed with the potential inhibitor in vitro. When the membrane fraction from a firA+ strain (SH5014) was mixed with the membrane of a firA mutant strain (SH7622), the resulting lipid A 4' kinase activity was equal to the sum of the enzyme activities from both strains. This suggests that the loss of activity was not due to an inhibitor of the lipid A 4' kinase found in the firA mutant cells.
Note that S. typhimurium strains produced additional products derived from tetraacyl-disaccharide-1-phosphate (Fig. 6). These additional lipids probably contain an additional hexadecanoic acid on the C-2 ,B-hydroxymyristate, as has been observed previously (6,36). Helander and coworkers (18) have shown that the lipopolysaccharide produced from the S. typhimurium firA mutant strain contains this additional fatty acid. As seen in Fig. 6, the apparent activity of this hexadecanoate acyltransferase was not increased in the firA mutant strain.
The half-life of the lipid A 4' kinase is unaffected in firA mutant strains. One possible explanation for the decrease in 4' kinase activity is a reduction in 4' kinase enzyme stability in firA mutant strains. To determine whether this is the case, we incubated the membrane preparations (diluted to 5 mg/ml) at 42°C for 10, 60, 120, and 180 min. After incubation at 42°C, the activity of the 4' kinase was assayed as described previously. These experiments demonstrated that for S. typhimurium SH5014 (firA+) and SH7622 (firA mutant), the half-life of the lipid A 4' kinase is unaffected by the firA mutation. Both strains possess a lipid A 4' kinase with an in vitro half-life of 4.8 h at 42°C. firA strains are hypersensitive to novobiocin. Defects in the synthesis of the outer membrane cause the cell to become hypersensitive to hydrophobic antibiotics such as rifampin and novobiocin (30,44). The activity of the lipid A 4' kinase does not correlate with the rate of synthesis of lipopolysaccharide in [32P]tetraacyl-disaccharide monophosphate was incubated with membrane fractions as described in the legend to Fig. 6. The number of units of lipid A 4' kinase (nanomoles per minute per milligram of protein) was determined for each sample and then normalized to the number determined for the isogenic wild-type strain (the kinase value for the firA+ strains is 1.00 by definition). Standard deviations are indicated. Plasmids pID2 (E. coli) and pUCHS14-II (S. typhimurium) overexpress firA, and plasmid pGAH317 overexpresses bothfirA and upstream gene hlpA. firA66 is from JMfir,firA200 is from JCR20, firA53 is from JCR53, firA1 is from JCR111, ssc is the firA mutation in S. typhimurium SH7622, and AfirA is the firA::Kanr mutation in JCR4 (after growth at 43°C for 8 h).
the different firA mutant strains (compare Fig. 3 and 7). To determine whether the sensitivity of firA mutant strains to hydrophobic antibiotics correlates with lipid A 4' kinase activity or lipopolysaccharide biosynthesis, we determined, by the method of Bauer et al. (3), the relative resistance of each strain to novobiocin. Figure 8 shows the zone of clearing, relative to that of the isogenic firA+ strain, for each of the firA alleles. This demonstrates that resistance to hydrophobic antibiotics is directly related to the rate of lipopolysaccharide biosynthesis, with the notable exception of the S. typhimurium firA mutant (compare Fig. 3 and 8). The greater resistance of the S. typhimurium firA mutant strain may be due to the greater production of heptaacyl lipopolysaccharide. The strain background did not significantly alter the results of the antibiotic sensitivity assays when the results were expressed relative to the wild type. These data also demonstrate that mutations in two other genes known to be involved in lipopolysaccharide biosynthesis, lpxA and lpxB, increase sensitivity to novobiocin.  8. Sensitivity of firA mutant strains to novobiocin. Each zone of inhibition was measured in millimeters from the edge of the novobiocin-containing filter disc to the edge of the clearing zone. All values are expressed as the ratio of the zone of inhibition of the given strain to that of the isogenic wild-type strain. The values shown are averages of at least four determinations. firA66 is from SM66, firA200 is from JCR20, firA53 is from JCR53, firAIll is from JCR111 and SM111, ssc is thefirA mutation in S. typhimurium SH7622, lpxA is from SM101, and lpxB is from MN7.

DISCUSSION
The data presented in this study indicate that mutations in firA affect multiple steps in the synthesis and assembly of lipopolysaccharide. Although firA encodes UDP-3-0-acyl-GlcN acyltransferase (24), the firA mutants accumulate a heptaacyl lipid A at the nonpermissive temperature for growth. This is reminiscent of other lipid A biosynthesis mutants (6,36). In these cases, a lipid A precursor remains on the cytoplasmic side of the inner membrane for a prolonged time.
The enzyme responsible for the addition of a hexadecanoic acid to the lipid A precursor is most likely localized on the cytoplasmic surface of the inner membrane, where it modifies the accumulated precursor (6). Addition of a hexadecanoic acid to the lipid A of the firA mutants indicates the retention of a precursor in the cytoplasm for modification. Because the lipid A disaccharide synthase cannot use a hexadecanoic acid-modified substrate (34), this modification in the firA mutants must occur after this step (Fig. 1).
Clementz and Raetz (7) mapped and cloned the gene for the KDO transferase, kdtA (Fig. 1). kdtA maps to min 81 on the E. coli chromosome. The firA mutation did not significantly affect KDO transferase activity (data not shown). The only other enzymes required for synthesis of a viable lipopolysaccharide are the lauroyl and myristoyl lipid A acyltransferases and the lipid A 4' kinase. Qureshi and coworkers (33) have demonstrated the presence of a pentaacyl lipid A assembled in deep rough mutants of E. coli. Our analysis does not detect any lipid A lacking either a lauroyl or a myristoyl group. These results indicate that firA mutation does not cause a significant decrease in the activity of either of these two acyltransferases.
FirA may affect the transport of the lipid A molecule through the inner membrane. If this is the case, accumulation of lipopolysaccharide precursors in the cytoplasm may repress the production of lipid A biosynthetic enzymes. Retention of lipopolysaccharide on the cytoplasmic surface of the inner membrane would allow it to be a substrate for the lipid A hexadecanoic acid acyltransferase.
Our results show that the activity of the lipid A 4' kinase is decreased in all of the firA mutations studied, in both E. coli and S. typhimurium (Fig. 7). Overproduction of the firA gene product, UDP-3-0-acyl-GlcN acyltransferase, does not increase the activity of the lipid A 4' kinase. This indicates that the decrease of lipid A 4' kinase activity is a secondary effect of the altered UDP-3-0-acyl-GlcN acyltransferase. UDP-3-0acyl-GlcN acyltransferase may help to stabilize the active conformation of the lipid A 4' kinase, or the firA mutations may cause repression of the gene for the lipid A 4' kinase. If FirA does help to stabilize the active conformation of the lipid A 4' kinase, the effect of the stabilization is not seen in the in vitro assays. FirA may interact with RNA polymerase to regulate the transcription of certain genes, as proposed by Dicker and Seetharam (12). The present studies do not answer the question of what causes the decrease in lipid A 4' kinase activity. This question may be more easily answered once the lipid A 4' kinase gene or protein has been identified. Reduced lipid A 4' kinase activity may cause accumulation of a lipopolysaccharide precursor on the cytoplasmic side of the inner membrane that is a substrate for the lipid A hexadecanoic acid acyltransferase. Alternatively, the presence of heptaacyl lipid A may be a secondary result of some other stress on the cell in firA-defective strains.
firA mutations cause a much greater defect in UDP-3-0acyl-GlcN acyltransferase activity than in lipid A 4' kinase activity. For this reason, the precursor of the lipid A 4' kinase, tetraacyl-disaccharide-1-phosphate, did not significantly accumulate in firA mutant cells at the nonpermissive temperature (data not shown).
The reduced rate of lipopolysaccharide biosynthesis at the nonpermissive temperature in firA mutant strains can explain the lethality of these mutations and the supersensitivity to hydrophobic antibiotics. The percentages of lipopolysaccharide biosynthesis reduction in severefirA mutants (84% reduction) and severe lpxA mutants (90% reduction) (14) are similar. The product of the lpxA gene catalyzes the first step in lipopolysaccharide biosynthesis (Fig. 1). Cells with the lpx;A and lpxB mutations share with firA strains the hydrophobicantibiotics-supersensitive phenotype (Fig. 8) (44). Accumulation of the heptaacyl lipopolysaccharide may also contribute to the temperature-sensitive phenotype offirA mutant strains by altering the conformation of the lipid A (25). The presence of heptaacyl lipid A is not likely to be the sole cause of the lethal phenotype, since S. minnesota is able to survive at 42°C with 70% heptaacyl lipid A (13,18).
The presence of heptaacyl lipopolysaccharide may actually increase the resistance of bacteria to hydrophobic antibiotics.
Although firA mutant S. typhimurium SH7622 had a much greater decrease in lipopolysaccharide biosynthesis than the corresponding E. coli strains (Fig. 3), SH7622 was more resistant to the hydrophobic antibiotic novobiocin (Fig. 8). The difference in antibiotic susceptibility may be explained by the larger percentage of heptaacyl lipopolysaccharide found in the S. typhimurium firA mutant strain than in E. coli firA mutant strains (Fig. 5).
This work demonstrates that E. coli is capable of synthesizing heptaacyl lipopolysaccharide and that FirA is important for more than one step in the synthesis of lipid A. Ongoing work in our laboratory will better define the precise roles of FirA in lipopolysaccharide synthesis and assembly.