Isolation and Characterization of β-Ketoacyl-Acyl Carrier Protein Reductase (fabG) Mutants of Escherichia coli and Salmonella enterica Serovar Typhimurium

Abstract

FabG, β-ketoacyl-acyl carrier protein (ACP) reductase, performs the NADPH-dependent reduction of β-ketoacyl-ACP substrates to β-hydroxyacyl-ACP products, the first reductive step in the elongation cycle of fatty acid biosynthesis. We report the first documented fabG mutants and their characterization. By chemical mutagenesis followed by a tritium suicide procedure, we obtained three conditionally lethal temperature-sensitive fabG mutants. The Escherichia coli [fabG (Ts)] mutant contains two point mutations: A154T and E233K. The β-ketoacyl-ACP reductase activity of this mutant was extremely thermolabile, and the rate of fatty acid synthesis measured in vivo was inhibited upon shift to the nonpermissive temperature. Moreover, synthesis of the acyl-ACP intermediates of the pathway was inhibited upon shift of mutant cultures to the nonpermissive temperature, indicating blockage of the synthetic cycle. Similar results were observed for in vitro fatty acid synthesis. Complementation analysis revealed that only the E233K mutation was required to give the temperature-sensitive growth phenotype. In the two Salmonella enterica serovar Typhimurium fabG(Ts) mutants one strain had a single point mutation, S224F, whereas the second strain contained two mutations (M125I and A223T). All of the altered residues of the FabG mutant proteins are located on or near the twofold axes of symmetry at the dimer interfaces in this homotetrameric protein, suggesting that the quaternary structures of the mutant FabG proteins may be disrupted at the nonpermissive temperature.

β-Ketoacyl-acyl carrier protein (ACP) reductase catalyzes the first of the two reduction steps in the elongation cycle of fatty acid synthesis (1, 16, 29, 30, 38, 41). In the type II fatty acid synthetic pathway the intermediates are covalently linked by a thioester bond to the prosthetic group of ACP (22, 32). In Escherichia coli the elongation cycle begins with a Claisen condensation reaction catalyzed by one of the three β-ketoacyl-ACP synthases (FabB, FabF, or FabH) that adds two carbons to the C_n acyl chain of an acyl-ACP. The resulting C_n+2 β-ketoacyl-ACP is reduced by the NADPH-dependent β-ketoacyl-ACP reductase to yield a β-hydroxyacyl-ACP which is then dehydrated by a β-hydroxyacyl-ACP dehydrase (either FabA or FabZ) to produce enoyl-ACP. Finally, an NADH-dependent enoyl-ACP reductase (FabI) reduces enoyl-ACP to give a C_n+2 acyl-ACP, the substrate for the next elongation cycle. Elongation ceases when the acyl-ACP attains the chain length required for acylation of phospholipid or lipid A precursors (22, 32).

Rawlings and Cronan (30) isolated a cluster of fatty acid synthetic genes from E. coli genomic DNA and found that a centrally located and highly expressed gene within the cluster had a high degree of sequence identity to several acetoacetyl coenzyme A (CoA) reductases. This gene was proposed to encode a β-ketoacyl-ACP reductase and was called fabG. Heath and Rock (15) then demonstrated that FabG had the predicted enzyme activity and was active on chain lengths from C₄ to C₁₄ in a reconstituted in vitro fatty acid synthetic system. Zhang and Cronan (41) by an indirect transcriptional analysis showed that fabG is essential for the growth of E. coli, but the strains constructed were not appropriate for physiological studies. Strains with fabG mutations suitable for physiological studies are needed since FabG is a member of a very large family of enzymes, the short-chain alcohol dehydrogenase/reductase (SDR) family (29, 30). SDR enzymes carry out a wide variety of reduction and dehydrogenase reactions using NADH or NADPH. Therefore, annotation of fabG genes can be problematical and various annotators of the E. coli and other bacterial genomes have assigned β-ketoacyl-ACP reductase activity to open reading frames that seem unlikely to posses this enzyme activity. Moreover, other bacteria contain enzymes in addition to FabG that have β-ketoacyl-ACP reductase activity. These enzymes are thought to play important roles in other synthetic pathways. For example, the Pseudomonas aeruginosa rhlG gene encodes an NADPH-dependent β-ketoacyl-ACP reductase that is involved in the synthesis of rhamnolipids and poly-β-hydroxyalkanoates (8). The NodG nodulation protein of Rhizobium sp. N33 has β-ketoacyl-ACP reductase activity in vitro (21). A further complication is that FabG provides precursors for medium-chain-length poly-β-hydroxyalkanoate biosynthesis (27, 31) and 3-oxo-homoserine lactone synthesis (17).

Therefore, it would seem advantageous to have strains available in which putative fabG genes can be tested for activity in fatty acid synthesis. For example in Lactococcus lactis subsp. lactis, two fabG homologues were reported (6). Therefore, suitable strains to test putative fabG genes by genetic complementation are needed. Although a putative P. aeruginosa fabG(Ts) mutant was utilized in investigations of 3-oxohomoserine lactone synthesis (17), this strain was obtained from another laboratory and no validating genetic, physiological, or enzymological data have been published. Moreover, the mutant was isolated from a pathogenic strain, which would limit its distribution and use. The fabG(Ts) mutants isolated in the present work should serve as the means to test functions of genes from this and other organisms in order to validate the physiological relevance of the annotations. Such validation would be valuable not only for bacteria and plants but also in studies of several protozoa that cause diseases such as malaria, toxoplasmosis, and sleeping sickness. These organisms retain an essential type II fatty acid synthetic pathway thought to be descended from an algal symbiont (39). For these reasons and in order to more definitely establish the role of fabG in type II fatty acid biosynthesis, we have isolated and characterized temperature-sensitive fabG mutants of E. coli and Salmonella enterica serovar Typhimurium. Our data indicate that FabG is responsible for all β-ketoacyl-ACP reduction in the fatty acid synthetic pathway. Moreover, analyses of the fatty acid synthetic intermediates that accumulated in the absence of FabG activity verified the mutant phenotype.

MATERIALS AND METHODS

Bacterial strains and plasmids.

The bacterial strains and plasmids used in this study are listed in Table 1. All E. coli strains were derivatives of E. coli K-12, and all S. enterica serovar Typhimurium strains were derivatives of strain LT2. Strain SJ16 was used as the wild-type E. coli strain in the ACP labeling experiments. Strain CL33 was constructed by transduction of strain CAG12147 with a phage P1 lysate grown on strain AB1623 with selection for nicotinic acid auxotrophs followed by screening for a glutamate requirement. Strain CL35 was obtained by P1 transduction of strain CL33 with a lysate grown on strain MR52 with selection for kanamycin resistance. Strain CL36 was constructed by P1 transduction of strain CL35 with a lysate grown on strain TL225 with selection for resistance to both tetracycline and kanamycin. Strain CL50 was obtained by P1 transduction of strain MG1655 with a P1 lysate grown on CL36 with selection for tetracycline and kanamycin resistance. Strains CL115, CL104, and CL116 were obtained by P1 transduction of strains MG1655, CL37, and JP1111, respectively, with a lysate grown on strain NRD1 followed by selection for chloramphenicol resistance. A fabF deletion mutant of S. enterica LT2, strain CL54, was obtained through λ Red gene-mediated gene replacement (10). The PCR products containing a kanamycin cassette flanked by the FLP recognition target (FRT) sites were generated by amplification from pKD4 (10) using primers SalF-N (5′- GGACTGGGCATGTTGTCTCCTGTCGGCAATACCGTGTAGGCTGGAGCTGCTTCG) and SalF-C (5′-GGAGTTGCACAGAGCGTACTCCAGATCGCTGACCCATATGAATATCCTCCTTAG). These primers contained 34 bp at their 5′ ends homologous to the ends of fabF and 20 3′-end base pairs homologous to sequences in pKD4. Strain CL55 was constructed by transduction of wild-type strain LT2 with P22 grown on strain CL54 with selection for kanamycin resistance. Strain MST3543 contains a locked-in Mud-P22 inserted at min 25 (purB1879::MudP) of the linkage map oriented such that DNA packaging proceeds in the counterclockwise direction (5). Strain CL61 was derived by phage P22 transduction of strain MST3543 with a lysate grown on strain CL55 with selection for kanamycin resistance.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant characteristics	Source or reference
E. coli K-12
SJ16	metB1 relA1 gyrA216 panD2 zad-220::Tn10	19
CAG12147	nadA57::Tn10 of MG1655	34
MG1655	Wild type	Lab collection
AB1623	F−thi gltA ara lac gal xyl mtl Strrtsx tfr	3
MR52	F− ΔlacU169 araD139 metE thi gyrA rpsL fabF::Kan	30
TL225	zce-727::Tn10	20
NRD1	panD::Cat	N. De Lay
JP1111	fabI(Ts)	CGSCa
CL33	gltA of MG1655	This work
CL35	gltA fabF::Kan of MG1655	This work
CL36	gltA fabF::Kan zce-727::Tn10 of MG1655	This work
CL37	fabG(Ts) fabF::Kan zce-727::Tn10 of MG1655	This work
CL48	fabG(Ts) panD2 zad-220::Tn10 of MG1655	This work
CL50	fabF::Kan zce-727::Tn10 of MG1655	This work
CL62	CL37/pCL33	This work
CL104	fabG(Ts) panD::Cat of MG1655	This work
CL115	panD::Cat of MG1655	This work
CL116	fabI(Ts) panD::cat of MG1655	This work
CL118	CL37/pCL79	This work
CL119	CL37/pCL80	This work
S. enterica serovar Typhimurium
MST3543	purB 1879::MudP	5, 40; via S. Maloy
MST725	gltA3 ara-7	S. Maloy
CL55	fabF::Kan of LT2	This work
CL61	fabF::Kan of MST3543	This work
CL65	fabG(Ts) fabF::Kan of LT2	This work
CL67	fabD(Ts) fabF::Kan of LT2	This work
CL95	fabG(Ts) fabF::Kan of LT2	This work
Plasmids
pCL25	Insertion of the 900-bp fabG PCR product of the E. coli chromosome (amplified with primers G15F and G12R)b into pCR2.1; encodes wild-type FabG	This work
pCL27	Insertion of the 900-bp fabG PCR product of CL37 chromosomal DNA (amplified with primers G15F and G12R) into pCR2.1; encodes E233K/A154T FabG	This work
pCL30	Insertion of the 686-bp XbaI-EcoNI fragment of pCL27 into pCL25 cut with the same enzymes; encodes A154T FabG	This work
pCL31	Insertion of the 686-bp XbaI-EcoNI fragment of pCL25 into pCL27 cut with the same enzymes; encodes E233K FabG	This work
pCL32	Insertion of the 993-bp BamHI-XbaI fabG fragment of pCL25 into pBR322 cut with BamHI and NheI; fabG is transcribed from ptet	This work
pCL33	Insertion of the 984-bp NsiI-BamHI fabG fragment of pCL27 into pHSG576 cut with PstI and BamHI; fabG is transcribed from plac; encodes E233K/A154T FabG	This work
pCL34	Insertion of the 993-bp BamHI-XbaI fabG fragment of pCL30 into pBR322 cut with BamHI and NheI; fabG is transcribed from ptet; encodes A154T FabG	This work
pCL35	Insertion of the 993-bp BamHI-XbaI fabG fragment of pCL31 into pBR322 cut with BamHI and NheI; fabG is transcribed from ptet; encodes E233K FabG	This work
pCL38	Insertion of the 1,010-bp fabH PCR product of the S. enterica chromosome (amplified with primers SalH-N and SalH-C)c into pCR2.1	This work
pCL45	Insertion of the 1.1-kb BamHI-XbaI fabH fragment of pCL38 into pBR322 cut with BamHI and NheI; fabH is transcribed from ptet	This work
pCL46	Insertion of the 1-kb fabD PCR product of the S enterica chromosome (amplified with primers SalD-N and SalD-C)d into pCR2.1	This work
pCL50	Insertion of the 1-kb BamHI-XbaI fabD fragment of pCL46 into pBR322 cut with BamHI and NheI; fabD is transcribed from ptet	This work
pCL79	Insertion of the 984-bp NsiI-BamHI fabG fragment of pCL30 into pHSG576 cut with PstI and BamHI; fabG is transcribed from plac; encodes A154T FabG	This work
pCL80	Insertion of the 984-bp NsiI-BamHI fabG fragment of pCL31 into pHSG576 cut with PstI and BamHI; fabG is transcribed from plac; encodes E233K FabG	This work
pHSG576	lacZα, pSC101 origin, chloramphenicol resistant	37
pCR2.1	lacZα, Δrop pBR322 origin	Invitrogen
pBR322	Tetracycline and ampicillin resistant	4

^aColi Genetic Stock Center, Yale University, New Haven, Conn.

^bThe primer sequences were as follows: G15F, 5′-GCGCTCGAGCTTTAAAAGAG; G12R, 5′-AACTAAATCCCGGCAGGTCT.

^cThe primer sequences were as follows: SalH-N, 5′-CCGAAAAGTGACTGAGCGTA; SalH-C, 5′-ACAAATGCAAATTGCGTCAT.

^dThe primer sequences were as follows: SalD-N, 5′-CGCGCTGATTCGTTTCTAGT; SalD-C, 5′-CGCAATCTTTCCTTCAAAGC.

The plasmid constructions together with the primers used in their constructions are described in Table 1. DNA sequencing of both strands of the relevant fab genes of plasmids pCL38, pCL46, and pCL25 and the fabG gene of plasmid pCL27 was performed by the Genetic Engineering Facility, University of Illinois at Urbana-Champaign, with Taq DNA polymerase cycle sequencing on an Applied Biosystems 373 DNA sequencer, using the M13/pUC forward (−20) and reverse (−24) sequencing primers. These plasmids were transformed into the temperature-sensitive mutant strains in order to identify fab gene mutations by complementation. The primers used for PCR amplifications were synthesized by the Genetic Engineering Facility, University of Illinois at Urbana-Champaign.

Culture media and growth conditions.

Culture media used were rich broth (RB) medium (12), Luria-Bertani (LB) medium (11), minimal salts medium E (25), and M9 minimal salts (25) supplemented as indicated. Green plates (11) were used for testing of phage sensitivity against the c2 mutant phage P22-H5. Kanamycin (50 μg/ml), tetracycline (12 μg/ml), ampicillin (100 μg/ml), chloramphenicol (20 μg/ml), isopropyl-β-d-thiogalactopyranoside (IPTG) (1 mM), and EGTA (10 mM) were added as required to the final concentrations given.

Mutagenesis.

Nitrosoguanidine mutagenesis was conducted as described by Miller (26). The cells were treated with 50 μg of nitrosoguanidine/ml at 37°C for various times, and a dose at which about 30% of the cells retained colony-forming ability was chosen. For hydroxylamine mutagenesis a Mud-P22 lysate from strain CL61 (purB1879::MudP fabF::Kan) was prepared by induction with mitomycin C (40). For mutagenesis a Mud-P22 lysate (0.2 ml) was mixed with 0.4 ml of 0.5 M phosphate-EDTA buffer, 0.8 ml of 1 M hydroxylamine, 0.02 ml of 1 M MgSO₄, and 0.6 ml of sterile water in a final volume of 2 ml at 37°C without shaking (11). Since phage Mud-P22 cannot form plaques (35), a P22 lysate grown on CL55 was used to assay the efficiency of mutagenesis. The treated lysates were titered on strain LT2 following various periods of hydroxylamine treatment. When mutagenesis reached 0.1% survival, samples were centrifuged in a microcentrifuge for 30 min at 4°C to pellet the phage. The phage pellet was handled and stored at 4°C as described by Maloy (23) and used to transduce MST725 (gltA), with selection for kanamycin resistance in the presence of 10 mM EGTA. The resulting transductants were pooled and subjected to tritium suicide selections.

Tritium suicide selections.

Tritium suicide selections were performed according to Harder et al. (13). Briefly, a 2-ml culture of mutagenized E. coli cells was grown at 30°C to a density of approximately 2 × 10⁷ cells/ml in minimal medium E supplemented with 1 mM potassium acetate, 0.4% glycerol, 0.1% vitamin-free Casamino Acids (Difco), 10 mM monosodium glutamate, 2 mM proline, and 0.001% thiamine. The culture was then incubated at 42°C for 20 min, centrifuged, and washed, and the cells were suspended in minimal medium E. Half of this culture was supplemented with 0.5 mM sodium acetate, 0.5 mM sodium [³H]acetate (specific activity, 10 Ci/mmol), 0.4% glycerol, 1% vitamin-free Casamino Acids, 10 mM monosodium glutamate, 2 mM proline, 20 mM glutamine, 0.4% sodium succinate, and 0.001% thiamine. The other portion of the culture contained the above ingredients except that 0.5 mM nonradioactive sodium acetate replaced the radioactive acetate. Both cultures were incubated at 42°C for 4 h, centrifuged, and washed three times with cold M9 medium. The cells were then suspended in 1 ml of minimal medium E and stored at 4°C. After various periods of storage, samples were removed, serially diluted, and plated for surviving cells at 30°C on RB plates. When tritium suicide killing reached 0.1% survivors, the colonies formed from the surviving cells were tested for growth at 30 and 42°C on RB plates. Colonies that grew at 30 but not at 42°C were streaked on RB plates supplemented with oleate and palmitate to detect strains that required fatty acids for growth at 37°C. Phage P1 lysates were grown on the strains requiring fatty acids and used to transduce strain MG1655 with selection for resistance to both tetracycline and kanamycin (the antibiotic markers were inserted at opposite ends of the fab cluster). The resulting transductants, possessing the temperature-sensitive phenotype, were transformed with plasmids pCL38, pCL46, and pCL25, which contain the fabH, fabD, and fabG genes, respectively, to determine in which genes the mutations were located. Mutants defective in any of these genes were then PCR amplified with corresponding primers, and the PCR product was sequenced to identify the mutations. Strain CL37, defective in the fabG gene, was isolated through this procedure. Strain CL47 was constructed by transduction of strain MG1655 with P1 grown on the fabG(Ts) strain CL37 with selection for kanamycin resistance followed by screening for tetracycline sensitivity, whereas strain CL48 was constructed by transduction of strain CL47 [fabG(Ts)] with phage P1 grown on strain SJ16 with selection for tetracycline resistance.

Pools of the S. enterica transductants resulting from transduction with hydroxylamine-mutagenized phage stocks were grown in minimal medium E supplemented with 1 mM potassium acetate, 0.4% glycerol, 0.1% vitamin-free Casamino Acids, 10 mM monosodium glutamate, 2 mM proline, 0.2% sodium butyrate, 0.1 mM tryptophan, and 0.001% thiamine at 30°C. When the density of the culture reached 2 × 10⁷ cells/ml, the culture was shifted to 42°C for 20 min and centrifuged, and the pellet washed in minimal medium E. The cells were then resuspended in minimal medium E and divided into two equal portions. One portion was supplemented with 0.5 mM sodium acetate, 0.5 mM sodium [³H]acetate (specific activity, 10 Ci/mmol), 0.4% glycerol, 1% vitamin-free Casamino Acids, 10 mM monosodium glutamate, 2 mM proline, 20 mM glutamine, 0.4% sodium succinate, 0.2% sodium butyrate, 0.1 mM tryptophan, and 0.001% thiamine. The other portion of the culture was supplemented with same ingredients as described above except that 1 mM nonradioactive sodium acetate was substituted for the radioactive compound. Both cultures were grown at 42°C for 4 h and then treated as described above. Strains having a temperature-sensitive phenotype were grown, and then a phage P22 lysate was prepared and used to transduce strain LT2 to kanamycin resistance. The resulting temperature-sensitive transductant colonies were tested for phage sensitivity by cross-streaking against the clear c2 mutant phage P22-H5 on green plates (23). Phage-sensitive strains were then transformed with plasmids pCL38, pCL46, and pCL25, which contain the fabH, fabD, and fabG genes, respectively, to determine in which gene each mutation was located. Mutants defective in any of these genes were then amplified by PCR with corresponding primers, and the PCR products were sequenced to identify the mutations.

Preparation of cell extracts.

The in vitro fatty acid synthesis extracts were prepared as described by Heath and Rock (14). Strains SJ16 (panD) and CL48 [panD fabG(Ts)] were cultured in 500 ml of LB medium and grown to late log phase. The cultures were centrifuged, and the cells were resuspended in 5 ml of lysis buffer (0.1 M sodium phosphate [pH 7.0], 5 mM 2-mercaptoethanol, 1 mM EDTA) and lysed in a French pressure cell at 18,000 lb/in². The lysate was centrifuged in a JA-20 rotor at 16,000 rpm at 4°C for 1 h to remove cell debris. Ammonium sulfate was added to the supernatant to 45% of saturation, and the precipitated protein was removed by centrifugation for 30 min at 10,000 rpm. Additional ammonium sulfate was added to the supernatant to 80% of saturation, and the precipitated protein was collected by centrifugation. The protein pellet was dissolved in 2 ml of lysis buffer and dialyzed for 5 h at 4°C against 2 liters of the same buffer. Protein concentrations were determined by the Bradford assay (7) with bovine serum albumin as the standard.

In vitro fatty acid synthesis assay.

The fatty acid synthesis assay was performed according to the method of Jackowski and Rock (18). The assay mixtures contained 0.1 M LiCl, 0.1 M sodium phosphate, pH 7.0, 1 mM 2-mercaptoethanol, 50 μM acetyl-CoA, 0.175 mM NADH, 0.149 mM NADPH, 54 μM ACP, and 30 μg of protein extract from either strain SJ16 or strain CL48 [panD fabG(Ts)] in a final volume of 40 μl. Cerulenin, when present, was added to a final concentration of 1 mM. The reaction mixtures were incubated for 5 min at room temperature to allow cerulenin to inactivate FabB and FabF, and then 45 μM [2-¹⁴C]malonyl-CoA (specific activity, 55 mCi/mmol) was added to initiate the reaction followed by incubation at 37°C for 10 min. The reactions were stopped by placing the tubes in an ice slush. Samples (40 μl) of the assay mixtures were mixed with gel loading buffer and analyzed by conformationally sensitive gel electrophoresis on 15% polyacrylamide gels containing 2.5 M urea for approximately 3 h at 4°C (28). The gels were fixed, soaked in Enlightning (DuPont), dried, and exposed to X-ray film.

Analysis of fatty acid synthesis in vivo with [1-¹⁴C]acetate.

The rate of fatty acid synthesis in the E. coli strains CL50 (wild type) and CL37 [fabG(Ts)] at 42°C was analyzed by labeling of lipids with sodium [1-¹⁴C]acetate (specific activity, 55 mCi/mmol). The two strains were grown to a density of 10⁷ cells/ml in LB medium and then shifted to 42°C for 10 min. Then 2 ml of each culture was removed at various times after the shift and labeled with 2 μCi of sodium [1-¹⁴C]acetate for 5 min before addition of 3 ml of methanol-chloroform (2:1, vol/vol) followed by shaking for 1 hour at 30°C. The organic and aqueous phases were separated by addition of 2 ml each of water and chloroform. The mixture was then vortexed and centrifuged. The upper aqueous layer was discarded, and the lower organic layer was washed twice with an equal volume of 2 M KCl. After a final wash with an equal volume of water, the solvent was evaporated under a stream of nitrogen. Analtech Silica Gel G plates were activated by heating at 80°C overnight. The phospholipids were suspended in a small volume of chloroform-methanol (2:1, vol/vol) and applied to thin-layer chromatography plates which were developed with a mobile phase of chloroform, methanol, and acetic acid (65:25:8, vol/vol/vol). The plates were dried and exposed to a Molecular Dynamics PhosphorImager screen for quantitation.

Assay of β-ketoacyl-ACP reductase activity.

The spectrophotometric assay for reductase activity measures the rate of oxidation of NADPH at 340 nm (38). The enzyme will accept CoA thioesters in place of the physiological ACP thioesters (38). All samples were assayed at room temperature. The reaction mixtures contained 0.1 M sodium phosphate, pH 7.0, 1 mM 2-mercaptoethanol, 239 μM NADPH, 58.7 μM acetoacetyl-CoA, 40 μl of crude extract protein, and water added to a final volume of 500 μl. Protein concentrations were measured following the spectrophotometric assay since the FabG activities of strain CL48, CL62, and CL119 extracts decreased very quickly following extract preparation. The reaction was initiated by the addition of crude extract. The rate of NADPH oxidation prior to acetoacetyl-CoA addition served as a blank value and was subtracted from the rate observed in the presence of acetoacetyl-CoA. Samples of each crude extract were subjected to heat treatments and then placed on ice. The enzymatic activity of each sample was then measured by monitoring the rate of absorbance decrease at 340 nm. The rates of NADPH oxidation were calculated from the change in absorbance, assuming ε₃₄₀ = 6.3 × 10³ mol⁻¹.

In vivo labeling of ACP pools.

Labeling of ACP pools was performed as described by Heath and Rock (14). E. coli strain CL48 [panD fabG(Ts)] was grown at 30°C to a density of 1 × 10⁷ cells/ml in M9 medium containing 0.6 μM [β-³H]alanine (specific activity, 60 Ci/mmol), 0.4% glucose, 1% vitamin-free Casamino Acids, 0.001% thiamine, and 0.005% methionine. The culture was divided into two portions. One half was shifted to 42°C and grown to a density of 1 × 10⁸ cells/ml, at which time a portion of this culture was treated with cerulenin (1 mg/ml) for 10 min and the remaining untreated portion was harvested as a control. The other half was grown at 30°C to a density of 1 × 10⁸ cells/ml, the culture was split, and one half was treated with 1 mg of cerulenin/ml as described above. Strain SJ16 (panD) was grown at 37°C to a density of 1 × 10⁸ cells/ml, one half of the culture was treated with cerulenin, and the other half was incubated at 37°C for 10 min. The cells were harvested by centrifugation and then lysed on ice by the procedure described by Clewell and Helinski (9). The lysate was centrifuged to sediment the DNA, and the supernatant fluid was fractionated on a 15% polyacrylamide gel containing 2.5 M urea at 4°C. The gels were then subjected to fluorography.

Strains CL115 (panD::Cat), CL104 [panD::Cat fabG(Ts)], and CL116 [panD::Cat fabI(Ts)] were inoculated to a density of 5 × 10⁷ cells/ml in the medium given above and labeled with [β-³H]alanine (specific activity, 60 Ci/mmol). The cultures of strains CL104 and CL116 were labeled for 4 h at 30°C, whereas the strain CL116 culture was labeled at 37°C for 4 h. Half of the cultures of strains CL104 and CL116 were then shifted to 42°C for 1 h. All cultures were then divided in half, and one half of each sample was treated with 1 mg of cerulenin/ml for 10 min. The acyl-ACP species were then extracted and analyzed as described above.

RESULTS

Tritium suicide selection of fab mutants.

[³H]acetate radiation suicide selection was previously used to isolate mutants of E. coli having defects in fatty acid biosynthesis (13, 33). The principle is that use of a strain defective in citrate synthase and the proper growth medium confines radioactive acetate incorporation almost exclusively to fatty acids. Cultures of mutagenized cells are exposed to [³H]acetate of very high specific activity while growing at 42°C. Those cells having normal fatty acid synthesis accumulate [³H]acetate and are subsequently killed by radioactive disintegration during storage (presumably due to oxidative damage caused by the β particles), whereas mutants blocked early in the fatty acid synthesis pathway fail to incorporate radioactive acetate and are recovered as surviving cells (13). Our tritium suicide selections of mutagenized E. coli cultures showed about 6 log units of killing after 40 days of exposure, and the extent of labeling was 1.47 dpm/cell. From this selection, 93 temperature-sensitive colonies were isolated. Upon supplementation of the medium with saturated and unsaturated fatty acids, 37 colonies grew better at 37°C in the presence of the fatty acid supplement. These “fatty acid remedial” strains failed to grow at 42°C with fatty acid supplementation, indicating that at 37°C some residual synthesis of fatty acids remained that allowed synthesis of intermediates that cannot be provided exogenously (2, 13, 36). These intermediates are probably β-hydroxymyristoyl-ACP (13) and perhaps the other short-chain acids required for lipid A biosynthesis (2). Phage P1 lysates were grown on all fatty acid remedial strains and used to transduce strain MG1655 to tetracycline and kanamycin resistance. Only one strain, strain CL37, had a temperature-sensitive phenotype and was identified as a fabG(Ts) mutant by complementation with a fabG plasmid.

We attributed the meager results obtained with E. coli to the fact that we did not localize the mutagenesis to the fab cluster region of the chromosome. We therefore turned to S. enterica because this organism offers an excellent means to localize mutagenesis to a specific region of the chromosome. Localized mutagenesis is based on a collection of chromosomal Mud-P22 insertions (5, 40). Upon induction these prophages are unable to excise from the bacterial chromosome and phage DNA replication generates a large fraction of phage particles that carry chromosomal DNA segments. The chromosomal DNA segments are progressively packaged for 150 to 250 kb to one side of a given insertion (5, 40). Therefore, we constructed S. enterica strain CL61 (purB1879::MudP fabF::Kan). The Mud-P22 prophage of this strain is inserted into the purB gene and packages in the direction of the fab gene cluster, which is located only 50 kb from purB (24). Therefore, upon induction we expected that the phage particles of the resulting lysates should be rich in DNA segments that encode the fab gene cluster and that the DNA segments of these particles should be effectively mutagenized by hydroxylamine in vitro. The phage particles were then treated with hydroxylamine to mutagenize the packaged bacterial DNA and subsequently used to transduce strain MST725 (gltA) to kanamycin resistance. The resulting transductants were pooled and submitted to tritium suicide selections.

The S. enterica tritium suicide selections reached about 8 log units of killing after 23 days of storage (data not shown), and the extents of labeling were 1.85, 0.44l, and 0.25 dpm/cell. About 40 temperature-sensitive colonies were isolated, and 10 colonies had a temperature-sensitive phenotype after transduction of the LT2 wild-type strain to kanamycin resistance. Compared to nitrosoguanidine mutagenesis of the whole E. coli chromosome, localized mutagenesis with hydroxylamine was a more effective means to enrich for mutations within the fab gene cluster. Following transformation with plasmids containing fabH, fabD, or fabG, three S. enterica fatty acid synthetic mutants were identified: strains CL65 and CL95 had mutations in fabG, whereas the mutation of strain CL67 mapped in fabD.

Analysis of fatty acid synthesis in vivo.

De novo fatty acid synthesis of the wild-type E. coli strain CL50 and the E. coli fabG(Ts) strain CL37 was assayed by incorporation of [¹⁴C]acetate into phospholipids. Cultures were grown to early log phase at 30°C and then shifted to 42°C. Samples of the cultures were then taken and labeled with radioactive acetate to determine the rates of fatty acid synthesis (see Materials and Methods). As expected, mutant strain CL37 was defective in fatty acid synthesis (Fig. 1). After shift to the nonpermissive temperature, the rate of incorporation of [¹⁴C]acetate into lipids in cultures of the fabG(Ts) strain, CL37, progressively decreased, whereas the wild-type strain, CL50, had a much higher rate of synthesis until the inhibition of synthesis by entry into stationary phase (the mutant culture failed to reach a stationary-phase cell density).

FIG. 1.

Analysis of E. coli fatty acid synthesis in vivo. De novo fatty acid synthesis was measured by incorporation of [¹⁴C]acetate into phospholipids of the wild-type strain CL50 and the fabG(Ts) strain CL37 (see Materials and Methods). The squares denote the E. coli wild-type strain CL50, while the open circles represent the E. coli fabG(Ts) strain CL37.

Analysis of fatty acid synthesis in vitro.

Cell extracts of the wild-type E. coli strain SJ16 and the fabG(Ts) E. coli strain CL48 were used to determine the role of FabG in the in vitro elongation cycle (Fig. 2). Fatty acid synthesis reactions containing acetyl-CoA, [2-¹⁴C]malonyl-CoA, NADPH, NADH, and ACP were conducted as described in Materials and Methods. Identification of the acyl-ACP species was based on their relative electrophoretic migration rates (14). Extracts of the wild-type strain SJ16 incorporated [2-¹⁴C]malonyl-CoA into long-chain acyl-ACPs (Fig. 2, lane 1), whereas extracts of strain CL48 [fabGs(Ts)] failed to assimilate radioactive malonyl-CoA into the elongation cycle (Fig. 2, lane 3). In the presence of cerulenin, an antibiotic that blocks all β-ketoacyl-ACP (KAS) I and KAS II activities but which has no effect on KAS III activity (32), accumulation of acetyl-ACP (or malonyl-ACP) and butyryl-ACP was seen in the wild-type extracts (Fig. 2). It should be noted that acetyl-ACP comigrates with holo-ACP and malonyl-ACP on 2.5 M urea gels (15). Extracts of strain CL48 failed to produce either long-chain acyl-ACPs or butyryl-ACP and accumulated only acetyl-ACP and/or malonyl-ACP (Fig. 2), indicating that the first elongation cycle was blocked and this blockage was due to the loss of FabG activity (note that the β-ketobutyryl-ACP intermediate is unstable and is largely degraded during electrophoresis [14]). In the presence of cerulenin the extracts of strain CL48 accumulated acetyl-ACP (or malonyl-ACP) and butyryl-ACP (Fig. 2), although the levels of butyryl-ACP accumulation were much lower than those seen in extracts of the wild-type strain (Fig. 2). These results indicate that residual FabG activity remained in the fabG(Ts) strain extracts. In summary, these data indicted that FabG was required for the first cycle of fatty acid synthesis and was the sole enzyme that performed reduction of β-ketoacyl-ACP species.

FIG. 2.

Analysis of the ACP products of in vitro E. coli fatty acid synthesis. Fatty acid synthesis assays were done as described in Materials and Methods. Cerulenin (1 mM) was added to indicated reaction mixtures (+) to inhibit the activities of FabB and FabF. The reactions were stopped after 10 min by cooling on ice, and the products were fractionated on 15% acrylamide gels containing 2.5 M urea followed by fluorography as described in Materials and Methods. Abbreviations: Ac-ACP, acetyl-ACP; Mal-ACP, malonyl-ACP. Lanes 1 and 2, extracts of the wild-type (WT) strain SJ16; lanes 3 and 4, extracts of the fabG(Ts) strain CL48. Cerulenin was added to the extracts in lanes 2 and 4.

Composition of ACP pools in vivo.

To characterize the role of FabG in vivo, E. coli strains SJ16 (panD2) and CL48 [panD2 fabG(Ts)] were grown with [β-³H]alanine to uniformly label the ACP pool and the products were fractionated by conformationally sensitive gel electrophoresis (see Materials and Methods). The ACP thioesters were identified on the gels by reference to the work of Heath and Rock (14). The composition of the strain CL48 ACP pool at the permissive temperature (30°C) was essentially identical to that of the wild-type strain SJ16 (Fig. 3A). Acetyl-ACP and/or malonyl-ACP was a major species, and some medium-chain and long-chain acyl-ACP species were also present. Addition of cerulenin to strain CL48 grown at 30°C resulted in modest accumulations of butyryl-ACP and medium-chain acyl-ACPs and a decrease in long-chain acyl-ACPs. Similar, although much greater, accumulations of these species were seen for the control strain SJ16 upon cerulenin treatment (Fig. 3A). Upon temperature shift to 42°C, the ACP pool composition of strain CL48 changed markedly. Butyryl-ACP and medium-chain acyl-ACPs disappeared, and long-chain acyl-ACPs were dramatically decreased (Fig. 3A). These data were consistent with the distribution of products observed in the fatty acid synthase assay in vitro (Fig. 2) and confirmed that the first cycle of fatty acid synthesis was blocked in the fabG(Ts) mutant. Addition of cerulenin to strain CL48 at 42°C did not lead to the altered pools seen in strain SJ16 because the strain CL48 fatty acid elongation cycle was totally blocked under these conditions.

FIG. 3.

Compositions of ACP pools of the wild-type and fabG(Ts) mutant E. coli strains. (A) Strain SJ16 (panD2) and CL48 [panD2 fabG(Ts)] were grown at the indicated temperatures in the presence of 0.6 μM [β-³H]alanine (specific activity, 60 Ci/mmol). At a density of 10⁸ cells/ml, the culture of SJ16 was divided, and a portion was treated with 1 mg of cerulenin/ml for 10 min. The CL48 culture was grown to a density of 10⁷ cells/ml and split into two portions. One portion was shifted to 42°C, and the remaining portion was grown at 30°C. At a density of 10⁸ cells/ml, each culture was divided and one portion was treated with cerulenin for 10 min. ACP thioesters were extracted and resolved on a 15% acrylamide-2.5 M urea gel as indicated in Materials and Methods. Lanes 1 and 2, wild-type (WT) strain SJ16; lanes 3 to 6, strain CL48 grown at the temperatures indicated. Cerulenin was added to the extracts of the even-numbered lanes. (B) Strains CL104 [panD::Cat fabG(Ts) ] and CL116 [panD::Cat fabI(Ts)] were inoculated to a density of 5 × 10⁷ cells/ml. The cultures were labeled with 0.6 μM [β-³H]alanine at the indicated temperature for 4 h. Half of the CL104 and CL116 cultures were then shifted to 42°C for 1 h. All cultures were then split in half, and one half of each sample was treated with 1 mg of cerulenin/ml for 10 min. Acyl-ACP species were extracted and analyzed as described in Materials and Methods. Lanes 1 to 4, fabG(Ts) strain CL104; lanes 5 to 8, fabI(Ts) strain CL116. The strains were grown at the temperatures indicated, and the cultures in the even-numbered lanes were treated with cerulenin. Note that the cultures of strain CL104 grew more slowly than those of strain CL116 due to the lower permissive growth temperature but were labeled for the same time interval with [β-³H]alanine, resulting in a different cell densities and different quantities of ACP species.

Heath and Rock (14) studied an E. coli strain with a temperature-sensitive lesion of the fabI gene, which encodes enoyl-ACP reductase. FabI is the enzyme responsible for the last step of the elongation cycle in E. coli, and these workers reported that FabI was the sole enoyl-ACP reductase of this organism (14). Since our results indicated that FabG, like FabI, is solely responsible for an elongation cycle reaction, we predicted that the intracellular ACP pool compositions of the two mutant strains should be similar at the nonpermissive temperature. Indeed, the compositions of the ACP pools of the two strains were found to be very similar (Fig. 3B). At the permissive temperature, both strains contained low levels of long-chain acyl-ACPs whereas cerulenin treatment caused the accumulation of butyryl-ACP in both strains (Fig. 3B). Upon shift of E. coli strains CL104 [fabG(Ts)] and CL116 [fabI(Ts)] to 42°C the long-chain acyl-ACP levels decreased whereas the addition of cerulenin resulted in only a slight increase in butyryl-ACP levels (Fig. 3B). The main differences seen in the ACP pools of the two strains were that the fabG(Ts) strain CL104 synthesized lower levels of long-chain acyl-ACPs than did the fabI(Ts) strain CL116. This is readily attributed to the differences in the stringency of the mutations. These data indicate that fabG, like fabI, is essential for all fatty acid synthesis and hence cell viability.

Resolution of the mutations of the doubly mutant E. coli and S. enterica fabG temperature-sensitive mutants.

Sequence analysis showed that E. coli strain CL37 contained two point mutations within fabG, resulting in a protein having A154T and G233K substitutions. Of the two S. enterica strains the FabG of strain CL65 had a single amino acid change, S224F, whereas the FabG of strain CL95 contained two mutations, M125I and A223T. The fabD of S. enterica strain CL67 also contained two missense mutations, P9L and A20V. To identify the point mutations responsible for the fabG(Ts) phenotypes of strains CL37 and CL95, the two mutations of each of the fabG(Ts) double mutants were separated by inserting a DNA segment containing one mutation into the wild-type gene in vitro, thereby resulting in two mutant genes that each contained a single mutation. The constructed singly mutant genes were then moved into the low-copy-number vector pHSG576 under control of the vector plac promoter and tested for complementation of the fabG(Ts) phenotype of strain CL37. Upon glucose repression of expression from the lac promoter, plasmid pCL79, encoding the A154T FabG, complemented strain CL37 to give growth at the nonpermissive temperature whereas pCL80, encoding the E233K FabG, failed to allow growth of strain CL37 at 42°C (data not shown). Also, as expected plasmid pCL33, carrying both mutations, failed to complement strain CL37 at the nonpermissive temperature in the presence of glucose. However, upon IPTG induction, all three plasmids restored the growth of strain CL37 at 42°C (data not shown). Therefore, it is clear that the E233K and E233K A154T proteins retain residual activity at the nonpermissive temperature since overexpression of these proteins from the lac promoter (or from the pBR322 tet promoter in the case of the E233K FabG) allowed growth.

In the case of S. enterica strain CL95 moving the mutant gene from the chromosome to a low-copy-number plasmid resulted in loss of the temperature-sensitive phenotype. Introduction of plasmid pCL72, which carries the doubly mutant gene in pHSG576, allowed growth of strain CL95 at 42°C even with glucose repression of the lac promoter, and thus the individual mutations could not be tested (data not shown). The observation that complementation required less expression of the S. enterica strain CL95 fabG than of E. coli strain CL37 fabG was consistent with the greater residual fatty acid synthetic activity of the former strain at 42°C (data not shown).

Thermolability of FabG activity in E. coli mutant cell extracts.

Since FabG utilizes NADPH to reduce β-ketoacyl-ACP, the enzyme activity can be assayed spectrophotometrically by measuring the rate of oxidation of NADPH at 340 nm (38). In our first experiments crude extracts of the E. coli wild-type strain SJ16 and the doubly mutant strain CL48 were assayed for reductase activity after the extracts were heated at various temperatures followed by assay at room temperature (see Materials and Methods). The activity of E. coli strain CL48 extracts was completely lost after 5 min at 42°C, whereas the extracts of strain SJ16 retained over half of the initial activity (data not shown). In order to determine the individual contributions of the two mutations of strain CL48 to enzyme stability in vitro, we assayed extracts of E. coli strain CL48 carrying each of the plac fabG plasmids described above. Crude extracts were subjected to heat treatment, and the enzymatic activity of each sample was measured. The reductase activities of extracts of strains CL119 (encoding FabG E233K) and CL62 (encoding FabG E233K and A154T) were much more thermolabile than the activity of the wild-type strain, CL50 (Fig. 4), consistent with their temperature-sensitive growth phenotypes. Moreover, extracts containing either the E233K or E322K A154T enzyme lost FabG activity quickly at all temperatures and incubation times tested. It should be noted that unlike the wild-type extracts these mutant extracts could not be processed by ammonium sulfate precipitation since this resulted in complete loss of the mutant enzyme activity. The thermostabilities of reductase activities in extracts of strains encoding each of the single mutations were determined. The E233K enzyme showed marked thermolability, but the rates of inactivation observed were less than that of extracts of the double (E233K A154T) mutant. The thermolability of the A154T extract reductase activity was similar to that of the wild-type extract. These enzyme stability data are consistent with the observation that extracts of CL48 could not incorporate [2-¹⁴C]malonyl-CoA into acyl-ACP in vitro (Fig. 2, lane 3).

FIG. 4.

Thermolability of FabG activities in E. coli cell extracts. Crude cell extracts were heated for various time intervals at 30 (□), 37 (⋄), or 45°C (○) and then assayed for β-ketoacyl-ACP reductase activity (see Materials and Methods). (A) Crude extract of the wild-type E. coli strain CL50; (B) crude extract of strain CL62, encoding E233K/A154T FabG; (C) crude extract of strain CL118, expressing A154T FabG; (D) crude extract of strain CL119, expressing E233K FabG. The initial unheated activities (micromoles of NADPH per minute per milligram of protein) were 12.0, 7.4, 15.4, and 8.4 for the extracts of strains CL50, CL62, CL118, and CL119, respectively. Identical experiments done with strains SJ16 and CL48 gave data very similar to those shown in panels A and B, respectively.

DISCUSSION

The isolation of fabG(Ts) mutants of E. coli and S. enterica demonstrates that fabG is essential for growth and cell viability, thus confirming the transcriptional termination data of Zhang and Cronan (41). Our data together with prior biochemical investigations indicate that FabG is the sole β-ketoacyl-ACP reductase in E. coli. Upon shift of strain CL48 to the nonpermissive temperature no long-chain acyl-ACP species are seen in cerulenin-treated cells, whereas under these conditions long-chain acyl-ACPs accumulate in the wild-type strain and in strain CL48 at the permissive temperature (Fig. 3). Cerulenin blocks synthesis of long-chain β-ketoacyl-ACP species, but any β-ketoacyl-ACPs present at the time of cerulenin addition will complete their synthetic cycle (i.e., the fatty acyl chains will become fully saturated) to give acyl-ACP species that are stable to gel electrophoresis. We attribute the lack of long-chain acyl-ACPs seen in extracts of cerulenin-treated strain CL48 cells to an inability to complete synthetic cycles; the nascent acyl-ACPs remain as β-ketoacyl-ACPs which are degraded during electrophoresis (14). The inability to efficiently convert β-ketoacyl-ACPs to more stable forms also accounts for the lower levels of long-chain acyl-ACPs seen in the absence of cerulenin in strains CL48 and CL104 grown at the nonpermissive temperature (Fig. 3). FabG activity is clearly required for butyryl-ACP synthesis (Fig. 2 and 3), and thus we believe that FabG is essential for all of the elongation cycles required to synthesize long-chain fatty acids. This picture agrees well with prior biochemical characterizations. Heath and Rock (14) reported that purified FabG carries out the reduction of β-ketoacyl-ACPs of all chain lengths during de novo fatty acid synthesis in a reconstructed E. coli in vitro system, whereas Toomey and Wakil (38) purified a single β-ketoacyl-ACP reductase from E. coli cell extracts and showed the enzyme to be active with substrates of 2, 6, and 10 carbon atoms. Since we began our FabG work the 2.6-Å X-ray crystal structure of E. coli FabG has been reported (29), and thus we were provided with a framework to analyze the mutations we isolated. Strikingly, all of our mutations are located in or near the subunit interfaces of the FabG homotetramer.

The FabG monomer contains a typical Rossmann fold structure, with a twisted, parallel β sheet composed of seven β strands flanked on both sides by eight α helices (29). The protein exists as a tetramer with two types of dimerization interfaces (29). That is, the enzyme can be considered a dimer of dimers. One type of interface is located between helices α4/α5 and α4′/α5′ (the prime denotes the neighboring monomer of the tetramer), which build a helix bundle. These are the top and bottom interfaces as seen in Fig. 5A. Each helix interacts with the same helix of the adjacent monomer. While the α4-α4′ interactions are almost purely hydrophobic, the α5-α5′ interface relies on steric interactions. Helices α5 and α5′ are comprised of a string of alanine and glycine residues (A152, A153, A154, A156, G157, and G160) that sterically complement one another. These residues form a hydrophobic anchor in the dimer interface which may be important in stabilizing the FabG tetramer (Fig. 5D). The other type of interface (seen at the left and right interfaces of Fig. 5A) is positioned between strands β7/β7′, helices α8/α8′, and helices α6/α7 and α6′/α7′.

FIG. 5.

Modeling of the FabG(Ts) mutant proteins. (A) Tetrameric structure of FabG (29) with the monomer chains shown in different colors. The residues involved in the A154T, E233K, and S224F mutations are shown as balls and sticks. In each case the structures are shown in pairs (B plus C, D plus E, and F plus G) with the wild-type FabG structure to the left and the modeled mutant structure to the right. (B and C) Consequences of the E233K mutation. The salt bridge between E233 and H236′, indicated by a dashed red line in panel B, is missing in panel C due to the E233K mutation. (D) View of interactions around A154 at the other dimer interface of the protein. (E) Modeled replacement of A154 with Thr. The α5 helix was shortened due to the formation of a hydrogen bond (dashed green line) between Y151 and T154, thereby increasing the distance between α5 and α5′. (F and G) S224F mutation. Panel F is a view of the dimer interface interactions around Ser 224. Panel G suggests that replacement of Ser 224 with Phe could destabilize the stacking between Phe 221 and Phe 221′ due to loss of hydrogen bonding by the hydroxyl groups of Ser 224 and Ser 224′. This mutation could also disrupt hydrogen bonding between Ser 224 and Glu 226.

The FabG of E. coli strain CL37 contains two mutations, E233K and A154T. The E233K mutation maps to β-sheet 7 which is involved in one of the dimerization interfaces within the tetramer, whereas the A154T mutation maps to helix 5 in the other type of dimerization interface. Specific interactions among the side chains bridge the gap between the antiparallel strands β7 and β7′. One of these interactions is a salt bridge between E233 of β7 and H236′ of β7′ (and vice versa) (Fig. 5B). This interaction would be destroyed by the E233K mutation (Fig. 5C) of mutant CL37. Moreover, the introduced K233 should repel H236′. E233 is also hydrogen bonded to H236′ and T234 (Fig. 5B). One of these lost hydrogen bonds might be compensated for by a possible hydrogen bond formed between the mutant lysine residue and T234′ (Fig. 5C). The E233K mutation of the FabG strain CL37 has much more severe effects on the stability of the protein both in vitro and in vivo than does the second mutation, A154T. However, although the A154T protein supports growth and has at best a modest effect on the thermostability of FabG in vitro, it seems to act in synergy with the E233K mutation in the doubly mutant protein. This follows from the finding that the reductase activity of the doubly mutant protein is more thermolabile than the E233K activity (Fig. 4). The A154T mutation would interrupt the sterically complementary surfaces of helices α5 and α5′ and might also result in formation of a hydrogen bond between the threonine hydroxyl and the carbonyl oxygen of tyrosine-151 (Fig. 5E). This putative interaction is predicted to shorten the helix and increase the distance between G157/G157′ and A153/G160′, which normally form a hydrophobic core within the interface. It therefore seems that destabilization of both of the interfaces of FabG tetramers results in a greater loss of stability towards heating than is seen when only a single interface is disturbed.

Since the FabGs of S. enterica and E. coli are 95% identical, we used the E. coli FabG structure to analyze the S224F mutation of S. enterica strain CL65. In the wild-type enzyme, S224 stabilizes F221 by hydrogen bonding to its carbonyl oxygen. F221 from β8 stacks with F221′ from β8′ of the opposing monomer and is located on the interface twofold axis (Fig. 5F). Substitution of the large bulky hydrophobic phenylalanine side chain in place of the small hydrophilic serine could well distort the local conformation around F221 and prevent stacking of F221 and F221′ (Fig. 5G). The S224F mutation would additionally disrupt the hydrogen bond formed between S224 and E226.

The S. enterica strain CL95 FabG contains two mutations, M125I and A223T. Interpretation of these mutations is less straightforward than those of the other strains. The M125I mutation is located close to the interface formed by the α4/α5 and α4′/α5′ four-helix bundle. Residues M122, M125, and M126 from α4 and M96′ from the other monomer were proposed to form hydrophobic clusters stabilizing the dimer interface (29). Replacement of M125 with the bulkier branched isoleucine side chain might affect the packing of helix α4 in the interface. A223T, the second mutation of the strain CL95 FabG, also resides adjacent to the dimer interface. The T223 hydroxyl group could make hydrogen bonds with the carbonyl oxygen atoms of Al220 and V219, leading to altered packing of helix α8, which in turn could pull F221 away from F221′ and destabilize the stacking of the side chains of these two residues, a crucial association of the dimer interface. Although we could not establish the relative importance of the two mutations in the phenotype of strain CL95, it seems likely that the A223T mutation plays a more important role in the destabilization of the FabG tetramer since the M125I substitution is a much more conservative change. Note that the action of mutations (such as E233K and S224F) within the interface that includes α6/α7 and α6′/α7′ could be more subtle than disruption of the FabG quaternary structure since this interface undergoes significant repositioning upon binding of the NADP⁺ cofactor and the conformational changes that accompany cofactor binding are thought to organize the active-site triad (29).