The Escherichia coli lipB Gene Encodes Lipoyl (Octanoyl)-Acyl Carrier Protein:Protein Transferase

Bacterial strains and media.

The media used were described previously (23), as were strains TM136 (12) and KER184 (12) and plasmid pKR111 (17).

Isolation of temperature-sensitive lipB mutants.

Plasmid pKR111, which contains the complete lipB gene plus 127 bp of upstream sequence in a high-copy-number plasmid, was mutagenized with hydroxylamine for various time periods according to the protocol of Miller (10). A hydroxylamine treatment time yielding a 50% reduction in the number of ampicillin-resistant transformants was chosen. The hydroxylamine-treated plasmid preparations were then transformed into the lipB::Tn1000dKan lplA::Tn10 strain TM136, and the transformants were plated on glucose minimal E medium containing sodium ampicillin (100 μg/ml), tetracycline hydrochloride (15 μg/ml), and kanamycin sulfate (50 μg/ml), followed by incubation of the plates at 30°C for 2 days. Eight hundred of the resulting colonies were then tested on glucose-minimal plates containing ampicillin incubated at 30, 37, or 42°C. A parallel set of plates containing 5 ng of lipoate/ml was also prepared and was incubated at 37 or 42°C for 2 days. Three isolates were found that grew at 30°C without lipoate but which failed to grow at 37 or 42°C unless supplemented with lipoate. The plasmids from these strains were designated pSJ65, pSJ67, and pSJ68, and their lipB genes, together with that of the parental plasmid pKR111, were sequenced by the Keck Genomics Center of the University of Illinois.

Construction of a plasmid encoding a hexahistidine (His)-tagged LipB.

Two plasmids were constructed, each of which contained an NcoI site overlapping the lipB initiation codon to be tested. Plasmid pSJ118, which encoded a gene beginning with the previously assigned ATG initiation codon, was produced by PCR amplification from plasmid pCTV616 with the primer LIPBATG (5′-CCCCCATGGATGAATTCACCGATACCCGCCATG-3′) and primer LIPBSTOP (5′-CCCAAGCTTAAGGGGTAATATATTCGAAGTC-3′). The PCR product was cloned into plasmid pCR2.1 by using the TA cloning kit (Invitrogen) to produce plasmid pSJ118. Plasmid pSJ112 resulted from the same manipulations except that primer LIPBTTG (5′-CCCCCATGGATCAGGATAAAATTCTTGTC-3′) was used as the upstream primer and the encoded gene started at the TTG codon (which was changed to ATG in making the NcoI site). The NcoI-HindIII lipB fragments of plasmids pSJ112 and pSJ118 were ligated to vector pKK233-2 (Pharmacia) cut with the same enzymes to yield plasmids pSJ115 and pSJ119, respectively. The same manipulations, except that vector pET30(a) (Novagen) was used in place of vector pKK233-2, yielded plasmids pSJ120 and pSJ122 derived from plasmids pSJ112 and pSJ118, respectively.

Purification and assay of LipB and LplA.

His-tagged LipB was purified by using Qiagen immobilized nickel-agarose matrix according to the protocol supplied by Qiagen. His-tagged LipB was then further purified by chromatography on a Mono-Q anion-exchange column (Pharmacia). The sample was adsorbed to the Mono-Q column by using 50 mM Tris-HCl (pH 7.4) containing 5% glycerol, and His-tagged LipB was eluted by using a gradient of 0 to 250 mM ammonium sulfate in the same buffer. LplA was purified exactly as described previously (11). Lipoylation activity was measured by the pyruvate dehydrogenase activation and gel mobility shift assays described previously (7, 8). In the former assay, lipoylation is measured by conversion of the inactive apo form of pyruvate dehydrogenase to the active lipoylated form (7), whereas in the latter assay lipoylation or octanoylation is measured by shift of the electrophoretic mobility of a purified lipoyl domain derived from E. coli pyruvate dehydrogenase (6, 8). Upon acylation of the target lysine, the electrophoretic mobility of the lipoyl domain increases due to loss of the positively charged side chain.

Synthesis of lipol-adenylate.

The lipoyl-AMP analogue, lipol-adenylate (LipOH-AMP) was made as follows. Lipoic acid was reduced to the alcohol (lipol) with 1.25 eq of BH₃ in tetrahydrofuran (2). This reagent does not reduce disulfides (2). After hydrogen evolution was complete, 15 ml of a 1:1 mixture of water and tetrahydrofuran was added to decompose any remaining reducing agent. The solution was then saturated with potassium carbonate, and lipol was extracted with ethyl ether. The extracts were dried with sodium sulfate, and the solvent was removed by rotary evaporation to give a quantitative yield of product. The 5′-phosphate of 2′,3′-O-isopropylidene adenosine was synthesized from 2′,3′-O-isopropylidene adenosine and 2-cyanoethyl phosphate as described by Tenner (21) and purified on a column of DEAE-Sephadex A25 equilibrated in 5 mM ammonium bicarbonate. Any remaining 2′,3′-O-isopropylidene adenosine was eluted with 5 mM ammonium bicarbonate, whereas 2′,3′-O-isopropylidene adenosine-5′-phosphate was eluted with 0.2 M ammonium bicarbonate. This solution was taken to dryness by rotary evaporation, dissolved in water, and again evaporated to dryness in order to remove ammonium bicarbonate. The 2′,3′-O-isopropylidene adenosine-5′-phosphate (2.66 mmol) was then converted to the pyridinium salt (1), which was mixed with a solution of 10 mmol of lipol in dry pyridine. To this mixture 10 mmol of the coupling reagent 1,3-dicyclohexylcarbodiimide was added, and the solution was then stirred for 40 h at room temperature (1). Water was added to decompose any remaining coupling reagent. The solution was filtered and the filtrate was taken to dryness by rotary evaporation. The residue was dissolved in water and again evaporated in order to remove residual pyridine. The residue was then taken up in 150 ml of 20% acetic acid and refluxed for 2 h to remove the isopropylidene protecting group (21). After it cooled, the solution was again filtered and evaporated to dryness. Thin-layer chromatography (silica gel 60) in 2-propanol-concentrated NH₄OH-water (7/3/1, by volume) showed that the major contaminant was AMP. The mixture was then loaded on a DEAE-Sephadex A25 column equilibrated in 20 mM triethylammonium bicarbonate. LipOH-AMP was eluted with 0.2 M triethylammonium bicarbonate (AMP remained on the column) and evaporated to dryness. The product showed a UV absorption maximum at 259 nm for the adenine ring with a shoulder at 302 nm. The shoulder was attributed to the thiolane ring since it disappeared upon treatment with the disulfide reducing agent tris(2-carboxyethyl)phosphine. Electrospray mass spectroscopy gave a mass of 536.2 for the product versus the calculated value of 536.1 (ammonium salt).

Isolation of a temperature-sensitive lipB mutant.

Given our inability to prove by biochemical means that lipB did (or did not) encode the lipoyltransferase, we turned to a classic genetic approach: the isolation of a temperature-sensitive mutant. The argument was that if lipB was the lipoyltransferase structural gene, a temperature-sensitive lipB mutant strain should encode a lipoyltransferase of increased thermolability. If no change in thermal stability of the lipoyltransferase activity was seen, a regulatory role was indicated for LipB. Plasmid pKR111, which carries a functional lipB gene, was mutagenized with hydroxylamine, and the mutagenized plasmids were then transformed into the lipB mutant strain KER184 (lipB::Tn1000dKan). Among these transformants, three strains were found that grew at 30°C without lipoate but which failed to grow at the higher temperatures unless supplemented with lipoate. This phenotype indicated that the temperature-sensitive nature of these strains was due to a mutation in the plasmid-borne lipB gene. The plasmids of these three strains were purified and again used to transform strain KER184 to confirm that the temperature-sensitive mutation was carried by the plasmid. The lipB genes of the three of these plasmids—pSJ65, pSJ67, and pSJ68—plus the parental plasmid were sequenced. Plasmids pSJ65 and pSJ68 had mutations within the putative promoter region upstream of lipB, whereas the pSJ67 lipB gene contained a nonsense mutation in codon 74 (CAG to TAG) that we denote as the lipB67 allele.

The in vitro temperature stabilities of the lipoyltransferases of the mutant strains were tested by treating cell extracts at 55°C for various periods of time before being placed on ice. The extracts were then assayed for lipoyltransferase activity by the pyruvate dehydrogenase activation assay (Fig. 2). The lipoyltransferase activities of strains SWJ175, SWJ178, and SWJ181 (carrying plasmids pSJ65, pSJ68, and pKR111, respectively) had similar temperature stabilities, whereas that of strain SWJ177 (carrying plasmid pSJ67) had a severely decreased temperature stability (Fig. 2). Therefore, the temperature-sensitive lipB allele results in the production of a lipoyltransferase of increased thermal lability. These data demonstrate that lipB is the gene that encodes the lipoyl (octanoyl) transferase and precludes an indirect regulatory role for the protein in lipoate metabolism.

Overproduction of LipB and lipoyltransferase activity from an altered TTG start codon.

Encouraged by the genetic findings we returned to the problematic overproduction of LipB. Previous attempts in this laboratory to overproduce LipB from an open reading frame (ORF) that encoded a 191-residue protein had consistently failed despite testing a large number of transcriptional and translational constructs (17; S. W. Jordan and J. E. Cronan, Jr., unpublished data). However, we had recently tested two constructs containing a putative Arabidopsis thaliana lipB homolog for the ability to complement an E. coli lipB-null mutant (24). One A. thaliana construct contained an ORF with a length similar to that which we and others had assigned for the E. coli lipB gene, whereas the second construct included a putative N-terminal leader sequence thought to be a mitochondrial targeting sequence. We expected that only the shorter ORF would complement, but instead the opposite result was obtained; only the longer ORF had complementation activity (24). Inspection of the putative A. thaliana leader sequence showed that the deduced amino acid sequence matched the amino acid sequence found upstream of the E. coli lipB methionine residue assigned as the initiation codon (Fig. 3). Moreover, the deduced A. thaliana sequence contained a methionine codon that was within one residue of alignment with a TTG leucine codon, a known low-efficiency E. coli initiation codon (20). Therefore, it seemed probable that the initiation codon had been mistakenly assigned through the usual practice of selecting the first methionine codon. Indeed, by aligning the amino acid sequences deduced from the DNA sequences (rather than those of the annotated ORFs) of lipB homologues from many diverse bacteria we found excellent conservation of the amino acid sequences upstream of the initiation codons assigned to the genes. Strikingly in organisms closely related to E. coli (e.g., Salmonella enterica serovar Enteritidis and Klebsiella pneumoniae), the leucine codon was the infrequently used TTG codon despite the divergence at the third positions of codons often observed between these bacteria and E. coli. Moreover, other bacteria were found to have an ATG methionine codon or another alternative initiation codon, GTG (valine), which aligned within a codon or two of the E. coli TTG codon (Fig. 3). Therefore, the TTG codon seemed very likely to be the true E. coli lipB initiation codon and would yield a 213-residue protein rather than the 191-residue protein previously tested. We therefore constructed a plasmid in which a NcoI site was introduced that overlapped the putative TTG initiation codon of the 213 residue ORF (the TTG codon was converted to ATG by this manipulation) such that conventional protein expression vectors could be used. An analogous construct in which the NcoI site overlapped the putative methionine initiation codon of the 191-residue ORF was also made. Upon induction of strains carrying the two plasmids only the 213-residue ORF produced a detectable protein product of the expected size (Jordan and Cronan, unpublished). From these data we concluded that the 213-residue protein was the stable native form, whereas the shorter version was degraded by proteolysis in vivo.

We therefore proceeded to construct plasmids encoding N-terminal His-tagged versions of both the 191-residue and the 213-residue ORFs. Upon induction of lipB expression, sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 4A) analysis showed a strong band of ∼29 kDa in the strain that expressed the longer protein encoded by plasmid pSJ120. This corresponded well to the expected 28.7-kDa mass of the longer His-tagged LipB fusion protein (Fig. 4A). The His-tagged version of the 191-residue protein encoded by pSJ122 gave no obvious band in the expected region of the gel (predicted size of 26.1 kDa) (Jordan and Cronan, unpublished). Therefore, the shorter protein was unstable even when an artificial N-terminal extension (the His tag) was added. However, plasmids (pSJ112 and pSJ118, respectively) that expressed either the 191-residue ORF or the 213-residue ORF both complemented lipB null mutants (Jordan and Cronan, unpublished). These results suggest that the additional residues of the 213-residue ORF are not critical for enzyme activity but are necessary for stability of the protein in vivo.

Purified His-tagged LipB has octanoyl-ACP transferase activity.

Overexpression of LipB from plasmid pSJ120 increased the specific activity of lipoyltransferase relative to nonoverproduced extracts by ∼1,200-fold. The His-tagged LipB was purified directly from cell extracts by immobilized nickel-chelate column chromatography, followed by ion-exchange chromatography to obtain a homogeneous protein (Fig. 4). The purified LipB protein had high levels of lipoyltransferase activity in both assays. The gel mobility shift assay was used to test the ability of LipB to utilize octanoyl-ACP (Fig. 5) or lipoyl-ACP (data not shown) to modify the lipoyl domain. As expected from prior results with cell extracts (9), the purified His-tagged LipB converted the apo form of a lipoyl-domain to the holo domain by using either octanoyl-ACP or lipoyl-ACP. When LipB was tested for the ability to use ATP plus octanoic acid (Fig. 5) or lipoic acid (data not shown), no modification was detected. The ability of the purified LipB preparation to transfer lipoate from lipoyl-ACP to pyruvate dehydrogenase was determined by the activation of apo-pyruvate dehydrogenase. LipB was found to follow Michaelis-Menten kinetics (data not shown) with a K_m for lipoyl-ACP of ∼1 μM (Table 1).

Purified LplA has both lipoyltransferase activity and lipoate ligase activity.

Previous studies on LplA indicated that LplA utilized free lipoic acid plus ATP to modify lipoyl domains (11), but at that time lipoyl-ACP had not yet been shown to be a lipoyl donor and thus the ability of LplA to utilize this substrate had not been tested. However, overproduction of LplA was found to allow growth of a lipB-null mutant in the absence of exogenous lipoic acid (12), suggesting either that LplA might be able to catalyze the lipoyltransferase reaction at a low rate or that lipoyl-ACP was cleaved to give traces of lipoic acid that could then be salvaged by high levels of LplA. The activity of LplA in the postulated alternate reaction would be low; otherwise, lipB-null mutant strains with a normal lplA gene would not have a lipoate auxotroph phenotype. Consistent with a low activity, lipB-null mutant strains were found to grow very slowly in the absence of lipoic acid (17, 23) and to contain only low levels of lipoylated proteins (12, 17). In strains with null mutations in both lipB and lplA there is no residual growth in the absence of lipoic acid and no detectable protein lipoylation (12). Although these genetic and physiological data were consistent with LplA possessing a low level of lipoyltransferase activity, direct proof was lacking.

In order to test whether LplA possessed lipoyltransferase activity, we tested the ability of purified LplA to use octanoyl-ACP and lipoyl-ACP as substrates in vitro. We used the gel mobility shift assay to assess the ability of LplA to use octanoyl-ACP as a substrate in place of free octanoic acid plus ATP (Fig. 5). LplA did indeed use octanoyl-ACP as a substrate, although extremely inefficiently. The octanoyl-ACP reaction required ca. 100-fold more enzyme than was required for comparable activity with the ligase substrates, octanoic acid plus ATP. LplA also used lipoyl-ACP as a substrate, as detected by the both the gel mobility shift (data not shown) and pyruvate dehydrogenase activation assays. The latter assay gave a K_m of ∼9 μM for the reaction of LplA with lipoyl-ACP (Table 1). Since the activity observed with ACP-bound substrates required high levels of LplA, it seemed possible that activities observed with octanoyl-ACP and lipoyl-ACP as substrates were due to LipB contamination of our LplA preparation. To test this possibility, we purified LplA that was overproduced in a lipB-null mutant strain and obtained an enzyme preparation that had the same activity on the ACP-bound substrates as the preparations from a wild-type host strain (data not shown). In another test we synthesized a nonhydrolyzable lipoyl-AMP analogue, lipol-AMP (Fig. 6), that, based on the corresponding aminoacyl-AMP analogues (13, 18), was expected to inhibit the reaction with free lipoic acid (plus ATP) catalyzed by LplA by occupying the active site. Indeed, lipol-AMP was found to be a competitive inhibitor of the reaction, although the analogue was a rather impotent inhibitor (Fig. 7A), suggesting that the lipoate carboxyl group plays an important role in lipoyl-AMP binding. We expected that, if the two reactions catalyzed by LplA used the same active site, then lipol-AMP should also inhibit transfer of lipoate from lipoyl-ACP to apo-pyruvate dehydrogenase. Indeed, the analogue also competitively inhibited the lipoyl-ACP reaction, indicating that both reactions were catalyzed by the same active site (Fig. 7B). Therefore, we conclude that a low level of transferase activity with ACP thioester donors is an intrinsic property of LplA. It should be noted that parallel experiments with purified LipB and lipoyl-ACP showed no inhibition of the LipB-catalyzed lipoate transfer reaction (data not shown).