A Complex Lipoate Utilization Pathway in Listeria monocytogenes*

Although a complete pathway of lipoic acid metabolism has been established in Escherichia coli, lipoic acid metabolism in other bacteria is more complex and incompletely understood. Listeria monocytogenes has been shown to utilize two lipoate-protein ligases for lipoic acid scavenging, whereas only one of the ligases can function in utilization of host-derived lipoic acid-modified peptides. We report that lipoic acid scavenging requires not only ligation of lipoic acid but also a lipoyl relay pathway in which an amidotransferase transfers lipoyl groups to the enzyme complexes that require the cofactor for activity. In addition, we provide evidence for a new lipoamidase activity that could allow utilization of lipoyl peptides by lipoate-protein ligase. These data support a model of an expanded, three-enzyme pathway for lipoic acid scavenging that seems widespread in the Firmicutes phylum of bacteria.

Listeria monocytogenes is a Gram-positive bacterium of the phylum Firmicutes that is a serious human pathogen. Like a number of other bacteria, it is a natural lipoic acid auxotroph and relies upon exogenous sources of lipoic acid for growth. Utilization of lipoic acid requires that the cofactor become covalently attached to the ⑀-amino group of a conserved lysine residue of the lipoyl domain(s) of the enzyme complexes that require the cofactor for activity. The only known route for attachment of exogenous lipoic acid is by lipoic acid ligase action (1). L. monocytogenes employs two lipoate protein ligases for this purpose, LplA1 and LplA2 (2). Although either ligase suffices for lipoylation in cells grown in a rich medium, only lplA1 is essential for intracellular growth and virulence (2,3). The largest lipoyl peptide substrate utilized by L. monocytogenes is the DKA tripeptide, where lipoate is attached to the lysine ⑀-amino group (DK L A) by an amide linkage. LplA1 is required for efficient use of DK L A as a lipoate source (2). Prior work (2) indicated that expression of LplA1 but not LplA2 in Escherichia coli functionally replaced the host LplA ligase, the most thoroughly characterized lipoate ligase. The enzymatic properties that differentiate LplA1 from LplA2 were unknown.
It was proposed that LplA1 and LplA2 may have different protein interaction partners or that LplA1 may be able to transfer the lipoyl group from a synthetic lipoylated peptide, DK L A, to lipoate requiring enzymes by functioning as a lipoyl-amidotransferase (2). We conducted the present study to improve our understanding of lipoyl scavenge by L. monocytogenes and to determine the mechanism of DK L A utilization. It was possible that the DK L A lipoyl moiety could be utilized either by direct transfer of the lipoyl moiety or by hydrolysis with subsequent ligation of the lipoic acid released.
In Bacillus subtilis a novel amidotransferase, called LipL, was recently shown to be required for lipoic acid biosynthesis (4,5). Moreover, LipL homologues are present in all Firmicutes that use lipoic acid including those, such as L. monocytogenes, that lack the ability to synthesize the cofactor (and thus are natural lipoate auxotrophs). Moreover, although growth of B. subtilis lipL null mutants is supported by lipoic acid, the rate of growth is slow, and the E2 subunits of the pyruvate and branched chain dehydrogenase proteins (pyruvate dehydrogenase and BkdB, 2 respectively) are only partially modified (5). Because this slow growth requires the LplJ lipoate ligase, it seems that the presence of LipL increases the efficiency of scavenging by the ligase, although it is unclear if LipL transfers lipoic acid or octanoic acid (or both) to the dehydrogenase proteins of this bacterium. It also remains to be seen if LipL confers any competitive advantage beyond its requirement for biosynthesis of lipoic acid. The inability of L. monocytogenes to synthesize lipoic acid provides a more straightforward system than B. subtilis to study lipoate scavenge in a Firmicutes bacterium. Moreover, it seemed possible that the amidotransferase activity of the putative L. monocytogenes LipL might allow direct utilization of DK L A to modify the lipoic acid-requiring dehydrogenases essential for growth and pathogenesis of this bacterium.
It also remained possible that DK L A utilization as a lipoate source proceeds by hydrolysis of the amide linkage to give free lipoic acid. Such a lipoamidase activity is present in Enterococcus faecalis (6), another Firmicutes that lacks the ability to make lipoic acid. The E. faecalis lipoamidase activity was discovered as an activity that allowed a lipoate ligase to use various lipoamide compounds as substrates. Moreover, the lipoamidase was also able to cleave lipoic acid from intact lipoyl domains. E. coli, * This work was supported, in whole or in part, by National Institutes of Health Ruth L. Kirschstein National Research Service Award 5 T32 GM070421 (NIGMS) and Grants AI15650 (NIAID, to J. E. C.) and AI064540 (NIAID, to M. X. D. O.). 1 (2). The ability to use lipoylated peptides is dependent on LplA1 (2), but the mechanism of utilization was unknown (2). Although no obvious orthologue of the E. faecalis lipoamidase is encoded in the L. monocytogenes genome sequences, the possibility of utilization of DK L A by a lipoamidase of unrelated sequence remains.
In this study we report that L. monocytogenes LplA1 is a lipoyl ligase that is largely specific for modification of the H protein of the glycine cleavage system and that L. monocytogenese LipL is a reversible lipoylamidotransferase that participates in lipoic acid scavenging. We also provide evidence for a novel lipoamidase activity, which together with LplA1provides a pathway for utilization of the lipoyl moieties of lipoyl peptides.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids-Bacterial strains, plasmids, and oligonucleotides used are given in Table 1. Standard molecular biology methods were used unless otherwise indicated (8). Cloned genes were verified by sequencing performed by ACGT, Inc. L. monocytogenes EDGe gcvH (lmo2425), lipL (lmo2566), and the lipoyl domain of bkdB (lmo1374 or E2 BkdB ) were placed under the control of a T7 promoter for heterologous expression in E. coli. Gene sequences were amplified using Phusion poly-merase (New England Biolabs) according to the manufacturer's recommendations. The amplified products were inserted into the expression vector such that the protein products would be produced with a N-terminal hexahistidine tag The tag could be removed by cleavage with tobacco etch virus (TEV) protease. L. monocytogenes EDGe genomic DNA from the American Type Culture Collection was used as a template. Primers Q155 and Q156 were used to amplify gcvH, Q159 and Q160 for the bkdB lipoyl domain (E2 BkdB ), and Q161 and Q162 for lipL. The PCR products were inserted into plasmid pMCSG21 by ligation independent cloning as described (9) except for a modified T4 polymerase treatment. The treatment was performed with 1.5 g of DNA and 5 units of T4 polymerase in the supplied buffer (New England Biolabs) plus the appropriate deoxynucleotide triphosphates (2.5 mM) for 2 h at 37°C. The products were purified using a Qiagen plasmid mini kit (Qiagen).
L. monocytogenes lplA1 (lmo0931) and lplA2 (lmo0764) were cloned such that the proteins would be produced with an N-terminal Halo tag (Promega) under the control of phage T7 RNA polymerase. Primers Q131 and Q132 were used to amplify lplA2, whereas primers Q166 and Q167 were used to amplify lplA1 as described above. The PCR products and vector pFN18A were digested with SgfI and PmeI and ligated according to the manufacturer's instructions.
Complementation of an E. coli lipA lplA Strain-Strains QC242 and QC241 were derivatives of the lipA lplA strain BH285 (10) that contained plasmids expressing either LplA1 or LplA2, respectively. For complementation analyses the strains Halo-tagged LplA2 expression, TEV protease-cleavable This study pQC081 Hexahistidine-tagged LipL expression, TEV protease-cleavable This study pQC082 Hexahistidine-tagged GcvH expression, TEV protease-cleavable This study pQC084 Hexahistidine-tagged LDE2b expression, TEV protease-cleavable This study pQC085 Halo-tagged LplA1 expression, TEV protease-cleavable This study were cultured in M9 minimal medium with 0.4% glycerol, 5 mM acetate, 5 mM succinate, 10 M FeCl 3 (from a stock solution in HCl), and 100 g/ml sodium ampicillin overnight at 37°C. The cells were collected by centrifugation and washed three times with medium lacking acetate and succinate. The final culture density was measured and adjusted to an A 600 of 0.05 in the same medium lacking acetate and succinate with additional supplements as indicated. Growth assays were carried out in a Bioscreen C (Growth Curves USA) at 37°C with very strong shaking, and growth was measured by A 600 readings taken every 15 min.
To further purify lipoamide for use as a growth factor, it was dissolved in ethyl acetate and extracted three times with aqueous sodium bicarbonate, and the ethyl acetate solution was then taken to dryness. The product was dissolved in a minimal volume of ethanol at 42°C and filtered through a sintered glass funnel. The filtrate was then allowed to slowly cool first to room temperature and then to Ϫ20°C. The fluffy needle-like yellow crystals were collected by filtration and washed with cold ethanol. A second crystallization was then carried out.
Purification of Proteins-Buffered solutions were prepared at room temperature. Apolipoyl domains GcvH and E2 BkdB were prepared from strains QC235 and QC237, respectively, whereas apoLipL was prepared from strain QC243. The strains were cultured in 1 liter of LB supplemented with 5 mM acetate, 5 mM succinate, 0.2% glucose, 25 g/ml chloramphenicol, 50 g/ml streptomycin, and 50 g/ml spectinomycin overnight. These cultures were subcultured in the same medium except for the presence of 0.1% glucose and grown at 37°C to an A 600 of 0.5 when phage T7 RNA polymerase expression was induced (11) by the addition of arabinose to 0.2%. After 4 h of incubation, the cells were harvested by centrifugation at 8500 ϫ g for 7 min and frozen at Ϫ80°C. Hexahistidine-tagged proteins containing a TEV cleavage site (GcvH LM , E2 BkdB , LipL LM ) were purified by a modification of a subtractive immobilized metal affinity chromatography method previously described (11). The purification buffer was 50 mM sodium-MOPS (pH 7.6), 300 mM NaCl, 1 mM (tris(2-carboxyethyl)phosphine) (TCEP). TCEP was included in most purification and assay buffers because the octanoyltransferases and amidotransferases have active site cysteine residues. The cell pellets were thawed, and the cells were suspended in 25 ml of purification buffer containing 20 mM imidazole and 5 mM TCEP. The cells were lysed by passage twice through a French pressure cell at 20,000 p.s.i., and the extract was cleared by centrifugation at 46,000 ϫ g for 30 min. Hexahistidine-tagged proteins were batch-bound to 4 ml of agarose Ni-NTA resin (Qiagen) preequilibrated in buffer and mixed for 1 h at 4°C. The resin was packed into a 0.5-inch diameter column and washed with 50 ml of purification buffer containing 40 mM imidazole. The proteins of interest were then eluted with 10 ml of purification buffer containing 250 mM imidazole. The concentrations of the eluted proteins were estimated by measuring absorbance at 280 nm and the theoretical extinction coefficients for the cleaved proteins: 18,450 M Ϫ1 cm Ϫ1 for GcvH, 6,990 M Ϫ1 cm Ϫ1 for E2 BkdB , and 25,900 M Ϫ1 cm Ϫ1 for LipL. For hexahistidine tag cleavage, 1 mg of turbo TEV protease (Accelagen) was added for every 50 mg of protein, and the mixture was dialyzed against purification buffer for 14 h at 4°C. The cleavage reactions were then incubated with 2 ml of preequilibrated Ni-NTA resin for 1 h at 4°C with mixing. The resin should bind all proteins except those from which the hexahistidine tag had been cleaved. The resin was again packed in a column, and the flow-through was collected along with two 2-ml washes with purification buffer. The purified cleaved proteins were dialyzed against storage buffer: 50 mM sodium MOPS (pH 7.2), 100 mM NaCl, 1 mM TCEP, and 10% glycerol. The proteins were concentrated using a Vivaspin centrifugal concentrator (GE Healthcare), analyzed by SDS-PAGE, and flash-frozen in an ethanol dry-ice bath for storage at Ϫ80°C. Lipoyl domains were analyzed by electrospray ionization mass spectrometry-MS as previously described (12). Holo-LipL was purified from strain QC244. This strain was grown in LB medium containing 5 g/l lipoic acid to an A 600 of 0.5 at which time expression was induced by 50 M isopropyl ␤-D-1thiogalactopyranoside for 4 h. The rest of the purification is the same for apoLipL. Holo-LipL was analyzed by matrix-assisted laser desorption/ionization (MALDI-MS) mass spectrometry as previously described for LipM (12) and was found to have a mass of 31,147.8; this was 189.6 atomic mass units greater than that expected for the polypeptide chain, indicating that the protein was in the lipoylated form. Hexahistidine-tagged E2 BkdB was purified by the same method except without the TEV protease cleavage and second immobilized metal affinity chromatography steps. Lipoyl domains were further purified by anion exchange on a 1.8-ml POROS HQ 20 column using an AKTA Purifier 10 (GE Healthcare) at 5 ml/min in 25 mM sodium-MES (pH 6.1). The proteins were eluted with a 0 -1 M gradient of LiCl in the same buffer. The MALDI-MS determined masses of the E2 BkdB domain and the GcvH protein were 9,179.9 and 13,942.7, values that were greater by 1.7 and 0.4 atomic mass units, respectively, from the theoretical masses. The bands of the proteins that copurified with apoLipL were cut from an SDS-PAGE gel, the proteins were digested with trypsin, and the resulting peptides were subjected to liquid chromatographymass spectrometry analysis for identification as previously described for lipoyl domains (5).
LplA1 and LplA2 were purified from strains QC239 and QC240, respectively. The strains were grown at 37°C in LB with 50 g/ml streptomycin and 50 g/ml spectinomycin to an A 600 of 0.5 at which point they were shifted to 30°C for 1 h. Protein expression was induced with 10 M isopropyl ␤-D-1-thiogalactopyranoside for 16 h, and the cells were harvested and lysed as described above except in Halo buffer: 50 mM sodium HEPES (pH 7.5), 150 mM NaCl, 0.01% Tergitol Nonidet P-40, and 5 mM TCEP. Purification steps using Halo and Ni-NTA resins were carried out as recommended by the manufacturer. The pure untagged protein was dialyzed against storage buffer and flashfrozen for storage at Ϫ80°C. LplA1 was quantified using the calculated extinction coefficient of 34,380 M Ϫ1 cm Ϫ1 and had a mass of 38,382 as determined by MALDI-MS.
The multimeric state of LipL was determined using size exclusion chromatography with a Superdex 200 10/300 GL column (GE Healthcare) with 50 mM sodium phosphate buffer (pH 7.5) containing 150 mM NaCl using an AKTA Purifier10 at 0.5 ml/min. A standard curve was used to estimate size using chy-motrypsin A, albumen, aldolase, and catalase as standards. The reported value is the average of three injections.
Lipoic Acid Ligase Assay-The reactions were performed in assay buffer (50 mM sodium MOPS (pH 7.2), 100 mM NaCl, 5 mM TCEP) with 1 mM ATP, 1 mM MgCl 2 , 20 M GcvH, 1 M LplA1, and 1 mM acyl substrate. The acyl substrates were sodium lipoate, lipoamide, or DK L A where indicated. The reaction was separated by native PAGE on a 20% acrylamide gel and visualized with Coomassie Blue R-250. Modification with lipoic acid was assayed by Western blotting using anti-lipoic acid antibody (Calbiochem) and anti-mouse IgG conjugated to horseradish peroxidase (Calbiochem). The modification state of the lipoyl domain was also analyzed by electrospray ionization mass spectrometry as previously described (12).
Assay of Amidotransfer-To directly measure lipoyl amidotransfer by LipL, lipoyl-GcvH was synthesized as a substrate. The synthesis and purification was the same as for [1-14 C]octanoyl-E2 BkdB except GcvH and E. coli LplA were used in place of E2 BkdB and LplA1. Transfer of lipoate from 100 M lipoyl-GcvH to 20 M hexahistidine-tagged E2 BkdB by 1 M LipL was measured in assay buffer and incubated at 37°C for 1 h with a total reaction volume of 100 l. In place of lipoyl-GcvH, 1 mM lipoamide or 1 mM DK L A was used where indicated. Reactions were analyzed by SDS-PAGE and Western blotting as described for lipoyl ligation assays.
Octanoylation of lipoyl domains by LplA1 and subsequent amidotransfer by LipL was assayed using radiolabeled octanoate. The reaction was performed in assay buffer (50 mM sodium-MOPS (pH 7.2) and 5 mM TCEP with the addition of 1 mM ATP, 1 mM MgCl 2 , 250 M sodium [1-14 C]octanoate, 20 M E2 BkdB , 20 M GcvH, 1 M LplA1, and 1 M LipL) and incubated at 37°C for 1 h. Reaction components were omitted where indicated. Proteins were separated by SDS-PAGE on 4 -20% gradient acrylamide gels, stained with Coomassie R-250, soaked in Amplify (GE Healthcare), dried on filter paper, and exposed to film at Ϫ80°C for 8 to 20 h. The reaction products were analyzed by the method of Laskey and Mills (13).
To directly measure octanoyl amidotransfer by LipL in the reverse of previous assays, [1-14 C]octanoyl-E2 BkdB was synthesized as a substrate using E. coli LplA. E. coli LplA was purified by successive immobilized metal affinity chromatography and ion exchange chromatographic steps as described above for lipoyl domains (without cleavage and subtractive immobilized metal affinity chromatography). The reaction contained assay buffer with 1 mM sodium [1-14 C]octanoate, 2 mM ATP, 2 mM MgCl 2 , 0.5 mM E2 BkdB , and 10 M LplA and was incubated at 37°C for 2 h. [1-14 C]octanoyl-E2 BkdB was purified by ion exchange chromatography as described above for lipoyl domains. Transfer of octanoate from 20 M [ 14 C]octanoyl-E2 BkdB to 20 M GcvH by 1 M LipL was measured in assay buffer with incubation at 37°C for 1 h in a total reaction volume of 25 l. The reaction was analyzed as described above.
Assay of Lipoamidase Activity in Crude Extracts-Half-liter cultures of L. monocytogenes were grown to mid-log phase in brain heart infusion medium (Difco), pelleted, resuspended in 20 ml of 50 mM sodium MOPS (pH 7.2), 200 mM NaCl, 10% (v/v) glycerol, and 5 mM 2-mercaptoethanol and lysed by passage through a French pressure cell twice at 20,000 p.s.i.. Ammo-nium sulfate was added to 85% of saturation to precipitate proteins, and samples were gently agitated for an additional 30 min on ice. The precipitated protein was pelleted at 14,000 ϫ g for 10 min, flash-frozen in a dry ice ethanol bath, and stored at Ϫ80°C indefinitely. For use, the extract pellet was thawed on ice and resuspended in assay buffer containing only 1 mM TCEP and 0.1 mM EDTA. The slurry was dialyzed twice for 2 h, and insoluble material was removed by centrifugation at 14,000 ϫ g for 10 min. Protein contents were assayed by Coomassie Blue binding with the Bio-Rad protein assay kit using bovine gamma globulin as the standard. The lipoamidase assay reaction contained assay buffer with 0.5 mM MnCl 2 , 2.5 mM DK L A, and 100 g of extract protein. The assay was heated to 37°C for 5 min before the addition of extract. MnCl 2 was replaced with other metal salts where indicated, and to avoid oxidation of redoxsensitive metals, the reactions were performed in an anaerobic chamber (Coy Laboratory Products Inc.) under an atmosphere of 85% nitrogen, 10% hydrogen, and 5% carbon dioxide. Samples of 50 l were removed at various time intervals, and the reaction was quenched by the addition to 150 l of ice-cold acetone. Nonadecanoic acid (5 g) was added to each of the quenched samples as an internal standard. and the samples were acidified with 10 mol of HCl. The samples were derivatized for gas chromatography-mass spectrometry (GC-MS) by first drying under a stream of nitrogen, then adding 100 l of the silylation reagent N-methyl-N-(trimethylsilyl)trifluoroacetamide. The derivatization reaction was carried out for 1 h at 65°C. The derivatized samples were analyzed by GC-MS at the Carver Metabolomics Center of the University of Illinois. Samples (1 l) were injected in split mode (5:1) into a GC-MS system that consisted of an Agilent (Palo Alto, CA) 7890A gas chromatograph, an Agilent 5975 mass selective detector, and an Agilent 7683B autosampler. Injections were performed on a 60-m HP-5 column with a 0.25-mm inside diameter and a 0.25-mm film thickness (Agilent) with an injection port temperature of 250°C, the interface set to 250°C, and the ion source adjusted to 230°C. The helium carrier gas was set at a constant flow rate of 1.5 ml/min. The temperature program consisted of isothermal heating at 180°C for 5 min followed by an oven temperature increase of 5°C/min to 310°C for 2 min. The mass spectrometer was operated in positive electron impact mode at a 69.9-eV ionization energy in the m/z 50 -800 scan range. The spectra of all chromatogram peaks were evaluated using HP Chemstation (Agilent) and AMDIS (National Institute of Standards and Technology, Gaithersburg, MD)]. The spectra of all chromatogram peaks were compared with electron impact mass spectrum libraries NIST08 (National Institute of Standards and Technology) and WILEY08 (Palisade Corp.). This procedure was modified to detect the alanine and aspartate residues of the DK L A peptide using a temperature program consisting of isothermal heating at 70°C for 5 min followed by an oven temperature increase of 5°C/min to 310°C for 2 min.

RESULTS
The lplA1 and lplA2 Genes Differentially Complement an E. coli lplA Strain-To investigate the proposed mechanism of lipoyl peptide utilization, we conducted a series of genetic complementation experiments. LplA1 has been shown to be critical for lipoyl peptide scavenging during L. monocytogenes infection, whereas LplA2 was dispensable (2). Also, lplA1 expression was found to allow growth of a ⌬lipA ⌬lplA strain of E. coli (2). To further examine the functions of the lplA1 and lplA2 gene products, complementation of the growth of a ⌬lipA ⌬lplA strain of E. coli was assayed using differing derivatives of lipoic acid at different concentrations. The E. coli strain carried a ⌬lipA mutation because in this bacterium lipoic acid biosynthesis must be disrupted for growth to be affected by the ⌬lplA mutation (14,15). Expression of the L. monocytogenes lplA1 and lplA2 restored growth of the E. coli strain when the minimal medium was supplemented with 25 M sodium lipoate, which provided the first evidence that LplA2 functions as a lipoyl ligase. However, when the lipoate concentration was lowered to the more physiologically relevant concentration of 25 nM, only lplA1 allowed growth (Fig. 1A). LplA1 also allowed growth when the medium was supplemented with either lipoamide or DK L A at 25 M, although neither compound was active when added at 25 nM (Fig. 1, B and C). The finding that the amide-linked lipoate compounds were only active at high concentrations raised the possibility that the growth observed was due to the presence of contaminating lipoic acid in these preparations (either carried over from synthesis and/the result of hydrolysis). To test this possibility, we purified the commercial lipoamide by extraction followed by recrystallization and found that the purified lipoamide preparation was an appreciably poorer supplement. These data argued that LplA-supported growth of E. coli on the commercial lipoamide and DK L A preparations was due to trace amounts of contaminating lipoic acid (Fig. 1D).
L. monocytogenes LipL Is a Lipoyl Domain-specific Amidotransferase-Recent work has shown that the lipoate synthesis and scavenging pathways of B. subtilis are markedly more complex than the E. coli pathways (4,5). A key player in B. subtilis is LipL, an enzyme shown to catalyze amidotransfer of octanoate from octanoyl-GcvH to the pyruvate dehydrogenase lipoyl domain (4). The L. monocytogenes genome encodes a LipL homologue (lmo2566) that is 50% identical to B. subtilis LipL. However, given the diverse enzymatic activities of this family of proteins, it could not be assumed that the L. monocytogenes LipL has the same enzymatic activity as B. subtilis LipL. Moreover, given the parallel chemistry, it seemed possible that that the putative L. monocytogenes LipL might be able to catalyze transfer of lipoyl moieties from DK L A to apolipoyl domains.
To test this possibility in vitro, recombinant L. monocytogenes LipL was expressed in E. coli. Previous work with B. subtilis LipL showed that the protein expressed in E. coli was modified with lipoate and octanoate. Therefore, we attempted to purify the unmodified form (apoLipL) from an E. coli lipoate auxotroph and unexpectedly found that apoLipL copurified with several larger proteins ( Fig. 2A). Attempts to separate LipL from the larger proteins by ion exchange chromatography resulted in insoluble LipL aggregates. Excision of the bands of contaminating proteins from the gel, trypsin digestion, and LC-MS/MS analysis of the resulting peptides revealed the bands were the E1, E2, and E3 subunits of the E. coli pyruvate dehydrogenase complex. This suggests that like E. coli LipB (10), L. monocytogenes LipL binds to the pyruvate dehydrogenase complex. Upon LipL expression in a prototrophic E. coli strain, the protein was isolated with increased yield and purity (Fig.  2B). Size exclusion chromatography and MALDI-MS analysis of LipL indicates this preparation is monomeric (Fig. 2C) and is primarily in the single lipoylated holo form, although the spectra suggest that some apoLipL and some double-lipoylated LipL may also be present (Fig. 2D). In our analysis of B. subtilis LipL evidence was found for modification of a secondary site that was not required for catalysis, which may be the larger species detected in the mass spectra of Fig. 2D. The holo-LipL preparation was assayed for amidotransferase activity using lipoyl-GcvH as the lipoyl donor for modification of pyruvate dehydrogenase E2 subunit and was found to have good activity (Fig. 3A). However, DK L A and lipoamide were inactive as substrates (Fig. 3A), indicating that interaction with the substrate lipoyl domain was required for catalysis. As expected, free lipoic acid was also not a LipL substrate.
LplA1, LipL, and GcvH Are Required for a Lipoylation Relay-In vitro assays performed with the pure proteins showed that LplA1 modified GcvH but was unable to significantly modify E2 BkdB (Fig. 3B). These results are similar to those observed in in vitro assays with B. subtilis LplJ (4) but are in contrast to E. coli LplA, an enzyme that modifies many different lipoyl domains (13). Modification of E2 BkdB was only detected when both LipL and GcvH were present in the assay. Moreover, octanoyl-amidotransfer was reversible. LipL transferred the octanoyl moiety of pure [1-14 C]octanoyl E2 BkdB to GcvH, and the transfer required LipL (Fig. 3C). These data support that a lipoylation relay occurs during lipoic acid scavenging in L. monocytogenes and that two enzymatic activities are required for scavenging due to the specificity of the ligase.
LplA1 Is Unable to Utilize Amide-bound Lipoate Moieties as Substrates-The inability of DK L A and lipoamide at physiologically relevant concentrations to provide lipoyl moieties to E. coli strains expressing LplA1 or LplA2 could be explained by an inability of the compounds to enter the test bacterium. To more directly test the functions of LplA1 or LplA2, we expressed the proteins in E. coli and attempted their purification. The solubility of LplA1 was improved by expression at low temperatures and use of an N-terminal Halo tag (Promega) (Fig. 2B). We also attempted to purify LplA2 but found it was insoluble when expressed in E. coli under a variety of conditions. Lipoate-protein ligases require a lipoate moiety having a free carboxyl group to form the mixed anhydride intermediate, lipoyl-adenylate. One solution to this requirement would be to bypass lipoyl-adenylate formation by LplA1-catalyzed transfer of the lipoyl moiety to a target protein as previously suggested (2). In this scenario LplA1 would have both ligase and amido-transferase activities. The possibility of transfer of lipoate from DK L A or lipoamide to a GcvH target domain in vitro was directly tested with purified LplA1. When free lipoate was the substrate, lipoylation of GcvH was detected using both an anti-LA antibody and by mass spectrometry, whereas no lipoic acid attachment was seen when either DK L A or lipoamide were tested as substrates (Fig. 4). This together with the evidence of the presence of free lipoate in the lipoamide and DK L A preparations (see above) strongly suggested that L. monocytogenes contains a lipoamidase activity.
Crude Extracts Assay of L. monocytogenes Contain Lipoamidase Activity-Given that neither L. monocytogenes LplA1 nor LipL were able to use DK L A as a substrate, the most straightforward explanation for the prior observations was that L. monocytogenes encodes a lipoamidase activity that generates free lipoic acid by cleavage of the amide bond linking lipoate to the host peptides and thereby provides a substrate for LplA1. The only well characterized lipoamidase, a member of the Ser-Ser-Lys family of amidohydrolases, is found in E. faecalis, another Firmicutes bacterium that is a human pathogen and auxotrophic for lipoate (7). However, no recognizable lipoamidase orthologue is encoded in the L. monocytogenes genome.
To determine whether L. monocytogenes contains a divergent lipoamidase activity, we added DK L A to protein extracts of wild type L. monocytogenes and assayed release of free lipoate by GC-MS (Fig. 5A). The addition of L. monocytogenes protein extracts to DK L A followed by extraction and silyation of the products resulted in a peak having the retention time and mass spectra of the expected silylated lipoate, the formation of which was dependent on the presence of extract. Free amino acids  were also produced from DK L A. The lysine residue was completely silylated and eluted near lipoic acid (Fig. 5A), whereas the aspartate and alanine residues eluted earlier and were seen only when a lower starting temperature was employed (Fig. 5B). The major peaks from lipoic acid are the molecular ion of m/z 278 and the fragment of m/z 123 (Fig. 5C). A peak was found at m/z 155 indicating that the previous assignment (16) of this peak as a methyl ester fragment is incorrect. We also found strong peaks at m/z 73 and 75 that are characteristic TMS fragments (17). Consistent with what is observed with free fatty acids (17), the TMS derivative of lipoic acid yielded a more intense molecular ion peak than the methyl ester (16), allowing us to select for the m/z 278 species with good sensitivity.
To test if the lipoamidase activity was metal-ion dependent, the effects of EDTA on lipoamidase activity were tested, and the chelator was found to block formation of free lipoic acid. When EDTA-treated L. monocytogenes extracts were supplemented with various metal ions and lipoamidase activity was assayed by GC-MS, manganese ion proved to be the most effective ion, whereas magnesium ion was completely inactive (Table 2) (Fig.  5D). Cobalt and reduced iron ions gave intermediate activities.
These data suggest that L. monocytogenes, when located within eukaryotic host cells, relies on metal-dependent lipoamidase activity to liberate lipoate from lipoyl peptide substrates. The metal dependence of DK L A cleavage indicates that the lipoamidase activity of L. monocytogenes is distinct from E. faecalis lipoamidase, consistent with the lack of an lipoamidase orthologue. To confirm that lipoamidase activity is independent of LplA1, LplA2, and LipL, we assayed extracts of mutants deficient in these enzymes. Lipoamidase activity was present in a strain lacking both ligases as well as a strain lacking LipL (Table 2). These data confirm that the ligases are incapable of DK L A utilization in vitro and that utilization of lipoyl-peptide by L. monocytogenes requires a novel gene product (or products) having lipoamidase activity.

DISCUSSION
In this study we define a novel pathway for scavenging of lipoic acid by L. monocytogenes (Fig. 6). Previous work has shown that LplA1, but not LplA2, is required for intracellular utilization of lipoic acid (2). Although direct in vitro comparison of the two enzymes was precluded by the insolubility of LplA2, lplA1 was able to complement E. coli with much lower levels of lipoic acid supplementation than was lplA2. Thus, LplA1 may have a higher affinity for lipoate than LplA2, although there are other possible reasons for the apparent lower scavenging ability of LplA2. However, the apparent high affinity of LplA1 for lipoate would seem important for growth within host cells where the levels of free lipoic acid are limited. We found that LplA1-catalyzed lipoic acid attachment was specific for modification of GcvH. This specificity engenders a requirement for amidotransfer of lipoyl moieties from GcvH to the 2-oxoacid dehydrogenases required for growth (18). Lipoyl  Recently, a transposon insertion into the gene encoding the L. monocytogenes LipL ortholog was isolated. The strain was attenuated in its ability to grow in bile (19). Growth was not restored upon the addition of lipoic acid to the medium, which is surprising given the two lipoate protein ligases of L. monocytogenes. These results provide physiological evidence that LipL is required for lipoyl scavenging in Listeria and are consistent with our findings that lipoyl ligation is specific for GcvH, a specificity that engenders a requirement for LipL to activate the essential dehydrogenases (Fig. 4B). The reduced tolerance of the lipL mutant strain to acidic bile salts (19) is likely due to a decreased branched fatty acid content of the membrane phospholipids, which results when L. monocytogenes is deficient in lipoic acid (18).
Although much attention has been given to lipoic acid ligation, lipoamidase action is poorly defined. Lipoic acid is assembled covalently bound to cognate enzymes on the enzymes themselves (20), and yet lipoate-protein ligases are widely distributed. This suggests that lipoamidases, which cleave the amide bond that links lipoic acid to its cognate enzymes, must also be widely distributed. Early studies of E. faecalis found that lipoamidase activity was present in two separate fractions, suggesting there may be two enzymes (6). This was also seen in a later purification of lipoamidase (21). Jiang and Cronan (7) found only a single lipoamidase-encoding gene, but it is possible that other lipoamidase genes were not detected because their protein products have different specificities.
L. monocytogenes extracts contain significant lipoamidase activity. This activity is independent of lplA1, lplA2, and lipL, FIGURE 6. Model for intracellular scavenge of lipoyl groups by L. monocytogenes. Lipoic acid is cleaved from lipoyl peptides through the action of a lipoamidase (Lpd). Free lipoic acid is ligated to the unmodified glycine cleavage system H protein (GcvH), which is the preferred substrate of LplA1. GcvH encounters apoLipL tethered to apo-2-oxoacid dehydrogenase complexes where LipL transfers the lipoyl group to the lipoyl domain (LD) of requiring enzymes. demonstrating that at least one additional gene is required for lipoic acid scavenging. The strong manganese ion dependence of this lipoamidase activity indicates hydrolysis is performed by a mechanism that differs from that of the metal-independent E. faecalis enzyme lipoamidase (7). Note that recent evidence demonstrates under oxidative stress conditions manganese can functionally replace reduced iron in iron-dependent enzymes (22). Thus, bacteria can mitigate the oxidative stress defense that mammalian hosts use against infection (23). The use of manganese by the L. monocytogenes lipoamidase may be a means to combat oxidative stress and also an adaptation to the iron limitation of the intracellular environment (23). It is clear from this study that more work is required to fully appreciate the diversity and abundance of lipoamidases, although the activity has been known for more than 50 years. However because the known enzymes are from pathogenic bacteria, biosafety considerations make enzyme discovery by fractionation of cell extracts highly problematical.
An interesting possibility is that a LipL relative could function as a lipoamidase by transferring the acyl group to water instead of to another lipoyl domain. We considered that L. monocytogenes LipL may be capable of employing water as an acyl acceptor and, therefore, acting as a lipoamidase but did not see appreciable loss of acyl groups from lipoyl domains in the presence of LipL. However, it remains possible that this activity exists in one of the many uncharacterized clades of this protein family (see below). The likelihood of this possibility could be tested by engineering a lipoamidase from an amidotransferase, which would be essentially the reverse of peptide transfer by proteases (24).
Finally, from characterization of the LipL amidotransferases of L. monocytogenes and B. subtilis, we can now better predict the function of related proteins. Amidotransferases are members of the cofactor transferase family (Pfam entry PF03099) and are distinct from lipoic acid ligases (Fig. 7). Also, the LipL clade presented has a significant bootstrap value that defines the clade and its relationship to other clades (Fig. 7B). LipL orthologues can be found in all Firmicutes that use lipoic acid. Other more deeply branching LipL homologues are also in the highlighted clade and are from order Pseudomonadales, class Halobacteria, and class Chloroflexi. Given that the gene distribution is somewhat independent of taxonomy, lipL was presumably distributed by ancient horizontal gene transfer events.
The overall phylogeny of LipL and relatives combined with new functional information yields some interesting observations. In the LplA family, the more deeply branching clades also include LipL lipoyl-amidotransferases, the LipM octanoyltransferase (12), and bipartite lipoate protein ligases (25). These enzymes do not have an attached accessory domain, although the bipartite lipoyl ligases can use a separate protein as the accessory domain. The related LipB octanoyltransferases (Fig.  7A) also do not have an accessory domain. This suggests the ancestral protein lacked an attached accessory domain, which would presumably be required for efficient ligase activity. The function of the ancestral protein is not clear, although a transferase active in modification of lipoyl or biotinoyl domains would be expected. This function is common to all character-ized members of this family. The reversibility of this reaction is used by LipL to transfer modifications.
Another characterized transferase is present in the LplA family, the mammalian lipoyltransferase (Fig. 7B). Lipoyltransferase transfers lipoic acid from lipoyl-adenylate to lipoyl domains but unlike the bacterial ligases is unable to form the adenylate intermediate (26). However, we lack evidence demonstrating the physiological role of lipoyltransferase. An lipoyltransferase orthologue from yeast, LIP3, has been shown to be required for lipoic acid biosynthesis (27). Loss of LIP3 results in a yeast strain that accumulates lipoylated H protein but shows no lipoylation of 2-oxoacid dehydrogenases. This is the same result observed in B. subtilis upon inactivation of lipL (5). Because LIP3p and lipoyltransferase are present in the same small clade (Fig. 7B), it appears that these proteins may represent another class of lipoyl-amidotransferases.