Lipoylation, the covalent attachment of lipoic acid to 2-oxoacid dehydrogenase multi-enzyme complexes, is essential for metabolism in aerobic bacteria and eukarya. In Escherichia coli, lipoylation is catalysed by LplA (lipoate protein ligase) or by LipA (lipoic acid synthetase) and LipB [lipoyl(octanoyl) transferase] combined. Whereas bacterial and eukaryotic LplAs comprise a single two-domain protein, archaeal LplA function typically involves two proteins, LplA-N and LplA-C. In the thermophilic archaeon Thermoplasma acidophilum, LplA-N and LplA-C are encoded by overlapping genes in inverted orientation (lpla-c is upstream of lpla-n). The T. acidophilum LplA-N structure is known, but the LplA-C structure is unknown and LplA-C's role in lipoylation is unclear. In the present study, we have determined the structures of the substrate-free LplA-N–LplA-C complex and E2lipD (dihydrolipoyl acyltransferase lipoyl domain) that is lipoylated by LplA-N–LplA-C, and carried out biochemical analyses of this archaeal lipoylation system. Our data reveal the following: (i) LplA-C is disordered but folds upon association with LplA-N; (ii) LplA-C induces a conformational change in LplA-N involving substantial shortening of a loop that could repress catalytic activity of isolated LplA-N; (iii) the adenylate-binding region of LplA-N–LplA-C includes two helices rather than the purely loop structure of varying order observed in other LplA structures; (iv) LplAN–LplA-C and E2lipD do not interact in the absence of substrate; (v) LplA-N–LplA-C undergoes a conformational change (the details of which are currently undetermined) during lipoylation; and (vi) LplA-N–LplA-C can utilize octanoic acid as well as lipoic acid as substrate. The elucidated functional inter-dependence of LplA-N and LplA-C is consistent with their evolutionary co-retention in archaeal genomes.

INTRODUCTION

Aerobic metabolism of 2-oxoacids and C1 metabolism are dependent on LA (lipoic acid) in a highly conserved manner [1]. LA is an essential co-factor of the OADHCs (2-oxoacid dehydrogenase complexes), which include the PDHC (pyruvate dehydrogenase complex), OGDHC (2-oxoglutarate dehydrogenase complex) and BCOADHC (branched-chain 2-oxoacid dehydrogenase complex), and of the GCS (glycine cleavage system). OADHCs comprise multiple copies of three proteins: 2-oxoacid decarboxylase (E1), dihydrolipoyl acyltransferase (E2) and dihydrolipoamide dehydrogenase (E3). E2 comprises E2lipD [E2 lipoyl domain(s)], a PSBD (peripheral subunit-binding domain) and a catalytic domain. E2lipD is the post-translational modification target: LA is covalently attached to E2lipD via an amide linkage to the ϵ-amino group of a specific lysine located at the tip of a β-turn. Once attached, the lipoyl moiety acts as a swinging arm that shuttles substrates/intermediates between the active sites of E1, E2 and E3. In the GCS, LA is attached to a lysine of the H protein that is structurally homologous with E2lipD.

In Escherichia coli, E2lipD lipoylation is catalysed by LipA (lipoic acid synthetase) and LipB [lipoyl(octanoyl) transferase] or, if LA is present in the medium/environment, by LplA (lipoate protein ligase) [24]. LipB and LipA work in tandem: LipB catalyses the covalent attachment of OA (octanoic acid, derived from the fatty acid biosynthetic pathway) to E2lipD, and LipA introduces sulfur atoms at the C6 and C8 positions. LplA is typically a single polypeptide comprising an N-terminal domain (approximately 250 residues) that has an LA-binding site, and a smaller C-terminal domain (approximately 90 residues) [5]. In E. coli and Oryza sativa, LplA can catalyse both steps of the lipoylation process: conversion of LA into lipoyl-AMP (lipoate adenylation) and subsequent covalent attachment of the lipoyl moiety to E2lipD (lipoate transfer) [6,7]. Mammals achieve the equivalent process using two enzymes, lipoate-activating enzyme and an LplA-like LPT (lipoyltransferase) [8]. In yeast, four enzymes are involved in lipoylation, including homologues of LipA, LipB and LplA, but there are major differences compared with lipoylation in E. coli [9]. While LA metabolism in bacteria other than E. coli is not fully understood, details elucidated to date indicate that numerous variations exist [1012].

Available structures include E. coli LplA [13] (PDB entries 1X2G, 1X2H, 3A7A and 3A7R), streptococcal LplAs (PDB entries 2P0L and 1VQZ) and a mammalian LPT [8] (PDB entries 2E5A and 3A7U). The LplA N-terminal domain belongs to the α/β class of proteins [3,5] and is structurally homologous with and evolutionarily related to the central catalytic domain of biotin protein ligase and class II aminoacyl-tRNA synthetase [14,15]. LplA C-terminal domain comprises three α-helices and two 310-helices packed against a three-stranded β-sheet [5]. Bovine LPT resembles LplA in that it comprises a larger N-terminal domain and a smaller C-terminal domain, both with similar folds to their respective E. coli counterparts. The overall conformation of lipoyl-AMP-bound LPT is, however, stretched relative to unliganded E. coli LplA due to rotation of the C-terminal domain by approximately 180° with respect to the N-terminal domain [8]. A similar rotation of the C-terminal domain relative to its apo orientation was observed in the crystal structure of lipoyl-AMP-bound E. coli LplA [13]. In the same structure, it was noted that two important loops also undergo conformational change upon lipoate adenylation: the adenylate-binding loop (residues 165–184), which is either partially disordered or not close to the active site in apo-LplA, covers the adenylate of lipoyl-AMP and interacts intimately with it, and the lipoate-binding loop (residues 69–76) is pulled towards lipoyl-AMP [13]. In the reaction scheme proposed by Fujiwara et al. [13], these conformational changes allow E. coli LplA to accommodate the lipoate acceptor domain/protein (E2lipD or H protein) and hence to catalyse the lipoate transfer step. These authors note, however, that in their LplA–apoH complex crystal structure with octanoyl-AMP rather than lipoyl-AMP, the distance between octanoyl-AMP and the acceptor lysine residue (LysApoH64) is too great for initiation of lipoyl transfer. In the same study, Fujiwara et al. [13] also observed that, unlike E. coli LplA, bovine apo-LPT adopts the same relative N- and C-terminal domain orientations as lipoyl-AMP-bound LPT [13].

Archaeal LplA studies have been conducted largely in Thermoplasma acidophilum, a species that possesses genes encoding individual proteins that resemble the N- and C-terminal domains of non-archaeal LplA. We term these gene products LplA-N and LplA-C. Structures of T. acidophilum LplA-N in unliganded (PDB entries 2ARS and 2C7I), lipoyl-AMP-bound (PDB entry 2ART), LA-bound (PDB entry 2C8M) and ATP-bound (PDB entry 2ARU) forms exhibit the same overall fold as the non-archaeal LplA N-terminal domain [3,16]. It was shown in crystal soaking experiments (1 day soaks) that LplA-N can catalyse lipoate adenylation to form lipoyl-AMP [16], but LplA-N is unable to catalyse lipoate transfer in vitro and an accessory protein was suggested [3]. We subsequently showed that LplA-N requires LplA-C to carry out lipoylation (corroborated using complementation assays in E. coli [17]) and that lipoylation occurs in vivo [18].

Comparative genomic analyses across 115 archaeal genomes (M.G. Posner, A. Upadhyay, M.J. Danson, S. Dorus and S. Bagby, unpublished work) show that archaeal species capable of lipoylation retain either the LplA or LipA–LipB system with 81% (61 out of 75 species) retaining LplA. Despite the evolutionary predominance of LplA in the archaea, and the fact that LplA-C is essential for lipoylation, no mechanistic information exists concerning co-ordination of LplA-N and LplA-C function. In the present study we have used structural and biochemical methods to investigate the role of LplA-C. We present structures of the T. acidophilum LplA-N–LplA-C complex and of E2lipD, show that LplA-C folding is driven by association with LplA-N, and that LplA-C induces localized conformational change in LplA-N, and we use NMR to monitor LplA-N–LplA-C interactions with LA/ATP and E2lipD, and to monitor E2lipD lipoylation.

EXPERIMENTAL

Expression and purification of T. acidophilum LplA-N–LplA-C and T. acidophilum E2lipD

pET19b-lpla and pET24a-ctd [18] were co-transformed into E. coli BL21(DE3) cells and expression was induced with 0.25 mM IPTG (isopropyl β-D-thiogalactopyranoside) at 16°C overnight. Harvested cells were sonicated, the lysate was centrifuged at 21000 g for 40 min and LplA-N–LplA-C complex was purified using His MultiTrap™ FF and His MultiTrap™ HP columns (GE Healthcare). The final LplA-N–LplA-C complex purity was >95% as judged by SDS/PAGE. T. acidophilum E2lipD was expressed and purified as described previously [18].

Expression and purification of E. coli LplA and E2lipD

E. coli LplA and E2lipD were expressed using TM202 and pET11c plasmids. Expression in BL21(DE3) cells was induced with 0.5 mM IPTG for 3 h at 37°C. Cells were sonicated and proteins were purified using HiTrap QFF with a 0–0.5 M NaCl gradient in 20 mM Tris/HCl (pH 7.5).

Crystallization of T. acidophilum LplA-N–LplA-C complex, data collection and structural analysis

T. acidophilum LplA-N–LplA-C was exchanged into 10 mM Tris/HCl (pH 7.5), concentrated to 20 mg/ml and centrifuged at 13000 g for 20 min at 4°C. Sitting-drop vapour-diffusion crystallization screens were set up at 18°C using Molecular Dimensions screens with a Phoenix robot (Art Robbins Instruments). Crystals in 40% (v/v) MPD (2-methyl-2,4-pentanediol), 0.1 M sodium acetate (pH 4.6) and 0.02 M CaCl2 were suitable for X-ray diffraction without further cryoprotectant. Diffraction data were collected at Diamond Light Source (Harwell, UK) on an ADSC Q315 CCD (charge-coupled device) detector on station IO2 (λ=0.9795 Å; 1 Å=0.1 nm). In total, 360 images were collected at an oscillation angle of 1°. Raw data images were processed using HKL2000 [19].

Model building

Molecular replacement using BALBES [20] was followed by model building with Coot [21] and rounds of refinement using Refmac5, part of CCP4 [22]. Other software included Molprobity [23] and Procheck [24].

Structural analysis

Hydrogen bonds and ionic interactions were evaluated with Contact CCP4 [22], ProtorP [25] and PISA [26]. Molecular graphics Figures were prepared in PyMOL (http://www.pymol.org).

NMR spectroscopy

15N-labelled LplA-C and 15N- and 15N13C-labelled E2lipD were produced by expression in M9 minimal medium supplemented with 1 g/l 15NH4Cl as the sole nitrogen source or 1 g/l 15NH4Cl and 2 g/l [13C]glucose respectively. His-tagged proteins were purified as described previously [18]. Most E2lipD and all LplA-N–LplA-C NMR data were acquired at 37 or 50°C on a 600 MHz Varian Unity Inova spectrometer with an ambient temperature probe, processed using NMRPipe/NMRDraw [27] and analysed using CCPN Analysis [28]. 15N-edited NOESY and 13C-edited NOESY spectra of E2lipD were acquired on an 800 MHz Varian Inova spectrometer at the MRC Biomedical NMR Centre (Mill Hill, London, U.K.). 1H, 15N and 13C chemical shifts were referenced to DSS [29]. Structures were calculated as described previously [30]. (1H-15N)-HSQC (heteronuclear single-quantum coherence) spectra of uniformly 15N-labelled LplA-C, both with and without unlabelled LplA-N, were recorded in 20 mM Tris/HCl (pH 7.5) and 150 mM NaCl. E2lipD spectra were recorded in 50 mM Hepes (pH 7.5) and 50 mM NaCl. All 1H-15N HSQC spectra in the present study were recorded with 128 increments in the nitrogen dimension, unless otherwise stated.

NMR titration of LplA-N–LplA-C with LA, ATP and Mg2+, then with E2lipD

LA (racemic mixture unless otherwise stated), ATP and Mg2+ were titrated in combination against an NMR sample containing LplA-N–LplA-C (unlabelled LplA-N and uniformly 15N-labelled LplA-C) in 20 mM Tris/HCl (pH 7.5) and 150 mM NaCl. The molar ratio of LplA-N–LplA-C to LA at each titration point was 1:0, 1:0.25, 1:0.50 and 1:1.25; a (1H-15N)-HSQC spectrum (32 scans, 128 min recording time) was recorded at each titration point. Unlabelled E2lipD was then added to the same NMR sample with ratios of LplA-N–LplA-C to E2lipD of 1:0.25, 1:0.50 and 1:1.25; a (1H-15N)-HSQC spectrum (32 scans, 128 min recording time) was recorded after each E2lipD addition.

NMR titration of LplA-N–LplA-C with E2lipD, and then with LA, ATP and Mg2+

Unlabelled E2lipD was added to an NMR sample containing LplA-N–LplA-C (unlabelled LplA-N and uniformly 15N-labelled LplA-C with no LA/ATP/Mg2+) in 20 mM Tris/HCl (pH 7.5) and 150 mM NaCl. The molar ratio of LplA-N–LplA-C to E2lipD at each titration point was 1:0, 1:0.25, 1:0.50, 1:0.75 and 1:1; a (1H-15N)-HSQC spectrum (24 scans, 160 increments, 120 min recording time) was recorded at each titration point. LA (2 mM), ATP (2.5 mM) and Mg2+ (1 mM) were then added together; a (1H-15N)-HSQC spectrum (24 scans, 160 increments) was recorded both directly after this addition and on the following day after overnight storage of the NMR sample at 4°C.

NMR titration of E2lipD with LplA-N–LplA-C, LA and ATP

Catalytic quantities of LplA-N–LplA-C were added to 15N-labelled 1.1 mM E2lipD in four steps (molar ratio of E2lipD to LplA-N–LplA-C of 1:0.0025, 1:0.005, 1:0.0075 and 1:0.01), followed by two additions of LA to a final concentration of 2.25 mM, and then by two additions of ATP to a final concentration of 2.25 mM (1.5 mM Mg2+ was present in the initial NMR sample). A (1H-15N)-HSQC spectrum was recorded at each titration point (eight scans, 34 min recording time). Chemical-shift perturbations were calculated as a weighted average of 1H and 15N chemical shift changes, Δδav {Δδav (p.p.m.)=[(Δδ2HN+Δδ2N/25)/2]1/2} [31].

Lipoylation/octanoylation activity assay

The electrophoretic mobility of E2lipD before and after lipoylation/octanoylation was analysed by non-denaturing PAGE as described previously [18]. In the lipoylation/octanoylation assays and T. acidophilum/E. coli enzyme cross-reactivity assays, the ratio of lipoylated to non-lipoylated E2lipD was quantified by MS.

Synthesis of octanoyl-AMP

Synthesis of ocatnoyl-AMP was carried out as described previously [32,33].

Model of LplA-N–LplA-C–E2lipD complex

In order to model a possible end-point of conformational change in LplA-N–LplA-C that permits lipoylation of E2lipD, the relative orientation of LplA-N and LplA-C was first changed to that observed between the N- and C-terminal domains of E. coli LplA in its complex with apo H protein and octanoyl-AMP (PDB entry 3A7A). E2lipD was then docked with the reoriented LplA-N–LplA-C complex using ClusPro 2.0 [34,35]; in the resulting models, E2lipD orientation and acceptor lysine (LysE2lipD42) position were compared with those in PDB entry 3A7A of apo H protein and LysApoH64 respectively. Models comparable with PDB entry 3A7A (i.e. with LysE2lipD42 in proximity to and oriented towards the LplA-N active site) were selected and their quality assessed using QMean [36]. The model from this subset with the highest QMean score was selected as a representative structure.

RESULTS

Structure of T. acidophilum LplA-C

Despite the fact that LplA-C is essential for lipoylation by archaeal LplA [17,18], the functional and structural relationship between LplA-N and LplA-C is poorly understood. We examined whether LplA-N and LplA-C exist independently or form a stable complex by first studying LplA-C structure without LplA-N. In crystallization screens, LplA-C showed a high propensity to precipitate. Poor chemical-shift dispersion, variable peak intensity and low peak count (approximately 50 peaks observed compared with 84 expected on the basis of the LplA-C amino acid sequence) in (1H-15N)-HSQC NMR spectra showed that LplA-C is disordered and heterogeneous over a range of pH values (pH 6–8) and NaCl concentrations (50–150 mM NaCl) (Figure 1A). Upon stepwise addition of unlabelled LplA-N to 15N-labelled LplA-C (final LplA-N/LplA-C molar ratio of 1:1), the observed dramatic increase in dispersion, homogeneity and number of LplA-C (1H-15N)-HSQC peaks indicated that LplA-C undergoes LplA-N binding-induced folding (Figure 1B). In total, 76 distinct backbone amide NH peaks were observed in the LplA-N-bound LplA-C (1H-15N)-HSQC spectrum; this close correspondence with the expected total of 84 peaks indicated that LplA-N-bound LplA-C adopts a single dominant conformation on average, and indirectly supports the presence of a single predominant LplA-N–LplA-C complex conformation in solution. The LplA-N-induced LplA-C fold, and by inference the LplA-N–LplA-C complex, is stable to at least 50°C (Figure 1B).

(1H-15N)-HSQC NMR spectra of T. acidophilum LplA-C in the absence and presence of LplA-N

Figure 1
(1H-15N)-HSQC NMR spectra of T. acidophilum LplA-C in the absence and presence of LplA-N

(A) (1H-15N)-HSQC spectrum at 37°C recorded on uniformly 15N-labelled LplA-C by itself and (B) in a 1:1 molar ratio with unlabelled LplA-N; an overlay of (1H-15N)-HSQC spectra recorded at 37°C (blue) and 50°C (red) is shown.

Figure 1
(1H-15N)-HSQC NMR spectra of T. acidophilum LplA-C in the absence and presence of LplA-N

(A) (1H-15N)-HSQC spectrum at 37°C recorded on uniformly 15N-labelled LplA-C by itself and (B) in a 1:1 molar ratio with unlabelled LplA-N; an overlay of (1H-15N)-HSQC spectra recorded at 37°C (blue) and 50°C (red) is shown.

T. acidophilum LplA-N–LplA-C X-ray crystal structure: comparison with other LplAs

Subsequent to the LplA-C NMR studies described above, crystallization screens of co-expressed LplA-N and LplA-C produced LplA-N–LplA-C complex crystals in 40% (v/v) MPD, 0.1 M sodium acetate (pH 4.6) and 0.02 M CaCl2. The LplA-N–LplA-C structure (Figure 2A) was determined to 2.7 Å resolution by molecular replacement with LplA-N (PDB entry 2ARS) (Table 1). The overall fold of LplA-N [3,16] is maintained in the presence of LplA-C, as confirmed using Dali [37] (Table 2). The β-strands in LplA-N are β1 (residues 1–7), β2 (residues 35–39), β3 (residues 44–47), β4 (residues 68–71), β5 (residues 79–81), β6 (residues 85–93), β7 (residues 121–123), β8 (residues 128–131), β9 (residues 144–154) and β10 (residues 157–165); the α-helices are α1 (residues 13–27), α2 (residues 59–65), α3 (residues 98–117), α4 (residues 181–183), α5 (residues 184–194), α6 (residues 206–223), α7 (residues 232–246) and α8 (residues 249–254); and the 310- or η-helices are η1 (residues 52–56), η2 (residues 171–176) and η3 (residues 198–202). The β-strands in LplA-C are β11 (residues 2–10), β12 (residues 15–23) and β13 (residues 26–35), and the α-helices are α9 (residues 42–52), α10 (residues 58–68) and α11 (residues 79–86), where the LplA-C strand and helix numbering continues from the LplA-N numbering, but the residue numbers start anew at the LplA-C N-terminus.

Structure of the T. acidophilum LplA-N–LplA-C complex

Figure 2
Structure of the T. acidophilum LplA-N–LplA-C complex

(A) LplA-N (green) and LplA-C (blue). Secondary structure elements, and the LplA-N N-terminus and LplA-C C-terminus, are indicated. (B) Some of the ionic interactions and hydrogen bond network interactions (indicated by broken lines) between LplA-N (green) and LplA-C (blue) with side chains in pale yellow (different orientation of the complex to that in A). (C) Overlays of unliganded/apo T. acidophilum LplA-N–LplA-C (PDB entry 3R07), E. coli LplA (PDB entry 1X2G) and S. pneumoniae LplA (PDB entry 1VQZ) (left-hand panel), and of lipoyl-AMP bound E. coli LplA (PDB entry 3A7R) and bovine LPT (PDB entry 3A7U) (right-hand panel). The N-terminal domain orientation is the same in both overlays.

Figure 2
Structure of the T. acidophilum LplA-N–LplA-C complex

(A) LplA-N (green) and LplA-C (blue). Secondary structure elements, and the LplA-N N-terminus and LplA-C C-terminus, are indicated. (B) Some of the ionic interactions and hydrogen bond network interactions (indicated by broken lines) between LplA-N (green) and LplA-C (blue) with side chains in pale yellow (different orientation of the complex to that in A). (C) Overlays of unliganded/apo T. acidophilum LplA-N–LplA-C (PDB entry 3R07), E. coli LplA (PDB entry 1X2G) and S. pneumoniae LplA (PDB entry 1VQZ) (left-hand panel), and of lipoyl-AMP bound E. coli LplA (PDB entry 3A7R) and bovine LPT (PDB entry 3A7U) (right-hand panel). The N-terminal domain orientation is the same in both overlays.

Table 1
Data collection and structural refinement statistics for the crystal structure of the LplA-N–LplA-C complex (PDB entry 3R07)

Numbers in parentheses represent statistics at the highest resolution (2.7–2.8 Å).

Measurement  
Crystallographic data  
 Resolution (Å) 2.7 
 Space group P3121 
 Unit cell parameters  
  a=b (Å) 118.57 
  c (Å) 72.92 
  α=β (°) 90 
  γ (°) 120 
Merging statistics  
 Number of reflections 15063 
 Average redundancy 7.1 (7.1) 
II 37.1 (6.1) 
 Completeness (%) 96.0 (98.5) 
R-merge (%) 7.7 (43.2) 
Refinement statistics  
 RMSD bond length (Å) 0.015 
 RMSD bond angle (°) 1.547 
 Residues in disallowed regions (%) 
 Mean B value (Å251.561 
R-factor (%) 19.8 
R-free (%) 25.4 
Measurement  
Crystallographic data  
 Resolution (Å) 2.7 
 Space group P3121 
 Unit cell parameters  
  a=b (Å) 118.57 
  c (Å) 72.92 
  α=β (°) 90 
  γ (°) 120 
Merging statistics  
 Number of reflections 15063 
 Average redundancy 7.1 (7.1) 
II 37.1 (6.1) 
 Completeness (%) 96.0 (98.5) 
R-merge (%) 7.7 (43.2) 
Refinement statistics  
 RMSD bond length (Å) 0.015 
 RMSD bond angle (°) 1.547 
 Residues in disallowed regions (%) 
 Mean B value (Å251.561 
R-factor (%) 19.8 
R-free (%) 25.4 
Table 2
Summary of Dali similarity searches

The values as determined for T. acidophilum LplA-N and LplA-C are listed individually and have been obtained from pairwise Dali comparisons. PDB entry 2ARS was compared only with the most similar non-T. acidophilum LplA, which is S. pneumoniae LplA (PDB entry 1VQZ).

Organism Ligands PDB entry Dali search with structure 2ARS; Z-score (RMSD, Å) Dali search with T.acidophilum LplA-N–LplA-C; Z-score (RMSD, Å) 
T. acidophilum None 2ARS 45.3 39.1 (0.6) 
 None 2C7I 43.2 (0.7) 38.2 (0.7) 
 LA 2C8M 43.2 (0.7)  
 ATP 2ARU 44.12 (0.3) 38.9 (1.0) 
 Lipoyl-AMP 2ART 44.12 (0.5) 38.9 (1.0) 
S. pneumoniae None 1VQZ 26.7 (2.5) LplA-N: 26.9 (2.8) 
    LplA-C: 11.4 (1.5) 
E. coli None 1X2G  LplA-N: 25.9 (2.8) 
    LplA-C: 8.8 (2.3) 
 LA 1X2H  LplA-N: 25.4 (2.9) 
    LplA-C: 8.7 (2.3) 
 Lipoyl-AMP 3A7R  LplA-N: 28.5 (2.2) 
    LplA-C: 9.7 (2.1) 
 Octanoyl-AMP, ApoH 3A7A  LplA-N: 28.5 (2.3) 
    LplA-C: 9.7 (2.1) 
B. taurus  2E5A  LplA-N: 27 (2.1) 
    LplA-C: 7.7 (2.5) 
Organism Ligands PDB entry Dali search with structure 2ARS; Z-score (RMSD, Å) Dali search with T.acidophilum LplA-N–LplA-C; Z-score (RMSD, Å) 
T. acidophilum None 2ARS 45.3 39.1 (0.6) 
 None 2C7I 43.2 (0.7) 38.2 (0.7) 
 LA 2C8M 43.2 (0.7)  
 ATP 2ARU 44.12 (0.3) 38.9 (1.0) 
 Lipoyl-AMP 2ART 44.12 (0.5) 38.9 (1.0) 
S. pneumoniae None 1VQZ 26.7 (2.5) LplA-N: 26.9 (2.8) 
    LplA-C: 11.4 (1.5) 
E. coli None 1X2G  LplA-N: 25.9 (2.8) 
    LplA-C: 8.8 (2.3) 
 LA 1X2H  LplA-N: 25.4 (2.9) 
    LplA-C: 8.7 (2.3) 
 Lipoyl-AMP 3A7R  LplA-N: 28.5 (2.2) 
    LplA-C: 9.7 (2.1) 
 Octanoyl-AMP, ApoH 3A7A  LplA-N: 28.5 (2.3) 
    LplA-C: 9.7 (2.1) 
B. taurus  2E5A  LplA-N: 27 (2.1) 
    LplA-C: 7.7 (2.5) 

LplA-N–LplA-C is structurally similar to single polypeptide apo-LplAs from Streptococcus pneumoniae and E. coli, including similar domain orientations (Figure 2C and Table 2). Bovine (Bos taurus) LPT, however, has a different arrangement of domains in both apo and lipoyl-AMP-bound forms, as does lipoyl-AMP-bound E. coli LplA; in these cases, the C-terminal domain has undergone a 180° rotation (Figure 2C) [13]. The structures of E. coli LplA C-terminal domain and T. acidophilum LplA-C agree well, with both forming a canopy above the tunnel-like entry to the active site. With respect to the inverted gene orientation in T. acidophilum (lpla-c is upstream of lpla-n with a TATA box upstream of lpla-c but no cis-regulatory sequence in the proximity of lpla-n) [18], it is important to note that the LplA-C C-terminus and LplA-N N-terminus are located at opposite ends of the LplA-N–LplA-C complex, approximately 56 Å apart (Figure 2A), confirming that LplA-N and LplA-C are made as separate polypeptides.

The LplA-N–LplA-C interface has a buried surface area of 805 Å2 compared with 993 Å2 between the N- and C-terminal domains of the closest single protein homologue, S. pneumoniae LplA [3,18]. The LplA-N–LplA-C interface involves 50 residues and includes 12 hydrogen bonds plus five salt bridges involving three pairs of residues [26]. These include a five-residue network that forms salt bridges (GluLplA-N56–ArgLplA-C17, GluLplA-N55–HisLplA-C31) and three hydrogen bonds (GluLplA-N55–HisLplA-C31, GluLplA-N55–SerLplA-C33, GluLplA-N56–ArgLplA-C17) (Figure 2B). In the corresponding location, S. pneumoniae LplA has an interdomain three-residue (Arg45–Asp284–His46) network with two interdomain ionic interactions involving Asp284, and E. coli LplA has no obvious ionic interaction. In addition, the LplA-N–LplA-C interface has a substantial hydrophobic component with approximately 25 hydrophobic residues contributing to the interface.

LplA-C-induced conformational change of LplA-N: ‘capping’ loop and adenylate-binding loop

LplA-N undergoes a substantial local structural rearrangement upon binding LplA-C. In isolated LplA-N (i.e. without LplA-C), β8 consists of residues 138–141 and is connected to β9 (residues 144–154) by a short β-turn, whereas strands β7 and β8 are connected by a long loop consisting of residues 124–137 (orange and labelled as the ‘capping loop’ in Figure 3). Interestingly, this loop makes several contacts with lipoyl-AMP in PDB entry 2ART, and may play a role in ensuring that isolated LplA-N is catalytically inert. This region is reorganized in the LplA-N–LplA-C structure such that residues 128–131 form β8, and a short turn comprising residues 125–127 connects β7 to β8, whereas β8 is connected to β9 by a disordered loop comprising residues 132–142, for most of which electron density is not observed (Figure 3). This structural shift seems to be facilitated by the similarity of the two motifs that alternate as β8: residues 128–131 are Asp-Val-Ser-Ile, whereas residues 138–141 are Asp-Ile-Met-Ala. It is of note that in E. coli LplA, β8 is a fixed motif, connected to β7 and β9 by short loops on either side.

LplA-C-induced conformational change of LplA-N

Figure 3
LplA-C-induced conformational change of LplA-N

The structure of the LplA-N–LplA-C complex (PDB entry 3R07) is shown with LplA-N predominantly in grey and LplA-C in blue. The structure of isolated (i.e. without LplA-C) LplA-N (PDB entry 2ART) is superimposed on the LplA-N–LplA-C complex and is also shown predominantly in grey. Structural features of structure 3R07 are highlighted in green and structural features of structure 2ART are highlighted in yellow, orange and purple. Lipoyl-AMP from structure 2ART is shown in red. Secondary structure elements of structure 3R07 are labelled. The lipoate-binding loop conformations are almost identical in structures 3R07 (green) and 2ART (yellow). In the absence of LplA-C, residues 124–137 form a long loop (orange, labelled as ‘capping loop’); the corresponding loop (green) is much shorter in the LplA-N–LplA-C complex. Instead, the subsequent loop is much longer in the LplA-N–LplA-C complex than in isolated LplA-N (electron density is lacking between Val133 and Gly142 in the LplA-N–LplA-C complex). Residues identified by structure-based alignment to form the adenylate binding region are shown in green (PDB entry 3R07) and purple (PDB entry 2ART), although in 2ART much of the adenylate-binding region lacks defined electron density.

Figure 3
LplA-C-induced conformational change of LplA-N

The structure of the LplA-N–LplA-C complex (PDB entry 3R07) is shown with LplA-N predominantly in grey and LplA-C in blue. The structure of isolated (i.e. without LplA-C) LplA-N (PDB entry 2ART) is superimposed on the LplA-N–LplA-C complex and is also shown predominantly in grey. Structural features of structure 3R07 are highlighted in green and structural features of structure 2ART are highlighted in yellow, orange and purple. Lipoyl-AMP from structure 2ART is shown in red. Secondary structure elements of structure 3R07 are labelled. The lipoate-binding loop conformations are almost identical in structures 3R07 (green) and 2ART (yellow). In the absence of LplA-C, residues 124–137 form a long loop (orange, labelled as ‘capping loop’); the corresponding loop (green) is much shorter in the LplA-N–LplA-C complex. Instead, the subsequent loop is much longer in the LplA-N–LplA-C complex than in isolated LplA-N (electron density is lacking between Val133 and Gly142 in the LplA-N–LplA-C complex). Residues identified by structure-based alignment to form the adenylate binding region are shown in green (PDB entry 3R07) and purple (PDB entry 2ART), although in 2ART much of the adenylate-binding region lacks defined electron density.

The lipoate-binding loop adopts the same conformation with and without LplA-C (Figure 3), whereas in LplA-N–LplA-C a substantial portion of the region corresponding to the adenylate-binding loop strikingly forms contiguous α-helices (α4 and α5; residues 181–183 and 184–194). In other LplA structures, this region is either a loop or is largely disordered such that electron density is absent. In several structures of isolated LplA-N, for example, much of the adenylate-binding loop region is disordered (e.g. PDB entries 2ARS and 2C7I, both unliganded; PDB entry 2C8M, LA-bound; and PDB entry 2ART, lipoyl-AMP bound), although it is of note that following an electron density gap in the structures 2ARS, 2ART and 2C8M, there is a nascent α-helix that overlaps with part of LplA-N–LplA-C α5 (e.g. the region shown in purple in Figure 3). In the structure of unliganded E. coli LplA (PDB entry 1X2G), the adenylate-binding loop occupies a similar position to LplA-N–LplA-C helices α4 and α5, whereas there is again missing electron density in unliganded bovine LPT. The adenylate-binding loops of lipoyl-AMP-bound E. coli LplA and bovine LPT overlap closely with each other and are shifted towards the active site relative to the adenylate-binding loops of the respective unliganded enzymes.

LplA-N–LplA-C interaction with lipoyl-AMP and E2lipD

We used NMR titrations to investigate LplA-N–LplA-C interactions, monitoring the (1H-15N)-HSQC spectrum of 15N-labelled LplA-C in complex with unlabelled LplA-N. In one titration (Titration 1), LA, ATP and Mg2+ were added together to LplA-N–15N-LplA-C, then E2lipD was added (Figures 4A and 4B). Upon addition of LA, ATP and Mg2+ (and presumably therefore upon formation of the lipoyl-AMP-bound form of LplA-N–LplA-C), 15 out of the 76 distinct backbone amide NH peaks in the (1H-15N)-HSQC spectrum of LplA-N–15N-LplA-C were significantly broadened (intermediate timescale exchange), five exhibited slow exchange (two peaks observed per backbone amide NH), four underwent a chemical-shift change, four exhibited both chemical-shift change and broadening, and at least one peak increased significantly in intensity (Figure 4A). Upon subsequent titration of E2lipD into the LplA-N–15N-LplA-C NMR sample, most of the peaks that had been perturbed (Figure 4A) reverted to a state the same as or close to that observed before the addition of LA, ATP and Mg2+ (Figure 4B), consistent with lipoyl transfer to E2lipD and hence consumption of substrate.

(1H-15N)-HSQC spectra of LplA-N–15N-LplA-C: titration with LA, ATP and Mg2+, and with E2lipD

Figure 4
(1H-15N)-HSQC spectra of LplA-N–15N-LplA-C: titration with LA, ATP and Mg2+, and with E2lipD

The results of two titrations are shown. In (A) and (B), LA, ATP and Mg2+ were added before E2lipD. In (C) and (D), E2lipD was added before LA, ATP and Mg2+. (A) LA, ATP and Mg2+ were added together in a step-wise fashion to an NMR sample containing LplA-N–LplA-C (unlabelled LplA-N and uniformly 15N-labelled LplA-C) in 20 mM Tris/HCl (pH 7.5) and 150 mM NaCl. A (1H-15N)-HSQC spectrum was recorded at each titration point. The molar ratio of LplA-N–LplA-C complex to LA at each titration point was 1:0, 1:0.25, 1:0.50 and 1:1.25. The initial spectrum is shown in blue, the final spectrum in yellow. LplA-C peaks perturbed upon addition of LA, ATP and Mg2+ are highlighted as follows: intermediate exchange (broadening), solid rectangle; slow exchange (two peaks), broken line rectangle; chemical-shift change, solid ellipse; chemical-shift change and broadening, broken line ellipse; and increase in intensity, blue broken line rectangle. (B) E2lipD was then added to the same NMR sample as in (A) with molar ratios of LplA-N–LplA-C to E2lipD of 1:0.25, 1:0.50 and 1:1.25. The final spectrum is shown in red with the same peaks highlighted as in (A). Most of the peaks perturbed in (A) reverted to the pre-LA/ATP/Mg2+ state, which is again represented by blue peaks. (C) In the second titration, E2lipD was added in four steps (molar ratio of LplA-N–LplA-C to E2lipD at each titration point was 1:0, 1:0.25, 1:0.50, 1:0.75 and 1:1) with essentially no change in the LplA-C (1H-15N)-HSQC spectrum (not shown). LA (2 mM), ATP (2.5 mM) and Mg2+ (1 mM) were then added together. The spectrum after E2lipD addition is shown in blue, and the spectrum after LA, ATP and Mg2+ addition is shown in yellow. Perturbed LplA-C peaks are highlighted using the same scheme described for (A), except that some of the larger chemical-shift changes are indicated by an arrow (broken arrow where the connection between the shifted peak and the original peak is tentative), newly appearing peaks are indicated by a blue broken rectangle, and chemical-shift change plus increased intensity is indicated by a blue broken ellipse. The peak at around 6.4 p.p.m., 113 p.p.m. labelled with an asterisk moved to 6.1 p.p.m., 113 p.p.m. Peaks subject to smaller chemical-shift changes have not been highlighted. (D) (1H-15N)-HSQC spectrum of the same sample as in (C), recorded after leaving the sample overnight at 4°C.

Figure 4
(1H-15N)-HSQC spectra of LplA-N–15N-LplA-C: titration with LA, ATP and Mg2+, and with E2lipD

The results of two titrations are shown. In (A) and (B), LA, ATP and Mg2+ were added before E2lipD. In (C) and (D), E2lipD was added before LA, ATP and Mg2+. (A) LA, ATP and Mg2+ were added together in a step-wise fashion to an NMR sample containing LplA-N–LplA-C (unlabelled LplA-N and uniformly 15N-labelled LplA-C) in 20 mM Tris/HCl (pH 7.5) and 150 mM NaCl. A (1H-15N)-HSQC spectrum was recorded at each titration point. The molar ratio of LplA-N–LplA-C complex to LA at each titration point was 1:0, 1:0.25, 1:0.50 and 1:1.25. The initial spectrum is shown in blue, the final spectrum in yellow. LplA-C peaks perturbed upon addition of LA, ATP and Mg2+ are highlighted as follows: intermediate exchange (broadening), solid rectangle; slow exchange (two peaks), broken line rectangle; chemical-shift change, solid ellipse; chemical-shift change and broadening, broken line ellipse; and increase in intensity, blue broken line rectangle. (B) E2lipD was then added to the same NMR sample as in (A) with molar ratios of LplA-N–LplA-C to E2lipD of 1:0.25, 1:0.50 and 1:1.25. The final spectrum is shown in red with the same peaks highlighted as in (A). Most of the peaks perturbed in (A) reverted to the pre-LA/ATP/Mg2+ state, which is again represented by blue peaks. (C) In the second titration, E2lipD was added in four steps (molar ratio of LplA-N–LplA-C to E2lipD at each titration point was 1:0, 1:0.25, 1:0.50, 1:0.75 and 1:1) with essentially no change in the LplA-C (1H-15N)-HSQC spectrum (not shown). LA (2 mM), ATP (2.5 mM) and Mg2+ (1 mM) were then added together. The spectrum after E2lipD addition is shown in blue, and the spectrum after LA, ATP and Mg2+ addition is shown in yellow. Perturbed LplA-C peaks are highlighted using the same scheme described for (A), except that some of the larger chemical-shift changes are indicated by an arrow (broken arrow where the connection between the shifted peak and the original peak is tentative), newly appearing peaks are indicated by a blue broken rectangle, and chemical-shift change plus increased intensity is indicated by a blue broken ellipse. The peak at around 6.4 p.p.m., 113 p.p.m. labelled with an asterisk moved to 6.1 p.p.m., 113 p.p.m. Peaks subject to smaller chemical-shift changes have not been highlighted. (D) (1H-15N)-HSQC spectrum of the same sample as in (C), recorded after leaving the sample overnight at 4°C.

In a separate titration (Titration 2), unlabelled E2lipD was first titrated into a LplA-N–15N-LplA-C NMR sample to a final molar ratio of 1:1 LplA-N–LplA-C to E2lipD, producing essentially no change in the LplA-N–15N-LplA-C (1H-15N)-HSQC spectrum. Significant perturbation of the (1H-15N)-HSQC was observed, however, when LA, ATP and Mg2+ were then added (Figure 4C), at least 30 peaks underwent chemical-shift change, five peaks were broadened, three showed both broadening and chemical-shift change, one showed chemical-shift change plus increased intensity, and two new peaks appeared in the vicinity of an initially absent peak at around 8.8 p.p.m., 116.3 p.p.m. which is generally weak/absent in LplA-N–15N-LplA-C (1H-15N)-HSQC spectra. Notably, at least 75% of the peaks perturbed in Titration 1 were also perturbed in Titration 2. When a (1H-15N)-HSQC spectrum was recorded the next day on the same sample, some peaks had reverted towards a pre-LA /ATP/Mg2+ position/intensity, but some had not (Figure 4D).

Substrate promiscuity and recognition of E2 lipoyl domains by LplA-N–LplA-C

As its bipartite nature may affect substrate recognition and specificity, we tested T. acidophilum LplA-N–LplA-C activity with different acceptor domains and substrates, including LA, OA and octanoyl-AMP, with E. coli LplA serving as a positive control. In gel-shift lipoylation assays [18] with LA, co-expressed LplA-N–LplA-C and a 1:1 mixture of individually expressed and purified LplA-N and LplA-C showed equal activity. LplA-N–LplA-C showed activity with both OA and octanoyl-AMP, in agreement with previous findings that LplAs can catalyse the formation and transfer of octanoyl-AMP [38]. As judged by MS, E2lipD modification efficiency by LplA-N–LplA-C with OA as the substrate was 20–30% of that with LA as the substrate.

E2lipD cross-reactivity was analysed next. E2lipD residues both N-terminal (−) and C-terminal (+) of the target lysine residue are important for efficient lipoylation [39]. Previous large-scale sequence alignments identified a highly conserved aspartate residue at −1, hydrophobic residues at +1, +5 and −4, glutamate/aspartate enrichment at −3 and +4, and serine/alanine at +7 [40]. A glutamate residue at −3 and glycine residue at −16 are involved in LplA recognition. Sequence alignment with E. coli E2lipDs and E. coli H protein showed that a glycine residue at −16 is conserved in T. acidophilum E2lipD. A glutamate residue at −3 is conserved in E. coli E2lipDs and H protein, whereas T. acidophilum E2lipD has a methionine residue at −3; residues other than glutamate at −3 reduce lipoylation efficiency in E. coli [41]. Correspondingly, E. coli LplA lipoylated T. acidophilum E2lipD with approximately 50% efficiency relative to E. coli E2lipD, and T. acidophilum LplA-N–LplA-C lipoylated E. coli E2lipD with approximately 15–20% efficiency relative to T. acidophilum E2lipD.

NMR analysis of T. acidophilum E2lipD structure and lipoylation

In order to facilitate modelling studies of the complete T. acidophilum lipoylation system (see below), the T. acidophilum E2lipD structure was determined by NMR (Table 3). T. acidophilum E2lipD is similar overall to other lipoyl domains [DaliLite Z-score of 6.8 and RMSD (root mean square deviation) over all Cα atoms of 2.7 Å compared with E. coli E2lipD (PDB entry 1QJO; 27% sequence identity)]. In an NMR titration to monitor E2lipD lipoylation, T. acidophilum E2lipD (1H-15N)-HSQC did not change upon step-wise addition of catalytic quantities of LplA-N–LplA-C (final molar ratio of 100:1, E2lipD/LplA-N–LplA-C), or upon addition of LA (final molar ratio approximately 1:2, E2lipD/LA) (Figure 5). Upon subsequent addition of ATP (final molar ratio approximately 1:2, E2lipD/ATP), however, several E2lipD backbone amide peaks underwent chemical-shift perturbation (Figures 5A and 5B); the largest chemical-shift perturbations were observed for E2lipD residues 42–44 (LysE2lipD42 is the lipoylation target residue), followed by residues 9–10; the ThrE2lipD40 peak was broadened. When mapped on to the E2lipD structure, the pattern of largest chemical-shift perturbations (plus broadening for residue 40) indicates that lipoylation induces a localized conformational change in E2lipD (Figure 5C).

Table 3
Structural statistics for the ensemble of NMR-derived structures of E2lipD

The RMSD from the mean structure was calculated over residues 2–5, 16–20, 27–29, 37–39, 44–46, 53–58, 64–66 and 73–76. Ramachandran plot statistics were calculated with PROCHECK-NMR. NOE, nuclear Overhauser effect.

Measurement  
Total number of NOE restraints 648 
 Intraresidue 146 
 Sequential/medium range (i to i+1−4) 270 
 Long range 232 
Number of dihedral angle restraints 76 
Number of hydrogen bond restraints 12 
RMSD for backbone atoms 0.48Å 
RMSD for non-hydrogen atoms 1.11 Å 
Average number of NOE violations >0.5Å (per structure) 
Average number of dihedral angle violations >1o (per structure) 
Ramachandran plot statistics  
 Most favoured (%) 81.4 
 Additional allowed (%) 15.2 
 Generously allowed (%) 3.2 
 Disallowed (%) 0.1 
Measurement  
Total number of NOE restraints 648 
 Intraresidue 146 
 Sequential/medium range (i to i+1−4) 270 
 Long range 232 
Number of dihedral angle restraints 76 
Number of hydrogen bond restraints 12 
RMSD for backbone atoms 0.48Å 
RMSD for non-hydrogen atoms 1.11 Å 
Average number of NOE violations >0.5Å (per structure) 
Average number of dihedral angle violations >1o (per structure) 
Ramachandran plot statistics  
 Most favoured (%) 81.4 
 Additional allowed (%) 15.2 
 Generously allowed (%) 3.2 
 Disallowed (%) 0.1 

Chemical-shift perturbation upon lipoylation of E2lipD

Figure 5
Chemical-shift perturbation upon lipoylation of E2lipD

(A) Overlaid (1H-15N)-HSQC spectra of uniformly 15N-labelled E2lipD from titration of 15N-labelled 1.1 mM E2lipD with catalytic quantities of LplA-N–LplA-C in four steps (molar ratio of E2lipD to LplA-N–LplA-C of 1:0.0025, 1:0.005, 1:0.0075 and 1:0.01), followed by two additions of LA to a final concentration of 2.25 mM; the final spectrum from the titration to this point is shown in green (no change from original spectrum). ATP was then added to a concentration of 2.25 mM; the subsequent (1H-15N)-HSQC spectrum is shown in red. Peaks are labelled with amino acid assignments [assignments could not be made for Glu11, Gly12, Thr14, Glu30, Tyr60 and Thr71 due to lack of peaks in three-dimensional spectra; unlabelled backbone NH peaks are assumed to arise from these residues or from the N-terminal His6-tag residues; the Gly16 peak (1H and 15N chemical shifts of 8.22 p.p.m. and 104.75 p.p.m. respectively) is omitted to allow greater overall clarity]. Peaks showing the largest chemical-shift changes upon addition of ATP are highlighted with an ellipse drawn around the pre-ATP (green) and post-ATP (red) peak positions (the peak due to Thr40 was broadened out of the spectrum upon E2lipD lipoylation). (B) Lipoylation-induced chemical-shift changes {calculated using Δδav (p.p.m.)=[(Δδ2HN+Δδ2N/25)/2]1/2} plotted as a function of E2lipD residue number. * indicates an unassigned residue, P indicates a proline residue [proline does not produce a signal in (1H-15N)-HSQC spectra], and B indicates a peak that was broadened out of the spectrum upon E2lipD lipoylation. Red, Δδav (p.p.m.)≥0.1 p.p.m.; orange, Δδav (p.p.m.)≥0.04 p.p.m.; and yellow, Δδav (p.p.m.)≥0.02 p.p.m. E2lipD secondary structure elements (β-strands) are indicated by arrows above the histogram. (C) E2lipD surface with most of the residues perturbed upon lipoylation highlighted using the same colour scheme as in (B).

Figure 5
Chemical-shift perturbation upon lipoylation of E2lipD

(A) Overlaid (1H-15N)-HSQC spectra of uniformly 15N-labelled E2lipD from titration of 15N-labelled 1.1 mM E2lipD with catalytic quantities of LplA-N–LplA-C in four steps (molar ratio of E2lipD to LplA-N–LplA-C of 1:0.0025, 1:0.005, 1:0.0075 and 1:0.01), followed by two additions of LA to a final concentration of 2.25 mM; the final spectrum from the titration to this point is shown in green (no change from original spectrum). ATP was then added to a concentration of 2.25 mM; the subsequent (1H-15N)-HSQC spectrum is shown in red. Peaks are labelled with amino acid assignments [assignments could not be made for Glu11, Gly12, Thr14, Glu30, Tyr60 and Thr71 due to lack of peaks in three-dimensional spectra; unlabelled backbone NH peaks are assumed to arise from these residues or from the N-terminal His6-tag residues; the Gly16 peak (1H and 15N chemical shifts of 8.22 p.p.m. and 104.75 p.p.m. respectively) is omitted to allow greater overall clarity]. Peaks showing the largest chemical-shift changes upon addition of ATP are highlighted with an ellipse drawn around the pre-ATP (green) and post-ATP (red) peak positions (the peak due to Thr40 was broadened out of the spectrum upon E2lipD lipoylation). (B) Lipoylation-induced chemical-shift changes {calculated using Δδav (p.p.m.)=[(Δδ2HN+Δδ2N/25)/2]1/2} plotted as a function of E2lipD residue number. * indicates an unassigned residue, P indicates a proline residue [proline does not produce a signal in (1H-15N)-HSQC spectra], and B indicates a peak that was broadened out of the spectrum upon E2lipD lipoylation. Red, Δδav (p.p.m.)≥0.1 p.p.m.; orange, Δδav (p.p.m.)≥0.04 p.p.m.; and yellow, Δδav (p.p.m.)≥0.02 p.p.m. E2lipD secondary structure elements (β-strands) are indicated by arrows above the histogram. (C) E2lipD surface with most of the residues perturbed upon lipoylation highlighted using the same colour scheme as in (B).

Model of the T. acidophilum LplA-N–LplA-C–E2lipD complex

In E. coli LplA, lipoate adenylation causes conformational changes, including an 180° rotation of the C-terminal domain (Figures 2C and 6A), that prime the system for lipoyl transfer [13]. In T. acidophilum apo-LplA-N–LplA-C, LplA-C forms a canopy over the active site in a similar manner to the C-terminal domain of other LplAs, obstructing access to lipoate of LysE2lipD42, the lipoate acceptor residue (Figure 6A). Hence, a substantial change to the apo-LplA-N–LplA-C structure is required for lipoylation to occur. Using the E. coli LplA–octanoyl-AMP–ApoH protein complex crystal structure as a template [13] (Figure 6A), we have modelled a possible end-point of such a change with E2lipD incorporated (Figure 6B). The respective lipoate acceptor residues (T. acidophilum LysE2lipD42 and E. coli LysApoH64) are in similar positions (Figure 4B). In our model, E2lipD bridges LplA-N and LplA-C, and LplA-C has undergone a rotation of approximately 120°. In total, 35 LplA-N residues, eight LplA-C residues and 34 E2lipD residues are involved in the interface [24].

Lipoate protein ligase complexes with E2lipD or ApoH protein

Figure 6
Lipoate protein ligase complexes with E2lipD or ApoH protein

(A) Superimposition of E. coli octanoyl-5′-AMP-bound LplA (yellow) in complex with E. coli ApoH (grey) (PDB entry 3A7A) and T. acidophilum LplA-N (green)–LplA-C (blue) (PDBentry 3R07). (B) Comparison of the acceptor lysine residue positions in the E. coli LplA–ApoH complex and in the T. acidophilum LplA-N–LplA-C–E2lipD complex in which LplA-C has undergone a change in position and orientation relative to LplA-N–LplA-C. E. coli ApoH is positioned as in the LplA–ApoH complex, but E. coli LplA has been omitted for clarity. The lipoyl acceptor residue of E. coli ApoH, LysApoH64, is shown in black. T. acidophilum E2lipD is in magenta with its lipoyl acceptor residue, LysE2lipD42, in cyan. Octanoyl-5′-AMP (red) is positioned as in the E. coli LplA–ApoH complex.

Figure 6
Lipoate protein ligase complexes with E2lipD or ApoH protein

(A) Superimposition of E. coli octanoyl-5′-AMP-bound LplA (yellow) in complex with E. coli ApoH (grey) (PDB entry 3A7A) and T. acidophilum LplA-N (green)–LplA-C (blue) (PDBentry 3R07). (B) Comparison of the acceptor lysine residue positions in the E. coli LplA–ApoH complex and in the T. acidophilum LplA-N–LplA-C–E2lipD complex in which LplA-C has undergone a change in position and orientation relative to LplA-N–LplA-C. E. coli ApoH is positioned as in the LplA–ApoH complex, but E. coli LplA has been omitted for clarity. The lipoyl acceptor residue of E. coli ApoH, LysApoH64, is shown in black. T. acidophilum E2lipD is in magenta with its lipoyl acceptor residue, LysE2lipD42, in cyan. Octanoyl-5′-AMP (red) is positioned as in the E. coli LplA–ApoH complex.

DISCUSSION

Biochemical data from our laboratory and elsewhere indicate that LplA-C is essential for lipoylation of E2 in T. acidophilum [17,18]. Our genomic profiling, moreover, indicates that bipartite LplA-N–LplA-C is the evolutionarily predominant lipoylation system in the archaea (M.G. Posner, A. Upadhyay, M.J. Danson, S. Dorus and S. Bagby, unpublished work). There is, however, conflicting evidence as to the exact role of LplA-C: although LplA-N by itself was reported to catalyse lipoate adenylation to form lipoyl-AMP [16], it was more recently reported that LplA-C is required for lipoate adenylation [17]. Further structural and mechanistic analysis of LplA-C function is therefore warranted. The structures of T. acidophilum LplA-N–LplA-C and E2lipD described in the present study represent the first structural analysis of a complete archaeal lipoylation system and the first of a bipartite lipoate protein ligase. The structure of LplA-N was already known [3,16], but the nature of LplA-N association with the functionally essential LplA-C was unknown [17,18]. It was not clear, for example, whether LplA-C is always associated with LplA-N or only during lipoylation. We now know that LplA-C is probably not functional by itself as it is disordered and undergoes LplA-N-induced folding. The C-terminal domain of E. coli LplA, on the other hand, was found by limited proteolysis to be structurally stable [3]. We do not know, however, whether LplA-C retains its fold once it is released from LplA-N as we have been unable to establish a non-denaturing procedure to dissociate the LplA-N–LplA-C complex.

The current evidence indicates that the observed interface between LplA-N and LplA-C is a biological rather than crystal packing interface, and that LplA-N and LplA-C exist permanently as a complex. This evidence includes the observations that isolated LplA-C is disordered, the LplA-N–LplA-C complex is stable to at least 50°C (Figure 1), LplA-N and LplA-C associate strongly upon co-expression, there is an extensive buried hydrophobic surface between LplA-N and LplA-C, similar interfaces are observed in other LplA structures from several organisms, and LplA-N and LplA-C are functionally interdependent. There are thus mechanistic differences from the LipA–LipB system in which LipA and LipB operate sequentially, although we note that E. coli LipA and LipB have been found to form a tight non-covalent association with the E2 components of PDHC and OGDHC [4], presumably with resulting potential for greater processivity and for interaction between LipA and LipB themselves. The inherent robustness of T. acidophilum LplA, and presumably LplA from other thermophiles, makes these ligases attractive starting points for biotechnological and chemical biology applications, such as have been demonstrated for E. coli LplA [42,43].

The T. acidophilum LplA-N–LplA-C complex adopts the same fold and the same spatial arrangement of domains as the structurally characterised bacterial apo-LplAs (PDB entries 1X2G, 2P0L and 1VQZ; Figure 2). LplA-N–LplA-C does, however, possess at least two distinctive local conformational features that could be functionally important. First, in E. coli LplA and bovine LPT, the adenylate-binding region is a loop, often at least partially disordered, whereas in LplA-N–LplA-C the equivalent region includes two contiguous α-helices (α4 and α5; Figure 3). We cannot say whether these helices persist at the optimum temperature (55°C) for T. acidophilum, but their presence reduces the probability that the LplA-N adenylate-binding region undergoes the same transition as the E. coli adenylate-binding region upon lipoate adenylation, which includes formation of a new β-strand anti-parallel to β13 of the C-terminal domain [13]. Secondly, LplA-C-induced conformational changes in LplA-N around strands β7 and β8 result in substantial shortening of a loop (that we label as the capping loop in Figure 3) that in isolated LplA-N partially occupies the space that the adenylate-binding loop occupies in E. coli LplA. It is likely that the capping loop functions at least to repress catalytic activity of LpA-N in the absence of LplA-C. It would be interesting in future to establish whether replacement of the capping loop with the equivalent short loop from E. coli LplA shows gain-of-function effects in isolated LplA-N.

E. coli LplA undergoes significant structural changes upon lipoate adenylation, including reorientation of the C-terminal domain to produce a more stretched overall conformation (PDB entry 3A7R). Bovine LPT adopts this stretched domain arrangement in both apo- and lipoyl-AMP-bound forms (PDB entries 2E5A and 3A7U). We have been unable to produce crystals of LplA-N–LplA-C complexes with lipoyl-AMP and with E2lipD that diffract to sufficient resolution to investigate structural changes in LplA-N–LplA-C. We have, however, studied LplA-N–LplA-C interactions with LA/ATP/Mg2+ and E2lipD using (1H-15N)-HSQC NMR spectra, which are highly sensitive to conformational changes and interactions. We performed two titrations in which at each step we recorded (1H-15N)-HSQC spectra of 15N-labelled LplA-C in complex with unlabelled LplA-N. In Titration 1, LA/ATP/Mg2+ were added before E2lipD such that lipoate adenylation and lipoate transfer are monitored separately. The observed NMR spectral changes (Figure 4A) are not immediately suggestive of a substantial LplA-C conformational change upon formation of the lipoyl-AMP intermediate, but a LplA-C positional change, like the 180° rotation of the C-terminal domain observed in E. coli LplA [13], cannot be ruled out as the observation of exchange in approximately 25 peaks indicates that nearly a third of LplA-C residues sample more than one chemical environment.

In Titration 2, E2lipD was added before LA/ATP/Mg2+. The lack of change upon E2lipD addition indicates that in apo-LplA-N–LplA-C, LplA-C does not interact with E2lipD and, further, if there is any E2lipD interaction with LplA-N in apo-LplA-N–LplA-C, it does not occur in the vicinity of LplA-C. Chemical-shift perturbations were observed in more than one-third of LplA-C (1H-15N)-HSQC peaks when LA/ATP/Mg2+ were then added to the mixture of apo-LplA-N–LplA-C and E2lipD (Figure 4C). Thus a more substantial change in LplA-C occurs when E2lipD is already present at the time of adding lipoate adenylation ingredients. A reliable explanation of this observation would require extensive further investigation, but for now we note that lipoate adenylation and lipoate transfer are monitored simultaneously in Titration 2, rather than separately as in Titration 1. We note also that at least 75% of peaks perturbed in Titration 1 were also perturbed in Titration 2, indicating the involvement of substantially overlapping regions of LplA-C in any conformational/positional changes occurring during the two titrations.

Our NMR data, particularly from Titration 2, indicate that LplA-C does undergo significant conformational change at one or more stages of lipoylation. This is consistent with our structure-based hypothesis that rearrangement of the LplA-N–LplA-C complex, akin to that seen in E. coli LplA [13], is required to allow E2lipD access to the LplA-N active site and hence to allow lipoate transfer. In our model of a possible end-point of such a rearrangement, LplA-C has rotated through approximately 120°, compared with the approximately 180° rotation observed for E. coli LplA C-terminal domain upon lipoate adenylation [13]. One potential flaw in our model, in common with the E. coli LplA–octanoyl-AMP–ApoH protein complex crystal structure, is that the distance between the adenylated intermediate and the acceptor lysine (LysE2lipD42) is too great for initiation of lipoyl transfer. It remains to be seen whether, as Fujiwara et al. [13] suggest, this is rectified if a version of a LplA–E2lipD/H protein complex with ‘true substrates’ can be crystallized.

Assuming that LplA-C becomes disordered if it is released from LplA-N, there is no evidence from our NMR data that LplA-C dissociates from LplA-N during lipoate adenylation or lipoate transfer; we detect only folded LplA-C species during catalysis of both steps, suggesting that LplA-C is not competed off LplA-N by incoming substrate and that LplA-C remains bound to LplA-N during any rearrangement of the complex. There remains the possibility, however, of minor populations of disordered LplA-C species at any one time that are not detected by the techniques used in the present study. On the other hand, if LplA-C remains structured upon dissociation from LplA-N, then rearrangement of the LplA-N–LplA-C complex could clearly involve a simple release and rebind mechanism.

We also used NMR to monitor the effect of lipoylation on E2lipD. We believe that the observed chemical-shift perturbations upon addition of the lipoylation ingredients (Figure 5) are more likely to result from lipoylation-induced localized conformational change in E2lipD than from non-covalent E2lipD interaction with LplA-N–LplA-C (present only in catalytic quantities) or substrate. Previous NMR analysis did not indicate any conformational change in E2lipD from Bacillus stearothermophilus PDHC upon lipoylation [44]. The reason for the difference is not obvious, but it is clear that the results of this titration represent further evidence that the T. acidophilum LplA-N–LplA-C complex is functional.

We have previously described features of the genes encoding LplA-C and LplA-N, including the facts that their genes overlap by a single base pair, and a TATA box is readily identifiable upstream of lpla-c; however, no cis-regulatory sequence is observed in the proximity of lpla-n, suggesting that the genes are transcriptionally coupled. Given that the gene order is lpla-c then lpla-n [18], our structure-based observation that the LplA-C C-terminus and LplA-N N-terminus are located at opposite ends of the LplA-N–LplA-C complex confirms that LplA-N and LplA-C are made as separate polypeptides.

We have shown that, like other LplAs, T. acidophilum LplA-N–LplA-C can use OA and octanoyl-AMP as substrates, albeit less efficiently than LA. At first glance this may be unsurprising, but fatty acid synthesis, the source of the OA precursor, is thought to be absent in the archaea [45]. It has long been hypothesized that ancient enzyme promiscuity gave rise to the specialized enzyme activities known today [46]. Hence, the ability to use OA as well as LA may reflect an early evolutionary state. Notably, this promiscuity has a physiological advantage in E. coli where LipB mutants can still carry out lipoylation using LipA and LplA [38]. LplA is evolutionarily related to LipB and biotin protein ligase [14], but whether LplA may have served as an evolutionary platform for LipB or vice versa is uncertain.

Abbreviations

     
  • E2lipD

    E2 lipoyl domain

  •  
  • IPTG

    isopropyl β-D-thiogalactopyranoside

  •  
  • GCS

    glycine cleavage system

  •  
  • HSQC

    heteronuclear single-quantum coherence

  •  
  • LA

    lipoic acid

  •  
  • LipA

    lipoic acid synthetase

  •  
  • LipB

    lipoyl(octanoyl) transferase

  •  
  • LplA

    lipoate protein ligase

  •  
  • LPT

    lipoyltransferase

  •  
  • MPD

    2-methyl-2,4-pentanediol

  •  
  • OA

    octanoic acid

  •  
  • OADHC

    2-oxoacid dehydrogenase complex

  •  
  • OGDHC

    2-oxoglutarate dehydrogenase complex

  •  
  • PDHC

    pyruvate dehydrogenase complex

  •  
  • RMSD

    root mean square deviation

AUTHOR CONTRIBUTION

Mareike Posner and Abhishek Upadhyay did all of the protein expression, purification, crystallization and NMR sample preparation. Mareike Posner, Abhishek Upadhyay and Susan Crennell acquired, processed and analysed X-ray diffraction data. Stefan Bagby acquired, processed and analysed NMR data. Mareike Posner and Abhishek Upadhyay did lipoylation assays. Mareike Posner modelled the LplA-N–LplA-C–E2lipD complex. Andrew Watson synthesized octanoyl-AMP. Steve Dorus and Stefan Bagby did the comparative genomic analysis. Mareike Posner, Abhishek Upadhyay, Michael Danson, Steve Dorus and Stefan Bagby designed experiments and interpreted the results. All authors contributed to writing of the paper, with Mareike Posner and Stefan Bagby making the major contributions.

We thank Professor R. Perham (University of Cambridge, Cambridge, U.K.) for kindly providing TM202 and pET11c plasmids, and Dr Geoff Kelly (MRC Biomedical NMR Centre, Mill Hill, London, U.K.) for assistance with NMR data collection.

FUNDING

This work was supported by a UNESCO-L’Oréal FWIS Fellowship (to M.G.P.).

References

References
1
Perham
R. N.
Swinging arms and swinging domains in multifunctional enzymes: catalytic machines for multistep reactions
Annu. Rev. Biochem.
2000
, vol. 
69
 (pg. 
961
-
1004
)
2
Cronan
J. E.
Zhao
X.
Jiang
Y. F.
Function, attachment and synthesis of lipoic acid in Escherichia coli
Adv. Microb. Physiol
2005
, vol. 
50
 (pg. 
103
-
146
)
3
McManus
E.
Luisi
B. F.
Perham
R. N.
Structure of a putative lipoate protein ligase from Thermoplasma acidophilum and the mechanism of target selection for post-translational modification
J. Mol. Biol.
2006
, vol. 
356
 (pg. 
625
-
637
)
4
Hassan
B. H.
Cronan
J. E.
Protein-protein interactions in assembly of lipoic acid on the 2-oxoacid dehydrogenases of aerobic metabolism
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
8263
-
8276
)
5
Fujiwara
K.
Toma
S.
Okamura-Ikeda
K.
Motokawa
Y.
Nakagawa
A.
Taniguchi
H.
Crystal structure of lipoate-protein ligase A from Escherichia coli: determination of the lipoic acid-binding site
J. Biol. Chem.
2005
, vol. 
280
 (pg. 
33645
-
33651
)
6
Morris
T. W.
Reed
K. E.
Cronan
J. E.
Identification of the gene encoding lipoate-protein ligase-A of Escherichia coli. Molecular cloning and characterization of the lplA gene and gene product
J. Biol. Chem.
1994
, vol. 
269
 (pg. 
16091
-
16100
)
7
Kang
S. G.
Jeong
H. K.
Lee
E.
Natarajan
S.
Characterization of a lipoate-protein ligase A gene of rice (Oryza sativa L.)
Gene
2007
, vol. 
393
 (pg. 
53
-
61
)
8
Fujiwara
K.
Hosaka
H.
Matsuda
M.
Okamura-Ikeda
K.
Motokawa
Y.
Suzuki
M.
Nakagawa
A.
Taniguchi
H.
Crystal structure of bovine lipoyltransferase in complex with lipoyl-AMP
J. Mol. Biol.
2007
, vol. 
371
 (pg. 
222
-
234
)
9
Schonauer
M. S.
Kastaniotis
A. J.
Kursu
V. A. S.
Hiltunen
J. K.
Dieckmann
C. L.
Lipoic acid synthesis and attachment in yeast mitochondria
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
23234
-
23242
)
10
Christensen
Q. H.
Hagar
J. A.
O’Riordan
M. X. D.
Cronan
J. E.
A complex lipoate utilization pathway in Listeria monocytogenes
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
31447
-
31456
)
11
Martin
N.
Christensen
Q. H.
Mansilla
M. C.
Cronan
J. E.
de Mendoza
D.
A novel two-gene requirement for the octanoyltransfer reaction of Bacillus subtilis lipoic acid biosynthesis
Mol. Microbiol.
2011
, vol. 
80
 (pg. 
335
-
349
)
12
Christensen
Q. H.
Martin
N.
Mansilla
M. C.
de Mendoza
D.
Cronan
J. E.
A novel amidotransferase required for lipoic acid cofactor assembly in Bacillus subtilis
Mol. Microbiol.
2011
, vol. 
80
 (pg. 
350
-
363
)
13
Fujiwara
K.
Maita
N.
Hosaka
H.
Okamura-Ikeda
K.
Nakagawa
A.
Taniguchi
H.
Global conformational change associated with the two-step reaction catalyzed by Escherichia coli lipoate-protein ligase A
J. Biol. Chem.
2010
, vol. 
285
 (pg. 
9971
-
9980
)
14
Reche
P. A.
Lipoylating and biotinylating enzymes contain a homologous catalytic module
Prot. Sci.
2000
, vol. 
9
 (pg. 
1922
-
1929
)
15
Wood
Z. A.
Weaver
L. H.
Brown
P. H.
Beckett
D.
Matthews
B. W.
Co-repressor induced order and biotin repressor dimerization: a case for divergent followed by convergent evolution
J. Mol. Biol.
2006
, vol. 
357
 (pg. 
509
-
523
)
16
Kim
D. J.
Kim
K. H.
Lee
H. H.
Lee
S. J.
Ha
J. Y.
Yoon
H. J.
Suh
S. W.
Crystal structure of lipoate-protein ligase A bound with the activated intermediate: insights into interaction with lipoyl domains
J. Biol. Chem.
2005
, vol. 
280
 (pg. 
38081
-
38089
)
17
Christensen
Q. H.
Cronan
J. E.
The Thermoplasma acidophilum LplA-LplB complex defines a new class of bipartite lipoate-protein ligases
J. Biol. Chem.
2009
, vol. 
284
 (pg. 
21317
-
21326
)
18
Posner
M. G.
Upadhyay
A.
Bagby
S.
Hough
D. W.
Danson
M. J.
A unique lipoylation system in the Archaea: lipoylation in Thermoplasma acidophilum requires two proteins
FEBS J.
2009
, vol. 
276
 (pg. 
4012
-
4022
)
19
Otwinowski
Z.
Minor
W.
Processing of X-ray diffraction data collected in oscillation mode
Methods Enzymol.
1997
, vol. 
276
 (pg. 
307
-
326
)
20
Long
F.
Vagin
A.
Young
P.
Murshudov
G. N.
BALBES: a molecular replacement pipeline
Acta Crystallogr. Sect. D Biol. Crystallogr.
2008
, vol. 
64
 (pg. 
125
-
132
)
21
Emsley
P.
Cowtan
K.
Coot: model-building tools for molecular graphics
Acta Crystallogr. Sect. D Biol. Crystallogr.
2004
, vol. 
60
 (pg. 
2126
-
2132
)
22
Bailey
S.
The CCP4 suite: programs for protein crystallography
Acta Crystallogr. Sect. D Biol. Crystallogr.
1994
, vol. 
50
 (pg. 
760
-
763
)
23
Davis
I. W.
Leaver-Fay
A.
Chen
V. B.
Block
J. N.
Kapral
G. J.
Wang
X.
Murray
L. W.
Arendall
W. B.
III
Snoeyink
J.
Richardson
J. S.
Richardson
D. C.
MolProbity: all-atom contacts and structure validation for proteins and nucleic acids
Nucleic Acids Res.
2007
, vol. 
35
 (pg. 
W375
-
W383
)
24
Laskowski
R. A.
Macarthur
M. W.
Moss
D. S.
Thornton
J. M.
PROCHECK: a program to check the stereochemical quality of protein structures
J. Appl. Crystallogr.
1993
, vol. 
26
 (pg. 
283
-
291
)
25
Reynolds
C.
Damerell
D.
Jones
S.
ProtorP: a protein-protein interaction analysis server
Bioinformatics
2009
, vol. 
25
 (pg. 
413
-
416
)
26
Krissinel
E.
Henrick
K.
Inference of macromolecular assemblies from crystalline state
J. Mol. Biol.
2007
, vol. 
372
 (pg. 
774
-
797
)
27
Delaglio
F.
Grzesiek
S.
Vuister
G. W.
Zhu
G.
Pfeifer
J.
Bax
A.
NMRPipe: a multidimensional spectral processing system based on UNIX pipes
J. Biomol. NMR
1995
, vol. 
6
 (pg. 
277
-
293
)
28
Vranken
W. F.
Boucher
W.
Stevens
T. J.
Fogh
R. H.
Pajon
A.
Llinas
P.
Ulrich
E. L.
Markley
J. L.
Ionides
J.
Laue
E. D.
The CCPN data model for NMR spectroscopy: development of a software pipeline
Proteins: Struct., Funct., Bioinf.
2005
, vol. 
59
 (pg. 
687
-
696
)
29
Wishart
D. S.
Bigam
C. G.
Yao
J.
Abildgaard
F.
Dyson
H. J.
Oldfield
E.
Markley
J. L.
Sykes
B. D.
1H, 13C and 15N chemical shift referencing in biomolecular NMR
J. Biomol. NMR
1995
, vol. 
6
 (pg. 
135
-
140
)
30
Upadhyay
A.
Burman
J. D.
Clark
E. A.
Leung
E.
Isenman
D. E.
van den Elsen
J. M. H.
Bagby
S.
Structure-function analysis of the C3 binding region of Staphylococcus aureus immune subversion protein Sbi
J. Biol. Chem.
2008
, vol. 
283
 (pg. 
22113
-
22120
)
31
Grzesiek
S.
Stahl
S. J.
Wingfield
P. T.
Bax
A.
The CD4 determinant for downregulation by HIV-1 Nef directly binds to Nef. Mapping of the Nef binding surface by NMR
Biochemistry
1996
, vol. 
35
 (pg. 
10256
-
10261
)
32
Bartoli
G.
Bosco
M.
Carlone
A.
Dalpozzo
R.
Marcantoni
E.
Melchiorre
P.
Sambri
L.
Reaction of dicarbonates with carboxylic acids catalyzed by weak Lewis acids: general method for the synthesis of anhydrides and esters
Synthesis
2007
(pg. 
3489
-
3496
)
33
Schall
O. F.
Suzuki
I.
Murray
C. L.
Gordon
J. I.
Gokel
G. W.
Characterization of acyl adenyl anhydrides: differences in the hydrolytic rates of fatty acyl-AMP and aminoacyl-AMP derivatives
J. Org. Chem.
1998
, vol. 
63
 (pg. 
8661
-
8667
)
34
Comeau
S. R.
Gatchell
D. W.
Vajda
S.
Camacho
C. J.
ClusPro: a fully automated algorithm for protein-protein docking
Nucleic Acids Res.
2004
, vol. 
32
 (pg. 
W96
-
W99
)
35
Comeau
S. R.
Gatchell
D. W.
Vajda
S.
Camacho
C. J.
ClusPro: an automated docking and discrimination method for the prediction of protein complexes
Bioinformatics
2004
, vol. 
20
 (pg. 
45
-
50
)
36
Benkert
P.
Kuenzli
M.
Schwede
T.
QMEAN server for protein model quality estimation
Nucleic Acids Res.
2009
, vol. 
37
 (pg. 
W510
-
W514
)
37
Holm
L.
Kääriäinen
S.
Rosenström
P.
Schenkel
A.
Searching protein structure databases with DaliLite v.3
Bioinformatics
2008
, vol. 
24
 (pg. 
2780
-
2781
)
38
Morris
T. W.
Reed
K. E.
Cronan
J. E.
Lipoic acid metabolism in Escherichia coli: the lplA and lipB genes define redundant pathways for ligation of lipoyl groups to apoprotein
J. Bacteriol.
1995
, vol. 
177
 (pg. 
1
-
10
)
39
Gong
X. M.
Peng
T.
Yakhnin
A.
Zolkiewski
M.
Quinn
J.
Yeaman
S. J.
Roche
T. E.
Specificity determinants for the pyruvate dehydrogenase component reaction mapped with mutated and prosthetic group modified lipoyl domains
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
13645
-
13653
)
40
Puthenveetil
S.
Liu
D. S.
White
K. A.
Thompson
S.
Ting
A. Y.
Yeast display evolution of a kinetically efficient 13-amino acid substrate for lipoic acid ligase
J. Am. Chem. Soc.
2009
, vol. 
131
 (pg. 
16430
-
16438
)
41
Fujiwara
K.
Okamura-Ikeda
K.
Motokawa
Y.
Lipoylation of acyltransferase components of α-ketoacid dehydrogenase complexes
J. Biol. Chem.
1996
, vol. 
271
 (pg. 
12932
-
12936
)
42
Fernandez-Suarez
M.
Baruah
H.
Martinez-Hernandez
L.
Xie
K. T.
Baskin
J. M.
Bertozzi
C. R.
Ting
A. Y.
Redirecting lipoic acid ligase for cell surface protein labeling with small-molecule probes
Nat. Biotechnol.
2007
, vol. 
25
 (pg. 
1483
-
1487
)
43
Cohen
J. D.
Zou
P.
Ting
A. Y.
Site-specific protein modification using lipoic acid ligase and bis-aryl hydrazone formation
ChemBioChem
2012
, vol. 
13
 (pg. 
888
-
894
)
44
Dardel
F.
Laue
E. D.
Perham
R. N.
Sequence-specific 1H-NMR assignments and secondary structure of the lipoyl domain of the Bacillus stearothermophilus pyruvate dehydrogenase multienzyme complex
Eur. J. Biochem.
1991
, vol. 
201
 (pg. 
203
-
209
)
45
Falb
M.
Müller
K.
Königsmaier
L.
Oberwinkler
T.
Horn
P.
von Gronau
S.
Gonzalez
O.
Pfeiffer
F.
Bornberg-Bauer
E.
Oesterhelt
D.
Metabolism of halophilic archaea
Extremophiles
2008
, vol. 
12
 (pg. 
177
-
196
)
46
Khersonsky
O.
Tawfik
D. S.
Enzyme promiscuity: a mechanistic and evolutionary perspective
Annu. Rev. Biochem.
2010
, vol. 
79
 (pg. 
471
-
505
)

Author notes

1

These authors contributed equally to this work.

2

Present address: Department of Chemistry, University of Canterbury, Christchurch, New Zealand.

3

Present address: Department of Biology, Syracuse University, Syracuse, New York, NY 13244, U.S.A.

Co-ordinates and the structure factor file for the T. acidophilum LplA-N–LplA-C structure and co-ordinates for the E2lipD structures are in the PDB under accession codes 3R07 and 2L5T.