The A-type domain in Escherichia coli NfuA is required for regenerating the auxiliary [4Fe–4S] cluster in Escherichia coli lipoyl synthase

The lipoyl cofactor plays an integral role in several essential biological processes. The last step in its de novo biosynthetic pathway, the attachment of two sulfur atoms at C6 and C8 of an n-octanoyllysyl chain, is catalyzed by lipoyl synthase (LipA), a member of the radical SAM superfamily. In addition to the [4Fe–4S] cluster common to all radical SAM enzymes, LipA contains a second [4Fe–4S] auxiliary cluster, which is sacrificed during catalysis to supply the requisite sulfur atoms, rendering the protein inactive for further turnovers. Recently, it was shown that the Fe–S cluster carrier protein NfuA from Escherichia coli can regenerate the auxiliary cluster of E. coli LipA after each turnover, but the molecular mechanism is incompletely understood. Herein, using protein–protein interaction and kinetic assays as well as site-directed mutagenesis, we provide further insight into the mechanism of NfuA-mediated cluster regeneration. In particular, we show that the N-terminal A-type domain of E. coli NfuA is essential for its tight interaction with LipA. Further, we demonstrate that NfuA from Mycobacterium tuberculosis can also regenerate the auxiliary cluster of E. coli LipA. However, an Nfu protein from Staphylococcus aureus, which lacks the A-type domain, was severely diminished in facilitating cluster regeneration. Of note, addition of the N-terminal domain of E. coli NfuA to S. aureus Nfu, fully restored cluster-regenerating activity. These results expand our understanding of the newly discovered mechanism by which the auxiliary cluster of LipA is restored after each turnover.

Since its initial identification by Sofia et al. in 2001, the radical SAM (RS) 2 superfamily has expanded to include almost 114,000 individual sequences representing more than 85 distinct reactions (1,2). Members of the RS superfamily coordinate a [4Fe-4S] which, in its reduced state, is used to fragment S-adenosylmethionine (SAM) to methionine and a 5Ј-deoxyadenosin 5Ј-radical (5Ј-dA ⅐ ) (3). Lipoyl synthase (LipA) uses the 5Ј-dA ⅐ intermediate to catalyze the last step of the de novo pathway for the biosynthesis of the lipoyl cofactor, which is the attachment of sulfur atoms at C6 and C8 of an n-octanoyllysyl chain on a lipoyl carrier protein. LipA belongs to a subclass of RS enzymes that contain additional auxiliary iron-sulfur (Fe-S) clusters, which further diversifies the chemical repertoire of the superfamily (4). The [4Fe-4S] auxiliary cluster of LipA has been shown to be cannibalized by the protein during catalysis to provide the sulfur atoms in the lipoyl product. Although this sacrificial role for an Fe-S cluster has been controversial, given that its destruction inactivates the protein, an abundance of biochemical, spectroscopic, and crystallographic evidence support it (5)(6)(7)(8)(9)(10).
All organisms have established and highly regulated pathways for Fe-S cluster assembly which share some basic components (11)(12)(13), and it seemed likely that these pathways might be involved in regenerating LipA's auxiliary cluster after each turnover. Briefly, a cysteine desulfurase provides sulfur from cysteine, whereas a still-debated source supplies iron to a general scaffold protein upon which an Fe-S cluster is assembled. In a poorly understood process, the scaffold protein transfers a cluster to targeting proteins, which recognize their targets and transfer newly assembled clusters to them. Defects in Fe-S cluster assembly in humans result in a number of prevalent disorders, such as Friedreich's ataxia and multiple mitochondrial dysfunctions disorder (14).
Recently, E. coli NfuA, a previously characterized intermediate Fe-S cluster carrier protein, was shown to restore the auxiliary cluster in LipA after each turnover in a process that does not limit the overall rate of catalysis (15)(16)(17). E. coli NfuA is composed of two distinct domains: a "degenerate" A-type N-terminal domain and an Nfu-like C-terminal domain, and previous studies have provided evidence for a role for both of these two domains in NfuA's function (18). Because LipA has now been established as a target for NfuA, we can exploit this system to better understand the roles of these two domains in Fe-S protein targeting and Fe-S cluster regeneration. This work was also supported by the Howard Hughes Medical Institute EXROP program (to Z. R. D.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This article contains Figs. S1-S6 and Tables S1 and S2. 1

Role of NfuA cysteine residues in cluster coordination and transfer
NfuA contains four cysteine residues, two of which are in its N-terminal A-type domain (Cys-39 and Cys-44), and two of which are in its C-terminal Nfu-like domain (Cys-149 and Cys-152) (Fig. 1). To assess the importance of each of the cysteine residues on NfuA's ability to enhance LipA catalysis, each was individually changed to an alanine residue, yielding four variant proteins. When Cys-39 or Cys-44 is changed to alanine, the resulting variant protein is as capable as WT NfuA of ligating a [4Fe-4S] cluster (Fig. S1A) and can enhance LipA turnover, although to a lesser extent (Fig. 2). In the absence of NfuA, 25 M LipA catalyzes formation of ϳ20 M lipoyl product in 2.5 h (Fig. 2, closed circles) S1B) and to enhance LipA turnover (Fig. 2, blue open circles and red open triangles) is abrogated. In these instances, LipA catalyzes significantly less than one turnover, suggesting that these NfuA variants inhibit LipA turnover. These results confirm the role of the two C-terminal cysteines in [4Fe-4S] cluster coordination and show that that the two N-terminal cysteines do not function in cluster transfer to LipA under the given reaction conditions. An earlier study in which the two N-terminal cysteine residues were substituted with serine residues reported that these variants could still coordinate a cluster (15). However, this observed difference may be because serine residues can participate in cluster coordination, as is the case for the auxiliary cluster of LipA as well as other variants of Fe-S proteins (19,20).

E. coli NfuA N-terminal domain is essential for tightly interacting with LipA
Currently, there is no well-defined function for the so-called degenerate A-type N-terminal domain of E. coli NfuA, although it has been proposed to play a role in protein-protein interactions (18). We investigated the role of the N-terminal domain in regenerating the auxiliary cluster of LipA by isolating and characterizing a fragment of NfuA (N-terminal domain) containing amino acid residues 1-97. This truncated domain has been shown to exhibit a far-UV CD spectrum that is identical to that of the full-length protein (18). We subjected the NfuA N-terminal domain to UV-visible spectroscopy and analyzed its effect on the activity of LipA. As expected, the UVvisible spectrum of the NfuA N-terminal domain shows no features that are indicative of an Fe-S cluster (Fig. 3). Further, the N-terminal domain does not have any apparent effect on LipA's activity (Fig. 4). In the presence of 400 M (200 M active dimer) WT NfuA, 25 M LipA catalyzes formation of almost 200 M lipoyl product in 2.5 h (Fig. 4, closed black squares). By contrast, in the absence of NfuA (Fig. 4, closed red circles), or when fulllength NfuA is replaced by its N-terminal domain (Fig. 4, closed blue circles), 25 M LipA catalyzes formation of slightly less than 25 M product. To determine whether the N-terminal domain of NfuA might be involved in protein-protein interactions, as has been suggested previously (18), its tight association with LipA was assessed by molecular-sieve chromatography ( Fig. 5A and Fig. S2). LipA alone (dotted trace) elutes at 60.6 ml, corresponding to an experimental mass of 44.3 kDa (theoretical mass is 38.2 kDa). The N-terminal domain of NfuA elutes at 75.3 ml (dashed trace), corresponding to an experimental mass of 12.5 kDa (theoretical mass is 12.6 kDa). The sample containing an equimolar mixture of LipA and the NfuA N-terminal domain (solid trace) elutes at 58.6 ml as a major peak, corresponding to an experimental mass of 52.4 kDa, which suggests a 1:1 complex between the two proteins (theoretical mass is 50.8 kDa). Fractions corresponding to the major peak were subjected to analysis by SDS-PAGE, and two bands corresponding

Domain architecture of Escherichia coli NfuA
to the masses of LipA and NfuA are observed ( Fig. 5B and Fig.  S6). Therefore, although the C-terminal domain of NfuA is essential for the protein's activity, it is dispensable for the interaction between NfuA and LipA.

E. coli NfuA C-terminal domain coordinates a [4Fe-4S] cluster but does not bind tightly to E. coli LipA
A wealth of evidence has supported a role for [4Fe-4S] cluster coordination at a dimer interface by two conserved cysteinyl residues in the Nfu-like C-terminal domain of NfuA (15,16). We therefore explored whether the C-terminal domain could still coordinate its cluster in the absence of the N-terminal domain. We first overproduced, isolated, and characterized a truncated version of E. coli NfuA lacking the N-terminal domain (residues 98 -191). Similar to the N-terminal domain, the Nfu-like C-terminal domain was also shown to have a far-UV CD spectrum identical to that of the full-length NfuA protein (18). The C-terminal domain was stable and was overproduced in high yields for analysis. The UV-visible spectrum of the as-isolated protein reveals features of a [4Fe-4S] cluster that are nearly identical to those of the full-length WT protein (Fig. 3). We then tested the effect of the C-terminal domain on LipA's activity. Surprisingly, although the C-terminal domain coordinates a [4Fe-4S] cluster, its ability to restore LipA's auxiliary cluster in a catalytic fashion is largely diminished (Fig. 4, closed triangles). Interestingly, however, the NfuA C-terminal domain allows for one additional turnover, albeit at a rate that is six to nine times slower than that in the presence of full-length NfuA (compare Fig. 4, closed triangles and closed squares). This additional turnover may result from complete destruction of the auxiliary [4Fe-4S] cluster on LipA after one turnover, and the subsequent relatively slow reconstitution of a new [4Fe-4S] cluster during catalysis at the expense of the NfuA C-terminal domain [4Fe-4S] cluster.
Given that the N-terminal domain of NfuA is found to interact with LipA in the absence of the NfuA C-terminal domain, we hypothesized that the C-terminal domain alone may be unable to recognize LipA or does so poorly. Therefore, a similar interaction analysis by molecular-sieve chromatography was conducted to assess whether LipA and the NfuA C-terminal domain interact tightly (Fig. 5C). LipA alone (dotted trace) elutes at 61.1 ml, corresponding to an experimental mass of 45.2 kDa (theoretical mass is 38.2 kDa). The C-terminal domain of NfuA alone elutes at 77.0 ml (dashed trace), corresponding to an experimental mass of 11.8 kDa (theoretical mass is 12.8 kDa). The sample containing an equimolar mixture of LipA and the NfuA C-terminal domain (solid trace) contains two major peaks: one peak at 60.9 ml and another at 77.0 ml. The first peak corresponds to LipA alone, whereas the second peak corresponds to the NfuA C-terminal domain. The absence of a shift in the elution volumes suggests that the two proteins do not form a tight complex as was observed previously for the fulllength NfuA and LipA (17), and for the NfuA N-terminal domain and LipA, as demonstrated in this work. Fractions corresponding to the major peak were analyzed by SDS-PAGE, confirming that the two proteins do not co-elute (Fig. 5D). The inability to interact tightly with LipA may explain the observation that the C-terminal domain alone has a limited effect on LipA's activity, despite the observation that it coordinates a [4Fe-4S] cluster. Together, our analysis suggests that the NfuA N-terminal domain is essential for recognizing LipA and interacting with it, and that in its absence, NfuA is unable to regenerate LipA's auxiliary cluster in a catalytic fashion.

NfuA from Mycobacterium tuberculosis also activates E. coli LipA
Given the structural studies that have been carried out on LipA from Mycobacterium tuberculosis (Mt) (10), we assessed whether Mt NfuA can also regenerate E. coli LipA's auxiliary Fe-S cluster. Inspection of an alignment of the two protein sequences shows that the primary structure of Mt NfuA shares 89% amino acid identity with the primary structure of E. coli NfuA, containing two domains corresponding to the N-termi-

LipA turnover is not enhanced by NfuA when using a peptide substrate containing an 8-thiooctanoyllysyl moiety
Early studies of lipoic acid biosynthesis showed that 8-thiooctanoic acid is converted into lipoic acid when administered to growing E. coli, but not as well as when octanoic acid is administered (21-23). Fig. 7 shows lipoyl product formation when a peptide substrate containing an 8-thiooctanoyllysyl amino acid residue is acted upon by LipA in the presence of 400 M NfuA. As can be observed, the lipoyl product levels off at a concentration that is nearly stoichiometric with that of LipA (75 M) in the reaction (closed circles). By contrast, when a peptide substrate containing an octanoyllysyl amino acid is used (closed squares), additional turnovers take place. Our model predicts that in the absence of chemistry that results in sulfur insertion at C8, the 8-thiooctanoyl substrate is converted to a lipoyl product that is still bonded to the auxiliary Fe-S cluster. This appendage, and/or the lack of partial or full destruction of the auxiliary cluster, is expected to inhibit cluster transfer from NfuA to LipA.

The addition of an A-type domain to S. aureus Nfu supports multiple turnover by LipA
The observation that the E. coli NfuA N-terminal domain is required to regenerate E. coli LipA led us to ask whether other Nfu-type proteins that lacked this domain could restore E. coli LipA's auxiliary cluster during catalysis. To address this ques-

Domain architecture of Escherichia coli NfuA
tion, we initially overproduced and isolated Nfu from S. aureus and used it in LipA reactions. The sequence homology between E. coli NfuA and S. aureus Nfu is limited to the C-terminal domain (Fig. S4). The UV-visible spectrum of S. aureus Nfu in its as-isolated state contains features that support the presence of a [4Fe-4S] cluster (Fig. S5), consistent with what was reported previously (24). However, when S. aureus Nfu is included in activity assays, its effect on LipA's activity is minimal, as shown in Fig. 8. In the presence of 400 M S. aureus NfuA, 25 M LipA catalyzes formation of ϳ25 M product (Fig.  8, closed triangles), whereas in its absence, it catalyzes formation of ϳ15 M product (Fig. 8, closed circles).
The effect of S. aureus Nfu on the LipA reaction is similar to that of the C-terminal domain of E. coli NfuA. The A-type domain in E. coli NfuA, which is shown to be essential in this work, is absent in S. aureus Nfu. We therefore created a fusion protein consisting of the E. coli NfuA N-terminal domain (residues 1-117) attached to S. aureus Nfu. Upon addition of the E. coli NfuA N-terminal domain S. aureus Nfu fusion to an E. coli LipA reaction, significant enhancement of lipoyl product formation is observed, providing more evidence for the essential role of this domain. Now, 25 M LipA catalyzes formation of greater than 250 M lipoyl product in 2.5 h (Fig. 8, closed squares).

Discussion
The recent demonstration that E. coli NfuA engenders multiple turnovers by E. coli LipA by restoring its Fe-S cluster cofactor provided the groundwork for expanding our understanding of this complex system. In this work, we confirm the critical role of the cysteinyl residues in the C-terminal domain of E. coli NfuA as coordinating ligands to an Fe-S cluster that is housed at the interface between two NfuA monomers, and show that the cysteinyl residues in the N-terminal domain have no apparent effect on cluster coordination or cluster transfer to LipA. In addition, we provide a detailed investigation of the essential role of the E. coli NfuA N-terminal domain in interacting with E. coli LipA, which was previously proposed to aid in protein-protein interactions (18). The C-terminal domain, which is homologous to many Nfu proteins, is fully capable of coordinating its cluster in the absence of the N-terminal domain, yet is unable to allow LipA to become catalytic, likely because it lacks the domain responsible for recognizing and interacting with E. coli LipA. Mt NfuA, which is highly homologous to the E. coli NfuA, is another NfuA that also contains the A-type domain, and engenders an identical effect on E. coli LipA. Interestingly, Nfu from S. aureus, which lacks the N-ter-    Cumulatively, our results highlight the role of the so-called degenerate A-type N-terminal domain as having an essential function. Although our results provide further insight into the regeneration of E. coli LipA's auxiliary cluster, significant gaps still remain in our understanding of the cluster restoration by NfuA. Our initial evidence that E. coli LipA can use an 8-thiooctanoyllysyl-containing peptide as a substrate, but not in a catalytic fashion, suggests that its cluster must be significantly (a [2Fe-2S] cluster or smaller) degraded before NfuA intervenes. Studies to provide a more detailed mechanism of the state of the cluster on NfuA as it is being transferred to LipA are currently underway.

General methods and instruments
The PCR was conducted using a Bio-Rad S1000 thermocycler. DNA sequencing was performed at the Penn State Genomics Core Facility. Amino acid analysis was performed at the UC Davis Proteomics Core Facility. UV-visible spectra were recorded on a Cary 50 spectrophotometer from Varian (Walnut Creek, CA) with the associated WinUV software package. All anaerobic experiments were conducted in a Coy Anaerobic Chamber (Grass Lake, MI). High-performance liquid chromatography (HPLC) with detection by MS (LC-MS) was conducted using an Agilent Technologies (Santa Clara, CA) 1200 system coupled to an Agilent Technologies 6410 QQQ mass spectrometer. Data collection and analysis were performed using the associated MassHunter software. Analytical molecular-sieve chromatography was performed on an ÄKTA system (GE Healthcare) housed in a Coy Anaerobic Chamber equipped with a HiPrep 16/60 Sephacryl S-200 HR column (GE Healthcare). SDS-PAGE was performed using a mini vertical electrophoresis unit from Hoefer (Holliston, MA). NfuA C39A, C44A, C149A, and C152A variants into pET28a The construction of the E. coli lipA-pET28a and E. coli nfuA-pET28a expression vectors has been described previously (5,17). E. coli nfuA-pET28a C39A, C44A, C149A, and C152A variants were constructed by site-directed mutagenesis using the Stratagene QuikChange II kit (Agilent Technologies) along with primers listed in Table S1. However, Pfu polymerase was substituted with Vent polymerase in the polymerase chain reaction. E. coli nfuA-pET28a plasmid DNA was used as the DNA template, and DNA sequencing was used to confirm the desired substitutions.

Cloning of genes corresponding to E. coli NfuA N-terminal and C-terminal domains into pET28a
The gene fragments corresponding to E. coli NfuA N-terminal domain (amino acid residues 1-97) and C-terminal domain (amino acid residues 98 -191) were PCR-amplified using the E. coli nfuA-pET28a plasmid as a DNA template. The primers used for amplification of the N-terminal domain and C-terminal domain introduced an NdeI restriction site at the 5Ј end of the gene and a XhoI restriction site at the 3Ј end. The gene fragments were digested and ligated into a pET28a vector digested with the same enzymes. Primers used for the constructs can be found in Table S1.

Cloning of the Mycobacterium tuberculosis nfuA gene
The gene encoding NfuA from Mycobacterium tuberculosis was codon-optimized for expression in E. coli by GeneArt Gene Synthesis (Thermo Fisher Scientific) and was received as a construct in plasmid pMA-T. The gene contained an NdeI restriction site at the 5Ј end and an XhoI restriction site at the 3Ј end. The gene was digested with these two enzymes and then ligated into a pET28a vector that was similarly digested. The desired construct was confirmed by DNA sequencing.

Cloning of S. aureus Nfu and S. aureus Nfu-E. coli NfuA N-terminal domain fusion protein into pET28a
Genes encoding Nfu (SAUSA300_0839) from S. aureus and a genetic fusion consisting of the N-terminal domain of E. coli NfuA (amino acids 1-117) attached to S. aureus Nfu were codon-optimized for expression in E. coli and received as constructs in plasmid pMA-T from GeneArt Gene Synthesis (Thermo Fisher Scientific). Both genes contained an NdeI restriction site at the 5Ј end and an XhoI restriction site at the 3Ј end. They were each digested with NdeI and XhoI and then ligated into pET28a that was similarly digested. The desired constructs were confirmed by DNA sequencing.

Overproduction of WT E. coli NfuA and NfuA variants, as well as Mt NfuA, S. aureus Nfu, and the E. coli NfuA N-terminal domain-S. aureus Nfu fusion protein
The overproduction of full-length E. coli NfuA has been previously described (17). E. coli NfuA C39A, C44A, C149A, C152A, NfuA N-terminal domain, NfuA C-terminal domain, Mt NfuA, S. aureus Nfu, and the E. coli NfuA N-terminal domain-S. aureus Nfu fusion were overproduced exactly as described for the full-length E. coli NfuA. Briefly, BL21(DE3) cells were co-transformed with the construct encoding the desired gene and plasmid pDB1282, which encodes the genes in the isc operon from Azotobacter vinelandii (25,26). A single colony was used to inoculate 200 ml lysogeny broth supplemented with 50 g/ml kanamycin and 100 g/ml ampicillin, and the starter culture was incubated overnight at 37°C with shaking at 250 rpm. The following day, 20 ml of the starter culture was used to inoculate four 6-liter flasks containing 4 liters M9 minimal media supplemented with 50 g/ml kanamycin and 100 g/ml ampicillin. The E. coli strains were cultured at 37°C with shaking at 180 rpm. At an A 600 ϭ 0.3, 0.2% arabinose was added to induce expression of the genes encoded in the pDB1282 plasmid. At an A 600 ϭ 0.6, 50 M FeCl 3 was added to the cultures, and the flasks were placed in an ice-water bath for ϳ30 min. Once chilled, expression of the desired gene was induced by the addition of 200 M isopropyl 1-thio-␤-D-galactopyranoside (final concentration) and allowed to proceed for ϳ18 h at 18°C with shaking at 180 rpm. Bacterial cells were Domain architecture of Escherichia coli NfuA harvested by centrifugation at 7500 ϫ g for 12 min. The resulting cellular pellet was flash-frozen in liquid N 2 and stored in liquid N 2 until further use.

Isolation of E. coli NfuA and NfuA variants, as well as Mt NfuA, S. aureus Nfu, and the E. coli NfuA N-terminal domain-S. aureus Nfu fusion protein
The purification of E. coli WT NfuA has been described previously (17). NfuA C39A, C44A, C139A, C152A, NfuA N-terminal domain, NfuA C-terminal domain, Mt NfuA, S. aureus Nfu, and the E. coli NfuA N-terminal domain-S. aureus Nfu fusion protein were purified exactly as described for WT E. coli NfuA. All purification steps were carried out in an anaerobic chamber containing Ͻ1 ppm O 2 (Coy Laboratory Products, Grass Lake, MI), with the exception of the centrifugation steps. In these instances, solutions were loaded into bottles that were then tightly sealed before being removed from the chamber for centrifugation. The cell pellet was resuspended in Buffer A (50 mM HEPES, pH 7.5, 300 mM KCl, 20 mM imidazole, and 10 mM ␤-mercaptoethanol). Lysozyme (1 mg/ml) and DNase I (0.1 mg/ml) were added, and the solution was stirred at room temperature for 30 min. The cells were lysed by sonic disruption, and the lysate was centrifuged at 45,000 ϫ g for 1 h. The supernatant was applied to a nickel-nitrilotriacetic acid (Ni-NTA) resin (Qiagen) column pre-equilibrated in Buffer A. The column was washed with Buffer B (50 mM HEPES, pH 7.5, 300 mM KCl, 45 mM imidazole, 10 mM ␤-mercaptoethanol, and 10% glycerol), and the protein was eluted using Buffer C (50 mM HEPES, pH 7.5, 300 mM KCl, 500 mM imidazole, 10 mM ␤-mercaptoethanol, and 10% glycerol). The protein was concentrated to 2.5 ml and then exchanged into Storage Buffer (50 mM HEPES, pH 7.5, 300 mM KCl, 1 mM DTT, 15% glycerol) using a PD-10 column (GE Healthcare). The protein was aliquoted, flash-frozen in liquid N 2 , and stored in liquid N 2 until further use. The E. coli NfuA N-terminal domain was further purified by size-exclusion chromatography using a HiPrep 16/60 Sephacryl S-200 HR column (GE Healthcare) connected to an ÄKTA FPLC system housed within an anaerobic chamber. The column was equilibrated in Storage Buffer at a constant flow rate of 0.5 ml/min before application and subsequent elution of the protein at the same flow rate. The purity of the isolated proteins was analyzed by SDS-PAGE (12.5% gel) and determined to be Ͼ95%. The protein concentration was determined via the Bradford method (27). Amino acid analysis revealed that the Bradford method overestimates the protein concentration of E. coli NfuA by a factor of 1.18, Mt NfuA by a factor of 1.03, S. aureus Nfu by a factor of 1.26, and the E. coli C-terminal domain by a factor of 1.24 (UC Davis Proteomics Core). Colorimetric iron and sulfide analyses were used to estimate the iron and sulfide content per polypeptide (28, 29).

Overproduction and isolation of E. coli LipA
The overproduction and isolation of E. coli LipA has been described previously (5). This method was performed exactly as described with the following amendment. In the overproduction of E. coli LipA, the M9 minimal media was inoculated with 0.2 ml of starter culture, which was allowed to grow overnight at 37°C with shaking at 180 rpm. The following day, expression of genes on pDB1282 was induced by addition of arabinose (0.2% final concentration) at A 600 ϭ 0.3. At an A 600 ϭ 0.6, 50 M FeCl 3 was added to the cultures, and the flasks were placed in an ice-water bath for ϳ30 min. Once chilled, expression of the lipA gene was induced by the addition of isopropyl 1-thio-␤-Dgalactopyranoside (200 M final concentration) and allowed to proceed for ϳ18 h at 18°C with shaking at 180 rpm. Bacterial cells were harvested by centrifugation at 7500 ϫ g for 12 min. The resulting cell pellet was flash-frozen in liquid N 2 and stored in liquid N 2 until further use. E. coli LipA was isolated using methods described previously with no modifications (5).

Analytical molecular-sieve chromatography
Physical interactions between E. coli LipA and truncated versions of E. coli NfuA, as well as among E. coli ErpA, E. coli NfuA, and E. coli LipA, were probed by analytical molecular-sieve chromatography. A 500 l mixture of standards composed of cytochrome c, carbonic anhydrase, bovine serum album, alcohol dehydrogenase, ␤-amylase, and blue dextran was applied to a HiPrep 16/60 Sephacryl S-200 HR column (GE Healthcare) connected to an ÄKTA FPLC system housed within an anaerobic chamber (Ͻ1 ppm O 2 ). The column was equilibrated in 50 mM HEPES, pH 7.5, 300 mM KCl, 15% glycerol, and 5 mM DTT (freshly prepared immediately before use) at a constant flow rate of 0.5 ml/min before applying and eluting protein samples. The log molecular weight of each standard was plotted against its elution volume after correcting for the void volume of the column. The linear equation was then used to estimate the molecular weight of each sample. The chromatogram of the standards as well as the standard curve used for experimental calculations are shown in

LC-MS activity determinations
The experimental setup and assay conditions were described in detail in previous publications with a few notable amendments (8,26). The reactions were conducted under strictly anaerobic conditions in a Coy anaerobic glovebox. Reactions of NfuA variants C39A, C44A, C149A, and C152A (Fig. 2)

Domain architecture of Escherichia coli NfuA
LipA, 400 M NfuA, 400 M peptide substrate (control) or 8-thiooctanoyl-containing peptide substrate, 0.5 M SAH nucleosidase, and 1 mM SAM. Reactions testing the effect of S. aureus Nfu or the E. coli NfuA N-terminal domain-S. aureus Nfu fusion protein (Fig. 8) contained 25 M LipA, 400 M S. aureus Nfu or fusion protein, 400 M peptide substrate, 0.5 M SAH nucleosidase, and 1 mM SAM. Reactions were conducted at room temperature and were initiated by addition of sodium dithionite to a final concentration of 2 mM. The reactions were quenched at appropriate times with H 2 SO 4 at a final concentration of 100 mM. All reactions were performed in triplicate. Detection of substrates and products was performed using electrospray ionization MS in positive mode (ESI ϩ -MS) with the following parameters: A nitrogen gas temperature of 340°C and flow rate of 9.0 liters/min, a nebulizer pressure of 40 psi, and a capillary voltage of 4000 V. Substrates and products were detected using multiple reaction monitoring (Table S2). The reaction mixture was separated on an Agilent Technologies Zorbax Extend-C18 Rapid Resolution HT column (4.6 mm ϫ 50 mm, 1.8 m particle size) equilibrated in 98% solvent A (0.1% formic acid, pH 2.6) and 2% solvent B (100% acetonitrile). A gradient of 2-23% solvent B was applied from 0.8 min to 3.5 min and maintained at 23% solvent B until 8 min, before returning to solvent B to 2% from 8 to 10 min. A flow rate of 0.4 ml/min was maintained throughout the method. The column was allowed to re-equilibrate for 2 min under the initial conditions between sample injections.