The architecture of the diaminobutyrate acetyltransferase active site provides mechanistic insight into the biosynthesis of the chemical chaperone ectoine

Ectoine is a solute compatible with the physiologies of both prokaryotic and eukaryotic cells and is widely synthesized by bacteria as an osmotic stress protectant. Because it preserves functional attributes of proteins and macromolecular complexes, it is considered a chemical chaperone and has found numerous practical applications. However, the mechanism of its biosynthesis is incompletely understood. The second step in ectoine biosynthesis is catalyzed by l-2,4-diaminobutyrate acetyltransferase (EctA; EC 2.3.1.178), which transfers the acetyl group from acetyl-CoA to EctB-formed l-2,4-diaminobutyrate (DAB), yielding N-γ-acetyl-l-2,4-diaminobutyrate (N-γ-ADABA), the substrate of ectoine synthase (EctC). Here, we report the biochemical and structural characterization of the EctA enzyme from the thermotolerant bacterium Paenibacillus lautus (Pl). We found that (Pl)EctA forms a homodimer whose enzyme activity is highly regiospecific by producing N-γ-ADABA but not the ectoine catabolic intermediate N-α-acetyl-l-2,4-diaminobutyric acid. High-resolution crystal structures of (Pl)EctA (at 1.2–2.2 Å resolution) (i) for its apo-form, (ii) in complex with CoA, (iii) in complex with DAB, (iv) in complex with both CoA and DAB, and (v) in the presence of the product N-γ-ADABA were obtained. To pinpoint residues involved in DAB binding, we probed the structure-function relationship of (Pl)EctA by site-directed mutagenesis. Phylogenomics shows that EctA-type proteins from both Bacteria and Archaea are evolutionarily highly conserved, including catalytically important residues. Collectively, our biochemical and structural findings yielded detailed insights into the catalytic core of the EctA enzyme that laid the foundation for unraveling its reaction mechanism.

ADABA were obtained. To pinpoint residues involved in DAB binding, we probed the structure-function relationship of (Pl)EctA by site-directed mutagenesis. Phylogenomics shows that EctA-type proteins from both Bacteria and Archaea are evolutionarily highly conserved, including catalytically important residues. Collectively, our biochemical and structural findings yielded detailed insights into the catalytic core of the EctA enzyme that laid the foundation for unraveling its reaction mechanism.
Compatible solutes are a distinct group of highly water-soluble organic osmolytes that are compliant with the biochemistry and physiology of both prokaryotic and eukaryotic cells (1)(2)(3). The function-preserving attributes of these solutes for proteins and other cellular components (4 -9) led to their description as chemical chaperones (10 -12). Building on the special physico-chemical characteristics of these compounds, many members of the Bacteria, Archaea, and Eukarya use compatible solutes as cytoprotectants against different types of environmental and cellular challenges (1)(2)(3)(13)(14)(15).
Compatible solutes have been widely adopted by microorganisms as osmotic stress protectants (3, 16 -18). Their amassing, either through synthesis or uptake (13), increases the osmotic potential of the cytoplasm and prevents a long-lasting increase in the ionic strength of this cell compartment (19,20) under osmotically unfavorable conditions. As an immediate result of compatible solute accumulation, high osmolaritytriggered water efflux from the cell is counteracted. This in turn prevents drop of vital turgor to physiologically nonsustainable values and averts an undue increase in molecular crowding of the cytoplasm (19 -22).
Reflecting their function as chemical chaperones, ectoines have already found a considerable number of commercial applications (26,40,66). To satisfy the increased demand for ectoines, an industrial scale biotechnological production process has been developed that uses the highly salt-tolerant bacterium Halomonas elongata as a natural and engineered cell factory (67); it is able to deliver ectoines on the scale of tons annually (40). Hence, both from the perspective of basic science and the biotechnological production of ectoines, a deeper understanding of the properties of the ectoine/5-hydroxyectoine biosynthetic enzymes is desirable (34,36,68). Substantial advances in this context have recently been made though detailed biochemical and structural studies of EctB (69), EctC (25, 70), and EctD (71)(72)(73). In contrast, EctA, the focus of this study, is far less well-understood.
Basic biochemical properties of EctA enzymes from H. elongata and several methylotrophs have been reported (34,35,68,76). However, a thorough understanding of EctA is still lacking, and in particular, crystal structures in complex with its substrates and/or its reaction product are missing. To fill this gap, we report here biochemical and structural characteristics of EctA from the thermotolerant bacterium Paenibacillus lautus (Pl) (77) in its apo, substrate, and co-substrate-bound forms and a crystal structure trapping the reaction product. Collectively, this crystallographic analysis, combined with site-directed mutagenesis experiments, illuminates the architecture of the active site of the EctA L-2,4-diaminobutyrate acetyltransferase and allows a proposal for its enzyme reaction mechanism.

Overproduction and purification of (Pl)EctA
Ectoine/5-hydroxyectoine-producing microorganisms can populate ecological niches with rather different biological and physico-chemical characteristics (26,40,41). One of these ectoine-producing microorganisms is the P. lautus strain Y4.12MC10, a Gram-positive spore-forming intestinal bacterium that was originally isolated from the effluent of the Obsidian hot spring in Yellowstone National Park (77). We explored the suitability of the (Pl)EctA protein for biochemical and crystallographic analysis.
For the heterologous production of (Pl)EctA in Escherichia coli, we constructed expression vectors using a codon-optimized version of the (Pl)ectA gene (GenBank TM accession number MF327591.1). These constructs yielded (Pl)EctA proteins with either a N-terminal (NH 2 -WSHPQFEK-SG) or a C-terminal (SA-WSHPQFEK-COOH) Strep-Tag II peptide for their purification by affinity chromatography. The corresponding synthetic (Pl)ectA constructs were expressed under the control of the plasmid-based TetR-regulated tet promoter. Both the Strep-Tag II-(Pl)EctA and the (Pl)EctA-Strep-Tag II proteins could be purified to apparent homogeneity (Fig. 1B). Both versions of the (Pl)EctA protein show a dimeric state in solution, as assessed by size-exclusion chromatography coupled to multiangle light scattering (SEC-MALS) (78). This is documented for the (Pl)EctA-Strep-Tag II protein in Fig. 1C and revealed a molecular mass of 39.73 Ϯ 0.04 kDa. The calculated theoretical molecular masses of the recombinant monomers are 20.68 kDa (Strep-Tag II-(Pl)EctA) and 20.25 kDa ((Pl)EctA-Strep-Tag II), respectively. Thus, the SEC-MALS data suggest that the (Pl)EctA protein is a stable dimer in solution, in line with previous reports that the EctA proteins from H. elongata and from the methylotrophs Methylomicrobium alcaliphilum, Methylophaga alcalica, and Methylophaga thalassica are also dimers in solution (34,68). Collectively, these findings suggest that dimer formation is probably a general feature of EctA-type L-2,4-diaminobutyrate acetyltransferases.
Analysis of the purified Strep-Tag II-(Pl)EctA and (Pl)EctA-Strep-Tag II proteins by SDS-PAGE revealed a significant difference in their migration behavior (Fig. 1B) that was not expected from their almost identical calculated theoretical molecular masses. We therefore performed an electrospray ionization-MS analysis of both recombinant proteins. The molecular masses determined by this technique revealed values of 20.55 kDa for the Strep-Tag II-(Pl)EctA protein and 20.12 kDa for the (Pl)EctA-Strep-Tag II protein, respectively. Hence, in both cases, the experimentally determined masses correlate with the calculated theoretical masses of the recombinant pro-teins minus 0.13 kDa. This finding could possibly indicate a posttranslational elimination of the N-terminal methionine residues during heterologous production and purification of the two N-terminal and C-terminal Strep-Tag II-marked (Pl)EctA proteins. The reason for the notable difference in their migration behavior on 15% SDS-polyacrylamide gels is not apparent.

Kinetic parameters of (Pl)EctA
In the ectoine biosynthesis route, N-␥-ADABA is the expected product of the EctA-catalyzed enzyme reaction (34,36,41). In contrast, its isomer N-␣-acetyl-L-2,4-diaminobutyric acid (N-␣-ADABA) is formed as an intermediate during the catabolism of ectoine when microorganisms use this nitrogenrich compound as a nutrient (67,79). Notably, N-␣-ADABA, but not N-␥-ADABA, serves as the inducer for the MocR/ GabR-type EnuR regulatory protein (80) controlling the expression of many ectoine catabolic gene clusters (41,81). To ascertain that the (Pl)EctA enzyme exclusively synthesizes N-␥-ADABA from L-2,4-diaminobutyrate and acetyl-CoA (34, 36) (Fig. 1A), we performed an enzyme assay in which we subsequently benchmarked the formed reaction product(s) against chemically synthesized and purified N-␥-ADABA and N-␣-ADABA reference samples. We used for this experiment an HPLC analysis protocol allowing the separation of the two ADABA isomers (70,82). Our analysis showed that the (Pl)EctA enzyme indeed exclusively synthesizes N-␥-ADABA under in vitro assay conditions ( Fig. 2A). Therefore, we conclude that the EctA L-2,4-diaminobutyrate acetyltransferase transfers the acetyl group from the co-substrate to the substrate in a highly position-specific manner.
For the initial enzyme assays, we determined basic biochemical characteristics of the recombinant (Pl)EctA-Strep-Tag II protein, including its pH optimum, its salt tolerance, and its optimal temperature. The (Pl)EctA enzyme showed a broad pH optimum under alkaline conditions (pH 8.5-9.5), maintaining 75% of its activity at pH 10 ( Fig. S1A). The enzyme was sensitive against extreme acidic conditions, as less then 10% of enzyme activity was retained at pH 6.0. Based upon these data, we performed all subsequent enzyme assays at pH 7.5 (75% EctA activ-  The tolerance of the (Pl)EctA enzyme against increased concentrations of NaCl was rather modest: a content of 100 mM NaCl in the assay solution already led to a notable decrease of the enzyme activity (down to 75%), and only about 12% of the enzyme activity remained when the NaCl content of the reaction buffer was increased to 1.5 M (Fig. S1B). The salt sensitivity of the (Pl)EctA enzyme contrasts sharply with the enzymatic characteristics of the EctB L-2,4-diaminobutyrate transaminase (69) and of the EctC ectoine synthase (25) from P. lautus Y4.12MC10, as these are highly salt-tolerant enzymes. Differences with respect to the behavior of EctA-type enzymes from various microorganisms toward NaCl have already been reported. Moderate concentrations of NaCl (0.2-0.4 M) activate the corresponding enzymes from H. elongata and M. alcaliphilum. However, increased NaCl concentrations inhibit those of M. thalassica and M. alcalica (34,68), a feature shared by the P. lautus EctA enzyme (Fig. S1B).
Keeping the thermo-tolerant nature of the P. lautus Y4.12MC10 donor strain in mind (77), a substantial degree of thermo-resistance of the (Pl)EctA enzyme activity was expected. Predictably, the activity of the (Pl)EctA enzyme increased with increasing temperature (Fig. 1C). However, the purified protein was not very thermostable, as the enzyme was only active for very short times at temperatures higher than 40°C. For instance, at 50°C, the (Pl)EctA enzyme was highly active but only for a few seconds, whereupon the activity dropped precipitously.
Considering the results of these basic enzyme assays, and taking the spontaneous hydrolysis of acetyl-CoA under alkaline conditions and the instability of the (Pl)EctA protein at high temperatures into account, we devised an enzyme assay for the (Pl)EctA protein. It employed the following conditions: 100 mM TES buffer (pH 7.5), 4 mM acetyl-CoA or 5 mM DAB, respectively, and a temperature of 30°C. The following apparent kinetic parameters for the (Pl)EctA were determined: (i) a K m of 0.13 Ϯ 0.03 mM and a V max of 44.87 Ϯ 2.66 units mg Ϫ1 for the substrate DAB and (ii) K m of 2.79 Ϯ 0.73 mM and a V max of 58.27 Ϯ 6.56 units mg Ϫ1 for acetyl-CoA (Fig. 2, B and C).

Crystal structure of apo-form of EctA reveals a homodimer
To gain insights into the molecular mechanisms of the EctA enzyme, we determined crystal structures of the apo-form of (Pl)EctA in the presence of the two substrates or its reaction product (Fig. 1A). First, we determined the crystal structure of the apo-(Pl)EctA protein; the obtained structure had a resolution of 2.2 Å ( Table 1). The (Pl)EctA protein adopts the classical GNAT fold (74, 75) with a mixed parallel/antiparallel twisted ␤-sheet in its center, flanked by four ␣-helices. A short 3 10 -helix (␣5) is present on the extended outward-facing loop connecting ␤-strand 6 and ␤7 (Fig. 3, A and B).
To identify the structurally closest homologs of the (Pl)EctA protein, we performed a DALI search (83), which revealed a variety of acetyltransferases among the top hits. The structurally closest relative of (Pl)EctA is the EctA from the pathogen Bordetella parapertussis (Bp) in complex with the substrate DAB (PDB code 3D3S: r.m.s.d. of 1.3 Å over 158 C␣ atoms). No Crystal structure of L-2,4-diaminobutyrate acetyltransferase detailed description of this crystal structure or publication describing the salient features of the (Bp)EctA enzyme is currently available. We discuss the position of the DAB ligand in the 3D3S crystal structure below, as it differs substantially from that which we found in the (Pl)EctA crystal structure. Further inspection of the crystal packing, in line with our SEC-MALS analysis (Fig. 1C), revealed the presence of an (Pl)EctA homodimer within the asymmetric unit (AS). Two (Pl)EctA monomers are packed against each other mainly through helices ␣1 and ␣2, the short 3 10 -helix ␣5, and strand ␤5 and their corresponding loop regions. Both monomers are rotated against each other by about 180° (Fig. 4, A and B). The interface area, as calculated with the PDBePISA online server (http://www.ebi.ac.uk/pdbe/prot_int/pistart.html), 5 is 1505.3 Å 2 and comprises 27 hydrogen bonds or salt bridges (Table S1). Notably, the dimeric assembly of (Pl)EctA also revealed that the side chain of Tyr 38 from one monomer penetrates into the other and vice versa (Fig. 4B). This residue is part of the amino acid sequence 36 SPYCYMLLGD 45 that forms helix ␣2 (Fig. 3A). The aromatic side chain of Tyr 38 points toward the binding site of the DAB substrate ( Fig. 4B) (see below).

The crystal structures of (Pl)EctA in complex with the cosubstrate reveals an evolutionarily conserved CoA-binding site
Next, we determined the (Pl)EctA:CoA structure by molecular replacement using the apo-(Pl)EctA structure as a search model. This crystal structure has a resolution of 1.5 Å ( Table 1). The CoA molecule could be unambiguously placed in the crystal structure and refined into a well-defined density present in the active site of (Pl)EctA (Fig. 5). A comparison of the apo-and CoA-bound structures of (Pl)EctA revealed no major overall differences, as indicated by an r.m.s.d. of only 0.4 Å over 150 C␣ atoms.
In the (Pl)EctA:CoA complex, the CoA molecule (Fig. 5, A and B) is bound in a deep cleft formed by the loops connecting ␣1-␣2, ␤4-␣3, ␤5-␣4, and helices ␣3 and ␣4 (Fig. 3), creating a mainly positively charged surrounding for the co-substrate (Fig. 5, A and C). The 3Ј-phosphate group of the adenosine moiety interacts with the side chains of Lys 94 (at a distance of 2.9 Å) and Arg 125 (at a distance of 3.0 Å), whereas the side chain of Arg 89 acts as a clamp that, together with the helix ␣4, sandwiches the adenine moiety. As characteristic for GNAT superfamily enzymes, the "P-loop" is the signature motif for the CoApyrophosphate-binding site (74,75,84) and possesses the following consensus sequence: Gln/Arg-X-X-Gly-X-Gly/Ala. The amino acid sequence of this region present in the (Pl)EctA protein corresponds to residues Arg 88 -Arg 89 -Gln 90 -Gly 91 -Ile 92 -Ala 93 . The oxygen atoms of the pyrophosphate are involved in hydrogen bonding with the backbone nitrogen atoms of Arg 89 (at a distance of 2.8 Å), Gln 90 (at a distance of 3.3 Å), Gly 91 (at a distance of 2.9 Å), Ala 93 (at a distance of 2.9 Å), and Lys 94 (at a distance of 2.9 Å). The pantothenate unit is hydrogen-bonded by its amide O atom to the backbone of Val 83 (at a distance of 2.8 Å). The ␤-alanine carbonyl oxygen interacts with the side chain of Asn 120 (2.8 Å) and the cysteamine nitrogen with the backbone hydroxyl group of Val 81 (at a distance of 2.7 Å) (Fig. 5C). Although the overall fold of (Pl)EctA does not significantly differ upon substrate and/or cofactor binding, it is worth mentioning that the N-terminal part of the loop connecting helices ␣1 and ␣2 (i.e. amino acids 29 -32) folds slightly   Crystal structure of L-2,4-diaminobutyrate acetyltransferase closer toward ␣2 in the (Pl)EctA:CoA crystal structure, thereby enabling the tail of the cofactor to orient toward the substratebinding site.
CoA-binding sites are evolutionarily highly conserved (75,84). We therefore assessed whether this was also true for the corresponding binding site in the (Pl)EctA:CoA crystal structure. Fig. S2 (A-C) represents overlays of the (Pl)EctA:CoA structure with that of the Ard1 acetyltransferase from the archaeon Sulfolobus sulfataricus P2 (PDB code 2X7B), an acetyltransferase from the bacterium Agrobacterium tumefaciens (PDB code 2GE3), and the human acetyltransferase NAA50 (PDB code 4X5K). This comparison highlights that the CoA-binding site in the (Pl)EctA:CoA crystal structure corresponds closely with the architecture(s) of typical CoA-binding sites (75,84).

The (Pl)EctA homodimer forms the binding sites for the DAB substrate
To further understand the catalytic mechanism of (Pl)EctA, we determined its crystal structure bound to its substrate DAB. This structure has a resolution of 1.5 Å (Table 1) and was solved by molecular replacement using the apo-(Pl)EctA structure as a search model. Our structural analysis shows that the DAB substrate is located in a narrow grove that is buried inside the enzyme (Fig. 4A). Contrary to the CoA-binding site, parts of both monomers of (Pl)EctA build up the DAB-binding site. From monomer A, the loop connecting ␣1-␣2, ␤4, the C terminus of ␤5, and the C-terminal part of the loop to ␤7 (Fig. 3) contributes to the binding pocket. In addition, helix ␣2 (particularly residue Tyr 38 ) from monomer B completes the mainly negatively charged substrate-binding pocket ( Fig. 4B and Fig.  6A). DAB interacts via both oxygen atoms of the carboxyl group with the side chain of Gln 80 and Trp 79 (at distances of 3.0 and 3.2 Å) and the side-and main-chain atoms of Asp 33 (at distances of 3.2 and 2.8 Å). The side chain of Asp 33 in turn is coordinated via the side chain of His 155 (at a distance of 2.8 Å). The DAB nitrogen is hydrogen-bonded by Asp 33 and Glu 158 (at distances of 3.2 and 2.7 Å), whereas the distal nitrogen atom interacts with the backbone of Trp 79 (at a distance of 2.9 Å) (Fig.  6A).
To consolidate our structural assessments of the DAB-binding site, we probed the importance of seven DAB-contacting residues for (Pl)EctA enzyme activity via site-directed mutagenesis. We replaced each of these DAB-contacting residues separately with an Ala residue ( Table 2). In an alignment of 432 bona fide EctA proteins (41), these seven residues are either strictly conserved or conservatively replaced by amino acid residues with similar properties. This is documented in an abbreviated alignment of 15 EctA-type proteins when (Pl)EctA was used as the search query (Fig. S3). For all constructed (Pl)EctA variants, we observed no differences in the their purification compared with the WT protein. All mutants exhibited a significantly reduced enzyme activity. Whereas the H155A, D33A, and W79A (Pl)EctA variants showed strongly reduced enzyme activity (remaining activities were 29.8, 8.7, and 9.9%, respectively), the other four (Pl)EctA variants exhibited only residual enzyme activity (Ͻ2%) ( Table 2). Because Tyr 38 of monomer B protrudes into the DAB-binding site of monomer A (and vice versa) (Fig. 4B) and is apparently crucial for (Pl)EctA enzyme activity (Table 2), we asked whether the Y38A substitution would fundamentally disturb the dimer assembly of (Pl)EctA. We therefore carried out a SEC-MALS analysis of the (Pl)EctA-Y38A variant and found that it still forms a stable dimer in solution (Fig. S4). Collectively, our mutant studies support the position and salient characteristics of the DAB-binding site that we have captured in the (Pl)EctA:DAB crystal complex (Fig.  6A).

Crystal structure of L-2,4-diaminobutyrate acetyltransferase
Among the top hits from the DALI search with the (Pl)EctA crystal structure as the query, only the crystal structure of the L-2,4-diaminobutyrate acetyltransferase (EctA) from B. parapertussis (PDB code 3D3S) contains DAB. The (Pl)EctA and (Bp)EctA crystal structures possessed an r.m.s.d. of 1.3 Å over 156 C␣ atoms when the structures were overlaid with each other (Fig. S5). However, in the (Bp)EctA crystal structure, the DAB-binding site is located at the interface of the two monomers (Fig. S5), whereas it is deeply buried in the two monomers of the (Pl)EctA crystal structure that we present here (Fig. 4A). DAB needs to react with acetyl-CoA for the EctA-mediated transfer of the acetyl group to form the reaction products N-␥-ADABA and CoA (Fig. 1A). Hence, one would expect the two substrates to be positioned in close proximity in the active site. In the crystal structures of the (Pl)EctA protein, this is indeed the case (Fig. 4A) (see below). However, in the (Bp)EctA:DAB crystal structure, the predicted position of the CoA molecule would be located at a large distance from the DAB molecule. Because the crystallization condition and a detailed description of the (Bp)EctA crystal structure have not yet been published, we cannot distinguish whether this is due to crystallization procedures or whether this protein contains an active site whose architecture is substantially different from that of (Pl)EctA. This latter possibility seems unlikely to us because the amino acid sequences of bona fide EctA-type proteins are highly conserved (41) (note: the (Pl)EctA and (Bp)EctA proteins possess an amino acid sequence identity of 36.4% and amino acid sequence similarity of 49.7%).

The (Pl)EctA crystal structure in complex with CoA and DAB
We succeeded in crystallizing (Pl)EctA in complex with both the CoA cofactor and the DAB substrate at a resolution of 1.2 Å (Table 1). Whereas (Pl)EctA:CoA contains only one monomer in the ASU, the ternary complex ((Pl)EctA:CoA:DAB) crystallized as a dimer in the ASU. This is the functional dimer, which was observed in all crystal structures when we inspected the symmetry-related molecules. In the (Pl)EctA:CoA and (Pl)Ec-tA:CoA:DAB crystal structures, the overall fold of the protein and the binding of CoA were very similar, as evidenced by the low r.m.s.d. of 0.41 Å over 131 C␣ atoms between both crystal structures. The largest difference was observed for the tail of the CoA molecule and the loop between helix ␣1 and ␣2. The ␤-alanine-cysteamine tail of CoA is oriented slightly outward, thereby introducing an additional hydrogen bond that is formed by the oxygen of the Ser 31 side chain with the N4P atom in monomer A (at a distance of 2.7 Å) (Fig. 7). Through this stable interaction, the loop between helices ␣1 and ␣2 is shifted toward the CoA cofactor. In chain B of the (Pl)EctA:CoA:DAB structure, the CoA tail is bent more outward, such that the interactions with Asn 120 and Ser 31 are weakened by an increased distance to 3.7 Å. In the (Pl)EctA:CoA:DAB structure, the DAB substrate is located at the same position as found in the (Pl)EctA:DAB structure. Actually, the DAB molecule is bound by via the same set of interactions in both crystal structures (Fig. 6A). Hence, the (Pl)EctA:CoA:DAB crystal structure probably represents the transition state of the (Pl)EctA enzyme.

The N-␥-ADABAbinding site of the (Pl)EctA enzyme
One of the crystal structures that we obtained (PDB code 6SJY) contained the reaction product of the EctA enzyme N-␥-ADABA (Fig. 1A). This crystal structure had a resolution of 2.2 Å (Table 1). Three monomers of the (Pl)EctA protein are present in the ASU, and only in monomer C was the electron density in the substrate-binding pocket sufficiently defined to evaluate the correct orientation of the N-␥-ADABA molecule within the active site (Fig. 6B). N-␥-ADABA forms hydrogen bonds with Asp 33 , Glu 158 , Gln 80 , and Trp 79 (at distances of 3.2, 2.7, 3.3, and 3.5 Å, respectively), whereas the carbonyl oxygen of the acetyl group transferred from acetyl-CoA onto DAB to form N-␥-ADABA makes a hydrogen bond to the backbone nitrogen of Val 81 (at a distance of 3.1 Å) (Fig. 6B). Notably, the enzyme reaction product N-␥-ADABA occupies essentially the same position and orientation as the substrate DAB within the active site of (Pl)EctA protein (Fig. 6, compare A with B).

Structures of the apo-, secondary, and ternary complex of (Pl)EctA represent different steps of the catalytic cycle
By comparing all obtained (Pl)EctA crystal structures, crucial steps in the catalytic cycle of the L-2,4-diaminobutyrate acetyltransferase can be visualized ( Fig. 8 and Video S1). In the apoform of the enzyme, the binding sites for the substrate DAB and the co-substrate acetyl-CoA are present in an "open" conformation (Fig. 8A). In this structure, there is a surface-exposed extended tunnel in which acetyl-CoA will bind, and a deep cavity is present in which DAB will be bound (Video S1). We do not know in which sequence of events the (Pl)EctA protein recognizes its two substrates acetyl-CoA and DAB. However, the kinetic parameters of the (Pl)EctA enzyme (Fig. 2, B and C) suggest that DAB might bind prior to acetyl-CoA to the protein.
In the two secondary complexes that we obtained ((Pl)EctA: CoA and (Pl)EctA:DAB), the chemical groups of the two substrates involved in the acetylation reaction point toward each other (Fig. 8, B and C). In the next crystal structure, the reaction product of EctA, N-␥-ADABA, is captured ( Fig. 8D and Video 1).
In the crystal structure of the (Pl)EctA:DAB:CoA ternary complex (Fig. 8E), the CoA molecule is somewhat differently orientated from the position of the CoA molecule observed in the secondary (Pl)EctA:CoA complex. In particular, the SHgroup of CoA points in a different direction in these two complexes (Fig. 8, compare B with E). While keeping in mind that crystal structures only provide a snapshot of the various states a protein can adopt, the superimposable positions of the substrate DAB and the reaction product N-␥-ADABA in the (Pl)EctA active site (Fig. 8, compare C with D) suggest that the protein backbone and the amino acid side chains do not move substantially during enzyme catalysis. An overlay of the (Pl)EctA:DAB and (Pl)EctA:CoA crystal structures revealed a distance of Ͻ3.0 Å between the sulfur atom of the CoA molecule and the reactive nitrogen in the ␥-position of DAB (Fig. 8F). This structural comparison suggests that these two secondary complexes might represent stages of the L-2,4-diaminobutyrate acetyltransferase prior to catalysis.

Crystal structure of L-2,4-diaminobutyrate acetyltransferase
Because acetyl-CoA is highly reactive, we were not able to obtain crystal structures of the (Pl)EctA:acetyl-CoA, (Pl)EctA: acetyl-CoA:DAB, or (Pl)EctA:acetyl-CoA:N-␥-ADABA complex. Instead, in our crystal structures, the nonreactive CoA is always present (as it was added to the crystallization solutions). To visualize a possible tertiary complex in which the actual co-substrate of EctA L-2,4-diaminobutyrate acetyltransferase, acetyl-CoA, is captured ((Pl)EctA:acetyl-CoA:DAB), we substituted in silico the thiol hydrogen (-SH) of CoA with an acetyl group (-C(O)CH 3 ) (Fig. 8G). This in silico model visualizes how close the reactive groups of acetyl-CoA and DAB are juxtapositioned just before the transfer of the acetyl group to DAB occurs (Video S1). Release of the (Pl)EctA reaction products CoA and N-␥-ADABA from the active site would then restore the apo-form of the EctA enzyme (Fig. 8A).

Conclusions
The five crystal structures of the (Pl)EctA that we present here allow us to trace and visualize the steps of the L-2,4-diaminobutyrate acetyltransferase before and after enzyme catal-ysis ( Fig. 8 and Video S1). Both monomers of the (Pl)EctA dimer are crucial for jointly building the complete architecture of the two active sites of the dimeric enzyme (Fig. 4B). Bona fide L-2,4-diaminobutyrate acetyltransferases are closely related proteins, as evidenced by the considerable degree of amino acid sequence identity (Fig. S3). Using the (Pl)EctA protein as the search query, it ranges between 94% for Paenibacillus glutanolyticus DSM5162 and 25% for Oceanobacillus iheyensis HTE831 among 432 inspected EctA proteins retrieved from a data set of a recent comprehensive phylogenomic analysis of ectABC gene clusters (41). Despite the fact that these 432 EctA proteins originate from 10 major bacterial and two archaeal phyla (41), the amino acid residues involved in the binding of the substrate DAB and the reaction product N-␥-ADABA are highly conserved (Fig. S3). Furthermore, the architecture of the acetyl-CoA-binding site of (Pl)EctA corresponds to an evolutionarily well-conserved fold found in microbial and eukaryotic acetyltransferases (75,84). Collectively, these data suggest that the crystal structures of the EctA enzyme that we present here from P. lautus (77) can serve as the blueprint for a structural
From the four enzymes required for ectoine production (Ect-ABC) and 5-hydroxyectoine (EctD) biosynthesis (26,34,40,41,45), the crystal structure of EctA now complements those of the already reported structures of ectoine synthase EctC (25) and ectoine hydroxylase EctD (71). Furthermore, an in silico model of EctB has also been recently established and probed by sitedirected mutagenesis (69). The seminal discovery of ectoine by Galinski et al. (23) in the extreme halophile Ectothiorhodospira halochloris and of 5-hydroxyectoine by Inbar and Lapidot (24) in Streptomyces parvulus occurred over 30 years ago. Now, a structure-based view of the entire biosynthetic route of these remarkable stress protectants is finally at hand ( Fig. 9 and Video S1). Collectively, this should aid new structure-guided attempts to improve the catalytic efficiency or stability of individual enzymes of the ectoine/5-hydroxyectoine biosynthetic route to increase industrial-scale biotechnological production of these commercially valuable chemical chaperones.

Chemicals
Ectoine was a kind gift from the bitop AG (Witten, Germany). Anhydrotetracycline hydrochloride (AHT), desthiobiotin, and Strep-Tactin Superflow chromatography material for the purification of proteins fused to a Strep-tag II affinity peptide were purchased from IBA GmbH (Göttingen, Germany). The reaction product of EctA, N-␥-ADABA, was synthesized through alkaline hydrolysis of ectoine (82). It was purified from the by-product N-␣-ADABA by repeated chromatography on a silica gel column (Merck silica gel 60) using a gradient of ethanol/25% ammonia/water (50:1:2-10:1:2) as eluent (70). The identity and purity of N-␥-ADABA was monitored by TLC and NMR spectroscopy ( 1 H NMR and 13 C NMR) on a Bruker  Fig. 8). F, overlay of the (Pl)EctA:CoA crystal structure with that of the ligand DAB found in the (Pl)EctA:DAB crystal structure, thereby visualizing the distance (2.8 Å) between the sulfur atom from CoA and the ␥-nitrogen from DAB. G, model of (Pl)EctA where an acetyl group is added in silico to the CoA sulfur, thereby mimicking the positions of the actual co-substrates, acetyl-CoA and DAB, in the active site of the L-2,4-diaminobutyrate acetyltransferase.

Acetyl-CoA synthesis and purification
Acetyl-CoA was synthesized from acetic anhydride (85) using a slightly modified protocol. CoA (320 mg) was dissolved in 0.5 M (8 ml) sodium bicarbonate-HCl buffer (pH 7.4). The solution was cooled down to 4°C, and acetic anhydride (80 l) was added dropwise under stirring. The reaction mixture was stirred for 30 min, and the completion of the reaction was confirmed by use of dithionitrobenzoic acid (DTNB) detecting the remaining free thiol groups. The reaction mixture was then acidified with formic acid until a pH of 3.0 was reached. The solution was subsequently degassed and directly applied to preparative HPLC for purification, using a preparative 1260 Infinity system (Agilent Technologies, Walsbronn, Germany). Acetyl-CoA, CoA, and other contaminants were separated using a 100 ϫ 21-mm Gemini 10 M NX-C18 110-Å column (Phenomenex, Aschaffenburg, Germany) and a mobile phase system comprised of 25 mM ammonium formate (pH 4.2) and methanol. Separation was achieved using a gradient of 5-22% of methanol over 7.5 min at a flow rate of 25 ml min Ϫ1 . Acetyl-CoA was detected using a 1260 infinity diode array detector (at 260 nm) and a 6130 Quadrupole MS system (Agilent Technologies, Walsbronn, Germany). Fractions containing acetyl-CoA were pooled and lyophilized for 48 h. The dry powder of acetyl-CoA was freshly dissolved in water before use in enzymatic assays with the purified (Pl)EctA enzyme. The concentration of acetyl-CoA stock solutions was calculated from a molar extinction coefficient (⑀ 260 nm ) for saturated acyl-CoA thioesters of 16,400 M Ϫ1 cm Ϫ1 (86).

Bacterial strains, media, and growth conditions
The E. coli strain TOP10 (Invitrogen) was used for the propagation of plasmids carrying ectA genes. Cultures of the plasmid-carrying E. coli strain were grown at 37°C in Luria-Bertani liquid medium (87) containing ampicillin (100 g ml Ϫ1 ). Heterologous overproduction of plasmid-encoded P. lautus EctA proteins ((Pl)EctA) carrying a Strep-tag II affinity peptide either at the N or C terminus was carried out in the E. coli B strain BL21 in modified minimal medium A (87) containing 0.5% (w/v) glucose as the carbon source and 0.5% (w/v) casamino acids, 1 mM MgSO 4 , and 3 mM thiamine as supplements.

Crystal structure of L-2,4-diaminobutyrate acetyltransferase Recombinant DNA procedures and construction of plasmids
The DNA sequence of the ectA gene was retrieved from the genome sequence of P. lautus strain Y412MC10 (accession number NC_013406.1) (77) and was used as a template for the synthesis of a codon-optimized version of the gene for its heterologous expression in E. coli. Synthesis of the (Pl)ectA gene was conducted by Invitrogen GeneArt (Thermo Fisher Scientific), and its DNA sequence was deposited in the GenBank TM database under accession number MF327591.1. To allow the overproduction and affinity purification of the recombinant (Pl)EctA protein in E. coli, we genetically constructed C-and N-terminal fusions of the ectA coding region to DNA segments encoding a Strep-tag II affinity peptide. For this purpose, the ectA gene was amplified from the plasmid (pLC46) provided by the supplier of the synthetic (Pl)ectA gene constructs using custom-synthesized primers (Table S2). The resulting PCR fragment was inserted into a pENTRY vector (IBA Göttingen, Germany), resulting in plasmid pLC48. By applying Stargate combinatorial cloning technology, the ectA-coding region was then inserted into the expression plasmids pASG-IBA3 and pASG-IBA5 (IBA, Göttingen, Germany), respectively; the resulting EctA overproduction plasmids were pLC50 (EctA with N-terminal Strep-tag II; NH 2 -WSHPQFEK-SG) and pLC51 (EctA with C-terminal Strep-tag II; SA-WSHPQFEK-COOH). In these plasmids, expression of the recombinant (Pl)ectA gene is mediated by the tet promoter, whose transcriptional activity is regulated through TetR, an AHT-responsive repressor protein (IBA GmbH, Göttingen, Germany).

Overproduction, purification, and determination of the quaternary assembly of EctA proteins
Cells of the E. coli B strain BL21 harboring an (Pl)ectA expression plasmid (either pLC50 or pLC51) were inoculated into modified minimal medium A containing 100 g ml Ϫ1 ampicillin (1 liter of medium in a 2-liter Erlenmeyer flask) to an OD 578 of 0.1 from an overnight preculture prepared in Luria-Bertani medium. The cells were grown on an aerial shaker (set to 180 rpm) at 37°C until the cultures reached an OD 578 of 0.5. At this time point, the synthetic inducer AHT of the TetR repressor was added to a final concentration of 0.2 mg ml Ϫ1 to trigger enhanced transcriptional activity of the tet promoter and thereby boost the expression of the plasmid-encoded (Pl)ectA gene. After 2 h of further growth of the culture, the E. coli B strain BL21 cells were harvested by centrifugation (2,360 ϫ g, at 4°C for 15 min), resuspended in 10 ml of purification buffer (100 mM Tris-HCl (pH 7.5) and 150 mM NaCl), and disrupted by passing them through a French pressure cell (16,000 p.s.i.). A cleared cell lysate of the disrupted cell was prepared by centrifugation (31,870 ϫ g, at 4°C for 45 min). The cleared cell extracts of the (Pl)EctA-overproducing cultures were used to purify the recombinant Strep-tag II-marked proteins by affinity chromatography on Strep-Tactin affinity resin as detailed previously (88,89). The concentration of the (Pl)EctA protein in the individual fractions eluted from the Strep-Tactin Superflow affinity column was measured with a Nanodrop Photospectrometer ND1000 (Peqlab, Erlangen, Germany) (25,440 M Ϫ1 cm Ϫ1 ). The purity and apparent molecular mass of the (Pl)EctA protein was assessed by SDS-PAGE (15% polyacrylamide); the PageRuler prestained protein ladder (Thermo Fisher Scientific) was used as a reference to monitor the electrophoretic mobility of the (Pl)EctA protein. Purified (Pl)EctA protein preparations were concentrated to ϳ10 mg ml Ϫ1 with Vivaspin 6 columns (Sartorius Stedim Biotech, Göttingen, Germany) with a 10-kDa molecular mass cutoff value prior to crystallization trials.
The molecular mass of (Pl)EctA proteins carrying a Strep-tag II affinity protein attached either to its N or C terminus was determined by MS analysis. 1-10 l of a 25 mM protein solution (in purification buffer), was prepared by desalting the protein solution with a MassPrep column (Waters, Milford, MA) in a Waters ACQUITY H-Class HPLC system. Protein elution into the electrospray ionization source of a Synapt G2Si mass spectrometer (Waters) was performed at 60°C with a flow rate of 0.1 ml min Ϫ1 using isocratic elution with 5% A (water/0.05% formic acid) for 2 min, followed by a linear gradient to 95% B (acetonitrile/0.045% formic acid) within 8 min and a final holding of 95% B for 4 min. The range of detected positive ions was 500 -5000 m/z. For automatic drift correction, Glu-Fibrinopeptide B was measured every 45 s. Deconvolution of averaged spectra was performed after baseline subtraction and smoothing using MassLynx instrument software with MaxEnt1 extension.
To analyze the quaternary assembly of the (Pl)EctA protein, we used SEC-MALS. For these experiments, an Agilent Technologies system connected to a triple-angle light scattering detector (miniDAWN TREOS, Wyatt Technology Europe GmbH, Dernbach, Germany) followed by a differential refractive index detection system (Wyatt Technology) was used. Typically, 200 l of purified (Pl)EctA protein (2 mg ml Ϫ1 ) was loaded onto the Bio SEC-5 HPLC column, and the obtained data were analyzed with the ASTRA software package (Wyatt Technology).

EctA enzyme activity assays
To determine the precise reaction product of the (Pl)EctA enzyme (in other words whether N-␥-ADABA, or N-␣-ADABA (or both) are synthesized), we carried out enzyme assays in 20 l of 100 mM TES-HCl buffer (pH 7.5) containing 2 mM acetyl-CoA, 2 mM DAB, and 1 g of purified enzyme at a temperature of 30°C. The reaction was stopped after 5 min by the addition Crystal structure of L-2,4-diaminobutyrate acetyltransferase of 20 l of acetonitrile. The enzyme reaction product(s) were then derivatized with fluorenylmethyloxycarbonyl chloride (FMOC-Cl) using a procedure based on a previously published method (82). In brief, 2 l of the (Pl)EctA enzyme reaction sample was mixed with 3 l of an FMOC solution (25 mg ml Ϫ1 FMOC in acetonitrile) in a thermomixer (1 min, 900 rpm, at 20°C). Subsequently, the FMOC reagent that had not chemically reacted was quenched by adding 6 l of 1-aminoadamantane solution (7.6 mg ml Ϫ1 1-aminoadamantane and 50% acetone in 0.5 M sodium borate buffer, pH 7.7) and mixed (1 min, 900 rpm, 20°C). The entire solution was then diluted with 988 l of H 2 O and centrifuged (20,800 ϫ g, at room temperature for 10 min) to remove the denatured (Pl)EctA enzyme. 37.5 l of the supernatant was injected into an HPLC system (1260 Infinity, Agilent Technologies, Walsbronn, Germany) equipped with a 150 ϫ 4.6-mm Gemini 5 M C18 110-Å column (Phenomenex, Aschaffenburg, Germany) and a fluorescence-detecting module (Agilent Technologies, Walsbronn, Germany). The fluorescence-detecting module was set to an excitation wavelength of 266 nm and an emission wavelength of 305 nm. The mobile phase consisted of solvent A (20% acetonitrile and 0.5% tetrahydrofuran in 50 mM sodium acetate buffer (pH 4.2)) and solvent B (80% acetonitrile in sodium acetate buffer (pH 4.2)). To allow the separation of the FMOCmodified N-␥-ADABA or N-␣-ADABA isomers, the flow rate was set to 1 ml min Ϫ1 at 40°C using a gradient of solvent A and B similar to the procedure described by Kunte et al. (82).
To determine the basic parameters of the (Pl)EctA enzyme, activity was measured at 30°C using a continuous spectrophotometric assay. In this assay, formation of free CoA was followed using DTNB. An extinction coefficient (⑀ 412 nm ) for DTNB of 14,000 M Ϫ1 cm Ϫ1 was used to determine the amount of CoA released during the enzyme reaction. The reaction mixture (300 l) contained 100 mM TES-HCl buffer (pH 7.5), 1 mM DTNB, 1.2 mM acetyl-CoA, 5 mM DAB, and 0.1 g of purified (Pl)EctA enzyme. The enzyme reaction was started by the addition of DAB to the reaction vessel and was run for 1.5 min, in which the absorption was determined. All (Pl)EctA assays were performed using two independently produced and purified protein preparations, and each protein sample was assayed twice. During the screening for the temperature profile of the (Pl)EctA enzyme, the reaction was started by the addition of the enzyme and was run, for high temperatures (45-60°C), only for 0.3 min, due to reduced enzyme stability. The screening for pH optimum was performed in a 50-l reaction volume containing a buffer mixture (MES (pH 5.5), PIPES (pH 6.5), TES (pH 7.5), CHES (pH 7.9), HEPES (pH 8.5), and CAPS (pH 10.0)) with 50 mM each, 1.2 mM acetyl-CoA, 1.7 mM DAB, and 0.1 g of purified (Pl)EctA protein. The pH values of these buffer solutions and the resulting mixtures were adjusted with 37% HCl or 5 M NaOH at 30°C. The enzyme reaction of the (Pl)EctA protein was started by the addition of acetyl-CoA, was run for 100 s, and was then stopped by the addition of 50 l of 80% acetonitrile. To remove denatured (Pl)EctA protein, the samples were centrifuged (20,800 ϫ g, at 4°C for 5 min). For the reconstitution of a neutral pH value for the DTNB reaction, 50 l of the supernatant was added to 150 l of DTNB solution (0.2 M TES (pH 7.5), 2 mM DTNB), and the DTNB absorption was measured using a Tecan plate reader (Tecan Group Ltd., Männedorf, Switzerland) at 30°C.
The kinetic parameters of the (Pl)EctA enzyme were determined using the continuous spectrophotometric assay described above either with 5 mM DAB and varied concentrations of acetyl-CoA (0.05-8 mM) or with 4 mM acetyl-CoA and varied concentrations of DAB (0.05-1.6 mM). The enzyme activity of the various (Pl)EctA mutants was monitored with the same continuous assay in a reaction containing 100 mM TES-HCl buffer (pH 7.5), 1 mM DTNB, 2 mM acetyl-CoA, 5 mM DAB, and 0.1 g of purified (Pl)EctA enzyme. The enzyme activities of the (Pl)EctA variants were benchmarked against the WT protein whose activity was set to 100%. Under these conditions, the WT (Pl)EctA enzyme had an activity of 29.08 Ϯ 4.08 units mg Ϫ1 protein. One unit is defined as the enzymatic conversion of 1 M acetyl-CoA to 1 M free CoA min Ϫ1 correlating with the same amount of DAB converted by the (Pl)EctA enzyme.

In silico analysis of EctA-type proteins
In a recent phylogenomic analysis of the distribution of ect biosynthetic gene clusters present in Bacteria and Archaea, a curated nonredundant data set comprising ectoine biosynthetic genes from 437 microbial species/strains was generated (41). We relied on this data set to retrieve EctA-type proteins and compared their amino acid sequences with the MAFFT multiple-amino acid sequence alignment tool (http://mafft. cbrc.jp/alignment/server/) 5 (90) using the (Pl)EctA protein sequence (accession number AWH98098) as the template for a BLAST search (91).

Crystallization of the (Pl)EctA protein
Several crystals were found for the apo-(Pl)EctA protein and the ligand-bound forms using commercial screens (Nextal, Qiagen, Hilden, Germany; Molecular Dimensions, Suffolk, UK) in 96-well sitting-drop plates (MRC3, Swissci) at 12°C. Both the C-terminal and N-terminal Strep-tag II-marked forms of the (Pl)EctA protein were used in these crystallization trials. Crystals of the apo-(Pl)EctA protein were obtained using commercial screens and by slightly optimizing the composition of the crystallization solution. 0.1 l of (Pl)EctA protein solution (10 -15 mg of protein ml Ϫ1 ) and 0.1 l of reservoir solution were mixed and equilibrated against 40 l of reservoir solution. For the apo-form of (Pl)EctA, the first crystal appeared after 12 h. The most promising condition was found with a solution containing 0.2 M lithium sulfate, 0.2 M sodium acetate, 0.1 M HEPES (pH 7.5), and 25% (w/v) PEG 4000 from the Nextal PEG II suite (Qiagen, Hilden, Germany) after 8 days. The bestdiffracting crystals were grown in a solution consisting of 0. Apart from the apo-form, different ligand-bound complex crystals were obtained by adding either 5 mM CoA (Sigma-

Crystal structure of L-2,4-diaminobutyrate acetyltransferase
Aldrich), 20 mM DAB (Sigma-Aldrich), 20 mM N-␥-ADABA (25, 82), or a combination of CoA and DAB. The different substrates were preincubated with the protein for at least 30 min on ice. If two substrates were used, the first one was incubated for 5 min before the second one was added. For cryoprotection, all crystal-containing drops were overlaid with mineral oil before the crystals were harvested and flash-frozen in liquid nitrogen.

Data collection, processing, and structure determination
For the crystallographic analysis of apo-(Pl)EctA, crystals of the ligand-free form of (Pl)EctA diffracted to a maximum of 2.2 Å. The data set was collected at ID30B (ESRF, Grenoble, France) at 100 K, processed with XDS (92,93), and phased using the automated AUTORICKSHAW pipeline (http:// www.embl-hamburg.de/Auto-Rickshaw/) 5 with only the (Pl)EctA protein sequence as input. The resulting initial model was subsequently autobuilt using the ARPWARP webservice (https://arpwarp.embl-hamburg.de). 5 After several rounds of model building using COOT (94) and subsequent refinement using refmac5 (95) from the CCP4 suite (96), the structure of the full-length apo(Pl)EctA protein was modeled into the electron density.
The following procedures were used for the crystallographic analysis of the various forms of the (Pl)EctA protein. For the crystallographic analysis of (Pl)EctA:CoA: a high-resolution data set up to 1.5 Å was collected at ID29 (ESRF) at 100 K, processed with XDS, and phased via molecular replacement using the apo-(Pl)EctA structure as a search model. For the crystallographic analysis of (Pl)EctA:DAB, a high-resolution data set up to 1.5 Å was collected at ID30A-3 (ESRF) at 100 K, processed with XDS, and phased via molecular replacement using the (Pl)EctA:CoA structure as a search model. For the crystallographic analysis of (Pl)EctA:CoA:DAB, a high-resolution data set up to 1.2 Å was collected at ID29 (ESRF) at 100 K, processed with XDS, and phased via molecular replacement using the (Pl)EctA:CoA structure as a search model. For the crystallographic analysis of (Pl)EctA:N-␥-ADABA, a data set up to 2.2 Å was collected at ID23-1 (ESRF) at 100 K, processed with XDS, and phased via molecular replacement using the (Pl)EctA: CoA structure as a search model. Refinements of all complex structures were performed as described above.
Adding the substrate or the ortholog of acetyl-CoA, CoA, prior to the crystallization improved the crystal quality significantly, as reflected by the higher resolution of the obtained data sets. The (Pl)EctA:CoA crystal structure was solved at 1.5 Å (R work and R free values were 14.7 and 18.3%, respectively), the (Pl)EctA:DAB crystal structure at 1.5 Å (R work and R free values were 17.8 and 12.2%, respectively), the (Pl)EctA:CoA:DAB crystal structure at 1.2 Å (R work and R free values were 13.3 and 15.0%, respectively), and finally, the (Pl)EctA:ADABA crystal structure at 2.2 Å (R work and R free values were 16.7 and 20.1%, respectively). A summary of the data collection statistics, refinement details, and model content of these different (Pl)EctA crystal structures is given in Table 1. The crystal parameters, especially the unit cell dimensions and space group, differ between the crystals of apo-(Pl)EctA and the crystal with the substrates (Table 1), which implies that a different number of (Pl)EctA proteins are present in the asymmetric unit. In the asymmetric unit of apo-(Pl)EctA and (Pl)EctA: ADABA, three copies of EctA were found, whereas in (Pl)EctA: CoA and (Pl)EctA:DAB, only one monomer of the EctA protein was present, and in the crystals of (Pl)EctA:CoA:DAB and (Pl)EctA:ADABA, two EctA monomers were found.

Figure preparation of crystal structures
Figures of the crystal structures of the (Pl)EctA protein were prepared using the PyMOL software suite (97) and Chimera (98).