Structure and function of an N-acetyltransferase from the human pathogen Acinetobacter baumannii isolate BAL_212.

Acinetobacter baumannii is a Gram-negative bacterium commonly found in soil and water that can cause human infections of the blood, lungs, and urinary tract. Of particular concern is its prevalence in health-care settings where it can survive on surfaces and shared equipment for extended periods of time. The capsular polysaccharide surrounding the organism is known to be the major contributor to virulence. The structure of the K57 capsular polysaccharide produced by A. baumannii isolate BAL_212 from Vietnam was recently shown to contain the rare sugar 4-acetamido-4,6-dideoxy-d-glucose. Three enzymes are required for its biosynthesis, one of which is encoded by the gene H6W49_RS17300 and referred to as VioB, a putative N-acetyltransferase. Here, we describe a combined structural and functional analysis of VioB. Kinetic analyses show that the enzyme does, indeed, function on dTDP-4-amino-4,6-dideoxy-d-glucose with a catalytic efficiency of 3.9 x 104  M-1  s-1 (±6000), albeit at a reduced value compared to similar enzymes. Three high-resolution X-ray structures of various enzyme/ligand complexes were determined to resolutions of 1.65 Å or better. One of these models represents an intermediate analogue of the tetrahedral transition state. Differences between the VioB structure and those determined for the N-acetyltransferases from Campylobacter jejuni (PglD), Caulobacter crescentus (PerB), and Psychrobacter cryohalolentis (Pcryo_0637) are highlighted. Taken together, this investigation sheds new insight into the Type I sugar N-acetyltransferases.

Coenzyme A

thymidine diphosphate

N‐2‐hydroxyethylpiperazine‐N′‐2‐ethanesulfonic acid

2‐(N‐morpholino)ethanesulfonic acid

nickel‐nitrilotriacetic acid

Tobacco Etch Virus

tris‐(hydroxymethyl)aminomethane

INTRODUCTION

Acinetobacter baumannii is an opportunistic Gram‐negative bacterium that has recently emerged as a significant human health threat worldwide. , Due to the increasing rise of A. baumannii infections in hospital settings, and its extensive antibiotic resistance spectrum, it has been designated as a “red alert.” Indeed, in 2017, the World Health Organization published a global priority list for the development of new antibiotics and placed A. baumannii into the “Critical” category. It belongs to the so‐called “ESKAPE” group of pathogens, which include Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumonia, A. baumannii, Pseudomonas aeruginosa, and Enterobacter species. Members of this group are becoming increasingly resistant to the bactericidal effects of antibiotics. Of particular concern is the ability of A. baumannii to survive for extended periods of time on hospital surfaces and equipment. , Strikingly, some clinical isolates of A. baumannii have been shown to remain viable for ~100 days under dry conditions. Whereas a variety of glycans are thought to contribute to A. baumannii viability including the lipooligosaccharide, the capsular polysaccharide, and the exopolysaccharide poly‐β‐(1–6)‐N‐acetylglucosamine, it is thought that the capsular polysaccharide is the major contributor to virulence. Indeed, in one study, it was demonstrated that isolates lacking the capsular polysaccharide did not cause infections.

The structure of the K57 capsular polysaccharide produced by A. baumannii isolate BAL_212 from Vietnam was recently shown to contain the rare sugar 4‐acetamido‐4,6‐dideoxy‐d‐glucose (also referred to as N‐acetylviosamine or d‐Qui4NAc). The genes required for its biosynthesis (Scheme 1) were identified as rmlA (H6W49_RS17310), rmlB (H6W49_RS17315), vioA (H6W49_RS17305), and vioB (H6W49_RS17300), which encode for a thymidylyltransferase, a 4,6‐dehydratase, a pyridoxal 5′‐phosphate‐dependent aminotransferase, and an N‐acetyltransferase, respectively. In this report, we describe a combined structural and functional investigation of the N‐acetyltransferase encoded by the H6W49_RS17300 gene in A. baumannii isolate BAL_212 and hereafter referred to as VioB. Surprisingly, one of the three structures determined in this investigation revealed the presence of a tetrahedral transition state analogue trapped in the crystalline lattice. We also provide a comparison between the model of VioB with those structures previously reported for the N‐acetyltransferases from Caulobacter crescentus and Psychrobacter cryohalolentis, which also contained either a trapped transition state analogue or intermediate, respectively. ,

SCHEME 1

Predicted pathway for the biosynthesis of dTDP‐d‐Qui4NAc

MATERIALS AND METHODS

Protein expression and purification

The gene encoding VioB from A. baumannii (strain BAL_212 KL57) was synthesized by Integrated DNA Technologies. It was cloned into pET28t3g, a modified pET28b vector (Novagen), which yields a protein with an N‐terminal polyhistidine tag as previously described. It was also cloned into pET31b(+) to produce a protein with a non‐cleavable C‐terminal polyhistidine tag.

The constructs were utilized to transform Rosetta2(DE3) Escherichia coli cells (Novagen). Cultures were grown in lysogeny broth supplemented with kanamycin and chloramphenicol (both at 50 mg/L concentration) for pET28t3g‐viob or ampicillin and chloramphenicol (100 mg/L and 50 mg/L concentration, respectively) for pET31b(+)‐viob at 37°C with shaking until an optical density of 0.8 was reached at 600 nm. The flasks were cooled in an ice bath, and the cells were induced with 1 mM isopropyl β‐d‐1‐thiogalactopyranoside and allowed to express protein at 21°C for 24 h.

The cells were harvested by centrifugation and frozen as pellets in liquid nitrogen. The pellets from the cell expressing the N‐terminally tagged protein were subsequently disrupted by sonication on ice in a lysis buffer composed of 50 mM sodium phosphate, 20 mM imidazole, 10% glycerol, and 300 mM sodium chloride (pH 8.0). The lysate was cleared by centrifugation, and the enzyme was purified at 4°C utilizing Prometheus™ Ni‐NTA agarose (Prometheus Protein Biology Products) according to the manufacturer’s instructions. All buffers were adjusted to pH 8.0 and contained 50 mM sodium phosphate, 300 mM sodium chloride, and imidazole concentrations of 20 mM for the wash buffer and 300 mM for the elution buffer. Following Ni‐NTA purification, the pooled protein samples were split in half. One half was dialyzed against 10 mM Tris–HCl (pH 8.0) and 200 mM NaCl. For the other half, the polyhistidine tag was removed by digestion with TEV protease. The TEV protease and remaining tagged proteins were removed by passage over Ni‐NTA agarose, and the tag‐free protein was dialyzed against 10 mM Tris–HCl (pH 8.0) and 200 mM NaCl. The C‐terminally tagged protein was purified in the same manner as the N‐terminally tagged enzyme. All constructs were concentrated to approximately 25 mg/mL based on an extinction coefficient of 3.1 (mg/ml)−1 cm−1.

Crystallizations

Crystallization conditions were surveyed by the hanging drop method of vapor diffusion using a sparse matrix screen developed in the Holden laboratory. Both the His‐tagged and tag‐free enzymes were tested for crystallization properties. The C‐terminally tagged enzyme yielded X‐ray diffraction quality crystals at room temperature using a precipitant containing 8%–12% poly(ethylene glycol) 8000, 200 mM LiCl, 5 mM dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose, 5 mM CoA, and 100 mM MES (pH 6.0). All the crystals utilized in this investigation were initially grown under these conditions that included both the dTDP‐sugar and CoA. X‐ray data collection from 2‐day old crystals and subsequent structural analyses revealed that the dTDP‐sugar and CoA ligands were bound with low occupancy. Thus, in order to increase the occupancy of these ligands, crystals that were only 2 days old were then soaked in a ligand‐free synthetic mother containing 15% poly(ethylene glycol) 8000, 200 mM LiCl, 200 mM NaCl, and 100 mM MES (pH 6.0). They were then transferred to solutions containing either 20 mM dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose, or 5 mM dTDP and 5 mM acetyl‐CoA.

For X‐ray data collection, the crystals were transferred to a cryo‐protectant solution composed of either 24% poly(ethylene glycol) 8000, 250 mM LiCl, 250 mM NaCl, 20% ethylene glycol, 5 mM dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose, and 100 mM MES (pH 6.0) or 24% poly(ethylene glycol) 8000, 250 mM LiCl, 250 mM NaCl, 20% ethylene glycol, 5 mM dTDP, 5 mM acetyl‐CoA, and 100 mM MES (pH 6.0).

The transition‐state analogue model was obtained using crystals grown from 8%–12% poly(ethylene glycol) 8000, 200 mM LiCl, 5 mM dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose, 5 mM CoA, and 100 mM MES (pH 6.0), which were allowed to equilibrate for at least 2 weeks.

X‐ray data collection and processing

X‐ray data sets were collected at 100 K utilizing a BRUKER D8‐VENTURE sealed tube system equipped with Helios optics and a PHOTON II detector. These X‐ray data sets were processed with SAINT and scaled with SADABS (Bruker AXS). Relevant X‐ray data collection statistics are provided in Table 1.

TABLE 1

X‐ray data collection and model refinement statistics

	dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose	Acetyl‐CoA/dTDP	Intermediate analogue
Space group	R3	R3	R3
Unit cell (a, b, c, [Å])	97.0, 97.0,73.3	97.4, 97.4, 72.8	97.9, 97.9, 73.0
Resolution limits (Å)	50.0–1.45 (1.55–1.45) a	50.0–1.65 (1.75–1.65) a	50.0–1.25 (1.35–1.25) a
Number of independent reflections	45094 (8089)	30576 (4811)	70527 (14096)
Completeness (%)	98.7 (96.8)	98.6 (96.3)	97.7 (94.5)
Redundancy	4.2 (2.9)	4.8 (3.2)	4.0 (2.6)
avg I/avg σ(I)	10.4 (3.0)	12.0 (3.0)	12.2 (3.0)
R sym (%) b	6.9 (35.3)	6.1 (35.4)	6.1 (30.2)
R‐factor (overall)%/no. reflections c	17.7/45094	17.4/30576	15.2/70527
R‐factor (working)%/no. reflections	17.6/42854	17.2/28942	15.1/66908
R‐factor (free)%/no. reflections	19.4/2240	20.4/1634	16.5/3619
Number of protein atoms	1591	1584	1618
Number of heteroatoms	248	260	373
Average B values
Protein atoms (Å2)	12.4	15.6	11.7
Ligand (Å2)	11.2	24.2	12.4
Solvent (Å2)	23.7	26.6	26.9
Weighted RMS deviations from ideality
Bond lengths (Å)	0.010	0.008	0.010
Bond angles (°)	1.65	1.59	1.82
Planar groups (Å)	0.008	0.008	0.008
Ramachandran regions (%) d
Most favored	99.0	99.0	99.0
Additionally allowed	1.0	1.0	1.0
Generously allowed	0.0	0.0	0.0

Statistics for the highest resolution bin.

R sym = (∑|I − |/∑I) × 100.

R‐factor = (∑|Fo − F c|/∑|F o|) × 100 where F o is the observed structure‐factor amplitude and F c is the calculated structure‐factor amplitude.

Distribution of Ramachandran angles according to PROCHECK.

Structure solution and refinement

A partial model of the enzyme/intermediate analogue structure was initially built with MrBUMP using the N‐acetyltransferase domain of PglB from Neisseria gonorrhoeae (PDB entry 4 M98) as the starting model. Iterative cycles of model‐building with COOT , and refinement with REFMAC led to a final X‐ray model with an overall R‐factor of 15.2%. The other structures were solved via difference Fourier techniques. Refinement statistics are presented in Table 1.

Determination of kinetic parameters

The enzymatic activity of VioB was monitored spectrophotometrically by following the increase in absorbance at 412 nm due to the reaction of the sulfhydryl group of the CoA product with 5,5′‐dithiobis‐(2‐nitrobenzoic acid). This results in a disulfide interchange leading to the formation of 5‐thio‐2‐nitrobenzoic acid which possesses a characteristic absorbance at 412 nm with an extinction coefficient of 14 150 M−1 cm−1. Reactions were monitored continuously with a Beckman Coulter DU‐640B spectrophotometer, and enzyme activities were calculated from the initial rates. Assay reaction mixtures were 100 μl in volume and contained, in addition to enzyme and substrates, 100 mM HEPES (pH 7.5) and 5 mM 5,5′‐dithiobis‐(2‐nitrobenzoic acid).

To determine the steady‐state kinetic parameters for dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose, acetyl‐CoA was used at a concentration of 2.0 mM and the enzyme concentration of the tag‐free version was 0.00055 mg/ml. The dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose concentration was varied from 0.025 mM to 6.0 mM (K M = 0.29 ± 0.03 mM, k cat/K M = 39 000 ± 6000 M−1 s−1). These data were fitted to the equation: v 0 = (V max[S])/(K M + [S]) using Prism9 (GraphPad). The k cat values per active site were calculated according to the equation: k cat = V max/[ET].

Production of dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose

The ligand was synthesized and purified as previously reported from this laboratory.

RESULTS AND DISCUSSION

Kinetic properties of VioB

The kinetic properties of VioB were determined via a continuous spectrophotometric assay. A plot of the initial velocity versus dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose is presented in Figure 1. VioB displays a K M = 0.29 ± 0.03 mM and a k cat/K M of 3.9 x 104 M−1 s−1 (±6000) for dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose. The catalytic efficiency of VioB is significantly less than that previously reported for the N‐acetyltransferases from C. jejuni (using UDP‐2‐acetamido‐4‐amino‐2,4,6‐trideoxy‐d‐glucose, 2.0 × 107 M−1 s−1), C. crescentus (using GDP‐4‐amino‐4,6‐dideoxy‐d‐mannose, 3.5 × 106 M−1 s−1), or P. cryohalolentis (using UDP‐2‐acetamido‐4‐amino‐2,4,6‐trideoxy‐d‐glucose, 8.8 × 105 M−1 s−1). , ,

FIGURE 1

Plot of initial velocity versus substrate concentration. In presenting the data as we do, we are adhering to standard conventions in enzymology. Measuring velocities over a wide range of substrate concentrations allows us to obtain data that define both k cat and k cat/K M well, which is not accomplished by measuring replicates at fewer different concentrations. The graph shown allows for a qualitative appreciation of the quality of the data; the quantitative goodness‐of‐fit to the Michaelis–Menten equation is given by the standard errors as described in Materials and Methods. The line shown is the calculated best fit

Structure of VioB

The structure of VioB in complex with dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose was solved to 1.45 Å resolution and refined to an overall R‐factor of 17.7%. Previous investigations have shown that the nucleotide‐linked sugar N‐acetyltransferases function as trimers with C3 symmetry. , , , , , , , VioB packed in the crystalline lattice with its local three‐fold rotational axis coincident to a crystallographic threefold, thereby reducing the contents of the asymmetric unit to a single monomer. A ribbon representation of the trimer is displayed in Figure 2A. The observed electron density for the polypeptide chain backbone was continuous from Met 1 to Glu 209. Likewise, the electron density corresponding to the dTDP‐sugar was unambiguous as can be seen in Figure 2B.

FIGURE 2

The molecular architecture of VioB. Shown in (A) is a ribbon representation of the trimer with the bound dTDP‐sugar ligands depicted in space‐filling representations. The observed electron density for the bound ligand is presented in (B). The electron density map, shown in stereo, was calculated with (F o‐F c) coefficients and contoured at 3σ. The ligand was not included in the X‐ray coordinate file used to calculate the omit map, and thus, there is no model bias. A close‐up stereo view of the region surrounding the bound ligand is presented in (C). Ordered water molecules are displayed as red spheres. The dashed lines indicate possible hydrogen‐bonding interactions within 3.2 Å. Note that the active site of VioB is shared by two subunits. Those amino acid residue labels marked by an asterisk belong to Subunit B. This figure and Figures 3, 4, 5, 6 were prepared with PyMOL

FIGURE 3

Structure of VioB in complex with dTDP and acetyl‐CoA. The electron density map, shown in stereo in (A) was calculated with (F o‐F c) coefficients and contoured at 3σ. Those amino acid residues lying within 3.2 Å of the acetyl‐CoA cofactor are displayed in stereo in (B). Ordered water molecules are drawn as red spheres, and possible hydrogen‐bonding interactions between the ligand and the protein are indicated by the dashed lines

FIGURE 4

Structure of the VioB tetrahedral intermediate analogue. The observed electron density for the intermediate analogue is shown in stereo in (A). It was calculated with (F o‐F c) coefficients and contoured at 3σ. Presented in stereo in (B) is a close‐up view surrounding the tetrahedral intermediate model with potential hydrogen bonds to the oxyanion represented by the dashed lines

FIGURE 5

Comparison of VioB, PerB, Pcryo_0637, and PglD. Shown in (A) is a stereo representation of the ribbon drawings for VioB, PerB, Pcryo_0637, and PglD depicted in violet, wheat, cyan, and pink, respectively. An amino acid sequence alignment for these enzymes is provided in (B). Conserved residues are highlighted in red

FIGURE 6

Comparison of the tetrahedral intermediates in Pcryo_0637, VioB, and PerB. The regions surrounding the tetrahedral intermediates observed in Pcryo_0637, VioB, and PerB are shown in stereo in (A), (B), and (C), respectively

Each subunit folds into two distinct motifs formed by Met 1 to Phe 87 and Ala 88 to Glu 209. The N‐terminal globular domain is dominated by a five‐stranded mixed β‐sheet flanked on either side by an α‐helix. The architecture of the C‐terminal domain consists of a β‐helix composed of seven turns. Pro 201, which adopts a cis conformation, is located in the last turn of the β‐helix (Gly 199 to Ala 202).

A close‐up stereo view of the binding pocket for dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose is provided in Figure 2C. The thymine ring is anchored into the active site by the backbone amide of Asn 33 (Subunit A) and two water molecules. The ribosyl group, which adopts the C2′‐endo pucker, participates in hydrogen bonding interactions with the side chains of Asp 32 and Lys 37 (Subunit A). The pyrophosphoryl group does not interact with side chains, but rather with water molecules and the backbone amides of Gly 11 and Gly 67 (Subunit A). His 135, contributed by Subunit B, lies within 2.7 Å of the C‐4′ amino group. In addition to this hydrogen bond, there are five ordered water molecules that lie within 3.2 Å of the pyranosyl moiety.

The next structure determined in this investigation was that of the enzyme complexed with dTDP and acetyl‐CoA. The model was refined to an overall R‐factor of 17.4% at 1.65 Å resolution. Again, there was one molecule in the asymmetric unit. Shown in Figure 3A are the electron densities corresponding to the bound ligands. The electron density for the dTDP ligand was weak, whereas that for the acetyl‐CoA was unambiguous. A stereo view of the acetyl‐CoA binding pocket, which is positioned between subunits, is provided in Figure 3B. The side chains of Lys 206 (Subunit A), Arg 170 (Subunit B), and Arg 189 (Subunit B) serve to neutralize the negative charges on the phosphoryl oxygens of CoA, and the side chain of Glu 171 (Subunit B) forms a hydrogen bond with the nitrogen in the aminoethanethiol unit. Additional interactions between the cofactor and the protein are provided by the backbone amide groups of Ala 165 and Met 183 in Subunit A and Gly 153 in Subunit B. Five ordered water molecules also surround the ligand.

The third model of VioB solved in this investigation utilized X‐ray data from crystals that had equilibrated under the original crystallization conditions (5 mM dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose and 5 mM CoA) for greater than 2 weeks. The model was refined to an overall R‐factor 15.2% at 1.25 Å resolution. Shown in Figure 4A is the electron density observed in the active site. Strikingly, a tetrahedral adduct was obtained with the crosslinking group being CHCH3. This phenomenon was first observed in our structural investigation of GDP‐perosamine N‐acetyltransferase from C. crescentus. In that case, the crystals were grown using poly(ethylene glycol) as the precipitant, which is notorious for being contaminated with peroxides and aldehydes such as acetaldehyde. It was proposed that the C‐4′ amino group of the sugar attacked the carbonyl carbon of the contaminating acetaldehyde to form a tetrahedral intermediate. This intermediate was thought to collapse to a Schiff base, which was subsequently attacked by the sulfhydryl group of CoA. Likewise, the crystals of VioB were grown in the presence of poly(ethylene glycol) 8000 suggesting that the same chemistry had occurred. Given that the crystals were several weeks old before they were subjected to X‐ray data collection, this most likely allowed for the development of the tetrahedral intermediate analogue. A close‐up view of the region surrounding the tetrahedral adduct is presented in Figure 4B. The position of His 135 suggests that it functions as the active site base to abstract a proton from the C‐4′ amino nitrogen. Most likely the backbone amide of Gly 153 provides stabilization of the oxyanion that forms during catalysis. Interestingly, in the model of VioB tetrahedral intermediate, Ser 152 adopts two distinct conformations, one of which would lie within hydrogen‐bonding distance to the developing oxyanion at the transition state.

Comparison of the VioB structure with other family members

One of the first molecular models of a monofunctional sugar N‐acetyltransferase to be reported was that of PglD from Campylobacter jejuni. This enzyme catalyzes the acetylation of UDP‐2‐acetamido‐4‐amino‐2,4,6‐trideoxy‐d‐glucose. The α‐carbons for VioB and PglD correspond with a root‐mean‐square deviation of 1.8 Å. Our subsequent investigation of GDP‐perosamine N‐acetyltransferase (PerB) from C. crescentus revealed the presence of a trapped transition state analogue in the active site as mentioned above. The α‐carbons for VioB and PerB superimpose with a root mean square deviation of 1.4 Å. And recently, the structure of Pcryo_0637 from P. cryohalolentis was reported from this laboratory. It utilizes the same substrate as PglD. Remarkably, in one of the Pcryo_0637 structures, a true tetrahedral intermediate rather than analogue was trapped in the active site. The α‐carbons for VioB and Pcryo_0637 correspond with a root mean square deviation of 2.3 Å.

Shown in Figure 5A is a superposition of the α‐carbons for VioB, PerB, Pcryo_0637, and PglD. All adopt a bilobal‐type structure with approximately the first 90 residues folding into a globular domain dominated by a five‐stranded mixed β‐sheet. This domain is followed by a classical left‐handed β‐helix motif. Whereas VioB, PerB, and PglD are approximately the same size, Pcryo_0637 has two insertions (Leu 25 to Ile 36 and Ile 63 to Asp 66) in the N‐terminal global domain and an insertion in the β‐helix motif (Thr 180 to Pro 184) as indicated in Figure 5A.

An amino acid sequence alignment of these proteins, all of which belong to the Class I sugar N‐acetyltransferases, is presented in Figure 5B. Of the 16 absolutely conserved residues among these N‐acetyltransferases, nine are glycines. Gly 8 and Gly 11 are likely conserved given their location in the active site cleft. Specifically, the α‐carbon of Gly 8 is only 3.8 Å away from the endocyclic oxygen of the dTDP ribose as shown in Figure 2C. This region simply cannot accommodate a larger side chain. This is equally true for Gly 11 with its α‐carbon located within 3.9 Å of β‐phosphoryl oxygen of the ligand (Figure 2C). Gly 106 is the third residue in Type II turn formed by Glu 104 and Thr 107. The peptidic nitrogen of this residue forms a hydrogen bond with the carbonyl oxygen of Ala 88 which resides in the random coil region connecting the N‐terminal globular domain to the left‐handed β‐helix motif. Note that the α‐carbon of Gly 106 is located at 3.9 Å of the β‐carbon of Phe 87. Any larger side chain would significantly perturb the tertiary structure of the enzyme. Gly 153 resides within 3.6 Å of His 135, the presumed catalytic base. A larger side chain would alter its position considerably in the active site cleft. Gly 182 is the first residue in a Type II turn located in the β‐helix. Its α‐carbon resides within 3.3 Å of the carbonyl oxygen of the pantothenic acid portion of CoA. The conserved Gly 184 is located in the third position of this Type II turn, and importantly, both the backbone peptidic nitrogen and the carbonyl group of Met 183 hydrogen bond with the carbonyl oxygen of the pantothenic acid portion and the NH2 group of the adenine ring of the CoA, respectively (Figure 3B). The integrity of this turn is most likely critical for proper cofactor binding. Finally, Gly 199 initiates another turn that also contains the conserved Pro 201. This is the reverse turn that in certain N‐acetyltransferases undergoes a conformational change upon CoA binding. It can be speculated that this conserved glycine is required for such flexibility. The reasons for the conservation of Gly 140 and Gly 164 are not apparent.

The two conserved charged residues, Arg 72 and Lys 74, are located near the active site. Specifically, the side chain of Lys 74 resides near the thymine ring of the bound ligand, whereas the guanidinium moiety of Arg 72 serves to bridge the carbonyl oxygens of Ile 66 and Gly 67 which, in turn, abut the β‐phosphoryl group of the ligand (Figure 2C). His 135 is conserved because it functions as the catalytic base as discussed below. The conservation of Ala 65 and Asn 128 is not immediately obvious.

Previously, it was proposed that a carboxylic acid‐containing residue following the histidine, either an aspartate or glutamate, either serves to increase the pK a of the general base or to play a role in its proper positioning. , , In the case of VioB, the residue following His 135 is glycine, thereby shedding doubt as to the need for a catalytic His/Glu(Asp) dyad for activity among the Type I N‐acetyltransferases. Of interest is the conserved proline at the C‐terminus (Pro 201 in VioB). From previous investigations, it has been shown that there is a trans to cis conformational change of this conserved proline upon CoA binding. , Specifically, in the absence of CoA, domain swapping occurs with the proline in the trans conformation, thereby allowing the last C‐terminal β‐strand of the β‐helix to reach over to another subunit in the trimer. It is not known whether this same transition occurs in PerB because the structure of the enzyme is not known in the absence of CoA. Surprisingly, this trans to cis change does not occur in VioB where Pro 201 adopts the cis conformation regardless of the presence or absence of CoA.

In the first X‐ray crystallographic analysis of PglD reported in 2008, molecular modeling was utilized to address the structural features involved in the reaction mechanism of the enzyme since it was not possible to obtain the enzyme/UDP‐sugar complex. On the basis of these insightful experiments, it was suggested that the conserved histidine (His 125 in PglD) functions as the catalytic base to abstract a proton from the sugar C‐4′ amino group of the sugar, which was further supported by site‐directed mutagenesis experiments. Additionally, His 125 was thought to be activated by its close proximity to the side chains of Glu 124 and Glu 126. It was also hypothesized that the sugar amino group enters the active site in its unprotonated form, which was subsequently validated in studies conducted by this laboratory on PerB. Finally, the transition state model suggested that the tetrahedral intermediate would adopt the S absolute configuration, that the oxyanion would be stabilized by a hydrogen bond from the backbone amide of Gly 143, that the C‐4′ amino nitrogen would hydrogen bond to Asn 118 via a water molecule, and that this interaction would aid in the stabilization of the tetrahedral intermediate.

These features can now be structurally addressed on the basis of the tetrahedral intermediates determined by this laboratory. Shown in stereo is Figure 6A is a close‐up stereo view of the region surrounding the true tetrahedral intermediate in Pcryo_0637. Indeed, the intermediate adopts the S configuration, the oxyanion is stabilized by the backbone peptidic group of Gly 160, and the C‐4′ amino nitrogen lies within 3.2 Å of His 142 and a water molecule, which, in turn, hydrogen bonds to Asn 135. As predicted for PglD, His 142 is positioned between two carboxylate side chains (Glu 141 and Asp 143). Several of these features are not conserved in either VioB or PerB, however. Indeed, in VioB, the conserved histidine (His 135) is not surrounded by carboxylate side chains but rather by Ser 134 and Gly 153. The water bridging the nitrogen of the tetrahedral intermediate to the asparagine residue (Asn 128) is present, though, as shown in Figure 6B. Strikingly, in PerB, the conserved histidine is surrounded by carboxylate containing side chains (Figure 6C), the water bridging, the nitrogen of the tetrahedral intermediate to the asparagine residue is conserved, but the backbone peptidic nitrogen of Gly 159 is not in the proper position to provide a positive charge to the oxyanion of the intermediate. It appears that the only truly conserved features among these three transition state intermediates (or analogues) are the positions of the catalytic histidines and the presence of water molecules that bridge the nitrogens of the intermediates to the carboxamide groups of the strictly conserved asparagine residues (Figure 6A–C). As noted above, VioB displays a significantly lower catalytic efficiency than PerB, PglD, and Pcryo_0637. This may be, in part, due to the lack of carboxylate‐containing side chains surrounding the catalytic histidine that have been proposed to increase its pKa.

CONCLUSION

In summary, we have demonstrated biochemically that VioB from A. baumannii isolate BAL_212 catalyzes the acetylation of dTDP‐4‐amino‐4,6‐dideoxy‐d‐glucose to yield dTDP‐4‐acetamido‐4,6‐dideoxy‐d‐glucose. Acetyl‐CoA serves as the two‐carbon source. VioB demonstrates a catalytic efficiency of 3.9 × 104 M−1 s−1, which is significantly reduced compared to enzymes such as PglD, which displays a catalytic efficiency of 2.0 × 107 M−1 s−1. Indeed, the high catalytic efficiency of PglD suggests, according to diffusion theory, that it is nearing its evolutionary endpoint, which is not the case for VioB.

In all Type I sugar N‐acetyltransferases studied thus far, there is a conserved proline that resides in the final turn of the β‐helix motif. To date, this proline in the Type I family members has been shown to undergo a trans to cis conformational change upon CoA binding. In the case of VioB, however, the proline adopts the cis conformation regardless of the presence or absence of the cofactor. Additionally, it has been suggested that a conserved dyad, namely the catalytic histidine, followed by either an aspartate or a glutamate, was important for catalysis. Surprisingly, the residue following the catalytic histidine in VioB is a glycine. This change may explain, in part, the reduced catalytic efficiency of VioB in comparison with other members of the Type I sugar N‐acetyltransferases.

How common is 4‐acetamido‐4,6‐dideoxy‐d‐glucose in the microbial world? It is not clear, but the first published report of its existence was in 1959 from E. coli. Since that time, there have been sporadic reports of the sugar being isolated from the O‐antigens of Hafnia alvei 1205, A. baumannii, Pseudomonas fluorescens, Cellulophaga tyrosinoxydans, and Aeromonas hydrophila. , , , , , An elegant biochemical characterization of the sugar N‐acetyltransferase from E. coli O7 was described in 2007. The kinetic parameters for the E. coli enzyme were similar to those reported here for VioB (K M of 0.14 mM and a k cat/K M of 1.1 × 104 M−1 s−1), and interestingly, the N‐acetyltransferase from E. coli demonstrated catalytic activity over a temperature range of 4 to 65°C.

Other than these published reports, however, there remains a paucity of biological information regarding the presence and role of 4‐acetamido‐4,6‐dideoxy‐d‐glucose. Indeed, whether 4‐acetamido‐4,6‐dideoxy‐d‐glucose is important for bacterial virulence is unknown at the present time. What is known, however, is that in some strains of A. baumannii, it has been found in either the O‐antigen or in the capsular polysaccharide. , A. baumannii is an opportunistic pathogen responsible for nosocomial infections worldwide, and it is becoming increasingly resistant to the current arsenal of antibiotics. In light of this, a thorough understanding of its glycoconjugates and the enzymes responsible for their biosynthesis is critical. Indeed, despite all mandates imposed throughout the world since the start of the 2020 pandemic, there continues to be outbreaks in hospitals of antibiotic‐resistant A. baumannii infections, which represent a significant challenge in patient care. ,

CONFLICT OF INTEREST

The authors have no competing financial interests.