Investigation of the catalytic triad of arylamine N-acetyltransferases: essential residues required for acetyl transfer to arylamines

Abstract

The NATs (arylamine N-acetyltransferases) are a well documented family of enzymes found in both prokaryotes and eukaryotes. NATs are responsible for the acetylation of a range of arylamine, arylhydrazine and hydrazine compounds. We present here an investigation into the catalytic triad of residues (Cys-His-Asp) and other structural features of NATs using a variety of methods, including site-directed mutagenesis, X-ray crystallography and bioinformatics analysis, in order to investigate whether each of the residues of the catalytic triad is essential for catalytic activity. The catalytic triad of residues, Cys-His-Asp, is a well defined motif present in several families of enzymes. We mutated each of the catalytic residues in turn to investigate the role they play in catalysis. We also mutated a key residue, Gly¹²⁶, implicated in acetyl-CoA binding, to examine the effects on acetylation activity. In addition, we have solved the structure of a C70Q mutant of Mycobacterium smegmatis NAT to a resolution of 1.45 Å (where 1 Å=0.1 nm). This structure confirms that the mutated protein is correctly folded, and provides a structural model for an acetylated NAT intermediate. Our bioinformatics investigation analysed the extent of sequence conservation between all eukaryotic and prokaryotic NAT enzymes for which sequence data are available. This revealed several new sequences, not yet reported, of NAT paralogues. Together, these studies have provided insight into the fundamental core of NAT enzymes, and the regions where sequence differences account for the functional diversity of this family. We have confirmed that each of the three residues of the triad is essential for acetylation activity.

INTRODUCTION

The NATs (arylamine N-acetyltransferases) are a family of enzymes initially identified as being responsible for the metabolism of arylamines and arylhydrazines by transfer of an acetyl group from acetyl-CoA to the terminal nitrogen group of the substrate [1]. As a result of genome sequencing, a range of NAT homologues have now been identified, and a pattern has emerged [2]. The proteins are found in a range of organisms, and were proposed historically through enzymic studies to have an active-site cysteine residue [1]. Kinetic analyses using tissue homogenates [3], and more recently pure enzyme [4,5], are compatible with a Ping Pong mechanism, in which the active-site cysteine is acetylated initially by the acetyl donor. For such reaction mechanisms, a nonproductive interaction with the acetyl acceptor in advance of cysteine acetylation can occur, and this has been specifically observed for NAT family members [4,6,7]. Studies on the NAT enzyme from Salmonella typhimurium, which was initially identified as also carrying out the O-acetylation of hydroxylamines, were the first to indicate that the active site of NATs contains a cysteine residue (in S. typhimurium NAT, this is residue 69) [8]. The active-site cysteine was also found to be essential by site-directed mutagenesis of the human NAT2 isoform [9,10], and by chemical modification of the cysteine residue in NATs from rabbit [11], S. typhimurium [12] and hamster [4].

Subsequently, crystallographic analysis of the enzyme from S. typhimurium indicated that the cysteine residue in the active site is located adjacent to a histidine residue and an aspartate residue, which together form a proposed catalytic triad [12]. These residues forming the catalytic triad are highly conserved in NAT homologues for which sequence data are available. In addition, crystal structures are now available for NATs from other prokaryotes, including Mycobacterium smegmatis [13], Pseudomonas aeruginosa [14] and Mesorhizobium loti NAT1 [15]. The NAT enzymes all share a common fold, composed of three domains of approximately the same length. The first two domains, an α-helical bundle and a β-barrel (amino acids 1–85 and 86–174 respectively), are joined by a linker helix (amino acids 175–200) to the third domain, an α/β lid (residues 201–280). In each case the tertiary structures of the native enzymes are very similar, and the residues of the proposed catalytic triad can be superimposed. While it has been demonstrated that these NAT enzymes each contain the residues forming the catalytic triad, it has not been demonstrated whether each of the three residues, apart from cysteine, is required for enzymic activity [8,9,16]. In order to answer this question, we have generated NATs from M. smegmatis and S. typhimurium in which each of the residues of the catalytic triad has been mutated. We have investigated the effects of these mutations on the catalytic activity of the enzyme, and relate this to in silico studies of the sequences of all NAT homologues for which sequence information is currently available. In addition, because a glutamine residue is an approximate isostere of an acetylated cysteine, we have mutated the active-site cysteine of M. smegmatis NAT (Cys⁷⁰; equivalent to Cys⁶⁹ of S. typhimurium NAT) to a glutamine (mutant C70Q). This mutant provides a model for the structure of the acetylated intermediate that offers some insight into the structural stabilization of this species, and we have determined its crystal structure to a resolution of 1.45 Å (where 1 Å=0.1 nm).

MATERIALS AND METHODS

Site-directed mutagenesis

A series of mutants was created (Table 1) using the gene sequences of NATs from M. smegmatis and S. typhimurium as templates, and using the method described previously [17]. The final mutant products were cloned into pET28b and pBAD/gIII vectors as described previously [5], with the mutated residues being confirmed by sequencing. The pET28b vectors were propagated in Escherichia coli strain BL21(DE3)pLysS, as described previously [13], in LB (Luria–Bertani) medium supplemented with final concentrations of 1 M sorbitol and 0.25 mM betaine, with induction of recombinant protein being carried out at 27 °C in the presence of 0.1 mM isopropyl β-D-thiogalactoside. The pBAD/gIII vector was propagated in E. coli TOP10 cells, as described previously [5], except that induction was carried out at 27 °C in the presence of 0.02% (v/v) arabinose for M. smegmatis NAT and the derived mutants.

Table 1. Oligonucleotides used for generating mutants of NATs from M. smegmatis and S. typhimurium.

For NAT from M. smegmatis, two PCR products were ligated to produce the mutated open reading frame of msnat, since the high GC content precluded the use of the system employed for S. typhimurium NAT. M. smegmatis NAT cloned into pGEMT-EZ was used as template DNA. Each sense primer (S) was used with the cloning vector SP6 and each antisense primer (AS) was used with the cloning vector T7 (http://www.promega.com/vectors). The two products were then blunt-end-ligated before amplification and subsequent cloning into either pET28b or pBAD/gIII. For NAT from S. typhimurium, mutations in the open reading frame of stnat cloned into pBADgIII [5] were created using the GeneEditor™ kit (Promega Inc.) using the mutagenic oligonucleotide sense primers shown. Numbers correspond to nucleotides in relation to the first nucleotide of the open reading frame of the corresponding gene. Nucleotide mismatches compared with the wild-type template are underlined. All mutated DNA fragments were sequenced. Activity assays were carried out as described in the Materials and methods section. ‘−’ indicates that no activity was detected with 4-aminoveratrol, p-anisidine, isoniazid, 5-aminosalicylate or p-aminobenzoic acid as substrate with either acetyl-CoA or pNPA as acetyl donor after 1 h at 37 °C with up to 1 mg of enzyme and 2 mM substrate or acetyl donor; ‘+’ indicates that activity was detected (see Table 3).

Mutation	Activity	Oligonucleotide (5′→3′)	Sense/antisense
M. smegmatis NAT
C70→X	−	C226TGCTCGGCTACGTACTG243	S
H110→X	−	G346CGGTGCCGGGCGCC360	S
D127→X	−	C397AGACGCTGACCTCGCC413	S
Both enzymes
C70→A	−	G225CCGTTGTGTTCGTAGGCGTACCCGCCACGGCG193	AS
C70→Q	−	G225CCGTTGTGTTCGTACTGGTACCCGCCACGGCG193	AS
C70→S	−	G225CCGTTGTGTTCGTAGCTGTACCCGCCACGGCG193	AS
H110→R	−	G345ACCGACAGCACGTTGCGGGTCTGGGCGGGCAG313	AS
H110→W	−	G345ACCGACAGCACGTTCCAGGTCTGGGCGGGCAG313	AS
H110→A	−	G345ACCGACAGCACGTTGGCGGTCTGGGCGGGCAG313	AS
D127→W	−	T396CCGCCGAAGCCCACTTCGACCAGGTACCGGCC364	AS
D127→N	−	T396CCGCCGAAGCCCACGTTGACCAGGTACCGGCC364	AS
D127→A	−	T396CCGCCGAAGCCCACGGCGACCAGGTACCGGCC364	AS
S. typhimurium NAT
C69→A	−	C190GACGCGGTGGATACGCTTTTGAACTGAATGGC223	S
C69→Q	−	C190GACGCGGTGGATACCAATTTGAACTGAATGGC223	S
C69→S	−	C190GACGCGGTGGATACAGTTTTGAACTGAATGGC223	S
G126 →P	+	T398GGGGTTTCCCGGCCAAACGCTAACCG424	S
G126→A	+	T398GGGGTTTGCCGGCCAAACGCTAACCG424	S

The recombinant NAT enzymes, all of which had an N-terminal hexahistidine affinity tag, were prepared from the soluble fraction of whole-cell lysates, which includes the periplasmic space. The NAT enzymes were purified using a Ni²⁺-nitrilotriacetate affinity resin, with the M. smegmatis NAT variants eluting in 50 mM imidazole [13] and the S. typhimurium NAT variants eluting in 250 mM imidazole [5]. The hexahistidine affinity tag was removed by digestion with thrombin (5 units/mg of protein); the protein was then dialysed into 20 mM Tris/HCl, pH 8.0, 1 mM EDTA and 1 mM dithiothreitol, filtered through a 0.22 μm filter, stored at 4 °C, and either used for assay of activity or concentrated to 15 mg/ml for crystallization trials, as described previously [12,13].

Crystallization of C70Q mutant M. smegmatis NAT

The NAT protein was incubated with Molecular Dimensions Structure Screen I and II, and crystallization trials were set up using a Tecan Genesis ProTeam 150 liquid handling robot into 96-well plates. A drop size of 1 μl (protein/mother liquor 1:1, v/v) was used in all cases. Plates were sealed and incubated at 20 °C. Crystals generally appeared between 1 and 3 days.

Data collection and structure determination

X-ray diffraction data were collected at the ESRF (European Synchrotron Radiation Facility, Grenoble, France) on beamline 14.1 under cryogenic conditions. Diffraction images were analysed and integrated using MOSFLM [17]. Molecular replacement was carried out using Amore [17], with a modified ‘A’ chain from the native M. smegmatis NAT (PDB ID 1gx3) as a search model to give an initial R-factor after the ‘FITING’ stage of 0.42. Rigid body refinement was applied to the molecular replacement solution using REFMAC5 [17a], followed by several rounds of restrained refinement (also REFMAC5). Model rebuilding was carried out using O [18]. Solvent molecules were located using ARP/wARP [19]. Co-ordinates and structure factors have been deposited in the Protein Data Bank (PDB ID 1W5R).

Determination of enzymic activity

Protein samples were analysed by SDS/PAGE for purity, and the concentration of protein was determined either by measuring the absorbance at 280 nm using molar absorption coefficients of 1.15 litre·mol⁻¹·cm⁻¹ for S. typhimurium NAT and 0.95 litre·mol⁻¹·cm⁻¹ for M. smegmatis NAT (http://www.basic.nwu.edu/biotools/proteincalc.html), or by determining the concentration of protein using the Bradford colorimetric method [20] with BSA as a standard. Determination of protein precipitation was by measurement of protein concentration during storage at 4 °C in the supernatant following centrifugation at 14000 g for 10 min at 4 °C.

Enzymic activity to quantify arylamine substrate remaining was determined spectrophotometrically (on a Cecil 5500 spectrophotometer) by analysis of acetylation of the arylamines p-aminobenzoic acid, p-anisidine, 4-aminoveratrole and 5-aminosalicylate as substrates at 37 °C, and using acetyl-CoA as an acetyl donor. In experiments where pNPA (p-nitrophenyl acetate) was used as an acetyl donor, quantification was carried out by measurement of the hydrolysis of pNPA in the presence of arylamine substrate, as described previously [5].

Sequence analysis

Similar NAT sequences were obtained by a BLAST search [21] using the amino acid sequences from M. smegmatis and S. typhimurium NATs as search models. Sequences were then aligned using the CLUSTALW (http://www.ebi.ac.uk/clustalw/) [22] program (version 1.82) using the BLOSUM matrix, and redundant sequences were removed using the EXPASY reduce redundancy program (http://us.expasy.org/tools/redundancy/). To find consensus sequences, ESPRIPT [23] was used to identify different levels of sequence conservation (100%, 90%, 80% and 70%). Prokaryotic sequences were analysed separately from eukaryotic sequences in order to determine conservation among each group, and then all sequences were analysed together to find a global consensus sequence. Sequence conservation was also analysed using the AMAS server (http://barton.ebi.ac.uk/servers/amas_server.html). The alignment produced from CLUSTALW [22] was used with NAT sequences grouped as suggested in the AMAS manual. All numbering of residues in the text and Figures is based on the NAT sequence from S. typhimurium for ease of comparison. Although S. typhimurium NAT has only 280 amino acid residues, sequence numbering is from 1 to 310 because of the amino acid sequence of Neurospora crassa NAT, which contains an insertion in the sequence not found in any other NAT homologue.

RESULTS AND DISCUSSION

Expression of active-site cysteine mutants

Recombinant mutant M. smegmatis NAT enzymes in which the active-site cysteine was mutated to alanine, serine or glutamine were expressed as recombinant proteins using both the pBADgIII expression system and the pET28b expression vector (Table 1). Using the pBADgIII system, each of the NAT mutants was expressed as a soluble protein, such that the yield of purified protein was approx. 3 mg/litre of culture for each mutant. This is comparable with the yield of wild-type M. smegmatis NAT protein produced with the same expression vector under similar induction conditions [0.02% (v/v) arabinose for 4 h at 27 °C]. The recombinant proteins were equally as soluble as the native protein, as observed using SDS/PAGE (Figure 1). Although other bands were present on the gel, there were no differences in these minor contaminants or degradation products between the wild-type enzyme and the site-directed mutants. This suggests that the mutant recombinant proteins were folded in solution, such that they were not susceptible to proteolysis, as has been observed with other, unfolded mutant NAT enzymes [10,24–26].

Figure 1. Site-directed mutants of M. smegmatis NAT are expressed as soluble proteins.

An SDS/12%-PAGE gel stained with Coomassie Blue is shown. Each mutant was purified as described in the Materials and methods section, and then the hexahistidine tag was cleaved with thrombin. A sample of 10 μg of purified NAT enzyme was loaded in each lane.

The C70Q M. smegmatis NAT mutant recombinant protein was expressed using the pET28b expression system in order to generate sufficient protein for crystallization. The protein was found to crystallize under a range of conditions, and the best crystals were obtained with 0.2 M MgCl₂, 0.1 M Tris/HCl, pH 8.5, and 30% poly(ethylene glycol) 4000. Upon X-ray analysis, the mutant enzyme was found to have crystallized in the space group P4₁2₁2 and diffracted to a resolution of 1.45 Å (Table 2). The X-ray data were processed as described in the Materials and methods section, and molecular replacement was carried out using an ‘A’ chain from wild-type M. smegmatis NAT as a template. Once the molecular replacement solution was found, rigid-body refinement was carried out, followed by model building using O, and further rounds of restrained refinement.

Table 2. Crystallographic data for C70Q M. smegmatis NAT.

Values in parentheses indicate the specific values in the highest-resolution shell. R_merge is defined as:

where I_h,j is the intensity of the jth observation of unique reflection h. R_conv is defined as:

where Fo_h and Fc_h are the observed and calculated structure factor amplitudes for reflection h. R_free is equivalent to R_conv, but is calculated using a 5% disjoint set of reflections excluded from the maximum likelihood refinement stages. RMSD, root mean square deviation. 1 Å=0.1 nm.

Parameter	Value
Space group	P41212
Cell dimensions (Å)	98.9, 98.9, 130.9
Maximum resolution (Å)	1.45
Highest-resolution shell (Å)	1.53–1.45
Observed reflections	330604
Unique reflections	103538
Completeness (%)	90.5 (84.7)
I/σI	6.8 (1.8)
Mosaicity (°)	0.32
Redundancy	3.2 (2.7)
Rmerge (%)	6.0 (41.4)
Vm (Å3/Da)	2.5
Refinement
Protein atoms	1 molecule of C70Q NAT chain A residue 0–276*
Number of atoms	9326
Other atoms	300 water molecules
Resolution range (Å)	31.00–1.45
Rconv (%)	18.8
Rfree (%)	21.5
RMSD bond length (Å)	0.012
RMSD bond angle (°)	1.482

* Residue 0 is the last residue of the N-terminal cloning artefact.

The Cα structure of the C70Q NAT recombinant protein is illustrated in Figure 2(A). The structure of the C70Q mutant is isomorphous with that of the wild-type enzyme from M. smegmatis, and superimposition of the C70Q mutant with the wild-type enzyme using Deep View (SwissPDB Viewer) gives a root mean square difference of 0.35 Å over 1080 backbone atoms. The regions that are different between the two structures are in solvent-exposed loops that are remote from sites of crystal contacts. Since glutamine represents a structural isostere to an acetylated cysteine residue (Figure 2B), the C70Q NAT mutant therefore represents a model of the acetylated intermediate state of the NAT enzyme. The fact that the cysteine-to-glutamine mutation is accommodated within the active site without any structural perturbation of the wild-type fold suggests that an acetylated cysteine group could also be accommodated (Figure 2C). These data do not preclude a conformational change of the protein before and during the acetylation process, as has been proposed for S. typhimurium NAT [5] and has been inferred for the human NAT1 enzyme from the effects of acetylation on protein stability [10]. The C70Q mutant adopts a conformation whereby the amide oxygen of the glutamine forms a 2.85 Å hydrogen bond with its own peptide nitrogen. In doing this, it displaces a tightly bound water molecule observed in the wild-type structure that has been suggested previously to define the centre of the oxyanion hole that facilitates formation of a tetrahedral intermediate during nucleophilic attack [12]. Observation of this interaction in this structure indicates that the oxyanion hole may play an additional role in stabilizing and orienting the acetyl moiety for subsequent attack by the acetyl group acceptor of the NAT reaction. Other minor rearrangements are seen in the active site, notably in the orientation of the imidazole ring of the catalytic histidine, although we cannot exclude the possibility that this results from the different chemical nature of a glutamine compared with the true acetylated intermediate.

Figure 2. Structure of C70Q M. smegmatis NAT at 1.45 Å resolution.

(A) Comparison of the peptide backbone of wild-type M. smegmatis NAT (blue) and the C70Q mutant (green), showing the catalytic triad as a ball and stick representation. (B) Superposition in the model of acetylcysteine and glutamine. (C) Close-up view of the modified catalytic triad in the C70Q M. smegmatis NAT. Maps are contoured at 1.5 σ. Panels (A) and (C) were produced using AESOP (M. E. M. Noble, unpublished work; details available from M. E. M. N. on application; email martin.noble@biop.ox.ac.uk); (B) was produced using the Deep-View Swiss pdb viewer [51] (http://www.expasy.org/spdbv).

The acetylated intermediate of hamster NAT2 has been estimated to have a half-life of 88 s [27] using pure enzyme, which is longer than a previous estimate of less than 1 min for pigeon NAT in a liver homogenate [3]. Although the half-life of the acetylated intermediate is likely to be dependent on the specific NAT isoenzyme, these values, suggesting a transient acetylated intermediate, are compatible with the calculation that approx. 10% of active human NAT1 becomes labelled when reacted with radioactive acetyl-CoA [11,28]. It has been found, using an assay to detect the hydrolysis of acetyl-CoA directly, that the presence of arylamine substrate greatly increases the rate of hydrolysis of acetyl-CoA by NAT from M. smegmatis [29].

In view of the difficulty in producing sufficient quantities of the acetylated intermediate of a NAT enzyme for crystallization, the glutamine mutant therefore represents the best model currently available for the active-site-acetylated intermediate.

Activities of site-directed mutants

Open reading frames corresponding to the mutants of both M. smegmatis NAT and S. typhimurium NAT, in which Cys⁶⁹, His¹⁰⁷ and Asp¹²² (S. typhimurium NAT numbering) were mutated, were created (Table 1). The mutated NAT enzymes were expressed as soluble recombinant proteins. The rate of precipitation of each of these proteins was found, on storage in the presence of 1 mM dithiothreitol, to be comparable with that of the wild-type enzyme. In each case the mutants were expressed at similar levels of soluble protein as the recombinant wild-type enzyme [30]. Recombinant NAT enzymes were each used (for enzyme assays) within the first 20 days of storage (at 4 °C) for comparison with the wild-type enzymes. None of the mutant proteins exhibited any acetylation activity against p-aminobenzoic acid, 4-aminoveratrole, 5-aminosalicylate or p-anisidine (Table 1). The latter three substrates are effectively acetylated by the wild-type enzyme at pH 7.5. A range of conditions, including modification of pH across the range 5.0–8.5, concentrations of substrates or acetyl donors of up to 2 mM, and the presence of up to 1 mg of protein per assay, equally did not result in any acetylation activity by the mutant enzymes after 1 h at 37 °C (Table 1). These data suggest that the residues Cys⁶⁹, His¹⁰⁷ and Asp¹²², which form the catalytic triad, are each individually essential for acetylation activity with acetyl-CoA as the acetyl donor.

Each of the catalytic triad residues was mutated to an alanine residue. None of these mutant enzymes exhibited acetylation activity. Enzymes in which the cysteine residue was mutated to either an alanine or a glutamine would not be expected to catalyse acetyl transfer. However, substitution of a cysteine with a serine residue, such as is involved in acetyl transfer in acetylcholinesterase [31], was expected to produce an enzyme with residual activity. However, no activity could be detected, suggesting that the serine residue is insufficiently nucleophilic to support NAT activity (Table 1). In an artificial cysteine protease, thiosubtilisin, in which serine has been replaced by cysteine, it has been demonstrated that the enzyme, as well as having protease activity, can catalyse an acetylation reaction when ethyl acetate is used as the acetyl donor, albeit at a very low rate [32].

The His¹⁰⁷ residue was mutated to an arginine residue to determine the role of the potential proton relay created by the histidine within the catalytic triad. Substitution of histidine with a tryptophan residue was also carried out to mimic the aromatic properties of histidine. Neither the His¹⁰⁷→Arg nor the His¹⁰⁷→Trp mutant supported catalytic activity in M. smegmatis NAT (Table 1).

In the acetylcholinesterases, the third residue of the catalytic triad is glutamate [31]. This is also the case in several species of Bacillus NATs (see Supplementary Figures 1 and 2 at http://www.BiochemJ.org/bj/390/bj3900115add.htm); however, it is not known whether these NAT homologues are active in catalysis. The replacement of Asp¹²² with Asn in M. smegmatis NAT resulted in an inactive protein (Table 1). This substitution is found in one isoform of human NAT2 (NAT 2*12D) which produces an inactive enzyme [33]. The studies presented here complement refolding studies with the insoluble recombinant hamster NAT2 enzyme, in which it was reported that enzyme in which the active-site aspartate was replaced by either alanine or asparagine could not be refolded in an active form, unlike the refolded native hamster NAT2 [4]. A catalytic aspartate has been demonstrated to be important in prokaryotic and eukaryotic members of the transglutaminase family [34,35] and in the serine proteases, such as subtilisin, whereas in the cysteine proteases asparagine is often found as the third residue of the triad [36]. Whether the NAT enzyme can act as a protease has not been reported.

Modification of the putative acetyl-CoA binding site

Unlike acetyltransferases of the unrelated GNAT (GCN5-related NAT) family, for which structures of the enzyme complexed with acetyl-CoA have been obtained for several family members [37,38], a structure of NAT with acetyl-CoA bound has not been forthcoming. However, there are regions of NAT that have been suggested, on the basis of sequence conservation [2,12], to be involved in acetyl-CoA binding, particularly in comparison with the RifF enzyme, a NAT homologue involved in amide bond formation in the synthesis of rifamycin which does not use acetyl-CoA as a cofactor [39]. It has been proposed that Gly¹²⁶ in NAT from S. typhimurium forms part of a structural P-loop involved in acetyl-CoA binding [12]. The Gly¹²⁶ residue of NAT from S. typhimurium was mutated to either an alanine or a proline. Proline was chosen for substitution because glycine is replaced in RifF by a proline [39]. When Gly¹²⁶ of S. typhimurium NAT was converted into a proline residue, the enzyme lost the ability to catalyse the transfer of an acetyl group from acetyl-CoA; however, the enzyme could still catalyse the acetylation of an arylamine using the acetyl donor pNPA (Table 3). The RifF enzyme, in which Gly¹²⁶ is a proline residue, is unable to catalyse acetylation of substrate using either acetyl-CoA or pNPA as an acetyl donor [39]. When Gly¹²⁶ in S. typhimurium NAT was replaced by alanine, the protein was less active (Table 3), but still supported the transfer of an acetyl group from acetyl-CoA with either isoniazid or p-anisidine as an acetyl acceptor. These recombinant mutant enzymes were found to be as soluble as the wild-type S. typhimurium NAT enzyme.

Table 3. Effect of substitution of Gly¹²⁶ of S. typhimurium NAT on acetyl transfer activity.

Activity assays were carried out as described in the Materials and methods section using wild-type enzyme and the G126A and G126P mutants. nd, no activity detected.

	Acetyl-CoA		pNPA
Enzyme	Km (μM)	kcat (s−1)	Km (μM)	kcat (s−1)
Wild-type	<20	1.27	159	353
G126A	80	8.73×10−3	1100	0.16
G126P	nd	nd	2350	0.89

Gly¹²⁶ therefore appears to be involved in acetyl-CoA binding. It is likely that there are residues other than Gly¹²⁶ contributing to acetyl-CoA binding; however, these have not yet been identified.

It has also been demonstrated using random [40] and site-directed [41] mutagenesis that a range of amino acid substitutions can result in a decrease in activity despite a lack of effect on protein stability. These point mutations, resulting in substitution of a particular amino acid, are likely to be specific effects, rather than due to non-specific structural modifications resulting in poorly folded, and hence unstable, proteins.

G126P/A substitutions in naturally occurring NATs

A Gly¹²⁶→Ala substitution is found in NAT enzymes from two Bacillus species, namely B. cereus and B. anthracis [42] (see Supplementary Figure 2 at http://www.BiochemJ.org/bj/390/bj3900115add.htm). Interestingly, a Gly¹²⁶→Pro substitution is found in the NAT3 enzymes from mice and rats [43–45] as well as in the rifamycin amide synthase enzyme RifF [39]. This substitution in mouse and rat NAT3 may explain why these isoenzymes exhibit very low rates of acetylation activity.

In silico analysis of NAT sequences

The analysis of 29 prokaryotic NAT sequences provides a clearer view of which residues are conserved (Figure 3). There are only ten fully conserved residues among the prokaryotic NATs, with 17 highly conserved residues (90% level). This is in contrast with the eukaryotic NATs, where 32 residues are completely conserved among 30 sequences, with 56 residues being highly conserved (90% level). There are two main clusters of conservation within the prokaryotic sequences, P*EN (residues 37–40) and RGG*CYE (residues 65–71) (where * refers to non-conserved residues); however, the eukaryotic sequences show a much higher degree of identity throughout the entire sequence, as might be expected from relative evolutionary timescales (Figure 4).

Figure 3. Sequence conservation among prokaryotic NAT enzymes.

A total of 29 prokaryotic NAT sequences were found using a BLAST search and aligned using the CLUSTALW facility (www.ebi.ac.uk/clustalw/). Once aligned, the sequences were subjected to analysis using the ESPRIPT [23] server. AESOP (M. E. M. Noble, unpublished work; details available from M. E. M. N. on application) was used to highlight the Cα-backbone by indicating 100% (red), 90% (orange), 80% (yellow) and 70% (green) conservation across all prokaryotic sequences aligned (A and B). The corresponding ribbon diagrams (C and D) are shaded ramped from N-terminus (blue) to C-terminus (red). The catalytic triad residues are shown in stick format. (A) and (C) are in the same orientation as each other, and are orthologous to (B) and (D).

Figure 4. Sequence conservation among NAT enzymes.

All NAT sequences found from BLAST searches were aligned using the CLUSTALW facility (www.ebi.ac.uk/clustalw), and the alignment files were used for analysis using the AMAS server [52]. Prokaryote (A) and eukaryote (B) NATs were analysed separately, with the inclusion of the putative NAT sequence from Neurospora crassa in both analyses. The N. crassa NAT enzyme has 310 amino acids, and appears to have an extended loop region not present in any other NAT homologues, creating an insertion in the sequence alignment corresponding to residues 254–268 (A) and 240–249 (B). Mid-grey shading indicates the level of conservation between all species, and black shading indicates similarity between species. The light-grey shading indicates differences in amino acids between the sequences at each point along the alignment. The top bar represents the boundaries between domains based upon the S. typhimurium NAT enzyme domains: residues 1–85, α-helical bundle; 86–174, β-barrel domain; 175–200, linker helix; 201–280, α/β lid domain.

Among all prokaryotic and eukaryotic NAT sequences analysed here, only two do not have an arginine residue at position 65: Bacillus subtilis NAT [2] and the Amycolatopsis mediterranei NAT-like protein RifF [39] have a histidine and glutamine residue respectively. There is an allele of the human NAT (human NAT2*5) where a glutamine occupies this position [46]. Unlike the RifF enzyme, the human NAT isoform still exhibits acetylation activity, albeit at a low rate [46].

Multiple NAT sequences have been found in several prokaryotic species, including Mesorhizobium loti [47], for which structural information is available [15]. As a result of our bioinformatic study (Figures 4 and 5, and Supplementary Figures 2 and 3 at http://www.BiochemJ.org/bj/390/bj3900115add.htm), we have identified several other prokaryotic organisms that have more than one NAT homologue, i.e. B. cereus and B. anthracis, each of which has three NAT homologues (Supplementary Figures 2 and 3).

Figure 5. Fully conserved residues among NAT sequences.

A total of 59 NAT sequences were aligned using the CLUSTALW server to analyse sequence conservation. Amino acids which are conserved across all NAT sequences are shaded in black. Amino acids conserved among eukaryotes only are shaded in light grey; those conserved in prokaryotes only are shaded in mid-grey. The sequence is shown in three sections approximating to the three domains, as indicated in the legend to Figure 4. Numbering is relative to that of the NAT sequence from S. typhimurium. Although some NAT enzymes are longer than 280 amino acids (e.g. Neurospora crassa with 310 amino acids), no sequence conservation is seen in the final C-terminal region. Amino acids are labelled in single-letter format. The amino acid sequence of N. crassa NAT was omitted from this analysis, as it appears to differ from both the prokaryotic-only and the eukaryotic-only alignments.

Only seven invariant residues throughout the entire sequence have been identified; these are Leu²⁵, Gly⁶⁶, Gly⁶⁷, Cys⁶⁹, Asn⁷³, Gly⁸⁴ and His¹⁰⁷ (Figure 5). The third member of the catalytic triad (Asp¹²²) is not completely conserved, and is found as an asparagine in one human NAT2 isoform [33] which is inactive in acetyl-transfer activity. Glutamate has been reported at position 122 in four Bacillus species [B. thuringiensis (subspecies konkukian), B. cereus (subspecies A.T.C.C. 10987), B. cereus (subspecies A.T.C.C. 14579/DSM31) and B. anthracis] replacing the catalytically important Asp¹²², but it is not known whether these enzymes exhibit N-acetylation activity. These enzymes may play another, as yet unknown, role in their host organism.

Of the seven invariant residues observed in all NAT enzymes (Figure 5), Leu²⁵ is likely to be involved in structural stability, as it forms van der Waals forces with Leu⁷⁹ and a highly conserved Val¹¹². Leu⁷⁹ is completely conserved in all NAT enzymes analysed here, with the exception of NAT from Streptomyces murayamaensis, where an equally hydrophobic valine occupies the equivalent position (see Supplementary Figure 2 at http://www.BiochemJ.org/bj/390/bj3900115add.htm). Val¹¹² is also highly conserved (>70% conservation among all NATs), being replaced by cysteine (4/59), alanine (9/59), threonine (4/59) or isoleucine (9/59) (Supplementary Figure 4). These interactions are likely to contribute towards maintaining the β-barrel domain position, hence contributing towards the structural integrity of the NAT enzymes. The Gly⁶⁶ and Gly⁶⁷ residues also appear to play a role in stabilizing the structure of the NAT enzymes, as they are involved in forming hydrogen bonds with the well conserved Pro³⁷ and Glu³⁹ (part of the well conserved PFENL region) [2] (Figure 4 and Supplementary Figures 2 and 4). The amide backbone nitrogen of Gly⁶⁶ forms hydrogen bonds with the carboxy side chain of Glu³⁹, while the amide backbone nitrogen of Gly⁶⁷ forms hydrogen bonds with the backbone carbonyl oxygen of Pro³⁷.

Cys⁶⁹ is the essential active-site cysteine [8,9,11,12,16]. Previous mutagenesis experiments with human NAT1 suggested that this cysteine residue may exist in an acetylated form, protecting the enzyme from proteolysis [10], although no evidence has been found for this as yet in the prokaryotic NAT enzymes.

Asn⁷³ is likely to play a role in stabilizing the catalytic cysteine, keeping it in the correct orientation, as it forms hydrogen bonds from the amide nitrogen of the backbone with the backbone carbonyl oxygen of Cys⁶⁹. Gly⁸⁴ is situated at the end of helix 4 prior to the β-barrel domain (Figure 4), and is likely to be important in allowing the adoption of the appropriate relative disposition of the α-helical and β-barrel domains of NAT. Finally, His¹⁰⁷ is the last completely conserved residue in the consensus sequence, and is important in forming the catalytic triad of residues. There are no fully conserved residues C-terminal of His¹⁰⁷, and this may relate to a divergent role for NAT enzymes. The C-terminus has been demonstrated to have a controlling effect on substrate specificity and enzymic activity [5]. It is possible that the wide range of NAT homologues may have widely divergent endogenous functions controlled by the C-terminal sequence.

In prokaryotes, only one additional completely conserved residue exists, Gly¹²⁴ (Figure 5). This residue both stabilizes the catalytically important Asp¹²² by forming a hydrogen bond and forms part of the loop containing Phe¹²⁵, implicated in many substrate/inhibitor studies of NAT enzymes [5,6,29,48]. However, a consensus sequence for eukaryotic enzymes only extends further into the C-terminal domain and may reflect a greater similarity of function in this subgroup of NAT enzymes. Discussion of the implications of the conservation of these residues with respect to structure is difficult due to the lack of a eukaryotic NAT structure. Previous reports of modelling of eukaryotic NAT enzymes have only managed to model the first 131 amino acids due to the very low identity between the eukaryotic and prokaryotic NAT enzymes at the C-terminus [46,49].

Two arginine residues have been implicated as being essential for NAT activity (Arg⁹ and Arg⁶⁵) [41]; however, they are not completely conserved throughout the sequences highlighted in the present study. At position 9 there are examples of glutamine (Staphylococcus aureus), lysine (Bacillus subtilis, Vibrio parahaemolyticus, Staphylococcus epidermidis) and histidine (Pseudomonas aeruginosa, Neurospora crassa) residues (Supplementary Figure 2 at http://www.BiochemJ.org/bj/390/bj3900115add.htm). The NAT enzyme from Pseudomonas aeruginosa is extremely active with 5-aminosalicylate [50] and hydralazine substrates [14]. It is one of the most active prokaryotic NAT enzymes, suggesting that this arginine residue is not essential. The arginine residue at position 65 is conserved in all NAT enzymes apart from that found in Bacillus subtilis, where a histidine is present; however, there is evidence suggesting that this enzyme is active [2].

Conclusions

The results reported here demonstrate that each of the residues of the catalytic triad are essential for the activity of NATs from S. typhimurium and M. smegmatis. A C70Q mutant of the NAT enzyme from M. smegmatis is a good model for an acetylated enzyme intermediate. Site-directed mutagenesis and activity studies are compatible with Gly¹²⁶ of S. typhimurium NAT being involved in acetyl-CoA binding. In silico analysis of 59 NAT-like sequences has revealed that only seven amino acids are conserved among all species, and that some residues which were thought previously to be invariant amongst NATs may be replaced by similar residues without loss of activity, e.g. Arg⁹ to His in P. aeruginosa NAT. The active-site cysteine and histidine residues are, however, fully conserved among all NAT sequences analysed here. The lack of sequence identity in the C-terminal domain may reflect a difference in the endogenous roles of the different NAT homologues.