Research article

Characterization of an acetyltransferase that detoxifies aromatic chemicals in Legionella pneumophila

Xavier Kubiak, Delphine Dervins-Ravault, Benjamin Pluvinage, Alain F. Chaffotte, Laura Gomez-Valero, Julien Dairou, Florent Busi, Jean-Marie Dupret, Carmen Buchrieser, Fernando Rodrigues-Lima


Legionella pneumophila is an opportunistic pathogen and the causative agent of Legionnaires' disease. Despite being exposed to many chemical compounds in its natural and man-made habitats (natural aquatic biotopes and man-made water systems), L. pneumophila is able to adapt and survive in these environments. The molecular mechanisms by which this bacterium detoxifies these chemicals remain poorly understood. In particular, the expression and functions of XMEs (xenobiotic-metabolizing enzymes) that could contribute to chemical detoxification in L. pneumophila have been poorly documented at the molecular and functional levels. In the present paper we report the identification and biochemical and functional characterization of a unique acetyltransferase that metabolizes aromatic amine chemicals in three characterized clinical strains of L. pneumophila (Paris, Lens and Philadelphia). Strain-specific sequence variations in this enzyme, an atypical member of the arylamine N-acetyltransferase family (EC, produce enzymatic variants with different structural and catalytic properties. Functional inactivation and complementation experiments showed that this acetyltransferase allows L. pneumophila to detoxify aromatic amine chemicals and grow in their presence. The present study provides a new enzymatic mechanism by which the opportunistic pathogen L. pneumophila biotransforms and detoxifies toxic aromatic chemicals. These data also emphasize the role of XMEs in the environmental adaptation of certain prokaryotes.

  • acetylation
  • arylamine N-acetyltransferase
  • detoxification
  • enzyme variant
  • kinetics


Legionella pneumophila is an intracellular bacterium that is ubiquitous in aquatic environments, where it is part of multi-organismal biofilms. The environmental life cycle of L. pneumophila includes growth, survival and spreading within the planktonic phase, protozoan hosts and biofilms. The capacity to parasitize protozoa allows L. pneumophila to replicate in human cells, causing a severe pneumonia called legionellosis or LD (Legionnaires' disease) [1].

From its natural aquatic environments, L. pneumophila can contaminate man-made water systems such as cooling towers, thermal baths or hospital water systems [1]. In these varied ecological niches, L. pneumophila is exposed to numerous chemicals of anthropic origin (e.g. plastic-derived chemicals, pesticides, chlorine disinfectants and antibiotics) and/or compounds produced by other micro-organisms present in biofilms (e.g. pyocyanin or 5-methyl-phenazine-1-carboxylic acid) [2,3]. Although, L. pneumophila is known to express a glutathione transferase and an efflux pump involved in extrusion of certain toxic compounds, the molecular mechanisms by which this micro-organism detoxifies chemical compounds remain largely unknown [4]. In particular, the expression and functions of XMEs (xenobiotic-metabolizing enzymes) that could contribute to chemical detoxification in L. pneumophila have been poorly documented at the molecular and functional levels.

One potential family of XMEs that may be utilized by L. pneumophila is arylamine NATs (N-acetyltransferases). NATs catalyse the AcCoA (acetyl-CoA)-dependent acetylation of a large range of aromatic amine chemicals, including drugs, toxins and toxic environmental pollutants, such as pesticides and by-products of rubbers and pharmaceuticals. These enzymes were first identified in humans, where they play a role in the metabolism of certain drugs [5].

More recently, NATs have also been identified and/or characterized in prokaryotes [6,7]. Structural studies conducted on both bacterial and eukaryotic NATs have shown that this XME family shares a common overall fold containing a Cys-His-Asp catalytic triad [811]. However, differences have been observed between prokaryotic NATs at the structural level (such as insertions/deletions of residue) and are suggested to result in the potentially widely different functions that these enzymes appear to catalyse [7]. For instance, NAT enzymes of Mycobacterium tuberculosis and Bacillus anthracis have been reported to contribute to resistance to aromatic amine antibiotics, such as INH (isoniazid), by acetylating them [7,12]. In addition, Mycobacterium bovis NAT has been shown to be required for the synthesis of mycobacterial cell-wall-specific lipids [13]. Strikingly, RifF (rifamycin amide synthase), a NAT homologue expressed in Amycolatopsis mediterranei, is devoid of NAT activity, but catalyses amide bond formation through ring closure in rifamycin synthesis [7]. Therefore, although NAT family members may have common functions, the roles of NATs may differ depending on the organism they come from [7]. Interestingly, in contrast with vertebrate isoforms, for which polymorphisms that lead to variations in the catalytic activity are well-documented, bacterial NATs from different strains of one species are considered as strictly conserved [7]. The only exception to date was the M. tuberculosis NAT that displays two SNPs (single nucleotide polymorphisms) that lead to decreased activity towards the anti-tubercular drug INH [14]. Interestingly, L. pneumophila also seems to use NATs, as the analysis of the genome sequence of three characterized L. pneumophila strains (Paris, Lens and Philadelphia) [15] revealed the existence of one gene predicted to encode a NAT protein with unusual sequence/structure properties in each strain. Recent analyses showed that this gene is present in all seven L. pneumophila genomes sequenced to date [16] (Supplementary Figure S1 at, but not in the genome of the Legionella longbeachae species [17,18].

In the present paper we report the molecular, biochemical and functional characterization of the NAT protein encoded by three clinical strains of L. pneumophila (Paris, Lens and Philadelphia). Sequence, kinetic and CD analyses demonstrated that the L. pneumophila nat gene displays strain-specific sequence variations (point mutations) leading to enzymatic variants with different structural and catalytic properties. Functional inactivation and complementation experiments showed that the NAT enzyme allows L. pneumophila to detoxify toxic aromatic chemicals. Our data suggest that this NAT protein contributes to the environmental adaptation of L. pneumophila. More broadly, the present study provides new insight into the role that XMEs may play in the adaptation of bacteria, such as L. pneumophila, to certain toxic chemicals present in their environments.


Strains, culture conditions and materials

Escherichia coli BL21 strains (Stratagen) were grown at 37°C in LB (Luria–broth) Miller medium. L. pneumophila clinical strains Paris, Lens and Philadelphia were grown at 37°C in ACES [N-(2-acetamido)-2-aminoethanesulfonic acid; 10 g/l]-BYE (buffered yeast extract broth) supplemented with 0.4 g of L-cysteine and 0.25 g of iron PPi per litre (pH 6.9) or grown on ACES-BCYE (buffered charcoal-yeast extract) agar plates [19,20]. Unless specified, all reagents were purchased from Sigma. 5-AS (5-aminosalicylate) was provided by Acros. Substrates were solubilized to 100 mM in MilliQ water or DMSO (Euromedex) and stored at −20°C until use. Protease inhibitor cocktail was prepared with AEBSF [4-(2-aminoethyl)benzenesulfonyl fluoride; 100 mM] and leupeptin (2 mM) (Euromedex). Antibodies raised against Salmonella typhimurium NAT [α-(SALTY)NAT1] were kindly provided by Professor Edith Sim (Department of Pharmacology, University of Oxford, Oxford, U.K.).

Bioinformatic analysis of L. pneumophila NAT sequences

Identification of the nat gene from L. pneumophila strains Paris (lpp2516), Lens (lpl2369) and Philadelphia (lpg2451) was achieved using BLAST at NCBI (, with default parameters. Multiple sequence alignments were done using Muscle (, with default settings and alignment outputs generated with ESPript V2.2 ( Secondary structure predictions were achieved using PsiPred (

Phylogenetic analyses of the L. pneumophila NAT protein

Amino acid sequences were retrieved with BLASTp using the sequence of the NAT protein from L. pneumophila Paris as the query. The proteins with the best hits from different prokaryotic and eukaryotic organisms were used for phylogenetic reconstruction. The sequences of NAT proteins whose structures have already been characterized were added to this group. All sequences were aligned with the program T-Coffee (expresso) [21] that considers the structure of the molecule. Phylogenetic inference was obtained using two methods: a Bayesian approach as implemented in MrBayes v3.2 [22] and a distance method implemented in MEGA [23]. MrBayes uses Metropolis-coupled Markov chain Monte Carlo methods to calculate the posterior probabilities for the parameters of interest. Each analysis was run for 1×106 generations with four differently heated chains and using the Dayhoff model as the model of evolution. For the distance tree we used NJ (neighbour-joining) analysis that was computed using the Dayhoff model with the option partial deletion of alignment gaps. In total 1000 bootstrap replications were performed to determine statistical support for clades.

Cloning of lpp2516, lpl2369 and lpg2451 coding sequences

The lpp2516, lpl2369 and lpg2451 ORFs (open reading frames) were amplified using two different sense primers for lpp2516 and lpg2451 (5′-CGGGCATATGAAACCCATCGACTTAAA-3′) and lpl2369 (5′-CGGGCATATGAAATCCATCGACTTAAA-3′). The reverse primer was common to all L. pneumophila nat genes (5′-GGGCTCGAGCTATAGTCTAAAAATTAAAT-3′). Genomic DNA was amplified by high-fidelity PCR using Pfu-DNA polymerase (Promega) as described by Pluvinage et al. [12]. PCR products were double-digested using NdeI and XhoI restriction enzymes (New England Biolabs) and cloned into a pET28(a) plasmid to express His6-tagged fusion recombinant proteins. Cloning was verified by DNA sequencing.

Expression and purification of recombinant L. pneumophila NAT proteins from Paris, Lens and Philadelphia strains

NAT protein overexpression was induced by adding 500 μM IPTG (isopropyl β-D-thiogalactopyranoside; for 5 h at 25°C) to E. coli BL21 cultures that had been transformed with the plasmid carrying the respective nat gene, when A600 reached 0.7. Bacteria were resuspended in 40 ml of PBS (pH 7.4), 10 mg of lysozyme, 0.1% Triton X-100 and protease inhibitors, per litre of culture, then sonicated at 4°C for 3 min (8 s pulse on, 30 s pulse off, 20% maximum power; Branson Digital Sonicator model 450) to complete cell lysis. After 30 min of centrifugation at 12000 g and 4°C, the supernatant was incubated with Nickel Chromatrix Ni-NTA (Ni2+-nitrilotriacetate) beads (Jena Bioscience) for 2 h at 4°C with shaking (50 rev./min). Beads were successively washed with PBS and PBS with 20 mM and 30 mM imidazole; proteins were eluted with PBS with 300 mM imidazole. The His6 tag was removed by thrombin cleavage. Enzymes [reduced with 1 mM DTT (dithiothreitol)] were either dialysed or buffer-exchanged (PD-10 columns, Millipore) against 25 mM Tris/HCl (pH 7.5), and kept at −80°C until use. The protein concentration was determined by absorbance at 280 nm.

SDS/PAGE and Western blot analysis

Proteins were separated by SDS/PAGE (12.5% gel) and either stained with Coomassie Blue R-250 to assess protein purity, or transferred on to nitrocellulose membrane for Western blot analysis. Briefly, the membrane was blocked with 5% (w/v) dried skimmed milk in TBS-T [Tris-buffered saline (50 mM Tris and 150 mM NaCl) containing 0.2% Tween 20] (for 1 h), then washed with TBS-T before incubation with primary antibody (1:10000 dilution). After washing, secondary antibody was added (1:100000 dilution). Staining was developed by incubation for 2 min with 1 ml of chemiluminescent solution (SuperSignal West Pico, Pierce) and was visualized with an LAS-4000 imaging system (Fuji).

Enzyme assays and determination of kinetic parameters of L. pneumophila NAT variants

Kinetic parameters, Km (app) and kcat, for the acetylation of prototypic aromatic amine substrates were obtained by steady-state kinetics using the PNPA (p-nitrophenylacetate) assay as described by Mushtaq et al. [24]. To this end, 10 μg/ml enzyme was mixed with 10 μl of aromatic amine substrates at increasing concentrations, and the reaction was started by adding 20 mM PNPA as an acetyl donor. A405 was measured using an ELISA plate analyser. Initial velocities (Vi) were determined on the basis of a ∊PNPA at 405 nm of 0.0035 μM/cm. The DTNB [5,5′-dithiobis-(2-nitrobenzoic acid)] assay was used to determine the kinetic parameters for AcCoA and propionyl-CoA as described in [25]. To this end, 10 μg/ml protein was mixed with 500 μM HDZ (hydralazine chloride) and the reaction was initiated by adding various concentrations of the acetyl/acyl donor. DTNB (25 μl DTNB at 2 mg/ml in 6 M GdmCl (guanidinium chloride) was added to stop the reaction at different time points. A405 was measured using an ELISA plate analyser. Initial velocities (Vi) were determined on the basis of an ∊DTNB at 405 nm of 0.0032 μM/cm. The apparent kinetic parameters Km (app) and kcat were obtained by non-linear curve fitting against the Michaelis–Menten equation using Kaleidagraph 3.5. Steady-state kinetics were performed in triplicate.

CD spectra of L. pneumophila NAT variants

CD measurements were achieved in an Aviv215 spectropolarimeter. Protein samples were exhaustively dialysed against 10 mM sodium phosphate (pH 7.5). Far-UV CD spectra were acquired from 260 to 180 nm through a cylindrical cell with a 0.02 cm path length. Ellipticity was recorded every 0.5 nm with an averaging time of 1 s per step. The final protein concentrations were 559 μg/ml, 546 μg/ml and 615 μg/ml for (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 respectively. To minimize the signal-to-noise ratio, five successive scans were averaged and the dialysis buffer baseline was subtracted. Resulting spectra were normalized to the protein concentration as the differential molar absorption coefficient Δ∊ per residue. The secondary structure was inferred by deconvolution of the spectra against a 22 proteins base [26], using the CDPro package ( Near-UV CD spectra were acquired from 350 to 250 nm with samples at 1.68 mg/ml, 1.82 mg/ml and 1.64 mg/ml for (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 respectively, in a 1 cm path-length rectangular cell. Five successive acquisitions were averaged and normalized to the baseline and protein concentration as the Δ∊ per chain.

Thermal stability of L. pneumophila NAT variants and chemical unfolding followed by CD

The first order inactivation rate constant kinact of each L. pneumophila NAT variant (100 μg/ml) incubated at 42°C was obtained by plotting the logarithm of the percentage of the residual activity against the incubation time, as described in [27], using the PNPA assay with 500 μM 4-AS. Fully unfolded protein [10 mM potassium phosphate buffer (pH 7.4)] at 0.1 mg/ml was obtained by adding GdmCl to a final concentration of 5 M. Fully unfolded and native proteins (0.1 mg/ml) were mixed in a final volume of 1 ml to increase the final GdmCl concentration by 0.1 M steps, while maintaining the final enzyme concentration at 0.1 mg/ml. Samples were kept at 25°C for 24 h to reach the unfolding equilibrium. Samples were then placed in a cylindrical cell with a 0.02 cm path length and ellipticity at 222 nm was recorded. The ellipticity transition at 222 nm was fitted to a three-state transition model [double isomerization, eqn (1), see also Figure 2] that best fit the experimental data: Embedded Image(1) where C is the concentration of GdmCl.

Variation of ellipticity was normalized to the total amplitude difference as explained by eqn (2). Embedded Image(2) where Yo is the experimental ellipticity, and Yu and Yn correspond to the ellipticity of the fully unfolded and native proteins respectively.

Construction of L. pneumophila Paris Δnat strain and Δnat pBCKS-lpp2516 complemented strain

The Δlpp2516nat) deletion mutant was constructed as described previously [28]. Briefly, Δnat was inactivated by introducing a CmR (chloramphenicol-resistance) cassette into the chromosomal gene. The chromosomal region containing lpp2516 was PCR-amplified with the primers lpp2516-for (5′-GAGCAACTACGTGCAAAACA-3′) and lpp2516-rev (5′-AAAATCGGAGCATACCATGA-3′). The PCR products were cloned into the pGEMT-easy vector (Promega). On this template, an inverse PCR was performed using primers lpp2516_inv.for (5′-CGGGATCCTGGAGATATTCCTTTAAGTCGATGGG-3′) and lpp2516_inv.rev (5′-CGGGATCCCCCCTTGATCAAGCGGAGTTAAA-3′), each containing a BamHI restriction site. The amplified pGEMT-easy backbone was BamHI-digested and ligated with the CmR cassette. The CmR cassette was PCR-amplified from plasmid pGEM-CmR using primers containing BamHI restriction sites at the ends. Ligation of both digested PCR fragments was performed. For chromosomal recombination, a PCR product of the lpp2516 gene interrupted by the CmR cassette was introduced into L. pneumophila strain Paris by natural competence as described previously [29]. Strains that had undergone allelic exchange were selected by plating on BCYE containing chloramphenicol, then confirmed by PCR and sequencing. The Δnat mutant was complemented (Δnat pBCKS-lpp2516) with the full-length lpp2516 gene carried by pBC KSkan, constructed by replacing the CmR cassette of plasmid pBC KS (Agilent) by a KanR (kanamycin-resistance) cassette. The KanR was amplified from pGEM-KanR [30] using the primers Kana+PacI-for (5′-AGTCAGTTAATTAAGGAGACTCCAGCATGAGAT-3′) and Kana+PacI-rev (5′-CGATCCTTAATTAAGATCCCGCTATCTGGACAA-3′) and was ligated to the plasmid pBC KS where the CmR cassette was deleted by amplifying the complete plasmid by inverse PCR using the primers pKS+PacI-for (5′-AGTCAGTTAATTAAAGGGCTTCCCGGTATCAAC-3′) and pKS+PacI-rev (5′-CGATCCTTAATTAAGTGCCCTTAAACGCCTGG-3′). The resulting plasmid was named pBC KSkan. Gene lpp2516 was amplified using the primers Nat_universel_FPCR (5′-GCAAAGGGATATGTAGGAGTT-3′) and Lpp2516end (5′-AAGCCAACACAAACCACTGC-3′) and ligated into pBC KSkan.

Determination of endogenous NAT activity in L. pneumophila lysates

Fresh exponential (A600~1.5) and post-exponential (A600~4) phase cultures (50 ml) of L. pneumophila strains Paris [wt (wild-type), Δnat and Δnat pBCKS-lpp2516], Lens and Philadelphia were resuspended in 400 μl of 25 mM Tris/HCl (pH 7.5), protease inhibitor cocktail, 2.5 mg/ml lysozyme and 0.1% Triton X-100, and sonicated for 10 s (2 s pulse on, 10 s pulse off) at 4°C. Proteins were reduced with 5 mM DTT and the concentration was estimated using the Bradford assay (Bio-Rad Laboratories). Typically, 75 μl of cell extract was mixed with 500 μM 2-AF (2-aminofluorene) and 500 μM AcCoA and the reaction was stopped with 15% (v/v) perchloric acid. Specific endogenous NAT activity was measured by quantifying the amount of 2-AAF (acetylated 2-AF) using HPLC (C18 column, Shimadzu) and normalizing to the protein concentration and AcCoA hydrolysis rate. Three independent experiments were performed for each L. pneumophila strain.

L. pneumophila aromatic amine sensitivity test

Aromatic amine sensitivity was determined with 2-AF as a model for toxic aromatic amine using a disc diffusion method. L. pneumophila strains Paris (wt, Δnat and Δnat pBCKS-lpp2516), Lens and Philadelphia were grown on BCYE agar plates at 37°C in the presence of the 2-AF or 2-AAF (10 and 100 mM) drop-deposited on paper disks. Toxicity (growth impairment) was evaluated by measuring the radius of the circle where no bacterial growth was possible around the paper disks.

Acanthamoeba castellanii infection by L. pneumophila

Infection of A. castellanii with L. pneumophila Paris wt and Δnat strains were performed as described previously [30]. In brief, 3-day-old cultures of A. castellanii were washed in ‘A.c. buffer’ (PYG 712 medium without proteose peptone, glucose and yeast extract) and adjusted to 105–106 cells per ml. Stationary-phase L. pneumophila grown on BCYE agar and diluted in water were mixed with A. castellanii at an MOI (multiplicity of infection) of 0.01. After invasion for 1 h at 37°C, the A. castellanii layer was washed twice, defining the starting point of the time-course experiment. The number of colony forming units of L. pneumophila was determined by plating on BCYE agar, and expressing as a function of time of infection.


The L. pneumophila arylamine NAT homologues encoded by strains Paris, Lens and Philadelphia exhibit unexpected strain-dependent sequence divergence

In the L. pneumophila strains Paris, Lens and Philadelphia, the genes lpp2516, lpl2369 and lpg2451 code for putative NAT homologues that were named (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 respectively, according to the NAT nomenclature [31]. These three L. pneumophila clinical strains and their genomes are well characterized [32]. (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 share 21–25% sequence identity with characterized NAT sequences (21–37% with all NATs of the UniProt database) (Supplementary Figure S2 at The catalytic triad shaped by residues Cys81–His124–Asp139 is conserved in the L. pneumophila NAT sequences. Furthermore, the two functional motifs found in all NAT homologues and known to be important for the catalytic activity are also present in the NAT sequences of the three L. pneumophila strains [7,33]. However, the L. pneumophila NAT sequence is significantly longer (322 amino acids) than those of known NAT enzymes (~290 amino acids). These sequence differences are particularly pronounced in domain III [7] as there is a C-terminal extension (domain III) present in L. pneumophila NATs. Unexpectedly, the NAT sequences of the three L. pneumophila strains were found to display different punctual substitutions at 16 different amino acid positions (Supplementary Figures S1 and S2). At the DNA sequence level, we identified 14 synonymous and 16 non-synonymous nucleotide substitutions (results not shown). The amino acid positions affected by these variations are mainly located in the predicted domains II and III, with six and nine variable positions respectively. Only the amino acid at position 172 differs among all three L. pneumophila NAT sequences, specifically tyrosine, histidine and glutamine, in (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 respectively. In addition to the Paris, Philadelphia and Lens strains, the analysis of four additional recently sequenced L. pneumophila genomes [16,32] identified new substitutions, so that a total of 26 varying amino acid positions exist between the seven L. pneumophila NAT sequences known to date (Supplementary Figure S1).

Phylogenetic analyses showed the separation of the majority of the NAT proteins analysed according to the different taxonomic groups of the organisms of origin (Supplementary Figure S3 at However the distribution of some of the proteins did not correspond to the expected vertical evolution, probably revealing horizontal interchange among species. This is the case for the clade containing the L. pneumophila NAT sequences which were found to cluster with eukaryotic sequences from Dictyostelium and Paramecium. Three other bacterial species also cluster in this clade. Interestingly, these organisms are, like Legionella, present in aquatic environments. This suggests that organisms living in the same environment contain similar NAT sequences [34]. Overall, the clustering of the L. pneumophila and Dictyostelium NAT sequences in one clade substantiate a horizontal transfer of the L. pneumophila nat gene from a eukaryotic host.

The three L. pneumophila NAT variants possess significant NAT activity, but display different catalytic properties

Recent studies on the bacterial RifF enzyme, a NAT homologue of the rifamycin-producing bacterium A. mediterranei, have proven that certain NAT homologues may be devoid of NAT activity despite possessing the characteristic functional motifs of typical NAT enzymes [7].

To investigate whether the L. pneumophila nat gene encodes a ‘true’ NAT enzyme, we cloned, expressed in E. coli and purified (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 proteins (Supplementary Figure S4A at The purified recombinant proteins were recognized by antibodies raised against the NAT protein of S. typhimurium (Supplementary Figure S4B), further supporting their homology with other NAT proteins. We show in the present study that the three recombinant L. pneumophila NAT proteins indeed acetylate prototypic aromatic NAT substrates (including antibiotics and environmental pollutants) in an AcCoA- and time-dependent manner. The acetylation rates were within the order of magnitude reported for other eukaryotic and prokaryotic NATs [35,36]. However, large differences between the three L. pneumophila NAT variants were observed: specific activity rates varied from 2 to 263 nmol/min per mg for (LEGPA)NAT1, 15–5396 nmol/min per mg for (LEGPL)NAT1, and 19–406 nmol/min per mg for (LEGPH)NAT1 (Supplementary Figure S5 at Thus, to determine whether L. pneumophila NAT variants display different catalytic properties, we performed steady-state kinetic experiments. First, kinetic parameters were determined for the acetyl donors AcCoA and PNPA. The Km (app) values of (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 for AcCoA were 5694 μM, 252 μM and 614 μM respectively (results not shown). Compared with AcCoA, the Km (app) value for PNPA was 23-fold lower for (LEGPA)NAT1 (247 μM). For (LEGPL)NAT1 and (LEGPH)NAT1, the Km (app) values for PNPA were 2-fold higher as compared with AcCoA (601 and 1349 μM). Determination of the catalytic efficiency, kcat/Km (app), with AcCoA and PNPA revealed that, in contrast with the two other variants, for (LEGPA)NAT1 PNPA is a far better acetyl donor than AcCoA (Table 1).

View this table:
Table 1 Kinetic parameters for different acyl donors

Catalytic efficiency parameters [kcat/Km (app), M−1·s−1] for PNPA, AcCoA and propionyl-CoA were determined as described in the Experimental section using 500 μM HDZ as the aromatic amine substrate. For each L. pneumophila variant, triplicate experiments were performed. Kinetic parameters were obtained by non-linear fitting to the Michaelis–Menten equation. Values are means±S.E.M. (n=3).

Next, catalytic properties of the three L. pneumophila NAT variants were further investigated by determining the Michaelis–Menten parameter kcat/Km (app) towards 11 prototypic aromatic amine substrates. L. pneumophila NAT variants tended to more efficiently catalyse the acetylation of arylamine substrates and aniline derivatives (Table 2). Measured kcat/Km (app) values ranged from 7 M−1·s−1 to 1×106 M−1·s−1. More importantly, large differences were observed between the three L. pneumophila NAT variants, with (LEGPH)NAT1 being systematically less efficient (from 1.4- to 62-fold) compared with the other variants. In particular, (LEGPL)NAT1 had a strikingly high kcat/Km (app) value for 5-AS (1×106 M−1·s−1) compared with (LEGPA)NAT1 (0.6×105 M−1·s−1) and (LEGPH)NAT1 (0.2×105 M−1·s−1). Similar data were obtained with HDZ. Efficient acetylation of the folate catabolite pABA (p-aminobenzoic acid) is rare in prokaryotes and has only been reported for Pseudomonas aeruginosa NAT1 [36]. L. pneumophila NAT variants appear to efficiently acetylate pABA, with kcat/Km (app) values ranging from 4211 M−1·s−1 for (LEGPH)NAT1 to 23675 M−1·s−1 for (LEGPL)NAT1 (Table 2). Taken together, these data demonstrate that the amino acid substitutions impart the three L. pneumophila variants with differences in their catalytic properties.

View this table:
Table 2 Catalytic efficiency parameters of (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 for prototypic aromatic amine substrates

kcat/Km (app) (M−1·s−1) for the acetylation of 11 prototypic aromatic amine NAT substrates was determined in steady-state kinetic experiments with each L. pneumophila NAT variant. Kinetic parameters were obtained by non-linear fitting of initial velocities against the Michaelis–Menten equation. Experiments were performed in triplicate. Values are means±S.E.M. (n=3).

L. pneumophila NATs exhibit N-propionyltransferase activity in vitro

So far, M. tuberculosis NAT1 is the only NAT enzyme known to be able to transfer a propionyl group from propionyl-CoA to aromatic amine substrates [25]. As observed for the two acetyl donors PNPA and AcCoA, we found significant, albeit different, N-propionyltransferase activity for the three L. pneumophila NAT variants with catalytic efficiencies ranging from 321 M−1·s−1 for (LEGPA)NAT1 to 9586 M−1·s−1 for (LEGPL)NAT1 (Table 1). As propionyl-CoA is a product of cholesterol catabolism, the N-propionyltransferase activity of L. pneumophila NAT could play a role in the utilization and homoeostasis of this CoA species, as suggested previously [25].

CD and thermal stability analyses reveal differences in structural characteristics among the three L. pneumophila NAT variants

CD spectra were obtained to characterize the structural features of the NAT enzyme from the three L. pneumophila strains. Deconvolution of the far-UV spectra (Figure 1A) revealed a lower content of α-helices (18.9%) and a higher content of β-sheets (16.8%) in (LEGPA)NAT1 compared with (LEGPL)NAT1 (21.8% and 13.8%) and (LEGPH)NAT1 (21.3% and 13.7%). Similarly, near-UV spectra (Figure 1B) showed a higher ellipticity signal in the tryptophan zone (~295 nm) and a lower signal in the tyrosine zone (270–280 nm) for (LEGPA)NAT1, suggesting structural modifications in the near environment of these residues. These data indicate that the amino acid substitutions in the three L. pneumophila NAT variants affect their structure. Moreover, they emphasize the singular structural properties of the (LEGPA)NAT1 variant as compared with the two other variants. These observations are further supported by thermal inactivation kinetics and chemical unfolding experiments followed by CD (Figure 2). The first-order inactivation rate constants (kinact) of (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 at 42°C were 0.051 min−1, 0.011 min−1 and 0.011 min−1 respectively (corresponding to half-life values of 13.4 min and 63 min) (Figure 2A). These results are consistent with thermal unfolding experiments followed by CD, suggesting that (LEGPA)NAT1 undergoes thermal transitions at lower temperatures (Supplementary Table S1 at In parallel, GdmCl-induced unfolding experiments were performed on the three L. pneumophila NAT variants. As shown in Figure 2(B), (LEGPL)NAT1 and (LEGPH)NAT1 ellipticity transitions fit eqn (1) (r2=0.999) with two GdmCl transition concentrations (C1/2) of 0.609 M and 2.775 M, and 0.678 M and 2.999 M respectively. For (LEGPA)NAT1, unexpected ellipticity data that did not fit eqn (1) were recorded for low GdmCl concentrations (0–1.5 M). So far, no clear explanation has emerged to described this phenomenon, which could be due, for instance, to protein aggregation. The ellipticity signal for concentrations above 1.5 M were adjusted to eqn (1) (r2=0.999) and displayed two C1/2 of 1.937 M and 2.899 M. These results are in agreement with CD analysis and indicate that amino acid variations effect the structure of L. pneumophila NAT variants, with (LEGPA)NAT1 behaving differently from the other two variants.

Figure 1 CD analysis of L. pneumophila NAT variants

(A) Secondary structure of L. pneumophila NAT variants was analysed by CD analysis in the far-UV spectrum (180–260 nm) and shown as the differential molar absorption coefficient (Δ∊ per residue). Deconvolution using a 22 protein base [27] was achieved to predict the percentage of α-helices (H), 3/10 helices (H3/10), β-sheets (S), polyproline helices (PP2), turns (Turn) and random coils (Unrd) in each protein (inset). This reveals an increased content of β-sheets and a higher content of α-helices in (LEGPA)NAT1 compared with the other L. pneumophila NAT variants. (B) CD spectra obtained in the near-UV spectrum (250–350 nm). (LEGPA)NAT1 shows a significantly different signal (Δ∊ per chain) in the tryptophan (~295 nm) and tryptophan/tyrosine (~280 nm) absorbance zone, compared with (LEGPL)NAT1 and (LEGPH)NAT1. Spectra are the average results of five successive measurements normalized to the protein concentration.

Figure 2 Thermal stability at 42°C and chemical unfolding

(A) Each L. pneumophila NAT variant was incubated at 100 μg/ml at 42°C for 30 min, and a PNPA assay was achieved every 5 min. Natural logarithm (ln) of residual activity (with 0 min incubation corresponding to 100% activity) was plotted against incubation time to determine the inactivation rate constant kinact by linear fitting. Half-life (t1/2) at 42°C was calculated as ln(2)/kinact. The S.D. is shown for triplicate experiments with (LEGPA)NAT1 (□), (LEGPL)NAT1 (○) or (LEGPH)NAT1 (Δ). (B) Ellipticity at 222 nm measured for 0.1 mg/ml final concentration of (LEGPA)NAT1 (□), (LEGPL)NAT1 (○) or (LEGPH)NAT1 (Δ) proteins incubated (25°C, 24 h) with increasing concentrations of GdmCl (0–5 M). The CD signal was normalized to the protein concentration and to the total amplitude difference. Non-linear fit of the experimental data against a three-state transition [eqn (1)] showed two GdmCl state-transition concentrations (C1/2,1 and C1/2,2) of 0.609 M and 2.775 M for (LEGPL)NAT1 and 0.678 M and 2.999 M for (LEGPH)NAT1. The particular (LEGPA)NAT1 behaviour towards chemical unfolding did not allow us to clearly establish the transition concentrations (see the Discussion). Eqn (1) was used for the data fitting.

L. pneumophila Paris, Lens and Philadelphia strains exhibit endogenous NAT activity towards aromatic amines

Endogenous NAT activity in the three L. pneumophila strains was measured by HPLC in lysates using 2-AF as the aromatic substrate. The three lysates were found to display N-acetylation activity towards 2-AF with rate values of 0.45±0.06, 0.23±0.05 and 0.31±0.04 pmol/min per mg for L. pneumophila Paris, Lens and Philadelphia strains respectively. To confirm that endogenous NAT activity was due to L. pneumophila NAT proteins, knock-out strains (Δnat) lacking the lpp2516 and lpg2451 genes were constructed in L. pneumophila Paris and Philadephia respectively. No knock-out mutant could be obtained for the Lens strain, which is less amenable to genetic manipulation. As expected, no endogenous NAT activity was detected in the lysates of the nat knock-out mutant strains. Complementation of the Paris mutant strain with a plasmid carrying the nat gene (Δnat pBCKS-lpp2516) restored NAT activity towards 2-AF with a rate of 1.86±0.62 pmol/min per mg. The higher activity detected in this complemented strain is likely to be due to the presence of multiple copies of the pBCKS plasmid carrying the nat gene. Overall, these data indicate that L. pneumophila NAT is endogenously functional and provides the L. pneumophila strains with the NAT-dependent aromatic amine N-acetylation pathway.

Endogenous NAT activity is involved in the detoxification of the toxic aromatic amine 2-AF in L. pneumophila strain Paris

To test whether the L. pneumophila NAT enzyme contributes to the detoxification of toxic aromatic amine chemicals in L. pneumophila, the Paris wt, Δnat and Δnat pBCKS-lpp2516 strains were grown in the presence of 2-AF, a model for toxic aromatic amines. As shown in Figure 3, in contrast with the wt strain, the growth and survival of the Paris Δnat strain was significantly affected in the presence of 2-AF. As expected, the complemented strain Δnat pBCKS-lpp2516 [which overexpresses (LEGPA)NAT1 enzyme] showed no growth inhibition zone, with an even better tolerance towards 2-AF than the wt strain at high concentrations of 2-AF (100 mM). Similar experiments carried out with 2-AAF showed no detectable effect on the growth of the three L. pneumophila strains tested (results not shown). This further supports the idea that L. pneumophila NAT activity detoxifies aromatic amine chemicals.

Figure 3 NAT is involved in the resistance of L. pneumophila to 2-AF toxicity

2-AF (10 mM and 100 mM) was drop-deposited on paper disks and disposed on BCYE agar plates spread with a layer of L. pneumophila Paris wt, Δnat and Δnat pBCKS-lpp2516 strains. Plates were incubated at 37°C. Toxicity was determined by measuring the radius of the growth inhibition zone around the paper disk. Experiments were performed in triplicate. A representative experiment is shown.

NAT activity does not seem to contribute to in vitro growth or intracellular replication of L. pneumophila

To test whether the NAT enzyme plays a role in growth, we analysed the growth of the L. pneumophila Paris wt and Δnat strains in BYE. No significant difference was found between the Δnat mutant and the wt strain (results not shown). We also analysed intracellular replication in the natural host A. castellanii. Again, no difference was observed between wt and the Δnat strain (Supplementary Figure S6 at These data suggest that the enzymatic function of the NAT protein in L. pneumophila is related to environmental persistence and survival rather than to infection.


Although L. pneumophila is exposed to several chemicals in its natural and man-made environments, the molecular mechanisms by which this bacterium detoxifies and/or disposes of toxic compounds remains poorly documented. In the present paper we report the molecular and functional characterization of a XME that biotransforms and detoxifies aromatic amine chemicals in three well-studied L. pneumophila strains (Paris, Lens and Philadelphia) [32]. We show that this enzyme is a member of the NAT family of XMEs, whose members are found in several eukaryotic and prokaryotic organisms. Comparative analysis indicated that the L. pneumophila nat gene is present in all seven L. pneumophila strains sequenced. We analysed the function of the NAT protein in three of the sequenced strains named Paris, Lens and Philadelphia. The genes coding for the NAT protein [(LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 respectively] share ~22% sequence identity with the other characterized NAT proteins. In addition, (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 possess the catalytic triad and the functional motifs found in all NAT homologues characterized to date [7], suggesting that they encode NAT function. Interestingly, in depth sequence analyses of the three L. pneumophila NAT proteins identified uncommon sequence/structure properties with respect to known bacterial NAT proteins [7]. This particularity of the L. pneumophila NATs is further underlined by the results from our phylogenetic analyses, which indicates that the L. pneumophila NAT proteins are phylogenetically closer related to eukaryotic NAT sequences than to prokaryotic ones. As shown in Supplementary Figure S3, the L. pneumophila NAT protein clusters with the NATs encoded by the amoeba Dictyostelium discoideum and the ciliate Paramecium teraurelia. In the environment L. pneumophila is parasitizing aquatic protozoa and ciliates where it replicates intracellularly [37]. Many eukaryotic-like proteins have been identified in the L. pneumophila genome sequence [15,38], thus interaction with aquatic protozoa seems to be a driving force in the evolution of this bacterium [28,39]. Similar to previous findings for other genes, our analyses suggest acquisition of the L. pneumophila nat gene from an eukaryotic host via horizontal transfer. The evolutionary history of the L. pneumophila NAT gene may explain, in part, its atypical sequence/structure features.

The L. pneumophila NAT protein is significantly longer (322 amino acids) than all of the other characterized members of the NAT family [7,11]. This atypical length is mainly due to the presence of extra amino acids at the C-terminus of the protein. Interestingly, it is known that the C-terminal region of NAT enzymes has a major role in substrate specificity and enzyme activity. More importantly, we found that the L. pneumophila nat gene displayed high sequence heterogeneity leading to strain-specific NAT variants. As a result, 16 amino acid positions were found to differ between (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 (Supplementary Figure S2). The existence of polymorphic NAT variants displaying different catalytic properties is well documented in humans and rodents, but not in prokaryotes [7]. The prokaryotic nat genes appear to be more conserved [7,14]. So far, only Mycobacterium tuberculosis NAT1 was shown to display strain-specific amino acid changes due to two SNPs [14]. However, these two variants were only found in a minority of strains. In the case of the L. pneumophila nat gene, analysis of the seven L. pneumophila strains sequenced (Supplementary Figure S1) indicated that each sequence is unique and differs at several amino acid positions.

The purified recombinant (LEGPA)NAT1, (LEGPL)NAT1 and (LEGPH)NAT1 proteins catalysed the AcCoA-dependent acetylation of all characteristic aromatic substrates of NATs tested in the present study (including arylamine drugs, pesticides and industrial chemicals derivatives), thus indicating that, despite their atypical sequence/structure features, the L. pneumophila NAT variants are ‘true’ functional NAT enzymes. Nonetheless, significant differences in substrate preferences and catalytic efficiencies were found between these variants (Tables 1 and 2, and Supplementary Figure S5). (LEGPH)NAT1 was the least efficient variant towards the 11 prototypic aromatic amine substrates tested. Depending on the arylamine substrate, fold-variations in acetylation efficiency [kcat/Km (app)] between the three variants ranged between 1.4 and 62. These data clearly indicate that the strain-specific amino acid variations in the sequence of the L. pneumophila NATs lead to variants with different acetylation properties. Surprisingly, pABA appeared as a good arylamine substrate, particularly for (LEGPA)NAT1 and (LEGPL)NAT1, with kcat/Km (app) values ranging from 4 to 23×103 M−1·s−1. Acetylation of this folate catabolite is considered as detrimental for de novo folate synthesis and thus for bacterial growth [36]. Nonetheless, efficient acetylation of pABA has been reported for the P. aeruginosa NAT enzyme [36]. Further studies are needed to ascertain whether acetylation of folate in certain bacteria, and in particular in L. pneumophila, is of biological importance.

The kinetic analyses of the three L. pneumophila NAT variants towards AcCoA (the physiological acetyl-donor) and PNPA (the non-physiological acetyl donor) further underlined the different catalytic properties of each enzyme variant (Table 1). Overall these analyses further support that the strain-dependent sequence variations affect the catalytic properties of the variants towards both AcCoA and aromatic amine substrates. Furthermore, the three L. pneumophila NAT variants have aromatic amine substrate profiles that are a mix between the profiles of characterized prokaryotic NAT [resembling (HUMAN)NAT2 enzyme] and the profile of (HUMAN)NAT1. Indeed, the three variants possess a significant activity towards INH and sulfonamide antibiotics [as for (HUMAN)NAT2], but also towards pABA and 4-aminosalicylic acid which are acetylated by (HUMAN)NAT1 [7]. Such a diversity in aromatic substrates may come from the unusual length of the C-terminus of L. pneumophila NAT: the C-terminal region of NAT enzymes is known to be in close proximity with the active-site cleft and to contribute to substrate preference [7]. The biophysical analyses of the three variants (Figures 1 and 2, and Supplementary Table S1) are in agreement with the kinetic analyses and indicate that the three L. pneumophila NAT variants have different structural properties, thus confirming the effect of the strain-dependent amino acid sequence variations. In particular, the kinetic and biophysical data suggest that (LEGPA)NAT1 has clear different structural and catalytic properties when compared with the two other variants. L. pneumophila is known to be able to replicate in many different protozoan hosts and to survive in different multispecies biofilms [40]. These interactions with multiple hosts and the possible contact with different bacteriocins or other toxic compounds in the diverse biofilm communities may have led to the variations in the NAT sequences leading to differences in its activity and thus allowing adaptation to different environments. Also other L. pneumophila proteins are known to display strain-dependent sequence variations, such as the F-box-encoding protein Lpp2082/AnkB, which is shorter in strain Paris than in strain Philadelphia [30].

The diversity of aromatic amine substrates that can be metabolized by L. pneumophila NAT variants in vitro is compatible with the many related aromatic chemicals to which this organism may be exposed in its different environments. The presence of functional NAT enzyme in lysates from the three strains was confirmed using 2-AF, a model of toxic aromatic amine chemicals. These data were in agreement with mRNA expression data [20,28] and confirm that the NAT-dependent acetylation pathway exists in L. pneumophila. The deletion of the nat gene of the Paris L. pneumophila strain led to impaired growth of the bacteria in the presence of 2-AF in contrast with wt and complemented strains. All strains were able to grow in the presence of 2-AAF, clearly indicating that the expression of the nat gene in L. pneumophila leads to the detoxification of 2-AF through its acetylation. So far, solid evidence implicating a nat gene in the detoxification and survival of toxic aromatic amine compounds potentially present in their environment has been reported for two species of filamentous fungi [41,42]. The present study suggests that the L. pneumophila NAT probably contributes to the environmental adaptation through biotransformation of aromatic chemicals. Although infection studies of the wt and nat knock-out strain in A. castellanii suggests that the enzyme is not directly involved in host infection mechanisms, we cannot rule out that L. pneumophila NAT plays an endogenous role. Interestingly, as shown for M. tuberculosis NAT, we found that the L. pneumophila enzyme displayed high N-propionyltransferase activity, suggesting that it could play a role in acyl-CoA metabolism and lipid synthesis [13]. Recently, Verdon et al. [43] revealed that L. pneumophila strains adapted to antimicrobial peptide Warnerycin K exhibited significant up-regulation of their nat gene and significant changes in fatty acid size and proportion. Ongoing experiments with the knock-out strains will help identify the potential endogenous role of the L. pneumophila NAT protein.

In conclusion, we have characterized, at the molecular and functional level, three variants of an acetyltransferase enzyme encoded by L. pneumophila and have shown that they detoxify toxic aromatic compounds and may thus contribute to chemical environment adaptation. These studies help in the understanding of the molecular mechanisms that contribute to L. pneumophila environmental adaptation.


Xavier Kubiak designed and performed experiments, analysed the results, prepared the Figures and wrote the paper. Delphine Dervins-Ravault designed and performed experiments and analysed the data on L. pneumophila. Benjamin Pluvinage was involved in the cloning and expression experiments. Alain Chaffotte did the CD experiments. Laura Gomez-Valero carried out the phylogenetic studies. Julien Dairou performed HPLC analysis. Florent Busi and Jean-Marie Dupret analysed the results and commented on the paper. Carmen Buchrieser designed the study, analysed the results and commented on the paper. Fernando Rodrigues-Lima conceived the project, supervised its development and wrote the paper.


This work was supported by the Université Paris Diderot Paris 7, DGA (Délégation Générale de l'Armement), CAMPLP (Caisse d'Assurance Maladies des Professions Libérales de Province), the Institut Pasteur, the CNRS (Centre National de la Recherche) and the Institut Carnot-Pasteur MI. X.K. is supported by a fellowship from the Université Paris Diderot-Paris 7. B.P. was supported by a fellowship from the DGA. L.G.V. was holder of a Roux post-doctoral research fellowship financed by the Institut Pasteur and subsequently with support from the FRM (Fondation pour la Recherche Médicale). C.B. was funded by Pasteur MI and the French Region Ile de France (DIM Malinf).


We thank Professor Edith Sim (Department of Pharmacology, University of Oxford, Oxford, U.K.) for helpful discussions. We acknowledge the technical plateform ‘Bioprofiler-UFLC’ (Unité de Biologie Fonctionnelle et Adaptative, Université Paris Diderot) and plateform ‘Biophysique des macromolecules et de leurs interactions’ for provision of HPLC and CD facilities respectively.

Abbreviations: 2-AAF, acetylated 2-aminofluorene; AcCoA, acetyl-CoA; ACES, N-(2-acetamido)-2-aminoethanesulfonic acid; 2-AF, 2-aminofluorene; 5-AS, 5-aminosalicylate; BCYE, buffered charcoal-yeast extract; BYE, buffered yeast extract broth; CmR, chloramphenicol-resistance; DTNB, 5,5′-dithiobis-(2-nitrobenzoic acid); DTT, dithiothreitol; GdmCl, guanidinium chloride; HDZ, hydralazine chloride; INH, isoniazid; KanR, kanamycin-resistance; NAT, N-acetyltransferase; pABA, p-aminobenzoic acid; PNPA, p-nitrophenylacetate; RifF, rifamycin amide synthase; SNP, single nucleotide polymorphism; TBS-T, Tris-buffered saline containing 0.2% Tween 20; wt, wild-type; XME, xenobiotic-metabolizing enzyme


View Abstract