Discovery of Catalases in Members of the Chlamydiales Order

Abstract

Catalase is an important virulence factor for survival in macrophages and other phagocytic cells. In Chlamydiaceae, no catalase had been described so far. With the sequencing and annotation of the full genomes of Chlamydia-related bacteria, the presence of different catalase-encoding genes has been documented. However, their distribution in the Chlamydiales order and the functionality of these catalases remain unknown. Phylogeny of chlamydial catalases was inferred using MrBayes, maximum likelihood, and maximum parsimony algorithms, allowing the description of three clade 3 and two clade 2 catalases. Only monofunctional catalases were found (no catalase-peroxidase or Mn-catalase). All presented a conserved catalytic domain and tertiary structure. Enzymatic activity of cloned chlamydial catalases was assessed by measuring hydrogen peroxide degradation. The catalases are enzymatically active with different efficiencies. The catalase of Parachlamydia acanthamoebae is the least efficient of all (its catalytic activity was 2 logs lower than that of Pseudomonas aeruginosa). Based on the phylogenetic analysis, we hypothesize that an ancestral class 2 catalase probably was present in the common ancestor of all current Chlamydiales but was retained only in Criblamydia sequanensis and Neochlamydia hartmannellae. The catalases of class 3, present in Estrella lausannensis and Parachlamydia acanthamoebae, probably were acquired by lateral gene transfer from Rhizobiales, whereas for Waddlia chondrophila they likely originated from Legionellales or Actinomycetales. The acquisition of catalases on several occasions in the Chlamydiales suggests the importance of this enzyme for the bacteria in their host environment.

INTRODUCTION

Macrophages are often a target of intracellular bacteria (1). The bacteria can be obligate intracellular bacteria, like Rickettsia spp. (2), or facultative intracellular bacteria, such as Legionella pneumophila (3), Brucella abortus (4), and Mycobacterium tuberculosis (5). Among members of the Chlamydiales order, Chlamydia trachomatis (serovar K9) does not resist macrophage microbicidal effectors (6), whereas Waddlia chondrophila is able to replicate very efficiently in macrophages (7, 8) and Parachlamydia acanthamoebae is able to replicate to a lower extent, rapidly inducing the apoptotic death of the macrophage (9–11).

The ability to grow in a professional phagocyte offers several advantages to the invading pathogen. First, macrophages represent an interesting cell target due to their presence in almost all tissues. Moreover, infecting macrophages give the bacteria an opportunity to hamper macrophage activation, which delays the development of an effective cytotoxic T cell response. Therefore, it is crucial to determine which factors define the resistance of macrophages to certain bacteria, especially when macrophages exhibited a different permissivity to bacteria belonging to the same order.

One of the first lines of defense of macrophages is the rapid degradation of the bacteria by acidic pH, lysosomal hydrolases, and various other microbicidal effectors, including reactive oxygen species (ROS) produced by the transmembranous NADPH oxidase complex (NOX2). Catalases are bacterial enzymes that degrade ROS (H₂O₂). Catalases belong to a very diverse functional class of proteins that can be classified in four main groups: the heme-containing monofunctional catalases, the heme-containing bifunctional catalase-peroxidases, nonheme catalases, and unclassified catalases (12). Mutations in certain components of the NOX2 complex cause an immune disease called chronic granulomatous disease (CGD) (13). This genetic disease is associated with recurrent bacterial and fungal infections. Interestingly, pathogens that infect CGD patients are expressing a ROS-degrading enzyme catalase (14) belonging to the heme-containing mono- or bifunctional catalase group.

Catalase-peroxidases are thought to have been acquired through lateral gene transfer from archaea (15), while monofunctional heme-containing catalases are believed to be very ancient. The latter can be further subdivided into three main clades.

During annotation of the genome of Waddlia chondrophila in our group, a catalase-encoding gene was identified (16). Since none of the members of the Chlamydiaceae family encode a catalase, we investigated the presence of these genes in other members of the Chlamydiales order, including Waddliaceae, Parachlamydiaceae, Simkaniaceae, and Criblamydiaceae families. The functional properties and evolutionary history of identified catalases were then assessed.

MATERIALS AND METHODS

Phylogenetic analysis.

BLASTP was performed on the NCBI BLAST platform (17). Protein sequences used for searches were retrieved from PubMed and Uniprot. Sequences were aligned with MUSCLE (18) in Geneious v6. Domains were determined with PROSITE (19, 20). The consensus sequence display was created with WebLogo (21).

Prior to phylogenetic analyses, the amino acid substitution model was assessed with ProtTest v.3.2 and modelgenerator (22). Phylogenetic analysis was done with the following models. Maximum likelihood (ML) was performed with the PhyML 3.0 platform using the LG substitution matrix, with invariant gamma distribution (4 categories) and 500 bootstraps (23). The maximum parsimony (MP) tree was obtained using the close-neighbor-interchange algorithm (24) with search level 1, in which the initial trees were obtained with the random addition of sequences (10 replicates) and 500 bootstraps. Evolutionary analyses were conducted in MEGA5 (25). A third tree was constructed with MrBayes using LG and 1,000,000 generations with invariant gamma distribution (4 categories) and frequency (26). The quality of the Bayesian phylogenetic tree was assessed by AWTY (J. C. Wilgenbusch, D. L. Warren, and D. L. Swofford, 2004 [http://ceb.csit.fsu.edu/awty]). Trees were visualized with FigTree v1.3.1.

Protein modeling.

Protein structures were modeled with 3DJIGSAW (28–30) and SWISS-MODEL Workspace (31–34). The quality of the models obtained with SWISS-MODEL Workspace was assessed with QMEAN4 (35, 36). The model then was further analyzed in Deepview (32). Other clade 3 catalases from P. mirabilis (Protein Data Bank [PDB] code 1M85) (37) and E. faecalis (PDB code 1SI8) (38) were used as controls. Root mean square (RMS), threading energy, and conformation clashes were determined with Deepview (32). For clade 2 catalases, Escherichia coli HPII PDB codes 1GGE (39) and 1IPH (40) were used for modeling.

Catalase cloning and expression.

Genomic DNA of W. chondrophila, P. acanthamoebae, E. lausannensis, and P. aeruginosa ATCC 27853 have been extracted with a Promega Wizard SV genomic DNA kit (Promega Corporation, Madison, WI) by following the manufacturer's instructions. The following primers were used: for W. chondrophila catalase, Wch_KatA_F (5′-CAC CAA AAG AGA TCG CCC AGC CACT-3′) and Wch_KatA_R (5′-TTA GGA GGA ACA GCC TGC TGC TTT TTT GAT TCGC-3′; P. acanthamoebae, Pac_katA_F (5′-CAC CGA GAA TAA AGA TAC GCT GAC CAC CA-3′) and Pac_KatA_R (5′-TCA GTT TTT ACG AGA GAG TAG GGCA-3′); E. lausannensis, Ela_KatA_F (5′-CAC CAC AGA TAA GCC CCC CCT AT-3′) and Ela_KatA_R (5′-CTA TTT TTT TCT CTT ATC CAG CGC TT-3′); P. aeruginosa, Pae_KatA_F (5′-CAC CGA AGA GAA GAC CCG CCT-3′) and Pae_KatA_R (5′-TCA GTC CAG CTT CAG GCC GAG-3′). The following PCR conditions were used for gene amplifications: initial incubation at 98°C for 30 s, followed by 35 cycles of 98°C for 10 s, 68°C for 30 s, and 72°C for 90 s, and an extension of 10 min at 72°C. The annealing temperature was lowered to 61°C for E. lausannensis and P. acanthamoebae catalase and to 60°C for P. aeruginosa. High-fidelity Phusion (Fermentas)-amplified genes were cloned in pET200 TOPO vector (Invitrogen). The expression of protein was done in BL21Star (Invitrogen). Purification of protein was performed in nondenaturing conditions with MagneHis (Promega). Purified proteins were concentrated in phosphate-buffered saline (PBS) with Amicon columns according to the manufacturer's instructions (Millipore). Protein concentrations were determined by fluorescence with a Qubit protein assay by following the manufacturer's instructions (Invitrogen).

Enzymatic activity.

Catalase activity was assessed by decrease in absorption of H₂O₂ at 240 nm due to degradation as described by Li and Schellhorn, with minor modifications (41). Briefly, hydrogen peroxide was used at a concentration of 10 mM. Purified proteins were used at a concentration of 50, 200, or 500 ng depending on the enzymatic activity. PureGrade 96-well plates (Brand, England) were used in a SynergyH1 microplate reader (BioTek) at 240 nm.

Elementary bodies of P. acanthamoebae, E. lausannensis, W. chondrophila, and S. negevensis grown in amoebae were purified according to Bertelli et al. (16), with a minor modification. Prior to gastrographin gradient ultracentrifugation, the bacteria were treated for 1 h at 37°C with DNase to reduce amoebal protein contamination. An aliquot of frozen bacteria then was washed twice with PBS before exposure to hydrogen peroxide. The bacterial concentration was determined by quantitative PCR (qPCR) with the pan-Chlamydiales primer pair and probe (42). A P. aeruginosa (ATCC 27853) overnight culture was washed twice with PBS and pelleted for 10 min at 8,000 × g at room temperature before exposure to hydrogen peroxide. Bacterial concentration was assessed by determining CFU counts. A. castellanii liquid axenic cultures were counted in a kovar chamber, and 10⁶ amoebae were pelleted at 500 × g for 5 min and washed twice with PBS prior to exposure to hydrogen peroxide.

Nucleotide sequence accession numbers.

Newly determined chlamydial catalase sequences were deposited in GenBank under accession numbers KC514601 to KC514620 (see Table S1 in the supplemental material).

RESULTS

Genetic and phylogenetic analyses of chlamydial catalases.

A prototypic small-subunit catalase from Staphylococcus aureus was used to perform a BLASTP search on all Chlamydiales genomes available, including unpublished, in-house, unfinished genomes of Criblamydia sequanensis, Estrella lausannensis, Protochlamydia naegleriophila, and Neochlamydia hartmannellae. Catalases were found in two members of the Criblamydiaceae family (C. sequanensis [KatE] and E. lausannensis [KatA]) and in two members of the Parachlamydiaceae family (P. acanthamoebae [KatA] and N. hartmannellae [KatE]), as well as the previously annotated KatA of Waddlia chondrophila. When using the catalase sequences of all of these members of the Chlamydiales order to perform additional BLASTP and PSIBLAST searches, no additional catalases were identified in any of the available genomes of Chlamydiaceae, Simkaniaceae, or Protochlamydia spp.

The amino acid sequence identity of the Chlamydia-related bacterial proteins compared to Staphylococcus aureus KatA and to the prototypical small-subunit Pseudomonas aeruginosa KatA ranged between 35 to 60% and 38 to 64%, respectively. Compared to catalases from clade 2, the identity to KatE of C. sequanensis reached 45 to 68%, except to KatE of N. hartmannellae (75.9%) (see Fig. S1A in the supplemental material). Identity among members of clade 3 reached around 54% sequence similarity, that between KatA of E. lausannensis and KatA of P. acanthamoebae was 77.6%, and that between KatA of W. chondrophila and L. longbeacheae was 77% (see Fig. S1B). These differences in identity led us to perform a detailed phylogenetic analysis of the chlamydial catalase proteins.

In addition to reference sequences used by Klotz et al. and Zamocky et al., we added the monofunctional catalases found in other amoeba-resisting bacteria (Legionellales, Mycobacteria, and Bradyrhizobiales) (12, 43, 44). Moreover, catalases of amoebal origin were added to determine a possible lateral gene transfer from the host. Sequences were directly retrieved by BLASTP from amoebal genomes available from NCBI (Dictyosteliida, Acanthamoeba, Tetrahymena, and Naegleria). For the amoebae belonging to the Jakobidae family, the sequences were reconstructed from expressed sequence tags (ESTs) retrieved by tBLASTn, since no genome was available at the time. Other amoebal sequences found in the EST database did not cover the whole protein sequence; therefore, they were excluded. Moreover, the three highest BLAST hits for each chlamydial catalase was added to the alignment as well. Phylogeny of the bacterial catalases was performed with maximum likelihood (ML), maximum parsimony (MP), and MrBayes (MB), all resulting in clear separation of the three catalase categories (i.e., clade 1, 2, and 3) (Fig. 1; also see Fig. S2 in the supplemental material). Sampled trees from MB analyses showed constant posterior probabilities, excluding any problem with convergence (see Fig. S1C).

Fig 1.

Gene environment and phylogenetic tree of typical bacterial catalases, including chlamydial catalases. (A) Representative tree of bacterial catalases, obtained from http://peroxibase.toulouse.inra.fr/, with catalase sequences of Chlamydiales and other amoeba-resisting bacteria. The evolutionary history was inferred by using MB (1,000,000 generations) and confirmed by ML (500 bootstraps) and MP (500 bootstraps). Only posterior probabilities below 1 are marked (red, 0.99 to 0.95; yellow, 0.95 to 0.9; green, 0.85 to 0.8; blue, 0.77 to 0.76). The bar indicates 0.1 amino acid substitutions per site. (B) The gene environment of the catalases present in the different Chlamydiales members exhibited no homology. Only W. chondrophila encodes transposases and integrases near the catalase location, suggesting that this catalase was obtained by lateral gene transfer.

Tree topology was conserved between ML and MB in clade 1 and clade 2, although with lower confidence in some nodes of ML (see Fig. S2A in the supplemental material). For MP, the Bacillales branch of clade 2 catalases clustered with Mycobacteria instead of branching off after E. coli, as occurs in the two other analyses. However, the phylogenetic relationships of Bacillales could not be accurately determined, since the bootstrap support was lower than 0.5 (see Fig. S2B). The catalases present in C. sequanensis and N. hartmannellae were assigned to the clade 2 catalases that comprise large tetrameric catalases. As previously observed, the negibacteria catalases are divided into 2 branches (12). KatE of C. sequanensis and KatE of N. hartmannellae clustered with the branch comprising the posibacterial lineage (Fig. 1A).

Amoebal clade 3 catalases all clustered together with a deep rooting compared to other bacterial clade 3 catalases. There were small differences in node placement between the three methods. This was mostly due to the inability of MP and ML to infer the phylogenetic relationship of a few sequences, as can be seen by the low bootstrap values (see Fig. S2A and B in the supplemental material). The chlamydial clade 3 catalases separate into two distinct branches with all three methods. Within the W. chondrophila KatA branch, five nodes had very low values for both MP and ML (see Fig. S2A and B). However, with MB, the phylogenetic relationship was determined with high posterior probabilities (Fig. 1A). E. lausannensis and P. acanthamoebae cluster in the same branch, but not together.

To better understand the origins of these different catalases in Chlamydiales, we analyzed their genetic environment and observed seven transposase and integrase elements immediately upstream of katA of W. chondrophila (Fig. 1B), as well as seven transposase and integrase elements 15 genes downstream of waddlial katA. None of the genes located between these transposases exhibited a BLASTP hit with an L. longbeacheae gene, precluding confirmation that katA was acquired from L. longbeacheae. However, the presence of these mobile elements supports the hypothesis of a horizontal acquisition of katA by W. chondrophila. No transposases were identified around the other catalases. Moreover, the genetic environment of the KatA-encoding gene from E. lausannensis and P. acanthamoebae was not conserved, despite these proteins exhibiting a sequence similarity of 78% and clustering together in phylogenetic trees. However, the synteny is low between members of different families of the Chlamydiales order (45), explaining the absence of the conserved genetic environment despite a likely common origin for E. lausannensis and P. acanthamoebae katA.

Conservation of domains and motifs.

The heme binding sites are conserved among all clades of monofunctional catalases (43). Therefore, we analyzed the sequences of the chlamydial catalases to determine motif conservation. Differences in prevalence of a given amino acid within the motifs were observed depending on the catalase clade. Amino acid variants that were mostly found in clade 2 catalases are highlighted in boldface in Fig. 2. The proximal heme-binding site was very conserved and was detected in all of the chlamydial catalases (Fig. 2E). The active sites in both sites were conserved (Fig. 2D and E, red boxes). For the distal heme binding site (Fig. 2D), analysis by PROSITE did not give any hit for KatA of P. acanthamoebae and KatA of E. lausannensis. This was due to two mutations at positions F44Q and S59W, respectively (Fig. 2D). The phenylalanine is very conserved, and its replacement with glutamine could strongly affect the binding site, especially its topology. The second substitution is also quite problematic, since tryptophan is more reactive and bulkier than serine.

Fig 2.

Surface epitopes and heme binding domains. Amino acids in boldface are differently conserved in large-subunit catalases. (A to C) Surface epitopes conserved in Chlamydiales, previously determined by in silico modeling of H. hepaticus (46); note the N304D and F306H mutations (arrows; Pseudomonas numbering). (D) Distal site of the prosthetic heme with catalytic histidine (red box); note the F44Q and S59W mutations (arrows). (E) Proximal site of the prosthetic heme with catalytic tyrosine (red box).

Moreover, the sequences were analyzed for the conservation of surface epitopes determined in a previous study (46). Since the consensus sequence for these epitopes was determined with only four sequences from small-subunit catalases, we further confirmed them by analyzing the MUSCLE alignment of all bacterial catalases obtained from http://peroxibase.toulouse.inra.fr/ (see Fig. S3 in the supplemental material). All three epitopes were conserved in all clades of catalases, with minor differences between small- and large-subunit catalases, except for the second epitope. The second epitope had a more conserved sequence in clade 2 catalases than clade 3. Moreover, the consensus sequences differed substantially between the two clades (see Fig. S3, orange, green, and blue boxes). The first epitope (Fig. 2A) was conserved in all Chlamydiales catalases. The last surface epitope (Fig. 2C) was well conserved in all Chlamydiae members except KatA of W. chondrophila, which presented two mutations in the more conserved sites (N304D and F306H). Moreover, at position 305, KatA of W. chondrophila had a phenylalanine, like large-subunit catalases.

Enzymatic activity.

Since Chlamydiales present two developmental stages, the activity of the catalase would be required during the early steps of infection, when the bacteria are still in their elementary body form. The amount of bacteria present in each vial was determined by qPCR with the pan-Chlamydiales primers. A total of 10⁷ purified elementary bodies of P. acanthamoebae, E. lausannensis, C. sequanensis, N. hartmannellae, and W. chondrophila, all encoding a catalase, were exposed to 0.2 M hydrogen peroxide (Fig. 3A). All of them produced oxygen bubbles to an extent similar to that of the P. aeruginosa positive control. However, N. hartmannellae and W. chondrophila displayed reduced oxygen bubble formation. A mock negative control of the amoebal coculture was also tested and proved to be negative. Simkania negevensis was used as a second negative control, since it does not encode a catalase. Therefore, as expected, it only produced one small oxygen bubble. This could be due to a slight contamination of purified elementary bodies by Acanthamoebae castellanii catalase. However, the contamination is insignificant compared to the catalase activity displayed by the different bacteria. The clade 3 catalase of A. castellanii was also blocked by 0.1 M azide. The activity of the catalase was blocked in all bacteria with 0.1 M azide, a noncompetitive inhibitor of catalases (47). Only when the concentration was reduced to 1 mM did we again observe oxygen formation for P. aeruginosa bacteria, while E. lausannensis showed no catalase activity even at 0.1 mM azide (Fig. 3A).

Fig 3.

Enzymatic activity of chlamydial catalases in vitro. (A) Elementary bodies (EBs; 10⁷) of Chlamydiales and P. aeruginosa were exposed to 0.2 M H₂O₂. Catalase activity is blocked by 0.1 M azide. (B) Degradation of hydrogen peroxide by purified catalase was followed by a decrease in absorbance at 240 nm. Catalase of P. acanthamoebae exhibited the weakest activity. The activity of the enzymes was conserved even when the pH was lower. Results are means from at least six independent experiments with standard errors of the means for each condition and protein. Hsp60 of W. chondrophila was used as a negative control. (C) Catalytic units derived from a decrease in absorbance are in M⁻¹ · s⁻¹.

Once we observed this oxygen production, we determined the catalytic activity of these catalases at various pHs in order to define whether the activity correlates with host range and phagolysosome survival. As a positive control, we cloned KatA from P. aeruginosa. To control for eventual contamination by the Escherichia coli catalase, we used as a negative control Hsp60 from W. chondrophila purified under the same conditions as the catalases. The enzymatic activity was assessed by measuring the absorption of H₂O₂. For KatA of P. acanthamoebae and KatA of W. chondrophila, we had to increase the amount of protein from 200 ng/ml to 2 μg/ml and 800 ng/ml, respectively, to detect a sustained degradation of hydrogen peroxide (Fig. 3B). The catalytic units (Fig. 3C) derived from the assay differed quite substantially depending on the species. KatA of P. acanthamoebae had the least ability to degrade hydrogen peroxide. KatA of E. lausannensis was more efficient at pH 7.0, whereas KatA of W. chondrophila appeared to be less sensitive to changes in pH.

In silico modeling of chlamydial catalases.

Several crystal structures exist for bacterial clade 2 and 3 catalases. Therefore, we used this information to build an in silico model of the chlamydial catalases using 3DJIGSAW (28–30) and SWISS-MODEL Workspace (31–34). The tetrameric structure of the clade 3 catalases was built on the crystal structure of Enterobacter faecalis (1SI8) (see Fig. S4A in the supplemental material) for all three chlamydial catalases. The organization of the tetramer was retained for all chlamydial small-subunit catalases (see Fig. S4A). For the chlamydial large-subunit catalases, the crystal structure of E. coli HPII (1IPH) was used (see Fig. S4B). The organization of the tetramer of KatE of N. hartmannellae catalase was more similar to HPII from E. coli than KatE of C. sequanensis (see Fig. S4B).

To determine the conservation of the tertiary structure of the tetramer subunits, the RMS values for each chlamydial catalase were determined. Overall, the tertiary structures were well conserved, with only limited deviations in the loop regions (Fig. 4A to C; also see Fig. S4C in the supplemental material). KatE of C. sequanensis presented one helix with more than a 5-Å deviation from the crystal structure. However, this part of the subunit is involved in neither the contact site of the tetramer nor in the catalytic domain (see Fig. S4C, left). Moreover, the quality of the tertiary structure was further determined by looking at residues with bad backbone conformation and side chains without hydrogen bonds. There were 24 residues in the C terminus with clashes in the backbone. Of these 24, 15 are mutated compared to E. coli HPII and 4 were not aligned. The remaining residues were close to mutated residues that caused the distortion of the backbone (see Fig. S5A). Moreover, in the C terminal of KatE of C. sequanensis, nine residues were mutated to prolines that strongly influenced the backbone orientation. Only three prolines (P383, P522, and P543) were not in an optimal conformation. As for KatE of C. sequanensis, most of the residues (11) that clashed with the backbone were located in the C-terminal part of the protein.

Fig 4.

In silico modeling of small-subunit chlamydial catalases. Tertiary structure of small-subunit chlamydial catalases marked according to RMS values. (A) W. chondrophila KatA tertiary structure is strongly conserved. (B and C) Tertiary structures of KatA of P. acanthamoebae and KatA of E. lausannensis are conserved, except in some loops (arrows). (D) Catalytic site of H. pylori (white; PDB code 2IQF), P. mirabilis (salmon; PDB code 1M85), and E. faecalis (yellow, PDB code 1SI8). Numbering is according to that for H. pylori. (E) Catalytic site of W. chondrophila KatA (purple). All residues are conserved, except R53 (arrow). (F) Catalytic site of KatA of P. acanthamoebae (red) and KatA of E. lausannensis (white). The orientation of the active-site residues is conserved, except for H198 of P. acanthamoebae (arrow) and N128 in both bacteria (arrow). Numbering is according to that of KatA of the P. acanthamoebae sequence. Water molecules are marked W1 to W3, with hydrogen bonds shown in dotted green lines. The heme of H. pylori was introduced into the model structure of the small-subunit chlamydial catalases.

Chlamydial small-subunit catalases had no side chains that lacked hydrogen bonds. In KatA of P. acanthamoebae, KatA of E. lausannensis, and KatA of W. chondrophila there were only five residues, of which 3 were mutated compared to the Helicobacter pylori (2IQF) catalase sequence (see Fig. S4B in the supplemental material). The mutations were not all located at the same residues. Moreover, KatA of P. acanthamoebae catalase had one proline mutation (Q379P), KatA of W. chondrophila had two (V6P and V55P), and KatA of E. lausannensis had one (V6P) that affected the backbone organization.

The water molecules for both catalases were added to the catalytic site residues according to Diaz et al. (48) (Fig. 4A). For the small-subunit catalase, the residue orientation is almost completely conserved. For KatA of W. chondrophila R53, the side chain is in another orientation that replaces the hydrogen bond with the heme with the water molecule (W2) (Fig. 4B). The rotation of N128 in KatA of P. acanthamoebae and KatA of E. lausannensis caused a weakening of the interaction with the water molecule (3.2 Å) of the nitrogen of the histidine ring and the oxygen of the asparagine side chain (Fig. 4C). Moreover, the interaction most likely occurs with the oxygen and not the nitrogen of the asparagine side chain, like in H. pylori. In KatA of P. acanthamoebae, the slight tilt of H53 further increased the distance to the water molecule, weakening the interaction. All other hydrogen bonds were in the range of moderate interactions (2.5 to 3.2 Å). The flipping of the H198 side chain in KatA of P. acanthamoebae disrupted the hydrogen bond chain between the catalytic residues Y338, R334, H198, and N328. Indeed, the distances passing 4 Å did not allow for hydrogen bond formation. This could affect the stability of the catalytic site. For chlamydial clade 2 catalases, the catalytic site was also conserved, and all hydrogen bonds could be formed (see Fig. S4D in the supplemental material). Only Y415 in KatE of C. sequanensis had a 14° shift of the aromatic ring, lengthening the distance between the tyrosine and the iron of the heme from 2.11 to 2.35 Å. However, the tyrosine was still close enough to interact with the iron.

DISCUSSION

In this work, we identified catalase-encoding genes in the genomes of five different Chlamydia-related bacteria, and we could demonstrate their enzymatic activity and a significant conservation of their active sites. These catalases belonged to clade 2 (KatE of N. hartmannellae and KatE of C. sequanensis) and clade 3 (KatA of W. chondrophila, KatA of E. lausannensis, and KatA of P. acanthamoebae) catalases. Moreover, waddlial catalase was branching far from the other two clade 3 catalases, encoded by P. acanthamoebae and E. lausannensis, suggesting a complex evolutionary history of these proteins. The limited amount of sequences available for chlamydial catalases did not allow us to determine definitely which one is most likely to be the ancestral Chlamydiales catalase (Fig. 5A). However, according to phylogenetic analyses performed by Klotz et al., the first catalase was likely a large-subunit catalase that underwent gene duplication with loss of the C-terminal domain and partial loss at the N-terminal domain (12). Thus, it is likely that the common ancestor of the Chlamydiales order had a large-subunit catalase that was lost by most of the species. Moreover, lateral gene transfer of large-subunit catalases was only observed from bacteria to archaea and not among bacteria (12).

Fig 5.

Model of catalase acquisition and loss in Chlamydiales. (A) MrBayes phylogeny of fully sequenced Chlamydiales based on concatenation of GyrA, GyrB, RpoA, RpoB, RecA, SecY, TufA, and TopA amino acid sequences. Only nodes with posterior probability inferior to 1 are marked. The tree constructed with ML placed the Criblamydiaceae node before that for Waddliaceae. Families are delimited by colored boxes. Distribution of catalases in the Chlamydiales order is depicted in the column. (B) Two hypotheses for lateral transfer of clade 3 catalase in P. acanthamoebae and E. lausannensis. Transfer from a Rhizobiales member to one of the two chlamydial species and then internal transfer (1) or transfer from a Rhizobiales member to both species at about the same time (2). (C) Catalase of W. chondrophila could originate from a lateral transfer from Legionellales (1) or an Actinomycetales isolate (2).

The clade 3 catalase shared by E. lausannensis and P. acanthamoebae was probably acquired later, since this enzymatic clade originated later in posibacteria and was then distributed to other species by lateral gene transfer (12). This transfer in Chlamydiales likely has taken place after the separation of the Criblamydiaceae/Parachlamydiaceae ancestor from the Waddliaceae, since they encode clade 3 catalases of different origins (Fig. 5A). Since E. lausannensis and P. acanthamoebae were clustering in the same branch but not together, it is more likely that they both received the catalase from a Rhizobiales member on a different occasion than an internal transfer from one to the other after lateral gene transfer from a nonchlamydial bacteria (Fig. 5B). The transfer of the catalase present in E. lausannensis and P. acanthamoebae probably occurred from a Rhizobiales precursor, since the majority of the first 30 BLASTP hits belong to this order. For W. chondrophila, the origin of the lateral gene transfer is probably an Actinomycetales or Legionellales isolate rather than one of the Planctomycetes BLAST best hits. This hypothesis is based on the fact that Legionellales isolates and several members of the Actinomycetales are able to survive within amoebae (44). Interestingly, host-associated genera, like Brucella and Bordetella, lost the large-subunit catalase upon acquisition of the clade 3 catalase. Such a loss could explain the absence of the clade 2 catalase from the chlamydia-related bacteria that encode a clade 3 catalase. As proposed by Klotz et al., this might be due to the presence of clade 3 catalases in eukaryotes that are forcing the bacteria to adapt to the physiological selective pressure present within the host (12). When looking at the most common human bacterial pathogens, the majority encode a catalase or a catalase-peroxidase. As seen for other anaerobic bacteria, Streptococcaceae do not possess a catalase, since their exposure to reactive oxygen species is much lower. Mycoplasma, Rickettsia rickettsii, Borrelia burgdorferi, Treponema pallidum, and Chlamydiaceae all underwent strong genome size reduction and adaptation to a specific niche, making the presence of a catalase redundant. Since the reservoir of Chlamydia-related bacteria is much broader, these bacteria need a more diverse panel of virulence factors to adapt to each host. For W. chondrophila and E. lausannensis, the presence of a catalase could indeed prove useful during the infection of humans when encountering professional phagocytes. Even though the members of the Chlamydiales order all share the same replicative cycle, the composition of the bacterial inclusion varies considerably between the Chlamydiaceae and Chlamydia-related bacteria. The inclusion of Chlamydiaceae escapes the endocytic pathway early and associates with Golgi proteins and membrane (49–51). In macrophages, W. chondrophila is only transiently associated with the endocytic pathway but avoids lysosomal fusion, as it is associated with the endoplasmic reticulum (8). Conversely, in macrophages, P. acanthamoebae does acquire LAMP-1 but prevents further maturation of the bacterial inclusion into a digestive lysosome (52). Perhaps due to these differences in the early intracellular trafficking of the Chlamydiales, W. chondrophila and P. acanthamoebae require a catalase to counteract initial exposure to ROS, whereas Chlamydiaceae do not. A better understanding of the trafficking of other Chlamydia-related bacteria, like Protochlamydia naegleriophila, which does not encode a catalase, and C. sequanensis, which encodes a clade 2 catalase, is essential to define the roles of these enzymes.

Despite significant differences at the amino acid sequence level, tertiary structure modeling showed significant conservation, especially around the catalytic sites. Histidine 198 of KatA of P. acanthamoebae probably destabilizes the interaction between the heme and the catalytic tyrosine due to a disruption of the hydrogen bond chain. This correlates with the reduced enzymatic activity of KatA of P. acanthamoebae observed in vitro. Such lower activity of KatA of P. acanthamoebae compared to that of KatA of W. chondrophila may affect the ability of these bacteria to grow within macrophages, since a rapid response to ROS production is crucial to prevent damage and activation of inflammatory signaling pathways. Indeed, W. chondrophila has a productive and rapid growth cycle in macrophages (7, 8), whereas P. acanthamoebae grows only poorly in macrophages (10, 11); this might partially be due to their different fates in early steps of intracellular trafficking. However, the presence of a catalase is not essential for growth in free-living amoebae, since Protochlamydia spp. do not encode any catalases but nevertheless grow successfully in these protists (53, 54). As revealed by phylogenetic analyses, the horizontal transfer of clade 3 catalases probably occurred on several occasions and after the divergence of the different families within the Chlamydia-related bacterial branch. Moreover, within the Parachlamydiaceae, the Protochlamydia spp. that do not encode any catalase diverged prior to acquisition of catalases by Parachlamydia spp.