Molecular modeling and in silico characterization of Mycobacterium tuberculosis TlyA: Possible misannotation of this tubercle bacilli-hemolysin

Background The TlyA protein has a controversial function as a virulence factor in Mycobacterium tuberculosis (M. tuberculosis). At present, its dual activity as hemolysin and RNA methyltransferase in M. tuberculosis has been indirectly proposed based on in vitro results. There is no evidence however for TlyA relevance in the survival of tubercle bacilli inside host cells or whether both activities are functionally linked. A thorough analysis of structure prediction for this mycobacterial protein in this study shows the need for reevaluating TlyA's function in virulence. Results Bioinformatics analysis of TlyA identified a ribosomal protein binding domain (S4 domain), located between residues 5 and 68 as well as an FtsJ-like methyltranferase domain encompassing residues 62 and 247, all of which have been previously described in translation machinery-associated proteins. Subcellular localization prediction showed that TlyA lacks a signal peptide and its hydrophobicity profile showed no evidence of transmembrane helices. These findings suggested that it may not be attached to the membrane, which is consistent with a cytoplasmic localization. Three-dimensional modeling of TlyA showed a consensus structure, having a common core formed by a six-stranded β-sheet between two α-helix layers, which is consistent with an RNA methyltransferase structure. Phylogenetic analyses showed high conservation of the tlyA gene among Mycobacterium species. Additionally, the nucleotide substitution rates suggested purifying selection during tlyA gene evolution and the absence of a common ancestor between TlyA proteins and bacterial pore-forming proteins. Conclusion Altogether, our manual in silico curation suggested that TlyA is involved in ribosomal biogenesis and that there is a functional annotation error regarding this protein family in several microbial and plant genomes, including the M. tuberculosis genome.


Background
Tuberculosis (TB) is an infectious disease that mainly afflicts populations in third-world countries. Although most infected people will never develop an active form of the disease, the global TB death rate is substantial, being around two million people per year. The situation has been further exacerbated by the emergence and spread of extremely resistant Mycobacterium tuberculosis strains (XDR) [1] which, together with the lack of effective antibiotics, urges the development of new alternatives for controlling this worldwide public health threat. Consequently, most research efforts have focused on understanding M. tuberculosis biology with the aim of identifying new therapeutic and vaccine targets.
Current management of TB cases consists of a twomonth intensive treatment comprising first-line drugs such as Rifampicin (RIF), Isoniazid (INH), Pyrazinamide (PZA) and Ethambutol (EMB)/Streptomycin (SM) which seeks to ensure that mono-resistant strains do not proliferate [2]. When RIF and INH are ineffective, standard treatment guidelines recommend a combination of second-line drugs [3]. Among these therapeutic choices, aminoglycosides such as Capreomycin (CMN) and Viomycin (VMN) are given in combination with other antibiotics to treat multidrug-resistant strains (MDR); they are also effective against non-replicating M. tuberculosis strains, as well as being useful in the treatment of latent TB infections [4]. CMN-and VMN-resistant bacilli are classified as XDR-TB strains and are a cause of major concern since these drugs have more toxic side effects and might result in higher death rates, especially among HIV-infected persons [1].
Controversially, M. tuberculosis tlyA gene product has been annotated both as a virulence factor, due to its ability to lyse red blood cells, as well as a protein involved in ribosome biogenesis [5,6].
TlyA has been assumed to induce hemolytic activity due to its high amino acid similarity (37.3%) to the poreforming hemolysin/cytotoxin virulence determinant from Serpulina hyodysenteriae. Notably such activity was only experimentally demonstrated with bacterial lysates [6] and another study has even suggested that TlyA could be associated with hemolysin expression in Escherichia coli [7,8]. Additionally, recent study showed that a concentration-dependent hemolysis of rabbit and human erythrocytes to be induced on incubation with the recombinant Rv1694 protein (the putative M. tuberculosis TlyA) when over-expressed in E. coli [8]. The article reported hemolysis inhibition using specific antiserum against the same protein as well as the presence of the recombinant protein on E. coli cell wall; these data were supportive of TlyA's hemolytic property, combined with a preliminary in silico analysis. Rv1694 oligomers were also observed on lysed erythrocyte membranes, as well as the susceptibility of Rv1694-expressing E. coli to CMN [8]. Despite such thorough in vitro analysis in E. coli, the functional activities suggested for Rv1694 have still not been clearly demonstrated when using M. tuberculosis mutant strains.
By contrast with the proposed hemolytic function, Johansen et al., [5] have reported that M. tuberculosis H37Rv and Beijing TlyA confer susceptibility to CMN and VMN. These antibiotics are structurally similar and share the same mechanism of action, both inhibiting bacterial proliferation by blocking the peptidyl transferase reaction. VMN hinders A-site and P-site tRNA positioning on the 50S ribosomal subunit, therefore hampering mRNA translation [9]. CMN resistance arises due to mutations in the tlyA gene which encodes a RNA 2'-Omethyltransferase (2'-O-MTase) that methylates ribose moieties from nucleotide C1409 in 16S rRNA and C1920 in 23S rRNA. The absence of these rRNA methylations confers resistance against CMN and VMN given that binding sites for these antibiotics are delineated by the localization of such methylated riboses [5]. Additional evidence has been provided by mutant complementation, showing that CMN susceptibility is restored by introducing an active copy of the tlyA gene [10]. TlyA methylation clearly enhances CMN and VMN susceptibility and is responsible, at least in part, for their efficacy against mycobacterial infections as drug binding is impaired when the enzyme is not produced. Resistance thus emerges when TlyA activity is lost. These studies show the need for a reevaluation of TlyA's function in virulence [5,11].

TlyA related sequences
The BLAST search for TlyA-related sequences in the TB and Genbank databases showed that although TlyArelated sequences are found mainly in prokaryotic organisms, homologous protein sequences are also present in eukaryotic organisms such as plants and algae ( Figure 1A).
These M. tuberculosis-related TlyA protein sequences are mainly classified as hemolysin A inside the FtsJ-like class based on domain similarity and have been annotated as such in Genbank, EMBL, UniProt and SWISS-PROT databases according to sequence similarity with Serpulina hyodysenteriae TlyA annotated as being a hemolysin. Paradoxically, although hemolysin A contains similar domain architecture to TlyA, it is predicted to be a rRNA methylase instead of a contact-dependent hemolysin (Table S1, Additional file 1).
TlyA amino acid sequences from M. tuberculosis H37Rv, H37Ra, CDC1551, M. bovis and M. bovis-BCG reference strain alignment showed a 100% identity with the Harleem, F11 and C M. tuberculosis strains ( Figure 1B). The same conservation pattern was obtained when the corresponding nucleotide sequences were aligned (> 99% identity).
TlyA transcriptional unit revealed some rearrangements in the Mycobacterium IprJ-recN operon  Figure 2) it was evident that almost all gene products, except for tlyA, are involved in intermediary metabolism, cell wall biosynthesis and signal transduction rather than virulence. Moreover, no trafficking proteins were found within the operon or near tlyA that could be suggestive of TlyA transportation either to the cell membrane or the extracellular milieu.
The IprJ-recN operon also had common gene organization in M. tuberculosis and M. bovis, while in M. leprae most genes have lost their function (i.e. have become pseudogenes). Such high rate of loss of function in genes involved in intermediary metabolism, cellular respiration and cell-wall biosynthesis may suggest that the loss of such genes confers an adaptive advantage on Mycobacterium during host cell infection or transmission [12]. No other genes reported to be involved in bacilli virulence were found within this operon.

TlyA three-dimensional structure
The three-dimensional structure of a putative hemolysin from Streptococcus thermophilus (PDB ID: 3HP7 chain A, at 1.53 Å resolution) was used as template for homology modeling. The obtained identity (37.2%, RMSD 0.44 Å, E-value 4.7e-26) suggested that the selected hemolysin structure was a suitable template for TlyA, and that the obtained modeled structure could reflect an experimentally-obtained M. tuberculosis TlyA structure.
Although this protein seemed to be a hemolysin, its domain architecture was also similar to RNA methyltransferase family, fibrillarin homologues and TlyA proteins. A consensus sequence was established which showed the presence of a common core, comprising six β-sheets, the first five of which were found parallel and the sixth anti-parallel between two layers of α-helices and random coiled regions. Such structural organization is commonly referred as to an AdoMet-dependent methyltransferase (MTase) fold [13,14].
TlyA has an overall globular fold and comprises two domains exhibiting different spatial arrangements. The smaller domain, named S4, is N-terminally located and is overlapped by the catalytic domain which exhibits structural similarity to various unrelated RNA-binding proteins, while the large catalytic domain, named FtsJlike methyltransferase domain is common to 2'-O-MTases and exhibits a α/β fold with a deep pocket ( Figure 3).
The S4 domain consists of 64 residues, starting at A-5 and ending at H-68. This domain is structurally formed by one β-sheet and three α-helices predominantly composed of positively charged residues (11 out of 47 residues) whose probable function is to mediate RNA binding and provide stability. In line with this, the RNA-BindR server identified three short sequences (1-MAR-RAR-6, 13-RRGLARSRQQ-22 and 31-KVR-33) inside Figure 2 tlyA transcriptional unit gene syntheny. Genes are represented as arrows and are drawn according to transcriptional orientation and genome functional annotation. Arrow color representation for gene functional annotation is as follows: green for cell wall synthesis, red for signal transduction, brown for evolutionarily conserved, yellow for intermediary metabolism, white for tlyA, pink for pseudogenes, purple for PE/ PPE and blue for IS/phages. the S4 domain that might be implicated in RNA binding ( Figure 3). The S4 domain has been specifically identified in some bacterial and eukaryotic ribosomal proteins, pseudouridine synthases, RNA methylases, bacterial tyrosyl-tRNA synthetases and may also be involved in translation regulation [15].
The analysis of the protein's electrostatic surface showed that positively charged residues were localized towards the N-terminal extreme, forming bulges and long finger-like projections that probably extend into the rRNA core to stabilize its structure ( Figure 4). In fact, charged residues like R and K exhibited the highest interface propensities, which is consistent with their ability to participate in interactions both with nucleotide bases and the negatively charged RNA phosphate backbone [16].
The FtsJ-like methyltransferase domain is located between residues 62 and 247 and includes the catalytic tetrad responsible for TlyA's enzymatic activity [14]. Moreover, the FtsJ-like methyltransferase domain exhibits a Rossmann fold composed of five parallel β-strand layers linked by two α-helices following an α/β secondary structure topological order [17].
Our comparative analysis of TlyA with the Fibrillarin family showed the presence of a GASTG AdoMet binding motif comprising residues 90-94 and a D located at position 112 juxtaposed to the above mentioned motif ( Figure  3A), which might stabilize TlyA coenzyme binding [13]. AdoMet was located between β-strands 4 and 5 within the inner depression inside the catalytic pocket formed by GASTG residues' main-chain NH groups ( Figure 3B). Moreover, substrate positioning might be driven by the Rossmann configuration fold and the pertinent AdoMetbinding/catalytic sites, suggesting substrate selectivity in the small ribosomal subunit [17]. Likewise, domain structure was in agreement with S-adenosyl-methionine (AdoMet)-dependent methyltransferases. AdoMet binding and affinity might indicate how an enzyme such as TlyA may methylate the two adjacent ribose rings on both ribosomal subunits. In this reaction S-adenosyl-methionine (SAM) acts as methyl donor, as it is converted into S-adenosyl-l-homocysteine (SAH); TlyA could thus be also classified as being an RNA-AdoMet-MTase class I [18]. This hypothesis is supported by the high structural similarity (98.1%) between TlyA and FtsJ RNA-MTase complexed with AdoMet (ligand-binding template PDB entry 1EJ0) (RMSD 2.98, Similarity score: 368.77 Evalue: 1.30E-06) ( Figure S1, Additional file 2). Two expected structural motifs were also identified in this structure which was formed by six buried residues between G-90 and F-96 which are pivotal for AdoMet interaction and therefore conserved in Mycobacterium TlyA proteins ( Table 1). The TlyA model was validated by using the RAM-PAGE server which evaluates a protein's 3D-structure based on the stereochemistry quality of torsion angles and geometry [19]. The Ramachandran plot showed 95.1% feasibility (253 aa) in favorable positions and 3.8% (10 residues) in acceptable regions, with just 1.1% in an outlier region (3 residues), supporting the high quality of the 3D-model ( Figure S2, Additional file 3).
This in silico structure analysis of M. tuberculosis proteins allowed a deeper understanding of TlyA's function and challenges the initially described function of TlyArelated proteins. Rather than its hypothetical function as a hemolysin, our data suggested functions as an MTase based on in silico prediction.

Predicting subcellular localization
Subcellular localization was predicted with TBpred [http://www.imtech.res.in/raghava/tbpred/] and the results suggested TlyA localization in the cytoplasmatic compartment (Table 2), also supporting the fact that TlyA acts as a methyltransferase [5]. A recent mycobacterial protein analysis led to ascertaining several predictors' reliability for subcellular localization, providing highly accurate results for Gpos-PLoc and PA-SUB v.2.5 and SignalP 2.0 as a confirmatory tool [20]. However, in this study, PA-SUB v.2.5 predicted TlyA as being an extracellular protein (72.84%), which could have been biased as the training strategy used involved results from the SWISS-PROT [21] database where TlyA is mis-annotated. Likewise, in agreement with BLAST scores for the query sequence (Rv1694), 26 out of 28 homologous proteins were designated as putative ribosomal RNA methyltransferases and RNA binding proteins. Only one submitted sequence seemed to be similar to hemolysin A (the S. hyodysenteriae sequence).  Pattern branching of the phylogenetic tree showed that the TlyA protein formed a monophyletic cluster with AdoMet RNA MTases, suggesting a common evolutionary origin; these findings might also imply a similar functional role in translational machinery. PFP formed a distant cluster where diverse proteins were clustered on a branch with the most divergent proteins ( Figure 5).
The synonymous and non-synonymous substitutions were calculated to identify the action of neutral, positive or purifying selection on tlyA genes; their synonymous, Pi(a), and non-synonymous diversity, Pi(s), was 0.672 and 0.132, respectively, corresponding to a 0.196 Pi(a)/Pi (s) rate which can be inferred as purifying selection acting on this gene among Mycobacterium species. This suggests that conservation at sequence level might show  Mycobacterium tlyA gene homogeneity and show low inter-specific variation between the species compared in this study.

Discussion
The aim of the present study was to describe structural and phylogenetic features of M. tuberculosis TlyA, using bioinformatics tools. The structural and phylogenetic analysis shown here highlights the need for re-annotating this protein family and redefining it as an RNA-binding FtsJ-like methyltransferase in bioinformatics databases; this is based on the high degree of amino acid sequence similarity found in the Mycobacterium species analyzed here. It can be surmised that physical organization, domain spatial arrangement and protein folding on TlyA is structurally conserved. Previous sequence analysis of 2'-O-MTase families has indicated that they use a similar catalytic mechanism and have inherited a common function from the same ancestral ribose 2'-O-MTase. Feder et al., suggested that TlyA is a 2'-O-MTase, identifying a catalytic tetrad consisting of lysine-69, aspartic acid-154, lysine-182 and glutamic acid in position 238 (K69-D154-K182-E238) [14]. Based on amino acid sequence alignment, our comparative analysis with the RrmJ family and other related RNA 2'-O-MTases [13,14,18] showed changes in D154-K182-E238 residue positions (Figure 1). However, despite being located in different sequence positions, the scaffold formed by these residues was maintained at the catalytic site in the modeled protein 3D structure without affecting protein function [25].
Despite the diversity of existing MTases, the majority maintain the same catalytic mechanism where K-182 binds to a hydrogen from the ribose 2-OH' group; this nucleophilic attack generates a transition state (S N 2-like state). The lysine side chain shifts E-238 so that its negatively charged carboxyl group becomes available to promote the attack of the methyl donor [18]; furthermore, this conserved amino acid residue pattern was also observed in our evolutionary trace analysis.
TlyA three-dimensional structure resembled an RNA methyltransferase The S. hyodysenteriae TlyA postulated mechanism for hemolysin action is said to be similar to that for poreforming protein (PFP) mechanisms [6]; TlyA folding, however, is inconsistent with PFPs due its lack of membrane-binding domains, cholesterol-recognition and insertion motifs, and other characteristic features such as a conserved undecapeptide (ECTGLAWEWWR) near the C-or N-terminal ends. It also lacks repeat domains in the toxin (RTX) and calcium-binding glycine-rich motifs necessary for oligomerization [26]. This in silico structure analysis of M. tuberculosis proteins led to a deeper understanding of TlyA's role and challenged the initially described function for TlyA-related proteins.
A combination of bioinformatics analysis and experimental structure elucidation may suggest alternative functional roles for previously-annotated enzymes, as with (SAM)-dependent methyltransferase, phosphatase and N-acetyltransferase structures from M. tuberculosis [11,27,28]. Theoretical evidence has thus been provided for TlyA's role in RNA modification instead of it just being a hemolytic factor.

Subcellular localization suggested a cytoplasmatic localization consistent with ribosome-associated activity
Predicting subcellular localization is important since several characteristics can be deduced, such as protein function and genome annotation. It also aids experimental design for proteomics platforms, particularly for identifying new candidates for vaccine development and drug targets [29,30]. Protein sequence analysis did not reveal a signal peptide, suggesting that TlyA is not secreted; however, some authors have described TlyA as being a probable PFP [6], supported by the presence of over-expressed TlyA on the E. coli wall surface [8].
TlyA lacks transmembrane helices and β-barrels (2.877 value), suggesting that this protein might not be a membrane-embedded or contact-dependent hemolysin, as has previously been proposed [6,23]. TlyA also has a hydrophobic region in the C-terminal domain (residues 115-133) which may not form a membranespanning α-helix. Nonetheless, such a region might allow the adenine present in AdoMet to fit properly inside the catalytic pocket, thus suggesting a different role for these hydrophobic residues. Virulence-related functional annotation, based on all the above-mentioned data, remains controversial.
Conservation of tlyA genes and functions was expected according to our phylogenetic and evolutionary trace analyses, the initial hypothesis being that homologous proteins have related functions. However, the extent to which this is true has not been assessed in detail up to now; annotation based on protein homology might thus lead to finding unexpected variations in function.

Conclusions
The main concern of our work was to decipher structural information about the TlyA family by homology comparison. Our structural analysis and finding residue conservation within the active site supports recent experimental work [8] about TlyA's role as RNA 2`-O-MTase in M. tuberculosis. This is further supported by the domain composition also suggesting methyltransferase activity and RNA-binding. The distribution of charged residues also defines a likely RNA interaction instead of hemolytic activity, little similarity being exhibited with hemolysin proteins, perhaps representing inaccurate annotations in several databases, as observed with other protein families [31].
Since genome annotation is based on sequence comparison, it is the most commonly used approach for determining functional homology. Annotations are assigned if proteins surpass similarity cut-off and the propagation of original errors could be increased by each new entry. In such cases, protein function needs to be inferred from their common three-dimensional structures and manual curation.  [36].

Methods
An M. tuberculosis TlyA (Rv1694) secondary structure consensus was built based on the predictions obtained with SAM [37], PSIPRED [38] and JNet secondary structure prediction [39] servers. This consensus allowed poorly and highly structured regions to be compared for selecting the best tridimensional (3D) structure model according to structural homology and folding prediction.

Phylogenetic and evolutionary trace analysis
The molecular phylogenetic tree for TlyA was built with Neighbor Joining (NJ), using p-distance as substitution model, and Maximum Parsimony (MP) methods in MEGA Version 4.0, with 5,000 iterations for calculating bootstrap confidence levels [53]. Phylogenetic tree construction included the sequences for Mycobacterium TlyA proteins and/or putative cytotoxin/hemolysins reported in the TB-database for the following 11 mycobacterial species, including 14 different strains. Pertinent sequences were found in 21 M. tuberculosis-related species (Table S1, Additional file 1).
Evolutionary Trace Report Maker [54] and Evolutionary Trace Analysis [55] and DnaSP [56] were used for identifying residues under evolutionary pressure and substitution rates were determined using MEGA 4.0.

Additional material
Additional file 1: Table S1. Descriptions of TlyA proteins used in the phylogenetic analysis. Protein sequences were gathered from the GenBank [http://www.ncbi.nlm.nih.gov/] by BLAST search [32].
Additional file 2: Figure S1. Sequence alignment and secondary structure representation for TlyA and Ftsj-like RNA methyltransferase complexed with AdoMet, defined as the ligand-binding template [PDB:1EJ0] [31] showing the matched residues with the template. Alpha helices are represented by red spirals, beta strands by yellow arrows and random structures by straight lines. The red bar indicates the region with greater structural similarity (152 residues-long). Amino acid regions sharing the highest similarity are enclosed in boxes.
Additional file 3: Figure S2. Ramachandran plot of TlyA threedimensional modeled structure. Dispersion zones are shown in blue and orange representing proline and glycine favored and allowed regions (Figure obtained by RAMPAGE server).
Colombia. 3  Authors' contributions NEA, CV and LMS wrote the manuscript, validated the tools and carried out the data analysis and interpretation. CYS, MEP, MAP and AG contributed to the methodological design, supervised its development and critically revised the manuscript's content. All authors read and approved the final version of the manuscript.