Mycobacterium tuberculosis CysA2 is a dual sulfurtransferase with activity against thiosulfate and 3-mercaptopyruvate and interacts with mammalian cells

Cyanide is a toxic compound that is converted to the non-toxic thiocyanate by a rhodanese enzyme. Rhodaneses belong to the family of transferases (sulfurtransferases), which are largely studied. The sulfur donor defines the subfamily of these enzymes as thiosulfate:cyanide sulfurtransferases or rhodaneses (TSTs) or 3-mercaptopyruvate sulfurtransfeases (MSTs). In Mycobacterium tuberculosis, the causative agent of tuberculosis, the gene Rv0815c encodes the protein CysA2, a putative uncharacterized thiosulfate:cyanide sulfurtransferase that belongs to the essential sulfur assimilation pathway in the bacillus and is secreted during infection. In this work, we characterized the functional and structural properties of CysA2 and its kinetic parameters. The recombinant CysA2 is a α/β protein with two rhodanese-like domains that maintains the functional motifs and a catalytic cysteine. Sulfurtransferase activity was determined using thiosulfate and 3-mercaptopyruvate as sulfur donors. The assays showed Km values of 2.89 mM and 7.02 mM for thiosulfate and 3-mercaptopyruvate, respectively, indicating the protein has dual activity as TST and MST. Immunological assays revealed that CysA2 interacted with pulmonary cells, and it was capable to activate macrophages and dendritic cells, indicating the stimulation of the immune response, which is important for its use as an antigen for vaccine development and immunodiagnostic.


Results
M. tuberculosis cysA2 gene encodes one of four putative thiosulfate sulfurtransferases found in the bacillus genome. The analysis of the genome of M. tuberculosis showed four genes encoding for putative thiosulfate sulfurtransferases, Rv0815 (CysA2), Rv3117 (CysA3), Rv3283 (SseA) and Rv2991 (SseB) (Fig. 1a). cysA2 and cysA3 are located in different regions of the genome but encode identical proteins. Genes related to the metabolism of sulfur and molybdenium are close to cysA2 and cysA3 24 . On the other hand, Rv3283 (sseA) and Rv2991 (sseB) are isolated in the genomes with putative functions as thiosulfate sulfurtransferases that share 50% and 26% of amino acid sequence identity with CysA2/CysA3, respectively. The superimpositions of the available three-dimensional structures of CysA2, CysA3 and SseA (PDB codes 3HWI, 3AAY, 3HZU, respectively) revealed they have identical folding (Fig. 1b). Moreover, the amino acid sequence alignments showed that from the four sulfurtransferases, only SseB has the MSTs motif CG(S/T) (Fig. 1c). Proteomics studies and two-dimensional gel electrophoresis followed by mass spectrometry in standing cultures 25,26 showed that although there were no differences in the expression levels of genes cysA2, cysA3 and sseB in ancient and modern Beijing strains, the average rhodanese activity observed in the cell lysate of ancient Beijing strain was 23% higher than the rhodanese activity observed in the modern one 26 . A mutagenesis analysis using Himar1-based transposon showed the genes are non-essential in M. tuberculosis CDC1551 strain 27 . On the other hand, the M. tuberculosis H37Rv sseA mutant showed slow growth rate 28 and impairment of survival of the bacillus in primary murine macrophages 29 .
Based on the amino acid sequence alignment, the four proteins show the two signatures of rhodanese enzymes 12 30 (Fig. 1b).
Based on the available three-dimensional structure of CysA2, we performed a comparison with putative orthologues studied in the literature. A search in the Protein Data Bank for the orthologues revealed the TSTs SseA of M. tuberculosis (50%), E. coli Rhobov, Azotobacter vinelandii RhdA (both with 30%) and E. coli YnjE (27%). Moreover, two MSTs that shared around 30% of amino acid sequence identity with CysA2 were identified in Saccharomyces cerevisiae (TUM1) and E. coli (SseA), respectively (Fig. S1a, Supplementary Material). Despite of the low amino acid sequence conservation, all the proteins maintained the structural α/β folding. The structural alignment of CysA2 showed a very low RMSD (0.5 to 0.6 Å) when compared to TST proteins, and higher values when compared to 3-mercaptopyruvate sulfurtransferases (RMSD from 1.4 to 2.6 Å) (Fig. S1b,c, Supplementary Material).
The structure of CysA2 was solved at 2.29 Å and revealed two α/β domains (137 amino acids of the N domain and 130 belonging to the C domain) (Fig. 2a). Based on the TSTs from literature, the N domain is inactive, whereas the C domain is responsible for the catalytic activity. The conserved catalytic cysteine of the active site 31 (Cys233) is present in M. tuberculosis CysA2 structure (Fig. 2a, in yellow sticks) as well as the signature sequences located at the N-and C-terminal domains (Fig. 2b, green and orange boxes, respectively). The residues CRI (amino acids 233-235) form the active site in CysA2, which is more consistent with TST motifs than MSTs (Fig. 2b,I and II).
CysA2 alternates the dimer/monomer states upon substrates binding. M. tuberculosis CysA2 was purified by immobilized nickel affinity chromatography followed by size-exclusion chromatography. According to the analytical gel filtration, the protein eluted as a dimer of 74 kDa (Fig. S2, Supplementary Material). Similarly, when submitted to dynamic light scattering analysis CysA2 showed a hydrodynamic radius (r) of 3.8 nm, which was consistent with a dimer. All the experiments were performed just about the purification steps.
CysA2 was purified as a structured protein that showed a circular dichroism (CD) spectrum characteristic of α/β folding (Fig. 3a). Deconvolution of the CD data showed a slight difference in the content of α-helices (18% to 21%) and β-sheets (39% to 36%) between the recombinant protein (CysA2r) and the protein in the presence of cyanide (CysA2r + KCN) (Fig. 3b), indicating changes in the structure upon interaction. Compared to the PDB crystal structure, the values of CysA2 samples in solution presented lower α-helices and higher β-sheet contents (29% and 15%, respectively). These differences can be attributed to the significant variation between the protein purification and crystallization conditions. In our experiments, the protein was maintained in absence of salts and 20 mM Tris-HCl pH 8.0 buffer while the crystals were obtained in 25% PEG 3350, 0.1 M BisTris pH 5.5, 0.2 M ammonium acetate.
To avaluate if the CysA2 was stable or partially folded, the thermal stability was monitored by the CD signal at 222 nm during the increasing of the temperature from 20 °C to 100 °C (Fig. 3c). The calculated temperature of melting (Tm) was equivalent to 58.3 °C indicating the protein was stable. The presence of thiosulfate or cyanide did not affect this result and the protein exhibited irreversible thermal unfolding. The behavior of CysA2 in absence of co-substrates, presence of cyanide, thiosulfate and cyanide +thiosulfate also was avaluated by dynamic light scattering (Fig. 3d). In this essay, the protein in absence of co-substrates showed a hydrodynamic radius (rh) of 3.8 nm and a calculated molecular mass of 72 kDa, which was equivalent to a dimer in solution (Fig. 3d, 10 min). With addition of the substrates CysA2 showed a time-dependent rh reduction that reached a minimum of 3 nm and a molecular mass of 44 kDa, which was followed by reestablishment of the original parameters Enzymatic activity of CysA2 shows it is capable to use both thiosulfate and 3-mercaptopyruvate as substrate. Enzymatic analysis of CysA2 showed that the enzyme was able to use both thiosulfate and 3-mercaptopyruvate as substrates ( Fig. 4a-h, respectively). Assays defined the enzymatic activity of CysA2 through the kinetic parameters observed in the Michaelis-Menten constant (K m ) and the maximum velocity of enzymatic catalysis (V max ) determined using the Hanes equation. The K m for thiosulfate was 2.97 mM +/− 0.05, while the K m for cyanide was 1.66 mM +/− 0.03. V max for thiosulfate was 7.40 mmol SCN − /mg.min +/− 0.72 mmol SCN − /mg.min, while V max for cyanide was 8.69 mmol SCN − /mg.min +/− 0.93 mmol SCN − /mg.min (Fig. 4a-d). An unpaired T test was used to compare the two maximum velocities obtained and no significant statistical difference was found. Enzymatic kinetics experiments using 3-mercaptopyruvate showed an additional ability of the enzyme to also utilize this substrate, The efficiency of the catalysis of cyanide, thiosulfate and 3-mercaptopyruvate by CysA2, was evaluated in comparison with kinetic parameters obtained for other sulfurtransferases ( Table 1). The results indicated that low concentrations of the substrate are not able to fully occupy the active site of the enzyme CysA2, as showed by the relatively low K m , especially when considering the same parameters obtained for Rhobov protein, a model of rhomboid structure and function. Once the rate of reaction is directly related to site occupancy, we have a V max with a value in the order of nmolar while other enzymes show units in the μmolar range. In fact, it is very unlikely that the natural substrates (in vivo) of these enzymes be cyanide and thiosulfate due to the high toxicity of these molecules to the cell. This feature might explain the low capacity of the conversion of the substrates to the expected product 15,32 . Although CysA2 conserves the functional motifs of thiosulfate sufurtransferases, its kinetic activity against thiosulfate was very low, suggesting that in vivo, different substrates might be used. The ability of CysA2 to use 3-mercaptopyruvate as a thiol group donor was also analysed. Protein activity was monitored in a constant concentration of cyanide potassium (5 mM) revealing K m and V max values for mercaptopyruvate of 7.02 mM and 0.461 μmol/min.mg, respectively.
Once the results showed low values for activity both as TST and MST, we also evaluated the pH and cations influence on CysA2 activity (Fig. 5). As for the effect of pH on CysA2 activity, it is observed that the enzyme is capable of operating efficiently over a wide pH range (7.0 to 9.0) (Fig. 5a). Rhodaneses of other species also have an optimal pH range around 8 and even more basic pH 31,33 .
The effect of cations on CysA2 activity was evaluated since classical studies with the bovine rhodanese showed a divalent cation-binding region in its structure 34 . The CysA2 analyses were performed in increased concentrations of Cu 2+ , Zn 2+ , Mg 2+ and Ca 2+ . The results showed that concentrations of 0.2 to 0.5 mM of cations have positive effect on the CysA2 enzymatic activity but not 1 mM (Fig. 5a,b). Exception was evidenced for CysA2 in presence of 3-mercaptopyruvate and 1 mM CaCl 2 that showed a significant increasing in the activity. In general, the effect of cations on CysA2 activity was more evident with 3-mercaptopyruvate than thiosulfate.
CysA2 can bind to pulmonary mouse cells and has the ability to activate dendritic cells and macrophages. The identification and presentation of antigen by target cells, mostly antigen presenting cells www.nature.com/scientificreports www.nature.com/scientificreports/ (APCs), involves the expression of specific molecules. Exogenous pathogens proteins are usually processed and presented via MHC class-II molecules by APCs, which in turn became activate and upregulate the expression of co-stimulatory molecules, as CD86 35,36 . Additionally, pathogens can be recognized by specific receptors from epithelial cells. Therefore, to understand some mechanisms involved in the M. tuberculosis recognition by target cells, we evaluate the immunogenic properties of CysA2 and explored its interaction with epithelial and immune cells. Since the target infection site of the M. tuberculosis is the lung, we first evaluated if CysA2 was able to bind to the TC-1 cells (Fig. 6), a transformed pulmonary cell from C57BL-6 mouse. Binding of CysA2 to TC-1 cells was assess by its labeling with an anti-histidine monoclonal antibody followed by a secondary labeling with an anti-IgG (Fig. 6a). Interestingly, in our experimental settings, CysA2 showed a great ability to bind to TC-1 cells (Fig. 6b), highlighting the possibility of its role in bacterial adhesion and, consequently, its importance in the M. tuberculosis recognition by epithelial cells. Furthermore, using the purified protein, we also evaluated the ability of CysA2 in activating two important antigens presenting cells (APCs), dendritic cells and macrophages (Fig. 7). The ex-vivo dendritic cells belonging to total the heterogeneous population of splenic cells from C57BL/6 mice were evaluated by multiple parameters after immunophenotyping (Fig. 7a). The murine macrophage cell line J774 activation was evaluated after gated as single cells for further analysis of the expression of the co-stimulatory molecule CD86 (Fig. 7b). Fortunately, CysA2 showed to be immunogenic, once it was able to activate the expression of CD86 in both murine dendritic cells (Fig. 7c) and murine J774 macrophages (Fig. 7d). Interestingly, the potential of CysA2 to activate immune cells was more pronounced in dendritic cells than J774 macrophages, since CD86 expression in this experimental group was similar to the positive control group (LPS). The co-stimulatory molecules CD40 and CD80 were not upregulated upon CysA2 treatment. These results clearly state that CysA2 stimulates the cellular immune response, responsible for antigen presentation and/or microbial degradation.

Discussion
The few literature data that may be associated with M. tuberculosis CysA2 have shown that this enzyme plays an important role in the physiology of bacillus. Its three-dimensional structure reveals the classical sulfurtransferase/ rhodanese domains, characteristics of enzymes importantly active in the metabolism of different organisms, as well as a shape with pockets and cavities, which make it an excellent target for the development of inhibitors using, for example, the methodology of fragment based drug discovery. In addition, its potential as an immunological agent draws attention to the use of this protein as a vaccine epitope or even for diagnosis, once the current     diagnostic methods exhibit low sensitivity, require a long time to perform the microbiological cultures and low specificity for M. tuberculosis. In this work, we described the functional features that set it as sulfurtransferase and performed biophysical characterization to define the kinetic activity. Moreover, we explored its potential as immunogenic target. CysA2 shares high sequence similarity with TSTs and its capability to use 3-mercaptopyruvate as substrate was shown, although the enzymatic parameters were significatively better in the presence of thiosulfate as substrate. On the other hand, the K m values in the order of millimolar show that both thiosulfate and 3-mercaptopyruvate should not be the physiological substrates. Moreover, M. tuberculosis CysA2 presented low activity when compared to other TSTs and also MSTs. Considering these results and comparing to similar studies with proteins from sulfurtransferase family, it was found that while the enzymes of the sulfurtransferase group are specific for the type of donor in the thiol group, other enzymes of this class seem to show some promiscuity for the donor type 30,37,38 . In MST1 and MST2 enzymes of A. thaliana that are able to catalyze both thiosulfate (TST) and 3-mercaptopyruvate (MST), the cysteine in the "CR(Y/R)" motif is essential for the enzyme's activity in the presence of any substrate. Different results were found for sulfurtransferases from men and cattle, the latter considered as a model of TST 37 . In these organisms the enzymes showed specificity for each of the substrates, being well defined as MST or TST. Similar to the enzyme identified in rat that showed ability to transit between the two substrates, M. tuberculosis CysA2 also performed as MST and TST, with better activity with thiosulfate as substrate. In the case of protozoa, Leishmania major, Leishmania mexicana and Trichoderma harzianum, also eukaryotes, the behavior of the enzymes is also dubious, with preference for substrates defined on characteristics signatures 17 .
The sequence of the catalytic site of all the MSTs analyzed by the kinetic activity in the literature presents high conservation of the amino acids and we can not identify with this analysis the permissive point for the use of one or the other substrate. In sulfurtransferases, alpha-helix macropoles surrounding the catalytic cysteine are important to create a redox environment that will support hydrogen bonds between sulfur atom of a persulfide and surrounding amino acids stabilizing catalytic intermediates. As for the effect of pH on CysA2 activity, it was observed that the enzyme is able to operate efficiently over a wide pH range with optimum activity at pH 8.0, which is in accordance with rhodaneses from other species 33 .
Regarding the M. tuberculosis pathogenesis, M. tuberculosis can infect epithelial cells both in vitro and in vivo [39][40][41][42] . It is believed that the bacillus interactions during infection may play a relevant role during the onset of the infectious process and that they may influence the immune response of the host 43,44 . Tuberculosis is basically a lung disease, the lung being the entry point of the bacillus and the main site of infection manifestation. Interestingly, although CysA2 belongs to a cytoplasmic pathway, it has been identified by mass spectrometry in other cell regions such as cell wall, membrane and M. tuberculosis culture supernatants 10,[45][46][47] . Concerning this last feature, it is likely that this protein can be recognized by the immune system. During bacterium pathogenesis, M. tuberculosis is phagocytosed by alveolar macrophages and dendritic cells that migrate through the lymphatic system and simultaneously can penetrate the lung parenchyma, attracting other macrophages and initiating the inflammatory focus 48 . Unraveling other properties of CysA2, this is an immunogenic protein, capable of being recognized and phagocytosed by dendritic cells and macrophages, leading to the upregulation of co-stimulatory molecules. Since CysA2 was able to bind to a pulmonary cell line, we hypothesized that this protein may be important to bacterial adhesion and/or recognition by host cells, suggesting a role during bacillus infection. Furthermore, the binding of CysA2 to TC-1 cells corroborates with previous studies that suggested its relevance as a serological marker for detection of tuberculosis in patients 8 . www.nature.com/scientificreports www.nature.com/scientificreports/ Given that CysA2 is secreted only under special culture conditions, that it has high levels of expression in infected patients 44 and the redundancy in the genome of M. tuberculosis 4 , CysA2 might be involved in the formation of such lipid rafts and could be the candidate that would play a relevant role in the process of infection of lung epithelial cells. Therefore, the CysA2 protein may behave as a good candidate for tuberculosis diagnosis, since it was recognized in serum from patients infected with M. tuberculosis 8 . Faced with these different properties of CysA2, this molecule might be a moonlighting protein, performing more than one function 49 . Many of moonlighting proteins have as their main characteristic, the activity in enzymatic catalysis and aggregation, and secondary, role in signal transduction, transcriptional regulation and apoptosis 50 . Furthermore, the presence of a gene duplication (cysA3) highlights the relevance of the respective proteins to the cell.
In conclusion, the dataset presented in this work consist of the first characterization of M. tuberculosis CysA2 evidencing relevant immunological features for its exploration in assays to reach a significant pool of proteins for diagnosis at the various stages of TB infection and manifestation. Moreover, due the importance of the protein for the bacillus, its kinetic activity and three-dimensional structure, which is highly drugable, CysA2 is an interesting target for the development of inhibitors. www.nature.com/scientificreports www.nature.com/scientificreports/ Methods Bioinformatics analyses. CysA2 amino acid sequence was obtained in the TubercuList database (http:// tuberculist.epfl.ch). Searching for orthologues in different species and related proteins with solved three-dimensional structures was performed using the BlastP tool (http://blast.ncbi.nlm.nih.gov) against the non-redundant protein sequence database (nr) and PDB, respectively. Amino acid sequences alignments were performed using ClustalW server (http://www.genome.jp/tools-bin/clustalw). Superimposition of the three-dimensional structures from CysA2 and orthologues was performed using in COOT 51 and analyzed in Pymol 52 .
Gene cloning, protein expression and purification. cysA2 gene was amplified by PCR using the M. tuberculosis genomic DNA as template and the oligonucleotides 5′-GAATTCATGGCACGCTGCGATGTCCTG-3′ (FcysA2) and 5-CTCGAGTCA GCTTCCCAACTCGATCGG-3′(RcysA2). Restriction sites for enzymes EcoRI and XhoI (underlined) were introduced in the forward and reverse oligonucleotides, respectively. The digested PCR product was cloned into pET-28a (Novagen, LA, CA, EUA), dowstream of the encoding His 6 tag sequence and checked by DNA sequencing. The plasmid was introduced in E. coli BL21(DE3) by chemical transformation. Protein expression was obtained after growth of the cells up to mid-log phase (OD 600 ~ 0.6) at 37 °C in LB medium with kanamycin (30 μg/ml) and induction by 4 hours after addition 0.5 mM isopropyl thio-β-D-galactopyranoside (IPTG). Soluble extracts were obtained after sonication of the cells (8 cycles x 15 s in a Brandson Sonifier 450) in lysis buffer (25 mM Tris-HCl pH 8.0, 500 mM NaCl, 100 μM PMSF) and centrifugation at 14.000 g for 40 minutes. The supernatant was immediately submitted for immobilized metal affinity chromatography (IMAC) using a column HiTrap IMAC HP (Qiagen, Boston, MA, EUA) and eluted with an imidazole gradient (0-500 mM). Dynamic light scattering analysis. Experiments were performed using a DynaPro 810 system (Wyatt Technology Corporation, Santa Barbara, CA, EUA) and DYNAMICS v.6.3.40 software. DLS experiments were carried out in a wide range of protein concentration (1-10 mg/mL) to analyze aggregation and to estimate CysA2 hydrodynamic radius. DLS measurements (100 measurements at 5 s intervals) were conducted at 291 K and the hydrodynamic radius (Rh) was extrapolated from the translational diffusion coefficient (Dt) using the Stokes-Einstein equation 53 .

Circular dichroism analysis.
Far-UV circular dichroism (CD) spectra of the enzyme in presence and absence of substrates were recorded on a Jasco J-810 spectropolarimeter (Jasco International Co., Tokyo, Japan). CD measurements were carried out in a 1 mm quartz cuvette using the wavelength range of 200-260 nm. The protein concentration was set to 5 μM in Tris-HCl 20 mM pH 8.0 with 5 mM of the substrats thiosulfate, 3-mercaptopyruvate and KCN. The data collection parameters were set to scan rate of 50 nm/min, response time of 4 s, sensitivity of 100 mdeg, scan step of 0.5 nm and 20 accumulation rounds. To study the effect of temperature on CysA2 structure, the samples were heated from 10 °C to 103 °C. The data collection parameters were exactly as described before, including the heating rate of 1 °C/min and delay time for spectrum collection of 60 s. The reversibility of the temperature effect was monitored from 103 °C to 20 °C using the same parameters described above. Baseline subtraction, smoothing and data normalization were carried out using the graphical software ORIGIN (http://www.originlab.com/). The CD data are shown as mean residue ellipticity units (deg.cm 2 .dmol −1 ) and evaluated by deconvolution of the CD spectrum using the DichroWeb server 54 .
Kinetic assays. The CysA2 sulfurtransferase activity was measured following the SCN-formation colorimetrically determined according to Sorbo's method 55 using thiosulfate or 3-mercaptopyruvate as sulfur donors and cyanide as sulfur acceptor. The kinetic parameters of CysA2 were obtained with 4 μg of protein in Tris-HCl 0.1 M pH 8.0 buffer with 50 mM KCN and 40 mM thiosulfate (STS assay) or 30 mM KCN and 30 mM 3-mercaptopyruvate (MTS assay). The reaction mixtures were incubated at 37 °C for 30 minutes and stopped with addition of 0.5 ml formaldehyde (15%), followed the addition of 2.5 ml of ferric nitrate (1%). The proportion of ferric ion complex and thiocyanate was measured spectrophotometrically at 460 nm using a spectrophotometer Epoch2 (Biotek Instruments, Vermont, USA). The steady-state kinetics was evaluated by Michaelis-Menten equation 56 . The influence of the pH in the enzymatic activity was checked in 100 mM Tris-HCl pH 7.0 and 9.0. The behavior of the enzyme was also evaluated in absence and presence of 0.2 mM, 0.5 mM and 1 mM of the different ions: CaCl 2 , CuCl 2 , MgCl 2 , or ZnCl 2 and 30 mM of substrate.
Pulmonary TC-1 mouse cells binding assay. The binding assays were performed as previously described 35 . Briefly, TC-1 cells (4 × 10 5 /well) 57 , generously provided by Dr. T. C. Wu from Johns Hopkins University, Baltimore, USA, in 2002, were incubated with CD16/CD32 antibodies (FcBlock, BD Biosciences) for 15 min on ice. Then, 5 µg/mL of CysA2 were added to cells, followed by an additional incubation of 40 min on ice. After washing cells with MACS buffer [phosphate-buffered saline (PBS), pH 7.2, 0.5% bovine serum albumin (BSA), and 2 mM EDTA], cells were incubated with anti-His-tag IgG monoclonal antibody (Thermo Fischer) for 40 min on ice. Then, after washing cells twice, TC-1 cells were stained with anti-IgG-PE antibody (Jackson ImmunoResearch) for 30 min on ice. Finally, cells were washed twice and resuspended in MACS buffer for further analysis in LSR Fortessa flow cytometers (BD Bioscience). Data were analyzed by FlowJo software (version 10.0.7 -Tree Star). Experiments were done twice, in triplicate.