Mycobacterium tuberculosis H37Rv was grown at 37°C in a shaking incubator (120 rpm) in Difco^TM Middlebrook 7H9 broth containing 0.2% glycerol, 0.05% Tween 80, and 10% albumin-dextrose-catalase growth enrichment (Becton Dickinson and Company, Diagnostic Systems, Sparks, MD, USA) to OD₆₀₀ of 0.5. M. tuberculosis H37Rv was cultured in a BSL3 facility at the Laboratório Central de Saúde Pública – Brasília – DF (Lacen – DF).

The popmt gene was amplified by PCR from M. tuberculosis H37Rv genomic DNA using the primers: forward, 5′-AGATTACATATGACATTTGAGCCTGCCC-3′ (Nde I site, underlined; initiation codon, bold) and reverse, 5′-GTTATAGGATCCTTAGCCGGCCAGCATCCG-3′ (BamH I site, underlined; stop codon, bold). The 2022 bp PCR product was digested and ligated into similarly digested pET-28a(+) (Novagen) using T4 DNA ligase (Invitrogen) in frame with N-terminus 6xHis-tag vector. The cloned popmt sequence was checked by DNA sequencing of both strands using T7 primers (Genscript, USA). For overexpression of his-tagged recombinant protein, the pET-28a(+):popmt plasmid was transformed into Escherichia coli BL21 (DE3) cells.

Overnight culture of E. coli BL21 (DE3) carrying pET-28a(+):popmt was inoculated into 500 ml of LB broth supplemented with 30 μg/ml kanamycin and incubated at 37°C, 200 rpm until the OD₆₀₀ reached 0.4–0.6. Standard conditions of expression were established with 0.05 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 20°C for 16 h. Recombinant POPMt (POPMt) was purified from the soluble fraction through nickel affinity chromatography (His Bind^® Kit – Novagen) according to the manufacturer’s protocol. Purified POPMt was stored in 50% glycerol at -20°C. POPMt purification was analyzed on 10% SDS-PAGE followed by Coomassie Brilliant Blue R staining (Sigma-Aldrich). The protein was quantified using the molar absorption coefficient 𝜀 value of 148,865 (M^-1cm^-1) at 280 nm measured in water.

The membrane was blocked by incubation in 5% (w/v) non-fat milk/PBS overnight at 4°C. Blot was incubated for 2 h with 1:100 POPMt diluted in 1% non-fat milk/PBS. After several washes with PBS, the membrane was incubated for 1 h with 1:1000 alkaline phosphatase-conjugated goat anti-mouse IgG (Invitrogen). Immunocomplexes were revealed with the alkaline phosphatase substrate 5-bromo-4-chloro-3-indolyl-phosphate/Nitro Blue Tetrazolium (BCIP/NBT Color Development Substrate – Promega).

Recombinant POPMt enzymatic activity was measured using 7-amido-4-methylcoumarin (AMC). The released of AMC was monitored up to 20 min in 96-well SpectraMax M5 microplate reader (Molecular Devices) at 25°C as previously described (Grellier et al., 2001). The POPMt enzymatic activity was assayed on several AMC-containing substrates: Ala-Ala-Phe-AMC, L-Proline-AMC hidrobromide, N-Succinyl-Ile-Ala-AMC, N-Succinyl-Leu-Tyr-AMC, N-Succinyl-Leu-Leu-Val-Tyr-7-AMC, L-Leucine-AMC hydrochloride, Gly-Pro-AMC, N-Succinyl-Gly-Pro-AMC (N-Suc-Gly-Pro-AMC) and N-Succinyl-Gly-Pro-Leu-Gly-Pro-AMC (N-Suc-Gly-Pro-Leu-Gly-Pro-AMC), which were purchased from Sigma-Aldrich. Enzymatic reactions were performed with 25 mM Tris pH 7.5 (reaction buffer) containing 20 μM of fluorogenic substrate with or without additives (NaCl – 0 to 400 mM and DTT – 0 to 20 mM) in 100 μl final volume. The temperature assay was carried out by incubating the enzyme at different temperatures: 20, 28, 37, 40, 45, 60, or 80°C for 20 min and afterward 20 μM of substrate were added. The pH activity optimum of POPMt was determined in AMT buffer (100 mM acetic acid, 100 mM MES, and 200 mM Tris-HCl) at pHs ranging from 5.0 to 10.0 (Bastos et al., 2010). All experiments were performed in triplicate and repeated three times independently.

Kinetic parameters were determined by incubation of POPMt in reaction buffer with different concentrations (6.25–150 μM) of N-Suc-Gly-Pro-Leu-Gly-Pro-AMC. K_m and V_max were determined by hyperbolic regression according to Cornish-Bowden (Cornish-Bowden, 1976). The k_cat was calculated by = V max/[E]t, where [E]_t is the total enzyme concentration.

Different concentrations of the inhibitors tosyl-lysylchloromethane (TLCK), N-p-Tosyl-L-phenylalanine chloromethyl ketone (TPCK), bestatin, EDTA, L-trans-epoxysuccinylleucylamido-(4-guanidino) butane (E-64), 1,10-phenanthroline and leupeptin (0.01–100 nM) were incubated with POPMt in 100 μl reaction buffer for 15 min at room temperature before the addition of 20 μM N-Suc-Gly-Pro-Leu-Gly-Pro-AMC. The enzymatic reactions were monitored as described above. The inhibition profile (IC₅₀) using Z-Pro-Prolinal, a specific inhibitor of POPs, was determined by non-linear regression analysis from the residual activity versus inhibitor concentrations curve and the Ki values determined using the Cheng-Prusoff method K_i= IC50/[1 + ([S]/Km)], where [S] is the concentration of the substrate (Yung-Chi and Prusoff, 1973).

Fluorescence measurements were performed using an ISS K-2 (Champaign, IL, USA) spectrofluorimeter at the same temperatures used in activity assay as described above. Spectra were recorded from 305 to 450 nm using an excitation wavelength of 295 and 2 nm bandwidth for both excitation and emission. Solutions of 0.30 μM POPMt were prepared in AMT buffer at pHs ranging from 5 to 10. Measurements were carried out in a 1.0 × 1.0-cm cuvette. The final spectra were baseline corrected by subtracting buffer spectrum.

The POPMt 3D-model was built based on the Myxococcus xanthus POP crystal structure, PDB code 2BKL (Shan et al., 2005), using the MODELLER 9v8 software (Eswar et al., 2006). Briefly, using the NCBI BLASTP webservers (Altschul et al., 1997), the POPMt sequence was submitted to a search query against a structure database [Protein Data Bank (PDB) (Berman et al., 2000)]. The PSI-BLAST algorithm was used to perform an alignment profile between all the homologous sequences and facilitate the alignment between the target sequence (POPMt) and the template. The alignment was then subjected to the MODELLER software, resulting in 5 comparative models. The one with the lowest DOPE score was selected to further refinements by MDs simulations.

The MD simulation was performed using the computational package GROMACS 4 – Groningen Machine for Chemical Simulations (Hess et al., 2008). The simulated ensemble was composed of the POPMt model immersed in 14,46 Single Point Charge (SPC) water molecules (Berendsen et al., 1981) in a cubic box with edges of 77 Å. Four sodium atoms were also included in order to neutralize the system charges. The lysine residues of the protein were protonated and an amide group was added to the C-terminus. The SETTLE algorithm was used to constrain the geometry of water molecules, and the LINCS (Hess et al., 1997) algorithm was used to constrain bond lengths. The electrostatic corrections were performed by the Particle Mesh Ewald (PME) algorithm with a cutoff radius of 1.4 nm, in order to minimize the computational time of the simulation. Derived from the Ewald summation, PME estimates in the Fourier space the long-range interactions that happen in real space (Darden et al., 1993; Essmann et al., 1995). The same radius value cutoff was also used in the van der Waals interactions. The neighbors list of each atom was updated every 10 simulation steps of 20 fs each.

The system was contained. Two steps of energy minimization were performed (2 ns each), the first used the conjugate gradient algorithm and the other one used the steepest descent algorithm. After the energy minimization step, the system went through a pressure and temperature normalization process using the integrator stochastic dynamics (SD), also for 2 ns. Thereafter, the system went through a position restrain step using the MDs integrator for another 2 ns. This ensemble was subjected to a relaxation MDs run at GridUNESP computers. The 50 ns simulation was divided into 25,000,000 steps of 2 fs each.

Peritoneal macrophages from naive C57BL/6 mice were harvested by lavage with sterile Roswell Park Memorial Institute (RPMI) 1640 medium. Macrophages (10⁶ cells/mL) were cultured overnight in RPMI containing 2% fetal calf serum. Non-adherent cells were removed after phosphate-buffered saline (PBS) wash. Macrophages were stimulated with POPMt (0.01–10 mg/mL), either native or denatured by boiling, for 24 h at 37°C in CO₂ atmosphere. Macrophages viability assessed by trypan blue exclusion at the end of each experiment was always >95%.

Cell-free supernatants from in vitro POPMt stimulated macrophages were collected after 24 h and stored at -20°C until analysis. TNF-α, IL-6, IL-12p70, IL-23, IL-10, IL-1b, and MCP-1 levels were measured in supernatants from in vitro POPMt stimulated macrophages by enzyme-linked immunosorbent assay, in accordance with manufacturer instructions (R&D Systems).

Bars graphics were generated with Prism software (GraphPad). All data was expressed as mean and standard deviation. Statistical analyses were performed by ANOVA followed by the Newman-Keuls-Student and Student’s t-tests. The significance level was set at p < 0.05.

The popmt gene from M. tuberculosis has an open read frame of 2,022 bp and encodes a protein of 673 amino acids residues with predict molecular mass of 74.40 kDa. POPMt has about 65 and 78% identity compared to M. smegmatis and M. avium POPs, respectively. On the other hand, POPMt amino acid sequence identity compared to human POP is about 24%. POPMt was expressed in E. coli BL21(DE3) as a soluble and active enzyme which allowed us to proceed to its purification (Figure 1A) and biochemical characterization. When subjected to intra-dermal immunization, isogenic BALB/c yielded specific anti-POPMt serum capable of recognizing a single band of approximately 75 kDa in the total extract of M. tuberculosis proteins. This result confirms that POPMt is immunogenic in mice and is expressed in H37Rv, a M. tuberculosis human pathogenic strain (Figure 1B).

Some fluorogenic peptides were used to define the substrate specificity of POPMt (Table 1). Among these substrates, the enzyme cleaved specifically at the carboxyl terminus of proline residues, feature similar to other POP, which are known to cleave a Pro-Xaa bond in peptides, where Xaa is not a proline residue (Koida and Walter, 1976). With respect to these substrates, the enzyme was active on N-Suc-Gly-Pro-Leu-Gly-Pro-AMC and N-Suc-Gly-Pro-AMC but totally inactive toward L-Pro-AMC and Gly-Pro-AMC. The K_M and K_cat/K_M values of POPMt toward N-Suc-Gly-Pro-Leu-Gly-Pro-AMC were 108 μM and 21.83 mM^-1 s^-1, respectively.

Thiol-reacting reagents and salts are known to improve POP enzymatic activity (Usuki et al., 2009). Based on that, NaCl and DTT were added into buffer reaction to determine their influence on POPMt enzymatic activity. As shown in Figure 2A, up to 20 mM DTT did not affect the enzyme activity. As for DTT, increasing concentrations of NaCl (up to 400 mM) did not alter POPMt enzymatic activity toward N-Suc-Gly-Pro-Leu-Gly-Pro-AMC (Figure 2B).

The characterization of POPMt revealed that the enzyme had a strong dependence on slightly alkaline pH 7.0–8.5. It was most stable at pH 7.5 and retained more than 90% of the residual activity at pH 8.0 and 8.5 (Figure 3A). We also analyzed POPMt tertiary structure changes under different pHs by fluorescence spectroscopy (Figure 3B). It is possible to observe changes in emission spectra of tryptophan in response to protein conformational transitions, subunit association, substrate binding or denaturation, all of which can affect the local environment surrounding the tryptophan indole ring (Lakowicz, 2004). POPMt contains 20 tryptophan residues and, as shown in Figure 4B, the emission maximum was 335 nm, indicating that most POPMt tryptophans are partially buried in the protein. The shape of the enzyme intrinsic fluorescence spectra did not change from pH 7.5 to 8.5 and a limited red-shift happened at pH 9.0 and 10.0.

The effect of temperature on the enzymatic activity of POPMt was examined over a range of 20–80°C at pH 7.5. The enzyme showed maximal enzymatic activity between 20 and 37°C. While at 45°C, the enzyme activity decreased nearly 50% (Figure 4A). As for pH analysis, we also applied the tryptophan residue fluorescence as a reporter for structural changes under different temperatures and, according to all the adjusted spectra, temperatures up to 60°C did not alter tryptophan emission pattern (Figure 4B).

Table 2 summarizes the effects of protease inhibitors on the enzymatic activity of POP from M. tuberculosis. Although POPMt is a serine protease, its activity was partially inhibited by AEBSF and TLCK, 33 and 35%, respectively. On the other hand, TPCK, a chymotrypsin-like protease inhibitor, inhibited 66% of the enzyme activity. Leupeptin, E64, pepstatin A and metalloproteinase inhibitors such as o-phenanthroline, bestatin and EDTA had no effect on the enzyme activity. The canonical POP inhibitor Z-Pro-Prolinal (Wilk and Orlowski, 1983; Yoshimoto et al., 1987) was also assayed on POPMt, showing a Ki value of 16.87 nM. Z-Pro-Prolinal was a less efficient inhibitor when compared to mouse and human POPs with Ki value of 0.35 and 0.50 nM, respectively (Bakker et al., 1990).

Concerning the homology modeling, the top ranked homolog structure returned by BLAST was that of M. xanthus (PDB ID 2BKL), with a query coverage of 64 and 37% identity. As we could not obtain higher query coverage, we opted to perform MDs simulations to both refine and validate our model. We predicted that an equilibrium MDs simulation would allow a relaxation of the macromolecule, in which the POPMt features would be adopted instead of those from M. xanthus enzyme.

The POPMt macromolecular structure is divided in two domains with a catalytic α/β hydrolase domain and a β-propeller domain (Figure 5A). The propeller domain is based on a radially arranged seven-fold repeat of four stranded antiparallel β sheets. In the case of POPs, this domain is considered to be of the “open-velcro” topology, where first and seventh blades are connected only through hydrophobic interactions, although their primary sequences diverge (Kaushik and Sowdhamini, 2011). The POPMt catalytic triad residues (Ser 532, Asp 615, e His 647) position is conserved like in other POPs and it is localized at the interface of the catalytic and propeller domains (Figure 5B) (Rea and Fülöp, 2006). The amino acid sequence of POPMt with the assigned secondary structure comprising coil, β sheet, α helix and Pi helix is shown in Figure 5B.

The superposition of the predicted and template structures (Figure 6A) shows an average RMSD of 4.86 Å (Figure 6B, red line). This might be explained by the fact that the M. xanthus POP template is a crystal structure, hardened due to the compression inherent in the crystallization process, which results in the inability of structure accommodation and its respective adaptation to the environment. Structures subjected to MDs simulations are expected to exhibit RMSD changes, as they go through a process of relaxation and adjustments, always driven by biochemical and structural restrictions.

Albeit resulting in conformational changes when compared to the initial structure, as observed by the RMSD shift, the POPMt relaxation process led the structure to a smaller radius of gyration (Figure 6B, blue line), indicating the contraction of the system. This can be interpreted as a sign of conformational stability, as during protein unfolding we would expect the opposite, and may be explained due to the increasing number of intramolecular hydrogen bonds during the simulation (Supplementary Figure S1). This kind of interactions cooperates to reduce the volume of the system and thus increase the degree of compactness of the protein.

Another sign of compactness is the solvent accessible area (Supplementary Figure S1), which corresponds to the area of solvent removed due to the relationship of the protein with itself. By folding, the protein contracts, resulting in the reduction of the solvent accessible area, decreasing the protein volume characterized by the separation of hydrophobic and hydrophilic portions.

Data extracted from this MD trajectory will be applied to identify clusters of conformational families as recently accomplished to the Dengue virus NS3 protease (de Almeida et al., 2013), which will be used in ensemble docking campaigns aiming the identification of new drug-like hits against POPMt.

To investigate the immunomodulatory properties of POPMt in triggering immune responses in vitro, peritonial macrophages were stimulated with purified POPMt for 24 h. POPMt significantly triggered production of the proinflammatory cytokines TNF, IL-12p70, IL-6, IL-23 and IL-1b, as well as chemokine MCP-1 after 24 h (Figures 7A–F). However, POPMt failed to trigger the anti-inflammatory and immunosuppression cytokine IL-10 (Figure 7G). In addition, all analyzed cytokines and chemokine secretion induced by POPMt were dependent on the enzyme intact structure since cytokine and chemokine production induced by boiled POPMt were greatly reduced.