L. infantum promastigotes were maintained in Schneider's Insect medium (Sigma Aldrich) supplemented with 20% (v/v) heat-inactivated fetal bovine serum (FBS) and 100 μg/mL gentamycin sulfate at 28°C.

Plasmids (pET19-b) containing the WT otuli or the sequence carrying the site-directed mutations that generated the F182S, F182S/L265P, and F82S/F182S/L265P variants (T545C, T545C/T794C, and T245C/T545C/T794C as nucleotide sequence respectively) were synthesized by GenScript. The constructions were expressed in BL21 (DE3) E. coli and produced the C-terminal His-tagged OtuLi under induction by 0.1 mM IPTG at 18°C for 5 or 16 h (F82S/F182S/L265P variant). The cells were lysed with BugBuster™ (Merck) and the supernatant was applied into the His-Bind® column (Novagen), washed with increasing concentration of imidazole and harvested with elution buffer (500 mM NaCl, 20 mM Tris-HCl, 200 mM imidazole, pH 7.5). After enzyme purification, WT rOtuLi and the 3 variants were dialyzed and concentrated with 25 mM Tris-HCl, pH 7.5. Purified enzymes was stored in 50% glycerol at −20°C and analyzed by SDS/PAGE (12% polyacrylamide gel) under reducing conditions with coomassie blue or silver staining.

The purified WT rOtuLi (5 μg) in Freund's complete adjuvant was used to immunize male BALB/c mice, followed by three biweekly immunizations in Freund's incomplete adjuvant. Immune sera were collected 5 days after the last immunization, diluted in glycerol (1:1) and analyzed by western blotting to evaluate the specific antibody production against the protein extract from L. infantum and the purified recombinant protein.

L. infantum promastigotes previously washed with PBS were submitted to the freezing/thawing process repeatedly in order to lyse cells and to obtain the soluble (SE) and insoluble (IE) protein extract. The proteins were transferred to a nitrocellulose membrane, blocked for unspecific antibody's interactions with TBS containing 5% milk followed by anti-OtuLi antibody (1:100) incubation and goat anti-mouse IgG horseradish peroxidase conjugated (1:30,000). Blotting analyses were developed using the ECL™ Prime Western blotting detection reagent (GE Healthcare) and performed in ImageQuant LAS 4,000 (GE HealthCare).

The enzyme reaction was performed with 250 ng of rOtuLi and 250 ng of K48- or K63-linked tetra-ubiquitin in 25 mM Tris pH 7.5 or 25 mM MES pH 5.5 for 5 h. For inhibition profile, 100 μM N-Ethylmaleimide (NEM) and 100 μM trans-Epoxysuccinyl-L-leucylamido(4-guanidino)butane (E-64) were incubated with rOtuLi for 30 min before addition of tetra-ubiquitin. Reactions were interrupted by addition of sample buffer (62.5 mM Tris–HCl pH 6.8, 2% m/v SDS, 5% v/v β-mercaptoethanol, 0.08% m/v bromophenol blue and 30% v/v glycerol) and analyzed by 12% SDS/PAGE, followed by silver staining.

The search for template was made in PDB—Protein Data Bank (www.rcsb.org/pdb) and the crystal structure of human otubain (2ZFY) (Edelmann et al., 2009) was chosen. The alignment of the target sequence (OtuLi) with the template (HsOtu1) was made using the T-Coffee software resulting in 32% identity and 44% similarity. Then, a model was constructed using the Swiss Model Server and the structural figures were generated with PyMOL system. The OtuLi model was validated using PROCHECK software (Laskowski et al., 1996).

The simulated ensembles are comprised of the macromolecule otubain from L. infantum, in both WT and inactive (F82S/F182S/L265P) forms, immersed in TIP3P water molecules (Berendsen et al., 1981). For each conformation, all-atom models of the non-bonded conformation and the OtuLi-Ubiquitin complex were constructed in GROMACS 4.6.7 using the CHARMM22 force field (Lazaridis and Karplus, 1999; Brooks et al., 2009; Pronk et al., 2013). Each protein was amidated at its C-terminus. All four different conformations were contained using periodic boundary conditions with unit cells corresponding to dodecahedral boxes with dimensions that enable adequate sampling without added complications arising from the proteins directly interacting with their mirror images. The four systems were then individually solvated with TIP3P water (wild non-bonded: 8,365 molecules TIP3P; wild OtuLi-Ubiquitin complex: 19,117 molecules TIP3P; Inactive non-bonded: 8,208 molecules TIP3P; inactive OtuLi-Ubiquitin complex: 20,124 molecules TIP3P), and salt ions were added to achieve electro-neutrality at a final salt concentration of 1.5M. After generating the solvated structures, the system was energy minimized using steepest descent until the change in potential energy between time-steps converged to machine precision. After energy minimization, the system underwent two phases of equilibration by using the Berendsen pressure and temperature coupling methods; it was first equilibrated with a time-step of 2 fs for 10 ps under constant pressure, and then was equilibrated with a time-step of 2 fs for 10 ps under constant pressure (Berendsen et al., 1984). Finally, the system underwent a 2 ns simulation of position restraint, in which the protein was restrained and the solvent molecules were relaxed through an energy minimization step under constant temperature by using the Berendsen temperature coupling method. All simulations were run under constant pressure and temperature (NPT). We performed a total of 200 ns of molecular dynamics simulations to sample the largest possible conformational space. During the production runs, the LINCS algorithm was used to constrain the lengths of bonds involving Hydrogen atoms, which vibrate faster than our simulation time-step of 2 fs (Hess et al., 1997), and the SETTLE algorithm was used to constrain the bond lengths within water molecules (Miyamoto and Kollman, 1992). Electrostatic calculations were performed using the Particle Mesh Ewald (PME) method with a cut-off radius of 1.4 nm (Darden et al., 1993). The same radius value cut-off was also used in the van der Waals calculations. The non-bonded list of each atom was updated every 10 simulation-steps. All simulations were performed using GROMACS 4.6.7 with the CHARMM22 force field with a machine temperature of 300 K.

Murine peritoneal macrophages (10⁶ cells/mL) from six C57BL/6 mice were harvested by peritoneal lavage and adhered for 16 h in RPMI medium supplemented with 2% FBS and non-adherent macrophages were removed by washing with PBS. Cells were stimulated with 1 μg of rOtuLi or the boiled denatured enzyme for 24 and 48 h at 37°C and 5% CO₂ atmosphere.

Peritoneal macrophages seeded in 24-well plate coverslips stimulated or not with rOtuLi were fixed in 3.7% (v/v) formaldehyde diluted in PBS. Then, the coverslips were washed three times with ultrapure water and incubated with Propylene Glycol (PEG) for 15 min. Subsequently, the coverslips were incubated with 0.5% Oil Red for 30 min and with 60% PEG for 1 min. After three washes, cells were stained with hematoxylin solution for 5 s. Finally, the slides were mounted with Aqua Poly/Mount medium (Vector). Lipid droplet biogenesis was assessed by quantification of lipid droplet structures on 50 consecutive cells by bright field microscopy.

The supernatants of the peritoneal macrophages stimulation assay with rOtuLi were collected at 24 and 48 h. IL-6 and TNF-α cytokines were measured by ELISA, using a commercial kit (eBioscience). The assay was performed following manufacturer's instructions and cytokine levels were demonstrated in absolute values (pg/mL).

For mitochondrion labeling, L. infantum promastigotes were washed with PBS, incubated in 80 nM MitoTracker Red CMXRos (Molecular Probes) for 1 h. Then, parasites were washed with PBS and immunofluorescence assay was conducted by fixing them with 3.7% paraformaldehyde with subsequent permeabilization with 0.1% Triton X-100 and then spread onto a poly-L-lysine coated glass slide. For OtuLi labeling, parasites were incubated with the primary antibody anti-OtuLi (1:100) for 2 h. For PDI labeling, an anti-PDI antibody from Gallus gallus was used in a 1:500 dilution for 2 h. After washing with PBS, parasites were incubated with one of the following antibodies: goat anti–mouse IgG conjugated with Alexa Fluor 488 (1:500) for OtuLi labeling, goat anti-rabbit IgG conjugated with Alexa Fluor 568 (1:300) for PDI labeling both for 2 h. For DNA staining, 100 ng/mL DAPI were used for 10 min and the slides were mounted with ProLong® Gold antifade (Invitrogen) and observed under a Nikon Eclipse TE 300 DV inverted microscope using appropriate fluorescence emission filters at room temperature. Data acquisition and processing were performed with MetaMorph and ImageJ software, respectively.

This study was carried out in accordance with institutional guidelines. The protocol was approved by Comissão de Ética de Uso Animal from the University of Brasilia, No. 27764/2016.

The graphics were generated using the Software GraphPad Prism. The results of lipid bodies and cytokines quantification were expressed as mean ± standard error of the mean (SEM) and statistically analyzed with analysis of variance (ANOVA) followed by t-Student test. The p significance level was broken down in the results.

Analysis of the L. infantum genome was performed using the Database of Ubiquitinating and Deubiquitinating Enzymes (http://www.DUDE-db.org/) (Hutchins et al., 2013), and provided 27 predicted proteins of which 16 are from the USP family, 5 from the JAMM family, 4 from the OTU family, and 2 from the UCH family. There were no predicted genes coding to proteases belonging to the MJD family. The otubain cysteine peptidase herein named OtuLi (XP_001464809.1) was selected among the predicted OTU enzymes for further characterization based on its low amino acids identity compared to the human otubain-1 (33%; AAO27702.1) and otubain-2 (26%; AAO27703.1).

The otuli gene has an open reading frame of 813 bp that codifies a protein of 270 amino acids corresponding to a predicted molecular weight of 30.3 kDa. The catalytic triad consists of the Cys81, His261, and Asn263 residues with an ubiquitin interaction motif (UIM) preserved. The ubiquitin-associated (UBA) domain and the nuclear localization signal which are commonly conserved in OTU family proteins and an extended N-terminal present in HsOtu1 are absent (Figure 1). Otubain is conserved between the subgenus Leishmania, mainly with L. donovani (99%) and L. major (98%). Regarding its counterpart in Trypanosoma cruzi, OtuLi shows only 39% identity.

Prior the enzymatic characterization of OtuLi, we obtained its recombinant enzyme (rOtuLi) by expression in prokaryotic system using vector pET19-b, which fused the enzyme to a histidine tail used for nickel-agarose purification. In order to better understand the catalytic mechanism, site-directed mutations were introduced into otuli sequence generating a single (F182S), a double (F182S/L265P), and a triple (F82S/F182S/L265P) variant. These mutations affected conserved amino acids that are located next to the Cys residue from the active site (F82), inside OTU domain (L265), or part of a hydrophobic surface (F182) important to the interaction with the ubiquitin molecule (Nanao et al., 2004). Variants were also obtained in prokaryotic system using the same expression vector. WT OtuLi, F182S and F182S/L265P variants were produced in abundance as both soluble and insoluble forms in BL21 (DE3) E. coli. However, the F82S/F182S/L265P variant was produced in small amount as soluble enzyme (Figure 2A). Nevertheless, WT and the 3 variants were successfully obtained as purified enzymes and used to perform enzymatic assays (Figure 2B). Native OtuLi was identified in soluble extract from promastigote forms of L. infantum using the polyclonal antibody against the WT rOtuLi produced in mice (Figure 2C).

The deubiquitinating activity was assessed by rOtuLi capability of cleaving K48- and K63-linked tetra-ubiquitin (4Ub) substrates at pH 7.5. rOtuLi cleaves preferentially K48-4Ub over K63-4Ub, as indicated by the complete cleavage to monoubiquitin (1Ub) in the K48-4Ub reaction and the partial activity of OtuLi against K63-4Ub with hydrolysis to tri-ubiquitin and di-ubiquitin (Figure 3A). Enzymatic activity was also assessed with fluorescent substrates such as Ub-AMC and Z-LRGG-AMC but rOtuLi was inactive (data not shown).

Considering the rOtuLi specifity for K48-4Ub, the inhibition pattern of the enzyme was assessed with this same substrate. Being OtuLi classified as a cysteine peptidase, the standard cysteine peptidase inhibitors E-64 and NEM were tested. Surprisingly, rOtuLi was not inhibited by E-64 and inhibited in a moderate manner by NEM, as we could see an intense band corresponding to 3Ub (Figure 3B). We also assessed the enzymatic activity in acidic environment and rOtuLi remained active at pH 5.5 (Figure 3C).

According to the homology modeling, the OtuLi model comprises 10 α-helixes and 4 β-strands that altogether constitute the β-sheet responsible for stabilizing the core of the protein. This portion provides the catalytic triad formed by C81, H261, and N263 residues (Figure 4A). There were significant changes between the WT OtuLi and the variants' models, mostly with F82S/F182S/L265P as shown by the overlapped models (Figures 4B–D) and by the RMSD values. Comparing the RMSD values, it was observed a difference of 1,512 Å (WT and F182S), 1,690 Å (WT and F182S/L265P), whereas the main difference (2,484 Å) was between WT and F82S/F182S/L265P. The F182S variant remains active against K48-linked substrate. However, the addition of L265P mutation (F182S/L265P) abolishes the enzyme activity against K48-4Ub, since no cleavage was seen with this mutant (Figure 4E, left). Regarding the enzymatic activity against K63-4Ub substrate, the F182S mutation was enough to confer inactivity to OtuLi (Figure 4E, right), implying that F182 residue is crucial to OtuLi activity against K63-linked substrates. Molecular dynamics simulations of the OtuLi-Ub complex showed an elevated degree of flexibility for some residues from the WT enzyme when compared to its free form. This flexibility is observed when we compare the root mean square fluctuation per residue of both the free and complex forms. Moreover, the same was not observed on F82S/F182S/L265P variant and Ub complex simulation (Figure 5). This lack of flexibility may explain the large amount of inclusion bodies observed on F82S/F182S/L265P variant produced in bacteria (Figure 2A), besides its inactivity against both substrates (data not shown).

In order to investigate the localization of OtuLi in promastigotes, a mitochondrion probe, or the Protein Disulfide-Isomerase (PDI) antibody (Florent et al., 2000) were employed to label mitochondrion and endoplasmic reticulum, respectively. OtuLi is presented in small vesicles along the cytoplasm of the cell, although an intense fluorescence can be observed near kinetoplast. There was no co-localization with mitochondrion or PDI (Figure 6). To examine whether rOtuLi could have an immunomodulatory effect, murine peritoneal macrophages were incubated with the enzyme and analyzed after 24 and 48 h of stimulation. rOtuLi triggered prominent lipid droplet biogenesis (a structural marker of inflammation) in macrophages compared both to unstimulated cells and macrophages stimulated with denatured rOtuLi, which presented three-fold less amount of lipid droplet after 48 h (Figure 7A). In addition, rOtuLi was able of triggering significant TNF-α and IL-6 secretion in macrophages compared to unstimulated cells after 24 h (Figure 7B).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.