Structural Characterization of Acidic M17 Leucine Aminopeptidases from the TriTryps and Evaluation of Their Role in Nutrient Starvation in Trypanosoma brucei

Leucine aminopeptidases (LAPs) catalyze the hydrolysis of the N-terminal amino acid of peptides and are considered potential drug targets. They are involved in multiple functions ranging from host cell invasion and provision of essential amino acids to site-specific homologous recombination and transcription regulation. In kinetoplastid parasites, there are at least three distinct LAPs. The availability of the crystal structures provides important information for drug design. Here we report the structure of the acidic LAPs from three kinetoplastids in complex with different inhibitors and explore their role in Trypanosoma brucei survival under various nutrient conditions. Importantly, the acidic LAP is dispensable for growth both in vitro and in vivo, an observation that questions its use as a specific drug target. While LAP-A is not essential, leucine depletion and subcellular localization studies performed under starvation conditions suggest a possible function of LAP-A in the response to nutrient restriction.

N eglected tropical diseases (NTDs) are a medically diverse group of infectious diseases mainly affecting people in the developing world. Among the NTDs are the diseases caused by the TriTryps, unicellular parasites of the order kinetoplastida, namely, Trypanosoma brucei, the causative agent of African sleeping sickness, Trypanosoma cruzi, which is responsible for Chagas' disease, and Leishmania species, which cause leishmaniases. Taken together, they are responsible for the suffering of millions of people.
African sleeping sickness (or human African trypanosomiasis [HAT]), is caused by T. brucei parasites and is ultimately fatal if left untreated. Estimates by the World Health Organization (WHO) state that approximately 20,000 current infections with 2,804 cases were recorded in 2015 (1). Additionally, T. brucei and the closely related species Trypanosoma vivax and Trypanosoma congolense have huge impacts on agriculture, causing a disease called nagana in cattle and other domestic and wild animals. Chagas' disease, caused by T. cruzi parasites, infects an estimated 8 million people in Central and South America. The leishmaniases are caused by parasites of the genus Leishmania and include different forms of the disease, i.e., cutaneous, mucocutaneous, and visceral leishmaniases. The WHO estimates 700,000 to 1 million new cases and 20,000 to 30,000 deaths annually (2).
Control of these diseases is difficult, as the parasites have complex life cycles involving insect vectors. Current treatments for HAT (3), Chagas' disease (4), and leishmaniasis (5) are highly toxic, the treatment regimens are difficult to implement, and the availability of some of them is limited. Emerging resistance and limited efficacy of available drugs emphasize the need for the development of new drugs with novel targets.
Peptidases are central enzymes in parasite metabolism, and some were shown to be potential drug targets in parasitic protozoa (6)(7)(8), as in the case of Plasmodium falciparum, the causative agent of malaria. Furthermore, the aminopeptidase inhibitor arphamenin A was shown to inhibit the growth of T. brucei subsp. brucei cells in vitro, suggesting aminopeptidase as a potential drug target in these parasites (9).
Leucine aminopeptidases (LAPs) are found in all kingdoms of life, with many species encoding more than one LAP. They catalyze the metal-dependent hydrolysis of the N-terminal amino acid of peptide or amino acyl substrates (EC 3.4.11.1, EC 3.4.11.10, EC 3.4.11.5, and EC 3.4.11.23). The structures of LAPs from many different species have been solved (10)(11)(12)(13), and all are homologous with overall root mean square deviations (RMSDs) between 0.27 and 0.35 Å and a sequence identity to the TriTryp LAPs of 32 to 45%. LAPs of the M17 family of aminopeptidases are bilobal proteins with a highly conserved C-terminal domain harboring the aminopeptidase activity. They form homohexamers of approximately 330 kDa with the active sites facing a central cavity separated from the bulk solvent but accessible through solvent channels. The aminopeptidase mechanism of LAPs is well characterized (14), with the activity being reported to be dependent on the presence of two cocatalytic zinc ions, although activity was shown for other divalent metal ions, e.g., manganese and magnesium (15). In contrast, the biological functions are diverse and less well understood. In bacteria, LAPs were shown to be involved in site-specific homologous recombination and transcription regulation (16,17); in plants, LAPs moonlight as molecular chaperones independently of the aminopeptidase activity (18); in mammalian cells, they are implicated in the N-terminal processing of certain proteins (19) and have been proposed to have a role in the determination of cell redox status (20); while in some parasites, they participate in host cell invasion and the provision of amino acids essential for growth (21,22).
There are four distinct LAPs in the TriTryp genomes, which can be classified according to their theoretical isoelectric points as acidic (LAP-A and LAP-A2), neutral (LAP-N), and basic (LAP-B). The sequence identity among the paralogues LAP-A, LAP-N, and LAP-B within a species is low, with the C-terminal domain and active site being more conserved. However, inspection of the lap-A2 genes reveals what appears to be an incomplete LAP in all three species and their genome surroundings might suggest a gene duplication forming a nonfunctional pseudogene. The subcellular localization of each of the paralogues has been predicted with PSORT II (23) to be different, indicating distinct functions. The biochemical characterization of TriTryp LAPs to date is limited; the enzymes studied include LAP-A from T. cruzi (24) and LAP-B from L. major, L. donovani, L. amazonensis (25), and Trypanosoma brucei, where it has recently been shown to be required for efficient kinetoplast DNA (kDNA) division (26).
This work focused on the LAP-As of L. major (LmLAP-A), T. brucei (TbLAP-A), and T. cruzi (TcLAP-A). We report the X-ray structures of TbLAP-A in the metal-free (apo) and unliganded (metal-containing, holo) forms and in complex with the inhibitors bestatin and actinonin, as well as the structures of holo TcLAP-A and LmLAP-A in complex with actinonin. In addition, we discuss the requirements of LAP-A for survival in T. brucei by using RNA interference (RNAi), double knockout (dKO), and overexpression of the protein in bloodstream form (BF) and procyclic form (PF) T. brucei subsp. brucei parasites in culture. Finally, in vivo experiments with mice were carried out to investigate the effect of LAP-A knockdown on parasite infectivity.

RESULTS
Expression, purification, and confirmation of the proteolytic activity of the LAP-As. The TriTryp LAP-As were expressed in Escherichia coli and purified successfully by affinity and size exclusion chromatography (SEC). The proteolytic activities of the purified enzymes against the fluorescent substrate analogues Leu-L-leucine-(4-methyl-7-coumarinylamide) (Leu-AMC), L-valine-7-amido-4-methylcoumarin (Val-AMC), and L-proline-7-amido-4-methylcoumarin (Pro-AMC) were confirmed by native gel zymography (see Fig. S1 in the supplemental material). All three LAP-As were active against Leu-AMC, while little or no activity against Val-AMC and Pro-AMC was detected. These findings are in agreement with the expected preference for substrates containing leucine as the N-terminal residue, as previously reported for this family of aminopeptidases (27).
Crystallization of the LAP-As. TbLAP-A crystals in space group P2 1 2 1 2 1 with six chains in the asymmetric unit grew under citrate-containing conditions and comprise apo TbLAP-A with no metal ions in the active site. TbLAP-A complexes crystallized under many different conditions, preferentially in space group P2 1 with 6 or 12 chains per asymmetric unit. All attempts to cryoprotect the crystals prior to vitrification resulted in substantial loss of diffraction quality, and only the apo TbLAP-A crystals were successfully cryoprotected. TcLAP-A crystallized only under citrate-containing conditions, resulting in crystals in space group R32 with one or both metal ions missing from the active site. None of the cocrystallizations with ligands yielded diffracting crystals. LmLAP-A crystallized under many different conditions, forming hexagonal plate-like crystals, one of which, grown in the presence of actinonin, diffracted to 2.5 Å and was in space group P321. The crystallization conditions leading to structure solutions are summarized in Table S1.
The TriTryp LAP-A structures. The TriTryp LAP-A structures were solved by molecular replacement (MR), and the crystallographic data and statistics are summarized in Tables S2 to S5. The structures show a close structural similarity to each other and LAPs of other species in the Protein Data Bank (PDB). Their nearly identical core structures have overall RMSDs for TcLAP-A and LmLAP-A toward TbLAP-A of 0.31 and 0.33 Å, respectively, with differences limited to loop regions and minor changes in the length of secondary structure elements.
Overall and quaternary structures in the crystals. The structures of all three LAP-As show the typical hexameric architecture of the M17 metallopeptidases (Fig. 1A), and analysis with PISA (28) supports the hexamer being the stable oligomeric state in solution. All of the conformational differences seen between the TriTryp LAP-As are located at the surface of the hexamer, away from the active site or protomer-protomer interfaces.
Like other M17 family LAP structures in the PDB, the TriTryp LAP-As have two domains (N and C terminal) (Fig. 1B). The smaller N-terminal domain (amino acids [aa] 2 to 197) consists of a six-stranded ␤ sheet (␤1, ␤6, ␤5, ␤2, ␤4, and ␤3) flanked by five ␣ helices (␣2 to ␣6). Unlike other LAPs in the PDB, TriTryp LAP-As have an ␣ helix at the N terminus (␣-1) that interacts with the neighboring protomer of the hexamer. The N-terminal domain is connected to the C-terminal domain via a long ␣ helix (␣7). The highly conserved C-terminal domain (aa 215 to 521), nearly identical to the structures of other LAPs in the PDB, shows a central eight-stranded ␤ sheet (␤7, ␤8, ␤10, ␤9, ␤13, ␤17, ␤14, and ␤15) surrounded by ␣ helices (␣8 to ␣17) and a small ␤ sheet (␤11, ␤12, and ␤16). The C-terminal domain contains the active site and forms the core of the hexamer, arranged around a central 3-fold axis with the active sites facing the central cavity, secluded from the surrounding bulk solvent.
Structural differences in LmLAP-A compared to TbLAP-A and TcLAP-A are mostly located in the loop regions. The secondary structure elements and their positions in LmLAP-A are nearly identical to those in TbLAP-A, with the exceptions of ␤1, ␤3, and ␤4, which are shortened, and ␣2, which is replaced by a loop. The N terminus of LmLAP-A is extended and forms a loop before the start of the ␣1 helix. The loop regions noted as different for TcLAP-A are extended in LmLAP-A and again adopt different conformations. Additionally, the loop between ␤6 and ␣7 is extended, harboring a small ␣ helix, and helix ␣11 in the C-terminal domain is extended by one turn.
Metal coordination in the active site and apo LAP-A. The metal-coordinating amino acid residues in the TriTryp LAP-As are conserved and identical to those in previously reported LAP structures. The two metal binding sites were shown to have different binding affinities for the metal ions, with site 1 generally having a lower affinity than site 2 (15,29). In TbLAP-A, the carboxylate oxygen atoms of D294 and E373 and the main-chain and carboxylate oxygens of D371 coordinate the metal ion in site 1 and the carboxylate oxygens of D294, E373, and D312 and the -amino group of K289 coordinate that in site 2 (Fig. 2). The corresponding residue numbers in TcLAP-A are K288, D293, D311, D370, and E372, and those in LmLAP-A are K299, D304, D322, D389, and E391.
The holo TbLAP-A (crystallized in the presence of leucine), TbLAP-A complexes, and LmLAP-A structures have two metals bound in the active site; the TcLAP-A structures, in contrast, clearly show only one metal ion in site 1 and no metal in site 2. However, the active-site residues adopt the same conformations as if two metal ions were bound.
Most LAP-A crystals were grown in the presence of manganese ions. As LAPs are reported to be zinc-dependent enzymes, crystals of EDTA-treated TbLAP-A were grown in the presence of ZnCl 2 . Nevertheless, X-ray fluorescence (XRF) scans of the resulting crystals indicated the presence of manganese ions, suggesting that EDTA may not to be able to efficiently chelate the Mn 2ϩ ions at the concentrations used. A second data set was collected from the same crystal at a wavelength of 1.3052 Å, between the absorption edges of manganese and zinc, and an anomalous difference Fourier synthesis was calculated for the data sets at both wavelengths by using phases from the refined model (Fig. 2). The map for the 1.3052-Å wavelength data showed peaks at around 10 for both metal sites, substantially larger than the 7 peaks for the original data recorded with a wavelength of 0.97949 Å, clearly indicating that neither site was occupied by zinc. The metals were therefore modeled as manganese ions.
The structures of LAP-As in their apo and holo forms and in complex with inhibitors are practically identical. The exception is the metal-free apo TbLAP-A form, whose active-site loop (aa 293 to 305) is disordered and in which no electron density for bound metal ions could be detected, most likely because of the high concentration of citrate in the crystallization buffer. Apo TcLAP-A and the other LAP-A structures show completely ordered active sites independent of metal or ligand binding.
The TbLAP-A--bestatin complex. TbLAP-A was cocrystallized with the inhibitor bestatin. The electron density of the ligand was well defined in all 12 chains in the asymmetric unit and was modeled in the same conformation (Fig. 3A). The bestatin conformation mimics the gem-diolate intermediate of the peptide substrate in the process of its cleavage, and its binding in TbLAP-A is identical to that in PfLAP (PDB code 3KR4). The coordination of bestatin is mainly facilitated through the Mn 2ϩ ions and hydrogen bonding. The Mn 2ϩ ions coordinate the N-2 nitrogen atom and O-2 and O-3 oxygen atoms. Additionally, N-2 is coordinated through hydrogen bonds with the side chain oxygens of D294 and D312, as well as the main-chain oxygen of T401. The main-chain oxygen of D371 and the side chains of K289 and D294 form hydrogen bonds with O-2, while the side chain of K301 forms a hydrogen bond with O-3. Furthermore, bound HCO 3 Ϫ forms a hydrogen bond with the O-2 atom. While hydrophobic residues surround the phenyl ring of bestatin, its carboxyl group forms hydrogen bonds with G404 and R467.
The actinonin complexes. For both the TbLAP-A and LmLAP-A complexes, the electron density was sufficient to model the ligand, albeit with less than 100% occupancy. The limited resolution and quality of the data sets do not allow accurate determination of the ligand's conformation. The TbLAP-A--actinonin complex had six molecules in the asymmetric unit, and local noncrystallography symmetry was imposed for the protein chains. The electron density for the ligand was generally poor but sufficient to build the actinonin in all six chains. Their densities varied slightly, so that the ligand was modeled in each chain individually, with variation in their conformations. Nevertheless, the actinonin coordination was identical in all chains and differed only slightly from that in LmLAP-A (Fig. 3B).
Actinonin in LmLAP-A and TbLAP-A is coordinated by the Mn 2ϩ ions and hydrogen bonds. In both structures, the two Mn 2ϩ atoms coordinate the O-2 and O-4 oxygen atoms of the ligand. In LmLAP-A, the side chains of K299, D304, K311, and D389 form hydrogen bonds with O-2 and O-4, and the main-chain oxygen of L420 and the main-chain nitrogen of G422 form hydrogen bonds with N-1 and O-13, respectively. The N-14 atom of actinonin coordinates a water molecule. The corresponding residues in TbLAP-A are K289, D294, K301, D371, L402, and G404, and A405 forms an additional hydrogen bond with O-27 in four of the six chains. Because of the limited data quality, no water molecules were modeled in the TbLAP-A structure, with the exception of those replacing the HCO 3 Ϫ molecule. Superposition of the active sites including the actinonin molecules reveals the high similarity of the conformations actinonin adopts, with the only difference being in the hydroxymethyl-pyrrolidine moiety, which forms a Amino acid residues and ligands are colored by atom type, with the C in ice blue (TbLAP-A) or gold (LmLAP-A). In both, the proteins were first superposed by using the SSM option in CCP4mg for residues from the C-terminal domain (300 to 500), followed by the "match close residues" option. hydrogen bond in TbLAP-A but not in LmLAP-A, where it is bent away from the corresponding residue (A424).
TcLAP-A in complex with citrate. TcLAP-A crystallized exclusively under citratecontaining conditions, which led to the absence of one or both Mn 2ϩ ions from the active site and hence no electron density for the inhibitors. The structure obtained from cocrystallization with the inhibitor amastatin showed clear density for one partially occupied Mn 2ϩ ion and a citrate molecule bound in the active site. There was no electron density for amastatin in this structure, and HCO 3 Ϫ is replaced by SO 4 2Ϫ (Fig. S2) The LAP-As crystallized as hexamers, as reported for other LAP structures in the PDB. To confirm the hexamer as the oligomeric state in solution, SEC-multiangle laser light scattering (SEC-MALLS) analyses of all three LAP-As were carried out at a range of protein concentrations. The expected molecular masses for the different oligomeric states are summarized in Table S6 and were used to annotate the elution peaks obtained from the SEC-MALLS data allowing for an error of Ϯ10%. An overlay of the differential refractive index traces at a protein concentration of 1.0 mg·ml Ϫ1 with annotated peaks according to the oligomeric states corresponding to the molecular masses is shown in Fig. S3. The analysis of TcLAP-A showed the presence of two well-separated peaks, one with a calculated molecular mass of 320 kDa and the other with a mass of~170 kDa, corresponding to the hexamer and trimer, respectively. At all of the concentrations analyzed, the hexamer was the most abundant species, while as the concentration was lowered, the proportion of trimers increased. Unlike for TbLAP-A, the hexameric and trimeric species were separated well on the column, which suggests a slower interconversion of the hexamer and trimeric TcLAP-A species than in TbLAP-A. Analysis of LmLAP-A showed three distinct molecular mass species, corresponding to the hexamer, trimer, and monomer, plus a considerable amount of aggregation. The three species were well separated from each other and the aggregates in the void volume. The lower the protein concentration, the greater is the proportion of monomers. Whether the monomers are a result of instability and subsequently result in their aggregation or if the monomeric state is physiologically relevant remains unclear and requires further investigation.
Knockdown and knockout of LAP-A do not impair T. brucei proliferation. To explore the potential of TbLAP-A as a drug target and assess its essentiality for T. brucei survival, the expression of the enzyme in both BF and PF T. brucei cells was knocked down by using an RNAi stem-loop/stuffer strategy. This created trypanosome lines that expressed a 465-bp fragment of the TbLAP-A gene under the control of a tetracyclineinducible promoter.
TbLAP-A gene knockdown in BF and PF parasites was monitored by Western blotting (Fig. 4A and B, respectively), and the resulting phenotype was examined in different media. The RNAi-mediated depletion of LAP-A approached 95 and 100% in BF and PF parasites, respectively, and was maintained throughout the period studied. With regard to the effect of LAP-A depletion on parasite proliferation, RNAi-depleted strains of BF parasites cultured in HMI-9 medium showed a growth rate similar to that of controls (Fig. 4C). When dialyzed fetal bovine serum (FBS) was used, the growth rate was slightly diminished yet similar in both control and LAP-A-depleted cells (Fig. 4C). For PF parasites, RNAi-mediated knockdown of the enzyme did not result in significant growth defects in SDM-79 medium (Fig. 4D). In addition, a LAP-A knockout strain was achieved for PF cells. Efficient elimination of enzyme expression is shown in Fig. 5A. In minimal essential medium (MEM) containing dialyzed FBS and therefore devoid of glucose, a growth rate reduction of~20% was observed in both control and dKO parasites (Fig. 5B). In summary, LAP-A is not an essential enzyme for the in vitro growth of either form of the T. brucei subsp. brucei parasite when it is cultured in a range of media.
Role of LAP-A in leucine starvation. LAPs cleave amino acids, preferentially leucine, from the N terminus of proteins or peptides and are important for protein recycling and turnover. It is reasonable to speculate that the depletion or overexpression of an aminopeptidase will have an increased effect on cell fitness upon exposure to suboptimal extracellular amino acid concentrations. In the event of LAP-A depletion, extracellular leucine levels might be particularly important. To test the effect of leucine  limitation on the growth and cell cycle of PFs of T. brucei subsp. brucei, cultures were grown in glucose-free MEM prepared without leucine and with dialyzed FBS to ensure that no small peptides or undefined amino acids were present. Parental and LAP-A-dKO cells from cultures grown in MEM containing leucine (with dialyzed FBS) were transferred to MEM without leucine and incubated at culture temperature. In the absence of leucine, both the parental and LAP-A-dKO parasites grew deficiently (Fig. 5C).
Additionally, to provide an insight into the function of LAP-A in trypanosomes, overexpressing cell lines were generated and the resulting phenotype was analyzed. Overexpression of LAP-A fused to a myc tag in both BF and PF parasites was induced by the addition of doxycycline. LAP-A overexpression in BFs was minor (~1.2-fold) and had no significant effect on proliferation, while LAP-A-overexpressing (LAP-A-OE) PF parasites showed significantly increased levels of LAP-A (2.5-to 3-fold), as assessed by Western blotting (Fig. 5D), yet grew at a rate similar to that of controls (Fig. 5E). Overexpressing PFs were further grown in MEM containing dialyzed FBS to monitor whether the presence of high levels of the enzyme confers a selective advantage under conditions of restricted nutrients. Lower growth rates were obtained with all of the cell lines tested compared to SDM-79, and LAP-A overexpression appeared to be slightly detrimental under these conditions (Fig. 5E). However, when cultured in glucose-free MEM without leucine, LAP-A-OE parasites grew at a higher rate than controls (Fig. 5C).

Subcellular localization of LAP-A in T. brucei.
BFs and PFs of T. brucei were grown to mid-log phase and fixed on poly-L-lysine-coated glass slides for determination of the intracellular localization of TbLAP-A. Mitochondria were stained with fluorescent Mito-Tracker Red, and nuclear DNA and kDNA were visualized with 4',6-diamidino-2phenylindole (DAPI). TbLAP-A was detected with affinity-purified rabbit anti-TbLAP-A antiserum. Cells were imaged by confocal and wide-field fluorescence microscopy (Fig. 6). Parasites overexpressing c-Myc-tagged LAP-A (2 days of induction) were also analyzed with a c-Myc-specific antibody. c-Myc-tagged LAP-A exhibited a cytosolic distribution similar to that obtained with the anti-TbLAP-A antibody in the parental cell line. Colocalization analysis indicated a minor overlap with mitochondria (Pearson's coefficient, 0.5 Ϯ 0.1; Manders' overlap coefficient, 0.27 Ϯ 0.07). Thus, while TbLAP-A is mainly cytosolic, a minor level within the mitochondrion in BFs and PFs cannot be fully excluded. The staining disappeared upon RNAi-mediated knockdown and in LAP-A-dKO cells, validating the specificity of the polyclonal antibody and the efficacy of RNAi-mediated depletion (Fig. 6).
Effect of nutritional starvation on LAP-A subcellular localization. Under stress conditions such as nutritional starvation, eukaryotic cells activate several mechanisms to recycle the cellular content, which may cause changes in the localization of the enzymes involved in these processes. Considering the potential participation of LAP-A in protein recycling and amino acid supply, we sought to examine the effect of nutritional starvation on its subcellular localization and expression. T. brucei PF parasites were starved for 60, 120, or 180 min by incubation in phosphate-buffered saline (PBS) and then subjected to immunofluorescence analysis with the anti-TbLAP-A antibody. Under normal conditions (untreated cells), LAP-A is homogeneously spread all over the cytoplasm (Fig. 7A and B). However, upon nutritional starvation, the enzyme progressively relocalized and accumulated in specific regions, especially in the cytoplasmic perinuclear area (Fig. 7A and B). This phenomenon was quantified by determining the number of particles containing LAP-A in each cell while applying an automatic threshold. In untreated cells, LAP was present at an average of three particles per cell, one of which corresponds to almost the entire cytoplasm, while during starvation, the number of particles significantly increased (Fig. 7C). The shape of the particles resulting from LAP-A accumulation in foci during starvation was also analyzed by measuring circularity, which corresponds to 1 in the case of a perfect circle and 0 in the case of an infinitely elongated polygon. LAP-A-containing particles were significantly more circular in starved cells than in untreated cells (Fig. 7D). Finally, quantification of the LAP-A signal intensity in the perinuclear region showed that significant accumulation of the protein occurs in this area during PBS treatment ( Fig. 7E and F). However, the total expression of TbLAP-A remains almost unchanged or slightly decreases during starvation (Fig. 7G). In mammalian cells, it has been reported that lysosomes and autophagic vesicles cluster in the perinuclear area, driven by intracellular pH changes in response to nutritional starvation. Likewise, a perinuclear localization of autophagosomes has been observed in trypanosomatids. Thus, TbLAP-A relocalization during nutritional starvation suggests a possible involvement of TbLAP-A in protein degradation that occurs in response to nutrient restriction.

Importance of LAP-A in vivo.
To evaluate the role of TbLAP-A in parasite survival within the mammalian host, C57BL/6J mice were infected with the TbLAP-A RNAi and parental strains, downregulation in vivo was induced by doxycycline administration, and the progression of the infection was monitored. After a few days, both the parental and TbLAP-A RNAi cell lines showed high levels of parasitemia (10 9 parasites ml Ϫ1 ) in mice that consequently resulted in death within 5 to 6 days postinfection (Fig. 8A), indicating that TbLAP-A is not essential for parasite viability in vivo. Efficient depletion of the enzyme in vivo was monitored by Western blotting (Fig. 8B).

DISCUSSION
LAP-A as a potential drug target in T. brucei. The data obtained indicate that LAP-A is not essential for survival and proliferation of BF and PF T. brucei subsp. brucei parasites under the conditions tested either in vitro or in vivo. RNAi-mediated depletion in BF parasites and both RNAi-mediated depletion and the generation of dKO mutants in PFs support this conclusion. Whether this is the result of redundancy caused by the presence of additional M17 LAPs, paralogues of the one investigated, requires further research. Recent evidence obtained with other parasites also questions the essential character of LAP-A. Thus, bestatin, a broad-spectrum metalloaminopeptidase inhibitor (30) reported to inhibit the M17 LAP of T. cruzi (24), was not cytotoxic against T. cruzi after treatment at 300 M for several hours (31) under conditions where multiple dipeptides were shown to be increased, confirming in situ inhibition of the LAP. In addition, a study with Toxoplasma gondii has shown that LAP knockout has a moderate effect on growth but does efficiently block development or virulence, suggesting that it is inadequate as a drug target also in this parasite (32). Conversely, TbLAP-B has been recently shown to be required for correct segregation of kDNA, with knockdown resulting in delayed cytokinesis and growth defects (26).
The observation that LAP-A is not an essential protein in T. brucei contrasts with its function in P. falciparum, where its drug target potential has been demonstrated (33,34). Likewise, it does not have an important role in survival in the mammalian host since the infectivity of RNAi-depleted parasites in mice was similar to that of controls. Because of the profound differences in metabolism and life cycle stages among the TriTryps, the results obtained with T. brucei subsp. brucei parasites do not allow conclusions to be drawn about LAP-A's potential as a drug target in the intracellular parasites T. cruzi and L. major, where it may be involved in host cell invasion.
TbLAP-N is predicted to localize to the endoplasmic reticulum by PSORT II (23), while TbLAP-B has been shown to localize to the kinetoplast (26) and has a function distinct from that of LAP-A, which is cytosolic. The establishment of subcellular LAP-A as a cytosolic enzyme may contribute to the understanding of its physiological function, which would be distinct from the potential endoplasmic reticulum neutral aminopeptidase. Thus, while LAP-A would perform metabolic or housekeeping functions in the cytosol, in the endoplasmic reticulum, it could be involved in the posttranslational modification of N-terminal signal peptides of specific proteins, performing a regulatory role. In the present study, the enzyme appears to be mainly cytosolic and mostly excluded from the nucleus, and hence, an involvement in site-specific homologous recombination, as reported in E. coli (35,36), also seems improbable. A similar localization of LAP-B was described in immunocytochemical studies with antibodies in L. amazonensis promastigotes, where diffuse fluorescence throughout the parasite cytosol was observed (25). While not essential for efficient proliferation, LAP-A may be involved in sensitivity to leucine and nutrient starvation. In mammalian cells, leucine starvation has been shown to induce increased lysosome-dependent protein breakdown to replenish amino acid pools (37). In T. brucei PF cells, starvation stress induces the presences of a functional autophagy pathway (38). PF parasites exposed to leucine depletion grew at very low rates and entered a short stationary phase before undergoing cell death. LAP-A-OE parasites appear to exhibit consistently a growth advantage under these conditions, which suggests a role for the enzyme in protein breakdown and the maintenance of leucine pools. Interestingly upon nutrient starvation, the location of LAP-A changed and concentrated in vesicles in the nuclear periphery. While the nature of these vesicles remains to be established, we postulate that LAP-A is recruited for specific functions related to protein and amino acid homeostasis and may be important for the cellular response to events related to a limited nutrient supply.
Structure and oligomeric state of LAP-A and their effects on possible physiological substrates. The TriTryp LAP-As form hexamers in solution and in the crystal, and homology to other LAPs indicates that this oligomeric state is a key feature of this family of enzymes. The active site of each protomer faces the inner cavity, sequestered from the bulk solvent. Although access channels allow diffusion into the inner cavity and the active sites, the accessibility seems restricted to small, unfolded substrates. This favors LAP-A's function as a mainly metabolic enzyme whose function is housekeeping or trimming of the N termini of specific unfolded peptides. Nevertheless, the bacterial M17 family LAPs PepA and PepB were shown to cleave the loosely folded protein casein in vitro (39), indicating LAP's potential to process larger substrates.
The physiological metal ion in the active site of LAP-A. While early studies on the bovine lens LAP showed Zn 2ϩ to be the physiological metal ion in the active site of the M17 family LAPs (15), the enzyme was also shown to be active with other divalent metal ions (24,25,(40)(41)(42). It was noted earlier (12) that Pseudomonas putida LAP produced in E. coli has Mn 2ϩ in its active site, with the substrate specificity being influenced by the identity of the metal in the active site (20), with sulfhydryl-containing substrates only being accepted by Mn 2ϩ -containing LAPs and not by Zn 2ϩ -containing LAPs.
Within this study, the best crystals of the TriTryp LAP-As grew in the presence of Mn 2ϩ and indeed, crystals grown under Zn 2ϩ -containing conditions still retained Mn 2ϩ in the active site. This might indicate that Mn 2ϩ is the physiological metal copurified with the enzyme expressed in E. coli. Given the similar estimates of manganese and zinc ion concentrations on the order of 10 to 100 M in the cytosol, further studies are required to confirm which of these is the physiological metal in these enzymes.
In summary, the structures of the TriTryp LAP-As were solved in the apo form and several ligand complexes, which gives insight into ligand binding and confirms the homology to other family members. The biological data show that LAP-A is not essential either for growth in vitro or for infectivity in mouse models, therefore precluding its exploitation as a drug target. Protection against leucine depletion and the behavior of the enzyme during starvation suggest a possible role in the response to nutrient restriction.

MATERIALS AND METHODS
Cloning of LAP-A genes and protein expression in E. coli. The genes encoding the LAP-As were amplified from genomic DNA (T. brucei Lister427, T. cruzi Sylvio X10/1, and L. major Friedlin) and cloned into the YSBLIC3C vector by ligase-independent cloning (43). Successful cloning was confirmed by restriction endonuclease digestion and sequencing. The resulting constructs (pTbLAP-3C, pTcLAP-3C, and pLmLAP-3C) encoded the LAP-A proteins with an N-terminal fusion to a hexahistidine tag separated from the N terminus of the protein by the recognition site of the human rhinovirus (HRV) 3C protease. Recombinant full-length TbLAP-A was overexpressed from the pTbLAP-3C vector into BL21(DE3) cells grown in LB medium supplemented with 35 g·ml Ϫ1 kanamycin. TcLAP-A and LmLAP-A recombinant proteins were expressed from the pTcLAP-3C and pLmLAP-3C vectors into pLEMO(DE3) cells grown in LB medium supplemented with 35 g ml Ϫ1 kanamycin and 34 g ml Ϫ1 chloramphenicol. TbLAP-A expression was induced with 1 mM isopropyl-␤-D-thiogalactopyranoside (IPTG) for 4 h at 37°C. TcLAP-A and LmLAP-A expression was induced overnight at 20°C with 1 mM IPTG. Cells were harvested by centrifugation, and the pellet was frozen at Ϫ20°C.
Crystallization, data collection, and structure determination. Prior to crystallization, the LAP-As were thawed at room temperature and preincubated with the desired metal salts for 2 h and subsequently with ligand for another 2 h on ice. Precipitated protein was removed by centrifugation, and the final protein concentration was determined. All crystals were grown at 20°C. Crystals were optimized by sitting-and hanging-drop vapor diffusion with a drop size of 1 l of protein plus 1 l of well solution. For LmLAP-A, the additive screen (Hampton Research) was used to identify a compound that enhanced the diffraction of the crystals. The conditions leading to LAP-A crystals that allowed structure solution and the ligands bound in the active sites of these structures are listed in Table S1. The crystal of apo TbLAP-A was cryoprotected in well solution containing 20% (vol/vol) glycerol prior to vitrification. The other LAP-A crystals were not cryoprotected.
All computations were carried out with programs of the CCP4 suite (44), unless stated otherwise. The data were processed with MOSFLM (45) and AIMLESS (46). Crystallographic data and statistics are shown in Tables S2 to S5.
The structures were solved by MR with PHASER (47) or MOLREP (48). The highest structural homology to TbLAP-A was predicted to be the Caenorhabditis elegans LAP by BLAST searching against entries in the PDB, and the protomer of PDB entry 2HB6 was used in MR to generate initial phases from which the TbLAP-A model was built. For all subsequent LAP-A structures, the first TbLAP-A model (apo) was used for MR. Model building was carried out with BUCCANEER (49), and manual fitting of the electron density maps and model building and fitting of the ligands were carried out in Coot (50). The structures were refined with REFMAC5 (51) and validated with Coot and MolProbity (52).
Identification of metal ions in the TbLAP-A structure. The crystal grown in the presence of ZnCl 2 and L-leucine used protein treated with EDTA during purification to remove any metals, and no other metal except ZnCl 2 was added prior to crystallization. XRF scans of this crystal were collected and analyzed with the PyMca software (53), but a specific edge scan for zinc gave no signal. X-ray diffraction data were collected at each side of the theoretical zinc edge at wavelengths of 0.9795 and 1.3051 Å and after data processing with Xia2 (54) and structure solution and refinement as described above. The reflection data from REFMAC5 (including the phases) were combined with the Friedel difference (DANO) and the standard deviation of F2 (SIGDANO) from the unmerged data file (from Xia2) by CAD from CCP4, and anomalous difference maps were generated with the FFT program in CCP4.
Anti-TbLAP antibody generation. Ten milligrams of pure recombinant TbLAP-A was lyophilized, resuspended in PBS, mixed with Freund's adjuvant (1:1 ratio), and injected into a rabbit in four inoculations of~400 g of protein each to obtain an anti-TbLAP polyclonal antibody. Antiserum was extracted before the inoculation and tested for the recognition of TbLAP-A in T. brucei lysates and purified recombinant protein. The antibody was affinity purified by using pure recombinant TbLAP-A coupled to Affi-Gel 15 Gel resin (Bio-Rad) in accordance with the manufacturer's instructions. Antibody sensitivity and specificity were tested via Western blot analysis (Fig. S4).

Generation of LAP-A-RNAi, LAP-A-dKO, LAP-A-OE, and LAP-A-myc lines.
A fragment of 465 bp (positions 10 to 475 of the coding region) of the TbLAP-A open reading frame (ORF) sequence was amplified by PCR from genomic DNA (T. brucei Lister427) with forward primer 5= GCGGATCCAAGCTTT ACCCAAGGCCGAAGCAAAA 3= and reverse primer 5= GCGTTAACGGGCCCCGTGCTACAGCAGCAGCGAT 3= containing the recognition sites (underlined) for BamHI-HindIII and HpaI-ApaI, respectively. The primers were designed by using the sequence found in the GeneDB database (Tb427tmp.02.4440) and validated with the TrypanoFAN-RNAit software (55). After a first cloning into the HindIII and ApaI sites of plasmid pGR19 (56), the fragment was cloned in the BamHI and HpaI restriction sites of the previous construct in the antisense orientation to create a hairpin structure of the transcribed RNA. The plasmid obtained was used to induce downregulation of TbLAP-A in T. brucei by RNAi. To obtain stable TbLAP-A depletion, two replacement cassettes for the dKOs of the gene were generated. A 538-bp fragment of the 5= untranslated region (UTR) upstream of the TbLAP-A gene (positions Ϫ61 to Ϫ599) and a 467-bp fragment of the 3= UTR downstream of the TbLAP-A gene (positions ϩ7 to ϩ475) were amplified from genomic DNA with 5= UTR forward primer 5= CGCGGCCGCGGGAGGCGGTAATGACAAGTC 3= (NotI restriction site underlined), 5= UTR reverse primer 5= GCCTCGAGGCTTCAGGTCCCCTTATGCAC 3= (Xho1 restriction site underlined), 3= UTR forward primer 5= GCAGGCCTGGCGTGCAAGTATCGGTTCCC 3= (Stu1 restriction site underlined), and 3= UTR reverse primer 5= GGCTAGCGCGGCCGCGTAGCGAGTAAACGAGTAAACTCA 3= (Nhe1 and Not1 sites underlined). The 5= UTR fragment was cloned into the pHD887 vector (57) with the XhoI and NotI endonucleases. Subsequently, the 3= UTR fragment was cloned into the resulting vector from the previous step with StuI and NheI. The resulting vector (pHD887B/UTRs) contained the blasticidin S transferase resistance gene and was used to transform parental T. brucei parasites for the suppression of the first allele. To remove the second TbLAP-A allele from the genome, the hygromycin phosphotransferase (HYG) selectable marker was subcloned into pHD887B/UTRs. The resulting vector, pHD887H/UTRs, was used to transfect the positive single-knockouts clones. In addition, the TbLAP-A gene (Tb427tmp.02.4440) ORF was amplified by PCR and cloned into vectors pGRV23b and pGRV33 (57) with ECC forward primer 5= GCCATATGATGCCTACTTTACCCAAGGCC 3= (Nde1 site underlined) and ECC reverse primer 5= GCGGATCCCTACAATTTTCGGAAGTACTCGG 3= (BamH1 site underlined). The vectors obtained were used to transfect both BF and PF T. brucei parental lines.
Trypanosome growth and transfection. The cell lines used were T. brucei subsp. brucei BF S16 (58) and PF cell line 449 (59). The BF parasites were cultured at 37°C and 5% CO 2 in HMI-9 medium containing 10% (vol/vol) FBS. The PF trypanosomes were cultured at 28°C in SDM-79 medium supplemented with 10% (vol/vol) FBS and 7.5 g·ml Ϫ1 hemin. To generate LAP-A dKO and knockdown (RNAi) parasite cells lines, as well as LAP-A and LAP-A-myc overexpression lines, parental BF and PF cells were transfected by electroporation with 10 g of the respective targeting vectors as previously described (58). Recombinants were selected in the presence of the corresponding drug used as a selectable marker. LAP-A-dKO and LAP-A-OE lines were also grown in leucine-free minimal medium (produced in our laboratory) supplemented with 10% (vol/vol) dialyzed FBS and 7.5 g·ml Ϫ1 hemin.
Three-dimensional immunofluorescence studies. Immunofluorescence studies were performed as previously described (60). Briefly, for each sample 10 ϫ 10 6 parasites were harvested by centrifugation, resuspended in 1 ml of PBS containing MitoTracker Red (Invitrogen), and incubated in the dark at culture temperature for 15 min. After one wash with 1 ml of PBS, cells were resuspended in 4% paraformaldehyde and wash solution (PBS plus 0.2% [vol/vol] Tween 20) and fixed on a poly-L-lysine-coated glass slide. The fixed cells were washed twice and permeabilized with a solution of PBS, 1% Blocking Reagent (Roche), and 1% IGEPAL (Sigma). The immunofluorescence assay was performed with anti-TbLAP-A primary polyclonal antibody (affinity purified, diluted 1:500 for overexpressing cells and 1:100 for all others) or anti-c-Myc monoclonal antibody (Sigma, 1:100) and fluorescein isothiocyanate-conjugated anti-rabbit secondary antibody (Sigma, 1:40), Alexa Fluor 488-conjugated anti-rabbit antibody (Sigma, 1:100), or Alexa Fluor 488-conjugated anti-mouse antibody (Sigma, 1:40). The slides were finally stained and mounted with Vectashield-DAPI (Vector Laboratories, Inc.). Vertical stacks of 15 to 20 slices (0.2-m steps) were captured with either an Olympus wide-field microscope and Cell R IX81software or a Leica SP5 confocal microscope. Images were deconvolved and pseudocolored with the Huygens Essential software (version 3.3; Scientific Volume Imaging) and Fiji software (version 1.5e; ImageJ) (61), respectively. Colocalization analysis (Pearson's and Manders' coefficients) was performed by with the JACoP plugin (62) in ImageJ. For nutritional stress studies, 25 ϫ 10 6 parasites were washed twice in 1ϫ PBS and incubated for 60, 120, or 180 min in 1ϫ PBS. After starvation treatment, immunofluorescence assays were performed as described above.
In vivo studies. Three groups of four female C57BL/6J mice (each around 8 weeks old) were used to test TbLAP-A-RNAi cell line infectivity in vivo. One group was infected with the parental BF line, while the two remaining groups were infected with the TbLAP-A-RNAi line. Each mouse was injected intraperitoneally with 1,000 monomorphic parasites from a logarithmic culture. Drinking water containing doxycycline (1 mg·ml Ϫ1 ) was given to one of the groups infected with the TbLAP-A-RNAi line to induce protein knockdown, while normal water was given to the other groups. Mouse survival was monitored for 1 week. The animal research described in this report complied with Spanish (Ley 32/2007) and European Union (2010/63/UE) legislation. The protocols used were approved by the Animal Care Committee of the Instituto de Parasitología y Biomedicina "López-Neyra," CSIC (protocol CEEA/2014/DGP/1).