Autophagy Is Involved in Nutritional Stress Response and Differentiation in Trypanosoma cruzi*

Autophagy is the major mechanism used by eukaryotic cells to degrade and recycle proteins and organelles. Bioinformatics analysis of the genome of the protozoan parasite Trypanosoma cruzi revealed the presence of all components of the Atg8 conjugation system, whereas Atg12, Atg5, and Atg10 as the major components of the Atg12 pathway could not be identified. The two TcATG4 (autophagin) homologs present in the genome were found to correctly process the two ATG8 homologs after the conserved Gly residue. Functional studies revealed that both ATG4 homologues but only one T. cruzi ATG8 homolog (TcATG8.1) complemented yeast deletion strains. During starvation of the parasite, TcAtg8.1, but not TcAtg8.2, was found by immunofluorescence to be located in autophagosome-like vesicles. This confirms its function as an Atg8/LC3 homolog and its potential to be used as an autophagosomal marker. Most importantly, autophagy is involved in differentiation between developmental stages of T. cruzi, a process that is essential for parasite maintenance and survival. These findings suggest that the autophagy pathway could represent a target for a novel chemotherapeutic strategy against Chagas disease.

Autophagy is a major mechanism for bulk degradation of proteins and organelles and is essential for maintaining cellular homeostasis and for cellular development during differentiation, metamorphosis, and aging. Although the process is believed to have arisen primarily as a response to starvation and stress in unicellular organisms, such as yeast, in multicellular organisms autophagy has been also linked to neurodegeneration, cardiomyopathies, pathogen infection, muscular disorders, cancer, and non-apoptotic cell death. During autophagy portions of cytoplasm are engulfed in double membrane vesicles called autophagosomes, which subsequently fuse with the lysosomes/vacuoles, thereby enabling degradation of engulfed material by lysosomal hydrolases (1)(2)(3).
The molecular basis of autophagy was elucidated through Saccharomyces cerevisiae genetic screens, where 16 essential genes (ATGs) were identified (4). Among these genes, two unique ubiquitin-like systems play an important role in the early stages of autophagosome biogenesis (5). The first one is the Atg12 system, where the ubiquitin-like Atg12 protein is conjugated to Atg5 protein in a process mediated by Atg7, an E1 6 -like enzyme (ubiquitin-activating enzyme), and Atg10, an E2-like enzyme (ubiquitin-conjugating enzyme). The Atg12-Atg5 conjugate forms a complex with Atg16 protein (6) that is thought to form a transient coat that drives the deformation of the sequestering membrane during vesicle formation. The second ubiquitin-like protein that acts on vesicle expansion and completion is Atg8. This protein is proteolytically processed by the Atg4 protease (autophagin), thereby exposing a Gly residue which is then covalently attached to a phosphatidylethanolamine (PE) moiety by the concerted action of Atg7 and Atg3 proteins, the latter being a specific E2-like conjugating enzyme (7). This enables the previously cytosolic Atg8 protein to tightly associate with the membranes (8) making Atg8-PE a suitable autophagosomal membrane marker. Before fusion with the vacuole, Atg8 is deconjugated from PE in the outer membrane by Atg4 and released to the cytosol to be reused for new vesicle formation (9). In yeast, autophagy overlaps with a biosynthetic process termed the cytoplasm-to-vacuole targeting (Cvt) pathway. The Cvt pathway is an example of a specific type of autophagy where some proteins that are destined to become resident vacuolar hydrolases are synthe-sized in the cytosol followed by specific packaging into vesicles and delivery to the vacuole (10,11). The molecular mechanism of autophagy was found to be strikingly similar in mammalian cells, with conservation of all the genes essential for the autophagosome formation (12).
Very little is known about autophagy in primitive unicellular organisms such as protozoan parasites, known to infect hosts of diverse origin. The recent completion of the genome sequencing projects of several trypanosomatids (13) now offers new opportunities to study autophagy on the molecular level also in these evolutionarily ancient eukaryotes. One of them is Trypanosoma cruzi, the causative agent of the American trypanosomiasis or Chagas disease, whose genome was completed in 2005 (14). T. cruzi has a complex life cycle, including two replicative forms, the epimastigote present in the gut of the insect vector and the amastigote, an intracellular form in the infected mammal, and two infective, non-replicative forms, the metacyclic trypomastigote in the insect vector and the bloodstream trypomastigote released from infected cells into the blood of the mammal (15). Transitions between the hosts as well as changes in the replicative environment are accompanied by extensive metabolic and morphological changes of the parasite. Proteasome and cruzipain were suggested to be involved in these differentiation processes (16,17); however, the molecular mechanism has not been elucidated. Because differentiation in higher organisms is known to involve autophagy, it was reasonable to hypothesize that autophagy could be involved in T. cruzi differentiation.
In this work we report that the T. cruzi genome contains all the major genes of the Atg8 conjugation system (Atg3, Atg4, Atg7, Atg8), whereas the major components of the Atg12-Atg5 conjugation system (Atg12, Atg5, Atg10) are apparently lacking. The two recombinant T. cruzi autophagins (Atg4 proteases) were found to process the two recombinant Atg8 homologues at the Gly residue. Moreover, all the T. cruzi Atg4, and to a lesser extent Atg8 homologues were found to substitute the yeast homologues in functional assays. Apart from starvation, autophagy was also substantially enhanced during differentiation of the parasite, a process that is essential for survival of the parasite.
Bioinformatics Analysis of T. cruzi Genome-BLAST searches were performed against the predicted ATG-related proteins contained in version 3 of the T. cruzi genome sequences (14). Amino acid sequences corresponding to the C54 family peptidase domain (21) in ATG4-like genes were used for alignment with ClustalW program, and their score values were taken as percentage of identity between two sequences. A similar analysis was performed for the ATG8-like genes analysis on the basis of the deduced amino acid sequences from the N-terminal Met to the conserved Gly residue at the predicted autophagin cleavage site.
Production of Polyclonal Antisera-Anti-TcAtg8.1 and anti-TcAtg8.2 antibodies were raised in rabbits by standard protocols using the purified recombinant proteins.
In Vitro Cleavage Assay of Recombinant TcAtg8 by Recombinant TcAtg4 Proteins-Purified TcAtg8.1 or TcAtg8.2 (17.2 g) recombinant proteins were mixed with different amounts of purified recombinant autophagins (750 ng-7.5 pg for autophagin-1 and 1.0 -0.01 g for autophagin-2) in 100 l of TBS containing 1 mM EDTA and 1 mM 1,4-dithiothreitol. The reaction mixtures containing autophagin-1 were incubated at room temperature for 35 min and those containing autophagin-2 for 18 h. The reactions were stopped by the addition of Laemmli sample buffer and 5 min of boiling. 20 l of each sample were separated on a 15% SDS-PAGE gel and stained with Coomassie Brilliant Blue. To determine the protease class, the experiments were carried out in the presence of different inhibitors (2 mM iodoacetamide, 2.5 mM N-ethylmaleimide, 100 M E-64, 39 ng/l cystatin C, 2.5 mM EDTA, 2 mM o-phenanthroline, or 10 M pepstatin). To determine the exact cleavage site in TcAtg8 proteins, similar reaction solutions were applied to a C8 reverse phase Aquapore RP-300 Brownlee TM column (Applied Biosystems) equilibrated with 0.1% trifluoroacetic acid. Peptides were eluted with an acetonitrile gradient 0 -90% (v/v) in 0.1% trifluoroacetic acid (v/v). The fraction corresponding to the peak that only appeared after treatment with autophagin was freezedried. N-terminal amino acid sequence was determined with Procise Protein Sequencing System 492 (Applied Biosystems) following the manufacturer's instructions.
Enzymatic Assays-Enzymatic activity of recombinant autophagins was tested using Abz-TFGQ-EDDnp, where Abz is the ortho-aminobenzoic acid fluorescent group, and EDDnp is N-(ethylenediamine)-2,4-dinitrophenylamide, a quenching group (kindly provided by Dr. Luiz Juliano, Sao Paulo, Brazil) (23). This peptide was designed based on the TcAtg8.1 cleavage site (TFG2). Activity measurements were performed in 96-well plates in TBS buffer containing 2 mM 1,4-dithiothreitol, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 10 g/ml substrate (final concentration) at 30°C. The reactions were started by the addition of 2 g of either autophagin to the reaction mixture. Fluorescence was measured in a Safire plate reader (Tecan) for 30 min at excitation and emission wavelengths of 320 and 420 nm, respectively.
Detection of Autophagin Activity in T. cruzi Cell-free Extracts-Extracts of the four T. cruzi developmental stages were prepared by resuspending the parasites in TBS containing 1 mM 1,4-dithiothreitol, 100 M E-64, 2 mM phenylmethylsulfonyl fluoride, and 2.5 mM EDTA. Cells were broken by three cycles of freezing at Ϫ20°C and thawing, and the cell-free extracts were obtained by centrifugation at 26,900 ϫ g. Lysates (100 g of protein) were then added to 17.2 g of recombinant TcAtg8.1 or TcAtg8.2 in TBS containing 1 mM 1,4-dithiothreitol, 2 mM phenylmethylsulfonyl fluoride, 2 mM EDTA, 2 mM o-phenanthroline, and 10 M E-64. Where indicated, equivalent reactions were inhibited by 3 mM iodoacetamide. The reactions were incubated for 18 h at room temperature and stopped with Laemmli sample buffer. One-third of each mixture was loaded on a 15% SDS-PAGE gel and transferred to a nitrocellulose membrane (Amersham Biosciences). To detect the cleavage fragments, antisera raised against recombinant TcAtg8.1 or TcAtg8.2 were used as primary antibodies, and alkaline phosphatase-conjugated goat anti-rabbit antibody was used as secondary antibody. The blots were developed using 5-bromo-4chloro-3-indolyl phosphate p-toluidine salt and nitro blue tetrazolium chloride.
Yeast Complementation Studies-BamHI and NotI sites were introduced at the 5Ј-and the 3Ј-ends of TcATG4.1, TcATG4.2, TcATG8.1, and TcATG8.2 genes by PCR using the following primers: for TcATG4.1, sense primer 5Ј-TTAGGATCCATG-CAAGGTACAATGACG-3Ј and reverse primer 5Ј-AATGCG-GCCGCTCAAGAAAAAGTGTC-3Ј; for TcATG4.2, sense primer 5Ј-TTAGGATCCATGGAGTGGTTGAAAATTG-3Ј and reverse primer 5Ј-AATGCGGCCGCTACTCCGCCACG-TCC-3Ј; for TcATG8.1, sense primer 5Ј-TTAGGATCCATG-GCGCCAAAGAAA-3Ј and reverse primer 5Ј-AATGCGGCC-GCTCACCAAGAACCAAAAGTTG-3Ј; for TcATG8.2, sense primer 5Ј-TTAGGATCCATGGCACGCAAATATC-3Ј and reverse primer 5Ј-AATGCGGCCGCTCAATTCCGCCGCA-CAGC-3Ј. Inserts were cloned into the pCM190 vector (24) and used to transform atg4⌬ and atg8⌬ WCG strains of S. cerevisiae (kindly provided by Dr. M. Thumm) by the lithium acetate method (25). Transformant cultures and wild type control WCG yeast strain were grown in yeast nitrogen base media containing 0.5% ammonium sulfate containing amino acids but lacking uracil until an optical density of 1 at 600 nm was reached. Cells were subsequently split into two aliquots. The first was centrifuged, and cells were resuspended in 1% potassium acetate and grown for additional 7 h to induce starvation response. Fresh growth medium was added to the second aliquot as a control. Cultures were centrifuged, Laemmli sample buffer and glass beads were added to the harvested cells, and the mixture was vortexed vigorously to break cell walls. Yeast proteins were separated on a 10% SDS-PAGE gel and transferred to a nitrocellulose membrane. Immunodetection was performed with anti-(pro)aminopeptidase I primary antibodies (gift from I. V. Sandoval and M. J. Mazón) and horseradish peroxidase-conjugated anti-rabbit secondary antibodies. The antigens were visualized with ECL TM detection system (Amersham Biosciences).
Immunofluorescence Studies-The parasites were fixed in 4% paraformaldehyde in PBS for 15 min. Next, the cells were washed twice with PBS, incubated for 10 min with 25 mM ammonium chloride, and washed again twice with PBS. Coverslides were saturated in the blocking buffer (2% bovine serum albumin, 0.1% saponin in PBS) containing 3% goat serum for 30 min and incubated for 2 h with the primary antibody diluted in the blocking buffer. Parasites were then washed with PBS and incubated with the appropriate secondary antibody diluted in the blocking buffer for 1 h. After extensive washing with PBS, coverslides were mounted using Fluor-Save TM reagent (Calbiochem) containing 5 g/ml DAPI. The following primary antibodies were used: rat anti-HA high affinity monoclonal antibodies (Roche Applied Science) (1/500 dilution), rabbit anti-Atg8.1 polyclonal antibodies (1/700 dilution), and mouse anti-TcSCP (T. cruzi serine carboxypeptidase) polyclonal antibodies as a lysosomal/reservosomal marker (1/1000) (28). The secondary antibodies used were AlexaFluor 546-conjugated goat anti-rat, Alexa-Fluor 546-conjugated goat anti-rabbit, AlexaFluor 488-conjugated goat anti-rabbit, and AlexaFluor 546-conjugated goat anti-mouse immunoglobulins (Molecular Probes), all diluted 1/1000. Preparations were analyzed using a fluorescence microscope (Nikon Eclipse E600), and image capture was performed by a Spot RT Slider Model 2.3.1 digital camera (Diagnostic Instruments). The number of Atg8.1 positive vesicles within untransfected T. cruzi epimastigotes was quantified by observing at least 70 cells from three independent experiments.
Transmission Electron Microscopy-Parasites were fixed for 1 h at room temperature with 2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.2, post-fixed for 30 min with 1% OsO 4 , 1.25% potassium ferrocyanide, and 5 mM CaCl 2 in 0.1 M cacodylate buffer, pH 7.2, dehydrated in an ascending ethanol series, and embedded in Epon resin. Ultrathin sections were stained with uranyl acetate and lead citrate and then examined in a Zeiss 902 electron microscope operating at 80 kV.

Identification of Candidate Autophagosome Formation
Genes in the T. cruzi Genome-The remarkable conservation of the genes involved in autophagy from yeast to human provides an excellent starting point toward identification of the autophagy-related genes in various organisms. Using BLAST analysis, the T. cruzi genome (14) was searched for sequences related to the Atg8-PE and Atg12-Atg5 conjugation systems, which represent the two major components of the autophagic machinery.
Surprisingly, BLAST analysis failed to identify most of the genes specifically involved in Atg12 conjugation. These included the ATG12 gene encoding an ubiquitin-like protein, which is the most important, as well as the genes encoding two other components of this pathway, ATG5 and ATG10. The only two sequences identified were both homologous to ATG16, which is, however, a downstream gene in the pathway (supplemental Table 1). This further suggests that the Atg-12-Atg5 conjugation system is most likely lacking in T. cruzi or at least significantly diverse from the yeast system. However, all the essential genes involved in the Atg8 conjugation system could be identified in the T. cruzi genome using the same approach (supplemental Table 1). The analysis, thus, revealed single genes homologous to the E1-enzyme Atg7 (34% identity with yeast Atg7) and the E2-enzyme Atg3 (28% identity with yeast Atg3). In contrast, two candidate genes were identified for each of the two early genes of the pathway, the ubiquitin-like modifier Atg8 and its processing protease Atg4. A detailed analysis of the two candidate genes for Atg4 protease (TcATG4.1 and TcATG4.2) revealed that they encode proteins with 30 and 20% identity to the yeast Atg4, respectively. The two genes are 21% identical with each other within the region corresponding to the catalytic domains characteristic for C54 (autophagin) family (supplemental Fig. 1). Both candidate proteases were found to contain all the key amino acid residues involved in substrate hydrolysis: the Cys-Asp-His catalytic triad, the Tyr residue, which participates in the formation of the oxyanion hole, and the conserved Trp residue critical for the binding of Atg8 (29). Interestingly, orthologs of both TcATG4.1 and TcATG4.2 could be found in the Trypanosoma brucei (Tb11.01.7970 and Tb927.6.1690) and in the Leishmania major (LmjF30.0270 and LmjF32.3890) genomes. A similar analysis of the two Atg8 homologues, TcATG8.1 and TcATG8.2, revealed that they are 43% identical to each other. Besides an overall similarity to the yeast Atg8, they also share the conserved Gly residue at the putative Atg4 cleavage site as well as the two crucial Phe residues suggested to be important for recognition by Atg4 (supplemental Fig. 2). TcAtg8.1 displays higher sequence identity (53%) to the yeast Atg8 protein and has two orthologs in T. brucei (Tb07.10C21.40 and Tb07.10C21.50) and one in L. major (LmjF19.1630). In contrast, TcAtg8.2 has only one ortholog in T. brucei (Tb07.28B13.800), whereas none could be found in L. major. However, the L. major genome contains several more distantly related ATG8-like sequences whose direct orthologs do not seem to be present in T. cruzi and T. brucei.
This immediately raised an important question, whether the apparently functional Atg8 conjugation system would be sufficient for the formation of autophagosomal vesicles in T. cruzi. To address this question, all the subsequent work was focused on the functional analysis of the two early genes, Atg4 and Atg8.
T 2) terminus were purified in one step using Ni-NTA affinity chromatography. Next, both recombinant TcAtg4 proteins (autophagins) were tested for their ability to process both potential recombinant TcAtg8 substrates in in vitro cleavage assays using fixed amounts of TcAtg8.1 or TcAtg8.2. Autophagin-1 (TcATG4.1) was shown to rapidly process both Atg8 proteins with a minor preference for TcAtg8.1 already at low protease concentration (Fig. 1, A and B). Processing was inhibited by iodoacetamide and N-ethylmaleimide but not by E-64, cystatin Autophagy in T. cruzi FEBRUARY 8, 2008 • VOLUME 283 • NUMBER 6 C, EDTA, o-phenanthroline, or pepstatin (not shown), consistent with the inhibition pattern of the yeast cysteine peptidase Atg4. In contrast, only minor amounts of Atg8 proteins were cleaved by autophagin-2 despite considerably higher concentration of the protease (Ͼ250-fold) and substantially increased incubation time (18 h) (Fig. 1, C and D). The possibility that autophagin-2 had been inhibited by metal cations during the Ni-NTA purification procedure was excluded since in a control experiment performed with crude extracts from E. coli cells overexpressing the recombinant peptidase the same results were obtained (not shown). Despite dramatically lower proteolysis efficiency of autophagin-2 (ϳ10,000-fold), which suggests that autophagin-1 is the main enzyme responsible for Atg8 protein processing in T. cruzi, both autophagins cleaved the two Atg8 proteins after the same Gly residues (Fig. 1G, Gly-121 in Atg8.1 and Gly-131 in Atg8.2), as determined by N-terminal sequencing of the C-terminal proteolytic products (Fig. 1, E  and F). This is also consistent with the predicted cleavage sites, as cleavage at these Gly residues is highly conserved throughout evolution among Atg8 proteins (supplemental Fig. 2).
TcATG4.1, TcATG4.2, and TcATG8.1 Restore Autophagy in Autophagy-deficient Yeast Strains-After confirming that both autophagins and their Atg8 substrates are functionally active in vitro, their possible implication in autophagy was studied in the S. cerevisiae model system, which offers the advantage of the overlapping autophagy and Cvt pathways. As a consequence, S. cerevisiae atg4⌬ and atg8⌬ strains were shown to be deficient in both autophagy and the Cvt pathway and, therefore, failed to deliver proaminopeptidase I (pAPI) from the cytosol to the vacuole where pAPI is normally processed to the mature enzyme (mAPI) (30). Processing of pAPI to mAPI was, thus, used as an indicative marker of functional autophagy under normal growth conditions and under nitrogen starvation. As shown in Fig. 2, transforming the S. cerevisiae atg4⌬ strain with either TcATG4.1 or TcATG4.2 complemented the yeast deletion almost completely, generating mAPI in both nutrient rich and starvation conditions. In contrast, neither of the TcAtg8 proteins restored the Cvt pathway in the atg8⌬ strain in nutrient-rich conditions. However, under nitrogen starvation conditions a minor mAPI band was observed in TcATG8.1 transformant, suggesting that autophagy is at least partially reestablished. Western blot analysis of the transformant yeast extracts using anti-TcAtg8.1 and anti-TcAtg8.2 polyclonal antibodies demonstrated that both pro-  2-4). Lane 1, control without autophagin-2. E and F, Atg8.1 (E) and Atg8.2 (F) were processed in vitro with recombinant autophagin-1. Reaction solutions were applied to reverse-phase high performance liquid chromatography, and the eluted peptides were detected by absorbance at 215 nm. As controls, unprocessed TcAtg8 were applied to the column and are shown in the upper half of each panel. The peaks corresponding to the cleaved C-terminal peptides are marked with arrows. Abs, antibodies. G, autophagin cleavage sites in TcAtg8.1 and TcAtg8.2 are marked with arrowheads. The conserved Gly residues are shown in bold.
teins were efficiently expressed (not shown), excluding the possibility that inefficient processing was a consequence of low protein expression. This is, however, in agreement with the results of an analogous experiment with Arabidopsis thaliana ATG8 homologues, where pAPI processing was partially restored under starvation conditions only (31). This difference between Atg4 and Atg8 can probably be explained by the fact that Atg8 protein has to interact with several host cell proteins to mediate autophagosome formation and not just with a single one as Atg4, giving a much higher chance that these Atg8 interactions are not optimal due to species divergence.
Autophagin Activity Is Expressed in all Four T. cruzi Developmental Stages-Having shown that T. cruzi Atg4 and Atg8 could function in the yeast autophagy model, we next wanted to investigate their functionality in the parasite. In the next experiment, expression of autophagins and Atg8 proteins was investigated in all four developmental stages of the parasite, i.e. epimastigotes, metacyclic trypomastigotes, amastigotes, and cell-derived trypomastigotes. Therefore, in the first experiment cell-free extracts of all four developmental stages were mixed with the recombinant substrate proteins Atg8.1 or Atg8.2. As shown in Fig. 3, both substrates were proteolytically processed with all four extracts, suggesting that autophagins are expressed constitutively throughout the parasite life cycle and not limited to a specific developmental stage of the parasite. Processing was completely abolished in the presence of iodoacetamide, confirming that the processing enzymes were cysteine proteases.
TcAtg8.1 Is a Functional Homologue of Yeast Atg8 Protein-To establish the role of TcAtg8.1 and TcAtg8.2 proteins in T. cruzi, N-terminal HA-tagged TcATG8.1 and TcATG8.2 genes were cloned into the pRibotex expression vector. T. cruzi epimastigotes were then transfected with these constructs, and the intracellular localization of the two proteins was investigated by indirect immunofluorescence microscopy using anti-HA antibodies (Fig. 4A). In nutrient-rich BHT growth medium, both proteins were located in the cytosol. However, when transfected epimastigotes were exposed to prolonged starvation in PBS, Atg8.1 localization changed considerably, revealing the presence of several large bright spots, probably resulting from Atg8.1 concentration in the membranes of autophagosomal vesicles. In a control experiment, when epimastigotes were transfected with the G121A mutant of HA-Atg8.1, no such change in localization could be observed. This further suggests that, analogous to the yeast system, the autophagosome localization of Atg8.1 is dependent on the conjugation of the C-terminal Gly residue probably to a PE moiety. Although after prolonged starvation Atg8.2 protein was also partially localized to several bright spots, the effect was less pronounced than in the case of Atg8.1. Furthermore, the TcAtg8.2 G131A mutant exhibited similar vesicle membrane-associated localization, suggesting that membrane binding can occur independently of conjugation to PE. This further implies that TcAtg8.2 might be partially bound to autophagy-related vesicles, although the possibility that it might have another localization and/or another function cannot be excluded. This is, however, more similar to the human Atg8 homologues GABARAP and GATE-16, whose physiological functions and roles in autophagy are also not clear (2,32).
Based on these results, it can be concluded that Atg8.1 is the functional homologue of the yeast Atg8 protein and the mammalian LC-3 protein and, hence, has a potential for use as an autophagosomal marker in T. cruzi. To verify this possibility, rabbit polyclonal antibodies were generated against the purified recombinant TcAtg8.1 protein and tested on whole cell lysates of the different forms of the parasite where they recognized a single protein band of about 17 kDa corresponding to the size of Atg8.1 in epimastigotes, amastigotes, and cell-derived trypomastigotes but not in metacyclic trypomastigotes (Fig. 4B). The polyclonal antibodies were then used for immunofluorescence microscopy studies of untransfected parasites (Fig. 4C). Under nutrient-rich growth conditions, several tiny dots distributed all over the parasite were observed. Additionally, large dots were present in a median number of 1 per cell (range, 0 -3) under this condition, in agreement with Atg8.1 being present in   FEBRUARY 8, 2008 • VOLUME 283 • NUMBER 6 epimastigotes under these conditions. In contrast, after 16 h of starvation in PBS, the median number of large spots per cell increased to 3 (range, 0 -8) and resembled the pattern obtained with the anti-HA antibodies for Atg8.1. This suggests that starvation-induced autophagy occurred also in nontransfected T. cruzi epimastigotes and confirmed that Atg8.1 could be used as an autophagosomal marker in the parasite.

Autophagy in T. cruzi
To confirm the appearance of autophagic structures under starvation conditions, routine transmission electron microscopy analysis of epimastigotes maintained in PBS for 16 h were performed. Double-membrane vacuoles containing material with an electron density similar to the cytoplasm in their interior, characteristic for autophagosomes, were observed under these conditions (Fig. 4D), whereas they were not observed in cells grown in rich medium, in agreement with previous observations in yeast and mammalian cells.
Because starvation-induced autophagy is often linked with increased protein degradation, we next tried to block the appearance of autophagosomes by the classical autophagy inhibitor 3-methyladenine, which similarly to wortmannin, blocks phosphatidyl inositol 3-kinase. However, both inhibitors seem to be toxic for the parasite, consistent with recent observations (33), which precluded us from performing the experiment.
T. cruzi Differentiation Is Accompanied by Autophagosome Formation-T. cruzi experiences profound morphological changes during development in the vertebrate and invertebrate hosts. The process of T. cruzi metacyclogenesis in volves the transformation of noninfective epimastigotes into metacyclic trypomastigotes, which are the naturally infective form of the parasite. During in vivo metacyclogenesis, differentiating epimastigotes adhere to the epithelium of the insect rectum before transforming into metacyclic trypomastigotes. It is not clear how adhesion triggers the differentiation process, but it has been demonstrated that both adhesion and metacyclogenesis are triggered by nutritional stress (34).
Because autophagy has been widely described to be involved in cellular remodeling during development and differentiation and is a common response to nutrient deprivation (35), as also demonstrated in T. cruzi (see above), the involvement of autophagy in the process of T. cruzi metacyclogenesis was studied. The next set of experiments was, therefore, performed using samples derived from cultures undergoing spontaneous differentiation from epimastigotes to metacyclic trypomastigotes in BHT medium. Based on bright field microscopy combined with fluorescence microscopy, three different cell-type populations could be defined: epimastigotes, differentiating epimastigotes, and fully differentiated metacyclic trypomastigotes (Fig. 5). The three populations have the following characteristics; epimastigotes (E) are spindle-shaped organisms (a) with a rounded nucleus and a slightly concave disk, the kinetoplast, located in the proximity of the nucleus and anterior to it, based on DAPI Immunofluorescence studies were performed using rat anti-HA monoclonal antibody and AlexaFluor 546-conjugated goat anti-rat secondary antibody. Nuclear and kinetoplast DNA were visualized by DAPI staining. The HA fluorescence image (red) has been merged with the corresponding DAPI staining (blue), and the resulting merged images are shown. The upper half represents the subcellular distribution of the overexpressed proteins in nutrient rich BHT media, whereas the lower half corresponds to the pattern observed after starvation in PBS. Arrowheads indicate autophagosomes. B, Atg8.1 expression in T. cruzi developmental stages. 20 ϫ 10 6 epimastigotes (E), metacyclic trypomastigotes (M), amastigotes (A), and cell-derived trypomastigotes (T) were resuspended in Laemmli sample buffer and boiled for 10 min. Proteins were resolved in 12.5% SDS-PAGE and analyzed by immunoblotting with a 1/500 dilution of polyclonal anti-TcAtg8.1 serum. C, subcellular localization of endogenous Atg8.1 in nutrient-rich BHT medium (a) and after prolonged starvation in PBS (b) was evaluated by indirect immunofluorescence experiments using anti-Atg8.1-specific antibodies and Alexa 546-conjugated secondary antibody (red). The nucleus and kinetoplast were visualized by DAPI staining (blue). D, routine transmission electron microscopy of epimastigotes starved in PBS for 16 h. The arrows depict the double membrane vacuoles similar to the previously identified yeast and mammalian autophagosomes. staining (b). Lysosome-like organelles called reservosomes, characteristic for this form, could also be seen at the posterior end of the parasite (c). Differentiating epimastigotes (D) are smaller than epimastigotes with a rounded posterior end (f and k). DAPI staining revealed that the nucleus is still spherical, but the kinetoplast has begun to migrate and is located beside or at the posterior side of the nucleus (g and l). The number of reservosomes is diminished in this stage based on carboxypeptidase staining (h and m). Finally, metacyclic trypomastigotes (T) are slender (p) with an elongated nucleus and a round-shaped kinetoplast occupying the posterior position in the cell as revealed by DAPI staining (q). Small lysosome-like vesicles located next to the kinetoplast at the posterior end of the parasite were identified based on carboxypeptidase staining (r). In addition to the differences described above, remarkable differences were observed based on Atg8.1 staining. Extremely intense staining was observed in differentiating epimastigotes (i and n), suggesting that these cells were undergoing massive autophagy. Because of this very intense signal, the exposition times used were much shorter as in Fig. 4, thereby precluding us from accurately detecting the levels of Atg8.1 in epimastigotes (d) and metacyclic trypomastigotes (s), where almost no signal was observed. However, based on these results, the results of the Western blot (Fig. 4B), and the findings in rich medium (Fig.  4C), it can be suggested that the basal level of Atg8.1 is not very high in epimastigotes and very low (below the detection limit) in metacyclic trypomastigotes. Merged images of serine carboxypeptidase and Atg8.1 staining of the same parasites revealed that both proteins partially colocalized in the reservosomes during differentiation, a fact that is likely due to the delivery of the autophagosome content to the reservosomes/lysosomes (j and o).

DISCUSSION
The apparent conservation of the mechanism of autophagy through billions of years of evolution emphasizes its indispensable role regardless of the eukaryotic organism in question (35). Although double membrane-containing structures and vacuolization had been previously observed in T. cruzi upon treatment with toxic compounds and in adverse conditions (33,36,37), the molecular features of this process had never been characterized. The major goal of the study was, therefore, to reveal the existence of autophagy in T. cruzi and to possibly characterize some of the molecules involved. A bioinformatics analysis of the T. cruzi genome unambiguously confirmed the existence of all four critical genes of the Atg8 conjugation pathway, ATG3, ATG4, ATG7, and ATG8. However, among the genes of the Atg12 conjugation system, we could only identify sequences related to ATG16, whereas the sequences related to ATG5, ATG10, and ATG12 genes were not found. These results are in agreement with a recent bioinformatics survey of trypanosomatid genomes for autophagy related genes (ATG), which revealed that several core genes, including those encoding the Atg8 lipidation pathway, are present in these ancient eukaryotes, whereas the Atg12 conjugation system, a very important factor of autophagosome formation, was reported to be missing (38). Although the components of this system were suggested to be found in the L. major genome (39), the similarity with Atg12-related genes in other organisms is extremely low. Therefore, their possible role in autophagy remains to be elucidated by functional studies.
Lipidation of Atg8 in yeast or LC3 modification in mammals is an essential step in correct autophagosome formation (40). Indeed, Nakatogawa et al. (41) have recently shown that Atg8 conjugated to PE mediates tethering between adjacent membranes and stimulates membrane hemifusion, an event that may mimic expansion of the autophagosomal membrane during autophagy. A prerequisite for this is, however, proteolytic removal of the C terminus of Atg8/LC3 by the cysteine protease Atg4 (autophagin) to expose a Gly residue, which is then FIGURE 5. Autophagosome formation during in vitro epimastigote to metacyclic trypomastigote transformation. Epimastigotes (E), differentiating epimastigotes (D), and metacyclic trypomastigotes (T) were observed under a fluorescence microscope. Left panels correspond to the bright field images. The nucleus (n) and kinetoplast (k) were distinguished by DAPI staining (blue). Reservosomes (red) were revealed using specific polyclonal antibodies raised against a T. cruzi lysosomal serine carboxypeptidase (anti-TcSCP). The occurrence of autophagosomes (green) was evaluated by immunofluorescence using specific anti-Atg8.1 polyclonal antibodies. Reservosomes are indicated by arrows, and autophagosomes are indicated by arrowheads. Merged pictures are shown in the right. Note that weak cytosolic staining of Atg8.1 could be observed in epimastigote and metacyclic trypomastigote stages; however, due to the much more intense vesicular staining of the intermediate stages, the microscope and camera settings had to be adjusted differently.
covalently modified (7). Atg4 and Atg8 homologues in T. cruzi and their possible implications in autophagy were, therefore, the focus of further work. In contrast to yeast, two homologues of both Atg4 (TcAtg4.1 and TcAtg4.2, supplemental Fig. 1) and Atg8 (TcAtg8.1 and TcAtg8.2, supplemental Fig. 2) were found in T. cruzi, resembling more the situation in mammals, where four homologues of each gene can be found (42,43). Because in humans only one Atg4 homologue, HsAtg4B, and one Atg8 homologue, LC3, are principally involved in autophagy (44,45), it was important to identify their functional TcAtg4 and TcAtg8 homologues.
Based on in vitro cleavage assays, TcAtg4.1 was found to be most likely responsible for processing the TcAtg8 proteins. Nevertheless, when expressed in the ATG4-deficient yeast strain, TcATG4.2 was as efficient as TcATG4.1 in reconstituting the Cvt pathway and autophagy. One possibility is that low levels of autophagin activity are sufficient to trigger substantial proaminopeptidase processing. This could also be linked with overexpression of both autophagins in yeast, which are probably much higher than the endogenous levels of Atg4. Alternatively, TcAtg4.1 and TcAtg4.2 may be similarly effective in processing yeast Atg8, as both autophagins were found to have the same specificity. Moreover, this result also suggests that both autophagins are capable not only of processing Atg8 but also deconjugating it from PE. TcAtg4.2 may, thus, be an orphan protease remaining in the genome and eventually serving as a backup system, although the possibility that TcAtg4.2 might be involved in the processing and/or deconjugation of other ubiquitin-like proteins remains open. A recent study performed in L. major suggested the LmATG4.2 gene to be important for autophagy and differentiation (46). The L. major atg4.2⌬ mutants had an increased proportion of the lipidated form of Atg8 relative to the unlipidated one, which led to speculation for a role of Atg4.2 in Atg8 deconjugation. However, no further studies were performed with Leishmania Atg4.2 or Atg4.1 to provide insight into the roles of these proteases.
Autophagins were found to be constitutively expressed in all life-cycle stages of T. cruzi, which is in agreement with the earlier findings on human autophagins (43). Despite its constitutive expression and apparent lack of existing endogenous inhibitors, the very strict specificity of autophagins may be sufficient to prevent unspecific proteolysis in the cytosol. This can be explained by the recently determined crystal structure of HsAtg4B (29). The active site cleft of this family of enzymes was found to be masked by a loop of four amino acid residues and could be exposed only after Atg8 binding, which might also explain the apparent lack of activity of both T. cruzi autophagins on the short peptidic substrate Abz-TFGQ-EDDnp, designed on the basis of the cleavage sequence in TcAtg8.1.
Functional studies with TcATG8.1 and TcATG8.2 clearly revealed TcAtg8.1 to be the critical molecule. TcATG8.1 was the only one able to partially replace yeast Atg8 function when expressed in an Atg8-deficient yeast strain (Fig. 2). Moreover, only TcAtg8.1 was found to concentrate in several large vesicles in HA-tagged TcATG8.1-transfected epimastigotes exposed to starvation conditions (Fig. 4A). Mutating the conserved Gly residue to Ala (G121A mutant) abolished this effect, suggesting that the observed pattern is a consequence of Atg8.1-PE conjugates insertion into the lipid membranes and not of unspecific protein aggregation. Finally, this finding was confirmed with polyclonal anti-TcAtg8.1 antibodies in wild-type nontransfected epimastigotes exposed to starvation.
These findings strongly suggest that the Atg8 conjugation system in T. cruzi functions very similarly to the Atg8 conjugation systems in yeast and mammals. Moreover, these experiments clearly demonstrated induction of autophagy as a response to starvation and validated TcAtg8.1 as a suitable marker for autophagy in T. cruzi, in agreement with previous data on LC3 and Atg8 (8,47). Autophagy could, thus, represent a crucial survival mechanism of T. cruzi epimastigotes as a response to starvation in the gut of the insect vector. The latter are known to suffer long periods of starvation, which can last for more than 12 months depending on species, instar, and climatic conditions (48). Furthermore, nutritionally poor growth media such as triatomine artificial urine and M16 induce metacyclogenesis, a process during which epimastigotes differentiate to metacyclic trypomastigotes and which occurs in nature in the insect rectum (49,50). The process, therefore, requires fast and extensive protein degradation and recycling of building blocks for the synthesis of new macromolecules. Initial studies suggested that cruzipain, the most abundant lysosomal/reservosomal cysteine peptidase of T. cruzi belonging to the CA clan (21) of proteases, is critical for the differentiation as the cell-permeable cruzipain inhibitors effectively inhibited the process (19,51), whereas overexpression of this enzyme is associated with enhanced metacyclogenesis (52). Our results clearly demonstrate that autophagy is also involved in cell remodeling most probably by delivering large parts of the cytoplasm and organelles to lysosomes/reservosomes. Blocking cruzipain is, thus, a downstream effect associated with defects in the terminal turnover of proteins. An analogous situation can be found in mammals where lysosomal cathepsins are responsible for the terminal protein degradation within autolysosomes after the autophagosome-lysosome fusion. Knocking out aspartic protease cathepsin D or, simultaneously, cathepsins B and L resulted in defects in autophagy and accumulation of autophagosomes (53). A similar effect was also observed in Leishmania, where inhibition of cysteine peptidases (CPA, CPB, CPC) homologous to cruzipain and cysteine cathepsins with broad spectrum reversible or irreversible inhibitors, which killed the parasite, was accompanied with defects in the lysosomal compartment resembling lysosomal storage disease (54). Moreover, CPA and CPB from Leishmania mexicana were recently found to be essential for both autophagosome degradation and differentiation of that parasite. In addition, LmAtg4.2 was suggested to be involved in autophagosome formation, thus being upstream of CPA and CPB (39,46).
Because differentiation is critical for the pathogenicity of the parasite, these findings could also be important for the treatment of Chagas disease. This infection is endemic in Central and South America, where its prevalence is estimated at 16 -18 million cases with 120 million people at risk. No vaccines have been developed so far, and the low effectiveness and serious adverse effects of the chemotherapeutic agents available makes treatment of Chagas disease very difficult. There is, therefore, a pressing need for the identification of novel drug targets and virulence factors to improve the prevention and treatment of this disease (55). Proteases are very attractive targets for infectious diseases (56), and cruzipain has already been validated as a target for Chagas disease (17). Our data suggest that autophagins, which act upstream of cruzipain in T. cruzi differentiation, should be considered as potential targets, as targeting an upstream protease in the signaling pathway is often beneficial (56).
In conclusion, we have found that conservation of the autophagy mechanism is even broader than believed so far, spanning all of eukaryotic evolution. However, autophagosome formation seems to be driven primarily by the functional Atg8 conjugation system, as Atg12 is apparently absent from the T. cruzi genome. Autophagy was shown to be important for the survival of the parasite during starvation and differentiation, which suggests that inhibition of autophagy might be considered a novel strategy for treatment of infections caused by T. cruzi.