A leucine aminopeptidase is involved in kinetoplast DNA segregation in Trypanosoma brucei

The kinetoplast (k), the uniquely packaged mitochondrial DNA of trypanosomatid protists is formed by a catenated network of minicircles and maxicircles that divide and segregate once each cell cycle. Although many proteins involved in kDNA replication and segregation are now known, several key steps in the replication mechanism remain uncharacterized at the molecular level, one of which is the nabelschnur or umbilicus, a prominent structure which in the mammalian parasite Trypanosoma brucei connects the daughter kDNA networks prior to their segregation. Here we characterize an M17 family leucyl aminopeptidase metalloprotease, termed TbLAP1, which specifically localizes to the kDNA disk and the nabelschur and represents the first described protein found in this structure. We show that TbLAP1 is required for correct segregation of kDNA, with knockdown resulting in delayed cytokinesis and ectopic expression leading to kDNA loss and decreased cell proliferation. We propose that TbLAP1 is required for efficient kDNA division and specifically participates in the separation of daughter kDNA networks.


Author summary
Trypanosomes bear a single mitochondrion with its genome (kinetoplast, or kDNA) arranged as a network of circular molecules. kDNA is essential for the Trypanosoma brucei life cycle, as it is required for proper functioning of the mitochondrion and progression through the insect vector. Division of kDNA is synchronized with the replication of nuclear DNA and additional cell cycle events. Though the mechanism of kDNA replication is well understood and proteins mediating this process identified, several key steps, including segregation, remain poorly known, in part due to the absence of characterized proteins specifically functioning at this stage. Here we report leucine aminopeptidase 1 (LAP1) as the first identified component of the nabelschnur, or umbilicus, a structure observed during the separation of daughter kDNA networks, whose expression is required for correct kinetoplast segregation. Introduction rotates, distributing the re-attaching minicircles around the periphery of the kDNA network [26]. In T. brucei, however, the disk remains stationary, and the replicated network divides by an unknown mechanism. It is assumed that this mechanism is unique to T. brucei and its close relatives due to the particular nature of the kDNA segregation process present in this species. Situated in the posterior region of the cell, the kDNA disk is physically linked to the basal and pro-basal bodies by the tripartite attachment complex (TAC) [2]. It has been postulated that, at the moment of cell division, the basal bodies align and direct kDNA segregation, a process orchestrated via the TAC [27]. The physical separation of the progeny kDNA networks as observed by electron microscopy, has been associated with the formation of a filament-resembling structure termed the nabelschnur or umbilicus [28]. This structure, so far observed exclusively in T. brucei, constitutes the final physical connection between the newly replicated daughter kDNA networks [28,29]. Although the kDNA of T. brucei is one of the best-studied mt genomes, the mechanism(s) governing this highly precise division remain largely unknown.
Here we characterize the function of a leucyl aminopeptidase, TbLAP1, in the segregation of dividing kDNA. LAPs are homohexameric metallopeptidases classified into either the M1 or M17 protease families [30] that cleave N-terminal amino acids from proteins, particularly, but not exclusively, L-leucine. Members of the M1 family carry a canonical HEXXH motif in their active site, whereas the M17 family members lack this motif and require two metal ions per monomer for activity [31]. LAPs have diverse subcellular localizations; initially found in the cytosol, they have been subsequently encountered in chloroplasts [32], on bacterial surfaces [33], or teguments of parasitic helminths [34]. Moreover, in Escherichia coli LAPs bind DNA [35], while they are amongst secreted proteins in other bacteria such as Mycoplasma [33]. Despite high sequence conservation, members of the M17 protease family perform a range of moonlighting functions in diverse organisms. Amongst these functions, LAPs regulate meiosis in fungi [36], are involved in infectivity of various bacteria, yeast and parasitic protists [37][38][39], regulate stress responses and signal transduction [40], act as molecular chaperones that protect proteins from heat inactivation and assist in their refolding in plants [41] and finally, are required for glutathione metabolism and recycling [42]. Regardless of these diverse roles of LAPs, the exact mechanisms of their many moonlighting functions remain to be clarified.
In this paper we report that TbLAP1 is a mt protein that dynamically associates with kDNA during the cell cycle (Fig 1). The protein localizes to kDNA and the nabelschnur, the proteinaceous link connecting progeny kDNAs at late stages of segregation, indicating a role of TbLAP1 in resolving kDNA replication.  [46]) predict mt localization for TbLAP1 based on the presence of an N-terminal mt targeting signal. Due to this we chose to investigate TbLAP1 further. Indeed, TbLAP1, in situ tagged at the C-terminus with either green fluorescent protein (GFP; S6 Fig) or V5 was throughout the cell cycle exclusively associated with the kDNA disk (Fig 1).

Subcellular localization of TbLAP1
Significantly, TbLAP1 distribution within the kDNA undergoes substantial changes during cell division. During the G 1 phase, TbLAP1 co-localizes with the kDNA but not with TAC102, a component of the TAC, a filamentous structure that connects kDNA with both the basal body and the flagellum (Fig 1). The onset of mitosis in T. brucei is marked by basal body division followed by duplication of the TAC. Replicating in parallel with the TAC, kDNA initially forms a dumbbell-like structure, where TbLAP1 bifurcates into two lobes that bind each side . DAPI (blue) shows the location of the nucleus and kDNA. Zoom shows enlargement of the kDNA and associated region from the merge. The V5 signal co-localizes with the kDNA disk but does not colocalize with the anti-TAC102 antibody (A). Once kDNA division has been completed (the "dumbbell" kDNA shape), TbLAP1 appears in two foci, one at each tip of the structure, meanwhile TAC102 is found in an elongated form close to segregation (B). Once kDNA segregation is initiated, the TbLAP1 signal extends into the nabelschnur, and at the same time co-localizes with the kDNA disk (C). As the segregation process of the divided but as yet unsegregated kDNA disk (Fig 1B). Shortly afterwards, the kDNA disk commences segregation into two daughter networks, positioned perpendicular to one another (Fig 1C, 1D and 1E). At this stage, the characteristic nabelschnur becomes apparent between the newly replicated kDNA networks [28,29]. As the daughter kDNA discs progressively separate, the nabelschnur extends with a small, yet prominent, TbLAP1 focus at its center (Fig 1C  and 1D). Once the daughter kDNA disks have completed their realignment, the TbLAP1 signal within the nabelschnur decreases and eventually remains confined to a spot overlaying each newly synthesized kDNA disk (Fig 1G). It is important to highlight that TbLAP1 localization overlaps with kDNA, but is clearly distinct.
The division of the basal body was also followed as a landmark for the dynamic localization of TbLAP1 during the division of procyclic T. brucei. This structure has been shown to divide prior to, and promote segregation of the daughter kDNA disks [47]. Immunolocalization of the YL1/2 epitope, specific for mature basal bodies, does not overlap with that of TbLAP1 ( Fig  2), demonstrating that these structures indeed segregate prior to the separation of the kDNA disks (Fig 2B and 2C). Staining of the basal bodies depicts localization of TbLAP1 to kDNA upon division and its dynamic localization along the duplicating disk ( Fig 2C). The TbLAP1 signal in Fig 2D and 2E indicates that the nabelschnur emerges coincident with immunostaining of anti-TAC102 antibody (Fig 1). After segregation of these newly divided disks, the basal bodies align perpendicularly, standing at 180˚angle to each other, as both kDNAs separate and align for cytokinesis (Fig 2E and 2F).
Localization of in situ V5-tagged TbLAP1 into the kDNA disk was further confirmed by cryosectioned transmission electron microscopy via immunodecoration with colloidal goldlabeled anti-V5 antibodies, with abundant gold particles found exclusively on the electrondense kDNA disk. Unlike with immunofluorescence, except for the "dumbbell-shaped" kDNA, it is not possible to assess at which stage of cell division these kDNAs are, but based on their sizes these are most likely interphase networks (Fig 3). The spatial distribution of the signal in the observed kDNAs was evaluated by Ripley's K function [48,49]. The statistical analysis indicates that the anti-V5 antibody signal exists in two modes and either forms clusters ( Fig  3A and 3B), or is randomly distributed throughout the kDNA (Fig 3C and 3D). In the analyzed samples (n = 20), 40% of the gold particles found display cluster formation, while the remaining 60% are randomly distributed (n = 342; S1 Table).

Knockdown of TbLAP1 by RNAi leads to accumulation of 2K2N cells
RNAi-mediated down-regulation of the TbLAP1 protein induces a growth defect (Fig 4A and  4D), which is nonlethal but causes a delay in cytokinesis. Real-time qPCR analysis confirmed the TbLAP1 transcript is reduced by 80% at 48 and 96 hrs post-induction, but seems to increase at 144 hrs to 50% when normalized to 18S rRNA levels (Fig 4C). At this time a substantial proportion of 2K2N cells was observed ( Fig 4B). Propidium iodide (PI) staining by FACS analysis indicated that the variability on the fluorescence of this compound associated with the cell population was greater for the RNAi-induced cells than the uninduced cells. Nevertheless, overall fluorescence does not vary between induced and uninduced cell lines, an indication that DNA content is not significantly changed upon silencing of TbLAP1 ( Fig 4E).
continues, the nabelschnur elongates, and concentration of TbLAP1 at the nabelschur centre increases along with the protein located in kDNA (D). kDNA segregation is completed following fading of the nabelschnur, with the remaining TbLAP1 signal co-localized with kDNA (E). The TAC signal has clearly divided into two structures and co-migrates with the kDNA/TbLAP1 association (F). 2K2N cells display similar localisations for both proteins as do the 1K1N type (G). Scale bars: 1 μm.
https://doi.org/10.1371/journal.ppat.1006310.g001 Immunofluorescence assay of cells expressing in situ-tagged TbLAP1-V5. Cells were double-labeled with polyclonal anti-V5 (green) and YL1/2 (red) antibodies. DAPI (blue) shows the location of the nucleus and kDNA. Zoom shows enlargement of kDNA and associated region from merge. TbLAP1 (detected with anti-V5 antibody) colocalizes with kDNA, which is positioned next to the basal body (A). TbLAP1 forms two foci overlapping with the dividing kDNA; the basal body has also divided (B). As kDNA begins segregation, the signal of the TbLAP1 arranges in a bi-lobular, dumbbell-like structure (C). The nabelschnur becomes apparent at the start of progeny kDNA segregation (D). At an advanced stage of kDNA segregation, the nabelschnur begins to fade, with only a small accumulation of TbLAP1 evident in the center of the fading link (E). Once the kDNAs have segregated, the The actual evidence is that cell duplets accumulate in culture, represented by 2K2N cells, in accordance with the DAPI staining results. RNAi-ablated cells display a dividing kDNA, prior to cytokinesis (Fig 4F).

Effects of inducible ectopic expression of TbLAP1-HA
Ectopic expression of TbLAP1-HA in procyclic T. brucei induced a significant growth defect (Fig 5A), which was manifested by the accumulation of 0K1N cells ( Fig 5B) and a defect in mt membrane potential (Fig 5D). Ectopic expression of TbLAP1-HA was monitored by Western nabelschnur completely disappears and the TbLAP1 signal remains localized exclusively to the kDNA disks, with the basal body juxtaposed next to it (F). Scale bars: 1 μm.
The staining of procyclic trypanosomes expressing TbLAP1-HA with MitoTracker resulted in an uneven, patchy distribution of the probe throughout the reticulated mitochondrion (S8 Fig), suggesting a defect in mt membrane potential. Indeed, we observed a very rapid effect on this parameter, evident after only 2 hrs of induction of TbLAP1-HA. Over-polarization of the mt membrane continued until 72 hrs of induction, after which the mt membrane potential decreased and fell below values in uninduced cells, presumably due to failure of the mt membrane and/or loss of viability ( Fig 5D).
When TbLAP1-HA expressing trypanosomes were immunodecorated with anti-TAC102 antibody, TAC segregation was heavily affected by the halt in kDNA separation (Fig 6). Cells expressing TbLAP1-HA are able to duplicate their kDNA, but proper separation fails (Fig 6B,  6C and 6D). Aberrant kDNA segregation was observed as early as 6 hrs post-induction (Fig 6B). Neither the division of the basal body, nor its segregation were significantly affected ( Fig  7). Immunodecoration of the same cells with anti-mtHsp70 antibody, a marker for the mt matrix [50], demonstrated accumulation of mtHsp70 around the kDNA disk after 48 hrs of induction, with a concomitant loss of the reticulated network structure of the mitochondrion ( Fig 8B). Expression of TbLAP1-HA was associated with extensive kDNA loss that peaked after 72 hrs of induction, when over 60% of cells became akinetoplastic ( Fig 8A) and lost the reticulated mt network, as well as the focal distribution of TbLAP1-HA ( Fig 8B).

Discussion
Here, we report the first case of a mitochondrion-localized LAP, namely TbLAP1 in T. brucei, which has a unique and unexpected function. In this compartment, TbLAP1 is present in both the kDNA disk and the nabelschnur, where it undergoes dynamic relocations during kDNA division and segregation. Dynamic repositioning during the kDNA replication is known for a number of kDNA-associated proteins, such as the one observed for DNA polymerase ID [1,6,9,51,52], yet to the best of our knowledge TbLAP1 is the first protein known to be localized within the nabelschnur. This morphologically prominent structure was first observed and defined by transmission electron microscopy under treatment with ethanolic phosphotungstic acid, a technique that highlights highly basic proteins [28]. The nabelschnur is formed by two parallel filaments that form a bridge between the segregating kDNA networks [28]. TbLAP1 has a basic pI of 9.8, which suggests the potential to interact with nucleic acids. It is noteworthy that the localization of TbLAP1 to the kDNA disk usually does not cover the entire network. This is particularly evident at the G 1 phase and immediately before cytokinesis, when progeny kDNA networks have completed segregation and the TbLAP1 foci are asymmetric, yet invariably colocalize with the kDNA disk.
Cytosolic forms of LAP are present in most eukaryotes, but the expansion into three paralogs in T. brucei, revealed by phylogenetic analysis (S1 Fig, S2 Fig) is notable and suggests lineage-specific evolution of these proteins. The division of functions between LAPs following their expansion is apparently unique for trypanosomes and does not inform us about the functions of LAP paralogs in other lineages. In T. cruzi LAP has been described as a cytosolic protein displaying enzymatic activity [53] being orthologous to the protein studied here (S2 Fig). Down-regulation of TbLAP1 resulted in a growth defect, as well as in the accumulation of cell doublets. DAPI counts and PI staining of the TbLAP1-depleted cells suggests delayed cytokinesis. On the other hand, ectopic expression of TbLAP1-HA causes the loss of kDNA and concomitant growth defects. Since the process of kDNA segregation finishes with the onset of cytokinesis, both the effect of down-regulation and ectopic expression of TbLAP1 indicate its clear involvement in segregation of kDNAs. Although our results cannot clearly establish a molecular model for the process driven by TbLAP1, we propose that this protein is involved in the machinery that orchestrates and possibly times the final stages of cell division, led by the movement of the basal bodies. Similar phenotypes have been observed upon RNAi-mediated depletion of katanins and spastin in bloodstream form T. brucei, with these proteins known to be involved in cytokinesis [54]. However, the down-regulation of TbLAP1 does not seem to cause a defect in cell division, since as long as 8 days post RNAi-induction, no multinucleated cells were observed in the culture. While it may be argued that the tag impairs the function of the ectopic copy in a dominant-negative fashion, we have found that in situ tagging of TbLAP1, with either short V5 or long GFP tag does not produce the phenotype caused by the ectopic copy, though it exhibits exactly the same localization (S6 Fig). Overexpression of a cytochrome b 5 reductase-like protein renders kDNA incapable of duplication prior to cell division [55], whereas excess levels of two of six kDNA helicases induce a gradual loss of kDNA [27]. Ectopically expressed TbLAP1-HA accumulates around the kDNA disk, inducing a rapid loss of the structure. Several proteins are known to be important for kDNA segregation [56,57] or its maintenance [22,27,58,59], yet none of them display a localization similar to that of TbLAP1, nor a kDNA loss of this type.
Ectopically expressed TbLAP1 disrupted kDNA disk segregation, but division and segregation of the basal bodies were not affected. The basal body has been described to mediate kDNA segregation [60] and the same has been implied for TAC-associated proteins, such as p166 [56]. Moreover, it is well established that nuclear division and segregation are independent of the mechanism of kDNA segregation [47,61]. Throughout the cell cycle, the TbLA-P1-HA signal dynamically changes, and its movement resembles that of the basal bodies ( Fig  9). The stress experienced by trypanosomes expressing TbLAP1-HA is further reflected by the recruitment of mtHsp70 and its colocalization with TbLAP1-HA. Moreover, the viability of the newly divided cells was seriously hampered, as observed by the accumulation of 0K1N cells. It is likely that the collapse of mt membrane potential is a secondary effect of kDNA loss, as the same phenotype occurs following the depletion of several kDNA polymerases [20]. On the other hand, the down-regulation of TbLAP1 affects the separation of 2K2N cells but does not seem to affect division. These results strongly associate the protein with the segregation process and with the basal bodies as its orchestrators. Moreover, they provide further support for the independence of the division and segregation processes in the parasite.
In an attempt to determine proteins interacting with TbLAP1, we constructed a C-terminal in situ GFP-tagged version of the protein and the corresponding cell line was used to immunoprecipitate TbLAP1-GFP using anti-GFP nanobodies (S6 Fig). Although the eluate did not contain any other proteins than TbLAP1-GFP, we noticed that the protein was not able to enter the polyacrylamide gel. Reversion of this phenomenon upon DNAse treatment and/or sonication suggested binding to kDNA (S7 Fig)[53]. Multiple attempts and alternative immunoprecipitation conditions did not yield any TbLAP1-interacting proteins, although both the tagged and untagged proteins were efficiently pulled down, likely a consequence of its capacity to form oligomers, as described in other eukaryotes [65]. We explain this result as a consequence of the short-lived nature of the nabelschnur, which is observed only at the moment of kDNA segregation. Since kDNA replication commences before that of the nucleus and lasts for only a short period of time, at any point in a non-synchronized culture a very small percentage of cells are undergoing the G 2 /M phase transition, during which the newly synthesized kDNAs separate. While our experiments cannot assert that TbLAP1 binds kDNA directly, the association with the structure is evident.
Interaction of LAP with DNA has been observed in other organisms. DNA-associated PepA from E. coli regulates the pyrimidine-dependent repression of the carbamoylphosphate operon transcription [35] and participates in the site-specific Xer recombination system [62,63]. It is worth noting that proteolytic activity is not necessary for recombination activity  [64]. Analysis of the TbLAP1 sequence indicates that not all of the amino acids involved in DNA binding and recombination activity of the E. coli orthologue are present in the T. brucei protein (S3 Fig, S4 Fig and S5 Fig). Of the nine amino acids required for exclusive binding of PepA to DNA [65], TbLAP1 contains only two (S3 Fig). Furthermore, TbLAP1 displays extra 100 amino acids that are absent in all the other LAPs assessed, and this sequence does not correspond to any known domain (S1 Fig, S3 Fig). In other organisms such as basidiomycetes, LAP promotes meiosis and may also be involved in DNA repair [36]. DNA binding of LAP has also been observed in human esophageal carcinoma, where it promotes G 1 /S transition, yet the underlying mechanism remains unknown [66]. The cysteinyl-glycyl activity is a more recently described function of LAPs. Although highly specialized when compared to that of the cleavage of N-terminal peptides, this LAP activity ranging from bacteria to mammals was proposed to recycle cysteinyl-glycyl in the γ-glutamyl cycle [42,67]. However, many components of both the γ-glutamyl cycle and the urea cycle with which it is closely associated, are absent from the T. brucei genome [68]. Hence, this enzymatic pathway likely does not take place in trypanosomatids, where glutathione is just one of several precursors for the subsequent formation of trypanothione. In summary, a new function and subcellular localization of LAP has been described in T. brucei, where the protein is involved in a unique mechanism.

Phylogenetic reconstruction and sequence analysis
Dataset for phylogenetic analysis was created from publicly available sources, using BLASTP at an E-value cut-off of 1x10 -20 . The dataset was aligned by MUSCLE [69] and informative positions were selected using Gblocks [70] with manual adjustment in Seaview 4.6.1 [71]. Maximum likelihood tree was constructed using RAxML 8.2.1 [72] with LG+GAMMA model and 1 000 bootstrap replicates. TriTrypDB and Genbank accession numbers for the sequences used are listed as follows: Trypanosoma brucei Tb927. 8

Construction of cell lines
A TbLAP1 RNAi cell line was constructed by cloning into the p2T7 Ti -177 vector [73] a PCR amplified 400 bp-long region of the TbLAP1 gene using primers TTGGTGTTGAGCTTCTGG TG and TTGCCAGACCTTTTCTTTCC. The final construct was linearized with NotI and transfected into procyclic SMOXP9 T. brucei [74]. For overexpression, the full-size TbLAP1 gene, amplified with primers AAAAGTAAAATTCACGGGCCCATGCTCAAGAGAGT and CAGATTTTCGTTTCTGGTACCTCAATTGCCAGACCT (inserted ApaI and KpnI restriction sites are underlined, respectively), was cloned into the HindIII+BamHI pre-digested p2329 vector [75] using the Geneart Seamless Cloning and Assembly kit (Life Technologies).

Cell cultivation and transfection
All experiments were conducted SDM79 medium. The procyclic 13-13 and SMOXP9 (which we refer to as SMOX) cell lines were used as parental lines for the ectopic expression of TbLA-P1-HA (referred to as TbLAP1-HA throughout the paper) and RNAi against TbLAP1, respectively. In situ tagging of TbLAP1 was performed in procyclic 427 and SMOX cell lines. The SMOX cell line was cultivated in SDM79 with puromycin (0.5 μg/mL) for the stable expression of tetracycline (Tet) repressor and T7 polymerase. The 13-13 trypanosomes were grown with phleomycin (5 μg/mL) for the maintenance of stable expression of pHD13-13-encoded Tet repressor [78], and the SMOX cell line was cultivated in the presence of puromycin (0.5 μg/ mL). Transfections were performed with 10 μg of linearized vector or PCR product and 2 x 10 7 cells per transfection using a BTX electroporator. Clones were selected as described previously [79]. Induction of RNAi and overexpression was initiated by the addition of 1 μg/mL Tet to the cultures. Cell numbers were monitored using a Beckman Coulter Z2 counter. All growth curves were started with 2 x 10 6 cells/mL and subcultured to the same cell number every 24 hrs.

Immunofluorescence and DAPI staining
Ectopically expressed and in situ tagged TbLAP1 was followed by immunofluorescence assay as described elsewhere [80], with minor modifications. Expression was induced with 1 μg/mL Tet and samples were taken at several time points, with parental cell lines used as controls. Cells were fixed with 4% (w/v) paraformaldehyde in phosphate buffered saline (PBS), permeabilized with 0.2% (v/v) TX-100 in phosphate buffered saline (PBS) on microscopy slides and then probed with primary antibodies in PBS/gelatin. Polyclonal anti-TbLAP1 and anti-HA antibodies produced in rabbits (Sigma-Aldrich), and monoclonal anti-mtHsp70 antibody (kindly provided by Ken Stuart) were used at 1:1,000 dilution. Monoclonal anti-TAC102 and YL1/2 antibodies (kindly provided by Torsten Ochsenreiter and Keith Gull, respectively) were used at 1:2,500 and 1:500 dilutions, respectively. Rabbit anti-V5 antibody (Sigma-Aldrich) was used at a dilution of 1:8,000. As secondary antibodies, Alexa Fluor 488 anti-rabbit and Alexa Fluor 555 anti-mouse (Life Technologies) were used. DNA was visualized using ProLong Gold antifade reagent with DAPI (Life Technologies), and DAPI counts were performed as described previously [81]. Immunofluorescence analysis was performed using a Zeiss microscope Axioplan 2, equipped with an Olympus DP73 digital camera and detection was carried out with cellSens software (Olympus). Image analysis was performed using Magic Montage plugin for ImageJ [82] and FIJI [83]. All zoomed image sections are approximately 1μm x 1μm.

Transmission electron microscopy of in situ tagged TbLAP1
Procyclic T. brucei expressing in situ tagged TbLAP1-V5 were fixed in 4% formaldehyde and 0.1% glutaraldehyde in 0.1 M 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid buffer (HEPES) for 1 h at room temperature. After washing in 10 mM glycine in HEPES, cell pellets were embedded into 10% (w/v) gelatin at 37˚C, and left on a rotating wheel in 2.3 M sucrose at 4˚C for 4 days, after which they were frozen by immersion in liquid nitrogen. Ultrathin cryosections were obtained using an ultramicrotome EM FCS equipped with UCT cryochamber (Leica Microsystems). Sections were transferred onto formvar-carbon-coated grids using a drop of 2.3 M sucrose/2% methyl cellulose (1:1). They were then washed in HEPES, blocked in a solution containing 5% low-fat milk, 10 mM glycine and 0.05% (v/v) Tween 20 for 1 h and incubated overnight with anti-V5-tag antibody (50 μg/ml; Invitrogen) in the blocking solution at 4˚C. After a wash in 2.5% (w/v) low-fat milk, 5 mM glycine and 0.025% (v/v) Tween 20 in HEPES, sections were incubated for 1 h in goat anti-mouse IgG conjugated to 10 nm gold particles (Aurion), diluted 1:40 in the washing solution. Sections were then washed in HEPES, distilled water, contrasted and dried using 2% methyl cellulose with 3% aqueous uranyl acetate solution diluted at 9:1, and examined in either 80 kV JEOL 1010 or 200 kV 2100F transmission electron microscopes. Background labeling was tested by a negative control, in the absence of primary antibody, under the same conditions as those described above.

Western blot analysis and FACS analysis
Cell lysates were prepared in NuPAGE LDS sample buffer (Invitrogen) using 5 x 10 6 cells per lane separated on Bolt 4-12% Bis-Tris polyacrylamide gels (Invitrogen) and transferred to a Amersham Hybond P PVDF membrane (GE Healthcare), which was subsequently hybridized with monoclonal anti-GFP (Roche), polyclonal anti-HA (Sigma) or monoclonal anti-tubulin antibodies at 1:1,000, 1:2,000 and 1:10,000 dilutions, respectively. After hybridizing with an appropriate secondary antibody conjugated with horseradish peroxidase (Sigma), Clarity ECL substrate (Bio-Rad) was used to visualize the proteins. Band densitometry analysis was analyzed using ImageJ software [82].
FACS analysis was performed using a FACS Canto II flow cytometer (BD Biosciences). For mt membrane potential measurement, 5 x 10 6 cells procyclic trypanosomes overexpressing TbLAP1 were incubated with MitoTracker Red CMXRos or TMRE (Tetramethylrhodamine) at 27˚C for 20 min in SDM79, spun, and the pellet was resuspended in 1 mL PBS. Ten thousand events were measured per sample and each set of samples was measured at least three times in independent experiments [84]. Propidium iodide (PI) stained cells were prepared as described elsewhere [85]. Briefly, 2 x 10 7 RNAi-induced cells were collected by centrifugation, washed with PBS, resuspended in 200 μL ice-cold 0.5% formaldehyde/PBS and incubated for 5 min on ice. Next, they were fixed with 2 mL of ice-cold 70% ethanol in vortex and allowed to stand 1 hr on ice. To stain cells with PI, the sample was centrifuged at 1500 x g, resuspended in 1 mL PBS and incubated with 50 μg PI and 200 μg of RNase A at 37˚C for 1 hr. Data were analyzed using Flowing Software (Turku Centre for Biotechnology).