ATPaseTb2, a Unique Membrane-bound FoF1-ATPase Component, Is Essential in Bloodstream and Dyskinetoplastic Trypanosomes

Summary

The presence of the FoF1-ATP synthase in every aerobic organism suggests that evolution has settled on a basic blueprint for the complex rotary motor capable of synthesizing life’s universal energy currency—ATP. However, compared to yeast and mammalian models of the FoF1-ATP synthase, several recent studies have reported unique structural and functional features of this complex from organisms representing the clades of Chromalveolata, Archaeplastida and Excavata. One of the most striking cases is observed in trypanosomes, important parasites of humans and animals. Notably, the FoF1-ATP synthase/ATPase switches from synthesizing ATP in the insect vector life stage to hydrolyzing ATP in their mammalian hosts to generate the essential mitochondrial membrane potential (Δψm). Moreover, this indispensable FoF1-ATPase contains up to 14 trypanosome-specific subunits. Here we characterize one such novel subunit, ATPaseTb2. We demonstrate that this subunit is crucial for the survival of the infectious stage of trypanosomes, part of the fully assembled FoF1-complex and it is essential for maintaining the Δψm. Given the enzyme’s irreplaceable function and extraordinary composition, we believe that the FoF1-ATPase is an attractive drug target.

Introduction

Trypanosomes are unicellular flagellates from the order Kinetoplastida, which is comprised of some of the most devastating human pathogens in the world. For example, in sub-Saharan Africa, infection from Trypanosoma brucei gambiense and T. b. rhodesiense causes Human African Trypanosomiasis, which is almost always fatal if left untreated [1]. The latest WHO reports estimate that there are 10,000 new cases annually in endemic regions. Meanwhile, a third subspecies, T. b. brucei, infects livestock and therefore negatively affects the human population through malnutrition and economic hardships [2,3].

T. brucei parasites have a complex life cycle, alternating between the mammalian host and the insect vector, a tse-tse fly. During this environmental switch, the protist undergoes rapid and dramatic changes in cell morphology and metabolism [4–6]. In particular, the single mitochondrion undergoes extensive remodelling, which reflects the adaptability of the parasite to consume different carbon sources based on their availability [4]. The procyclic (insect) form (PF) of trypanosomes catabolizes amino acids and maintains a well-developed mitochondrion with abundant cristae, Krebs cycle enzymes and a complete oxidative phosphorylation pathway. This pathway includes enzymatic complexes that generate a mitochondrial (mt) membrane potential (Δψ_m) that is coupled to ATP synthesis by the F_oF₁-ATP synthase [5]. In contrast, the bloodstream form (BF) of this parasite populates the glucose-rich fluids (e.g. blood and spinal fluid) of its vertebrate host, allowing them to utilize just glycolysis for ATP production. This results in a drastically reduced mitochondrion that lacks significant cristae, key enzymes of the Krebs cycle and the cytochrome-containing respiratory complexes that pump protons into the inner mt membrane space [6,7]. Despite this reduction, the BF mitochondrion is still an active organelle, holding vital processes e.g. lipid metabolism [8], ion homeostasis [9], calcium signalling [10,11], FeS cluster assembly [12] and acetate production for de novo lipid biosynthesis [13]. Importantly, in the absence of proton-pumping respiratory complexes III and IV, the indispensable Δψ_m is sustained mainly by the hydrolytic activity of the F_oF₁-ATPase. Thus, this complex possesses an essential, unique and irreplaceable function in BF mitochondria [14].

In other eukaryotes, this reverse activity of the F_oF₁-ATP synthase is observed only rarely, for very brief moments of time and under very specific conditions (i.e. during oxygen deprivation or in response to damaged or mutated mt respiratory proteins). When the function of the respiratory complexes is compromised, the Δψ_m falls below a physiological threshold and is restored by the reverse proton pumping activity of the F_oF₁-ATPase, which is powered by ATP hydrolysis. The hydrolytic activity of the catalytic F₁-ATPase is also essential for exceptional cells that lack mtDNA (ρ° cells). These cells do not express several core subunits of the membrane embedded F_o-moiety (subunits 6, 8 and 9 in yeast, subunits a and A6L in bovine) of the F_oF₁-ATPase, notably those that are components of the proton pore. Thus, the matrix protruding F₁-ATPase energizes the inner mt membrane by coupling ATP hydrolysis with the exchange of ADP^3- for ATP^4- by the ATP/ADP carrier (AAC) [15]. The same mechanism for producing the Δψ_m is utilized by trypanosomes that lack a mt genome, which is called a kinetoplast [16]. These naturally occuring dyskinetoplastic forms (Dk) of T. brucei (e.g. T. b. evansi or T. b.equiperdum) cause economically significant diseases in horses, camels, and water buffaloes. Remarkably, these parasites are not able to switch to the insect stage and are transmitted mechanically by bloodsucking insects or by coitus [17]. The mtDNA-lacking trypanosomes can also be chemically induced in the laboratory (e.g. Dk T. brucei EATRO164) [18]. Interestingly, each of the Dk cell lines characterized so far, bear one of several different compensatory mutations in the nuclear encoded subunit γ that enable the Δψ_m to be generated independently of the F_o-moiety [14,16,19].

In general, the F_oF₁-ATP synthase complex consists of two functionally distinct enzymatic segments: the hydrophilic F₁ catalytic moiety and the membrane-bound F_o pore. Both of these subcomplexes are linked together by the central and peripheral stalks. The central stalk rotates with the c-ring when protons are allowed to pass through the F_o pore, located between the c-ring and subunit a. In contrast to the rotation of the central stalk, the stationary peripheral stalk plays a crucial role in keeping the catalytic F₁ headpiece static, thus resisting the rotational torque. The eubacterial F₁-moiety consists of the catalytic domain and the central stalk, which are comprised of five subunits in a stoichiometry of α₃, β_3, γ_1, δ_1, ε₁. The F_o-moiety is composed of the oligomeric c_10–15 ring and a single subunit a joined together with two copies of subunit b, which extend from the membrane and form the base of the peripheral stalk. The composition of the eukaryotic enzyme has been determined mainly from detailed studies of F_oF₁-ATP synthase purified from the mitochondria of Bos taurus and Saccharomyces cerevisiae, members of the clade Opisthokonta. These enzymes contain homologous prokaryotic-like core components, but also incorporate additional eukaryotic specific subunits involved in the structure, oligomerization and regulation of the complex [20]. These novel subunits have been assigned to various regions of the F_oF₁-ATP synthase: i) subunit ε and inhibitory factor 1 (IF₁) bind to the F₁-moiety; ii) the small hydrophobic subunit 8 (subunit A6L in bovine) is located in the membrane interacting with subunit a; iii) the soluble subunit h (subunit F₆ in bovine), oligomycin sensitivity-conferring protein (OSCP) and hydrophilic subunit d are assigned to the peripheral stalk; iv) subunits e, f, g, all containing one transmembrane helix, are regarded as accessory peripheral stalk proteins that predominantly reside in the inner membrane where they presumably stabilize the utmost hydrophobic subunit 6 (or subunit a in bovine), thereby assisting the peripheral stalk with its stator function [21–23].

Recently, however, several studies have reported unique structural and functional features of the F_oF₁-ATP synthase from medically relevant parasites and other organisms that represent the clades of Chromalveolata [24], Archaeplastida [25] and Excavata [26]. In Plasmodium falciparum, a member of the phylum Apicomplexa, the genomic data indicates the likely presence of all eukaryotic F₁ subunits and some of the F_o and stator subunits [27]. Conspicuously absent are the subunits a and b, which are crucial to the function of all known F_oF₁-ATP synthases. These subunits are also missing in the ATP synthase of P. yoelii and Tetrahymena thermophila, the latter a member of the sister phylum Ciliophora, both of which have been shown experimentally to possess ATP synthase activity [28–30]. This suggest that ciliates and apicomplexan species employ highly divergent or novel subunits to fulfil the functions of the classical subunits a and b. Another example of an unusual F_oF₁-ATP synthase complex comes from studies of the chlorophycean algea Chlamydomonas reinhardtii and Polytomella sp. that determined the complex contains up to 9 unique subunits (Asa1-Asa9) that either form an innovative peripheral stator or are responsible for complex dimerization [31].

Trypanosoma F_oF₁-ATP synthase consists of the well conserved F₁-moiety comprised of subunits α, β, γ_, δ_, ε and the trypanosome-specific subunit p18 [26,32], and the less characterized F_o pore and peripheral stalk where only subunits c, a and OSCP were identified at the gene or protein level [26,33]. Additionally, the complex contains up to 14 Kinetoplastida-specific subunits that lack homology to any of the previously described subunits. Therefore, their position within the complex and their function is unknown. Interestingly, these recently identified F_oF₁-ATP synthase subunits specific for T. brucei (ATPaseTb) may represent early evolutionary attempts to create the functional and structural components of the eukaryotic accessory subunits for this early divergent species. Previously, two of these novel subunits, ATPaseTb1 (Tb10.70.7760) and ATPaseTb2 (Tb927.5.2930), were shown to be essential for the proper function and structural integrity of F_oF₁-ATP synthase in the procyclic form of this parasite [26].

Here, we have extended this study and functionally characterized ATPaseTb2 in the disease-relevant stage of T. brucei and Dk T. b evansi. Our results demonstrate that the membrane-bound ATPaseTb2 protein is indispensable for the survival of BF and Dk trypanosomes because it is crucial for maintaining the Δψ_m in these cells. Furthermore, the unexpected observation of assembled F_oF₁-ATPase complexes in two different strains of trypanosomes lacking mt DNA suggests that the composition of the Dk F_oF₁-ATPase complex is more similar to the BF F_oF₁-ATPase complex than previously thought.

Results

Bioinformatics reveals that ATPaseTb2 contains a putative transmembrane domain and a region homologous to the bovine subunit d

ATPaseTb2 is annotated as a hypothetical protein in the TriTrypDB database (www.tritrypdb.org) and its mt localization was predicted by the presence of an N-terminal mt targeting sequence assigned by Mitoprot II v1.101 [34] (probability 0.823). Strikingly, this subunit has no detectable homologs outside Kinetoplastida based on a similarity sequence search (e.g. BLAST). To uncover any homologous relationships for protein sequences that do not share high enough sequence identity, but might have similar secondary structure, we employed the HHpred toolkit (http://toolkit.tuebingen.mpg.de), which is based on an HMM-HMM comparison that reveals homologous relationships even if the sequences share less than 20% sequence identity [35]. All of the available Kinetoplastida ATPaseTb2 homologs obtained from the TriTrypDB (S1A Fig.) were analyzed using HHpred software. Among the first five structural homologs detected was the bovine F_oF₁-ATP synthase subunit d, with a probability of 44.7 and a P-value of 0.0021. Then a consensus sequence generated from an alignment of this 92 amino acids (aa) region from all identified kinetoplastida homologs was resubmitted for HHpred analysis. This search returned a significant hit to the bovine subunit d with a probability of 78.1 and a P-value of 0.0002 (S1B Fig.). This bovine F_oF₁-ATP synthase subunit is comprised of 160 aa and resides within the peripheral stalk where it interacts with all the other three components of the peripheral stalk (b, F₆ and OSCP). This subunit is predominantly hydrophilic, contains several long alpha-helices and is essential for the enzymatic function [23,36,37]. The region of similarity to subunit d falls in the middle of the ATPaseTb2 sequence and it is followed by a putative trans-membrane (TM) domain (aa 239–254) that was predicted by MEMSAT-SVM software (http://bioinf.cs.ucl.ac.uk/psipred/) [38] with a -0.738 prediction score (Fig. 1). This prediction is intriguing because a typical eukaryotic subunit d is not directly attached to the membrane, but rather it interacts with the soluble region of the membrane-bound subunit b via three coiled-coil interactions [23]. Thus, ATPaseTb2 may represent a novel subunit of the membrane-bound peripheral stalk that possibly depicts an early divergent functional alternative of the stator by creating a chimera of subunits b and d.

ATPaseTb2 is a mt membrane-bound protein expressed in various forms of T. brucei

The mitochondrial localization, as well as the structural and functional association of ATPaseTb2 with the F_oF₁-ATP synthase in PF T. brucei cells was previously described [26]. Because the activity of this complex significantly differs between the insect and mammalian stage of the parasite, we were interested if the ATPaseTb2 protein is also expressed in trypanosomes residing in the bloodstream of their host. Therefore, we cultured PF427 (insect stage), BF427 (bloodstream forms) and the laboratory-induced dyskinetoplastic Dk164 T. brucei cells, in addition to the naturally occurring Dk T.b.evansi. Whole cell lysates from PF427, BF427, Dk164 and T.b.evansi cells were fractionated by SDS-PAGE and analyzed by western blot using a specific polyclonal antibody against ATPaseTb2 (Fig. 2A). Interestingly, ATPaseTb2 is expressed in all four cell types; however, its expression is significantly decreased in BF427 and even more reduced in Dk164 and T.b.evansi cells. A similar expression pattern for F₁-ATPase subunit β and mt chaperone Hsp70 was also observed between PF, BF and Dk cells (Fig. 2A). These results are in agreement with the proposed reduction of the mitochondria size, function and activity in the various bloodstream forms of T. brucei [39].

To experimentally investigate the subcellular localization of ATPaseTb2 within the bloodstream T. brucei cell, the ATPaseTb2 gene product was C-terminally tagged with a v5-tag and inducibly expressed using tetracycline (tet). Cytosolic and mitochondrial fractions were isolated from a hypotonic lysate of cells over-expressing the ATPaseTb2_v5 protein. The mitochondrial fractions were then treated with digitonin to create mitoplasts that were further incubated with sodium carbonate (Na₂CO₃) to obtain mt membrane and mt matrix fractions. The subsequent western analyses established the purity of the extracted fractions as the cytosolic enolase, the mt inner membrane carrier protein (AAC) and the matrix-localized guide RNA binding protein (MRP1) were confined within their respective subcellular fractions. Notably, ATPaseTb2 was predominantly localized in the mt membrane fraction, while subunit β of the matrix-soluble F₁-ATPase was released into the matrix fraction (Fig. 2B), demonstrating that ATPaseTb2 is firmly embedded in the inner mt membrane.

ATPaseTb2 is a bona fide subunit of F_oF₁-ATPase monomers and higher oligomers

To confirm that ATPaseTb2 is a genuine subunit of both the BF and Dk F_oF₁-ATPase, we employed the same strategy applied previously for PF cells [26]. Therefore, we first created BF cell lines in which either the ATPaseTb2 or the F₁ subunit p18 were inducibly expressed with a C-terminal tag. Complexes assembled with a tagged protein were then purified using a one-step IgG affinity purification, followed by treatment with TEV protease to release bound complexes from the IgG beads. The TEV eluates were further analyzed by western blot for the presence of the tagged proteins, known F_oF₁-ATPase subunits, and AAC (Fig. 3A). Importantly, the presence of the tagged subunits ATPase Tb2 and p18 in the respective cell lines were confirmed using a specific antibody against the c-myc epitope that comprises part of the tag. In addition, the ATPaseTb2_TAP eluate contained F_oF₁-ATPase subunits β, p18 and ATPaseTb1. Vice versa, the p18_TAP eluate contained subunits ATPaseTb2, ATPaseTb1 and subunit β. While AAC has been described to associate with mammalian F_oF₁-ATPase under certain conditions [40], it is not a core component of the enzyme and the lack of its signal indicates the stringency of this technique. The TAP tag purification from non-induced cells was used as a negative control to verify that the detected proteins do not bind non-specifically to the charged beads. To analyze the composition of the Dk F_oF₁-ATPase, p18 tagged complexes were purified and subjected to the same set of antibodies described for the BF complexes (Fig. 3B). These results also depicted the same interactions seen in Fig. 3A. Therefore, the co-purification of ATPaseTb2 with known ATPase subunits p18 and β validates that it is an authentic constituent of the F_oF₁-ATPase complex in both BF and Dk cells.

The sedimentation profile of F_oF₁-ATP synthase in 10–30% glycerol gradients (GG) was previously published and revealed two distinct regions representing the F₁-moiety and the F_oF₁-monomer and multimeric complexes [26]. In order to specify if ATPaseTb2 is a component of the F₁-moiety or the F_oF₁-complex, the glycerol gradient sedimentation profile was determined using mt lysates from all four cell types: PF427, BF427, Dk164 and T.b.evansi. In correlation with the literature, a new specific antibody raised against the F₁-subunit β depicted both regions, which are defined as ∼10S (fractions 6–10), and ∼40S (fractions 14–22) (Fig. 3C). A different sedimentation pattern was revealed when a specific antibody recognizing ATPaseTb2 detected only the 40S region (fractions 14–22), representing the F_oF₁-monomer and -multimers. These findings suggest that the novel subunit is not a component of the F₁-moiety, but rather a member of the fully assembled complex. Furthermore, compelling results hinting at the function of this hypothetical protein were also obtained when the mt lysates of Dk164 and T. b. evansi cells were fractionated on similar gradients. While the majority of the signal for subunit β was predictably detected in fractions 6–10, representing the F₁-ATPase complex, there was also a weak signal detected in higher S values. These same ∼40S complexes were identified using the ATPaseTb2 antibody (Fig. 3C). This indicates that in the absence of subunit a, the F₁-moiety is still attached to the mt membrane via a central stalk, a ring of subunit c and the peripheral stalk. Since the F_o subunit a is missing in these cells, the attachment of this complex to the membrane is presumably not as strong as in BF cells. Therefore, upon treatment with detergent to lyse the mitochondria, a majority of F₁-ATPase is released and detected in lower S values while a small portion of the F_oF₁-complexes is preserved and sediments at higher S value fractions. Consequently, our data suggests that previously undetected membrane-bound F_oF₁-complexes are present in the mitochondria of the Dk trypanosomes.

To verify our intriguing results, the F_oF₁-ATP synthase/ATPase complex was examined by an alternative method involving high resolution clear native electrophoresis (hrCNE) followed by western blot analysis. Results obtained using the antibody against subunit β revealed four predominant bands in both the PF427 and BF427 samples, representing various forms of the F₁-moiety and F_oF₁-complexes (Fig. 3D). The lowest band likely corresponds to the F₁-ATPase (subunits α, β, γ,δ,ε), while the one above it seemingly represents F₁-ATPase with a ring of subunit c, as it has been described for the mammalian complex [41]. Noticeably, in each sample the antibody against ATPaseTb2 immunodetected only two major bands corresponding to the F_oF₁-monomer and -multimers. When the mt lysates of Dk164 and T. b. evansi cells were fractionated on a hrCNE, subunit β was mainly detected in the lower bands representing the F₁-moiety and the F_oF₁-monomer, while only a weaker band was observed at the size depicting F_oF₁-multimers. But western blot analysis with the ATPaseTb2 antibody clearly demonstrated the existence of higher assemblies of the F_oF₁-ATPase complexes (Fig. 3D). Thus, ATPaseTb2 is a membrane-bound subunit of the monomeric and multimeric F_oF₁-ATP synthase/ATPase complex in all four T. brucei cell types examined. Furthermore, no significant differences in size were observed between the F_oF₁-monomeric and -multimeric complexes detected in PF427, BF427 and Dk cells, indicating that subunit a might be the only missing component of the assembled F_oF₁-ATPase complex in dyskinetoplastic cells.

Silencing of ATPaseTb2 inhibits cell growth by decreasing the Δψ_m in BF T. brucei

To assess the importance of ATPaseTb2 in the infective stage of T. brucei and to evaluate its functional association with F_oF₁-ATPase, an ATPaseTb2 RNA interference (RNAi) cell line was created. The expression of dsRNA was triggered by the addition of tet to the culture medium and the in vitro growth of the ATPaseTb2 knock-down (KD) cells was inspected for seven days (Fig. 4A). Strikingly, a significant growth phenotype was already detected in RNAi induced cells on just the second day of tet addition. Presumably, the powerful selection forces acting on these cells missing a critical protein resulted in the emergence of a subpopulation that was no longer responsive to RNAi induction after only 120 hours, leading to the growth recovery of the culture (Fig. 4A). This phenomenon is often reported for RNAi experiments in BF T. brucei [42].

The targeted KD of ATPaseTb2 was confirmed by western blot analysis of whole cell lysates harvested from an equivalent number of cells for ATPaseTb2 RNAi non-induced and induced cells throughout the RNAi time course (Fig. 4B). Western blot analysis using the antibody against ATPaseTb2 exhibited a reduction of the targeted protein by 73% at day 2 of RNAi induction. A non-specific band detected on the same membrane was used to determine equal loading of the samples.

With the efficient knockdown of ATPaseTb2 verified, an in vivo assay was employed to measure the Δψ_m in the ATPaseTb2 KD cell population (Fig. 4C), since the Δψ_m is generated by the F_oF₁-ATPase in BF T. brucei [9]. Flow cytometry analysis was used to determine the changes observed in these cells stained with Mitotracker Red CMX-Ros, whose fluorescent intensity is proportionally dependent on the strength of the Δψ_m. A substantial decrease of the Δψ_m in the KD cell population was observed two days (IND2) after RNAi induction, coinciding with the first time point to display a significant growth inhibition. Importantly, this result confirms the vital function of ATPaseTb2 within the F_oF₁-ATPase.

Depleting ATPaseTb2 does not significantly alter ATP hydrolysis capabilities

To assess if the decrease in the Δψ_m is the result of an impaired ATP hydrolytic activity of the enzyme, the total ATPase activity was measured in mt lysates with or without azide, an inhibitor of the catalytic F₁-ATPase [43]. Typically, the specific F₁-ATPase activity represents ∼ 35–45% of total mt ATPase activity. Our results indicate that neither the total ATPase nor azide-sensitive activities were significantly altered between the non-induced and ATPaseTb2 RNAi-induced cells (Fig. 5A). ATPase activity can also be visualized by in-gel activity staining when mitochondria from these cells are solubilized with dodecyl maltoside and separated on a 2–12% blue native PAGE (BNE) [41]. The specific ATPase staining revealed two major bands in non-induced cells, the predominant lower one representing the activity produced from the F₁-moiety and the less significant higher band representing the hydrolytic activity of the monomeric F_oF₁-ATPase (Fig. 5B). It is important to note that solubilizing mitochondria with any detergent can result in unintended structural consequences not found under normal physiological conditions, especially for a large complex comprised of a matrix moiety and a membrane embedded region. Although detergents tend to dissociate the F₁-moiety from F_oF₁-ATPase complexes, it is intriguing that there is still a portion of ATP hydrolysis being produced from the complete F_oF₁-ATPase in the non-induced BF cells. Conspicuously, this activity is absent when ATPaseTb2 RNAi is induced (Fig. 5B).

Loss of ATPaseTb2 disrupts the F_oF₁-ATPase complexes in BF T. brucei

The absence of activity for the F_oF₁-monomeric complex might be explained by complex instability. Thus, the structural integrity of the coupled F_oF₁-complex was further analyzed using hrCNE, which were loaded with digitonin-lysed mitochondria from non-induced and ATPaseTb2 RNAi induced cells. After transferring the separated native protein complexes to a nitrocellulose membrane, a specific antibody against the p18 subunit detected the F₁-moiety as well as both monomeric and oligomeric F_oF₁-complexes (Fig. 5C). Interestingly, we repeatedly observed that the abundance of the monomeric and oligomeric complexes is decreased after RNAi induction and by day 3 these complexes are not detected at all. These results were complemented by the sedimentation profile of F_oF₁-ATPase complexes on glycerol gradients. Mitochondria were isolated from non-induced RNAi cells and cells induced for two days (IND2). These were then lysed by 1% Triton X-100 and an equal amount of mt protein from each sample was fractionated on a 10–30% GG. The resolved fractions were then analyzed by western blot and the anti-β and anti-p18 antibodies demonstrated the effect of the ATPaseTb2 KD on the F_oF₁-ATPase sedimentation profile. As indicated in Fig. 5D and the corresponding scanning densitometry results listed in S2 Fig., two days of RNAi induction resulted in a diminished amount of β and p18 signals from the higher S-values (IND2 panels), while the GG sedimentation pattern of RNAi non-induced cells was more similar to wild type BF427 (see Fig. 3C). These results are in agreement with the PF RNAi ATPaseTb2 cell lines, in which the stability of the monomeric and multimeric F_oF₁-ATP synthases was significantly affected upon RNAi [26].

Using a variety of methods employing various detergent types, we have established that the ATPaseTb2 subunit is an essential component of the F_oF₁-ATPase monomer and higher oligomers in the bloodstream stage of T. brucei. Furthermore, the depletion of ATPaseTb2 diminishes the abundance of membrane-bound F_oF₁-ATPase complexes, which directly initiates a substantial decrease in the Δψ_m that manifests as a strong growth phenotype.

ATPaseTb2 depletion affects cell growth, the Δψ_m and F_oF₁-ATPase integrity in Dk T. b. evansi

To further dissect the primary role of the membrane-bound ATPaseTb2 subunit, we investigated the effect of targeted gene silencing in transgenic Dk T. b. evansi, which facilitate the tet inducible expression of dsRNA. These Dk cells were transfected with an ATPaseTb2 RNAi vector and several positive clones were screened for a growth phenotype. As shown in Fig. 6A, the surprising inhibition of cell growth in RNAi induced cells appeared two days after tet induction and the doubling time remained reduced throughout the whole ten day experiment. Furthermore, western analysis of whole cell lysates harvested over the RNAi time course exhibited a significant reduction of ATPaseTb2 in the RNAi induced cells beginning on day 1 (Fig. 6B), while the analysis of a non-specific band served as the loading control. Interestingly, if the only role of the subunit was to stabilize the proton pore, it presumably would not be essential in cells that lack mtDNA and have adapted to the loss of a functioning proton pore. However, these results suggest that ATPaseTb2 is important for F_oF₁-ATPase complexes in these unique cells and thus this cell line was further exploited to characterize the function of ATPaseTb2.

Since the Δψ_m in Dk cells is generated using the hydrolytic activity of the F₁-moiety coupled with ATP/ADP translocation, the Δψ_m was examined in ATPaseTb2 non-induced and RNAi induced cell populations. As shown in Fig. 6C, a reduction of the Δψ_m was detected in the RNAi induced cell population after 2 (IND2) and 3 days (IND3). Since the reduction of the Δψ_m already appears at day 2 of tet induction and the growth phenotype does not truly manifest until day 3, we postulate that the observed decrease of the Δψ_m is a direct effect of ATPaseTb2 silencing and not a consequence of decreased cell viability. Furthermore, to assess if this decreased Δψ_m is caused by an impairment of the F_1-ATPase hydrolytic activity, the total ATPase activity was measured in mt lysates with or without azide. Similar to BF RNAi ATPaseTb2 cells (Fig. 5A), no changes in total or azide-sensitive activities were observed between the non-induced and ATPaseTb2 RNAi-induced cells (Fig. 6D).

The structural integrity of Dk F_oF₁-ATPase complexes was then investigated using mild non-denaturing hrCNE to fractionate mitochondrial proteins purified from BF427 cells and Dk cells either induced for ATPaseTb2 RNAi or not. A western blot containing BF427 mitochondria as a reference sample was probed with an anti-β antibody and used as marker to visualize each state of the F₁-moiety, either isolated by itself or being partnered with the monomeric and multimeric F_oF₁-complexes. Strikingly, while a majority of the β subunit signal in Dk RNAi non-induced cells was observed in the region of isolated F₁-ATPase, we also consistently detected an irrefutable signal corresponding to the F_oF₁-monomeric complex, presumably assembled intact with the exception of subunit a (Fig. 6E). While there is evidence of mammalian rho cells containing a similar complex [44–46], this is the first time it has been demonstrated in the comparable Dk trypanosomatids. Furthermore, the induction of ATPaseTb2 RNAi in the Dk cells led to the disappearance of the monomeric complex, leaving only the F₁-ATPase to be detected on the western blot (Fig. 6E). Once again, the evidence indicates that ATPaseTb2 is important for the structural integrity of the F_oF₁-ATPase and its depletion in Dk T.b. evansi cells leads to the disruption of the unique membrane-bound F_oF₁-ATPase complexes.

To corroborate this observation, we purified mitochondria from Dk T. b. evansi non-induced and ATPaseTb2 RNAi-induced cells and lysed them with Triton X-100. These gently disrupted lysates were then fractionated on glycerol gradients, which revealed that a majority of the signal from subunits β and p18 was located within fractions 6–10, representing the F₁-moiety (Fig. 6F). Nevertheless, weaker bands representing the F_oF₁-ATPase complexes (fractions 14–22) were again detected in the non-induced samples (Fig. 6F NON panels). Noticeably, these same protein markers in the mt fractionation of ATPaseTb2 RNAi induced cells displayed a significant reduction in fractions 14–22 (Fig. 6F IND4 panels), where the F_oF₁- monomer/oligomer sediments. This observation is quite surprising since the main role of this enzyme is to create the Δψ_m, presumably using only the hydrolytic activity of the matrix facing F₁-moiety. In an attempt to determine if a portion of the F₁-ATPase is membrane associated and that the disruption of this attachment is the reason for the observed decrease in the Δψ_m and the detected growth phenotype, we decided to visualize these complexes in situ using immunogold labelling followed by electron microscopy.

Loss of ATPaseTb2 leads to the dissociation of F₁-ATPase from the mt membrane

Immunogold labeling with a primary anti-β antibody was performed on ultrasections of Dk T. b. evansi non-induced (NON) and ATPaseTb2 RNAi induced cells for 3 (IND3) and 5 (IND5) days. Electron micrographs illustrate that Dk trypanosomes have a simple and reduced mitochondrion (S3A Fig.), lacking both cristae and a typical kinetoplast structure, as previously reported for the dyskinetoplastic T. b. evansi [17]. Whenever these reduced double-membrane organelles could be unequivocally identified, we tallied the number of immunogold beads either in the close proximity of the mt membrane or in the matrix (S3A Fig.). Strikingly, 70% of the gold particles identified were found associated with the mt membrane in the NON images (S3B Fig.). From the 113 images captured from either NON, IND3 or IND5 samples, we performed a Chi-squared analysis on the immunogold beads associated with the mt membrane to determine if their distribution is random (S3C Fig.). With two degrees of freedom, the calculated χ² = 43.5 (p < 0.0001), which signifies that the difference between the observed and expected particles was statistically significant. These values were then plotted to determine the relative labelling index (RLI = N_obs/ N_exp) for the mt membrane associated gold particles (S3D Fig.). This analysis reveals that the observed number of gold particles found in close proximity to the mt membrane for the NON electron micrographs is significantly greater than if the particles were randomly distributed (RLI = 1), while the opposite is true for the IND5 samples. Since we understand the general subjectivity of this method, we performed this experiment in a blinded study and only considered a gold bead to be associated with the mt membrane if it was in the immediate proximity of the membrane. For the very few scored beads that could potentially fall in a membrane-bound grey zone, it should be noted that the physical distance from the subunit β antigen to the gold particle can be up to 15–20 nm, not to mention that subunit β itself is projected from the membrane by subunit γ by as much as 13 nm [47,48]. Furthermore, very few immunogold beads were detected outside of defined mt structures and each mt image only contained 1–8 immunogold beads (average of 2.4 beads/mt image), suggesting that the labelling was very specific for our abundant mt antigen. It is important to note, that utilizing an assay that does not require the use of any detergents, we demonstrated that a majority of the F₁-ATPase is located within close proximity of the mt membrane and this association is disrupted when ATPaseTb2 is depleted.

Discussion

The T. brucei F_oF₁-ATPase has obtained unique properties in comparison to its eukaryotic counterparts. The occurrence of novel subunits combined with the lack of typical eukaryotic subunits that compose the F_o membrane-bound moiety and the peripheral stalk is intriguing from an evolutionary viewpoint. Because these atypical subunits lack significant homology to known proteins, questions arise concerning their authenticity as genuine subunits of the complex and what their function and localization within the complex might be. Here, we focused on the functional characterization of one of the trypanosomatid specific subunits, ATPaseTb2 in the bloodstream stages of T. brucei. Notably, we determined that its function is essential for maintaining the normal growth rate of the infectious stage of T. brucei and also for the important veterinary dyskinetoplastic parasite, T. b. evansi. Additional analyses demonstrated that the ATPaseTb2 is membrane embedded and a component of the F_oF₁-ATPase that is present in both BF and Dk cells. Furthermore, the depletion of this F_o-moiety subunit results in a decreased Δψ_m and a loss of F_oF₁-ATPase complexes. Combined with bioinformatic tools that predict a transmembrane domain and identify a low homology to subunit d, we suggest that ATPaseTb2 might be a component of the peripheral stalk of the F_oF₁-ATPase.

The peripheral stalk is indispensable for ATP synthesis as it serves to impede the movement of the catalytic α₃β₃ headpiece while the proton motive force rotates the c-ring and the connected asymmetrical central stalk in a way that imposes conformational changes in the catalytic nucleotide binding sites of the β subunits. The peripheral stator is also essential when the complex harnesses the energy provided from ATP hydrolysis to drive the rotation of the enzyme in reverse, allowing protons to be pumped across the mt inner membrane to produce the Δψ_m when physiological conditions dictate.

The bovine and yeast peripheral stalk is composed of four subunits, OSCP, b, d, and F₆/h (bovine/yeast nomenclature). Subunit b (∼20kDa protein) contains two trans-membrane segments at the N-terminus, while the rest of the protein is hydrophilic—protruding into the matrix and interacting with OSCP and subunit d [23,49–51]. The interaction between subunit b and OSCP is stabilized by subunit F₆ [22]. The predominantly hydrophilic subunit d interacts with all three mentioned subunits and it has been shown to be essential for the function of the F_oF₁-complex. The yeast knock-out of subunit d was characterized by the de-attachment of the catalytic F₁-ATPase from the protonophoric sector, the loss of detectable oligomycin sensitive ATPase activity and the absence of subunit a in the F_o-moiety [36]. Notably, of these four subunits, the only homolog identified in the T. brucei genome is OSCP and its protein product was identified as a bona fide subunit of the purified F_oF₁-ATP synthase [26]. Homologs for subunits b, d, and F₆ are missing and most likely have been replaced by the novel proteins that associate with the T. brucei F_oF₁-ATP synthase. Here we demonstrated that the ATPaseTb2 subunit is a membrane-bound protein, containing one predicted transmembrane domain, with a large hydrophilic region extending into the matrix that possesses a low homology to the bovine subunit d. In accordance with other yeast or bovine subunits of the peripheral stalk, the down-regulation of ATPaseTb2 affects the stability of the F_oF₁-complex and its oligomycin sensitivity (this study, [26]).

Considering the relatively large molecular mass of ATPaseTb2 (43 kDa), it is plausible to speculate that ATPaseTb2 functionally represents the membrane-bound subunit of the peripheral stalk fused with subunit d, offering an early attempt by eukaryotes to add layers of complexity that will allow for greater adaptability. It is noteworthy to mention that a species-specific architectural variant of the peripheral stalk was also proposed for colorless and green algae, where the novel subunits Asa2, Asa4 and Asa7 fulfil a structural role in forming the peripheral stalk [52,53]. Other discrepancies from the typical eukaryotic enzyme model can be found in the ciliate Tetrahymena thermophila, representing the superphylum of Alveolates, where the homolog of the conserved subunit b has so far not been identified in the genome. Instead, three novel proteins detected in this ATP synthase complex have been proposed to substitute for subunit b [28]. Thus, it seems that the composition and overall structural appearance of the F_oF₁-ATP synthase from organisms representing different lineages other than Opisthokonts, might be more diverse than previously thought and deserve more attention from basic researchers to elucidate alternate evolutionary solutions to a common thread of life.

T. brucei is a member of the Excavata clade and represents a desirable model to study the function and regulation of the F_oF₁-ATP synthase/ase complex since it utilizes the proton motive force to produce ATP in the PF life stage, while in the BF stage it hydrolyzes ATP to pump the protons necessary to create the Δψ_m. Moreover, dyskinetoplastic T. b. evansi employs the hydrolytic activity of the F₁-ATPase and the electrogenic exchange of ADP^3- for ATP^4- by the AAC as yet another strategy to generate the Δψ_m (Fig. 7). Therefore, it was surprising that the ATPaseTb2 protein was detected in all four examined cell types (PF, BF, Dk 164 and T. b. evansi), albeit at a much lower abundance in BF and Dk cells, but that is in full agreement with the reduced mitochondria previously defined for these two forms [39]. Furthermore, when PF and BF mt lysates were resolved by GG and hrCNE, ATPaseTb2 was only detected when co-localized with the monomeric and oligomeric F_oF₁-ATP synthase/ase complexes. Interestingly, similar results were obtained from the mt lysates of cells lacking mtDNA. Considering that only the F₁-ATPase was previously assumed to be important for maintaining the Δψ_m in the Dk trypanosomes, the presence of these complexes is intriguing. Nevertheless, a proteomics study performed with osteosarcoma 143B ρ° cells revealed that in addition to the F₁-ATPase subunits, subunit d of the peripheral stalk was also identified [54]. In a similar project involving fibroblast MRC5 ρ° cells, the oligomycin insensitive complex was purified and determined to contain several F_o (b and c) subunits along with a couple of peripheral stalk subunits (OSCP, d). Notably, this complex was loosely associated with the mt membrane [45]. Furthermore, several additional studies have shown that the F₁-ATPase in yeast or mammalian cells lacking a mt genome (i.e. mammalian subunits a/ATP6, A6L; yeast subunits 6, 8 and 9) is membrane associated, with the attachment most likely occurring via the central and/or peripheral stalk [44,46,55,56]. Our data also suggest that since ATPaseTb2 membrane-bound complexes (monomeric and multimeric) do not display significant differences in their native size or sedimentation values, then the mt encoded subunit a might be the only missing subunit from these complexes. Such a uniquely structured complex can be explained by the current model for the assembly of the F_oF₁-ATPase, in which the F_o subunit a is the last protein incorporated into the enzyme to ensure that it is able to function properly before introducing a complete proton pore. Furthermore, this integration is dependent on the presence of subunit b of the peripheral stalk [57].

RNAi silencing of ATPaseTb2 in BF cells caused a strong growth phenotype concurring with a decreased Δψ_m. Importantly, ATPaseTb2 also proved to be essential for maintaining the normal growth rate and the Δψ_m of Dk cells. Combined with the hrCNE and glycerol gradient sedimentation assays that revealed a decrease of the high molecular weight complexes, but not the F₁-ATPase, our studies suggest that these membrane-bound enzymes might significantly contribute to the membrane potential in Dk cells. The biological significance of this localization of the F₁-ATPase might be to efficiently coordinate the activity of this enzyme with its substrate transporter, both of which are responsible for producing the Δψ_m in Dk. By restricting the F₁-ATPase to the mt membrane, it keeps it within the close vicinity of its biochemically functional partner, the AAC. In this way, it helps ensure that the hydrolysis of ATP results in a relatively concentrated region of substrate for the mt membrane embedded AAC, especially in a mitochondrion lacking defined cristae and the microenvironments they create. Indeed, there is precedence for such a close spatial collaboration to increase efficiency, as an actual physical interaction between these two enzymes has been reported as the ATP synthasome in mammals [40,58]. We are currently building a set of tools to perform additional functional assays and imaging techniques that we hope will further resolve the function of the F_oF₁-ATPase complexes in these pathogenic trypanosomes.

Materials and Methods

Plasmid construction

The ATPaseTb2 (Tb927.5.2930) RNAi constructs targeted an 825 bp fragment of the gene that was PCR amplified from T. brucei strain 427 genomic DNA with the following oligonucleotides: FW—CACGGATCCATGCGCCGTGTATC, REV-CACCTCGAGTTCGGCCCGATC. Utilizing the BamHI and XhoI restriction sites inherent in the primers (underlined), this fragment was cloned into the p2T7-TA blue plasmid [59] (used to create a BF RNAi cell line) and the p2T7–177 plasmid [60] (used to generate a Dk RNAi cell line). For the inducible expression of ATPaseTb2 fused with a C-terminal 3xv5 tag, the ATPaseTb2 coding sequence was PCR amplified (Fw: ACAAAGCTTATGC GCCGTGTATC, Rev: ATAGGATCCGTGATGGGCCTTTTTC) and cloned into the pT7_v5 vector using HindIII and BamHI restriction enzymes [61]. The pLEW79MHTAP vectors for the tet inducible expression of ATPaseTb2 and Tb927.5.1710 (F_oF₁-ATPase subunit p18) TAP-tagged proteins were previously described in [26].

Trypanosoma culture conditions and generation of cell lines

Bloodstream form T. brucei brucei Lister 427 strain, stable acriflavin-induced dyskinetoplastic T. b. brucei EATRO 164 [18] and dyskinetoplastic T. b. evansi Antat 3/3 [62] were all grown in vitro at 37°C and 5% CO₂ in HMI-9 media containing 10% FBS. The BF427 single marker cell line [63] and a T. b. evansi cell line [14], both constitutively expressing the ectopic T7 RNA polymerase and tet repressor (TetR were used for the tet inducible expression of dsRNA and the TAP- and v5-tagged proteins. Each plasmid was linearized with NotI enzyme and transfected into the appropriate cell line as described previously [63]. Linearized p2T7–177 was targeted to the minichromosome 177 basepair repeat region, while linear pTv5 and pLEW79MHTAP vectors were integrated into the rDNA intergenic spacer region [60,61,63]. The induction of RNAi and ectopically expressed tagged proteins was triggered by the addition of 1 μg/ml of tet into the media. Cell densities were measured using the Z2 Cell Counter (Beckman Coulter Inc.). Throughout the analyses, cells were maintained in the exponential mid-log growth phase (between 1x10⁵ to 1x10⁶ cells/ml).

Tandem affinity purification (TAP) of tagged complexes

The TAP protocol was generously provided by L. Jeacock and A. Schnaufer (personal communication) and optimized for BF cells. Briefly, 2x10⁸ cells were harvested and lysed for 20 min on ice with 1% Triton X-100 in an IPP150 buffer (150 mM NaCl, 0,1% NP40, 10mM Tris-HCl pH 8.0) containing Complete protease inhibitors (Roche). The lysates were then cleared by centrifugation (16,000 g, 15 min, 4°C). Meanwhile, IgG antibodies were covalently bound to Dynabeads M-270 Epoxy (Invitrogen) using a protocol previously described [64]. Charged beads were blocked with 1% BSA and then incubated with the cleared cell lysates (4°C for 2 hours, constantly rotating). The beads were then washed three times with IPP150 and once with a TEVCB buffer (150 mM NaCl, 0.1% NP40, 0.5 mM EDTA, 1 mM DTT, 10 mM Tris-HCl, pH 8.0). Finally, the bound protein complexes were released by AcTEV protease (Invitrogen) cleavage and the eluate was analyzed by SDS-PAGE.

Isolation of mt vesicles

Crude mt vesicles were obtained by hypotonic lysis as described in detail earlier [65]. Briefly, cell pellets from ∼2x10⁹ cells were washed with SBG (150 mM NaCl, 20 mM glucose, 20 mM NaHPO₄, pH 7.9), resuspended in DTE (1 mM Tris, 1 mM EDTA, pH 8.0) and homogenized in a Dounce homogenizer. Alternatively, for a smaller scale hypotonic isolation, cell pellets from ∼1x10⁹ cells were washed in NET buffer (150 mM NaCl, 100 mM EDTA, 10 mM Tris, pH 8.0), resuspended in DTE and homogenized by passing through a 25G needle. To re-create the physiological isotonic environment, 60% sucrose was immediately added to the lysed cells to reach a final concentration of 250 mM. Samples were spun down at 15,000 g, 10 min, 4°C to clear the soluble cytoplasmic material from the lysates. The resulting pellets were resuspended in STM (250 mM sucrose, 20 mM Tris pH 8.0, 2 mM MgCl₂), supplemented with a final concentration of 3 mM MgCl₂ and 0.3 mM CaCl₂ before incubating with 5 μg/ml DNase I for 1hr on ice. Then an equal volume of STE buffer (250 mM sucrose, 20 mM Tris pH 8.0, 2 mM EDTA pH 8.0) was added and the material was centrifuged at 15,000 g, 10 min, 4°C. Pellets enriched with the mt membrane vesicles were washed in STE and kept at -70°C.

SDS-PAGE and Western blot

Protein samples were separated on SDS-PAGE, blotted onto a PVDF membrane (PALL) and probed with the appropriate monoclonal (mAb) or polyclonal (pAb) antibody. This was followed by incubation with a secondary HRP-conjugated anti-rabbit or anti-mouse antibody (1:2000, BioRad). Proteins were visualized using the Pierce ECL system (Genetica/Biorad) on a ChemiDoc instrument (BioRad). When needed, membranes were stripped at 50°C for 30 min in a stripping buffer (62.5 mM Tris pH 6.8, 100 mM mercapthoethanol, 2% SDS) and re-probed. The PageRuler prestained protein standard (Fermentas) was used to determine the size of detected bands. Primary antibodies used in this study were: mAb anti-v5 epitope tag (1:2000, Invitrogen), mAb anti-mtHsp70 (1:2000) [66], pAb anti-AAC (1:2000) [67], pAb anti-MRP1 (1:1000) [68], pAb anti-APRT (1:1000) and pAb anti-enolase (1:1000) [69]. Antibodies against ATPaseTb2 (1:1000), ATPaseTb1 (1:1000), subunit β (1:2000) and subunit p18 (1:1000) were prepared for the purpose of this study. The open reading frames of the respective genes without their predicted mt localization signal were cloned into the E. coli expression plasmid, pSKB3. The proteins were overexpressed in BL21 E. coli cells and purified under native conditions (subunit β) or denatured conditions (ATPaseTb2, ATPaseTb1 and p18) using a 6 M guanidinium lysis buffer and 8 M urea binding buffer. The denatured proteins were then refolded with a step-wise dialysis procedure that included 0.5 M arginine in the refolding buffer. Native and refolded antigens were sent to Pineda (Antikörper-Service, Germany) or Davids Biotechnologie (Regensburg, Germany) for polyclonal antibody production.

Mt membrane potential (Δψ_m) measurement

The Δψ_m was determined utilizing the red-fluorescent stain Mitotracker Red CMX-Ros (Invitrogen). Cells in the exponential growth phase were stained with 100 nM of the dye for 30 min at 37°C. Cells were pelleted (1,300 g, 10 min, RT), resuspended in 2 ml of PBS (pH 7.4) and immediately analyzed by flow cytometry (BD FACS Canto II Instrument). For each sample, 10,000 events were collected. Treatment with the protonophore FCCP (20 μM) was used as a control for mt membrane depolarization. Data were evaluated using BD FACSDiva (BD Company) software.

F₁- ATPase assay

ATPase activity was measured with the Sumner assay, which is based on the release of free phosphate when ATP is hydrolyzed by the enzyme as described earlier [14,70]. Briefly, crude mt lysates were obtained from 2x10⁸cells by SoTe/digitonin extraction (0.015% digitonin, 0.6 M Sorbitol, 2 mM EDTA, 20 mM Tris-HCl pH 7.5). Mt pellets were resuspended in an assay buffer (200 mM KCl, 2 mM MgCl₂, Tris-HCl pH 8.0) and the 20 min reaction was initiated by the addition of ATP to a final concentration of 5 mM. Where indicated, samples were pre-treated with the F₁-ATPase specific inhibitor, sodium azide (2 mM) for 10 min at 37°C. The 100 μl enzymatic reactions were deproteinated by the addition of 1.9 μl of 70% perchloric acid. After a 30 min incubation on ice, the samples were spun down (16,000g, 10 min, 4°C) and 90 μl of the supernatant was incubated for 10 min with 0,5 ml of Sumner reagent (8.8% FeSO₄.7H₂O, 375 mM H₂SO₄, 6.6% (NH₄)Mo₇O₂₄.4H₂O)[62]. 200 μl was then transferred to a 96 well plate and the absorbance was measured at 610 nm using a Tecan Infinite plate reader (Infinite M200Pro, Tecan). To calibrate the assay, a standard curve was calculated from the absorbance values of linear inorganic phosphate samples (0–2 mM).

In-gel histochemical staining of F₁-ATPase activity

Blue native PAGE (BNE) of mt lysates, followed by in-gel activity staining was adapted from published protocols [26,32]. Briefly, mt vesicles from ∼2x10⁹ cells were resuspended in a mt lysis buffer (0,75 M amino-n-caproic acid—ACA, 50 mM Bis-Tris, 0,5 mM EDTA, pH 7.0), lysed with 2% dodecylmaltoside (DDM) for 1hr on ice and then cleared by centrifugation (16,000g for 30 min, 4°C). The protein concentration of each mt lysate was determined by a Bradford assay (BioRad), so that 100 μg of total mt protein could be mixed with 1 M ACA and 5% Coomassie brilliant blue G-250 before being loaded on a 2–12% Bis-Tris BNE gel. Immediately after the run (3 hr, 100 V, 4°C), the gel was incubated overnight in an ATPase reaction buffer (35 mM Tris pH 8.0, 270 mM glycine, 19mM MgSO4, 0.3% Pb(NO₃)₂, 11 mM ATP).

High resolution clear native PAGE (hrCNE)

The protocol for hrCNE was adapted from published studies [71,72]. In summary, crude mt vesicles from ∼5x10⁸ cells were resuspended in a mt lysis buffer (2 mM ACA, 50 mM Imidazole-HCl, 1 mM EDTA, 50 mM NaCl, pH 7) and lysed for one hour on ice with 4 mg digitonin/1 mg protein. The samples were spun down at 16,000 g for 30 min and the cleared lysate protein concentrations were determined by a Bradford assay. Samples were mixed with a 5x loading dye (0.1% Ponceau-S, 50% glycerol) and loaded onto a 3%-12% native gradient gel. After electrophoresis (3 hr, 100 V, 4°C), the resolved mt lysates were transferred onto a nitrocellulose membrane (overnight, 20 V, 4°C) and probed with selected antibodies.

Na₂CO₃ submitochondrial fractionation

Na₂CO₃ extraction of mt membranes was adapted from a previously published protocol [73]. Mt vesicles from 1x10⁹ cells were isolated by hypotonic lysis as described above. The resulting supernatant from the 25G needle homogenization step was kept as a cytosolic fraction (CYTO). The mt pellet was further treated with digitonin (80 μg/ml) for 15 min on ice to disrupt the mt outer membrane. The material was then cleared by centrifugation (12,000 g, 20 min, 4°C) and the pelleted mitoplasts were resuspended in 0.1 M Na₂CO₃ buffer (pH 11.5) and incubated for 30 min on ice. A final ultracentrifugation step (100,000 g, 4°C for 1 hr) performed in an SW50Ti rotor of a Beckman Instrument resulted in a supernatant comprised of proteins from the mt matrix (MX), including stripped peripheral membrane proteins, and a pellet containing integral proteins isolated from the mt membrane fraction (M).

Glycerol gradient (GG) sedimentation

Hypotonically purified mt vesicles from ∼2.5x10⁹ cells were resuspended in a GG lysis buffer (10 mM Tris, pH 7.2, 10 mM MgCl₂, 200 mM KCl, 1mM DTT) and lysed with 1% Triton X-100 (30 min, on ice). The lysates were cleared by a centrifugation step (2x 16,000 g, 30 min, 4°C) and the protein concentration was determined by a Bradford assay. Mt cleared lysates were resolved by ultracentrifugation (Beckman Instrument, SW40 rotor) at 38,000 g for 5 hr on 11ml 10–30% GG, which was poured manually or using the Gradient Station (Biocomp) according to the manufacturer’s protocol. The glycerol gradients were then fractionated either manually or with the Gradient station and 500μl fractions were stored at -70°C.

Immunogold staining of ultrathin sections and transmission electron microscopy

Cells (∼5x10⁷) were pelleted (1300 g, 10 min, RT), washed in PBS (pH 7.4) and immediately fixed in a 4% formaldehyde/ 0.1 M phosphate buffer. Samples were then dehydrated at -10°C through a series of seven steps that increased the concentration of ethanol from 30% to100%, pausing at each step for 1 hour. Next, the samples were embedded in LR White Resin (Electron Microscopy Sciences) and polymerized (UV light, 48 hours at -10°C). Ultrathin sections were prepared by the Ultracut UCT ultramicrotome (Leica) and mounted on nickel grids. Prepared sections were blocked in 5% BSA, labelled with primary anti-β antibody (1:10 dilution), washed three times with PBS and incubated with protein A conjugated to 10 nm gold particles (1:100 dilution, Aurion). The immunogold labelled grids were contrasted, carbon coated and examined by the TEM (JEM-1010, Jeol). Grids, which were immunolabeled with only the protein A conjugated to 10nM gold particles, were used as a negative control.

Quantification and statistical analysis of immunogold labelling

The number of gold particles was statistically evaluated as described earlier [12,74]. Using ImageJ software (NIH, USA), a grid consisting of squares (test points, P) with constant size (16,105 μm²) was superimposed randomly on the electron micrographs of identified mitochondria. All test squares and immunogold particles within the immediate proximity to the mt membrane were counted separately for each micrograph. In order to statistically evaluate the labelling, all observed gold particles (N_obs) and all test points (P) from 113 images captured from NON, IND3 and IND5 samples were tallied. Expected numbers of gold particles (N_exp) for each sample were calculated as (total sum N_obs x P)/total sum P. To determine if the difference between the observed and expected particles was significant and not due to random fluctuations, a Chi-squared analysis χ² = (N_obs—N_exp)²/ N_exp was performed using GraphPad QuickCalcs calculator (www.graphpad.com/quickcalcs). In addition, the relative labeling index (RLI), where the predicted RLI = 1 for random labelling, was calculated as RLI = N_obs/ N_exp.

Gene IDs for the genes and proteins mentioned in this study

ATPaseTb2 (Tb927.5.2930), ATPaseTb1 (Tb927.10.520), TbAAC (Tb927.10.14820/30/40), p18 (Tb927.5.1710), mtHsp70 (Tb927.6.3740), MRP1 (Tb927.11.1710), APRT (Tb927.7.1780), enolase (Tb427.10.2890), subunit β (Tb927.3.1380).

Supporting Information

Attachment 1

Attachment 2

Attachment 3