The Mitochondrial Ca2+ Uniporter Complex (MCUC) of Trypanosoma brucei Is a Hetero-oligomer That Contains Novel Subunits Essential for Ca2+ Uptake

Trypanosoma brucei causes human African trypanosomiasis and nagana in animals. The finding of a mitochondrial calcium uniporter (MCU) conserved in this parasite was essential for the discovery of the gene encoding the pore subunit. Mitochondrial Ca2+ transport mediated by the MUC complex is critical in Trypanosoma brucei for shaping the dynamics of cytosolic Ca2+ increases, for the bioenergetics of the cells, and for viability and infectivity. We found that one component of the complex (MCUb) does not act as a dominant negative effector of the channel as in vertebrate cells and that the TbMCUC possesses two unique subunits (MCUc and MCUd) present only in trypanosomatids and required for Ca2+ transport. The study of the interactions between these four subunits (MCU, MCUb, MCUc, and MCUd) by a variety of techniques that include coimmunoprecipitation, split-ubiquitin membrane-based yeast two-hybrid assays, and site-directed mutagenesis suggests that they interact through their transmembrane helices to form hetero-oligomers.

T he Trypanosoma brucei group of parasites causes nagana in cattle and African trypanosomiasis or sleeping sickness in humans. Two of the best-studied life cycle stages of T. brucei are the procyclic form (PCF), which is found in the tsetse fly vector, and the bloodstream form (BSF), which is present in the blood of infected animal hosts. Although both stages have a single mitochondrion, the PCF mitochondrion has a respiratory chain, while the BSF mitochondrion does not possess a functional respiratory chain or oxidative phosphorylation. BSF trypanosomes rely on the reverse action of the ATP synthase to maintain a mitochondrial membrane potential (Δ⌿ m ) (1)(2)(3)(4), which is required for protein (5) and Ca 2ϩ (2) uptake. Both stages have a functional mitochondrial Ca 2ϩ uniporter (MCU) (6,7), which is essential for growth and infectivity (7).
The finding of a MCU in trypanosomatids with characteristics similar to that present in mammalian cells (8,9) was important for the discovery of the molecular nature of a modulator of the channel, mitochondrial calcium uptake 1 (MICU1) (10), and the pore subunit of the uniporter or MCU (11)(12)(13). After this significant discovery, other subunits of the MCU complex (MCUC), such as MCU regulator 1 (MCUR1) (14), MICU2 and MICU3 (15), MCUb (16), and essential MCU regulator (EMRE) (17), were described in mammals. Trypanosomatids lack orthologs to EMRE, MCUR1, and MICU3 (18). Current models of the metazoan uniporter indicate that MCU spans the mitochondrial inner membrane forming the pore and is surrounded by the other regulatory subunits as a large complex (11,17). Interestingly, the recombinant pore domains of MCU from Caenorhabditis elegans (19) form homo-pentamers, while those from several fungi, like Neurospora crassa (20), Neosartorya fischeri (21), Metarhizium acridum (22), Fusarium graminearum (22), and Cyphellophora europaea (23), and those from zebra fish (23) form homotetramers in vitro. However, which is the oligomeric state in vivo and how MCU interacts with its membrane partners MCUb and EMRE remain to be investigated.
In this work, we report that T. brucei MCUC (TbMCUC) has two additional subunits not found in mammalian cells that we have named TbMCUc and TbMCUd. These subunits are essential for mitochondrial Ca 2ϩ uptake, and their physical interaction with the other subunits (TbMCU and TbMCUb) of the MCUC was studied by coimmunoprecipitation and split-ubiquitin membrane-based yeast two-hybrid (MYTH) assays. The results suggest that the MCUC is a hetero-oligomer which is composed of four distinct pore-forming and Ca 2ϩ -conducting subunits (TbMCU, TbMCUb, TbMCUc, and TbM-CUd). This is the first report in which the oligomeric state of a hetero-MCU complex has been defined using MYTH technology.

RESULTS
Identification of novel subunits of the TbMCUC. To screen for genes encoding TbMCU orthologs, the amino acid sequence of TbMCU (TriTryp Database accession number Tb427tmp.47.0014) was used to search TriTrypDB using BLASTp. This search yielded three putative proteins, each with two transmembrane domains, that we designated TbMCUb (Tb427. 10.300) (18), TbMCUc (Tb427tmp.02.1760), and TbMCUd (Tb427. 10.2150), which have high identity to TbMCU (25 to 31% identity in 62 to 191 amino acids). The open reading frames predict proteins of 254, 249, and 214 amino acids, with apparent molecular weights of 28.4, 27.8, and 24.7 kDa, respectively, for TbMCUb, TbMCUc, and TbMCUd. Interestingly, TbMCUc and TbMCUd, along with TbMCUb, exhibited 16 to 19% overall identity and 28 to 34% similarity with TbMCU and contained most of the conserved domains of TbMCU, including two transmembrane domains and one modified putative Ca 2ϩ selectivity filter, and belong to the MCU family (Pfam: PF04678). Most of them contain a putative mitochondrial targeting signal (MTS), with the exception of TbMCUb, and some have coiled-coil motifs (Fig. 1A). TbMCUb does not contain a typical N-terminal MTS, as predicted by MitoProt ( Fig. 1A and see Fig. S1A in the supplemental material), but it has a possible cleavage site for a signal sequence between amino acids 51 and 52, as predicted by PSORT II. To search for any MCU orthologs in other organisms, the amino acid sequences of TbMCU, TbMCUb, TbMCUc, TbMCUd, and human MCU and MCUb were used to search TriT-rypDB and GenBank using BLASTp (iterative PSI-BLASTp). TbMCU and TbMCUb orthologs are widely distributed among most eukaryotes, but those of TbMCUc and TbMCUd are found only in trypanosomatids ( Fig. 1C and Fig. S1B), including the free-living trypanosomatid Bodo saltans, which has orthologs to MCU (BS46740), MCUb (BS74080), and MCUc (BS20060) but not to MCUd. Significantly, the WDXXEPXTY motif found in the pore region of eukaryotic MCU and MCUb was WNXXEPXTY in both MCUc and MCUd in almost all trypanosomatid species examined ( Fig. 1B and Fig. S1B). The serine (S) residue responsible for the sensitivity to Ru360 in the pore region of mammalian MCUs (11) was changed to D/G in the trypanosomatid MCU paralogs ( Fig. 1B and Fig. S1B). The critical conserved substitution R to W between MCU and MCUb near the pore of mammalian MCUb (16) was instead R to Y in trypanosomatid MCUb orthologs (Fig. S1B). Additionally, 11 of 28 MCU orthologs from trypanosomatids did not have a typical N-terminal MTS, as predicted by MitoProt (Fig. S1A).
Localization of TbMCUb, TbMCUc, and TbMCUd to the mitochondria of PCF and BSF trypanosomes. To investigate the localization of the novel MCUC subunits, the C terminus of each protein was tagged in PCF trypanosomes with a hemagglutinin (HA) tag using homologous recombination with the endogenous gene locus. However, we were unable to detect the proteins by Western blot or immunofluorescence analyses, probably as a result of their low expression. We then generated cell lines overexpressing each gene tagged with HA, as described in Materials and Methods. All these proteins colocalized with MitoTracker to the mitochondria of PCF trypanosomes ( Fig. S2A to C). Western blot analysis detected bands of the expected size in the overexpressing cell lines that were not visible in the absence of tetracycline induction (Fig. S2D). To confirm the localization of these proteins, we modified the pMOTag vectors for in situ epitope tagging (24) and tagged the C terminus of each endogenous protein with a high-performance tag (spaghetti monster fluorescent protein [smFP]) with FLAG, HA, or V5 epitope tags (Fig. S2E) that has recently been described to enhance the detection of weakly expressed proteins (25)(26)(27). Figure 1D to I show that all the proteins, along with TbMCU ( Fig. S2F), colocalize with MitoTracker (MT) to the mitochondria of PCF (Fig. 1D to F) and BSF ( Fig. 1G to I) trypanosomes. Western blot analyses confirmed the expression of the proteins with the expected apparent molecular weights ( Fig. S2G and Fig. 2G and H).
Colocalization and coimmunoprecipitation of TbMCU, TbMCUb, TbMCUc, and TbMCUd. To investigate whether TbMCU can form hetero-oligomers in T. brucei, we generated a single PCF transgenic cell line in which TbMCUb, TbMCUc, and TbMCUd were endogenously C-terminally tagged with smFLAG, smHA, and smV5, respectively. Immunofluorescence analyses of the triple-smFP-tagged transgenic cell line revealed that TbMCUb, TbMCUc, and TbMCUd colocalized with the TbMCU to the mitochondria of PCF trypanosomes, using anti-TbMCU (␣-TbMCU), ␣-FLAG, ␣-HA and/or ␣-V5 antibodies, respectively ( Fig. 2A to F). The cells were lysed, and immunoprecipitations were done with ␣-TbMCU, ␣-FLAG, ␣-HA, or ␣-V5 antibodies, respectively. TbMCUb, TbMCUc, and TbMCUd, but not T. brucei voltage-dependent anion channel (TbVDAC) or cytochrome c 1 (TbCyt c 1 ), which are localized to the outer and inner mitochondrial membranes, respectively, were detected by protein immunoblotting after immunoprecipitation of TbMCU from the triple-smFP-tagged T. brucei PCF cell line (Fig. 2G). Similarly, TbMCU, TbMCUc, and TbMCUd were pulled down with TbMCUb using ␣-FLAG; TbMCU, TbMCUb, and TbMCUd were pulled down with TbMCUc using ␣-HA; and also TbMCU, TbMCUb, and TbMCUc were pulled down with TbMCUd using ␣-V5, but neither TbVDAC nor TbCyt c 1 was immunoprecipitated with any of the antibodies (Fig. 2H). To confirm that these TbMCUC subunits belong to the same large protein complex, crude PCF mitochondrial vesicles were isolated from the triple-smFP-tagged TbMCUC cell line and lysed with dodecylmaltoside (DDM) for blue Native PAGE (BN-PAGE) and then immunoblotted with ␣-TbMCU, ␣-FLAG, ␣-HA, or ␣-V5 antibodies, respectively. TbMCU, TbMCUb, TbMCUc, and TbMCUd migrated with the same separation pattern, with one dominant band at ϳ500 kDa and three weak bands ranging from 420 to 720 kDa (Fig. 2I), suggesting that they exist in a large protein complex with a net molecular weight of approximately 380 kDa after removal of the triple smFP tags. Hence, the results suggest that the TbMCU complex is a hetero-oligomer containing at least 4 subunits: TbMCU, TbMCUb, TbMCUc, and TbMCUd.
RNAi and overexpression of new TbMCUC subunits affect mitochondrial Ca 2؉ uptake but do not alter ⌬⌿ m . We used RNA interference (RNAi) to knock down the Immunoblot analyses were performed using antibodies against TbMCU, FLAG, HA, and V5. Antibodies against TbCyt c 1 were used as a loading control but detected on a SDS-PAGE gel. Arrowheads indicate one dominant band at ϳ500 kDa and three weak bands at 420 to 720 kDa. Multiple bands probably reflected different combinations of tagged and nontagged subunits in the complex, since only one allele of each endogenous gene was generally tagged. Markers are shown on the left, and the antibodies used in immunoblots are shown on the right. expression of each protein, which, in contrast to the results reported with TbMCU (7), did not result in growth defects in either PCF (Fig. S3A) or BSF (Fig. S3C) trypanosomes. However, a significant growth defect was observed when PCF TbMCUc and TbMCUd mutants were grown in a glucose-deficient medium (SDM-80) (31) (Fig. S3D). Relative reverse transcription-PCR (RT-PCR) and ImageJ analyses showed that the mRNA was downregulated by 52 to 79% after 4 days of tetracycline addition to PCF trypanosomes (Fig. 3A). Similar results were obtained using BSF trypanosomes (Fig. 3C). To confirm the protein downregulation by RNAi, we tagged each of the novel TbMCUC subunits at the C terminus of each endogenous protein with a high-performance smFP epitope using the corresponding TbMCUb, TbMCUc, or TbMCUd gene RNAi cell line of PCF and BSF trypanosomes. Western blot analyses revealed a correlative decrease by 76 to 91% of TbMCUc and TbMCUd in both PCF (Fig. 3B) and BSF (Fig. 3D) trypanosomes. However, TbMCUb had only an 18% reduction (Fig. 3D). Further phenotypic analyses were done after 2 days and 4 days of growth for BSF and PCF trypanosomes, respectively.
To investigate the ability of the knockdown cell lines to take up Ca 2ϩ , we monitored Ca 2ϩ uptake with calcium green-5N in digitonin-permeabilized PCF and BSF trypano- somes. A decrease in fluorescence indicates decreasing medium Ca 2ϩ or increasing vesicular Ca 2ϩ . Figure 3 shows that addition of 50 M digitonin in the presence of 5 mM succinate in the case of PCF trypanosomes ( Fig. 3E and F) or 1 mM ATP in the case of BSF trypanosomes ( Fig. 3G and H) and 20 M Ca 2ϩ produced a fast decrease in Ca 2ϩ concentration starting after a period of about 30 s. This Ca 2ϩ uptake activity was fully eliminated by the addition of 16 M ruthenium red, indicating that it is due to the uniporter. Knockdown of TbMCUc and TbMCUd significantly decreased the mitochondria's ability to take up Ca 2ϩ in both PCF and BSF trypanosomes ( Fig. 3E to H), while knockdown of TbMCUb only significantly decreased uptake in PCF trypanosomes, which is consistent with the weak effect of RNAi on protein expression in BSF trypanosomes (Fig. 3D).
The mitochondria of permeabilized control PCF trypanosomes were capable of buffering multiple pulses of exogenously added Ca 2ϩ , and overexpression of the TbMCUc or TbMCUd gene increased significantly the ability of their mitochondria to accumulate Ca 2ϩ in response to Ca 2ϩ pulses (Fig. 4). The lack of a dominant negative effect of the overexpressed TbMCUb gene (Fig. 4A and 4D) is consistent with results reported for T. cruzi (28) and different from those reported for HeLa cells (16).
To investigate whether down-or upregulated expression of the new TbMCUC subunits affects Δ⌿ m , we used Safranin O to measure Δ⌿ m in digitonin-permeabilized PCF trypanosomes in the presence of succinate as the mitochondrial substrate. When Safranin O was used, an increase in fluorescence after addition of digitonin indicates stacking of the dye to the energized inner mitochondrial membrane ( Fig. 3I and 4E). Addition of ADP induced a small decrease in membrane potential, indicating ADP phosphorylation. Δ⌿ m returned to its initial level after addition of the mitochondrial ADP/ATP carrier inhibitor carboxyatractyloside (CAT) (Fig. 3I and Fig. 4E). Ca 2ϩ uptake was also observed in BSF in the presence of ATP (Fig. 3K). Addition of the uncoupler carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) in PCF ( Fig. 3I and 4E) or the ATP synthase inhibitor oligomycin in BSF (Fig. 3K) collapsed the membrane potential. Knockdown ( Fig. S3A and C) or overexpression (Fig. S3B) of the TbMCUb, TbMCUc, or TbMCUd gene, which does not affect growth, altered only mitochondrial Ca 2ϩ uptake but did not affect the Δ⌿ m at the steady state or ADP phosphorylation ( Fig. 3I and J and Fig. 4E and F). Collectively, the newly identified TbMCUC subunits have the same properties of TbMCU (7) and other characterized eukaryotic MCUs (11,12) upon Ca 2ϩ uptake and Δ⌿ m .
Physical direct interactions between TbMCUC subunits. To better understand the possible organization of TbMCUC subunits in the hetero-oligomeric complex of T. brucei, we used the split-ubiquitin membrane-based yeast two-hybrid (MYTH) assays (29) (Fig. S4A) to determine the physical direct interactions between TbMCUC subunits in Saccharomyces cerevisiae. The split-ubiquitin system allows detection of in vivo interaction between membrane proteins that have their N and/or C terminus located in the cytosol (Fig. S4A). The membrane topology of TbMCU, TbMCUb, TbMCUc, and TbMCUd predicted with Protter showed that these membrane proteins could be localized to the yeast plasma membrane with both their N and C termini facing the cytosol (Fig. S5A). In the MYTH assays, TbMCU, TbMCUb, TbMCUc, or TbMCUd (the bait, without the mitochondrial targeting signal [MTS]) was fused to the C-terminal half of ubiquitin (C ub ) and the artificial transcription factor LexA-VP16 (TF). TbMCU, TbMCUb, TbMCUc, or TbMCUd (the prey, without the MTS) was fused to the mutated half of ubiquitin (N ub G), and the interaction of the protein partners was monitored by the release of the TF, which translocates to the nucleus, where it binds to LexA operators situated upstream of reporter genes (HIS3, ADE2, lacZ) via its Lex DNA binding domain. The reporter genes enable the yeast to grow on defined media lacking histidine or/and adenine, while lacZ encodes the enzyme ␤-galactosidase (␤-Gal), resulting in the growth of yeast in selective medium and color development in ␤-galactosidase assays. Figure 5F showed that each of the TbMCUC subunits was expressed as a bait, with the yeast beta-fructofuranosidase or invertase (SUCrose 2 [SUC]) signal sequence instead of the MTS targeted correctly to the yeast plasma membrane. The yeast reporter strain expressing the bait TbMCU, TbMCUb, TbMCUc, or TbMCUd alone or with the empty prey vector did not grow on the selective synthetic dropout (SD) medium plates (SD medium with a triple dropout [SD-3DO], SD-4DO, and SD-4DO plus X-Gal [5-bromo-4chloro-3-indolyl-␤-D-galactopyranoside]) (Fig. 5A), indicating that the baits were not self-activated. The strain expressing TbMCU as a bait and TbMCUc or TbMCUd as a prey enabled growth on the high-stringency selective SD-4DO plates and had high ␤-galactosidase activities ( Fig. 5A and B), suggesting that TbMCU interacted strongly with TbMCUc and TbMCUd, respectively. Similarly, MYTH screens identified that TbM-CUb interacted strongly with TbMCUc and TbMCUd and that TbMCUc or TbMCUd also strongly interacted with itself ( Fig. 5A and B). However, MYTH screens showed that TbMCU interacted weakly with TbMCUb, that TbMCUc interacted weakly with TbMCUd, and that TbMCU or TbMCUb interacted weakly with itself ( Fig. 5A and B). Expression of each of the baits or the bait-prey pairs in yeast was confirmed by Western blot analyses using antitag antibodies from the MYTH expression vectors pBT3-SUC and pPR3N (29), i.e., antibodies ␣-VP16 for the bait and antibodies ␣-HA for the prey (Fig. 5C). To eliminate false-positive interactions, the specific interactions among the TbMCUC subunits were confirmed by bait-prey swapping ( Fig. 5A and B), coimmunoprecipitations ( Fig. 5D and E), and immunofluorescence subcellular colocalization ( Fig. S4C and  D). The interaction strength of TbMCUC subunits in yeast may reflect the relative proximity of the subunits in a large protein complex; thus, our data suggest a heterohexameric model for the putative organization and composition of the TbMCU complex in T. brucei (Fig. S4B).

TMHs are determinant of the interactions between TbMCUC subunits.
To identify specific interacting domains or motifs that mediate the interactions between TbMCUC subunits, truncated or substitution TbMCU mutants (Fig. 6A) were generated and expressed as baits for MYTH assays ( Fig. 6B and C). Deletion of the N-and/or C-terminal regions of TbMCU, designated TbMCUΔ1, TbMCUΔ2, and TbMCUΔ3 ( Fig. 6A and Fig. S5B), did not affect the interaction with the subunit TbMCUc or TbMCUd ( Fig. 6B and C), suggesting that the regions flanking the transmembrane helices (TMHs) of TbMCU are not involved in the protein-protein interactions. In contrast, TbMCU mutations of the conserved residues in TMH1 (Q213A, V216F, I217F, and F222A) or in TMH2 (Y235A, F236A, T241E, and Y248A) (Fig. S5C), named TbMCUΔ4 and TbMCUΔ5 (Fig. 6A), reduced the interaction with TbMCUc or TbMCUd ( Fig. 6B and C), indicating that these residues are important for the TbMCU association with TbMCUc or TbMCUd. To confirm the critical role of the TMHs of TbMCU in the protein-protein interaction, TMH1, TMH2, or both TMHs of TbMCU were replaced with an artificial transmembrane "WALP" (19 amino acid residue long peptide) helix (GWWLALALALALALALWWA) (30) to generate the mutants TbMCUΔ6, TbMCUΔ7, and TbMCUΔ8 (Fig. 6A). The topology of these mutants did not change, as predicted with Protter (data not shown). Strikingly, the substitution of two TMHs did not alter their plasma membrane localization in yeast ( Fig. S6B) but significantly disrupted their interaction with TbMCUc or TbMCUd ( Fig. 6B and C). These results suggest that the TMHs of TbMCU are essential for the proteinprotein interactions. Expression of the mutant proteins in yeast lysates was confirmed by Western blot analyses using ␣-VP16 and ␣-HA antibodies to detect the baits and preys, respectively ( Fig. 6D and E).
To determine whether the regions flanking the TMHs of TbMCUb, TbMCUc, and TbMCUd are involved in the protein-protein interactions, we generated the truncated mutations TbMCUbΔ1, TbMCUcΔ1, and TbMCUdΔ1 ( Fig. 6F and Fig. S5B), which, like TbMCUΔ3, contained only the two conserved TMHs and pore regions of these proteins. MYTH assays revealed that the mutations, like that of TbMCUΔ3, did not affect the protein-protein interactions ( Fig. 6G and H). The interactions among these TbMCUC truncation mutations were confirmed by coimmunoprecipitation ( Fig. 6K and L) and immunofluorescence subcellular colocalization ( Fig. 6I and F). On the other hand, to validate the important role of TMHs in the interactions, we replaced the TMH1 and/or TMH2 of TbMCUb, TbMCUc, and TbMCUd with the artificial transmembrane WALP helices to generate the mutations TbMCUbΔ2/Δ3/Δ4, TbMCUcΔ2/Δ3/Δ4, and TbMCUdΔ2/Δ3/Δ4 (Fig. 6F). Interestingly, the replacement of the TMHs of TbMCUb, TbMCUc, or TbMCUd by the WALP helices significantly disrupted the interactions (Fig. 6G and H), while the expression and localization of these proteins to the yeast plasma membrane were not altered (Fig. S6C to E). Like TbMCUΔ8 (Fig. 6B and C), the The scheme depicts wild-type TbMCU and truncated and substitution mutant constructs. Coil, coiled-coil domain; TM1 and TM2 (black rectangles), transmembrane domains 1 and 2; M (black ellipse), the conserved WDXXEPXTY motif; Δ, truncated or substitution mutants; WALP, artificial TM sequence GWWLALALALALALALWWA. Substitutions of the conserved residues of TMH1 or TMH2 of TbMCU are indicated (multiple substitutions were generated, because single substitutions did not significantly alter protein-protein interaction). (B) Growth assay of the yeast NMY51 strain expressing the bait (TbMCU or TbMCUΔ1 to -Δ8) together with the prey (TbMCUc or TbMCUd) on SD selection agar plates as described for Fig. 5A. (C) A quantitative ␤-Gal activity assay of strain NMY51 coexpressing the bait-prey pairs as described for panel B determines their interaction strength. Each column represents the mean Ϯ standard deviation (n ϭ 3; 8 colonies for each independent experiment). (D and E) Expression level of each bait or prey as determined by immunoblot analysis using antitag antibodies, VP16 for the bait and HA for the prey, and hexokinase (HK) antibodies used as a loading control. Lanes M, molecular markers. (F) The scheme depicts wild-type TbMCUb, TbMCUc, TbMCUd, and their truncated or substitution mutant constructs as described for panel A. (G) Growth assay of the yeast strain NMY51 expressing the mutated bait (TbMCUΔ3, TbMCUΔ6-Δ8, TbMCUbΔ2-Δ4 TbMCUcΔ2-Δ4, or TbMCUdΔ2-Δ4) together with the mutated prey (TbMCUΔ3, TbMCUΔ6-Δ8, TbMCUbΔ2-Δ4 TbMCUcΔ2-Δ4, or TbMCUdΔ2-Δ4) on SD selection agar plates as in Fig. 5A. (H) Quantitative ␤-Gal activity assay of the strain NMY51 coexpressing the TbMCUC mutant bait-prey pairs to determine their interaction strength. Each column represents a mean Ϯ standard deviation (n ϭ 3; 8 colonies were used in each independent experiment). (I and J) Immunofluorescence validation of colocalization of the bait-prey interaction pairs, as in panels G and H, to the yeast plasma membrane. The PCC of the bait-prey pairs TbMCUΔ3-TbMCUcΔ1, TbMCUΔ3-TbMCUdΔ1, TbMCUcΔ1-TbMCUcΔ1, (Continued on next page) Novel Hetero-oligomeric MCUC replacement of both TMHs of these proteins (TbMCUb4, TbMCUcΔ4, and TbMCUdΔ4) with the WALP helices abolished the protein-protein interactions (Fig. 6G and H). These studies therefore suggest that the TMH-TMH interactions of TbMCUC subunits, particularly the TMH2-TMH2 interaction, play a pivotal role in the oligomerization of TbMCUC subunits that form a hetero-hexameric MCU complex in vivo. Expression of the mutant proteins in yeast lysates was confirmed by Western blot analyses using ␣-VP16 and ␣-HA antibodies to detect the baits and preys, respectively (Fig. S6A).

DISCUSSION
This study has discovered two novel pore-forming subunits (TbMCUc and TbMCUd) of the MCUC of T. brucei that are essential for mitochondrial Ca 2ϩ uptake. We also defined their direct physical interactions with other subunits (TbMCU and TbMCUb), which suggest the formation of a hetero-hexameric MCU complex.
These two new paralogs of MCU are absent in the mammalian host, and examination of GenBank databases found them only in trypanosomatids. Both TbMCUc and TbMCUd have mitochondrial targeting signals, two transmembrane domains, and a modified Ca 2ϩ selectivity filter (conserved WDXXEPXTY motif). Downregulation of their expression by RNAi significantly decreased while their overexpression significantly increased mitochondrial Ca 2ϩ uptake without affecting cell growth in rich media or the mitochondrial membrane potential.
In agreement with results reported for T. cruzi (28), but in contrast to results described for HeLa cells (16), the T. brucei ortholog of mammalian MCUb (TbMCUb) did not have a dominant negative effect on mitochondrial Ca 2ϩ uptake when overexpressed, and it reduced instead of increased mitochondrial Ca 2ϩ uptake when its expression was downregulated. TbMCUb does not have a typical MTS, although it has a possible cleavage site for a signal sequence between amino acids 51 and 52, and our in situ tagging and overexpression studies localized it to the mitochondrion. The lack of a typical MTS also occurs in other subunits of the MCU complex in other trypanosomatids (Fig. S1A). Protein expression of TbMCUb could only weakly be downregulated by RNAi in BSF trypanosomes, explaining its weak inhibitory effect on mitochondrial Ca 2ϩ uptake in the mutants. As happens with TbMCUc and TbMCUd, downregulation or overexpression of TbMCUb did not affect growth in rich media or the mitochondrial membrane potential. However, growth was significantly affected when TbMCUc and TbMCUd mutants were grown in glucose-deficient media, in agreement with the relevance of mitochondrial metabolism in PCF trypanosomes under low-glucose conditions (31).
All four subunits involved in mitochondrial Ca 2ϩ uptake, TbMCUb, TbMCUc, TbM-CUd, and TbMCU, colocalize with MitoTracker to the mitochondria of PCF and BSF trypanosomes. They interact with each other, as revealed by the formation of large protein complexes detected by blue native gel separation of mitochondrial proteins and labeling with antibodies against all the subunits and by coimmunoprecipitation of the subunits. To confirm these interactions, we used the MYTH technology. Although this technique detects mostly binary interactions between integral membrane proteins, by transforming yeast with genes encoding two subunits at a time, we were able to demonstrate the strength of their interactions, providing proof of concept for a new application of this technique. There were strong interactions of TbMCU with TbMCUc and TbMCUd, of TbMCUb with TbMCUc and TbMCUd, and of TbMCUs and TbMCUd with themselves, and there were weak interactions of TbMCU with T. cruzi MCUb (TcMCUb) or of TbMCUc with TbMCUd. The results suggest the formation of a heterohexameric complex, as shown in Fig. S4B. Our approach could be adopted to study other oligomeric complexes. It is interesting to note that previous studies with recom- binant MCU from Caenorhabditis elegans showed that the purified protein with its amino-terminal domain (NTD) deleted formed homo-pentamers in vitro (19).
We could also obtain evidence that TM helix 1 (TMH1) and especially 2 (TMH2) of each of the four subunits of the complex are important for their interaction. Deletions of the C-and N-terminal regions of the proteins did not affect their membrane localization or their protein-protein interactions, while mutations in TbMCU TMH1 and TMH2 or their replacement with an artificial WALP helix (30) in any of the subunits greatly decreased their interactions without affecting their membrane localization. It is interesting to note that the N-terminal domain of the C. elegans MCU was shown to be nonessential for Ca 2ϩ uptake but that TMH2 forms the inner core of the Ca 2ϩ channel (19), suggesting a dual role for this helix. Only TbMCU and TbMCUd contain coiled-coil motifs. These are ubiquitous protein domains that mediate specific homo-and heteromeric protein-protein interactions among a wide range of proteins (32). However, split-ubiquitin MYTH assays with mutants lacking these domains showed that they are not relevant for interactions between the MCU complex subunits.
MCU paralogs are weakly expressed in trypanosomes, and endogenous tagging with high-performance probes was necessary for localization and protein-protein interaction studies. These tags consist of green fluorescent protein scaffolds containing numerous copies of peptide epitopes that simultaneously bind IgG antibodies at each location (spaghetti monsters) (25)(26)(27). This method allowed the use of cells with three of the subunits tagged with different epitopes and will be useful for localization and coimmunoprecipitation studies with other oligomeric and weakly expressed proteins in trypanosomes.
In summary, we detected two novel subunits of the T. brucei MCU complex, TbMCUc and TbMCUd. The results indicate that these proteins, as well as TbMCU and TbMCUb, are essential for mitochondrial Ca 2ϩ uptake and suggest that they form a heterohexameric complex. Identification of the new components of the TbMCU complex may not only help to fully elucidate its structure but also provide new insights into understanding their evolutionary diversity.
Ca 2؉ uptake by digitonin-permeabilized T. brucei. The uptake of Ca 2ϩ by permeabilized T. brucei was assayed by fluorescence measurements at 28°C using calcium green-5N (7). For PCF, trypanosome cells were collected by centrifugation at 1,000 ϫ g for 7 min and washed twice with cold buffer A with glucose (BAG) (Text S1). PCF cells were resuspended to a final density of 1 ϫ 10 9 cells per ml in BAG and kept on ice. For BSF, trypanosome cells were collected by centrifugation at 1,000 ϫ g for 7 min and washed twice with cold separation buffer, which contained 44 mM NaCl, 55 mM D-glucose, 57 mM Na 2 HPO 4 , and 3 mM KH 2 PO 4 at pH 8.0. BSF trypanosomes were resuspended to a density of 2 ϫ 10 7 cells per ml in the separation buffer and kept on ice. Before each experiment, a 10-ml aliquot of the BSF cells was concentrated to 50 l by centrifugation at 1,600 ϫ g at room temperature for 3 min. A 50-l aliquot of PCF (5 ϫ 10 7 cells) or BSF (2 ϫ 10 8 cells) trypanosomes of the cell suspension was added to the standard reaction buffer (125 mM sucrose, 65 mM KCl, 10 mM HEPES-KOH buffer, pH 7.2, 1 mM MgCl 2 , 2.5 mM potassium phosphate; 2.45 ml) containing 1 M calcium green-5N and the reagents indicated in the figure legends. Ca 2ϩ uptake by the cells was initiated by the addition of 50 M digitonin (for PCF trypanosomes) or 40 M digitonin (for BSF trypanosomes). Fluorescence changes were monitored in an F-7000 fluorescence spectrophotometer (Hitachi), with excitation at 490 nm and emission at 525 nm. Permeabilized cells grown in the presence of tetracycline under the assay conditions used did not show any fluorescence (excitation at 390 nm and emission at 400 to 650 nm) that could be attributed to tetracycline or its complexes.
(ii) MYTH bait and prey constructs. The full-length or truncated cDNAs of the TbMCU, TbMCUb, TbMCUc, and TbMCUd genes without 5= nucleotide sequences encoding the putative mitochondrial targeting signals (MTS) were amplified from T. brucei genomic DNA by PCR using the corresponding specific forward and reverse primers (see Table S1 in the supplemental material), which were introduced SfiI sites, digested with SfiI at 50°C overnight, and then cloned in frame into SfiI-digested MYTH assay bait (pBT3-SUC) and prey (pPR3N) expression vectors (29) to generate a set of MYTH constructs (Table S2) as described in the legend of Fig. 6A and F. The double-stranded sequences of the cloned cDNA inserts that express proteins of TbMCUC subunits C-terminally fused to Cub-LexA-VP16 in pBT3-SUC or N-terminally fused to NubG-HA in pPR3N were confirmed by sequencing as indicated above. The mutated amino acid residues or artificial WALP (GWWLALALALALALALWWA) sequence(s) was introduced into or replaced the TM domains of TbMCUC subunits in the MYTH bait or prey expression vectors (as described above) by fusion PCR (37) using a Phusion site-directed mutagenesis kit, according to the manufacturer's instructions.
(iii) MYTH assays of interaction between TbMCUC baits and preys. The recombinant MYTH bait and prey plasmids (Table S2) harboring full-length, truncated, or mutated TbMCUC subunits with the empty vectors as negative controls were cotransformed into the yeast NMY51 strain by lithium acetate (LiOAc)-mediated transformation as described previously (38) and cultured successively on the dual, triple, and quadruple SD media (SD medium minus Leu and Trp, SD medium minus Leu, Trp, and His, and SD medium minus Leu, Trp, His, and Ade [shortened to SD-2DO, SD-3DO, and SD-4DO, respectively]). After incubation at 30°C for 3 to 4 days, colonies grown on the selective SD medium plates were further screened by cultivating them on SD-4DO-5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-Gal) medium, and ␤-galactosidase (␤-Gal) activity was measured as described below to test the expression of the reporter gene lacZ. MYTH colonies were analyzed by Western blotting, immunofluorescence microscopy, and coimmunoprecipitation as described below.
(iv) ␤-Gal activity assays. To detect ␤-Gal expression by lacZ with X-Gal (39), MYTH colonies were grown onto SD-DO selection plates containing 2 mM 3-amino-1,2,4-triazole (AT), 0.8 mg/ml ␤-Gal (X-Gal), and 1ϫ buffered (BU) salts (0.7% Na 2 HPO 4 , 0.3% NaH 2 PO 4 ·H 2 O, pH 7.0) at 30°C. Blue staining was recorded for 1 to 2 days. To verify or quantify the interaction strengths of proteins, ␤-Gal activity for lacZ with the substrate of o-nitrophenyl-␤-D-galactopyranoside (ONPG) was measured by a ␤-Gal microplate plate assay using a yeast ␤-Gal assay kit according to the manufacturer's protocol. Briefly, single MYTH colonies were transferred with pipette tips to microcentrifuge tubes, each containing 150 l SD liquid medium without Leu or Trp, mixed gently with a vortex mixer to create homogeneous solutions, and incubated with shaking at 30°C. After a 30-min incubation, 100 l of each cell culture was added to a 96-well plate, and the optical density at 660 nm (OD 660 ) of the solution was determined with a SpectraMax plate reader. Subsequently, 100 l of fresh working solution containing a mixture of 1 volume of yeast protein extraction reagent (Y-per) with an equal volume of ␤-Gal assay buffer was added to each test well and mixed gently with a multichannel pipette. The reaction mixture was incubated at room temperature for approximately 30 min or until a color change was observed and then quenched by the addition of 1 M Na 2 CO 3 . The OD 420 of each well was measured with the SpectraMax, and ␤-Gal units were calculated as follows: [units ϭ 1,000 ϫ OD 420 /(time ϫ volume ϫ OD 600 )]. One unit of ␤-Gal is defined as the amount that hydrolyzes 1 mol of ONPG to o-nitrophenol and D-galactose per min per cell (40). The assay was repeated 3 times for a number of colonies (as indicated in the figures), followed by calculation of standard deviations. Statistical significance was calculated using Student's t test.
Statistical analyses. All values are expressed as means Ϯ standard deviations. Significant differences between treatments were compared using an unpaired Student t test. Differences were considered statistically significant at a P of Ͻ0.05, and n refers to the number of experiments performed. All statistical analyses were conducted using GraphPad Prism 5 (GraphPad Software, San Diego, CA).