The SCaMCs (small calcium-binding mitochondrial carriers) constitute a subfamily of mitochondrial carriers responsible for the ATP-Mg/Pi exchange with at least three paralogues in vertebrates. SCaMC members are proteins with two functional domains, the C-terminal transporter domain and the N-terminal domain which harbours calcium-binding EF-hands and faces the intermembrane space. In the present study, we have characterized a shortened fourth paralogue, SCaMC-3L (SCaMC-3-like; also named slc25a41), which lacks the calcium-binding N-terminal extension. SCaMC-3L orthologues are found exclusively in mammals, showing approx. 60% identity to the C-terminal half of SCaMC-3, its closest paralogue. In mammalian genomes, SCaMC-3 and SCaMC-3L genes are adjacent on the same chromosome, forming a head-to-tail tandem array, and show identical exon–intron boundaries, indicating that SCaMC-3L could have arisen from an SCaMC-3 ancestor by a partial duplication event which occurred prior to mammalian radiation. Expression and functional data suggest that, following the duplication event, SCaMC-3L has acquired more restrictive functions. Unlike the broadly expressed longer SCaMCs, mouse SCaMC-3L shows a limited expression pattern; it is preferentially expressed in testis and, at lower levels, in brain. SCaMC-3L transport activity was studied in yeast deficient in Sal1p, the calcium-dependent mitochondrial ATP-Mg/Pi carrier, co-expressing SCaMC-3L and mitochondrial-targeted luciferase, and it was found to perform ATP-Mg/Pi exchange, in a similar manner to Sal1p or other ATP-Mg/Pi carriers. However, metabolite transport through SCaMC-3L is calcium-independent, representing a novel mechanism involved in adenine nucleotide transport across the inner mitochondrial membrane, different to ADP/ATP translocases or long SCaMC paralogues.
The transport of metabolites, nucleotides and cofactors across the inner mitochondrial membrane is performed by structurally related proteins belonging to the MCF [MC (mitochondrial carrier) family] (reviewed in [1–3]). All MCF members have a tripartite structure comprising three repeated sequences of approx. 100 amino acids in length containing two transmembrane spanning segments and a characteristic sequence motif in matrix-facing loops . Previously, the significant sequence conservation of the MCs has permitted characterization of the repertoire of MCs in numerous genomes [3,5–9], and this suggests a common ancestor in evolution . Initial analysis of the Saccharomyces cerevisiae genome allowed identification of 35 MCF members [5,9]; subsequent studies have revealed a higher complexity in MCFs from pluricellular eukaryotes and vertebrates [3,6]. MCF complexity is mainly due to the generation of paralogues and, to a lesser extent, by the emergence of proteins with novel transport capabilities . In both processes, gene duplication has contributed as the major source for the generation of MCF functional diversity [3,10,11].
One of the largest subgroups of MCs is that formed by those involved in adenine nucleotide transport, which comprise the ADP/ATP translocases and SCaMCs (small calcium-binding MCs) (see  for a review) with four and three paralogues characterized in mammals respectively [3,12,13]. SCaMCs correspond to isoforms of the ATP-Mg/Pi carrier [12,13]. This transport was functionally identified 20 years ago and catalyses the net transport of adenine nucleotides across the inner mitochondrial membrane [14,15]. It mediates a reversible electroneutral exchange between ATP-Mg2− and HPO42− [13,16]. Likewise, SCaMCs are CaMCs (calcium-dependent MCs) (reviewed in ), a subgroup of MCs having a bipartite structure with a C-terminal half containing the MC homology domain and a long N-terminal extension harbouring EF-hand calcium-binding motifs [12,17–19], and whose transport activity is regulated by cytosolic calcium [20–23]. The complexity of SCaMCs is increased by the existence of several spliced-variants [12,17,24] and the recent identification of a shortened fourth paralogue, slc25a41 . It has been proposed that in liver mitochondria the ATP-Mg/Pi carrier regulates the matrix adenine nucleotide content , contributing to the regulation of mitochondrial activities dependent on adenine nucleotides, such as gluconeogesis, urea synthesis or mitochondrial biogenesis . In S. cerevisiae the mitochondrial ATP-Mg/Pi transporter, Sal1p, is recruited as a calcium-dependent mechanism to import ATP-Mg from the cytosol upon glucose addition to nutrient-starved yeast. It is also involved in ATP-Mg uptake from the cytosol in yeast growing exponentially in glucose, a condition in which yeast mitochondria are ATP consumers [26,27]. Sal1p is also involved, along with the ADP/ATP translocase Aac2p, in mitochondrial translation and mtDNA (mitochondrial DNA) maintenance .
In the present study, we report a comprehensive characterization of the smaller fourth SCaMC paralogue, SCaMC-3L (SCaMC-3-Like, also named slc25a41), which appears to be the result of one partial tandem duplication. We show that SCaMC-3L and its closest paralogue, SCaMC-3, display different expression patterns and genomic structures because the duplication event did not cover exons encoding the N-terminal-extension-containing calcium-binding domains. Transport assays in isolated mitochondria of yeast expressing mouse SCaMC-3L has confirmed that SCaMC-3L functions as a calcium-independent ATP-Mg/Pi exchanger, therefore providing a novel mechanism of ATP transport across the inner mitochondrial membrane different to that described for yeast Sal1p, the liver ATP-Mg/Pi carrier and the ADP/ATP translocases.
MATERIALS AND METHODS
Searching for homologous genes and the bioinformatics tools used
We have used the protein and nucleotide sequences of previously characterized human SCaMCs  to screen on-line genome databases (NCBI, Swiss-Prot and Ensembl) using the BLAST algorithms. Searches were carried out using default values with the low-complexity filter off. TBLASTN searches were performed for genomes whose assembly and annotation processes are not totally finished. Searches were performed in NCBI, and also BLAT searches in the UCSC (University of California Santa Cruz) Genome Browser website (http://genome.ucsc.edu/) were occasionally performed. We used WGS (whole-genome shotgun) assembly of the grey short-tailed opossum genome (Monodelphis domestica) (MonDom5) and from the monotreme platypus (Ornithorhynchus anatinus) (Build 1.1) (available at the NCBI). For some species whose genome annotation is still unfinished, such as rhesus monkey (Macacca mulatta, Build 1.1), or whose protein sequences were misassembled, such as bovine (Bos taurus), the identified sequences are the results of a genome-wide manual screening. Assembly of predicted sequences was undertaken using the Prophet 5.0 program. Finally, sequences were aligned by ClustalW followed by manual adjustments. To distinguish among orthologues and paralogues we performed reciprocal-best-blast hit analysis. Furthermore, orthology was verified on the basis of the syntenic relationships among genes from mammalian species. Chromosomal position and orientation of duplicated pairs and their neighbouring genes was determined using NCBI Entrez GeneView. In unfinished genomes, analysis was performed on overlapping genomic clones using the TBLASTN algorithm. Analysis of repeat elements was performed in FREP [the Functional Repeats database (http://facts.gsc.riken.go.jp/FREP)]. The SH3-Hunter program was used to find consensus signals interacting with proteins containing SH3 domains (http://cbm.bio.uniroma2.it/SH3-Hunter). Multiple sequence alignments were carried out with the ClustalW program and coloured using the Boxshade program (www.ch.embnet.org). Tree construction and bootstrap analysis were carried out with the Phylo-win program . We threaded the sequence of SCaMC-3L into the structure of ANT1 (adenine nucleotide translocator 1; PDB code 1OKC)  using the Phyre web tool (http://www.imperial.ac.uk/phyre). Secondary structure predictions were carrier out using APSSP available on the EXPASY server.
RNA isolation, RT (reverse transcription)–PCR analysis and cloning of mouse SCaMC-3L cDNA
Total RNA was extracted from mouse tissues using the acid guanidinium thiocyanate/phenol/chloroform procedure. First-strand cDNA was synthesized using 5 μg of total RNA from each tissue obtained as a template, 100 ng of random primer (Promega) and avian myeloblastosis virus reverse transcriptase (Promega). From first-strand cDNA synthesis, 5% of the yield was used as a template for subsequent PCR amplifications using TaqDNA polymerase (PerkinElmer) under standard conditions. Specific primers derived from predicted rat LOC301114 (sense r3c-5, 5′-ACTGTGCCAGGCAGATCTTGG-3′ and antisense, r3c-3, 5′-CTGCTCAGGACTTGTACACC-3′), mouse 4933406J04Rik gene (sense m3c-5, 5′-ATGGTACCCGAGCCCTATAC-3′ and antisense, m3c-3, 5′-TCTGCTCAGGACTTGTACAC-3′) and mouse SCaMC-3 (sense m3-5, 5′-TGACTCTACGCAGAACTGGC-3′ and antisense, m3-3, 5′-GACCTTCATGAAGTTGGGGG-3′) were used. For normalization, a pair of specific β-actin primers common for mouse and rat (sense, 5′-GGTATGGAATCCTGTGGCATCCATGAAA-3′ and antisense, 5′-GTGTAAAACGCAGCTCAGTAACAGTCC-3′), which amplify a 635 bp product, were used. The templates were amplified under the conditions described above. For β-actin levels, 28 cycles of amplification were carried out, and for SCaMC-3L sequences, 35 cycles of amplification were performed. PCR products were electrophoresed in 1.5% (w/v) agarose gels and stained with ethidium bromide. Also, the amplified products were cloned and sequenced using the ABI prism dye terminator cycle sequencing kit (Applied Biosystems).
The SCaMC-3L full-length cDNA was synthesized from 3 μg of total RNA of mouse testis with the Cells-to-cDNA™ II Kit (Ambion) using random primers. To obtain full-length cDNA, PCR was performed using specific primers. Primers were designed according to identified partial mouse EST (expressed sequence tag) clone BY716380 and the predicted 4933406J04Rik gene. We have designed the following two sets of PCR primers: m3c-1/m3c-2 pair [sense (containing ATG), 5′-CAACTATGGGAGTCCATCTCG-3′ and antisense, 5′-TCGGTATAGGGCTCGGGTAC-3′]; and the m3c-5/m3c-3 pair (sense, 5′-ATGGTACCCGAGCCCTATAC-3′ and antisense, 5′-TCTGCTCAGGACTTGTAC-3′). PCR fragments were cloned using pSTBlue1-blunt vector (Novagen) and sequenced. Finally, the whole cDNA of 941 bp (GenBank® accession number FM165286) was obtained by assembling those partial sequences. Rat SCaMC-3L full-length cDNA (GenBank® accession number FM165287) was also amplified from testis total RNA using r3c-1/r3c-2 primers (sense, 5′-CAACTATGGGAGTCCATCTGGAG-3′ and antisense 5′-TCTGCTCAGGACTTGTACACC-3′). The plasmid expressing mouse SCaMC-3L–FLAG was generated by using PCR to insert DNA encoding the FLAG epitope (DYKDDDDK). FLAG-epitope was added at the C-terminus immediately preceding the termination codon of mouse SCaMC-3L by PCR using primers m3c-1 and m3cFLAG (5′- ATCACTTGTCTACGTCGTCCTTGTAGTCTCTGCTCAGGACTTGTAC-3′) containing 27 nucleotides that encode the FLAG octapeptide (underlined in the sequence above) following a stop codon. The resultant cDNA was cloned into pSTBlue1 and verified by sequencing. Finally, an EcoRI fragment containing FLAG-tagged mouse SCaMC-3L was subcloned into the expression vector pCMV5 and was digested with EcoRI, to obtain pCMV5-SCaMC-3LFLAG. To carry out expression assays in yeast, this EcoRI fragment was subcloned into the EcoRI site of the centromeric pYX142 plasmid (Novagen) to generate pYX142-SCaMC-3L.
Cell culture and transfection
HEK (human embryonic kidney)-293T and COS-7 cells were cultured in DMEM (Dulbecco's modified Eagle's medium) supplemented with 5% (v/v) inactivated FBS (foetal bovine serum) at 37 °C in a 7% CO2 atmosphere. Cells were grown on coverslips and transiently transfected using the Lipofectamine™ reagent as described previously . After 16–36 h of incubation to allow expression, cells were harvested for Western blot analysis or immunofluorescence assays. For the study of mitochondrial location, living cells were incubated with 200 nM MitoTracker Red CMXRos (Molecular Probes) for 30 min at 37 °C or co-transfected with pDsRed2-Mito (Clontech). MitoTracker-loaded cells were fixed in 2% (w/v) paraformaldehyde [at room temperature (22 °C) for 4 min] and 100% methanol (at −20 °C for 3 min), washed and then used for immunofluorescence as described previously [12,18]. A monoclonal antibody against FLAG peptide, clone M2 (1:100; Sigma), was used. Fluorescence microscopy was performed using an Axioskop2 plus microscope (Zeiss) at a nominal magnification of ×63. Digital images were taken in parallel with a Coolsnap FX camera controlled via the MetaView software.
For Western blot analysis of expressed proteins, HEK-293T cells were plated in 10-cm Petri dishes and transfected with 10 μg of expression plasmids. At 36 h after transfection, cells were washed and harvested by centrifugation (500 g for 5 min at 22 °C). Mitochondria-enriched fractions were obtained as described previously . Proteins were separated by SDS/PAGE on 10% gels and the presence of FLAG-tagged proteins was determined by Western blotting using an anti-FLAG antibody (1:5000). Proteins were visualized by ECL (enhanced chemiluminescence; Amersham Biosciences). Polyclonal serum against human SCaMC-1 was used at 1:5000 .
Yeast strains and culture media
In the present study we used the W303-1B wild-type strain (MATα, ura3-1, trp1-Δ2, leu2-3,112, his3-11, ade2-1, can1-100) and the sal1Δ strain (W303; MATa, ura3-1, trp1-Δ2, leu2-3,112, his3-11, ade2-1, can1-100; sal1::kanMX4) in which SAL1, which codes for the mitochondrial ATP-Mg/Pi carrier, is disrupted by replacement of its complete coding sequence with the kanMX4 module [26,27,30]. Both contain the plasmid pYeDP-Cox4-Luciferase . The sal1Δ strain was also transformed with plasmid pYX142 or plasmid pYX142-SCaMC-3L. For the selection of Geneticin (G418; Life Technologies) resistance, cells were spread on YPD [2% (w/v) bactopeptone/1% (w/v) yeast extract/2% (w/v) glucose] plates containing 200 μg/ml G418 . For selection of pYX142 plasmid transformants, MM (0.17% yeast nitrogen base without amino acids and ammonium sulfate/0.5% ammonium sulfate/0.2% glucose as carbon source) plates without leucine were used. For isolation of mitochondria, yeast cells were precultured on MM (supplemented with the required appropriate auxotrophic requirements), diluted in YPGR [2% (w/v) bactopeptone/1% (w/v) yeast extract/2% (w/v) galactose/2% (w/v) rafinose] and grown until late exponential phase.
ATP transport assays
Mitochondria were isolated as described previously . Briefly, protoplasts were prepared by enzymatic digestion with zymolyase 100T (Seikagaku) and mitochondria were isolated by differential centrifugation after homogenization of protoplasts. The resuspension of the final protoplast pellet and all subsequent centrifugations were performed in mitochondrial buffer: 0.6 M mannitol, 10 mM Tris-maleate, 0.5 mM Na2HPO4, 0.2% BSA (pH 6.8) and protease inhibitors (1 mM PMSF and 1 μg/ml pepstatin A). The mitochondrial pellet was resuspended in the same medium and the protein concentration was determined by standard procedures. Assays for ATP transport were performed in 96-well microplates inside a FLUOstar OPTIMA microplate. The luminescence signals were transformed into the ATP concentrations using calibration curves as described previously . The initial ATP content in freshly isolated mitochondria from sal1Δ cells expressing luciferase from plasmid pYeDP-Cox4-Luc and SCaMC-3L from plasmid pYX142, in our assay conditions, is 0.238±0.037 nmol of ATP per mg of protein (means±S.E.M. of three independent determinations; determined as in ), identical with that of mitochondria from W303 and sal1Δ cells , and thus the same calibration curves were used.
RESULTS AND DISCUSSION
Identification of a novel SCaMC paralogue, SCaMC-3L, originated by a tandem duplication of a SCaMC-3 ancestor
During the inspection of human SCaMC-3 C-terminal variants, we detected immediately downstream of SCaMC-3 a predicted gene, LOC284427, containing sequences 52–75% identical with those of exons 5–10 of SCaMC-3. When similar analysis was performed on mouse and rat syntenic regions, equivalent counterparts, the hypothetical genes LOC301114 and 4933406J04Rik, were detected downstream of SCaMC-3 genes. In addition, corresponding EST clones were also detected in mouse and human databases, supporting the suggestion that they are expressed genes. We have named the genes SCaMC-3L given that they are closely related to SCaMC-3 sequences. Previously, a partial clone corresponding to rat LOC301114 sequences has been identified during a screening for MCs expressed in rat brain and numbered as slc25a41 . In this survey, human and mouse orthologues were also identified and positioned in close proximity to their respective SCaMC-3 paralogues .
A detailed analysis of these loci has confirmed that all mouse, rat and human SCaMC-3/SCaMC-3L loci share an identical genomic organization forming a head-to-tail tandem array. A diagram of the mouse genomic structure is shown in Figure 1(A). The predicted SCaMC-3L/slc25a41 genes, hereafter named SCaMC-3L, are composed of seven exons whereas SCaMC-3 genes have ten exons (Figure 1A). SCaMC-3 and SCaMC-3L are separated by short intergenic regions and exon/intron boundaries of SCaMC-3L exons 2–7 perfectly match those of exons 5–10 of their respective SCaMC-3 paralogues, suggesting that SCaMC-3L arose by a partial tandem duplication of a pre-existing SCaMC-3 ancestor. SCaMC-3 is most likely the ancestral locus for duplication as it contains exons absent in the shorter SCaMC-3L. In all SCaMC-3L loci analysed, sequences corresponding to SCaMC-3 exons 1–4 encoding the calcium-binding N-terminal extension characteristic of the SCaMC subfamily have been lost (Figure 1B). The predicted exons 2–7 of SCaMC-3L encode peptides highly homologous with amino acids 162 to the C-terminal end of their SCaMC-3 counterparts, and a novel exon 1, SCaMC-3-unrelated, provides the in-frame start codon and first amino acids (see Figure 1A). Searches using these novel exon 1 sequences failed to detect homologues belonging to other genes, implying that neighbouring sequences from its new genomic location have probably been recruited to generate this novel first SCaMC-3L exon.
SCaMC-3L, a novel shortened SCaMC paralogue generated by a tandem partial duplication of SCaMC-3
Subsequently, we confirmed mouse and rat SCaMC-3L full-length coding sequences by RT–PCR with specific primers designed to anneal to predicted exons 1 and 7. The sequences obtained matched well those of EST clones and encoded highly conserved proteins, 95% identical, of 312 amino acids. Mitochondrial localization of mouse SCaMC-3L was assayed in COS-7 cells by transient transfection of the full-length cDNA tagged at the C-terminal end with the FLAG epitope. Cells transiently transfected with SCaMC-3LFLAG were later analysed by immunofluorescence with anti-FLAG antibodies. In these assays, mitochondrial localization was verified by co-staining with MitoTracker Red  and by co-localization with the DsRed2-Mito protein (Figure 1C). Additionally, mitochondrial location in transfected cells was also confirmed by subcellular fractionation and Western blot analysis. Anti-FLAG antibodies detected a protein of the expected mass for SCaMC-3L, approx. 30–35 kDa, in both mitochondrial-enriched (Figure 1D, Mit lane) and total homogenate fractions (Figure 1D, H lane).
Interestingly, both mouse and rat SCaMC-3L differ from human SCaMC-3L, a predicted longer protein of 370 amino acids. In mouse and rat SCaMC-3L genes, the novel exon 1 displays a very short coding part covering only six amino acids, whereas the hypothetical human SCaMC-3L exhibits a long exon 1 which encodes an N-terminal extension of 60 amino acids not found in murine orthologues. The existence of transcripts of human SCaMC-3L is supported by several EST clones and sequenced cDNA fragments. Predicted SCaMC-3L orthologues in other non-rodent mammals such as chimpanzee, dog, cow or bat are also longer than murine ones, approx. 370 amino acids in length, harbouring conserved N-terminal extensions (Figure 1E). These N-terminal extensions possess proline-rich sequences, containing the core PxxP (showed in Figure 1E), a structural domain not described previously in any MC, which matches consensus sequences that interact with SH3 domains . Its high conservation in non-rodent mammals also suggests that SCaMC-3L exon 1 has been shortened exclusively in the rodent lineage. In agreement with this, we detected that sequences flanking the start codon are conserved between mouse and humans, but those downstream, encompassing a poly-proline-rich region, have been replaced by a repetitive element which belongs to the SINE (short interspersed element) family (results not shown). Therefore it is likely that the insertion of this repetitive element favoured the partial lost of exon 1 sequences in mouse SCaMC-3L.
Despite these differences at the N-termini, murine and human SCaMC-3L proteins exhibit high similarity with each other (92% similar), as well as with the C-terminal domain of SCaMC-3, their closest paralogue (approx. 76% similar), and with the more distant paralogues SCaMC-1 and SCaMC-2 (71% similar). When alignment between common regions of SCaMC-3 and SCaMC-3L counterparts was performed, we observed that substitutions were scattered along SCaMC-3L, albeit a slightly greater conservation is found in the transmembrane helixes H1, H2, H3 and H4 and in the long h3-4 matrix loop (see Figure 2A). On the other hand, amino acid substitutions appear more frequently in the short external loop that connects transmembrane helixes H2–H3 and in the C-terminal end region comprising amino acids 200–312. Interestingly, these residues are well-conserved between SCaMC-3L orthologues. Also, three single amino acid deletions conserved in all SCaMC-3L orthologues are observed. These single deletions are not derived from intron/exon boundary sequences and lie outside of transmembrane helixes. In addition, none of the substitutions affect residues proposed as substrate contact points for the yeast ATP-Mg/Pi carrier, Sal1p [33,34]. According to the proposed model, Sal1p binds the adenine group to residues Gly416 and Ile417 (equivalent to Gly188 and Ile189 in SCaMC-3L, contact point II), the phosphate groups form salt bridges with residues Lys314 and Lys523 (Lys95 and Lys286 in SCaMC-3L, contact points I and III respectively) and Arg242 (equivalent to Arg44 in SCaMC-3L) in the H1 helix, and an additional residue in contact point I, Glu318 (Glu99 in SCaMC-3L), are involved in the co-ordination of Mg2+ (Figure 2B). Although Arg242 in Sal1p has not been reported to make contact points, molecular simulations and mutagenesis approaches indicate that it participates in adenine nucleotide translocation [35,36]. In the yeast AAC (ADP/ATP carrier) Aac2p, mutation of Lys38, equivalent to residue Arg242, abolishes its transport activity . These interacting residues are conserved in all human ATP-Mg/Pi carrier paralogues identified until now (results not shown) [33,34]. In particular, mouse SCaMC-3L, in a similar manner to its human and rat orthologues, conserves equivalent residues to Arg242 (Arg44), and contact points I (Lys95 and Glu99), II (Glu188 and Ile189) and III (Lys286) (boxed in Figure 2A), as previously reported , suggesting that it could also perform an ATP-Mg/Pi counter-exchange [13,25–27].
SCaMC-3L and SCaMC-3 paralogues show high homology at the MC domain and conserve substrate contact points for ATP-Mg and Pi
SCaMC-3L is a calcium-independent ATP-Mg/Pi carrier
Although the high similarity of SCaMC-3L with other SCaMC paralogues suggested that this novel carrier probably carries out ATP-Mg/Pi exchange , its transport activity has not been tested to date. With this purpose, we have employed yeast expressing luciferase targeted to the mitochondrial matrix as a read-out system for matrix ATP levels . The transport activity of SCaMC-3L was studied after expressing mouse SCaMC-3L in a yeast strain deficient in Sal1p [26,27]. ATP transport in mitochondria isolated from this SCaMC-3L-expressing strain was tested with a system allowing the direct quantification of mitochondrial ATP import as recently described .
Figure 3(A) shows that SCaMC-3L-expressing mitochondria are able to take up ATP in the presence of 20 μM CAT (carboxyatractyloside), an inhibitor of the AACs, when ATP is added to the incubation medium, as shown by an increase in mitochondrial luminescence. This ATP transport activity is not present in mitochondria from the parental strain (sal1Δ mutant) (Figure 3A). ATP transport by SCaMC-3L is reduced by decreasing the Mg2+ concentration (Figure 3B), and also by the addition of 1 mM EDTA (see below), in agreement with the fact that ATP-Mg is the transported species in mammalian and yeast ATP-Mg/Pi carriers [13,15,27]. ATP transport depended on the presence of phosphate in mitochondria and was reduced when the phosphate concentration was lowered, as also seen for Sal1p (results not shown and ). We have found that SCaMC-3L is also an ADP transporter, as shown for other mammalian and yeast mitochondrial ATP-Mg/Pi carriers [13,15,27]. Figure 3(C) shows the increase in mitochondrial ATP obtained after ADP addition to mitochondria in the presence of CAT which follows its conversion into ATP by the H+-ATP synthase. It is not present in the parental strain (sal1Δ mutant) (Figure 3C). The rise in mitochondrial ATP after ADP addition was blocked by oligomycin, an inhibitor of the H+-ATP synthase, as also observed for ADP uptake by Sal1p (results not shown and ). ADP transport was, unlike ATP transport, insensitive to 1 mM EDTA addition (Figure 3D), in agreement with the fact that free ADP (and not ADP-Mg) is the transported species in other ATP-Mg/Pi carriers [13,15,27].
Transport activity of SCaMC-3L
The kinetics of ATP and ADP influx along SCaMC-3L over a range of external concentrations of ATP and ADP is shown in Figure 3(E). The apparent Km for ATP was 0.41±0.08 mM (mean±S.E.M. of three independent experiments performed in duplicate), similar to that obtained previously for a truncated version of human SCaMC-3, containing only the MC homology region, and reconstituted in proteoliposomes (0.22 mM) , confirming as well that the N-terminal extension is not required for its transport function. It was also similar to the Km of Sal1p in yeast mitochondria (0.20–0.24 mM) . For ADP the Km (0.90±0.12 mM; mean±S.E.M. of three independent experiments performed in duplicate), was higher than that for ATP, as also shown for SCaMC-3 (0.54 mM) , and unlike Sal1p and SCaMC-1 which have a similar Km for ATP and ADP [13,27]. On the other hand, the Vmax for ATP and ADP transport through SCaMC-3L was similar, as for the other ATP-Mg/Pi carriers studied so far [13,27]. Significantly, as expected from its lack of calcium-binding domains, metabolite transport mediated by SCaMC-3L was calcium-insensitive, in contrast with the behaviour of Sal1p, the only representative of the ATP-Mg/Pi carriers whose calcium regulation has been studied (see Supplementary Figure S1 at http://www.BiochemJ.org/bj/418/bj4180125add.htm). Taken together, our results prove that SCaMC-3L is a true mitochondrial transporter of ATP-Mg/Pi which, unlike its other paralogues, is calcium independent.
The tandem duplication event of the SCaMC-3 ancestor occurred prior to mammalian radiation
To date three ATP-Mg/Pi paralogues, SCaMC-1, -2 and -3, have been described in vertebrate genomes. All are probably generated by gene-duplication events, the major source of new genes in vertebrates . As a number of insects, sea urchin or Ciona intestinalis (a basal chordate and representative of outgroup to the vertebrates), all present a single SCaMC counterpart, it is reasonable to consider that the ATP-Mg/Pi paralogues arose during large-scale genome-amplification events that occurred around the time of the origin of vertebrates . In fact, human SCaMCs map at chromosomal loci enclosed in regions belonging to the characterized MHC paralogous region 6p/1p/9q/19p [12,39,40]. Analysis of these regions in human and amphioxus supports that MHC-paralogous regions are the result of en bloc duplications that occurred after the divergence of cephalochordates and vertebrates (approx. 750 Myr ago [40,41]). However, BLASTN searches using mouse SCaMC-3L as a query failed to detect SCaMC-3L orthologue sequences in either fish, amphibian or bird genomes, suggesting that this carrier may represent a more recent acquisition exclusive to mammalian genomes.
Thus an interesting question was whether its apparition was linked to the initial mammalian radiation or represented a more recent event. With this purpose, we searched for SCaMC-3L and SCaMC-3 orthologues in relevant mammalian and vertebrate genomes currently available to ascertain its phylogenetic relationships. The high conservation degree within each SCaMC subgroup facilitates the identification of orthologues, as opposed to paralogues, among species by reciprocal-best-blast tests. In addition, all SCaMC-3/SCaMC-3L loci analysed display an identical head-to-tail arrangement favouring their identification. SCaMC-3 orthologues are found in pufferfish (Tetraodon nigroviridis) and zebrafish (Danio rerio) and, although no SCaMC-3 counterparts are detected in frogs or chicken, highly conserved SCaMC-3 counterparts are detected in all mammals analysed including the monotreme platypus (Ornithorhynchus anatinus). In chicken, the homologous segment with the human region encompassing the SCaMC-3 locus has been lost in the corresponding synteny block on chicken chromosome 28 . On the other hand, we have identified SCaMC-3L orthologues in all placental mammals with available genomes, as well as in the marsupial opossum (Monodelphis domestica), but not in the platypus. In the platypus genome, a large genomic clone containing SCaMC-3 without adjacent SCaMC-3L-related sequences has been detected (NCBI accession number: AAPN01306014). This suggests that duplication of the SCaMC-3L ancestor could have occurred in mammals after radiation from monotremes. This hypothesis was tested by constructing a phylogenetic tree with representative SCaMC-3 and SCaMC-3L proteins using the neighbour-joining method (Phylo_win)  and with non-vertebrates SCaMCs as the basal outgroup. As is shown in Figure 4, the topology of the tree clearly rules out that the ancestral SCaMC-3 duplication occurred in monotremes. SCaMC-3L and SCaMC-3 mammalian orthologues (including the platypus counterpart) form two independent and well-defined clades and it is likely that the partial duplication event that generated the SCaMC-3L paralogue occurred prior to the mammalian radiation (80–430 Myr ago ), probably after divergence between fish and tetrapod lineages, by means of a small-scale tandem duplication not related to large-scale duplications common of vertebrates. Unfortunately, the lack of SCaMC-3 counterparts in amphibians and birds prevents the determination with more precision of when the SCaMC-3L subfamily emerged. Nevertheless, its maintenance in all mammalian groups indicates that SCaMC-3L has a relevant function in this lineage.
Phylogenetic relationship between SCaMC-3 and SCaMC-3L paralogues
Interestingly, although we detected high homology among SCaMC-3L paralogues, the SCaMC-3L cluster shows a branch longer than that of SCaMC-3, suggesting unequal evolution between paralogues. It is possible that the partial duplication created a non-functional product that underwent an accelerated rate of amino acid changes during the period following gene duplication. Later, the duplicated gene became functional by the acquisition of a novel first exon and cis-regulatory elements, resulting, finally, in a relatively distant subfamily whose members have evolved maintaining high homology inside the cluster (see the alignment in Supplementary Figure S2 at http://www.BiochemJ.org/bj/418/bj4180125add.htm).
SCaMC-3L expression is restricted to testis and brain
It has been hypothesized that duplicated versions would be evolutionarily retained if they diverge acquiring novel functions or if original functions are subdivided between ancestral and newly duplicated versions [43–47]. Indeed, a common fate postulated for the members of duplicated-gene pairs is the partitioning of tissue-specific expression patterns of the ancestral gene [43,44]. In order to compare expression patterns between duplicates, we analysed their distribution in mouse tissues by RT–PCR (Figure 5) and Northern blot analysis (Supplementary Figure S3 at http://www.BiochemJ.org/bj/418/bj4180125add.htm), confirming that SCaMC-3 and SCaMC-3L display non-coincident expression patterns. In mouse tissues, SCaMC-3L expression is mainly detected in testis and, at lesser levels, in brain (Figure 5), whereas SCaMC-3 is broadly distributed in most tissues, including testis and brain (Figure 5), as previously reported [12,13,24]. A similar expression pattern is also detected for SCaMC-3L in rat tissues (Figure 5, bottom panel). Although, it has been previously reported that rat SCaMC-3L is expressed mainly in brain and at low levels in testis , the observed testis-predominant pattern matches well with the origin of mouse SCaMC-3L EST clones found in databases where eight out of ten derived from testis (NCBI Unigene Mm.33647). Moreover, the mouse SCaMC-3L transcript harbours a short 3′ UTR (untranslated region), a common feature of testis-specific transcripts. In contrast, in other non-rodent mammals, such as humans or cattle, SCaMC-3L-clones derive principally from brain and liver, indicating that SCaMC-3L is differentially expressed among mammalian subgroups, but always shows a more limited expression pattern than SCaMC-3.
Murine SCaMC-3L and SCaMC-3 show different expression patterns
SCaMC-3L represents, therefore, a clear example of gene duplication followed by subfunctionalization, which takes place both at structural and expression levels. On the structural level, SCaMCs are proteins with two separated functional modules; this aspect makes them prone to subfunctionalization since it facilitates that newly duplicated copies undergo independent changes in each module to carry out only a fraction of the original functions [45,47]. The duplication event that generated SCaMC-3L was partial, as found for 50% of newborn duplications in Caenorhabditis elegans , and the exons of the ancestral gene encoding calcium-binding domains were not included in the duplicated segment. Thus in SCaMC-3L the entire regulatory calcium-binding N-terminal extension distinctive of SCaMC members has been lost, maintaining the intact transport module. A second mechanism of subfunctionalization at the expression level is based on the divergence and subdivision of the cis-regulatory elements of the parental gene, so that the newly duplicated gene has a more restricted distribution, probably in a subset of the tissues where its ancestors were distributed [38,40,41,43]. In mouse and rat tissues, SCaMC-3L expression is mainly detected in testis, whereas SCaMC-3 is expressed in most tissues. This predominant testis-expression is infrequent among MCF genes. A notable exception is the ADP/ATP translocases, where a newly discovered fourth member AAC4/slc25a31, is expressed exclusively in testis [10,49,50] and has been found to be essential for spermatogenesis . In addition, SCaMC-1a, an SCaMC-1 isoform found in primates, is expressed from a testis-specific promoter [13,17]. Collectively, these results suggest that mitochondrial transport of adenine nucleotides by either the testis-specific AACs that are classically involved in oxidative phosphorylation or the testis-specific ATP-Mg/Pi carriers that catalyse a net accumulation or depletion of mitochondrial adenine nucleotides is important in testicular development or spermatogenesis. The specific role of each of these transporters, and, particularly, the two ATP-Mg/Pi carriers enriched in primate testis, the calcium-dependent SCaMC-1a and the calcium-independent SCaMC-3L remains to be established.
adenine nucleotide translocator
expressed sequence tag
human embryonic kidney
0.17% yeast nitrogen base without amino acids and ammonium sulfate/0.5% ammonium sulfate/0.2% glucose as carbon source
small calcium-binding MC
This work was supported by grants from the Ministerio de Educación y Ciencia [grant numbers BFU2005-C02-01, GEN2003-20235-C05-03/NAC]; the European Union [grant numbers LSHM-CT-2006-518153]; the Comunidad de Madrid [grant numbers S-GEN-0269-2006 MITOLAB-CM]; CIBER de Enfermedades Raras an initiative of the ISCIII; and an institutional grant from the Fundación Ramón Areces to the Centro de Biología Molecular Severo Ochoa. J. T. was a recipient of an FPU fellowship from the Ministerio de Educación y Ciencia.