Background

The mitochondrial carrier family (MCF) comprises structurally related but functionally diverse proteins that are characteristic for and generally restricted to eukaryotes [1–5]. MCF proteins represent the main solute carriers in the inner mitochondrial membrane and catalyze the translocation of various metabolites, such as nucleotides, cofactors, carboxylates, amino acids etc (for review see [6]).

Mitochondrial ATP-Mg/phosphate carriers (APCs) represent a specific MCF subgroup comprising carriers from different eukaryotes that are phylogenetically related to the well characterized ADP/ATP carriers (AACs) required for mitochondrial energy passage (for review see [6, 7]). Over the past years the physiological and biochemical properties of the single yeast APC isoform Sal1p (suppressor of ∆aac2 lethality) as well as of various mammalian homologs became more and more clarified [8]. Initially, Sal1p was shown to suppress the growth phenotype of yeast impaired in mitochondrial energy transport (due to AAC deletion or inhibition). In a similar fashion, AAC compensates the loss of functional Sal1p [8]. Subsequent studies revealed that Sal1p and its mammalian homologs mediate the counter exchange of adenine nucleotides and phosphate [9–13]. Therefore, the redundant physiological function of Sal1p and AAC supposedly was not primarily energy exchange but adenine nucleotide translocation, most likely ATP entry into mitochondria [13, 14].

Alteration of the mitochondrial adenine nucleotide pool by adenine nucleotide exchange with phosphate was shown to affect different physiological processes, such as glucose metabolism, oxidative phosphorylation, mitochondrial biogenesis and DNA maintenance in yeast or mammals [9–13]. APC proteins apparently prefer two-fold negatively charged substrates, either ATP in complex with Mg2+ (ATP-Mg2−), protonated ADP (HADP2−) or HPO4 2−, which makes the catalyzed transport electroneutral [15]. The composition (respective concentrations) of the different substrates at the matrix and cytosolic sides of the carrier determine whether adenine nucleotides preferentially become exported or imported [11, 15].

Interestingly, addition of Ca2+ to isolated mitochondria as well as metabolic situations that result in increase of free cytosolic Ca2+ were shown to enhance mitochondrial adenine nucleotide levels by stimulation of APC activity in mammals and yeast [8, 16–18] (for review see [19]). In one aspect APC proteins considerably differ structurally from typical MCF proteins; they are N-terminally extended by a domain that is exposed to the inter-membrane space of the mitochondrion and contains up to four putative Ca2+-binding EF-hand motifs [20–22]. Very recent structural studies with the N-terminal domain of human APC isoform 1 (also termed SCaMC1 for short Ca2+-dependent mitochondrial Carrier 1) showed that the Ca2+-bound state is quite compact and rigid whereas the apo (Ca2+-free) state appeared more flexible [21, 22]. Moreover, interaction studies with the two individual SCaMC1 domains, the Ca2+-binding part and the C-terminal transmembrane region, led to the assumption that the apo state of the N-terminal domain forms a cap that closes the translocation pathway whereas Ca2+-binding causes cap removal/opening and thus transporter activation [21, 22].

In contrast to yeast and mammals [8, 12, 16, 18, 23–25] analyses concerning the net adenine nucleotide transport of mitochondria in plants are still rudimentary. Previous studies led to controversial results but have indicated that plant mitochondria are capable of net adenine nucleotide uptake [26–31]. Arabidopsis thaliana possesses three putative APC proteins (AtAPC1-3) that exhibit high amino acid sequence similarities to their human and yeast counterparts. Phylogenetic analysis of MCF proteins shows that APCs cluster together and that plant APCs form a sister group to the human and yeast orthologs [6]. Similar to yeast or mammalian APCs, the plant pendants contain an N-terminal extension with four putative EF-hand motifs and were recently shown to interact with Ca2+ at least in vitro [32]. Moreover, all three plant isoforms were able to rescue the specific growth phenotype of ∆sal1p yeast mutants [32]. Therefore, AtAPC1-3 isoforms were suggested to represent Ca2+-regulated ATP-Mg/phosphate transporters. To gain first insights into the biochemical characteristics of the three APCs from A. thaliana we reconstituted the heterologously expressed proteins into liposomes and investigated their capacity for adenine nucleotide transport. Our data indicate that plant APCs mediate antiport of ATP, ADP and phosphate and therefore might be involved the alteration of the mitochondrial adenine nucleotide pool. Moreover, the determined transport characteristics suggest that in the in vitro system, the plant APCs preferentially import the Ca2+- and not the Mg2+-complexed form of ATP.

Methods

Generation of expression constructs

The coding sequences of AtAPC1-3 were amplified from Arabidopsis cDNA with specific primers via Pfu-polymerase-mediated PCR. For generation of the truncated AtAPC2 mutant protein lacking its Ca2+-interacting N-terminus a sense primer was chosen that internally hybridizes with the corresponding full-length sequence resulting in a recombinant protein starting at amino acid position 164 directly after the fourth predicted EF-hand motif coding region. The isopropyl β-D-thiogalactopyranoside (IPTG)-inducible T7 RNA polymerase pET-vector/Rosetta™ 2 expression system (Merck Biosciences, Novagen®, Darmstadt, Germany) was used for heterologous protein synthesis. Accordingly, the primers were adapted to allow insertion into the expression vector pET16b in frame with the histidine-tag coding sequence. The coding sequence of AtAPC1 was inserted via NdeI (sense primer) and XhoI (antisense primer) whereas the remaining sequences were inserted via XhoI (sense primer) and BamHI (antisense primer). Correctness of the respective expression constructs was verified by sequencing.

Heterologous protein synthesis and detection

For heterologous protein synthesis Rosetta™ 2 cells were transformed with the expression constructs and cultured in 50 mL standard Terrific Broth (TB) medium at 37 °C under vigorous shaking. At an OD600 of 0.5, expression was induced by addition of 1 mM IPTG. Two hours after induction, cells were concentrated by centrifugation (3000 g, 5 min, 4 °C) and rapidly frozen (in liquid nitrogen). The frozen cell pellet was resuspended in buffer R (25 % sucrose, 50 mM Tris, pH 7.0, 1.5 % Triton X-100, 18.75 mM EDTA) supplemented with 1 mM PMSF, a pinch of DNAse and RNAse and incubated for approximately 30 min at 37 °C to stimulate autolysis by the endogenous lysozyme which was released from the cells due to the freeze/thaw procedure. Subsequent sonication additionally supported cell disruption. Inclusion bodies were separated from soluble and membrane proteins of the cell homogenate by centrifugation (20,000 g, 15 min, 4 °C).

For documentation of heterologous protein synthesis, an aliquot of the inclusion bodies fraction was used for SDS-PAGE, Western-blotting and immune detection. For this, inclusion bodies were resuspended in buffer R and an appropriate volume of 6 x concentrated sample buffer medium (375 mM Tris/HCl, pH 6.8, 0.3 % SDS, 60 % glycerol, 1.5 % bromophenol blue) was added. Protein separation was performed in a discontinuous, denaturing system with a 3 % stacking and a 12 % separating polyacrylamide gel [33]. Following electrophoresis, the gel was coomassie stained or used for Western-blotting. Immune detection was performed using a monoclonal anti poly His IgG (Sigma; http://www.sigmaaldrich.com) combined with a secondary alkaline phosphatase conjugated anti-mouse IgG (Sigma). Alkaline phosphatase activity was detected by staining with nitro blue tetrazolium chloride/5-bromo-4-chloro-3’-indoly phosphate toluidine salt.

Purification of inclusion bodies

Basically, purification of inclusion bodies as well as their solubiliztaion, refolding and integration into lipid/detergent micelles was performed according to [34]. For this, the cell pellet of the inclusion body fraction washed in buffer W1 (20 ml 1 M urea, 1 % Triton X-100 and 0.1 % β-mercapto-ethanol). After centrifugation (20,000 g, 15 min, 4 °C) inclusion bodies were additionally washed in buffer W2 (20 mM Tris, pH 7.0, 0.5 % Triton X-100, 1 mM EDTA, 0.1 % β-mercapto-ethanol) and finally in buffer W3 (50 mM Tris, pH 7.0, 1 mM EDTA, 0.1 % β-mercapto-ethanol). Solubilization of the purified inclusion body proteins was achieved by resuspension in buffer medium S (10 mM Tris, pH 7.0, 0.1 mM EDTA, 1 mM DTT, 0.05 % polyethylene glycol 4000) containing 1.67 % of the detergent n-lauroylsarcosine and incubation for 15 min on ice. The protein fraction was diluted (threefold) with 10 mM Tris (pH 7.0) and finally, the solubilized proteins were separated from insoluble aggregates by centrifugation (12,000 g, 4 min, 4 °C).

Preparation of proteoliposomes and transport measurements

For preparation of proteoliposomes 100 μg of the solubilized proteins were mixed with 20 mM Hepes, pH 7.0 and 1 mM PMSF. To obtain vesicles with internal counter exchange substrates 5 mM of phosphate or adenine nucleotides were added to the protein mixture. Preparation of mixed detergent-lipid micelles (100 mM PIPES, pH 7.0, 20 mg phosphatidylcholine, 1.6 mg cardiolipin, 28 mg C10E5) and detergent removal by amberlite XAD-2 beads was performed exactly as given by Heimpel et al., [34]. Overnight incubation with biobeads completed protein refolding and proteoliposome formation. External buffer medium and loading substrates were removed from the vesicles (500 μL) by desalting with NAP-5 columns (GE Healthcare; http://www.gehealthcare.com). Columns were equilibrated and liposomes were eluted with of import buffer (50 mM NaCl, 10 mM PIPES, pH 7.5). For transport measurements 50 μL of these proteoliposomes were mixed with 50 μL of import buffer supplemented with the indicated concentrations of [α32P]-ATP, [α32P]-ADP, [45Ca], MgCl2 and CaCl2 and incubated at 30 °C. At the given time points import was terminated by removal of external import medium via vacuum filtration as described in [35]. Briefly, liposomes were loaded to pre-wetted filters (mixed cellulose ester, 0.45-μm pore size; Whatman) and washed rapidly with phosphate buffer. Imported radioactivity was quantified by scintillation counting (Beckman LS6500; Beckman Coulter). For [45Ca] uptake measurements import was terminated and non-imported Ca2+ was removed by EGTA addition (2 mM) and incubation for 15 s prior to vacuum filtration and washing.

Results

Recombinant plant APCs act as ATP, ADP and Pi antiporters

To determine functional properties of the different APCs from A. thaliana we used the heterologous Escherichia coli expression system for production of the respective isoforms and performed transport measurements after carrier reconstitution into artificial lipid vesicles, so called liposomes. This approach was previously successfully applied to biochemically characterize several MCF proteins, including two selected human SCaMC isoforms [12, 34, 36–38].

The three plant APCs were heterologously expressed as N-terminal His-tag fusions. Like previously observed for many MCF proteins [12, 34, 36–38] also plant APCs were synthesized at high levels and accumulated in form of insoluble inclusion bodies (Additional file 1: Figure S1A and B). The aggregated proteins were enriched, purified, solubilized and finally refolded during their integration into liposomes.

Import measurements were performed on proteoliposomes either harboring or lacking selected possible counter exchange substrates in the lumen (Fig. 1, Additional file 2: Figure S2). This allowed investigation of in vitro transport activities and hence functionality of the reconstituted proteins as well as of the catalyzed transport mode. All recombinant plant APCs mediated time dependent uptake of [α32P]-ATP into phosphate (Pi) loaded liposomes (Fig. 1a, c, e, black rhombs) and no comparable accumulation of radioactivity was observable with corresponding vesicles lacking Pi in the lumen (Fig. 1a, c, e, open rhombs). This observation already demonstrates that plant APCs can act as antiporters; ATP/Pi exchange by the different APC isoforms was linear for at least 5 min. Maximal uptake via AtAPC1 of ~ 6 nmol/mg protein was reached after 10 to 15 min (Fig. 1a, black rhombs), whereas AtAPC2 and AtAPC3 show marginally or considerably higher transport rates that approached a maximum of ~ 9 nmol/mg protein and ≥ 17 nmol/mg protein after 20 min, respectively (Fig. 1c and e, black rhombs).

Fig. 1
figure 1

Time dependent ATP transport via AtAPC1-3. Transport of 50 μM [α32P]-ATP into Pi (a, c, e) and into ATP (b, d, f) loaded proteoliposomes with reconstituted AtAPC1 (a, b), AtAPC2 (c, d) and AtAPC3 (e, f). ATP uptake was measured in absence (black rhombs) and presence (gray circles) of 500 μM externally applied MgCl2. Non-loaded liposomes (non-filled rhombs; negative control) showed only marginal accumulation of radioactivity and the corresponding rates were unaffected by MgCl2 addition. Data represent mean values of at least three independent replicates, standard errors are given

Full size image

Yeast Sal1p and mammalian SCaMCs were shown to discriminate against free ATP as substrate or at least to prefer the Mg2+-complexed form of ATP over free ATP [12, 15, 16, 39]. To check whether this is also true for the plant APCs, the influence of Mg2+ on ATP transport was analyzed. To this end, the ATP transport medium was supplemented with 500 μM Mg2+ to convert ~ 80 % of free ATP (ATP4−) into the Mg2+-complexed form (ATP-Mg2−) (http://maxchelator.stanford.edu/CaMgATPEGTA-TS.htm [40]). ATP-Mg2− and HPO4 2− exchange results in an electroneutral transport. In case of AtAPC1 and AtAPC3 addition of Mg2+ caused moderate (~1.6-fold to 2.0-fold) increase in adenine nucleotide/Pi exchange compared to ATP without Mg2+ (Table 1; Fig. 1a and e, compare gray circles and black rhombs) whereas transport by AtAPC2 was stimulated to a lesser extent (Table 1; Fig. 1c, compare gray circles and black rhombs).

Table 1 Comparison of counter exchange rates of AtAPC1-3
Full size table

To unravel whether the stimulatory influence of Mg2+ on ATP uptake is due to general preference for ATP-Mg as substrate or rather due to the electroneutrality of the corresponding transport process we investigated Mg2+-effects on ATP homo-exchange. Homo-exchange of ATP is electroneutral but becomes electrogenic when ATP-Mg2− is exchanged with ATP4−. Comparison of the transport rates indicates that AtAPC1 highly, AtAPC3 markedly and AtAPC2 slightly prefer ATP homo-exchanges over the corresponding ATP/Pi hetero-exchanges (Table 1; compare Fig. 1a, c, e with b, d, f, black rhombs). Moreover, ATP homo-exchanges of all three AtAPCs became further enhanced by Mg2+ (Fig. 1b, d, f, compare gray circles and black rhombs) and the degree of Mg2+-dependent stimulation was nearly identical to that of the ATP/Pi hetero-exchange (Table 1). The observed stimulatory effects of Mg2+ on ATP/Pi and ATP/ATP transport indicate that AtAPC1 and 3 generally prefer ATP-Mg as substrate whereas AtAPC2 apparently only slightly favors the Mg2+-complexed form.

Because ADP represents an additional substrate of yeast Sal1p and human SCaMCs [12, 15, 18, 39] we verified whether this nucleotide is also transported by the plant orthologs in our in vitro system. For this, uptake of radiolabeled ADP into differentially loaded liposomes was measured. All plant APCs transported ADP in hetero-exchange with Pi as well as in homo-exchange with ADP and no import occurred into non-loaded vesicles (Additional file 2: Figure S2). The rates of ADP transport (in exchange with Pi or ADP) largely resemble the rates of the corresponding Mg2+-stimulated ATP transport (in exchange with Pi or ATP) (Table 1; compare Additional file 2: Figure S2 and Fig. 1). Just like observed for ATP transport, all plant APCs favor the homo-exchange of ADP over the corresponding ADP/Pi hetero-exchange and this preference is highly pronounced for AtAPC1 followed by AtAPC3 and finally AtAPC2 (Table 1). Moreover, comparison of the rates of ATP and ADP homo-exchanges with those of the corresponding Pi hetero-exchanges (Table 1) suggests that AtAPC2 in contrast to AtAPC1 and 3 does not strongly discriminate between nucleotides or Pi as internal counter exchange substrate. The ineffectiveness of non-loaded vesicles to induce significant import of ATP or ADP also demonstrates that vesicles do not allow carrier-independent passage of the labeled compounds.

Calcium differentially affects ATP and ADP transport properties of the plant APCs

Diverse physiological data indicate a Ca2+-dependent regulation of mitochondrial net adenine nucleotide passage [16–18, 39, 41]. In the native environment many factors such as activity of adenylate kinases, Ca2+-induced metabolic processes, the mitochondrial membrane potential, respiration, Mg2+ complexation of ATP, etc. influence internal and external adenylate and Pi pools and consequently also mitochondrial adenine nucleotide translocation in general [42–45].

Transport studies with reconstituted APCs might provide a suitable tool to overcome interfering metabolic and physiological effects and to study the impact of Ca2+on this process in more detail. However, it is important to mention that transport of reconstituted human SCaMC1 was not stimulated by Ca2+ addition [12] and also AtAPC1-3 are already active in the absence of any Ca2+ addition (Fig. 1 and Additional file 2: Figure S2). These findings suggest that Ca2+ is not essentially required for carrier activation or that Ca2+contaminations exist in the buffer media. Determination of cations (by ion chromatography) revealed that in fact traces of both, Ca2+ and Mg2+, are present in the media (~9 μM, respectively).

If Ca2+ is essential for carrier activation and under the assumption that the proteoliposomes still contain a certain amount of inactive (Ca2+-free) APC proteins, an addition of extra Ca2+ should result in transport stimulation. To investigate a possible Ca2+-induced increase in transport activation we performed uptake studies with and without 200 μM Ca2+. Elevated Ca2+ availability generally stimulated nucleotide uptake of all three plant APCs (Table 2). This observation might point to a Ca2+-induced activation of previously inactive (Ca2+-free) carrier proteins. Studies with the two separately expressed sub-domains (N-terminal domain and membrane spanning part) of the human SCaMC1 led to the conclusion that the N-terminal domain acts as a lid that either opens or closes the translocation pathway in response to Ca2+ availability [22]. Given that Ca2+ exclusively causes removal of the N-terminal domain and hence activation of previously closed carriers, the same degree of stimulation would be expected independent of the kind of substrate exchanged. However, direct comparison of Ca2+ influence on different exchanges shows that for the reconstituted plant APCs the degree of stimulation is higher for ATP than for ADP or ATP-Mg uptake (Table 2).

Table 2 Stimulation of the given exchanges by addition of 200 μM CaCl2
Full size table

We thus determined the apparent biochemical parameters of ATP/Pi and ADP/ATP exchange for all three AtAPCs in more detail (Table 3). Velocity of transport of all recombinant carriers approached saturation with increasing ATP or ADP concentrations and conformed to simple Michaelis-Menten kinetics (Additional file 3: Figure S3). The individual AtAPC isoforms differed in their respective ADP affinities (AtAPC1: 180 μM, AtAPC2: 374 μM and AtAPC3: 72 μM) whereas the ATP affinities were more similar (ranging from 68 to 113 μM). Affinities of AtAPC1 for ATP and ADP remained more or less unaffected by Ca2+ addition whereas ATP affinities of AtAPC2 and AtAPC3 increased (1.6- and 2.0-fold) and ADP affinities decreased (1.4- and 1.9-fold), respectively. All AtAPC isoforms generally exhibit lower maximal velocities (Vmax) for ATP than for ADP transport (Table 3). Since the Vmax is proportional to the amount of actively transporting carrier proteins enhanced Ca2+-dependent activation of the APCs should be reflected by an identical increase in the Vmax of both, ADP and ATP transport. However, addition of extra Ca2+ caused only a moderate increase (by approximately 1.2- to 1.5-fold) in maximal ADP Vmax but stimulated the respective ATP Vmax (2.0- to 2.5-fold) of all three APCs to a greater extend.

Table 3 Effects of 200 μM calcium on KM-Values of adenine nucleotide transport
Full size table

The different effects of Ca2+ on ATP and ADP transport properties indicate that besides its proposed function in cap removal and carrier activation Ca2+ fulfills an additional role in substrate transport/recognition.

Ca2+ effects override Mg2+ effects on ATP transport

To approach the function of Ca2+ during plant APC mediated transport it is important to keep in mind that ATP can form a complex with Mg2+as well as with Ca2+ and it is thus imaginable that plant APCs are capable of ATP-Ca transport in vitro.

Comparison of Ca2+ effects on ATP and ATP-Mg transport indeed revealed interesting results that support this assumption. Mg2+ addition causes marginal (AtAPC2) to moderate (AtAPC1 and 3) increase in ATP transport when no extra Ca2+ is present (Figs. 1 and 2). With rising Ca2+ concentration the positive impact of Mg2+ becomes abolished and even reverted into a negative one (Fig. 2). More precisely, with higher Ca2+ concentrations (>10 μM AtAPC2; > 50 μM AtAPC1 and 3) the rates of ATP transport in absence of Mg2+ exceed the rates of the corresponding exchange in presence of Mg2+. Accordingly, in presence of Mg2+ higher concentrations of Ca2+ are apparently required to achieve ATP-transport saturation.

Fig. 2
figure 2

Ca2+-impact on ATP transport via AtAPC1-3. Effect of rising Ca2+-concentrations (0–500 μM) on transport mediated by recombinant AtAPC1 (a, b), AtAPC2 (c, d) and AtAPC3 (e, f). Transport of 50 μM [α32P]-ATP was conducted in absence (black rhombs) and presence of supplemental MgCl2 (gray rhombs). Transport was allowed for 5 min and is given in nmol mg protein−1 h−1. Ca2+-dependent stimulation of ATP/Pi hetero-exchanges (a, c, e) and ATP/ATP homo-exchanges (b, d, f). Non-loaded liposomes (non-filled rhombs; negative control) showed only marginal accumulation of ATP and the corresponding rates were unaffected by MgCl2 addition. Data represent mean values of three independent replicates, standard errors are given

Full size image

ATP transport stimulation by Ca2+ does not involve the N-terminal domain

We choose AtAPC2 for a more detailed analysis of the proposed ATP-Ca transport because ATP uptake of this transporter was markedly stimulated by Ca2+and particularly because Ca2+ stimulation was only slightly affected by Mg2+ presence (Fig. 2). To investigate ATP-Ca transport disconnected from possible Ca2+-dependent carrier activation we generated an AtAPC2 mutant protein lacking the predicted N-terminal domain (Additional file 4: Figure S4A and B). ATP uptake measurements verified that truncated AtAPC2 is functional (Additional file 4: Figure S4C), however, the uptake rates were slightly lower than those of the full-length protein.

Determination of Ca2+ impact on transport activity showed that ATP/Pi exchange via the mutated carrier was considerably stimulated by increasing Ca2+ concentrations (~3-fold). Moreover, the degree of Ca2+-dependent stimulation and the general course of the corresponding transport basically resembled that of the full-length protein (Fig. 3, black squares). Investigation of ADP uptake into ATP loaded liposomes revealed slight transport stimulation of the full-length protein by low Ca2+ concentrations (~35 % at 50 to 100 μM Ca2+), which approached saturation at higher concentrations (+60 %) (Fig. 3a, gray circles), whereas the corresponding transport of the truncated carrier version remained rather unaffected by moderate Ca2+ concentrations (+/− 10 % until 200 μM Ca2+) and became stimulated only at higher Ca2+ concentrations (+50 %) (Fig. 3b, gray circles).

Fig. 3
figure 3

Ca2+-impact on ATP and ADP transport of full-length and N-terminally truncated AtAPC2. Transport via recombinant AtAPC2 (a) and via the mutated version lacking its N-terminal domain (b). Import of [α32P]-ATP into Pi loaded proteoliposomes (black squares) and of [α32P]-ADP into ATP loaded vesicles (gray circles) was allowed for 5 min. Transport without CaCl2 was set to 100 % and transport in presence of rising concentrations of externally added CaCl2 (0 - 500 μM) was calculated accordingly. Data represent net values of ATP/Pi and ADP/ATP uptake minus the respective control (non-loaded vesicles) of three independent replicates. Standard errors are given

Full size image

Although slight differences in the Ca2+-impact are detectable, the higher influence of Ca2+ on ATP than on ADP import is apparently independent of the presence or absence of the N-terminal domain. This result verifies that the observed Ca2+-dependent ATP transport stimulation does not primarily result from carrier activation and might rather be caused by increased ATP-Ca formation and substrate availability.

ATP but not ADP import of AtAPC2 requires the presence of divalent cations

Because full-length AtAPC2 already exhibits basic ATP/Pi exchange activity without extra Ca2+ addition and particularly because ADP uptake becomes not highly stimulated by rising Ca2+-concentrations (Fig. 3), it might be assumed that the majority of reconstituted carriers is already opened/activated due to contaminating Ca2+.

The cation chelator EGTA efficiently chelates Ca2+ (with significant higher affinity than to Mg2+) and accordingly should remove residual Ca2+ from the medium. We thus used addition of EGTA to the transport medium to investigate whether and how Ca2+ depletion affects carrier activities. ATP/Pi exchange of full-length AtAPC2 becomes significantly reduced by addition of 10 μM EGTA and further increase of its concentration causes total inhibition (Fig. 4a, black squares). Interestingly, a similar inhibitory effect was also observed for the truncated carrier version (Fig. 4b, black squares). Given that the N-terminal domain forms a lid that virtually closes the translocation pathway when free Ca2+ is missing, efficient Ca2+-removal should impede transport activity of AtAPC2 but not of the “un-capped” mutant. Moreover, ADP/ATP exchange of both, full-length and truncated, AtAPC2 variants remained more or less unaltered by EGTA addition (Fig. 4a and b, gray circles). Accordingly, Ca2+ removal from the medium did not cause inhibition of the overall transport capacity by deactivation of the reconstituted carrier.

Fig. 4
figure 4

Impact of rising EGTA concentrations on adenine nucleotide transport of full-length and N-terminally truncated AtAPC2. Transport via recombinant AtAPC2 (a) and via the mutated version lacking its N-terminal domain (b). Import of [α32P]-ATP into Pi loaded proteoliposomes (black squares) and of [α32P]-ADP into ATP loaded vesicles (gray circles) was allowed for 5 min. Transport without EGTA was set to 100 % and transport in presence of rising concentrations of externally added EGTA (0 - 500 μM) was calculated accordingly. Data represent net values of ATP/Pi and ADP/ATP import minus the respective control (non-loaded vesicles) of three independent replicates. Standard errors are given

Full size image

Interestingly, transport via AtAPC2 was not only blocked by EGTA but also by the divalent cation chelator EDTA. Moreover, activity of the EGTA-inhibited carrier could be fully restored by either Ca2+ or Mg2+ (Fig. 5). However, when compared to Ca2+ higher concentrations of Mg2+ are required for transport reactivation/stimulation.

Fig. 5
figure 5

EGTA-inhibition of ATP transport via AtAPC2 and reactivation by MgCl2 and CaCl2. Transport of 50 μM [α32P]-ATP into Pi loaded vesicles in absence of EGTA, MgCl2 and CaCl2 was set to 100 % (control; red line). Transport in presence of 200 μM EGTA, 200 μM EDTA and the given concentrations (in mM) of MgCl2 or CaCl2 was calculated accordingly. Generally transport was allowed for 10 min. However, EGTA-inhibited transport was reactivated after 10 min of uptake by subsequent addition of MgCl2 or CaCl2 and transport was again allowed for 10 min (EGTA pretreatment and reactivation; light gray bars). Data represent net values (ATP import in exchange with Pi minus background values of non-loaded proteoliposomes) and are the mean of three independent experiments. Standard errors are indicated

Full size image

So far we cannot explain explicitly why solely ATP transport, but not general carrier activity, becomes inhibited by EGTA. It is imaginable that full-length AtAPC2 proteins are primarily or exclusively inserted in an inside-out orientation, exposing the N-terminal domain to the liposomal interior. This orientation would clearly hinder EGTA access to the regulatory sites (EF-Hands). However, AtAPC2-proteoliposomes loaded with Pi and 200 μM EGTA were still capable for ATP import (78 % of the corresponding EGTA-unaffected transport) (Additional file 5: Figure S5). Moreover, inhibition of ATP uptake into these EGTA-loaded liposomes by external EGTA as well as its (re)activation by 500 μM external Ca2+ were nearly identical when compared to standard AtAPC2-proteoliposomes lacking internal EGTA (Additional file 5: Figure S5).

Together, the obtained results indicate that ATP transport but not ADP or Pi transport of AtAPC2 essentially requires the presence of divalent cations and this requirement is independent of the N-terminal domain and thus not connected to carrier activation.

Plant APC2 can mediate Ca2+-transport in vitro

The observed Ca2+ and EGTA effects on AtAPC2 activity led us to the conclusion that Ca2+ might act as an important co-substrate in ATP transport. To verify the proposed capacity of AtAPC2 for ATP-Ca transport in the liposomal system we performed uptake studies with 20 μM [45Ca] and 100 μM non-labeled ATP. Preliminary analyses revealed that the read-out of the import rates was hampered due to the high degree of nonspecific [45Ca]-interaction with the phospholipids at the liposomal surface (causing high radioactive background values). However, reduction of these non-specific background counts by removal of the vast majority of [45Ca] from the liposomal surface was achieved by additional EGTA treatment of the vesicles subsequent to the uptake measurements (prior to vacuum filtration and washing). The correspondingly modified transport assay allowed determination of small but significant time dependent Ca2+ uptake by full-length and truncated AtAPC2.

Ca2+ uptake into Pi loaded vesicles (Fig. 6a and b, black rhombs) always exceeded the corresponding rates obtained with non-loaded proteoliposomes (Fig. 6a and b, gray squares) indicating that Ca2+ accumulation is directly connected to the antiport activity of the carrier. The full-length protein exhibits higher Ca2+ transport rates and also the back-ground values of the non-loaded vesicles are enhanced when compared to the truncated version (compare Fig. 6a and b). So far it cannot be discriminated whether - albeit EGTA treatment - a certain amount of Ca2+ still binds to the N-terminal domain of recombinant AtAPC2 or/and the functionality of the truncated protein is generally slightly impaired.

Fig. 6
figure 6

Determination of Ca2+ transport via AtAPC2. Time dependent uptake of [45Ca] via full-length AtAPC2 (a) and via N-terminally truncated AtAPC2 (b) reconstituted into Pi (black rhombs) and non-loaded liposomes (gray squares). (c) Effects of rising MgCl2 concentrations on [45Ca] transport into Pi loaded (dark gray bars) and non-loaded (light gray bars) AtAPC2 proteoliposomes. Transport media contained 20 μM [45Ca] and were additionally supplemented with 100 μM non-labeled ATP and the indicated concentrations of MgCl2. For determination of the Mg2+-effects on Ca2+ transport via AtAPC2 uptake was allowed for 10 min (given as nmol mg protein−1 h−1). Data represent mean values of three independent replicates. Standard errors are indicated

Full size image

Lastly, we analyzed effects of Mg2+ on Ca2+ import via recombinant AtAPC2. For this, Pi loaded and non-loaded AtAPC2 proteoliposomes were incubated in transport medium containing 20 μM [45Ca], 100 μM non-labeled ATP and increasing concentrations of Mg2+. [45Ca] import into phosphate loaded vesicles became significantly reduced by Mg2+ whereas the corresponding rates of the non-loaded vesicles remained more or less unaffected by Mg2+ addition (Fig. 6c). Quite high amounts of Mg2+ (200 μM) are required to cause approximately half maximal transport inhibition whereas 25-fold excess of Mg2+ completely blocks Ca2+ uptake. Because of the generally low [45Ca] transport rates of the truncated AtAPC2 reliable interpretation of the corresponding results obtained with this protein is complicated. Nevertheless, the tendency of Mg2+ impact on Ca2+ uptake generally resembles that of the full-length protein (Additional file 6: Figure S6). The obtained data suggest that Mg2+ competes with Ca2+ during ATP complex formation and thereby can reduce ATP-Ca availability and hence Ca2+-import via the reconstituted carrier.

Discussion

Transport capacities of plant APCs allow energy exchange as well as net adenine nucleotide provision

Diverse biological conditions, such as ATP-loading during mitochondrial biogenesis or physiological and environmental changes, require modulation of the mitochondrial adenine nucleotide pool size [9, 17, 18, 46]. During the past decades net influx or efflux of adenine nucleotides into or out of the organelle as well as the involved carriers have been well studied in mammals and yeast [9, 11, 12, 14–18, 46]. However, much less is known about these processes in plants.

It is quite obvious that also plant mitochondria have to adapt the adenine nucleotide concentration in the mitochondrial matrix in accordance to the respective metabolic demands. Already in the 1970s isolated corn and cauliflower mitochondria were shown to exhibit (carboxy)atractyloside insensitive (AAC independent) uptake of adenine nucleotides [26–28]. In the beginning, net import of ADP into plant mitochondria was identified to occur via exchange with Pi [31]. Later on, ADP transport was shown to be influenced by Mg2+ and Ca2+ and it was suggested that exogenous rather than endogenous Pi drives net ADP uptake [29]. These inconsistencies might be due to the fact that mitochondria harbor various carriers and enzymes directly or indirectly involved in adenine nucleotide transport and metabolism and that these proteins are differently affected by the respective test conditions and metabolic states of the organelle.

Arabidopsis thaliana encodes three MCF proteins (AtAPC1-3) that represent promising candidates for net adenine nucleotide transport. First of all, AtAPC1-3 exhibit significant amino acid similarities to APCs from animals or yeast and contain the characteristic N-terminal domain with EF-hand motifs (Additional file 7: Figure S7 and Additional file 8: Figure S8). Secondly, these proteins can compensate the growth defect of yeast ∆sal1p mutants inhibited in AAC mediated transport [32]. Thirdly, transport assays performed in this work with the reconstituted, recombinant carriers revealed that AtAPC1-3 act in a strict antiport mode (Fig. 1, Additional file 2: Figure S2); they can catalyze homo-exchanges of ATP and ADP as well as ATP/ADP hetero-exchange but most importantly also ATP and ADP hetero-exchange with Pi in vitro (Fig. 1, Additional file 2: Figure S2, Tables 1, and 3). The latter capacity was also shown recently in a study by Palmieri and coworkers that was published while this manuscript was in revision [47]. Based on the in vitro characteristics growth-restoration in the yeast complementation assay by the three AtAPC isoforms [32] can be attributed to their capacity for net adenine nucleotide supply (complementation of Sal1p activity) and/or for energy provision (complementation of AAC activity).

Plant mitochondria possess a high affinity ADP uptake system that is sensitive to AAC-specific inhibitors and a low affinity ADP uptake system that apparently does not involve AAC activity [30]. Biochemical characterization of single isoforms suggest that AAC proteins mediate the high affinity ADP transport [48] whereas APCs catalyze or contribute to the low affinity ADP transport (Table 3) [47].

Interestingly, APC genes show more or less ubiquitous expression with highest rates in growing tissues of enhanced mitochondrial propagation (Aramemnon, BAR eFP browser; [49, 50]). The recent work by Palmieri and coworkers showed that the promoter of Atapc1 exhibits enhanced activity when compared to the remaining two APC isoforms [47]. Moreover, expression of specific isoforms (Aramemnon, GENEVESTIGATOR [49, 51]) is induced by growth-promoting plant steroids (brassinosteroides) or in response to abiotic stressors, like hypoxia or phosphate limitation; conditions assumed to be associated with altered mitochondrial metabolism/respiration [45, 47, 52–54]. In future studies it will be interesting to determine whether specific developmental stages or stress situations characterized by enhanced or reduced APC expression correlate with the establishment or alteration of the mitochondrial adenine nucleotide pool.

Substrate preferences and impact of divalent cations on transport

The fact that recombinant AtAPC3 and AtAPC1 apparently prefer homo-exchanges of ATP and ADP over the corresponding hetero-exchanges with Pi (Fig. 1, Additional file 2: Figure S2) might be indicative of transport reduction due to unfavorable charge imbalances generated in the liposomes by the electrogenic hetero-exchange. Similar to net ATP uptake by yeast and mammalian mitochondria [11, 15, 16, 18] ATP transport of AtAPC1 and AtAPC3 is markedly stimulated by Mg2+ (Fig. 1, Table 1). This stimulation occurs during homo- and hetero-exchange and suggests that AtAPC1 and AtAPC3 generally prefer ATP-Mg2− over ATP4− as import substrate independent of the generation of charge imbalances.

In contrast to AtAPC1 and AtAPC3, rates of homo- and hetero-exchange of recombinant AtAPC2 are quite similar (Fig. 1, Additional file 2: Figure S2) and ATP uptake was only slightly enhanced by Mg2+ addition (Fig. 1c and d). These observations suggest that either a strong preference of AtAPC2 for Pi as exchange substrate compensates possible negative effects of the charge imbalance of ATP/Pi (and ADP/Pi) hetero-exchange or that hetero-exchange with Pi is not electrogenic at all. Interestingly, ATP transport of AtAPC2 was totally inhibited by EGTA or EDTA and could be restored by Mg2+ or Ca2+ (Fig. 5). This result strikingly argues for the requirement of divalent cations for ATP translocation. Whether this is due to their function as co-substrate and/or as effectors of the carrier protein cannot be unambiguously stated yet.

In contrast to our studies, Palmieri and coworkers investigated the capacity of ATP-Mg to act as export and not as import substrate and under those conditions ATP-Mg transport is rather unfavorable when compared to ATP [47]. Summarily, the current data therefore suggest that the plant APCs possess different substrate preferences at their exterior and interior side (Fig. 1, Table 1) [47].

Although Ca2+-dependent activity regulation of human and yeast APCs has been well known for a long time, first insights into the mechanistic principle were only gained recently. Sophisticated interaction studies with human SCaMC1 suggest that in absence of Ca2+ the quite flexible N-terminal domain caps the transmembrane part whereas Ca2+-binding turns the N-terminal domain into a more rigid state which leads to its dissociation and opening of the translocation pore [21, 22]. Superimposition of the corresponding regions in a structural alignment visualizes a high degree of conservation among the N-terminal domains of plant APCs and human SCaMC1 (Fig. 7). These structural similarities as well as computer based docking analyses (Additional file 8: Figure S8) suggest that the N-terminal domains of the plant APCs also interact with four Ca2+ ions. Moreover, amino acid sequence similarity to Sal1p and human SCaMC isoforms suggest that plant APCs are likewise regulated by Ca2+ (Additional file 7: Figure S7, Fig. 7).

Fig. 7
figure 7

Structural alignment of the N-terminal domains of AtAPC1-3 and human SCaMC1. Three-dimensional homology models of AtAPC1 (residues 34–189, green), AtAPC2 (residues 38–194, yellow) and AtAPC3 (residues 35–189, orange). N-terminal domains were built using HHPred server and Modeller based on the crystal structure of the Ca2+-binding N-terminal domain of human SCaMC1 (blue; PDB ID: 4N5X) in complex with four calcium ions (gray spheres). The sequence alignment followed by a structural superimposition of the models was carried out using PyMOL (version 1.3)

Full size image

The fact that reconstituted APC isoforms from human [12] and A. thaliana were already active without extra Ca2+-addition led to the assumption that Ca2+ contaminations in the buffer media were sufficient for carrier activation. Because increase in Ca2+-concentrations resulted in transport stimulation of all recombinant AtAPC isoforms one might conclude that under the reconstitution conditions a mix of active and non-active carries occurs and addition of Ca2+can thus activate additional carriers (Fig. 2). However, the rates of Ca2+-stimulation were not identical and varied depending on the kind of substrate transported (Table 2).

Assuming that Ca2+ exclusively operates in carrier activation by displacement of the N-terminal domain from the translocation pathway we would expect the same degree of (i) Ca2+-dependent transport stimulation, (ii) Vmax increase (proportional to the amount of functional carriers), and (iii) transport reduction by Ca2+-depletion (with EGTA) independent of the exchanged substrates. Moreover, truncation of the N-terminal domain should cause constantly active carriers that are no longer influenced by Ca2+. However, the data obtained in this work suggest that this is not the case. We therefore hypothesize that in the in vitro system ATP-Ca acts as substrate of the plant APCs and is even favored over ATP-Mg or free ATP. By contrast, ADP-Ca seems to be rather discriminated against when compared to free ADP. Ca2+-induced alterations of the apparent transport affinities most likely reflect these specific substrate preferences of the respective APC isoforms e.g. higher preference for ATP-Ca (when compared with the Mg-complexed or free ATP) and lower preference of ADP-Ca (when compared to free ADP) (Table 3). Accordingly, Ca2+ complexation of ATP enhances and that of ADP reduces the amount of favored substrates and by this the respective transport capacity of the reconstituted protein. It is also imaginable that in the liposomal system, Ca2+ co-transport with ATP prevents charge accumulation of the ATP/Pi hetero-exchange and with ADP3− (ADP-Ca1−) enhances the imbalance caused by the ADP/ATP hetero-exchange. In addition, effects of EGTA, EDTA, Mg2+ and Ca2+ on ATP transport inhibition, stimulation or reactivation suggest a competition between these cations during complex formation and provide further evidences for ATP-Ca as a potential in vitro substrate of recombinant plant APCs (Table 2 and Figs. 3, 4, 5). We conclude that the influence of Ca2+ on transport by the reconstituted APCs is a consequence of diverse factors, such as substrate preferences, charge accumulation/compensation and competition with Mg2+ during complex formation.

Transport characteristics obtained with AtAPC2 and the N-terminally truncated version support the assumption that ATP-transport stimulation by Ca2+ is not (or not exclusively) caused by activation of previously inactive (Ca2+-free) carriers. ATP transport of both, the full-length carrier and the truncated version, can be stimulated by Ca2+ and inhibited by EGTA whereas ADP transport was not significantly affected (Figs. 3 and 4). These results verify that solely ATP but not ADP transport activity is highly dependent on the presence of Ca2+ and that removal of this cation did not cause carrier deactivation in general. The ineffectiveness of EGTA in the inhibition of total transport activity is surprising. The possibility that plant APCs are generally not regulated in a Ca2+-dependent manner is apparently not applicable. Important structural similarities of the plant, yeast and mammalian isoforms are suggestive for a similar regulatory principle but most importantly, a corresponding regulation could be demonstrated in the recent study by Palmieri and coworkers [47]. It remains unclear whether in our in vitro system the functionality of the N-terminal domain of the recombinant AtAPC2 is somehow impaired or its affinity for Ca2+ is higher than that of EGTA. However, the possibility that insight-out orientation of reconstituted AtAPC2 and hence inaccessibility of the N-terminal domains caused ineffectiveness of EGTA in transport inhibition can be ruled out since proteoliposomes internally loaded with EGTA were still capable to import ATP in exchange with Pi (Additional file 5: Figure S5).

The fact that external but not internal EGTA caused inhibition of AtAPC2 mediated ATP import in exchange with Pi demonstrates that ATP but not Pi transport requires the presence of Ca2+ (or divalent cations). Moreover, this observation also demonstrates that the chelator at the liposomal interior is apparently physically separated from Ca2+ at the exterior (at least during the analyzed time span) which indicates that both, EGTA and Ca2+, do not pass the lipid barrier freely.

Notwithstanding or even because of the missing Ca2+-dependent regulation, we were able to identify the in vitro function of Ca2+ as co-substrate with the applied system.

Although uptake studies with α[32P]-ATP provided evidence for a possible ATP-Ca transport it would still have been imaginable that Ca2+ stimulates transport of unchelated ATP and impedes ATP-Mg transport in a different way. However, the specifically adapted uptake assay using [45Ca] provided a direct proof that ATP-Ca is de facto transported via reconstituted (Fig. 6). Time dependent uptake of [45Ca] via AtAPC2 is tightly connected to its antiport activity because Pi loaded proteoliposomes accumulated higher amounts of [45Ca] than non-loaded vesicles. Competition experiments further verified that ATP-Ca transport is favored over ATP-Mg transport in vitro since quite high concentrations of Mg2+ are required to reduce ATP-transport associated Ca2+ uptake (Fig. 6c, Additional file 6: Figure S6). When compared to full-length AtAPC2 the N-terminally truncated carrier shows reduced Ca2+ import capacity (Fig. 6b). Whether absence of the N-terminal domain affects transport activity directly or rather indirectly (via impairments in refolding and membrane insertion) cannot be deduced from these experiments.

Further studies with the reconstituted proteins as well as with transgenic APC plants and isolated mitochondria will be required to completely decipher, evaluate and compare in vitro and in vivo characteristics of APC proteins. Moreover, it will be interesting to determine the stoichiometry of the ATP and Ca2+ co-transport. Preliminary estimation suggests that these substrates are not transported in a 1:1 stoichiometry. However, in this context it is important to mention that uptake assays had to be adapted to make Ca2+ transport determination feasible and furthermore that Ca2+ and Mg2+ contaminations of the media have to be considered. Therefore, in future studies we want to further optimize Ca2+-transport measurements in liposomes and intent to decipher the impact of divalent cations on AtAPC1-3 function in vivo.

Can ATP-Ca transport via plant APCs occur in vivo?

SCaMCs as well as yeast Sal1p seem to prefer ATP-Mg whereas our initial studies indicate that at least one of the AtAPC isoforms clearly favors ATP-Ca over both, ATP-Mg and ATP, as import substrate in the liposomal system. Due to the high structural similarity to ATP-Mg it is - from a biochemical point of view - not surprising that at least certain APCs can in principle accept ATP-Ca as substrate in vitro. However, the intriguing question arises whether ATP-Ca formation and correspondingly APC mediated Ca2+-transport can and will take place under physiological conditions. Generally, ATP-Ca formation is a rather unlikely phenomenon in plant cells. The concentration of free Ca2+ is usually low when compared to Mg2+, which represents a dominating divalent cation and also is Mg2+ favored over Ca2+ in ATP-complex formation. However, one could envision specific situations that might support possible ATP-Ca formation in close proximity to the carrier.

Although plant mitochondria contribute to Ca2+ storage, the majority of internal Ca2+ is probably transiently fixed as amorphous phosphate precipitate and thus the resting concentration of free Ca2+ in the matrix only slightly exceeds that of the cytosol (200 nM vs. 100 nM) [55–57]. Moreover, due to high Mg2+ concentrations within plant mitochondria ATP is nearly completely complexed with Mg2+, which argues against any potential ATP-Ca formation in the matrix [45]. Although lower Mg2+ levels in the cytosol increase the accessibility of free ATP, it is unclear whether conditions or microdomains of high Ca2+ availability at the mitochondrial surface might allow ATP-Ca formation [55, 58–63]. In the liposomal system Ca2+ uptake via AtAPC2 was low and completely blocked by 25-fold excess of Mg2+. If these characteristics (the biochemical properties in combination with a high Mg2+ to Ca2+ ratio next to the carrier) also represent the in vivo situation, ATP-Ca transport via plant APCs is highly unlikely to occur.

Although, a direct role of plant APCs in ATP-Ca transport is therefore arguable, recent data suggest an indirect function of a mammalian isoform in Ca2+ translocation. SCaMC3 was shown to physically interact with the (low affinity) Mitochondrial Calcium Uniporter (MCU) and lack of SCaMC3 apparently decreases ATP and Ca2+ import into mitochondria [24, 64]. Accordingly, SCaMC3 was supposed to represent an important component of the mitochondrial Ca2+ uptake system, a supercomplex formed by channels and carriers in microdomains for enhanced Ca2+-sensitivity [64]. Whether certain plant APC isoforms fulfill a function related to that described for SCaMC3 is unclear, however, physical proximity to proteins involved in Ca2+ release might be advantageous to guarantee fast Ca2+-dependent activation and response of plant APCs.

Conclusions

Determination of the biochemical characteristics of three putative APC isoforms from A. thaliana in the liposomal system revealed that the recombinant carriers mediate ATP, ADP and phosphate exchange. Accordingly, plant mitochondria harbor a subset of carriers capable of net adenine nucleotide translocation, however in contrast to yeast and mammalian orthologs they show no high preference for ATP-Mg as import substrate. Surprisingly, we instead obtained evidence for a possible ATP-Ca transport by the reconstituted plant APCs in the liposomal context but it is arguable that physiological Mg2+ and Ca2+ concentrations most likely prevent ATP-Ca formation and its subsequent transport in vivo. Although we were not able to detect EF-hand based Ca2+-dependent carrier regulation, this was shown recently to exist in plant APCs [47]. Summarily, the current data suggest that low Ca2+ concentrations regulate activity of plant APCs via EF-hands of the N-terminal domain whereas high Ca2+ concentrations can induce its own transport as co-substrate of ATP in vitro. While this study deepens our knowledge about mitochondrial net nucleotide transport of plants it also gives rise to new intriguing questions. In the future, it is important to investigate the in vivo function of plant APCs and the impact of divalent cations on the corresponding transport.