Nonmitochondrial ATP/ADP transporters accept phosphate as third substrate.

Chlamydiales and Rickettsiales as metabolically impaired, intracellular pathogenic bacteria essentially rely on "energy parasitism" by the help of nucleotide transporters (NTTs). Also in plant plastids NTT-type carriers catalyze ATP/ADP exchange to fuel metabolic processes. The uptake of ATP4-, followed by energy consumption and the release of ADP3-, would lead to a metabolically disadvantageous accumulation of negative charges in form of inorganic phosphate (Pi) in the bacterium or organelle if no interacting Pi export system exists. We identified that Pi is a third substrate of several NTT-type ATP/ADP transporters. During adenine nucleotide hetero-exchange, Pi is cotransported with ADP in a one-to-one stoichiometry. Additionally, Pi can be transported in exchange with solely Pi. This Pi homo-exchange depends on the presence of ADP and provides a first indication for only one binding center involved in import and export. Furthermore, analyses of mutant proteins revealed that Pi interacts with the same amino acid residue as the gamma-phosphate of ATP. Import of ATP in exchange with ADP plus Pi is obviously an efficient way to couple energy provision with the export of the two metabolic products (ADP plus Pi) and to maintain cellular phosphate homeostasis in intracellular living "energy parasites" and plant plastids. The additional Pi transport capacity of NTT-type ATP/ADP transporters makes the existence of an interacting Pi exporter dispensable and might explain why a corresponding protein so far has not been identified.

Most organisms possess the capacity to resynthesize the fundamental energy currency ATP by fusion of ADP and Pi. Generally, in eukaryotes the major part of energy is produced in specialized organelles, the mitochondria. Mitochondrial ADP/ATP carriers (AACs)2 mediate the export of newly synthesized ATP in strict counter-exchange with cytosolic ADP and therefore provide energy to the cellular metabolism (1). Plants additionally generate high amounts of ATP during photosynthesis in chloroplasts. However, under conditions of limiting or missing photosynthetic activity, plant plastids depend on external energy supply (2–4). Specific nucleotide transporters (NTTs) located in the inner plastid envelope membrane mediate the required energy import (5). These transporters structurally, functionally, and phylogenetically differ from mitochondrial AACs. They catalyze the import of cytosolic ATP in exchange with stromal ADP, are monomers consisting of 12 predicted transmembrane helices, and are related to the functionally heterogeneous group of bacterial NTTs (5).

Although most prokaryotic organisms are able to regenerate ATP and therefore are considered as energetically self-sustaining, the obligate intracellular living bacterial orders Chlamydiales and Rickettsiales are impaired in energy and nucleotide synthesis or even completely lost the corresponding pathways (6–8). Therefore, these bacteria, which comprise important human pathogens (9, 10), essentially rely on nucleotide and energy import. Bacterial NTTs catalyze the required import of a broad range of nucleotides and NAD or facilitate the counter-exchange of ATP and ADP (5, 11–15). The latter process has been termed “energy parasitism” and obviously is of high importance for the survival of rickettsial and chlamydial cells (5, 16–18).

Although import measurements on intact Escherichia coli cells expressing the corresponding proteins allowed characterization of many bacterial and plastidial NTTs (12–15, 19–24), a very important physiological question is still not clarified. The uptake of ATP4- in exchange with ADP3- in absence of a concerted Pi export would result in a charge difference and a phosphate imbalance in the bacterial cell. In mitochondria, phosphate carriers metabolically cooperate with AACs because they provide Pi for ATP synthesis (25). Similarly, it was assumed that NTT-type ATP/ADP transporters cooperate with phosphate exporters to guarantee phosphate homeostasis in the bacterium or plastid. However, a Pi exporter interacting with ATP/ADP transporters is not known in “energy parasites” or plant plastids. Bacterial and plant phosphate transport systems rather facilitate Pi import or the counter-exchange of Pi and phosphorylated compounds and therefore do not allow net Pi export (26–29). Furthermore, the newly identified plastidial (proton-driven) phosphate transporters are not preferentially expressed under conditions or in tissues that require ATP provision to the plastid (30, 31).

Recently, we succeeded in the purification of the first recombinant NTT from Protochlamydia amoebophila (PamNTT1), a parachlamydial endosymbiont of the protist Acantamoeba (32). The functional reconstitution of the highly pure PamNTT1 into artificial lipid vesicles for the first time allowed the biochemical characterization of a representative nonmitochondrial ATP/ADP transporter unaffected by the complex metabolic situation of the bacterial cell. We demonstrated that in contrast to mitochondrial AACs, PamNTT1 catalyzes a membrane potential independent, electroneutral adenine nucleotide hetero-exchange (32, 33). The latter could argue for a cotransport of a counterion compensating for the electrogenic ATP4-/ADP3- exchange.

Here, we investigated possible ions accompanying ATP or ADP transport. Interestingly, we uncovered that PamNTT1 and also rickettsial and plastidial ATP/ADP transporters accept an additional important substrate, which is Pi. We performed a comprehensive characterization of the Pi transport and gained new insights into the transport properties of ATP/ADP transporters.

EXPERIMENTAL PROCEDURES

Heterologous Synthesis of Selected NTTs and Mutant Proteins in E. coli—The heterologous synthesis of PamNTT1 and of two ATP/ADP transporters from the Rickettsia-related bacteria Caedibacter caryophilus (CcNTT) and Holospora obtusa (HoNTT) was performed on the basis of existing pET16b constructs (13, 22), which were transformed into BLR cells (Merck). Heterologous expression of the full-length cDNA encoding the plastidial ATP/ADP transporter from Arabidopis thaliana (AtNTT1) led to very low amounts of recombinant transporters in BLR cells (data not shown). However, the truncation of the N-terminal extension (the putative leader peptide of 87 amino acid residues) and application of Rosetta2™ (DE3) pLysS expression cells resulted in an increased synthesis of AtNTT1. The truncated AtNTT1 sequence was amplified from the expression construct (34) by PCR using Pfu polymerase and the following oligonucleotides: AtNTT1shortNdeI_sense, 5′-GGCCGCGGCTCATATGGACGGAGCTG-3′, and the standard primer T7-term. The PCR construct contained one NdeI restriction site inserted by the sense primer and the XhoI restriction site of the plasmid. The amplificate was cloned into an appropriately (NdeI/XhoI) restricted pET16b vector.

For generation of K446 mutant proteins, the PamNTT1 encoding sequence (inserted in pET16b) was modified by site-directed mutagenesis using the QuikChange site-directed mutagenesis kit (Stratagene). The triplet encoding lysine 446 (AAA) was substituted by the triplets CGC, CAG, and GAA, respectively, by PCR. To eliminate possible PCR-caused mistakes in the amplified expression vector, we excised the mutated sequences by using BamHI (13) and inserted them into the correspondingly prepared original pET16b vector. The correctness of all constructs was proven by complete sequencing (NanoBioCenter, TU Kaiserslautern, Germany).

Heterologous expression was conducted as previously described (32). Briefly, transformed bacterial cells were grown at 37 °C in TB medium under selective conditions. At an A600 of ∼0.5, heterologous expression was induced by the addition of isopropyl-β-d-thiogalactopyranoside (1 mm). Four hours post-induction the cells were harvested by centrifugation (5,000 × g, 10 min, 20 °C), resuspended in buffer medium (1 mm EDTA, 15% glycerol, 10 mm Tris, pH 7.0), and immediately frozen in liquid N2.

Purification and Reconstitution of Heterologously Expressed NTTs—Enrichment of E. coli membranes and purification and reconstitution of heterologously expressed NTTs were conducted as described by Trentmann et al. (32). In brief, frozen cells were disrupted by thawing and sonication. The membranes were fractionated by different centrifugation steps and solubilized with 1% N-dodecyl-β-maltoside (Glycon Biomedicals, Luckenwalde, Germany). Nonsolubilized membrane proteins were removed by ultracentrifugation. The recombinant, histidine-tagged proteins were purified by immobilized metal affinity chromatography with nickel-Sepharose 6 Fast Flow (GE Healthcare) according to the manufacturer's instructions. Buffers used for washing and elution contained 0.5% N-dodecyl-β-maltoside. To rule out Pi incorporation during reconstitution, we omitted Pi not only in the elution buffer but also in the last washing step. Additionally, the remaining liquid at the immobilized metal affinity chromatography material and in the column tip was removed thoroughly prior to elution. 50 μl of the eluate were reconstituted into liposomes loaded with the given interior substrates (10 mm ATP or ADP or 5 mm Pi). Protein concentrations of the purified NTTs were quantified in a Bio-Photometer (Eppendorf, Hamburg, Germany) according to the method described by Bradford (35). Purity of the recombinant NTTs was analyzed by SDS-PAGE (3% stacking and 12% separating gel) as described by Laemmli (36).

Import of Radioactively Labeled Substrates into Proteoliposomes—Uptake studies were performed as previously described (32). For transport measurements, proteoliposomes were incubated at 30 °C in presence of the given concentrations of α-32P-labeled nucleotides or radioactive Pi [32P] (PerkinElmer Life Sciences). Optionally, import was conducted in the presence of the given concentrations of nonlabeled nucleotides or Pi. Removal of substrates by anion exchange chromatography (Dowex 1 × 8 Cl, 200–400 mesh; Sigma) was used to terminate import. Liposomes were eluted, and internal radioactivity was quantified in a scintillation counter (Canberra-Packard, Rüsselsheim, Germany).

RESULTS

P—The observation that PamNTT1 mediated adenine nucleotide exchange is electroneutral (32) was surprising. This carrier obviously transports a counterion that compensates for the charge difference of the hetero-exchange. The catalytic activity of bacterial and plastidial ATP/ADP transporters in contrast to bacterial proton-driven NTTs is not affected by the proton gradient across the membrane (12–15, 19–23). This argues against a cotransport of ATP and protons.

To detect possible cotransported ions, we analyzed the influence of diverse salt compositions on PamNTT1-mediated nucleotide transport in the liposomal system. We identified that the comparably low rates of ADP import into ATP loaded vesicles (ADPim/ATPex) were increased by 500 μm external Pi (3.7 times) and also by its structural analogue, arsenate (2.6 times) (supplemental Table S1). Higher concentrations (2 mm) of latter compounds further increased the stimulatory effects. The addition of other components had lower stimulatory or even inhibitory influences on the corresponding transport (supplemental Table S1).

In a previous study we demonstrated that only those reconstituted PamNTT1 proteins, which were inserted in the native orientation (right side out), displayed catalytic activity (32). To mimic the presence of Pi in the bacterium, we tested whether Pi application at the liposomal interior also influences adenine nucleotide hetero-exchange. When compared with the control (0 mm luminal Pi), the addition of internal Pi stimulated ATP import in exchange with ADP (ATPim/ADPex) ∼4-fold, whereas ADPim/ATPex transport was not highly affected by internal Pi application (Fig. 1). Accordingly, not only external but also internal Pi supports ATP/ADP exchange. However, stimulation of the hetero-exchange exclusively occurred when ADP and Pi were present at the same side of the liposome (ADP plus Pi) (supplemental Table S1 and Fig. 1).

FIGURE 1.

Influence of the presence of P ATPim/ADPex transport and ADPim/ATPex transport was analyzed in absence (light gray bars, set to 100%) and in presence of internal Pi (dark gray bars, calculated according to the transport in absence of Pi). For this, PamNTT1 was reconstituted into liposomes loaded with 10 mm ADP or ATP or into vesicles containing 5 mm Pi plus 10 mm ADP or 5 mm Pi plus 10 mm ATP. External nucleotides and Pi were removed by gel filtration. Import of 50 μm α-32P-labeled ATP or ADP was allowed for 5 min and stopped by anion exchange chromatography. The data represent net values calculated by subtraction of uptake into vesicles lacking counter-exchange nucleotides and are the means of three independent experiments. Standard errors are displayed. For better comparison the presence of Pi and adenine nucleotides at the liposomal interior and exterior is graphically displayed.

It had been suggested that Pi plays a regulatory role in ATP/ADP exchange across the rickettsial membrane (37). This based on the observation that high Pi concentrations marginally enhanced ATP but highly stimulated ADP import into isolated bacteria cells. Furthermore, only in the presence of Pi, ADP was able to compete with ATP for import, leading the authors to the assumption that Pi might increase (at least) the affinity of the rickettsial ATP/ADP transporter for ADP (37). To investigate the role of Pi in NTT-mediated adenylate exchange in more detail and separated from the complex cellular metabolism, we analyzed the effect of rising external Pi concentrations on adenine nucleotide import into PamNTT1-proteoliposomes loaded with interior Pi plus the given counter-exchange nucleotides.

Changes in external Pi availability had low impact on ATP homo-exchange parameters and only slightly affected ADP homo-exchange (4-fold increased Vmax at 5 mm external Pi) (data not shown). In contrast to the homo-exchanges, the hetero-exchanges were substantially influenced by Pi addition (Fig. 2). As observed in the effector analysis (supplemental Table S1) ADPim/ATPex hetero-exchange was significantly stimulated by exterior Pi. The highest rise in stimulation was obtained between 50 μm and 2 mm Pi (Fig. 2, dark gray bars). Increasing the external Pi concentration above 2 mm caused a further but comparably slight stimulation of the ADP import rates and suggests the beginning of a Pi saturation phase. Determination of the kinetic parameters revealed that the stimulatory influence of external Pi on ADPim/ATPex exchange can be traced back to an enhanced affinity (from 0 to 5 mm: 14-fold lower K) and an increased Vmax (from 0 to 5 mm: 3-fold higher Vmax) of PamNTT1 for ADP.

FIGURE 2.

Influence of rising exterior P PamNTT1 was reconstituted into liposomes loaded with 5 mm Pi plus 10 mm of the given nucleotides. The effects of rising external Pi concentrations on ATP import into ADP plus Pi-loaded proteoliposomes (light gray bars) and on ADP import into ATP plus Pi-loaded vesicles (dark gray bars) were analyzed. Import of 50 μm α-32P-labeled ATP or ADP was stopped after 5 min. The applied external Pi concentrations (in mm) are displayed at the x axis. K and Vmax values (in parentheses) were calculated for adenine nucleotide hetero-exchanges in the absence (0 mm) and in the presence of 5 mm external Pi, respectively, and are given above the bars of the corresponding exchanges. The data represent net values calculated by subtraction of uptake into vesicles lacking counter-exchange nucleotides and are the means of three independent experiments. The error bars are given. Standard errors of the kinetic parameters were below 12% and of ADPim/ATPex transport in absence of external Pi (0 mm) were at ∼20%. In the graphic depicting the presence of Pi and nucleotides at the liposomal interior and exterior, the radioactively labeled nucleotides are marked with asterisks.

Concentrations of more than 2 mm Pi in the transport medium led to a substantial reduction of the ATPim/ADPex transport rates. The Vmax of this exchange was high in the presence as well as in the complete absence of external Pi, and thus the lowered import rates were mainly caused by a decreased affinity of PamNTT1 for ATP import (from 0 to 5 mm: 27-fold lower K) (Fig. 2).

It has to be mentioned that the transport rates of optimal ATPim/ADPex exchange (0 mm external and 5 mm internal Pi, ∼1000 nmol mg protein-1 5 min-1) were much higher than that of Pi stimulated ADP import into ATP plus Pi-loaded vesicles (5 mm external and 5 mm internal Pi, ∼350 nmol mg protein-1 5 min-1) (Fig. 2). In this context it might be assumed that Pi in the vesicle lumen hindered optimal ATP export. However, depletion of luminal Pi did not result in a further relevant stimulation of ADPim/ATPex exchange (data not shown; see also ADPim/ATPex in Fig. 1). The opposed effects of external Pi on ADPim/ATPex versus ATPim/ADPex exchange could be explained by (i) a cotransport of ADP and Pi and by (ii) a possible competition of Pi with ATP during hetero-exchange.

P—To analyze whether PamNTT1 is able to transport Pi, we performed import studies using radioactively labeled Pi (50 μm 32P). Pi import was conducted with proteoliposomes loaded with interior Pi, Pi plus ADP, or Pi plus ATP. In absence of exterior nucleotides no Pi uptake into the vesicles was measurable when only Pi, or Pi plus ATP were internally present (Fig. 3, white circles and black diamonds). To our surprise Pi was imported into liposomes loaded with Pi plus ADP (Fig. 3, gray squares). In a similar experiment we tested the effect of externally added ADP (Fig. 3) or ATP (Fig. 3) on Pi transport. Application of external ADP did not markedly affect the rates of Pi import into Pi plus ADP loaded liposomes (Fig. 3, compare , gray squares) but stimulated Pi import into liposomes containing solely Pi, or Pi plus ATP (Fig. 3, compare , white circles and black diamonds). External ATP led to no or to comparably low Pi import (compare the import rates of Fig. 3). The latter effect further argues for a possible competition of Pi and ATP for import.

FIGURE 3.

Analysis of P Time-dependent import of 50 μm radioactive Pi into PamNTT1-proteoliposomes loaded with 5 mm Pi or 5 mm Pi plus the given adenine nucleotides (10 mm). Pi import was allowed for the given time spans and stopped by anion exchange chromatography. Pi uptake studies were performed in the absence of external nucleotides (A), in the presence of 50 μm nonlabeled external ADP (B), or in the presence of 50 μm nonlabeled external ATP (C) to investigate a possible influence of external adenine nucleotides on Pi transport. Pi uptake into proteoliposomes loaded with 5 mm Pi (open circle), 5 mm Pi plus 10 mm ADP (gray square), or 5 mm Pi plus 10 mm ATP (black diamond). The data are the means of at least three independent experiments. Standard errors are displayed. The insets summarize the applied substrate conditions at the liposomal interior and exterior. Radioactive Pi is marked with an asterisk.

Our results show that Pi indeed is a substrate of PamNTT1 but only when ADP is present at least at one side of the liposomal membrane. The observation that Pi transport occurs also in the absence of adenine nucleotide counter-exchange (Fig. 3) suggests that PamNTT1 catalyzes, in addition to a possible Pi plus ADP cotransport in exchange with ATP, an ADP-dependent Pi homo-exchange. To provide further evidence for a Pi homo-exchange, we analyzed Pi import into ADP-loaded liposomes lacking luminal Pi. The measured Pi import rates were ∼10-fold lower than the corresponding rates obtained with liposomes containing ADP plus Pi (data not shown). This is a further argument supporting the postulated Pi homo-exchange. To analyze whether Pi homo-exchange is accompanied by a simultaneous unidirectional translocation of ADP, we additionally investigated [α-32P]ADP import into proteoliposomes solely loaded with Pi. Because in the absence of counter-exchange nucleotides Pi but not ADP import occurred (data not shown), we concluded that ADP-dependent Pi homo-exchange is not associated with a unidirectional ADP transport.

PamNTT1 Catalyzes P—In a further study we analyzed ADP and Pi transport in more detail to address two important questions. First, is PamNTT1 capable of transporting Pi as cosubstrate of ADP during hetero-exchange, and second, what is the stoichiometrical ratio of the Pi plus ADP cotransport.

ATP import in strict stoichiometrical exchange with one ADP plus one Pi (H2PO-4, with one negative charge) would be electroneutral. In this context it is important to mention that under the applied conditions an important part of Pi carries one negative charge.

Unfortunately, determination of the ADP plus Pi export parameters of the reconstituted PamNTT1 was impossible. We were not able to measure the slight decrease of the high internal label. Furthermore, traces of liposomal contaminations hampered the determination of exported radioactivity. However, we demonstrated above that exterior Pi stimulates ADPim/ATPex transport (supplemental Table S1 and Fig. 2, black bars) and that similarly interior Pi stimulates ATPim/ADPex transport (Fig. 1). Therefore, it seems justified to assume that Pi acts as cosubstrate of ADP during hetero-exchange independent of the transport direction (ADPim+Piim/ATPex or ATPim/ADPex+Piex). Accordingly, we investigated the coimport of Pi and ADP.

For this, import measurements were performed with radioactively labeled ADP in the presence of nonlabeled Pi as well as with labeled Pi in the presence of nonlabeled ADP. One set of experiments was done with liposomes containing ATP plus Pi, and one was done with vesicles loaded with solely ATP (to disconnect the ADP-dependent Pi homo-exchange from coimport with ADP). For determination of the ADP to Pi coimport stoichiometry, we calculated Pi import in relation to ADP import (ADP import was set as 1).

The presence of Pi in the liposomal lumen led to a high Pi to ADP import ratio [4.03 (± 0.45) Pi to 1 ADP] (supplemental Fig. S1). Therefore, Pi homo-exchange substantially exceeds coimport with ADP. However, depletion of luminal Pi led to a significantly reduced Pi homo-exchange and allowed the calculation of a nearly one-to-one stoichiometry for Pi and ADP coimport (1.21 (± 0.16) Pi to 1 ADP) in exchange with interior ATP (supplemental Fig. S1). It has to be mentioned that Pi contaminations in the liposomal lumen or a marginal unidirectional Pi transport might have caused a slight overestimation of the exact Pi-to-ADP ratio.

Determination of the Kinetic Properties and Identification of the Phosphate Binding Center of PamNTT1—As demonstrated above, already low external ATP concentrations inhibited Pi import (Fig. 3, compare ), whereas high exterior Pi concentrations were required to substantially reduce ATPim/ADPex transport (Fig. 2). These characteristics suggest that the two anions, ATP and Pi, compete for uptake and that ATP is the preferred import substrate. On the other hand, Pi exhibits positive effects on ADP transport (import and export) rather than competitive interference (Figs. 1 and 2). Accordingly, Pi and ADP simultaneously fit into the binding center, whereas Pi probably interacts with the NTT domain, which is otherwise occupied by the γ-phosphate of ATP.

In a previous study we showed that mutations of a lysine residue at position 527 in the plastidial ATP/ADP transporter from A. thaliana (AtNTT1) reduced ATP transport to a higher extent than ADP transport (34). Remarkably, this lysine residue is conserved in all plastidial and bacterial ATP/ADP transporters (supplemental Fig. S2). To test whether this amino acid residue is not only important for ATP but also for Pi transport, we generated different PamNTT1 mutant proteins and analyzed their biochemical properties in the liposomal system. This was mandatory because E. coli possesses endogenous Pi transporters that obscure NTT mediated Pi translocation. We substituted the positively charged lysine by a positively charged arginine (K446R), by the neutral glutamine (K446Q), or by the negatively charged glutamate (K446E) to investigate a correlation between the charge of the amino acid residue 446 and the kinetic properties (K and Vmax values) of PamNTT1.

The biochemical parameters of unmodified PamNTT1 for ATP and ADP transport were in the same range as previously described (32). Furthermore, the reconstituted wild type protein exhibited moderate affinities (K in the range of 140 μm) and high Vmax values (∼9–16 μmol mg protein-1 h-1) for ADP-dependent Pi transport (Table 1). Application of internal ATP slightly and of external ATP highly reduced the affinity of PamNTT1 for Pi uptake.

TABLE 1

Comparison of the kinetic constants of reconstituted Kinetic parameters of nucleotide or Pi exchanges (exterior substrates/interior substrates) mediated by reconstituted PamNTT1 and the three mutant proteins (PamNTT1-K446R, PamNTT1-K446Q, and PamNTT1-K446E) were determined by application of rising concentrations of the respective labeled import substrates (5–1,500 μm). The used proteoliposomes contained 5 mm internal Pi and the given nucleotides (10 mm). Kinetic parameters of Pi import were analyzed in the absence or presence of 50 μm nonlabeled external nucleotides (added external nucleotides in squared brackets). K values are given in μm, and Vmax values (in parentheses) are in μmol mg of protein–1 h–1, respectively. Transport was allowed for time spans in the linear phase of the corresponding import at 50 μm. The data are the means of three independent experiments. Standard errors of the low affinity and low velocity imports were below 20% and for the remaining imports below 14%.

Exchanges	PamNTT1	PamNTT1 mutant proteins
		K446R	K446Q	K446E
ATP/ATP + Pi	95 (27.6)	123 (0.5)	187 (6.5)	620 (2.3)
ATP/ADP + Pi	8.6 (31.3)	66 (1.0)	274 (7.2)	552 (2.4)
ADP/ATP + Pi	1401 (5.5)	179 (0.3)	68 (4.5)	29 (1.6)
ADP/ADP + Pi	120 (7.1)	96 (0.7)	44 (4.0)	159 (15.4)
Pi/ADP + Pi	125 (13.6)	271 (2.1)
Pi[+ADP]/Pi	154 (10.0)	152 (0.5)
Pi[+ADP]/ADP + Pi	139 (15.9)	220 (1.9)	544 (0.4)
Pi[+ADP]/ATP + Pi	295 (9.0)	275 (0.7)	550 (0.4)
Pi[+ATP]/ADP + Pi	506 (13.3)	598 (2.0)

The conserved exchange of lysine 446 (K446R) led to a remarkable reduction of the Vmax, in particular of the ATP homo-exchange (Table 1). Furthermore, the affinity for ATP import in exchange with interior ADP plus Pi was reduced, and that for ADP import into ATP plus Pi-loaded vesicles was enhanced. Nevertheless, the basic properties of the mutant protein PamNTT1-K446R, such as the affinities for (ADP-dependent) Pi import or ATP and ADP import during homo-exchange, as well as the preference of the ATPim/ADPex transport over ADPim/ATPex exchange, still resemble that of the unmodified PamNTT1.

In comparison with the wild type protein, the mutant protein with the neutral substitution (PamNTT1-K446Q) exhibited reduced affinities (and Vmax values) for ATP import, whereas the affinities for ADP import, in particular in exchange with interior ATP, were significantly enhanced (Table 1). These changed biochemical characteristics led to a general adjustment of the import rates, the ATP homo-exchange equals ATPim/ADPex transport, and also the high difference between ADP homo-exchange and ADPim/ATPex transport of the wild type protein was reduced in PamNTT1-K446Q (Table 1 and supplemental Fig. S3). Because Pi transport of PamNTT1-K446Q was nearly completely diminished (supplemental Fig. S3), determination of the corresponding kinetic parameters was hampered. Solely in the presence of external ADP and internal Pi plus adenine nucleotides, very low affinities and Vmax values for Pi import were calculable (Table 1).

Interestingly, the replacement of the positively charged lysine by the negatively charged glutamate completely blocked ADP-dependent Pi uptake (supplemental Fig. S3 and Table 1). In comparison with the unmodified PamNTT1 and the mutant PamNTT1-K446Q, the mutation in PamNTT1-K446E led to a further reduction of the K and Vmax values of the ATP import (Table 1). Furthermore, the affinity for ADP import into ATP-loaded proteoliposomes was enhanced, whereas the Vmax of this exchange was reduced. The moderate affinity for ADP import into ADP plus Pi-loaded vesicles, however, was accompanied by the highest Vmax. Generally, the observed differences in the biochemical characteristics between the wild type protein PamNTT1 and the mutant PamNTT1-K446Q are more pronounced in the mutant PamNTT1-K446E. Our analyses of the mutant proteins revealed that a positively charged amino acid residue at position 446 is required for proper Pi import, for the preference of ATPim/ADPex transport, and for the discrimination of ADPim/ATPex exchange.

P—The fact that PamNTT1 accepts Pi as third substrate might resolve inconsistencies in the phosphate metabolism of P. amoebophila. To analyze whether also phylogenetically more distantly related ATP/ADP transporters from other organisms possess the capacity to transport Pi, we investigated two carriers from the rickettsial species C. caryophilus (CcNTT) and H. obtusa (HoNTT) and one plastidial transporter from the higher plant A. thaliana (AtNTT1). These representative rickettsial and plastidial NTTs were heterologously expressed in E. coli, purified (supplemental Fig. S4), and reconstituted into liposomes. All recombinant carriers mediated ATP and ADP transport and therefore were functional in the liposomal system (supplemental Fig. S4). Generally, HoNTT exhibited highest net uptake rates that suggest a high activity of this reconstituted protein (Fig. 4 and supplemental Fig. S4). Import studies with radioactive Pi showed that AtNTT1 and also the two selected rickettsial NTTs were able to import Pi. In Fig. 4 the time-dependent Pi uptake in presence of exterior ADP is presented. Pi import into vesicles loaded with ADP plus Pi exceeded Pi import into proteoliposomes loaded with ATP plus Pi or with solely Pi. The rates of AtNTT1- and CcNTT-mediated Pi homo-exchange were lower than the rates of Pi (plus ADP) import into vesicles loaded with ATP plus Pi, whereas HoNTT exhibited higher rates for Pi homo-exchange than for Pi import in presence of exterior ADP and interior ATP plus Pi. Apart from slight differences in the substrate preference pattern, our results clearly demonstrate that ATP/ADP exchanging NTTs from distantly related organisms are capable for an ADP-dependent Pi transport, which resembles that of PamNTT1 (Figs. 3 and 4).

FIGURE 4.

P Heterologously expressed and purified ATP/ADP transporters from A. thaliana (AtNTT1), C. caryophilus (CcNTT), and H. obtusa (HoNTT) were reconstituted into liposomes loaded with 5 mm Pi or with 5 mm Pi plus the given nucleotides (10 mm). Time-dependent import of 50 μm radioactive Pi in presence of 50 μm nonlabeled ADP mediated by AtNTT1 (A), CcNTT (B), and HoNTT (C). Import into proteoliposomes loaded with Pi (open circle), Pi plus ADP (gray square), or Pi plus ATP (black diamond). The given values are net values (calculated by subtraction of Pi import in absence of external and internal nucleotides). The data are the means of at least three independent experiments, and the standard errors are displayed. The insets summarize the applied substrate conditions at the liposomal interior and exterior. Radioactive Pi is marked with an asterisk.

DISCUSSION

Metabolically impaired energy parasites depend on ATP import from the host cell, and also plant plastids rely on energy supply from the surrounding cytosol when photosynthetic activity is reduced or missing. ATP/ADP exchange would lead to substantial Pi accumulation in the organelle or cell if no interacting export mechanism for phosphate exists. In this study we identified that Pi transport is a previously not identified intrinsic feature of nonmitochondrial ATP/ADP transporters (Figs. 3 and 4). Pi is transported simultaneously with ADP, and this cotransport is facilitated in a one Pi to one ADP stoichiometrical exchange with ATP (supplemental Fig. S1). The cotransport of one H2PO-4 (with one negative charge) would be in complete agreement with the recently documented electroneutrality of PamNTT1-mediated ATP/ADP exchange (32). During hetero-exchange, Pi could act as a counterion compensating for the generation of a charge difference across the membrane.

A stimulatory influence of high Pi concentrations (in the millimolar range) on ADP import into isolated Rickettsia prowazekii cells had been observed more than three decades ago (38). Although a regulatory effect of Pi on the transport characteristics was discussed, the exact role of Pi in ATP/ADP exchange was not known (37). Interestingly, the simplest explanation, a possible cotransport of Pi and ADP, was never suggested for NTTs. Detailed analyses of PamNTT1 (Figs. 2 and 3 and supplemental Fig. S1) and first studies with two carriers from Rickettsiales (Fig. 4, ) suggest that the regulatory principle of Pi is tightly associated to its function as a cosubstrate of ADP. The addition of Pi indeed increased the Vmax and the affinity of PamNTT1 for ADP transport into ATP loaded proteoliposomes. It is imaginable that a simultaneous entry of ADP and Pi might mimic the presence of a triphosphorylated adenine nucleotide (ATP), which is the preferred import substrate.

The capacity of the phylogenetically different NTTs to transport ADP plus Pi in exchange with ATP (i) is in line with the electroneutrality of nucleotide exchange (32), (ii) prevents harmful Pi accumulation in intracellular living bacteria and plant plastids, and (iii) implies that the phosphate exporter that was proposed for a long time in these bacteria and in plastids is represented by the “bifunctional” ATP/ADP plus Pi transporter.

Very recent analyses demonstrated that energy parasitism is not only restricted to Chlamydiales and Rickettsiales but also occurs in further important mammalian pathogens (23, 24). NTT-type carriers reside in the plasma membrane and in the mitochondrial relict (the so called mitosome) of the intracellular protist Encephalitozoon cuniculi (24). By catalyzing a highly specific ATP/ADP exchange, these NTTs provide energy to the cell or to the mitosome that lacks ATP synthesis. It is tempting to speculate that also these eukaryotic NTTs catalyze a Pi coexport with ADP.

To our surprise, NTT-type ATP/ADP transporters mediate a nucleotide exchange-independent but ADP-induced homo-exchange of Pi in addition to the Pi cotransport with ADP (Figs. 3 and 4, white circles). The Pi homo-exchange was induced by external as well as by internal ADP (Fig. 3). In the following we propose a scenario that could explain the induction of Pi homo-exchange independent of the side of ADP application.

ADP and Pi enter the binding center, for example at the proteoliposomal lumen, and are transported to the opposed side (Fig. 5). In absence of exchange nucleotides, ADP stays at the binding center, whereas the nonlabeled Pi is displaced by radioactive Pi. A subsequent reversely directed translocation of the two bound substrates and replacement of radioactive Pi by nonlabeled Pi at the interior causes the observed Pi homo-exchange (Fig. 5). This hypothesized mechanism presupposes that only one single binding center exists for import and export that opens from one side of the membrane to the other where nucleotides and/or Pi are displaced by other substrates.

FIGURE 5.

Model explaining the ADP-dependent P ADP (Ado-P-P, P-P-Ado) supports entry of interior Pi (oval 1), ADP plus Pi are translocated (white arrow) to the exterior. Pi is displaced by radioactive Pi (Pi), whereas ADP remains at the binding center (oval 2). The substrate complex is translocated (white arrow) to the interior and radioactive Pi is displaced by nonlabeled Pi (oval 3).

By the help of mutant proteins we identified a correlation between the charge of the amino acid residue at position 446 and the transport properties of PamNTT1. A positive amino acid residue at position 446 allowed the insertion of ATP or Pi (plus ADP), a neutral amino acid residue or even more a negative amino acid residue stimulates ADP but suppresses Pi or ATP entry (Table 1). Substitution of Lys446 by arginine (K446R) entailed substantially lowered import rates for all substrates. We assume that structural differences between lysine and arginine rather than an unfavorable charge might have caused the decreased transport velocity (Table 1). The biochemical properties of the mutant proteins suggest that lysine 446 is an essential component of the ATP and Pi binding center or translocation pathway. We conclude that the positively charged lysine 446 interacts with the negative charge introduced either by the γ-phosphate of ATP or alternatively by Pi.

The fact that ATP import was highly reduced, whereas ATP export still occurred (ADPim/ATPex transport) in the mutant protein PamNTT1-K446E might argue against a single binding center for ATP import and export. However, the Vmax of ADPim/ATPex exchange was remarkably low when compared with that of ADP homo-exchange (Table 1). This characteristic could be explained by an impaired ATP export capacity leading to a return transport of previously imported ADP. This would be in agreement with the characteristics of the mutated plastidial ATP/ADP transporter, which exhibited a highly reduced capacity for both ATP import and ATP export (34).

It becomes clear that additional studies are required to get more insights into structure/function relationships of NTTs, and it will be interesting to focus on this topic in the future. Furthermore, we wish to analyze transport parameters under conditions mimicking the energy state of intracellular living bacteria in their host cells. However, for this it is a precondition to determine Pi, ATP, and ADP concentrations in the bacterium and the infected cell.