Adenine Nucleotide Translocator Transports Haem Precursors into Mitochondria

Haem is a prosthetic group for haem proteins, which play an essential role in oxygen transport, respiration, signal transduction, and detoxification. In haem biosynthesis, the haem precursor protoporphyrin IX (PP IX) must be accumulated into the mitochondrial matrix across the inner membrane, but its mechanism is largely unclear. Here we show that adenine nucleotide translocator (ANT), the inner membrane transporter, contributes to haem biosynthesis by facilitating mitochondrial accumulation of its precursors. We identified that haem and PP IX specifically bind to ANT. Mitochondrial uptake of PP IX was inhibited by ADP, a known substrate of ANT. Conversely, ADP uptake into mitochondria was competitively inhibited by haem and its precursors, suggesting that haem-related porphyrins are accumulated into mitochondria via ANT. Furthermore, disruption of the ANT genes in yeast resulted in a reduction of haem biosynthesis by blocking the translocation of haem precursors into the matrix. Our results represent a new model that ANT plays a crucial role in haem biosynthesis by facilitating accumulation of its precursors into the mitochondrial matrix.


Introduction
Haem, an endogenous porphyrin derivative, is a prosthetic group for haem proteins, which play an essential role in oxygen transport, respiration, signal transduction, and detoxification. Haem biosynthesis is catalyzed by eight enzymes. The porphyrin ring structure is generated in the cytosol, and then the haem precursor protoporphyrinogen IX is transported into the intermembrane space of mitochondria. PP IX is generated from protoporphyrinogen IX by protoporphyrinogen oxidase (PPO) on the outer surface of the mitochondrial inner membrane. Finally, haem is generated by ferrochelatase (FECH)-catalyzed iron insertion into PP IX on the inner surface of the mitochondrial inner membrane [1].
Mitochondria are enclosed by two layers of membranes, the outer membrane and the inner membrane. In haem biosynthesis, haem precursors must be transported into the matrix across the two membranes. ABCB6 and PBR are candidate porphyrin transporters across the outer membrane [2,3]. Particularly, ABCB6 is thought to play a crucial role in haem biosynthesis by facilitating the accumulation of porphyrins into mitochondria [3]. On the other hand, it is controversial how PP IX reaches the mitochondrial matrix through the inner membrane. It has been suggested that PPO binds to FECH across the inner membrane, and that this association permits the direct transfer of PP IX from PPO to FECH [4].
However, Proulx et al. reported that the PPO/FECH complex formation is not essential for haem biosynthesis, because exogenous PP IX could be converted into haem independently of the PPO/ FECH complex [5]. It is therefore likely that there is another inner membrane transport system for PP IX.
We have studied on the mitochondrial accumulation of phosphorescent porphyrin derivatives. Using affinity latex beads, the mitochondrial transporter 2-oxoglutarate carrier (OGC) was identified as a protein binding to phosphorescent porphyrins and was shown to be responsible for the accumulation of phosphorescent porphyrins in mitochondria [6]. However, it remains unclear whether OGC has any effect on the transport of natural porphyrins and on haem biosynthesis. Here, we identified ANT as a novel mitochondrial protein binding to haem and its precursors. ANT is an inner membrane transporter that facilitates the exchange of ATP and ADP [7]. We show evidence that ANT contributes to haem biosynthesis by transporting its precursors into mitochondria.

Results
Haem and PP IX directly bind to ANT First, we attempted to identify haem-or PP IX-binding proteins in mitochondria using affinity beads termed as SG beads.
Carboxyl groups of haem or PP IX were succinated and conjugated to amino-modified SG beads. Using haem-or PP IX-conjugated beads, haem-or PP IX-binding proteins were purified from rat mitochondrial extract. Two proteins with apparent molecular weights of 33 and 30 kDa bound to haemconjugated beads, while only the 30 kDa protein was found associated with PP IX-conjugated beads (Fig. 1A). Although the 33 kDa protein was not identified successfully, the 30 kDa protein was identified as ANT2 by Q-TOF MS analysis. The identity was confirmed by Western blotting (Fig. 1B). We examined the haembinding activity of three ANT isoforms (ANT1, 2 and 3) using recombinant proteins and found that haem bound to all the ANT isoforms similarly (Fig. 1C). The amount of recombinant protein that bound to the haem-conjugated beads was calculated to be 10% of the input amount for ANT1, 2, and 3. We also showed that apo-myoglobin bound to the haem-conjugated beads, but that glutathione-S-transferase did not (Fig. S1).

Modeling of the binding between haem and ANT
ANT is a mitochondrial carrier that transports ADP across the inner mitochondrial membrane. Recently, an atomic model of the complex between Bos taurus ANT1 (BtAAC1) and the ANT inhibitor carboxyatractyloside (CATR) has been proposed [8]. To elucidate the binding properties of haem for ANT, we used computational analysis based on the BtAAC1 structure. As shown in Fig. 2A and B, haem docked in the center pore domain of BtAAC1. Furthermore, we also showed that the ADP binding site was formed by K22, R79 and R279 residues which associate with the phosphate moieties of ADP. The residues between G182 and I183 associated with the adenine ring structure (Fig. 2C), as seen in the previous in silico analysis [9]. Interestingly, haem also associated with the K22 and R79 residues of BtAAC1 via its carboxy residue and with I183 and R279 residues via its pyrrol ring structures ( Fig. 2D and E). To confirm the binding model, we elucidated the haem-binding activity of ANT1 mutated in the residues of K22, R79 or R279. As shown in Fig. 2F, while ANT1 R79A mutant remained the haem-binding activity, K22A and R279A could not bind to haem. These results strongly support our docking model between haem and ANT1.

Haem and its precursors accumulate into mitochondria via ANT
Since the above analysis predicted that haem binds to ANT though a common ADP binding site, we hypothesized that ANT would contribute to the mitochondrial transport of haem as well as that of ADP. To elucidate this hypothesis, we analyzed the effect of porphyrins on the mitochondrial uptake of ADP using isolated rat liver mitochondria. [ 3 H] labeled-ADP uptake into mitochondria reached saturation after 5 min, whereas pretreatment of atractyloside (ATR) blocked its uptake completely. Notably, haem inhibited an initial rate of ADP uptake (Fig. 3A). Furthermore, PP IX and CP III also strongly inhibited ADP uptake in a dose-dependent manner (Fig. 3B). In addition, ADP uptake was not changed by treatment with FCCP or NaN 3 , or by omitting succinate (Fig. S2A). We also confirmed the disruption of membrane potential or loss of oxygen consumption by treatment with FCCP or NaN 3 , respectively ( Fig. S3 and Text. S1). Lineweaver-Burk plot analysis revealed that the inhibition of ADP uptake by haem is competitive with an inhibition constant (K i ) of 7.3 mM (Fig. 3C and 3D). PP IX or CP III also inhibited ADP uptake in a competitive manner with similar Ki values (Fig. 3D). We previously reported that haem binds to the transporter OGC and inhibits the mitochondrial uptake of its substrate, 2oxoglutarate [6]. We therefore compared Ki for the inhibition of ADP or 2-oxoglutarate uptake by haem. As a result, the inhibitory Figure 1. Identification of mitochodrial haem-binding proteins. Purification of haem or PP IX-binding proteins. Haem-or PP IX-conjugated or unconjugated SG beads were incubated with mitochondrial extracts of rat liver. Bound proteins were eluted with Laemmli dye and subjected to SDS-PAGE, followed by silver staining (A) or Western blotting with anti-ANT antibody (B). C, FLAG-tagged ANT1, 2, or 3 was incubated with haemconjugated (+) or unconjugated (2) SG beads. The eluates were separated by SDS-PAGE, followed by western blotting. doi:10.1371/journal.pone.0003070.g001 effect of haem was found to be more potent on ADP uptake than on 2-oxoglutarate uptake, suggesting that haem and haem precursors are preferentially accumulated in the matrix by ANT.
Next, to examine the mitochondrial uptake of haem precursor PP IX, we analyzed the Zn-PP IX formation in the isolated mitochondria. PP IX and Zn-acetate were incubated with rat liver mitochondria, and then Zn-PP IX was detected by HPLC analysis. Zn-PP IX generation was significantly decreased by the addition of ADP or ATP (Fig. 3E), whereas it was not changed by treatment with AMP, FCCP or NaN 3 , or by omitting succinate (Fig. 3E, Fig. S2B). These results suggested that PP IX is transported into the mitochondrial matrix along the same route as ADP.

Haem biosynthesis is reduced in ANT-deficient yeast
We examined the effect of ANT on haem biosynthesis in vivo using an ANT-deficient yeast strain (DAAC), in which all the ANT homologous genes, AAC1, AAC2, and AAC3, were disrupted [10]. We found using affinity beads that Aac1p, Aac2p, and Aac3p all bind to haem (data not shown). In the DAAC strain, the amount of mitochondrial haem was found to be significantly decreased compared to the wild-type strain (Fig. 4A). Furthermore, we investigated whether exogenous expressed catalase A activity, which is known as haem protein, is affected in the DAAC strain. HA-tagged catalase A was immunoprecipitated, and its activity was determined. As shown in Fig. 4B, HA-catalase A activity was significantly reduced in the DAAC strain. These results indicated that ANT contributes to haem biosynthesis.
If the haem reduction is due to defects of the mitochondrial translocation of haem precursors, their accumulation should be observed in the DAAC strain. To examine this hypothesis, we extracted haem precursors from the wild-type and DAAC strains, followed by HPLC analysis. As shown in Fig. 4C and D, PP IX and CP III were dramatically increased in the DAAC strain while uroporphyrin, coproporphyrin I, and other haem precursors were not. In addition, peaks appearing at 260 nm were the same in the different yeast strains (Fig. S4). Furthermore, we also examined haem biosynthesis in an ATP synthase component ATP3-deficient yeast strain (DATP3) [11]. As shown in Figure 4A and B, the amount of mitochondrial haem and catalse A activity was the same in the DATP3 yeast strain as in the wild type. Also, whereas CP III accumulation was observed in the DATP3 yeast strain, the amount of PP IX was the same as in the wild type. Although the reason for CP III accumulation is unclear, we conclude that the mitochondrial transport of the haem precursor is not impaired in the DATP3 yeast strain, since PP IX must be transported into mitochondria for haem biosynthesis. These results strongly support our hypothesis that ANT contributes to haem biosynthesis by accumulating haem precursors independently of energy production or membrane potential in mitochondria. In addition, we also analyzed haem biosynthesis using an OGC-deficient yeast strain (DODC), in which OGC homologous genes, ODC1 and ODC2, were disrupted [12]. Neither haem reduction nor accumulation of haem precursors was observed in the DODC strain (Fig. S5). These results suggested that ANT, but not OGC, contributes to haem biosynthesis and the translocation of haem precursors into the matrix.

Discussion
In the present study, we identified ANT as a haem or PPIX binding protein by affinity purification (Fig. 1). In vitro biochemical analyses revealed that ADP uptake was also inhibited by haemrelated porphyrins in a competitive manner (Fig. 3). Furthermore, PP IX uptake into the mitochondrial matrix was inhibited specifically by the addition of ADP (Fig. 3E). These results indicate that haem and ADP are transported into mitochondria through a common pathway comprising ANT. In support of this, haem associates with the center pore site of BtAAC1, as shown by by in silico docking analysis (Fig. 2). Interestingly, haem makes contacts with BtAAC1 through the same residues (K22, R79, I183, R279) of ANT as those used for ADP docking. A previous study showed that mutation of residues K22, R79 or R279 abolished the transport activity of the yeast ANT [13,14]. X-ray crystallography analysis of BtAAC1 also suggested that these residues are important for ADP binding and would contribute to its transport into the inner matrix [8]. We also showed that mutation of ANT1 K22A and R279A resulted in the disruption of its binding activity with haem (Fig. 2F). Taken together, these results show that haem and its precursors are incorporated into mitochondria in the same way as ADP via ANT. On the other hand, atractyloside did not affect the Zn-PP IX formation, whereas ADP or ATP inhibited strongly (Fig. S2B). Although the detailed mechanism is not elucidated yet, there are several differences of the ANT-binding site between haem and atractyloside. It has been known that atractyloside interacts with R79, N87, K91, L127, V130, I183, R187, D231 and R234 of ANT1 from an atomic model of the complex [8], while haem interacts with K22, R79 G182, I183 and R279 of ANT1 in our docking analysis (Fig. 2). ANT1 R79A mutant, which overlaps the binding site with haem or atractyloside, remained the haem-binding activity, but K22A and R279, the haem-specific binding residues, disrupted its binding completely. Haem might be transported into the matrix in the presence of atractyloside through recognizing the different ANT1 residues from the atractyloside-binding site.
We previously found that OGC binds to phosphorescent porphyrin derivatives and accumulates them into mitochondria. In vitro analysis showed that the uptake of the ANT substrate ADP was more severely impaired by haem than the uptake of the OGC substrate 2-oxoglutarate (Fig. 3D). On the contrary, 2-oxoglutarate uptake was inhibited more strongly than ADP uptake by phosphorescent porphyrin derivatives such as palladium mesotetra(4-carboxyphenyl)porphyrin (PdTCPP) and palladium mesotetra(4-aminophenyl)porphyrin (PdTAPP) in a competitive manner, with Ki constants of 15 mM and 1.9 mM, respectively (data not shown). Moreover, neither haem reduction nor haem precursor accumulation was detected in OGC-deficient yeast. These findings suggest that ANT is primarily involved in the transport of haem precursors and haem biosynthesis, while OGC is involved in the mitochondrial accumulation of synthetic phosphorescent porphyrins. ANT may recognize the haem-related porphyrin structure, while OGC may bind to the tetraphenyl-porphyrin structure.
We showed that the disruption of ANT homologues in yeast resulted in a 40% reduction in haem content and a two-fold increase in the level of PP IX (Fig. 4). On the other hand, ATP synthase component ATP3-deficient yeast had the same levels of haem and PP IX as the wild-type strain. This indicates that ANT contributes to haem biosynthesis by accumulating the haem precursor in mitochondria independently of energy production or metabolic events. On the other hand, CP III accumulation was observed in ATP3-deficient yeast, but amount of PP IX and haem was not changed. However, the detailed mechanism is not elucidated yet, the CP III accumulation was observed commonly in ANT-or ATP3-deficient yeast, both which were abolished the energy production of mitochondria. CP III amount might be affected by energetic status of mitochondria, but haem or PP IX amount would not. Furthermore, we also showed that haem and ADP uptake in mitochondria was not affected by NaN3 or the uncoupling reagent FCCP, or by the omission of the electron transport substrate succinate (Fig S2). This suggested that the inner mitochondrial membrane potential is not necessary for haem or ADP transport. Previously, it has been suggested that PP IX accumulation in the mitochondrial matrix is mediated by the PPO/FECH transmembrane complex. However, our results strongly suggest that ANT contributes to haem biosynthesis by serving as an alternative PP IX transport pathway.

Materials
Haemin, PP IX, and CP III were purchased from Porphyrin Products. Anti-ANT antibody was kindly provided by Dr. Catherine Brenner-Jan [15]. Anti-FLAG antibody (M2), anti-FLAG resin, FLAG peptide, anti-HA antibody (HA-7), and anti-HA resin were purchased from Sigma.

Plasmids construction and yeast strains
cDNAs of human ANT1, ANT2, and ANT3 were obtained from a HL60 cDNA library. ANT1 mutagenic primers were showed in Table S1. For expression of FLAG epitope tagged ANT, the cDNAs were inserted into the mammalian expression vector pCMV-tag2. cDNA of yeast catalase A was obtained from W303-1B genomic DNA. For expression of catalase A with a carboxylterminal HA epitope, the cDNA was inserted into the yeast expression vector pRS424 carrying the TFA1 promoter.

Purification of haem binding proteins
Amino-modified SGNEGDEN beads were prepared as described [16]. To generate haem-or PP IX-conjugated beads, 0.3 mM haem or PP IX was succinated with an equal amount of N-hydroxysuccinimide in dimethylformamide and then reacted with SGNEGDEN beads at room temperature overnight as described [6].

Measurement of ANT activity
Measurement of ADP uptake in mitochondria was performed as described previously [17]. Rat liver mitochondria (1 mg/ml) were pretreated with or without haem, PP IX, or CP III for 5 min in buffer (10 mM Tris-HCl (pH 7.4), 250 mM sucrose, 5 mM succinate, and 10 mM EGTA) and then incubated with 3 Hlabeled ADP (2.5-20 mM) on ice for 0.5, 1, 2, 5, 10 min. The reaction was terminated by the addition of 100 mM atractyloside (final concentration) and centrifuged at 10,0006 g for 2 min at 4uC. Mitochondria were lysed with 50 ml of 1% SDS, and [ 3 H] ADP uptake was measured using a liquid scintillation counter [17]. The mitochondrial uptake of 2-oxoglutarate was measured as described in Kabe et al. (2006) [6].
For analysis of Zn-PP IX formation in the isolated mitochondria, PP IX (50 mM) was incubated with 0.5 ml rat liver mitochondria (1 mg/ml) in the presence or absence of ADP on ice for 30 sec in buffer (10 mM Tris-HCl (pH 7.4), 250 mM sucrose, 5 mM succinate, 10 mM EGTA and 50 mM Zn-acetate) and then centrifuged at 10,000 6g for 2 min at 4uC. The generated Zn-PP IX was extract with 1 ml ethyl acetate/acetic acid (3/1, v/v). After centrifugation at 5000 rpm for 10 min, the 10 ml supernatants were subjected to chromatography on a C 18 reverse-phase HPLC column equipped with a fluorometric detector [21].

Measurements of haem and its precursors in yeast
Yeast mitochondria were isolated as described in Daum et al. (1982) [18]. The mitochondrial haem was converted to PP IX in 2 M oxalic acid at 100uC for 30 min, and the amount of PP IX was determined by its fluorescence intensity at 600 nm [19,20].
To detect haem precursors in yeast, yeast were grown in 100 ml of YPD medium overnight. Yeast were recovered by centrifugation at 160 6g for 5 min, and spheroplasts were prepared as described previously [18], and then homogenated with 5 ml ethyl acetate/acetic acid (3/1, v/v). After centrifugation at 5000 rpm for 10 min, the supernatant was concentrated to 1 ml, and 10 ml samples were subjected to chromatography on a C 18 reverse-phase HPLC column equipped with a fluorometric detector [21].

Measurement of catalase activity
Yeast cells transformed with an expression vector of HA-tagged catalase A were harvested and lysed with glass beads in lysis buffer (50 mM potassium phosphate, 1 mM EDTA). After centrifugation at 15,000 rpm for 15 min at 4uC, HA-tagged catalase A was purified from the supernatant with anti-HA resin. Catalase activity was measured using Catalase Assay Kit (Cayman Chemical) according to the manufacturer's instructions.

Docking analysis between haem and ANT1
In silico work was performed on a IntelH PentiumH D (2. Figure S1 Purification of apo-myoglobin or glutathione-Stransferase using haem-conjugated beads. 1 mg of Apo-myoglobin (A) or glutathione-S-transferase (GST) (B) was incubate with haem-conjugated (+) or unconjugated (2) SG beads. The eluates were separated by SDS-PAGE, followed by silver staining (A) or western blotting with anti-GST antibody (B). Equine apomyoglobin was prepared as described previously [27].