OUP user menu

The delivery of ADP/ATP carrier protein to mitochondria probed by fusions with green fluorescent protein and β-galactosidase

Katarína Polčicová, Petra Kempná, L'udmila Šabová, Gabriela Gavurníková, Peter Polčic, Jordan Kolarov
DOI: http://dx.doi.org/10.1016/S1567-1356(03)00170-3 315-321 First published online: 1 December 2003


The import of proteins into mitochondria is an essential process, largely investigated in vitro with isolated mitochondria and radioactively labeled precursors. In this study, we used intact cells and fusions with genes encoding two reporter proteins, green fluorescent protein (GFP) and β-galactosidase (lacZ), to probe the import of the ADP/ATP carrier (AAC). Typical mitochondrial fluorescence was observed with AAC-GFP fusions containing at least one complete transmembrane loop. This confirms the results of in vitro analysis demonstrating that an internal targeting signal was present in each one of the three transmembrane loops of the carrier. The fusions of AAC fragments to β-galactosidase demonstrated that the targeting signal was capable of delivering the reporter molecule to the mitochondrial surface, but not to internalize it to a protease-inaccessible location. The delivery to a protease-inaccessible location required the presence of more distal sequences present within the third (C-terminal) transmembrane loop of the carrier molecule. The results of our study provide an alternative for investigation in a natural context of mitochondrial protein import in cells when the isolation of intact, functional mitochondria is not achievable.

  • ADP/ATP carrier
  • Mitochondrion
  • Biogenesis
  • Protein import

1 Introduction

The mitochondrial ADP/ATP carrier (AAC) is the most abundant representative of a large family of inner mitochondrial membrane proteins (mitochondrial carrier family) that serve as solute carriers involved in the traffic of metabolites across the inner mitochondrial membrane. The common structural feature of this exclusively mitochondrial family of proteins is that their polypeptide chains consist of three related sequence repeats of approximately 100 amino acids in tandem [1]. Each sequence repeat might be folded into a large loop containing two putative transmembrane α-helices linked with a polar region, forming a structure with six transmembrane α-helices and extremities extended to the cytosol (Fig. 1). Like other members of this protein family, AAC is synthesized on the cytosolic ribosomes without a cleavable N-terminal extension and is imported into the mitochondrial membrane by a pathway that is different from the pathway for proteins with cleavable N-terminal presequences [24]. The carrier proteins are recognized on the mitochondrial surface by the receptor Tom70 [5,6] and transported across the outer membrane via a general import pore that is also employed by mitochondrial cleavable preproteins. The mechanism of translocation of carrier proteins is, however, different as they are most probably translocated in a loop formation and not as a linear polypeptide chain led by a cleavable presequence [7,8]. In the intermembrane space, the import pathways of the two classes of precursor proteins diverge completely. Here, the carrier proteins are initially bound by an oligomeric Tim9–Tim10 complex [9,10] and then transferred to Tim12 and to the protein insertion complex of the inner membrane formed by Tim22, Tim54, and Tim18 [9,1114].


Schematic structure of the AAC1 gene, AAC1-GFP, AAC1-lacZ, and AAC3-lacZ fusion proteins. The positions of the primers used to clone different segments of the AAC1 gene in front of GFP are shown. The individual constructs are numbered according to the position of amino acids within AAC proteins. TM1, TM2, …, TM6 indicate the position of putative transmembrane α-helices present in the first (I), second (II), and third (III) loops of the carrier protein.

In spite of the extensive investigations on the import mechanism, the character of the internal targeting sequence in carrier proteins remains unclear [15,16]. Early work on the in vivo mitochondrial import using fusions between AAC1 and β-galactosidase has revealed that the mitochondrial targeting and delivery of the different hybrid gene products into a protease-inaccessible compartment required the information present within the first 115 residues of the protein [17]. On the other hand, from the results of in vitro import studies with isolated mitochondria, it has been concluded that targeting information is not restricted to the N-terminal region. The truncated AAC containing the two C-terminal transmembrane loops has also been found to be imported into mitochondria [2]. More recently, by the same in vitro approach, a mitochondrial targeting signal has been found in each of the three transmembrane loops [7] and it was suggested that the three transmembrane loops function in a concerted manner [8] at several distinct steps of the import process.

The aim of this work was to corroborate the conclusions obtained on the basis of in vitro import studies using an in vivo approach with reporter proteins and intact cells. In-frame fusions of different segments of AAC proteins with the green fluorescent protein (GFP) from Aequorea victoria and β-galactosidase (lacZ) from Escherichia coli were prepared and their targeting to the mitochondria was investigated. The in vivo studies demonstrate that each of the three transmembrane loops of the AAC protein contains sufficient information for targeting the AAC-GFP fusion to the mitochondria. β-Galactosidase is also directed to the mitochondria by hybrids containing at least one complete transmembrane loop, but the import to a protease-inaccessible mitochondrial location requires the participation of the third transmembrane loop.

2 Materials and methods

The wild-type Saccharomyces cerevisiae strain W303-1B (MATα, ade2, leu2, his3, trp1, ura3, can1) and the aac1, aac2 disrupted strain Jd1 (MATα, ade2, leu2, his3, trp1, ura3, can1, aac1::LEU2, aac2::HIS3) [18] were used.

2.1 AAC-GFP fusions

To prepare GFP fusion constructs of either full-size AAC1, or fragments containing a single or double transmembrane loop (Fig. 1), the gene sequences were amplified by polymerase chain reaction (PCR) using the following oligonucleotide primers equipped with restriction sites for EcoRI or BamHI: GG1: 5′-CGG GAT CCA TGT CTC ACA CAG AAA CA-3′ (+1 +18); GG2: 5′-CGG AAT TCC TTG AAT TTT TTG CC-3′ (+913 +927, reverse); 12AC1: 5′-CGG GAT CCA TGT TGA GTT ACG ACA GA-3′ (+310 +328); 23AC1: 5′-CGG AAT TCA GCC CCC GTC AAC AG-3′ (+639 +654, reverse); 33AC1: 5′-CGG GAT CCA TGT TGA CGG GGG CTC TA-3′ (+625 +643).

The PCR-amplified sequences were digested with BamHI and EcoRI and ligated into the corresponding sites of a pGFP-C-FUS vector (provided by J.H. Hegemann, Justus-Liebig University, Giessen, Germany) [19]. The plasmid containing the fusion of the incomplete third transmembrane loop (AAC1–230–309G) was prepared by XbaI digestion and religation of the construct (AAC1–209–309G) containing the third loop (Fig. 1a). ATG at position +230 served as a start codon. Cells transformed with the GFP fusion-containing plasmids were induced to express the fusion proteins by omitting methionine from the growth medium. At the indicated time, the cells were loaded with 4′,6′-diamidino-2-phenyl-indole (DAPI, Molecular Probes, Eugene, OR, USA) and observed using an Olympus BX-50 fluorescence microscope equipped with filters recommended by the manufacturer.

2.2 AAC-lacZ fusions

To prepare AAC-lacZ hybrids we explored the restriction sites present in the AAC1 and AAC3 genes. AAC1-lacZ fusions were prepared from the plasmid YEp 6-19-28 containing a 2.8-kb BamHI fragment of the AAC1 gene [20]. SalI, HincII, and PstI, which cleave the AAC1 gene-coding region at +113, +474, and +861, respectively, were used to obtain fragments cloned in-frame with the lacZ gene in YEp 354, 355, and 358 vectors, giving AAC1(1–37)Z, AAC1(1–158)Z, and AAC1(1–287)Z fusions (Fig. 1b). To prepare the AAC3-lacZ fusions the plasmid pGEM3ZfA4 containing a 1.7-kb XmnI fragment (−98/+1602) of the AAC3 gene [21] was used. HindIII, PvuII, HincII, and RsaI (+269, +376, +485, +760) restriction sites were used to clone AAC3 gene fragments in YEp vectors, yielding AAC3(1–89)Z, AAC3(1–125)Z, AAC3(1–163)Z, and AAC3(1–253)Z fusions (Fig. 1b). Cells transformed with AAC-lacZ fusion plasmids were cultivated overnight in synthetic medium (SM) containing 0.5% glucose. β-Galactosidase activity was assayed [22] in crude cell extracts, in isolated mitochondria treated with Triton X-100, and in postmitochondrial supernatants.

Yeast mitochondria were isolated from protoplasts as described previously [23]. For protease treatment the mitochondria were diluted to 1 mg protein ml−1 in 0.6 M sorbitol, 10 mM Tris–HCl, pH 7.4 and incubated with proteinase K (20 μg ml−1) for 30 min at 23°C. The reaction was stopped by addition of 1 mM phenylmethylsulfonyl fluoride and transfer of the samples on ice.

Published procedures [24] were used for isolation and manipulation of DNA. All fusion constructs were verified by DNA sequencing. Yeast cells were transformed by the standard lithium acetate procedure [25]. Standard procedures were used for Western blotting onto nitrocellulose and immunodetection with β-galactosidase antibody and secondary antibodies conjugated with alkaline phosphatase (Promega, Madison, WI, USA).

3 Results

3.1 AAC-GFP fusions

To investigate the functionality of AAC-targeting signals in vivo, sequences containing different parts of AAC1 were fused to the GFP gene (Fig. 1) and introduced into either the wild-type (W303-1B), or in the carrier deletion (Δaac1aac2) mutant strain (Jd1). Cells transformed with fusion plasmids were induced to express the fusion proteins by omitting methionine from the growth medium and subsequently subjected to fluorescence microscopy. A homogeneous bright green fluorescence within the cytoplasm of cells expressing GFP alone was in contrast with the particulate fluorescence yielded by fusion of the full-size AAC1 to GFP (Fig. 2). The latter co-localized with blue fluorescence of the complexes formed by DAPI with mtDNA. The specific mitochondrial staining was observed in the wild-type, as well as in the oxidative-phosphorylation-deficient strain lacking the two aerobic isoforms of the AAC (Δaac1aac2). The lack of AAC function in this mutant, however, was not complemented by the expression of plasmids containing AAC-GFP fusions (not shown).


Fluorescence microscopy of yeast cells expressing GFP and AAC1-GFP fusion. The double deletion (Δaac1aac2) mutant cells (Jd1) transformed with control (pGFP-C-FUS) plasmid (a), or AAC1-GFP fusion-containing plasmids were grown overnight in synthetic media without methionine, stained with DAPI, and analyzed by fluorescence microscopy using appropriate filters for visualization of green (a,b) and blue (c) light.

We asked whether different parts of the AAC protein would be capable of targeting the GFP to the mitochondria of intact cells. As shown in Fig. 3a,b, AAC1(1–213)G and AAC1(104–309)G fusions, containing two complete transmembrane loops (I–II, or II–III) exhibited distinct fluorescence which coincided with the DAPI staining of mtDNA, like the GFP fusion of full-size AAC1. This indicates that the targeting information is not limited to the first transmembrane loop as was suggested earlier by others [17].


Fluorescence microscopy of yeast cells expressing different parts of AAC1 fused to GFP. The Jd1 deletion mutants transformed with plasmids containing AAC1(1–213)G (a), AAC1(104–309)G (b), AAC1(104–213)G (c), AAC1(209–309)G (d), and AAC1(230–309)G (e) fusions were grown and analyzed as in Fig. 2. f: The same cells were examined under visible light.

ATG codons introduced at positions +310 or +625 were used to express AAC1(104–213)G and AAC1(209–309)G fusions containing a single complete transmembrane loop (II or III). Based on the microscopic appearance, both fusions targeted the GFP to the mitochondria (Fig. 3c,d). On the other hand, the expression of a truncated transmembrane loop III present in the AAC1(230–309)G fusion, which contains the last 80 amino acids of AAC1, yielded only faint fluorescence (Fig. 3e,f). The above results, together with the results of early work on in vivo carrier import [17], allow us to conclude that each of the three transmembrane loops of AAC protein possesses sufficient information for targeting of the carrier to the mitochondria.

3.2 AAC-lacZ fusions

For construction of β-galactosidase fusions advantage was taken from the presence of restriction sites within the coding region of the AAC1 and AAC3 genes. Both genes encode carrier isoforms, which are 75% identical (96% similar) and exhibit identical enzymatic properties [17,21]. Moreover, the position and the amino acid sequences of the transmembrane α-helices in both proteins are almost 100% identical [29]. Fusions of these proteins with β-galactosidase made it possible to distinguish whether the passenger protein was successfully internalized to the protease-protected location, or was only delivered to the surface of the organelle. To determine the role of the hydrophobic transmembrane loops in the delivery and import processes, the cells expressing fusions of individual AAC fragments with lacZ were disintegrated and fractionated to mitochondrial and postmitochondrial (cytoplasmic) fractions. The presence of β-galactosidase in these fractions was then examined by activity measurements and Western immunoblotting.

The AAC-lacZ fusions, AAC1(1–37)Z and AAC3(1–89)Z, containing incomplete parts of the first transmembrane loop delivered 3% and 30%, respectively, of the total enzyme activity to the mitochondrial fraction (Fig. 4). The addition of proteinase K to this fraction led to the complete loss of β-galactosidase activity, indicating that the enzyme is loosely associated with the outer surface of the mitochondria. Western blot analysis is in good agreement with the enzyme activity assay, confirming that the fusion with the longer N-terminal fragment of AAC was more efficient in delivering proteins to the mitochondria. Fig. 4 also shows that more peptides with electrophoretic mobility higher than that of the fusion protein have interacted with anti-β-galactosidase serum. Such peptides were not detected in cells transformed with the control plasmid and most probably they are β-galactosidase breakdown products.


Identification of the AAC-lacZ hybrid proteins in mitochondrial and postmitochondrial fractions by Western immunoblotting analysis. Jd1 mutant cells transformed either with the control (YEp354), or AAC1(1–37)Z and AAC3(1–89)Z fusion-containing plasmids were grown overnight in SM supplemented with 0.5% glucose. Mitochondrial (M) and postmitochondrial (P) fractions were prepared as described in Section 2 and the presence of β-galactosidase was determined either enzymatically, or by Western immunoblotting with anti-β-galactosidase serum. The portion of β-galactosidase imported into mitochondria was estimated by proteinase K treatment. The numbers at the bottom represent the percentage of total β-galactosidase activity present in the respective fractions.

The next three fusions, AAC1(1–158)Z, AAC3(1–125)Z, and AAC3(1–163)Z, carried one complete transmembrane loop, but they differed in the location of the fusion joints with respect to the sides of the membrane. Regardless of that, all three fusions delivered about 80% of the total enzyme activity to the mitochondria. Almost exclusive delivery of β-galactosidase to the mitochondrial fractions was confirmed by the Western blots analysis. Nevertheless, the treatment of the mitochondrial fraction with proteinase K showed that in all cases the enzyme activity was sensitive to the added enzyme (Fig. 5), indicating that the reporter molecule was not internalized into the organelles.


AAC sequences localizing β-galactosidase hybrids to mitochondria. The Jd1 mutant cells transformed with AAC3(1–125)Z, AAC1(1–158)Z, AAC3(1–163)Z, and AAC1(1–287)Z fusion-containing plasmids were grown and processed as in Fig. 4. M, mitochondrial fraction; P, postmitochondrial fraction.

The longest lacZ fusion, AAC1(1–287)Z, containing full-size carrier without the last 20 amino acids, was not so efficient in targeting the enzyme activity to the mitochondria. Less than half of the enzyme activity co-purified with mitochondria. In contrast to all other fusions, however, more than 90% of mitochondria-associated β-galactosidase activity was inaccessible to proteinase K. The enzyme activity was abolished after simultaneous addition of proteinase K and detergent, indicating that the enzyme fraction associated with the mitochondria was imported into mitochondria. Identical results were obtained with a shorter fusion, A3(1–253)Z (not shown), which contained only half of the third transmembrane loop. It should be noted that all longer AAC-lacZ fusions were expressed to a relatively low level as compared to those containing a single or double loop.

4 Discussion

The import of cytoplasmically synthesized proteins into mitochondria is an essential process and pathological conditions associated with defects in the import machinery have been reported in humans [26]. To investigate the various factors, compounds, and conditions influencing the import machinery it is necessary to employ an assay enabling the examination of the import process in intact cells.

In the present study we examined the in vivo mitochondrial import of the AAC, the most abundant and most intensively studied representative of the family of mitochondrial carrier proteins. Different carrier domains were fused to GFP and the bright green fluorescence of mitochondria evidenced the presence of specific mitochondrial targeting sequence(s) within the fusion. By this technique we were able to show that an internal targeting signal, which most probably directed the proteins to mitochondrial import receptors, was present in each of the three transmembrane loops. This is in agreement with the results of in vitro import studies demonstrating the presence of three internal targeting signals in the AAC [8].

To find out if the targeting signals are sufficient to drive the insertion of AAC-derived precursors to a protease-inaccessible location, we used fusions with another reporter: β-galactosidase. The latter allowed us to follow the intracellular distribution of hybrids by both enzymatic and immunochemical techniques. Analysis of the localization of gene fusion products harboring only one transmembrane α-helix (incomplete transmembrane loop) revealed that the passenger reporter protein is predominantly delivered to the postmitochondrial fraction. The hydrophobic segment at the N-terminal part of the longer fusion (A3–89) increases the association with the mitochondria probably through non-specific interaction with the membranes.

Three other gene fusion products, AAC1(1–158)Z, AAC3(1–125)Z, and AAC3(1–162)Z, possessing a single complete transmembrane loop, delivered as much as 80% of the total β-galactosidase activity to the mitochondrial fraction. Two of them, AAC1(1–158)Z and AAC3(1–162)Z, have an additional transmembrane α-helix, which puts the AAC C-terminal amino acids to the matrix side of the membrane. Independent of that, all three precursor fragments brought β-galactosidase to a proteinase K-accessible location. This indicates that the targeting information that is present in the first complete transmembrane loop is not sufficient for internalization. The latter results are in apparent contradiction with those of the early studies [17,27] showing that the first 115 amino acid residues of the carrier contained both targeting and internalization information. According to others [28], the central (second) loop of the uncoupling protein (Ucp1p), which is a close homolog of AAC, contained both the targeting and the import information. More recent in vitro import studies on isolated mitochondria [8] have demonstrated that the distal sequences of the carrier (third transmembrane loop) were required for the insertion into the inner mitochondrial membrane. The latter conclusion is fully supported by our results obtained with the longest gene fusion products, AAC1(1–287)Z and AAC3(1–253)Z. The last two fusion proteins are targeted to the mitochondria and the reporter protein activity is not sensitive to added protease. This strongly suggests that, in contrast to the fusion products presented to the mitochondrial surface by targeting sequences, the fusion product containing the third transmembrane loop has a different conformation and/or mitochondrial localization. It is conceivable that the intramitochondrial localization of the later fusion product is due to a distinct membrane potential (ΔΨ)-responsive signal as suggested in [7].

Taken together the results of in vivo analysis of carrier protein import presented in this study are in basic agreement with the mechanisms implied by in vitro import studies with isolated organelles concerning the function and localization of targeting and internalization sequences. The prerequisite for in vitro studies, however, is the use of fully intact, coupled mitochondria, isolated from the standard respiratory-competent cells. This excludes the possibility of investigating the import process in cells where the mitochondrial functions are compromised due to mutations, growth conditions, or presence of inhibitors. Therefore the use of intact cells and fusion proteins should be a useful, and probably the only, alternative for analysis of mitochondrial protein import at the above conditions.


This work was supported by grants from VEGA (No. 7285/20), and Science and Technology Assistance Agency (No. APVT-20-017202) to J.K.

ADP/ATP carrier
green fluorescent protein
synthetic medium
translocase of outer mitochondrial membrane
translocase of inner mitochondrial membrane


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
  14. [14]
  15. [15]
  16. [16]
  17. [17]
  18. [18]
  19. [19]
  20. [20]
  21. [21]
  22. [22]
  23. [23]
  24. [24]
  25. [25]
  26. [26]
  27. [27]
  28. [28]
  29. [29]
View Abstract