INTRODUCTION

Although circularity is accepted as the most common form of mitochondrial DNA (mtDNA), there are numerous examples of organisms with linear mtDNA, raising questions about their evolutionary origin and mode of replication (1). Mitochondrial genomes of many yeast species are represented by linear DNA molecules with defined terminal structures (2-5). Two different types of yeast linear mtDNA have been described. Type 1 (Pichia pijperi, Pichia jadinii, Williopsis mrakii) possesses inverted terminal repeats with a covalently closed single-stranded hairpin at each ends (6). In contrast, terminal structures of type 2 linear mtDNA (Candida parapsilosis, Candida salmanticensis, Pichia philodendra) have inverted terminal repeats consisting of units repeated in tandem. The variable number of repeat units generates a population of mtDNA molecules of heterogeneous size. The largest fraction of mtDNA molecules of C. parapsilosis is represented by the shortest molecules, which contain only one incomplete tandem repeat unit. Regardless of the number of repetitions, both ends of the mtDNA molecule terminate at a unique position in the last repeat unit sequence thus leaving an open structure with a 5 single-stranded extension of about 110 nucleotides (7). This is in contrast to typical nuclear telomeres of most eukaryotes consisting of simple arrays of tandem G-rich repeats that run 5 to 3 at the chromosome ends and protrude as 3 overhangs (8). The structure of the mitochondrial telomeres of C. parapsilosis is more reminiscent of those of Tetrahymena mtDNA, although the telomeric tandem repeat unit of the latter is considerably shorter (9, 10). The role of terminal structures in the replication and maintenance of C. parapsilosis mtDNA is not known. Their unique organization raised many intriguing issues, e.g. (i) how the terminal 5 single-stranded extensions are stabilized in mitochondria; (ii) why DNA polymerase stops at the 3 end and does not fill the protruding 5 overhang; (iii) how the 5 overhang is generated; and (iv) how the shortest molecules restore the missing sequence, to name a few.

The search for proteins specifically recognizing mitochondrial telomeres of C. parapsilosis may reveal essential components of the telomere replication machinery (e.g. analog of the nuclear telomerase) and proteins that play a role in the stabilization, maintenance, and proper segregation of the linear mtDNA. For nuclear genomes, specific telomere-binding proteins (TBPs)¹ have been identified in protozoa (11-16), slime molds (17, 18), yeasts (19-26), algae (27), plants (28), and vertebrates (29-32). These proteins presumably protect the telomere DNA from nucleolytic degradation, mediate telomere-telomere and telomere-nuclear matrix interactions, promote the formation of a typical telomeric structure, and regulate the accessibility of telomeres to the replication machinery (33). They may also participate in the events associated with the cell cycle (34) and cellular senescence (35).

In this paper we describe the identification and purification of the first mitochondrial telomere-binding protein (mtTBP). MtTBP from yeast C. parapsilosis is a heat- and protease-resistant protein that interacts specifically with the sequence of the 5 single-stranded extension from the extreme ends of a linear mtDNA molecule and protects it against various DNA-modifying enzymes. The possible role of mtTBP in telomere stabilization is also discussed.

EXPERIMENTAL PROCEDURES

Oligonucleotides were synthesized by Genset (Paris, France) (their sequences and positions in C. parapsilosis mtDNA are shown in Fig. 1). DNase I and proteinase K were purchased from Boehringer (Mannheim, Germany). Cyanogen bromide-activated Sepharose 4B and RNase A were obtained from Sigma. Trypsin was from Serva (Heidelberg, Germany), exonuclease III and bacterial alkaline phosphatase were provided by Life Technologies, Inc. (Paisley, UK). T4 polynucleotide kinase and mung bean nuclease were from New England Biolabs (Beverly, MA), and S1 nuclease was obtained from Pharmacia (Uppsala, Sweden). [-³²P]ATP was from Amersham (Braunschweig, Germany) or New England Nuclear (Les Ulis, France). Rabbit antisera against mitochondrial malate dehydrogenase and porin of Saccharomyces cerevisiae were generously provided by Dr. Gottfried Schatz (Biocenter, University of Basel, Switzerland). Rabbit antisera against aconitase of S. cerevisiae were obtained from Dr. P. Haviernik (Department of Biochemistry, Comenius University, Bratislava). Restriction enzymes were from New England Biolabs.

Strain SR23 (CBS 7157) of C. parapsilosis is a stock from the Department of Biochemistry, Comenius University, Bratislava. Yeasts were grown in complete liquid medium (1% Bacto-yeast extract (Difco, Detroit, MI), 1% Bacto-peptone (Difco), 2% glucose) at 28 °C with shaking until early stationary phase. For standard preparation, 4 liters of culture were harvested by centrifugation, and cells were washed with cold water and resuspended in 300 ml of 25 mM EDTA (pH 7) containing 4.5% (v/v) 2-mercaptoethanol. After a 20-min incubation at room temperature cells were washed with SCE buffer (1.2 M sorbitol, 0.1 M sodium citrate, 60 mM EDTA (pH 7)), resuspended in 300 ml of SCE containing 50 mg of zymolyase 20T (Kirin Brewing Ltd., Tokyo, Japan) and incubated at 37 °C until about 90% of cells converted to spheroplasts (60-90 min). The reaction was stopped by the addition of an equal volume of ice-cold 1.2 M sorbitol, and spheroplasts were washed twice with 1.2 M sorbitol.

The pellet was resuspended in 200 ml of 0.7 M sorbitol, 60 mM Tris-Cl (pH 7.4), 1 mM EDTA and mixed with a magnetic stirrer for 10 min at 4 °C. Spheroplasts were mechanically broken in Kenwood mixer (30 s at maximal speed), the pH was adjusted to 7.0 with several drops of 1 M Tris base, and cell debris was pelleted by centrifugation for 10 min at 2,000 × g. Mitochondria were pelleted from the supernatant by a 10-min centrifugation at 10,000 × g and washed three times with 0.5 M sorbitol, 50 mM Tris-Cl (pH 7.4), 1 mM EDTA. Fractions from spheroplast crude lysate, postmitochondrial supernatant, and mitochondria were stored in small aliquots at 70 °C.

For the extraction of mtTBP, mitochondria were resuspended in 10 volumes of buffer A (10 mM Tris-Cl (pH 7.4), 1 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 10 µg/ml leupeptin, 5 µg/ml pepstatin A) and mixed for 20 min at 4 °C with a magnetic stirrer. The suspension was centrifuged for 15 min at 10,000 × g, and the pellet was resuspended in 10 volumes of buffer A using a Potter homogenizer. After a 15-min incubation on ice, the suspension was homogenized using a Potter homogenizer and incubated on ice for an additional 30 min. Triton X-100 was added to a final concentration of 0.1% followed by a 60-min incubation on ice. Triton-insoluble material was pelleted by centrifugation for 45 min at 55,000 × g. The resulting pellet was resuspended in 5 volumes of buffer A containing 2 M NaCl using a Potter homogenizer and incubated on ice for an additional 60 min. The suspension was centrifuged for 45 min at 55,000 × g. The supernatant containing the majority of mtTBP was dialyzed overnight against 200 volumes of 10 mM Tris-Cl (pH 7.4), 1 mM EDTA, aliquoted, and frozen at 70 °C. Protein concentration was determined as described by Bradford (36). The preparation is called crude mitochondrial membrane extract.

Oligonucleotides were 5 end labeled using [-³²P]ATP by T4 polynucleotide kinase and separated from nonincorporated nucleotides by chromatography on a BioSpin-6 column (Bio-Rad).

5 µg of crude mitochondrial membrane protein extract or 1 ng of the purified mtTBP was incubated with 1 ng of labeled oligonucleotide in 10 µl of DNA-binding buffer (10 mM Tris-Cl (pH 7.4), 50 mM NaCl) containing 1 mg/ml poly(dI·dC) for 10 min at room temperature. Poly(dI·dC) was omitted when treated with DNA-modifying enzymes (see below). DNA-protein complexes were resolved by electrophoresis on 4% polyacrylamide gel in 0.5 × TBE (4.5 mM Tris borate, 1 mM EDTA) at 10 V/cm for 90 min. Gels were dried under vacuum at 80 °C and exposed to Kodak X-Omat film. In some experiments, samples were treated with different protein- or nucleic acid-modifying enzymes before or after the DNA-protein binding reaction as indicated in corresponding figure legends. If necessary, different ions were added prior to the addition of enzymes.

Samples indicated in corresponding figure legends were after the DNA binding reaction (10-µl total volume, see above) illuminated by UV light (312 nm, Vilbert-Lourmat transilluminator) from a distance of 3 cm for 45 min. DNA-protein complexes were resolved by 12% SDS-PAGE by the method of Laemmli (37). Gels were heat dried under vacuum and exposed to Kodak X-Omat film at 70 °C between intensifying screens.

For UV cross-linking in situ, a modification of the protocol of Williams et al. (38) was employed. Briefly, after the mobility shift assay, gel was irradiated for 30 min from a distance of 5 cm by UV light (312 nm), positions of the free and shifted probe were visualized by autoradiography, and bands corresponding to both free and complexed probe were excised. Gel slices were boiled for 5 min in 3 ml of a modified SDS-PAGE sample buffer (1% SDS, 3 mM dithiothreitol, 125 mM Tris-Cl (pH 6.8)) and then placed between gel plates in the stacking gel area. Covalently bound DNA-protein complexes were separated from a free probe by 12% SDS-PAGE by the method of Laemmli (37), and the gel was dried at 80 °C under the vacuum and exposed to Kodak X-Omat film at 70 °C between intensifying screens.

TEL51-Sepharose was prepared according to a protocol described by Kadonaga and Tjian (39). The concentration of TEL51 was estimated to be 15 µg/ml of Sepharose. 3 ml of beads was packed into a column and equilibrated by 20 ml of buffer B (10 mM Tris-Cl (pH 7.4), 50 mM NaCl, 0.1 mM EDTA). Dialyzed mitochondrial membrane protein extract (total protein of about 10 mg) was combined with poly(dI·dC) (final concentration 65 µg/ml) and incubated at room temperature for 15 min. The DNA-protein mixture was loaded on the column and incubated with beads for 5 min. The flow-through was reloaded on the column, and this step was repeated three times. Beads were then washed with 50 ml of buffer B. Bound proteins were eluted stepwise with buffer B containing 0.3, 0.5, 0.7, 1.0, 1.5, 2.0, and 3.0 M NaCl. In some experiments, beads were washed with 50 ml of buffer B containing 0.5 M NaCl to remove contaminants weakly bound to the ligand. Selected samples were dialyzed against 1,000 volumes of 10 mM Tris-Cl (pH 7.4), 0.1 mM EDTA, 20% glycerol and stored at 70 °C. For SDS-PAGE analysis, selected samples were dialyzed against double distilled water, concentrated by lyophilization, and dissolved in a SDS-PAGE sample buffer. Proteins were then separated by 15% SDS-PAGE according to the method of Laemmli (37) and stained with silver using a silver stain kit (Pharmacia).

Proteins (50 µg) from the corresponding fractions were separated by 13.5% SDS-PAGE according to the method of Laemmli (37) and electrotransferred to a nitrocellulose filter (BA85, Schleicher & Schuell) in buffer consisting of 25 mM Tris, 192 mM glycine (pH 8.3) for 60 min at 250 mA. The blots were first blocked for 2 h in TBS buffer (10 mM Tris-Cl (pH 7.4), 150 mM NaCl) containing 2% skim milk (Difco) and then probed with anti-malate dehydrogenase (diluted 1:1,000 in a blocking solution), anti-porin (1:500), and anti-aconitase (1:500) antisera, respectively, followed by anti-rabbit immunoglobulin G-alkaline phosphatase conjugate (Sigma). The blots were developed with 0.1 M NaHCO₃, 1 mM MgCl₂ (pH 9.8) containing 0.15 mg/ml 5-bromo-4-chloro-3-indolyl phosphate toluidinium (Sigma), 0.3 mg/ml nitro blue tetrazolium (Sigma).

Published procedures were used to assay glucose-6-phosphate dehydrogenase (40) and cytochrome c oxidase (41).

Appropriate fractions were treated with proteinase K in 50 mM Tris-Cl (pH 7.5), 5 mM EDTA, 150 mM NaCl, 1% N-laurylsarcosine for 2 h at 50 °C. Nucleic acids were extracted by phenol:chloroform:isoamyl alcohol (25:24:1) and precipitated with ethanol. Samples were treated with RNase A following the phenol:chloroform:isoamyl alcohol extraction and ethanol precipitation. DNA samples (250 ng) were digested with restriction enzymes HindIII and PvuII according to the manufacturer's instructions and separated by agarose gel electrophoresis (42).

A gel retardation assay was performed as described above except that 100 ng of the probe and a corresponding amount of protein were subjected to DNA-protein binding reaction. After separation of the DNA·mtTBP complex, the gel was exposed, and the shifted band and bands from the corresponding areas in control lanes (probe without mtTBP and mtTBP without a probe) were excised and denatured by boiling for 5 min in 3 ml of a modified SDS-PAGE sample buffer (1% SDS, 3 mM dithiothreitol, 125 mM Tris-Cl (pH 6.8)). The excised band was then placed between gel plates in the stacking gel area, proteins were separated by 15% SDS-PAGE by the method of Laemmli (37) and stained with silver using a silver stain kit.

Elution of mtTBP from the gel, SDS removal, guanidine hydrochloride treatment, and renaturation were performed essentially as described by Hager and Burgess (43) except that renaturation of the protein was achieved by overnight dialysis of the denatured sample against 50,000-fold excess of 10 mM Tris-Cl (pH 7.4), 1 mM EDTA at 4 °C.

The apparent molecular mass of native mtTBP was determined using gel filtration FPLC on a Superose 12 column (Pharmacia) equilibrated with 10 mM Tris-Cl (pH 7.4), 0.1 mM EDTA, 150 mM NaCl. The column was calibrated with catalase (240 kDa), aldolase (158 kDa), bovine serum albumin (68 kDa), ovalbumin (45 kDa), chymotrypsin (25 kDa) and cytochrome c (12.5 kDa).

All experiments were repeated at least twice with similar results.

RESULTS

Preliminary experiments using the Southwestern assay indicated that most of the DNA binding activity is associated with the inner mitochondrial membrane (data not shown). The majority of membrane-associated DNA-binding proteins were extracted by treatment with 2 M NaCl. To assess the presence of a putative mtTBP in the extract, an oligonucleotide identical to the sequence of 51 nucleotides (TEL51) from the extreme 5 end of the mitochondrial telomere (Fig. 1) was used in a gel retardation assay (Fig. 2). Mixing of TEL51 with mitochondrial membrane extract resulted in a DNA-protein complex that migrated slower than a free probe (Fig. 2, lanes 1 and 2). A 5-100-fold molar excess of unlabeled TEL51 competed quantitatively with the labeled probe for complex formation (Fig. 2, lanes 3-5). Similarly, an oligonucleotide representing the last 31 telomeric bases (TEL31) proved to be a competitor (Fig. 2, lanes 6-8). In contrast, oligonucleotides derived from other parts of mtDNA of C. parapsilosis did not compete with labeled TEL51 as effectively as homologous competitors (Fig. 2, lanes 9-17). Moreover, oligonucleotides complementary to TEL51 (oligonucleotide 4 in Fig. 1), yeast tRNA, and total mitochondrial RNA from C. parapsilosis were found to be ineffective competitors (data not shown). In addition to a major DNA-protein complex representing a specific interaction between the telomeric probe and a putative mtTBP, a slower migrating complex was also observed in some experiments (Fig. 2, lane 2). However, in this case all nonspecific competitors competed successfully with a probe (Fig. 2, lanes 3-17). Moreover, the formation of this complex was not observed when purified mtTBP was used for the gel retardation assay (see below), suggesting that the slower migrating complex probably resulted from nonspecific DNA-protein interaction.

Several assays were performed to assess the purity of mitochondria used for the extraction of mtTBP (Fig. 3). First, mtTBP activity was shown to be completely absent in the postmitochondrial supernatant (Fig. 3A). Second, the mitochondrial fraction was highly enriched for the mitochondrial immunological markers porin, malate dehydrogenase, and aconitase (Fig. 3B). Third, mitochondrial preparation was almost devoid of the cytosolic enzymatic marker glucose-6-phosphate dehydrogenase, which was in contrast to the distribution of cytochrome c oxidase activity (Fig. 3C). Finally, mtDNA isolated from the mitochondrial fraction was not substantially contaminated by nuclear DNA (Fig. 3D). All of these results strongly suggest that mtTBP is localized in mitochondria of C. parapsilosis.

C. parapsilosis is a petite negative yeast, and the stability of its mtDNA is probably essential for cell viability. Resistance of the mitochondrial telomeric structure to various damaging agents including proteases and DNA-modifying enzymes might reflect the need to maintain the integrity of the mitochondrial genome. The susceptibility of mtTBP to proteinase K and trypsin and the effect of heat on mtTBP activity were assayed by pretreatment of the extract followed by a gel retardation analysis using labeled TEL51 (Fig. 4A). Interestingly, mtTBP was highly resistant to proteinase K (Fig. 4A, lane 4) and partially resistant to both trypsin and to 15-min incubation at 62 °C (Fig. 4A, lanes 5 and 6). Moreover, neither an overnight incubation at 30 °C nor treatment with 0.1% SDS destroyed the mtTBP activity (data not shown).

Next, the susceptibility of the mtTBP·TEL51 complex to various DNA-modifying enzymes was evaluated (Fig. 4B). None of the tested enzymes completely destroyed the complex, whereas a free probe that served as an internal control was degraded completely in each case. To demonstrate that the enzymes are active in the mitochondrial extracts, reactions were also performed in the presence of the molar excess of homologous competitor (Fig. 4B, 3rd, 6th, 9th, 12th, 15th, and 18th lanes). Since the probe was terminally labeled by T4 polynucleotide kinase, resistance of mtTBP·TEL51 to bacterial alkaline phosphatase (Fig. 4B, 4th-6th lanes) suggests that mtTBP protects the very end of the mitochondrial telomere of C. parapsilosis. These results imply that mtTBP may play a similar protective role in vivo thus preventing mtDNA from degradation.

To estimate the apparent molecular mass of a mtTBP·TEL51 complex, UV cross-linking experiments both in solution and in situ were performed (Fig. 5, A and B). Both of these assays gave rise to two prominent bands with apparent molecular masses of ~32 and 40 kDa, respectively (Fig. 5, panel A, lane 4, and panel B, lanes 2 and 4). The larger complex was more abundant in the sample after UV cross-linking in solution, whereas irradiation of mtTBP·TEL51 complex in mobility shift gel resulted in a higher proportion of the faster migrating complex (Fig. 5B). The presence of a 100-fold excess of either TEL51 or TEL31 completely competed out a labeled probe from the complex (Fig. 5A, lanes 5 and 6). On the other hand, oligonucleotides derived from different regions of mtDNA of C. parapsilosis were not as effective as specific competitors (Fig. 5A, lanes 7-9), which is consistent with the results obtained from gel retardation assays.

To purify mtTBP from a crude mitochondrial membrane extract, DNA affinity chromatography based on TEL51 covalently bound to Sepharose was employed. MtTBP activity was eluted from the column with relatively high salt concentrations as shown by both gel retardation (Fig. 6A) and UV cross-linking (Fig. 6B). Formation of a novel complex observed in a sample eluted by 0.5 M NaCl (Fig. 5A, lane 4) was probably due to the presence of a less specific TEL51-binding protein and the absence of mtTBP in this fraction. Silver staining of proteins after SDS-PAGE revealed the presence of a major protein band with an apparent molecular mass of about 15 kDa (Fig. 6C). Strong interaction between mtTBP and immobilized telomeric oligonucleotide caused most contaminants to elute with 0.3-0.5 M NaCl. Purified preparations of mtTBP from TEL51-Sepharose contained only minor impurities with a molecular mass above 40 kDa.

Because of an apparent contradiction between the results of the UV cross-linking assay (which gave rise to two prominent bands with apparent molecular masses of ~32 and 40 kDa, respectively) and the abundance of 15-kDa protein (p15) in the purified fractions, two experiments were performed to prove that p15 represents mtTBP. First, the mtTBP·TEL51 complex was excised from the mobility shift gel, denatured, and separated by SDS-PAGE. As controls gel slices were used from the same regions of the lanes with a probe and mtTBP activity containing fraction alone, respectively. A protein band comigrating with purified p15 was detected only in a sample containing both probe and mtTBP (Fig. 7A).

Second, affinity-purified p15 was isolated from the gel after SDS-PAGE, precipitated with acetone to remove SDS, resuspended in 6 M guanidine hydrochloride, renatured by dialysis, and tested for mtTBP activity (Fig. 7B, lane 3). As a control, the same protocol was performed on a gel slice in which no protein was present (Fig. 7B, lane 4). Although it lost some of the mtTBP activity, renatured p15 was able to bind the TEL51 probe in a gel mobility shift assay. Taken together these data demonstrate that p15 represents mtTBP and is responsible for the generation of the complexes in both gel retardation and UV cross-linking assays.

It was shown previously that under standard SDS-PAGE conditions, covalently linked protein-DNA complexes usually migrate with the same electrophoretic mobility as the protein alone (38). However, purified mtTBP migrated in SDS gel as a 15-kDa protein band (Fig. 6C), whereas the UV cross-linked mtTBP·TEL51 complex exhibited an apparent molecular mass above 30 kDa (Fig. 5). The possibility that this discrepancy might be due to an oligomeric state of the native mtTBP was tested using FPLC gel filtration chromatography on Superose 12 (Fig. 8). TEL51-retarding activity was eluted between bovine serum albumin (68 kDa) and ovalbumin (45 kDa), suggesting that native mtTBP is either a trimer or tetramer. The absence of any major protein band other than p15 in purified mtTBP preparations suggests that mtTBP is a homo-oligomer.

DISCUSSION

Several lines of evidence indicate that nuclear telomeres form specific nucleoprotein structures (telosomes), consisting of telomeric DNA and various proteins involved in both replication and stabilization of the chromosome ends. Moreover, these structures are responsible for distinct interactions in the nucleus and influence the expression of subtelomeric genes (8, 44-47).

Specific nuclear TBPs from a variety of organisms were recently identified (33, 48). The first isolated and best characterized is TBP from Oxytricha nova which is a heterodimer of two different polypeptides  and  (12, 13). It binds specifically to the 3 overhang both in vitro and in vivo, protects the telomeric DNA from chemical modification and nuclease digestion, and promotes the formation of a G4 structure typical for telomeric DNA (49). Another well characterized telomeric protein is RAP1 of S. cerevisiae which is a multifunctional polypeptide that binds to the duplex region of the telomeric sequence both in vitro and in vivo and plays not only a protective but also regulatory role at telomeres (20-22, 26, 50-53). The crystal structure of its DNA binding domain in complex with telomeric DNA was reported recently (54). Although several TBPs were isolated from vertebrates (29-32), only the protein described recently by Chong et al. (29) has been shown to be at telomeres in vivo.

Linear mtDNAs were described recently in more organisms than were expected previously. Their genomic organization and terminal structures differ greatly. Different types of mitochondrial telomeres may imply different strategies of their replication which do not resemble the replication of typical nuclear telomeres. For example, linear mtDNAs found in several species from yeast genera Pichia and Williopsis possess a terminal hairpin that may allow the formation of circular replication intermediates (2, 6). In contrast, termini of linear mtDNA of C. parapsilosis consist of tandem repeats that end with a 110-nucleotide-long 5 single-stranded overhang that does not permit the formation of circles (7). It is not clear how the 5 overhang is formed, how it is stabilized, and why DNA polymerase does not use this sequence as a template.

Although data about nuclear TBPs accumulate rapidly, no similar proteins were identified from mitochondria with linear genomes. We considered the possibility that specific mtTBPs are responsible for the stability and/or replication of mtDNA termini of C. parapsilosis. By both gel retardation and UV cross-linking experiments using a 51-base-long oligonucleotide derived from the 5 overhang of the extreme end of linear mtDNA we detected a specific mtTBP in the salt extracts from mitochondrial membranes of this yeast. The interaction of mtTBP with the membrane was relatively strong since lower salt concentrations (<2 M NaCl) or nonionic detergents (1% Triton X-100) did not quantitatively extract mtTBP from the membrane environment. It is possible that the high salt is needed to remove mtTBP from endogenous DNA. However, treatment of the membranes with nucleases did not result in the extraction of mtTBP (data not shown). The mtTBP of C. parapsilosis may at least partly mediate the association of mtDNA with the inner mitochondrial membrane, which might be necessary for its replication and proper segregation into newly formed mitochondria. Association of mtDNA with the mitochondrial membrane was demonstrated for Physarum polycephalum and S. cerevisiae, and several protein components responsible for this interaction were detected (55-57).

One of the interesting characteristics of mtTBP was its resistance to protease and heat treatments. This resistance was higher in a crude fraction than in purified form. The crude fraction did not contain general protease inhibitors since more than 90% of other proteins were degraded by the two proteases tested (data not shown). It is possible that mtTBP in a crude fraction is in association with a partner that protects it from the action of proteases or which in a diluted state mtTBP is more susceptible to proteolytic cleavage. Moreover, mtTBP activity was not affected by pretreatment of the sample with RNase A (data not shown), excluding the possibility that mtTBP is a ribonucleoprotein complex similar to nuclear telomerase (58). The resistance of mtTBP to heat (<7 min at 62 °C) was helpful in removal of contaminating Mg²⁺-dependent nucleases. This allowed us to test the sensitivity of the mtTBP·TEL51 complex to various DNA-modifying enzymes, some of which require magnesium ions (DNase I, exonuclease III). None of the used enzymes modified DNA in the complex. Surprisingly, free TEL51 was sensitive to exonuclease III (Fig. 4B, 16th lane) as this enzyme is specific for double-stranded DNA molecules with protruding 5 ends. This suggests that TEL51 may adopt a double-stranded structure under the reaction conditions employed in this experiment. The relevance of such structures for in vivo function(s) of mitochondrial telomere remains to be determined. Since the probe was labeled terminally with polynucleotide kinase and the label was not lost by the action of bacterial alkaline phosphatase, mtTBP probably protects the very end of the telomere. Tight interaction between mtTBP and mitochondrial telomere may also play a role in preventing DNA polymerase from resuming synthesis of the 3 end. In addition, mtTBP may be involved in regulating the length of the mitochondrial telomere.

Preliminary experiments demonstrated that the formation of a complex between mtTBP and a telomeric oligonucleotide is resistant to relatively high salt concentrations (data not shown). This allowed us to purify mtTBP to near homogeneity in a single step using DNA affinity chromatography. Identification of a single protein with a molecular mass of about 15 kDa in the active fractions was in apparent contradiction with the data from UV cross-linking experiments, where the mtTBP·TEL51 complex migrated as two bands with molecular masses of about 32 and 40 kDa, respectively. Three different chemical cross-linkers (3,3-dithiobis(sulfosuccinimidylpropionate), bis(sulfosuccinimidyl)suberate, potassium permanganate) gave similar results when compared with cross-linking by UV light (data not shown). Since these two bands were formed also by UV cross-linking of a mobility shift gel in situ and since the 15-kDa protein was extracted from the retarded band after gel retardation assay we tested the possibility that native mtTBP is a homo-oligomer. Data from FPLC gel filtration chromatography supported this hypothesis. The nature of two differently migrating bands after UV cross-linking of mtTBP to a probe remains to be determined.

MtTBP identified in this study is the first example of a protein specifically recognizing the terminal structures of linear mtDNA. It remains to be shown that the mtTBP functions effectively in vivo as a termini-binding protein or fulfills some other tasks. To answer this question, the gene encoding mtTBP will be cloned,² and mutations will be created for functional analysis. Such studies are complicated by the absence of a genetic transformation system and by the asexual mode of reproduction of C. parapsilosis which does not allow the use of powerful molecular and genetic tools available for S. cerevisiae. In spite of the lack of in vivo data the specificity of mtTBP binding to a telomeric sequence in vitro suggests that it plays a specific role in mitochondria. Beside the protective role that is implied by the resistance of the mtTBP-telomere complex to various DNA-modifying enzymes mtTBP may play a part during the replication of mtDNA termini. Because of their unusual structure, mitochondrial telomeres of C. parapsilosis may reveal a novel strategy to solve the end replication problem associated with linear DNA genomes. In addition, mtTBP might be involved in the recombination and segregation of mtDNA molecules in C. parapsilosis, thus serving as a guard of mtDNA integrity of this petite negative yeast. It will be also interesting to find the mtTBP-encoding gene and its homologs in other organisms. At this stage we cannot rule out the possibility that mtTBP is encoded by mtDNA since only one-third of the C. parapsilosis mtDNA sequence is known, and linear cytoplasmic genomes (viruses, plasmids) often code for their terminal proteins (59). All of these questions are addressed in experiments that are currently in progress in our laboratory.