Functional Domains of Chicken Mitochondrial Transcription Factor A for the Maintenance of Mitochondrial DNA Copy Number in Lymphoma Cell Line DT40*

,

Nuclear and mitochondrial (mt) forms of chicken mt transcription factor A (c-TFAM) generated by alternative splicing of a gene (c-tfam) were cloned. c-tfam mapped at 6q1.1-q1.2 has similar exon/intron organization as mouse tfam except that the first exons encoding the nuclear and the mt form-specific sequences were positioned oppositely. When cDNA encoding the nuclear form was transiently expressed in chicken lymphoma DT40 cells after tagging at the C terminus with c-Myc, the product was localized into nucleus, whereas the only endogenous mt form of DT40 cells was immunostained exclusively within mitochondria. c-TFAM is most similar to Xenopus (xl-) TFAM in having extended C-terminal regions in addition to two high mobility group (HMG) boxes, a linker region between them, and a C-terminal tail, also found in human and mouse TFAM. Similarities between c-and xl-TFAM are higher in linker and Cterminal regions than in HMG boxes. Disruption of both tfam alleles in DT40 cells prevented proliferation. The tfam ؉ /tfam ؊ cells showed a 50 and 40 -60% reduction of mtDNA and its transcripts, respectively. Expression of exogenous wild type c-tfam cDNA in the tfam ؉ /tfam ؊ cells increased mtDNA up to 4-fold in a dose-dependent manner, whereas its transcripts increased only marginally. A deletion mutant lacking the first HMG box lost this activity, whereas only marginal reduction of the activity was observed in a deletion mutant at the second HMG box. Despite the essential role of the C-terminal tail in mtDNA transcription demonstrated in vitro, deletion of c-TFAM at this region reduced the activity of maintenance of the mtDNA level only by 50%. A series of deletion mutant at the tail region suggested stimulatory and suppressive sequences in this region for the maintenance of mtDNA level.
Transcription of mitochondrial (mt) 1 DNA is best understood for Saccharomyces cerevisiae (1,2), in which a core RNA polymerase of M r ϳ 145 with sequence similarity to viral RNA polymerases of T3, T7, and SP6 (3) initiates transcription together with a dissociable 39-kDa mt transcription factor B (sc-mtTFB) (4,5). This protein functions like a bacterial factor (6 -8), but amino acid sequence comparisons (9) and mutational analyses (10) do not strongly support the hypothesis that sc-mtTFB is homologous to bacterial factor. The Xenopus equivalent (xl-mtTFB) having this activity has been characterized (11)(12), but direct evidence confirming the structural similarity of xl-mtTFB to sc-mtTFB has not been reported. Human mt-TFB has been cloned and shown to be structurally related to RNA adenine methyltransferase (13). These two classes of proteins bind cooperatively to a nonanucleotide promoter sequence (4,14) present at multiple locations in the 75-kb circular yeast mt genome (15). Thus, mtTFBs acts as a specific factor enabling the RNA polymerase to locate the appropriate start sites.
Our knowledge about the transcription of human mtDNA, mainly based on the in vitro transcription system, is somewhat different and depends on a transcription factor (h-TFAM; formerly referred to as h-mtTFA) containing two HMG boxes (2,16,17). The yeast homologue sc-mtTFA is an abundant protein composed almost entirely of two HMG boxes separated by a rather short linker region (18). sc-mtTFA was originally described as ABF2, which could bind to the replication origin of yeast nuclear DNA but was later found to be localized in mitochondria (19). Disruption of the ABF2 gene led to a loss of respiratory competence and mtDNA when cells were grown in the presence of glucose. Because expression of h-TFAM in the yeast abf2 strain improved the phenotype, a potential functional homology of h-TFAM to sc-mtTFA was confirmed (20). Nonetheless, the role of sc-mtTFA in mtDNA transcription remains unclear since in vitro it can hardly activate transcription. By replacing the various regions of h-TFAM with the corresponding parts of the yeast homologue, the linker region and the C-terminal tail abbreviated in sc-mtTFA were shown to be significant for the recognition and transcriptional activation of the human mtDNA (21). Mouse (22) and Xenopus mtTFA (25) (will be referred to as m-and xl-TFAM hereafter, respectively) have been cloned and shown to have the long linker region and the C-terminal tail as h-TFAM does. xl-TFAM stim-ulates transcription in vitro up to 10-fold for the H-strand, but only ϳ3-fold for the L-strand (23).
For better understanding of the function of TFAM, the evolutionary variation of the control region of mtDNA may be worthwhile to be considered. This is the region between the genes for tRNA Pro (tRNA Glu in birds) and tRNA Phe containing the H-strand replication origin, the promoters for transcription activation, and H-and L-strand start sites (24). In Xenopus, the distance of the transcription start sites of the L-and H-strands is rather short, with 50 bp (25) versus 150 bp in the case of human (26). Lower dependence of Xenopus mtDNA transcription on xl-TFAM compared with the human system can be due to this difference of the control region. From this aspect, chicken mtDNA provides an opportunity to expand our knowledge of mtDNA transcription in higher vertebrates. The chicken mitochondrial genome has a characteristic organization of the genes compared with other vertebrates (27), and its H-and L-strand sites have no separating base pair (28). We have cloned and determined the cDNA sequence encoding chicken (c-)TFAM, which is more similar to xl-TFAM than to the human and mouse proteins. Targeted disruption of both c-tfam alleles in a chicken cell line DT40 showed that TFAM is essential for proliferation of this cell line. In cells with heterozygous disrupted c-tfam, mtDNA and its transcripts were reduced. Expression of exogenous wild type c-tfam cDNA in those cells increased mtDNA up to 4-fold in a dosage-dependent manner, whereas its transcripts increased only marginally. We thus show that c-TFAM is important for maintenance of the mtDNA copy number.

MATERIALS AND METHODS
Cell Culture, Gene Transfection, and Screening-DT40 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 50 units/ml penicillin, 50 g/ml streptomycin at 37°C under a humidified atmosphere of 5% CO 2 . Cells were split to 1 ϫ 10 5 /ml every third day. For DNA transfection, 10 7 cells were suspended in 0.5 ml of PBS containing 30 g of linearized plasmid and electroporated with a Gene Pulser (Bio-Rad) at 550 V and 25 microfarads. The cells were transferred to 20 ml of fresh medium and incubated for 24 h. After resuspending in 100 ml of medium containing either 30 g/ml mycophenolic acid or 2 mg/ml G418, cells were divided into five 96-well plates and cultured for 7-10 days. Drug-resistant colonies were transferred to 50 ml of T-flasks for preparation of genomic DNA to confirm heterozygous disruption of tfam alleles by Southern blotting. For screening of DT40 clones having both tfam alleles disrupted, the colonies with heterozygous disruption were subjected to another round of transfection/screening cycle using a medium supplemented with 50 g/ml uridine and 110 g/ml sodium pyruvate. To obtain the colony expressing exogenous c-tfam cDNA and its deletion mutants under the background of heterozygous tfam allele disruption, plasmids having either of these sequences and neomycin-resistance gene (neo r ) in tandem, both under the control of ␤-actin promoters, were transfected to the cells with either tfam allele targeted by MPA, and the G418resistant colonies were selected. The colonies positive to exogenous c-TFAM were further selected by Western blotting of the cell lysate.
Screening and Sequencing of cDNA and Genomic DNA Clones-A chicken cDNA library in which mRNAs isolated from the day 5 embryos were reverse-transcribed using oligo-dT primers, and double-stranded cDNA were inserted into EcoRI-XhoI sites of ZAPII vector was a gift from Dr. Atsushi Kuroiwa. ϳ1 ϫ 10 7 phages were screened using the SalI fragment of h-tfam cDNA (17) as a probe in a buffer containing 50 mM Tris-HCl (pH 7.5), 5ϫ SSC, 5ϫ Denhardt's solution, 1% SDS, and 100 g/ml salmon sperm DNA. After hybridization (16 h, 53°C), filters were washed 3 times with 2ϫ SSC containing 0.1% SDS at room temperature and once with 0.2ϫ SSC containing 0.1% SDS at 53°C for 30 min and exposed to x-ray film (Fuji). Because the cDNA clones isolated from this library were found to encode only nuclear c-TFAM (GenBank TM accession number AB21166), the 5Ј sequence encoding the N terminus of mt c-TFAM (GenBank TM accession number AB059657) was obtained by reverse transcription of RNA extracted from DT40 cells with SMART rapid amplification of cDNA ends cDNA amplification kit (Clontech) using a primer of 5Ј-CACCAGCTCCAAGCTGTTCAG-3Ј. The amplified fragment was sequenced with a primer of 5Ј-GCTGGT-TGTCCCTCAGGAAG-3Ј. For transient expression of the nuclear or mt c-tfam cDNA tagged with a c-Myc sequence at 3Ј end, either cDNA clone was amplified with 3Ј primer of 5Ј-GCGGGATCCTTACAAGTCTTCT-TCAGAAATAAGCTTTTGTTCTTCTTCAGATTTTTTCAGCTTCAGTT-TCGCCAG-3Ј paired with 5Ј-primer of 5Ј-CCCggTCgACACCGCGGTG-GCGGCCGCCATGGGGCGGAGGGGCGGGCTG-3Ј or 5Ј-GGCGGTC-GACTCCGTGGGGAGGGTGGGCTGCGGCATG-3Ј, respectively. The amplified fragments were inserted downstream of ␤-actin promoter after SalI/BamHI digestion.
A DT40 genomic DNA library with Sau3A partial digests (15-25 kb) inserted into the XhoI site of a FIXII vector was a gift from Dr. Tatsuo Nakayama. ϳ1 ϫ 10 6 phages were screened at a hybridization temperature of 65°C using a c-tfam cDNA clone as a probe. Genomic DNA clones were analyzed by restriction endonuclease mapping and Southern blot analysis. Pertinent restriction fragments were subcloned into pBluescript. DNA was sequenced by the dideoxynucleotide method on an Applied Biosystems (model 377) DNA sequencer.
Preparation of Anti-chicken TFAM Antiserum-To express c-TFAM in Escherichia coli, the SacI-HindIII fragment of c-tfam cDNA coding for amino acid residues Ala-42 to Leu-248 was cloned into the SacI-  (42). The secondary structure of the second HMG box of rat HMG1 as evaluated by NMR (Protein Data Bank code 1hme (43)) is depicted as a schematic below the alignment. In the structure of the corresponding domain of hamster HMG1, helix 1 is interrupted near the middle, and helix 3 is shortened by five residues as indicated by dashed lines (44). Motif boundaries were chosen based on deletion mutants in DNA binding assays (45)(46)(47) and extensive sequence alignments. Arrows denote residues for which point mutation studies performed on different HMG box-containing proteins have shown a loss of DNA binding capacity. Conserved residue types (h, hydrophobic; ϩ/Ϫ, charged; P, proline) shown in the bottom line are assessed qualitatively from more complete alignments. Triangles mark the positions of exon boundaries within c-and m-tfam. Columns on the right give the BLOSUM62 scores resulting from pairwise comparison of the first (H1) and second (H2) c-TFAM HMG box regions with the adjacent sequences as well as the sum of the two values.
plete Freund's adjuvant. Sera were collected 4 weeks after initial injection.
Western Blotting-Cell extracts corresponding to 10 g of total protein were separated in 15% SDS-polyacrylamide gels and transferred to Hybond enhanced chemiluminescence (ECL) nitrocellulose (Amersham Biosciences). Filters were preincubated with 5% skim milk in PBS, incubated with anti-c-TFAM (1:500 in buffer A) for 1 h, washed 4 times with buffer A, incubated with horseradish peroxidase-conjugated antirabbit IgG (Amersham Biosciences) for 1 h, and washed with buffer A.
Protein bands were visualized using the ECL Western blotting reagents (Amersham Biosciences).
Chromosome Preparation and in Situ Hybridization-Chromosomal assignment of c-tfam was performed using the direct R-banding fluorescence in situ hybridization method as described (29,30) with minor modifications. Mitogen-stimulated splenocytes were cultured at 39°C and synchronized by a thymidine block, and 5-bromodeoxyuridine was incorporated during the late replication stage for differential replication staining after release of excessive thymidine. R-bands were stained by exposure of Hoechst 33258-stained chromosome slides to UV light. Chromosome slides were heated at 65°C for 2 h, denatured at 70°C in 70% formamide in 2ϫ SSC, and dehydrated in a 70, 85, 100% ethanol series at 4°C. The 15-kb chicken genomic DNA fragment inserted into the SalI site of pBluescript was labeled by nick translation with biotin 16-dUTP (Roche Applied Science), ethanol-precipitated together with chicken whole genomic DNA, salmon sperm DNA, and E. coli tRNA, and denatured at 75°C in 100% formamide for 10 min. The denatured probe was mixed with an equal volume of hybridization solution to a final concentration of 50% formamide, 2ϫ SSC, 10% dextran sulfate, and 2 mg/ml bovine serum albumin. 20 l of the solution containing 250 ng of labeled DNA were put on the denatured slide, covered with Parafilm, and incubated overnight at 37°C. Slides were washed in 50% formamide in 2ϫ SSC at 37°C for 20 min, and 2ϫ and 1ϫ SSC at room temperature for 20 min. After rinsing in 4ϫ SSC, slides were incubated with Cy2-labeled streptavidin (Amersham Biosciences) at 1:500 dilution in 1% bovine serum albumin, 4ϫ SSC at 37°C for 1 h. After washing with 4ϫ SSC, 0.1% Nonidet P-40 in 4ϫ SSC, and 4ϫ SSC for 10 min each on a shaker, slides were rinsed with 2ϫ SSC and stained with 0.75 g/ml propidium iodide. Hybridization was visualized by excitation at 450 -490 nm (Nikon filter set B-2A) and ϳ365 nm (UV-2A), and recorded on Kodak Ektachrome 100 film.
Southern Blotting-Genomic DNA was purified from DT40 cells with DNAZOL (Invitrogen). 10 g of DNA were separated by 1% agarose gel electrophoresis after digestion with EcoRI, SalI, and/or PstI and transferred to Hybond Nϩ (Amersham Biosciences). Hybridization was performed in a buffer containing 50 mM Tris-HCl (pH 7.5), 5ϫ Denhardt's solution, 1% SDS, and 0.1 mg/ml salmon sperm DNA at 65°C for 16 h. Filters were washed 3 times with 2ϫ SSC containing 0.1% SDS at room temperature for 10 min and once with 0.2ϫ SSC containing 0.1% SDS at 65°C for 30 min and analyzed with a Fuji BAS 2000 image analyzer. For the screening of DT40 clones, of which tfam was disrupted, a 0.5-kb probe (probe 1 in Fig. 3A) was prepared by amplifying the SalI-EcoRI fragment inserted in pBluescript KS(ϩ) with a paired primer of 5Ј-TAGTTCCCTTTCTGTCAAAG-3Ј and T7 primer and digesting it with EcoRI. For semi-quantification of mtDNA, the EcoRI-StuI fragment of chicken ATPase 6/8 cDNA was used. A 5-kb EcoRI fragment of tfam (probe 2 in Fig. 3A) was used for the normalization of the quantification of mtDNA. Because this probe sequence is localized outside of the targeted region, its signal cannot be affected by the disruption of tfam in DT40 cells.
Targeting Vector Construction-As a source of targeting vectors, plasmids with MPA (32) and neo r (33) under the control of the chicken ␤-actin promoter were inserted into pBluescript II. Plasmids were gifts of Dr. Shunichi Takeda. Selection marker genes were sandwiched by a 1.8-kb sequence upstream of exon I M and a 4.6-kb sequence spanning the 3Ј terminus of exon II through intron VI as left and right arms, respectively (cf. Fig. 3A). For the left arm insertion, a PstI site of the 1.8-kb SalI-PstI fragment was modified to a SalI site. For the right arm insertion, a BamHI-BamHI fragment was prepared by PCR using paired primers of 5Ј-CCCGGGATCCCTTCCTGAGGGACAACCAGC-3Ј and 5Ј-CCCGGGATCCCACCTGTGCAGTACTCATCC-3Ј.
Expression Vectors for TFAM Deletion Mutants-Paired c-tfam cDNA fragments that sandwich the domain to be deleted were prepared by PCR so as to have restriction ends of 5Ј-SalI/blunt-3Ј or 5Ј-blunt/BamHI and were inserted downstream of ␤-actin promoter by a three-piece ligation technique. The deleted sequences of c-TFAM were Lys-46 -Try-112 for the first HMG box, Lys-z13-Val-146 for the linker, Leu-147-Trp-213 for the second HMG box, and Glu-214 -Glu-262E for the tail. A series of deletion mutants at the tail was also prepared; Val-218 to Glu-262 for del M, Ser-228 to Glu-262 for del R, Ser-244 to Glu-262 for del G, and Lys-254 to Glu-262 for del A. The vectors for expression of the mutants had neo r under the control of ␤-actin promoter for the selection of positive colonies.
Sequence Analysis-General sequence editing and analysis were performed using MacDNASIS, version 3.7 (Hitachi Software). Similarity searches against the non-redundant GenBank TM data base (June 1999) were done using BLAST 2.0 (34). HMG boxes were compared with those of the Pfam data base (version 4.0; Ref. 35). Multiple sequence alignments were performed with ClustalW (36). Pairwise comparisons were done using the BLOSUM62 matrix and default values for gap penalties.
Quantification of the Blot Signals-A Southern blot was analyzed by a Fuji Film BAS 2000 Image Analyzer, and the radioactivity was quantified by determining the photostimulated luminescence count with a Fuji Film Image Gauge, version 3.4. The ECL signal Western blotting was semiquantified with Image ID program (Amersham Biosciences).

RESULTS AND DISCUSSION
cDNA Cloning of Nuclear and mt c-TFAMs-Low stringency screening of 1 ϫ 10 7 independent cDNA clones of mRNAs expressed in the day 5 chicken embryos with h-tfam cDNA probes yielded five selected clones. Partial sequencing and restriction fragment analysis of the inserts showed that they share identical 3Ј-sequences with different extensions in the 5Ј direction. Three clones had the same sequence (AB021166) of 1324 bp, whereas the other two clones contained sequences of 1174 and 1008 bp, respectively. The longest sequence encodes an open reading frame of 792 bp starting with ATG at the fourth base and terminating at TAA at the 769th base to code for a 264-residue amino acid sequence. Neither poly(A)-adding signal nor poly(A) tail was found.
Attempts of rapid amplification of cDNA ends (37) using mRNA extracted from DT40 cells as the template yielded 5Јextended sequence (AB059657) encoding different N-terminal sequence (Fig. 1A), suggesting that chicken cells express two forms of c-TFAM. When these cDNA sequences obtained from the day 5 chicken embryos were transiently expressed in DT40 cells after tagging with the c-Myc sequence at the C terminus, indirect immunofluorescence staining of the cells with anti-c-Myc polyclonal IgG showed localization of the product into the nucleus, whereas the endogenous form in DT40 cells was local-  Fig. 3B) and tfam ϩ /tfam Ϫ clones (lanes 10 -12) was digested with SalI and EcoRI. 10 g were separated on a 1.0% agarose gel and transferred to a nylon membrane. As an internal control, the membrane was first hybridized with 32 P-labeled probe 2 (see Fig. 3A) (A) and then rehybridized with a 32 P-labeled mtDNA fragment spanning the ATPase 6/8 genes (B). C, mtDNA signals of tfam ϩ /tfam ϩ and tfam ϩ /tfam Ϫ cells were quantified as described under "Materials and Methods" and normalized by the internal control. ized in mitochondria ( Fig. 1B; see below). We thus cloned mt and nuclear forms of c-TFAM.
For mouse spermatocytes, a testis-specific TFAM isoform resulting from alternative splicing of a different exon I T has been characterized in addition to mt mouse m-TFAM, and this isoform is targeted to the nucleus (22). Screening of a DT40 genomic DNA library with the mt c-TFAM cDNA sequence as a probe yielded three overlapping fragments. Analysis of restriction sites and partial sequencing resulted in the exon/intron organization of c-tfam as shown in Fig. 1A. The N-terminal sequence of nuclear c-TFAM is encoded by exon I N , which is directly connected to exon II without insertion of an intron, whereas that of mt c-TFAM is encoded by exon I M and continued to the sequence encoded by exon II by splicing out of the exon I N . Thus, the mechanism of generating the nuclear and mt c-TFAMs is similar to m-tfam, but the exons encoding the mt and nuclear form-specific sequences (exon I N and exon I M ) are positioned oppositely. The downstream cDNA sequence is encoded by additional six exons. All exon/intron boundaries had well conserved consensus sequences for splicing both at the 5Јand 3Ј-end of introns (not shown). Immunostaining of DT40 cells with antiserum prepared against c-TFAM showed staining of mitochondria well overlapping with that of living dye for mitochondria (MitoTracker Red), and no signal was indicated in the nuclear region (Fig. 1C). Fig.  2A shows an identity of 42, 37, and 40%, respectively. In contrast to h-and m-TFAMs, c-TFAM shares with xl-TFAM the C terminus-extended regions. Its calculated molecular mass (M r ϭ 29,985) is larger than that of h-(M r ϭ 29,097) and m-TFAM (M r ϭ 27,988) but smaller than that of xl-TFAM (M r ϭ 35,501). Its theoretical isoelectric point of 10.7 is comparable with that of Xenopus (pI ϳ 10.2), which is slightly more basic than that of h-and m-TFAMs (pI ϳ 9.7). As do other TFAMs, the chicken sequence contains two HMG boxes separated by a ϳ35-amino acid residue linker. Whereas the HMGbox sequences show high correlations between the different species, the closest relationship between c-and xl-TFAMs in the whole sequences is mainly due to the similarities of the C-terminal extensions and the linker region.

Sequence Comparison of c-TFAM with xl-, m-, and h-TFAMs-Alignment of the deduced c-TFAM amino acid sequence with those of xl-(23), h-(17), and m-(22) TFAMs in
By replacing the various regions of h-TFAM with the corresponding parts of the yeast homologue, the significance of the linker region and the C-terminal tail for the recognition and transcriptional activation of the human mtDNA promoter has been established (21). For c-TFAM, the isoelectric point of the linker region is predicted as pI ϳ 11.9, which is similar to that estimated for xl-and m-TFAMs and higher than for h-TFAM (pI ϳ 10.0). Assuming that the two HMG regions recognize the repeated half-sites in the control region of mtDNA, as suggested for human, mouse, and Xenopus (21,38), it is interesting to note that the mouse and Xenopus half-sites are separated by three to six A/T-rich sequences, whereas those of human consist of G/C rich regions. Chicken mtDNA promoter sequences are not yet specified in such detail, but the highly basic character of the linker region shared with that of mouse and Xenopus might indicate a similarity in the binding sites. Besides, the chicken and Xenopus linker regions have a 60% sequence identity, and the C-terminal tail of both is similarly extended by ϳ20 (chick) and ϳ30 (Xenopus) residues over that of mammalian TFAMs (Fig. 2A). Such similarity could be of importance with respect to the considerably shorter distance of 50 (25) and 0 bp (28) between the transcription start sites of the Land H-strands found for Xenopus and chicken, respectively, which in human extends over 150 bp (26).
HMG domains have the ability to bend DNA and to wrap distorted DNA with low nucleotide sequence specificity (39). The 367 HMG boxes collected in the Pfam data base (35) have an average length and sequence identity of 66 residues and 36%, respectively. Although in most cases they occur as single motifs, in 34 proteins they appear twice in close proximity, and four proteins contain three or four HMG domains. Fig. 2B shows an alignment of the two c-TFAM HMG regions with those of some other proteins, selected on the basis of high similarity. The combined similarity values as expressed as BLOSUM62 scores are highest when compared with xl-TFAM, which is due to the close similarities between the second HMG boxes, whereas the first HMG box has a lower similarity, which is generally found for other proteins having two HMG boxes. The structure of several HMG domains has been solved to high resolution by multidimensional NMR techniques (43,44). As one example, the sequence of the second HMG box of rat HMG1 (r-1hme) of known structure is included at the bottom of Fig.  2B. It contains three ␣-helices of which the first (N terminus) runs antiparallel to the second (middle) helix, and the third (C terminus) helix bends at an angle of 70 -80°relative to the middle helix, forming a V-shaped configuration. The aromatic and basic character of residues corresponding to Phe-14, Phe-17, Trp-45, Lys-53, and Tyr-56 of the r-1hme structure is also found in both c-TFAM HMG boxes (printed in reverse font in Fig. 2B), which in rat and hamster HMG1 cluster to a concave surface formed between the three helices (43,44). Point mutation studies on a number of HMG box-containing proteins indicate the importance of several residues for its DNA binding capacity (arrows in Fig. 2B; Refs. [45][46][47]. Most of the corresponding residues within the c-TFAM HMG boxes appear consistent with those of active DNA binding sequences. Exon/Intron Organization and Chromosomal Assignment of c-tfam-When compared with the cDNA-derived amino acid sequence, all boundaries are well aligned with those of m-and h-TFAM (Fig. 2B). Within the HMG domains, exon boundaries II/III and V/VI are localized within the turn region separating helices 1 and 2 (Fig. 2B). Boundaries III/IV and VI/VII correspond to the beginning of helix 3, with exons IV and VII including the linker and C-terminal tail regions, respectively. This suggests that tfam arose by gene duplication of an ancestral gene covering either exons II to IV, which show a higher sequence variation between species, or exons V to VII.
Chromosomal assignment of tfam was carried out by fluorescence in situ hybridization using a 15-kb genomic fragment as probe. Direct R-banding localized the signal to chicken chromosome 6q1.1-q1.2 (not shown).
Targeted Disruption of c-tfam in DT40 Cells-DT40 is a chicken lymphoma cell line established by infection of chicken with avian leucosis virus. Because this cell line has a high efficiency of targeted integration of foreign DNA (33,48), it provides a convenient system for targeted gene disruption at the cellular level. We prepared the targeting vectors in which MPA (MPA-Tfam) or neo r (Neo-Tfam) were sandwiched by a 1.8-kb sequence upstream of exon I M and a 4.6-kb sequence spanning the 3Ј terminus of exons II through intron VI as left and right arms, respectively (Fig. 3A). In these constructs, exon I M and the 5Ј-terminus of exon I N including the ATGs for the nuclear and the mt c-TFAMs, respectively, were deleted together with the upstream sequence. Probe 1 (Fig. 3A) was prepared for monitoring of the targeted allele by Southern blotting of SalI-EcoRI-digested DNA from drug-resistant clones. This probe is expected to show a 9-and 5-kb fragment for the wild type (tfam ϩ ) and targeted (tfam Ϫ ) allele, respectively, for either of the constructs, whereas no signal was expected for any random integration.
After introduction of MPA-Tfam by electroporation, 30 mycophenolic acid-resistant clones could be isolated, of which 12 were targeted at either of the tfam alleles. Fig. 3B shows the Southern blotting on a half of mycophenolic acid-resistant clones. Additional integration of Neo-Tfam into three expanded clones resulted in 0 of 100 neomycin-resistant clones showing a Southern blotting pattern consistent with a heterozygous tfam deletion. These results were essentially the same when the cells were targeted first with Neo-Tfam followed by MPA-Tfam. With a 40% efficiency for targeted disruption of either of the two tfam alleles, it is unlikely that the second disruption should be difficult to be achieved. This demonstrates that a functional tfam allele is essential for the proliferation of DT40 cells. Although DT40 cells with a heterozygous deletion of Tfam showed no apparent phenotype, the tfam transcript level (see Fig. 5) as well as the amount of c-TFAM was reduced to about half that of wild-type cells (Fig. 3C).
The level of mtDNA was significantly reduced in tfam ϩ / tfam Ϫ DT40 cells, which is in agreement with the data found for heterozygous mice (49). Quantification of the Southern blotting signals of mtDNA in three independent tfam ϩ /tfam Ϫ clones showed a reduction of ϳ50% (Fig. 4). Although the results show a direct influence of c-TFAM on the mtDNA copy number, heterozygous DT40 cells showed no sign of an opposite influence on the level of c-TFAM (see Fig. 3C) as was observed in cells that have been artificially depleted of mtDNA (50,51). In that case, cells exhibited low levels of TFAM concomitant with artificial DNA depletion without any decrease of tfam mRNA, which might be due to rapid degradation of the protein when not bound to DNA.
In tfam ϩ /tfam Ϫ DT40 cells, levels of mtDNA transcripts (ATPase 6/8 mRNA, ND6 (NADH dehydrogenase subunit VI) mRNA, and 12 S/16 S rRNA) tended to be reduced, whereas levels of ATPase ␣ and ␥ mRNA encoded in the nuclear genome FIG. 6. Dosage-dependent increase of mtDNA copy number in DT40 cells by exogenous c-TFAM. Wild type c-tfam cDNA under the control of ␤-actin promoter and with neo r as the selection marked was introduced into the tfam ϩ /tfam Ϫ DT40 cells, and the drug-resistant clones were selected. A, Southern blot analysis on DNA extracted from three independent clones (tfam) together with a control clone (c) with 32 P-labeled mtDNA fragment (top panel) or 32 P-labeled control probe (lower panel) was carried out as in Fig. 4. B, Western blot analysis on c-TFAM in the proteins extracted from individual clones was carried out as in Fig. 3C. In this experiment, the exogenous and the endogenous c-TFAM had the same size and could not be distinguished. C, normalized mtDNA signal of an individual clone quantified as in Fig. 4C is plotted against c-TFAM signal quantified as described under "Materials and Methods." A regression line with regression coefficient (slope) and cutting point of the vertical axis of 1.01 and 0.02, respectively, is drawn to connect the plots by the least squares method. The correlation coefficient () of the line was 0.98. D, Northern blot analysis on RNA extracted from individual clones was carried out as in Fig. 5. remained unchanged (Fig. 5). Reproducible reduction of mtDNA transcripts by 40 -60% together with unchanged levels of the transcripts encoded in the nuclear genome was observed in independently repeated Northern blotting (not shown). This suggested an apparent absence of nuclear-mitochondrial communication for the adjustment of the level of respiratory chain components, which are of dual origin. Thus, our results are in contrast with the insignificant changes in transcript levels observed in the mitochondria of heart, kidney, and muscle of heterozygous knock-out mice of m-tfam (49).
In experiments not shown, wild type DT40 cells were cultured in the presence of ethidium bromide (50 ng/ml) using a medium supplemented with 50 g/ml uridine and 110 g/ml sodium pyruvate. Retardation of cell growth was observed 48 h after the addition of ethidium bromide, and the cells died eventually. Southern blotting of mtDNA showed a reduction of 60% after 48 h, whereas Northern blotting of the same cells showed no change and only a 30% reduction of c-tfam and ATPase 6/8 mRNAs, respectively. These results suggest that maintenance of mtDNA copy number is essential for the survival of DT40 cells.
Dosage-dependent Increase of mtDNA Copy Number by Exogenous c-TFAM-We found that the reduced mtDNA level in tfam ϩ /tfam Ϫ DT40 cells could be restored in a dosage-dependent manner by expression of exogenous mt c-TFAM (Fig. 6A). For this, a plasmid having cDNA coding for mt c-TFAM and neo r in tandem both under the control of ␤-actin promoters was transfected to DT40 cells of which either tfam allele was targeted by MPA, and the G418 resistant colonies were selected. The level of mt c-TFAM detected by Western blotting of the cell FIG. 7. Functional domains of c-TFAM for the maintenance of mtDNA copy number. A series of cDNA constructs encoding c-TFAM deleted at various domains was prepared and inserted downstream of ␤-actin promoter. These constructs were TFAM, the wild type cDNA; del HMG1, tfam cDNA lacking sequence encoding the first HMG box; del link, lacking sequence for the linker between HMG boxes; del HMG2, lacking sequence for the second HMG box; c-myc, tfam cDNA tagged with c-myc sequence at 3Ј-end; del A, lacking the C-terminal tail downstream of 253A; del G, lacking downstream of 243G; del R, lacking downstream of 227R; del M, lacking downstream of 217M; del tail, lacking the tail sequence entirely. The c-tfam cDNA constructs were introduced into the tfam ϩ /tfam Ϫ DT40 cells together with neo r , and the drug-resistant clones were selected. Assays detailed in Fig. 8 for del M cDNA construct were carried out using three independent colonies of tfam ϩ /tfam Ϫ DT40 cells stably expressing individual mutant at different level, and regression coefficient (slope) and correlation efficient () obtained by drawing regression lines are listed at the right margin. In the case of del HMG1 construct, the slope of the regression line gave minus value since the mtDNA/control values for three independent clones were equal to or slightly smaller than 1. We, thus, denoted its regression coefficient as ϳ0. B, Southern blot analysis on DNA extracted from the clones introduced with none, the deletion c-TFAM mutants at HMG box 1, linker, HMG box 2, or c-TFAM tagged with c-Myc using 32 P-labeled mtDNA fragment (top panel) or 32 P-labeled control probe (second panel) was carried out as in Fig. 4. Relative mtDNA signals of individual clones (third panel) quantified and normalized by the internal control in Fig. 4C are plotted. Western blot analysis on c-TFAM and its deletion mutants in the proteins extracted from individual clones (bottom panel) was carried out as in Fig. 3C. Migration position of the endogenous c-TFAM is indicated in the right margin. C, results for the deletion series at the C-terminal tail are summarized. lysate was variable, probably depending on the context of the insertion site of the exogenous c-tfam in chromosome (Fig. 6B). Interestingly, the copy number of mtDNA was linearly proportional to the amount of mt c-TFAM (endogenous plus exogenous), as demonstrated in Fig. 6C for three such colonies. Restoration of mtDNA copy number was more than 4-fold, which is beyond the level in the tfam ϩ /tfam ϩ cells, suggesting that stoichiometric binding of c-TFAM to mtDNA can be the background mechanism of increased copy number. mtDNA transcripts increased marginally by the expression of exogenous c-TFAM and remained within the levels of 1.2-2fold of the tfam ϩ /tfam Ϫ cells (Fig. 6D). The increase was too small to be correlated with the reinforced levels of c-TFAM and/or mtDNA (reaching up to 4-fold). This suggests that dosage of mtDNA as the template and amount of transcription factor are not the primary factors, but some post-transcriptional regulation (for example, stability of mt mRNAs) is more important for the maintenance of the transcript levels.
For the molecular dissection of functional domains in c-TFAM, a series of deletion mutants (Fig. 7A) was stably expressed in the tfam ϩ /tfam Ϫ DT40 cells, and the level of mtDNA was quantitated by Southern blotting. In the case of the mutant deleted at the first HMG box, considerable amounts of the protein remained with the size of precursor before localization into mitochondria, but the mature protein was also observed in the amount comparable with the endogenous c-TFAM (the bottom panel of Fig. 7B). Nonetheless, the reduced level of mtDNA in the tfam ϩ /tfam Ϫ was not restored by this mutant. Interestingly, deletion at the second HMG box impaired only marginally (see below) the ability of c-TFAM to increase the copy number of mtDNA (Fig. 7B). Compared with close similarities of the second HMG box to those of other species, the first HMG box of c-TFAM has a lower similarity. Almost complete loss of the activity by the deletion of the first HMG box implies that the mtDNA-maintaining activity is unique to TFAMs. More careful assay described below suggested that the deletion mutant at linker region retains almost full activity of mtDNA maintenance (Fig. 7, A and B). Surprised at the considerable activity of the c-TFAM to which the c-Myc tag was added at the C terminus (Fig. 7B), we also tested a deletion series of the C-terminal tail region (Fig. 7C) and found that most part of the C-terminal tail region is dispensable for maintaining mtDNA activity.
As was the case for exogenously expressed wild type c-TFAM (Fig. 6), mtDNA increased proportionally to the amount of expressed mutant protein, as shown in Fig. 8 for three colonies of the tfam ϩ /tfam Ϫ cells expressing the del M mutant at different levels. This indicated that evaluation of functional domain using only one colony for each deletion mutant as in Fig.  7, B and C, is not quantitative and that the assay needs to be normalized by the amount of mutant protein expressed in each colony. When the ratio of mtDNA/control (mtDNA level in each colony divided by that in the control cell) was plotted against exogenous TFAM/endogenous TFAM for the del M mutant (Fig.  8C), a regression line cutting the vertical axis at 1.14 (very close to 1) with a slope (regression coefficient) of 0.99 could be drawn by the least square method. This implies that the value of the regression coefficient gives relative activity of the mutants having that of endogenous c-TFAM to be 1, and the value of correlation coefficient () gives the statistic reliability of this assay.
We repeated this assay for all deletion mutants of c-TFAM and listed the values of regression coefficients and correlation coefficients at the right margin of Fig. 7A. It is now clear that deletion at the linker region does not impair the activity. Activity of the mutant deleted at the whole tail region was 51% of the wild type but was restored by the addition of amino acid residues 214 -217 (EAKM) as observed in the del M mutant. The addition of the residues 218 -227 (VELGREDLIR) was suppressive as in the del R mutant, but it was recovered to the wild type level by further addition of the tail sequence. It is worth noting that the addition of c-Myc tag to the tail sequence impaired the activity (the c-Myc mutant).
Even if the mtDNA copy number was considerably increased by the expression of some deletion mutants, the level of mtDNA transcripts remained within 1.2-1.8-fold of the tfam ϩ /tfam Ϫ cells (not shown), as it was for the expression of wild type c-TFAM (Fig. 6D). We thus demonstrated that the structural requisite of c-TFAM for its mtDNA maintenance activity is different from that of h-TFAM established for the transcriptional activation of mtDNA in vitro (21).
Conclusions-Vertebrate mtDNAs have variation mainly in the control region, which is the sequence between the genes encoding tRNA Pro and tRNA Phe and contains the H-strand replication origin and the binding domains of TFAMs for transcriptional activation of H-and L-strands (52). Xenopus and chicken mt genomes are similar in the distance of transcription start sites of L-and H-strands. They are separated by ϳ150 bp in human (26), ϳ100 bp in bovine (53) and mouse (54), ϳ50 bp in Xenopus (25), and 0 bp in chicken (28). c-TFAM cloned in this study has higher similarity to xl-TFAM than to h-and m-FIG. 8. Dosage-dependent increase of mtDNA copy number in DT40 cells by exogenous c-TFAM. Del M cDNA construct was introduced into the tfam ϩ /tfam Ϫ DT40 cells, and the drug-resistant clones were selected. A, Southern blot analysis on DNA extracted from three independent clones (del M) together with a control clone (c) with 32 Plabeled mtDNA fragment (top panel) or 32 P-labeled control probe (lower panel) was carried out as in Fig. 6. B, Western blot analysis on endogenous c-TFAM and del M mutant in the proteins extracted from individual clones was carried out as in Fig. 6. C, normalized mtDNA signal of an individual clone quantified as in Fig. 6 is plotted against del M signal divided by endogenous c-TFAM signal. A regression line with regression coefficient (slope) and cutting point of the vertical axis of 0.99 and 1.14, respectively, is drawn to connect the plots by the least squares method. The correlation coefficient () of the line was 0.99.
TFAMs. It is possible that the structural similarity between xland c-TFAMs is related to the proximity of two start sites in xland c-mtDNAs. The binding of xl-TFAM to a single binding domain between the two start sites of xl-mtDNA was sufficient for activation of bi-directional transcription (23). The mechanisms of c-mtDNA transcription from the L-and H-strand start sites without any intervening base have not been studied, but future comparison of TFAMs for their transcriptional activation of c-mtDNA may reveal the reason for the structural similarity of xl-and c-TFAMs.
Disruption of tfam in DT40 cells resulted in essentially the same consequence as tfam mutant mice (49). The tfam Ϫ /tfam Ϫ cells could not proliferate, and mtDNA copy number was reduced in the tfam ϩ /tfam Ϫ cells. Minor differences were that the reduction of mtDNA was ϳ50% in the tfam ϩ /tfam Ϫ cells compared with 34% in the tfam ϩ /tfam Ϫ mice, and mitochondrial transcripts reduced to 40 -60% in the tfam ϩ /tfam Ϫ cells, whereas there was small change in the tissues of mutant mice. Thus, TFAM appears to have dual functions. Minimum expression of tfam is essential for the proliferation of DT40 cells, whereas massive expression supports the normal mtDNA copy number.
For the regulation of mtDNA copy number, many factors such as DNA polymerase, RNA polymerase and transcription factors, which create the replication primer, single-stranded DNA-binding proteins, and RNA-processing enzymes may be involved. Identification of human diseases associated with drastically reduced levels of mtDNA suggested the significance of nucleotide pool size in the organelle for the maintenance of normal level of mtDNA (55). There has been no report that increases in TFAM would lead to increases in mtDNA levels. Our surprising observation was that expression of c-tfam as an exogenous gene under the background of tfam ϩ /tfam Ϫ resulted in 4 -8-fold increases in mtDNA number. The increase was proportional to the total amount of c-TFAM expressed in the cell. Considering the general ability of vertebrate TFAM and also its yeast homologue (sc-mtTFA) to wrap, bend, and condense DNA by HMG-boxes (18,23,56), the effect of exogenous c-TFAM was probably due to its activity of stabilizing mtDNA. Consistent with this understanding, deletion of the first HMG box from the exogenous c-TFAM canceled the effect. Because sc-mtTFA, which lacks the C-terminal tail, could bind and bend double-stranded DNA (56), it is not surprising that exogenous c-TFAMs deleted at the C-terminal tail differentially could support the increased mtDNA copy number. Considering the essential role of the C-terminal tail in supporting the in vitro transcription of mtDNA (21), however, the consequence might be different if these deletion mutants of c-TFAM could be introduced to the tfam Ϫ /tfam Ϫ cells to rescue the lethal phenotype.