Characterization of the basic transcription machinery of mammalian mitochondrial DNA (mtDNA)1,2 is of fundamental biological interest and may also lead to therapeutic interventions for human diseases associated with mitochondrial dysfunction3,4,5,6. Here we report that mitochondrial transcription factors B1 (TFB1M) and B2 (TFB2M) are necessary for basal transcription of mammalian mitochondrial DNA (mtDNA). Human TFB1M and TFB2M are expressed ubiquitously and can each support promoter-specific mtDNA transcription in a pure recombinant in vitro system containing mitochondrial RNA polymerase (POLRMT)7 and mitochondrial transcription factor A8,9. Both TFB1M and TFB2M interact directly with POLRMT, but TFB2M is at least one order of magnitude more active in promoting transcription than TFB1M. Both factors are highly homologous to bacterial rRNA dimethyltransferases, which suggests that an RNA-modifying enzyme has been recruited during evolution to function as a mitochondrial transcription factor. The presence of two proteins that interact with mammalian POLRMT may allow flexible regulation of mtDNA gene expression in response to the complex physiological demands of mammalian metabolism.
Initiation of transcription of Saccharomyces cerevisiae mtDNA requires only two proteins, the yeast homolog of POLRMT (Rpo41)10 and its specificity factor (Mtf1)11. In contrast, transcription of mammalian mtDNA is also dependent on mitochondrial transcription factor A (TFAM; previously known as mtTFA), a high-mobility group box protein2,12,13. Mammalian mtDNA contains only two promoters, the light-strand (LSP) and the heavy-strand (HSP) promoters, from which transcripts of almost genomic length are produced and then processed to yield the individual mRNAs, tRNAs and rRNAs14,15,16,17. Transcription from LSP also produces an RNA primer that is required for initiating mtDNA replication. Germline disruption of the mouse gene Tfam leads to loss of mtDNA, probably because transcription-dependent priming of mtDNA replication is abolished18. Recombinant TFAM protein and a partially purified fraction containing human POLRMT are sufficient for activating transcription from LSP and HSP in vitro19; however, attempts to reconstitute human mtDNA transcription with recombinant POLRMT and TFAM proteins have been unsuccessful. An activity that co-purifies with a protein of 40 kD and stimulates transcription by interacting with the POLRMT has been described in Xenopus laevis20,21.
We used the profile-based PSI-BLAST method to search the sequence database of the National Center for Biotechnology Information using the putative Mtf1 homolog from Schizosaccharomyces pombe, and identified a predicted human protein that we termed TFB1M (Fig 1a). Notably, TFB1M also showed sequence similarity to a second human protein, termed TFB2M (Fig. 1a). Homologs to both proteins, termed Tfb1m and Tfb2m, were found in the mouse (Fig 1b). TFB1M and Tfb1m (and also TFB2M and Tfb2m to some extent) show strong sequence similarity to bacterial rRNA dimethyltransferases (Fig. 1b).
Computer algorithms predicting subcellular localization suggested that TFB1M and TFB2M contained amino-terminal mitochondrial targeting signals (data not shown). Similarily, mitochondrial import of yeast Mtfl is directed by amino-terminal sequences that are retained after import22. Confocal microscopy studies of human cells transfected with plasmids encoding mouse Tfb1m and Tfb2m fused to green fluorescent protein (GFP) showed that both Tfb1m-GFP and Tfb2m-GFP had a localization pattern indistinguishable from that of ornithine transcarbamylase (OTC-GFP), a protein that is known to localize in the mitochondria23 (Fig. 2a). Northern blot analyses showed that the genes encoding TFB1M and TFB2M were expressed ubiquitously (Fig. 2b), which is consistent with the expression patterns of other nucleus-encoded components of the mitochondrial transcription machinery24,25.
We developed a pure in vitro transcription system that comprised mtDNA fragments containing HSP and LSP, and recombinant TFAM, TFB1M, TFB2M and POLRMT proteins purified after expression in insect cells (Fig. 3a). Expression of POLRMT on its own was not successful, because most of the protein (>95%) was insoluble. However, co-expression of POLRMT with TFB1M or TFB2M resulted in the formation of highly soluble heterodimeric POLRMT–TFB1M or POLRMT–TFB2M complexes.
Both TFB1M and TFB2M could independently support promoter-specific initiation of transcription in the presence of TFAM and POLRMT (Fig. 3b,c). The levels of transcription with POLRMT–TFB1M were at least tenfold lower than with POLRMT–TFB2M. Given their effects on POLRMT solubility, it was possible that TFB1M and TFB2M were required only for purification of POLRMT and were not involved directly in transcription. We therefore examined transcription with different combinations of pure TFAM, POLRMT and TFB2M and found that all three factors are needed for promoter-specific initiation of transcription (Fig. 4a). Similar results were obtained for TFB1M (data not shown). We also monitored the effects of increasing amounts of TFB2M on LSP transcription with constant amounts of TFAM and POLRMT (Fig. 4b). Increasing amounts of TFB2M stimulated transcription, reaching a maximum at a TFB2M to POLRMT molar ratio of about 1:1; higher TFB2M concentrations did not stimulate transcription further.
Most RNA polymerases can initiate transcription on duplex DNA that has a short single-stranded 3′ tail, which makes it possible to monitor effects on transcription elongation independent of promoter recognition. Notably, POLRMT alone, POLRMT–TFB1M and POLRMT–TFB2M were almost equally efficient in stimulating the transcription of tailed templates (Fig. 4c). Therefore, the marked difference in activating transcription from mitochondrial promoters by POLRMT–TFB1M and POLRMT–TFB2M (Fig. 3b) must be caused by differences in promoter-dependent initiation of transcription.
We carried out immunodepletion experiments on transcriptionally active mitochondrial extracts by using an antiserum that reacted with TFB2M (Fig. 5a). Removing TFB2M protein from mitochondrial extracts (Fig. 5b) resulted in a loss of transcriptional activity (Fig. 5c). We used the TFB2M-depleted extract in reconstitution experiments in which we added back recombinant pure proteins. Adding TFB2M (Fig. 5c), TFB1M (data not shown) or POLRMT (data not shown) had no effect, whereas adding the POLRMT–TFB2M complex reconstituted mtDNA transcription. These results suggested that the removal of TFB2M also resulted in loss of the associated POLRMT.
Given the absolute requirement for TFAM in in vitro transcription, we monitored the effect of increasing amounts of TFAM, with constant amounts of POLRMT and TFB2M, on the initiation of transcription from LSP and HSP (Fig. 6a). LSP transcription was stimulated by small amounts of TFAM and remained highly active over a range of TFAM concentrations. In contrast, HSP transcription was activated only at a high concentration of TFAM. A sharp decline in both HSP and LSP transcription occurred when TFAM concentrations were increased further. This specific pattern of TFAM-stimulated POLRMT–TFB2M transcription prompted us to monitor TFAM stimulation of POLRMT–TFB1M transcription (Fig. 6b), because the presence of two functionally overlapping transcription-initiation factors might be explained by different effects at the LSP and HSP promoters. Although the levels of transcription were significantly lower with POLRMT–TFB1M than with POLRMT–TFB2M, the overall patterns of TFAM stimulation of transcription from LSP and HSP were very similar.
Three observations suggest that both TFB1M and TFB2M interact directly with POLRMT to form a heterodimer. First, POLRMT is highly unstable when expressed on its own but markedly more stabile when co-expressed with either TFB1M or TFB2M. Second, purification of His-TFB1M or His-TFB2M on Ni+-columns allows co-purification of roughly equimolar amounts of untagged POLRMT. Last, stimulation of POLRMT activity by TFB2M reaches a maximum at a molar ratio of POLRMT to TFB2M of about 1:1.
Both TFB1M and TFB2M have strong sequence homology with bacterial rRNA methyltransferases. Consistent with our findings, the crystal structure of the related Mtf1 protein of yeast has strong structural homology to the bacterial ErmC′ rRNA methyltransferase26. A human protein denoted mtTFB, which is identical to TFB1M, has been isolated and shown to stimulate transcription in a mitochondrial extract27. Although the molecular mode of transcriptional activation by TFB1M was not described, it was shown that TFB1M binds the methyl group donor S-adenosylmethionine27. So far, we have not detected methyltransferase activity by in vitro methylation assays using recombinant TFB1M or TFB2M and different mitochondrial RNA substrates (data not shown). It is, of course, possible that we have failed to identify the correct conditions for these enzymes or that other unknown factors are needed to stimulate the methyltransferase activity. But the strong sequence similarity of TFB1M and TFB2M with their bacterial counterparts does not necessarily imply an identical enzyme activity. During the course of evolution, TFB1M and TFB2M may have evolved the ability to recognize specific DNA structures to assist POLRMT in promoter recognition. Notably, the crystal structure of the accessory subunit of the mitochondrial DNA polymerase shows that this protein has strong structural similarity to aminoacyl tRNA synthetases28. It is thus possible that both the mitochondrial transcription and replication machineries evolved by recruiting two different enzymes that were originally involved in RNA modification and used the sequence recognition features of these proteins for other purposes.
The identification of TFB1M and TFB2M has allowed us to reconstitute fully the basal mammalian mtDNA transcription machinery in vitro. It should now be feasible to develop a system in which transcription and replication are coupled. This might lead to insights into the regulation of mammalian mtDNA replication and the control of mtDNA copy number. The in vitro transcription system could also be used to identify substances that impair or stimulate mtDNA transcription, which may be beneficial in the development for treatments for some types of human disease.
We constructed plasmids encoding the complete amino acid sequence of mouse Tfb1m and Tfb2m, fused in frame to green fluorescent protein (GFP). We cloned a XhoI–SacII fragment encoding the complete Tfb1m protein sequence (345 residues) into the XhoI and SacII restriction sites of the EGFP-N3 plasmid (Clontech). The resulting plasmid encodes a fusion protein consisting of Tfb1m with an in-frame addition of GFP to its carboxy terminus. We cloned a XhoI–XmaI fragment encoding the complete Tfb2m amino acid sequence (396 residues) into the XhoI and XmaI restriction sites of the plasmid EGFP-N3. The resulting plasmid encodes a fusion protein consisting of Tfb2m with an in-frame addition of GFP to its C terminus. We used an OTC–GFP control plasmid containing the OTC mitochondrial targeting peptide fused in-frame to the N terminus of GFP23. We used a laser scanning confocal microscope to monitor expression of GFP in transfected HeLa cells. We observed excitation and emission of GFP at 488 nm and 400–440 nm, respectively. We added MitoTracker Red CMXRos (Molecular Probes) to living cells at 25 nM for 20min, and observed the cells at an excitation wavelength of 568 nm and an emission wavelength of 580–640 nm.
Expression and purification of recombinant proteins.
We maintained and propagated Spodoptera frugiperda (Sf9) cells in suspension in SFM 900 medium (Gibco-BRL) containing 5% fetal calf serum at 27 °C. We amplified genes encoding TFB1M, TFB2M, POLRMT (without leader peptide) and TFAM (without leader peptide) from cDNAs by PCR, and cloned them into the vector pBacPAK9 (Clontech). We also made plasmid constructs in which a 10×His tag had been introduced at the N terminus (POLRMT) or a 6×His tag had been introduced at the C terminus (TFAM, TFB1M, TFB2M). We prepared Autographa californica nuclear polyhedrosis viruses recombinant for the individual proteins as described in the BacPAK manual (Clontech). For protein expression, we grew Sf9 cells in suspension and collected them 60–72 h after infection. We froze the infected cells in liquid nitrogen and thawed them at 4 °C in lysis buffer containing 25 mM Tris-HCl (pH 8.0), 20 mM 2-mercaptoethanol, and 1× protease inhibitors (for all purifications the 100× stock of protease inhibitors contained 100 mM phenylmethylsulfonyl fluoride, 200 mM pepstatin A, 60 mM leupeptin and 200 mM benzamidine in 100% ethanol). After incubation on ice for 20 min, we transferred the cells to a Dounce homogenizer and lysed them by 20 strokes of a tight-fitting pestle. After adding NaCl to a final concentration of 0.8 M, we swirled the homogenate gently for 40 min at 4 °C. We cleared the extract by centrifugation at 45,000 r.p.m. for 45 min at 4 °C using a Beckman TLA 100.3 rotor.
For purification of His-POLRMT–TFB1M or His-POLRMT–TF2BM complexes, we made extracts from cells infected with 5 plaque-forming units (p.f.u.) of His-tagged POLRMT together with either 5 p.f.u. of TFB1M or 5 p.f.u. of TFB2M. We diluted the extracts 1:1 with buffer A (25mM Tris-HCl (pH 8.0), 10% glycerol, protease inhibitors and 20 mM 2-mercaptoethanol) containing 20 mM imidazole. We added 2 ml of Ni2+-NTA matrix superflow (APBiotech) pre-equilibrated with buffer A, supplemented it with 10 mM imidazole and 0.3 M NaCl, and incubated it for 60 min at 4 °C with gentle rotation. We collected the Ni2+-NTA matrix by centrifugation (1,500g, 10 min, 4 °C), resuspended it in buffer A (40 mM imidazole, 0.3 M NaCl), poured into it a column and washed it with 10 column volumes of the same buffer. We eluted the POLRMT–TFB1M and POLRMT–TFB2M complexes with buffer A (250 mM imidazole, 0.3M NaCl). We dialyzed the peak fractions for 6 h against buffer B (25 mM Tris-HCl pH 8.0, 10% glycerol, 1 mM dithiothreitol (DTT), protease inhibitors, 0.5 mM EDTA). We supplemented the peak fractions with 0.1 M NaCl, froze them in liquid nitrogen and stored them at −80 °C.
To purify the isolated polymerase, we co-expressed His-POLRMT with TFB2M and used the purification method described for His-POLRMT–TFB2M with the following modifications. We did not dilute the cellular extract but supplemented it with 10 mM imidazole. Buffer A for the Ni2+-NTA column contained 1.0 M NaCl, which gave an effective separation of His-POLRMT from TFB2M. We further purified His-POLRMT on a 1-ml HiTrap heparin column (APBiotech) equilibrated in buffer B (0.1 M NaCl). After washing the column with three column volumes of buffer B (0.1 M NaCl), we used a linear gradient (10 ml) of buffer B (0.1–1 M) to elute the His-POLRMT protein at 0.8 M NaCl. The yield of His-POLRMT protein from 400 ml of culture was ∼2 mg. We estimated the purity of the protein to be at least 95% by SDS–PAGE with Coomassie blue staining.
We purified His-TFB2M by the method used for His-POLRMT with the following modifications. We infected the Sf9 cells with 10 p.f.u. of recombinant virus and eluted the TFB2M protein from the Hi-Trap Heparin column at 0.6 M NaCl. We purified His-TFB1M by the method used for His-POLRMT with the following modifications. We infected the Sf9 cells with 10 p.f.u. of recombinant virus. We eluted the HiTrap heparin column with a linear gradient (10 ml) of buffer B (0.5–1.5 M NaCl). His-TFB1M eluted at about 1.3 M NaCl, and the yield of protein was ∼6 mg from 400ml of culture. The purity of the protein was at least 95%. We purified TFAM by the methods used for His-POLRMT with the following modifications. We infected Sf9 cells with 10 p.f.u. of recombinant virus. We ran dialyzed His-TFAM from the Ni2+-NTA step through a Mono-Q column equilibrated with buffer B (0.1 M NaCl), and His-TFAM eluted in the flow through fractions. The yield of His-TFAM from 400 ml of culture was ∼5mg with a purity of at least 95%. We froze all proteins in aliquots in liquid nitrogen and stored them at −80 °C.
In vitro transcription reactions.
We cloned DNA fragments corresponding to bp 1–741 (LSP and HSP), 1–477 (LSP) or 499–741 (HSP) of human mtDNA14 into pUC18. After linearization, we used the plasmid constructs to measure promoter-specific transcription in a run-off assay8,9. Individual reaction mixtures (25 μl) contained 85 fmol of digested template, 10 mM Tris-HCl (pH 8.0), 10 mM MgCl2, 1 mM DTT, 100 μg/ml bovine serum albumin, 400 μM ATP, 150 μM CTP and GTP, 10 μM UTP, 0.2 μM α-32P UTP (3,000 Ci/mmol), 4 U of RNasin (APBiotech), and the indicated concentrations of proteins. We stopped the reactions after 30 min at 32 °C by adding 200 μl of stop buffer (10 mM Tris-HCl (pH 8.0), 0.2 M NaCl, 1 mM EDTA and 0.1 mg/ml glycogen). We treated the samples with 0.5% SDS and 100 μg/ml proteinase K for 45 min at 42 °C, and precipitated them by adding 0.6 ml of ice-cold ethanol. We dissolved the pellets in 10 μl of gel loading buffer (98% formamide, 10 mM EDTA pH 8.0, 0.025% xylene cyanol FF, 0.025% bromophenol blue), heated them at 95 °C for 5 min, and analyzed the samples on a 5% denaturing polyacrylamide gel in 1× TBE buffer. We fixed the gels in 10% acetic acid then dried and exposed them.
We generated tailed templates comprising a stretch of double-stranded DNA (71 bp), preceded by a single-stranded 3′ tail of 15 nt (dC) by annealing two oligonucleotides of 71 and 86 nt. We carried out the transcription reactions as described for promoter-dependent transcription.
Production of antisera and immunodepletion of TFB2M.
We separated pure recombinant TFB2M protein by SDS–PAGE and used it to immunize rabbits. Antisera were collected 6 weeks after the third booster injection (Agrisera AB). We prepared mitochondrial extracts as described29. We coupled antiserum against TFB2M (antiserum 2, Fig. 5a) to protein A–Sepharose beads as described30 and incubated 80 μl of a mitochondrial extract with 40 μl of the beads for 6 h at 4 °C. We carried out the transcription reactions as described for promoter-dependent transcription, using 5μl of crude or immunodepleted extracts instead of recombinant proteins. Reconstitution experiments were done by adding 300 pmol of pure TFB1M, TFB2M, POLRMT or POLRMT–TFB2M proteins to the TFB2M-immunodepleted extract.
We thank R. Wibom and E. Holme for the kind gift of mitochondrial extracts. M.F. is supported by a postdoctoral fellowship from the Karolinska Institutet. N.G.L. is supported by grants from the Swedish Research Council, Funds of Karolinska Institutet, Torsten and Ragnar Söderbergs stiftelse, Human Frontiers Science Program, the Swedish Heart and Lung Foundation and the Swedish Foundation for Strategic Research (Functional Genomics and INGVAR). C.M.G. is supported by grants from the Swedish Cancer Society, the Swedish Research Council, Human Frontiers Science Program, the Swedish Foundation for Strategic Research (INGVAR), the Swedish Society for Medical Research, the Ake Wiberg Foundation, and the Magn. Bergwall Foundation.