Transcription termination at the mouse mitochondrial H-strand promoter distal site requires an A/T rich sequence motif and sequence specific DNA binding proteins


  • Vijayasarathy Camasamudram,

    1. Laboratories of Biochemistry, Department of Animal Biology, School of Veterinary Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Ji-Kang Fang,

    1. Laboratories of Biochemistry, Department of Animal Biology, School of Veterinary Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
    Search for more papers by this author
  • Narayan G. Avadhani

    1. Laboratories of Biochemistry, Department of Animal Biology, School of Veterinary Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA
    Search for more papers by this author

V. Camasamudram, Department of Biochemistry, Faculty of Medicine and Health Sciences, United Arab Emirates University, PO Box: 17666, Al-Ain, UAE. Fax: + 971 3 7672033, Tel.: + 971 3 7039502, E-mail:


Termination of mitochondrial (mt) H-strand transcription in mammalian cells occurs at two distinct sites on the genome. The first site of termination, referred to as mt-TERM occurs beyond the 16 S rRNA gene. However, the second and final site of termination beyond the tRNAThr gene remains unclear. In this study we have characterized the site of termination of the polycistronic distal gene transcript beyond the D-loop region, immediately upstream of the tRNAPhe gene. This region, termed D-TERM, maps to nucleotides 16274–16295 of the mouse genome and includes a conserved A/T rich sequence motif AATAAA as a part of the terminator. Gel-shift analysis showed that the 22 bp D-TERM DNA forms two major complexes with mouse liver mt extract in a sequence-specific manner. Protein purification by DNA-affinity chromatography yielded two major proteins of 45 kDa and 70 kDa. Finally, the D-TERM DNA can mediate transcription termination in a unidirectional manner in a HeLa mt transcription system, only in the presence of purified mouse liver mt D-TERM DNA binding proteins. We have therefore characterized a novel mt transcription termination system, similar in some properties to that of sea urchin, as well as the nuclear RNA Pol I and Pol II transcription termination systems.




promoter distal terminator sequence


H strand promoter


L strand promoter


mitochondrial transcription termination factor


promoter proximal terminator sequence

The mitochondrial (mt) genome in mammalian cells is a double-stranded circular DNA, which encodes two rRNAs, 22 tRNAs, D-loop primer RNAs for DNA replication and 13 polypeptides that are components of the mt electron transport-coupled oxidative phosphorylation system [1,2]. Both the H- (heavy) and L- (light) strands of the mt DNA in vertebrate cells are transcribed symmetrically, and nearly completely [3,4] as polycystronic precursor RNAs starting from strand-specific promoters, HSP1 and LSP, respectively [5–10]. Until recently, the H-strand of the D-loop region, between tRNAPro and tRNAPhe was thought to be a noncoding region of the genome. The D loop is created by the displacement of one of the parental strands by 0.5–1 kb nascent DNA strand needed for the replication of the L strand. The 0.8–1 kb long D loop is a ubiquitous feature of the vertebrate mt genomes [4,11]. The D-loop of the vertebrate mt DNA also houses HSP and LSP organized in opposite orientations, but within about 100 nucleotides of each other.

The genes encoded by the H-strand, believed to be the leading strand, can be classified into two categories: the promoter-proximal region encoding the tRNAPhe, 12 S rRNA and the 16 S rRNA genes [1] and the promoter-distal region which encodes the majority of mRNAs, tRNAs, and the 0.8 kb D-loop region RNA of unknown function [2]. This region spans a 13.6-kb sequence downstream of the tRNALeu gene. A majority of the mt RNA species is processed from larger polycistronic precursors [12]. The promoter-proximal and -distal regions of the H-strand are expressed at distinctly different rates: the promoter-proximal region is transcribed at 40- to 80-fold higher rate than the promoter-distal region [13,14]. A combination of biochemical and mutational analysis coupled with the analysis of mt DNA from human patients with mitochondrial diseases led to the identification of a tridecamer DNA sequence that supports partial transcription termination at the end of the 16 S rRNA gene. The putative tridecamer terminator sequence has been mapped to the 5′ end of the tRNALeu gene, which occurs immediately downstream of the 16 S rRNA gene [15,16]. A 36-kDa protein, which binds to the tridecamer sequence motif, has been purified and characterized by cDNA cloning [17]. This protein, termed the mt transcription termination factor (mTERF) binds to the promoter proximal terminator sequence (mt-TERM), and promotes transcription termination, under in vitro conditions, albeit on a partial basis. mTERF, a leucine zipper protein, appears to bind to the DNA as a monomer, possibly through a novel coiled-coil structure created by interactions between intramolecular zipper domains and terminates transcription by a DNA bending mechanism. Additionally, this terminator sequence exhibits bi-directional activity as described by Shang and Clayton [18], thus invoking its possible role in L-strand transcription termination as well. However, it is not clear as to how small populations of H-strand transcripts escape termination at the mt-TERM and continue through the distal sites of the genome.

Studies from two laboratories showed that the D-loop regions from the rat and mouse mtDNAs encode a stable 0.8-kb poly(A) containing RNA of as yet unknown function [19,20]. The 5′ end of this RNA maps to nucleotide 15 417 of the mouse genome, which marks the start of tRNAPro on the L-strand. The 3′ end maps to nucleotide 16295 of the genome, which is immediately upstream of the tRNAPhe gene on the H-strand [20]. Our results also showed that in the mouse mt system, the 3′ end of this stable RNA is preceded by a conserved A/T-rich sequence motif, AAUAAA [20]. This canonical nuclear polyadenylation signal is believed to have a role in the 3′ end formation of nuclear RNA polymerase II transcripts and also yeast mt mRNAs [21,22]. Similar A/T-rich sequences have been shown to occur at or close to the 3′ end of D-loop coded RNAs from the human and rat systems [1,23]. Based on this, we proposed that the conserved AAUAAA motif, along with its flanking sequences, function as the transcription termination site (D-TERM) for the promoter distal H-strand genes. In the present study we provide evidence that this A/T-rich D-TERM motif with its flanking sequences (16274-5′-ATTACGCAATAAACATTAACAA-3′-16295) binds to mt-specific proteins, different from the previously characterized 36-kDa mTERF [17], and terminates transcription in an in vitro reconstituted system in a unidirectional manner.

Materials and methods

Preparation of mt extract capable of transcription initiation

Mt extract from HeLa cells rich in RNA polymerase activity was prepared as described by Shadel and Clayton [24]. HeLa cells were grown in suspension culture in Joklik's medium supplemented with 10% new born calf serum (both from Sigma Chemical Co). The cells (about 1 g wet weight) in logarithmic phase were harvested and disrupted by homogenization in a buffer containing 134 mm NaCl, 5 mm KCl, 1 mm Na2HPO4, and 2.5 mm Tris/HCl pH 7.5. Mitochondria were isolated from the homogenate by differential centrifugation, and further purified by sucrose density banding. Mt particles banding at the interphase of 1.2–1.6 m sucrose was recovered and used for preparing the mt extract. The RNA polymerase activity from the mt lysate was enriched by successive chromatography on heparin agarose and DNA–Sephacel columns [24]. The polymerase activity was monitored by a filter-binding assay, which measures the incorporation of [α32P]UTP into nascent RNA chains programmed on denatured calf thymus DNA templates [25].

Preparation of mt extract for DNA binding

Mt protein extracts for gel shift analysis were prepared from freshly isolated mouse liver mitochondria and subjected to heparin agarose chromatography as described earlier [20]. Briefly, mitochondria were suspended in buffer A (20 mm Hepes pH 7.9, 50 mm KCl, 10 mm MgCl2, 1 mm EDTA, 1 mm dithiothreitol), lysed by adding 0.6 m KCl and the soluble fraction was separated by centrifugation at 105 000 g for 60 min at 4 °C. The supernatant fraction was dialysed against buffer B (20 mm Hepes, 50 mm KCl, 10 mm MgCl2, 1 mm EDTA, 15% glycerol, 2 mm dithiothreitol, 1 mm phenylmethanesulfonyl fluoride and 1 µg·mL−1 each of leupeptin, pepstatin and antipain) for 3–4 h, at 4 °C and used for binding to heparin–agarose resin. About 1 mL heparin–agarose, pre-equilibrated with buffer B, was added to the extract (100 mg protein) and mixed with gentle agitation for 1 h at 4 °C. The mixture was poured into a 1-mL column and the flow-through fraction was collected. The column was subsequently washed with 15 mL buffer B, followed by a step gradient of 0.1–1.0 m KCl in buffer B. The fractions eluted between 0.5 and 1.0 m KCl were pooled and dialysed against buffer B. DNA–protein binding was assayed by gel mobility shift analysis.

Partial purification of mt transcription termination factor by DNA affinity chromatography

The putative termination factor was purified by affinity chromatography [26]. The 22-bp synthetic double-stranded DNA − termed D-TERM DNA − contains the promoter-distal termination sequence of the mouse mt genome immediately upstream of tRNAPhe (16274-5′-ATTACGCAATAAACATTAACAA-3′-16295). About 1.5 mg 5′ biotinylated synthetic double-stranded D-TERM DNA of 22 bp was bound to avidin-agarose resin (1 mL swollen resin, Sigma) in a buffer containing 10 mm Tris/HCl pH 7.4, 50 mm NaCl, 1 mm dithiothreitol, 1 mm EDTA and 5% glycerol. The affinity DNA matrix (1 mL) was mixed with 50 mg heparin–agarose purified mouse liver mt protein extract by gentle mixing in loading buffer, containing 10 mm Tris/HCl pH 7.4, 50 mm NaCl, 1 mm dithiothreitol, 1 mm EDTA, 5% glycerol, 1 mm phenylmethanesulfonyl fluoride, 1 µg·mL−1 each of leupeptin, pepstatin and antipain and 25 µg poly dI-dC. The binding mixture was incubated for 60 min at 4 °C on a rotating wheel. The contents were then poured into a 0.6 × 5 cm column and the unbound material was collected as flow-through. The column was washed twice with 10 mL loading buffer, and sequentially with a step gradient (4 mL each) containing 0.2, 0.5 and 1.0 m KCl in loading buffer. Protein eluted in each step was tested for binding to the putative termination sequence motif (D-TERM) by gel mobility shift assays.

Gel mobility shift assays

The 22 bp synthetic double-stranded D-TERM DNA was 5′ end labelLed using [γ-32P]ATP and polynucleotide kinase. Protein–DNA binding reactions were carried out in 20 µL volume containing 10 mm Tris/HCl pH 7.4, 50 mm NaCl, 1 mm dithiothreitol, 1 mm EDTA, 5% glycerol, 100 ng dI:dC, 0.1–0.2 ng 32P-labelled gel purified double-stranded DNA probe (30 000 c.p.m.) and 2–5 µg mt protein extract or 1–2 µg affinity purified protein fraction [20]. Synthetic oligonucleotides with nucleotide replacements targeted to various positions of D-TERM were used as mutant probes to assess the specificity of protein–DNA binding. Unlabelled or mutant oligonucleotides were preincubated for 10 min on ice with added proteins under normal assay conditions before adding the probe. Binding was carried out for 25 min at room temperature and the DNA–protein complexes were resolved by electrophoresis on 4% acrylamide gels in Tris/Acetate/EDTA buffer at 4 °C [20].

DNaseI footprint analysis

The putative transcription termination region DNA (D-TERM DNA) cloned in the BamH1 site of Bluescript vector (Stratagene) was linearized with either Xba1 or III and 3′ end-labelled with Klenow enzyme in the presence of α-32P dCTP. After a second digestion with either Xba1 or HindIII the resulting fragments were gel purified and electroeluted. Gel purified 3′ end-labelled DNA probe (190 000 c.p.m. Xba1 end-labelled H strand fragment or HindIII end-labelled L strand fragment), was incubated with 25 µg mt protein extract or 25 µg albumin using the binding reaction conditions as described for the gel mobility shift assay. The protein-bound DNA complexes and free DNA probe were resolved by electrophoresis on 4% acrylamide gels in Tris/acetate/EDTA at 4 °C and were recovered from the gel following autoradiography and electroelution. Protein–DNA complex (complex II) and the free DNA probe, recovered from the gel (20 000 c.p.m.) were subjected to DNAse 1 (Boehringer Mannheim) treatment at a concentration ranging from 0.025 to 0.100 U per 50 µL reaction, in the presence of 0.5 mm CaCl2, 1 mm MgCl2 for 2 min at 25 °C as described by Henninghausen and Lubon [27]. Samples were phenol extracted, concentrated by ethanol precipitation and resolved on 8% sequencing gels as described before.

Methylation interference analysis

The 3′ end-labelled D-TERM DNA probe prepared as described above was partially methylated using dimethylsulphate essentially as described by Maxam and Gilbert [28]. Binding reactions contained 500 000 c.p.m. methylated DNA and 30 µg mt extract. The resolution of the protein bound (complex II) and free DNA probes by electrophoresis on 4% acrylamide gels under gel shift conditions, treatment of gel recovered DNA with piperidine and analysis of cleaved DNA strands on a sequencing gel, were carried out as described before [26,27].

UV induced protein-DNA cross-linking

A photoreactive 32P-labelled D-TERM DNA probe was prepared from a template that encompasses the mouse mt genome sequences 16274–16295 upstream of tRNAPhe gene. The 22-bp template was annealed to a 8-base primer oligo 5′-TTGTTAAT-3′ and filled using Klenow fragment in the presence of 10 µCi [α-32P]dATP/dCTP, 200 µm dGTP, 100 µm TTP and 100 µm Brd UTP as described [26,29]. Protein–DNA binding reactions containing photoreactive probe were carried out in a 500-µL eppenderof tube as described in gel mobility shift assays. The reaction mixtures were then placed on a bed of ice and irradiated with UV light (wavelength 256 nm) for 10 min at a distance of 5 cm. After the addition of SDS sample buffer and heat denaturation at 95 °C for 5 min, the reaction mixtures were subjected to electrophoresis on a SDS/12% polyacrylamide gel and the DNA–protein complexes were visualized by autoradiography. For competition experiments, unlabelled competitor oligonucleotides or mutants were added at 20–100-fold molar excess and incubated for 10 min on ice prior to the addition of the photoreactive probe. To identify the proteins in complex II, gel shift reactions containing the photoreactive probe were run on gels as described in previous sections. The protein–DNA complexes were cross-linked by UV irradiation of the gel for 30 min on a transilluminator (Fotodyne, Hartland, USA), vacuum dried and autoradiographed. The individual complexes were excised, and subjected to SDS/PAGE (12% acrylamide) followed by autoradiography.

South-western blotting

The South-western protocol was modified from Mangalam et al. [30]. The proteins resolved on 12% SDS/PAGE were electroblotted onto nitrocellulose membrane using 40 mm glycine, 50 mm Tris and 20% methanol buffer. The proteins on the blot were denatured with 6 m guanidine/HCl, in binding buffer (10 mm Tris HCl pH 7.5, 50 mm NaCl, 1 mm dithiothreitol, 0.1 mm EDTA, 0.01% NP40, 5% glycerol), for 5 min and renatured by stepwise dilution of guanidine/HCl solution from 6 m to 0.375 m in binding buffer. Washed membranes were equilibrated with 3% BSA in binding buffer for 5–15 min. The excess BSA was rinsed off with binding buffer and the membranes were incubated with γ-32P-labelled D-TERM or mutant DNA probes (8 × 105 c.p.m.·mL−1) in binding buffer containing 2 µg·mL−1 poly(dI/dC) for 12 h at 4 °C on a rotating wheel. The filters were removed, washed with binding buffer and exposed to X-ray film for autoradiography.

Construction of transcription vectors and assay of transcription termination

Because of the known specificity of the human mt RNA polymerase for human mt promoters LSP and HSP [31], we constructed chimeric templates containing the human LSP (pLSP) for transcription initiation and wild-type or mutated mouse D-TERM motifs for assaying transcription termination (Fig. 7A). A 526-bp human mt D-loop DNA fragment (nucleotides 530–534) was amplified by PCR using the pKB741SP (a kind gift from D. A. Clayton) DNA template and cloned in the EcoR1 site of pGEM 7Z plasmid DNA. This region of DNA is selected for the presence of LSP segment as well as the L-strand transcription start site (nucleotide 407). The chimeric templates were constructed by introducing the mouse D-TERM sequences (listed in Fig. 2B) at Mfe1 site of pLSP (at nucleotide 242; 165 bases downstream of L-strand start site) either in the forward pD-TERM (F) or in reverse orientation pD-TERM (R). The mutant versions containing the nucleotide replacements in the D-TERM sequence (see Fig. 2B; pD-TERM Mut1, pD-TERM Mut2 and pD-TERM Mut3) were cloned in a similar way at the Mfe1 site of pLSP in the forward orientation.

Figure 7.

Factor-dependent termination of human LSP initiated transcripts, in an in vitro HeLa mt transcription system. (A) Schematic representation of the DNA templates tested for transcription termination (mt: mitochondria; the numbers in light print represent human and the numbers in bold italics represent the mouse mt genome sequences). The template DNAs consisted of human LSP DNA [pLSP: human mt DNA sequences (nucleotides 5′-530–4-3′) without or with mouse mt putative terminator sequence containing D-TERM DNA (5′−16 274–16 295-3′). In all the chimeric transcription vectors used in termination reactions, mouse D-TERM DNA was inserted at the single Mfe1 site (nt 242), which is 165 bp downstream of L strand start site at nt 407 [5,7]. pD-TERM (F) contained mouse D-TERM sequences cloned in forward orientation (5′-16 274–16 295-3′) while pD-TERM (R) contained the mouse terminator sequences cloned in reverse orientation (5′-16 295–16 274-3′). Although, not shown pD-TERM Mut1, Mut2, and Mut 3 transcription vectors contained mutant forms of D-TERM cloned at the same Mfe1 site in forward orientation (see Fig. 2B). The transcription vectors were linearized with Ssp1, which results in run-off (RO) transcripts of 230 and 265 nucleotides from pLSP and pD-TERM templates, respectively. Also the two possible D-TERM dependent termination transcripts of 189 and 196 nucleotides arising from pD-TERM templates are also shown. (B) Patterns of in vitro run-off transcripts using the control pLSP and pD-TERM (F) DNA that were digested with either Mfe1 or Ssp1 to generate 3′ tailed templates. Run-off transcription was carried out using HeLa cell mt RNA Pol (4 µg) and α32P UTP as described in Materials and methods. Lanes 1 and 2 show the run-off (RO) transcription with pLSP plasmid templates (20 µg·mL−1) linearized with Mfe1 and Ssp1 restriction enzymes, respectively. Lane3, run-off transcript of the pD-TERM (F) linearized with Ssp1. (C) Effects of affinity purified protein fractions (4 µg) on in vitro transcription termination using Ssp1 linearized pD-TERM (F) or pD-TERM (R) or mutant (pD-TERM Mut1, pD-TERM Mut2 and pD-TERM Mut3) templates. Run-off transcription reactions were carried out with HeLa mt RNA Pol fraction alone (lane 2) or that supplemented with the various DNA-affinity column fractions (4 µg protein each lanes 3–9). Lane 3, wash fraction, lanes 4 and 5, 0.5 and 1.0 m KCl eluted fractions, respectively. Lanes 6, same as lane 4, except that 20 µg·mL−1 pD-TERM (R) DNA was used as a template. Lanes 7–9, same as in lane 4, except that pD-TERM (F) DNA carrying various nucleotide replacements (Mut1, Mut2, and Mut3 as shown in Fig. 2B) were used as templates. Although not shown, sequence ladders with primers starting from the transcription initiation nucleotide at +1 were run alongside for determining the exact positions of in vitro terminated RNA.

Figure 2.

Protein binding property of the D-TERM DNA. (A) DNA-protein binding by gel mobility shift analysis was carried out using 32P-labelled D-TERM probe (0.2 ng, 30 000 c.p.m.) encompassing the mouse mt genome sequence 16 274–16 295 as described in Materials and methods. Lane 1 represents control with no added protein. In lanes 2–14, 5 µg mouse liver mt extract was used. A 10–50 molar excess of unlabelled DNAs, D-TERM (lanes 3–5) and mt-TERM (lanes 12–14) as well as mutant versions of D-TERM (Mut) were also used for competition: Mut1 (lanes 6–7), Mut 2 (lanes 8–9), Mut 3 (lanes 10–11). (B) Nucleotide sequences of the L-strand D-TERM DNA and H-strand mt-TERM DNA probes as well as their positions on the mouse mt genome are shown. Mutant probes are the synthetic D-TERM DNA probes with nucleotide replacements in polyadenylation signal sequence (Mut1) or upstream (Mut2) or downstream (Mut3) of the signal sequence. The nucleotide replacements are highlighted in bold letters.

Prior to use in in vitro transcription reactions, the affinity-purified fractions were tested for DNAse and RNAse contamination. The mt DNA template and run-through RNA product were incubated with increasing concentrations of affinity-purified fractions and tested by agarose gel electrophoresis. Treatment of DNA or RNA with 4 and 10 µg of 0.5 m or 1 m KCl eluted fractions did not cause any DNA or RNA degradation thus indicating the absence of nuclease activity in the affinity purified protein fractions.

Run-off transcription assays were carried out in 25 µL reaction mixture containing, 10 mm Tris/HCl pH 8.0, 10 mm MgCl2, 1 mm dithiothreitol, 100 µg·mL−1 BSA, 400 µm ATP, 200 µm GTP, 200 µm CTP, 40 µm UTP, 20 µg·mL−1 linearized transcription template, 4 µg HeLa mt RNA polymerase and 0.25 µm[α-32P]UTP (5 µCi from a stock of 400 Ci·mmol−1 specific activity), at 28 °C for 30 min [24,25]. In some experiments, the reaction mixture was supplemented with 4 µg affinity-purified moue liver mt protein fraction. After a 15-min chase with cold 200 µm UTP, the reactions were terminated by the addition of equal volume of stop buffer (20 mm EDTA, 1% SDS, 0.5 mg·mL−1 Proteinase K) and incubation continued at 37 °C for 30 min. The in vitro transcription products were recovered by phenol extraction followed by ethanol precipitation and resolved on 6% polyacrylamide, 8 m urea gels.


Locations and sequence properties of the proximal and distal transcription terminators

Fig. 1A shows the locations of the promoter proximal transcription terminator (mt-TERM) downstream of the 16 S rRNA gene, and at the 5′ end of the tRNALeu gene on mouse mt genome [15,16]. Studies on S1 nuclease analysis, coupled with the 5′- and 3′-end mapping of D-loop H-strand coded RNA suggested that the distal gene transcription proceeds until the end of the D-loop, overlapping with the HSP [19,20]. In the mouse system, the 3′ end of the D-loop H-strand RNA maps to the CAA sequence ending at nucleotide 16 295 of the genome ([20] see Fig. 1). Since the RNA is polyadenylated, it is uncertain if the termination occurs at the C residue at 16 293 or the A residue at 16 295 of the genome. The 22-bp nucleotide sequence (16 274-5′-ATTACGCAATAAACATTAACAA-3′-16 295) containing the mouse mt H strand transcription start site and also the putative termination site will be referred to as promoter distal transcription termination sequence, D-TERM (see Fig. 1B).

Figure 1.

Location of the putative D-TERM element for H-strand transcription termination on the mouse mt genome. (A) Location of the D-TERM element on the mouse mt genome. The triangles represent the tRNA genes designated by the single-letter amino acid code. The polarity of triangle indicates the direction of the transcription. OL and OH represent the origin of L and H strand replication, respectively. LSS, L strand start site; HSS, H strand start site. The map positions of the rRNA, tRNAs and mRNAs (ND1 to ND6, A6 and A8 and COX I to III) are based on Bibb et al. [2]. The continuous outer arrow indicates the promoter proximal transcript terminating at mt-TERM [15,16], while the discontinuous outer arrow indicates the promoter distal transcript terminating at D-TERM sites. (B) Schematic representation of the mouse mt D-loop region encompassing the conserved sequence boxes I-III (CSB), tRNA genes, L and H strand promoters (LSP and HSP), L and H strand transcription start sites (LSS and HSS). The l-strand sequence and position of the putative promoter distal transcription termination element (D-TERM) is also shown.

Protein binding property of the distal terminator sequence

In a previous study we showed that the 22-bp D-TERM DNA forms two differently migrating complexes with protein from mt extract as tested by gel mobility shift analysis [20]. In the present study, the sequence specificity of D-TERM DNA binding to mt proteins was tested in gel mobility shift assays by competition with wild-type and synthetic mutant D-TERM DNA and mt-TERM DNAs. The sequences of the normal and mutant versions of the promoter distal putative transcription terminator D-TERM, as well as the promoter proximal terminator motif mt-TERM (mouse H strand sequence 2681–2660) are shown in Fig. 2B.

As shown in Fig. 2A, both complexes I and II were effectively competed with 10–20-fold molar excess of unlabelled D-TERM DNA (lane 3–5), while even a 50-fold molar excess of mt-TERM DNA did not compete significantly with either of the complexes (lanes 12–14). Thus, the mt-TERM DNA containing A/T rich sequence seems to have binding specificity different from that of the D-TERM DNA. The D-TERM DNA contains the cononical polyadenylation signal sequence AATAAA. Further, 20 and 50 molar excesses of Mut1 DNA, with nucleotide replacements targeted to the AATAAA sequence region failed to compete for protein binding to D-TERM DNA (Fig. 2A, lanes 6 and 7). Similar concentrations of Mut2 and Mut3 DNAs with nucleotide replacements targeted upstream or downstream of the AATAAA sequence, respectively, competed slightly differently (Fig. 2A, lanes 8–11) from Mut1 with both complexes. These results show that the putative polyadenylation signal AATAAA, and also the sequences upstream and downstream of the canonical polyadenylation signal are important for protein binding to D-TERM DNA.

The binding specificity of D-TERM DNA and also the possible protein–DNA contact sites were further investigated using DNAse1 footprinting and methylation interference analysis. Fig. 3A shows the DNAse1 footprint of complex II (see gel shift Fig. 2A) using the H-strand labelled 135-bp DNA probe. It is seen that relatively low level of protection against DNAase1 is afforded in the absence of added mt protein (lane 7). Reactions with added mt protein (lanes 3–5) however, showed a window of protection spanning the entire D-TERM region, against 0.025–0.1 U DNAse1 per 50 µL reaction. Although not shown, the L-strand labelled probe showed a similar generalized protection of the entire D-TERM sequence region. Additionally, analysis of complex I from the gel shift in Fig. 2A showed a comparable footprint pattern. These results suggest a surprising possibility that complexes II and I may be closely related. The methylation interference analysis of complex II was carried out using the L-strand labelled 135-bp D-TERM DNA probe.Fig. 3B shows that a G (nucleotide 16 279) and two A (nucleotides 16 282 and 16 291) residues, as indicated by arrows, are protected suggesting possible protein–DNA contact sites (lanes 3–4). While one of the protected residues is localized in the polyadenylation signal sequence (A at 16 282) the remaining two residues lie in the flanking sequences. Additionally, two A residues (nucleotides 16 294 and 16 295, respectively), immediately 3′ to the protected A residue of D-TERM DNA become hypermeyhlated. Hypermethylation is thought to be the result of conformational changes in the DNA leading to enhanced sensitivity to dimethyl sulfoxide. These results together suggest that the protein complex probably spans the entire D-TERM region with three purine residues in contact with the protein(s).

Figure 3.

Protein binding specificity and protein contact sites of the D-Term DNA. (A) DNAse 1 footprinting using mt protein extract. Binding reactions contained 3′ end-labelled H-strand D-TERM DNA probe (190 000 c.p.m.) and 25 µg BSA or mt extract. Following gel mobility shift assay, the wet gel was autoradiographed and the free and protein bound complex II or complex I were excised, electro eluted and digested with DNAse1 as described in Materials and methods. Maxam and Gilbert sequencing reactions of the D-TERM probe (GA and CT) were used as markers in lanes 1 and 2. A vertical bar shows the protected sequence: lanes 3–6, reactions containing protein–DNA complexes (complex II, 20 000 c.p.m.) were treated with 0.025 U, 0.050, 0.075 and 0.1 U DNAse1 per 50 µL of reaction, respectively. The free probe treated with 0.075 U of DNAse1 per 50 µL reaction is shown in lane 7. (B) Methylation interference analysis of the protein-bound D-TERM DNA. The 3′ end-labelled L-strand D-TERM DNA fragment was partially methylated and subjected to gel mobility shift analysis using mt extract as described in Materials and methods. F and B indicate piperidine cleavage products of DNA recovered from free and protein-bound DNA (complex II), respectively. The arrows on the right and asterisks on the left hand sequence bar denote the protected nucleotides. The Maxam and Gilbert sequence reactions were loaded in lanes marked as GA and TC.

Nature of mt proteins binding to the D-TERM DNA sequence

UV-induced DNA cross-linking was carried out to investigate the size and complexity of mt proteins binding to the D-TERM DNA. 32P-labelled photo-sensitive DNA probe was bound to mouse liver mt protein extract subjected to UV irradiation in solution and the cross-linked products were analysed by SDS/PAGE following denaturation at 95 °C for 5 min. As expected, bound complexes without UV irradiation did not yield any radioactive bands on the gel (see Fig. 4, lane 1). UV cross-linked protein complexes with mouse liver mt extract yielded three distinct bands of about 77, 75 and 55 kDa (lane 2). Assuming a molecular mass of ≈ 7 kDa for single-stranded DNA probe, the apparent size of the bound proteins correspond to ≈ 70, ≈ 68, and ≈ 48 kDa, respectively. Results also show that protein cross-linking with labelled DNA probe was effectively competed by a 10-fold molar excess of unlabelled D-TERM DNA (lane 3), while a 20-fold molar excess of mt-TERM DNA did not affect the level of protein cross-liking (lane 4). These results further indicate the sequence specificity of protein cross-linking.

Figure 4.

Nature of protein binding to D-TERM DNA by UV cross-linking. Brd UTP-substituted and 32P end-labelled D-TERM DNA was incubated with mouse liver mt extract under conditions of protein–DNA binding reactions as described in Fig. 2. Following the binding reaction at room temperature for 30 min, the reaction mixture was irradiated at 254 nm of a UV lamp for 10 min at a distance of 5 cm. The cross-linked products were incubated in Laemmli's sample buffer at 95 °C for 5 min and subjected to electrophoresis on a 10% SDS polyacrylamide gel. DNA–protein complexes were visualized by autoradiography. For competition experiments unlabelled competitor DNAs were added at 50-fold molar excess and incubated for 10 min prior to the addition of labelled photoreactive DNA probe. Lanes 1–4: mouse liver mt extract without (lane 1) and with UV exposure (lanes 2–4). Lanes 3 and 4 contained 50-fold molar excess of unlabelled D-TERM(DT) and mt-TERM (MT) DNAs, respectively.

Purification of D-TERM binding proteins

With a view to understand the nature of D-TERM DNA binding proteins and their possible role in transcription termination, we have partially purified the proteins by DNA affinity chromatography. To this end heparin agarose-bound mouse mt protein fraction was subjected to DNA affinity chromatography. The ability of the affinity column-purified protein fractions to bind to D-TERM DNA was tested by gel mobility shift analysis. The results presented in Fig. 5A show that the input protein fraction yielded both complex I and II, as described in Fig. 2A. The flow-through fraction yielded both of the complexes, though at reduced intensities (lane 3). Further, the two wash fractions (lanes 4 and 5) and also the fraction eluted with 0.2 m KCl (lane 6) did not form complexes with the DNA probe. The 0.5 m and 1 m KCl eluates, on the other hand, formed complexes with D-TERM DNA (Fig. 5A, lanes 7 and 8, respectively). The latter fraction (1 m KCl eluted), however, formed negligible complex I suggesting possible loss of proteins or a change in the protein composition.

Figure 5.

Purification of D-TERM binding protein by DNA affinity chromatography. The preparation of D-TERM DNA affinity column and chromatography using heparin–agarose purified mouse liver mt extract were as described in Materials and methods. After washing the column with 10 vols loading buffer, the DNA–Resin-bound proteins were eluted sequentially with a step gradient of 4 column vols buffers, each containing 0.2, 0.5, and 1 m KCl as indicated. The fractions were tested for the presence of D-TERM DNA binding proteins by gel mobility shift analysis. (A) Lanes 1–8, gel mobility shift patterns of the D-TERM DNA binding to proteins from various column fractions. Reactions were run using 4 µL of input protein heparin–agarose fraction (HA fraction) and 8–9 µL each of flow-through, wash-1 (W1), wash-2 (W2) and 0.2 m KCl eluted fractions. In lane 7, 0.5 µL of 0.5 m KCl eluted fraction and in lane 8 1 µL of 1 m KCl eluted fractions were used for binding. Lanes 9–11, gel mobility shift assays were carried out using 32P labelled mt-TERM DNA (0.2 ng, 30 000 c.p.m.) and 5 µL each of HA and 0.5 m KCl eluted fractions. (B) SDS/PAGE of various column fractions indicated in (A). Proteins were resolved by electrophoresis through SDS/12% polyacrylamide gels and visualized by silver staining. Lane 1, crude mouse liver mt extract (30 µg); lane 2, input heparin–agarose column purified extract (HA Fr, 20 µg); lane 3, 0.5 m KCl eluted (4 µg); and lane 4, 1.0 m KCl eluted (2 µg) fractions.

The D-TERM DNA binding activity was calculated based on gel shift analysis using excess probe followed by gel quantification. The combined intensity of complexes I and II extrapolated for 1 mg of HA fraction was considered as 1. Quantitative results of protein recovery at different steps of purification and relative DNA binding properties of purified proteins show that the flow through fraction and the wash fraction-1, which together represent < 90% of the input protein, show reduced DNA binding efficiency of 0.3. The wash fraction-2 and 0.2 m KCl eluted fractions which represent ≈ 5–6% of the input protein also show very low DNA binding activity in the range of 0.05–0.1. The 0.5 m KCl eluted fraction, which represents < 1% of input protein showed ≈ 20-fold higher DNA binding activity as compared with control input protein. The 1 m KCl eluted fraction representing > 0.5% of the input protein showed DNA binding activity of ≈ 4.7-fold that of control input protein.

The binding specificity of the affinity purified termination factor(s) was tested by gel mobility shift using radiolabelled mt-TERM DNA probe. As shown in Fig. 5A (lane 11) the mtTERM DNA bound to the partially purified factor (0.5 m KCl elute) with ≈ 20-fold lower efficiency as compared with the D-TERM DNA probe. Additionally, the input mouse liver mt extract also did not form the characteristic complexes I and II, with the mt-TERM DNA probe (lane 10). These results further support the view that the D-TERM DNA has a distinct protein binding property and suggest that protein factors binding to the D-TERM DNA are distinctly different from the 36-kDa mt-TERM protein factor mTERF [17].

The SDS/PAGE patterns of affinity purified proteins is presented in Fig. 5B. It is seen that the heparin–agarose binding step selectively enriched proteins in the range of 29 kDa to > 95 kDa with prominent bands of 70 kDa and 45 kDa. The 0.5 m KCl fraction eluted from affinity column contained predominantly the 70-kDa and 45-kDa components, while the 1 m KCl eluted fraction contained only the 45-kDa species. Further, the 45-kDa species in both the 0.5 m and 1.0 m KCl eluted fractions resolve as doublets. Currently it remains unknown if these doublets represent post-translationally modified versions of the same protein. The difference in the electrophoretic patterns of the 0.5 m and 1.0 m KCl eluted fractions (Fig. 5B) is consistent with the observed differences in D-TERM DNA binding properties of these purified fractions (Fig. 5A).

The DNA binding ability of the purified 70-kDa and 45-kDa proteins was tested by South-western blotting, in which the heparin–agarose bound fraction (20 µg) and the proteins eluted from the affinity column (1–10 µg) were probed with 32P-labelled wild-type and mutant D-TERM DNAs. As shown in Figs 6 D-TERM DNA specifically bound to two protein components of 70 kDa and 45 kDa from heparin–agarose purified protein fraction (HA Fr). The DNA probe also bound to two similarly migrating proteins from the affinity purified 0.5 m KCl eluted fraction. However, the 1 m KCl eluted fraction yielded a prominent band at 45 kDa. Because the SDS gel profile in Fig. 5B shows a 4- to 5-fold higher abundance of the 45-kDa species in the 0.5 m KCl eluted fraction, these results suggest that the 45-kDa protein is a weak DNA binding protein while the 70-kDa species is a high affinity DNA binding protein. Under identical experimental conditions, the Mut1 D-TERM DNA probe with nucleotide replacements targeted to the AATAAA sequence region did not bind to these protein fractions thus indicating sequence specificity. These results collectively show that the 70- and 45-kDa proteins purified by DNA affinity chromatography bind to D-TERM DNA in a sequence-specific manner, but with different affinities.

Figure 6.

DNA binding properties of affinity purified proteins by South-western analysis. HA fraction (25 µg protein), 0.5 m KCl eluted fraction (2 µg protein) or 1 m KCl eluted fraction (15 µg) were resolved, on SDS/12% acrylamide gels, the proteins were transferred to nitrocellulose membranes and probed with 32P labelled D-TERM or Mut1 DNA probes (8 × 105 c.p.m.·mL−1 each), as described in Materials and methods. Figures represent autoradiograms of blots.

Although not shown, UV-mediated DNA–protein cross-linking with gel eluted complex II, showed a 68-kDa protein in addition to a major 70- and a minor 48-kDa component, a pattern similar to that obtained with mt extracts (see Fig. 4). It is likely, that the ≈-48 kDa species identified by UV cross-linking is probably an overestimate of 45-kDa protein. The closely migrating 70- and 68-kDa cross-linked products may represent the same protein bound to the H and L-strands of the probe. Some of the discrepancies between the South-western and UV cross-link analyses as well as the failure to obtain the sequence of the poly(vinylidine difluoride) (PVDF) membrane-bound protein clearly suggest the need for much higher protein purification levels.

Factor-dependent transcription termination by D-TERM DNA under in vitro conditions

We used the well-established HeLa mt lysate system for in vitro transcription initiation and termination assays because our attempts to develop an active in vitro system from the mouse liver/heart mt extracts were unsuccessful. To test the ability of the putative termination signal D-TERM DNA to terminate transcription, we constructed chimeric DNA templates containing human LSP and mouse D-TERM sequences. As shown in Fig. 7A, we placed mouse D-TERM DNA sequence downstream of the human LSP (pLSP) at an Mfe1 site in forward orientation [pD-TERM (F)]. In addition, the D-TERM DNA sequence was also placed in the reverse orientation [pD-TERM (R)] to see if the termination is bi-directional. Plasmid DNAs linearized with Ssp1 were used as templates in run-off transcription reactions using HeLa mt RNA polymerase.

Fig. 7B shows the read-through transcription with pLSP plasmid templates linearized with Mfe1 and Ssp1, respectively. As expected, pLSP template DNAs yielded run-off transcripts of 165 and 230 nt in length, respectively (lanes 1 and 2). Introduction of the putative terminator DNA in pD-TERM (F) resulted in a longer read-through transcript of 265 nt with Ssp1 digested DNA, consistent with the 35-nt D-TERM added to the template (lane 3). However, no detectable transcription termination was observed at the site of inserted D-TERM sequence. We therefore decided to test the effects of affinity purified mouse mt protein fractions that were devoid of any contaminating DNAse or RNAse activity on transcription termination with Ssp1 linearized pD-TERM DNA templates.

As shown in Fig. 7C, HeLa polymerase fraction alone or the reaction mixture supplemented with the ‘wash fraction’ yielded transcripts terminating at the Ssp1 site with no significant termination at the D-TERM site (lanes 2 and 3). Addition of 0.5 m KCl fraction (4 µg·reaction−1), however, yielded a major termination downstream of AAUAAA signal at the end of CAA*, A being the terminal nucleotide 16 295 of the mouse mt genome (189 nt transcript; Fig. 7A and C, lane 4). An additional major termination was also observed at the downstream vector site corresponding to a CAA sequence motif (196 nt transcript). The significance of the termination at the latter downstream site remains unclear, although it corresponds to a CAA sequence motif similar to the upstream D-TERM site. Transcription termination at the second CAA site, indicates the importance of this sequence motif in addition to AAUAAA sequence, in the termination process. Quantitation of the gel by radiometric imaging showed that nearly 60% of the RNA chains were terminated at the two sites marked ‘premature temination sites’ in Fig. 7C. Finally, the protein fraction eluted with 1.0 m KCl also caused transcription termination at the same two sites, though at vastly reduced rates (lane 5). The D-TERM sequence cloned in reverse orientation [D-TERM (R)] was unable to induce transcription termination in the presence of added affinity-purified factor suggesting that the termination is unidirectional or polar (lane 6). Results also show that pD-TERM-Mut1, Mut2 and Mut3 sequences with negligible to marginal ability to compete for protein binding with D-TERM DNA yielded vastly reduced transcription termination (lanes 7–9). These results provide direct evidence that D-TERM DNA functions as a transcription terminator under in vitro conditions and that its activity is dependent on the presence of a novel mt protein factor. Termination was concentration dependent and was also inhibited by more than 60% by 50 ng D-TERM DNA added to the transcription reaction mixture.


It is well established that ≤ 60-fold higher steady-state levels of rRNAs as compared to distal gene coded mRNAs in the vertebrate mitochondria, result from a partial termination of transcription at the end of the 16 S rRNA gene [13]. Several mechanisms, including transcription attenuation at the putative hairpin structure of precursor RNA (at the 5′ end of the tRNALeu gene) have been proposed for the partial termination at the end of the rRNA genes [32,33]. Use of in vitro transcription systems, coupled with extensive mutagenesis at the putative termination region, led to the identification of a tridecamer sequence termed mt-TERM. A 36-kDa protein termed mTERF has been to shown to bind to the mt-TERM DNA in a sequence-specific manner and promote partial termination of transcription [15–17]. Despite extensive characterization of the mt-TERM mediated partial termination past the rRNA genes, details as to the specific site and the mechanism of termination of the H-strand distal gene transcription remain to be elucidated.

Insight into the possible site of distal H-strand transcription termination came from studies on the characterization of novel H-strand coded polyadenylated RNAs, mapping to the D-loop regions of the rat and mouse mtDNAs [19,20]. Occurrence of such RNAs of relatively high abundance in the mouse and rat tissues was surprising since the D-loop region of the vertebrate mtDNA was believed to be the only nontranscribed region of the genome. We identified a 0.8-kb poly(A)-containing RNA, whose 3′ end maps to the CAA sequence at nucleotide 16 295 of the mouse mt genome [20]. The 3′ terminus of the 0.8-kb RNA is preceded by a putative polyadenylation signal AAUAAA [21]. Analysis of the 3′ end polyadenylation sites of the H-strand encoded RNAs by cDNA sequencing, has revealed that the putative polyadenylation signal, AAUAAA, is conserved in human, mouse and rat mt genomes [20]. A dodecamer sequence AAUAA(U/C)AUUCUU was also shown to be the site of pre-mRNA processing and 3′ end formation in yeast mt mRNAs [34]. The occurrence of polyadenylated and oligoadenylated rRNAs in animal cell mitochondria is well documented and known to be coupled to mt RNA processing in the vertebrates [35]. The role of polyadenylation in the nuclear RNA Pol II transcription termination is a well established entity. In view of these facts, we postulated that the conserved sequence motif, AAUAAA might be the site of termination of distal region H-strand transcription, and that the termination may be linked to polyadenylation [20].

In the present study, we demonstrate the ability of the 22-bp putative D-TERM sequence (nucleotide 16 274–16 295 of the mouse genome), containing the polyadenylation signal and the flanking sequences to terminate transcription in an in vitro mt transcription system. In a human mt transcription system driven by HeLa mt RNA polymerase, transcription termination was dependent on the addition of DNA-affinity purified mouse liver mt protein fraction. Further confirmation of the need for affinity purified protein(s) comes from experiments showing that nucleotide replacements targeted to the D-TERM sequence, which affect protein binding also yield reduced factor-dependent transcription termination (Fig. 7C). Finally, D-TERM sequence cloned in the reverse orientation failed to induce significant termination suggesting the specificity of the in vitro system.

The D-TERM-mediated transcription termination exhibits some similar, yet a number of distinct features as compared to the mt-TERM dependent transcription termination. Although both of these DNA motifs contain A/T rich sequences, they show distinct protein binding properties. The D-TERM DNA probe formed two complexes (complex I and II), both of which were not competed by even 50-fold molar excess of mt-TERM DNA (Fig. 2A). Termination by both D-TERM and mt-TERM sequences appear to be linked to polyadenylation. However, we do not have experimental evidence to indicate whether the polyadenylation is a termination-linked or a processing event. While D-TERM mediated transcription termination characterized in the present study is unidirectional or polar, the mt-TERM dependent termination is reported to be bidirectional [18]. In keeping with their different binding specificities, the two terminator sequences seem to bind to distinctly different proteins. The D-TERM DNA binding proteins purified by affinity chromatography contain two major protein components of 70 and 45 kDa, while the mt-TERM DNA binding protein is of 36 kDa [17]. It should also be noted that the 1 m KCl eluted fraction containing predominantly the 45-kDa protein, shows a weak DNA binding and only a marginal termination activity under in vitro conditions, while the 0.5 m KCl eluted fraction, containing both the 45-kDa and 70-kDa, proteins exhibits full activity (Fig. 7C). The precise roles of the two proteins in transcription termination remain to be elucidated. In the case of mt-TERM mediated termination, the DNA affinity purified mTERF factor was fully active in promoting termination in an in vitro system [16]. However, the bacterially expressed, purified 36-kDa factor showed no significant termination activity [17]. These results suggest the possibility that mt-TERM dependent transcription termination requires multiple protein factors including the well-characterized 36-kDa protein. It is possible that the H-strand transcription termination system also requires multiple protein factors.

The DNA binding properties of the proteins were studied by multiple approaches. Initially, use of DNA-affinity chromatography resulted in the purification of two major proteins of 45 and 70 kDa (Fig. 5B). UV-mediated DNA–protein cross-linking with crude extracts as well as with complex II, however, showed a 68-kDa protein in addition to a major 70- and a minor 48-kDa component. We believe that the ≈ 48-kDa species identified by UV cross-linking is probably an overestimate and may be the same as that purified as a 45-kDa protein by affinity binding. Furthermore, the affinity purified 45-kDa species migrated as a doublet on SDS/PAGE, although the South-western analysis showed a single protein band interacting with the DNA probe. The precise nature of these two proteins remains unclear, though they may represent post-translationally modified forms. Based on the relative DNA binding affinities we postulate that the 70-kDa protein is the major DNA binding component and the 45-kDa protein may associate with the DNA bound protein complex through protein–protein interaction.

Most interestingly, another protein that binds to D-loop region of mt DNA with a role in transcription termination has recently been identified and cloned [36]. Studies by Fernandez-Silva [36] have established that the binding of sea urchin mt displacement (D)-loop binding protein (mtDBP) to the noncoding region (D-loop: 133 bp region between tRNAThr and tRNAPro) of sea urchin mt genome leads to transcription termination. Interestingly, the 40-kDa sea urchin mtDBP shows a significant sequence homology with the mammalian mTERF [37] and functions as a bipolar transcription termination factor. They propose a regulatory role for mtDBP even in mt DNA replication [38].

In summary, we describe a novel sequence-specific termination of the mouse mt H-strand distal transcripts, which is reminiscent of the sea urchin mt transcription termination system as well as the nuclear Pol I-dependent rRNA [39] and Pol II-dependent mRNA transcription termination systems [21]. Similar to the recently identified sea urchin mtDBP, the D-TERM binding proteins bind in a sequence specific manner to terminator motif localized in the noncoding region (D-loop) of mouse mt DNA. However, while both sea urchin mtDBP and mTERF function as bipolar transcription termination factors, D-TERM proteins show polarity, similar to the TTF1/Reb-1 mediated termination of rRNA [39]. Whether this is due to a structurally asymmetric protein–DNA complex is not known. Further details of the termination mechanism and the nature of the D-TERM binding proteins are currently under investigation.


We thank D. A. Clayton, Howard Hughes Medical Institute, Chevy Chase, MD, USA, for providing the human LSP DNA used in this study. We also thank M.-A. Robin, G. Biswas and C. Chandran (FMHS, UAE University) for helping with the illustrations and M. Higgins for editorial assistance. This research was supported in part by NIH grant GM49683 and Common wealth and General Assembly of Pennsylvania grant awarded to N. G. Avadhani.