Replication-transcription switch in human mitochondria

See allHide authors and affiliations

Science  30 Jan 2015:
Vol. 347, Issue 6221, pp. 548-551
DOI: 10.1126/science.aaa0986

Switching transcription and replication

Because mitochondrial DNA is circular, the transcription and replication machinery might be expected to collide. A single mitochondrial RNA polymerase (mtRNAP) transcribes the mitochondrial DNA and also generates primers for replication. Agaronyan et al. now show that transcription and replication are kept separate in human mitochondria, with the mitochondrial transcription elongation factor TEFM serving as a key player in the switch. In the absence of TEFM, mtRNAP terminates downstream from the promoter, forming primers to promote replication. In the presence of TEFM, the primers are not formed, and the overall processivity of mtRNAP elongation complexes is enhanced, promoting genome transcription. These mutually exclusive mechanisms allow the processes to proceed independently as needed by the cell.

Science, this issue p. 548


Coordinated replication and expression of the mitochondrial genome is critical for metabolically active cells during various stages of development. However, it is not known whether replication and transcription can occur simultaneously without interfering with each other and whether mitochondrial DNA copy number can be regulated by the transcription machinery. We found that interaction of human transcription elongation factor TEFM with mitochondrial RNA polymerase and nascent transcript prevents the generation of replication primers and increases transcription processivity and thereby serves as a molecular switch between replication and transcription, which appear to be mutually exclusive processes in mitochondria. TEFM may allow mitochondria to increase transcription rates and, as a consequence, respiration and adenosine triphosphate production without the need to replicate mitochondrial DNA, as has been observed during spermatogenesis and the early stages of embryogenesis.

The maternally inherited circular mitochondrial DNA (mtDNA) encodes subunits of complexes of the oxidative phosphorylation chain, as well as transfer RNAs (tRNAs) and ribosomal RNAs (1, 2). Transcription of human mtDNA is directed by two promoters, the LSP (light-strand promoter) and the HSP (heavy-strand promoter) located in opposing mtDNA strands, which results in two almost-genome-sized polycistronic transcripts that undergo extensive processing before polyadenylation and translation (3, 4). Note that transcription terminates prematurely about 120 base pairs (bp) downstream of LSP at a vertebrate-conserved G-rich region, called conserved sequence block II (CSBII), as a result of formation of a hybrid G-quadruplex between nascent RNA and the nontemplate strand of DNA (57). This termination event occurs near the origin of replication of the heavy strand (oriH) (8) and generates a replication primer. According to the asymmetric model (9), replication then proceeds through about two-thirds of the mtDNA, until the oriL sequence in the opposing strand becomes single stranded and forms a hairpin structure. The oriL hairpin is then recognized by mitochondrial RNA polymerase (mtRNAP), which primes replication of the light strand (10). Because replication of mtDNA coincides with transcription in time and space, collisions between transcription and replication machineries are inevitable and, similarly to bacterial and eukaryotic systems, likely have detrimental effects on mtDNA gene expression (11).

We analyzed the effects of a mitochondrial transcription elongation factor, TEFM, recently described by Minczuk and colleagues (12), on transcription of mtDNA. This protein was pulled down from mitochondrial lysates via mtRNAP and was found to stimulate nonspecific transcription on promoterless DNA; however, its effect on promoter-driven transcription had not been determined (12). We found that in the presence of TEFM, mtRNAP efficiently transcribes through CSBII (Fig. 1, A and B). Thus, TEFM acts as a factor that prevents termination at CSBII and synthesis of a primer for mtDNA polymerase. We identified the exact location of the termination point in CSBII (fig. S1). MtRNAP terminates at the end of a U6 sequence (positions 287 to 283 in mtDNA), 16 to 18 nucleotides (nt) downstream of the G-quadruplex (Fig. 1A). At this point, the 9-bp RNA-DNA hybrid in the elongation complex (EC) is extremely weak, as it is composed of only A-U and T-A pairs. This is reminiscent of intrinsic termination signals in prokaryotes—where the formation of an RNA hairpin is thought to disrupt the upstream region of the RNA-DNA hybrid—and is followed by the run of six to eight uridine 5′-monophosphate residues that further destabilizes the complex (5, 13).

Fig. 1 TEFM prevents termination at CSBII.

(A) Schematic illustration of the human mtDNA region downstream of the LSP promoter. The numbers below the scheme correspond to the reference human mtDNA; the transcribed sequence of CSBII is shown at the top. (B) MtRNAP does not terminate at CSBII in the presence of TEFM. Termination efficiency (%) is indicated below the panel. (C) A rare polymorphism in a reference mtDNA results in a decreased efficiency of transcription termination at CSBII.

Human mtDNA is highly polymorphic in the CSBII region; coincidently, the reference mitochondrial genome (Cambridge) contains a rare polymorphism in the G-quadruplex—namely, G5AG7—whereas the majority of mtDNAs from various haplogroups have two additional G residues (G6AG8) (14). We found that the termination efficiency of mtRNAP was substantially lower at G5AG7-CSBII (Fig. 1C), which suggested an effect of G run length on quadruplex formation and underscored the importance of further studies of various polymorphisms in this region.

In considering a putative mechanism of TEFM antitermination activity, we investigated whether it can interact with the nascent transcript and, thus, interfere with the formation of the quadruplex structure. We assembled ECs on a nucleic acid scaffold containing a photoreactive analog of uridine, 4-thio-uridine, 13 nt downstream from the 3′ end of RNA, and walked mtRNAP along the template by incorporation of appropriate substrate nucleoside triphosphates (NTPs) (Fig. 2A). We observed efficient cross-linking between TEFM and RNA when the photoreactive base was 15 to 16 bp away from the 3′ end of RNA. Additionally, using a template DNA containing the LSP promoter and CSBII, we probed interactions of TEFM with RNA having a photoreactive probe 6-thio guanosine-5′-monophosphate (6-thio GMP) 9 to 16 or 18 to 25 nt away from the 3′ end of RNA (fig. S2). We found that TEFM efficiently cross-linked to the G-quadruplex region of RNA, which further confirmed its interactions with the nascent transcript (fig. S2).

Fig. 2 Interactions of TEFM with the components of the elongation complex.

(A) TEFM interacts with the RNA in a scaffold EC. (Top) Cross-linking of labeled RNA to the proteins indicated; (bottom) RNA extension. (B) TEFM interacts with the downstream DNA. (C) Putative region of interaction with TEFM. Structure of human mtRNAP EC (surface representation, PDB ID 4BOC) is shown with the major domains indicated (fingers, pink; palm, light green; N-terminal domain, dark green; thumb, orange; DNA template strand, blue; nontemplate strand, cyan; and RNA, red). The 9-nt-long RNA in the EC is extended by 9 nt to model the trajectory of the nascent RNA.

The catalytic domains of mtRNAP and the single-subunit RNAP from bacteriophage T7 share high sequence and structural homology (15). When we compared the structural organization of human mtRNAP EC (15) and T7 RNAP EC (16), one striking difference became apparent—the former polymerase makes very few contacts with the downstream DNA duplex (fig. S3). In contrast, in T7 RNAP EC, the N terminus of the polymerase makes extensive interactions with the entire downstream duplex. These extensive interactions with the downstream DNA may explain the high processivity of T7 RNAP (16). Because TEFM binds the C terminus of mtRNAP (12), we hypothesized that it may also bind the downstream DNA to compensate for the lack of mtRNAP-DNA interactions. To evaluate this, we incorporated 4-thio-2′-deoxythymidine-5′-monophosphate (4-thio dTMP) into the DNA template in the downstream region and assembled the EC using RNA-DNA scaffolds. We found that TEFM efficiently cross-linked to the +7 and +8 bases in the DNA template strand (Fig. 2B and fig. S4A). The DNA-TEFM cross-linking was species-specific, as no efficient cross-linking was detected when yeast mtRNAP was used (fig. S4B).

The binding site of TEFM on mtRNAP has been previously identified in the C-terminal region, between residues 768 and 1230 (12). These data, taken together with the RNA and DNA cross-linking data, suggest that TEFM likely binds to the palm subdomain of mtRNAP in proximity to the PPR domain and interacts with the emerging RNA transcript (Fig. 2C). It is therefore likely that TEFM antitermination activity is due to direct interference with the formation of the bulky G-quadruplex structure between the palm subdomain and the PPR domain of mtRNAP or due to the allosteric effect of quadruplex formation on complex stability. Similar mechanisms of antitermination have been suggested for lambda phage antiterminators Q and N and a number of bacterial antiterminators that were shown to interact with both RNAP and its nascent transcript that would otherwise form a hairpin structure (17, 18).

Because TEFM interacts with the downstream DNA (which may stabilize the EC), we decided to investigate its effect on mtRNAP processivity. We found that the efficiency of runoff synthesis by mtRNAP was low if the size of the product exceeded 500 nt (Fig. 3, A and B). Indeed, the processivity of mtRNAP was substantially lower than that of a related T7 RNAP (which is known to transcribe large fragments of DNA) and was apparently not sufficient to transcribe a 16,000-bp-long mtDNA. In the presence of TEFM, the efficiency of synthesis of long RNA was significantly increased (Fig. 3, A and B), which suggested that TEFM indeed acts as a processivity factor, as was originally proposed (12).

Fig. 3

TEFM increases processivity and stability of mtRNAP EC. (A and B) TEFM increases the EC processivity. Transcription was performed using LSP templates lacking CSBII sequence to generate 500-nt (A) or 1000-nt (B) runoff products. Efficiency of synthesis of 500 nt of RNA in the presence of TEFM is taken as 100%. (C) TEFM increases the EC stability. Complexes halted at +35 in the presence or absence of TEFM were incubated for the times indicated and chased with CTP. Numbers below indicate efficiency of a halted EC extension (%).

We next probed if TEFM contributes to the stability of the elongating mtRNAP and, as a consequence, to its antitermination activity. To demonstrate this, we halted the EC 35 bp downstream of the promoter start site by omitting one of the substrate NTPs [cytidine 5′-triphosphate (CTP)] (Fig. 3C and fig. S5), in the presence or absence of TEFM. After 10 to 40 min, the complexes were chased to allow synthesis of a runoff product. In the absence of TEFM, accumulation of a 35-oligomer RNA product predominated, with little production of runoff, which indicates that the halted EC is unstable, rapidly dissociates and reinitiates, but is inefficiently chased (Fig. 3C and fig. S5). In contrast, in the presence of TEFM, little, if any, of 35-oligomer RNA product accumulates, but more efficient synthesis of full-length products is observed, which indicates that, under these conditions, the EC is stable and may subsequently be extended.

Our in vitro data demonstrate that TEFM increases the overall processivity of ECs. In the absence of TEFM, mtRNAP is unable to efficiently produce long transcripts and, instead, terminates at CSBII, which primes replication of the heavy strand of mtDNA (Fig. 4). Consistent with this, a knockdown of TEFM in human cells results in a decrease of promoter-distal RNA transcripts (12). We therefore propose that replication and transcription are mutually exclusive processes in human mitochondria that allow the corresponding machineries to avoid the detrimental consequences of head-on collisions (Fig. 4). As such, TEFM serves as a molecular switch that would allow the organelles to either replicate the mtDNA and regulate its copy number or to elevate its transcription rates. This switch can be important for a number of developmental processes, such as embryogenesis and spermatogenesis, during which an active transcription of mtDNA is observed without replication (1925). Although TEFM may not be the only factor affecting mtDNA copy number (26), the proposed mechanism provides a solution to these important biological processes. Further in vivo analysis will be required to address the question as to how this switch is regulated in mitochondria of actively metabolizing cells.

Fig. 4 Model for a replication-transcription switch in mitochondria.

In the absence of TEFM, mtRNAP initiates transcription at the LSP promoter but terminates at CSBII and produces a 120-nt replication primer at the replication origin oriH. The primer is used by replisome for replication of the heavy strand of mtDNA, during which a separate initiation event involving mtRNAP generates a replication primer at oriL. Because of low processivity, transcription from the HSP does not produce a full-size transcript (bottom left). When TEFM is bound (bottom right), transcription from HSP produces genome-length transcripts. At the same time, LSP-driven transcription is not terminated at CSBII and proceeds to the site of factor-dependent termination (MTERF).

Supplementary Materials

Materials and Methods

Figs. S1 to S5


References and Notes

  1. Acknowledgments: We thank K. Schwinghammer for help in cloning and isolation of TEFM and W. McAllister and M. Gottesman for the critical discussion and reading of the manuscript. This work was supported in part by NIH R01GM104231.

Stay Connected to Science

Navigate This Article