Introduction
The regulation of mammalian oxidative phosphorylation capacity in response to physiological demands and disease states is complex and requires the concerted action of both nuclear and mtDNA‐encoded genes (
Scarpulla, 2008). The mtDNA genome only encodes 13 proteins, but these are essential for the oxidative phosphorylation system (
Larsson et al, 1998). Reduced mtDNA expression is a well‐recognized cause of human mitochondrial disease (
Tuppen et al, 2010) and is heavily implicated in age‐associated diseases and ageing (
Larsson, 2010;
Wallace, 2010). Nuclear genes are necessary for maintenance and expression of mtDNA, for example, by controlling mtDNA copy number (
Ekstrand et al, 2004), transcription initiation (
Falkenberg et al, 2002) and translation (
Metodiev et al, 2009;
Camara et al, 2011).
Control of mtDNA transcription initiation is thought to have a key role in regulation of oxidative phosphorylation capacity. The basal machinery for transcription of mtDNA consists of the nuclear‐encoded mitochondrial RNA polymerase (POLRMT), mitochondrial transcription factor A (TFAM;
Parisi and Clayton, 1991) and mitochondrial transcription factor B2 (TFB2M;
Bogenhagen, 1996;
Falkenberg et al, 2002), which together are sufficient and necessary for
in vitro transcription initiation from mtDNA fragments containing the heavy and light strand promoter (HSP and LSP;
Falkenberg et al, 2002). Mitochondrial transcription generates large polycistronic transcripts, which undergo RNA processing to release 13 mRNAs, 2 rRNAs and 22 tRNAs. In the polycistronic transcripts, mRNAs are often flanked by tRNAs and endonucleolytic processing to release tRNAs will therefore also release mRNAs, according to the so‐called tRNA punctuation model (
Ojala et al, 1981). The enzymatic excision of tRNAs involves two enzymatic activities, that is, RNase P at the 5′ end (
Holzmann et al, 2008) and RNase Z suggested to process the 3′ end (
Takaku et al, 2003;
Dubrovsky et al, 2004). Most mRNAs are subsequently polyadenylated by the mitochondrial polyA polymerase (mtPAP;
Tomecki et al, 2004) and polyadenylation is often necessary to generate the stop codon at the 3′ end of the open reading frame encoded by the mRNA. A number of enzymes are involved in rRNA (
Metodiev et al, 2009;
Camara et al, 2011) and tRNA modification (
Nagaike et al, 2001;
Suzuki et al, 2011). The function of polyadenylation, besides generating stop codons in some transcripts, is not fully understood. Polyadenylation is implicated in regulation of mitochondrial mRNA stability (
Nagaike et al, 2005;
Slomovic and Schuster, 2008;
Wydro et al, 2010) and a mutation in the
mtPAP gene has been reported to cause impaired mitochondrial function and ataxia in humans (
Crosby et al, 2010).
The mechanism whereby mature mRNAs are recognized by the ribosome for subsequent translation initiation is well characterized in prokaryotes. Most prokaryotic mRNAs have an untranslated region (UTR) upstream of the start codon containing a so‐called Shine–Dalgarno (SD) sequence. This SD sequence is complementary to a sequence in the 16S rRNA of the 30S bacterial ribosomal subunit and allows the mRNA start codon to find the correct position at the P site of the ribosome (
Shine and Dalgarno, 1974). In yeast mitochondria, mRNA recognition by the ribosome takes advantage of the affinity between the 5′ UTR of the mRNA and transcript‐specific translational activators. One such example is PET309, a proposed homologue of leucine‐rich pentatricopeptide repeat containing (LRPPRC), which acts as a specific translational activator for the COXI mRNA to promote translation initiation (
Tavares‐Carreon et al, 2008). Mammalian mitochondrial mRNAs do not have 5′ UTRs and an alternate mechanism must therefore be responsible for mRNA recognition by mammalian ribosomes.
The pentatricopeptide repeat (PPR) protein family was first discovered in plants and is characterized by a canonical, often repeated, 35 amino acid motif involved in RNA binding. A surprisingly large number of PPR proteins have been reported in plants, where they are implicated in regulating processing, editing and stability of organelle genome transcripts in chloroplasts and mitochondria (
Schmitz‐Linneweber and Small, 2008;
Zehrmann et al, 2011). Mammals have only seven PPR proteins and while the function of some has been at least partly elucidated (
Holzmann et al, 2008;
Xu et al, 2008;
Davies et al, 2009;
Rackham et al, 2009), the molecular mechanisms remain unclear. One of the mammalian PPR proteins, LRPPRC, was first discovered as being highly expressed in hepatoma cancer cell lines (
Hou et al, 1994). Subsequent papers have associated LRPPRC with a ribonucleoprotein complex responsible for shuttling mature mRNAs from the nucleus to the cytosol (
Mili and Pinol‐Roma, 2003). LRPPRC has also been proposed to be a cofactor of the eukaryotic translation initiation factor 4E, which is involved in control of nuclear gene expression by regulating the export of specific mRNAs from the nucleus to the cytosol (
Topisirovic et al, 2009). In addition, a nuclear role for LRPPRC has been reported as it has been shown to interact with the co‐activator PGC‐1α to regulate the expression of nuclear genes involved in mitochondrial biogenesis (
Cooper et al, 2006). Recessive mutations of
Lrpprc cause the French‐Canadian type of Leigh syndrome (LSFC;
Mootha et al, 2003), a mitochondrial disease which is characterized by infantile onset of severe neurodegeneration in the brain stem and a profound cytochrome c oxidase deficiency in liver and brain (
Merante et al, 1993;
Debray et al, 2011). Studies of the subcellular distribution of LRPPRC have demonstrated that it is mainly present in mitochondria (
Tsuchiya et al, 2002;
Mili and Pinol‐Roma, 2003;
Xu et al, 2004;
Cooper et al, 2006;
Sasarman et al, 2010;
Sterky et al, 2010). We recently reported that transcription of the
Lrpprc gene only seems to produce a single mRNA isoform, which is encoding an LRPPRC protein with a mitochondrial targeting sequence that is cleaved after import to the mitochondrial matrix (
Sterky et al, 2010). Results from studies of cell lines indeed suggest that LRPPRC has an intramitochondrial role in regulation of mtDNA expression (
Gohil et al, 2010;
Sasarman et al, 2010), although the mechanism by which it acts is controversial (
Sondheimer et al, 2010). On the one hand, it has been shown that knockdown of LRPPRC in tissue culture cells causes a general decrease in mitochondrial mRNA levels, impaired translation and a general decrease of the respiratory chain complexes (
Sasarman et al, 2010). On the other hand, fibroblasts from LSFC patients have a respiratory chain deficiency mainly affecting complex IV. Immunoprecipitation experiments and blue native polyacrylamide gel electrophoresis (BN–PAGE) gel analyses have demonstrated that LRPPRC interacts with the stem‐loop interacting RNA binding protein (SLIRP) in an RNA‐independent way (
Sasarman et al, 2010). SLIRP was initially described as a protein binding the nuclear RNA augmenting co‐activation of nuclear receptors, but recent results suggest it is mainly present in mitochondria and has a role in maintaining mitochondrial mRNAs (
Baughman et al, 2009). Recent studies have implicated LRPPRC in apoptosis (
Michaud et al, 2011) and in autophagocytosis (
Xie et al, 2011); however, these effects may be secondary because deficient oxidative phosphorylation is known to increase apoptosis (
Wang et al, 2001;
Kujoth et al, 2005) and has been reported to induce autophagy (
Narendra et al, 2008).
We have characterized the in vivo function of LRPPRC by generating and characterizing conditional knockout mice. We report here that LRPPRC is essential for embryonic survival and that loss of LRPPRC in the heart leads to a drastic reduction in steady‐state levels of all mitochondrial mRNAs, except ND6. LRPPRC forms an RNA‐dependent complex with SLIRP, which is necessary for maintaining a pool of non‐translated mitochondrial mRNAs. Loss of LRPPRC does not only lead to decreased mRNA stability but also causes loss of mRNA polyadenylation and the appearance of a misregulated mitochondrial translation pattern. Thus, LRPPRC has important roles in post‐transcriptional regulation of mtDNA expression and is an essential regulator of oxidative phosphorylation capacity in mammals.
Discussion
We report here a number of new and unexpected insights into the function of LRPPRC in the post‐transcriptional regulation of mitochondrial expression
in vivo. Studies of knockdown cell lines have previously indicated that decrease of LRPPRC levels causes a global decrease in mRNA stability and translation. Surprisingly, we show here that loss of LRPPRC in the mouse
in vivo causes an unexpectedly complex phenotype. LRPPRC is necessary for stability of all mtDNA‐encoded mRNAs except ND6. Without LRPPRC proper coordination of translation is lost and translation pattern becomes aberrant. Some mRNAs are translated at much higher levels than others and in several cases the newly produced translation products are unstable. We also report that LRPPRC is necessary for mRNA polyadenylation and in its absence mRNAs are only oligoadenylated. Studies of cell lines have indicated that LRPPRC forms an RNA‐independent complex with SLIRP. At variance with these results, we show here that the LRPPRC/SLIRP complex is RNA dependent and that it maintains a pool of extra‐ribosomal non‐translated transcripts. Without LRPPRC the extra‐ribosomal pool of mRNAs disappears and mRNAs are found aberrantly bound to free ribosomal subunits, in addition to the normally occurring binding to the assembled ribosome. Taken together, these results define LRPPRC as a key post‐transcriptional regulator of mtDNA expression. The use of knockout mice has the advantage that insights are provided into the physiological function of LRPPRC in a differentiated tissue. Continuously dividing transformed or primary culture cells are mainly glycolytic and are usually grown at much higher oxygen tensions than those present in real tissues, which may explain why previous
in vitro studies have not identified a role for LRPPRC in translation coordination (
Xu et al, 2004;
Sasarman et al, 2010).
Regulation of mRNA stability is important in respiratory chain dysfunction (
Wang et al, 1999;
Wredenberg et al, 2002), but the molecular mechanisms of this process are largely unknown. Our findings demonstrate that LRPPRC is required for the regulation of mRNA levels, but its importance differs between the individual mitochondrial transcripts. Among the most LRPPRC‐dependent transcripts are COXI, COXII and COXIII, which are dramatically reduced upon loss of LRPPRC. Patients with LSFC have a profound COX deficiency, similarly to the
Lrpprc knockout mice, and our findings argue that this is due to the drastic decrease of COXI–III mRNA levels causing reduced synthesis of these subunits. In addition, we report a high turnover of newly synthesized COXI subunits, which may make complex IV especially vulnerable to decreased synthesis of its subunits.
The levels of ND6 mRNA, the only mRNA transcribed from LSP, remain normal in the absence of LRPPRC, whereas the levels of all other mRNAs are decreased. The ND6 mRNA is interesting as it is normally not polyadenylated in heart mitochondria, but instead it contains a long 3′ UTR. Surprisingly, there is thus no need for polyadenylation to regulate stability and translation of the ND6 mRNA. LRPPRC forms a physical complex with SLIRP and mitochondrial mRNAs. The mRNA component is essential for the stability of the complex because RNase A treatment leads to disruption of the LRPPRC–SLIRP interactions. The levels of the LRPPRC–SLIRP complex correlates nicely with total mRNA levels in different mouse models and in human cell lines. The only mRNA not found to be associated with the LRPPRC–SLIRP complex is ND6 and changes in LRPPRC levels do not affect ND6 mRNA stability, again demonstrating the unique nature of this transcript.
There is a dramatic drop in mRNA levels upon loss LRPPRC, but no compensatory increase of mitochondrial transcription. This finding stands in stark contrast to previous reports of other mutants, where impaired translation leads increased
de novo transcription (
Metodiev et al, 2009;
Camara et al, 2011;
Kolanczyk et al, 2011). The lack of a stimulatory effect on transcription could be explained if LRPPRC is itself directly involved in sensing mRNA levels in mammalian mitochondria. Results from studies of mouse knockouts for TFAM, TFB1M and MTERF4, which have decreased, normal and increased steady‐state levels of mtDNA transcripts, respectively, show that the LRPPRC levels follow the mRNA levels. In addition, we show that knockout of
Lrpprc leads to a rapid decrease of mRNA levels and that LRPPRC overexpression in cell lines increases steady‐state levels of mRNAs. Thus, the LRPPRC protein binds and stabilizes mRNAs and the LRPPRC protein may under normal physiological circumstances be saturated with mRNA molecules. A reduction in mRNA levels, for example, due to decreased transcription, may decrease the mRNA saturation of LRPPRC and thereby elicit a response that increases mtDNA transcription. This hypothetical model would explain why loss of LRPPRC, due to knockout of its gene, leads to decreased mRNA levels without a concomitant increase of mtDNA transcription.
Mutations that inactivate the yeast mitochondrial degradosome, responsible for mRNA degradation cause an accumulation of transcripts, which in turn impairs mitochondrial gene expression (
Rogowska et al, 2006). This effect can be suppressed by partial loss‐of‐function mutations in genes that encode for the yeast homologues of POLRMT and TFB2M, thus suggesting that the absolute levels of transcripts must be tightly regulated in yeast mitochondria. If transcript levels need to be tightly controlled also in mammalian mitochondria remains to be established, but based on our findings, LRPPRC could be a component of a system that senses transcript levels and signals to the transcription apparatus in order to maintain a balance between RNA synthesis and degradation.
LRPPRC does not only bind to mitochondrial mRNA species, but the protein also influences polyadenylation as its loss causes a substantial shortening of polyA tails. In mitochondria, polyadenylation is performed by a mitochondrial isoform of the polyA polymerase (mtPAP) (
Tomecki et al, 2004;
Nagaike et al, 2005). There have been reports that polyadenylation is a two‐step process, with oligoadenylation preceding the addition of long polyA tails, but the mechanisms behind this effect and what constitutes the switch between oligoadenylation and polyadenylation remain obscure. Based on our findings, it is possible that LRPPRC plays a role in helping to extend the polyA tail of oligoadenylated mitochondrial transcripts. In some support of this suggestion is the finding that oligoadenylation of the rRNAs is normal. In nuclear mRNA processing, efficient polyadenylation is not only dependent on polyA polymerase (PAP), but also requires a polyA binding protein (PABP), which helps recruit and stabilize PAP to its substrate RNA. In fact, the polyA tail length in human cells is controlled via the stabilization or destabilization of a ternary complex containing PAP and PABP. It is possible that LRPPRC performs related roles in mitochondria, perhaps by stabilizing an interaction between mtPAP and the oligoadenylated transcript. In support of this model is the finding that the non‐polyadenylated ND6 mRNA is the only mRNA not bound by LRPPRC and not affected by loss of LRPPRC. Interestingly, PPR proteins in trypanosomes have recently been shown to affect polyadenylation of mitochondrial mRNA transcripts. In this unicellular protozoa, the PPR proteins KPAF1 and KPAF2 associate with the mtPAP and stimulate mRNA polyadenylation, thereby coordinating stability and translation of mRNA (
Aphasizheva et al, 2011).
The polyA tail is required for translation of nuclear transcripts. In contrast, our data suggest that mitochondrial translation can be very effective even in the absence of mRNA polyadenylation. However, LRPPRC does seem to have an important role in coordinating the translation of different transcripts. In the Lrpprc knockout heart, translation is misregulated with massive translation of some transcripts and no translation of others. Interestingly, our results demonstrate that neither LRPPRC nor SLIRP associates directly with the ribosome. Instead, the two proteins maintain an extra‐ribosomal pool of mRNAs, which may be presented to the ribosome in an orderly fashion. The LRPPRC–SLIRP–mRNA complex is lost in the Lrpprc knockout tissues and the remaining levels of mRNAs are bound to the ribosomal subunits or the assembled ribosome. Thus, a pool of translationally inactive mRNAs is maintained and stabilized by the LRPPRC–SLIRP complex. Controlled release of mRNA from this pool may be necessary for coordinated translation of different respiratory chain components in mammalian mitochondria. In the absence of LRPPRC, the pool is lost and the translation becomes misregulated as mRNAs may enter the translation machinery at random. In addition, the lack of polyadenylation also has different effects on the individual mRNA transcripts, with levels of some mRNAs decreasing much faster than others. A combination of these effects, that is, loss of the non‐translated mRNA pool and loss of polyadenylation, may cause a drastically increased translation of some mRNA molecules, which blocks the translation machinery and thereby prevents translation of other transcripts. This effect may explain why ND6 translation and protein levels are severely decreased in Lrpprc knockout hearts despite the observation that the levels of the ND6 transcript are unaffected by the loss of LRPPRC.
In summary, our findings define a novel role for LRPPRC and shows that this protein is necessary for mRNA stability and polyadenylation. In addition, we report that LRPPRC is necessary for maintaining a pool of non‐translated transcripts and for coordination of mitochondrial translation. These findings point to the existence of an elaborate machinery that is regulating mammalian mtDNA gene expression at the post‐transcriptional level.
Acknowledgements
This study was supported by Swedish Research Council grants (2010‐2766 to NGL and 2009‐4848 to CG), Leducq Foundation to NGL, Lundberg foundation to CG, an ERC Advanced Investigator joint grant to NGL and CG, Deutsche Forschungsgemeinschaft, SFB 829 (NGL), SFB815 (Z1 to UB and ZI 552/3‐1 to VZ) and the Cluster of Excellence ‘Macromolecular Complexes’ EXC 115 (UB) at the Goethe University. We thank the Karolinska Center for Transgenic Technologies (KCTT) for help with production of knockout mice. We thank Ilka Wittig for valuable help and Vahid Edrisi, Karin Siegmund and Andrea Duchene for excellent technical assistance. AW is funded by a FEBS Long‐Term Fellowship.
Author contributions: BR, CMG, UB and NGL designed the experiments; BR, MDM, AW, AB, JBS, CBP, YC, DM, VZ, RW, KH, HEB and PT performed the experiments; BR and NGL wrote the paper.