Mice homozygous for a defect in the PTCD2 (pentatricopeptide repeat domain protein 2) gene were generated in order to study the role of this protein in mitochondrial RNA metabolism. These mice displayed specific but variable reduction of ubiquinol–cytochrome c reductase complex activity in mitochondria of heart, liver and skeletal muscle due to a decrease in the expression of mitochondrial DNA-encoded cytochrome b, the catalytic core of the complex. This reduction in mitochondrial function has a profound effect on the myocardium, with replacement of ventricular cardiomyocytes by fibro-fatty tissue. Northern blotting showed a reduction in the mRNA for the mitochondrial DNA encoded proteins cytochrome b (cytb) and ND5 (NADH dehydrogenase subunit 5) and an elevation in a combined pre-processed ND5-CYTB transcript. This suggests that the PTCD2 protein is involved in processing RNA transcripts involving cytochrome b derived from mitochondrial DNA. This defines the site for PTCD2 action in mammalian mitochondria and suggests a possible role for dysfunction of this protein in the aetiology of heart failure.
- cytochrome b
- pentatricopeptide repeat (PPR) protein
- pentatricopeptide repeat domain protein 2 (PTCD2)
- RNA processing
In 2000, Small and Peeters  described a set of proteins with 35-amino-acid repeat sequences that were dubbed “PPR (pentatricopeptide repeat) cassette proteins”. These repeats, which occurred in variable numbers (average 9.1 per protein), were present in 213 proteins, that included a number of organellar proteins present in mitochondria or chloroplasts. These included the Saccharomyces cerevisiae protein PET309, the Neurospora crassa protein cya5 and the plant protein CRP1p, all of which were known to be involved in the assembly of chloroplast cytochrome bf6 or its counterpart in mitochondria, COX (cytochrome oxidase) [2–5].
The PPR motif is predicted to consist of two α-helices based on an alignment of 1000 putative motifs, with the helix pairs packing together to form a groove [1,6] (Pfam PF013535). Although initially predicted to bind protein, it has now become evident that PPR proteins are RNA-binding proteins with a certain degree of specificity. The LRPPRC (leucine-rich PPR cassette) protein is a distant mammalian homologue of the yeast protein PET309, a protein required for the efficient expression of COX [7,8]. The LRPPRC gene was identified as the cause of LSFC (Leigh Syndrome French Canadian) due to a founder mutation producing a protein with an A354V substitution [7,8]. The consequence of this mutation is a reduction of the import of LRPPRC protein into the mitochondrial matrix and, since this protein is necessary for the stability and translation of the COX1 and COX3 transcripts in the mitochondria, a subsequent failure of assembly of the COX complex .
Other PPR proteins have been shown to have roles in RNA protection and editing, particularly in chloroplasts [9–15]. The mammalian genome has a smaller number of PPR proteins compared with the several hundred identified so far for Arabidopsis thaliana alone. There are six PPR proteins thought to be present in mammalian mitochondria. The largest of them is the 130 kDa LRPPRC protein . Two smaller proteins, the 70 kDa PTCD1 (PPR domain protein 1) and PTCD3 are of unknown function and have homology with factors required for assembly of complex I in N. crassa and for complex V in plant mitochondria respectively. PTCD2 and MRPS27 (mitochondrial ribosomal protein 27) are genes encoding proteins with single PPR motifs which are transcribed in opposite directions from loci within 10 kb on chromosome 5q.13.2. The MRPS27 and PTCD2 proteins possess 17.8% identity, including one stretch of 32 amino acids with 50% identity. Mitochondrial RNA polymerase also has two PPR motifs .
In the present paper we investigate the effect of ‘gene-trap’ modification of the PTCD2 gene in mice homozygous for the insertion of the β-geo construct, and we show that PTCD2 is involved in the expression of cytochrome b, the overall assembly of mitochondrial complex III and the phenotypic development of a lipid cardiomyopathy.
MATERIALS AND METHODS
For mitochondrial localization studies, HeLa human cervical carcinoma cells were maintained in DMEM (Dulbecco's modified Eagle's medium) with glucose (4.5 g/l), glutamine (2.5 mM), penicillin (100 μg/ml) and streptomycin (100 μg/ml) at 5% CO2 at 37 °C.
Tissue RNA was prepared using TRIzol® reagent (Invitrogen Life Technologies, Burlington, ON, Canada). Briefly, 100 mg of mouse tissue was cut into small pieces on ice and homogenized in 4 ml of TRIzol® solution. The homogenate was agitated for 5 min at 4 °C on a constant rocking platform. Chloroform (0.8 ml) was then added and mixed thoroughly by vortexing. The mixture was centrifuged at 12000 rev./min in an SS-34 rotor for 10 min at 4 °C. The supernatant was collected and RNA in the supernatant was precipitated with an equal volume of propan-2-ol.
Reverse transcription and PCR
Reverse transcription of total RNA was carried out using Superscript III reverse transcriptase (Invitrogen Life Technologies, ON, Canada). For cDNA amplification, Vent DNA polymerase (New England BioLabs, ON, Canada) was used. For genomic DNA amplification, the Expanded Long Template PCR System (Roche Diagnostics, Laval, QC, Canada) was used.
PTCD2–GFP (green fluorescent protein) vector construction, cell transfection and image capture
Human PTCD2 cDNA was amplified by PCR and subcloned into the pEGFP-N1 vector (Clontech, Palo Alto, CA, U.S.A.). After DNA sequencing to confirm correct insertion, the construct was purified using a Qiagen Plasmid Mini kit (Qiagen, Mississauga, Burlington, ON, Canada). The purified plasmid (5 μg) was used to transfect cultured human HeLa cells using Lipofectamine™ 2000 (Invitrogen Life Technologies, Burlington, ON, Canada) according to the recommended manufacturer's protocol. After 24 h of transfection, Hoechst 33342 and MitoTracker Red were added to a final concentration of 1 μM and 100 nM in medium respectively and the culture continued for 10 min. The fluorescent signals in living cells were observed under a fluorescence microscope (Richardson Technologies, Bolton, ON, Canada) with immersion objectives.
Search of the gene-trap site of PTCD2 and mouse chimaera generation
Genomic DNA was prepared from mouse PTCD2 gene-trapped ES (embryonic stem) cells (Cell line ID RRF537) obtained from the Mutant Mouse Regional Center (MMRRC) at the University of California, Davis, CA, U.S.A. (http://www.genetrap.org/) . The PTCD2 gene and β-geo insertion site was amplified by PCR using primers for mouse chromosome 13, 5′-TGCCAGTTTAATGCCTCCATCTT-3′ and the β-geo gene-trap vector 5′-CGCTCCACAGTTTCGGGTTTTC-3′. The PCR product was gel-purified and sequenced. The RRF537 ES cells (strain SF129) were injected into mouse C57BL/6 blastocysts. Embryos were transferred into pseudopregnant mice. Resultant coat colour chimaeras were mated to C57BL/6 mice.
Genomic DNA isolation and Southern blot analysis
Mice were killed by cervical dislocation. All procedures were approved by the Animal Care Committee of the Hospital for Sick Children to minimize potential suffering in compliance with the recommendation of the Canadian Council on Animal Care. Whole spleens were surgically removed from homozygous mutant, heterozygote and wild-type mice. Genomic DNA was prepared following the standard classical protocol of SDS protein denaturation, proteinase K digestion, phenol/chloroform extraction and ethanol precipitation. DNA samples (10 μg) were digested with EcoRI at 37 °C overnight. The fragmented DNA was separated by electrophoresis on an 0.8% agarose gel containing 0.5 μg/ml of ethidium bromide in TAE buffer (40 mM Tris/acetate and 1 mM EDTA), examined under UV light and transferred on to Hybond-N nylon membranes (Amersham Biosciences). The membranes were dried for 1 h at 80 °C on a gel drier and then hybridized with [α-32P]dCTP-labelled probe targeted upstream of exon 8, and capable of detecting both wild-type and mutant DNA. Following hybridization, the membranes were exposed to Kodak XAR film at −80 °C.
Biochemical assays of NADH dehydrogenase (complex I), rotenone-sensitive NADH–cytochrome c reductase (complexes I and III), COX (complex IV) and citrate synthase were performed as described . Complex III was measured as ubiquinol–cytochrome c reductase (antimycin sensitive).
Antibodies and Western blot analysis
Anti-human complex I 49 kDa (NDUFS2) subunit and PTCD2 antibodies were prepared by immunizing rabbits with peptides corresponding to the C-terminal 14 amino acids of the protein. Antibodies to cytochrome b were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.) and complex III core I (UQCRC1) protein and the Reiske FeS (UQCRFS1) protein were obtained from Mitosciences (Eugene, OR, U.S.A.). Citrate synthase antibody and antibodies against COX2, COX5A and COX6C were prepared by immunizing rabbits against the purified citrate synthase or bovine heart COX. Protein samples were prepared either by lysis of isolated mitochondria or by diluting a nuclear extract in 1% Nonidet P40 lysis buffer [50 mM Tris/HCl, pH 8.0, 100 mM NaCl, 10 mM NaF, 0.5 mM DTT (dithiothreitol), 0.1 mM EDTA, 1 mM Na3VO4 and a cocktail of protease inhibitors (Sigma–Aldrich)]. Protein (50 μg) from each sample was used and SDS/PAGE separation was carried out as described previously, followed by transfer to PVDF membranes . Immunoblotting was carried out as described previously with the secondary antibody as donkey anti-rabbit, anti-goat or anti-mouse IgG-conjugated HRP (horseradish peroxidase) followed by development with ECL (enhanced chemiluminescence) solution .
Probes and Northern blot analysis
All probes were amplified by PCR and all PCR products were sequenced before labelling. Primers for the PTCD2 cDNA were 5′-CTCCGCCTGCTGCCGCTGCACTCT-3′ and 5′-CACACTATGCTTTCTGGATTCAC-3′. Those for CYTB (cytochrome b gene) were 5′-CAGTAGACAAAGCAACCTTGAC-3′ and 5′-ACTAAGGCTAGGACACCTCCT-3′. Primers for ND5 (NADH dehydrogenase subunit 5 gene) were 5′-TCCTACTGGTCCGATTCCACC-3′ and 5′-AATGCTAGGCGTTTGATTGGG-3′. Primers for ND6 were 5′-CACCCAGCTACTACCATCATT-3′ and 5′-ACGACTGCTATAGCTACTGAG-3′. Primers for UQCRFS1 were 5′-TTCCTGCTTCTGTCCGTTTTTCC-3′ and 5′-ACACAAGGCTTACTCTCCACTC-3′. Primers for 18S RNA were 5′-GACGGAAGGGCACCACCA-3′ and 5′-CGCTGAGCCAGTCAGTGT-3′.
Probes were labelled by PCR with [α-32P]dCTP (Amersham Biosciences) and purified using the QiaQuick PCR Purification Kit (Qiagen, Mississauga, ON, Canada). Mouse heart, skeletal muscle and liver tissues were surgically removed from wild-type and mutant mice respectively. Total RNA was prepared and dissolved in DEPC (diethyl pyrocarbonate)/water. RNA (20 μg of each sample) was heated in loading dye at 65 °C for 15 min and separated on 1% formaldehyde-agarose gel by electrophoresis. After transferring RNA to Hybond-N nylon membranes (Amersham Biosciences), the membranes were hybridized with probes following standard procedures as described previously . After each hybridization experiment, the membranes were stripped in 1% (w/v) SDS solution at 70 °C for 1 h. The strip efficiency was examined by exposure to Kodak XAR film at −80 °C for 48 h. Membranes without visible signals were used for the next round of hybridization.
Native blue gel electrophoresis
Native blue gels were prepared using mitochondria isolated from heart, muscle and liver, using the methods described by Wittig et al.  and Schagger et al. [20,21]. Gels were transferred to PVDF membrane (Hybond P, Amersham Biosciences) and stripped of Serva Blue G by soaking in methanol for 15 min, TBST (25 mM Tris/HCl, 137 mM NaCl, 27 mM KCl and 0.1% Tween 20, pH 7.4) for 5 min and stripping buffer (10% SDS and 10 mM 2-mercaptoethanol in TBST) for 20 min. The membranes were then used for immunoblotting studies.
Light and electron microscopy
Brain, liver, heart, muscle and kidney from both mutant and wild-type mice were harvested immediately after mice were killed and were fixed in either 10% neutral buffered formalin overnight and embedded in paraffin for light microscopy or minced into mm3 pieces and fixed in 2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4, and fixed for an additional 2–4 h. Tissues for electron microscopy were then post-fixed in 2% osmium tetroxide dehydrated in acetone and embedded in Embed 812-araldite. For light microscopy, hearts were frozen in OCT in liquid nitrogen and cryostat sections cut and stained with Oil Red O. The other part of the heart was fixed in formalin, dehydrated and embedded in paraffin. Sections were then cut and stained with Elastin-Trichrome stain. Ultrathin sections were cut from the Embed 812-araldite embedded tissues, mounted on grids and stained in uranyl acetate and lead citrate prior to examination in a JEOL JEM 1230 transmission electron microscope (JEOL U.S.A., Peabody, MA, U.S.A.). Images were then recorded with a CCD (charged-coupled device) camera (AMT, Danvers, MA, U.S.A.).
Tissues from both wild-type and PTCD2 mutant mice were harvested and fixed in 4% (w/v) paraformaldehyde and 0.1 M phosphate buffer (pH 7.4) containing 0.1% glutaraldehyde, for a minimum of 4 h. The tissues were then washed several times with phosphate buffer. They were then stored at 4 °C in PBS containing 20 mM sodium azide until further processing. Prior to cryoultramicrotomy, the cells were infused overnight in 2.3 M sucrose. The tissues were then frozen in liquid nitrogen on aluminum pins and sections were cut on a diamond knife in a cryoultramicrotome (Ultracut R, Leica Canada, Willowdale, ON, Canada). Sections were transferred to formvar nickel-coated grids in a loop of molten sucrose. The grids were then blocked in PBS/BSA (PBS containing 0.5% BSA and 0.15% glycine). After several rinses in PBS/BSA, the sections were incubated in PTCD2 antibody diluted 1:10 in PBS for 1 h. Following several rinses in PBS/BSA, the specimens were then incubated with 10 nm gold goat anti-rabbit IgG particles (Amersham, Oakville, ON, Canada) for an additional hour, rinsed in PBS and distilled water and stabilized in a thin layer of methyl cellulose containing 0.2% uranyl acetate. Controls included incubation in a cocktail of PTCD2 antibody and PTCD2 peptide instead of just the antibody and the omission of the primary antisera. The grids were then examined in a transmission electron microscope (JEOL JEM 1200, JEOL U.S.A.) and images captured using CCD camera (AMT).
PTCD2 is located in mitochondria
The Mitoprot II algorithm (http://ihg.gsf.de/ihg/mitoprot.html) predicts that the human PTCD2 protein has a 93.2% probability of being targeted to the mitochondria, with mouse and zebrafish proteins having 86% and 97% probabilities respectively . In order to investigate the subcellular localization of PTCD2, we made a PTCD2–GFP construct by removing the PTCD2 stop codon and fusing the C-terminus to a GFP cDNA construct in order to track its location in live cells. After transfecting HeLa cells and staining with Hoechst and MitoTracker Red CMXRos fluorescent dyes to indicate locations of nuclei and mitochondria, we found that the PTCD2–GFP protein green fluorescence overlapped substantially with the CMXRos red fluorescence of the mitochondria. This suggests that the sublocalization of PTCD2 is mitochondrial (Figure 1).
Immunogold labelling with PTCD2 antibody in the wild-type mouse heart also showed that PTCD2 protein localizes specifically to the mitochondria (Figures 2A and 2B). There was no PTCD2 protein identified in the nuclei, sarcomeres or cytoplasm. PTCD2-immunogold labelling of PTCD2 mutant mouse heart showed no full-length PTCD2 protein was present (Figure 2C).
Mouse PTCD2 is significantly expressed in heart and liver
Mouse PTCD2 (381 amino acids) has a similar domain structure to the human (388 amino acids) protein, there being one PPR motif of 35 amino acids at residues 158–183. In order to obtain information on tissue-specific expression of PTCD2, we performed Northern blotting using multiple mouse tissue RNAs. The expression level of PTCD2 was robust in heart and liver, with weaker signals in kidney, brain and testis (Figure 3, top panel, and Figure 4). On rehybridization of the same membrane with a probe for CYTB, a similar pattern of distribution was seen (Figure 3, middle panel). The blot was finally hybridized with a probe for ACTB (β-actin) as a control; the upper band (2 kb) represents the cytoskeletal actins, and the lower band (1.6–1.8 kb) represents the muscle (skeletal/cardiac) actins (Figure 3, bottom panel). Densitometry measurements using ImageJ software confirmed that heart and liver PTCD2 transcripts were increased compared with β-actin (heart was increased 2.3-fold compared with muscle actins, liver was increased 1.4-fold compared with cytoskeletal actins) . We conclude that PTCD2 is highly expressed in heart and liver relative to the expression of β-actin. Densitometry of the CYTB band showed a 2-fold increased heart expression compared with muscle actins, and 0.9-fold increased liver expression compared with cytoskeletal actins, a similar pattern of expression as seen with PTCD2.
Disruption of mouse PTCD2 gene with β-geo
The mouse PTCD2 gene has ten exons (Figure 4A, +/+). Although the insertion of the gene-trap vector had been defined by 5′RACE (rapid amplification of cDNA ends) as following exon 8 of PTCD2, we confirmed the insertion site at the genomic DNA level. By designing a series of PCR primers linking sequence in the β-geo construct to the PTCD2 gene and trying different PCR conditions, we successfully amplified a 3 kb genomic fragment (Figure 4B). The gene-trap vector was confirmed to be inserted within intron 8. The insertion thus precedes exons 9 and 10 so that upon transcription, the β-geo construct becomes spliced on to exon 8 (Figure 4D). Exons 9 and 10 become redundant as they follow the stop codon in the β-geo construct, even though we showed these exons still to be present in genomic DNA from homozygous mutants (results not shown). DNA sequencing revealed the exact β-geo insertion site. With this information, we designed genotyping primers and used them to genotype progeny obtained after blastocyst injection and mating of the resultant generation of chimaeras. To confirm the genotypes identified by PCR, we performed a Southern blot assay. EcoRI-digested DNA was probed with the PCR-amplified DNA product covering the fusion site of the trapped vector. We detected a 7.4 kb migrating band in the wild-type (+/+) lane, and the major band in the mutant lane was a 12 kb band that comprised of the PTCD2–β-geo fusion product (−/−) as predicted (Figure 4C). Both bands were found in the heterozygote lane (+/−).
Mutant mice produce a chimaeric PTCD2–β-geo fusion product
By Northern analysis, as expected, the chimaeric RNA of PTCD2–β-geo appeared as a slow migrating band only in the lanes of mutants (Figure 4E). This slower migrating band was easier to visualize in heart than in muscle and liver. Faster migrating bands present in some lanes could be due to degraded PTCD2–β-geo. In contrast, wild-type PTCD2 mRNA appeared as a strong band in lanes of wild-type mice. Comparing the band intensity among tissues, it was darkest in heart, fainter in muscle, and was faintest in liver. Re-hybridizing the same membrane with a probe for 18S RNA revealed a similar density of bands in all lanes.
Complex III of the respiratory chain is specifically affected
Mitochondria were prepared from heart, skeletal muscle, liver and kidney and the activity of the mitochondrial respiratory chain complexes were measured. The activities of NADH–cytochrome c reductase (complex I+III) and ubiquinol–cytochrome c reductase (complex III) were variably reduced in heart, liver and skeletal muscle of the PTCD2 mutant mice, the biggest reduction being seen in heart (Figure 5A). The effect on complex III was greatest in the heart at 19% of wild-type activity in isolation and 13.5% in relation to citrate synthase. COX (complex IV), on the other hand, showed variable increases in activity in heart, muscle and liver. Relative to citrate synthase, the activities in the mutants were not significantly increased except in liver. Further detailed analysis of NADH–CoQ1 (coenzyme Q1) reductase and succinate–DCIP (2,6-dichlorophenol-indophenol) reductase showed no decrease in activity, but succinate–cytochrome c reductase was significantly deficient in heart mitochondria (Figure 5B).
Blue native gel electrophoresis of heart mitochondria isolated from mutant mice showed a marked reduction of complex III in relation to the other complexes (Figure 6A). When immunoblotted for the Rieske FeS protein of complex III (UQCRFS1), the hearts of mutant mice showed a major reduction in titre, whereas the complex I 39 kDa subunit (NDUFA9) showed an unchanged titre. Mitochondria from liver and muscle showed less discernable decreases in titre in keeping with the less-marked reduction in activity of complex III. Immunoblotting for cytochrome b and the complex III core I protein (UQCRC1) showed assembled complex III to be slightly decreased in heart from PTCD2 mutant mice, with less effects in muscle and liver. The amount of complex I appears to be elevated in muscle and liver from PTCD2 mutant mice, which is in keeping with the slight elevation of NADH–CoQ1 reductase. There are no degradation or subassembly products visible that are specific to the mutant mice.
Immunoblotting of isolated mitochondria from heart under denaturing conditions showed that full-length PTCD2 protein was completely absent in all tissues in PTCD2 mutant mice. There are also major decreases of cytochrome b in heart, with smaller decreases in muscle and liver. The Rieske FeS protein (UQCRFS1) and the core I protein (UQCRC1) of complex III show lesser, but still noticeable, decreases in heart (Figure 6B). Subunits COX2, COX5A and COX6C of complex IV showed similar protein levels compared with wild-type, as did the complex I 49 kDa subunit (NDUFS2) and citrate synthase, a matrix enzyme.
Disruption of the PTCD2 gene delays mRNA maturation of CYTB but not COX1
Northern blots probed with a DNA probe for CYTB showed two bands in all lanes of both wild-type and mutant mice (Figure 7B). One band representing mature CYTB migrated rapidly, the slower migrating band corresponds to the combined pre-processed ND5–CYTB transcript [24,25]. In all lanes of wild-type mice, the upper pre-processed transcripts were much fainter than lower bands of mature CYTB. Conversely, the upper pre-processed transcripts were much stronger than lower ones for processed CYTB in all lanes of mutant mice. This difference suggested that the pre-processed ND5–CYTB transcripts were accumulating in the tissues of the PTCD2 mutated mice. The same blot probed for COX1 and UQCRFS1 transcripts showed little difference between wild-type and mutated PTCD2 mice (results not shown). Using a probe for ND5 again highlighted the same slow moving band identified by CYTB with decreased titre of the ND5 transcript in PTCD2 mutant mouse RNA (Figure 7B).
Microscopic analysis of tissues in PTCD2 mutant mice
Sections of heart, muscle, liver and kidney from PTCD2 mutant mice were prepared and examined by light microscopy (heart) and electron microscopy (all tissues) after appropriate staining. Light microscopy of the PTCD2 mutant heart showed partial thinning of the ventricular wall which had been replaced by fibro-fatty tissue (Figure 8B, *). The outer layers of ventricular cardiomyocytes appeared to be infiltrated with macrovesicular fat deposits, having the appearance of adipocytes. Consistent with a cardiac mitochondrial cytopathy was the presence of lipid droplets seen by Oil Red O stain and also within the sarcoplasm among the mitochondria by electron microscopy (Figure 8A). Focal degenerating cardiac myocytes and macrophages were detected in the fibrous tissue (Figure 8B, arrows). Mitochondria throughout the heart were increased in number with disorganized cristae. In approx. 2–3% of the mitochondria, electron opaque trilamellar structures were detected (Figures 8C and 8D). Similar structures have been seen in muscle mitochondria of mice subjected to 24 h of muscle ischaemia .
Light microscopy of skeletal muscle showed no obvious pathology beyond a modest increase in the number of subsarcolemmal mitochondria (results not shown). By electron microscopy the mitochondria were quite pleomorphic with a high percentage of large mitochondria (Figure 9B). Skeletal muscle sections showed an increase in the number of subsarcolemmal mitochondria. Cristae variations included a reduction in numbers and often a total rearrangement. Many mitochondria had voids in the intracristae spaces and the matrix was filled with amorphous material.
In many of the animals, the kidneys were enlarged compared with wild-type. The average size of the wild-type kidneys was 0.16±0.01 g (n=4) and the mutant kidneys was 0.23±0.02 g (n=6). An unpaired t test confirmed that these two means are significantly different (t=2.732, degrees of freedom=8, P=0.0258) with the mutant kidneys 1.4-fold bigger than wild-type. Histological examination of these organs revealed a proliferation of tubules with enlarged lumina throughout the medulla (results not shown). The combined number of tubules in both the cortex and medulla were approx. 1.6-fold the number of those found in the wild-type animals. In the nephron, the distal tubules were enlarged. Electron microscopy of these structures revealed pleomorphic mitochondria (Figures 8G–8I).
By light microscopy the liver, especially in zone 1 and 2, was quite steatotic (results not shown). Under the electron microscope, liver mitochondria were increased in number and many were elongated and serpentine in appearance. Mild alterations in cristae structure were frequently detected (Figure 9D).
In other tissues where the effects on cytochrome b are not as severe, the lack of assembled complex III appeared to cause some mitochondrial proliferation and pleomorphism.
In the present study we examined an unknown functional gene, PTCD2, by establishing the subcellular location of its protein product and the tissue distribution of its mRNA. We established a mouse model for defective PTCD2 derived using gene-trapped ES cells and observed its potential function in mitochondrial RNA maturation and mitochondrial respiratory chain function.
Examination of the gene-trap ES cell strain that we selected revealed the following information. The modified PTCD2–β-geo chimaeric protein produced in this mouse model possessed eight out of ten exons, retaining the single PPR motif ensuring mitochondrial targeting and PPR-dependent function. The lower titre of PTCD2–β-geo chimaeric RNA seen on Northern blotting compared with the wild-type PTCD2 suggests some possible instability of the transcript.
In mice homozygous for the gene-trap PTCD2 construct, the combined activities for complexes I+III and complexes II+III were decreased, particularly in heart, as was complex III itself. In muscle and liver tissue, there were increases in the activities of citrate synthase and COX, which suggests that the titre of oxidative enzymes per mg of mitochondrial protein has increased, as has been observed in cases of pathogenic mitochondrial DNA mutations. Native blue gels confirmed the complex III specific decrease in heart. The lower levels of complex III components, particularly in cardiac tissue, exemplified by cytochrome b, complex III core I protein and the Rieske FeS protein, and the mitochondrial location of PTCD2, suggest that it has a function in the mitochondria affecting the assembly of complex III. Complex III assembly centres around the transcription, translation and mitochondrial membrane insertion of cytochrome b encoded by mitochondrial DNA.
Northern analysis showed a significant accumulation of the pre-processed transcripts of ND5–CYTB with very low levels of mature CYTB and ND5 mRNA in mutated mice. The origin of the increased ND5–CYTB transcript level in the PTCD2 mutant mouse tissues could be due either to a primary failure to process this transcript to mature cytochrome b, or it could represent an accumulation of ND5–CYTB by default, because of improper processing of the H transcript to produce mature cytochrome b RNA . Currently we do not have definitive evidence to say which of these mechanisms is correct.
The processing of the polycistronic RNA into the component tRNA and protein-coding transcripts in mammalian cells occurs with the creation of at least two large combined RNAs, one containing ND5 and CYTB and the other containing COX3, ATP6 and ATP8. These long intermediate transcripts appear to be a function of the lack of tRNA in between two protein-coding transcripts. There are no tRNAs between ATP6, ATP8 and COX3, nor between ND4L and ND4 and there is only a short non-coding sequence between ND5 and CYTB. Interference with the yeast gene Aep3, which encodes a PPR-containing protein with homology with mammalian PTCD3, was shown to reduce expression of COX and ATPase subunits due to loss of stability of a combined ATP6-ATP8 mRNA transcript and COX1 mRNA transcript while increasing the titre of a combined COX1-ATP6-ATP8 mRNA . The mouse mutant PTCD2 shows a similar profile in processing the RNA transcript for ND5 and CYTB. The present study shows that defective PTCD2 leads to accumulation of unprocessed transcripts and a lowering of CYTB and ND5 RNAs. In an earlier study we showed that LRPPRC is essential for the stability of the mRNAs for COX1 and COX3 .
The unusual pattern of tissue-specific response to the absence of normal PTCD2 expression is intriguing. We were able to show heart tissue with excessive lipid within the cardiomyocytes, thinning of the ventricular wall and replacement by fibro-fatty tissues. This appearance is similar to that reported for some cases of arrythmogenic ventricular dysplasia or hystiocytoid cardiomyopathy [28–30]. One of these cases was found to be due to a heteroplasmic mutation in the CYTB gene, which affected the titre of cytochrome b . The mutated PTCD2 gene in the case of these mice mimics this situation allowing only a small amount of functional cytochrome b to be translated for assembly into complex III.
This work was supported by funds from the Canadian Institutes for Health Research [grant number MT 6573], the MitoMarch for Kirkland and Association de l'Acidose Lactique-Saguenay-Lac-Saint-Jean (Québec).
Abbreviations: CCD, charge-coupled device; CoQ1, coenzyme Q1, COX, cytochrome oxidase; CYTB, cytochrome b gene; ES, embryonic stem; GFP, green fluorescent protein; LRPPRC, leucine-rich pentatricopeptide repeat cassette; ND5, NADH dehydrogenase subunit 5; PPR, pentatricopeptide repeat; PTCD, PPR domain protein
- © The Authors Journal compilation © 2008 Biochemical Society