Genetic modifications of Mecr reveal a role for mitochondrial 2-enoyl-CoA/ACP reductase in placental development in mice

Abstract

Mitochondrial fatty acid synthesis (mtFAS) is an underappreciated but highly conserved metabolic process, indispensable for mitochondrial respiration. It was recently reported that dysfunction of mtFAS causes childhood onset of dystonia and optic atrophy in humans (MEPAN). To study the role of mtFAS in mammals, we generated three different mouse lines with modifications of the Mecr gene, encoding mitochondrial enoyl-CoA/ACP reductase (Mecr). A knock-out-first type mutation, relying on insertion of a strong transcriptional terminator between the first two exons of Mecr, displayed embryonic lethality over a broad window of time and due to a variety of causes. Complete removal of exon 2 or replacing endogenous Mecr by its functional prokaryotic analogue fabI (Mecr^tm(fabI)) led to more consistent lethality phenotypes and revealed a hypoplastic placenta. Analyses of several mitochondrial parameters indicate that mitochondrial capacity for aerobic metabolism is reduced upon disrupting mtFAS function. Further analysis of the synthetic Mecr^tm(fabI) models disclosed defects in development of placental trophoblasts consistent with disturbed peroxisome proliferator-activated receptor signalling. We conclude that disrupted mtFAS leads to deficiency in mitochondrial respiration, which lies at the root of the observed pantropic effects on embryonic and placental development in these mouse models.

Introduction

Fatty acid synthesis (FAS) in mammalian cells takes place in two compartments; in the cytosol where it follows the canonical integrated pathway (FAS I) and in mitochondria (mtFAS). The mtFAS pathway resembles bacterial type II FAS where each step is carried out by separate enzymes (1). Currently, the only mtFAS product with a confirmed physiological function is octanoic acid, which is used in endogenous lipoic acid (LA) synthesis. LA acts as a cofactor of enzymes of oxidative decarboxylation reactions like pyruvate dehydrogenase (PDH), α-ketoglutarate dehydrogenase (KGD) as well as two dehydrogenases in amino acid metabolism, the glycine cleavage system and branched chain dehydrogenase (2). Studies in Saccharomyces cerevisiae indicate that inactivation of any enzyme in the mtFAS pathway leads to respiratory defects, lack of cytochromes due to disturbed mitochondrial translation and respiratory complex assembly, rudimentary mitochondria and complete loss of protein lipoylation in yeast (3–6). In mouse models, adult mice subjected to tamoxifen–induced inactivation of the Mcat gene, encoding malonyl-CoA transferase in mtFAS, suffered from loss of lipoylation, premature aging symptoms and disturbed energy metabolism (7).

The 2-enoyl-thioester reductase (ETR) catalyzes the NADPH-dependent reduction of enoyl-ACP (acyl carrier protein) to saturated acyl-ACP in the mtFAS pathway. The disruption of the ETR1 gene encoding this reductase in yeast leads to the typical mtFAS defect phenotype described earlier, while overexpression of ETR1 results in dramatically enlarged mitochondria (3). The mitochondrial enoyl-CoA/ACP reductase (Mecr) is the mammalian orthologue of ETR and is highly conserved in higher eukaryotes (8). The overexpression of Mecr in mouse resulted in cardiac myopathy, as well as mitochondrial enlargement in this organ (9) similar to the phenotype observed in yeast. Recently, the first human individuals suffering from a disorder caused by recessive mutations in the MECR gene have been described. These patients display childhood onset dystonia, optic atrophy and basal ganglia signal abnormalities on MRI together with decreased MECR function and mitochondrial protein lipoylation (10). Due to the defects observed in the nervous system of the patients, the term MEPAN (mitochondrial enoyl-CoA reductase protein-associated neurodegeneration) was proposed as name for the disorder.

Etr1, Mecr/MECR and other eukaryotic homologs belong to the family of medium-chain dehydrogenase/reductase proteins whereas most prokaryotic functional analogues, like fabI, are of the short-chain dehydrogenase/reductase type (1). Expression of MECR or mitochondrially targeted bacterial fabI in the yeast etr1Δ mutant alleviated the respiratory defect of these cells, indicating that it is the enoyl reductase activity which is required for growth (3,8). A role of Mecr in transcriptional regulation mediated by peroxisome proliferator-activated receptors (PPAR) has recently been proposed independently by two different groups (11,12) although it is controversial whether this effect is due to a product of mtFAS or direct action of Mecr as a transcriptional co-activator.

In the study presented here, we demonstrate that Mecr inactivation in mice leads to embryonic lethality. We report the results of our studies of the time points and causes of embryonic lethality in three different Mecr models: embryos produced by a leaky ‘knock-out-first’ mouse line, a complete knock-out line generated by removal of a critical exon, and synthetic hypomorphic embryos carrying a replacement of mammalian Mecr with the bacterial fabI analogue. This is the first report on the roles of Mecr or mtFAS in mouse embryonic development.

Results

Generation of ‘knock-out-first’ type (Mecr^tm1a) and Mecr exon 2 deleted (Mecr^tm1d) mice

Mecr^tm1a mouse embryos were obtained from the European Mouse Mutant Archive (EMMA) consortium. In this model, Mecr expression is abrogated by insertion of a marker gene, followed by a strong transcriptional terminator, between exons 1 and 2 of Mecr. This construct also introduced FLP recombinase target (FRT) and loxP sites allowing for subsequent removal of the marker cassette and exon 2. These tools were used to generate the Mecr^tm1dmodel missing exon 2, as described in Materials and Methods (Fig. 1A and B). During this process the second exon of Mecr was deleted by crossing mice carrying a floxed second exon of the Mecr gene with Cre-expressing mice. The lack of this sequence results in removal of a part of Mecr encoding a conserved N-terminal section of the protein. In addition, splicing between the remaining exons 1 and 3 would generate an out-of-frame fusion and yields an inactive protein. The cross breeding of heterozygous Mecr^+/− did not produce any homozygous Mecr^−/− pups from either the Mecr^tm1a or the Mecr^tm1d line.

Figure 1.

 Modification of mouse Mecr gene. (A) A FRT-flanked selection cassette inserted in intron 1 and lox P site on both side of exon 2 (Mecr^tm1a). (B) The Mecr^tm1d—line was generated from Mecr^tm1a mice as described in ‘Materials and Methods’ section. The selection cassette and exon 2 were deleted by breeding with the appropriate recombinase-expressing mice. (C) For the Mecr modification 54 nucleotides from 5′- end of exon 2 was fused with fabI sequence appended with double HA tag (Mecr^tm(fabI)). This sequence is followed by transcription terminator BGH attached to an FRT-flanked selection cassette. The figure is not drawn to scale.

Generation of Mecr^tm(fabI) mice

As mitochondrially targeted bacterial fabI complements respiratory growth of etr1 deficient yeast cells and can be selectively inhibited by the drug triclosan (3,6), we sought to test whether an analogous construct would rescue the Mecr deficiency in mice. Therefore, a sequence encoding the bacterial fabI ETR was introduced into exon 2 of mouse Mecr. The construct was C-terminally tagged with a double HA epitope allowing visualization of fabI expression using western blot analysis. The splicing acceptor sites of exon 2 of Mecr as well as 54 nucleotides of the exon 2 itself were maintained in frame with the fabI coding region. As the Mecr mitochondrial targeting sequence (MTS) is encoded by exon 1, splicing between the latter and the modified exon 2 is predicted to yield in a chimeric mRNA encoding HA-tagged fabI N-terminally extended with the Mecr MTS. This construct is followed by the strong transcriptional terminator, BGH (13) and was designed to exclude the possibility of producing intact Mecr (Fig. 1C). The corresponding cDNA, encoding a fabI variant N-terminally appended with the Mecr MTS and a C-terminal double HA tag, complemented the respiratory growth defect of a yeast etr1Δ strain (Supplementary Material, Fig. S2). The indirect immunofluorescence studies demonstrated that the MecrfabI protein chimera expressed in cultured mammalian HEK 293T cells localized to mitochondria (Supplementary Material, Fig. S3A). Furthermore, when tested in separate experiments, plasmid borne fabI activity was observed in mitochondrial enriched fraction prepared from pcDNA6 MmMecr-fabI HAHA transfected HEK 293T cells. Taken together, these results show that this fabI variant can be expressed and targeted into mammalian mitochondria in active form (Supplementary Material, Fig. S3B).

Mecr and fabI in the Mecr mutant mouse lines

Production of the fabI transcript in E 9.5 embryos of homozygous mutants, heterozygous and wild type embryos from heterozygous Mecr^tm(fabI) females was studied by RT-PCR. We robustly detected the fabI transcript in homozygous knock-in embryos. In comparison, the fabI mRNA levels were about 25% in heterozygous embryo and fabI transcription was essentially absent in wild type embryos (Fig. 2A). In addition, Mecr mRNA expression in all the embryos derived from pregnant heterozygous Mecr^tm(fabI), Mecr^tm1a and Mecr^tm1d were analyzed by RT-PCR and found reduced to background levels in all the homozygous mutants compared with heterozygous or wild type controls (Supplementary Material, Fig. S4).

Figure 2.

Mecr and fabI expression in Mecr mutants. (A) qPCR analysis for fabI expression in total RNA samples from Mecr^tm(fabI) embryos. The amount of mRNA was normalized using actin (Actb) as reference gene. Results are shown as means ± SD, n = 5 per group, independent samples t test, ***P < 0.001, **P < 0.01, *P < 0.05. (B–E) Western blot analysis of different Mecr embryos. HA expression in Mecr^tm(fabI) embryos (B) and Mecr protein expression in embryos from Mecr^tm(fabI) (C), Mecr^tm1a (D) and Mecr^tm1d (E) breedings. Mecr protein expression was not detected in any of the three homozygous mutants.

Embryos of identical genotype were pooled to obtain cell extract material sufficient for immunoblot analysis. The expression of fabI in these extracts was analyzed by western blotting probing for the C terminal HA tag to confirm presence of the chimeric protein in Mecr^tm(fabI) embryos. A strong signal was observed when probing extract of homozygous fabI knock-in embryos with the HA antibody, a less intense signal was observed in extracts from heterozygous embryos. A weak signal corresponding to a protein of similar molecular mass was also observed in samples from homozygous wild type (Fig. 2B). However, the signals from samples containing the fabI construct were consistently and reproducibly much stronger in repeated experiments. Therefore, we suggest that the band observed in wild type embryo extract is a signal from an unspecific interaction with a protein of similar size as fabI-HA. An identical signal could be seen from extracts prepared from wild type embryos developing in Mecr^+/+ mothers (data not shown), indicating that the cross-reacting band is not a result of contamination with maternal tissue expressing fabI-HA. Immunoblotting for Mecr expression demonstrated the presence of Mecr protein in protein extract from wild type and heterozygous embryos, whereas no Mecr protein could be visualized in samples from mutant embryos homozygous for any one of the three types of Mecr modification (Fig. 2C–E).

Embryonic lethality studies of mecr mutants

The breeding of heterozygous Mecr^tm1a, Mecr^tm1d and Mecr^tm(fabI) lines did not produce any homozygous mutant pups, indicating that the mutant embryos died during embryonic development and that fabI complementation did not rescue the embryonic lethality phenotype. Therefore, triclosan inhibition of fabI activity for induction of a phenotype as initially planned was unnecessary and the analysis of the breedings was extended to embryonic lethality studies of all three mutant lines.

When pregnant heterozygous Mecr^tm1a females were dissected between E 7.5 and E 18.5 (Fig. 1D, Supplementary Material, Table S1 and Supplementary Material, Fig. S1), we found that homozygous knock-out embryos were smaller in size, displayed opaque yolk sac membranes and were adherent to membranes by E 8.5, whereas wild type yolk sac membranes were translucent and clear and embryos were easy to separate from membranes. Follow-up dissections indicated that the time of embryonic death was variable, occurring between E 9.5 and E 15.5 as determined by visible heart beats or pulsing blood in umbilical vessel and necrotic signs. Long-surviving homozygous knock-out embryos of the Mecr^tm1a line found alive at E 14.5 were pale and had hemorrhages (Fig. 3A). Due to the variability in embryonic lethality, we suspected that Mecr^tm1a construct could not completely inactivate the Mecr gene. Although the Mecr^tm1a construct inserts a strong transcriptional terminator between exons 1 and 2, the exons themselves as well as the Mecr ORF remain intact.

Figure 3.

Morphology of control and mutant embryos. (A) A wild type (left) and a homozygous Mecr^tm1a mutant (right) which was paler and had multiple hemorrhages at E 14.5. (B, C) Mecr^tm1d embryos control (left) and knock-out (right) embryos at E 9.5 (B) and E 10.5 (C). (D) A small homozygous Mecr^tm(fabI) mutant (right) and regular-sized control littermate (left) at E 10.5 (E) A homozygous Mecr^tm(fabI) embryo at E 10.5 with pale vitelline vessels. The homozygous mutants in (B–D) were smaller and developmentally delayed than control littermates. All the embryos in this figure were alive at the time of dissection as justified by pulsing blood in umbilical vessel (A) or beating heart (B–E). Scale bar 1mm.

The apparent leakiness of the Mecr^tm1a allele was the motivation for the generation of the Mecr^tm1d mouse line (described earlier and in ‘Materials and Methods’ section). In our initial studies of embryonic lethality of Mecr^tm1d embryos, we observed that the window of period of embryonic death was much narrower than for Mecr^tm1a mutants. The homozygous Mecr^tm1d embryos were dead (based on absence of heart beat) and at varying stages of resorption at E 11.5. Dissections focusing on the period E 7.5–E 12.5 (Fig. 1D, Supplementary Material, Table S2 and Supplementary Material, Fig. S1) confirmed the time window of embryonic death ranged from E 9.5 to E 11.5. The homozygous 1d knock-out embryos displayed abnormal morphology similar to 1a knock-out embryos at E 8.5. At E 9.5, some of homozygous mutants were dead. The live homozygous Mecr^tm1d mutant embryos were smaller in size and displayed a developmental delay characterized by retention of dorsal curvature, resembling the morphology of wild type E 8.5 embryos and adherent fetal membranes (Fig. 3B). The live 1d knock-out embryos at E 10.5 were also paler than control littermates (Fig. 3C).

Dissections of pregnant females at specific time point of gestation varying from E 8.5 to E 13.5 were also carried out for Mecr^tm(fabI) mice (Fig. 1D, Supplementary Material, Table S2 and Supplementary Material, Fig. S1). The homozygous knock-in embryos were found to develop normally and undistinguishable from wild type and heterozygous littermates until E 8.5. At E 10.5, all the homozygous knock-in embryos were paler, fragile and smaller than their heterozygous or wild type littermates (Fig. 3D), with pale and empty yolk sac vessels (Fig. 3E) and their heart beats were weak or absent. All the homozygous Mecr^tm(fabI) embryos were dead and at varying stages of resorption at E 11.5, establishing a period of embryonic lethality in knock-in mice from E 10.5 to E 11.5.

Chi square test analysis of embryos at E 8.5 indicated that the numbers of genotypes in each of generated modified mouse lines were not significantly different from expected Mendelian segregation. These results show that there was no genotype specific death among the experimental lines before E 8.5. Despite numerous efforts, we were unable to produce mouse embryonic fibroblasts from either the homozygous Mecr^tm1a or fabI knock-in embryos.

Protein lipoylation

Total protein extracts from all the Mecr knock-out and fabI gene replacement embryos (either E 11.5 or E 9.5) were probed with LA antibody. Presence of lipoylated proteins was used as indicator for functionality of the mtFAS pathway (5). The lipoylated E2 subunits of PDH (Dlat) and KGD (Dlst)) were present in extracts from all wild type samples as well as in corresponding heterozygotes (Fig. 4A). Interestingly, a clear lipoylated Dlat signal was observed in protein extracts from homozygous Mecr^tm1a embryos at E 11.5 (Fig. 4A). The observation is in line with the proposed leakiness of the Mecr^tm1a construct in homozygous embryos, allowing for synthesis of low levels of LA used for modification of Dlat. The result also agrees with the observations of the embryonic lethality studies (Supplementary Material, Table S1 and Fig. 4A). No lipoylated proteins could be detected in homozygous Mecr^tm1d knock-out embryos, consistent with complete loss of Mecr function in this embryo line (Fig.4 and Supplementary Material, Fig. S5A, lane 1). Extracts from the Mecr^tm(fabI) repeatedly revealed a faint band cross-reacting with the antiLA antibody in homozygous mutants. This band corresponds to lipoylated Dlat and the presence of this signal indicated a mild rescue of the enoyl-reductase deficiency in these specimens (Fig. 4A and Supplementary Material, Fig. S5B, lane 1). This is in line with the brief extension of embryonic survival of these fabI knock-in embryos.

Figure 4.

Protein lipoylation in modified Mecr embryos and cellular respiration in fabI knock-in embryos. (A) Western blot analysis for lipoylated E2 subunits of PDH (Dlat) and KGD (Dlst) in protein samples from embryos of different genotype. Mecr^tm1a (WT/WT and 1a/1a), Mecr^tm1d (WT/WT, 1d/WT, 1d/1d), 6 Protein molecular mass marker, Mecr^tm(fabI) (WT/WT, fabI/WT, fabI/fabI).The arrow points to the lipoylated Dlat band visualized weakly in samples from fabI/fabI embryos. (B) Ponceau staining for protein loading control of A. (C) Titration of oxygen consumption by permeabilized homogenized embryos at E 9.5 from Mecr^tm(fabI) breedings with various substrates. G + M+P is referring to respiratory rate in incubations using glutamate (G), malate (M) and pyruvate (P) as described in Materials and Methods. Note that basal rates of oxygen consumption were similar in all embryos tested at E 9.5. Relative respiratory rates after addition of ADP (D), cytochrome c (c), succinate (S) and F max after addition of the uncoupler FCCP into the incubations are indicated. The mutant knock-in embryo was unresponsive between titration of each substrate compared with wild type and heterozygous embryos. *P < 0.05, **P < 0.01, ***P < 0.001.

Mitochondrial respiratory chain activities in Mecr^tm(fabI) embryos

Mitochondrial dysfunction has been shown to lead to embryonic lethality (14–17), and mutations in mtFAS of S. cerevisiae have been reported to result in severe respiratory chain (RC) defects (6). We therefore investigated whether Mecr mutations are linked to mitochondrial RC dysfunction. As Mecr^tm1d embryos that were available at E 8.5 were too small for this type of analysis and were frequently already dead or dying at E 9.5, we turned to the Mecr^tm(fabI) model for which embryos were robustly alive at E 9.5. Mitochondrial respiratory function was analyzed from homogenized embryonic materials with an Oxygraph-2k.

Incubation of homogenized, permeabilized embryos in a mitochondrial respiration media supplemented with glutamate, malate, and pyruvate revealed similar basal O₂ consumption rates per embryo with 11.34 ± 2.98 pmol/s per embryo, 11.96 ± 2.04 pmol/s per embryo, 9.14 ± 2.22 pmol/s per embryo for wild type, heterozygous and fabI knock-in, respectively (values in means ± SD). The rate of oxygen consumption increased to 17.92 ± 4.64 pmol/s per embryo, 22.94 ± 5.10 pmol/s per embryo after addition of adenosine diphosphate (ADP) in wild type and heterozygous embryos whereas the increase in rate in homozygous fabI knock-in was not statistically significant (Fig. 4C). All the embryo preparations were unresponsive to the addition of cytochrome c, confirming the integrity of the outer mitochondrial membrane in these preparations. Further addition of succinate, wherefrom electrons are transferred to coenzyme Q bypassing the complex I in the RC, led to a further increase in oxygen consumption in wild type (28.42 ± 6.28 pmol/s per embryo) and heterozygous (33.46 ± 9.88 pmol/s per embryo) embryos. The fabI knock-in embryos also responded (12.90 ± 4.24 pmol/s per embryo), but the rate of oxygen consumption did not reach the level found in wild type and heterozygous embryos. The maximum rates of respiration determined by titration with the uncoupler carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) were 40.76 ± 5.58 pmol/s per embryo in wild type, 43.02 ± 11.68 pmol/s per embryo in heterozygotes and 15.72 ± 1.30 pmol/s per embryo in homozygous mutants.

Placental defects in homozygous Mecr^tm1d and Mecr^tm(fabI) mutant embryos

Placental insufficiency at E 10.5 is known to result in developmental failure (18). Because homozygous Mecr^tm1d and Mecr^tm(fabI) embryos showed pale vitelline vessels and experienced death at a matching time period, histological analyses of placenta were carried out at E 9.5. We found that the fetal part of the placentas of homozygous Mecr^tm1d and Mecr^tm(fabI) mutants was much thinner compared with wild type controls (Fig. 5A–D). Moreover, the organization of trophoblast giant cells, spongiotrophoblasts and the labyrinthine layer of the fetal part of placenta in these specimens was poor and maternal and fetal blood spaces in labyrinth were not evident (Fig. 5C and D).

Figure 5.

 Histological and ultrastructural characterization of placentas of control and Mecr mutants (A–D) Histology of placenta at E 9.5. Placentas associated with wild type (A), heterozygous (B) and homozygous fabI knock-in (C) embryos from Mecr^tm(fabI) breedings and homozygous Mecr^tm1d knock-out (D). The fetal placenta is thinner and disorganized in homozygous mutant of Mecr^tm(fabI) and Mecr^tm1d mouse lines. Scale bar 200 μm. Dotted lines demarcate the placental layers; de, decidua, sp, spongiotrophoblast and la, labyrinth. (E and F) transmission electron microscopy image of mitochondria (arrows) in labyrinthine trophobalst of wild type (E) and mutant (F) placenta from Mecr^tm(fabI) breedings. Scale bar 1μm.

Analyses of electron micrographs of placenta from Mecr^tm(fabI) embryos further confirmed failure of differentiation of labyrinthine trophoblasts as well as maternal and fetal blood spaces in labyrinth, which were indistinguishable. Furthermore, electron microscopic analyses revealed evenly distributed mitochondria of oval shape in wild type cells with clearly visible and organized cristae (Fig. 5E). In contrast, mitochondria in labyrinthine trophoblasts of homozygous mutant embryos were clustered, irregular in shape and electron-lucent without clear cristae structure (Fig. 5F).

Mecr deficiency is associated with disturbed transcriptional control by PPAR in fetal placenta

The histological and ultrastructure evidence clearly pointed to a role of Mecr in placental development. Immunohistochemical analysis revealed that Mecr expression was very weak in the fetal part of placentas connected to wild type and heterozygous embryos, and undetectable in the corresponding structures of homozygous fabI knock-in embryos. A strong immunohistochemical staining for Mecr was found in decidual sites of placentas irrespective of embryo genotype. Because the decidual sites are a maternal contribution to the placenta, the data indicate that the genotype of the embryos (wild type, heterozygous or fabI knock-in) does not affect Mecr expression in maternal decidua (Supplementary Material, Fig. S6D).

To obtain further information on effects of Mecr mutations on placental development, we analyzed the population of trophoblast giant cells and spongiotrophoblast cells in fetal placenta associated with wild type and homozygous fabI knock-in embryos. We carried out RNA in situ hybridization using probes specific for trophoblast giant cells (Pl1) as well as spongiotrophoblast cells (Tpbpa or 4311). The cellular linings of both trophoblast cell types were thinner in fetal placenta of homozygous knock-in embryos (Fig. 6A–D). In addition, folliculotropin (Fol) was also undetectable by immunohistochemical means in fabI knock-in placenta (Fig. 6F), whereas a strong signal was observed in wild type and heterozygous placenta (Fig. 6E).

Figure 6.

 Fetal placental analysis. RNA in situ hybridization (A–D) and immunohistochemistry (E and F) of placentas associated with wild type (top) and fabI knock-in (bottom) embryos from Mecr^tm(fabI) breedings. Riboprobes for a trophoblast giant cell marker (Pl 1) and a spongiotrophoblast marker (Tpbpa) are used. Trophoblast giant cell (A and B) and spongiotrophoblast (C and D) layers are markedly thinner in the mutants (B and D) than in wild type (A and C). Dotted lines demarcate maternal and fetal placenta. Scale bar 500 μm. (E and F) The Fol expression was negligible in fetal placenta of homozygous mutant (F) compared with wild type (E). Scale bar 200 μm. (G and H) RT-PCR analysis for PPAR target gene expression in total RNA samples from Mecr^tm(fabI) placenta. (G) Muc1, PPARγ target gene, n = 5 per group. (H) Mdfi, PPARβ target gene, n = 6 per group. The amounts of mRNA were normalized using actin (Actb) as reference gene. Results are shown as means ± SD, independent samples t test, *P < 0.05.

PPARγ and PPARβ/δ mediated signalling are essential for placental development (19,20) while no placental abnormalities in PPARα knock-outs were reported (21). Mecr has recently been implicated in PPARα and γ mediated signalling (11). Therefore, we analyzed PPAR target genes expression by RT-PCR using total RNA samples isolated from whole placentas of embryos of Mecr^tm(fabI) breedings. The Muc1 transcript was chosen as a representative PPARγ target gene (22), while the Mdfi transcript is a marker specific for PPARβ signalling (23). Both transcripts (Fig. 6G and H) were found to be reduced to 50% in homozygous mutant placenta compared with WT controls. This result suggests that a negative effect of Mecr on placental development is mediated in part by disturbed PPAR signallings.

Cardiac developmental defects in Mecr^tm(fabI) embryos at E 10.5

The histological analysis of hearts from wild type embryos revealed a clearly distinguishable atrial chamber, ventricular chamber, and endocardial cushion associated with atrio-ventricular canal at E 10.5 (Fig. 7A). In homozygous fabI knock-in mutants, hearts were underdeveloped and resembled embryonic heart at E 8.5–9.5. The primitive atrial chamber and ventricular chamber were observed in homozygous mutants (Fig. 7B). The trabeculation in myocardial wall of ventricular chambers was visible in both wild type and mutant heart (arrows, Fig. 7A and B).

Figure 7.

Morphology of heart at E10.5 in wild type and fabI knock-in embryos. (A and B) Histology of the embryo hearts. The atrium (a), ventricle (v) and endo-cardial ushion (e) are distinguishable in wild type embryos (A). The primitive atrial chamber (a) and ventricle (v) are the only distinguishable chambers in the heart of mutant embryos (B). Even though ventricular trabeculations can be seen in both embryos (arrows), heart development in mutant embryo was delayed compared with wild type control. Scale bar 200 µm. (C and D) µCT analysis of heart. The embryonic heart had three to four chambered appearance in control embryos (C) whereas two chambered in mutant hearts (D). The encircled heart region is shown in insert. The distance between each purple dot is 500 µm.

The embryos in uterine swelling were taken for micro computed tomography (µCT) imaging. Based on the phenotypes of more than 100 genetically confirmed embryos from Mecr^tm(fabI) breeding at E 10.5, the small and pale uterine swellings were taken to present homozygous fabI knock-in mutants and large and red uterine swellings were chosen as controls (wild type/heterozygous). The µCT analysis revealed that in mutant embryos heart and cavity volumes were 19 and 21% smaller than in controls, respectively, and the fabI knock-in embryo had a two chambered heart (Fig. 7D) whereas control embryo had three to four chambered heart (Fig. 7C).

Discussion

Mitochondrial defects are associated with many diseases like obesity, diabetes, cancer, cardiovascular and neurodegenerative disorders, demonstrating the vital role of these organelles in cellular function. Functional mitochondria are also essential for embryogenesis (14–17).

Aiming to understand the physiological role of mtFAS in mammals, we here generated several mutated Mecr mouse lines and analyzed the causes for the ensuing developmental defects. Unlike the mutations that cause MEPAN, which results from compromised function of human MECR (10), our knock-out and fabI knock-in constructs were unable to support development of an organism sufficiently healthy to survive until birth. Although the value of our results is limited in terms of in understanding the progression of MEPAN in patients, which may be more appropriately addressed in neuron-specific Mecr knock-out models, they yielded in some intriguing insights in the role of Mecr and mtFAS in embryonic development.

The surprisingly wide time window of embryonic death of the homozygous Mecr^tm1a mouse line embryos, varying from E 9.5 to E 15.5, was indicative of aborted embryonic development due to a variety of etiological factors (18). The live homozygous Mecr^tm1a ‘knock-out-first’ embryos at E 14.5 had hemorrhages and were paler. Paleness can be attributed to defective erythropoiesis (24) and hemorrhage may be due to failure in either platelet development (25) or angiogenesis (26). These defects may all contribute to embryonic death at this stage of development. Based on several observations, we propose that homozygous Mecr^tm1a embryos still have residual Mecr function. In this model, embryonic survival was prolonged compared with embryos lacking exon 2 or fabI replacement mutants. In addition, although neither a Mecr transcript nor Mecr protein can be detected by qPCR or visualized by western blots of homozygous Mecr^tm1a embryo extracts, respectively, the clear presence of lipoylated Dlat indicated mitochondrial synthesis of LA in these mutants. Because octanoic acid synthesized by mitochondria is used for synthesis of LA and all the exons of Mecr are intact, we conclude that this knock-out is leaky. This conclusion is supported by the complete absence of lipoylated Dlat in the Mecr^tm1d knock-out lacking exon 2 (Fig. 4A). If maternally derived LA could explain this observation, no difference in lipoylation should be detected between Mecr^tm1a and Mecr^tm1d lines.

Our studies on Mecr^tm1d and Mecr^tm(fabI) embryonic lethality show more narrow windows of embryonic death. Mecr^tm1d embryos died between E 9.5 and E 11.5, while the knock-in Mecr^tm(fabI) embryos perished between E 10.5 and E 11.5. Notably, also the retardation of embryonic growth was delayed in Mecr^tm(fabI) embryos compared with homozygous Mecr^tm1d embryos (E 9.5 and E 8.5, respectively). This difference in growth delay suggests that partial complementation of Mecr function is due to the ETR activity of fabI. Congruently, weak lipoylation of the Dlat in Mecr^tm(fabI) mutants could be detected by western blotting. Earlier studies by Yi and Maeda reported that LA synthase knock-out embryos die between E 7.5 and E 9.5 (27). The slight difference in period of embryonic lethality between Mecr^tm1d KO (C57BL/6) and Lias^−/− (129S) may be due to differences in the genetic background of mice strain used in these studies.

The growth delay and time point of death of homozygous mutants of Mecr^tm1d and Mecr^tm(fabI) embryos, match symptoms described as typical for indication of placental developmental defects in mice (28,29). Our investigation of the placental development in our mouse lines confirmed that these organs were hypoplastic and disorganized. The trophoblast cells have many functions in developing placenta, as for example secretion of hormones regulating the growth of fetal and maternal components of this organ (23,30–32). The overall smaller size and the irregular arrangement of the fetal part of the placenta (Fig. 5A–D) can be the result of trophoblast differentiation defects (23,33). Our RNA in situ hybridization studies for markers of trophoectoderm (Pl1) and ectoplacental cone descendants (Tpbpa/4311) (23,34,35) indicated a highly reduced cell population expressing Pl1 and Tpbpa/4311 in homozygous mutant embryonic placenta (Fig. 6B and D). The diminished trophoblast lining is congruent with the thinning of placenta we observed in histological analysis, as is the apparent absence of Fol in trophoblast cells of placenta associated with fabI knock-in embryos.

The PPAR signalling pathways play important roles in placental development (23,29,33,36,37). The transcriptional control of PPAR regulated genes by Mecr has been reported to be mediated by an mtFAS product (11) or a cytosolic isoform of Mecr (12). The evidence presented in the latter investigation suggests that the cytosolic Mecr variant arises from a splicing variant of the Mecr transcript that results in translation initiation in exon 2 further downstream of the coding region, producing a Mecr isoform devoid of the mitochondrial targeting signal. The expression of non-mitochondrial variant of Mecr might not be completely abolished by insertion of a transcriptional terminator after exon 1 in our Mecr^tm1a model. The presence of this Mecr variant in the embryos, which cannot be produced in either Mecr^tm1d or Mecr^tm(fabI) mutants, may explain in part the survival of ‘knock-out-first’ embryos until E 14.5. However, as protein lipoylation occurs in the Mecr^tm1a embryos, a certain amount of Mecr protein must be targeted to mitochondria also in these mutants.

Both the PPARβ/δ and PPARγ knock-out embryos die around E 10 with evident placental defects (23,36,38). In both of these mutants, disturbed trophoblast differentiation was reported as cause of death (23,33). Our analysis revealed a reduction of PPARγ and β/δ target gene expression in placentas of fabI knock-in embryos compared with wild type controls. As the RNA samples were isolated from placenta carrying both maternal and fetal contributions, the actual reduction of PPAR target gene expression is likely to be much more severe in the fetal fraction than evident in our data. Although our results present no definite proof of a role of Mecr in the PPAR signalling pathway, they are consistent with the previous reports of a possible regulatory role of Mecr mediated by PPAR pathways (11,12). It is therefore not completely ruled out that the possible disrupted PPAR signalling might contribute to placental developmental failure in the mutants. If such a link exists, it is worth of noting that the physiological ligands of PPARs are fatty acids and eicosanoids (39,40) and PPARs are located in the extramitochondrial space. However, free standing ACP of mtFAS is localized in mitochondria. Therefore, a role for a non-mitochondrial isoform of Mecr protein or unknown downstream events would be more plausible as effector to PPAR signalling than a product of mtFAS.

Another possible cause for embryonic lethality between E 9 and E 12 is a cardiovascular developmental defect (25). In line with the generalized retarded growth, the heart was found underdeveloped in the homozygous mutants compared with controls. Further histological analysis of homozygous fabI knock-in embryos at E 10.5 showed delayed cardiac development in comparison to wild type controls (Fig. 7A and B). The µCT analysis confirmed that heart appeared tubular in homozygous Mecr^tm(fabI) mutants, whereas chamber-like organization appeared in wild type controls at E 10.5 (Fig. 7C and D).

An important finding of our study was the effect of Mecr inactivation on mitochondrial function in embryos and placentas of Mecr^tm(fabI) mutants. The results of our electron microscopy studies on placentas of mutant embryos, revealing an abnormal mitochondrial structure in labyrinthine trophoblast, in conjunction with our observation of a reduced rate of mitochondrial respiration in permeabilized cells from fabI knock-in embryos indicate that absence of Mecr or diminished ETR activity yields in respiratory dysfunction also in mammals. This conclusion is in line with data reported for experiment on knocking down ACP in cultured HEK293T cells and in mitochondria isolated from tamoxifen inducible conditional Mcat knock-out mice (7,41) and congruent with results reported for yeast (6).

Several mouse models deficient in mitochondrial energy metabolism and embryonic development have been described in the literature. For instance, Johnson et al. (42) reported that a knock-out of the Pdha1 subunit of PDH yields in embryos with disrupted mitochondrial energy metabolism and developmental delay at E 9.5 stage followed by subsequent death. This time period is close to the findings reported here, suggesting that oxygen dependent mitochondrial metabolism plays an important role at this stage of development. This is in agreement with Larsson et al. (14), who reported embryonic death at E 8.5 to E 10.5 with loss of mitochondrial DNA and cytochrome c oxidase activity in mitochondrial transcription factor A knock-out embryos. Another example for mitochondrial dysfunction associated embryonic lethality is embryonic death at mid-gestation caused by cytochrome c deficiency (15). Although the conceptus is able to oxidize pyruvate during pre-implantation, there are several reports indicating that this stage of development can be successfully completed in spite of a respiratory defect (43). Energy metabolism of early post-implantation embryos depends on anaerobic glycolysis which is maintained until days E 6 to E 7.5 (44).

Our current work on mtFAS applying three different genetic modifications in Mecr gene showed that disruption of activity of this pathway leads to embryonic lethality due to pleiotropic effects including disturbances in fetal placental and cardiovascular developments and retardation of embryonic growth at the stage when the embryos become depending on aerobic metabolism. These defects coincided with disturbed PPARγ and PPARβ/δ-mediated signalling. Based on the analysis of several factors involved in mitochondrial function, we propose that diminished mitochondrial respiratory capacity is upstream of this pantropic morphological phenotype. In view of the data presented in this work, mtFAS defects should be considered as yet another potential factor for embryonic loss also in humans.

Materials and Methods

Mouse strains

Mice were maintained in the Experimental Animal Center of University of Oulu according to accepted criteria for humane care and use of experimental animals. The experimental protocols were approved by the Animal Care and Use Committee of National Animal Experiment Board, Finland (license number-ESAVI/7696/04.10.03/2012).

Production of Mecr^tm1a and Mecr^tm1d mice

‘Knock-out-first’ type Mecr mutant mice, B6Dnk; B6Brd; B6N-Tyr^c-BrdMecr^tm1a(EUC°^MM)Wtsi (EM: 04825) were derived from cryopreserved heterozygous embryos, obtained from the EMMA as a free of charge Transnational Access service. These animals harbor a construct where an FRT-flanked neo selection cassette, followed by a strong transcriptional terminator signal was placed between exons 1 and 2 of the Mecr gene, with loxP sites introduced 5′ and 3′ of exon 2 (Fig. 1A). The pups born after embryo transfer were back-crossed with C57BL/6JOlaHsd. Genotyping was done according to the supplier’s instructions.

To generate the Mecr^tm1d mice, heterozygous Mecr^tm1a mice were bred with transgenic mice Tg(ACTFLPe)9205Dym/J, expressing the FLP1 recombinase gene under the direction of the human ACTB promoter (45) (backcrossed to C57BL/6JOlaHsd), to delete the FRT-flanked selection cassette. The resulting mice (Mecr^tm1c) carry a conditional knock-out allele with Mecr exon 2 flanked by loxP sites. The heterozygous Mecr^tm1c strain mice were bred with Tg(CAG-Cre)13Miya mice with ubiquitous Cre expression (46) to generate Mecr^tm1d mice in which the floxed exon 2 was deleted (Fig. 1B). Genotyping was done using primers 5′-CAGGGCTGACCCAGAGTTTC- 3′ and 5′-GACCCTG CTCTCATG AGCTGTCC- 3′. The mutant amplicon was about 500 bp and wild type about 1000 bp.

Generation of Mecr^tm(fabI) mice

Starting with a commercial knockout cassette (loxP-FRT-PGK-gb2-neo-FRT, Gene Bridges, Heidelberg, Germany), a knock-in cassette for insertion of fabI in frame with base pair sequence of 5’ region of Mecr exon 2 was constructed. In vitro gene synthesis was used to produce a cassette with a fabI double HA tag replacing the loxP site in this construct, as well as homologous arms (Gene Script USA, Piscataway, NJ). The construct includes a NotI site followed by 26 bp from intron 1 and 54 bp from exon 2 of mouse Mecr, whole fabI with double HA tag, BGH transcription terminator and FRT-flanked neomycin selection cassette having another transcription terminator, and BamHI restriction site (Supplementary Material, Fig. S6A). The construct was verified by sequencing and then recombineered into a BAC containing Mecr (clone coordinate RP 24-76P5, BAC PAC, Children’s Hospital Oakland Research Institute, Oakland, CA). BAC clones harboring the insert were selected by chloramphenicol and kanamycin resistance and confirmed by PCR and agarose gel electrophoresis. The construct was then recombineered into targeting vector p15AmpDTA (47). The targeting construct (Fig. 1C) harbored by the p15AmpDTA vector contained 1.2 kb of Mecr sequence homology upstream of the insertion site (5′ arm) and 8 kb Mecr sequence 3′ of the insertion point (3′ arm). A NotI site was introduced to allow linearization of the vector for electroporation into embryonic stem (ES) cells. The correct clones were selected by resistance to ampicillin (targeting vector) and kanamycin (knock-in selection cassette) and further confirmed by PCR/agarose gel analysis (Supplementary Material, Fig. S6B) as well as sequencing of the final construct.

The NotI-linearized targeting vector was electroporated into PRX-B6N ES cells (Primogenix, Boulevard, MO) in the Biocenter Oulu Transgenic Core Facility. The ES cell clones containing the proper recombination product were first screened by PCR, followed by Southern blotting (Supplementary Material, Fig. S6C). The primers used for PCR screening of ES cell clones were (i) 5′- ATATGCCAGAGCCTGACATC- 3′, (ii) 5′- CTAATCCTGAG CAC TTC A  GAC- 3′ and (iii) 5′- GCGTTAACATAGTCACCATCC- 3′. (i) and (ii) yield in about 1400 bp product from wild type, and (i) and (iii) about 1600 bp product from fabI inserted DNA. Two ES cell clones were selected for injection into C57BL/6JOlaHsd blastocysts.

The male chimeras harboring a knock-in insert were bred with C57BL/6JOlaHsd females to generate heterozygous mice. The heterozygotes appeared healthy and were interbred to generate homozygous Mecr^tm(fabI) mice. A Mecr^tm(fabl) mouse line devoid of the selection cassette was also generated by deleting the FRT-flanked selection cassette by breeding with Tg(ACTFLPe)9205Dym/J (45) as above. The primer sequences used for genotyping of Mecr^tm(fabI) mice were (i) 5′- ATGATT TTT AGAACCAAGCAGGTTG- 3′, (ii) 5′-AATAGTAA CGGTACGGCGAA T CG- 3′ and (iii) 5′-CTAATCCTGAGCACTTCAGACTCTAC- 3’. (i) and (ii) yield in a 960 bp product from mutant allele, (i) and (iii) a 450 bp product from wild type allele.

Southern blotting

A 200 bp fragment of the upstream region of the 5′ homologous arm of Mecr is labelled by random priming was used as a probe (Random primed DNA labelling kit, Roche, Espoo, Finland). The restriction enzymes used were HindIII and EcoRV and routine procedure was done (48). The wild type bands ran close to the 3 kb whereas mutant DNA bands were almost 2.4 kb (Supplementary Material, Fig. S6C).

Western blot analysis of embryos

Total proteins were extracted using T-Per tissue protein extraction reagent (Thermo Scientific, Waltham, MA) according to manufacturer instructions. Immunoblotting was done with antibodies recognizing the HA epitope (1:4000, Sigma Aldrich, St Louis, MO), or Mecr antibody (1:2000, Cat. Nr 51027-2-AP; Proteintech, Chicago, IL), or antiLA antibody (1:1000, EMD Millipore, San Diego, CA). The antiMecr antibody specifically recognizes mouse and human Mecr protein as judged by detection of human MECR expressed in yeast, disappearance of the cross reacting band in patients with destabilizing MECR mutations (10) or upon shRNAi mediated knock-down of Mecr in mouse NIH3T3 cells (Supplementary Material, Fig.S7). The same antiLA antibody has been used successfully in previous publications by us (6,10) and others. For loading controls either β actin (Abcam, Cambridge, UK) or Ponceau S (Sigma Aldrich) solutions were used.

RT-PCR studies

Total RNA from Mecr^tm1d and Mecr^tm(fabI) mouse embryos at E 9.5 and Mecr^tm1a at E 10.5 were isolated with an RNeasy Mini Kit (Qiagen, Hilden, Germany) and all RNA samples were DNase-treated with RNase- Free DNase Set (Qiagen). cDNA was produced with a Revert Aid First Strand cDNA Synthesis Kit (Thermo Scientific). qPCR was performed with a 7500 Real Time PCR System (Applied Biosystems, Foster City, CA) using fluorogenic probe-based Taqman chemistry with the Taqman probes for mouse Mecr and Taqman Universal PCR Master Mix (Applied Biosystems) according to the manufacturer’s instructions. For relative quantification, the results were normalized with actin, Actb, as endogenous controls. To study PPAR target gene expression, RNA samples were isolated from placenta at E 9.5 as described earlier. cDNA was synthesized from 1 µg RNA. Muc1 (22) and Mdfi (23) probes were used as markers for PPARγ and PPARβ/δ target genes, respectively. The protocol was same as described above. For the fabI transcription level analysis, qPCR was done in CFX 96 Real-Time PCR system (Bio-Rad, Helsinki, Finland) using iTaq Universal SYBR Green Supermix and specific primers. The fabI primers were 5′-CATCGACACCATGTTCGC- 3′ and 5′- GCGTTAACATAGTCACCATCC- 3′’. The endogenous control, β-Actin, primers were 5′-GGCTGTATTCCCCTCCATCG-3′and 5′-CCAGTTGGTAACAATGCCATGT-3′.

Mitochondrial respiratory complex activity studies

The respiratory complex activity measurements were done by modification of the protocol described previously in (49). Briefly, the embryos were dissected from pregnant mothers and kept in PBS on ice until the measurements were carried out. Embryo material was gently homogenized in mitochondrial respiration medium MiR05 (0.5 mM EGTA; 3 mM MgCl₂; 60 mM lactobionic acid; 20 mM taurine; 10 mM KP_i; 20 mM HEPES; 110 mM D-sucrose; 1 g/l BSA, essentially fatty acid free; pH 7.1) before measurement.

The homogenate prepared from each embryo was placed into the Oxygraph-2k chamber (OROBOROS Instruments, Innsbruck, Austria) in a total volume of 2 ml. The chamber was closed and the cells were permeabilized by titration of 5 µg digitonin into the chamber. Glutamate (final concentration of 10 mM), malate (2 mM) and freshly prepared pyruvate (5 mM) were titrated into the chamber and the baseline respiration of the permeabilized cells in the homogenate was measured. Further substrates were titrated in the chamber in the following order: ADP (5 mM), cytochrome c (10 µM), succinate (10 mM) and FCCP (titrated into the incubation mixture in 0.5 µM steps until maximum respiration is reached) as uncoupler. The change of oxygen consumption in the chamber after each titration was recorded.

Embryonic lethality studies

The heterozygous females and males of ‘knock-out-first’ (Mecr^tm1a), Mecr null mutants (Mecr^tm1d) and Mecr^tm(fabI) lines were bred in this study. The period of embryonic lethality in Mecr^tm1a, Mecr^tm1d and Mecr^tm(fabI) mice were determined by timed mating and dissection of pregnant heterozygous females of all lines, at various days of gestation. The morphological appearance and viability of the embryos were assessed. To determine the genotype of embryos, a tail sample or whole embryos (if they were E 8.5 or younger) were collected. Embryos from E 9.5 and older were collected for histological, biochemical and molecular studies. Placenta samples were collected for molecular, ultrastructural and histological analysis.

Histological analysis

Each uterine swelling was separated. The outer layers of the uterus were removed and the embryos were harvested. The rest of myometrium with placenta was fixed in 4% paraformaldehyde in 100 mM NaP_i buffer (pH 7.4) over night and the uterus was cut into halves mid-sagittally. One of the half was used for regular histology processing and embedded in paraffin. Sagittal sections of 5 µm were used for haematoxylin and eosin (H&E) staining or immunohistochemistry with Mecr antibody (1:250). Sections of 2.5 µm were used for immunohistochemistry probing for Fol with antibody raised against human chorionic gonadotropin ((1: 4000, Dako, Glostrup, Denmark) and for RNA in situ hybridization respectively. The 5 µm sections from wild type and fabI knock-in embryos at E 10.5 from Mecr^tm(fabI) breedings were also processed similarly to analyze the heart morphology. The stained sections were analyzed with Olympus BX 51 microscope equipped with Cell^ M imaging software (Olympus, Planegg, Germany).

Transmission electron microscopy

Placenta samples were fixed in a glutaraldehyde (1%), formaldehyde mixture (4%) in 100 mM NaP_i. They were post fixed in 1% osmium tetroxide, dehydrated in acetone and embedded in Epon LX 112 (Ladd Research Industries, Williston, VT). Thin sections (70 nm) were cut with Leica Ultracut UCT ultramicrotome, stained with uranyl acetate and lead citrate and examined in Tecnai G2 Spirit 120 kV transmission electron microscope (FEI Europe, Eindhoven, The Netherlands). Images were captured by Quemesa CCD camera operated with iTEM software (Olympus Soft Imaging Solutions, Munster, Germany).

Micro computed tomography

The uterine swelling with embryos at E 10.5 were dissected and prepared for high-resolution µCT imaging. For microstructural imaging samples were paraformaldehyde fixed, dehydrated with graded ethanol series and moved to hexamethyldisilazane (HMDS). HMDS treated uterine swellings were attached with dental wax to holder and placed in µCT scanner (Skyscan 1272, Brüker microCT, Kontich, Belgium). For imaging low energy X-rays with 40 kV and no filtering was applied. Pixel size was set to 1.75 µm and 1800 projections were collected over 360°. Each projection was collected for 1815 ms and averaged three times. Projection images were reconstructed with NRecon-software (v.1.6.9.8, Brüker microCT) and ring artefact and beam hardening corrections were applied. 3D data sets were visualized with CTVox (3.2.0 r1294). Heart morphology was analyzed using CTAn software, where heart was first manually segmented by drawing regions of interest (ROIs). ROIs were refined with shrink-wrap function and original image. Chambers were extracted with threshold and morphological operations and 3D model and analyses were performed. This provided measures of heart, heart muscle volume and chambers. Results from analysis were rendered in CTVol software.

RNA in situ hybridization

The 7 µm sagittal sections of myometrium with placentas from E 9.5 embryos were used. Briefly, antisense and sense riboprobes for Pl1 (34,35) and Tpbpa or 4311 (23), were generated by in vitro transcription of the corresponding cDNA clones with SP6, T7 or T3 RNA polymerase, using DIG RNA labelling mix (Roche). In situ hybridizations were done following the routine method (50). After hybridization and post hybridization washing, digoxigenin labelled sections were detected by alkaline phosphatase conjugated antiDIG antibody and Fast red as substrate (Roche). Counterstaining was done with Mayer’s hematoxylin and coverslips mounted with ImmuMount (Thermo Scientific).

Statistical analysis

Chi square test was applied to analyze whether the genotypes of embryos in each Mecr modified mouse lines followed Mendelian segregation. The transcription levels are given by means and SDs and the statistical significances of RT-PCR analysis were evaluated with independent samples t test. The outcomes of the oxygen consumption variables are grouped by genotype and presented graphically. Independent samples t test was used to evaluate the statistical significances of differences in outcome levels across genotype groups. To analyze the change in outcome levels of substrate titrations within the genotypes, we applied t test for repeated measurements. IBM SPSS Statistics version 22 and GraphPad Prism version 5.00 were used for the statistical analyses.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We are grateful to Dr Raija Soininen for the continuous support at various stages of this project and critical review of this article, Dr Reetta Vuolteenaho and Dr Harri Elamaa for helpful discussions. Dr Raija Sormunen and Biocenter Oulu EM laboratory, part of Biocenter Finland, is acknowledged for the support of EM analysis. We thank Leena Ollitervo and Heli Ylisuutari for technical help with histology samples and Jana Muschhammer and Mahmoud Elkhwanky for their assistance in genotyping samples, Dr Janet Rossant (University of Toronto) for kindly gifting Pl1 and Tpbpa plasmids, Raman Devarajan for the cybergreen method of RT-PCR and Dr Antti Hassinen for confocal microscopy.

Conflict of Interest statement. None declared.

Funding

The project was funded by the Academy of Finland, the Sigrid Juselius Foundation and Biocenter Finland. Part of this work has been supported by the INFRAFRONTIER-I3 project under the EU contract Grant Agreement Number (3123250 of the EU FP7 Capacities Specific Programme) and the European commission FP-INFRASTRUCTURES Grant 227490 (Transnational access to EMMA mice).