Introduction

Eukaryotic cells regulate mitochondrial shape and function through the complementary processes of mitochondrial fission and fusion. Normal fission and fusion protect mitochondria from becoming isolated, autonomous organelles that would be susceptible to depletion of substrates and metabolites or mutations in the mitochondrial DNA (mtDNA). When these dynamics are lacking, mitochondria develop abnormal morphologies (reviewed in Detmer and Chan 2007). The biological significance of these dynamics is clear, since disrupting them leads to alterations in apoptosis, calcium buffering, and ATP production (Duchen 2000; Kroemer and Reed 2000; Rizzuto et al. 2000). Mutations in a number of genes that regulate mitochondrial dynamics have been associated with several human neurodegenerative diseases (reviewed in Chen and Chan 2009; Knott et al. 2008). Furthermore, neurons are found to be especially susceptible to disruptions in fission and fusion, although it remains unclear why this is so given the fundamental role of mitochondria in all cells.

The core proteins regulating fission and fusion are well known. Fission is regulated by two GTPases: dynamin-related protein 1 (Drp1) and fission 1 (Fis1) (Bleazard et al. 1999; Fekkes et al. 2000; James et al. 2003; Labrousse et al. 1999; Mozdy et al. 2000; Smirnova et al. 2001; Tieu and Nunnari 2000; Yoon et al. 2003). Fusion is mediated by the mitochondrial outer-membrane proteins mitofusin 1 and 2 (Mfn1 and Mfn2), as well as the mitochondrial inner-membrane protein Opa1 (Chen and Chan 2004; Hales and Fuller 1997; Hermann et al. 1998; Rapaport et al. 1998). Mitochondria lacking the Mfns or Opa1 are unable to fuse and therefore appear fragmented, whereas loss of either Drp1 or Fis1 causes mitochondria to elongate and form long, continuous tubules (Chen et al. 2005; Cipolat et al. 2004; Griparic et al. 2004; Lee et al. 2004; Smirnova et al. 2001).

Opa1 is a dynamin family GTPase containing two coiled-coil domains, a mitochondrial import sequence, the GTPase domain, and an uncharacterized region called the middle domain. Opa1 is known to function independently in mitochondrial fusion and cellular apoptosis (Frezza et al. 2006; Spinazzi et al. 2008). There are eight isoforms of Opa1 that exist due to alternative splicing of exons 4, 4b, and 5b (Delettre et al. 2001; Olichon et al. 2007; Song et al. 2007). Exon 4 mediates the fusion of the mitochondrial network, whereas exons 4b and 5b are essential in the release of cytochrome c, an apoptotic response. In addition, Opa1 post-translational processing cleaves the protein into long and short isoforms that oligomerize to mediate fusion (Delettre et al. 2001; Griparic et al. 2007; Herlan et al. 2003; Merkwirth et al. 2008; Park et al. 2005; Song et al. 2007).

Opa1 has been well studied as the gene most frequently mutated in autosomal dominant optic atrophy (ADOA), one of the most common forms of inherited blindness affecting 1 in 12,000-50,000 individuals (Alexander et al. 2000; Delettre et al. 2000; Kivlin et al. 1983; Kjer et al. 1996). While the cause of ADOA is widely accepted as haploinsufficiency, the reason for the specific sensitivity of the retinal ganglion cells remains unclear. In an effort to develop animal models for ADOA, two groups have examined the heterozygous mouse phenotype associated with distinct point mutations in Opa1 (Alavi et al. 2007; Davies et al. 2007). In both alleles, they found less Opa1 in protein extracts from heterozygous mutants than in extracts from wild-type controls, supporting the notion that haploinsufficiency underlies ADOA. Furthermore, the animal models showed progressive phenotypes in the eye that paralleled the human disease. In the course of these analyses, the groups reported that mice homozygous for either Opa1 mutation are embryonic lethal, since there were no homozygous mutant pups resulting from heterozygous intercrosses; however, in neither case did they specifically characterize the phenotype (Alavi et al. 2007; Davies et al. 2007).

Here we identify and provide the first molecular characterization of the embryonic phenotype of a novel ENU-induced Opa1 mouse mutation, lilR3. lilR3 mutant embryos die at E11.5 due to loss of mitochondrial retention of Opa1. The lilR3 mutation disrupts a conserved amino acid in an uncharacterized domain of the protein, indicating that this domain is required for Opa1 localization to mitochondria. Embryos with mislocalized Opa1 specify the basic body plan normally; however, as the embryos begin the growth and differentiation phase of embryogenesis, they die. We find abnormal signaling in pathways known to be necessary for the differentiation of specific cell fates in the posterior of the embryo. Surprisingly, the same pathways are normal in the anterior of the embryo, where we see increased cell death. These results indicate that the lack of Opa1 affects different cell types in distinct ways, reminiscent of the human disease state in which specific neural cells are most sensitive to Opa1 dosage.

Results

Meiotic mapping and sequencing identify Opa1 mutation in lilR3

In the course of a chemical mutagenesis screen for recessive mutations critical in mammalian embryogenesis, we found the lilR3 mutant line as a result of its small size, failure to properly rotate along the axis, and exencephaly at E9.5 (Garcia-Garcia et al. 2005). As we induced the lilR3 mutation on a C57BL/6 background, we performed a backcross to C3H/HeJ to map and identify the mutation by sequencing candidate genes (see Experimental Procedures for details). From this sequencing we identified a single mutation: an A → G transition mutation in intron 19 of Opa1 that created a cryptic splice acceptor site (Fig. 1). From the sequencing of Opa1 cDNA extracted from lilR3 embryos, we verified that the predominant transcript produced from the mutant allele used the cryptic splice site, resulting in a 6-bp insertion of intronic sequence. This aberrant transcript was predicted to result in the translation of a cysteine (C) in place of a tryptophan (W) at amino acid 616, as well as the addition of amino acids leucine (L) and arginine (R) due to the translation of intronic sequence (W616CinsLR). These amino acids lie within the middle domain, a highly conserved but uncharacterized domain of the Opa1 protein. The disrupted tryptophan residue is an invariant amino acid in Opa1 homologs in flies, worm, and human. Furthermore, there are three ADOA patients whose mutations disrupt W616 in Opa1 (Alexander et al. 2000; Ferre et al. 2009).

Fig. 1
figure 1

The putative causative mutation of lilR3 is located in Opa1. Illustration of wild-type and Opa1 lilR3 sequences of the mutated region. An A → G transition in intron 19 creates a cryptic splice site that results in the insertion of six intronic bases in frame with the Opa1 coding sequence. A new codon results from the use of the cryptic splice acceptor between exons 19 and 20. This results in a C in place of a W at amino acid 616. The addition of the intronic sequences creates codons for L and R. These mutated amino acids are within an uncharacterized domain of the Opa1 protein called the middle domain

Full size image

Opa1 is one of three GTPases known to regulate mitochondrial fusion. When Opa1 is knocked down in cells, mitochondrial fragmentation results (Chen et al. 2005; Cipolat et al. 2004; Griparic et al. 2004; Olichon et al. 2003). To determine whether the mutation we found in Opa1 disrupts Opa1 function, we examined mitochondrial morphology in vivo. To circumvent technical difficulties associated with doing this in whole embryos, we isolated primary mouse embryonic fibroblasts (MEFs) from E10.5 embryos and used MitoTracker to visualize the morphology of the mitochondria using confocal microscopy (Fig. 2). In the wild-type MEFs, we saw morphologically normal mitochondria. However, in lilR3 heterozygous and homozygous MEFs, we observed fragmentation of the mitochondria consistent with the lilR3 mutation disrupting Opa1 function. The importance of Opa1 dosage was underscored by a more severe fragmentation phenotype in the homozygous mutant MEFs.

Fig. 2
figure 2

The lilR3 mutation leads to mitochondrial fragmentation. Using MitoTracker dye to visualize the mitochondria in primary mouse embryonic fibroblasts (MEFs). a, b Wild-type MEFs display an abundance of tubular mitochondrial networks in their cytoplasm, with only a small population of the mitochondria being fragmented. c, d Heterozygotes for the lilR3 mutation show a mixture of fragmented and tubular mitochondria. e, f Homozygous lilR3 mutant MEFs display mostly fragmented mitochondria in their cytoplasm

Full size image

The dosage effect and fragmentation phenotype of the lilR3 mutation suggested that it is a loss-of-function mutation. To investigate this we performed Western blot analysis on protein extracts from wild-type, heterozygous, and lilR3 mutant E10.5 embryos using an antibody that recognizes the C-terminus of the Opa1 protein, a domain common to all isoforms (Misaka et al. 2002) (Fig. 3). As Opa1 localizes to the inner mitochondrial membrane, we separated the mitochondrial protein fraction from the cytosolic cellular fraction by centrifugation. As expected, in wild-type embryos we found Opa1 protein predominantly in the mitochondrial fraction, which was confirmed by the presence of cytochrome c. In heterozygous protein extracts we saw Opa1 protein in both the cytosolic and mitochondrial fractions, suggesting that the lilR3 mutation caused Opa1 to mislocalize from the mitochondria. This was confirmed in our analysis of the homozygous fractions where we observed all Opa1 protein in the cytosolic fraction. We confirmed that we were analyzing equivalent amounts of total protein in all lanes by reprobing the blot with antibody against tubulin. Thus, Opa1 protein containing the W616CinsLR mutation results in stable mislocalized protein.

Fig. 3
figure 3

Mislocalization of Opa1 protein in lilR3 embryos. Western blot analysis of mitochondrial (left panels) and cytosolic (right panels) E10.5 embryo protein lysates using antibodies against Opa1, α-tubulin (loading control), and cytochrome c (mitochondrial marker, demonstrating the efficiency of fractionation). Normalized Opa1 expression is designated for each lane as the ratio of Opa1/tubulin expression relative to a wild-type sample set to 1. Opa1 is enriched in the wild-type (+/+) mitochondrial fraction, consistent with its known localization to the mitochondrial inner membrane. Opa1 is present in both the mitochondrial and cytosolic fractions from heterozygous (+/lilR3) extracts, whereas Opa1 is exclusively in the cytosolic fraction from homozygous (lilR3/lilR3) mutant extracts, indicating that the lilR3 mutation causes Opa1 to mislocalize from the mitochondria. Upon overexposure, a faint band of Opa1 is visible in the mitochondrial fraction of the homozygous mutant extract (data not shown). Thus, the W616CinsLR mutation in the Opa1 middle domain disrupts the mitochondrial localization of Opa1, consistent with the lilR3 allele being a strong loss-of-function mutation

Full size image

Morphological defects of the Opa1 lilR3 homozygous mutant

We identified Opa1 lilR3 due to its obvious morphological defects at E9.5: developmental delay, failure to properly rotate the body along the axis, small size, and exencephaly (Fig. 4a–d). At E8.5, Opa1 lilR3 heterozygous and homozygous mutant embryos are indistinguishable from their wild-type littermates. By E10.5, Opa1 lilR3 homozygous mutants appeared to have grown but were still much smaller than littermate controls. Craniofacial defects, including hypomorphic arches and clefting of the nasal process, were apparent (Fig. 4a, b upper inset). We also found hemorrhaging, indicating defects in vascular integrity (Fig. 4c, d). The Opa1 lilR3 somites were smaller and seemed disorganized compared with the solid cube structures of the littermate controls (Fig. 4a, b lower inset). At E11.5, the hearts stopped beating indicating the time of lethality in the Opa1 lilR3 embryos. The fact that the Opa1 lilR3 homozygous mutants displayed overall developmental and growth delay reflects the ubiquitous expression of Opa1 (Rochais et al. 2009).

Fig. 4
figure 4

Morphological defects of the E10.5 homozygous Opa1 lilR3 mutant embryo. Wild-type littermate control (a) and homozygous Opa1 lilR3 (b-d) embryos at E10.5. a Lateral view of a littermate control. b Ventral view of Opa1 lilR3 mutant. Inset is a higher-magnification view of the hypomorphic somites. c Lateral view of the craniofacial and pharyngeal structures. Note the blood pooling in the frontonasal process, cranial folds, and the pharyngeal arches. d Dorsal view of exencephaly. pa pharyngeal arches, fn frontonasal process, som somites, ht heart tube, ba branchial arch, ov otic vesicles. Anterior is up, posterior is down

Full size image

Key signaling pathways in embryonic development are disrupted in the Opa1 lilR3 homozygous mutant

Considering the morphological defects of the Opa1 lilR3 embryos, we examined genes known to regulate the development of the disrupted tissues. For example, hypomorphic pharyngeal arches and vascular defects can be the result of improper neural crest development. Some of the pathways important in neural crest development also regulate neural tube closure, as well as somitogenesis, all of which are disrupted in Opa1 lilR3. Therefore, we examined the expression of several genes that are required markers of these important processes: Sox10, Pax3, Wnt1, and Msx1.

Sox10 is a transcription factor required for proper neural crest development. It maintains early-stage multipotency and directs late-stage non-neuronal fates in neural crest cells (Kim et al. 2003). In mice, Sox10 is expressed in pre- and post-migratory neural crest cells all along the A–P axis. By E9.5, the neural crest has almost completed its emigration from the dorsal neural tube to ventral target tissues. We saw Sox10-positive neural crest in the pharyngeal region migratory streams and along the dorsal neural tube in both the control and Opa1lilR3 mutant E9.5 embryos (Fig. 5a, b). However, the posterior boundary of Sox10 expression is further anterior in the Opa1lilR3 mutant embryos (Fig. 5b inset). Considering the growth delay, it was possible that Sox10 expression had simply not been induced yet in the posterior trunk region of the mutant embryo at E9.5. However, by E10.5, we still detected no Sox10 expression in the posterior trunk of the Opa1lilR3 mutant, whereas there was robust Sox10 expression along the entire length of the A-P axis in the control (Fig. 5c, d). Furthermore, we found that Sox10 expression was diminished in the craniofacial tissues of the Opa1lilR3 mutant at E10.5.

Fig. 5
figure 5

Misexpression of key genes involved in neural crest development. Whole-mount in situ hybridization of a, c, e, g, i, i′ wild-type control littermates and b, d, f, h, j, j′ Opa1 lilR3 homozygous mutant embryos. Expression of a neural crest marker Sox10 at E9.5 in a control and b Opa1 lilR3 embryos and in E10.5 c control and d Opa1 lilR3 embryos. Pax3 expression in E9.5 e control and f Opa1 lilR3 embryos. Expression of Wnt1 in E10.5 g control and h Opa1 lilR3 mutant embryos. Msx1 expression at E10.5 in i control and j Opa1 lilR3 mutant embryos

Full size image

Pax3, Wnt1, and Msx1 are all expressed in the neural tube and are instrumental in neural closure and neural crest development. We examined Pax3 expression at E9.5 and found relatively normal expression in the head, anterior neural tube, and lateral mesoderm of the Opa1lilR3 mutant embryos, but no expression in the somites (Fig. 5e, f). However, analogous to the axial difference we saw in Sox10 expression, Pax3 expression was absent in the posterior of Opa1lilR3 embryos. We found this same pattern with expression of both Wnt1 and Msx1 at E10.5, when both are missing in the posterior (Fig. 5g–j).

Anterior-posterior patterning of the axis is grossly normal in the Opa1 lilR3 mutant

An interesting aspect of the expression analysis above is that Sox10, Wnt1, Pax3, and Msx1 were most severely disrupted in the posterior of the Opa1 lilR3 mutant embryos. It is unclear why particular subsets of genes are not expressed in specific cell types or at certain axial positions, while others appear normal. Axial-specific defects could be explained if A–P patterning of the embryonic body plan was disrupted by the Opa1 lilR3 mutation. Since Hox genes regulate the establishment of the A–P axis, we examined the expression of Hoxb6 and Hoxb9 and found they were normal (Fig. 6). Thus, at a gross level, the initial A-P patterning is not disrupted in the Opa1 lilR3 mutants.

Fig. 6
figure 6

Expression of Hoxb6 and Hoxb9 appears grossly normal in the Opa1 lilR3 mutant embryo. Whole-mount in situ hybridization of E10.5 (a, c) control littermates and (b, d) Opa1 lilR3 mutants. Arrows indicate anterior limits of expression

Full size image

Somitic expression of proliferative genes is diminished or lost, explaining the failure in somite development

The somites in Opa1 lilR3 embryos formed but looked abnormal; so, in addition to Pax3 we investigated somitic development by examining Pax1 and Pax9 expression. Pax1 and Pax9 are paired box transcription factors expressed in the sclerotome, the ventromedial region of the somites. Consistent with the morphological defects we observed in somites, we found that like Pax3 expression, expression of Pax1 and Pax9 was diminished or lost in all somites of the Opa1 lilR3 mutants (Fig. 5f). The absence of Pax1, 3, and 9 indicated that the somites do not follow their normal developmental program in Opa1 lilR3 mutants (Fig. 7a–d), and unlike what we observed with Wnt1 and Msx1 expression, Pax expression in the somites was disrupted regardless of axial position.

Fig. 7
figure 7

Hypomorphic somite development is likely due to a failure to express Pax1 and Pax9. Pax9 expression in E9.5 a control and b Opa1 lilR3 embryos. Pax1 expression at E10.5 in c control and d Opa1 lilR3 mutant embryos. Note that Pax9 and Pax1 are expressed in the pharyngeal endoderm but not in the somites of the Opa1 lilR3 mutant compared with the control littermates. Both genes are required for somite growth

Full size image

Cell death analysis of Opa1 lilR3 mutant embryos

Several factors prompted us to examine cell death in the Opa1 lilR3 mutants. First, mitochondrial dynamics have been linked to apoptosis. The loss of Opa1 has been found to correlate with increased apoptosis under certain cell culture conditions (Cipolat et al. 2006; Frezza et al. 2006; Olichon et al. 2007). Second, Pax3 homozygous mutants display increased apoptosis in somites and the neural tube (Goulding et al. 1991). Third, the loss of Pax1 and Pax9 results in more apoptosis in the sclerotome (Peters et al. 1999). Finally, the lack of Sox10, Wnt1, Pax3, and Msx1 expression in the posterior of Opa1 lilR3 embryos could have been explained by the death of those cells. Thus, we analyzed cell death using LysoTracker Red, a vital dye that monitors phagolysosomal activity during the engulfment of apoptotic bodies (Zucker et al. 1999). LysoTracker Red has the advantage of permeating whole embryos well, enabling us to visualize cell death throughout the embryo.

At E9.5 we found that the pattern of cell death was relatively similar in the E9.5 Opa1 lilR3 mutant and the wild-type control embryos, indicating that the aberrant gene expression we observe between Opa1 lilR3 mutant and wild-type littermates is not simply a consequence of general cell death (Fig. 8a, b). By E10.5 we observed an expansion in the pattern of cell death in Opa1 lilR3 mutant embryos compared with wild-type (Fig. 8c, d). From dorsal views, it was apparent that the cell death was concentrated in the somites, consistent with the absence of Pax1, 3, or 9 expression in somites along the A-P axis.

Fig. 8
figure 8

Cell death is not responsible for the growth delay in the lilR3 homozygous mutant embryos. Cell death as assessed in whole embryos using LysoTracker. The pattern of cell death is comparable to that seen in a control and b Opa1 lilR3 mutant embryos at E9.5. A more broad pattern of cell death is visible in both c control and d Opa1 lilR3 embryos at E10.5. d′ A higher-magnification view of cell death in the somites of the Opa1 lilR3 mutant embryo in (d). However, cell death is visible along the neuraxis of the mutant and not in the control. 1 and 2, arches 1 and 2; ht, heart

Full size image

Discussion

Here we characterize a novel allele of Opa1 and for the first time ascribe a function to the middle domain of the Opa1 protein. Based on our observation of the mitochondrial fragmentation phenotype in lilR3 MEFs and the mislocalization of Opa1 protein in the embryos, we are confident that the A → G transition in Opa1 is responsible for the lilR3 phenotype. Mitochondrial fragmentation in the MEFs was similar to mitochondrial fragmentation in cell lines from patients with heterozygous mutations in Opa1 as well as cell lines in which Opa1 expression was knocked down by siRNA (Amati-Bonneau et al. 2005, 2008, 2009; Chen et al. 2005; Cipolat et al. 2004; Griparic et al. 2004; Olichon et al. 2003, 2007; Zanna et al. 2008). Three ADOA patients have been identified whose mutations disrupt W616 as the lilR3 allele does (Alexander et al. 2000; Ferre et al. 2009). Furthermore, the embryos die at midgestation, which is about the same time as embryos with mutations in either one of the other components of the fusion machinery (Mfn1 and Mfn2) or the proteins that process Opa1 (Phb2) (Chen et al. 2003; Merkwirth et al. 2008; Park et al. 2005). Thus our data are consistent with Opa1 function being dependent on its mitochondrial localization.

The mislocalization of Opa1 protein in the lilR3 mutant raised the possibility that gain of function due to the accumulation of Opa1 in the cytosol contributes to the phenotype. While this remains a possibility, it seems unlikely given the previous analysis of two distinct ENU-induced point mutations in Opa1 (Alavi et al. 2007; Davies et al. 2007). In both cases, the authors found a decrease in Opa1 protein expression in the heterozygous mutants. One group analyzed the cytosolic and mitochondrial cell fractions of their ENU allele and saw no mislocalization of Opa1 protein (Alavi et al. 2007). Thus, these two alleles appear to result in less Opa1 protein. Both groups focused their work on careful characterization of the heterozygous mice as animal models for ADOA; however, they did report that homozygous embryos died around E11.5, and in one case published a photo of the mutant embryo that displayed features similar to those we observed (Davies et al. 2007). Specifically, their photograph showed an embryo with growth delay, exencephaly, and a body shape that is characteristic of the lilR3 phenotype. While their analysis was on a distinct genetic background, the identical times of death in all three alleles, the overlapping phenotypes, and the suggestion that the other two alleles result in the loss of Opa1 protein are most consistent with the lilR3 mutation being a loss-of-function allele. Although technically it is possible that one of the 13 remaining genes in the lilR3 interval is also affected, the facts that the distinct Opa1 alleles phenocopy one another and were discovered independently along with the mislocalization of Opa1 protein in lilR3 embryos are most consistent with lilR3 affecting Opa1 alone.

The precise defect underlying the lethality of Opa1 lilR3 mutants remains unclear. Mitochondrial dysfunction in humans has been linked to a variety of cardiomyopathies, and other mouse mutants with defects in mitochondrial fission and fusion display abnormal cardiac function, so it is possible that Opa1 lilR3 lethality may be due to abnormal heart development (Marin-Garcia et al. 1995, 1996, 2001; Zell et al. 1997).

Our analysis provides the first molecular characterization of the homozygous phenotype involving loss of Opa1 function. Curiously, we saw that gene expression was more affected posteriorly despite a normal anterior-posterior axis. The phenotype we observed raises the possibility that distinct types of neural crest may have distinct requirements for Opa1, or even for mitochondrial dynamics in general. Ever since the discovery linking Opa1 to ADOA, it has been clear that retinal ganglion cells are more sensitive to decreased Opa1 than other cells (Alexander et al. 2000; Delettre et al. 2000). Several factors are proposed to explain these differences. First, retinal ganglion cells may preferentially express certain Opa1 isoforms. Second, cells have distinct energy needs, and the speculation is that those with the highest needs would be most sensitive to perturbations in mitochondrial dynamics. Furthermore, while mitochondria are most famous for ATP production, they also function to buffer intracellular calcium and are central to apoptosis. Thus, the loss of Opa1 and the resulting change in mitochondrial shape could influence any of these processes.

A reasonable interpretation of the distinct patterns of cell death and gene expression we saw in Opa1 lilR3 embryos is that these patterns correlate to the relative importance of these distinct mitochondrial processes in different cell populations at specific times in development. Our analysis points to tissue-specific functions of Opa1. The somites give rise to the skeleton and muscle of the adult. Several ADOA patients have been reported to have additional phenotypes, including myopathy, deafness, and ataxia (Chen et al. 2007; Ferraris et al. 2008; Stewart et al. 2008; Treft et al. 1984). Other human conditions associated with mitochondrial dysfunction display abnormal processing of Opa1 and skeletal defects (Duvezin-Caubet et al. 2006). The somite-specific cell death we see in our Opa1 lilR3 mutants would be consistent with somatic lineage being sensitive to the loss of the apoptotic role of Opa1.

In sum, the Opa1 lilR3 mutant mouse provides an excellent in vivo model for exploring the cell-specific roles of Opa1. While it remains unclear if any of the identified Opa1 alleles are null alleles, the Opa1 lilR3 allele permits investigation of Opa1 function that depends on its localization to mitochondria. Our data suggest that tissues not typically believed to be affected in ADOA patients could be involved in the disease. There are at least ten mitochondrial proteins implicated in neurogenerative diseases such as Charcot-Marie-Toothe, hereditary spastic paraplegia, Friedrich’s ataxia, and familial Parkinson’s. Defining the mammalian loss-of-function roles of specific Opa1 isoforms could reveal fundamental mechanisms shared by these disorders.

Experimental procedures

Mutant phenotypic and genotypic identification

The lilR3 mutation was generated by ENU mutagenesis as described previously and analyzed in C3H/HeJ mice (Garcia-Garcia et al. 2005; Kasarskis et al. 1998). Genetic mapping of Opa1 lilR3 was performed by linkage analysis of 2,876 opportunities for recombination with SSLP markers (www.informatics.jax.org and http://mouse.ski.mskcc.org). We obtained physical map information from Ensembl (http://www.ensembl.org/Mus_musculus/index.html).

cDNAs of four genes in the lilR3 interval (Opa1, Hes1, ApoD, and Lrrc15) were amplified by RT-PCR (Superscript One-Step RT-PCR, Invitrogen, Carlsbad, CA) using RNA from E10.5 lilR3 embryos. DNAs from the corresponding genomic regions were also amplified. All amplicons were sequenced and, when compared with the C57BL/6 reference sequence (http://www.ensembl.org/Mus_musculus/index.html), an A-to-T mutation was found in intron 19 of Opa1. Subsequent analysis of C57/B6 DNA has verified this mutation. We found no mutations in any of the other sequenced genes in the interval.

Western blotting

E10.5 Opa1 lilR3 embryos were homogenized to obtain the crude mitochondrial fraction as described in Alavi et al. (2007). The total protein concentration was determined through Bradford assay (Bio-Rad, Hercules, CA) following the manufacturer’s recommendations. Equal amounts of total protein were loaded on a gradient 4-20% SDS-PAGE gel (Bio-Rad) and transferred onto a nitrocellulose membrane (Amersham Bioscience, Piscataway, NJ). The membrane was subjected to Western blot analysis with antibodies against Opa1 (BD Pharmingen, Franklin Lakes, NJ), α-tubulin (Sigma, St. Louis, MO), and cytochrome c (BD Pharmingen), followed by immunodetection using an ECL chemiluminescence system (GE Healthcare Biosciences, Piscataway, NJ).

MitoTracker assay

Opa1lilR3/Opa1lilR3, Opa1lilR3/+, and wild-type MEFs were grown on coverslips and treated with MitoTracker Red CM-H2XRos (Invitrogen, M-7513) in DMEM (high glucose, 10% FBS, penicillin/streptomycin) at a final concentration of 250 nM for 45 min at 37°C, 5% CO2. After treatment, cells were washed in DMEM (high glucose, 10% FBS, penicillin/streptomycin) for 1 h at 37°C, 5% CO2. The coverslips were then fixed for 10 min in 4% paraformaldehyde at room temperature, rinsed in PBS for 5 min, and mounted in ProLong Gold antifade reagent (Invitrogen, P36934). Images were captured using the Zeiss LSM 510 META confocal at 63× with and without optical zoom.

Analysis of mutant phenotype

Embryos were dissected in PBS containing 0.4% BSA at different stages as assessed by the presence of vaginal plugs in mothers. Whole-mount RNA in situ hybridization activity was performed as described (Belo et al. 1997; Schaeren-Wiemers and Gerfin-Moser 1993). Probes for Sox10, Pax3, Pax1, Pax9, HoxB5, HoxB6, Wnt1, and Msx1 have been previously reported (Becker et al. 1996; Chalepakis et al. 1991; Goulding et al. 1991; Hakami et al. 2006; Krumlauf et al. 1987; Parr et al. 1993; Peters et al. 1995).

Analysis of cell death

To analyze cell death, freshly dissected embryos were washed two times for 5 min in 1× PBS/0.1% Triton X-100 at room temperature, then incubated in 1× PBS/0.1% Triton X-100 containing 50 nM LysoTracker Red (Invitrogen) for 30 min at room temperature. Embryos could be visualized immediately or fixed in 4% paraformaldehyde overnight at 4°C. The embryos were rinsed in 1× PBS, and labeled cells were visualized with a Leica MZFLIII microscope under UV light using a GFP filter. Images were captured using a Q-imaging Retiga EXi camera.