Distinct functional consequences of ECEL1/DINE missense mutations in the pathogenesis of congenital contracture disorders

In addition to DINE-deficient mice [23] and C760R knock-in mice [24], newly developed G607S and C760G knock-in mice were used in this study (see below for detailed information). Homozygous mutant mice were obtained by intercrossing heterozygous parents. All mutant mice were crossbred with Hb9::eGFP mice [35] to label embryonic motor nerves with GFP. Noon on the day of vaginal plug detection was considered embryonic day 0.5 (E0.5). The genotype of DINE-deficient mice, Hb9::eGFP mice and C760R knock-in mice was determined using allele-specific primers as previously described [24].

DINE knock-in mice with a pathogenic mutation (G607S: replacement of Gly607 with Ser) were generated using the CRISPR/Cas9 system. The plasmid vector pX330 was a gift from Feng Zhang (Addgene plasmid # 42230) [7], expressing single-guide RNA (sgRNA), as well as Cas9. The CRISPR Design Tool (http://​tools.​genome-engineering.​org) was used to design sgRNA in silico. The target sequence and protospacer adjacent motif (PAM) sequence was TCTCTGAACTACGGGGGTATTGG (The PAM sequence is shown in bold letters). The sgRNA and Cas9 mRNA were synthesized in vitro using commercially available kits as previously described [24]. The sgRNA (10 ng/μl) and Cas9 mRNA (10 ng/μl) were injected into C57BL/6 mouse zygotes in the presence of 90 bp single-stranded oligonucleotides (ssODN) (10 ng/μl) using a microinjection system under standard conditions. The ssODN sequence was: CAGTCGTCATAGCCATGGGTCAGTTCGTGCCCAATGATGGTGCtAATACCCCCGTAGTTCAGAGACCTGGGCCCACAGCAGCAGGTTAAA (the mutation site is shown as a lowercase letter). The zygotes were cultured in culture medium at 37 °C in 5% CO₂ up to the two-cell embryo stage and then the embryos were transferred into the oviducts of recipient mice. Tail genomic DNA was amplified using a specific primer set: forward 5′-ACTATCCGTCCTTCCCCTCC-3′ and reverse 5′-AGATCTCTGGGGCCTCTCTG-3′. The PCR product was used to confirm the mouse genotype by sequencing with the forward primer or restriction fragment length polymorphism analyses with Ban I.

Similar methods were used for generation of another DINE knock-in mouse with an artificial mutation (C760G: replacement of Cys760 with Gly) with appropriate modification of the target sequence as well as the ssODN sequence: the target and PAM sequences were CCCAGTTTGAGGAATTCGGCCGG. The ssODN sequence was: CACCCCAGGGTCCTGGGCAGCGTATCCCAGTTTGAGGAATTCGGCCGaGCCTTCCACgGTCCCAAGGACTCTCCCATGAACCCCGTCCAT. Two mutation sites are shown as lowercase letters and the first mutation (a) is a synonymous mutation to inhibit re-cutting by Cas9.

We chose five potential off-target sites (OT1-OT5) of the target sequence for G607S knock-in generation (TCTCTGAACTACGGGGGTATTGG) using the CRISPR design tool (http://​tools.​genome-engineering.​org): OT1 TCTaaGAACTACtGGGGTATGGG, OT2 agTgTGAACTACGGGGGTAaCAG, OT3 TCTtTGAACcACGGGGGgATGAG, OT4 TCTaaGtACTACaGGGGTATAAG, OT5 aCaCTGAAgTACaGGGGTATGAG. Mutation sites and PAM sequences are shown as lowercase letters and bold letters, respectively. The surveyor assay was performed to detect the CRISPR/Cas9-induced mutations, as previously described [24]. Target fragments on the off-target sites were amplified using specific primer sets: OT1 forward 5′-TGTTAACAAAATGGAAATGATTCAA-3′ and OT1 reverse 5′-TCAGAGTTCCATGTGGCAGTA-3′, OT2 forward 5′-TCCTTCTCAGATCCCTTGTCA-3′ and OT2 reverse 5′-TGCCATGGATGTAAATCATCA-3′, OT3 forward 5′-CGGTGGGTGGTGTTTCTTAT-3′ and OT3 reverse 5′-GGTGGCAGGAGTTCCTTCTT-3′, OT4 forward 5′-GGCTGCAGGCAGGTAGTTCT-3′ and OT4 reverse 5′-TCCCAAACAGTTAATGAATCAGTG-3′, OT5 forward 5′-TTCTTCTGGAGTCCCCAATG-3′ and OT5 5′-reverse 5′- GCACAGGTTTTTGGAGGAAA-3′.

E12.5 mouse embryos were collected, fixed in 4% paraformaldehyde (PFA) at 4 °C for 2 h, then immersed in PBS containing 30% sucrose for two additional days. The tissue samples were embedded in optimal cutting temperature (OCT) compound (Sakura Finetek, Torrance, CA, USA), and stored at −80 °C until use. Serial 20 μm sections were cut using a cryostat microtome. Immunohistochemistry was performed as described previously [17], with minor alternations. Briefly, the sections were rinsed three times in PBS, permeabilized by immersion in absolute methanol for 6 min at −30 °C, rinsed in PBS for 30 min, and blocked for 30 min in 0.3% Triton X-100 and 0.2% bovine serum albumin (BSA) in PBS. The sections were then incubated with goat anti-DINE primary antibody (1:500; Santa Cruz Biotechnology, Dallas, TX, USA) in the blocking solution at room temperature overnight. After washing, anti-goat secondary antibody conjugated with Alexa Fluor 546 (1:500; Invitrogen, Carlsbad, CA, USA) was applied for 50 min and then tissues were rinsed three times in PBS. The sections were visualized using confocal microscopy (FV1200; Olympus, Tokyo, Japan).

Whole-mount immunohistochemistry was performed as previously described [24]. Embryonic tissues were incubated with rabbit anti-GFP primary antibody (1:500; Life Technologies, Carlsbad, CA, USA) to detect the motor nerves. For detection of cranial nerves, embryonic heads were incubated with mouse monoclonal antibody 2H3 (1:500; Developmental Studies Hybridoma Bank, University of Iowa, IA, USA). Subsequently, tissues samples were incubated with Alexa Fluor 488 goat anti-rabbit secondary antibody (1:500; Invitrogen) and Alexa Fluor 546 goat anti-mouse secondary antibody (1:500; Invitrogen). After PBS washes, tissue samples were dehydrated through a methanol series (30%, 50%, 80%, and 100% for 30 min each), cleared with benzyl alcohol-benzyl benzoate (BABB), and imaged using confocal microscopy FV1200 (Olympus).

An FV1200 laser-scanning confocal microscope (Olympus) was used to acquire the confocal images. We used 10x and 20x dry objective lenses to visualize motor nerves. Multiple adjacent regions of embryonic mouse limb, head, or individual muscles were imaged using a motorized xy stage module. Image analyses were performed using IMARIS software (Bitplane, Zurich, Switzerland). To evaluate the extent of motor innervation in each individual skeletal muscle, all stacked images were converted into 3D images and the motor nerve terminals were semiautomatically traced using the filament tracer function. The number of terminal points was automatically calculated. The measurement of the nerve length of ocular motor nerves was performed using the measurement point function. To appropriately compare results between samples, all data were processed using the same criteria.

Total RNA was isolated from embryonic spinal cords using the RNeasy mini kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s protocol. For quantification of DINE mRNA expression, total RNA (1 μg aliquots) was converted to cDNA by ReverTra Ace (Toyobo, Osaka, Japan) and an oligo (dT) primer. The resultant cDNA was diluted 1:50 with distilled water. The quantitative PCR (qPCR) procedures were performed as previously described [24]. TaqMan Gene Expression Assays (Applied Biosystems, Waltham, MA, USA) for mouse actin beta (ACTB) (Mm00607939_s1) and ECEL1 (Mm00469610_m1) were used for specific target amplification. Relative mRNA expression was calculated by using the comparative cycle threshold (CT) method, and then normalized to endogenous ACTB mRNA expression for each sample. The CT value was obtained from the amplification plot with the aid of SDS software (Applied Biosystems).

For qualitative analysis of the mutant DINE transcript, total RNA (1 μg aliquots) was converted to cDNA using ReverTra Ace (Toyobo, Osaka, Japan) with random primers. The cDNA was amplified using the following specific primers: forward 5′-CCACCCTGTATGACCCAGAC-3′, reverse 5′-ATAGAGGCGAACGATGCACT-3′.

E17.5 mice spinal cords were homogenized in 50 mM Tris-HCl (pH 7.4) containing protease inhibitor cocktail complete mini (Roche Diagnostics, Indianapolis, IN, USA) (homogenizing buffer). After one centrifugation, the supernatant was centrifuged at 20,000 g for 30 min at 4 °C. The pellet fraction was collected and then solubilized with homogenizing buffer containing 1% Triton X-100 for 30 min at 4 °C and centrifuged again at 20,000 g for 30 min at 4 °C. The supernatants served as protein samples for further analyses.

Protein samples (100 μg) were boiled in Glycoprotein Denaturing Buffer (NEB, Ipswich, MA, USA) for 10 min. For Endoglycosidase H (Endo H) treatment, samples were added to a reaction containing GlycoBuffer 3 (NEB) and endo H (1000 units, NEB) and incubated at 37 °C for 1 h. For PNGase F treatment, samples were added to a reaction containing GlycoBuffer 2 (NEB), 1% Nonidet P-40 (NEB), and PNGase F (1000 units, NEB) and incubated at 37 °C for 1 h.

Protein samples (50 μg) with or without glycosidase treatment were separated by 5–20% gradient sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), followed by transfer to polyvinylidene fluoride (PVDF) membranes. After incubating with 2% enhanced chemiluminescence (ECL) blocking reagent (GE Healthcare, Buckinghamshire, UK), the membranes were incubated with goat anti-DINE antibody (1:500; Santa Cruz Biotechnology) at 4 °C overnight. The membranes were repeatedly washed then incubated with horseradish peroxidase-conjugated anti-goat IgG secondary antibody (1:5000; Vector, Burlingame, CA, USA). Anti-GAPDH antibody (1:5000; Trevigen, Gaithersburg, MD, USA) was used for the control experiments. Each set of experiments was repeated at least three times to confirm results.

Data were first analyzed for normal distribution and equal variance. When normally distributed, two independent samples were statistically analyzed using a two-tailed Student’s t test or Welch’s t test. If the data did not pass normality testing, the Mann-Whitney U test was used. For three independent samples, the data was statistically analyzed using one-way ANOVA for normal distributions or the Kruskal-Wallis test followed by the Steel-Dwass test for non-normal distributions, with p < 0.05 considered significant. All analyses were completed with Statcel 3 (add-in software for Excel, Microsoft, USA).

We previously generated a DINE knock-in mouse line carrying C760R as a relevant model of ECEL1-mutated DA, and detected axonal arborization defects of motor nerves in homozygous C760R mutant limb muscles [24]. However, given that the observed expressivities vary in some affected regions in patients with ECEL1 mutations [2, 30], phenotypic comparison with another knock-in mouse with a distinct pathogenic mutation is necessitated to judge whether axonal arborization defects are a common mechanism in the pathogenesis of ECEL1-mutated DA. Notably, Shaaban et al. have reported two siblings with a missense c.1819G > A mutation (p.G607S) (Fig. 1) in the ECEL1 gene that presented with significant ophthalmoplegia and less pronounced contractures in the distal joints of lower limbs [30]. Because the symptoms did not meet the major criteria for the diagnosis of DA, the authors concluded that the two siblings differed from other patients with different ECEL1 pathogenic mutations. To experimentally compare the pathogenic effects between the C760R and G607S mutations, we have generated a DINE knock-in mouse line carrying G607S using the CRISPR/Cas9 system. We designed a target sequence of sgRNA in the region close to the mutation site, as well as a 90 bp single-stranded DNA (ssDNA) with the pathogenic mutation as the DNA template (Fig. 2a). The CRISPR/Cas9 tools were injected into 200 mouse zygotes and then 158 normally developed two-cell embryos were transferred into recipient female mice. A total of 71 mice were born normally. We performed sequencing analyses using the PCR amplified target region to verify the genotype of the CRISPR-injected mice and successfully obtained 7 F0 mutant mice. We selected two male mice (Founder 1 and Founder 2) with a dominant mutated peak in electropherograms (Fig. 2b) and used these founder mice for expansion of the mouse colony. Off-target analyses using a mismatch cleavage enzyme showed no off-target mutations at five potential sites in the founders (Fig. 2c). The results were also confirmed using direct sequencing analyses (data not shown). After confirmation of the transmission of the desired mutation from the founders to their offspring, we generated G607S mutant mice with motor neuron-specific expression of GFP by crossing with Hb9::eGFP mice [35].

First, we performed whole-mount immunostaining with an anti-GFP antibody in the homozygous G607S mutant diaphragm. Identical to the DINE-deficient [23] and C760R diaphragm [24], homozygous G607S mutant mice displayed impaired axonal arborization of motor nerves in the diaphragm, possibly leading to perinatal lethality (Additional file 1: Figure S1). In fact, we never obtained any homozygous G607S mutant pups by mating heterozygous mutants. Next, we performed the same morphological analyses in hindlimb muscles at E17.5 to detect the morphology of embryonic motor nerves and to compare the morphological phenotypes of motor innervation between C760R and G607S mutant mice. Compared with wild-type hindlimbs, abnormal motor innervation was detected in a number of G607S mutant hindlimb muscles (data not shown). In order to quantify the number of motor nerve terminals, we selected two hindlimb muscles, the gracilis anterior muscle and the rectus femoris muscle, which are severely affected in DINE-deficient embryos and C760R mutant embryos [24]. The number of motor nerve terminals was clearly decreased in the G607S mutant gracilis anterior muscles (46.8 ± 5.0 in G607S mutant embryos, n = 5, vs. 110.6 ± 8.4 in wild-type embryos, n = 5; p = 0.0002, two-tailed Student’s t test) (Fig. 2d–g), as well as in the rectus femoris muscle (46.4 ± 4.9 in G607S mutant embryos, n = 5, vs. 166.4 ± 9.5 in wild-type embryos, n = 5; p = 0.000003, two-tailed Student’s t test) (Fig. 2h–k), similar to the C760R mutant muscles. We also performed the same morphological analyses in foot muscles in order to directly compare the innervation of motor nerves between G607S and C760R mutants. In contrast to the phenotypic discordance in foot muscles between patients with G607S [30] and DA patients with C760R [8], G607S mutant mice displayed the same clear axonal arborization defects in the foot muscle as the C760R mutant (median terminal number 26 in wild-type embryos, n = 5; 6 in G607S mutant embryos, n = 5; 10 in C760R mutant embryos, n = 5; Kruskal-Wallis test followed by the Steel-Dwass test, p < 0.05) (Fig. 3a–d). These results suggested that G607S and C760R mutations similarly affected intramuscular axonal arborization of motor nerves in the hindlimbs.

Next, we explored whether abnormal innervation also occurred in the mutant cranial motor nerves. Given that the ocular phenotype is the most prominent feature of ECEL1-related DA [14] as well as the patients with G607S [30], it is important to assess the function of DINE in cranial motor nerves, especially ocular motor nerves. To this end, whole-mount immunostaining with anti-GFP antibody and anti-neurofilament antibody was performed on embryonic head samples. Among the three ocular motor nerves (oculomotor, trochlear and abducens nerves) innervating extraocular muscles, only the abducens nerves were fluorescently labeled in Hb9::eGFP mice (Fig. 4a–c), consistent with previous results [12]. Both oculomotor nerves and trochlear nerves were visualized only with anti-neurofilament antibody (Fig. 4a–c). At E11.5, all wild-type ocular motor nerves correctly chose their pathway toward the target muscles around the eye (Fig. 4a, d). In contrast, the abducens nerves stopped extending their axons toward the target muscles in a number of homozygous G607S mutant embryos (Fig. 4b, e). The oculomotor nerves and trochlear nerves showed a normal trajectory to the target muscles in these mutant embryos (Fig. 4b, e). Interestingly, the axon guidance defects of the abducens nerves were similarly observed in homozygous C760R mutants (Fig. 4c, f). Subsequent quantitative analyses revealed that a drastic reduction of nerve length could be detected in a substantial number of homozygous mutant abducens nerves (median nerve length 855.4 μm in wild-type embryos, n = 21; 685.2 μm in G607S mutant embryos, n = 16; 642.4 μm in C760R mutant embryos, n = 12; Kruskal-Wallis test followed by the Steel-Dwass test, wild-type vs. G607S mutant, p < 0.01) (Fig. 4g). In contrast, such a reduction was not observed in the homozygous mutant oculomotor nerve, although the difference in nerve length between wild-types and the homozygous G607S mutants reached statistical significance (median nerve length 1280 μm in wild-type embryos, n = 22; 1216 μm in G607S mutant embryos, n = 20; 1198 μm in C760R mutant embryos, n = 12; Kruskal-Wallis test followed by the Steel-Dwass test, wild-type vs. G607S mutant, p < 0.05) (Fig. 4h). The length of the trochlear nerve was not significantly different between wild-type and homozygous mutant embryos (median nerve length 1607 μm in wild-type embryos, n = 22; 1479 μm in G607S mutant embryos, n = 20; 1469 μm in C760R mutant embryos, n = 12; one-way ANOVA, p > 0.05) (Fig. 4i). To exclude the possibility that the observed phenotype in some mutant embryos was just due to developmental delay, we performed the same morphological analyses in abducens nerves at a later stage. At E12.5, all wild-type abducens nerves reached their target muscles. While some of the homozygous G607S mutant abducens nerves normally innervated the extraocular muscles (Fig. 4j), substantial numbers of abducens nerves followed the wrong pathway in some homozygous mutant embryos (‘wandering’ phenotype; Fig. 4l). In addition, a number of mutant abducens nerves stopped extending their axons toward their target muscles without any correct innervation (‘stalled’ phenotype; Fig. 4n). These axon guidance defects were also observed in homozygous C760R mutant embryos with variable penetrance (Fig. 4k–o). Quantitative analyses with a substantial number of embryos (n = 78) clearly revealed that abnormal axon guidance could be detected in over half of the homozygous mutants (Fig. 4p). These results indicated that both G607S and C760R mutations frequently led to axon guidance defects in abducens nerves.

In contrast to nonsense mutations, which lead to loss of full-length protein, the biological interpretation of a missense mutation with only one amino acid exchange is often challenging. In order to characterize the consequences of G607S on DINE function, we first performed immunohistochemical analyses of G607S mutant embryonic spinal cords with an anti-DINE antibody. In contrast to wild-type samples (Fig. 5a–c), only a faint positive signal could be detected in homozygous G607S mutant spinal cords (Fig. 5d–f), similar to DINE-deficient spinal cords (Fig. 5g-i). The immunohistochemical data was subsequently confirmed using western blotting analysis. DINE protein expression levels were significantly decreased in homozygous mutant spinal cords (0.098 ± 0.012 in G607S mutant embryos, n = 4 vs. 0.87 ± 0.039 in wild-type embryos, n = 4; p = 0.000002, two-tailed Student’s t test) (Fig. 5j, k). We hypothesized that the missense mutation could lead to reduced protein expression via the response against formation of misfolded proteins. To validate the possibility, we further performed quantitative real time PCR in order to examine whether DINE expression was maintained at a transcriptional level. Unexpectedly, the mRNA expression level was significantly reduced in homozygous mutant samples compared with wild-type samples, similar to the protein level (0.003 ± 0.001 in DINE-deficient embryos, n = 3 vs. 0.023 ± 0.002 in wild-type embryos, n = 3; p = 0.002, two-tailed Student’s t test) (Fig. 5l). To further explore the mechanism underlying the reduced mRNA expression, DINE transcripts containing exon 13 with the pathogenic mutation site were reverse transcribed and then amplified using a forward primer on exon 12 and a reverse primer on exon 13 (Fig. 5m). The mature mRNA product was clearly observed with a faint pre-mRNA product in wild-type samples (Fig. 5n). In contrast, in homozygous mutant samples, a strong pre-mRNA product could be detected in addition to the mature mRNA, which was of negligible intensity (Fig. 5n). We also detected a similar tendency in heterozygous adult G607S mutant hypothalamus (Additional file 2: Figure S2), although the band intensity of the pre-mRNA product was much weaker than that in the homozygous mutant spinal cords. We subsequently performed sequencing analyses on the TA cloned RT-PCR products and confirmed that the homozygous mutant transcript contained additional sequences of intron 12 and intron 13 (Fig. 5o), indicating that splicing defects occurred in the mutant samples. Importantly, due to the second to fourth nucleotide sequences of intron 12, the aberrant transcript with intron retention contained TAA, a nucleotide sequence corresponding to a translation termination codon, on the reading frame (Fig. 5o). As it is widely accepted that an in-frame premature translation termination codon induces an mRNA degradation pathway termed nonsense-mediated decay [18], these results provide the first evidence that splicing defect-induced mRNA degradation could be the primary functional consequence of the G607S mutation.

Finally, we explored the functional consequence of the pathogenic C760R missense mutation. Given that the homologous residue to cysteine 760 in the DINE protein has been shown to form a disulfide bond in a family member protein [19], we hypothesized that this residue plays crucial roles in the appropriate protein conformation of DINE. To test the possibility, we first examined whether the posttranslational modifications of mutant DINE protein had occurred appropriately. Previous cell culture experiments showed that most DINE protein is Endoglycosidase H (endo H)-sensitive and retained in the endoplasmic reticulum (ER) but some DINE protein is endo H-resistant and plasma membrane-associated [4]. Consistent with the previous cell culture study, we could discriminate some endo H-resistant DINE protein from the endo H-sensitive in wild-type spinal cords (Additional file 3: Figure S3). In contrast, all mutant DINE protein was endo H-sensitive (Additional file 3: Figure S3). These results suggested that the mutant DINE protein was not appropriately glycosylated, which may result in abnormal protein conformation. We could not detect any different patterns in samples digested with another glycosidase PNGase F, which cleaves glycosylation added in both the ER and the golgi complex (Additional file 3: Figure S3). Next, we performed immunohistochemical analyses with an anti-DINE antibody to gain further information about the localization of the mutant DINE protein. In wild-type E12.5 embryos, DINE was expressed in motor neuron soma and axons in ventral spinal cords and all DINE positive signals were colocalized with GFP positive motor neurons (Fig. 6a–c). In contrast, DINE expression was drastically decreased in homozygous C760R mutant motor axons, which were clearly visualized with GFP signals, but not in the soma (Fig. 6d–f). We also explored whether mislocalization of the mutant protein occurred at distal parts of motor nerves. Whereas DINE expression was colocalized with GFP positive motor axons in wild-type diaphragm muscle (Fig. 6j–l), such positive immunoreactivity could not be detected in mutant diaphragm muscle (Fig. 6m–o).

The homologous residue in a family member protein to cysteine 760 in the DINE protein has been shown to form a disulfide bond [19], but there is no direct evidence showing that cysteine 760 of DINE actually forms a disulfide bond. Thus, it remains unclear whether the pathogenic effect of C760R results from the lack of a disulfide bond or an arginine-induced conformational change. To further discriminate the pathogenic possibilities, we generated a new DINE knock-in mouse with an artificial substitution of cysteine 760 with glycine, another small non-charged amino acid. This C760G mutant produced identical immunohistochemical results to the C760R mutant (Fig. 6g–i, p–r). These results indicate that cysteine 760 contributes to the appropriate conformation of DINE protein, possibly via a disulfide bond, and this conformational change possibly promotes the axonal transport of DINE.