A -Acting Protein Effect Causes Severe Eye Malformation in the Mouse

Summary

Mp is an irradiation-induced mouse mutation associated with microphthalmia, micropinna and hind limb syndactyly. We show that Mp is caused by a 660 kb balanced inversion on chromosome 18 producing reciprocal 3-prime gene fusion events involving Fbn2 and Isoc1. The Isoc1-Fbn2 fusion gene (Isoc1^Mp) mRNA has a frameshift and early stop codon resulting in nonsense mediated decay. Homozygous deletions of Isoc1 do not support a significant developmental role for this gene. The Fbn2-Isoc1 fusion gene (Fbn2^Mp) predicted protein consists of the N-terminal Fibrillin-2 (amino acids 1–2646, exons 1–62) lacking the C-terminal furin-cleavage site with a short out-of-frame extension encoded by the final exon of Isoc1. The Mp limb phenotype is consistent with that reported in Fbn2 null embryos. However, severe eye malformations, a defining feature of Mp, are not seen in Fbn2 null animals. Fibrillin-2^Mp forms large fibrillar structures within the rough endoplasmic reticulum (rER) associated with an unfolded protein response and quantitative mass spectrometry shows a generalised defect in protein secretion in conditioned media from mutant cells. In the embryonic eye Fbn2 is expressed within the peripheral ciliary margin (CM). Mp embryos show reduced canonical Wnt-signalling in the CM – known to be essential for ciliary body development - and show subsequent aplasia of CM-derived structures. We propose that the Mp “worse-than-null” eye phenotype plausibly results from a failure in normal trafficking of proteins that are co-expressed with Fbn2 within the CM. The prediction of similar trans-acting protein effects will be an important challenge in the medical interpretation of human mutations from whole exome sequencing.

Introduction

The accurate prediction of the phenotypic consequences of individual mutations is of increasing medical importance with the rapid development of whole genome sequencing technologies. Different mutations can have distinct effects on expression of the gene, its function, or localisation of its product. It is therefore not surprising that allelic heterogeneity can result in multiple distinct disorders from mutations in a single gene (e.g. LMNA [1], FLNA [2], FGFR1 [3]). Such heterogeneity, in combination with stochastic effects and genetic and/or environmental modifiers, also accounts for phenotypic variability within an individual disorder.

Particular mutations cause disease due to a failure of normal post-translational processing and protein folding. For example, the mutation-specific induction of endoplasmic reticulum (ER) stress responses seen in disorders including late onset neurodegenerative diseases [4], congenital hearing loss [5] and skeletal dysplasias [6]. The mechanisms proposed for pathogeneisis associated with ER stress include: loss of function of the mutant protein [7], alteration of the cell specific differentiation state [6] or uncharacterised generalised cytopathy [4].

Anophthalmia (absence of the eye) and microphthalmia (small eye) are important causes of congenital visual impairments in developed countries with a live birth prevalence rate of 0.2 and 1.7 per 10,000 live births respectively [8], [9]. In most cases the cause of anophthalmia/microphthalmia remains unknown [10] although several single gene causes have been discovered including SOX2 [11], [12], OTX2 [13], STRA6 [14], FOXE3 [15], SMOC1 [16], [17], [18] and PAX6 [19]. Of these, heterozygous loss-of-function mutations in SOX2 are the most common, accounting for 20–40% of bilaterally affected cases [20]. SOX2 functions at multiple stages during eye development, including during lens induction [21], [22] and formation and maintenance of the ciliary margin (CM) at the distal peripheral rim of the optic cup [23]. The CM forms the ciliary epithelium and the inner layer of the iris and controls ciliary muscle and stroma development [24]. At the boundary of the CM and the neural retina (NR), multipotent cells make a binary decision to commit to either CM or NR fate [23]. SOX2 expression, which marks the boundary between CM and NR, is dependent on canonical Wnt-signaling [25], [26]. Loss of either Wnt-signalling or SOX2 expression in the CM leads to failure of the development of the ciliary body structures and thinning and rosetting of the neural retina [23], [27], [28].

The Mp mouse was generated through the irradiation mutagenesis programme at the Oak Ridge National Laboratory in the 1960's. Mp homozygotes appeared to be anophthalmic and to have syndactylyyly of the hindlimbs [29]. Here, we identify the cause of Mp as a balanced 660 kb inversion on chromosome 18. The consequent gene fusion events result in the production of a C-terminally truncated fibrillin-2 protein that is retained in the rough-ER(rER) of expressing cells. Null mutations of Fbn2 accurately phenocopy the Mp limb anomalies but are not associated with any ocular malformation [30], [31], [32]. Fbn2^Mp inclusions trigger the unfolded protein response (UPR) in a subset of cells within the CM resulting in aplasia of the ciliary apparatus with thinning and rosetting of the neural retina. The UPR seen in Mp is associated with signalling and patterning anomalies and a reduction in general protein secretion in Fbn2^Mp expressing cells. This mechanism has broader significance in the interpretation of human disease-causing mutations. It suggests that the phenotypic effect of specific UPR-inducing mutations may not result from loss of that gene product but rather from loss of co-expressed gene products that are also processed through the ER.

Results

Main adult phenotypes of Mp

We bred Mp onto both the C57BL/6J inbred, and the CD1 outbred background strains. The phenotype in these mice was consistent with that documented in the previous report ([29]; Figure 1A–H), with small eyes and ears in adult heterozygotes (Figure 1B; Mp/+) and small ears, apparent anophthalmia and hindlimb syndactyly in homozygotes (Mp/Mp; Figure 1C). The eyelids in Mp/Mp animals never opened, however contrary to the original report, microphthalmia rather than anophthalmia was identified on dissection (Figure 1D). Further analyses revealed that the Mp/Mp limb phenotype ranged in severity from osseous fusions of the entire phalanges of digits 2–3–4 (Figure 1H), to simple soft tissue syndactyly affecting these digits (not shown). Digits 1 and 5 were never affected and no metatarsal abnormalities were identified. Consistent with the original report in which homozygotes were described as runted, we observed significant weight differences between mutant and Wt animals (Figure S1). Mp mutant phenotypes were fully penetrant throughout crosses of all genetic backgrounds tested (Figure S1).

Ciliary body malformation and aberrant developmental Wnt signalling in the Mp eye

Histological analysis at adult stages (P21) revealed pan-ocular structural defects in Mp (Figure 2A–C). In particular, the neural retina cell layers displayed severe rosetting and the vitreous was absent from the anterior chamber and posterior eye. Immunofluorescence studies revealed the nature of the retinal disruption with rosettes composed primarily of disorganised rod photoreceptors, and reduced numbers of cells at the inner nuclear layer (Figure S2). Furthermore, ectopic ganglion cells were identified in multiple regions of the Mp/Mp retina and expression of GFAP was observed, indicative of gross damage to the retina. In both Mp/M and Mp/+, the ciliary body was consistently absent however the iris appeared to form normally (Figure 2D). Although retinal lamination is not fully complete until approximately three weeks after birth, the ciliary body develops during mid gestation from non-pigmented ciliary epithelial cells. The anterior region of the early developing retina, and future ciliary epithelium, is marked specifically by canonical Wnt signalling and we therefore crossed Mp to BAT-gal [33], a Wnt-reporter strain that expresses beta-galactosidase in the presence of activated beta-catenin. In Mp/Mp-BAT-gal^+/− E14.5 mice, we detected significantly reduced canonical Wnt signalling in the anterior retina as shown by the loss of positively stained cells (Figure 2E–F, and Figure S2). Furthermore, at E15.5 a marked reduction in the size of the developing non-pigmented ciliary was observed upon histological staining as well as staining for the anterior retina marker Pax6 (Figure 2G–J). Furthermore, we revealed reduced total retinal size in Mp/Mp eyes at the same developmental stage (Figure S2). We therefore concluded that the Mp eye phenotype resulted from perturbations to normal ocular development, and first identifiable at the region of the developing ciliary epithelium.

Genetic mapping of Mp to the Fbn2 locus

We performed a whole-genome scan using microsatellite markers on heterozygous Mp mice backcrossed to C57BL/6J, and identified enrichment for background strain C3H-specific alleles co-segregating with Mp at three consecutive markers on chromosome 18 (data not shown). During fine mapping, no recombination events between markers D18Mit74 and D18Mit184 were found in 43 Mp/+ animals tested (Figure 3A). The Mp mutation was thus likely to be located between approximately 53 Mb and 67 Mb on chromosome 18. Fbn2 (RefSeq Gene NM_010181), an extracellular matrix protein, maps within this interval. The hind-limb phenotype of Mp was consistent with previously published loss of function Fbn2 mutations [30], [31], [32], [34]. Human mutations in FBN2 are associated with crumpled ear helices in affected patients [35], [36], [37] and comparable ear phenotypes were observed in Mp. However, heterozygous FBN2 mutations in humans do not cause ocular malformations and no ocular phenotype has been reported for Fibrillin-2 null mice [30], [31], [32], [34]. Furthermore, our histological examination of adult and neonatal Fbn2-null eyes revealed no ocular phenotype on genetic backgrounds analogous to Mp (Figure S3).

Nevertheless, we considered Fbn2 to be a good candidate for Mp. Standard RT-PCR was unable to amplify a 3′- region of the Fbn2 cDNA (Figure 3B). Using 3′-RACE we identified a fragment present in Mp/Mp but absent from Wt mRNA, and a corresponding absence of the Wt Fbn2 amplicon in Mp/Mp (Figure S3). Cloning and sequence analysis of the Mp-specific Fbn2 amplicon identified that Fbn2^Mp was transcribed up to and including all of exon 62, following which the mRNA contained the full sequence of the terminal exon from Isoc-1 (RefSeq ID NM_025478) (Figure 3C), a 5-exon gene of undetermined function located <1 Mb from Fbn2 on chromosome 18. Using combinations of Fbn2 and Isoc1 specific primers we discovered that this fusion was reciprocal and balanced so that the terminal Fbn2 exons (exons 63–65) replaced the final exon (exon 5) of the Isoc1 transcript (Figure S3), creating a 7-exon mRNA. Sequencing genomic DNA confirmed this to be a balanced inversion of chromosome 18 [NCBI37/mm9: Chr18:58,178,418–58,835,146] (Figure 3D). The predicted consequence for the translated Fibrillin-2^Mp protein was replacement of the C-terminal 261 amino acids (aa) from position 2646, with an 11 aa out-of-frame extension followed by a termination codon, encoded by the out-of-frame final exon of Isoc1 (Figure 3E). Fibrillin-2^Mp lacks the evolutionarily conserved predicted furin cleavage site present in the full-length protein (Fbn2 2769–73 aa; RRKR) [38].

In contrast, the inversion predicted nonsense-mediated decay of the Isoc1^Mp transcript due to the frameshift-induced stop codon in the 5th of the 7 exons of this fusion gene [39], [40]. Consistent with this we observed approximately 50% reduction in Isoc1^Mp mRNA levels by quantitative RT-PCR from developing eye tissue (Figure S3). The Slc27a6 gene, encoding a member of the fatty-acid transport family protein, is located between Fbn2 and Isoc1 and its orientation was changed by virtue of its position at the centre of the Mp genomic rearrangement, possibly affecting its cis-regulation. However the gene itself was not structurally altered and we observed no change in the low level of Slc27a6 expression that could be detected in Wt and Mp samples by RT-PCR (Figure S3). Additionally, signal was not detectable for Slc27a6 or Isoc1 at any stage when analysed by whole mount in situ hybridisation (WISH) in Wt and mutant samples (Figure S4). Although WISH is not sensitive enough provide conclusive proof, this data suggests that the developmental expression of these genes was not significantly changed or ectopically upregulated in Mp embryos. However, we could exclude loss of low-level expression of either of these genes as a cause of the ocular phenotype based on previous genetic studies. A chromosomal deletion causes the shaker-with-syndactylism (sy) phenotype extends at least between the microsatellite markers D18Mit124 at 57.61 Mb (NCBI37/mm9 C57BL/6J) and D18Mit205 at 58.95 Mb, encompassing Isoc1 (58.82–58.84 Mb) and Slc27a6 (58.72–58.77 Mb) as well as Slc12a2 and Fbn2 [34]. Mice heterozygous for sy have no morphological phenotype, and homozygotes have fused digits due to loss of Fbn2 and are deaf because they lack Slc12a2 [41]. The truncation of Fibrillin-2^Mp thus remained the sole candidate and we therefore focussed our investigations on the Fbn2^Mp gene product.

Fibrillin-2 distribution was disturbed in Mp

We performed whole-mount and tissue section in situ hybridisation to determine the embryonic expression patterns of Fbn2 Wt and Mp mouse embryos. Fbn2 showed site and stage specific expression in the developing eye, limb and tail in early embryos and gene expression patterns did not differ between Wt and Mp in any tissue or at stage analysed (Figure 4A & B; and Figure S4). Expression of Fbn2 has recently been well documented in the developing mouse eye, with strong expression observed in the developing non-pigmented ciliary epithelium at the anterior retina, with expression also observed in cells of the cornea and the periocular mesenchyme [42]. Similarly, we observed Fbn2 expression in our analysis of Wt and Mp/Mp eyes at E13.5, with no differences observed between genotypes (Figure 4C–D). Consistent with these sites of expression, using the Fibrillin-2 specific polyclonal pAb868 antibody, we observed Wt Fibrillin-2 protein in extracellular regions of the periocular mesenchyme and in the anterior region of the eye between the neural retina and lens (Figure 4E). In contrast, Fibrillin-2^Mp was absent from these extracellular areas and instead was observed within cells of the anterior eye (Figure 4F), with the strongest signal detected in the non-pigmented ciliary region. Fibrillin-2^Mp was also observed in the retinal pigmented epithelium (RPE) and in the corneal mesenchyme (consistent with Fbn2 mRNA distribution). We then looked at Fibrillin-2 distribution in other tissues known to express Fbn2 and revealed similarly marked differences in protein localisation between Wt and Mp (Figure S4). In all samples analysed (mouse embryonic fibroblasts (MEFs); femoral articular cartilage and skin), Wt Fibrillin-2 protein was detectable in typical extracellular microfibril structures, but mutant protein was confined to intracellular foci adjacent to cell nuclei and was not observed in extracellular regions.

Fibrillin-2^Mp formed stable fibrils within the endoplasmic reticulum

To further identify the consequence of the Mp truncation to Fibrillin-2 distribution, and to explore possible consequences to expressing cells, we performed transmission electron microscopy (TEM) of eye scleral cells on mutant samples. These revealed inclusions located within the enlarged rough endoplasmic reticulum (rER) lumen that were organised into thick fibril-like aggregates, with apparent structural periodicity (Figure 5A–B). The inclusions were confirmed as Fibrillin-2 positive by immuno-EM (Figure 5C) and by immunofluorescence (Figure 5D). Deposits of Fibrillin-1-containing fibrils with similar ultrastructural appearance to the Fibrillin-2^Mp inclusions have been previously identified in extracellular normal human cartilage matrix [43], however the inclusions in our study did not contain Fibrillin-1 when tested by immunofluorescence (Figure S4), or by immuno-EM (not shown), and were exclusively intracellular. Nevertheless, both these results suggest that under some physiological conditions, Fibrillins can produce atypically banded fibrils. Immunofluorescence confirmed that the Fibrillin-2^Mp inclusions were ER-specific, and did not localise within the golgi (figure 5E–G).

Fibrillin-2^Mp inclusions triggered ER-stress

We hypothesised that the presence of large inclusions could disrupt normal ER mechanisms and secretory function. First, we analysed expression of Xbp1, a widely used marker ER-stress or the unfolded protein response (UPR)[44]. Specifically, a quantitative assay was devised to measure the ratios of unspliced versus UPR-specific Xbp1 transcripts. Validation was performed with cells cultured with increasing concentrations of tunicamycin (not shown). We then assayed Mp/Mp and Wt RNA samples from MEFs and from dissected E13.5 and E16.5 embryonic eye tissues, and found that relative levels of Xbp1s were significantly increased in all Mp/Mp samples compared to Wt (Figure 5H). We next analysed embryonic ocular expression of the UPR chaperone protein Hspa5, and observed expression in Mp, but not in Wt, specifically in the ciliary epithelium (Figure 5I–J). Immunostaining for Protein Disulphide Isomerase (PDI), revealed that Mp embryos had enhanced staining in the non-pigmented ciliary epithelium and adjoining RPE, co-localising with the Fibrillin-2^Mp inclusion-positive cells, whereas equivalent Wt cells were negative for PDI expression (Figure S5). In combination, these data confirmed ER-stress in developing Mp eyes, in regions consistent with the distribution of Fibrillin-2^Mp inclusions at the anterior developing ciliary margin.

We then investigated whether the presence of Fibrillin-2^Mp inclusions resulted in an increase in apoptosis in the developing Mp eye. We reasoned this could provide a straightforward explanation for the loss of retinal and ciliary epithelial cells observed in Mp. Immunostaining for activated Caspase-3 revealed increased numbers of apoptotic cells in the embryonic mutant retinas compared to Wt (Figure 5K). The positive cells were evenly dispersed throughout the entire neural retina, but surprisingly these were not enriched in the areas with the highest number of Fibrillin-2^Mp-positive cells, i.e. the non-pigmented ciliary epithelium and adjacent RPE (Figure S5). This suggested that additional or alternative pathological mechanisms account for the disruptions to the developing anterior eye in Mp.

Protein secretion defects in UPR-affected cells

We then considered whether the presence of Fibrillin-2^Mp inclusions and the consequential induction of UPR prevented the secretion of other extra-cellular proteins expressed in the same cells. We took multiple independent primary Wt and Mp MEF cultures and analysed their conditioned media using quantitative label-free mass-spectrometry. We found reduction in the secretion of many proteins, of which the most severely affected were collagens (Col6a1; Col3a1; Col1a2; Col1a1; and Col12a1), Periostin and Follistatin-like 1 (Figure 6A). Confirmation of the MS data was performed by immunoblotting for one of the identified enriched proteins, Col6a1, which revealed elevated intracellular protein levels in Mp (Figure 6B). Albumin supplemented into the culture media was detected at equivalent levels by these analyses, and Coomassie-stained SDS-PAGE gels of total secreted proteins were used as a further experimental control. To ensure that transcriptional down-regulation was not responsible for the differences seen in protein levels, we assessed the relative mRNA levels of the affected proteins and found the expression of all the genes to be equivalent to Wt (Figure 6C). As a complementary assay, we transfected Wt human RPE1 cells with a plasmid encoding a tagged secreted protein (Wnt3A-FLAG) and added tunicamycin (Tn) to the cell culture medium to induce the UPR. Increased levels of Hspa5 in cell lysates confirmed that the Tn-treated cells were undergoing the UPR, and we saw a coincident reduction to secreted Wnt-FLAG in these cells compared to similarly transfected cells without Tn (Figure S6). We also observed a reduction in the mass of the migrating Wnt-FLAG protein, consistent with reduced glycosylation in the Tn-treated cells. Thus, Fibrillin-2^Mp inclusions have a negative effect on protein secretion, and this is a likely consequence of ER-stress.

Discussion

In humans, mutations in the FIBRILLIN gene family result in fibrillinopathies, diseases that display a broad phenotypic spectrum consistent with their developmental expression patterns [45], [46]. Autosomal dominant (AD), and less commonly autosomal recessive, mutations in FBN1 result in the skeletal, cardiovascular and ocular features of Marfan syndrome (MIM #154700), whereas AD FBN2 mutations typically affect skeletal regions in arthrogryposis, in which patients usually present with long digits, distal joint contractures and crumpled ears, but rarely with ocular or cardiovascular complications. Haploinsufficiency for FBN1 appears to be the mechanism responsible for the majority of cases of Marfan syndrome [47]. However, dominant-negative effects appear to be responsible for the very severe phenotypes associated with neonatal and infantile Marfan syndrome. In these cases the causative mutations cluster within exons 24–32 [48]. The cognate region of FBN2 is the most commonly mutated in Beal syndrome and would plausibly be associated with a similar dominant negative effect [49]. An interesting phenotype of neonatal progeroid features, Marfan syndrome and generalised lipodytrophy (Wiedemann-Rautenstrauch syndrome) has recently been described in association with FBN1 mutations located in the C-terminal domain of the protein [50], [51], [52]. The pathogenic mechanism is not yet clear but it will be important to determine if the mutant protein is normally secreted.

Mp displayed phenotypic overlap with the human fibrillinopathies, including contractures, small ears and skeletal malformations in the limbs. Studies in mice have shown that the Fibrillins are differentially expressed through embryonic development, with Fbn2 transcripts appearing first and coinciding with early morphogenesis and elastic fibre assembly. The overlapping, but delayed expression of Fbn1 is consistent with a more structural role after defined organ structures have been established. Although both Fibrillin-1 and Fibrillin-2 are present during embryogenesis, neither is absolutely necessary, as microfibrils and elastic fibres are formed in the absence of either protein. To date, five separate loss of function mouse mutations in Fbn2 have been described, all with highly penetrant homozygous hind-limb syndactyly but no ocular phenotype [30], [31], [32]. Thus Mp is likely a Fibrillin-2 null mutation in limb development but its ocular phenotype clearly resulted from a separate mechanism, specific to its genomic inversion.

It cannot be ruled out that the extraneous amino acids added by the out-of-frame fusion to Isoc1 conferred a gain-of-function to this chimeric protein, preventing its secretion. Although no pro-protein cleavage by furin has previously been shown directly for Fibrillin-2 at the C-terminal domain, the inclusion data presented here, together with the high degree of conservation of furin-cleavage motifs among all Fibrillins [53], strongly suggests that this processing is common to all Fibrillins but is lost in Fibrillin-2^Mp. It seems likely then that the failure of proprotein processing is the cause of the intracellular accumulations, and it is the failure of secretion of Fibrillin-2^Mp that is the cause of the UPR.

In addition to the primary secretion failure, ER-stress is likely to reduce the protein synthesis and/or secretory capacity of affected cells [54]. We observed a significant reduction in several secreted proteins from Mp cultured primary cells. The cells in the developing Mp eye that contain the highest number of Fibrillin-2^Mp inclusions and display a UPR are those at the anterior rim of the optic cup in the non-pigmented ciliary epithelium. These cells produce various secreted factors including collagen proteins required for the accumulation and maintenance of ocular vitreous [55], [56]. In this context it is interesting that the quantitative MS analysis of the Mp versus Wt MEFs identified several collagen molecules among the most altered of the secreted proteins. The reduction of posterior vitreous chamber is an early and prominent feature of the developmental pathology in the Mp eye. While this feature could be due to failure of collagen secretion, is it also plausibly the consequence of a failure in morphogenesis of the ciliary apparatus itself. Cells in the Fbn2-expressing region of the embryonic eye also secrete Wnt2b, which is of interest in the context of the reduction to BAT-gal reporter activity that we observed in Mp in this region. Disruption of Wnt2b production results in pan-ocular phenotypes overlapping with those observed in Mp, including retinal rosettes, lens abnormalities, iris hypoplasia and the absence of vitreous [27], [57]. Wnt2b from the anterior retina has also been shown to confer layer-organising properties to the central retina [58]. In Mp, a reduction of Wnt2b from the ciliary body may be predicted as a result of UPR-mediated apoptosis (i.e. fewer cells to produce), but may also result from reduction to the normal secretion of Wnt2b through the ER of Fbn2 expressing cells, thus reducing the amount of bioactive paracrine signal available to the developing eye.

One question that arises from this study is why are developing eyes the only tissues that are sensitive to Fibrillin-2^Mp ER-stress, and not other expressing cells? If we trust the phenotype as an assay both in the Fbn2 null and Mp animals, this would suggest that Fibrillin-2 has few non- redundant roles (digit development) and perhaps in Mp only the developing retinal cells that normally secrete Fbn2 are co-expressing gene products that are also critical for development. Skin tissue appeared surprisingly lax compared to Wt (data not shown), and animals were runted. These data suggest there may also be pathogenic consequences to other tissues, but that these do not manifest as profoundly as in the developing eyes, indicating either elevated sensitivity to ER-stress, or intolerance to reductions of co-secreted proteins in the developing eye.

It is possible that there are other mechanisms mediating the protein secretion abnormalities observed in Mp in addition to the ER stress. One such mechanism could relate to the capacity for the ER-localised Fibrillin-2^Mp to bind other proteins. Mutations in COMP cause human pseudoachondroplasia (PSACH; MIM#177170) and multiple epiphyseal dysplasia (EDM1; MIM#132400), diseases of short stature and osteoarthiritis, both of which are associated with retention of mutant COMP within the rER. The mutations cause a variety of plausibly pathological effects including: a decrease in cellular viability, failure of COMP to be secreted into the extracellular matrix, and a trans-acting effect on co-expressed proteins [59], [60], [61]. Mutant COMP accumulates within the enlarged rER of in vivo cartilage and associates with co-expressed type IX collagen, with the secretion and extracellular localisation of these factors correspondingly affected. However, recent analyses of PSACH mouse models have shown that these effects are not mediated via UPR but rather through a novel form of chondrocyte stress [62]. The best-known endogenous Fibrillin-2 binding partners are the family of latent TGFbeta binding proteins (LTBPs). In this regard, the observed reduction in secretion of the modulator of BMP signalling factor, Follistatin-like 1, may be interesting, although there is currently no direct evidence that this protein binds to either LTBPs or Fibrillin-2. Interestingly Fibrillin-1 is also a binding partner of Fibrillin-2 and is similarly processed through the ER, but we were not able to find any evidence of Fibrillin-1 colocalisation to the Mp inclusions (see Figure S4). However, we did observe increased intracellular retention of Col6a1, suggesting that intracellular association of Collagens to Fibrillin-2^Mp may occur. Future work will be required to determine whether collagens, or Wnt2b, directly associate with Fibrillin-2^Mp inclusions in the developing eye.

We have shown here that the ocular phenotypes in Mp are caused by a truncating mutation in Fbn2. Perturbation of normal ER function, via mutations similar to Mp in proteins that functionally pass through the ER, may lead to abnormal phenotypes that cannot be predicted from knowledge of the primary mutant allele. The terms antimorphic, and the more commonly used dominant-negative, are used to describe a worse than heterozygous effect on the function of the protein encoded by the mutated gene or on the cognate functional pathway. Here we propose that the Fbn2^Mp mutation is disrupting one or more developmental functions in which the gene product itself plays no direct part. The compromised gene products are simply expressed in the same cell and share a secretory pathway. We suggest the term synodiporic (fellow traveller) effect for this phenomenon. Thus, the Mp mouse provides clues towards a genetic mechanism of activating disease mutation, which may be phenotypically more severe than those predicted simply by loss of protein function, or may elicit an entirely novel phenotype due to ER perturbation or any other mechanism affecting the production or availability of secondary co-expressed proteins. This has wide-ranging consequences for disease-gene mapping, and for disease management and therapy in human conditions.

Materials and Methods

Ethics statement

The Mp mouse was re-derived from frozen sperm obtained from the Jackson Lab (http://www.jax.org/mmdb/results/Sp_Mp.htmlx). All work was carried out under UK Home Office Project License 60/3785 (IJ Jackson, MRC Human Genetics Unit).

Rederivation and phenotyping of Mp

Heterozygote (Mp/+) sperm was used to perform intra-cytoplasmic sperm injection using oocytes from a donor female F1 (CBA x C57BL/6J). Phenotypically affected offspring were crossed to C57BL/6J to facilitate genetic mapping and to CD1 for phenotype and penetrance analysis. The Mp mutation phenotype has been described [29]. Heterozygote animals were identified by small eyes with/or without reduced pinnae size. Apparent anophthalmia (demonstrated by failure to open eyelids) in combination with distal limb syndactyly was used to identify homozygote animals. Records of each phenotype were accumulated for penetrance analysis in crosses to both C57BL/6J and CD1 strains.

Skeletal analysis

Adult animals were culled with lethal intravenous doses of Euthatal (Merial Animal Health Ltd. Essex, UK), treated in 95% ethanol for 24 hours; then 72 hours in 100% acetone; a minimum of 3 days in stain solution (1 part 0.3% alcian blue in 70% ethanol; 1 part 0.1% alazarin red in 95% ethanol; and 1 part acetic acid, in 17 parts 70% ethanol); 3 days in 1% potassium hydroxide (KOH); 3 days in 1% KOH/30% glycerol; and 24 hours in each 1% KOH/50% glycerol and 1% KOH/70% glycerol and finally stored in 100% glycerol.

Mouse eye processing for histology and immunofluorescence

Embryos and eyes were dissected and rinsed in cold phosphate buffered saline (PBS) and then fixed in 4% PFA overnight at 4°C; rinsed again in PBS and dehydrated in serial dilutions of ethanol and paraffin embedded. All adult eye sections were cut using a standard microtome at 12 µm and oriented in the sagittal plane; embryonic sections were cut at 6 µm and oriented in the coronal plane. Haemotoxylin and Eosin staining was performed according to standard methods.

qRT-PCR and PCR

Details of all primers and sequences used in this study are in Table S1. DNA was extracted according to established protocols. RNA was isolated from tissues or from cultured mouse embryonic fibroblast (MEF) cells using TRIzol Reagent (Invitrogen) according to the manufacturers instructions. For cDNA, 1^st Strand cDNA Synthesis Kit for RT-PCR (Roche), or Superscript III (Invitrogen) were used according to the manufacturers instructions. All PCR reactions were carried out with primers at 0.2 µM. Cycle conditions: 95°C×3 min; followed by 30 cycles of 95°C×30 s; 58°C×45 s; 72°C×45 s; followed by 10 min at 72°C. Quantitative RT-PCR was performed as above using Brilliant II SYBR Green QPCR Master Mix (Agilent) and an ABI HT7900.

Genome-wide mapping of Mp

Microsatellite markers were used for PCR-based mapping using genomic DNA samples from heterozygous (n = 38) offspring from the 3^rd generation of outcrosses to C57BL/6J (for Wt, n = 30). In total 73× markers were selected that were informative between C3H and C57BL/6J alleles. A list of the markers used and peak sizes for each allele are available in Table S1. Equimolar samples were pooled together as either wild type or heterozygote DNA. Amplicons were analysed on the ABI PRISM 310 Genetic Analyzer. Peak heights for each allele were recorded and the ratios of C57BL/6J:C3H were calculated. For the chromosome 18 locus, non-fluorescently labeled microsatellite PCR products were resolved on 4% agarose gels. Oligonucleotides for fine mapping chromosome 18 are listed in Table S1.

3′-End cDNA amplification

Rapid Amplification of cDNA Ends (RACE) was used for the identification of the 3′-end of the Fbn2 transcript using an adaptor oligonucleotide primer (5′-GACTCGAGTCGACATCGATTTTTTTTTTTTTTTTT-3′) as the only primer in separate reverse transcription reactions with RNA extracted from phenotypically mutant and Wt animals as described above, using 1^st Strand cDNA Synthesis Kit for RT-PCR (Roche) with standard conditions. The single-stranded cDNA species (cDNA) were then used in gene-specific PCR reactions with an adaptor-only primer (5′-GACTCGAGTCGACATCG-3′) and the Fbn2-specific primer (5′-GTCTCAGCCTTCCCTCTGTG-3′). Products from these separate reactions were then gel electrophoresed and candidate bands were excised and TA-cloned into the pGEM-Easy vector system (Promega), according to the manufacturers instructions. Sequence analysis was performed using T7 and SP6 vector-specific primers and Sequencher Version 4.8 (Gene Codes Corp. MI, USA) and then used as query sequences in BLAT search alignments against the mouse genome (http://genome.ucsc.edu/cgi-bin/hgBlat). A PCR-based genotyping assay was devised based on the genomic rearrangement identified at the Mp locus.

Transmission-Electron Microscopy (TEM)

Tissues were collected from freshly culled neonates (P2 and P8) and placed into sterile serum-free DMEM (Gibco). For ultrastructural observation, the tissue was fixed for 1–3 hours in 1.5% Glutaraldehyde/1.5% Paraformaldehyde with 0.05% tannic acid in DMEM, rinsed in DMEM then immersed in 1% OsO₄ in DMEM for one hour. Tissues were dehydrated in a graded series of ethanol to 100%, washed in propylene oxide, then infiltrated and embedded in Spurr's epoxy. 60–90 nm ultrathin sections were contrasted with saturated uranyl acetate in 50% ethanol for 15 minutes followed by lead citrate for 3 minutes. For immunocytochemistry, fresh tissue was fixed in 0.1% glutaraldehyde/4% paraformaldehyde for 30 minutes on ice, rinsed in DMEM then 0.15 M TRIS-HCl (pH 7.4) overnight. Tissues were then exposed to a graded series of ethanol dilutions to 90% at progressively lower temperature to −20°C, infiltrated in LR White embedding media at −20°C, then polymerized at 60°C in a nitrogen atmosphere. Ultrathin sections mounted on formvar coated 1×2 mm nickel slot grids were immunolabeled using antibody pAb0868 specific to Fbn2, pAb9543 specific to Fbn1 and also control antibodies of irrelevant specificity followed by a combination of −5 and −10 nm secondary gold conjugates as previously described [42]. Sections were examined on either a Philips EM410LS TEM or an FEI G20 TEM operated at 120 kV.

Primary MEF cultures

Genotyped embryos were collected at E13.5 and the limbs and skin from each embryo were then finely chopped and cultured in DMEM (Gibco) with 20% foetal calf serum and 1% penicillin/streptomycin at 37°C; 5% CO₂; 3% O₂. The Wnt-FLAG containing plasmid was a kind gift from Dr Angela Lee and modified from a PCMxGFP2FLAG vector. Transfection of RPE1 cells with Wnt-FLAG were performed using Lipofectamine-LTX (Invitrogen), with pcDNA3.1 containing enhanced-GFP to control for transfection efficiency. Tunicamycin (Sigma) was used at a concentration of 0.5 µg/ml and carried in DMSO. Non-Tn treated cells were exposed to carrier alone.

Xbp1 splicing assay

Alternatively spliced Xbp1transcripts were differentially identified using RT-PCR and the ABI PRISM 310 Genetic Analyzer, using a FAM-labeled forward, and an unlabelled reverse primers specific for the spliced region of Xbp1 (Table S1). RNA samples were collected from embryonic eyes and MEF cultures and cDNA was synthesized. The unspliced (Xbp1u) transcript was 289 bp and the spliced transcript (Xbp1s) was 263 bp. The assay was verified using tunicamycin to induce ER-stress in cultured MEFs (data not shown). Peak heights for each amplicon were measured using the ABI PRISM 310 Genetic Analyzer, and means were calculated and displayed graphically.

Immunohistochemistry

All antibodies used in this study are listed in Table S2. For immunofluorescence, primary MEFs were cultured on glass coverslips as described above for 72 hours and media was replaced with serum-free media for 16 hours. Coverslips were fixed (i) for PDI and GM130 staining- with 3.7% formaldehyde on ice for 10 minutes, or (ii) for Fbn2 with acetone at −20°C for 15 minutes, rinsed once in PBS and then permeabilised in PBS+0.5% Triton-X-100 (Sigma) for 10 minutes, blocked in PBS+10% serum for 1 hour, and incubated with primary antibodies overnight at 4°C. Coverslips were washed 3×5 minutes in PBS, and incubated in Alexa Fluor F(ab′) fragment (Invitrogen) secondary antibodies diluted 1∶1000 in blocking buffer for 1 hour and mounted with ProLong Gold Antifade reagent (Invitrogen). For immunohistochemistry, paraffin sections were de-waxed and antigen retrieval was performed by boiling slides in 0.1 M Citrate buffer (pH 6.0) for 5 minutes. Paraffin or cryosection samples were then rinsed in PBS and blocked for 1 hr in buffer containing 10% serum (heat inactivated at 60°C for 1 hr); 1% BSA; 0.1% Tween-20; 0.05% Triton-X-100, all in PBS. Primary and secondary antibodies were applied as before. Colorimetric immunohistochemistry was performed using Vectastain ABC kits (Vector Laboratories) with alkaline phosphatase detection and BM Purple (Roche). Imaging was performed using a Zeiss Axioplan II fluorescence microscope with Plan-neofluar objectives. Images were captured using a Coolsnap HQ CCD camera (Photometrics Ltd, Tucson, AZ). Image analysis was performed using IPLab Spectrum Software (Scanalytics Corp, Fairfax, VA).

RNA In Situ analysis and riboprobe synthesis

Riboprobe synthesis primers are listed in Table S1. Purified PCR products were incubated for 37°C for 2 h in a 20 µl reaction containing: 2 µl transcription buffer; 1 µl RNase inhibitor; 2 µl DIG RNA labelling mix and T7 RNA polymerase (all Roche); in ultra-pure H₂O. Riboprobes were then DNAseI treated and Sodium-acetate/Ethanol precipitated. Whole-mount in situ hybridization to mouse embryos was carried out as previously described [63]. For section in situ, tissue sections were prepared as described for immunohistochemistry, and processed as described in J. Rainger's doctoral thesis (available on request). X-gal staining was performed on dissected embryos for beta-galactosidase activity as described in [18].

Secretion assays

For secretion assays, 1×10⁵ cells were seeded into 6-well plates and cultured under normal conditions for 3 days. Five independent cell lines were used per genotype. Media was removed and replaced with 1 ml DMEM (Gibco) with reduced serum and cultured for a further 72 hours. The conditioned media was collected and 500 µl was concentrated 10 fold using an Amicon Ultra 30 K (Millipore) centrifugal filter according to the manufacturers instructions. Concentrated conditioned media samples were then both (i) run on 10% NuPAGE Novex Bis-Tris Mini Gel SDS polyacrylamide gel (Invitrogen) and stained with 0.05% Coomassie Brilliant Blue (40% methanol, 10% acetic acid), and (ii) measured for total protein by Bradford assay, to ensure equal protein content per sample. Their protein composition in these media were analysed using a label-free shotgun proteomics method according to a previous report [64] with some modifications. First, the database search was performed using Proteome Discover (version 1.2) or Peaks (Peaks Scientific, version 6.0) with peptide false discovery rate constrained at 1%. Second, parent ion mass tolerance and fragment ion mass tolerances were set at 5 ppm and 0.6 Da. Third, Sieve software (version 2.0; Thermo Scientific) and Peaks (Peaks Scientific) were used to compare protein abundance with a total ion current normalization. Only more than two peptides per protein detectable was considered and each protein was expressed as a ratio Mp/Mp:Wt according to Asara et al [65]. For transcript analyses, cells were dissociated and counted using a haemocytometer and total RNA extractions were performed with Trizol reagent (Invitrogen) and cDNA prepared. qRT-PCR was performed on ABI HT7900 using 0.5 M sequence-specific primers for Col6a1; Col3a1; Col1a2; Col12a1; Fstl1; Postn; and Actn1 with Brilliant II SYBR Green QPCR mix (Agilent) and 0.4 µl of cDNA template per 20 µl reaction. Each sample was run in triplicate per qRT-PCR reaction and the data presented represents the mean of 5 independent samples per genotype and three separate qRT-PCR technical repeat. For relative expression, all CT values were normalized to CT-ß-Actin.

Immunoblotting

Protein samples were run on NuPage SDS polyacrylamide gels and transferred to a PVDF membrane, blocked for 1 hour in 5% non-fat dry milk diluted in PBS and incubated overnight in primary antibodies diluted in block. Membranes were washed extensively in PBS and incubated in the relevant secondary IgG-HRP antibody diluted in block for 1 hour at room temperature. Membranes were then washed again and immuno-complexes were visualized with Amersham ECL Western Blotting Detection Reagent (GE Healthcare Life Sciences).

Supporting Information

Attachment 1

Attachment 2

Attachment 3

Attachment 4

Attachment 5

Attachment 6

Attachment 7

Attachment 8