Mutations in Zebrafish lrp2 Result in Adult-Onset Ocular Pathogenesis That Models Myopia and Other Risk Factors for Glaucoma

The glaucomas comprise a genetically complex group of retinal neuropathies that typically occur late in life and are characterized by progressive pathology of the optic nerve head and degeneration of retinal ganglion cells. In addition to age and family history, other significant risk factors for glaucoma include elevated intraocular pressure (IOP) and myopia. The complexity of glaucoma has made it difficult to model in animals, but also challenging to identify responsible genes. We have used zebrafish to identify a genetically complex, recessive mutant that shows risk factors for glaucoma including adult onset severe myopia, elevated IOP, and progressive retinal ganglion cell pathology. Positional cloning and analysis of a non-complementing allele indicated that non-sense mutations in low density lipoprotein receptor-related protein 2 (lrp2) underlie the mutant phenotype. Lrp2, previously named Megalin, functions as an endocytic receptor for a wide-variety of bioactive molecules including Sonic hedgehog, Bone morphogenic protein 4, retinol-binding protein, vitamin D-binding protein, and apolipoprotein E, among others. Detailed phenotype analyses indicated that as lrp2 mutant fish age, many individuals—but not all—develop high IOP and severe myopia with obviously enlarged eye globes. This results in retinal stretch and prolonged stress to retinal ganglion cells, which ultimately show signs of pathogenesis. Our studies implicate altered Lrp2-mediated homeostasis as important for myopia and other risk factors for glaucoma in humans and establish a new genetic model for further study of phenotypes associated with this disease.


Introduction
The multi-factorial nature of many ocular diseases poses a major challenge in understanding their molecular etiology and in engineering animal models to study mechanisms of pathology. Macular degeneration, myopia, and glaucoma are examples of prevalent and disruptive complex ocular diseases. While characterization of complement factor genes has provided insight into most cases of macular degeneration [1], no major genetic pathway has been found to underlie myopia or glaucoma. Myopia is the most common human ocular disorder worldwide and is caused by abnormal growth of the eye resulting in refractive error [2,3]. Myopia also increases risk for other visual impairing diseases including glaucoma [4]. The glaucomas are a heterogeneous group of progressive blinding disorders that result from damage to retinal ganglion cells and their axons [5]. Important risk factors for glaucoma include elevated intraocular pressure (IOP), age, family history, and myopia [6]. Although traditional human genetic analysis has been limited in identifying causative genes for complex disorders, mutational screens in animals can provide insights into disease etiology. Recently, progress has been made on establishing the zebrafish model to study phenotypes associated with glaucoma. From a forward-genetic perspective, zebrafish offer a major advantage in studying complex disease, in that large pedigrees can be efficiently generated with moderate space and time requirements.
Through a mutational screen for adult ocular defects, we identified a complex mutant, bugeye, that manifests multiple adultonset phenotypes associated with glaucoma including enlarged eyes with myopia, elevated IOP, and damage to retinal ganglion cells. Using linkage analysis we discovered non-sense mutations in low density lipoprotein receptor-related protein 2 (lrp2) for bugeye, as well as within a non-complementing allele. Lrp2 is a large transmembrane protein of the LDL-receptor related protein (Lrp) family [7]. Lrp2 participates in receptor-mediated endocytosis and has a host of identified ligands including signaling molecules like Sonic hedgehog and Bone morphogenetic protein 4, vitamin and hormone binding proteins, apolipoproteins, among others [8]. Lrp2 is expressed on cells of the renal proximal tubule, choroid plexus, developing neural tube, intestine, thyroid, and inner ear. Within the eye, Lrp2 is expressed on retinal pigment epithelial cells as well as ciliary epithelial cells [7][8][9]. In humans, mutations in LRP2 result in Donnai-Barrow syndrome [10], a rare disease characterized by a spectrum of phenotypes including agenesis of the corpus collosum, diaphragmatic hernia, sensonurial deafness, hypertelorism, buphthalmia (enlarged eye globes) and high myopia [11,12]. As the eyes of bugeye zebrafish are also highly myopic, Lrp2 may be critical in regulating emmetropic eye growth across species. The strong association of myopia with glaucoma [13] makes bugeye an attractive model to study the genetic and molecular pathways involved in these ocular diseases.

Identification of an enlarged eye mutant with elevated intraocular pressure
The bugeye zebrafish mutant was identified in a three-generation forward-genetic screen for adult ocular abnormalities. Mutants were easily identified by 6 months as their eyes were visibly enlarged ( Figure 1A, 1C). Interestingly, the degree of eye enlargement often varied between the two eyes of a single fish (Figure S1A-S1G). Occasionally the phenotype presented only in one eye, and the other eye remained normal in size ( Figure S1B). To address whether ocular enlargement in mutants might represent a retinoblastoma phenotype, we analyzed eyes by histology. Instead of obvious cellular overgrowth we found that the retina was notably thinner in all layers ( Figure 1B, 1D). As buphthalmia is often associated with elevated IOP, we used servonull electrophysiology to measure the eye pressures in mutants and wild-type siblings [14]. Compared to wild-type fish, bugeye mutants consistently showed elevated IOPs ( Figure 1I). In addition, the rare fish that presented the phenotype in a unilateral manner had normal pressure in the wild-type sized eye and elevated pressure in the enlarged eye ( Figure S1H). IOP is maintained by the balance of aqueous humor production and drainage. Like mammals, aqueous humor in zebrafish is produced in the ciliary epithelium and drained at the iridocorneal angle. However, unlike mammals where drainage occurs circumferentially throughout the angle region, aqueous outflow for zebrafish is facilitated through a discrete ventrally localized canalicular network [15]. Histology did not reveal obvious disorganization in either the dorsal ciliary epithelium ( Figure 1E, 1G) or in the ventral canalicular outflow network ( Figure 1F, 1H). However, the ciliary epithelium occasionally appeared mildly hypertrophied ( Figure 1G, arrow) and the angle region of mutants was more prone than wild-type specimens to separation between the iris and corneal tissues during histological preparation ( Figure 1H, asterisk). Additional characterization of these regions at the time of phenotype onset confirmed these observations ( Figure S2).

Mutations in lrp2 underlie the complex bugeye phenotype
The original bugeye mutants presented in the third generation of a three-generation screen, suggesting the mutation was recessive. However, only 3 fish out of a family of 28 showed the phenotype and therefore the penetrance was lower than predicted for a

Author Summary
Complex genetic inheritance, including variable penetrance and severity, underlies many common eye diseases. In this study, we present analysis of a zebrafish mutant, bugeye, which shows complex inheritance of multiple ocular phenotypes that are known risk factors for glaucoma, including high myopia, elevated intraocular pressure, and up-regulation of stress-response genes in retinal ganglion cells. Molecular genetic analysis revealed that mutations in low density lipoprotein receptor-related protein 2 (lrp2) underlie the mutant phenotypes. Lrp2 is a large transmembrane protein expressed in epithelia of the eye. It facilitates transport and clearance of multiple secreted bioactive factors through receptor-mediated endocytosis. Glaucoma, a progressive blinding disorder, usually presents in adulthood and is characterized by optic nerve damage followed by ganglion cell death. In bugeye/ lrp2 mutants, ganglion cell death was significantly elevated, but surprisingly moderate, and therefore they do not model this endpoint of glaucoma. As such, bugeye/ lrp2 mutants should be considered valuable as a genetic model (A) for buphthalmia, myopia, and regulated eye growth; (B) for identifying genes and pathways that modify the observed ocular phenotypes; and (C) for studying the initiation of retinal ganglion cell pathology in the context of high myopia and elevated intraocular pressure. simple recessive mutation (,9% vs. 25% predicted). Moreover, incrossing 2 of those original mutant fish resulted in 25 progeny that showed large eyes and 18 that never developed the phenotype. Again, if the mutation was a simple recessive mutation, incrossing should have resulted in all progeny showing the phenotype. To better characterize inheritance and establish recombinant mapping panels to genetically position the mutant locus, we set up a series of test-crosses. Table 1 summarizes the results of incross, outcross and backcross matings over multiple generations and genetic backgrounds ( Table 1). The data indicate that the bugeye phenotype is most likely caused by a single recessive mutation, but like many multi-factorial complex diseases, the penetrance was modified by common wild-type backgrounds and/ or by non-genetic factors.
To map the mutant locus, progeny from single pair backcross matings were used for whole-genome linkage analysis. Cosegregation for markers on chromosome 9 and the mutant phenotype was found ( Figure 2A). Informatively, no other linkage in the genome was noted, consistent with the single recessive causative mutation hypothesis. Public databases revealed that the lrp2 gene was within the critical recombinant interval. Given the similarity of the bugeye phenotype to those caused by LRP2 mutations in humans, we sequenced this candidate gene. Analysis of lrp2 cDNA from bugeye mw1 mutants revealed a T to A conversion that changes a cysteine to a stop codon at predicted amino acid position 23 (C23X) ( Figure 2B, 2C). Through an independent genetic screen we identified a second large eye mutant that like the bugeye mw1 allele, presented in adulthood and showed reduced penetrance. Intercrosses between this mutant (allele p5bnc) and bugeye mw1 were non-complementing and suggested that lrp2 may also be affected in the p5bnc mutant. Indeed, sequencing of p5bnc cDNA revealed a separate non-sense mutation, also very early in the coding region of lrp2 (bugeye p5bnc , Q413X) ( Figure 2B, 2C). To test whether somatic reversion or alternate splicing around the non-sense mutations might underlie the reduced penetrance or variability often observed between the left and right eyes, we sequenced ocular cDNA in affected and unaffected eyes. However, we did not find evidence of mosaicism or alternate splicing surrounding the mutations, suggesting the penetrance and phenotype variability is influenced by other genes, epigenetics, and/or unpredictable changes in physiology which affects the phenotypes.
In mammalian eyes, the multi-ligand receptor Lrp2 is known to be expressed in the developing and adult retinal pigment epithelium (RPE) and ciliary epithelium. We therefore analyzed Lrp2 expression in wild-type, bugeye mutant larvae treated with phenyl-thio-urea (PTU), which blocks pigmentation and allows visualization of potential RPE immunoreactivity. As predicted, strong immunoreactivity was found in wild-type RPE and ciliary epithelium. Other regions of expression noted in wild-type fish included forebrain ventricles, regions of the inner ear, proximal pronephros, and gut epithelium (data not shown). All Lrp2 immunoreactivity was completely absent in mutant larvae for both bugeye alleles ( Figure 2D, 2E and data not shown). We next developed genotyping protocols for both mutant alleles and confirmed that large-eyed fish never showed wild-type lrp2 genotypes ( Figure 2F, 2G). We also used this assay to test whether the reduced penetrance of the ocular phenotype could be explained by increased larval lethality of lrp2 mutants. However, we found that all genotypes were represented in Mendelian ratios in the adult progeny of either heterozygous or backcross pairwise matings, despite the fact that some homozygous mutants never developed the enlarged eye phenotype. Cumulatively, these data indicate that lrp2 mutations are responsible for the large-eyed phenotype in bugeye and that the reduced penetrance and variability in eye enlargement are due to either common (yet unknown) genetic background differences and/or non-genetic factors such as physiological modifiers of the mutation.

Lrp2 mutants show adult onset buphthalmia and progressive myopia
Having established the causative gene for bugeye, we next investigated the onset of the ocular phenotype and quantified the pathology. To characterize the development of enlarged eyes in bugeye/lrp2 mutants we performed longitudinal studies tracking wild-type and mutant fish from 1-12 months. The zebrafish eye reaches its final adult anatomy by approximately 1 month of age [16]. Because overall growth rates can vary between equally aged fish -even within the same tank -we used the ratio of eye size to body length (E:B) to determine the relative size of the eye. This ratio remained constant in wild-type fish, allowing comparison of relative eye size between individuals regardless of the overall growth of the fish. Although this ratio remained flat as wild-type fish grew, the E:B ratio increased over time for most lrp2 mutants ( Figure 3A). Despite individual variability, the average body length growth rates between wild-type and lrp2 mutant fish were indistinguishable ( Figure S3). For the E:B ratio, no wild-type fish had a value greater than 0.05 (most fell between 0.02 and 0.04), and mutants with visibly enlarged eyes had an E:B ratio $0.07. The onset of large eyes was variable both within shared tanks of siblings and between generations, but a statistically significant difference between mutant and wild-type fish was consistently found at 2 months ( Figure 3A and data not shown). In general, lrp2 mutant eyes become visibly enlarged in adults between 2-6 months and eye growth often plateaus between 8-12 months.
Histological cross-sections of lrp2 mutant eyes revealed that the region with the greatest increase in size was the depth of the vitreous chamber ( Figure 3B). This suggests that in large-eyed mutants, the retina lies behind the point at which the lens focuses light and the eyes are therefore myopic. We calculated the relative refractive error (RRE), an estimate for the degree of myopia, for lrp2 mutant eyes relative to wild-type eyes at 1 and 2 months using retina and lens radius measurements collected from histological sections. Using the RRE equation, a myopic eye has a negative value and a hyperopic eye is positive ( Figure 3C; Methods). We found that lrp2 mutant eyes are slightly myopic at 1 month, but become significantly more myopic by 2 months ( Figure 3D).

Lrp2 mutants show reduced retinal cell density with increased eye size
Histology suggested retinal cell density was affected in lrp2 mutant eyes (Figure 4). At 1 month, before eyes of mutant fish were visibly enlarged, there was a small reduction in retinal cell density as compared to wild-type fish ( Figure 4A, 4B). By 2 months, when the onset of large eyes had occurred in some mutants but not in others, there was a significant difference in cell density in all layers of the retina ( Figure 4C, 4D). As expected, at 6 months when relative eye size was greater overall, there was a further decrease in cell density ( Figure 4E, 4F). When considering retinal cell density for each layer as a function of relative eye size (as measured by the ratio of the retinal cross-section length to body length), we found that for mutants, the relation between neuron density and relative eye size decreased in a linear manner ( Figure 5A). The same was true when considering just the absolute  . Eye growth and relative refractive errors in bugeye/ lrp2 mutants. A bugeye mutants have a greater eye area to body length ratio (E:B) than age-matched wild-types beginning at 2 months (p,0.05, t-test), which becomes more pronounced with age. Each dot represents the E:B for an individual eye. B Histological transverse sections through the whole eye show an increased depth of the vitreous chamber (from the lens to retina) in bugeye (right) as compared to wild-type fish (left). C Diagram showing the calculation of relative refractive error (RRE): lens radius (l) and retina radius (r), using a focal length of the lens (f) as 2.32 x lens radius. D The RRE measurements revealed mild but significant myopia in 1-month bugeye fish (p = 0.0002, t-test), that becomes more dramatic at 2 months of age (p,0.0001, ttest). n = 6 eyes each for 1 month TL and bugeye; n = 16 and 22 eyes for 2 month TL wild-type and bugeye, respectively. doi:10.1371/journal.pgen.1001310.g003 size of eye (as measured by retinal cross-section length, Figure 5B). Interestingly, there was an increase in photoreceptor density in larger eyes for wild-type fish ( Figure 5B). When considering cell density for wild-type and mutant eyes of the same absolute size, but of different ages in order to match size, density was still reduced in lrp2 mutant fish ( Figure 5C). For this comparison we evaluated retinal cell density of 6-month old wild-type fish and 2month old lrp2 mutant fish, each that had retinal lengths that fell between 2-3 mm. Importantly, there was no significant change in cell density for the retinal ganglion cells layer between 2-6 months in wild-type fish. For the inner nuclear and photoreceptor layers, there was a small, but significant change (ANOVA, p,0.001), where the cellular densities increased with age. Together, these data suggest that the reduced neuron density seen in lrp2 mutant retinas is not simply due to an acceleration of normal ocular growth.
Despite the reduced cell density in mutant eyes, total retinal cell number was estimated to be greater than wild-type, owing to the much larger eye size overall. We estimated total retinal cell numbers by considering the retina area as that of the surface area for half a sphere and extrapolated total cell numbers using density data. These calculations showed that mutant eyes with E:B ratios .0.07 had significantly increased numbers of total neurons. More directly, analysis of DNA content, which is proportional to total cell number, confirmed that large-eyed mutant fish (EB ratio .0.07) had more cells, even though retinal cell density was much lower (data not shown).
The altered retinal cell density in lrp2 mutants could be due to either insufficient cell generation to match scleral growth and remodeling, or through increased cell death. To address these possibilities we analyzed by immunofluorescence the number of proliferating cells within the ciliary margin zone (using Minichromosome maintenance homolog 5, Mcm5 antibodies) and the number of apoptotic cells across the retina (using activated-Caspase3 antibodies). Mcm5 is required for DNA replication and is expressed throughout the cell cycle in all proliferating cells, but the protein is rapidly lost in post-mitotic cells. Proteolytic cleavage of Caspase3, recognized by the activated-Caspase3 antibody, is one of the last steps in the apoptosis cascade and marks cells committed to die in a number of contexts, including glaucoma. At 1 month, proliferation in both wild-type and lrp2 mutant retinas was primarily confined to the ciliary marginal zone, a stem cell niche where ongoing proliferation from multipotent elongated neuroepithelial cells is known to occur in fish [17] (Figure 6A-6C). For each genotype, occasional Mcm5-positive cells were also located in the inner nuclear layer, which have previously been shown to be rod progenitor cells in teleost fish [18][19][20]. At 2 months, cell counts indicated a reduction in Mcm5-positive cells per CMZ niche in bugeye fish, suggesting maintenance of stem cells was inadequate to match eye globe growth ( Figure 6D-6F). Consistent with this observation, a role for Lrp2 in maintaining neuronal stem cells of the adult mouse forebrain has been recently described [21].
Similar to analysis of proliferation, cryosections of wild-type and lrp2 mutant retinas were used to investigate cell death. However, very few dying cells were noted in sections of retina from either condition. Similar results were obtained using the TUNEL assay to characterize dying cells. We therefore used activated-Caspase3 immunoreactivity on control and lrp2 mutant flat-mounted retinas  to observe all neurons from individual samples. Even by flatmount analysis, there was little apoptosis up to 6 months of age ( Figure 6G), although at these times bugeye mutants showed trends towards increased numbers of activated-Caspase3-positive cells. By 12 months, apoptosis in bugeye retinas was significantly elevated. We also noted that activated-Caspase3 immunoreactivity from all ages was restricted to the retinal ganglion cell layer ( Figure 6H). It is possible, however, that some cells, including those outside of the ganglion cell layer die by Caspase3-and TUNEL-independent mechanisms. Overall, these data indicate that initially, as lrp2 mutant eyes expand, proliferation is not sufficient to maintain proper cell density and later, perhaps following mechanical stress imposed by retinal stretch, retinal ganglion cells begin to die.
To investigate whether the markers of retinal ganglion cell stress correlated with optic nerve pathology, we first compared sagittal sections of wild-type and lrp2 mutant optic nerve heads from 6month-old fish by light microscopy. We then analyzed crosssections of wild-type and lrp2 mutant optic nerves, just posterior to the optic nerve head from 7-and 12-month-old fish by transmission electron microscopy (TEM). In zebrafish, like other teleost fish as well as some rodents, the optic nerve head is comprised of an astroglial lamina without obvious elastin-collagen rich laminar plates as observed in primates [44][45][46]. In addition, as the optic nerve exits the fish eye, it is initially unmyelinated, like that in humans and most mammals [47][48][49][50]. Histology of the optic nerve head did not reveal excavation or cupping in lrp2 mutants, but did indicate mutant nerves were larger, consistent with increased total numbers of retinal ganglion cells in the large-eyed fish ( Figure 7B). Optic nerve cross-sections for TEM were collected distal to the exit point from the eye within the myelinated region of the optic nerve, which is adjacent to the site of axonal injury in glaucoma [51][52][53]. Nerve damage was scored as 1) degenerating axons, as noted by electron-dense appearance, 2) axons having an unraveled myelinated sheath, or 3) space left behind by a shrunken and degenerating axon. At both ages, examples for each type of pathology were found in wild-type and lrp2 mutant optic nerves ( Figure 7C). Surprisingly, when total counts were normalized to area (mm 2 ) there were no differences between genotypes or ages ( Figure 7D).
Because the ultrastructural signature of degenerating axons following a crush injury is relatively short-lived in the optic nerve tract of teleost fish as compared to mammals [54], we utilized a genetic tool to label damaged and regenerating axons over a longer period of time [55]. We crossed Tg(3.6Frgap43:GFP) mil1 transgenic fish with lrp2 homozygous mutants and then used the resulting progeny to backcross with non-transgenic lrp2 mutant fish. This breeding scheme resulted in families with equal proportions of lrp2 heterozygous and homozygous mutant fish carrying single insertions of the 3.6Frgap43:GFP transgene. This transgene contains 3.6 kb of regulatory sequence (59 flanking region and first intron) from theTakifugu rubripes gap43 locus driving GFP. Importantly, in these transgenic fish, GFP is expressed in axons following injury [55]. For our analysis, we compared largeeyed lrp2 homozygous mutant fish (.0.07 E:B ratio) to normaleyed heterozygous siblings ( Figure 8K-8T). In all large-eyed mutant fish we observed strong activation of GFP in a sub-set of retinal ganglion cells. In the majority of mutant retinas examined (6 of 6 at 6 months, Figure 8P-8T; and 10 of 12 at 12 months, data not shown), there was a characteristic axon 'wandering' and 'circling' around the optic nerve head. This axon phenotype, where GFP-positive axons approached the optic nerve head in a disorganized and circuitous fashion, was never observed in retinas from age matched lrp2 heterozygotes ( Figure 8K-8O) or from 12month wild-type fish that carried the 3.6Frgap43:GFP transgene (data not shown). The transgene was activated with variability at 2 months in both wild-type or lrp2 mutant fish ( Figure 8A-8J), but the wandering axon phenotype was only rarely observed in mutants at this early timepoint. Weak expression of the transgene was noted in the nerve fiber layer of non-mutant retinas, consistent with the ongoing neurogenesis of zebrafish. In addition, older wildtype fish occasionally showed stronger GFP-positive axons, suggesting sporadic age-related degeneration. In wild-type eyes, all of the low-GFP expressing axons, as well as the occasional high-GFP expressing axons, exited the eye directly without wandering or circling the optic nerve head like those of mutants. To address whether the chronic stress conditions of lrp2 mutants differ from acute injury, we performed optic nerve crushes on adult gap43:GFP fish. At 6 days post-crush there was significant upregulation of GFP across the retina ( Figure S4). By 5 weeks postcrush, when axons had regrown [56], there was only an occasional wandering axon. Most samples following nerve crush, however, showed accurate and direct axon targeting through the optic nerve head. By 11 weeks post-crush, there was significant reduction in transgene activation and no axons showed wandering or circling at the optic nerve head like age-matched lrp2 mutants. This comparison highlights the differences between the chronic stresses caused by the lrp2 mutation versus the acute, crush injury model, in which the genetic model results in changes at the optic nerve head that are not evident in the post-nerve head crush paradigm.

Discussion
A major challenge with research on either myopia or glaucoma is identifying genetic lesions that impact the diseases. Recently, genome wide association studies for both diseases revealed noncoding associated changes, but the predicted effects on phenotypes were small and the actual gene products affected by the intergenic alterations have yet to be identified [57][58][59]. In addition to defining genetic susceptibilities for glaucoma, there is a need to understand and model how other risk factors like age, raised IOP, and myopia itself affect the onset, severity, and progression of neuropathology. In our studies we identified non-sense mutations in zebrafish lrp2 that lead to phenotypes that are known risk factors for glaucoma. These phenotypes included increased IOP, enlarged eye globes with significant refractive errors, decreased retinal neuron density, activation of retinal ganglion cell stress genes, and distinct axon pathology at the optic nerve head. The zebrafish lrp2 mutants have similar heritable phenotypes to the black moor goldfish [60,61] and the RCS;rdy-rat [62]. It will be interesting to see if lrp2 or pathway genes are affected in either of those models. Similarly, it is possible that alterations to genes that control pathways affected by loss of Lrp2 might influence myopia or forms of glaucoma. To date, however, polymorphisms in Lrp2 have only been linked with urate and cholesterol levels in serum [63,64] and the molecular and cellular pathways affected by loss of Lrp2 that impact the ocular phenotypes remain uncharacterized. In general,  Lrp2 functions in regulation and homeostasis of multiple bioactive molecules including vitamins, hormones, nutrients, and growth factors through localized tissue delivery or reuptake by epithelia. In knowing the affected gene, the zebrafish mutants hold promise in shedding light on how de-regulated signaling and homeostasis affect phenotypes such as elevated IOP or excessive eye growth.
While it is tempting to speculate that the excessive eye growth in lrp2 mutants is due to the elevated IOP, our studies do not rule out the possibility that these two phenotypes are distinct. In fact, the only two Donnai-Barrow patients who have had their IOPs reported (each with non-sense mutations in LRP2), showed values in the normal range [65]. Despite normal IOPs, the eyes of the two young siblings were enlarged and showed high myopia. Furthermore, as an endocytic receptor found on the RPE, Lrp2 is an interesting candidate as a direct regulator of emmetropization [66]. Potentially, Lrp2 mediates the availability or transport of signaling molecules from the retina to affect remodeling within the sclera. In this context Lrp2 might be key in facilitating the matching of visual input with axial length of the eye. Nonetheless, relationships between eye pressure and size are established and the elevated IOP in zebrafish lrp2 mutants is likely to be at least contributory to the observed buphthalmia. Consistent with this possibility, in the few mutant fish where the eye phenotype presented in a unilateral manner, IOPs were normal in unaffected eyes, yet elevated in enlarged ones. Indeed, expression of Lrp2 on the ciliary epithelium suggests a direct role in IOP regulation, particularly considering the function of Lrp2 at other sites of fluid regulation. For example in mice, Lrp2 has been shown to regulate glomerular filtration in the proximal tubule of the kidney and in the choroid plexus the receptor modulates homeostasis of cerebrospinal fluid [67][68][69][70][71].
A significant characteristic of lrp2 mutant fish is the strong relationship between abnormal eye globe growth, retinal thinning, and activation of retinal ganglion cell stress markers. In this context, lrp2 mutants have value as a genetic model for studying the effects of protracted mechanical stress on retinal ganglion cells, their axons, and the associated glia. As this phenotype relates to glaucoma, it was surprising that mutant fish did not show significantly elevated optic nerve pathology with TEM analysis. It is possible that the stresses induced by lrp2 mutations simply do not reach a threshold to cause ultrastructural pathology. Alternatively, low-grade stress may actually ''pre-condition'' and promote protective mechanisms in the mutant neurons [72,73]. However, the lack of a difference in ultrastructure pathology between mutant and wild-type siblings could also be explained by the surprisingly high number of pathological events noted in the wild-type fish. This perhaps relates to the regenerative capacity of teleosts [56,74] and a relaxation of selective pressure to maintain nerve health with normal aging. Through evolution, fish may have lost highly-robust nerve protective mechanisms against age-related stresses, and instead rely on ongoing growth and regeneration to maintain vision, perhaps accounting for the unexpected pathology scored in wild-type optic nerves. In addition, because a higher proportion of the ganglion cell axons in lrp2 mutant fish are in fact younger than those of wild-type siblings (due to the excessive ongoing generation of neurons in their eyes), many of the optic nerve profiles might be expected to in fact look healthier in a relative manner.
The modest death of retinal ganglion cells in lrp2 mutants was less surprising. First, extended retinal ganglion cell soma survival, despite axonal damage and dysfunction, is known for the DBA/2 mouse glaucoma model. DBA/2 mice show a pigment dispersionrelated glaucoma with elevated IOP [75,76]. In young DBA/2 mice, axons at the nerve head often show focal insults with many having dystrophic features [51]. In many aged animals, axons are clearly degenerative [51]. Most retinal ganglion cells, however, survive for extended periods of time and their disconnected proximal (intra-retinal) axons take on reactive and stressed characteristics [51,77,78]. Second, the resilient nature of retinal ganglion cells in teleosts has been well characterized. In fact for goldfish, experimental axotomy or optic nerve crush results in less than 10% death of retinal ganglion cells [79], and in zebrafish only 20% of the lesioned neurons are reported to die [80]. In contrast, optic nerve axotomy in mammals results in apoptosis of nearly all retinal ganglion cells [81][82][83]. The regrowth of axons in teleosts occurs over a course of weeks and results in correct axon pathfinding and appropriate tectal innervation [84,85]. In contrast, in lrp2 mutants, retinal ganglion cells appear to be under prolonged mechanical stress from the stretching and growth of the eye globe. This was evident from the changes in retinal density with eye enlargement and the activation of retinal ganglion cell stress markers. Of interest, axon regrowth through the optic nerve was affected in lrp2 mutants. The wandering and circling phenotype of the gap43:GFP axons in large-eyed mutants is reminiscent of the EphB3-dependent 'reactive plasticity' following optic nerve injury in mice [86,87]. Regardless, of why bugeye/lrp2 mutants do not show dramatic retinal ganglion cell death, this fact emphasizes that while these fish model initiating risk factors for glaucoma, they do not model the end stages of the disease.
Lrp2 mutations in humans and mice are often lethal, but always developmentally relevant, particularly within the nervous system [71,88,89]. Our analyses of both bugeye alleles indicate Lrp2 is dispensable for survival in zebrafish. Furthermore, we did not detect morphological phenotypes in mutant embryos, similar to the observations following oligonucleotide knock-down of zebrafish lrp2 [67]. The total lack of lethality in zebrafish lrp2 mutants may be due to species differences in respiration, as mice mutants often die from respiratory failure at birth. Alternatively, there may be compensation from other Lrp family members in zebrafish. Compensation from Lrp family members may also explain the lack of obvious developmental defects. More detailed studies of the zebrafish mutant embryos and larvae are warranted to assess whether subtle defects exist.
In summary, we have identified mutations in lrp2 that cause adult-onset ocular pathogenesis in zebrafish. While mutants appear normal during larval stages of development, as young adults they develop enlarged eyes with elevated IOP. Over time, retinal cell density becomes significantly reduced due to insufficient proliferation of marginal zone stem cells and increased neuronal cell death. Markers of retinal ganglion cell stress become elevated and damaged and/or regenerating axons at the optic nerve head show a characteristic wandering and circling phenotype. These fish will be valuable for future studies on the signaling and cellular mechanism of myopia and other risk factors for glaucoma.

Fish maintenance
Wild-type and mutant zebrafish (Dano rerio) were maintained at 28uC with a 14 on/10 off light cycle and were feed a standard diet [90]. All animal husbandry and experiments were approved and conducted in accordance with the guidelines set forth by the Institutional Animal Care and Use Committee of the Medical College of Wisconsin.

Measurement of intraocular pressures
Servo-null electrophysiology was used to measure IOPs as described previously [14].

Recombinant linkage mapping
Mapping panels of 6 month adult mutant fish (obviously enlarged eyes) were collected from backcross pedigrees. Bulked segregant analysis, using pooled samples of mutant genomic DNA and individual parental DNA, was conducted with simple sequence repeat (SSR) markers to establish linkage to Chromosome 9. For higher-resolution mapping, sequencing of parental genomic DNA in regions associated with the closest linked microsatellite markers was done to find additional SSRs. These new SSRs were then used to refine the critical interval by analyzing single mutant fish.
Genotyping PCR was performed on DNA isolated using the Puregene kit (Qiagen, Germantown, MD) from tailfin-clips, using primers designed to amplify the allele specific mutations in lrp2: bug mw1 F: CGTTATTTTCTGTCTAGGTTCAGGTTA, bug mw1 R: GAAAAGAAAAGATTGATACATACGG bug p5bnc F: GTGTGTTTTCTGAAAACTGTCAAGC, bug p5bnc R: CTTTGCAGCTGGTAATGAAAATCCACAC-CAACAGCGGCTCCTCTGTCCTA. Underlined letter in primer denotes mutant nucleotide, bolded letter denotes a single nucleotide change in the primer to generate a novel restriction site for each allele (bug mw1: MseI; bug p5bnc AvrII).

Eye size and body length measurements
Fish were anesthetized with 0.05% Tricaine and body lengths were measured in side-view from the tip of the head to the end of the trunk (before the caudal fin). To measure eye size, anesthetized fish were imaged at a fixed magnification from a dorsal perspective using a Nikon CoolPix995 camera attached to a Leica MZFLIII microscope. These images were imported into Metamorph software (Universal Imaging Corp, Philadelphia, PA), and the area of each eye from the dorsal view was traced using the Region Measurements function.

Relative refractive error
Lens radius (L) was measured from histological cross sections; retina radius (R) was back-calculated by assuming the retina to be a semi-circle, measuring the length of the retina, and taking that measurement as half the circumference of a circle (so R = length of retina/p). Sections with minimal distortion from processing were used and no attempts to correct for distortions were made. A focal length (F) of 2.32 x L for the lens was used as in studies with goldfish [91]. RRE was calculated as 1-(R/F). By this calculation, all wild type fish were predicted to be slightly hyperopic (RRE .1), likely due to fixation artifact. To adjust this, the ratio of (R/F) was multiplied by a constant factor for both genotypes at each age (1 month, 1.15; 2 months, 1.18), so that on average, the wild type fish were emmetropic (RRE = 0).

Histology
Heads were removed from terminally anesthetized fish and fixed overnight in gluteraldehyde/paraformaldehyde at 4uC, washed three times in PBS, and dehydrated in increasing ethanol solutions (50%, 70%, 80%, 90%, 95%, 100%, 100%, 100%) for 10 minutes each, all at RT. The heads were then infiltrated with propylene oxide for 15 minutes twice, then a 1:1 mix propylene oxide:epon for 2 hours at RT. An additional equal volume of epon was added to the samples and these were incubated overnight with culture tube caps off so that the propylene oxide would evaporate. Heads were bisected when necessary to fit in block-molds, embedded in epon, and baked for at least 24 hours at 65uC. Semi-thin sections were cut on a Leica RM2255 microtome and stained with 1% Toluidine, 1% Borax.

Cell counts
For each eye, 5 non-consecutive sections were imaged from the central retina (sections with the largest lens diameter) with a 40X objective on a Nikon E600FN microscope with a Photometrics CoolSnap camera attached. Each image was printed and the nuclei in each layer of the retina were counted. The average of the 5 sections was calculated and represented 1 data point. For sample condition, between 6-12 eyes were scored in this manner.

Immunostaining
Zebrafish embryos or isolated eyes were fixed overnight at 4uC in 4% PFA (pH 7.4, in PBS), washed three times for 10 minutes in PBS, then infiltrated with increasing concentrations of sucrose (15%, 30%) for 2 hours each at 4uC, followed by overnight incubation in HistoPrep freezing media (Fischer Scientific, Pittsburgh, PA). Cryoprotected embryos were embedded in HistoPrep and flash frozen, sectioned at 10-12 mM and collected on Supercharge Plus slides (Fischer Scientific). Cryosections were allowed to dry on the slide for 1hr at RT, and the edge of the slide was traced with a PAP pen. Slides were rinsed briefly with PBTD (PBS +1% DMSO +1% Tween-20) to rehydrate the tissue, and then incubated in block (5% donkey serum in PBTD) for 2 hours at RT. Primary antibody was diluted in block (Sheep-anti-Lrp2 1:1000, gift from Dr. Thomas Willnow (Max Delbruck Center, Berlin, Germany)) and incubated on slides overnight at 4uC. Antibody was removed and slides were washed three times with PBTD rinses, and secondary antibody diluted in block (Cy3-Donkey anti-Sheep 1:250, Jackson ImmunoResearch, Westgrove, PA) was incubated at RT for 1.5 hours. Secondary antibody was removed with three washes of PBTD, and slides were mounted in 1:1 PBS to glycerol with 0.1% Hoechst nuclear stain (cryosections). Images were collected using a Nikon C1 confocal microscope. The same procedure was followed for dissected whole adult retinas prior to flat-mount analysis, using anti-cleaved caspase-3 primary antibody (1:500, Cell Signaling Technology, Danvers, MA) and DyLight 488 secondary (1:1000, Jackson ImmunoResearch).

Real-time PCR
1 month fish measuring between 10-12mm were anesthetized in Tricaine, and both eyes were removed and placed immediately in TRIzol (Invitrogen). Each sample was a pool of 3 pairs of eyes (6 eyes per sample), and 4 samples were used for each genotype. RNA was isolated following the Invitrogen protocol. Reversetranscription PCR was carried out following the protocol for SuperScript III First Strand Synthesis (Invitrogen). Gene specific primers were used as follows to amplify the genes of interest: Aif1l Initially, PCR products were amplified with Accuprime Taq HighFidelity (Invitrogen), and sequenced to verify specificity. Real-time analysis was performed on a Bio-Rad iCycler using iQ SYBR Green SuperMix (Bio-Rad). 3-step PCR with a 57uC annealing temperature was used for all primer sets except Arg1, Atf3, and Thy1, which used a 2-step PCR with a 54uC annealing temperature to eliminate a non-specific product. All samples were run in triplicate, and fold change was calculated using the DDCt method, with Ef1a as the housekeeping gene for all primer sets.

TEM of optic nerves
Heads were removed from terminally anesthetized fish. In a Petri dish filled with buffer, the optic nerves were dissected from the heads first by removing the skin, skeleton, and connective tissue, leaving the eyes and attached nerves and tectum intact. The tectum was cut from the nerves, leaving the nerves intertwined at the chiasm. The nerves were separated by gently pulling on the eye globes with forceps, and making a cut with an 8 mm Spring Scissors (Fine Science Tools) when necessary. Dissected nerves with attached eyes were then incubated overnight at 4uC in gluteraldehyde/paraformaldehyde fixative. Heads were washed three times in 0.1M PO 4 buffer, and then most of the eye globe removed by using the 8 mm scissors to make a circumferential cut around the optic nerve head, leaving a small portion of the posterior eye attached to the dissected nerve. The nerves were post-fixed in gluteraldehyde/paraformaldehyde for 1 hr at room temperature, washed 3X in 0.1M PO 4 buffer, fixed in 1% buffered Osmuium for 1 hr on ice, and washed 3X with ice cold water. The following steps were all done at room temperature: nerves were dehydrated in an increasing series of MeOH (30%, 50%, 70%, 95%, 100%, 100%, 100%), then infiltrated with acetonitrile, 2X for 15minutes each, followed by 2 hours in a 1:1 mix of acetonitrile and EM Epon, and finally incubated in 100% EM Epon overnight, embedded in molds, and baked for at least 24 hours at 65uC. The blocks were trimmed to between 100-200 microns past the optic nerve head on a Leica RM2255 microtome, and ultra-thin sections were cut and plated on a grid, and imaged using a Hitachi H600 transmission electron microscope. The entire nerve cross-section was canvassed at 8000X, and 10-16 representative images were collected from each nerve at this magnification. Quantitative assessment of nerve pathology was conducted in a double-blinded manner in which both the TEM microscopist and the individual scoring pathology for the samples was unaware of the sample genotype.

Retinal flat-mount analysis
Eyes were dissected from terminally anesthetized adult fish and fixed overnight at 4uC in 4% PFA (pH 7.4, in PBS), then washed three times in PBS. In a Petri dish filled with PBS, a circumferential cut was made at front of the eye with a scalpel, near the border of the anterior and posterior segments. The anterior segment was discarded, followed by removal of the sclera from the posterior segment. The remaining retina with RPE was post-fixed 1-2 hrs with 4% PFA (pH 7.4, in PBS), washed in PBS, and then laid flat on a slide by making incisions through the retina so that it would lay flat. Whole retinas were mounted on the slides with 20 ml of Vectashield Mounting Medium (Vector Labs, Burlingame, CA), and coverslipped. For retinas used for antiactivated-caspase-3 immunofluorescence, antibody incubations were done after removal of the anterior segment and sclera, but prior to flat-mount analysis. Figure S1 A-G bugeye mutants show varying degrees of asymmetry in eye size (B-G), with the smaller of the two eyes only occasionally falling within the range of the wild-type eye size to body length ratio (E:B; wild-type E:B , 0.05, bugeye E:B .0.06). For each eye, the E:B is indicated. H In asymmetric eyes of bugeye fish, intraocular pressure (IOP) was elevated only in the enlarged eyes. In TL wild-type, the eye measured for IOP first had a higher value than the second eye measured in the same fish. Black symbols represent the first eye of a fish that was measured for IOP and gray symbols represent the second eye. The vertical bar connecting symbols shows eyes from the same fish. Circles represent normal sized eyes; squares represent enlarged eyes.