Ocular Abnormalities in Mice Lacking the Ski Proto-oncogene

Abstract

purpose. Persistent hyperplastic primary vitreous (PHPV) is a developmental ocular malformation often associated with additional ocular abnormalities. This study involved a novel mouse model of PHPV, generated by a null mutation of the Ski proto-oncogene, that displays other anterior segment and retinal malformations often found in human cases of PHPV.

methods. Morphologic and histologic analyses of Ski ^{− / −} mice were used to document ocular abnormalities in comparison to those of normal littermates. Immunohistochemical studies were used to examine the expression of relevant markers of ocular and vascular development including Pax6, β-III tubulin, and Flk1.

results. PHPV and microphthalmia were found in 100% of Ski ^{− / −} fetuses. Other abnormalities included anterior segment and lens dysgenesis, retinal folds, chorioretinal coloboma, and Peters anomaly. The severity was variable, even in a highly homogeneous genetic background. PHPV was characterized by the presence of retrolental fibrous and vascular tissue that did not express the neuronal marker β-III tubulin, but was positive for Flk1 expression and contained no obviously pigmented cells.

conclusions. The results show that normal ocular development requires the function of the Ski proto-oncogene, and mice lacking Ski have many features associated with PHPV, and some similarities with Peters anomaly in humans. Defects in Ski ^{− / −} mice closely resemble those described in animals lacking several of the retinoic acid receptor genes, or in animals exposed to excess retinoic acid during gestation. Ski has been shown to repress transcription induced by retinoic acid signaling, and may thus affect ocular development by regulating RA signaling.

The fetal vitreous tissue and vasculature begin to involute perinatally in the mouse, ^{1 2} and late during human fetal development. ³ Excessive residual or proliferating mesenchymal tissue in the vitreous is considered a developmental abnormality. Persistence of the hyaloid vascular system in its severe form can be accompanied by hyperplastic fibrous tissue, microphthalmia, retinal dysplasia, and lens degeneration. These abnormalities have been described in humans and in mouse models of PHPV. ^{3 4}  

The etiology of PHPV is not well understood, but several mouse models indicate that genetic abnormalities may play a role. Disruptions of retinoic acid signaling during embryogenesis also result in PHPV. Exposure to retinoic acid from embryonic day (E)7.5 to E11.5 can lead to many of the features characteristic of PHPV ^{5 6} ; conversely, mice lacking the β2 retinoic acid receptor (RAR), or compound RARβ2γ2 mutants, also exhibit persistence of the primary vitreous mesenchyme. ^{7 8} Mice lacking the p53 tumor suppressor gene show strain-specific persistence of the fetal vasculature, which is incorporated into retrolental fibrovascular tissue. ^{9 10} Mice lacking the Arf tumor-suppressor gene, which plays a role in stabilizing p53, also display ocular abnormalities reminiscent of severe PHPV. They have microphthalmia, fibrovascular retrolental tissue containing RPE cells and remnants of the hyaloid vascular system, and posterior lens capsule defects with lens degeneration and opacity and severe retinal dysplasia and detachment. ^{11 12} In addition, persistence of the hyaloid vasculature has been reported in Bmp4 ^{+/ −} and in TGF-β2 ^−/− mice, ^{13 14} as well as in mice in which apoptosis is impaired by the absence of both bax and bak. ¹⁵  

Originally discovered as a transforming gene in the avian Sloan-Kettering retroviruses, the v-Ski gene was shown to induce oncogenic transformation of chicken embryonic fibroblasts. ¹⁶ Its cellular homologue, c-Ski has been identified in several vertebrate species including chicken, ¹⁷ Xenopus, ¹⁸ zebrafish, ¹⁹ mouse, ²⁰ and human. ²¹ Ski expression is ubiquitous but is regulated in a tissue- and stage-specific manner, with elevated expression reported in differentiating skeletal muscles, migrating neural crest cells, and cells derived from the neural ectoderm. ^{20 22} The Ski gene family includes one other closely related member, Sno (or Skil). ²¹ Several studies have shown that Ski and Sno function as corepressors or coactivators, interacting with multiple transcription factors to regulate transcription. Both Ski and Sno interact with Smad2, -3, and -4 to repress signaling by TGF-β ^{23 24 25 26} and by BMP. ²⁷ Ski has also been shown to downregulate RA signaling by repressing RA-responsive transcription. ²⁸ Furthermore, recent evidence has shown that the Sno gene product SnoN can interact with p53 to regulate transcription. ²⁹ Therefore, many of the pathways reported to result in persistence of the fetal hyaloid vasculature are subject to regulation by Ski. 

Given the wide range of functional interactions displayed by the Ski protein, it is not surprising that mice lacking Ski show highly pleiotropic defects. Of interest is that many of the developmental abnormalities occur in regions that show elevated Ski expression during development, including neural crest derivatives, skeletal muscles, and neural structures in both the central and peripheral nervous systems. The salient phenotype of homozygous Ski-deficient mice is a cranial neural tube defect accompanied by excessive apoptosis. ³⁰ Ski ^−/− mice also show a dramatic reduction in skeletal muscle mass as well as skeletal patterning defects. The penetrance and expressivity of the phenotypes have been found to be highly strain dependent, and change from a neural tube defect to midline facial clefting when the Ski ^−/− mutation is backcrossed into the C57BL6/J background. ³¹ Other defects, including depressed nasal bridge, digital abnormalities, skeletal muscle defects, and eye abnormalities show increased penetrance in the C57BL6/J background. 

Similarities have been reported between the phenotypes observed in Ski-null mice and the defects in patients with 1p36 monosomy (deletion syndrome), in which SKI appears to lie within the minimal commonly deleted region and may be responsible for some features of the syndrome. ³¹ A complete description of the phenotypes associated with loss of Ski function in mice may therefore be important in identifying other human syndromes that could involve loss of SKI function. The specific eye abnormalities in Ski-deficient mice have not been reported in detail. In this study, we describe the complex ocular malformation phenotype in mice lacking the Ski gene, and show that it reflects many of the clinical findings associated with PHPV, as well as other anterior segment and retinal malformations that can be associated with PHPV. 

Materials and Methods

Mouse Genotyping and Genetics

The Ski-null allele was generated on a 129P2 genetic background (E14 ES cells ³¹ ), and maintained by backcrossing to C57BL/6J females for 10 to 15 generations, followed by backcrossing to C57BL/6J males for two generations; alternatively, the mutation was also maintained by backcrossing to 129S6 males for six to eight generations. Mice used to generate the (129S6 X C57Bl/6J) F2 animals were N6 (to 129S6) or N12 (to C57Bl/6J). Southern blot analysis and/or PCR were used for genotyping as described previously, ³¹ using the following primers: forward 5′AGGGGAGACCATCTCTTGTTTC 3′ (starting at codon 104 of exon 1), and reverse: 5′GACTTTGAGGATCTCCAGCTGG 3′ (starting at codon 161 of exon 1). These produce a 171-bp fragment from the wild-type allele and a 1277-bp fragment from the mutant allele. 

Morphology and Histology

Embryos or fetuses were collected from timed matings (day of plug = 0.5); the uteri were placed in cold PBS, and the fetuses were euthanatized by decapitation. ^{32 33} They were then evaluated for morphologic abnormalities, and genotyped retrospectively, as described earlier herein. Embryos or fetuses were processed for histologic analysis as described previously. ³⁰ Briefly, they were fixed for 6 to 12 hours in phosphate-buffered 3.7% formaldehyde, dehydrated through a graded ethanol series, and embedded in paraffin. Four-micron sections were used for hematoxylin and eosin (H&E) staining, immunostaining, or in situ hybridization. 

Immunohistochemistry and In Situ Hybridization

Antibodies included monoclonal antibodies against Pax-6 (1:250; Developmental Studies Hybridoma Bank, Iowa City, IA), β-III tubulin (1:400; Covance, Vienna, VA), Flk-1 (1:50; Research Diagnostics, Flanders, NJ), Tie-2 (1:80; Santa Cruz Biotechnology, Santa Cruz, CA), and CD31 (1:200; Santa Cruz Biotechnology). The sections were incubated with primary antibodies for 2 hours at room temperature (Pax6, β-III tubulin) or overnight at room temperature (Flk1, Tie2, CD31) in humidified chambers, followed by extensive washing. Secondary antibody staining and detection were performed (VectaStain Elite or ABC kit; Vector Laboratories. Burlingame, CA), and the sections were counterstained with fast green or hematoxylin. 

Animal Use

These experiments were performed according to the institutional animal care and use committee (IACUC)–approved protocol, with strict adherence to the ARVO Statement for the use of Animals in Ophthalmic and Vision Research and the NIH Guide on the Care and Use of Laboratory Animals. 

Results

Morphologic Observations

To examine the ocular phenotype of Ski-deficient mice, we compared the eyes of nine Ski ^−/− animals from nine separate litters with those of Ski ^{+/ −} littermates at E16.5 to E18.5. The mutant eyes showed several abnormalities. 

General

The size of the globe overall was consistently smaller in the Ski knockout mice when compared with the control mice. Size was evaluated as shown in 1 Figures 2A and 2B , by measuring the greatest anteroposterior diameter of each eye. By this criterion, the size of the knockouts’ eyes ranged from 80% of normal, to as little as 50% to 60% of normal (Figs. 1A 1B , Figs. 2A 2B ). The iris and pupil were invariably misshapen, although the expressivity of this phenotype was variable, even in a homogeneous, C57BL/6J background. One embryo had evidence of a small cyst located behind the globe (Fig. 2C) . 

Abnormalities were always bilateral but usually asymmetric in severity. 

Lids.

Adhesions between the lids and sclera or cornea were observed in both mutant and normal fetuses, but were more pronounced in mutant, Ski ^−/− eyes (Figs. 3C 3D) . Many eyelids in Ski ^−/− pups remained open (Figs. 3A 3E 4A)well past E16.5, when the lids were fully closed in their normal littermates. 

Anterior Segment.

In several embryos, there was apposition of the lens to the back of the cornea, with poor angle formation (Fig. 3E) , and there was absence of an anterior chamber proper (Figs. 3A 3C 3D 3E) . Adhesions of the lens to the corneal endothelium (Figs. 3A 3C 3D 3F)were common, and in some embryos the corneal epithelium appeared to be continuous with the lens epithelium (Fig. 3B) . Many embryos had irregularly shaped lenses (Figs. 3A 3D 3E 3F) ; most lenses were generally smaller than in the control eyes and in some cases partially were absorbed (Figs. 3E 3F)or even nonexistent (Figs. 4A 4B) . Some lenses had irregular epithelial layers or had vacuoles and cataract formation (Figs. 3B 3D) . 

In some of the mutant fetuses, formation of the inner epithelial layer at the anterior margin of the optic cup, which would be necessary for the differentiation of the iris and ciliary body, was abnormal in some eyes (Figs. 3D 3F ; 5 6A 6C , curved arrows). In these cases, the retina did not thin down to a double epithelial layer. This defect would be consistent with abnormal formation of the iris, which could result in partial or complete aniridia. 

Vitreous.

Excessive fibrovascular retrolental tissue was observed in the knockout mice, occupying the space directly behind the lens (Fig. 2A)and sometimes extending posteriorly to appose the retina and optic nerve head (Figs. 3A 3D) . This fibrovascular hyperplasia was never present in heterozygous or wild-type littermates. 

The abnormal vitreous tissue included no obviously pigmented cells. 

Retina.

Colobomas were present with some dysplastic changes in the area of the ciliary body and ciliary processes (Fig. 3F , arrowhead). Many of the Ski-deficient embryos showed retinal folds (Figs. 3E 3F 5) . Abnormalities were present that were interpreted as chorioretinal colobomas, in which the retina and choroid appeared fused, with hyaloid remnant tissue extending from that area to the center of the globe (Figs. 3E 3F) . 

Eye Muscles.

No significant changes were observed in the extraocular muscles (not shown). 

Penetrance and Expressivity of Ocular Abnormalities

A summary of the incidence of abnormalities documented is presented in Table 1 . Major defects observed consisted of microphthalmia (100%), hyperplastic hyaloid vasculature (100%), retinal folds and abnormalities (78%), chorioretinal coloboma (78%), and cornea-lens adhesions (44%). Some Ski ^−/− mice also had open lids, lens abnormalities including microphakia and cataracts, iris coloboma, and anterior segment dysgenesis, but these defects were less penetrant. 

As previously reported, the penetrance of ocular defects in Ski ^−/− mice was high in the C57Bl/6J genetic background at N8 and N12, with 5 out of 5 mice showing microphthalmia, PHPV, and coloboma. However, these abnormalities were also observed in 4 mice on a mixed, C57Bl/6J and 129S6 genetic background (Table 1) . Defects were always observed bilaterally, but the severity of the defects frequently differed, and either side could be more or less severely affected. 

Expression of Informative Cell Markers

Pax6 is a key regulator of ocular development, and mutations in Pax6 have been reported to result in multiple abnormalities in the mouse, including colobomatous microphthalmia, aniridia, PHPV, and anterior segment dysgenesis, whereas in humans, PAX6 mutations result in aniridia in most patients and in Peters anomaly and PHPV in only a few cases. ^{34 35} We found that expression of Pax6 was irregular and patchy in the mutant eyes compared with those of normal littermates (Fig. 6) . Within the anterior margin of the optic cup, in the region containing the precursors to the iris and ciliary body, fewer cells expressed Pax6 (for example, Fig. 6A 6C , white arrowheads), while retinas also had regions of uneven Pax6 expression which were not observed in the eyes of Ski ^+/− or ^+/+ littermates (Figs. 6B 6D) . 

To determine the nature of the abnormal vitreous tissue, we used an antibody to β-III tubulin, which is expressed after neuronal differentiation in the retina, but not in the RPE. ³⁶ We also used antibodies that detect markers of vascular and endothelial cells, including Flk1, Tie2, and CD31. The retrolental tissue did not show expression of β-III tubulin (Figs. 7A 7B) . In addition, expression of the neuronal marker in the inner retinal layer was decreased compared with the normal eyes of heterozygotes. The retrolental tissue did not express CD31, but was positive for Flk1/VEGFR2 expression (Figs. 7D 7E) . Tie2 expression was observed only in close association with blood vessels within the fibrovascular tissue (data not shown). 

Discussion

We present the first description of the developmental ocular defects that result in the mouse from loss of function of the Ski proto-oncogene, which encodes a transcriptional cofactor. Previous studies have indicated that Ski plays a complex role in the development of the mouse nervous system, ^{30 37} and our findings now show that Ski is also required for normal eye development. This result is consistent with previous data showing that expression of Ski is ubiquitous and is relatively upregulated during development in many cell types derived from the neural ectoderm. ²² Our studies provide additional insight into the molecular and cellular mechanisms underlying the pathogenesis of the ocular defects described, which are analogous to congenital abnormalities that occur in humans. 

One of the most penetrant defects in the Ski ^−/− eyes is the hyperplasia of the embryonic hyaloid vasculature, which in the normal mouse begins to regress shortly before birth. ^{3 38} Ski ^−/− fetuses had a marked excess of tissue in the vitreous compared with their normal littermates, and we propose that Ski-deficient mice are predisposed to the development of a condition analogous to persistent hyperplastic primary vitreous (PHPV), also termed persistent fetal vasculature (PFV) in humans. ³ This condition occurs in the context of a generalized abnormal ocular development, and eyes with PHPV frequently have other abnormalities such as microcornea, Peters anomaly, cataracts, chorioretinal colobomas, and microphthalmia. ³ We observed all these abnormalities in Ski-null mice, although progression of these defects could not be documented in adult mice, as the Ski-null mutation is lethal shortly after birth. ³⁰  

The hyaloid vessels, which include the hyaloid artery and tunica vasculosa lentis, are a normal part of the developing mouse (and human) eye and nourish the developing lens at a time when aqueous production and formation of the anterior chamber have not yet begun. ³ At E16.5 in the normal mouse fetus, the vessels that occupy the vitreous cavity begin to regress and become less prominent, except along the lens and inner retinal surface. ^{39 40} Regression of these vessels has been shown to result from the apoptosis that is triggered by macrophages that invade ocular tissues during development. ^{1 2} In agreement with these data, mice deficient in several proapoptotic genes including bax, bak, and p53 also show persistence of the fetal vasculature. ^{10 15} Excessive apoptosis has been reported in the Ski knockouts in the neural tube and craniofacial mesoderm ^{30 41} ; however, in this study we found no differences in apoptosis between the Ski knockouts and their normal littermates at several stages of ocular development (data not shown). 

The absence of β-III tubulin expression indicates that the retrolental, hyperplastic vitreous tissue is of mesenchymal rather than neural origin. As previously shown, ^{42 43} we found this tissue to express Flk1, a receptor for vascular endothelial growth factor (VEGF), suggesting that VEGF signaling may be involved in the fibrovascular hyperplasia. Although Flk1 expression has also been reported in retinal precursors, ⁴⁴ the absence of neuronal tubulin strongly suggests that the hyperplastic retrolental tissue observed in the knockout eyes is of mesodermal origin. However, the absence of expression of additional endothelial cell markers, such as CD31, suggests that the retrolental tissue present in the Ski knockouts may represent an immature, endothelial precursor-like cell, in which VEGF signaling has not yet induced complete endothelial cell differentiation. ⁴⁵ It is tempting to speculate that this incomplete endothelial cell phenotype could result from aberrant TGFβ or RA signaling, both of which have been shown to be modulated by Ski and to regulate VEGF signaling negatively. ⁴⁶  

The lens completely separates from the surface ectoderm, which develops into the cornea, by E15.5 to E16. At this time, the cornea has a homogeneous structure with an inner limiting layer destined to become Descemet’s membrane and an outer layer of endothelial cells. In humans, the presence of a central corneal leukoma, absence of the posterior corneal stroma and Descemet’s membrane, and a variable degree of iris and lenticular attachments to the central aspect of the posterior cornea are combined to characterize Peters anomaly. ⁴⁷ Forty-four percent of Ski-null mice showed evidence of lens–corneal adhesions, poorly formed angles suggesting anterior segment dysgenesis, and iris hypoplasia that we believe to be analogous to Peters anomaly. As mentioned previously, Peters anomaly is associated with PHPV and could be the result of decreased expression of developmental genes that are important for anterior segment development. ⁴⁷ Some cases of Peters anomaly have been found to result from mutations in PAX6, ³⁵ a gene shown to play a major role in ocular morphogenesis. ^{35 48}  

Ski-null mice showed a decrease in cells expressing Pax6 in the ciliary body, as well as patchy expression of Pax6 in the outer layer of the retina. This aberrant expression may be a secondary defect resulting from abnormal distribution of Pax6-positive cells, but it may also be related to the observed anterior segment abnormalities, ³⁵ based on the phenotypes of hypomorphic Pax6 alleles in both mice and humans. Currently, no direct interaction between Ski and Pax6 has been reported, but our results here suggest that an indirect functional relationship is possible. These observations are consistent with our interpretation of the anterior segment defects as corresponding to Peters anomaly. 

Chorioretinal colobomas and microphthalmia, which were present in 100% of Ski ^−/− mice, are rather nonspecific and can be associated with several mouse and human diseases and genetic abnormalities. Retinal folds and dysplasia have previously been found in eyes with PHPV. ³ Their presence in this case further strengthens the similarity between the human and mouse phenotypes. 

It is of interest to note that severe PHPV was found in p53-null mice on the C57BL/6J background, but not in other strains ⁹ ; by contrast, the incidence of exencephaly in p53-null mice is higher in the 129 background. ⁴⁹ Similarly, Ski-null mice show much lower penetrance of exencephaly, and higher penetrance of ocular defects, in the C57BL/6J background, suggesting that similar modifiers may affect the activity of Ski and p53. In this regard, recent work has shown that interaction between the SnoN gene product and p53 is essential for appropriate regulation of gene expression by p53. ²⁹ Because SnoN and Ski are closely related and have overlapping functions, these data suggest that Ski may also be involved in transcriptional regulation by p53. Therefore, the absence of Ski may disrupt p53 functions that result in the PHPV-like phenotype we have described. 

The Arf tumor-suppressor gene, a positive regulator of p53, is also required for hyaloid vascular regression in the mouse eye during development ¹² ; however, its role in this context has recently been shown to be independent of p53 and Mdm2. ⁵⁰  

Alterations in signaling via components of two other pathways shown to be subject to regulation by Ski has also been found to result in persistence of the fetal vasculature: TGFβ/BMP signaling, and retinoic acid (RA) signaling. Both knockout mice lacking Tgfβ2, and heterozygotes carrying a targeted Bmp4 allele, have been reported to have excessive vitreous vasculature, and a Peters-like anomaly or anterior segment dysgenesis, respectively. ^{14 51} However, Ski has been reported to act as a corepressor of both TGFβ and BMP signaling, ^{23 24 25 26} and so loss of function of Ski would be predicted to result in a net gain of function of TGFβ/BMP signaling. Therefore, this pathway may not be involved in the ocular abnormalities described in Ski ^−/− mice. The case for a potential effect on RA signaling is stronger, since both gain and loss of function of RA signaling can induce PHPV. Compound null mutants lacking several combinations of two of the RA receptors were found to have ocular abnormalities that included microphthalmia, retinal coloboma, corneal and eyelid abnormalities, anterior segment dysgenesis, and lens agenesis with low penetrance. ^{7 8} Although these represent significant overlaps with the Ski-null ocular phenotype, the expected outcome of a loss of Ski corepression ²⁸ would again be a gain-of-function abnormality. In this case, exposure to excessive RA around midgestation, between E7.5 and E11.5, results in PHPV, microphthalmia, treinal coloboma, and a Peters-like anomaly involving lack of separation between the corneal and lens epithelia. ⁶ Therefore, the Ski-null mutation could most likely affect either RA signaling, or p53-dependent gene expression, or both. 

In 2002, Colmenares et al. ³¹ reported the human Ski gene to be located at distal 1p36.3. Furthermore, the study found the Ski gene to be located within the commonly deleted region in patients with 1p36 monosomy and suggested that Ski may contribute to some of the abnormalities common to this syndrome. Anomalies associated with 1p36 monosomy include cleft lip and/or cleft palate, large anterior fontanelle, hypotonia, moderate to severe mental retardation, seizures, growth delay, pointed chin, eye and vision problems, deep-set eyes, hearing deficits, low-set ears, clinodactyly, flat nasal bridge, abusive behavior, and thickened ear helices. ^{52 53} Specific eye and vision abnormalities found in 1p36 deletion syndrome are strabismus, sixth-nerve palsy, amblyopia, hyperopia, myopia, astigmatism, anomalous optic disks, and lacrimal defects. ⁵² Some of these conditions could be caused by ocular muscle abnormalities; however, these were not observed histopathologically in Ski-deficient mice. 

Our present studies suggest a complex role for Ski in ocular development. The facts that Ski acts as a transcriptional cofactor, and has been shown to modulate expression of many genes regulated by key signaling pathways in growth and development, provide an embarrassment of riches in the search for physiologically relevant targets. If several of these potential target pathways are indeed affected in the Ski knockouts, then the complex cross-talk among signaling pathways may be the likely explanation for the highly variable penetrance and expressivity of the observed phenotypes. Future investigations will focus on the interactions between Ski and Pax6 in relation to retinoic acid signaling, between Ski and p53, and on the effect of Ski on expression of targets of TGFβ/BMP signaling. 

 Supported by National Institute of Dental Research Grant DE15198 (CC).

 Submitted for publication December 4, 2005; revised May 3 and June 6, 2006; accepted August 3, 2006.

 Disclosure: P. McGannon, None; Y. Miyazaki, None; P.C. Gupta, None; E.I. Traboulsi, None; C. Colmenare s, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

External morphology of the eyes in Ski ^{− / −} fetuses. (A) Appearance of the eye at E15.5 in a Ski-null mouse and its normal littermate. At this stage, the lids are just beginning to close, and the optic fissure has closed in the heterozygous fetus. (B) Iris and pupil abnormalities and the degree of microphthalmia differed in severity in E18.5 fetuses. Magnification, ×20. (A and part of B reprinted with permission from Colmenares C, Heilstedt HA, Shaffer LG, et al. Loss of the SKI proto-oncogene in individuals affected with 1p36 deletion syndrome is predicted by strain-dependent defects in Ski ^−/− mice. Nat Genet. 2002;30:106–109. http://www.nature.com. © Nature Publishing Group.)

Figure 2.

 

Microphthalmia and cyst. Histologic sections of eyes stained with H&E. Serial sections were scanned and a section selected in the region of the greatest anteroposterior diameter. Micrographs were taken at identical magnification, and the blue bars are exact replicas of each other. (A, B) Ocular sections from littermates at E15.5; arrows: vascular tissue in the vitreous, which is hyperplastic in the absence of Ski. (C) Section shows the ventral aspect of the eye in E16.5, Ski ^−/− fetus; arrowhead: a cyst. Magnification: (A, B) ×100; (C) ×64.

Figure 3.

 

Anterior segment abnormalities. (A, B) Both eyes from a Ski-null fetus at E18.5 show prominent hyperplastic vitreous tissue. Arrowheads: cornea–lens adhesions; black arrow: open eyelid; white arrow: hyperplastic vitreous tissue occupying the entire vitreous cavity. (C, D) Eyes from two different Ski-null fetuses at E16.5. Arrowheads: lens adhesions; white arrow: hyperplastic primary vitreous. (D) A combination of PHPV and corneal–lens adhesion. (E, F) Eyes from two different Ski-null fetuses at E18.5 (different magnifications) showing anterior segment abnormalities and severe retinal dysplasia. Arrowheads: colobomas; white arrows: retinal folds; black arrows: open eyelids; curved arrows: abnormal iris. Magnification: (A, E) ×40; (B, D) ×100; (C, F) ×200.

Figure 4.

 

Lens abnormalities. (A, B) Eyes from two different Ski-null fetuses at E18.5. Serial sections from the entire eye were examined in each case, and no evidence of a lens was observed. Both eyes also show dysplastic retinas. Some degenerating lens material remains. Magnification: (A) ×100; (B) ×200.

Figure 5.

 

PHPV, coloboma (arrowhead), and retinal fold. Magnification, ×200.

Figure 6.

 

Pax6 expression. (A, B) Sections of eyes from Ski-null and heterozygous littermates, respectively, stained with antibody to Pax6 and counterstained with fast green. Distribution of Pax6-positive cells in retinas of Ski-null fetuses is patchy and uneven (white arrowheads). (C, D) Higher magnification showing boxed areas in (A) and (B). In the Ski-null eye, note decrease in the number of Pax6-positive cells in the anterior margin of the optic cup, in region that will form the iris and ciliary body. In this region there is no maintenance of the inner epithelial layer that will form the non-pigmented epithelium of the ciliary processes (curved arrow). Magnification: (A, B) ×64; (C, D) ×400.

Table 1.

 

Incidence of Ocular Defects in Ski-Deficient Mice of Homogeneous or Mixed-Strain Genetic Background

Strain	Microphthalmia	Lens/Corneal Adhesions	PHPV	Coloboma	Retinal Folds
C57BL/6J (n = 5)	5/5	3/5	5/5	5/5	4/5
C57BL/6JX 129S6 F2 (n = 4)	4/4	1/4	4/4	2/4	3/4
Total incidence	9/9	4/9	9/9	7/9	7/9

Figure 7.

 

Expression of endothelial and neural cell markers. (A–C) Sections of eyes from littermates at E18.5 stained with antibody to β-III tubulin, a marker of neuronal differentiation, and counterstained with fast green. The inner layer of the retina shows β-III tubulin expression in both Ski-null and heterozygous eyes. However, the vitreous fibrovascular tissue shows no expression. (C) Staining of collagenous material in the vitreous is nonspecific, as shown by the control stained with no primary antibody. (D) Staining of retrolental fibrous tissue with antibody to Flk1, a marker of endothelial cells, indicates endothelial origin. (E) No primary antibody control. Magnification: (A–C) ×100; (D, E) ×400.

The authors thank John Porter for the analyses of ocular muscles and Joe Hollyfield and Ed Stavnezer for insightful suggestions on the manuscript.