Varied manifestations of persistent hyperplastic primary vitreous with graded somatic mosaic deletion of a single gene

• Introduction
• Methods
• Results
• Discussion

Abstract

Purpose: Persistent hyperplastic primary vitreous (PHPV) represents a developmental eye disease known to have diverse manifestations ranging from a trivial remnant of hyaloid vessels to a dense fibrovascular mass causing lens opacity and retinal detachment. PHPV can be modeled in mice lacking individual genes, but certain features of such models differ from the clinical realm. For example, mice lacking the Arf gene have uniformly severe disease with consistent autosomal recessive disease penetrance. We tested whether the graded somatic loss of Arf in a subset of cells in chimeric mice mimics the range of disease in a non-heritable manner.

Methods: Wild type ↔ Arf ^−/− mouse chimeras were generated by morulae fusion, and when the mice were 10 weeks old, fundoscopic, slit-lamp, and histological evaluations were performed. The relative fraction of cells of the Arf ^−/− lineage was assessed with visual, molecular genetic, and histological analysis. Objective quantification of various aspects of the phenotype was correlated with the genotype.

Results: Sixteen chimeras were generated and shown to have low, medium, and high contributions of Arf ^−/− cells to tail DNA, the cornea, and the retinal pigment epithelium (RPE), with excellent correlation between chimerism in the tail DNA and the RPE. Phenotypic differences (coat color and severity of eye disease) were evident, objectively quantified, and found to correlate with the contribution of Arf ^−/− cells to the RPE and tail-derived DNA, but not the cornea.

Conclusions: Generating animals composed of different numbers of Arf ^−/− cells mimicked the range of disease severity observed in patients with PHPV. This establishes the potential for full manifestations of PHPV to be caused by somatic mutations of a single gene during development.

During ocular development, the hyaloid vascular system (HVS) in the primary vitreous nourishes the developing vitreous and lens [1,2]. The HVS regresses in later stages of development to leave the largely acellular mature vitreous space. Vision is compromised when the process goes awry in certain ocular diseases such as persistent hyperplastic primary vitreous (PHPV), also known as persistent fetal vasculature (PFV) [1,3]. PHPV treatment is largely limited to surgical intervention (such as lensectomy, vitrectomy, membranectomy) to prevent late sequelae such as glaucoma or phthisis, but these approaches have less capacity to restore vision when the disease is severe [4,5]. Relatively little is known about the molecular pathogenesis of PHPV/PFV. The rare occurrence of familial disease has implied a genetic cause [6-8], and certain molecular genetic abnormalities have been correlated with cases of PHPV that are accompanied by other developmental defects [9-11].

To gain more insight into possible disease-causing aberrations, we and others have leveraged mouse models to elucidate the fundamental processes driving primary vitreous maturation and HVS regression. It is clear that disrupting cellular or genetic mechanisms that promote endothelial cell apoptosis can lead to persistence of the hyaloid vessels. Examples include loss of Wnt7b signals in macrophage-like hyalocytes [12], and the absence of Angiopoietin 2 [13,14] or p53 [15,16], which may be linked to endothelial cell apoptosis in the postnatal period [16]. HVS regression is also blocked by hyperplasia of perivascular cells in the primary vitreous. This occurs in mice with deficient Tgfβ signaling [17-19], ectopic expression of the pseudorabies virus protein IE180 [20], deregulated vascular endothelial growth factor (Vegf) [21], and the absence of the Arf tumor suppressor. In this last model, the Arf gene product, p19^Arf, blocks proliferation signals stemming from Pdgfrß [22,23]. The resulting hyperplastic perivascular cells provide trophic factors, such as Vegf, which may impede the HVS involution [24].

We have focused on the Arf model of primary vitreous hyperplasia because Arf is normally expressed in perivascular cells enveloping the mouse HVS, and the robust phenotype recapitulates many aspects of severe PHPV [22,24]. Human ARF abnormalities have not yet been described in patients with PHPV, and two features of the human disease are not reflected in the Arf ^−/− model. First, disease penetrance is essentially complete in the mouse, clearly tracking as an autosomal recessive trait. Second, uniformly severe disease develops in both eyes in Arf ^−/− mice, whereas human disease manifestations vary widely and are often unilateral [1,3]. We considered that if the disease were due to a somatic mutation of a critical gene in only a subset of cells in development, it might not be heritable. Further, the disease manifestations might also vary, perhaps in proportion to the number of cells lacking the critical gene. We previously showed that mosaic mice composed of a mixture of Arf wild type and Arf ^−/− cells displayed primary vitreous hyperplasia at birth, proving that p19^Arf uses primarily cell-intrinsic mechanisms to control accumulation of Pdgfrß-expressing cells in the primary vitreous [25]. Thus, most of the cells that populate the PHPV tissue at birth are drawn from the Arf ^−/− pool. We conducted the current analyses to test whether the somatic mosaic loss of Arf could lead to the full “clinical” manifestations of PHPV and whether the relative proportion of Arf-deficient cells in the chimeric animal correlated with disease severity.

Fused morulae, implanted in the uteri of four pseudopregnant female mice resulted in 16 wild type ↔ Arf ^−/− chimeric pups and nine Arf ^−/− control weaned pups. Visual inspection of the coat revealed four pups with mostly white coats, five with mostly agouti coats, and seven judged to have approximately equal mixing of coat color (Figure 1A-C). Because the wild type mice and Arf ^−/− mice had white and agouti coats, respectively, these data indicated that the Arf ^−/− contribution to these chimeras ranged from relatively low to high.

A semiquantitative PCR assay (see Methods) assessed the relative ratio of wild type and Arf ^−/− alleles in genomic DNA derived from the tail biopsy samples. Because the wild type allele was amplified more efficiently (Figure 1D, lane 2), the relative intensity of each band was normalized to Arf ^+/− heterozygous DNA, which was amplified in parallel. Quantitative analysis from the 13 chimeric animals with measurable DNA indicated that the contribution of wild type and mutant Arf ^−/− cells in the tail-derived DNA ranged from nearly all wild type (Figure 1E, lane D) to nearly all Arf ^−/− (Figure 1E, lane F).

We next quantified the relative contribution of wild type and Arf-deficient cells to individual tissues of the eye. Direct green fluorescence was used to assess chimerism in the cornea. In this case, the total length of the corneal epithelium was determined using 4',6-diamidino-2-phenylindole dihydrochloride (DAPI)–stained images (Figure 2B,D,F). The total length of green fluorescence versus non-green fluorescence (Figure 2A,C,E) was calculated to determine the relative number of wild type and Arf ^−/− cells, respectively. To assess the RPE, we took advantage of the fact that the Arf wild type mice were albino. Hematoxylin and eosin stained eyes showed regions of the RPE that had or lacked pigment (Figure 2G,H, arrowhead versus arrows) and therefore were derived from the Arf ^−/− or wild type lineage, respectively. The relative linear distance of pigmented versus non-pigmented RPE was calculated from these images. Midline sections were of sufficient quality to allow quantitative analysis of 16 eyes taken from nine animals (Figure 2I). The degree of chimerism in the cornea and the RPE varied widely across the cohort, as it did in the tail-derived DNA and coat color analysis. However, the intraclass correlation coefficient for the RPE was 0.92 and for the cornea was 0.77, suggesting that variation between two eyes in the same mouse was less than the mouse-to-mouse differences.

We previously showed that the wild type and Arf ^−/− contribution in tail-derived DNA, cornea, and RPE correlate with each other in newborn pups [25]. Similarly, linear regression analyses showed that the Arf ^−/− lineage contribution to the RPE correlated well with its contribution to tail-derived DNA (p<0.001) (Figure 3A,D) but less well when the tail-derived DNA (p = 0.014) and the RPE (p = 0.010) were compared to the cornea (Figure 3B,C,E,F).

We also tested whether the degree of chimerism correlated with various aspects of the mouse phenotype. We first compared the relative Arf ^−/− content to the coat phenotype. Regardless of whether it was based on genomic DNA from the tail biopsy or the assessment of the RPE, the average Arf ^−/− content increased step-wise in mice that were judged to have mostly white, mixed, or mostly agouti coats (Figure 4A,B). In contrast, coat color did not correlate with chimerism based on the cornea analysis (Figure 4C; also see summary in Table 1).

As with the coat color assessment, ocular disease manifestations varied across the cohort of chimeric mice (Figure 5). Disease manifestations were felt to be of low, moderate, and high severity based on slit-lamp and fundoscopic evaluations, but all chimeric mice had some evidence of disease, based on the presence of a retrolental membrane on the slit-lamp exam or visual evidence of remnants of the hyaloid vasculature in the vitreous on the fundus exam (Figure 5B). In some cases, the vitreous membrane was darkly pigmented whereas, in other cases, pigmentation was minimal or absent on “clinical” examination. The slit-lamp exam findings ranged from the lens being essentially normal in two eyes to having the posterior surface completely covered by a retrolental membrane. The fundus findings ranged from observing small remnants of the hyaloid vessels to the presence of a retrolental mass fully obscuring the fundus. Disease severity assessed with the slit-lamp exam correlated with the severity on the fundoscopic exam in 25 of 26 eyes. Further, disease scoring was concordant in both eyes of 11 of 13 (slit-lamp) and 12 of 13 (fundoscopic) animals.

The severity of eye disease was also assessed microscopically by scoring the midline sections for the presence or absence of a retrolental mass and by measuring the size of the mass (Figure 6). We detected a retrolental mass in the midline sections of eyes from all Arf ^−/− animals, whereas a mass was detected in only 62.5% of the chimeric eyes. We noted that 11 of 12 retrolental masses from chimeric mice contained densely pigmented cells, indicating their derivation from the non-albino, Arf ^−/− lineage (Figure 6C-F). None of the eyes with low disease severity on the fundus exam had pathological evidence of a retrolental mass. In contrast, 67% and 100% of the eyes judged to have moderate or severe fundoscopic evidence of disease, respectively, had a retrolental mass on histological assessment. The average size of the retrolental mass was smaller in the chimeric animals as compared to the Arf ^−/− eyes (Figure 6J).

Last, we determined whether the severity of ocular disease correlated with the contribution of Arf ^−/− cells in the animal. The quantity of Arf ^−/− cells in the tail-derived DNA and RPE correlated with the ocular disease severity based on either fundoscopic or slit-lamp exams in chimeric mice (Figure 7A,B; additional data not shown). Greater Arf ^−/− contribution to these tissues also correlated with the presence of a larger retrolental mass size on histological analysis (Figure 8A,B). The roughly linear relationship between the retrolental mass size and Arf ^−/− contribution to these tissues speaks to the cell-intrinsic mechanism by which p19^Arf controls the accumulation of primary vitreous cells [25]. However, differences in Arf ^−/− content in the cornea did not correlate with ocular disease severity (Figure 7C) or with larger retrolental mass size (Figure 8C; also see summary in Table 2).

In general, PHPV pathogenesis centers on either the disruption of proapoptotic pathways needed to directly promote endothelial cell regression in the HVS or the override of mechanisms for preventing hyperplasia of perivascular cells enveloping the hyaloid vessels. Numerous mouse models support the concept that this disease has a genetic basis [12,15-19,21,22,24], and in most, the disease exhibits an autosomal dominant or recessive inheritance pattern with a phenotype penetrance that can be complete. In that regard, though, such models fail to reflect certain clinical features of human PHPV, including the fact that it is only rarely reported to be heritable [6-8,29].

In the case of Arf loss as a mouse model for PHPV, Pdgfrß-dependent signals drive excess proliferation of vascular mural cells enveloping the hyaloid stalk and vasa hyaloidea propria in the posterior eye [23,24]. Thus, this model principally models primary vitreous hyperplasia that derails the involution of only the posterior components of the hyaloid vasculature. The usual clinical scenario for patients with PHPV, though, contrasts three features of the Arf ^−/− model: Ocular disease in the Arf ^−/− mouse is heritable with complete penetrance, it is consistently bilateral and severe, and it is associated with cancer susceptibility. The incongruence between the clinical specimen and this model can be mostly eliminated if we consider that PHPV might be caused by somatic loss of human ARF. First, although we did not formally test whether the PHPV-like phenotype is heritable in the mouse chimeras we generated, we recently demonstrated that mouse Arf deletion in cells derived from the Wnt1-expressing neural crest results in PHPV [30]. Thus, although there is as yet no evidence from clinical specimens, one could envision that human ARF disruption in just the neural crest might be sufficient for PHPV. As a somatic mutation, the germ line could remain unaffected, and the disease would not be heritable.

Second, by varying the number of Arf ^−/− cells in the chimeric mouse, we recapitulated the wider spectrum of disease that occurs in the clinical setting. In the 26 chimeric eyes we examined “clinically,” nearly one third had minimal evidence of disease. In some cases, the “clinical” exam findings were essentially normal or displayed only a small retrolental membrane and persistence of the hyaloid stalk exactly modeling milder forms of the disease. Beyond demonstrating that disease severity varied with the relative number of Arf-deficient cells, we also occasionally generated chimeric mice with little or no detectable disease in one of the eyes, which may reflect unilateral distribution of Arf ^−/− cells in the fused morulae. One could also envision unilateral defects due to mitotic recombination later during embryogenesis in a human heterozygous “carrier” of a single mutant ARF allele. Mouse Arf is expressed within the eye from the point when HVS vessels are initially forming (approximately E12.5), and the expression persists until the vessels begin to regress in the first postnatal week [22,24]. Thus, this would represent the developmental window in which disrupting ARF or its expression could cause unilateral disease, perhaps with even less severe manifestations.

Third, one might expect cancer susceptibility if somatic ARF deletion contributes to PHPV, but an association between PHPV and cancer has not been described. It is widely accepted that p19^Arf suppresses tumor formation by exerting cell-intrinsic control over cell proliferation and apoptosis in nascent cancer cells [31,32], and we proved that this protein controls the accumulation of Pdgfrß-expressing cells in the primary vitreous in a cell autonomous manner [25]. As has been observed in individuals with somatic deletion of other tumor suppressors, including p53 [33], human ARF ^−/− cells should be at risk for neoplastic growth in a chimeric setting. However, two important points should be noted. First, there are few reports of ARF haploinsufficiency in the human germ line, and ARF disruption is often accompanied by deletion of the overlapping INK4A tumor suppressor gene [34-36]. Second, it is commonly felt that INK4A has greater tumor suppressor activity than ARF in humans [37]. Even considering disruption of both genes, germ line haploinsufficiency at this locus does not lead to wide cancer susceptibility similar to that described for the p53 mutation [34-36]. This suggests that tumor susceptibility in the absence of ARF/INK4A is strongly influenced by cell or developmental contexts. This is analogous to cancer susceptibility in the setting of germ-line RB haploinsufficiency. Loss of a second allele leads to retinoblastoma formation with essentially complete penetrance, but the cancer risk is largely (though not exclusively) limited to the retina, with the retina tumor risk diminishing after the first 2 years of life [38]. For these reasons, somatic loss of just ARF in a narrowly defined lineage during human eye development may not unmask a clinically relevant cancer predisposition.

For somatic deletion of human ARF to represent a tenable pathogenetic mechanism for PHPV, both ARF alleles must be disrupted in a cell lineage ultimately giving rise to perivascular cells in the primary vitreous (likely, the neural crest, based on our studies of Wnt1-Cre, Arf ^fl/fl mice [30]); and the deletion must occur within the critical developmental window alluded to. This concept is not without precedent because an heritable retinoblastoma forms when a second RB gene mutation occurs during embryonic or early postnatal development in a child with germ-line RB haploinsufficiency [38]. The Arf locus seems particularly prone to recombination events leading to gene deletions in cultured fibroblasts [39], in mouse models of cancer [40,41], and in certain human tumors [42,43]. High-density arrays for detecting single nucleotide polymorphisms and ultrasensitive, massively parallel DNA sequencing now offer opportunities to detect somatic mutations in diseased tissue [33,44-46]. Such approaches could be leveraged to explore the possibility that sporadic cases of PHPV and perhaps other vitreoretinopathies are due to somatic mutation.

Beyond demonstrating the potential that somatic deletion of a single gene causes PHPV, our findings provide new insight into disease manifestations and the broader roles that p19^Arf might play during mouse eye development. First, we previously demonstrated that the retrolental mass in Arf ^−/− animals usually becomes pigmented as the mice age [22,24]. In our current series of chimeric animals, the retrolental mass was composed of densely pigmented cells in nearly every case. Because the wild type lineage was albino, this allows us to conclude that the densely pigmented cells must be of Arf ^−/− origin. We do not know, though, whether this was due to selective migration of pigmented cells from the Arf ^−/− RPE or choroid, as implied by findings from the Nathans laboratory in Fzd5 ^−/− mice [47]. It is also conceivable that the pigmented cells in the mass differentiate directly from non-pigmented, and primarily Arf ^−/− [25], cells populating the mass at birth. Regarding the latter concept, one might speculate that p19^Arf may normally act to block melanocytic differentiation in the neural crest-derived cells that migrate to the developing primary vitreous.

We also previously showed that the relative contribution of wild type and Arf ^−/− cells in tail-derived DNA correlates closely with their contribution to the cornea and the RPE at birth [25]. This implies that p19^Arf played no specific role in these tissues during development. In the 10-week-old chimeric animals studied here, the tail-derived DNA and RPE genotypes again correlated closely, but their correlation with the corneal epithelium was attenuated: The Arf ^−/− cells were underrepresented in the cornea in some cases and overrepresented in others. One straightforward explanation is that our one-dimensional analysis cannot reliably quantify cell lineage in the radially oriented cornea at 10 weeks, as opposed to the stippled lineage distribution in the newborn mouse [48]. Of note, we observed a relatively high intraclass correlation coefficient in the RPE and cornea analysis, and this finding helps to support the accuracy of our approach. Speculating that non-developmental factors control the relative number of wild type and Arf ^−/− cells in the cornea in older mice implies a functional role for p19^Arf in this tissue. However, at least at the level of light microscopy, our studies of Arf ^−/− mice have failed to show corneal defects at any point through 10 weeks of age [22,24].