Persistent Fetal Ocular Vasculature in Mice Deficient in Bax and Bak

Background  The ocular fetal vasculature normally regresses by apoptosis but for unknown reasons fails to regress in the human disease persistent fetal vasculature.

Objective  To investigate whether proapoptotic Bcl-2 members, Bax and Bak, are involved in fetal vasculature regression.

Methods  Adult eyes from mice deficient in Bax and/or Bak were examined grossly and histologically for persistence of fetal vasculature. Vessels were identified by the presence of lumens and erythrocytes and by Factor VIII labeling. Eyes from postnatal day 7 mice were processed for terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling (TUNEL) analysis to determine if deficiency of Bax and Bak results in defective developmental apoptosis.

Results  Only bax^−/−bak^−/− eyes retained fetal vasculature into adulthood. This vasculature consisted of a hyaloid artery emerging from the optic nerve head and intravitreal and perilental vessels but not a pupillary membrane. At postnatal day 7, wild-type but not bax^−/−bak^−/− eyes had TUNEL-positive cells in the fetal vasculature.

Conclusions  These data demonstrate that Bax and Bak serve overlapping functions in fetal vasculature regression, emphasizing the importance of apoptosis in developmental remodeling.

Clinical Relevance  Disruption of Bax and Bak results in persistent fetal vasculature in knockout mice, providing a model of the human disease persistent fetal vasculature to investigate its etiology and potential therapies.

The human disease persistent fetal vasculature (PFV) results from a failure of the fetal vasculature to regress properly.¹ One of the most common congenital ocular malformation syndromes, PFV is not associated with diseases in other tissues of the body and is often found in an otherwise normal child. Persistent fetal vasculature usually presents unilaterally with characteristic exam findings including retrolental white vascularized tissue with microcornea or microphthalmia of variable severity. Secondary glaucoma, intravitreal hemorrhage, retinal detachment, and lenticular and corneal opacification may also occur. While minimally affected eyes may survive to adulthood without surgical intervention, current treatment of more profoundly affected eyes involves surgical removal of intravitreal fibrovascular tissue.¹ Surgical treatment of PFV is difficult, and understanding mechanisms of normal fetal vascular regression may shed light on novel therapeutic options for PFV.

The fetal vasculature nourishes the developing mammalian eye and consists of the pupillary membrane and hyaloid vasculature. The pupillary membrane arises from the annular vessel in the rim of the neuroectodermal optic cup. The hyaloid vasculature originates at the optic nerve and sends branches called the vasa hyaloidea propria through the vitreous to anastomose with the tunica vasculosa lentis around the lens. These intravitreal vessels do not normally persist to adulthood; in a normal human, these vessels regress in utero¹, while in a normal mouse, the vasa hyaloidea propria regress by postnatal day (P)5 to P7, the tunica vasculosa lentis by P7 to P21, and the pupillary membrane by P3 to P9.^2,3

Regression of the ocular fetal vasculature occurs by apoptosis,⁴ and the specific apoptotic factors involved have recently been investigated. Apoptosis is a process of programmed cell death closely regulated by several gene families, including the bcl-2 gene family.⁵ This family contains both proapoptotic members, including bax and bak, as well as anti-apoptotic members, which collectively regulate outer mitochondrial membrane stability. Opposing members heterodimerize, and a cell’s fate may be regulated in part by the relative concentrations of each member.

We recently showed that 2 proapoptotic Bcl-2 family members, Bax and Bak, play a major role in apoptotic tissue remodeling. Because of impaired developmental apoptosis, mice deficient in both Bax and Bak have retained interdigital webs, imperforate vaginas, and supernumary central nervous system neural progenitor cells as well as rod photoreceptors in the retina.^6,7 Retinas deficient in Bax alone are developmentally impaired, although to a lesser extent than bax^−/−bak^−/− retinas, also leading to supernumerary retinal neurons.^7-9 Additionally, overexpression of Bcl-X_L blocks Bax-mediated lead-induced developmental photoreceptor apoptosis.¹⁰ Mice deficient in Bax and Bak are resistant to light damage–induced photoreceptor death.¹¹

Given the important roles of Bax and Bak in inducing apoptosis, we investigated their functions in eye development with particular attention to regression of the fetal vasculature. The hyaloid vessels and pupillary membrane have been shown to persist in a macrophage-ablated mouse,³ and macrophages surround regressing vessels in the normal mouse,^2,12 implicating a macrophage-driven, cell-extrinsic model in regression of these vascular tissues. Through histological comparisons, we demonstrate that mice deficient in cell-intrinsic proapoptotic factors, Bax and Bak, retain fetal vasculature in the vitreous, providing a model of PFV to better understand its pathologic mechanisms and investigate potential therapies.

C57BL/6 mice with targeted mutations in bax, bak, and bax/bak were generated and genotyped as previously described.^6,13 For bromodeoxyuridine (BrdU) analysis, mice were labeled in vivo with BrdU.¹⁴ Three-week-old mice were injected intraperitoneally with BrdU (Sigma-Aldrich, St. Louis, Mo) at 50 μg/kg body weight 1 hour before being humanely killed, or 8- week-old mice were injected intraperitoneally at 50 μg/kg body weight daily for 12 days and killed 12 days later. All eyes were enucleated immediately after the mice were killed and fixed in 4% paraformaldehyde, 0.5% glutaraldehyde overnight.

All procedures in this investigation conformed to the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research, and the procedures were approved by the Animal Care Committee of the University of Pennsylvania.

Fixed wild-type (n = 14) and bax^−/−bak^−/− (n = 12) globes were rinsed in phosphate-buffered saline and eyecups prepared. The vitreous and associated vasculature were photographed at low magnification with a Nikon SMZ-U dissecting microscope and Nikon CoolPix 990 camera (Nikon Inc, Melville, NY).

Wild-type (n = 4) and bax^−/−bak^−/− (n = 4) fixed globes were rinsed in phosphate-buffered saline, and whole eyes were dehydrated and embedded in paraffin. Sections 5 μm thick were cut through the optic nerve head and stored at room temperature. Sections were stained with either periodic acid–Schiff/hematoxylin or with hematoxylin. Factor VIII labeling (A0082, 1:500; DAKO Corp, Carpinteria, Calif) was performed by standard techniques with an initial 0.06% Pronase (Roche Diagnostics Corp, Indianapolis, Ind) digestion at 37°C and a peroxidase/diamino benzidine detection method using the Envision+ System (DAKO Corp) followed by hematoxylin counterstain. Bromodeoxyuridine labeling (G3G4, 1:50; Developmental Studies Hybridoma Bank, Iowa City, Iowa) was performed by standard immunofluorescence techniques with an initial antigen retrieval (Antigen Unmasking Solution, Vector Laboratories Inc, Burlingame, Calif). Digitized images were acquired with a Nikon Eclipse TE-300 light microscope and a Spot RT Slider camera (Diagnostic Instruments Inc, Sterling Heights, Mich) with ImagePro Plus 4.1 software (Media Cybernetics, Silver Springs, Md).

Terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling (TUNEL) analysis was performed using a commercial kit (Roche Diagnostics Corp, Indianapolis, Ind) on paraffin sections from wild-type (n = 2) and bax^−/−bak^−/− (n = 2) eyes harvested from mice at P7. Nuclei were counterstained with 4′, 6-diamidino-2-phenylindole (1.5 μg/mL)–supplemented mounting media (Vector Laboratories Inc), and sections were analyzed by fluorescence microscopy.

On gross examination of adult eyes following removal of the anterior segments, a stalk of branching tissue was observed emerging from the optic nerve head of bax^−/−bak^−/− eyes only (Figure 1B and C). This branching tissue was observed in all examined adult (>3 weeks old) bax^−/−bak^−/− eyes (n = 12) up to 8 months old, the age of the oldest mouse analyzed, but was not present in eyes from adult wild-type mice (n = 14; Figure 1A) or from adult Bax or Bak single-deficient mice.

To identify and further study this persistent tissue, paraffin sections through whole eyes were prepared (Figure 2A-M). Adult wild-type retinas (n = 4) occasionally displayed a Bergmeister papilla as observed grossly (Figure 1A). The Bergmeister papilla is a small remnant of the fetal vasculature often found in mice and sometimes in humans, representing the occluded base of the hyaloid artery surrounded by fibrous tissue (Figure 2A and E); no patent vessels were observed in any papillae. In contrast, the fibrovascular stalks emerging from the optic nerve head of adult bax^−/−bak^−/− eyes (n = 4; Figure 2B, F, and H) were significantly larger and contained erythrocytes (Figure 2G). Furthermore, while adult wild-type eyes did not contain intravitreal or perilenticular vessels (Figure 2A, E, and I), bax^−/−bak^−/− eyes exhibited persistent vessels in the vitreous (Figure 2B-D) and around the lens capsule extending as far anterior as the lens equator (Figure 2J-M). Vessels were identified by visualization of erythrocytes or by immunohistochemistry for Factor VIII. The brown color of the Bergmeister papilla in the wild-type eye (Figure 2A and E) is also present in unstained sections and represents normal intracellular pigmentation; no erythrocytes were observed in these papillae. Remnants of a pupillary membrane were not observed in any adult eyes examined. Indeed, the entire anterior segment was unremarkable in adult bax^−/−bak^−/− eyes as well as in eyes of all other genotypes studied.

To determine if the fetal vasculature observed in adult bax^−/−bak^−/− eyes resulted from defective regression, as would be expected given the proapoptotic roles of Bax and Bak, or from proliferation, TUNEL analysis was performed on bax^−/−bak^−/− (n = 2) and wild-type (n = 2) eyes at P7. Three to 5 TUNEL-positive endothelial cells were observed per wild-type section; no TUNEL-positive cells were observed in the bax^−/−bak^−/− fetal vasculature (Figure 3).

Proliferation of fibrovascular tissue is associated with the human disease persistent fetal vasculature and also occurs after P14 in association with the persistent fetal vasculature of a p53-deficient mouse. To determine if ongoing proliferation similarly occurs in bax^−/−bak^−/− eyes, bax^−/−bak^−/− and wild-type mice were pulsed with BrdU at P21 and P8 weeks. No BrdU incorporation was detected in the fetal vasculature of bax^−/−bak^−/− mice (not shown).

In the human disease PFV, the ocular fetal vasculature fails to undergo normal developmental regression for unknown reasons. The balance of angiogenic factors including vascular endothelial growth factor and placental growth factor is important in this regression,¹⁵ and mice deficient in angiopoeitin 2, which modulates angiogenesis through its binding to the Tie2 receptor, retain ocular fetal vessels.¹⁶ Initial studies investigating mechanisms of regression have implicated the macrophage as a central mediator. Persistence of the hyaloid vessels and the pupillary membrane to postnatal day 23 was observed in a macrophage-ablated mouse and provided the first direct evidence that the macrophage actively induces the regression of these vessels.^3,17 We report that cell-intrinsic proapoptotic Bcl-2 family members, Bax and Bak, are similarly essential for fetal vasculature regression such that mice deficient in Bax and Bak retain ocular fetal vessels into adulthood.

The primacy of cell-intrinsic vs cell-extrinsic influences on developmental cell death has long been debated. Early evidence from C. elegans indicates that mutations in cell engulfment genes do not affect apoptotic death in most cell lineages, supporting a cell-intrinsic model.¹⁸ In contrast, recent studies have demonstrated that mutations in genes involved in cell engulfment can result in survival of cells that otherwise would have undergone apoptosis, supporting a cell-extrinsic model.^19,20

While macrophage involvement has implicated cell-extrinsic factors as critical in inducing fetal vasculature regression, we provide support for involvement of a cell-intrinsic mechanism. Mice deficient in proapoptotic Bcl-2 members, Bax and Bak, have developmental abnormalities from impaired apoptosis, and we report here that their eyes displayed an abnormal persistence of normally regressing fetal vessels into adulthood, for at least 8 months. This network of vessels, absent in eyes of other genotypes, included a large stalk emerging from the optic nerve head (hyaloid vessel) that divided into smaller vessel branches in the vitreous (vasa hyaloidea propria) and surrounding the lens (tunica vasculosa lentis). These persistent vessels were identified by their morphology, including the presence of a lumen containing erythrocytes, and by anti-factor VIII labeling (Figure 2). The fetal hyaloid vasculature normally regresses completely by P21,^2,3,21 but was found in all adult bax^−/−bak^−/− eyes examined (n = 12) from 3 weeks to 8 months old. The presence of TUNEL-positive cells within the fetal vasculature in wild-type eyes at P7 but not in bax^−/−bak^−/− eyes suggests that the persistence of fetal vasculature in bax^−/−bak^−/− eyes resulted from defective apoptosis. Neither bax^−/− nor bak^−/− eyes retained more vasculature than wild-type mice, indicating that Bax or Bak is sufficient to induce apoptosis for full vessel regression.

Our findings identify a cell-intrinsic apoptotic mechanism controlled by proapoptotic Bcl-2 members, Bax and Bak, as critical to the developmental regression of the hyaloid vasculature, suggesting that hyaloid regression involves both cell-extrinsic and cell-intrinsic influences. One possible mechanism is that after macrophages trigger apoptosis, intracellular components such as Bax or Bak execute the apoptosis. Alternatively, Bax or Bak may trigger apoptosis with subsequent involvement of macrophages recruited by chemotactic factors.

Persistent fetal vasculature in bax^−/−bak^−/− eyes did not include retention of the pupillary membrane, consistent with a lack of involvement of Bax or Bak in pupillary membrane regression. In contrast, studies on regression of the pupillary membrane have indicated that macrophages are initiators of apoptosis in these vessels.^12,17 It appears, therefore, that Bax and Bak are not required for pupillary membrane regression while macrophages are required. Regression of the hyaloid vasculature and the pupillary membrane thus occur through different mechanisms.

Persistent fetal vasculature also occurs in p53-deficient mouse eyes.²² These mice have a hyaloid artery, vasa hyaloidea propria, and tunica vasculosa lentis. Like macrophage-ablated mice, they also develop associated fibrovascular retrolental plaques. While this retrolental fibrous tissue in p53-deficient mouse eyes results from proliferation observed after P14, the persistence of the hyaloid vasculature was shown by TUNEL analysis to result from decreased apoptosis in the developing eye.

To determine whether any ongoing fibrous proliferation occurred in the bax^−/−bak^−/− fetal vasculature, we examined eyes from BrdU-injected mice and found no BrdU incorporation in the fetal vasculature, consistent with the lack of retrolental fibrous tissue in these eyes. Thus, unlike mice deficient in p53, which is involved in both apoptosis and cell cycle control, mice deficient in proapoptotic members, Bax and Bak, do not have ongoing fibrous proliferation associated with the fetal vasculature. The stimulus for fibrous proliferation present in some mouse models of persistent fetal vasculature is unclear, but evidence from an apoptotic protease activating factor 1 knockout²³ suggests that defects in the postmitochondrial apoptotic program can result in necrotic cell death. In the eye, these necrotic cells could stimulate inflammation and fibrous proliferation. This explanation would be consistent with a lack of fibrous inflammation in the eyes of mice lacking the premitochondrial proapoptotic members, Bax and Bak.

Apoptosis within the eye is a normal component of development. These normal apoptotic processes can be pathologically disturbed, as in the activation of apoptosis in retinal degenerations or in the presumed failure of apoptosis in the human disease PFV. The partially overlapping functions of Bax and Bak mediate apoptosis within the eye, as at least 1 must be present for regression of hyaloid vessels. Specifically, Bax or Bak is necessary for apoptotic cell death of the hyaloid vasculature, which otherwise persists into adulthood in bax^−/−bak^−/− mice but regresses normally in bax^−/−, bak^−/−, and wild-type mice. A similar persistence of fetal vasculature occurs in PFV, a human disease of unknown pathogenesis. The bax/bak mouse provides a model of PFV and suggests that a failure to activate apoptosis through Bax or Bak may contribute to its pathogenesis.

Correspondence: Joshua Dunaief, MD, PhD, 305 Stellar Chance Labs, 422 Curie Blvd, Philadelphia, PA 19104 (jdunaief@mail.med.upenn.edu).

Submitted for Publication: February 16, 2004; final revision received August 18, 2004; accepted August 30, 2004.

Financial Disclosure: None.

Funding/Support: This study was supported by grant EY00417 from the National Institutes of Health, Bethesda, Md; grant T32-GM-07170, Medical Scientist Training Program, University of Pennsylvania, Philadelphia; a career development award from Research to Prevent Blindness, New York, NY; the Steinbach Foundation, New York; International Retina Research Foundation, Birmingham, Ala; and the Paul and Evanina Bell Mackall Foundation Trust, Wilmington, Del.

Acknowledgment: We thank T. Dentchev for advice and assistance.