|Home | About | Journals | Submit | Contact Us | Français|
Current therapies directed at controlling vascular abnormalities in cancers and neovascular eye diseases target VEGF and can slow the progression of these diseases. While the critical role of VEGF in development has been well described, the function of locally synthesized VEGF in the adult eye is incompletely understood. Here, we show that conditionally knocking out Vegfa in adult mouse retinal pigmented epithelial (RPE) cells, which regulate retinal homeostasis, rapidly leads to vision loss and ablation of the choriocapillaris, the major blood supply for the outer retina and photoreceptor cells. This deletion also caused rapid dysfunction of cone photoreceptors, the cells responsible for fine visual acuity and color vision. Furthermore, Vegfa deletion showed significant downregulation of multiple angiogenic genes in both physiological and pathological states, whereas the deletion of the upstream regulatory transcriptional factors HIFs did not affect the physiological expressions of angiogenic genes. These results suggest that endogenous VEGF provides critical trophic support necessary for retinal function. Targeting factors upstream of VEGF, such as HIFs, may be therapeutically advantageous compared with more potent and selective VEGF antagonists, which may have more off-target inhibitory trophic effects.
Phototransduction by rod and cone photoreceptors in the retina converts light into electrical energy. This process, forming the basis for vision, is facilitated by the vascular plexus (the choriocapillaris) and retinal pigmented epithelial (RPE) cells, both of which underlie and are intimately associated with the photoreceptors. Degeneration of photoreceptors, RPE cells, and the choriocapillaris is associated with age-related macular degeneration (AMD), the leading cause of vision loss in industrialized nations. While VEGF clearly plays a role in physiological vascular development (1–4) and pathological neovascularization (5, 6) in the retina and the choroid, there is increasing evidence that it is also important for the trophic maintenance of neurons (7–9). Inhibition of VEGF has become the mainstay of therapy for treating AMD (10–12), and increasingly potent VEGF antagonists have recently been introduced into the clinics. In order to better understand the trophic activity of VEGF in the adult retina and choriocapillaris and to assess the effect of deleting its expression by RPE cells, we utilized inducible Cre/loxP technology to genetically delete floxed Vegf alleles in vivo with an RPE-specific VMD2-Cre transgenic line (13).
Three days after RPE-specific Vegf inactivation, we observed complete ablation of the choriocapillaris (Figure (Figure1A).1A). Since almost the entire blood supply to the outer two-thirds of the retina (including the photoreceptors) is provided by the choriocapillaris, we predicted that its loss would induce dramatic secondary effects on the retinal neurons. In fact, following Vegfa inactivation in RPE cells, we observed profound retinal dysfunction, especially in cone photoreceptors. Within 3 days after induction, rapid and progressive dysfunction of cone photoreceptors was detected using electroretinography (ERG) (Figure (Figure1B)1B) compared with multiple controls, including Vegfaf/f (no Cre), Vegfaf/f;VMD2-Cre, Vegfaf/+;VMD2-Cre with doxycycline, and Vegfaf/f;VMD2-Cre without doxycycline (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI65157DS1). Using immunohistochemistry to selectively stain for the cone photopigment opsin, we observed a nearly complete absence of cone photoreceptor outer segments (Figure (Figure1C).1C). Cone dysfunction was observed by at least 7 months after Vegfa deletion in RPE (Supplemental Figure 2). These data demonstrate that RPE-derived VEGF is required for maintenance of the choriocapillaris and for cone photoreceptors, the most metabolically demanding cells in the central nervous system (14, 15). Photoreceptors also express VEGF receptors (9, 16), and paracrine-derived VEGF may have a direct role in their maintenance. In humans, loss of cones results in legal blindness, since fine (or “reading”) visual acuity is maintained by cones located in the central retina, or macula.
In contrast, rod photoreceptor function (responsible for achromatic and night vision) is insensitive to Vegfa inactivation at the same time points when we observe cone degeneration. Immunohistochemical analyses revealed that the layer of rod photoreceptor outer segments was as thick as that of controls (Figure (Figure2A).2A). Since the murine retina is composed largely of rods and the thickness of the entire retina does not change as late as 7 months after Cre recombinase induction (Figure (Figure2B),2B), this suggests that there is no significant rod photoreceptor atrophy. Scotopic (dark adapted) ERGs demonstrate that rod photoreceptors are fully functional, since there was no change in rod photoreceptor light responsiveness in recordings before and after induction in Vegfa mutant mice (Figure (Figure2C).2C). The preservation of rod photoreceptors for several months after gene deletion suggests that these cells are supported by a minimal vasculature not dependent on RPE-derived Vegfa for its survival. Indocyanine green angiography demonstrated 2 blood supply sources for photoreceptors in Vegfa mutants; intraretinal and outer, major choroidal vessels (Figure (Figure2D).2D). Utilizing electron microscopy, we determined that the choriocapillaris remained atrophic through 7 months and choroidal vessels were observed only sporadically (Figure (Figure2E).2E). These findings are consistent with results from a previous study that examined the effects of global deletion of the soluble VEGF isoforms (VEGF188/188 mice) (17).
Atrophy and thinning of the choriocapillaris is a known age-related event in humans (18). To compensate for choriocapillaris thinning, the RPE is known to synthesize proangiogenic factors that maintain the choriocapillaris (19, 20). Our findings suggest that VEGF may act both directly and indirectly to regulate this phenomenon. Using gene-profiling assays, we demonstrated that the loss of Vegfa affects the expression of multiple angiogenesis-related genes (most are downregulated) in isolated RPE/choroid preparations (Figure (Figure2F2F and Supplemental Table 1). Vegfa is tightly regulated by HIFs during development and pathologically in disease states (21, 22). To determine the role for HIFs in maintenance of the choriocapillaris, we genetically ablated Hif1a (encoding HIF-1α protein), Epas1 (HIF-2α), and Hif1a/Epas1 using VMD2-Cre in adult RPE. No obvious morphological, functional, or transcriptional differences, including Vegfa mRNA, were observed in naive-state adult HIF mutants (Supplemental Figures 3 and 4, and Supplemental Table 1). Several pathways, such as NF-κB/JunB and PGC1α/ERRα are known to induce Vegfa gene expression independently of HIFs (23, 24). These factors, or others, rather may maintain VEGF levels at sufficient levels in healthy retinas. Using laser photocoagulation to induce choroidal neovascularization (CNV), we determined that the extent of CNV was significantly reduced in Vegfa and Epas1 mutants (and partially reduced in Hif1a mutants; Figure Figure3,3, A and B). The upregulation of angiogenic genes was observed in wild-type C57BL/6 mice, but normal expression of these genes was observed in HIF mutants and expression was attenuated in Vegfa mutants compared with naive controls (Figure (Figure3C3C and Supplemental Table 2). These data strongly suggest that HIFs become activated in pathological states to promote VEGF-mediated neovascularization, but are not required to maintain physiological vasculature.
VEGF-mediated neovascularization is associated with vision loss in diseases such as AMD (10–12). The majority of patients receiving currently approved anti-VEGF drugs to treat AMD experience no adverse events. However, long-term safety studies of the effects of VEGF antagonism in the eye have not yet been completed, and there are isolated clinical reports showing progressive RPE/photoreceptor atrophy in patients on anti-VEGF therapies (25). Due to the differences in rod and cone photoreceptor distribution in humans and mice, it is difficult to speculate on what the effect of near-complete VEGF antagonism may be in human subjects. However, if a certain threshold of VEGF could be neutralized, the fovea would be very susceptible and the loss of macular function could result in the devastating loss of central vision. These observations, coupled with reported renal, gastrointestinal (GI), and mucosal complications in patients on systemic VEGF inhibition, contribute to the concern that VEGF antagonism may not be completely effective or safe (26, 27). The data in this study reinforce this concept and suggest that HIF-targeted or combination therapies targeting multiple angiogenic pathways may be safer and more effective than increased VEGF suppression therapy. Newer VEGF antagonists that bind VEGF with higher affinities and less rapid clearing from the eye are coming into use to treat a variety of neovascular diseases (28). Our results suggest that “off-target” effects of potent, persistent VEGF antagonism in the eye may exacerbate degeneration of the choriocapillaris, RPE, and photoreceptors due to the diminished trophic effect of RPE-produced VEGF.
Transgenic mice carrying the human vitelliform macular dystrophy-2 (VMD2) promoter-directed reverse tetracycline-dependent transactivator (rtTA) and the tetracycline-responsive element–directed (TRE-directed) Cre recombinase (VMD2-Cre mice) (13) (provided by Y.Z. Le of University of Oklahoma Health, Sciences Center, Oklahoma City, Oklahoma, USA) were mated with Vegfaf/f mice (29), Hif1af/f mice (30), or Epas1f/f mice (31) (provided by R.S. Johnson of University of Cambridge, Cambridge, United Kingdom).
Electron microscopy, ganzfeld ERG, immunohistochemistry, in vivo imaging, and laser-induced CNV were performed as previously described (32–34). mRNA PCR arrays for 84 angiogenic genes (Mouse Angiogenesis RTβ Profiler PCR Array, PAMM-024; QIAGEN) were performed according to the manufacturer’s instructions. Please see Supplemental Methods for more details.
Comparisons between the mean variables of 2 groups were performed by a 2-tailed Student’s t test. P < 0.05 was considered significant. Please see Supplemental Methods for more details.
All procedures involving animals were approved by the Scripps Research Institute Animal Care Committee, which ensures that all federal animal experimentation guidelines are met.
We would like to thank Malcolm Wood, Melissa S. Henrie, and Mollie S. Friedlander at the Scripps Research Institute for the excellent technical assistance, and Yun-Zheng Le at University of Oklahoma Health Sciences Center and Randall S. Johnson at University of Cambridge for providing transgenic mice. This work was supported by grants to M. Friedlander from the National Eye Institute (EY-11254) and the Lowy Medical Institute (MacTel). T. Kurihara is supported by a fellowship from the Manpei Suzuki Diabetes Foundation and The Japan Society for the Promotion of Science Postdoctoral Fellowships for Research Abroad. P.D. Westenskow is a Ruth L. Kirschstein Fellow of the NIH (EY021416).
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2012;122(11):4213–4217. doi:10.1172/JCI65157.
See the related Commentary beginning on page 3849.