Normal retinal vascular development and homeostasis is achieved by a fine temporal and spatial balanced production of proangiogenic and antiangiogenic factors. A number of proangiogenic and antiangiogenic factors have been identified in the eye and several appear to play major roles in normal development and in the disease states. Insulin-like growth factor-1 (IGF-1), basic fibroblast growth factor (bFGF) and VEGF are examples of stimulators of angiogenesis in the eye. PEDF, a protein originally identified in conditioned medium from RPE cells, has received a great deal of attention as a potent inhibitor of angiogenesis (Bouck, 2002
). However, its physiological role during retinal vascular development and neovascularization requires further investigation.
Expression of PEDF in the neural retina is spatially and temporally regulated and may play important roles in retinal vascularization and homeostasis. In mouse eyes, PEDF is not detected in the neural retina until late in gestation (E18.5). PEDF expression detected in retinal ganglion cell layer increases after birth and remains high through P14 and decreases thereafter, but persists through adulthood (Behling et al., 2002
). Here we did not observe significant β-galactosidase staining in retinal ganglion cell layer during postnatal vascularization of retina until later postnatal days (P14-P17; ). The strongest staining was observed in the outer plexiform layer followed by some sporadic staining in the inner retina. The staining pattern decreased after P14 and was almost undetectable in the neural retina by P42. Thus, normal postnatal retinal vascularization corresponds to the time points of higher PEDF expression in the neural retina. Given the antiangiogenic nature of PEDF, it is reasonable to speculate that expression of PEDF may act as a feed back mechanism to dampen proangiogenic signaling and promoting a more quiescent, differentiated state of the endothelium. These expression patterns are also consistent with the reported VEGF expression pattern which also decreases as retinal vascularization reaches completion (Ohno-Matsui et al., 2001
; Suzuma et al., 1999
). In addition, PEDF inhibits VEGF expression in retinal EC and muller cells (Zhang et al., 2006
), and its expression is up-regulated by exposure to hyperoxia when VEGF is minimally expressed during OIR ((Dawson et al., 1999
), and ). Furthermore in the absence of PEDF, VEGF levels remained higher during exposure to hyperoxia or normoxia in OIR (), thus, further supporting a role for PEDF in down regulation of VEGF expression under hyperoxic conditions.
PEDF inhibits EC proliferation, migration, and capillary morphogenesis resulting in death of EC by apoptosis (Chen et al., 2006
; Ho et al., 2007
; Kanda et al., 2005
). In contrast, PEDF promotes PC survival and proliferation, and protects PC from oxidative damage in culture (Amano et al., 2005
; Yamagishi et al., 2005
). We compared the EC and PC density of retinal vessel trypsin digests from PEDF+/+ and PEDFminus;/minus; mice at P21 and P42. At P21, when formation of the retinal primary vascular plexuses is complete, the mean EC density was significantly higher in PEDFminus;/minus; retinal microvessels compared to PEDF+/+ mice. However, the mean density of PC was not significantly affected in the absence of PEDF. Thus, increased number of EC results in a higher EC/PC ratio in PEDFminus;/minus; mice. The retinal vascular remodeling and pruning occurs after P21 and is complete by P42 (Wang et al., 2003
). We observed a similar EC and PC ratios and densities in P42 PEDF+/+ and PEDFminus;/minus; (–). Thus, our data suggest that PEDF deficiency promotes the early vessel growth in the inner neural retina by permitting more EC growth during the formation of retinal vascular plexuses without impacting its pruning and remodeling at later postnatal age when PEDF level is decreased.
We used retinal wholemount and frozen eye sections stained for collagen IV to further evaluate retinal vascular development during early postnatal days. During the first week of life (P7), when a superficial layer of vasculature spreads from the optic disc towards the peripheral retina, vessel spreading occurred at a significantly faster rate in PEDFminus;/minus; mice compared to PEDF+/+ mice (). After P7, when sprouting of the retinal superficial layer of blood vessels into the deep retina occurs, we observed significantly more intermediate retinal vessels (between the superficial and the deep layer) developing in P10 PEDFminus;/minus; mice compared to PEDF+/+ mice (). This is consistent with the increased rate of proliferation and apoptosis observed in retinal vasculature of PEDFminus;/minus; mice ( and ). Our results are consistent with those reported in mice where increased expression of PEDF resulted in a delay of deep retinal vascularization and vessel maturation (Wong et al., 2004
). However, our results are in contrast to those reported in 3-month-old mice where lack of PEDF was associated with an increase in retinal vascular density (Doll et al., 2003
). Our analysis failed to detect significant differences in vascular density at P42, and later time points (– and not shown). The analysis of retinal vasculature in the study by Doll and colleagues was limited to a single time point (3 months, when PEDF is normally undetectable). In addition, no images of stained retinal vasculature, demonstrating these differences, were provided. Perhaps strain differences and the exact methods used to generate the null mice may contribute to these differences.
Here we show that early postnatal retinal vascularization occurs at a faster rate in PEDFminus;/minus; mice. However, the development of retinal vasculature in PEDF+/+ mice catches up with that of PEDFminus;/minus; mice at later postnatal days, exhibiting spatial and density patterns similar to those of PEDFminus;/minus; mice at P42. Thus, PEDF deficiency may disturbs the normal balance between proliferation and apoptosis impacting early retinal vascularization. The most significant effect on the developing retinal vasculature occurred during the first two weeks of life, when PEDF expression is normally at its highest level ((Behling et al., 2002
) and ), perhaps acting as a feed back mechanism to dampen the VEGF proangiogenic signaling in an ischemic retina. This is consistent with increased rate of proliferation, and its compensation by increased rate of apoptosis, in PEDFminus;/minus; mice when active angiogenesis is occurring (P7-P10). Later, retinal angiogenesis reaches completion, and as a result, the degree of proliferation and apoptosis is significantly reduced, especially in the absence of PEDF (P14-P21). This is perhaps due to exaggeration of these activities in the absence of PEDF during earlier postnatal days. PEDF expression normally decreases in the retina after P21, and the deficiency of PEDF appear to play a less significant role in retinal vascularization. This is consistent with significant decrease in rates of proliferation and apoptosis observed in retinal vasculature of P21 PEDF+/+ and PEDFminus;/minus;mice. However, it is possible that other antiangiogenic factors, such as thrombospondin-1, may play a significant role during later stages of retinal vascular development and homeostasis (Bhutto et al., 2004
; Uno et al., 2006
; Wang et al., 2003
). We did not observe a significant impact on expansion and organization of astrocytes, PC, the number of endothelial tip cells, or VEGF levels during early retinal vascularization (, , and not shown). Thus, a tightly balanced production of pro- and anti-angiogenesis factors is essential for retinal vascular homeostasis.
The pupillary membrane and hyaloid vasculature (hyaloids arteries, tunica vasculosa lentis, and vasa hyaloidea propria) provide nourishment in the immature lens, retina and vitreous (Ito and Yoshioka, 1999
). However, they regress during the later stages of ocular development by apoptosis. The contribution of PEDF to these processes has not been previously addressed. To determine the potential role of PEDF in the regression of hyaloid vasculature, an apoptosis-dependent process, we examined the hyaloid vasculature at different postnatal days. Our results indicated that the regression of hyaloid vessels, mainly the tunica vasculosa lentis, was not significantly affected in the PEDFminus;/minus; mice compared to PEDF+/+ mice. Thus, the regression of ocular embryonic vessels is not compromised in the absence of PEDF.
We next examined the response of developing retinal vasculature to oxygen-induced ischemic retinopathy. P7 PEDF+/+ and PEDFminus;/minus; mice were exposed to 75% oxygen for 5 days. Hyperoxia negates the increase in VEGF expression during vascular development resulting in underdeveloped retinal vasculature and obliteration by apoptosis of existing vessels. This is mainly attributed to decreased VEGF level (Pierce et al., 1995
). An increase in PEDF level, at P12 (5 days of hyperoxia) when the most vascular obliteration occurs, has been reported ((Dawson et al., 1999
) and ). We observed that the developing retinal vasculature of PEDFminus;/minus; mice is more sensitive to oxygen-induced vessel obliteration resulting in increased nonperfused areas compared with the PEDF+/+ mice despite higher levels of VEGF. This may be attributed, at least in part, to the underdeveloped glial cells in the retina of PEDFminus;/minus; mice (Eichler et al., 2004
; Yafai et al., 2007
). We consistently observed lower levels of GFAP in retinal extracts prepared from PEDFminus;/minus; mice during room air or OIR (not shown). However, whether the presence of astrocytes with reduced levels of GFAP in PEDFminus;/minus; mice contributes to enhanced sensitivity of the retinal vasculature to hyperoxia remains to be determined.
PEDF is also a neuroprotective factor, which in its absence dying signals, especially under hyperoxia, may be exaggerated (Becerra, 1997
; Cao et al., 1999
; Tombran-Tink and Barnstable, 2003
). PEDF protects the retina from the stress of hyperoxia through its anti-oxidant and/or PC protective activity (Amano et al., 2005
; Yamagishi et al., 2006
). Increased oxidative stress, in the absence of PEDF, may also account for the increased VEGF levels observed in PEDFminus;/minus; mice during hyperoxia or normoxia in OIR. However, the molecular and cellular mechanisms which mediate PEDF antioxidant and/or protective effects require further delineation.
Exposure of mice after 5 days of hyperoxia to room air for 5 days (normoxia) stimulates growth of new blood vessels into the vitreous (Smith et al., 1994
). Our results indicated that preretinal neovascularization occurred similarly in the PEDF+/+ and PEDFminus;/minus; mice at P17. This was consistent with similar levels of VEGF expression observed in P15 PEDF+/+ and PEDFminus;/minus; (5 days of hyperoxia and 3 days of normoxia), when VEGF is maximally induced ((Pierce et al., 1995
) and ). Thus, lack of PEDF does not affect VEGF expression during normal postnatal retinal vascularization or active neovascularization in OIR. However, increased levels of VEGF were observed after exposure to hyperoxia or after retinal neovascularization in PEDFminus;/minus; mice during OIR. Thus, although the impact lack of PEDF has on retinal neovascularization during OIR is minimal, its expression may be essential for appropriate down-regulation of VEGF in response to hyperoxia and excessive neovascularization.
In summary, our results indicate that PEDF is an important modulator of retinal vascular homeostasis. It provides a protective role during formation of primary retinal vascular plexuses by maintaining the pro- and anti-angiogenic balance. Thus, alterations in PEDF levels during development and/or under pathological conditions may contribute to retinal vasculopathies. Modulation of PEDF levels has already proven as a suitable target for manipulation of retinal vascularization under some pathological conditions. However, additional studies are needed to elucidate the molecular and cellular mechanisms underlying the antiangiogenic and neuroprotective activities of PEDF.