T2-TrpRS Inhibits Preretinal Neovascularization and Enhances Physiological Vascular Regrowth in OIR as Assessed by a New Method of Quantification

Abstract

purpose. A carboxyl-terminal fragment of tryptophan tRNA synthetase (T2-TrpRS) has demonstrated potent angiostatic activity during retinal developmental neovascularization in vivo. The effects of T2-TrpRS on pathologic neovascularization were tested and compared with a potent VEGF antagonist using the mouse model of oxygen-induced retinopathy (OIR).

methods. C57BL/6J mice were transiently exposed to hyperoxic conditions (75% O₂) between postnatal day 7 (P7) and P12 and then returned to room air. Retinas were isolated, blood vessels stained with isolectin Griffonia simplicifolia, images of retinal whole-mounts acquired, and the area of vascular obliteration and extent of preretinal neovascularization quantified. This method was compared to the commonly used method of OIR quantification in which the number of pre-inner limiting membrane (ILM) nuclei is counted in serial sections of whole eyes. To assess the angiostatic activity of T2-TrpRS, mice were injected intravitreally at P12 with either T2-TrpRS, a VEGF aptamer, or vehicle (PBS) alone, and the effects on area of obliteration and on preretinal neovascular tuft formation were assessed.

results. Using a modified method of quantification in the mouse OIR model based on images of isolectin-stained retinal wholemounts, we were able to assess reliably and consistently both vascular obliteration and preretinal neovascular tuft formation in the same specimen. T2-TrpRS demonstrated potent angiostatic activity, reducing the appearance of pathologic neovascular tufts by up to 90%. Surprisingly, T2-TrpRS also enhanced physiological revascularization of the obliterated retinal vasculature, reducing these areas by up to 60% compared with PBS-injected eyes. In contrast, the VEGF antagonist, while similarly reducing preretinal neovascular tuft formation, did not enhance revascularization of the obliterated areas.

conclusions. Use of a rapid, quantifiable method to assess the effect of T2-TrpRS on retinal angiogenesis in the OIR model demonstrates the importance of a quantification system that permits simultaneous analysis of a drug’s effect on vascular obliteration as well as on preretinal neovascularization. The results obtained using this method suggest enhanced clinical value for compounds such as T2-TrpRS that not only inhibit pathologic neovascularization, but also facilitate physiological revascularization of ischemic tissue.

Vascular diseases of the retina, including diabetic retinopathy, exudative age-related macular degeneration, retinopathy of prematurity, and vascular occlusions, are major causes of vision loss. This group of diseases is the focus of intense research aimed at identifying novel treatments that will help prevent or reduce pathologic ocular neovascularization. Over the past decade, several angiostatic compounds have been identified and tested in preclinical models. ^{1 2} Some of these compounds have been applied clinically in recent years, with varying efficacy. ^{3 4} Whereas several demonstrate an effect on reducing the extent of neovascularization or a reduction in morbidity, complete inhibition of angiogenesis and/or cure of the underlying disease has not yet been achieved. To enhance the clinical efficacy of angiostatic therapy, it may be necessary to combine angiostatic drugs and/or use molecules with both angiostatic and trophic activities. Indeed, several compounds have been reported to exhibit trophic effects beyond their angiostatic activity ^{5 6} and this dual action may have clinical utility. 

Testing potential angiostatic treatments has been greatly facilitated by the development of models of oxygen-induced retinopathy (OIR) in several animal species, including the kitten, the beagle puppy, the rat, and the mouse. ⁷ These models mirror the events that occur during retinopathy of prematurity (ROP), a condition involving pathologic neovascularization that can affect premature infants. ^{8 9 10} Over the past decade, the mouse model of OIR as described by Smith et al. ¹¹ has become the most commonly used model for studying abnormal angiogenesis associated with oxygen-induced retinopathies. The vascular changes observed in this model are highly consistent, reproducible, and quantifiable, and the use of this model has recently been extended to the study of the effects of many potential therapeutic agents on ischemic vasculopathies. 

In this study, we present a technique that allows same-eye quantification of several aspects of the neovascular processes occurring in OIR, including obliteration and intraretinal revascularization and pathologic preretinal neovascular formations (tufts). Isolectin Griffonia simplicifolia (isolectin GS) staining, fluorescence microscopy and widely available computer image analysis programs can be used to quantify the retinopathy efficiently and consistently in wholemount retinal preparations from mice with OIR. This method was found to be highly reliable and consistent and correlated well with the current, standard method of quantification, which is based on the counting of pre-inner limiting membrane (ILM) nuclei in retinal cross-sections. ¹¹  

Wholemount quantification was subsequently used to analyze the angiostatic activity of T2-TrpRS, a carboxyl-terminal fragment of tryptophanyl tRNA synthetase, on pathologic neovascularization associated with OIR. Angiostatic activity of T2-TrpRS has been demonstrated in various in vitro models, ^{12 13 14} as well as a developmental model of retinal neovascularization. ^{15 16 17} The quantification method described in this study has shown T2-TrpRS to have potent angiostatic activity in the OIR model, inhibiting formation of neovascular tufts (pre-ILM correlates) by more than 90%. Surprisingly, revascularization of the obliterated superficial retinal vasculature was accelerated, indicating a novel activity of T2-TrpRS, and demonstrating the importance of a quantification method that incorporates measures of both vascular obliteration and pathologic neovascularization observed in the OIR model. 

Materials and Methods

OIR Model

All animal studies adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. OIR was induced according to the protocol described by Smith et al. ¹¹ Briefly, postnatal day (P)7 pups and their mothers were transferred from room air to 75% oxygen for 5 days and then returned to room air. 

Retinal Wholemount Preparation and Immunohistochemistry

For FITC-dextran angiography, animals were anesthetized, and approximately 300 μL of high-molecular-mass (2 × 10⁶ Da) FITC-dextran (concentration 50 mg/mL; Sigma-Aldrich, St. Louis, MO) was injected into the left ventricle and allowed to circulate for approximately 2 minutes before the animals were euthanatized and the eyes enucleated. Retinas were dissected and prepared for immunohistochemical analysis as previously described. ¹⁸ Briefly, they were blocked with 20% FBS and 20% normal goat serum (NGS) in PBS for 1 hour, incubated with the appropriate primary antibody and isolectin GS (1:150 dilution; Invitrogen, Carlsbad, CA) overnight in blocking buffer, and visualized using fluorescently labeled secondary antibodies (1:200 dilution, 2 .5-hour incubation). Because isolectin GS will stain not only vascular endothelial cells but also activated murine macrophages, polyclonal antibodies specific for the endothelial marker CD31 (platelet-endothelial cell adhesion molecule [PECAM]; 1:100 dilution; BD Biosciences, San Diego, CA) or against the macrophage marker F4/80 (1:150; Caltag Laboratories, Burlingame, CA) were used to characterize further the isolectin GS-positive cells. 

Imaging and Quantification

Images were taken using a 4× objective lens with a detector resolution of 512 × 512 pixels. Focusing just above the inner limiting membrane of the retina using confocal microscopy (Radiance 2100; Bio-Rad, Hercules, CA), the prelaminar neovascular tufts were prominent and readily distinguished from the underlying superficial vascular plexus. Four overlapping images (each representing one retinal quadrant) were acquired from each retina. No image manipulation was performed in terms of background, intensity, or color curves, and the resultant images were used “as is. ” With image analysis software (Photoshop 6.0; Adobe Systems, Mountain View, CA), each individual image was converted to 2 × 2 in., with 300 pixels/in. resolution. Then, whole retina montages were created by using retinal landmarks such as the optic disc and major vessels to align the images of the four quadrants. Subsequent quantification was performed on these montages, ensuring that size and resolution parameters were identical for each retina and that overlapping areas from each individual image used to create the montages were quantified only once. 

Quantifications were performed by “masked” individuals to prevent bias. To measure areas of vascular obliteration, the freehand selection tool of the software (Photoshop; Adobe Systems) was used. During analysis, care was taken not to include areas in which a dissection or mounting artifact was present. The borders of the avascular areas were traced, and the total area of obliteration in pixels was computed. Measurement of the areas of neovascular tufts was also performed in the wholemount configuration, based on the higher intensity of isolectin staining and the characteristic appearance of the tufts. The neovascular tufts were specifically selected by hand with the “magic wand ” tool in the software (Photoshop; Adobe Systems). The total area (in pixels) was then determined and a conversion factor, based on image-acquisition (4× magnification lens), montage assembly parameters, and resolution (300 pixels/in.) was applied to generate absolute areas in square micrometers (for our parameters, the conversion factor was calculated to be 27.8 μm²/pixel). 

Retinal Cross-Sections and Pre-ILM Nuclei Counting

In a subset of eyes, after quantification in the wholemount configuration, the retinas were removed from the slides, postfixed in 4% paraformaldehyde (PFA), embedded in paraffin, and serially sectioned in their entirety for pre-ILM nuclei counting, as described in Smith et al. ¹¹ Standard hematoxylin and eosin (H&E) staining was performed, and all nuclei present on the vitreous side of the ILM were counted in every fourth section across the retina. This method was used to include the entire retina for correlation with the wholemount analysis where all tufts across the retina are also quantified. The average number of nuclei per section was computed. 

Comparison of the Wholemount Quantification Method with Standard Quantification Methods

Mice were injected subcutaneously with 10 mg/kg per day of a specific iNOS inhibitor (iNOS inhibitor 51; GE Healthcare) from P12 to P17. This compound has been shown to be specific for iNOS and has previously demonstrated angiostatic activity (personal communication, Lois Smith, MD, PhD, Harvard Medical School, July 2004). On P17, fellow eyes were enucleated and processed for quantification by either wholemount quantification as described earlier (left eye), or by counting pre-ILM nuclei in H&E-processed sections (right eye). For counting of pre-ILM nuclei, whole eyes were embedded in paraffin, cross-sectioned, and stained with H&E. Prelaminar nuclei were counted according to the standard protocol described by Smith et al. ¹¹ in a sample of centrally located cross sections on both sides of the optic nerve. 

Preparation and Intravitreal Injection of Angiostatic Compounds

T2-TrpRS was made as a recombinant compound. Purified product was stored in 50% glycerol at −20°C. Before injection, an aliquot of T2-TrpRS was dialyzed overnight into sterile 1× PBS. Protein concentration was determined by the Bradford assay (Bio-Rad) with BSA (Sigma-Aldrich) as a standard. The VEGF antagonist was synthesized as a 40-kDa PEG-conjugated aptamer (Transgenomic Inc., Boulder, CO) based on published information. ^{19 20} This compound has been shown to inhibit the activity of VEGF. After synthesis and purification, the compound was determined to be pure by reversed-phase liquid chromatography. The VEGF antagonist was stored as a lyophilized powder in a desiccator at −20°C and was reconstituted in RNase-free PBS immediately before injection. All masses and concentrations refer to the active VEGF antagonist, as analyzed by spectrophotometric analysis using 260/280 nm wavelengths, rather than the total PEG-conjugated compound. T2-TrpRS, the VEGF antagonist or PBS alone were injected intravitreally at a volume of 0.5 μL with a syringe (Hamilton, Reno, NV) and a 33-gauge needle, as previously described. ^{17 18} Intravitreal injections were performed at P12 (on return to normoxia) for the single injection experiments, or at P13 and P16 for the double injection experiments. 

Results

The normal process of postnatal retinal vascular development in the mouse is well characterized ^{18 21} and includes centripetal growth of superficial vessels emerging from the optic disc at birth and reaching the far periphery approximately 9 to 11 days later. During the second and third weeks after birth, vessels sprout from this superficial network and form the deep and intermediate vascular plexuses. The developmental vascularization process is accompanied by waves of remodeling and pruning and is largely completed by P21 to P23. In C57Bl6/J mice, exposure to high levels of oxygen significantly alters this orderly sequence of events. ¹¹ After exposure to 75% oxygen between P7 and P12, large areas of the vascular network regress in the center of the retina, leaving only the major vessels and practically no capillary network. This can be seen in both dextran-perfused (Figs. 1A 1B)and isolectin-stained (Figs. 1D 1E)wholemount preparations. The peripheral retina still showed evidence of a vascular network, but, in general, the deep vascular plexuses had completely failed to form. On return to normoxia at P12, a relative state of ischemia in the poorly vascularized retina prompted the excessive regrowth of superficial vessels, leading to abnormal sprouting at the interface between retina and vitreous (pre-ILM vascular tufts). It is important to note that after perfusion of the retinal vessels with a fluorescent dye (in this case, dextran FITC), the areas of vascular obliteration were clearly delineated, but the tufts could not be easily identified. Staining with isolectin-GS allowed visualization of both processes (compare Fig. 1Bto 1E ). 

Quantifying Areas of Vascular Obliteration and Rate of Retinal Revascularization

To quantify the extent of vascular obliteration and rate of retinal revascularization, we imaged dextran-injected or isolectin-stained retinal wholemount preparations by fluorescent microscopy at low magnification. Areas of obliteration (yellow) as well as the total retinal areas (blue line) were measured (Figs. 2A 2B) . Because the border between vascular and avascular retina was usually well defined in these preparations (Fig. 2C) , interobserver variability was very low (Fig. 2G) . Isolectin-positive cells were often observed in OIR retinas, particularly in regions of obliteration (Fig. 2C) . Isolectin is known to stain microglia and activated macrophages, and the involvement of such cells in oxygen-induced vascular obliteration has been shown. ²² Retinal growth apparently continued during hyperoxia, with the total retinal area increasing by approximately 20% between P8 and P13. After this time, the rate of change became negligible (Fig. 2D , top). Obliteration occurred rapidly after exposure to hyperoxia at P7 (Fig. 2D , bottom) and occupied approximately 30% to 35% of the total retinal area between P8 and P12. Revascularization after return to room air occurred at a slower rate, with extensive neovascularization beginning around P16, 4 days after return to normoxia. The process was symmetrical, as demonstrated by the high degree of correlation between fellow eyes of the same animal (Fig. 2F) . This provides support for the use of fellow eyes as the control in future studies based on this model. 

Quantifying the Extent of Abnormal Neovascularization

Between P15 and P22, abnormal vascular growth occurred at the interface between the retina and vitreous, with most preretinal neovascular tufts forming at the border between the vascularized peripheral and obliterated central regions of the retina. These misdirected vascular elements are commonly quantified in H&E-stained, paraffin-embedded cross sections by counting nuclei of cells protruding into the vitreous side of the inner limiting membrane (pre-ILM nuclei). However, these areas are also clearly seen in isolectin GS-stained wholemount images as collections of cells overlying the superficial vascular network (Figs. 1E 3A) . The area of these strongly staining isolectin GS-positive regions was readily measurable with either intensity-threshold software or image-analysis software selection (Fig 3B , neovascular areas marked in red). Note that these areas of proliferation are only partially perfused (Fig. 3C) , precluding angiography as a tool to visualize and quantify the tufts adequately. Cell-specific markers showed that most of the isolectin GS-positive cells within the tufts expressed the endothelial marker CD31, confirming their vascular identity (Fig. 3D) . Of note, cells of macrophage lineage were also present in the neovascular tufts, as shown by F4/80 immunostaining (Fig. 3E) , supporting recent evidence that has suggested a role for macrophages during the progression of pathologic neovascularization and the formation of neovascular tufts in the OIR model. ²³ Between P16 and P20, approximately 12% to 16% of the total retinal area was covered by neovascular tufts, peaking at P17 (Figs. 3F 3G) . In the wake of the retinal revascularization that was concomitantly proceeding, these neovascular tuft areas also eventually resolved. The relatively large, protruding groups of cells seen earlier (Figs. 3A 3B 3C 3D 3E)were pruned (Fig. 3F)and ultimately disappeared between P22 and P25 (Fig. 1F) . 

Although the correlation of tuft formation between fellow eyes was less than that seen for vascular obliteration, it remained a largely symmetrical process (Fig. 3H) . Interobserver variability was again low, regardless of the time point studied and the extent of apparent tuft formation (Fig. 3I) , attesting to the relative consistency and clarity of this quantification method. Finally, to examine whether the neovascular areas measured in whole mounts correlate with pre-ILM nuclei counting in cross sections, several previously quantified wholemounted retinas were paraffin embedded and cross-sectioned, and the pre-ILM nuclei were counted across the whole retina (Fig. 2J) . It is evident that these two methods of quantification are generally in agreement (i.e., an eye with a larger tuft area tends to also have a higher nuclei count and vice versa). 

Correlation with the Established Method of Quantifying Preretinal NV

The utility of the new quantification technique was further assessed and compared with the current method of pre-ILM nuclei counting by examining the effect of an inhibitor of iNOS (Fig. 4) . This agent (or control saline) was delivered from P12 to P17 by once-daily subcutaneous injections, and at P17, fellow eyes were enucleated and processed for quantification by either wholemount quantification (left eye) or by counting of pre-ILM nuclei in H&E processed sections (right eye). When quantified from wholemounts, the neovascular tuft area was found to be reduced by 30% (center bars), similar to the magnitude of effect observed when pre-ILM nuclei were counted in cross sections (36%, Fig. 4B , right). Thus, the observed effects of an angiostatic compound are in general agreement, using either the standard quantification methods or wholemount analysis. 

Effect of T2-TrpRS on Neovascular Tuft Formation in the OIR Model

T2-TrpRS is a proteolytic, carboxyl-terminal fragment of TrpRS that has demonstrated potent angiostatic potential in several models of angiogenesis, including an in vitro shear stress model ¹² and the neonatal mouse retinal angiogenesis model. ¹⁷ The mouse OIR model was used to test the activity of T2-TrpRS on pathologic neovascularization by using the described method of quantification. When injected intravitreally at P12, immediately after the mice were returned from hyperoxia (75% O₂) to normoxia, T2-TrpRS markedly reduced the area of neovascular tuft formation at P17 in a dose-dependent manner (Fig. 5A) . Although the full-length tryptophan tRNA synthetase (FL-TrpRS) also demonstrated some activity at the highest dose, the observed levels of inhibition were significantly lower than equivalent injections of T2-TrpRS. As the FL-TrpRS has been found to have no angiostatic activity in other models of angiogenesis, ^{15 17} it is possible that the observed activity resulted from natural cleavage of the FL-TrpRS to a product similar to T2-TrpRS after in vivo injection. 

The obliterated areas were also quantified in control and T2-TrpRS-treated retinas, to analyze the effects on physiological revascularization after oxygen-induced obliteration (Fig. 5B) . Although T2-TrpRS potently inhibited neovascular tuft formation, revascularization of the normal superficial and deep retinal vascular plexuses was not inhibited. In fact, T2-TrpRS-injected retinas consistently showed significantly enhanced revascularization compared with the PBS controls, even at lower doses. In T2-TrpRS-injected retinas, the obliterated areas at P17 were consistently reduced between 40% and 60% compared to the obliterated areas in P17 PBS control-injected retinas (Fig. 5B) , suggesting a novel dual activity for T2-TrpRS in promoting physiological revascularization while simultaneously inhibiting pathologic neovascular tuft formation. Even at P22, when the neovascular tufts have naturally regressed and the retina is healing, the neovessels, formed by physiological revascularization in control-injected OIR retinas, still appear severely abnormal, characterized by their tortuosity and a few remaining neovascular tufts (Fig. 5E) . In contrast, at P17, the vascular morphology of 1.25 μg/eye T2-TrpRS-treated retinas appears nearly normal (Figs. 5C 5D 5E) . Retinas injected with FL-TrpRS also demonstrated significant enhancement of the revascularization process, but only at the higher doses. This finding again suggests either a lower level of activity for FL-TrpRS, or in vivo proteolytic cleavage to the active T2-TrpRS form. 

The activity of T2-TrpRS in the OIR model was directly compared with that of a VEGF antagonist that has demonstrated potent inhibition of OIR-induced neovascular tuft formation. Dosage studies were performed with this VEGF antagonist to determine the optimal concentration. At the optimal concentrations, 2 μg (215 picomoles) per eye of the VEGF antagonist and 1.25 μg (26 picomoles) per eye of T2-TrpRS, both angiostatic molecules were found to be potent inhibitors of neovascular tuft formation. Eyes injected with T2-TrpRS exhibited a slight, but significant reduction in the neovascular tuft areas compared with the VEGF antagonist-treated retinas (Fig. 6) . Thus, even after injection of molar concentrations 10-fold below the VEGF antagonist, T2-TrpRS exhibited highly competitive inhibitory activity on neovascular tuft formation in the OIR model. Unlike T2-TrpRS, the VEGF antagonist did not enhance revascularization of the superficial plexus. Although the VEGF antagonist did not prevent physiological revascularization (these retinas eventually fully revascularized), the revascularization was somewhat delayed compared with control vehicle (PBS)–injected retinas and was substantially delayed compared with those injected with T2-TrpRS (Fig. 6) . 

Because of the observed differences in the effects of T2-TrpRS and the VEGF antagonist on the revascularization process, we wanted to assess the longer-term effects on retinal neovascularization after hyperoxia-induced obliteration. To this extent, two separate lower-dose injections of either T2-TrpRS (0.25 μg; [5.2 picomoles] per injection) or the VEGF antagonist (0.5 μg [53.9 picomoles] per injection) were performed at P13 and P16, and the resultant retinal vasculatures were analyzed at P19. Again, both compounds significantly inhibited neovascular tuft formation. The revascularization process was again significantly enhanced in T2-TrpRS-treated retinas, whereas the delay in the revascularization process was even more apparent at P19 after treatment with the VEGF antagonist, despite the use of lower doses (Figs. 6B 6C 6D) . 

Discussion

The mouse model of OIR has been valuable for the study of ischemic vasculopathies and antiangiogenic agents. ⁷ We have used the OIR model and modified the described method of quantifying the associated neovascular processes to fully assess the angiostatic activity of T2-TrpRS in a model of pathologic angiogenesis. We have demonstrated a strong, dose-dependent angiostatic effect of T2-TrpRS on pathologic neovascularization in the OIR model. Injection of 1.25 μg/eye T2-TrpRS resulted in nearly complete inhibition of neovascular tuft formation. After two lower-dose injections, T2-TrpRS inhibited pathologic neovascularization and reduced tuft formation by more than 75%. These observed effects were comparable, if not superior to, the observed effects after injection of a VEGF antagonist, another potent inhibitor of pathologic neovascularization in the OIR model, despite the injection of approximately 10-fold higher molar concentrations of the VEGF antagonist. Unlike the VEGF antagonist, T2-TrpRS also enhanced physiological revascularization after oxygen-induced obliteration. Although the areas of obliteration were comparable at P12, when the retinas were removed from hyperoxia and injections were performed, the obliterated areas in T2-TrpRS treated retinas at P17 were consistently reduced between 40% to 60% compared with control-treated retinas. In addition, injection of T2-TrpRS led to faster and more complete retinal healing, and a resultant vasculature that anatomically more closely resembled normal retinal vasculature. 

Although this accelerated retinal revascularization effect of T2-TrpRS that accompanied its potent angiostatic activity was surprising, such dual properties (angiostasis of “pathologic” neovascularization and facilitation of “physiological” revascularization), while unusual, are not without precedent. Angiogenesis is a complex process regulated by an intricate balance of stimulators and inhibitors. ²⁴ Multiple signaling pathways converge to regulate the growth of new blood vessels during both physiological and pathologic neovascularization, and some factors have demonstrated seemingly conflicting, environmentally dependent roles in this complex process. For example, extracellular matrix (ECM) remodeling by matrix metalloproteinases (MMPs) promotes cell migration, a critical event in the formation of new vessels. ²⁵ MMP-mediated ECM remodeling can also promote angiogenesis by releasing matrix-bound growth factors that enhance endothelial migration and growth. ²⁶ In contrast, certain ECM molecules released by MMP degradation, such as thrombospondin-1 and -2, and proteolytic fragments of matrix molecules created by MMP-dependent ECM cleavages, such as endostatin, can exert antiangiogenic effects by inhibiting endothelial cell proliferation, migration, and tube formation. ^{5 27 28} T2-TrpRS, a fragment of a naturally occurring, larger “parent” molecule, may also exert context-dependent roles during angiogenesis. It is angiostatic, as demonstrated by potent angiostatic properties in several in vivo models of angiogenesis. However, in appropriate situations such as the regrowth of the normal, anatomic retinal vasculature, T2-TrpRS may also participate in the growth and stabilization of normal vasculature. The fact that T2-TrpRS enhances revascularization in the OIR model, although it potently inhibits deep vascular formation during normal vascular development, suggests that the circumstances of vascular growth under each condition may be vastly different. Although both processes proceed by angiogenesis, normal vascular development is highly controlled and occurs under moderate levels of hypoxia, ^{18 29} whereas retinal revascularization after oxygen-induced obliteration occurs in a highly hypoxic environment. ^{30 31} This is associated with rapid and often uncontrolled (as evidenced by the concomitant formation of preretinal neovascular tufts) neovascularization in the OIR model. Experiments investigating the different properties of T2-TrpRS in relationship to different contextual angiogenic situations are currently ongoing. 

A possible explanation for the apparent paradoxical effects of T2-TrpRS (promoting more rapid revascularization of obliterated areas while concomitantly inhibiting preretinal neovascular tuft formation) could be that the latter effect actually stems from the former. In other words, T2-TrpRS may primarily function to enhance revascularization of the retina, thus lessening the mass of ischemic tissue and reducing the hypoxic drive, resulting in a reduction of pathologic preretinal tuft formation. This explanation, however, seems unlikely, given that previous studies in our laboratory and others have demonstrated potent angiostatic activity for T2-TrpRS in other in vivo models of neovascularization. ^{15 17} The apparent dual action of T2-TrpRS is likely to be more complex and may indeed reflect different microenvironment-dependent activities. 

Further data supporting the concept of a context-dependent dual activity of T2-TrpRS is the differential effects with respect to dose. Enhanced revascularization was observed at low doses that had little or no effect on pathologic tuft formation (Fig. 5) . At higher doses, inhibition of tuft formation became apparent. As the dose continue to increase, the effect on tuft formation reached its maximum, whereas enhancement of revascularization plateaus and even began to decrease. Similar dose-dependent dual activities have been reported for other compounds. For example, in a laser-induced model of choroidal neovascularization (CNV), pigment epithelial derived factor (PEDF) was shown to decrease CNV at low doses (90 μg/mL administered subcutaneously via a miniosmotic pump), while at higher doses (360 μg/mL), PEDF significantly increased CNV. ³² Plasminogen activator inhibitor (PAI)-1 also has a bell-shaped efficacy curve with proangiogenic properties at physiological (lower) doses and antiangiogenic activity at higher doses. ^{33 34} However, it should be noted that although the dual activities of T2-TrpRS appear to be somewhat dose dependent, in the OIR model both properties occur concomitantly. Similar simultaneous dual activities have also been reported for inhibitors of tumor necrosis factor-α. In a recent study by Gardiner et al., ⁶ intraperitoneal injection of the cytokine inhibitor semapimob was shown to promote vascular recovery in a model of OIR, accelerating physiological revascularization, while concomitantly inhibiting neovascular tuft formation. The mechanism of this dual function for TNF-A is not yet known. 

In contrast, treatment with other antiangiogenic compounds (e.g., VEGF antagonists that inhibit all VEGF isoforms) may reduce pathologic tufts, but may also prolong vaso-obliteration in ischemic retinas. ^{23 35} This could lead to long-term vascular abnormalities because the hypoxic drive that initiates pathologic neovascularization may persist. VEGF is known to be critical for the normal development of the retinal vasculature. Thus, it is not surprising that the inhibition of VEGF may delay revascularization of the superficial plexus. Also, VEGF activity is known to be necessary for the maintenance of the normal adult retinal vasculature. The loss of VEGF activity below a critical threshold is associated with pericyte loss, which can result in subsequent vascular degeneration often associated with diabetic retinopathy. ³⁶ In addition, a critical level of VEGF activity is required for normal development and maintenance of retinal neurons. ³⁷ Thus, as the angiostatic properties of anti-VEGF therapies are tested in clinical settings, the long-term affects of VEGF inhibition on superficial revascularization, retinal morphology, and retinal function should be considered and experimentally examined. These studies should be performed with a method of quantification that allows whole retina analysis of obliteration as well as pathologic tuft formation in the same eye, such as the method presented herein. 

Our data also suggest that VEGF is not the sole mediator of OIR-associated pathologic neovascularization. T2-TrpRS, which does not directly block VEGF or its receptors, ^{14 15 17} has potent angiostatic effects on long-term pathologic neovascularization. Thus, even in the OIR model, in which VEGF₁₆₅ has been shown to be a critical mediator of pathologic neovascularization, ^{23 38} other factors are likely to make critical contributions as well. ^{23 39 40}  

Advances in our understanding of angiogenesis have led to the development of medical therapies to inhibit pathologic neovascularization and to movement away from thermal laser-mediated vaso-obliteration. Although elimination of abnormal vessels is desirable to avoid vision-threatening complications, such as macular edema, traction retinal detachments, and vitreous hemorrhage, inhibition of neovasculature could also lead to exacerbation of the underlying ischemia that initiates the vascular proliferation to begin with. In this regard, it is important to think about vascular reconstruction and/or maturation to relieve hypoxia-driven neovascularization, perhaps by the development of compounds that may both inhibit abnormal vessel development and promote physiological revascularization. ⁴¹ Such modified gene products with multiple static-trophic functions may provide an alternative to pure angiostatic therapy, an activity that may have unwanted local and systemic side effects in patients with ischemic and/or age-related retinopathies. ⁴² Fragments of tRNA synthetases appear to have such dual properties, and the use of quantifiable imaging techniques to evaluate their effect on ocular tissue, such as those discussed herein, should facilitate the development and assessment of such drugs. 

³  Contributed equally to the work and therefore should be considered equivalent authors.

 Supported by Grants EY11254 and EY014174 from the National Eye Institute, The Robert Mealey Program for the Study of Macular Degenerations, the V. Kann Rasmussen Foundation, the MacTel Foundation, and the Kovner Family Fund (MF). MR was supported by a National Research Service Award from the National Institutes of Health.

 Submitted for publication August 18, 2005; revised October 18 and November 29, 2005, and January 9, 2006; accepted March 7, 2006.

 Disclosure: E. Banin, None; M.I. Dorrell, None; E. Aguilar, None; M.R. Ritter, None; C.M. Aderman, None; A.C.H. Smith, None; J. Friedlander, None; M. Friedlander, 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.

 

Wholemounts of retinas at different stages of progression through the OIR model. The vasculature was visualized by dextran perfusion (A-C) or by staining with isolectin-GS (D–F). (A, D) P9 retinas showed marked regression of the central vascular network after exposure to 75% oxygen at P7. (B, E) P18 retinas showed partial revascularization from the periphery inward after return to normoxic conditions at P12, accompanied by the formation of pathologic neovascular tufts (E). (C, F) Revascularization of the retina is largely complete by P22, and the areas of neovascular tufts have mostly resolved.

Figure 2.

 

Time course of vascular obliteration and retinal revascularization at different stages of the mouse OIR model. (A) A representative retinal wholemount of a P17 retina is shown and, in (B), the areas of vascular obliteration (yellow) and total retina area (blue line) are marked, demonstrating the ability to identify and measure these areas with image-analysis software. (C) A higher-magnification image demonstrates that the avascular borders can be easily identified using isolectin GS to stain retinal vessels. (D) Changes in total retinal area and the extent of obliteration throughout OIR progression (each point represents mean ± SD). (E, F) Correlation between fellow eyes of the same animal was high for both total retinal area (E) and the area of vascular obliteration (F) at different time points throughout the progression of OIR. These processes were largely symmetrical, particularly for area of obliteration, as indicated by linear regression parameters. Each point represents a pair of retinas (left eye versus right eye) from a single mouse. (G) The degree of interobserver variability was low between four masked observers quantifying vascular obliteration in six different eyes. Hatched bar: mean area ± SD for each eye.

Figure 3.

 

Appearance, composition, and quantification of preretinal neovascular tufts in OIR. (A) Confocal images focused slightly above the superficial vascular layer assist in identifying areas of abnormal preretinal neovascular tufts in a P18 eye stained with isolectin GS. (B) The tufts were easily identified (red) and quantified with commercially available image-analysis software. (C) Neovascular tufts were only partially perfused, as shown after systemic infusion of dextran-FITC. (D) The identified tufts were endothelial in origin, as demonstrated by colocalization with CD31. (E) Cells of macrophage lineage expressing F4/80 were also a major component of tufts. (F) Resolving areas of neovascularization in a P22 isolectin-stained retina demonstrate that tufts were diminishing, but the retinal vasculature was still not normal. (G) Time course of neovascular tuft formation and regression (each bar represents mean retinal tuft area ± SD. A total of 84 retinas were quantified: between P15 and P19, n ≥ 10 retinas at each time point; P20 to P22, n ≥ 6; P25, n = 4). (H) Although less symmetrical than the areas of obliteration between fellow eyes, the areas of neovascular tufts were relatively symmetrical from eye to eye. (I) The degree of interobserver variability between four masked observers quantifying neovascular tuft formation in four different eyes (hatched bar: mean area ± SD for each eye). (J) Quantification of abnormal neovascularization by tuft area in P17 wholemounts compared to subsequent counting of pre-ILM nuclei in cross sections obtained from the same retinas. The results were normalized to the average for each method among the six eyes, demonstrating that similar trends were observed regardless of the method of quantification used.

Figure 4.

 

Comparison of the wholemount quantification method with standard quantification methods (A) Wholemount preparations and cross sections of P17 retinas from a control animal (top row) and an iNOS inhibitor-treated animal (bottom row). The area of obliteration (left, yellow) and neovascular tuft formation (middle, red) were quantified by wholemount analysis. Arrows, right: examples of pre-ILM nuclei. The accepted method of quantifying abnormal neovascularization in the mouse OIR model relies on counting such cells. (B) Subcutaneous injection of an iNOS inhibitor reduced the area of vascular obliteration at P17 by 28% compared with controls (left). Area of pre-neovascular tufts quantified from wholemounts was reduced by 30%. Data are the mean ± SEM. Control: n = 10; treated: n = 12. Counting of pre-ILM nuclei in cross-sections from a separate group of animals showed a similar magnitude of effect, 36% (mean ± SEM). Control: n = 55; treated: n = 13.

Figure 5.

 

(A) Injection of T2-TrpRS at P12 inhibited neovascular tuft formation in a dose-dependent manner and (B) also promoted faster retinal revascularization as demonstrated by a reduction in the area of obliteration. FL-TrpRS had similar activities, but only at higher doses. A representative image of a P17 retina that was treated with 1.25 μg/eye of T2-TrpRS (C) showed significant reduction in neovascular tufts (red) and obliterated areas (yellow) compared with a control, PBS-injected retina (D). (E) The vasculature in OIR retinas treated with T2-TrpRS looked relatively normal (top) and lacked the characteristic tortuosity that was characteristic of revascularized regions in control OIR retinas, even at a much later stage (bottom).

Figure 6.

 

Comparison of the effects of T2-TrpRS with a VEGF antagonist. (A) Both T2-TrpRS and the VEGF antagonist were potent inhibitors of neovascular tuft formation, but only T2-TrpRS reduced the areas of obliteration compared with control PBS injections. (B) At P19, after two injections, the areas of obliteration were significantly larger in the VEGF-treated retinas, indicating that the VEGF antagonist may slow the revascularization process. (C, D) Representative images of P19 retinas after double injection with either T2-TrpRS (C) or the VEGF antagonist (D) demonstrate the differences in tuft areas (red) and obliteration areas (yellow).

The authors thank members of the Friedlander laboratory for many helpful and insightful discussions; Paul Glidden of Angiosyn for providing recombinant T2-TrpRS; Lois Smith and members of her laboratory for generously providing technical assistance with regard to the iNOS experiments and helping us correlate the different methods of OIR quantification.