Persistent Inhibition of Oxygen-Induced Retinal Neovascularization by Anthrax Lethal Toxin

Abstract

Purpose.: To evaluate the role of mitogen-activated protein kinase kinase (MKK) signaling in a mouse model of oxygen-induced retinopathy (OIR) that mimics retinopathy of prematurity (ROP).

Methods.: Postnatal day 7 mice were exposed to elevated oxygen for 5 days to induce retinopathy. Anthrax lethal toxin (LeTx), an MKK inhibitor, was injected into the vitreous after restoration to normoxia, and its effects on vascular growth were analyzed by whole mount immunofluorescence and confocal microscopy. Pericyte coverage was determined by PDGFR-β and α-SMA staining. Macrophage presence was determined by F4/80 staining. Vitreal cytokine secretion was measured by ELISA and multianalyte profiling.

Results.: Intravitreal injection of LeTx over a restricted time interval after return to normoxic conditions blocked the progression of OIR. This block was independent of vascular endothelial growth factor (VEGF) release and did not alter the release of cytokines and growth factors associated with OIR. VEGFR2 expression and activation were similarly unaffected. LeTx had no statistically significant effect on macrophage recruitment. LeTx sensitivity correlated with vessel maturity, extent of hypoxia, and growth of the deep vascular plexus network.

Conclusions.: Correlation among pericyte coverage, deep vascular plexus growth, and hypoxia after LeTx treatment indicate immature vessels in a hypoxic environment are preferentially sensitive to LeTx-mediated MKK inhibition. The persistence of VEGF without concomitant induction of neovascular growth or revascularization of vaso-obliterated zones suggests MKK inhibition causes an inability of the cells that are present, or a failure to recruit cells able, to respond to proangiogenic stimuli. These results indicate the inhibition of MKK signaling presents a novel strategy for the inhibition of vascular retinopathies such as OIR and ROP.

Retinopathy of prematurity (ROP) caused by exposure to elevated oxygen is a leading cause of blindness in infants. ROP, as well as other retinopathies such as diabetic retinopathy and age-related macular degeneration, are caused by unregulated and abnormal blood vessel growth, resulting from the increased release of proangiogenic factors, the most notable of which is vascular endothelial growth factor (VEGF). ^1,2 A central component of these diseases is the growth of abnormal blood vessels that are leaky and tortuous, lack pericytes, and have an irregular distribution, resulting in severe vision loss and blindness. 

In a mouse model of ROP (oxygen-induced retinopathy [OIR]; ³ ), exposure of mice to high (75%) oxygen levels (hyperoxia) at postnatal day (P)7 for 5 days results in rapid and profound vaso-obliteration, followed by loss of vascular density in the retina. ⁴ On return to room air, the incompletely vascularized retina becomes increasingly hypoxic as it matures and becomes more metabolically active. ⁵ The hypoxic state of the retina leads to the upregulation of proangiogenic factors in an HIF-1α–dependent manner, resulting in extensive neovascularization, with maximal neovascularization occurring between P17 and P21. ⁵ Since its development, this OIR model has been used to analyze a variety of critical pathways and factors involved in pathologic angiogenesis in the retina (see Ref. 6 for review). Importantly, it reflects the events that occur during ROP and mimics some of the pathologic events of the proliferative phase of diabetic retinopathy, ⁶ providing a relevant model with which to examine potential novel therapeutics for these diseases. 

The mitogen-activated protein kinase (MAPK) signaling pathways are a group of related protein kinase signaling cascades widely expressed in eukaryotes, involving a three-tiered kinase cascade comprising a mitogen-activated protein kinase (MAPK), a MAPK kinase (MAPKK, MEK, or MKK), and an MKK kinase (MAPKKK, MEKK, or MKKK). ^7,8 In response to extracellular stimuli such as growth factors and cytokines, ligand/receptor binding initiates the activation of MKKK that phosphorylates MKK, which in turn phosphorylates MAPK. Phosphorylated (and hence activated) MAPK then phosphorylates effector molecules which may act at the transcriptional, posttranscriptional, or posttranslational levels. The most extensively studied vertebrate MAPK pathways to date are those regulated by extracellular signal-related kinases 1 and 2 (ERK1/2), which are activated by MEK1 and MEK2, c-Jun N-terminal kinases (JNK), which are activated by MEK4 and MEK7, and p38 MAPKs, which are activated by MKK 3 and 6 (see Ref. 8 for review). The ERK pathway is preferentially activated by growth factors, whereas the JNK and p38 pathways respond to cellular stresses such as osmotic shock and inflammatory cytokines (see Ref. 9 for review). 

Recent observations suggest MAPK may play a role in OIR. Pathologic angiogenesis is reduced in JNK knockout mice in a mouse model of OIR, ¹⁰ and intravitreal injection of a MEK1/2 inhibitor has been reported to reduce the formation of neovascular tufts in OIR in rats. ¹¹ Based on these observations we hypothesize that MAPK signaling pathways play a critical role in hypoxia-induced retinal neovascularization. 

To test this hypothesis we administered intravitreal injections of a global MKK inhibitor, anthrax lethal toxin (LeTx). LeTx is derived from an exotoxin produced by the Gram-positive bacterium Bacillus anthracis. It is composed of two proteins: protective antigen (PA) and lethal factor (LF). PA alone is nontoxic and serves to translocate LF to the cytosol. LF is a zinc-dependent metalloprotease that inactivates all MKK with the exception of MEK5. ^{12 –14} We found that MKK inhibition did not alter levels of vitreal VEGF but did prevent the induction of neovascular growth and revascularization of the vaso-obliterated zone. Sensitivity to LeTx was linked to vessel maturity, with immature retinal vasculature more sensitive than mature vessels. Together these data indicate that MKK signaling plays a critical role in the initial development of blood vessels in OIR. MKK inhibitors alone, or in combination with VEGF-targeted inhibitors, present a novel strategy for the treatment of retinopathies such as ROP. 

Methods

Cell Culture and Reagents

Simian virus 40–transformed mouse microvascular endothelial cells (SVECs) were obtained from Silvio Gutkind (National Institute of Dental and Craniofacial Research, Bethesda, MD). SVECs were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. The ARPE-19 cell line is a spontaneously immortalized retinal pigment epithelial cell line and was acquired from American Type Culture Collection (CRL-2302; ATCC, Manassas, VA). ARPE-19 cells were cultured in DMEM/F12 1:1 media (214012; Invitrogen, Carlsbad, CA), supplemented with 10% FBS and 1% penicillin/streptomycin. Human umbilical vein endothelial cells (HUVECs) were cultured in endothelial cell growth medium (EGM-2, CC-3156; Lonza, Walkersville, MD) supplemented with additional medium (EGM-2 Single Quots, CC-4176; Lonza) except when serum starved, when cells were cultured in endothelial cell growth medium alone. All cells were maintained at 37°C in a humidified 5% CO₂ incubator. 

PA, LF, and LF-E687C (E687C), an inactive mutant of LF, ¹⁵ were expressed in an attenuated strain of B. anthracis (BH445) and were purified as previously described. ^16,17 Protein concentrations were estimated using the bicinchoninic acid method and by densitometric analyses of Coomassie blue–stained polyacrylamide gels. 

In Vitro Lethal Toxin Activity

ARPE-19 and SVECs were seeded in six-well plates at approximately 30% confluence (200,000 cells/well). After overnight incubation, cells were transiently treated with LeTx (1 μg PA) and LF (100 ng). After 30-minute incubation, cells were rinsed and media were replaced by fresh media lacking LeTx. Cells were then collected at 0 hour (no treatment) or 30 minutes, 1 hour, 3 hours, 6 hours, 12 hours, 24 hours, 72 hours, or 168 hours (1 week) after LeTx treatment. ERK activity was then assessed by immunoblot analysis with phosphospecific antibodies. 

OIR and Intravitreal Injections

All experiments were conducted in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. All procedures were approved by the Van Andel Institute Institutional Animal Care and Use Committee. C57BL/6 mice were obtained from the Van Andel Institute's breeding colony. Mice were maintained with a 12-hour light/12-hour dark cycle and had free access to food and water. The mouse model of OIR was performed exactly as described. ^3,18 Briefly, litters of 1-week-old mice were exposed to 75% oxygen in a hyperoxygen chamber (model 1300; Sechrist Industries, Anaheim, CA) for 5 days (P7–P12) and then were removed to room air. Mice aged P14 and older were anesthetized by intraperitoneal injection of anesthesia (Avertin; 600 μL/20 g body weight). Mice younger than P14 were anesthetized by hypothermia, as described. ¹⁹ Intravitreal injections were performed as described. ^19,20 When necessary, the eyelids were separated (Straight 15° Optimum Knife; BD Beaver, Becton Dickinson, Franklin Lakes, NJ), and eyes were proptosed with gentle pressure. Intravitreal injections were performed under a dissecting scope using a 32-gauge Hamilton needle and syringe to deliver 1 μL injectate. For E687C/PA and LeTx treatments, intravitreal injections containing 0.1 μg PA plus 0.02 μg LF or 0.02 μg E687C were delivered in a volume 1 μL (1 standard dose; 1 SD). The dose of VEGF-neutralizing antibody (VEGF-NA; R&D Systems, Minneapolis, MN) used was based on the ND50 value range of 0.05 to 0.15 μg/mL. Therefore, VEGF-NA was used at a 1× dose (150 pg/mL) and a 10× dose (1500 pg/mL), in conjunction with either 1 SD LeTx or 1 SD E687C/PA (0.5 μL each for a final delivery volume of 1 μL). After treatment delivery, eyes were repositioned, lids were approximated over the cornea as necessary, and a topical antibiotic (Neomycin and Polymyxin B Sulfates and Dexamethasone Ophthalmic Ointment; Bausch & Lomb, Inc., Rochester, NY) was applied. 

Immunoprecipitation and Immunoblotting

To analyze the level and phosphorylation status of VEGFR2, we performed immunoprecipitation of HUVEC lysates and retinal lysates as follows. HUVECs were serum starved for 24 hours and then treated for 6 hours with either E687C/PA or LeTx, as described. ¹⁷ To induce the phosphorylation of VEGFR2, cells were pulsed with 50 ng/mL VEGF (rhVEGF165; R&D Systems) for 2 minutes at 37°C. Cells were then lysed in ice-cold RIPA buffer, sonicated, and quantified for protein concentration as described. Retinas from mice that underwent OIR were treated with either E687C/PA or LeTx on P14 and collected 3 days after treatment (P17). Lysates were prepared as described. HUVEC lysates (200 μg) or retina lysates (500 μg) were precleared with Protein A-agarose slurry (Roche Applied Science, Indianapolis, IN) and incubated overnight at 4°C with an antibody against VEGFR2 (1:100; Cell Signaling, Beverly, MA). Protein A-agarose slurry was added, and samples were reincubated for another hour at 4°C. Samples were washed in RIPA buffer, pelleted, and resuspended in 50 μL of 2× sample buffer. ¹⁷  

Immunoblotting was performed as described previously. ¹⁹ For immunoprecipitations, 25 μL each sample was run on an 8% denaturing SDS-PAGE gel, transferred to polyvinylidene difluoride (PVDF) membranes, and probed with the following antibodies: antiphosphotyrosine (pTyr clone 4G-10; 1:1000; Millipore, Billerica, MA) to detect phosphorylated VEGFR2, and VEGFR2 (1:1000; Cell Signaling) to verify the immunoprecipitation procedure. In addition, whole cell lysates from which the immunoprecipitations were generated were analyzed for VEGFR2 to determine the levels of VEGFR2 after LeTx treatment. An antibody against α-tubulin was used as a loading control (1:5000; Cell Signaling). To verify LeTx activity in HUVECs, lysates were probed with an antibody against the N-terminal region of MEK1 to verify MEK cleavage (1:2000; Upstate Biotechnology, Lake Placid, NY) and pERK to verify loss of MAPK activity (1:1000; Cell Signaling). To determine the kinetics of LeTx activity in vitro, cells were collected at the indicated time points after LeTx treatment in ice-cold RIPA buffer. ¹⁷ To analyze the basal activity of the MAPK pathways during OIR, eyes were collected from C57BL/6 mice at various time points after removal to room air, and retinas were dissected. Retinal tissue was homogenized in ice-cold RIPA buffer with a micropestle. Cells or homogenized tissue were then lysed in RIPA buffer, sonicated, and cleared by centrifugation. Protein concentrations were determined using the bicinchoninic acid protein assay (Pierce Biotechnology, Rockford, IL). Five micrograms of cell lysate, or the entire lysate of one retina for each time point, was run on a denaturing SDS-PAGE gel and transferred to PVDF membranes. Membranes were immunostained with the following antibodies: pERK 1:1000 (Cell Signaling), ERK 1:1000 (Cell Signaling), pp38 1:1000 (Cell Signaling), p38 1:1000 (Cell Signaling), pJNK 1:1000 (Cell Signaling), and JNK 1:1000 (Cell Signaling). Membranes were probed with appropriate horseradish peroxidase–conjugated secondary antibodies at 1:2000 (KPL, Inc., Gaithersburg, MD). Specific proteins were visualized by chemiluminescence (LumiGLO Chemiluminescence Substrate; Cell Signaling), followed by exposure to Kodak film (Biomax Light; Eastman Kodak, Rochester, NY). Immunoblots were stripped and reprobed with antibodies against either β-actin (1:2000; Sigma, St. Louis, MO) or GAPDH (1:2000; Cell Signaling) for a loading control. 

Immunofluorescence and Immunohistochemistry

For retinal whole mounts, eyes were enucleated at the indicated time points and were fixed in 4% paraformaldehyde for 2 hours on ice. Retinas were dissected as described. ^19,21 Briefly, eyes were bisected equatorially, the cornea and the lens were removed, and the retina was separated from the optic cup using a dissecting microscope. Retinas were then postfixed for 1 hour. Retinas were blocked in 5% normal goat serum for 2 hours at room temperature, followed by incubation at 4°C overnight in Alexa 594–conjugated isolectin Griffonia simplicificolia-IB4 (GSA; 1:200, Invitrogen) and FITC-conjugated α-smooth muscle actin (αSMA; 1:200, Sigma), or F4/80 (1:10; Abcam, Cambridge, MA) in 0.3% Triton X-100 in PBS. An Alexa Fluor 488 anti-rat IgG secondary antibody was used to visualize F4/80 staining (1:400; Invitrogen). Retinas were radially relaxed and flat mounted on glass slides with mounting medium (Fluoromount Aqueous Mounting Media; Sigma). For retinal cross sections, eyes were fixed and embedded in paraffin. Immunohistochemistry was performed with the use of optimized standard protocols on an automated stainer (Discovery XT; Ventana Medical Systems, Oro Valley, AZ) using a mouse-specific collagen, type IV antibody (AB756P, 1:50; Chemicon, Temecula, CA), PDGFRβ (1:100; Abcam), or F4/80 (1:10; Abcam). Horseradish peroxidase–conjugated anti-rabbit IgG (for collagen, type IV, and, PDGFRβ; UltraMap; Ventana Medical Systems) or anti-rat IgG (for F4/80) (Ventana Medical Systems) was used with diaminobenzidine to stain the slides. Hematoxylin and eosin staining was performed by standard methods. 

Vitreous Collection and Analysis for Cytokine Secretion

Vitreous fluid was collected as described. ¹⁹ Forty-eight hours after their removal to room air, mice were injected intravitreally with 1 SD of E687C/PA or 1 SD LeTx, or they were sham injected by inserting and then removing a 32-gauge needle. Mice were euthanized, and eyes were collected 3 days after injection (2 eyes per sample, 5 samples per treatment per time point). Eyes were rinsed in ice-cold isolation buffer (80 mM NaCl, 8 mM KCl, 2 mM EDTA pH 8.0, 50 mM Tris-Cl pH 7.2; Complete EDTA-free Protease Inhibitor Cocktail [Roche], 1 tablet per 50 mL). Eyes were then homogenized in 200 μL isolation buffer using a micropestle. Vitreal supernatant was collected by centrifugation at 12,000g for 30 minutes at 4°C and then was snap frozen. Samples were sent to Rules Based Medicine, Inc. (Austin, TX) for cytokine screening (Rodent MAP v2.0). 

Enzyme-Linked Immunosorbent Assay

Forty-eight hours after their removal to room air, mice were injected intravitreally with 1 SD of E687C/PA or LeTx. Mice were euthanized and collected 3 days after injection (2 eyes per sample, 3 samples per treatment per time point). Vitreous fluid was collected as described. The levels of vitreous angiopoietin 1 (Ang1) and angiopoietin 2 (Ang2) were measured using the ELISA kit for mouse Ang1 (E90008Mu) and Ang2 (E90009Mu) from Cedar Lane Laboratories (Burlington, NC). Twenty-five microliters of vitreous samples were run in duplicate for each kit according to the manufacturer's instructions. 

Microscopic Analysis

Epifluorescent images were acquired with a microscope (E600; Nikon, Tokyo, Japan) equipped with a camera (Spot Insight QE; Diagnostic Instruments, Sterling Heights, MI). Identical illumination and camera settings were used to obtain all images. Optimization for color, brightness, and contrast was performed using graphics editing software (Photoshop CS2; Adobe, San Jose, CA); all optimization was performed identically for all images. Areas of vaso-obliteration and neovascularization after OIR were quantified using Imagine software (developed in-house; available as freeware at https://github.com/erikor/Imagine) as follows. The total area of the retina per image was selected as the region of interest (ROI) using the free select tool and the ROI Area feature in the Analysis Methods section. Vaso-obliterated zones were measured in a similar fashion and are presented as a percentage area of the total ROI. We used a similar method to measure neovascular regions. Using the Free Select tool and the zoom feature as necessary, neovascular regions were traced based on visual inspection of the image. Neovascular areas were defined as areas that tended to fluoresce more brightly than surrounding normal vessels, had characteristic neovascular-tuft morphology, and appeared connected to normal vessels. ^18,22 These were added together to generate a total neovascularization for each image. Percentage neovascular area was then calculated based on the total area of the retinal ROI. This method of analysis was chosen because it avoids subjective bias that is introduced in setting intensity thresholds to compensate for batch-to-batch variation in staining intensity when measuring neovascularization based on pixel intensity. Moreover, side-by-side comparison of the two methods showed that measurement of pixel intensity alone resulted in the inclusion of nonneovascular regions that stain with high intensity or the exclusion of neovascular regions that stain with lower intensity but are clearly neovascular by visual inspection. Given that the use of pixel intensity is a common method to quantify neovascular regions, ^18,22,23 we requantified all images based on the methods described by Connor et al. ¹⁸ As a consequence of the issues we encountered using pixel intensity as described, this approach consistently generated smaller values than we observed with Imagine software. Despite this, at least four images per retina were analyzed, and values from all images per retina were averaged. 

For macrophage quantification, at least three retinas per treatment per time point were analyzed. Retinas were whole mounted after immunofluorescent staining for GSA and F4/80, as detailed. F4/80, a microglia and macrophage-specific marker, and GSA, which stains activated macrophages in addition to vasculature, can be used to effectively colabel activated macrophages in the retina. ^{24 –26} Double-positive cells, as identified by their ameboid, globular morphology in the nerve fiber layer (NFL), were counted in a masked fashion and expressed as the number of cells per square millimeter of retina. 

Statistical Analysis

Data are presented as mean ± SD. Statistical significance was evaluated using the Student's t-test. P < 0.05 was considered significant. 

Results

MKK Pathways Are Activated in OIR

To determine whether MKK signaling is activated in OIR, we performed immunoblots with antibodies against their substrate MAPKs over the course of OIR. We were unable to detect active MAPK signaling on the day mice were removed to room air (P12) (Fig. 1). However, using phosphospecific antibodies, we observed the activation of ERK 1/2 and the p38 MAPK pathways within 1 day of exposure to room air (P13). Phosphorylation of ERK 1/2 and p38 MAPK was sustained through the period of neovascularization (which is maximal at P17) and into the regression and recovery phase of this model (P21). A similar initial pattern of activation was observed using antibodies directed against phosphorylated JNK (Supplementary Fig. S1). However, in this case JNK signaling decreased past P17. These data demonstrate MKK signaling pathways are active in OIR. 

Figure 1.

 

MAPK pathways were activated after OIR. MAPK pathway activation during OIR was analyzed by Western blot analysis. Retinal lysates were made from mice that underwent OIR. Eyes were collected at indicated time points, from P12 through P21. Immunoblots indicate the activation of MAPK signaling by phosphorylated ERK (pERK) and phosphorylated p38 (pp38). Total ERK (ERK) was analyzed to ensure no change in MAPK protein levels, and β-actin was used as a loading control.

Inhibition of MKK Signaling Causes Sustained Inhibition of OIR

To test the necessity of MKK signaling in OIR, we injected LeTx at time points after the return to room air. LeTx inhibits MAPK signaling by cleaving and inactivating MKKs, the upstream activators of MAPK. ^{12 –14,27,28} Initially, we chose to target retinas 48 hours after removal to room air (P14), shortly after MAPK signaling was activated but before the initiation of neovascularization. Three days after treatment (P17), we observed significantly elevated vaso-obliteration and reduced neovascularization compared with eyes injected with inactive LeTx (E687C/PA) or sham-injected eyes (Figs. 2A, 2D, 2G). Five days after LeTx treatment (P19), we observed that though control-treated retinas had recovered the majority of their vaso-obliterated zones (Fig. 2B), LeTx-treated retinas maintained large areas of vaso-obliteration (Fig. 2E). Histologic analysis of retinas at these time points showed no evidence of LeTx affecting neuronal organization, and all cell layers of the retina looked comparable to control-treated eyes (Supplementary Fig. S2). Two weeks after the administration of LeTx (P28), we observed that LeTx-treated retinas still had a significant amount of vaso-obliteration, whereas control-treated retinas had fully revascularized their vaso-obliterated zones and regressed neovascular lesions (Figs. 2C, 2F, 2I). Moreover, those regions of LeTx-treated eyes that retained vessels had a substantially reduced density compared with control-treated eyes (Fig. 2F). Collectively, these results indicate that the blockade of MKK signaling causes sustained inhibition of OIR. 

Figure 2.

 

MAPK pathway inactivation by LeTx persistently delays the development and progression of OIR. One-week-old mice were subjected to 75% oxygen for 5 days and then removed to room air (P7–P12). Forty-eight hours after removal to room air (P14), mice were intravitreally injected either with E687C/PA (0.02 μg E687C and 0.1 μg PA in 1 μL) (A–C) or with LeTx (0.02 μg LF and 0.1 μg PA in 1 μL) (D–F). Eyes were collected 3 days (A, D), 5 days (B, E), or 14 days (C, F) after treatment, stained with GSA, and flat mounted. Areas of vaso-obliteration (% vaso) and neovascularization (% nv) were quantified for retinas analyzed 3 days after treatment (G) or 5 days after treatment (H) and represented as a percentage area of the total retina. The area of vaso-obliteration for all analyses after treatment is also represented as a percentage from P12, which was set to 100% (I). At least four retinas per treatment were analyzed, and error bars indicate SD. *P < 0.02; **P = 0.05; §P = 0.005; ¥P < 0.04. All statistical values were compared with E687C/PA treatment. Scale bar, 100 μm.

These results may be explained if LeTx caused a persistent inhibition of MKK signaling. However, the injection of LeTx immediately after reintroduction to room air (P12), when MAPK signaling pathways were not active (Fig. 1), had no effect on neovascular growth or vaso-obliterated area up to 5 days after treatment (Fig. 3). If LeTx did cause a persistent inhibition of MKK signaling, we would have predicted P12 injections would have an effect similar to that of P14 injections. These results suggest the effects of LeTx are short-lived. This may be explained if free LeTx was rapidly removed from the vitreous or if cells became impervious to toxin and MKK activity was restored. This interpretation is consistent with in vitro observations showing the restoration of MKK activity in ARPE19 and SVECs within 6 hours and 72 hours, respectively, after transient (0.5 hour) treatment with LeTx (Supplementary Fig. S3). 

Figure 3.

 

The development and progression of OIR are unaffected by MAPK pathway inactivation at the onset of hypoxia. One-week-old mice were subjected to75% oxygen for 5 days. On removal to room air, mice were intravitreally injected either with E687C/PA (0.02 μg E687C and 0.1 μg PA in 1 μL) (A, B) or LeTx (0.02 μg LF and 0.1 μg PA in 1 μL) (C, D). Eyes were collected 3 days (A, C) or 5 days (B, D) after treatment, stained with GSA, and flat mounted. Areas of vaso-obliteration (% vaso) and neovascularization (% nv) were quantified for retinas analyzed 3 days after treatment (E) or 5 days after treatment (F) and were represented as a percentage area of the total retina. At least four retinas per treatment were analyzed. Error bars indicate SD.

To determine how long the retina retained dependence on MKK signaling for revascularization, we performed injections at 4 days (P16) or 6 days (P18) after removal to room air. At 4 days, the hypoxic retina is approaching maximal neovascular growth, whereas at 6 days, revascularization of the vaso-obliterated zones is beginning. ⁶ LeTx treatment at 4 days after removal to room air had no effect on the extent of neovascularization but prevented revascularization of vaso-obliterated regions up to 12 days after treatment (Figs. 4A–D, 4K). Specifically, we noted that though control-treated retinas had no measurable vaso-obliteration or neovascular regions (Fig. 4K; E687C at P28), LeTx-treated retinas had significant vaso-obliteration and minor areas of neovascularization (Fig. 4K; LeTx at P28). Interestingly, LeTx administration at 6 days after removal to room air did not significantly alter the extent of neovascular regression or revascularization of the vaso-obliterated regions compared with E687C/PA-treated eyes (Figs. 4E–J, 4L). Taken together, these data indicate that MKK signaling pathways are critical for the formation of pathologic angiogenesis between P12 and P16, coincident with a time of extensive hypoxia in the retina. However, once neovascularization or revascularization has begun, the involvement of these pathways is less critical. 

Figure 4.

 

MAPK pathway inhibition persistently maintains vaso-obliteration during the neovascular phase of OIR, whereas regression and recovery are unaffected. One-week-old mice subjected to 75% oxygen for 5 days were intravitreally injected either 4 days (P16) (A–D) or 6 days (P18) (E–J) after removal to room air. Retinas were injected with either with E687C/PA (A, C, E, G, I) or LeTx (B, D, F, H, J). For P16 injections, eyes were collected 3 days (A, B) or 12 days (C, D) after treatment, whereas P18-injected eyes were collected 3 days (E, F), 5 days (G, H), or 7 days (I, J) after treatment. Retinas were isolated, stained with GSA, and flat mounted. Extent of vaso-obliteration (% vaso) and neovascularization (% nv) was quantified and represented as a percentage area of the total retina (K, 3 days after treatment; L, 12 days after treatment). At least four retinas per treatment were analyzed. *P < 0.03; **P < 0.003; §P = 0.01. All statistical values are compared to E687C/PA treatment. Scale bar, 100 μm.

Persistent Inhibition of OIR by LeTx Does Not Act through Modulation of the Angiopoietin or VEGF Pathway

In tumor studies, treatment of cultured cells with LeTx prevents the release of proangiogenic cytokines, including VEGF. ²⁹ Thus, our observations may be explained if LeTx is shown to prevent the release of such factors in the retina. Accordingly, we measured levels of proangiogenic cytokines in the vitreous of LeTx-treated eyes. 

The Tie2 receptors and their ligands, the angiopoietins (Ang), have been shown to play critical roles in vessel growth and maturation (see Refs. 30, 31 for reviews). We therefore asked whether secreted levels of Ang1 and Ang2 are altered in response to LeTx treatment. Analysis of vitreous 3 days after LeTx administration on P14 during OIR demonstrated no change in either Ang1 or Ang2 secretion compared with controls (Supplementary Fig. S4), indicating that the persistent inhibition of vessel growth during OIR by LeTx is not due to modulation of these factors. 

VEGF is another important regulator of angiogenesis and has been shown to be a critical factor in the development of diabetic retinopathy, age-related macular degeneration, and retinopathy of prematurity (see Ref. 32 for review). In previous studies we have observed that treatment of cells with LeTx in vitro causes decreased release of VEGF to culture medium. ^29,33 By analogy, we hypothesized that LeTx-mediated persistent inhibition in OIR was caused by a loss of VEGF signal in the hypoxic retina. Therefore, we measured the levels of VEGF in vitreous 3 days after LeTx treatment. Contrary to our expectations, we observed VEGF was sixfold elevated after LeTx treatment (Fig. 5). These results indicate that not only is VEGF present in the retina after LeTx treatment but also that the LeTx-treated retina failed to appropriately respond to the increased levels of VEGF. 

Figure 5.

 

Vitreal levels of VEFG were significantly higher after LeTx treatment during OIR. Mice were subjected to 75% oxygen for 5 days, removed to room air for 48 hours, and either sham treated (sham) or treated with E687C/PA (E687C) or LeTx. Vitreous was collected 3 days after treatment, and VEGF levels were measured as detailed in Methods. Five vitreal samples per sham and E687C treatment and six vitreal samples per LeTx treatment were analyzed. *P < 0.03 compared with sham and E687C treatments.

Because the overexpression of growth factors can lead to effects opposing their physiologic functions, ^34,35 we wondered whether the dramatic increase in VEGF secretion after LeTx treatment could suppress retinal angiogenesis. Therefore, we coinjected a VEGF-neutralizing antibody (VEGF-NA) along with LeTx 2 days after return to room air (P14). As expected, the administration of VEGF-NA alone inhibited the revascularization of vaso-obliterated regions of the retina (Fig. 6A). However, coinjection of up to 10-fold more VEGF-NA did not rescue the persistent vaso-obliteration in LeTx-treated retinas (Fig. 6B). Together, these data indicate that LeTx blocks the retina's ability to respond to VEGF. 

Figure 6.

 

Persistent inhibition of neovascularization by LeTx does not act through the VEGF pathway. (A, B) Mice were subjected to 5 days of 75% oxygen and then removed to room air. On P14, mice were intravitreally injected with E687C/PA, LeTx, or LeTx in conjunction with 1× VEGF-neutralizing antibody (1× + LeTx) or 10× VEGFNA (10× + LeTx). Eyes were collected 5 days after treatment, stained with GSA, and flat mounted for microscopic analysis and quantitation of vaso-obliteration and neovascularization. (A) Effect of 1× (150 pg/mL) dose of VEGF neutralizing antibody (VEGFN-NA) compared with sham treatment (no treatment delivered) on development and progression of OIR. Mice were intravitreally injected on P14 and analyzed on P17. (B) Extent of vaso-obliteration (% vaso) and neovascularization (% nv) after cotreatment of LeTx with VEGF-NA. *P < 0.003; **P < 0.009; §P = 0.01. All statistical analyses are compared with E687C/PA treatment. (C) Extent of phosphorylation of VEGFR2 in HUVECs was analyzed by immunoprecipitation followed by Western blot analysis. Serum-starved HUVECs were left without treatment (NT) or were treated with E687C/PA (E687C) or LeTx followed by VEGF treatment (V). Lysates were immunoprecipitated with a VEGFR2 antibody and blotted for pTyr or VEGFR2. Levels of VEGFR2 were measured in whole cell lysates. α-Tubulin was used as a loading control. (D) LeTx activity in HUVECs was confirmed by Western blot analysis for loss of N-terminal MEK1 epitope and decreased pERK expression.

The inability of the retina to respond to elevated VEGF after LeTx treatment may be due to the altered regulation of VEGF receptors, most notably VEGFR2, ^36,37 either by regulation of the receptor levels or activation of the receptor. To test this, we analyzed the levels of phosphorylated VEGFR2 by immunoprecipitation (for VEGFR2), followed by Western blot analysis (for pTyr). HUVECs were used as a surrogate for retinal cells because whole retinal tissue from mice that underwent OIR had minimally detectable levels of VEGFR2 (Supplementary Fig. S5). Treatment of HUVECs with VEGF caused the phosphorylation of VEGFR2 (Fig. 6C). Exposure of HUVECs to LeTx for 6 hours before VEGF treatment did not alter VEGFR2 expression or activation (Fig. 6C). Analysis of cleavage of MEK1 (by loss of N-terminal epitope) and pERK levels after LeTx treatment verified the activity of LeTx in HUVEC cultures. These data indicate that the inability of the retina to respond to elevated VEGF cues after LeTx administration was not due to changes in VEGFR2 levels or activation. 

Levels of Vitreal Growth Factors and Cytokines Are Unaltered by LeTx-Treatment whereas Macrophage Recruitment Is Minimally Affected

Although LeTx does not prevent angiopoietin or VEGF release, it may block the release of other growth factors or cytokines responsible for recruiting macrophages or circulating endothelial progenitors to sites of neovascularization. To test this we examined the expression of a panel of 58 growth factors and cytokines in the vitreous 3 days after LeTx treatment. Surprisingly, though the levels of 23 of these factors were significantly altered in the OIR model system compared with mice kept at room air (Table 1; Supplementary Table S1), none of these was affected by LeTx-treatment (Supplementary Table S2). These include a variety of macrophage-recruitment factors such as MCP and MIP family members (Supplementary Table S2), which have previously been shown to be elevated in OIR retinas using this model, ³⁸ and in other cytokines not previously known to be regulated in OIR, including interleukin (IL)-4, IL-7, leukemia inhibitory factor (LIF)-1, and serum amyloid P (SAP). These results indicate that OIR alters the vitreous levels of numerous cytokines and growth factors. However, with the exception of VEGF, LeTx treatment does not significantly alter the expression of cytokines and growth factors induced by OIR. 

Table 1.

 

Measurement of Secreted Cytokines Comparing OIR with Normoxic (Normoxia) Retinas

Cytokine	Normoxia*	OIR*	Fold Change	P
CRP, μg/mL	0.011 ± 0.002	0.025 ± 0.007	2.3	0.01
EGF, pg/mL	4.02 ± 0.21	8.15 ± 0.96	2.0	0.0004
Factor VII, ng/mL	0.575 ± 0.06	12.46 ± 2.63	21.7	0.0005
bFGF, ng/mL	21.83 ± 2.76	349.2 ± 188.61	16.0	0.02
GCP-2, ng/mL	0.08 ± 0.01	0.03 ± 0.02	0.375	0.006
Haptoglobin, μg/mL	0.45 ± 0.02	0.90 ± 0.22	2.0	0.01
IFN-γ, pg/mL	4.60 ± 1.13	11.42 ± 3.04	2.5	0.005
IL-4, pg/mL	7.89 ± 1.60	25.82 ± 7.27	3.3	0.004
IL-7, ng/mL	0.025 ± 0.003	0.065 ± 0.026	2.6	0.02
IL-12, ng/mL	0.018 ± 0.008	0.04 ± 0.007	2.2	0.02
IL-18, ng/mL	0.30 ± 0.07	1.12 ± 0.54	3.7	0.03
IP-10, pg/mL	9.96 ± 2.01	21.70 ± 8.74	2.2	0.04
LIF-1, pg/mL	95.30 ± 6.71	298.0 ± 30.20	3.1	0.00005
MCP-3, pg/mL	8.88 ± 0.87	15.7 ± 3.32	1.8	0.008
MCP-5, pg/mL	1.14 ± 0	2.79 ± 1.29	2.4	0.05
M-CSF, ng/mL	0.05 ± 0.006	0.08 ± 0.02	1.6	0.03
MIP-1α, ng/mL	0.06 ± 0.01	1.02 ± 0.22	17.0	0.0006
MIP-1β, pg/mL	13.33 ± 0.98	51.40 ± 15.59	3.9	0.005
MIP-2, pg/mL	1.58 ± 0.36	5.03 ± 1.79	3.2	0.01
SAP, μg/mL	0.07 ± 0.015	0.14 ± 0.05	2.0	0.02
TIMP-1, ng/mL	0.11 ± 0.004	0.44 ± 0.12	4.0	0.004
VEGF-A, pg/mL	60.87 ± 1.45	165.80 ± 52.26	2.7	0.01

 CRP, C-reactive protein; EGF, epidermal growth factor; bFGF, basic fibroblast growth factor; GCP-2, granulocyte chemotactic protein-2; IFN-γ, interferon gamma; IP-10, inducible protein 10; MCP-3, monocyte chemoattractant protein 3; MCP-5, monocyte chemoattractant protein 5; M-CSF, macrophage colony stimulating factor; TIMP-1, tissue inhibitor of metalloproteinase type 1.

*  ±SD of three independent experiments for normoxia and five independent experiments for OIR.

Immunohistochemical staining for F4/80 indicated that although LeTx-treated retinas lack neovascular tufts (compare Figs. 7A and 7B), macrophages are still present in the ganglion cell layer, indicating that LeTx treatment did not affect the presence of macrophages in OIR retinas. To confirm that macrophage recruitment is unaffected by MKK inhibition, we quantified their presence in LeTx-treated retinas that were fluorescently stained with GSA and F4/80 (Figs. 7C, 7D). Although analysis of LeTx-treated retinas revealed a decreasing trend in the number of macrophages present in OIR retinas at a variety of time points after LeTx treatment (Fig 7E), the differences did not reach statistical significance. These observations show LeTx has modest to no effect on the ability of retinas to recruit and maintain macrophages during OIR. 

Figure 7.

 

Macrophage recruitment is unaffected by LeTx during OIR. (A, B) Mice subjected to 5 days of 75% oxygen were intravitreally injected 48 hours after removal to room air with E687C (A) or LeTx (B) and were analyzed for the presence of macrophages by F4/80 immunohistochemical staining of paraffin-embedded sections. Arrows: F4/80-positive staining in both retinas. (C, D) Macrophages were visualized in retinal whole mounts by immunofluorescence for GSA (red) and F4/80 (green). Arrows: macrophages. (E) Macrophages were quantified from whole mounts of retinas stained with GSA and F4/80 at the indicated ages after treatment on P14 with sham injection, E687C, or LeTx. There are no statistically significant differences in the number of macrophages between the treatments at any analyzed age after injection.

Sensitivity of Vessels during OIR Corresponds to Maturation Status

Previously we have shown that MKK inhibition by LeTx can delay sprouting angiogenesis into the deeper layers of the retina without affecting the mature superficial plexus. ¹⁹ This suggests LeTx may preferentially target nascent blood vessels and would explain why LeTx opposes OIR 2 to 4 days after removal to room air (P14–16), but not immediately after removal (P12) or after the peak in neovascular growth (P18). To test whether targeted vessels are immature, we compared the extent of maturation of normal developmental vessels (at room air) and vessels that develop during OIR by immunofluorescence or immunohistochemistry with three markers of vessel maturation: collagen IV, platelet-derived growth factor receptor β (PDGFR-β), and α-smooth muscle actin (αSMA). Collagen IV is a component of the basement membrane surrounding blood vessels and is an early marker of vessel integrity. PDGFRβ is expressed in pericyte progenitor cells and is required for PDGF-mediated recruitment of pericytes to the endothelium of growing capillaries. ³⁹ αSMA is expressed by vascular smooth muscle cells and pericytes associated with mature arteries and arterioles. ^40,41 Comparison of αSMA staining indicated no difference in arterial maturation between retinas maintained at room air and retinas that underwent 5 days of high oxygen (P12 retinas under both conditions; Supplementary Fig. S6). Given that these vessels within the retina are unaffected by LeTx treatment during OIR (see Figs. 2, 4), these data suggest that mature vessels are insensitive to MAPK inhibition by LeTx. During OIR, at P14 collagen IV was detected on vessels in the ganglion cell layer and in vessels extending through the inner plexiform layer (Fig. 8A). In comparison, PDGFRβ staining at this stage was concentrated on vessels in the ganglion cell layer of the inner vascular plexus only (Fig. 8B, arrowheads). Compared with room air retinas at this same time (P14), there is remarkably reduced deep vessel plexus growth (collagen IV staining) and reduced PDGFRβ staining in the ganglion cell layer of the inner vascular plexus (Supplementary Figs. S7C, S7D). In contrast, after P16, OIR (P17 OIR) retinas unaffected by LeTx treatment showed PDFGRβ staining within all regions of the retina that contained vessels (Figs. 8C, 8D), including through the inner nuclear layer (Fig. 8D, arrowheads). These results indicate vessels of the P14 OIR retina are immunohistochemically most similar to immature vessels observed in the developing P3 retina. 

Figure 8.

 

Lack of pericyte coverage corresponds to LeTx sensitivity in OIR. Eyes from mice subjected to 5 days of 75% oxygen (P14 OIR: A, B; P17 OIR: C, D) were collected at the indicated ages, paraffin-embedded, and stained with an antibody against collagen IV, a component of the basement membrane (A, C), or PDGFRβ, a marker of pericytes (B, D). Tissue sections were counterstained with hematoxylin to identify all the cellular layers of the retina. Arrowheads: regions of superficial vascular growth with pericyte coverage. Scale bar, 20 μm.

The preceding data suggest immature retinal vessels may be particularly sensitive to LeTx. To test whether LeTx preferentially targets immature vessels, its effects on the retinas of P3 mice and P21 (fully mature) mice were compared. Examination of P3 retinas 4 days after LeTx treatment (P7) revealed a complete destruction of vessels (Figs. 9, 9E). This effect persisted up to 8 days after LeTx administration (P11; Supplementary Fig. S8). Histologic analysis revealed the vitreous cavity had filled with blood in LeTx-treated eyes compared with control-treated eyes, suggesting a loss of vascular integrity within the retina and continued destruction of developing vessels at this time (Figs. 9B, 9F). Closer inspection of LeTx-treated retinas revealed a lack of organized superficial plexus formation compared with control-treated retinas (Figs. 9C, 9G). Consistent with an earlier report, ⁴⁰ these vessels contained both collagen IV and PDGFRβ staining in the superficial vessels (Supplementary Figs. S7A, S7B). Interestingly, the extent of PDGFRβ staining in the superficial plexus at P3 looked remarkably like that of OIR P14 (compare Supplementary Figs. S7B and S8B). Although the superficial staining of PDGFRβ at these immature and hypoxic time points was strong, covering the length of the superficial plexus, staining in the less hypoxic, more mature time points (OIR P17 [Fig. 8D] and WT P14 [Supplementary Fig. S7D]) was less intense and more sporadic over this region of the retina. This staining pattern suggests a shift in maturation level of the vessels, which correlates with LeTx sensitivity. Surprisingly, PDGFRβ was also found in regions of the retina that lacked collagen IV staining (open arrow in Supplementary Fig. S7A), suggesting the recruitment of nonfunctional pericytes at this time. In contrast, LeTx had no effect on the organization of retinal vasculature when injected into the vitreous of adult (P21) mice (Figs. 9D, 9H). Although it has been suggested that pericyte coverage (as demonstrated by PDGFRβ staining) alone is not sufficient to determine vessel maturation, ⁴² the correlation of PDGFRβ staining and LeTx sensitivity during OIR suggests that MKK activity is critical for activities associated with the formation of immature vessels in this model system. 

Figure 9.

 

MAPK pathways are critical for superficial plexus development in the retina. Mice at P3 (A–C, E–G) or P21 (D, H) were intravitreally injected with either E687C/PA (A–D) or LeTx (E–H). Mice treated on P3 were collected 4 days after treatment (P7), whereas mice treated on P21 were collect 7 days after treatment (P28). Retinas were dissected, stained with GSA, and flat mounted (P3: A, E; P21: D, H) to visualize the extent of vascular growth of the superficial plexus. Eyes were collected, paraffin embedded, sectioned, and either analyzed for pathology by hematoxylin and eosin (B, F), or immunostained for GSA to visualize pathology across the entire thickness of the retina (C, G). DAPI was used as a nuclear stain to visualize all cellular layers of the retina. Scale bars: (A, D) 100 μm; (B, C) 20 μm.

Discussion

Angiogenesis, the process by which new blood vessels form from a preexisting network of vessels, is essential during both normal development and disease. This complex process initiates when cells respond to hypoxia, either because of increased metabolic demands as tissues grow or when oxygen delivery is inhibited because of injury or an underlying pathologic insult. The upregulation of angiogenic factors, such as vascular endothelial growth factor (VEGF), under hypoxic conditions is controlled by the transcriptional factor HIF-1α (see Ref. 43 for review). HIF-1α is essential for angiogenesis during development, for tumor growth, and after ischemic insult in tissues, including the retina. Although developmental angiogenesis is tightly controlled to allow for the proper growth and formation of a functional vascular network, during pathologic insult the demand for an increased oxygen supply results in the deregulation of angiogenic factors, leading to the formation of aberrant vasculature that is functionally impaired. ⁴⁴ Such is the case with retinal neovascular diseases, such as ROP and proliferative diabetic retinopathy, which are caused by unregulated and abnormal blood vessel growth, resulting from the increased release of proangiogenic factors, the most notable of which is VEGF. ^1,2 A central component of these diseases is the growth of abnormal blood vessels that are leaky, tortuous, lack pericytes, and have an irregular distribution, resulting in severe vision loss and blindness. Current treatment strategies focus on targeting VEGF; however, it is clear that the process of retinal angiogenesis is complex, involving many other growth factors and signaling pathways that are potential targets for novel therapeutics (see Ref. 32 for review). 

In this report, we show that MKK signaling is essential for hypoxia-driven pathologic neovascularization in the retina. MKK signaling was activated in OIR, and inhibition of these pathways with anthrax LeTx resulted in the persistent inhibition of neovascularization in an experimental model of ROP. The requirement for MKK signaling was restricted to a period marked by hypoxic conditions, vessel immaturity, and the onset of MAPK activation. The induction of hypoxia was accompanied by marked increases of several secreted proteins in the vitreous (Supplementary Table S2). These included MCP and MIP family members previously been linked to macrophage recruitment in OIR, in addition to others, such as IL-4, IL-7, LIF-1, and SAP, not previously demonstrated to change in OIR. The multifaceted cytokine response induced by OIR suggests an underlying complexity that we do not yet fully understand. Contrary to an earlier report on the role of JNK signaling in OIR, ¹⁰ MKK inhibition did not cause decreased VEGF expression. Rather, the LeTx-treated retina failed to generate vessels in the presence of elevated VEGF. Furthermore, LeTx treatment had no effect on the secretion of any of the 23 other factors significantly altered during OIR or on Ang1 or Ang2. However, based on our results, the mechanism by which LeTx prevents vascular growth during OIR is independent of VEGF expression and VEGFR2 activation. 

The persistent delay in vascular growth occurred only during a narrow time frame during OIR, when neovascularization had not yet begun but the retina had already become hypoxic. This time frame corresponds with the onset of neovascular growth, before pericyte coverage of newly formed vessels, suggesting that immature vasculature is uniquely sensitive to MAPK pathway inhibition. Similarly, we and others ^{28,29,33,45,46} have shown that LeTx blocks perfusion and decreases vascular density in tumor xenograft models. Like immature vessels, tumor-associated blood vessels grow under hypoxic conditions, and have reduced pericyte coverage. The hypoxic nature of the tumor environment and the lack of pericytes may render these vessels sensitive to LeTx. The observation that endothelial cells lacking pericytes in a hypoxic environment are particularly sensitive to MKK inhibition indicates this is a vulnerability that may also be targeted for the control of tumor growth. Interestingly, physiologically developing vessels (P3) were also sensitive to LeTx administration, even though areas of retinal vascular growth stained positive for PDGFRβ, a pericyte marker. Although PDGFRβ staining is not indicative of pericyte function, this suggests that pericyte presence alone does not confer protection from the effect of LeTx. It is interesting that collagen IV, an early marker of vessel maturity, is also reported to be a ligand for CMG2, ⁴⁷ one of the cellular receptors for LeTx. ⁴⁸ This raises the intriguing possibility that the insensitivity of mature vessels is caused by receptor occupancy by collagen IV or other components of the basement membrane surrounding blood vessels. Finally, because there are indications that the vaso-obliterated zone in OIR differs from the avascular retina during normal development, ⁶ our results imply that there are different mechanisms by which the superficial plexus forms during normal vascular development and during proliferative retinopathies, as modeled by OIR. Regardless, the extent of hypoxia correlated with LeTx sensitivity, suggesting that other signals provided by the hypoxic environment during early vascular development and during OIR also contribute to the persistent effects by LeTx. 

Vascular damage and dysfunction are hallmarks of anthrax infection (see Refs. 49, 50 for reviews), and the classic anthrax hemorrhagic pathology is attributed to LeTx. ⁴⁹ LeTx induces vascular collapse and hemorrhage in rodent models, ⁵¹ vascular permeability in zebrafish, ⁵² and delays in sprouting angiogenesis in the developing mouse retina. ¹⁹ In vitro studies suggest endothelial cell apoptosis, ⁵³ endothelial cell barrier dysfunction, ⁵⁴ and reduced endothelial cell invasion and tube formation ⁵⁵ are due to a direct action of LeTx on endothelial cells. However, the relevant in vivo targets have not yet been identified. The fact that LeTx inhibits tumor angiogenesis in xenograft models without affecting the quiescent vascular system, while rapidly preventing perfusion to tumors, ²⁹ also suggests a direct action on endothelial cells or cells supporting their function. The data presented here show that the existing vascular network during OIR is unable to respond to hypoxia and VEGF-related cues to generate either pathologic or physiologic vessels (neovascularization or regrowth of vaso-obliterated zones, respectively) after LeTx treatment. Whether this was due to a direct effect on endothelial cell function is unclear; however, our data suggest the effect was likely not caused by direct targeting of VEGFR2 on endothelial cells themselves. 

The persistent inhibition of vascularization during OIR by LeTx is not caused by alterations in macrophage recruitment signals or the presence of macrophages in the hypoxic retina. These data do not rule out a potential effect of LeTx on macrophage function in the retina during OIR. Furthermore, though macrophages have been studied as a major cellular target of LeTx ^56,57 (and see Ref. 49 for review), more recent work ⁵⁸ has shown that macrophages are not required for the establishment of anthrax infection. Therefore, it is unlikely that the effects seen by LeTx during OIR are a consequence of a direct action on macrophages themselves. 

Although we did not find evidence to support a role for macrophage recruitment in the LeTx-treated retina during OIR, we have not excluded the possibility that LeTx prevents the recruitment of circulating endothelial progenitor cells (EPCs) to the retina. Although angiogenesis appears to be the major driving force behind pathologic angiogenesis, circulating EPCs have been implicated in contributing to new blood vessel formation in vivo under pathologic conditions. Although their role remains controversial in tumor angiogenesis (see Ref. 59 for review), recent evidence suggests that bone marrow–derived cells participate in retinal neovascularization. ⁶⁰ The persistent delay in neovascular growth in the OIR model by LeTx treatment may be due to a block in the recruitment or the participation of bone marrow–derived cells that are required for this neovascular growth. Further studies are required to determine what effect LeTx has on bone marrow recruitment to the retina. 

In summary, we report that MKK signaling plays a key role in the development and progression of neovascular growth using a mouse model of OIR. MKK inhibition does not alter the release of angiogenic signals but, rather, prevents response to them. Our results indicate that the inhibition of MKK signaling may form the basis of a therapeutic strategy to treat ROP and other retinopathies. Significantly, MKK inhibition affects only immature blood vessels, particularly those under hypoxic conditions in which a deep vessel plexus is lacking; therefore, existing mature blood vessels would not be impacted by this treatment. Moreover, because MKK inhibition prevents OIR by a mechanism independent of the VEGF-targeting strategies, this approach may be used in conjunction with existing strategies using VEGF inhibitors to prevent retinal neovascularization. 

Supplementary Materials

Figure sf01, PDF - Figure sf01, PDF 

Table st1, PDF - Table st1, PDF 

The authors thank Jim Resau and the staff at the Laboratory of Analytical, Cellular, and Molecular Microscopy, particularly Bree Berghuis and Lisa Turner, for excellent technical assistance in preparation and histologic staining of samples; members of the Alberts and Duesbery laboratories for helpful discussions; and Louis Glazer and Patrick Droste for critical reading of the manuscript. 

Footnotes

 Supported by The Van Andel Institute.

Footnotes

 Disclosure: J.L. Bromberg-White, None; E. Boguslawski, None; D. Hekman, None; E. Kort, None; N.S. Duesbery, None