NK1 Receptor Antagonists as a New Treatment for Corneal Neovascularization

Abstract

Purpose.: To determine whether the inhibition of Substance P (SP) activity can reduce corneal neovascularization (CNV) by means of local administration of high-affinity, competitive, tachykinin 1 receptor (NK1R) antagonists Lanepitant and Befetupitant.

Methods.: We performed a safety and efficacy study by using (1) two different C57BL/6 mouse models of CNV: alkali burn and sutures; (2) different concentrations; and (3) different routes of administration: topical or subconjunctival. Clinical examination endpoints, SP levels, CNV index, and leukocyte infiltration were measured.

Results.: Substance P increased after injury in the corneal epithelium of both CNV models, and later in the suture model. Topical Lanepitant was nontoxic to the ocular surface and effective in reducing hemangiogenesis and lymphangiogenesis, corneal SP levels, and leukocyte infiltration, as soon as 4 days later in the alkali burn model. Topical Lanepitant, up to 7 days, was ineffective in the suture model. However, subconjunctival Lanepitant was effective in reducing lymphatic CNV, leukocyte infiltration, and SP levels in the suture model, after 10 days. Additionally, in the alkali burn model, subconjunctival Lanepitant significantly reduced blood CNV, corneal perforation rate, opacity, and leukocyte infiltration, and improved tear secretion. Finally, topical application of Befetupitant reduced CNV in the alkali burn model but was toxic owing to the vehicle (dimethyl sulfoxide [DMSO]); hence, Befetupitant was not tested in the suture model.

Conclusions.: The NK1R antagonist Lanepitant is safe for the ocular surface and effective in reducing both corneal hemangiogenesis and lymphangiogenesis, and leukocyte infiltration. We suggest that inhibition of NK1R may represent an adjunctive tool in the treatment of CNV.

Italian Abstract

Introduction

The normal cornea is one of the few avascular tissues in the human body. Several ocular diseases, however, can cause corneal hemangiogenesis and lymphangiogenesis.¹ Indeed, corneal neovascularization (CNV) affects 4% of the US population, and it is a constant finding in severe corneal diseases, and represents the second cause of blindness worldwide. The growth of vessels into the cornea is clinically relevant because it induces loss of corneal transparence, due to infiltration of calcium, lipids, and inflammatory cells. Additionally, it significantly reduces effectiveness of the key surgical procedure used to restore corneal clarity (i.e., keratoplasty), as it increases the rate of graft rejection.² 

For these reasons, significant effort is being spent to find novel therapies targeting CNV. 

Although totally devoid of vessels, the cornea receives dense sensory innervation. Corneal nerves originate in the trigeminal ganglion, where neural bodies reside.³ Corneal nerve endings contain several neuropeptides.⁴ Between these, Substance P (SP), an 11-amino-acid polypeptide, member of the tachykinin family, is highly expressed.^5,6 Substance P is derived from preprotachykinin A protein and secreted by both neuronal and nonneuronal cell types (i.e., macrophages and dendritic cells).^7–9 It exerts its actions by binding to tachykinin receptors (NK1, NK2, and NK3),¹⁰ having the highest affinity for tachykinin 1 receptor (NK1R).¹¹ This is expressed on a variety of cell types including neurons and immune, endothelial, epithelial, and glial cells.^12–14 

Substance P is well known as a key mediator of inflammation^15,16 and wound healing.^17–19 More specifically, SP levels are increased in the murine cornea after alkali burn (Son Y, et al. IOVS 2004;45:ARVO E-Abstract 1423) and infection,²⁰ conditions commonly associated with CNV. Moreover, SP appears to stimulate CNV in vivo via a specific action on microvascular NK1R,²¹ since NK2 and NK3 receptor agonists have no significant effects on endothelial cell proliferation.²² 

Substance P receptor antagonists have been tested in several inflammatory diseases, including inflammatory bowel disease,²³ neurogenic²⁴ and liver²⁵ inflammation, and polymicrobial sepsis.²⁶ In the cornea, inhibition of SP ameliorated Pseudomonas keratitis.²⁷ Additionally, a recent study²⁸ has shown that treatment with the SP antagonist Spantide I significantly reduces the severity of herpes simplex virus-1, including reduced angiogenesis. 

However, the effect of selective NK1R inhibition in the cornea, and specifically in animal models of CNV, has not been described yet. The aim of this study was to test safety and efficacy of high-affinity, nonpeptide, competitive NK1R antagonists (Lanepitant and Befetupitant) in two different murine models of CNV. 

Materials and Methods

Animals

Female, 6- to 8-week-old, C57BL/6 mice (Charles-River, Calco, LC, Italy) were used for all experiments (total: 283 mice). Animals were allowed to acclimatize to their environment for 1 week before experimentation. Each animal was deeply anesthetized with intraperitoneal injection of Tribromoethanol (250 mg/kg) before all surgical procedures. Postoperatively, all animals received a single dose of Carprofen at 5 mg/kg subcutaneous. Carbon dioxide inhalation and subsequent cervical dislocation were applied to euthanize the animals. All experimental protocols were approved by the Animal Care and Use Committee of the San Raffaele Scientific Institute, in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Substance P Quantification in the Cornea

To quantify the amount of SP in healthy, alkali burn, or sutured corneas (before and after Lanepitant treatment), six samples per group (total: 81 mice) were pooled in pairs on days 4, 7, and 10 post injury in 500 μL RIPA buffer (Sigma-Aldrich Corp., St. Louis, MO, USA) containing protease inhibitor cocktail (Sigma-Aldrich Corp.). Each of three pools was homogenized on ice with UltraTurrax T8 (IKA, Wilmington, NC, USA). The samples were then clarified by centrifugation at 10,000g for 5 minutes at 4°C. An aliquot of each supernatant (1:500 dilution) was assayed in triplicate for SP protein by using a commercially available EIA (Enzyme Immunoassay) kit (Cayman Chemical, Ann Arbor, MI, USA), according to the manufacturer's instruction. Results were expressed as picogram per cornea. 

Lanepitant Treatment in the Corneal Alkali Burn Model

A corneal alkali burn was created in the left eye of each mouse by means of a paper disc (3-mm diameter) soaked in 1 N NaOH for 10 seconds, under slit-lamp examination. The ocular surface was then washed with 15 mL normal saline. To increase reproducibility, a single investigator applied the burn to all animals. 

Previous literature proposed the use of Lanepitant (LY303870) at 20 mg/kg/d, via intraperitoneal injection, in rats.²⁹ Considering an average mouse weight of 35 g, we extrapolated a dose of 0.7 mg/d/mouse. We arbitrarily decided to administer topically approximately half of the systemic dose. Specifically, the highest topical dose we used was 6.4 mg/mL, six times per day, in a 10-μL volume (total: 0.384 mg/d). Considering that we treated only one eye, the total amount of Lanepitant we used was approximately half the intraperitoneal dose. We also tested 1:4 scalar dilutions: 6.4, 1.6, and 0.4 mg/mL. 

Animals were randomized into four groups (n = 6, each) receiving Lanepitant 0.4, 1.6, or 6.4 mg/mL (MW = 632.66, Dompé S.p.A., L'Aquila, Italy) dissolved in a total volume of 10 μL phosphate-buffered saline (PBS, Sigma-Aldrich Corp.) or 10 μL vehicle as control, topically six times a day for 4 days (Table). A second experiment was designed, extending treatment duration to 7 days with the lower concentrations, 0.4 or 1.6 mg/mL, in comparison to the vehicle. 

Table

 

Schematic Representation of Different Drug Application Strategies

Corneal Model	Drug	Route of Administration	Concentration, mg/mL	Days of Treatment
Alkali burn	Lanepitant	Topical (six times daily)	0.4	4
			1.6	4
			6.4	4
			0.4	7
			1.6	7
Alkali burn	Lanepitant	Subconjunctival (every 2 d)	12.8	10
Sutures	Lanepitant	Topical (six times daily)	0.4	4
			1.6	4
			6.4	4
			0.4	7
			1.6	7
			6.4	7
			12.8	7
Sutures	Lanepitant	Subconjunctival (every 2 d)	12.8	10
Alkali burn	Befetupitant	Topical (six times daily)	0.4	4
			1.6	4
Healthy	Lanepitant	Topical (six times daily)	1.6	9
Healthy	Befetupitant	Topical (six times daily)	0.4	9
Alkali burn	Substance P	Subconjunctival (every 2 d)	0.01	10
Sutures	Substance P	Subconjunctival (every 2 d)	0.01	10

Topical Lanepitant toxicity was evaluated in two groups of six healthy animals receiving in the left eye 10 μL topical Lanepitant 1.6 mg/mL or PBS, six times a day for 9 days. 

A third experiment was performed by testing subconjunctival injection of Lanepitant (12.8 mg/mL, in 3 μL, every 2 days for 10 days) versus subconjunctival PBS in the alkali-burned left eye of eight mice. 

Befetupitant Treatment in the Corneal Alkali Burn Model

A corneal alkali burn was created as described above for the Lanepitant experiment. Animals were then randomized into three groups (n = 6), receiving 10 μL Befetupitant 0.4 or 1.6 mg/mL (MW = 565.55, Dompé S.p.A.) in 100% dimethyl sulfoxide (DMSO, Sigma-Aldrich Corp.) or 10 μL vehicle (DMSO) as control, topically six times a day for 4 days (Table). 

Topical Befetupitant and DMSO toxicity was evaluated in two groups of six healthy animals receiving in the left eye 10 μL topical Befetupitant 0.4 mg/mL or 100% DMSO, six times a day for 9 days. 

Lanepitant Treatment in the Corneal Suture Model

Mice were anesthetized and three 10.0 nylon sutures were placed intrastromally as previously described with some modifications.³⁰ Briefly, a 2-mm corneal trephine was placed on the cornea and centered on the pupil and then three 10.0 nylon sutures were placed intrastromally 120° apart with knots left unburied. Animals were randomized into four groups (n = 6, each), receiving 10 μL Lanepitant (0.4, 1.6, or 6.4 mg/mL) dissolved in PBS or 10 μL vehicle as control, topically six times a day for 4 days (Table). A second experiment was performed, extending treatment duration to 7 days with topical 0.4, 1.6, 6.4, and 12.8 mg/mL Lanepitant or topical vehicle (n = 6). A third experiment was performed with subconjunctival Lanepitant (12.8 mg/mL, in 3 μL, every 2 days for 10 days) or subconjunctival vehicle (n = 10). 

Substance P Treatment in Two Models of Corneal Neovascularization

Subconjunctival injection of SP (Bachem, Torrance, CA, USA) was performed in the left eye of 10 mice (0.01 mg/mL, in 3 μL, every 2 days for 10 days) and compared with subconjunctival injection of PBS or Lanepitant (12.8 mg/mL) in both models of CNV (alkali burn or suture models), as described in the Table. 

All treatments started immediately after caustication or suturing. 

Clinical Endpoints

Clinical examination was performed to check for drug toxicity and/or gross pathologic changes. Corneas were examined and photographs were taken under the slit-lamp microscope SL 990 (C.S.O., Florence, Italy) every day in a blinded fashion. In vivo corneal fluorescein staining was used to evaluate drug toxicity after Lanepitant and Befetupitant treatment. A scoring system was used to evaluate corneal opacity, as previously described.³¹ Briefly, corneal opacity was scored on a scale of 0 to 4, where 0 = completely clear; 1 = slightly hazy, iris and pupils easily visible; 2 = slightly opaque, iris and pupils still detectable; 3 = opaque, pupils hardly detectable; and 4 = completely opaque with no view of the pupils. Tear production was quantified with phenol red thread test, as demonstrated before.³² Briefly, tear production was measured with the phenol red thread test (Zone-Quick; Lacrimedics, Eastsound, WA, USA). Under a microscope, the threads were held with jeweler forceps and placed in the lateral cantus of the conjunctival fornix of the left unanesthetized eye for 30 seconds after the excess tears were removed for a standard time of 4 seconds. The tear distance was measured in millimeters under a microscope. 

Immunohistochemical Analysis

Mouse corneas without (control) or with neovessels (4 days after alkali burn or suture placement) and human corneas affected with keratoconus were frozen in OCT (Optimum Cutting Temperature) compound (Killik; Bio-Optica, Milan, Italy) and stored at −80°C until ready for sectioning. Cryosections (7 μm) were fixed with paraformaldehyde 4% (Sigma-Aldrich Corp.), blocked in 2% bovine serum albumin/0.3% Triton X-100 (Sigma-Aldrich Corp.), as previously described.³³ Mouse sections were immunostained with rabbit polyclonal antibody binding the entire SP (1:200, AB1566; Chemicon, Temecula, CA, USA) and rabbit anti-TUJ1 (1:200, neuron-specific class III beta-tubulin, 1 mg/mL; Chemicon) at 4°C overnight, and subsequently with Alexa Fluor-488 donkey anti-rabbit IgG, 2 mg/mL (Invitrogen-Moleular Probes, Paisley, UK) in a 1:500 dilution for 2 hours at room temperature. 

Human cornea sections were immunostained with the same antibody anti-SP and with mouse anti-TUJ1 (1:200, 1 mg/mL; Covance, Princeton, NJ, USA), and subsequently with Alexa Fluor-488 donkey anti-rabbit IgG and Alexa Fluor-546 donkey anti-mouse IgG (Invitrogen-Moleular Probes). 

The rabbit IgG antibody (1:200, 1 mg/mL, 011-000-003; Jackson ImmunoResearch, West Grove, PA, USA) was used as isotype control. 

Human and mouse sections were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories, Inc., Burlingame, CA, USA), mounted, and photographed by using, respectively, an epifluorescence microscope (Leica CTR5500; Leica Microsystems, Wetzlar, Germany) and a confocal microscope (Leica TCS SP5). Confocal imaging was performed at 1 μm-slide thickness (5-μm z-stack) by using ×63 oil-immersion objective. 

In toxicity experiments, six eyes per group were included in OCT compound and the sections were stained for hematoxylin-eosin to detect inflammatory cell infiltration and epithelial damage. 

Immunostaining of Corneal Neovascularization

At the end of any experiments, corneas were dissected, fixed with iced acetone, and immunostained for blood (CD31) and lymphatic (LYVE1) vessel markers, and for leukocyte CD45 marker (except for Befetupitant treatment) as previously described.³³ We used rat anti-CD31 (BioLegend, San Diego, CA, USA), goat anti-LYVE1 (AbCam, Cambridge, UK), and rabbit anti-CD45 (R&D System, Minneapolis, MN, USA). Corneas were flat-mounted by using a mounting medium with DAPI and analyzed by epifluorescence microscope for CNV analysis. Digital pictures of the flat mounts were analyzed by a Java-based image processing program (ImageJ 1.44p; http://imagej.nih.gov/ij/; provided in the public domain by the National Institutes of Health, Bethesda, MD, USA). The total corneal area was outlined by using the innermost vessel of the limbal arcade as the border, and the area of neovascularization was then calculated and normalized to the total corneal area (CNV index, expressed as a percentage of the cornea covered by green CD31+ or red LYVE1+ vessels). CD45+ cell infiltration was quantified by counting the nucleated CD45-positive cells per field; six peripheral fields (×40) per cornea were taken with confocal microscope. 

Statistical Analysis

One-way ANOVA, following Bonferroni post hoc tests, were performed to analyze the difference in SP level, tear secretion, corneal transparence, cellular infiltration, and CNV index between groups. Kruskal-Wallis test, following Dunn's Multiple Comparison test, was applied to analyze the difference in corneal transparence. The χ² test (without Yates correction) was performed to evaluate the incidence of corneal perforation. 

Significance was defined as a P value < 0.05. Results are presented as mean ± standard error of the mean (SEM). All data were processed with GraphPad Prism software 5.0 (GraphPad Software, Inc., San Diego, CA, USA). 

Results

Topical Lanepitant Treatment Reduced Substance P Levels, Increased in the Epithelium of Mouse CNV Models

In avascular human and mouse corneas, the neuropeptide SP was localized in the epithelium and only sporadically in the stroma (Figs. 1A, 1B), as previously reported.²⁷ TUJ1, expressed principally in basal layer of the epithelium, colocalized with SP, suggesting its expression in corneal sensory nerves. Four days after alkali burn and suture placement, the staining for SP was more marked in the corneal epithelium of the mouse. 

Figure 1

 

Substance P localization in the neovascularized cornea and its reduction after Lanepitant treatment. In human (A) and mouse (B) corneal cross sections, SP was localized in the epithelium (E) and sporadically in the stroma (S), and it colocalized with TUJ1, principally localized in the basal layer of the epithelium. The isotype control immunostainings were negative (lower panel). The inserts show a detail at ×2 magnification. (C) Correlation between CD45+ cells and SP levels in both CNV models on days 4 and 7, expressed as fold-change increase with respect to healthy corneas. The two measures had the same trend over time. The amount of SP and leukocyte infiltration was significantly higher in the alkali burn model than the suture model on day 4. (D) Substance P protein level in the cornea was significantly increased in two mouse models of CNV (alkali burn and suture placement) on days 4 and 7. The alkali burn produced the highest increase of SP on day 4. Topical Lanepitant treatment (6.4 mg/mL) significantly reduced the amount of SP, only in the alkali-burned cornea. Histograms represent mean values ± SEM. *,§P < 0.05, **,§§P < 0.01, ###,§§§P < 0.001 (n = 6).

The increased expression of SP in the cornea after injury was confirmed by EIA analysis. The alkali burn model induced an early increase of SP level, already on day 4, whereas the suture model stimulated a later production of the neuropeptide in cornea, on day 7 (Fig. 1C). Notably, the amount of CD45+ cell infiltration correlated with SP levels (both expressed as fold-change increase with respect to healthy corneas) after injury. The two measures had the same trend over time. The amount of SP and leukocyte infiltration in the alkali burn model was almost four times that of the control and 2.5 times (P < 0.01 and P < 0.001, respectively) that of the suture model on day 4. After a week, the fold-change increase for SP and CD45+ cells was almost 2.5 times that of the control and similar in both CNV models. The amount of SP in the alkali burn model remained significantly higher (P < 0.05) than in the suture model. 

In detail, SP protein level was significantly increased in both neovascularized alkali-burned (P < 0.001) and suture-induced (P < 0.001) mouse corneas (Fig. 1D) on days 4 and 7. The increase was most prominent in the alkali burn model on day 4 (12,435 ± 1483 pg/cornea), compared to the suture model (4862 ± 418 pg/cornea) and to the control (3255 ± 140 pg/cornea). Substance P protein reached high levels later (day 7) in the suture models (9080 ± 557 pg/cornea). Topical NK1 receptor antagonist (6.4 mg/mL) was significantly effective in reducing the amount of SP only in the alkali burn model on day 4 (5457 ± 319 pg/cornea; P < 0.01) and day 7 (6046 ± 447 pg/cornea; P < 0.05). It was ineffective in the suture model at these time points. 

Topical Application of Lanepitant Reduced Corneal Neovascularization in the Mouse Model of Alkali Burn, but Not in the Suture Model

In the alkali burn model (Fig. 2A), topical Lanepitant treatment for 4 days induced a significant 50% reduction (P < 0.01) of blood and lymphatic corneal neovessels (Figs. 2B, 2C) at both concentrations (1.6 and 6.4 mg/mL). Moreover, this treatment preserved corneal transparence, as observed at slit-lamp examination (Fig. 2A). Finally, topical Lanepitant significantly reduced the CD45+ cell infiltration (P < 0.05) on day 4 (Fig. 2D). Lower concentrations of Lanepitant (0.4 and 1.6 mg/mL) for a longer time (7 days), confirmed the efficacy of topical Lanepitant in reducing CNV and leukocyte infiltration (P < 0.05, data not shown). 

Figure 2

 

Topical Lanepitant treatment in the alkali burn model. (A) After 4 days of topical Lanepitant treatment, corneas showed a better outcome in terms of opacity at slit-lamp examination. (B) Corneal whole-mount showed fewer CD31+ blood vessels and LYVE1+ lymphatic vessels. (C) Both concentrations (1.6 and 6.4 mg/mL) of topical Lanepitant significantly reduced hemangiogenesis and lymphangiogenesis. (D) CD45+ cell quantification showed a significant reduction of leukocyte infiltration in the alkali-burned cornea after Lanepitant (LA) treatment (6.4 mg/mL for 4 days), compared to the vehicle (VE). Histograms represent mean values ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 (n = 6).

In the suture model, topical application of 0.4 and 1.6 mg/mL Lanepitant for 4 or 7 days did not affect CNV (data not shown). Treatment with higher concentrations, 6.4 and 12.8 mg/mL, of Lanepitant for 7 days were similarly ineffective, as observed at the slit-lamp (Fig. 3A) and with immunofluorescence staining for blood vessels and lymphatics (Figs. 3B, 3C). Also, CD45+ cell infiltration was not affected by Lanepitant treatments in the suture model (Fig. 3D). 

Figure 3

 

Topical Lanepitant treatment in the suture models. (A) Topical application of 6.4 and 12.8 mg/mL Lanepitant on sutured corneas for 7 days did not reduce the opacity at slit-lamp examination. (B, C) Moreover, the treatment did not affect CNV, as shown in whole-mount images. (D) The number of CD45+ cells remained unchanged after treatment. Histograms represent mean values ± SEM (n = 10).

Topical Lanepitant (1.6 mg/mL for 9 days) treatment was not toxic to the ocular surface as assessed with slit-lamp examination (no opacity or perforations detected); the absence of vital fluorescein staining confirmed that Lanepitant is not toxic to the corneal epithelium (Fig. 4A). Hematoxylin-eosin stain did not detect inflammatory cell infiltration or epithelial damage, confirming Lanepitant safety (Fig. 4B). 

Figure 4

 

Evaluation of Lanepitant toxicity. (A) Topical application of Lanepitant (1.6 mg/mL) on healthy corneas for 9 days was not toxic: no opacity or fluorescein staining was observed at slit-lamp examination (n = 6). (B) Hematoxylin-eosin stain confirmed that Lanepitant was nontoxic to the ocular surface: no inflammatory cell infiltration or epithelial damage was observed.

Subconjunctival Application of Lanepitant Reduced SP Levels Only in the Suture Model

Later, on day 10, the amount of SP was significantly higher in the suture model (P < 0.001) than in the alkali burn model (Fig. 5A). Moreover, the suture model showed a significantly higher cell infiltration (P < 0.05) than the alkali burn model. In both CNV models, the fold-change increase of SP was significantly lower (P < 0.05) than CD45+ cell infiltration (Fig. 5A). 

Figure 5

 

Subconjunctival Lanepitant and Substance P treatments in the alkali burn model. (A) Correlation between CD45+ cells and SP levels in both CNV models on day 10, expressed as fold-change increase with respect to healthy corneas. The levels of SP and leukocyte infiltration were significantly predominant in the suture model as compared to the alkali burn model on day 10. Moreover, the fold-change increase of CD45+ cells was significantly higher than SP level in both models. (B) On day 10, SP protein levels in the injured cornea remained significantly higher than in the healthy cornea, above all in the suture model. The alkali burn produced the highest increase of SP, on day 4. After 10 days of subconjunctival (s.c.) Lanepitant treatment (12.8 mg/mL), the amount of SP significantly decreased in the sutured cornea. (C) In the alkali burn model, the subconjunctival Lanepitant–treated eyes showed less perforations and opacity at slit-lamp examination, in comparison with subconjunctival injection of vehicle and of SP (0.01 mg/mL). (D) CD31 staining showed a significant reduction of CNV after Lanepitant treatment in comparison to the vehicle. (E) CD45+ cell quantification showed a significant reduction of leukocyte infiltration in the alkali-burned cornea after Lanepitant treatment, compared to the vehicle. (F) Lanepitant treatment significantly reduced the onset of corneal perforation after 2 days of alkali burn. (G) Corneal opacity measure was significantly reduced after Lanepitant treatment, compared to vehicle, while it significantly increased after SP injection, compared to Lanepitant. (H) Tear secretion was significantly higher in the Lanepitant group versus vehicle and SP groups; it significantly decreased in the SP group. Histograms represent mean values ± SEM. *,§P < 0.05, **P < 0.01; ***,§§§P < 0.001 (n = 10).

As stated before, topical Lanepitant treatment was ineffective in reducing the amount of SP only in the suture model after 4 and 7 days. Instead, administration of NK1 receptor antagonist was effective in the suture model (P < 0.01) after subconjunctival administration on day 10 (Fig. 5B). In the alkali burn, the SP levels were comparable between vehicle and Lanepitant groups. 

Subconjunctival Application of Lanepitant Reduced Corneal Blood Neovascularization in the Alkali Burn Model

In the alkali burn model (Fig. 5C), Lanepitant (12.8 mg/mL) administered subconjunctivally for 10 days significantly reduced blood vessel growth (P < 0.05), compared to the vehicle (Fig. 5D). 

Staining for CD45 showed a significant reduction of CD45+ cells in the cornea after Lanepitant treatment (P < 0.05) compared to the vehicle (Fig. 5E). 

Moreover, Lanepitant was effective in reducing perforation onset and corneal opacity in the alkali burn model, as shown in slit-lamp photographs (Fig. 5C), in comparison to vehicle and SP (0.01 mg/mL). Specifically, the perforation onset was delayed in the Lanepitant group (Fig. 5F), reaching statistical significance (P < 0.05) on day 2 (30% of perforations in the Lanepitant group versus 80% to 100% in the vehicle-SP groups, respectively). Corneal opacity was significantly (P < 0.05) reduced after Lanepitant treatment (2.50 ± 0.22) in comparison to controls (3.20 ± 0.20), while the SP group showed a higher opacity index (3.90 ± 0.10), which was significantly different from the Lanepitant group (Fig. 5G). Finally, Lanepitant significantly increased tear secretion (1.90 ± 0.07 mm) in comparison to vehicle (1.40 ± 0.16 mm; P < 0.05) and SP (0.75 ± 0.14 mm; P < 0.001) groups (Fig. 5H). 

Subconjunctival Application of Lanepitant Reduced Corneal Lymphatic Neovascularization in the Suture Model

In the suture model (Fig. 6A), Lanepitant (12.8 mg/mL) administered subconjunctivally for 10 days reduced blood vessel growth, although not significantly in comparison to the vehicle (Fig. 6B). Substance P treatment induced a significant (P < 0.05) increase of blood CNV in comparison to the Lanepitant group. Lanepitant, however, was significantly effective (P < 0.01) in reducing lymphatic CNV in the suture model in comparison to both vehicle and SP groups (Fig. 6B). Analysis for CD45 staining showed a significant reduction (by half) of CD45+ cells in the cornea after Lanepitant treatment (P < 0.05) compared to the vehicle (Fig. 6C). 

Figure 6

 

Subconjunctival Lanepitant and Substance P treatments in the suture model. (A) Lanepitant (12.8 mg/mL) administered subconjunctivally for 10 days to sutured eyes resulted in a reduced opacity score at the slit-lamp examination, in comparison with subconjunctival injection of vehicle and of SP (0.01 mg/mL). (B) Flat-mounted corneas stained for CD31 showed a reduction of CNV after Lanepitant treatment in comparison to the SP group (P < 0.05). Moreover, Lanepitant treatment significantly reduced LYVE1+ lymphatic vessels, compared to the other two groups. (C) CD45+ cell quantification showed a significant reduction of leukocyte infiltration in the cornea after Lanepitant treatment. (D) Corneal opacity scores were significantly reduced after Lanepitant treatment, compared to vehicle and to SP. Histograms represent mean values ± SEM. *P < 0.05, **P < 0.01 (n = 10).

Finally, the treatment was effective in reducing corneal opacity, as shown in slit-lamp photographs (Fig. 6A), in comparison to vehicle and to subconjunctival SP (0.01 mg/mL). The corneal opacity index was significantly (P < 0.05) reduced after Lanepitant treatment (1.15 ± 0.11), in comparison to the two groups (vehicle: 1.65 ± 0.15, SP: 1.75 ± 0.13; Fig. 6D). 

Topical Befetupitant Reduced Corneal Neovascularization in the Alkali Burn Model, but It Was Toxic to the Ocular Surface

Finally, we tested Befetupitant, a different, highly selective NK1R antagonist, in the alkali burn model (see Supplementary Fig. S1A). Topical application of Befetupitant for 4 days was effective (P < 0.05) in reducing hemangiogenesis and lymphangiogenesis (see Supplementary Figs. S1B, S1C) at both concentrations (0.4 and 1.6 mg/mL). Befetupitant and its vehicle DMSO, however, induced corneal opacity even in healthy controls, as observed at slit-lamp examination (see Supplementary Fig. S2A). Moreover, fluorescein and hematoxylin-eosin staining showed epithelial damage and inflammatory cellular infiltration in the stroma, respectively (see Supplementary Figs. S2A, S2B), confirming DMSO toxicity. We limited the experiment just to the alkali burn model and to the topical administration because Befetupitant was toxic. 

Discussion

The cornea receives the densest sensory innervation of the entire body. Among nerve-secreted peptides, SP, which binds mainly to the NK1R, is one of the most prevalent. Hence, it is tempting to speculate that it may have a key role in modulating corneal inflammation. Since CNV is a constant finding in persistent ocular surface inflammation, we studied the effect of selective NK1R antagonism. Here, we confirmed that SP was expressed in the epithelial layer of avascular cornea and colocalized with corneal nerve endings, specifically in the subbasal nerve plexus of the epithelium. Further, we showed that SP was significantly increased in two different CNV mouse models (alkali burn and suture models), with a different time course. In the alkali burn model SP peaked earlier (day 4), while in the suture model this occurred later (7–10 days). Since SP is also secreted by macrophages,^7–9 the SP expression pattern we observed could possibly be due to the early leukocyte infiltration observed in the alkali burn model (peak on day 4), while in the suture model the infiltration was delayed (peak on day 10). Previous literature³⁴ has reported a similar trend of inflammatory cell infiltration in the corneal tissues in these two animal models. The corneal suture model seems to maintain long-term inflammation, which is possibly elicited by longstanding sutures. Notably, we found a good correlation between SP increase in the cornea and CD45+ cell infiltration, in both CNV models. Although the extensive epithelial and stromal disruption induced by the burn may result in ulceration and perforation,³⁵ with consequent loss of epithelium and nerve fibers,³⁶ the alkali burn cornea presented the highest levels of SP, when compared to the suture model. 

To evaluate safety and efficacy of SP inhibition in the setting of CNV, we tested two different, highly selective NK1R antagonists: Befetupitant and Lanepitant. These are both potent and selective NK1R antagonists, with a high binding affinity for NK1R in both mice and humans.^37–39 Befetupitant was effective in the treatment of CNV, reducing the ingrowth of both corneal blood and lymphatic vessels already after 4 days. However, its topical application was toxic, causing corneal epithelial damage and inflammatory cell infiltration. Similar results were obtained with DMSO, which we used to dissolve Befetupitant, suggesting that toxicity was largely due to the vehicle used, as previously reported.⁴⁰ 

Differently from Befetupitant, Lanepitant is freely soluble in water. Topical Lanepitant was effective in reducing SP expression in the alkali-burned cornea, while subconjunctival Lanepitant reduced SP in the suture model. Additionally, Lanepitant application significantly reduced both hemangiogenesis and lymphangiogenesis in the alkali burn model (topical and subconjunctival administration) and only lymphangiogenesis in the suture model (subconjunctival administration only). Interestingly, topical administration decreased leukocyte infiltration in the alkali-burned cornea, while subconjunctival injection reduced the cellular infiltration in both CNV models. 

Substance P, through its receptor NK1R, has promitotic and antiapoptotic effects and favors epithelial wound closure.¹⁹ For this reason, we tested whether its inhibition with Lanepitant may alter corneal epithelial integrity. Indeed, Lanepitant was not toxic to the epithelium when administered for up to 9 days. 

Topical administration of Lanepitant did not affect CNV and inflammatory cell recruitment in the suture model, even at higher concentrations and with longer treatment duration. This was possibly due to the limited drug penetration, as the corneal epithelial barrier remains in place in the suture model, which is not the case after alkali burn. In this regard, it should also be noted that the presence of two strong basic groups induces a decrease in lipophilicity and, hence, poor Lanepitant penetration.⁴¹ Additionally, although the suture placement was associated with lower SP levels in comparison to the alkali burn at earlier time points, topical Lanepitant treatment was ineffective. These findings may be explained by the limited penetration of Lanepitant through intact corneal epithelium. On the contrary, SP expression was higher in the suture than in the alkali model at later time points. Since we administered the same dose in the two models, but these exhibited different kinetics in SP expression, it may be possible that a higher concentration/different administration regimen could be effective in reducing hemangiogenesis even in the suture model. 

Finally, the different pathophysiologic mechanism inducing CNV in the two models should also be considered. We previously demonstrated that these CNV models had a similar pattern of induction for hemangiogenesis.⁴² For this reason, we chose the same time points to test treatment efficacy in the two models. Lymphangiogenesis was more pronounced in the suture model; and, the two types of corneal injuries were associated with different timing in inflammatory cell recruitment. It should be noted that inflammation and necrosis are prevalent by far in the alkali burn model.⁴³ In this vein, the well-known anti-inflammatory effects of NK1R antagonists may explain the significant CNV inhibition in the alkali burn model. In the suture model, the extension of hemangiogenesis and lymphangiogenesis did not differ significantly. In fact, NK1R promotes inflammation in many ways as it activates the proinflammatory transcription factor NF-kB, inducing expression of cycloxygenase 2 and production of prostaglandin E2.⁴⁴ 

In an effort to increase corneal penetration of Lanepitant, we used subconjunctival injection; additionally, we treated for a longer time (10 days) in both models. Indeed, subconjunctival Lanepitant inhibited blood vessel growth in comparison to the vehicle only in the alkali burn model. Moreover, it reduced corneal opacity score, delayed perforation onset, and promoted tear secretion. Interestingly, SP treatment induced worsening of all these parameters. It should be noted that all of these clinical endpoints are generally considered markers of active ocular surface inflammation. This is not surprising, since proinflammatory effects of SP have been described before. Substance P exerts its immunomodulatory activity through activation of macrophages and fibroblasts,^45,46 and it stimulates proinflammatory cytokine production (i.e., IL-1β, IL-6, and TNF-α).⁴⁷ Interestingly, SP treatment also induces an increase in VEGF levels in a model of skin wound healing, together with macrophage recruitment and angiogenesis.⁴⁸ In this vein, the key role of VEGF in CNV has been established already.⁴⁹ Finally, SP, either exogenously applied or released from endogenous sources, can stimulate angiogenesis in the synovium by binding to NK1R.⁵⁰ 

Subconjunctival Lanepitant improved corneal transparence in the suture model, as opposed to vehicle and to SP administration. Additionally, it was only effective in reducing lymphatic vessel growth, as compared to both vehicle and SP, possibly because of the high amount of SP protein present in the cornea. This may have relevant clinical implications, since lymphangiogenesis is a well-known risk factor for corneal graft rejection, possibly even more relevant than hemangiogenesis.⁵¹ The fact that Lanepitant was ineffective in the suture model, when administered subconjunctivally, may be explained by the higher amount of SP in the sutured cornea at later time points. Interestingly, this was associated with an increased number of inflammatory cells, which are known to produce proangiogenic factors by themselves. Higher concentrations and/or different administration regimen of Lanepitant could possibly reduce even blood CNV in the suture model. 

Intriguingly, leukocyte infiltration in whole mount corneas was reduced by 50% after subconjunctival Lanepitant treatment. This is critical as inflammatory cell infiltration favors corneal angiogenesis,³⁰ and its inhibition is associated with reduced CNV.⁵² In this vein, SP has been described as a potent chemoattractant,⁵³ inducing oxidative burst in macrophages.⁵⁴ Finally, it has been reported that human macrophages express NK1R and are themselves a relevant source of SP.⁹ 

To the best of our knowledge, this is the first study demonstrating the safety and efficacy of an NK1R antagonist administered locally for the treatment of CNV. This is paramount from a clinical translation perspective as, although previous literature has proposed the use of SP antagonists (Spantide) in mice, limited evidence exists on its human use. Additionally, by antagonizing SP, Spantide can also inhibit NK2 and NK3 receptors—together with their manifold actions—differently from Lanepitant, which blocks NK1R selectively. 

Acknowledgments

We thank Marcello Allegretti, MD, PhD, and Francesco Sinigaglia, MD (Dompé S.p.A., L'Aquila, Italy) for helpful suggestions and discussions. 

Supported by a Dompé S.p.A. research grant. GF is the inventor of a patent covering part of the results presented in this paper. 

Disclosure: F. Bignami, None; C. Giacomini, None; A. Lorusso, None; A. Aramini, None; P. Rama, None; G. Ferrari, Dompé S.p.A. (F, C), P