Cone Photoreceptor Neuroprotection Conferred by CNTF in a Novel In Vivo Model of Battlefield Retinal Laser Injury

Abstract

Purpose.: To develop a reproducible laboratory model to simulate a battlefield foveal laser injury and to test potential neuroprotective effects of a single injection treatment that might be administered in a military setting.

Methods.: Frequency-doubled 532-nm Nd:YAG laser was used to induce a threshold retinal injury bilaterally in transgenic reporter mice that have fluorescent cones. Intravitreal injection of ciliary neurotrophic factor (CNTF) was then administered to the lasered eye and compared with a contralateral sham injection of saline. The effect on fluorescent cone cell survival was quantified using a confocal scanning laser ophthalmoscope (cSLO), TUNEL assays, and quantitative real-time PCR (qPCR).

Results.: At 3 weeks post-laser, cSLO imaging showed that the proportion of surviving cones expressing green fluorescent protein (GFP) was greater in CNTF-treated (54.1 ± 5.15% of baseline count) than in sham-injected eyes (28.7 ± 4.4%), which was accompanied by a reduction in TUNEL-positive cells. This difference in cone survival persisted at the 6-week point (treated, 39.6 ± 3.2% versus sham, 18.0 ± 3.8%). These changes were accompanied by a reduction in TUNEL-positive cells. The Bcl-2/Bax ratio was increased in CNTF-treated eyes at 1 week postlaser exposure relative to controls.

Conclusions.: A single intravitreal injection of CNTF protein was shown to improve cone survival when administered immediately after laser exposure. Similar treatments with CNTF might also have a role in attenuating retinal laser damage sustained by combat personnel in the military setting.

Introduction

Even outside the military, the use of laser technology across a variety of occupations is expanding. In contrast to civilian life, however, high-powered military lasers are frequently used outdoors and at long ranges, posing additional safety risks to personnel who may not have adequate personal protection equipment. ^1–3 Retinal laser injury, for which there is currently no satisfactory treatment, represents an infrequent but potentially devastating cause of irreversible sight loss. ^4,5 Of further concern is the potential development of offensive laser weaponry to cause permanent blindness, which although against United Nations agreed conventions, ⁶ might be considered an asymmetric warfare threat. Currently there is no battlefield (role 1) treatment available to military personnel who might be exposed to an acute retinal laser injury. The development of a simple injection neuroprotection treatment would therefore be timely and might have a positive effect on the morale of troops if risks of laser exposure ever became significant in future combat deployments. 

Within the posterior segment of the eye, laser energy is primarily absorbed by melanin within the RPE and choroid, hemoglobin in the retinochoroidal circulation, and xanthophyll in the neurosensory retina. ⁷ However, it is damage to photoreceptor cells that is largely responsible for visual loss incurred following laser exposure. The pathogenesis of injury may be due to photothermal, photomechanical, or photochemical mechanisms, ^8,9 with severity being dependent on the site of insult. For example, a foveal injury, which almost exclusively involves cone photoreceptors, will result in maximal visual loss compared with a more peripheral injury, where less visual impact will be encountered. This is in part due to the refractive optics of the eye, where a beam of light focused on the fovea will have an irradiance level over 10⁴ times greater than the incident ray at the cornea. ¹⁰ Resultant cell death may occur by necrosis or apoptosis, the latter comprising a form of programmed cell death that may be physiological or pathological. ¹¹ Although it is extremely unlikely that necrosis can be reversed with secondary interventions after exposure, apoptosis may be inhibited under certain conditions. ¹² Therefore, following retinal laser injury, specific treatments that aim to target secondary apoptotic mechanisms activated in photoreceptors may form effective neuroprotective therapies to reduce the degree of visual field loss. 

Primary neuronal injury causes release of inflammatory mediators such as cytokines and nitric oxide into the extracellular environment, which may in part lead to the extension of the lesion to involve adjacent cells. ^13,14 Regarding laser injury, a number of neuroprotective compounds have been investigated and shown to protect secondary photoreceptor loss postexposure. ^15–18 One such protective protein, ciliary neurotrophic factor (CNTF), has completed phase I clinical trials in inherited retinal degeneration and dry age-related macular degeneration and showed promising results. ^19,20 Specifically, these early trials demonstrated that intravitreal administration of encapsulated cells secreting CNTF protein was safe and may preserve photoreceptor survival and function. An injection treatment delivering immediate access of a drug to the retina that might be applied in the battlefield would need to have proven efficacy in photoreceptor neuroprotection, be relatively stable during storage, and have an extensive safety profile. In this regard, CNTF has many advantages over other therapies. 

Since human retinal injury is most debilitating when it involves the fovea, the aim of this study was to develop an in vivo murine model in which to observe cone photoreceptor loss after retinal laser exposure, thus providing a suitable model system for longitudinal studies, including fundus autofluorescence (AF) imaging. Although the mouse retina does not have a macula or fovea, its cone to rod ratio is similar to that of the human retina and serves as a nonprimate model to investigate cone injury. Hence, herein we describe the development of an in vivo model in which to follow temporally the fate of cone photoreceptor cells following retinal laser injury. We then apply a single injection CNTF neuroprotection treatment and assess its effects on cone survival after laser exposure. 

Methods and Materials

Animals

All animal experiments were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and UK Home Office guidelines. B6.Cg-Tg(OPN1LW-EGFP)85933Hue/Mmmh mice (referred to herein as OPN1LW-EGFP) in which enhanced green fluorescent protein (EGFP) expression is restricted to medium wavelength-sensitive cones (M cones) ²¹ were obtained under a material transfer agreement (MTA) from the Mutant Mouse Regional Resource Centres, National Institutes of Health (NIH; stock number: 000043-MU). Wild-type (WT) C57BL/6 mice were provided by the Biomedical Sciences division of the University of Oxford. All animals were kept in a 12-hour light (<100 lux)/12-hour dark cycle, with food and water available ad libitum. 

Although explained in greater detail within each methods section, animal strains were used according to the type of experiment. OPN1LW-EGFP mice were used for AF imaging and the qualitative detection of the TUNEL assay. WT mice were used for histology, molecular analysis, and the quantitative detection of the TUNEL assay. 

Retinal Laser Photocoagulation

Mice were anesthetized by intraperitoneal injection of 1 mg/kg medetomidine (Dormitor 1 mg/mL; Pfizer, Sandwich, UK) and 60 mg/kg ketamine (Ketaset 100 mg/mL; Fort Dodge, Southampton, UK). Pupil dilatation was achieved with tropicamide (Mydriaticum 1%; Bausch & Lomb, Kingston-on-Thames, UK) and phenylephrine (phenylephrine hydrochloride 2.5%; Bausch & Lomb) eyedrops. Preliminary experiments identified the optimal settings required to create a threshold injury in the experimental setup, that resulted in a consistent window of GFP-positive cone loss visible by in vivo imaging. A frequency-doubled 532-nm Nd:YAG laser (Novus Spectra, Lumenis, San Jose, CA) mounted on a slit lamp biomicroscope (Haag-Streit, Koeniz, Switzerland) was applied directly through a flat contact lens (9-mm circular cover glass; VWR, Lutterworth, Leicestershire, England) with the aid of a coupling medium (Viscotears, Novartis, Switzerland) bilaterally to the retinae of OPN1LW-EGFP mice for in vivo study, and WT mice for histological confirmation of injury and molecular analysis. 

For in vivo fundus AF studies, retinae of OPN1LW-EGFP mice (n = 11) received three to four equidistant burns bilaterally approximately two disc diameters from the edge of the optic disc. This distance allowed for adequate imaging of GFP-positive cones without blur resulting from being too close to the optic disc. For histology, WT mice (n = 13) received eight burns bilaterally (16 burns in total) in a similar distribution as described. 

To investigate the effect of laser on retinal gene expression, WT mice (n = 7) received 20 pan-retinal burns unilaterally with the contralateral eye left untreated to serve as control. An identical laser protocol was employed to investigate the effect of intravitreal CNTF versus contralateral saline on gene expression postlaser (n = 28). 

Intravitreal CNTF

For in vivo study, human CNTF (0.5 mg/mL, prepared in PBS [Peprotech, London, UK]) was delivered by intravitreal injection into the left eye, whereas a sterile PBS sham injection was made into the right eye of OPN1LW-EGFP mice to control for the release of endogenous growth factors as a result of generalized eye injury. ²² In both cases, 1-μL injections were made at a dose of protein that was chosen based on previous work that showed improved photoreceptor survival following intravitreal CNTF injection in murine models of retinal degeneration. ²³ Injections were performed after confocal scanning laser ophthalmoscope (cSLO) imaging (within 30 minutes of retinal laser injury) using a Hamilton syringe and needle (Esslab, Essex, UK) and under direct visualization using a surgical microscope (M620 F20; Leica Microsystems GmbH, Germany). For histological and molecular analyses, a similar method was applied to WT mice. 

In Vivo Assessment

The effects of both laser treatment and CNTF neuroprotection in OPN1LW-EGFP mice (8-week old females) were observed using the 488 nm autofluoresecence (AF) and 820 nm near-infrared (NIR) reflectance modes of a cSLO (Spectralis HRA; Heidelberg Engineering, Heidelberg, Germany) following a recently described protocol. ²⁴ In this mouse model, 488 nm AF has been used to detect hyper-autofluorescent spots representing M cones. ²⁵ A custom-made, 3.2 mm diameter polymethyl methacrylate (PMMA) contact lens was used for all recordings (Cantor and Nissel, Brackley, UK), with lubricating eyedrops (Hypromellose BPC 0.3%; Martindale Pharmaceuticals, Romford, UK) as coupling fluid. For quantitative analysis, nonnormalized images were recorded with a detector sensitivity of 95 in all cases. Images were taken immediately before and after laser application, and subsequently at 1, 3, and 6 weeks. 

Histology

WT mice were sacrificed at 1, 3, and 6 weeks postlaser treatment by cervical dislocation (n = 10 per time point) and following enucleation, eyes were fixed in 4% (w/v) paraformaldehyde (PFA). After corneal excision and lens extraction, the remaining eyecups were fixed in 4% PFA overnight, before being transferred to 30% (w/v) sucrose for 2 hours and embedded in OCT compound (VWR). Blocks were serially sectioned on a cryostat to provide sagittal slices of 14-μm thickness. Selected sections were stained with 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen, Paisley, UK), and imaged using an inverted fluorescence microscope (DMIL; Leica Microsystems GmbH, Wetzlar, Germany) or on a confocal microscope (Zeiss LSM 710; Carl Zeiss MicroImaging GmbH, Jena, Germany). 

TUNEL of Fragmented DNA

DNA strand breaks in retinal cell nuclei were detected by TUNEL assay on retinal sections between 1 to 5 days postlaser application according to the manufacturer's instructions (In Situ Cell Death Detection Kit; Roche, Basel, Switzerland). Sections were further counterstained with DAPI to identify photoreceptor nuclei. Apoptotic cells were counted from three sections of the central most area of maximal lesion damage and the resulting numbers were averaged. 

Image Analysis

Histological and cSLO images were analyzed using Java-based imaging software (available in the public domain at http://rsb.info.nih.gov/ij/index.html; ImageJ; NIH, Bethesda, MD). ²⁶ For cSLO images, delineation of the lesion area was determined from the NIR-reflectance image which provided a well-defined and consistent demarcation in the area of laser injury. A freehand outline of this area was transposed to the nonnormalized image taken using the 488 nm AF setting. This area, derived using the 488 nm AF image, was used to calculate the number of GFP-positive cones and lesion size. Lesion size was given in disc areas (DA) by dividing the delineated area by the optic disc area delineated on NIR-reflectance images, the latter providing a reference in all images due to a variable magnification factor of individual images. ²⁴  

For histological sections, photoreceptor nuclei within the outer nuclear layer (ONL) of 137 lesions were counted. A photoreceptor count from the central 100 μm of the lesion was compared as a fraction of the photoreceptor count from nonlesioned internal control areas at 100 μm from the edge of the laser burn. For cell counting of both cSLO and histology-derived images, the Image-based Tool for Counting Nuclei mode of Java-based imaging software (ImageJ; NIH) was used after standardization of settings. 

Tissue Isolation and RNA Preparation

At 3 days postexposure, a comparison was made between gene expression in lasered versus nonlasered retinae. The 3-day time point was chosen to exclude changes associated with necrosis in the immediate aftermath of laser exposure. Previous reports have also shown that alterations in apoptotic gene expression occur 3 days postlaser treatment. ²⁷ Retinae were dissected free of all surrounding tissues under a microscope (M620 F20; Leica Microsystems GmbH) in a commercial RNA stabilization reagent (RNAlater; Ambion, Inc., Huntingdon, UK), before being transferred to a nucleic acid purification reagent (TRIzol; Life Technologies Corp., Grand Island, NY). Total RNA was extracted using a commercial RNA purification kit (RNeasy Micro Kit; Qiagen, Hilden, Germany), according to the manufacturer's instructions, and quantified by using a commercial spectrophotometer (NanoDrop; Thermo Scientific, Wilmington, DE). 

Quantitative Real-Time PCR (qPCR) Analysis

Oligo dT-primed complementary DNA (cDNA) was synthesized from 1 μg total RNA, using a commercial reverse transcriptase enzyme (SuperScript III Reverse Transcriptase; Life Technologies Corp.), according to the manufacturer's procedure. The equivalent of 150 ng cDNA was used to PCR amplify transcripts expressed by the following genes implicated in apoptosis ²⁸ : B-cell lymphoma protein-2 (Bcl-2); Bcl-2 associated protein X (Bax); cysteinyl aspartic acid-protease-3 (Caspase-3); cellular oncogene FBJ murine osteosarcoma viral oncogene homolog (c-Fos); cellular oncogene V-jun avian sarcoma virus 17 oncogene homolog (c-Jun); mitogen-activated protein kinase-11 (Mapk-11); Mapk-12; Mapk-13; Mapk-14; and signal transducer and activator of transcription-3 (Stat-3) genes. In murine tissues, the Bcl-2 gene is expressed as two isoforms, whereas the Bax gene is transcribed as a single mRNA transcript. Being related genes, Bcl-2 and Bax encode for transcripts that share a high sequence identity. To ensure cross-hybridization would not occur during the qPCR experiments, primers for these genes were carefully designed with specific 3′-ends, such that polymerization would only occur to amplify a specific product. All reactions used a commercial qPCR kit (SYBR Green PCR Master Mix; Applied Biosystems, Warrington, UK) and 5 μM final concentration of each forward and reverse primers (Table). All primer pairs were designed to span an intron of the target gene to minimize genomic DNA (gDNA) contamination. All qPCR experiments were performed in triplicate using a commercial real-time PCR instrument (StepOnePlus Real-Time PCR System; Applied Biosystems) with the following settings: an initial denaturation step of 95°C for 10 minutes, followed by 40 cycles of 95°C for 30 seconds, an annealing temperature of either 55°C (all genes except Bcl-2) or 60°C (for Bcl-2) for 30 seconds, and extension at 72°C for 30 seconds. Values obtained for the target genes were normalized to the geometric mean of three housekeeping genes, namely acidic repeat protein (Arp), beta-actin, and glyceraldehyde-3-phosphate dehydrogenase (Gapdh). Primer sequences were designed using primer design software (Primer3 program; provided in the public domain at http://frodo.wi.mit.edu/cgi-bin/primer3/) ²⁹ and standard curve analysis was performed to ensure primer efficiencies were close to 100%. 

Table

 

Oligonucleotides Used for qPCR Analysis of Apoptotic Gene Expression

Gene	Forward (5′ to 3′)	Reverse (5′ to 3′)
Bax	CATGTTTGCTGATGGCAACTTC	CACAAAGATGGTCACTGTCTGC
Bcl-2	GATGACTTCTCTCGTCGCTACC	CGGTTCAGGTACTCAGTCATCC
Caspase-3	CTCTACAGCACCTGGTTACTATTCC	ACAATACACGGGATCTGTTTCTTTG
c-Fos	GAACCCTTTGATGACTTCTTGTTTC	AAAGGAAGACGTGTAAGTAGTGCAG
c-Jun	AGAACTCGGACCTTCTCACGTC	TGTTCTGGCTATGCAGTTCAGC
Mapk-11	GATCATGCTAAACTGGATGCACTAC	GAGGCAGAGACTGGATGTATGTC
Mapk-12	AGACAGTGAGATGACAGGATATGTG	CTCTGCACTCTGTAGCTTCTGAAC
Mapk-13	GTCCAGTACTTGGTGTACCAGATG	GTCGACTGTCTGGTTGTAATGC
Mapk-14	TCAGTTTCTCATCTACCAGATCCTC	TGTCTGGTTATAGTGCATCCAATTC
Stat-3	AAGAACGTGAACTTCTTCACTAAGC	ATATTGTCTAGCCAGACCCAGAAG
Arp	GATCATCCAGCAGGTGTTTGAC	GTGTACTCAGTCTCCACAGACAATG
Beta-actin	GCTGTGCTATGTTGCTCTAGACTTC	CATAGAGGTCTTTACGGATGTCAAC
Gapdh	CTTCATTGACCTCAACTACATGGTC	GTGGTGAAGACACCAGTAGACTCC

Statistical Analysis

Histology, cSLO, and qPCR results were analyzed using paired two-tailed Student's t-tests (with Bonferroni correction where appropriate). TUNEL data were analyzed using one-way ANOVA and Tukey's post hoc test. Statistical significance was set where P ≤ 0.05. Statistical analyses were presented as mean ± SEM and conducted using commercial statistical analysis software (SPSS v. 19; IBM Corporation, Armonk, NY) with charts prepared using commercial scientific graphing and statistics software (GraphPad Prism v. 5.00 for Windows; GraphPad Software, San Diego, CA). 

Results

Optimization of Laser Settings

In order to establish an in vivo model investigating the effect of laser injury on cone photoreceptors, laser intensity settings and exposure time were determined that resulted in the relatively gradual loss of GFP-positive cones that was visible by cSLO imaging over a period of time. The aim was to create a large well-demarcated burn that would facilitate a longitudinal measurement of the lesion area for cone counting. Visible spectrum light funduscopy was used to establish the settings required to create a minimum visible lesion (MVL-ED50; Fig. 1). The MVL-ED50 as observed by cSLO imaging differed according to the following adjustments of exposure duration and spot size: 

Figure 1

 

Retinal laser burns surrounding the optic disc, as visualized under the operating microscope with the aid of a cover slip on the eye (A). Scale bar: 1 mm. Near-infrared reflectance imaging. (B) and 488 nm AF imaging pre- and 2 minutes postlaser (C, D, respectively) showed retinal lesion formation as evidenced by immediate change in the pattern and intensity of fluorescence at the laser injury site (with examples indicated by green arrows) in the OPN1LW-EGFP eye.

(1) Exposure duration—with a maximal spot size (500 μm), an MVL-ED50 was produced at 160 mW. On infrared (IR) reflectance imaging, the lesion area increased from 1.34 ± 0.14 to 2.53 ± 0.22 disc areas (DA) at an exposure time of 0.5 seconds and 3 seconds, respectively (***t ₅ = 9.31, P < 0.0001, n = 6). 

(2) Spot size—with a fixed duration (3 seconds), an MVL-ED50 was produced at 60 mW. The lesion area increased from 1.10 ± 0.04 to 1.36 ± 0.12 DA with a spot size of 100 μm and 500 μm, respectively (**t ₅ = 5.74, P < 0.01, n = 6). 

Previous reports that investigated treatments for laser-induced photoreceptor injury in murine models used laser settings with a smaller spot size (≤200 μm) and shorter duration (≤0.05 seconds). This was performed to produce discrete lesions with outer retinal changes that were quantifiable by histological methods. The application of similar settings in OPN1LW-EGFP mice resulted in immediate hypo-autofluorescence postlaser that was followed by an increase in autofluorescence from the third day onwards, as detected on 488 nm AF imaging. The latter signal, however, obscured any visualization of individual GFP-positive cone loss following laser, so alternative settings were adopted: optimal laser settings with a maximal duration of 3 seconds and a spot size of 500 μm, using 60 mW power, were shown to produce a well-demarcated, circular lesion with central loss of GFP-positive cells, without any explosive rupture of Bruch's membrane. Excessive power resulting in rupture of Bruch's membrane was characterized by an immediate reaction, causing a vaporization bubble and subsequent emission of a hyperfluorescent signal with 488 nm AF imaging. Retinae producing such an effect were excluded from further analysis, since this reaction is associated with the growth of choroidal new vessels, which might confound analysis of cone survival. 

AF Imaging of OPN1LW-EGFP Mice Postlaser Detects Loss of M Cones

With optimal settings, 488 nm AF imaging showed a reduction of GFP-positive cones within the area of laser application from 1 week postexposure (Fig. 2). Relative to the baseline, the number of GFP-positive cones decreased to 36.3 ± 2.7% (***t ₁₀ = 12.9, P < 0.001, n = 11) and 18.8 ± 3.8% (***t ₁₀ = 12.1, P < 0.001, n = 11) at 3 and 6 weeks postexposure, respectively. Similarly, the area of GFP-positive cone loss increased from 0.42 ± 0.19 DA to 0.94 ± 0.14 DA at 3 and 6 weeks postexposure, respectively (*t ₁₀ = 3.05, P < 0.05, n = 11). 

Figure 2

 

AF images of an OPN1LW-EGFP retina preceding (A) and at 1 (B), 3 (C), and 6 weeks (D) postlaser application. Each fluorescent spot represents a cone photoreceptor. ²⁵ Cone loss was observed from 1-week following laser exposure corresponding to the lesion area, with the zone of cone loss expanding over the ensuing weeks (with one lesion example indicated by a green arrow).

Histological Confirmation of Photoreceptor Loss Postlaser Photocoagulation

Light microscopy was used to determine the site of laser injury with maximum cytoarchitectural damage. From 1-week postexposure, ONL attenuation was visible at the lesion center with a loss of photoreceptor nuclei (Figs. 3A–C). A segment of detached RPE could be seen at the base of the lesion in most cases. The outer plexiform layer was disrupted with thickening of the inner nuclear layer (INL), presumably due to edema, while the retinal ganglion cell (RGC) layer was unaffected. As expected, the number of photoreceptor nuclei present at the center of the lesions decreased to 62.4 ± 4.1% and 55 ± 3.8% of nonlesioned internal control areas at 1 week (***t ₅ = 9.22, P < 0.0001, n = 6) and 6 weeks (***t ₆ = 7.65, P < 0.0001, n = 7), respectively. Bruch's membrane remained intact and no subretinal neovascularization was evident at any time point postlaser, confirming the consistency of the laser model. 

Figure 3

 

DAPI-stained sections of retinal laser lesions showing the central area of injury (A–C). Photoreceptor density was calculated by counting the number of outer retinal nuclei in the lesion area (Box 1 in [A]) and expressing this as a percentage relative to a nonlasered internal control area of equal size (Box 2 in [A]). Photoreceptor loss was greater at 6 weeks (C) than 1 week (B) postinjury (arrows). Scale bar: 100 μm. DAPI-stained (blue) sections of retinal laser lesions showing reduction in TUNEL-positive cells (red) between 1 and 5 days postlaser ([D, E], respectively). Scale bar: 50 μm. A confocal micrograph (F) showing the localisation of TUNEL stain predominantly to the photoreceptor layer. Loss of GFP-positive cones coincided with the area of maximal TUNEL staining. The vertical white arrow in the DAPI-stained image shows a TUNEL/GFP-stained cone and the horizontal white arrow in the corresponding Nomarski image indicates the approximate location of Bruch's membrane. Scale bar: 20 μm. Analysis of TUNEL count postlaser showing a reduction in the number of TUNEL-positive cells from 1 to 5 days postlaser exposure (G). Error bar ± SEM, n = 5 per time point. **P < 0.01. INL, inner nuclear layer; ONL, outer nuclear layer.

The greater loss of photoreceptor nuclei at weeks compared to 1 week implied another mechanism of cell death beyond necrosis might be active in the later stages following laser exposure. To assess the contribution of apoptosis histologically, TUNEL stain was applied and was found to be confined to the ONL, confirming that photoreceptor death occurred via the activation of apoptotic pathways (Figs. 3D, 3E). Confocal microscopy showed that TUNEL staining was located at different depths of the ONL (Fig. 3F), reflecting rod apoptosis throughout the ONL and cone apoptosis that was limited to the outer region of the ONL close to the external limiting membrane, where cone nuclei typically reside. The area of maximal TUNEL staining coincided with the loss of GFP-positive cones. The number of TUNEL-positive cells decreased from 1 to 5 days postlaser exposure (F _2,12 = 8.35, P < 0.01, n = 5 per time point, **P < 0.01 post hoc, Fig. 3G). 

CNTF Attenuates M Cone Loss Postlaser Exposure

AF imaging 3 weeks after laser application showed that there were more surviving M cones in CNTF-treated eyes (54.1 ± 5.15% of baseline count) than in sham-injected eyes (28.7 ± 4.4%; **t ₁₀ = 3.81, P = 0.03, n = 11, Fig. 4). This difference in cone survival persisted at the 6-week time point (treated, 39.6 ± 3.2% versus sham, 18.0 ± 3.8%; **t ₁₀ = 3.53, P = 0.005, n = 11, Fig. 4). There was no significant difference in the number of GFP-positive cones in nonlesioned internal control areas in either the CNTF or sham-injected experiments at any time point (paired t-test, P = 0.73 and 0.43 at 3 and 6 weeks, respectively), showing that the effect on cone survival was specific for the presence of CNTF and not secondary to trophic effects due to the intraocular injection procedure. Furthermore at 6 weeks, the lesion area was smaller in CNTF-treated (44.6 ± 0.07%) than in sham-injected (59.5 ± 0.08%) eyes (*t ₁₀ = 2.85, P = 0.017, n = 11). 

Figure 4

 

High-power cSLO images immediately (“baseline”) and 6 weeks following intravitreal CNTF administration versus sham-injection postlaser exposure. At 6 weeks, CNTF-treated eyes (A) showed greater preservation of GFP-positive cones on 488 nm imaging within the site of injury compared with contralateral sham-injected eyes (B). Areas of decreased reflectance seen on near-infrared reflectance imaging immediately following laser corresponded to the lesion area. Nonlesioned internal control areas remained unchanged between baseline and six weeks in both treated (C) and sham-injected (D) groups. Scale bar: 1 disc diameter. Analysis of cone survival by cSLO imaging showing cone photoreceptor number (indicated by GFP counts), following intravitreal CNTF versus sham-injection postlaser (E). The CNTF-treated group showed greater cone survival than sham-injected controls, equivalent to rescue of approximately 50% of cones at six weeks postlaser. Error bar ± SEM, n = 10 or 11 eyes in each group at each time point. **P < 0.01 for both groups.

As an additional validation, histological assessment (which includes cell counts of rod and cones) was also performed on retinal sections through the laser lesions. In keeping with the in vivo imaging, the proportion of surviving photoreceptors within lesions (normalized to the photoreceptor count in a nonlesion control area in each eye) was significantly greater in CNTF-treated eyes compared with sham-treated eyes (63.4 ± 2.3% vs. 50.8 ± 3.4%, treated and sham-injected group, respectively, by 6 weeks; **t ₉ = 3.78, P = 0.004, n = 10, Figs. 5A, 5B). Although there was a trend, no significant difference was detected in the photoreceptor count at 1-week postlaser, possibly due to greater variability at this stage (Fig. 5C). 

Figure 5

 

DAPI-stained sections of lesion centers at 6 weeks postretinal laser (A, B). A greater preservation of photoreceptors was observed within the CNTF group (A) compared with contralateral sham-injected (B) eyes. Scale bar: 50 μm. Analysis of rod and cone survival by comparing photoreceptor nuclei within the lesion center following intravitreal CNTF versus sham-injection postlaser (C). Photoreceptor density is given as a percentage within the lesion center relative to a nonlasered internal control area. An overall increase in the survival of photoreceptors was seen within the CNTF-treated group relative to the sham-injected experiment at 3 and 6 weeks postlaser. This result is consistent with the effect observed in cones by in vivo analysis. Error bar ± SEM, n = 10 eyes in each group at each time point. **P < 0.01.

CNTF Inhibition of Laser-Induced Photoreceptor Apoptosis Correlates With Modulation of the Bax Pathway in C57BL/6 Mice

Having established the positive neuroprotective effects of CNTF on photoreceptor survival following retinal laser injury, further experiments were performed to identify which molecular pathways of cell death might be involved. The expression of candidate genes involved in the apoptotic pathway in 14 retinae 3 days after laser exposure was investigated by qPCR (Fig. 6). Three days after laser application, the expression of retinal c-Fos mRNA was increased by 2.4-fold (*t ₆ = 3.60, P = 0.01, n = 7) relative to the nonlasered group (Fig. 6). A positive trend was seen for the proapoptotic gene, Bax, with relative induction in the laser compared with nonlaser group. Notably a negative trend was observed for the antiapoptotic gene, Bcl-2, in the laser-treated group compared with nonlaser controls. However, in both these latter cases, the trends were not statistically significant. The expression of Stat-3 tended to increase 3 days after laser exposure, although this trend was also not statistically significant. Generally considered to be antiapoptotic, Stat-3 can also act to promote apoptosis, which might be relevant following laser injury. ³⁰  

Figure 6

 

Apoptotic gene expression postlaser, determined by qPCR in C57BL/6 retinae at 3 days postexposure. Results showed a significant increase in the transcriptional expression of proapoptotic c-Fos, and positive and negative trends for Bax and Bcl-2 expression, respectively, in lasered eyes. Error bar ± SEM, n = 7 mice. *P < 0.05.

In a comparison of eyes receiving intravitreal CNTF versus contralateral sham injection (Fig. 7A), only Bax expression was significantly influenced by CNTF, which led to a 2-fold reduction in Bax expression at 1 week postlaser exposure (**t ₆ = 4.07, P = 0.007, n = 7, Fig. 7B) relative to the sham-injected group. In addition, positive and negative trends were observed in the expression of other apoptotic genes in response to CNTF treatment: in particular, Bcl-2 expression increased after 1 week; c-Fos expression decreased after 3 days, and Stat-3 expression decreased after 3 days and 2 weeks (Fig. 7A). Although not statistically significant, the trends with CNTF treatment were inverse to those resulting with laser treatment alone, suggesting that this neuroprotective agent may act on similar cell stress-related pathways to reverse any changes in apoptotic gene expression that are induced by laser injury. 

Figure 7

 

A heat map of differentially expressed apoptotic genes postintravitreal CNTF versus contralateral sham-injection following laser (A). The numbers used to generate the heat map are the mean fold changes (absolute values) in gene expression determined from sham-injected eyes compared with those calculated from the contralateral CNTF-injected group. Red and green indicate an increase or decrease in the mRNA level, respectively. The heat map colors represent the expression of each gene individually and are not comparable between genes. The expression of the Bax gene after intravitreal CNTF injection following laser treatment, compared with contralateral sham controls (B). qPCR analysis in C57BL/6 retinae at time points postlaser exposure. At 1 week postlaser, Bcl-2/Bax ratios were elevated (red line) while Bax was inhibited (black columns) following intravitreal CNTF compared to controls. Error bar ± SEM, n = 28 mice. **P < 0.01.

Discussion

In this study, we first defined a reproducible animal model to study secondary, apoptotic cone loss, following retinal laser injury. We subsequently showed that a single intravitreal injection of CNTF protein delivered immediately following laser exposure had a neuroprotective effect on cones. Translating this into the clinical scenario and to the military setting in particular, a foveal laser injury would cause immediate visual incapacitation and be potentially catastrophic if experienced on the battlefield. Identifying a treatment that could be used in conflict situations would reduce the threat of military laser weapons. To that end, this study demonstrates that intervention directed to modulate the apoptotic pathway in our model can prevent secondary cone death resulting from retinal laser damage and may thereby limit the size of a subsequent scotoma. 

Although fluorescein angiography, near infrared reflectance imaging, white light funduscopy, and electroretinographic techniques have been used to study the effects of laser injury in various animal models, ^31–35 the majority of previous investigations have drawn conclusions based on histology. Aside from the benefits of noninvasive imaging, as per our in vivo model, in helping to reduce the number of experimental animals required, histological techniques are severely limited due to a number of factors: such studies are necessarily cross-sectional and, therefore, subject to a source of biological and procedural error across animals and time points. ³⁶ Quantification of photoreceptor nuclei or ONL thickness in a lasered area depends on localisation of a focal area of injury. This is potentially inaccurate due to asymmetry of the injured area, variability of laser exposure and selection bias. However, coupled with a longitudinal imaging method, our approach provides for a more robust method of analysis. A limitation of this in vivo study is that media clarity needs to be maintained in the presence of the intravitreal agent, which, therefore, excludes certain opaque compounds (e.g., triamcinolone acetonide) from investigation that may otherwise be used in routine clinical practice via this route. 

In our in vivo model, we observe hypoautofluorescence of the outer retina immediately following laser with no changes in autofluorescence of the inner retina. Two possible mechanisms to explain the hypoautofluorescent changes in the outer retina could be due to, firstly, the immediate damage of GFP positive cones that would occur as a result of photothermal injury or, secondly, due to the presence of edema with resultant blockage of autofluorescence. ³⁷ The finding that the loss of GFP fluorescence over 6 weeks as the lesion expanded indicates that cones are progressively lost following an initial injury in our model. This result is consistent with previous reports of enlargement of retinal laser lesions in humans over time, where lesion enlargement by 42% at 32 months ³⁸ and 103% at 1 year ³⁹ was determined by measuring the macular laser lesion diameter following treatment for choroidal neovascularization. 

By using the TUNEL assay, we show that a mechanism of photoreceptor death occurring within the first week postinjury is by apoptosis. In order to target this secondary cell death, neuroprotective treatment would need to be initiated in the acute phase of injury. Since our data suggest that maximal apoptosis occurs between 1 and 3 days postlaser in line with previous reports investigating retinal laser injury, ¹⁷ this time period would provide the optimum window for therapeutic intervention. The neuroprotective effect of CNTF has been demonstrated in previous rodent models of retinal injury, ⁴⁰ and more recently, in a retinal explant model of cone degeneration. ⁴¹ Therefore, we chose to assess CNTF as a potential therapeutic agent in our model of cone loss induced by laser exposure. By using a multimodal approach consisting of both in vivo imaging and in vitro histological and molecular techniques, we were able to confirm the protective effect of CNTF in our model of laser injury. 

Bcl-2, Bax, and c-Fos have all been identified as key components in the intrinsic or mitochondrial apoptotic pathway, with the latter being activated by toxic light exposure. ^42,43 Bax protein is a potent proapoptotic initiator in the cell death cascade. Following homodimerization, the resulting configuration forms a channel across the mitochondrial membrane. This allows cytochrome-C to exit the mitochondria and enter the cytosolic space, where it is involved in conversion of pro-Caspase-9 to Caspase-9 and subsequent initiation of apoptosis. ⁴⁴ CNTF has previously been shown to inhibit Bax in studies of axotomized motor neurons following sciatic nerve transection. ⁴⁵ The antiapoptotic molecule, Bcl-2, which is an integral membrane protein localized to mitochondria, inhibits the activation of Bax following initiation of a death signal. ⁴³ As such, the ratio between these two proteins reflects the susceptibility of cells to a death signal, with positive correlations having been shown between higher Bcl-2/Bax ratios and injury. ^46,47 Additionally, the c-Fos gene, which encodes a nuclear phosphoprotein that constitutes the transcription factor activator protein 1 (AP-1), has been implicated as having a role in light-induced photoreceptor degeneration. ⁴⁸ Upregulation of c-Fos gene expression has been demonstrated in rodent models of retinal cell apoptosis following photic injury. ⁴² Our results are consistent with the aforementioned findings, showing a reduction of Bax mRNA, an increase in the Bcl-2/Bax ratio, and a trend toward reduction of c-Fos mRNA following CNTF administration postlaser trauma, thus suggesting that CNTF affects Bax, Bcl-2, and c-Fos at both the transcriptional and posttranslational levels. 

Nonetheless, the precise mechanism by which CNTF improves photoreceptor survival is unknown. CNTF has been shown to act on the alpha CNTR receptor (CNTFRα), the beta leukemia inhibitory factor receptor (LIFRβ), and the glycoprotein 130 receptor (gp130R). ⁴⁹ Photoreceptors lack a CNTF receptor, which is found on inner retinal neurons and Müller cells. Based on this, the postulated mechanism underlying CNTF-dependent photoreceptor protection is thought to be related to Müller cell activation and antiapoptotic Janus kinase/signal transducer and activator of transcription (JAK/STAT) and MAPK signaling pathways. ⁵⁰  

Given the interspecies differences in anatomy, genomes, and response to injury, in certain disease processes it may be difficult to make a direct correlation between the results observed in rodents and humans. ⁵¹ However, as we have discussed, in both the rodent and human eye, photoreceptor cell death following laser injury has been shown in previous work, as has the protective effect on photoreceptors with the use of CNTF. Therefore, our study combines the use of these established facts and thus allows the extrapolation of results specific to cone injury. 

In conclusion, although our research has a military focus with regard to developing a potential treatment for offensive laser weapons on the battlefield, the model we have developed might be relevant to assessing any treatment relevant to cone neuroprotection and diseases of the human fovea. 

Acknowledgments

We thank Mark Hankins, PhD (University of Oxford), for his support of W.I.L. Davies. 

Presented at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May 2012. 

Supported by the UK Ministry of Defence, the Special Trustees of Moorfields Eye Hospital, the NIHR Biomedical Research Centres at Moorfields Eye Hospital and Oxford University Hospitals NHS Trust, an Australian Research Council (ARC) Future Fellowship (WILD), and the Royal College of Surgeons of Edinburgh. 

Disclosure: S.A. Aslam, None; W.I.L. Davies, None; M.S. Singh, None; P. Charbel Issa, None; A.R. Barnard, None; R.A.H. Scott, None; R.E. MacLaren, None