Ccl2/Cx3cr1 Knockout Mice Have Inner Retinal Dysfunction but Are Not an Accelerated Model of AMD

Abstract

Purpose.: The chemokine, Ccl2, and the fractalkine receptor, Cx3cr1, have both been implicated in the pathogenesis of age related macular degeneration (AMD), with mice lacking both genes exhibiting features of AMD by 3 months of age. However, recent reports indicate that this ascribed phenotype is due to the presence of a retinal degeneration mutation (crb1^rd8/rd8 , rd8) on the background strain. Our aim was to characterize the retinal effects of lack of Ccl2 and Cx3cr1 (Ccl2^−/−/Cx3cr1^EGFP/EGFP, CDKO-mice), in mice without the rd8 mutation.

Methods.: Nine-month-old, CDKO and wildtype C57blk6J mice were investigated for retinal fundus appearance and histology. The function of the rod and cone pathways was assessed using the ERG.

Results.: The CDKO mice did not develop lesions in the retinal fundus, and the ultrastructure of Bruch's membrane and the RPE were similar to that of C57blk6J mice. From the ERG, there was no change in the amplitude of the rod photoreceptor response, or in the rod or cone post-photoreceptor b-wave. However, the rod and cone ERG oscillatory potentials were significantly reduced in the CDKO animals, a phenotype apparent in Cx3cr1^EGFP/EGFP- but not Ccl2^−/−-founder lines. This correlated with aberrant amacrine cell morphology in the CDKO mice. In addition, Müller cells were gliotic and microglial morphology subtly altered, indicative of retinal stress.

Conclusions.: These results suggest that in the absence of the rd8 mutation, the CDKO-mouse has a mild inner retinal phenotype characterized by altered amacrine cell function, but that it is not an accelerated model of AMD.

Introduction

AMD is one of the leading causes of blindness in the Western world. ¹ A clinical hallmark of the disease is the presence of white or yellow spots in the fundus, called drusen, atrophy of the RPE, and eventual death of the photoreceptors. ¹ Drusen are lipid-rich, highly immunogenic deposits that build up in the Bruch's membrane between the RPE and the choroidal vasculature. One of the main genetic factors predisposing individuals to the development of AMD is failure of the components of the innate immune system, which are thought to be important in clearing these deposits. Single nucleotide polymorphisms in genes such as complement factor H, ^2,3 C3, and the chemokine receptor Cx3cr1 ⁴ have all been found to predispose individuals to the development of AMD. 

Recent work on transgenic mice with altered innate immune systems indicates that these might be valuable models for studying the development of AMD in humans. ⁵ Indeed, mice lacking the chemokine Ccl2, also known as monocyte chemotactic protein-1, have been found to have many of the features of AMD by 15 months of age. ⁶ In the retina, Ccl2 is a chemokine that is likely released from glia, when the retina is under stress, to attract microglia/macrophages expressing the Ccr2 receptor to sites of retinal damage. ⁷ Loss of this chemokine has been shown to induce fundus lesions, accumulation of lipofuscin in the RPE, drusen, photoreceptor death, and choroidal neovascularization. ⁶ Similarly, Cx3cr1 is a receptor expressed only on microglia in the retina and it coordinates the response of the microglia to the chemokine, Cx3cl1 (fractalkine/neurotactin). ⁸ Cx3cl1 is commonly expressed by neurons and glia and is traditionally thought to have an anti-inflammatory or calming effect on microglia in the CNS. ⁸ Like Ccl2^−/−-mice, mice lacking the chemokine receptor, Cx3cr1 have been found to have white deposits in the retinal fundus and photoreceptor degeneration with age. ⁹ Recent work on both aged Ccl2^{−/− 10} and aged Cx3cr1^−/− animals ¹¹ indicates that the white deposits in the retinal fundus of these animals may not be the traditional drusen observed in human AMD, but instead subretinal microglia. Results from a study on aged Ccl2^−/− mice suggest that the subretinal microglia do not accelerate photoreceptor death or retinal functional loss above that seen in normal aging. ¹⁰ In contrast, subretinal microglia in the aged Cx3cr1^−/− mice have been suggested to be involved in traditional drusen formation, ¹¹ as subretinal macrophages have been reported in histological studies of human AMD. ¹²  

Mice lacking both Ccl2 and Cx3cr1 have been found to have an accelerated AMD phenotype, with white deposits in the fundus by 3 months of age. ^13–16 However, there is some variability in the reported phenotype of these mice and recent work indicates that this variability may be due to the presence of an underlying retinal degeneration mutation (Crb1, rd8) on the background C57blk6 strain. ^17,18 Indeed, Ccl2^−/−/Cx3cr1^−/− mice with the Crb1^rd8/rd8 mutation have been found to have white spots in the retinal fundus while Ccl2^−/−/Cx3cr1^−/− mice without this mutation do not. ¹⁷ Further work is required, however, as this recent study did not go on to investigate the effects of loss of Ccl2 and Cx3cr1 on either retinal structure or function in the absence of rd8. While it appears that the presence of fundus spots in the Ccl2^−/−/Cx3cr1^−/− mice may be due to the presence of the rd8 mutation, ¹⁷ it is unclear what effect the loss of Ccl2 and Cx3cr1 has on retinal function and morphology in a model free from rd8. 

The aim of the present study was to investigate the retinal phenotype of mice lacking Ccl2 and Cx3cr1, generated by crossing Ccl2^−/− and Cx3cr1^EGFP/EGFP mice on a C57blk6J background obtained from Jackson Laboratories (Bar Harbour, ME). Nine-month-old Ccl2^−/−/Cx3cr1^EGFP/EGFP mice (CDKO mice) and age-matched C57blk6J, wildtype (WT) mice were investigated for fundus appearance, retinal function using the ERG, and ocular morphology using histological techniques. As previously reported, CDKO mice were not found to have white spots in the retinal fundus, while mice harboring the rd8 mutation exhibited this classical phenotype. ¹⁷ The structural data show Bruch's membrane and RPE morphology to be qualitatively similar in CDKO and WT mice, suggesting that the CDKO mice on an rd8-free background are unlikely to be a good accelerated model of AMD. Despite not showing the classical AMD pathology, CDKO mice had reduced ERG oscillatory potentials suggesting altered amacrine cell function. This phenotype was apparent in age-matched Cx3cr1^GFP/GFP-, but not Ccl2^−/−-founder mice, suggesting that loss of Cx3cr1 signaling underlies this phenomenon. In addition, CDKO mice had altered amacrine cell morphology consistent with the ERG phenotype. Moreover, Müller cells were gliotic and the morphology of microglia in the inner plexiform layer (IPL) was altered in a manner consistent with an activated phenotype. These data suggest that intact chemokine signaling is required for normal inner retinal function. 

Methods

Animals

C57blk6J mice (WT mice) were obtained from the Animal Resource Center, Australia. Ccl2^−/− mice on a C57blk6J background were obtained from Jennifer Wilkinson-Berka and their original source was Jackson Laboratories, Maine (B6.129S4-Ccl2^tm1Rol /J, stock number 004434). Cx3cr1^GFP/GFP, which express EGFP under control of the endogenous Cx3cr1 locus, on a C57blk6J background were obtained from Paul McMenamin. ¹⁹ Ccl2^−/− mice were crossed with Cx3cr1^GFP/GFP mice to generate a double knockout, Ccl2^−/−/Cx3cr1^GFP/GFP mice, henceforth called CDKO-mice. Mice harboring the crb1^rd8/rd8 mutation (rd8 mice) were included for comparison of retinal fundus phenotype and histology. All mice were bred and housed at the University of Melbourne animal facility on a 12-hour light/dark cycle with cage illumination <10 lux during the light period. Food and water was available ad libitum. All experiments adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and the ethics committee standards of the University of Melbourne. All experiments were completed on 9-month-old animals. 

Genotyping

Small tail samples were collected from WT, CDKO, and rd8 mice when the mice were weaned for extraction of genomic DNA and rd8 genotyping. Standard PCR-based genotyping for Ccl2 and Cx3cr1 was completed according the protocol provided on the Jackson Laboratories' website. Genotyping for rd8 mutation was performed as previously described by Mehalow et al., ²⁰ using allele specific PCR incorporating the primers specified in this study. In addition, to ensure the CDKO lacked the rd8 mutation, specific primers flanking the rd8 point mutation were designed, and a PCR using the genomic DNA from WT, CDKO, and rd8mice completed. The resulting PCR product samples were sequenced (Australian Genome Research Facility, Melbourne, Australia; Fig. 1). A deletion of nucleotide 3481, a cytosine, of the Crb1 gene was observed in rd8 mice. This region of the Crb1 gene in the CDKO and WT controls was normal, suggesting neither of these strains harbored the rd8 mutation (Fig. 1; arrows indicate nucleotide 3481). 

Figure 1. 

 

Genotyping for Crb1^rd8/rd8 mutation. To ensure the CDKO mice lacked the rd8 mutation, specific primers flanking the Crb1^rd8/rd8 point mutation were used to amplify genomic DNA from CDKO and rd8 mice. The resulting PCR products were sequenced. It is known that the rd8 mutation arises due to a single base pair deletion at nucleotide 3481. In the CDKO mice a cytosine is present at nucleotide 3481 of the Crb1 gene indicating these mice did not harbor the rd8 mutation, while in the rd8-positive mice, this nucleotide was deleted. Arrows indicate nucleotide 3481.

Fundus Photography

Nine-month-old WT, CDKO, and rd8 mice were anesthetized using a mixture of ketamine (67 mg/kg) and xylazine (13 mg/kg). The corneal reflex of the mice was anesthetized with topical application of proparacaine hydrochloride eyedrops (Alcaine 0.5%; Alcon Laboratories, Frenchs Forest, Australia) and the pupils of the mice were dilated with topical application of tropicamide eyedrops (Mydriacyl 0.5%; Alcon Laboratories). Mice were placed in a specialty holder and the retinal fundi of mice was viewed and photographed using a fundus camera, (Micron III; Phoenix Research Laboratories, Inc., Pleasanton, CA). Fundus images were viewed and collected using specialty software (Phoenix Research Laboratories, Inc.). 

Gross Histology

The gross retinal morphology of WT, CDKO, and rd8 mice was ascertained using toluidine blue stained resin sections using previously published techniques. ²¹ Eye cups were fixed overnight in 1% paraformaldehyde, 2.5% glutaraldehyde, 3% sucrose, 0.01% calcium chloride in 0.1 M phosphate buffer pH 7.4 (PB). Eyecups were washed in PB, and then dehydrated in a graded series of methanol (75%, 85%, 95%, and 100%) and acetone (100%). Tissues were embedded in an epon resin (ProSciTech, Queensland, Australia) and polymerized overnight at 60°C. Retinae were sectioned (1 μm) on an ultramicrotome (Reichert-Jung Ultracut S; Reichert, Depew, NY) and stained using 1% toluidine blue. A microscope (Axioplan; Carl Zeiss, Göttingen, Germany) was used to view retinal sections, and images were captured using a digital camera and computer software (SPOT, version 3.5.2; Diagnostic Instruments, WA, Australia). Images were converted to grayscale and adjusted for white levels, brightness, and contrast with graphics editing software (Adobe Photoshop CS4; Adobe Systems, San Jose, CA). For analysis of retinal thickness in 9-month-old WT (n = 6) and CDKO mice (n = 6), sections within 150 microns of the optic nerve were measured using an image processing freeware (ImageJ 1.43; National Institutes of Health [NIH], Bethesda, MD) for the thickness (μm) of: the total retina (outer limiting membrane to inner limiting membrane); the photoreceptor inner and outer segments (IS and OS, respectively); the outer nuclear layer (ONL); the inner nuclear layer (INL); and the IPL. 

Electron Microscopy

Transmission scanning electron microscopy was used to assess the ultrastructural morphology of the retina, RPE, and Bruch's membrane of 9-month-old WT and CDKO mice. Eye cups were fixed overnight as for gross histology, washed in cacodylate buffer, incubated for 1 hour in 1% OsO₄ prepared in cacodylate buffer, dehydrated, and embedded in epon resin. Ultrathin sections (70 nm) were cut on an ultramicrotome (Reichert-Jung Ultracut S; Reichert) and collected on formvar-coated copper grids. Sections were contrasted with uranyl acetate and lead citrate solutions, and viewed on a Philips CM120 electron microscope (Philips CM120 BioTwin Transmission EM; Philips, North Ryde, NSW, Australia). 

ERG

Nine-month-old WT (n = 18); Ccl2^−/− (n = 9); Cx3cr1^GFP/GFP (n = 19); and CDKO mice (n = 14) were dark-adapted overnight. Under dim red illumination, the mice were anesthetized as for fundus photography using a mixture of ketamine and xylazine. The mice were administered topical proparacaine hydrochloride eyedrops (Alcon Laboratories) and tropicamide eyedrops (Alcon Laboratories) and placed on a heating pad to maintain body temperature. A custom-made AgCl recording electrode was placed centrally on the cornea and a reference electrode placed in the mouth of the mouse. Coordination of ERG stimulation and recording of electrical responses was completed using commercial software (Scope, version 3.6.9; ADInstruments, NSW, Australia) and the responses acquired were filtered for 60-Hz noise, amplified and digitized at 10 kHz over a 250 ms epoch (gain × 5000; −3 dB at 1 Hz and 1 kHz; ADInstruments). To elicit the ERG response, a full field flash of 2.1 log cd-s/m² was generated by a photography flash unit (Nikon SB-900; Nikon Australia, NSW, Australia) and delivered via a custom-made Ganzfeld. Specifically, two flashes were delivered with a 0.8-s inter stimulus interval, to elicit responses from the rod and cone pathways (mixed response) and the cone pathway (cone response) alone, respectively. The cone response was digitally subtracted from the mixed response to generate the rod response. ^22,23  

ERG component analysis was completed using previously published equations and techniques. ^24,25 The rod photoreceptor responses (rod a-wave) were isolated and analyzed using a modified PIII model, to derive the amplitude of the PIII response (PIII R _max in μV) and the sensitivity, which represents the gain of phototransduction cascade (S in m²cd⁻¹s⁻³). The rod post-photoreceptoral function (rod b-wave) was isolated by subtraction of the rod PIII from the raw rod waveform and then fitted using an inverted gamma function to generate the rod PII. From the rod PII fit, the amplitude of the PII response (rod PII R _max in μV) and the time to peak (implicit time in ms) were derived. 

Due to the small number of cone photoreceptors in the mouse, the cone photoreceptor response (cone a-wave) was too small to be analyzed. However, the cone post-photoreceptoral response (cone b-wave) could be assessed and the cone PII was analyzed by fitting an inverted gamma function to the raw cone waveform. From the cone PII fit, the amplitude of the cone PII response (cone PII R _max in μV), and the time to peak (implicit time in ms) were determined. 

The oscillatory potentials (OPs), which ride on the ascending limb of the rod and cone b-wave, are thought to primarily reflect the excitatory and inhibitory activities of the amacrine cells. ²⁶ The OPs were isolated by subtracting the fitted PIIs from the raw waveforms. In the case of the rod ERG, OP2, OP3, and OP4 were analyzed to assess their amplitude (μV) and their implicit time (ms). OP1 and OP5 were not assessed as they were difficult to consistently distinguish from noise due to their small amplitude. In the case of the cone ERG, OP1, OP2, and OP3 were analyzed to assess their amplitude (μV) and their implicit time (ms). 

Immunohistochemistry

Fluorescence immunohistochemistry was used to assess morphology of retinal cell classes using previously described techniques. ²⁷ The posterior eye cups of WT and CDKO mice were fixed for 30 minutes in 4% paraformaldehyde in PB, washed three times in PB, and cryoprotected in a series of graded sucrose solutions (10%, 20%, and 30% in PB). Tissues were then either processed for flat mount, immunohistochemistry, or embedded in optimal cutting temperature (OCT) (Tissue-Tek OCT compound; Sakura, Torrance, CA); frozen and sectioned transversely at 14 μm on a cryostat at −20°C (Microm, Walldorf, Germany). Sections were collected on slides coated with permanent adhesive (polysine; Menzel-Glaser, Braunschweig, Germany) and stored at −20°C. 

For labeling of sections on slides, slides were defrosted and washed in PB. Sections were coated in a blocking solution (10% normal goat serum [NGS], 1% BSA, 0.5% non-ionic detergent [Triton-X; Sigma-Aldrich Corp., St. Louis, MO] in PB) for 1 hour before incubation overnight, at room temperature, in primary antibody (Table 1) dissolved in antibody buffer (3% NGS, 1% BSA, 0.5% nonionic detergent [Sigma-Aldrich Corp.] in PB). After washing in PB, sections were incubated with secondary antibody: goat anti-guinea pig, goat anti-mouse, or goat anti-rabbit conjugated to fluorescent dyes (AlexaFluor 488, AlexaFluor 594, or AlexaFluor 643; diluted 1:500; Invitrogen, VIC, Australia) as required and a cell nuclei label, DAPI (diluted 1:500; Invitrogen) for 90 minutes. The sections were washed in PB, coated in a glycerol/Mowiol-based mounting media, and covered with a glass coverslip. 

Table 1. 

 

List of Primary Antibodies Used for Immunohistochemistry

Cell Type/Target	Antibody	Dilution	Source
Microglia	Rabbit anti-ionized calcium binding adaptor molecule 1 (IbA1)	1:1,500	Cat. No. 019-19741; Wako Pure Chemical Industries, Richmond, VA
Müller cells	Mouse anti-glutamine synthetase (GS, clone GS-6)	1:1,000	Cat. No. MAB302; Millipore, NSW, Australia
Blood vessels	Lectin from Bandeiraea simplicifolia BS-I isolectin B4 FITC conjugate	1:75	Cat. No. L2895; Sigma-Aldrich, NSW, Australia
Astrocytes and gliotic Müller cells	Rabbit anti-glial fibrilliary acidic protein	1:20,000	Cat. No. Z0334; Dako, Carpinteria, CA
Rod bipolar cells	Mouse anti-PKCα	1:400	Cat. No. P 5704; Sigma-Aldrich, NSW, Australia
Cone-ON & OFF bipolar cells (type 2 & 6)	Mouse anti-ZNP-1	1:2,000	Cat. No. ZDB-ATB-081002-25; ZIRC, Eugene, Oregon
Amacrine and ganglion cells	Mouse anti-calretinin 6B3	1:1,000	Cat. No. 63B; Swant, Bellinzona, Switzerland
Photoreceptor and bipolar cell terminals	Guinea pig anti-VGLUT1	1:1,000	Cat. No. AB5905; Millipore, NSW, Australia

For labeling of whole retinae, entire eyecups in 30% sucrose were frozen and thawed three times, washed in PB, and blocked for 1 hour using the blocking solution (above). Retinae were then incubated for 4 days in primary antibody to label microglia (rabbit anti-IbA1, 1:1500; Cat. No# 019-19741; Wako Pure Chemical Industries, Richmond, VA) dissolved in antibody buffer (above). After washing in PB, retinae were incubated with secondary antibody, goat anti-rabbit conjugated to fluorescent dye (diluted 1:500; Invitrogen) overnight. The sections were washed in PB, coated in a glycerol/Mowiol-based mounting media, and covered with a glass coverslip ganglion cell–layer side up. 

Retinae were viewed and imaged using a confocal microscope (Zeiss Meta or Pascal LSM-5; Carl Zeiss). Air objective (×20) or oil objectives (×40 and ×63) were used as required and fluorophore-labeled sections were captured at a resolution of 1024 × 1024 pixels using image browser software (Zen; Carl Zeiss) and an appropriate fluorescence filter (Alexa TM 594/CY3: excitation 568 nm, emission filter 605/32; Alexa TM 488/FITC: excitation 488 nm, emission filter 522/32). Far-red, red, green, and blue fluorescence were scanned separately and adjusted for black levels and contrast with graphics editing software (Adobe Systems). 

Analysis of Microglia Number and Morphology

Flat-mounted retinae from WT (n = 4) and CDKO mice (n = 4) were imaged for microglia at the level in the IPL. For cell counts, tile scans of the entire retina were collected using a ×20 air objective. The area of the retina in correct focal plane (mm²) and the number of microglia within that area was assessed using image processing freeware (NIH) and cell counts were completed using the macro, ITCN. For morphological analysis, at least three high-resolution Z-stack images containing four to six microglia per image were captured using an ×40 oil objective (×0.7 zoom) for each retina. Microglial morphology was assessed using microscopy image analysis software (MetaMorph Offline; Molecular Devices Corporation, Sunnyvale, CA). Soma area was assessed using cell scoring application (Multi Wavelength Cell Scoring Application; Molecular Devices Corporation) and all other morphological parameters were determined using the neurite outgrowth application module (Molecular Devices Corporation). For a given retina, the morphological properties of at least 15 microglia from that retina were assessed and averaged to generate a single result (n = 1). Morphological parameters from n = 4 animals in each group were assessed for statistical comparison including: soma area (μm²), number of processes, number of branches, and total outgrowth of processes, equating to the summed length of all processes (μm), which is a measure of arbor spread. 

Statistical Analysis

Manipulation and curve fitting of the ERG data was completed using spreadsheet software (Excel 2007; Microsoft, Redmond, WA). Graphing and statistical analysis was performed using scientific graphing software (GraphPad Prism 4; GraphPad Software, San Diego, CA). ERG and retinal thickness measures from WT and CDKO mice are presented as the mean ± SEM. Data for retinal thickness measures and the ERG PIII and PII parameters were compared between WT and CDKO mice using a Student's t-test performed using scientific graphing software (GraphPad Software) and differences between the mean responses were considered significant if P < 0.05. Data for ERG OP parameters were compared using a two-way ANOVA for genotype (WT, Ccl2^−/−, Cx3cr1^GFP/GFP, and CDKO mice) and OP number (e.g., OP2, OP3, OP4) performed using scientific graphing software (GraphPad Software) and differences between the mean responses were considered significant if P < 0.05. Data for cell counts and microglial morphological analysis, based on n = 4, were compared between WT and CDKO mice using a nonparametric, Mann-Whitney test performed using scientific graphing software (GraphPad Software) and differences between the responses were considered significant if P < 0.05. 

Results

Retinal Fundus and Gross Retinal Morphology

The retinal fundi of 9-month-old WT, CDKO mice and adult rd8 mice were imaged using a fundus camera (Phoenix Research Laboratories, Inc.; Figs. 2A–C). While the fundi of mice harboring the rd8 mutation showed multiple white spots across the retina, in particular in the temporal/inferior quadrant (Fig. 2C), the fundi of WT- and CDKO mice did not (Figs. 2A, 2B, respectively). In the rd8 mice, the white spots in the fundus likely corresponded to regions of pigment loss due to RPE and photoreceptor cell death, which could be observed in transverse sections of toluidine blue stained retina (Fig. 2F, black arrow). In contrast, the gross retinal morphology of both QT and CDKO mice was normal and qualitatively similar (Figs. 2D, 2E). This finding supports the work of Luhmann et al., ¹⁷ and suggests that CDKO mice that do not have the rd8 mutation have a normal retinal fundus at nine months of age. 

Figure 2. 

 

Fundus photography and gross retinal histology of WT, CDKO, and Rd8 mice. (A–C) Retinal fundi were imaged using a fundus camera and images are presented from 9-month-old (A) WT mice, (B) CDKO mice, and (C) adult Rd8 mice. On inspection of the fundus, rd8 mice had abnormal white spots while WT and CDKO mice fundi were apparently healthy and qualitatively similar. S, superior; T, temporal; I, inferior; N, nasal. (D–F) Eyecups were fixed, embedded in epon-resin, cut at 1 μm, and stained with toluidine blue to inspect the gross structural morphology of (D) WT, (E) CDKO, and (F) Rd8-retinae. While the gross morphology of WT and CDKO retinae was qualitatively similar, the photoreceptor layer and RPE (arrow) of rd8 retinae had degenerated. Scale bar, 20 μm.

The thickness of WT and CDKO retinal layers was quantified (Table 2, n = 6 for each) and there were no differences in the photoreceptors (inner segment length, outer segment length, or ONL thickness) or the inner retinal layers (INL or IPL). These data suggest that the absence of Ccl2 and Cx3cr1 does not significantly affect the integrity of the gross retinal architecture. 

Table 2. 

 

Retinal Thickness Measures in 9-Month-Old WT and CDKO Mice

Genotype	Total Retina	OS	IS	ONL	INL	IPL
WT	122 ± 5	28 ± 1	17 ± 1	42 ± 1	25 ± 1	37 ± 1
CDKO	134 ± 4	28 ± 1	18 ± 1	44 ± 1	26 ± 1	39 ± 1
P value	P = 0.054	P = 0.98	P = 0.31	P = 0.20	P = 0.14	P = 0.13

 Retina were embedded in epon-resin, stained with toluidine blue, and images captured with a ×40 oil objective. Thickness of each of the retinal layers: Total retina (outer limiting membrane to inner limiting membrane), OS, IS, ONL, INL, IPL for n = 6 animals in each group was assessed in microns (μm). Data are presented as the mean ± SEM. There were no statistically significant changes in any of the retinal thickness measures between WT and CDKO mice (n = 6 each genotype; Student's t-tests were used to generate P values presented).

To further explore the morphology of the outer retina of CDKO mice for signs of AMD-like pathology, sections of CDKO and WT retina were imaged using high-power magnification, light microscopy, and transmission scanning electron microscopy (Fig. 3). Figures 3A and 3B shows representative high power images (×100 oil objective) of the outer retina of toluidine blue–stained, WT and CDKO mice, respectively. There were no overt changes in photoreceptor inner or outer segments, the RPE, or the choroid in CDKO mice. In addition, the ultrastructural morphology of the RPE (Fig. 3C: WT; 3D: CDKO) and Bruch's membrane (grey line outside Fig. 3E: WT; 3F: CDKO) appeared similar in WT and CDKO mice. In particular, there were no instances of AMD-like pathology in the CDKO mice, such as RPE atrophy, overt thickening of Bruch's membrane, or choroidal neovascularization. 

Figure 3. 

 

Gross histology and transmission electron microscopy of eyecups from WT and CDKO mice. (A, B) Eyecups were fixed, embedded in epon-resin, cut at 1 μm, and stained with toluidine blue to inspect the gross structural morphology of the RPE and photoreceptor IS/OS in (A) WT and (B) CDKO mice. Scale, 20 μm. (C–F) Eyecups were fixed, embedded in epon-resin, cut at 70 nm, and prepared for transmission electron microscopy to inspect the ultrastructural morphology of the RPE in (C) WT and (D) CDKO mice, and Bruch's membrane delineated by a grey line outside micrographs in (E) WT and (F) CDKO mice. Scale bar, 5 and 1 μm, respectively. Ch, Choroid; BM, Bruch's membrane.

Function of Inner Retinal Neurons Is Altered in CDKO Mice

The general function of the retina was assessed in 9-month-old WT (n = 18) and CDKO mice (n = 14) using a twin flash ERG paradigm to isolate the rod and cone pathway responses. Representative, rod ERG waveforms are presented for WT (Fig. 4A) and CDKO mice (Fig. 4B) and show that the overall amplitude of the ERG response is not qualitatively different between the two, but that the OPs of the CDKO mice appear smaller. When quantified, there was no difference between WT and CDKO mice in the amplitude of the rod photoreceptor response (Fig. 4C; rod PIII amplitude, Student's t-test, P = 0.29) or the rod photoreceptor sensitivity (Fig. 4D; rod PIII sensitivity, Student's t-test, P = 0.28). Nor was there any difference between the genotypes, in the rod post-photoreceptor PII (b-wave); amplitude (Fig. 4E; rod PII amplitude, Student's t-test, P = 0.46); or timing (Fig. 4F; rod PII implicit time, Student's t-test, P = 0.92). However, when the Ops—which are thought to primarily represent the activity of the amacrine cells—were isolated from the rod ERG, they were found to be significantly smaller in the CDKO mice than in the WT mice (Fig. 4G; rod OP amplitude, two-way ANOVA, for genotype, P < 0.0001 and OP number, P < 0.0001; individual results for WT versus CDKO mice for OP2, P < 0.05; OP3, P < 0.001; OP4, P < 0.001) without a change in their timing (Fig. 4G; rod OP implicit time, Two-way ANOVA, for genotype P = 0.84). 

Figure 4. 

 

The function of the rod pathway, inner retinal response is altered in CDKO mice. Mice were dark-adapted overnight and the rod ERG responses were isolated from the mixed ERG by using a twin flash paradigm. (A, B) Representative waveforms of the rod ERG recorded from 9-month-old (A) WT (n = 14) and (B) CDKO mice (n = 18) in response to a 2.1 log cd-s/m² intensity flash. The photoreceptor derived, rod PIII (modeled a-wave) was not altered in either (C) amplitude (Student's t-test, P = 0.29) or (D) sensitivity (Student's t-test, P = 0.28) in CDKO mice when compared with the WT response. The post-photoreceptor, rod PII (modeled b-wave) was also not found to be different in either (E) amplitude (Student's t-test, P = 0.46) or (F) timing (Student's t-test, P = 0.92) between the two strains. The inner retinal derived OPs were significantly reduced in (G) amplitude, but not different in (H) timing in CDKO mice (two-way ANOVA, rod OP amplitude; for genotype, P < 0.0001 and OP number, P < 0.0001; individual results for WT versus CDKO mice for OP2, P < 0.05; OP3, P < 0.001; OP4, P < 0.001). * Indicates P < 0.05, a significant difference between WT, n = 18, and CDKO mice, n = 14.

The cone pathway responses were also isolated and although the cone photoreceptor response could not be assessed due to the small number of cones in mice, robust cone post-photoreceptor ERG responses were obtained for WT (Fig. 5A) and CDKO mice (Fig. 5B). As was seen for the rod pathway response, there was no change in the cone post-photoreceptor PII (b-wave) amplitude (Fig. 5C; cone PII amplitude, Student's t-test, P = 0.90) or implicit time (Fig. 5D; cone PII implicit time, Student's t-test, P = 0.18) between the genotypes. However, there was a significant loss of amplitude in the cone OPs in the CDKO mice (Fig. 5E; cone OP amplitude, two-way ANOVA, for genotype P < 0.0001 and OP number P < 0.0001, individual results for WT versus CDKO mice for OP1, P > 0.05; OP2, P < 0.001; OP3, P < 0.001) without a change in the kinetics of the response (Fig. 5F; cone OP implicit time, two-way ANOVA, for genotype P = 0.84). These data suggest that overall retinal function is intact in the 9-month-old CDKO mice, with no changes indicative of AMD such as photoreceptor/RPE dysfunction observed. However, the significant loss of both the rod and cone pathway OPs suggests that CDKO mice have inner retinal dysfunction, likely derived from alterations to bipolar or amacrine cell signaling. 

Figure 5. 

 

The function of the cone pathway, inner retinal response is altered in CDKO mice. (A, B) Representative waveforms of the cone ERG recorded from 9-month-old (A) WT (n = 18) and (B) CDKO mice (n = 14) in response to a 2.1 log cd-s/m² intensity flash. The post-photoreceptor, rod PII (modeled b-wave) was not found to be different in either (C) amplitude (Student's t-test, P = 0.90) or (D) timing (Student's t-test, P = 0.18) between WT and CDKO mice. The inner retinal derived OPs were significantly reduced in (E) amplitude, but not different in (F) timing in CDKO mice (two-way ANOVA, cone OP amplitude; genotype P < 0.0001 and OP number P < 0.0001, individual results for WT versus CDKO mice for OP1, P > 0.05; OP2, P < 0.001; OP3, P < 0.001). * Indicates P < 0.05, a significant difference between WT (n = 18) and CDKO mice (n = 14).

In order to determine the origin of the alteration in the OPs in the CDKO mice, rod and cone ERGs were recorded from Ccl2^−/− (n = 9) and Cx3cr1^GFP/GFP mice (n = 19). The rod and cone ERG OP amplitudes for WT, Ccl2^−/−, Cx3cr1^GFP/GFP, and CDKO mice are presented in Figures 6A (rod OPs) and 6B (cone OPs). Neither the rod nor the cone OP amplitudes of Ccl2^−/− mice were significantly different from the responses of WT mice. In contrast, both the rod and cone OP responses of Cx3cr1^GFP/GFP mice were significantly reduced compared with WT and instead similar to those recorded from CDKO mice (Fig. 6A; rod OP amplitude, two-way ANOVA, for genotype P < 0.0001 and OP number P < 0.0001, Individual results for WT versus Cx3cr1^GFP/GFP mice for OP2, P < 0.01; OP3, P < 0.001; OP4, P > 0.05; Figure 6B; cone OP amplitude two-way ANOVA, for genotype P < 0.0001 and OP number P < 0.0001, individual results for WT versus Cx3cr1^GFP/GFP mice for OP1, P < 0.05; OP2, P < 0.001; OP3, P < 0.001). This suggests that the inner retinal dysfunction observed in the CDKO mice is unlikely to have occurred due to the combined loss of both Ccl2 and Cx3cr1, but rather that it is due primarily to alterations in the ERG responses of the Cx3cr1^GFP/GFP-founder line. 

Figure 6. 

 

Loss of the rod and cone ERG OPs in the CDKO mice is due to the Cx3cr1^GFP/GFP-founder line. Rod and cone ERGs were recorded from Ccl2^−/−- and Cx3cr1^GFP/GFP mice in response to a 2.1 log cd-s/m² intensity twin flash. (A) The rod ERG OP amplitudes for OP2, OP3, and OP4 from WT (n = 18), Ccl2^−/− (n = 9), Cx3cr1^GFP/GFP (n = 19), and CDKO mice (n = 14) are presented. As was observed for the CDKO mice, the rod OP responses of Cx3cr1^GFP/GFP mice were significantly reduced compared with WT mice (two-way ANOVA, for genotype P < 0.0001 and OP number P < 0.0001, individual results for WT versus Cx3cr1^GFP/GFP mice for OP2, P < 0.01; OP3, P < 0.001; OP4, P > 0.05). (B) The cone ERG OP amplitudes for OP1, OP2, and OP3 from WT, Ccl2^−/−, Cx3cr1^GFP/GFP, and CDKO mice are presented. The cone OP responses of Cx3cr1^GFP/GFP mice were significantly reduced compared with WT mice (two-way ANOVA, for genotype P < 0.0001 and OP number P < 0.0001, individual results for WT versus Cx3cr1^GFP/GFP mice for OP1, P < 0.05; OP2, P < 0.001; OP3, P < 0.001). * Indicates P < 0.05 and a significant difference from WT.

Morphology Changes of Rod and Cone Bipolar Cells Do Not Underlie ERG Phenotype in CDKO Mice

The morphology of the rod and cone bipolar cells was assessed using immunohistochemistry in order to determine if structural alterations in these cell classes were contributing to the functional changes observed in the CDKO mice (Fig. 7). Rod bipolar cells were labeled with an antibody against Protein Kinase Cα (PKCα, pseudocolored green), the terminals of the photoreceptors and rod and cone bipolar cells were labeled with an antibody against vesicular glutamate transporter 1 (VGLUT1, red) and the cell nuclei were labeled with DAPI (blue). In both the WT (Fig. 7A) and CDKO retinae (Fig. 7B), rod bipolar cells extended dendrites into the outer plexiform layer (OPL), in close association with VGLUT1-positive photoreceptor terminals. The terminals of the rod bipolar cells of both genotypes extended down into the inner most sublamina of the IPL and overall rod bipolar cell morphology was qualitatively similar between the two. Similarly the cone bipolar cells, assessed using an antibody against ZNP-1, which labels an ON- and OFF-cone bipolar cell subtypes (type 2 and 6 ²⁸ ), were found to project terminals to the correct sublamina of the IPL and the morphology of these cells was comparable between WT and CDKO mice (Figs. 7C, 7D, respectively). This suggests that obvious changes in bipolar cell morphology do not underlie the reduction in the ERG OPs recorded in the CDKO mice. 

Figure 7. 

 

Immunohistochemistry for inner retinal neurons in WT and CDKO mouse retina. Eyecups from 9-month-old animals (n = 6 each) were fixed, cut at 14 μm on cryostat, and labeled with antibodies against inner retinal neurons. Rod bipolar cells were labeled with an antibody against PKCα (pseudocolored green); photoreceptor and bipolar cell terminals were labeled with an antibody against VGLUT1 (red); and cell nuclei were labeled with DAPI (blue) in (A) WT and (B) CDKO retinae. Cone bipolar cells (type 2 and 6) were labeled with an antibody against ZNP-1α (pseudocolored green) and photoreceptor and bipolar cell terminals were labeled with an antibody against VGLUT1 (red) in (C) WT and (D) CDKO retinae. Rod and cone bipolar cell morphology was qualitatively similar in WT and CDKO retinae. Amacrine and ganglion cells were labeled with an antibody against Calretinin (pseudocolored green) in (E) WT and (F) CDKO retinae. The calretinin-positive cell processes were frequently aberrant in CDKO retinae (F). Scale bar, 20 μm.

Morphology of Amacrine Cells Is Aberrant in CDKO Mice

The morphology of subpopulations of amacrine and ganglion cells was assessed using an antibody against the calcium-binding protein, calretinin (pseudocolored green). In the WT retinae, calretinin-positive amacrine cells in the INL and displaced amacrine cells and ganglion cells within the ganglion cell layer (GCL) were regularly spaced (Fig. 7E). Moreover, these cells projected processes into three distinct plexuses within the IPL. In contrast, within the CDKO retinae, although there was no significant differences in the number of calretinin-positive cells per mm of retinae from WT (WT, n = 4 versus CDKO, n = 4; Mann-Whitney nonparametric test, P = 0.114), there were many regions in which the morphology of the calretinin-positive processes were aberrant (Fig. 7F). Specifically, the processes of the calretinin-positive cells of CDKO mice often did not form into distinct laminae within the IPL and instead appeared disorganized. 

Morphology of the Inner Retinal Blood Vessels of CDKO Mice Is Similar to WT Mice, but Microglial Morphology Is Subtly Altered

Microglia play an important role in surveillance of the vasculature for bloodborne pathogens and in responding to chemokines as part of the innate immune response. 

To determine if lack of Ccl2 and Cx3cr1 had an effect on either the inner retinal vasculature or microglial morphology, the retinae of WT and CDKO mice were fluorescently labeled using IB4-lectin-fluoroscein (green, blood vessels) and rabbit anti-IbA-1 (red, microglia; Figs. 8A, 8B). The inner retinal vasculature of 9-month-old WT mice was found to be similar to that previously described for the mouse, ²¹ with vessels located in nerve fiber layer/GCL (superficial plexus), the border of the INL and IPL (intermediate plexus), and in the OPL (deep plexus; Fig. 8A). In the CDKO mice, there were no qualitative differences in the appearance of any of the inner retinal vessels when compared with WT retinae (Fig. 8B). 

Figure 8. 

 

Immunohistochemistry for blood vessels, micro- and macroglia in WT and CDKO mouse retina. (A, B) Eyecups from 9-month-old animals (n = 6 each) were fixed, cut at 14 μm on cryostat and labeled with antibodies against inner retinal vessels and glia. Blood vessels were labeled with IB4-lectin-fluoroscein (green), microglia were labeled with an antibody against IbA1 (red) and cell nuclei were labeled with DAPI in (A) WT and (B) CDKO retinae. In CDKO mice, all IbA1-labelled microglia were also EGFP positive. Both the blood vessel profiles and intraretinal location of microglia were qualitatively similar between WT and CDKO retinae. (C–J) Flat mounted, whole retinae from WT and CDKO mice (n = 4 each) were labeled for microglia using IbA1 and imaged at the plane of the IPL. (C–E) Tile scans of microglia (pseudocolored green) across the entire retina at ×20 were collected for analysis for cell counts from WT ([C], entire tile scan; [D], magnification of an area from [C]) and CDKO mice ([E], entire tile scan; [F], magnification of an area from [E]). (G–J) For microglial morphology, Z-stack images were captured (×40 oil). Representative microglia from WT ([G], original; [H], computer morphological analysis) and CDKO retinae ([I], original; [J], computer morphological analysis) are presented. (K, L) Müller cell bodies were labeled with an antibody against glutamine synthetase (green); and astrocytes and gliotic Müller cells were labeled with an antibody against GFAP (red) in (K) WT and (L) CDKO retinae. The number of GS-positive Müller cells that also expressed GFAP was higher in CDKO mice than WT mice, indicative of Müller cell gliosis and retinal stress in the CDKO retinae. Scale bars, 20 μm.

Retinal microglia in the WT retinae were found in the OPL, IPL, and GCL/nerve fiber layer, often in close association with blood vessels (Fig. 8A), as has been previously described. ²¹ Occasionally, aberrant microglial processes were found extending into the ONL in WT retina, an occurrence that we have noted increases with age (data not shown). In the CDKO mice, because EGFP is expressed in the cells that would usually express the Cx3cr1 receptor, which in the retina is only microglia, ²⁹ all microglial cells were positive for EGFP and also labeled with IbA1 (Fig. 8B). However, despite the lack of Cx3cr1 on microglia in the CDKO mice, the intraretinal location of these cells was qualitatively similar to that observed in age-matched WT mice. 

To quantitatively assess for changes in microglia number and morphology in the CDKO mice, whole retinae (WT and CDKO, n = 4 each) were labeled with IbA1 (pseudocolored green) and imaged in flat mount at the plane of the IPL. Previous studies have shown microglia are recruited, increasing in number in neural trauma and retinal pathology. ^21,30 Tile scans of the entire retina were collected for counting of microglial number and representative images of a WT (Figs. 8C, 8D) and a CDKO retina (Figs. 8E, 8F) are presented. There was no significant change in the number of microglia in the CDKO retinae compared with the WT retinae (Table 3; WT, 112 ± 9 versus CDKO, 128 ± 8 microglia/area of retina mm²; Mann-Whitney nonparametric test, P = 0.20). 

Table 3. 

 

Microglia Counts and Morphological Measures in 9-Month-Old WT and CDKO Mice

Genotype	Number of Microglia, per mm2	Soma Area, μm2	Number of Main Processes	Number of Branches	Total Outgrowth of Processes, μm
WT	112 ± 9	53 ± 6	20 ± 3	267 ± 25	839 ± 60
CDKO	128 ± 8	49 ± 5	20 ± 3	218 ± 25	667 ± 60
P value	P = 0.20	P = 0.68	P = 0.98	P = 0.34	P = 0.03*

 Retinae from WT and CDKO mice were labeled for microglia using an antibody against IbA1 and whole mount preparations at the plane of the IPL were imaged. For microglia counts (number per retinal area mm²), tile scans of the entire retina at ×20 were collected for analysis. For microglial morphology, at least 3 Z-stack images containing 4 to 6 microglia per image were captured at ×40 for each retina. For a given retina (n = 1 animal), the morphological properties of at least 15 microglia from that retina were averaged. Morphological parameters were assessed for n = 4 animals in each group and data are presented as the mean ± SEM. Microglia from CDKO mice had significantly smaller outgrowth of processes compared with WT mice (n = 4 each genotype; Mann-Whitney nonparametric tests were used to generate P values presented).

*  Indicates statistical significance.

Microglial morphology was also assessed as increases in soma area and/or decreases in arbor spread suggest microglial activation consistent with response to pathology. ^21,30 In Figure 8, micrographs of IbA1-immunolabeled microglia in the WT (G, original; H, computer analysis) and CDKO retina (I, original; J, computer analysis) are presented. For a given retina (n = 1 animal), the morphological properties of at least 15 microglia from that retina were assessed and averaged. In general, microglia from CDKO retinae had a similar soma area to WT (Table 3; soma area; WT, 53 ± 6 μm² versus CDKO, 49 ± 5 μm², Mann-Whitney nonparametric test, P = 0.68). In addition, there were no changes in the number of main processes or number of branches from the main processes in the CDKO retina (Table 3). However, the outgrowth of microglial processes was significantly reduced in CDKO retinae compared with WT retinae, suggesting a reduced arbor spread (Table 3; total outgrowth processes; WT, 839 ± 60 μm versus CDKO, 667 ± 60 μm, Mann-Whitney nonparametric test, P = 0.03). 

Müller Cells of the CDKO Mice Are Gliotic

The morphology of the retinal macroglia was also investigated in the WT and CDKO mice (Figs. 8K, 8L, respectively). The Müller cells, which are radial glial that span the transverse retina, were labeled with an antibody against glutamine synthetase (GS, green). In addition, an antibody against GFAP was used to label the astrocytes, which run along the nerve fiber layer, and to label the cell processes of gliotic Müller cells (red). The morphology of the GS-positive Müller cells was similar in WT and CDKO mice. In addition, the labeling of GFAP-positive astrocytes was similar in WT and CDKO mice, although in both mouse strains, occasional breaks were observed in astrocyte GFAP labeling, likely due to astrocyte loss, which occurs with age. ³¹ However, the number of GS-positive Müller cells that also expressed GFAP was higher in CDKO mice than WT-mice. This suggests that Müller cells in the CDKO retinae are gliotic, which is indicative of global retinal stress in the CDKO mice. 

Discussion

Mice lacking the chemokine, Ccl2, and the receptor for fractalkine, Cx3cr1, have been suggested to be an accelerated model of AMD, exhibiting white spots in the fundus, and photoreceptor and RPE cell loss by 3 months of age. ^13–16 However, other studies suggest that this phenotype may be due to a background rd8 mutation. ¹⁷ Our work in 9-month-old mice lacking both Ccl2 and Cx3cr1 that do not have the rd8 mutation suggests that while the CDKO mice do display signs of retinal stress such as Müller cell gliosis, they do not exhibit the hallmarks of an accelerated AMD model. In 9-month-old CDKO mice, there were no changes in fundus appearance, no overt changes in Bruch's membrane, and no loss of RPE or photoreceptor numbers or changes in photoreceptor function when compared with WT mice. In contrast, CDKO mice were found to have inner retinal dysfunction, possibly due to alterations in amacrine cell signaling. 

Mice Lacking Ccl2 and Cx3cr1 Are Not an Accelerated Model of AMD

AMD is a major cause of visual impairment affecting older adults ¹ and is characterized by drusen accumulation, RPE atrophy, photoreceptor death, and in some cases, choroidal neovascularization. Genetic variations in the innate immune system such as CFH ² and CX3CR1⁴ have been found to predispose individuals to the development of AMD. Although mice do not have a macula, mice lacking chemokine pathways involved in the innate immune system, Ccl2 and Cx3cr1, have been proposed to be a model of human AMD. Ccl2- and Cx3cr1-deficient mice display fundus lesions, RPE atrophy, drusen, photoreceptor death, and in some instances, choroidal neovascularization. ^13–16 However, other studies have suggested that this AMD phenotype could be occurring primarily as a result of the rd8 mutation in the Crb1 gene ²⁰ present on the C57blk6 background strain. ^17,18 Our findings are consistent with the work of this later study. Our results suggest that 9-month-old CDKO mice on a C57blk6J strain (Crb1^+/+ ) have no lesions in the fundus, while mice harboring the Crb1^rd8/rd8 mutation do. In mice with the rd8 mutation, the photoreceptors were found to degenerate and in some places, the RPE was found to atrophy, consistent with the previous description of this phenotype. ²⁰ This loss of RPE and resulting pigment loss would likely lead to the appearance of white spots upon inspection of the retinal fundus. However, another possibility is that these fundus lesions are occurring due to bloated subretinal microglia containing lipofuscin as has been described previously. ^10,11  

In contrast to mice harboring the rd8 mutation, CDKO mice did not display this phenotype. Quantitative analysis of the retinal layers showed that the photoreceptor inner and outer segments, and the number of photoreceptor nuclei were similar in CDKO mice and age-matched, WT mice. In addition, there were no qualitative differences between the RPE and Bruch's membrane in the WT and CDKO mice. In particular, there was no RPE atrophy, excessive thickening of Bruch's membrane or choroidal neovascularization in the CDKO mice. There were also no significant functional changes that would indicate photoreceptor dysfunction in the CDKO mice. There was neither a loss of the rod ERG PIII amplitude, which would indicate a loss in the number and/or length of the rod outer segments, nor were there changes in the PIII sensitivity, which would indicate alterations to the phototransduction cascade. ²⁵  

These findings suggest that the CDKO mice that we have generated are not a good accelerated model of AMD. One reason for this difference in phenotype from previous reports ^13–16 may be due to differences in the origin of our Ccl2^−/− and Cx3cr1^EGFP/EGFP mice from that of other studies. In the case of the Cx3cr1 mutation, our mice are a knock-in/knockout of EGFP/Cx3cr1, and thus, they express green fluorescent protein instead of the chemokine receptor. These mice are not from the same original line as that previously published for Ccl2^−/−/Cx3cr1^−/−-mice with an AMD-like phenotype. ^13–17 Thus, it is possible that when crossed with Ccl2^−/− mice, the combination of Cx3cr1^EGFP/EGFP instead of the Cx3cr1^−/−, may result in a different, more WT-like retinal phenotype than what has been described previously. More work is required to determine if this is the case. In addition, more work beyond the scope of this study would be required to determine if our CDKO mice develop AMD-like pathology in senescence in a manner similar to that described for aged Ccl2^−/− mice ⁶ and aged Cx3cr1^−/− mice. ⁹ However, given the recent findings that the original Ccl2^−/−/Cx3cr1^−/−-mice ¹⁵ do not have fundus lesions if they do not harbor the rd8 mutation, ¹⁷ our work on Ccl2^−/−/Cx3cr1^GFP/GFP mice provides further support for the idea that knockout of both Ccl2 and Cx3cr1 does not induce an AMD phenotype. Specifically, the CDKO mice examined in the present study do not display signs of AMD pathology at 9 months of age and, thus, cannot be used as an accelerated model of AMD. 

Loss of Ccl2/Cx3cr1 Causes Inner Retinal Dysfunction, Partial Microglial Activation, and Müller Cell Gliosis

Although the CDKO mice did not exhibit an AMD phenotype, there were other alterations in retinal function and morphology in these mice. Using the ERG to independently assess the function of the rod and cone pathways, it was found that although the overall amplitude of the ERG PIII and PII responses was similar between WT and CDKO mice, the OPs were significantly reduced in the CDKO mice. While the neural source of the OPs is not well established, they are largely thought to be driven by amacrine cell responses. ²⁶ This suggests that CDKO mice have inner retinal dysfunction, likely derived from alterations to bipolar or amacrine cell signaling. ²⁶  

This inner retinal dysfunction hailed primarily from the Cx3cr1^GFP/GFP-founder mice, suggesting that loss of Cx3cr1 signaling, rather than Ccl2 or the combined effects of loss of Ccl2 and Cx3cr1, underlies this phenomenon. It is possible that this effect could also be independent of the chemokine pathways altogether and instead be due to a different gene mutation that has not yet been characterized but that is transmitted from the Cx3cr1^GFP/GFP-founder line. However, given the Cx3cr1^GFP/GFP-founder mice had been backcrossed onto the WT (C57blk6J) background, this is unlikely. Thus, in the CDKO and Cx3cr1^GFP/GFP mice, such ERG OP dysfunction would more likely result from alterations to the Cx3cl1/Cx3cr1 chemokine pathway modulating the neurons themselves, specifically the bipolar and/or amacrine cells. While the gross morphology of the rod and cone bipolar cells appeared to be intact and similar between CDKO and WT mice, when the amacrine cells were investigated using the marker calretinin, it was found that their morphology was aberrant. Specifically, the calretinin-positive amacrine cell arbors—which usually laminate into the three synaptic layers within the IPL—were frequently disorganized in the CDKO retinae. 

These subtle alterations to synaptic connections within the inner retina of the CDKO might highlight a role for microglia. Despite the lack of two factors important for normal microglial function (Ccl2 and Cx3cr1) within the retina of CDKO mice, the location, number, and somal size of microglia within the 9-month-old retinae of CDKO mice was generally comparable with WT. However, in the CDKO retinae, the outgrowth of microglial processes was significantly reduced. This is a subtle morphological change characteristic of microglia that are partially activated. ³² This may indicate a change in immunosurveillance and retinal pathology consistent with retinal stress. ³² However, recent work has shown that microglia also play an important role in the development and maintenance of neuronal synapses, pruning synaptic buttons to aid in the removal and/or strengthening of synaptic connections. ^33,34 Thus, microglial activity in the CDKO mice may be altered, thereby modulating the synaptic connections between bipolar cells and amacrine cells. In particular, the Cx3cr1 is a receptor expressed only on microglia in the retina, and it is involved in coordinating the response of the microglia to the chemokine, Cx3cl1 (fractalkine/neurotactin). Recent work in the CNS indicates that synaptic pruning is altered during development in the absence of the Cx3cr1. ³³ Indeed, deficient synaptic pruning in Cx3cr1^GFP/GFP mice results in an excess of immature synapses and persistence of the electrophysiological hallmarks of immature brain circuitry. ³³ Further research is required to determine if microglial activation via the Cx3cr1 plays a similar role in synapse pruning in retinal development and in particular in the maturation and maintenance of the inner retinal circuitry. 

Another mechanism that may underlie the loss of the ERG OPs is poor vascular perfusion. Inner retinal hypoxia due to poor vascular perfusion has been shown to induce reductions in the ERG OPs in rodents. Previous studies have shown that loss of the ERG OPs occurs in the hypoxic retina in oxygen-induced retinopathy, ^35,36 diabetic retinopathy, ^37,38 and ischemia. ³⁹ In the case of the CDKO mice, the inner retinal vasculature is intact with the superficial and deep plexuses well developed and apparently similar to that of WT mice. Although we cannot rule out hypoxia as the underlying reason for the loss of the OPs in the CDKO mice, our findings suggest that the inner retinal vasculature has developed normally in these mice. 

In addition to the alterations in inner retinal neurons and subtle alteration in microglial morphology in the CDKO mice, the Müller cells were found to express GFAP. Expression of GFAP by Müller cell processes is a well accepted indicator of gliosis and occurs as a result of trauma, ^40,41 neural degeneration, ⁴² and hypoxia. ²¹ GFAP is an intermediate filament and its expression is important for resistance to mechanical stress in disease. ⁴³ In CDKO mice, upregulation of GFAP may indicate a reduction in Müller cell ability to maintain retinal fluid homeostasis ⁴⁴ or altered chemical metabolism within neurons. ⁴⁵ However, Müller cell gliosis may also be a protective mechanism involved in preserving the morphology and function of the retina. ⁴⁶  

Conclusions

The CDKO mice investigated in the current study were not found to be a useful accelerated model of AMD. They did not have drusen-like lesions in the retinal fundus, a thickened Bruch's membrane, or RPE atrophy. In addition, function of the rod and cone pathways was generally intact as there was no change in the amplitude of the rod photoreceptor response or of the rod or cone post-photoreceptor b-wave in the CDKO mice. However, there were reductions in the rod and cone ERG oscillatory potentials, likely indicating a change in bipolar or amacrine cell function. Amacrine cell morphology was found to be aberrant, microglia were partially activated, and Müller cells were gliotic, indicative of retinal stress in the CDKO mice. These results indicate that CDKO mice have an unusual inner retinal phenotype, which might relate to altered microglial synaptic pruning during development. 

Acknowledgments

The authors thank Jennifer Wilkinson-Berka and Paul McMenamin for their donation of the mice strains used in the present study. We also thank Andrea Rassell for technical assistance with this project. 

Footnotes

 Supported by the NHMRC project Grants 566814 and 1021918 (ELF); by the American Health Assistance Foundation; Macular Degeneration Research grant (ELF); and a Victoria's Science Agenda grant (ELF).

Footnotes

 Disclosure: K.A. Vessey, None; U. Greferath, None; A.I. Jobling, None; J.A. Phipps, None; T. Ho, None; M. Waugh, None; E.L. Fletcher, None