Differential Gene Expression in Eyecup and Retina of a Mouse Model of Stargardt-like Macular Dystrophy (STGD3)

Abstract

Purpose.: To investigate differentially expressed genes in eyecup and retina of the ELOVL4 transgenic mouse, a model of Stargardt-like macular dystrophy (STGD3).

Methods.: We examined gene and protein expression in known pathways relevant to retinal degeneration using PCR arrays, Western blotting, and immunohistochemistry. Investigations were performed on ELOVL4 transgenic mice at 9 months, when 50% of rod (but no cone) photoreceptors had degenerated. Age-matched wild-type littermates served as controls.

Results.: Significant expression level changes were found in only 17 of the 252 genes examined. Nine were upregulated (Fgf2, Fgfr1, Ntf5, Cbln1, Ngfr, Ntrk1, Trp53, Tlr6, and Herpud1), and eight were downregulated (Ccl22, Ccr3, Il18rap, Nf1, Ccl11, Atf6β, Rpn1, and Serp1). Overexpression of FGF2 was detected at 1 month, before rod loss onset, and was maintained at high levels until cone loss (18 months). By 9 months, FGF2 overexpression was seen in photoreceptor cell bodies. Increased glial fibrillary acidic protein (GFAP) expression due to glial cell reactivity followed the same time course. Levels of NGFR/p75NTR remained invariant. Although present in rod outer segments at 1 month, the macrophage chemoattracting chemokine CCL22 became undetectable by 9 months, a likely consequence of progressive rod outer segment truncation.

Conclusions.: At a mid-degeneration stage, major changes in gene expression in the ELOVL4 transgenic mouse retina included upregulation of Fgf2 and Fgfr1 and downregulation of Ccl22. Modulation of FGF2 occurred very early, concomitant with an increase in GFAP expression. Future studies will address which factors upstream of Fgf2 could provide potential therapeutic targets to slow photoreceptor degeneration in STGD3.

Stargardt-like macular dystrophy (STGD3, MIM 600110) is an autosomal-dominant juvenile form of progressive atrophic macular degeneration that results in lipofuscin accumulation, atrophy of the cone-dominated central retina and of the retinal pigment epithelium (RPE), and, finally, in visual loss. ^{1 –3}  

Mutations causing STGD3 have been identified in exon 6 of the ELOVL4 gene (AF277094, MIM 605512), which encodes a protein involved in fatty acid (FA) elongation. All these mutations introduce a premature stop codon, resulting in the loss of a C-terminal endoplasmic reticulum (ER) retention signal at the protein level, ⁴ thus, leading to cellular mislocalization of the ELOVL4 mutant protein. ^5,6 Although the mechanisms responsible for abnormal intracellular trafficking in the presence of mutant protein are not fully understood, ^6,7 it is thought that the mutant protein inhibits ER retention of the wild-type (WT) protein and sequesters the latter into stable aggresome-like complexes. ^5,6,8,9 In addition, recent in vitro experiments ¹⁰ have shown that mutant ELOVL4 also forms heteroligomers with other ELOVLs and interacts with 3-ketoacyl-CoA and trans-2,3-enoyl-CoA reductases that are part of the elongase complex. 

The production of abnormally folded proteins induced by ELOVL4 mutations has two main consequences. First, saturation of the proteasome with misfolded protein complexes leads to the activation of an unfolded protein response (UPR). Activation of UPR was demonstrated in hEK293 and COS cell lines transfected with mutant ELOVL4 by upregulation of ER resident chaperones, immunoglobulin heavy chain binding protein (GRP78/BiP), and C/EBP-homologous protein. ⁶ After UPR, the ER usually returns to its normal physiologic state. However, anomalous accumulation of misfolded proteins resistant to proteasomal degradation through ER-associated degradation can completely disrupt ER and cellular function and activate apoptotic signaling pathways. ¹¹ Second, sequestration of WT protein in a non-ER location, oligomerization with other ELOVLs, and disruption of the normal elongase complex can impair the elongation process of many FAs and result in altered cellular lipid content. Although its exact role remains to be clarified, additional genetic manipulation of Elovl4 in mice indicates that this FA elongase is involved in the synthesis of very long chain polyunsaturated fatty acids (VLC-PUFAs), primarily omega-3 FAs with 28 or more carbon chains. ^4,12,13 As with the ELOVL4 protein, these VLC-PUFAs are found at high levels in both rod and cone photoreceptors, where they are thought to take part in important structural functions such as stabilization of the curved membrane of the disks in the outer segments. ⁴ Over time, abnormal FA composition will undoubtedly cause structural disorganization of the photoreceptor outer segments, dysfunction of the transduction cascade, inflammatory reactions, impaired phagocytosis by RPE cells, and cell death. ⁹  

As evidenced with both anatomic and electroretinographic approaches, transgenic mice expressing a mutated ELOVL4 gene under the control of a photoreceptor-specific promoter ¹⁴ experience retinal abnormalities similar to those found in the human STGD3 pathology as well as in the dry form of age-related macular degeneration (AMD). ¹⁵ Degeneration in these mice proceeds in three major phases: first, undigested phagosomes and lipofuscin accumulate in the RPE and subtle homeostatic changes affect the outer retina (inducing reactivity of Müller cells); second, massive death of rods, ultimately followed by cones, occurs while second-order neurons undergo negative remodeling; and, finally, the remnant nonfunctioning inner retina is invaded by blood vessels, RPE cells, and macrophages, and is progressively eradicated. ^14,16 This general scheme is common to most photoreceptor-induced retinal degenerations, irrespective of the primary defect, according to an extensive literature review. ¹⁷ As a result, understanding the early changes in signaling pathways appears critical to characterize pathologic processes and to develop therapeutic strategies (besides gene therapy) able to prevent or delay the onset of subsequent retinal degeneration. 

The purpose of this study was therefore to examine changes in gene expression in the retina of the ELOVL4 transgenic mouse at 9 months, an age at which approximately 50% of rod photoreceptors are lost and cone photoreceptors are still unaffected. ¹⁶ By choosing this time point, we aimed to elucidate genes downstream to the primary defect that could serve as specific indicators of the initial degenerative process. Alterations in lipid metabolism have been reported in AMD. ¹⁸ Such changes alongside analysis of pertinent genes are part of an extensive ongoing study in which detailed quantification of fatty acid profiles will be analyzed as a function of retinal degeneration (for preliminary results, see Suh M et al., Applied Physiology, Nutrition, and Metabolism. 2011;36:Abstracts of the Canadian Nutrition Society's 2nd Annual Scientific Meeting, 485). 

Material and Methods

Animals

The present study was performed on heterozygous ELOVL4/TG1-2 mice (TG) and WT littermates from a colony at the University of Alberta. All offspring were PCR genotyped as previously reported. ¹⁶ Animals were maintained on a 12:12 light–dark cycle, temperature 21°C, relative humidity ≅ 50% and supplied without restriction with water and rodent diet (Laboratory Rodent Diet #5001 LabDiet; Nutrition International, Richmond, IN). Experiments were carried out in accordance with guidelines of the Institutional Animal Care and Use Committee (University of Alberta) and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Assessing the Retinal Phenotype

To clarify the degenerative events at early time points, photoreceptor rows were counted in retina cross-sections of TG and WT mice at 1, 2, and 3 months (TG, n = 5; WT, n = 4 at each time point). 

After a lethal dose of anesthesia (Euthanyl; Bimeda-MTC Animal Health Inc., Cambridge, Ontario, Canada), eyes were enucleated, corneas punctured, and eyes immersed for 30 minutes at room temperature in 4% paraformaldehyde (pH 7.4). After removal of the lenses, the eyes were fixed for an additional 30 minutes and cryoprotected in graded sucrose concentrations at 4°C. After overnight incubation in 30% sucrose, the eyes were embedded in embedding resin (Shandon Cryomatrix; Anatomic Pathology, Pittsburgh, PA), flash frozen in liquid nitrogen, and stored at −80°C until sectioned. Cross-sections of 20 μm were cut parallel to the temporonasal axis through the optic nerve head, mounted on glass slides (Superfrost Plus; Thermo Fisher Scientific, Waltham, MA), coated with an antifade reagent (Prolong Gold, Cat. #P36931; Molecular Probes, Eugene, OR) containing 4′,6-diamidino-2-phenylindole (DAPI), and coverslipped. 

Photoreceptor rows were counted in a single plane from confocal images (106.15 × 106.15 μm) taken at three retinal locations: center (400 ± 200 μm from the optic disc), periphery (500 ± 200 μm from the ora serrata), and middle (equidistant between the center and periphery). The number of DAPI-stained rows was determined as follows: one number was obtained for each animal by averaging counts from two retina sections each from both eyes. Counts from all animals of a given age were then averaged and the SD calculated for both TG and WT, respectively. 

RNA Isolation

Eyes from TG and WT (all at 9 months) were dissected out and cut behind the ora serrata; cornea and lens were removed, and posterior eyecups were stored in 500 μL RNA stabilization reagent (RNAlater; Qiagen, Mississauga, ON, Canada) at 4°C overnight. The following day, tissue was removed from this solution, quick frozen on liquid nitrogen, and stored at −80°C until all eyes had been collected. Tissue was pooled from three to four animals each for TG or WT (to maximize the amount of RNA and to minimize bias due to biological variation) ¹⁹ in 0.75 mL commercial reagent (TRIzol; Invitrogen, Burlington, ON, Canada), disrupted using a homogenizer and total RNA was isolated following manufacturer's instructions. RNA was dissolved in 100 μL nuclease-free water, purified using an RNA stabilization and purification reagent (Qiagen RNeasy Mini kit protocol, Cat. #74104) including an on-column DNase digestion step, eluted in 30 μL RNase-free water, and a second elution step was performed using the first eluate. RNA aliquots were frozen on dry ice and stored at −80°C. Quality of RNA was assessed by measurement of absorbance at 230, 260, and 280 nm and visualization on a denaturing formaldehyde gel. A total of three replicates each were prepared for both TG and WT. 

Real-Time Quantitative PCR Arrays

RNA was sent to a commercial company (SABiosciences, Frederick, MD) for quality control, reverse transcription (with RT² First Strand Kit; Cat. #330401), and real-time PCR data analysis on three PCR arrays (RT² Profiler: Unfolded Protein Response [Cat. #PAMM-089], Neurotrophins and Receptors [Cat. #PAMM-031], and Inflammatory Response and Autoimmunity [Cat. #PAMM-077]). These arrays were selected (1) because mutant ELOVL4 upregulated markers of UPR in vitro ⁶ ; and (2) because they included genes differentially expressed in retinal degeneration models ²⁰ such as growth factors (Ngf, Bdnf, Ntf3, Ntf5, Cntf) and indicators of apoptosis (Bcl2, Bax, and Myc) for the Neurotrophins and Receptors array; several chemokines, interleukins, and Toll-like receptors (Ccl2, Il6, Tnf, Ccr2, Tlr3, Tlr4) for the Inflammatory Response and Autoimmunity array. Overall, a total of 252 genes were screened. 

Results were returned as relative gene expression level [2^∧(−ΔC^t )] with genes of interest (GOI) normalized to housekeeping genes (HKG) using the formula ΔC^t = C^t (GOI) − average C^t (HKG). Reverse transcription controls (RTCs) passed efficiency criteria {ΔC^t [average RTC − average PPC (Positive PCR control)] ≤ 5 } for test and control samples and there was no genomic DNA contamination present in samples (C^t [GDC] ≥ 35). Fold difference test sample (TG) over control sample (WT) was provided as well as statistical P values as determined by t-test, and fold up- or downregulation test sample over control sample (TG/WT). 

Western Blot Analysis

Freshly dissected eyecups (minimum of four each from TG and WT animals) or retinas (minimum of six each), at ages ranging from 1 to 18 months, were homogenized in PBS buffer. Lysates were prepared by addition of sample buffer (4% [w/v] sodium dodecyl sulfate, 20% [v/v] glycerol, 0.13 M Tris, pH 6.8) with protease inhibitor cocktail (Proteoblock, Cat. #R1321; Fermentas Molecular Biology Tools/Thermo Fisher Scientific). Protein levels were determined using a protein assay kit (Pierce BCA, Cat. #PI-23227; Thermo Fisher Scientific). After addition of 2% (v/v) 2-mercaptoethanol and 1% (v/v) saturated bromophenol blue, protein extracts were boiled for 5 minutes. Samples (30 μg total protein) were resolved by SDS-PAGE on 8–10% acrylamide gels. Proteins were transferred to polyvinylidene fluoride membranes, blocked for 1 hour with 5% nonfat milk diluted in TBS-T (20 mM Tris, 137 mM NaCl, 0.1% Tween-20, pH 7.6), incubated overnight with specified antibodies (Table 1) diluted in the blocking solution. α-Tubulin was used as a loading control. The following day membranes were washed 3 × 10 minutes in TBS-T and reacted for 1 hour with anti-mouse or anti-rabbit IgG, horseradish peroxidase–conjugated chemiluminescent substrate antibody (ECL Reagent Kit; GE Healthcare, NA931 and NA934, respectively; 1:5000 in the blocking solution). After a final extensive washing, protein bands were visualized using the chemiluminscent reagent (Perkin Elmer, NEL 103, on a Kodak Image Station 440). Net intensity of bands was calculated using imaging software (Kodak, v.4.0.3). 

Table 1.

 

Antibodies Used and Western Blot Results Obtained

Antibody	Host	Source, Catalog Number	Dilution	Molecular Mass
FGF-2/basic FGF, clone of bFM-2	Mouse	Millipore, 05–118	1:1000	17.5 kDa
CCL22/MDC/ABCD-1	Rabbit	Abcam, ab53002	1:5000	No bands
GFAP (SMI22)	Mouse	Covance, SMI-22R	1:1000	50 kDa
p75NTR/LNGFR	Rabbit	Millipore, 07–476	1:5000	75 kDa
NF-κB p105/p50 (phospho S927)	Rabbit	Abcam, ab60936	1:500	105 kDa
Rhodopsin, clone 4D2	Mouse	Millipore, MABN15	1:500	39 kDa
OPN1SW (N-20)/s-opsin	Goat	Santa Cruz Biotech, sc-14363	1:500	40 kDa
α-Tubulin (TU-02)	Mouse	Santa Cruz Biotech, sc-8035	1:500	54 kDa
TRK (B-3)/NTRK1/TRK A	Mouse	Santa Cruz Biotech, sc-7268	1:200	Many bands
FGFR1 (phospho Y654)	Rabbit	Abcam, ab59194	1:500	Many bands
CKR-3 (H-52)/CCR3	Rabbit	Santa Cruz Biotech, sc-7897	1:200	Many bands

Immunohistochemistry

TG and WT were studied immunohistochemically at 9 months (with additional investigations at 1, 3, 6, and 18 months for specific products). 

Retina cross-sections were prepared as described earlier. After hydration in PBS, the sections were blocked for 1 hour in PBS + 0.3% Triton X-100 + 10% serum (same species as secondary antibody), and reacted overnight in a humid container with the following primary antibodies: mouse anti-FGF-2 (1:250), mouse anti-rhodopsin (1:500), goat anti-OPN1SW (1:200), rabbit anti-p75NTR (1:1000), rabbit anti-CCL22 (1:250), and rabbit anti–NF-κB (1:250). Hosts and sources of these antibodies are given in Table 1. Antibodies directed against FGF2, rhodopsin, glial fibrillary acidic protein (GFAP), and s-opsin have been shown previously to detect specific cell types in retina sections. ^16,21,22 The specificity of the remaining antibodies was assessed by Western blotting. The following day, sections were washed extensively in PBS, reacted for 1 hour with species-appropriate secondary antibodies conjugated to fluorescent dyes (1:1000, Alexa; Molecular Probes, Eugene, OR), and washed extensively in PBS. For FGF2, a fluorescein immunodetection kit (FMK-2201, Vector M.O.M.; Vector Laboratories) was used to minimize any background staining issues associated with using a mouse monoclonal antibody on mouse tissue. Finally, slides were coated with an antifade reagent (Prolong Gold) with DAPI and coverslipped. 

Additionally, the avidin-biotin-peroxidase method was used to better visualize the weak expression of CCL22. After hydration in PBS, the sections were blocked for 2 hours in PBS + 0.1% Triton X-100 + 5% goat serum, and reacted overnight in a humid container with anti-CCL22 (same as above; anti-macrophage-derived chemokine [MDC], 1:250). After blocking for 1 hour with the previous medium, sections were successively (1) reacted for 2 hours with rabbit biotinylated secondary antibody (1:200), (2) incubated for 2 hours with previously prepared reagent (ABC reagent, as per kit instructions; PK-6101 VECTASTAIN Elite ABC kit; Vector Laboratories), and (3) exposed to the peroxidase DAB-Ni²⁺ substrate solution for approximately 3 minutes. Finally, slides were rinsed extensively in PBS and mounted with aqueous mounting media (Fluoromount, F4680; Sigma). 

All antibodies were diluted in a 1:10 solution of the blocking medium; all immunoreactions were separated by extensive washing in PBS; and all processes were performed at room temperature. 

Images from fluorescent material were captured from the center of the retina (400 ± 200 μm from the optic disc) using either a confocal microscope (Zeiss LSM510 or LSM710, with a Plan-Neofluar ×40/1.3 oil objective; see Figs. 1, 3, 4) or a fluorescent microscope (Leica DMRE6000B, using a ×40/1.25 oil objective; see Figs. 5 6–7). Images were projections of z-stacks of 6 to 10 slices of 1 μm. Brightness and contrast levels were adjusted if necessary (Adobe Photoshop CS2 software version 9.0.2; Adobe, San Jose, CA). Images from the treated material (ABC-DAB) were captured on a brightfield microscope (Leica DMRE6000B, using a ×40/1.25 oil objective). 

TUNEL

TUNEL was performed on 20-μm cross-sections of retinas using a commercial detection kit (Roche In Situ Cell Death Detection Kit, TMR red; Roche 12 156 792 910) following the manufacturer's instructions for cryopreserved tissue. 

Results

Photoreceptor Counts

Counts performed at 3 months confirmed earlier data reporting approximately 30% photoreceptor loss in TG compared with WT at this time point. ¹⁶ Significant loss (36%) was observed in the central retina (loss of four rows). This loss was less pronounced in the middle retina (three rows) and even less in the peripheral retina (two rows). At 2 months (P60), photoreceptor loss was just beginning and again demonstrates a central-to-peripheral gradient (Fig. 1). At 1 month of age, both TG and WT retinas had the same number of photoreceptor rows at all eccentricities. We therefore conclude that, although driven by the interphotoreceptor retinoid-binding protein promoter, which is active as early as E11 in mouse, ²³ the ELOVL4 transgene has no detectable effect on photoreceptor survival during development up to P60 in the present transgenic line. 

Figure 1.

 

Onset of photoreceptor loss in ELOVL4 TG retina. (A) Photoreceptor loss begins at 2 months with a central to peripheral gradient. (B) Examples from central retina of same initial number of photoreceptors in TG at 1 month compared with WT and loss in TG at 3 months (blue, DAPI-stained cell bodies). Scale bar, 20 μm.

Gene Expression Levels

Scatterplots of expression level [2^∧(−ΔC^t )] of each gene in TG versus WT are provided in Figure 2. These plots illustrate that expression levels of most genes are unchanged, with greater variation apparent in genes expressed at very low levels. In fact, of the 252 genes examined in the three PCR arrays, only 6 genes were found to be differentially expressed with >2-fold change and values of P < 0.05 (Table 2). Three genes in the Neurotrophins and Receptors array were found to be upregulated and three genes in the Inflammatory Response and Autoimmunity array were found to be downregulated with respect to WT levels. The three upregulated genes were: Fibroblast growth factor 2 (Fgf2; basic Fibroblast growth factor), Fibroblast growth factor receptor 1 (Fgfr1), and Neurotrophin 5 (Ntf5). The three downregulated genes were Chemokine (C–C motif) ligand 22 (Ccl22), Chemokine (C–C motif) receptor 3 (Ccr3), and Interleukin 18 receptor accessory protein (Il18rap). None of the genes in the UPR array fell within these threshold criteria. 

Figure 2.

 

Scatterplots of expression level [2^∧(−ΔC^t )] of each gene in TG versus WT (provided by SABioscience). The black line represents fold changes [2^∧(−ΔΔC^t )] of 1 and the pink lines indicate fourfold change in gene expression threshold.

Table 2.

 

RT-PCR Array Results

Description	Gene Symbol	GenBank Accession Number	Fold Change	P Value
Fibroblast growth factor 2	Fgf2	NM_008006	2.55	0.0416
Fibroblast growth factor receptor 1	Fgfr1	NM_010206	2.25	0.0044
Neurotrophin 5	Ntf5	NM_198190	2.07	0.0071
Toll-like receptor 6	Tlr6	NM_011604	1.65	0.0475
Neurotrophic tyrosine kinase, receptor, type 1	Ntrk1	NM_283871	1.61	0.0259
Nerve growth factor receptor (TNFR superfamily, member 16)	Ngfr	NM_033217	1.50	0.0034
Homocysteine-inducible, ER stress-inducible, ubiquitin-like domain member 1	Herpud1	NM_022331	1.45	0.0439
Transformation-related protein 53	Trp53	NM_011640	1.30	0.0401
Cerebellin 1 precursor protein	Cbln1	NM_019626	1.28	0.0293
Chemokine (C─C motif) ligand 22	Ccl22	NM_009137	−10.60	0.0104
Interleukin 18 receptor accessory protein	Il18rap	NM_010553	−2.38	0.0138
Chemokine (C─C motif) receptor 3	Ccr3	NM_009914	−2.04	0.0332
Neurofibromatosis 1	Nf1	NM_010897	−1.62	0.0367
Chemokine (C─C motif) ligand 11	Ccl11	NM_011330	−1.50	0.0188
Activating transcription factor 6 beta	Atf6β	NM_017406	−1.44	0.0340
Stress-associated endoplasmic reticulum protein 1	Serp1	NM_030685	−1.43	0.0401
Ribophorin 1	Rpn1	NM_133933	−1.28	0.0280
Activating transcription factor 4	Atf4	NM_009716	−1.22	0.1845
Activating transcription factor 6	Atf6	NM_001081304	−1.55	0.3169
Adenylate cyclase activating polypedtide 1 receptor 1	Adcyap1r1	NM_007407	1.87	0.0625
Artemin	Artn	NM_009711	−1.04	0.9163
Bcl2-associated X protein	Bax	NM_007527	1.02	0.7126
B-cell leukemia/lymphoma 2	Bcl2	NM_009741	−1.35	0.2809
Brain-derived neurotrophic factor	Bdnf	NM_007540	1.27	0.4192
CAMP responsive element binding protein 3	Creb3	NM_013497	1.18	0.0752
CAMP responsive element binding protein 3-like 3	Creb3l3	NM_145365	−1.06	0.6089
Chemokine (C─C motif) ligand 2	Ccl2	NM_011333	−1.24	0.4268
Chemokine (C─C motif) ligand 20	Ccl20	NM_016960	−2.08	0.0738
Chemokine (C─C motif) receptor 2	Ccr2	NM_009915	−1.74	0.0864
Chemokine (C─C motif) receptor 4	Ccr4	NM_009916	−1.62	0.4050
Chemokine (C─X─C motif) ligand 3	Cxcl3	NM_203320	−2.80	0.0798
Chemokine (C─X─C motif) receptor 4	Cxcr4	NM_009911	1.34	0.0799
Ciliary neurotrophic factor receptor	Cntfr	NM_016673	1.04	0.6284
Chemokine (C─X3─C) receptor 1	Cx3cr1	NM_009987	1.10	0.8533
Complement component 3	C3	NM_009778	−1.13	0.6402
Complement component 4B (Childo blood group)	C4b	NM_009780	1.75	0.0846
Eukaryotic translation initiation factor 2a	Eif2α	NM_001005509	1.04	0.8454
Eukaryotic translation initiation factor 2 alpha kinase 3	Eif2αk3	NM_010121	1.10	0.8252
Endoplasmic reticulum chaperone SIL1 homolog (S. cerevisiae)	Sil1	NM_030749	−1.20	0.4322
Endoplasmic reticulum protein 44	Erp44	NM_029572	1.12	0.5859
Fas (TNF receptor superfamily, member 6)	Fas	NM_007987	1.27	0.3968
Fas ligand (TNF superfamily, member 6)	Fasl	NM_010177	1.37	0.6202
FBJ osteosarcoma oncogene	Fos	NM_010234	1.14	0.5720
Glial cell line-derived neurotrophic factor	Gdnf	NM_010275	1.12	0.5346
Heat shock protein, alpha-crystallin-related, B9	Hspb9	XM_894891	1.25	0.0577
HtrA serine peptidase 2	Htra2	NM_019752	−1.12	0.1497
HtrA serine peptidase 4	Htra4	NM_001081187	1.08	0.5839
Interleukin 6	Il6	NM_031168	−1.23	0.7948
Interleukin 10 receptor, beta	Il10rβ	NM_008349	1.37	0.0698
Leukemia inhibitory factor	Lif	NM_008501	2.19	0.0676
Myelocytomatosis oncogene	Myc	NM_010849	1.00	0.9990
Nerve growth factor	Ngf	NM_013609	1.44	0.0771
Neuropeptide Y	Npy	NM_023456	1.62	0.4840
Neurotrophin 3	Ntf3	NM_008742	1.02	0.9210
Neurotrophic tyrosine kinase, receptor, type 2	Ntrk2	NM_008745	1.39	0.1745
Nuclear factor of kappa light polypeptide gene enhancer in B-cells 1, p105	Nfκb1	NM_008689	1.77	0.0583
Persephin	Pspn	NM_008954	−1.22	0.2133
Ring finger protein 5	Rnf5	NM_019403	−1.32	0.0573
Sterol regulatory element binding transcription factor 1	Srebf1	NM_011480	−1.21	0.0654
Toll-like receptor 3	Tlr3	NM_126166	−1.16	0.6801
Toll-like receptor 4	Tlr4	NM_021297	1.01	0.9879
Transforming growth factor alpha	Tgfα	NM_031199	1.44	0.3803
Transforming growth factor, beta 1	Tgfβ1	NM_011577	−1.19	0.3175
Tumor necrosis factor	Tnf	NM_013693	−1.16	0.5583
X-box binding protein 1	Xbp1	NM_013842	−1.08	0.5847

 Bold type indicates genes differentially expressed with >2-fold change and P values <0.05. Underscored type indicates upregulated genes also of interest but borderline with significance.

An additional group of 11 genes displayed a weaker (1.3- to 1.6-fold) but still significant change. Upregulated genes were: Toll-like receptor 6 (Tlr6); Neurotrophic tyrosine kinase, receptor, type 1 (Ntrk1); Nerve growth factor receptor (Ngfr); Homocysteine-inducible, endoplasmic reticulum stress-inducible, ubiquitin-like domain member 1 (Herpud1); Transformation related protein p53 (Trp53); and Cerebellin 1 precursor protein (Cbln1). Downregulated genes were: Neurofibromatosis 1 (Nf1); Chemokine (C–C motif) ligand 11 (Ccl 11); Activating transcription factor 6 beta (Atf6β); Stress-associated endoplasmic reticulum protein 1 (Serp1); and Ribophorin 1 (Rpn1). 

Other genes of interest (with regard to their association with retinal degeneration), encoding proteins involved in growth and survival (Bdnf, Cntfr, Cx3cr1, Gdnf, Ngf, Ntf3, Pspn, Ntrk2, Npy, Tgfα, and Tgfβ1), apoptosis (Bcl2, Bax, Myc), inflammation and autoimmunity (C3, Ccl2, Ccr2, Il6, Tlr3, Tlr4, and Tnf), and unfolded protein response (Atf4, Creb3, Creb3l3, Eif2α, Eif2αk3, and Xbp1) remained unchanged. Of note, three genes in this series (Nfκb1, Adcyap1r1, and Lif) showed strong upregulation (1.77- to 2.19-fold) borderline with significance (P = 0.058–0.0672). 

Protein Expression Levels

Candidate genes for further examination at the protein level were chosen based on their expression level as assessed by the number of PCR cycles (threshold cycle, C^t ≤ 30), availability of specific commercial primary antibodies (one band of predicted molecular mass) for Western blotting and/or immunohistochemistry, as well as consideration of their interaction with other known genes involved in retinal pathologies. 

As done for the arrays, investigations were primarily undertaken in TG and WT mice at 9 months. By this age, FGF2 expression in eyecup preparations was found to be increased >3-fold in TG compared with WT mice. Similar measures at 18 months showed that this increase was maintained (Fig. 3A). When protein extracts were prepared from retina alone, FGF2 levels in TG mice were also systematically higher than those in WT littermates. Starting at 1 month (2.5-fold), FGF2 levels increased to approximately 5-fold at 3 and 6 months, and >7-fold at 9 and 18 months (Fig. 3B). Protein levels for the band of expected molecular mass of the activated (pY654) form of FGFR1 (≈117 kDa) were also increased approximately 6-fold at 9 months. 

Figure 3.

 

Increased FGF2 and GFAP levels. (A) >3-fold upregulation FGF2 in TG eyecup at 9 and 18 months. (B) FGF2 upregulation in retina; 2.5-fold at 1 month; approximately 5-fold at 3 and 6 months; >7-fold at 9 and 18 months. GFAP upregulation in retina is 1.5-fold at 1 month of age. (C) Examples of upregulation of FGF2 in photoreceptors at 9 and 18 months in TG (FGF2-labeled photoreceptors, green; DAPI-stained cell bodies, blue). Scale bar, 20 μm.

Early onset upregulation was also observed for GFAP with a 1.5-fold increase relative to WT already present by 1 month of age (Fig. 3B). GFAP expression continued to increase with age as seen for FGF2. At 9 months, GFAP levels in TG retinas were previously shown to be 8-fold that of age-matched WT retinas. ¹⁶  

NGFR/p75NTR protein levels in the eyecup and retina were found to be unchanged in TG compared with WT at 9 months (Fig. 4A). Protein levels for CCR3 (≈40 kDa) were decreased approximately 25% when compared with WT levels at 9 months (data not shown). Finally, we were unable to examine CCL22 expression on Western blot, likely due to very low expression levels of the protein, consistent with the low expression of Ccl22 (average C^t > 30 cycles). In all experiments, levels of α-tubulin (used as a loading control) were found similar in both TG and WT (Figs. 3A, 3B, 4A). Due to their low reliability (Table 1), antibodies against FGFR1, CCR3, and TrkA were not used further. 

Figure 4.

 

No changes to NGFR/p75NTR levels at 9 months. (A) Protein levels are unchanged in both eyecup and retina preparations. (B) There is no change in Müller cell expression pattern (NGFR-labeled Müller cells, red; DAPI-stained cell bodies, blue). Scale bar, 20 μm.

Retina Immunohistochemistry

Constitutive FGF2 expression by Müller cells in the inner nuclear layer was present in both TG and WT retinas at 9 months and did not vary with aging (data not shown). Furthermore, FGF2 was undetectable in WT photoreceptors at both 9 and 18 months (Fig. 3C). In contrast, FGF2 expression could be detected in photoreceptor cell bodies of 9-month-old TG mice (with approximately 50% of photoreceptors remaining), and was noticeably increased at 18 months (with only one row of photoreceptors remaining in the central retina). 

CCL22 staining was confined to rod photoreceptor outer segments (Fig. 5). CCL22 staining did not overlap with cone photoreceptor expressing s-opsin but did localize with rhodopsin (Fig. 6). The staining pattern for CCL22 was unchanged between TG and WT retinas from 1 month (Figs. 5A, 5B, 5E) through 3 months. At 6 months, outer segments in TG retinas were severely truncated, but CCL22 protein expression remained (Figs. 5C, 5F). CCL22 protein expression started to disappear by 9 months, concomitant with the loss of outer segments (Figs. 5D, 5G). In addition, at this age, the RPE monolayer was disrupted and subretinal deposits were evident (Figs. 5D, 5G). 

Figure 5.

 

CCL22 labeling in outer segments. (A) WT at 1 month; (B, C, D) TG at 1, 6, and 9 months, respectively. (E, F, G) Enlargement of insets in (B), (C), and (D) showing rod truncation and loss with concomitant loss of CCL22. (G) Example of RPE disruption (arrow) and subretinal deposit (asterisk) at 9 months in TG. RPE, retinal pigment epithelium; OS, outer segment; IS, inner segment; ONL, outer nuclear layer. Scale bar, 20 μm.

Figure 6.

 

CCL22 labeling in rod outer segments. Top: Photoreceptor outer segments labeled exclusively with CCL22 (red) or s-opsin (green). Bottom: Overlapping expression in outer segments labeled with CCL22 (green) and rhodopsin (red), showing overlapping expression in the merged image (yellow). Brightness and contrast was enhanced for CCL22 to better show localization of weak staining. Scale bar, 20 μm.

No changes were detected in NGFR/p75NTR retina labeling at 9 months between TG and WT mice (Fig. 4B). The expression pattern resembled that for glutamine synthetase, ¹⁶ with selective labeling of glial Müller cells. 

TUNEL

There was no evidence of apoptotic cell death in TG and WT retinas before 9 months. At this time point an occasional TUNEL-labeled cell could be found in TG retinas (Fig. 7), although the majority of retinal sections examined contained no TUNEL-labeled cells. 

Figure 7.

 

Absence of apoptotic cells in TG retina. (A) Evidence of few TUNEL-labeled cells in central retina at 9 months. (B) Enlarged inset. (C) TUNEL-positive control. (D) TUNEL-negative control. Scale bar, 25 μm.

Discussion

The present work describes, for the first time, gene and protein expression differences in the eyecup and retina of transgenic mice (ELOVL4/TG1–2 line) expressing a mutated gene responsible for an autosomal-dominant juvenile form of progressive atrophic macular degeneration, Stargardt-like dystrophy (STGD3). 

This study was performed before cone loss, at an age at which half of the rod population had already disappeared, 9 months. Abundant evidence (for a review, see Sancho-Pelluz et al. ²⁴ ) suggests that photoreceptor cell death in inherited retinal degeneration occurs primarily through a nonapoptotic mechanism. Our findings support this proposal: first, genes in the arrays that are indicators of apoptosis (Bcl2, Bax, Fas, FasL, Fos, Myc, Zfp110/Nrif) are expressed at similar levels in TG and WT mice; second, as in other models with slow photoreceptor degeneration, ^22,25 extremely few TUNEL-labeled cells were found in the retina of TG mice examined at 1, 3, 6, and 9 months; and third, there is a negligibly small upregulation of Trp53, which encodes a pivotal transcription factor in classical apoptosis. We cannot exclude, however, that photoreceptors undergo apoptotic death at time points other than those analyzed here. 

Although the present study involved a limited number of genes, many of them (such as C3, Crp, Cx3cr1, Il6, Tlr1–5, Serp1) have been previously implicated in retinal diseases, more specifically AMD. ²⁶ As with those involved in apoptosis, none of these genes seemed to be reacting to an actual loss of approximately 3.2 million rod photoreceptors (50% of total population estimates). ²⁷ Only a few genes, mostly involved in cell survival, displayed significant changes in expression compared with WT. 

FGF2 is a prosurvival factor. FGF2 supplementation in vitro, ²⁸ its acute in vivo delivery after light damage ²⁹ (but see Valter et al. ³⁰ ), and its sustained availability by gene transfer in S334ter-4 transgenic rats ³¹ have all been reported to promote photoreceptor survival. Of note, a variant of FGF2 was recently associated with the development of atrophic AMD in a Spanish population. ³²  

In the present study, FGF2 expression was already 2.5-fold upregulated at 1 month of age in TG retinas where there are no phenotypic differences with WT mice. This early FGF2 upregulation parallels that of GFAP (Fig. 3B), ¹⁶ a ubiquitous marker of retina stress and degeneration. ^33,34 Thus, the onset of pathologic events in the TG mouse involves subtle homeostatic changes that precede any detectable anatomic and/or functional alterations. 

FGF2 levels continue to increase with age in TG mice, reaching levels 7-fold that of WT from 9 to 18 months (Fig. 3B). FGF2 immunoreactivity observed in photoreceptor cell bodies is essentially due to local overexpression of Fgf2 mRNA. ³⁵ Similar results have been reported with various types of retinal injuries, ^{30,36 –40} inherited photoreceptor degeneration, ^35,41 or chemically induced diabetic retinopathy. ²¹ The survival-promoting activity of FGF2 is mediated in part by upregulation of its receptor FGFR1 ⁴² and involves a CREB-dependent BCL2 upregulation. ⁴³ FGF receptors are present on rodent photoreceptors ^30,44 and are necessary for the maintenance of their morphologic integrity. Fgfr1 upregulation in TG mice matches that seen for Fgf2 at 9 months. 

Given the fact that very few photoreceptors (all cones) remain at 18 months, the high expression level of FGF2 observed at this age may be a sign of the increasing resistance of the residual photoreceptors to maintain morphologic stability and function. On the other hand, contribution from other cell types cannot be ruled out. FGFR1 and FGF2 are naturally expressed in glial Müller cells and retinal ganglion cells (RGCs). ^{21,30,36,38,45} BDNF-induced production of FGF2 by Müller glia increases bipolar cell survival in vitro. ⁴⁶ High levels of FGF2 may also contribute to the expansion of inner retinal blood vessels ^47,48 observed in TG mice at late degeneration stage. ¹⁶ However, as with other models of retinal dystrophies, ^21,36 no significant increase in FGF2 expression could be detected by immunohistofluorescence in RPE and inner retina of old TG mice. 

Ntf5/Nt4, Ntrk1/TrkA, and Ngfr/p75Ntr were all found to be substantially upregulated in eyecups of TG mice at 9 months. In rodents, Trk mRNA and protein are expressed by RGCs, various amacrine cell types, and Müller glial cells. ^{49 –53} The marked upregulation of Ntf5/Nt4 and Ntrk1/TrkA may therefore contribute to the normal appearance of inner retina circuits at this mid-degenerative stage ¹⁶ as well as, indirectly, to photoreceptor rescue. ⁵⁴ Additional indirect prevention of photoreceptor loss is likely due to the behavior of p75NTR, which is expressed throughout the extent of Müller cells. ^{55 –59} Although there was a weak increase of Ngfr/p75Ntr in 9-month-old TG mice compared with WT, no difference was found at the protein level. This result contrasts with those referred to in rapidly progressing inherited or induced photoreceptor degenerations, ^54,57 but has been observed after ON axotomy. ^56,58 Interestingly, blockade of p75NTR on Müller cells prevents photoreceptor apoptosis during light-induced retinal degeneration by increasing synthesis of the photoreceptor survival factor FGF2. ⁵⁴  

We did observe a weak upregulation of Tlr6 and Nfκb1 (see the following text). This may denote an inflammatory response of RPE cells (which express Tlr6 constitutively at a very low level ⁶⁰ ) to accumulating degradation products from degenerating rods and an injured extracellular matrix. 

Herpud1 is the only gene induced by ER stress that is upregulated (although at extremely low levels) in TG eyecups. It encodes HERP, a protein suspected to delay the degradation of cytosolic proteins ⁶¹ (including those misfolded, in the present model). Of note, HERP accumulates in senile plaques of Alzheimer's patients ⁶² as well as in the substantia nigra of Parkinson's patients. ⁶³  

Finally, the present study confirms that the Purkinje cell marker, Cbln1 (a ligand for the ionotropic “orphan” glutamate receptor delta 2), is expressed at very low levels in the eyecup (“Genevestigator”; https://www.genevestigator.com/gv/index.jsp). Interestingly, (1) Delta1/2 subunits in the mammalian retina are localized at the rod bipolar–AI amacrine cell synapse ^{64 –66} ; (2) Cbln1 has been located in the mouse retina and is upregulated with age ⁶⁷ ; and (3) formerly identified Purkinje cell markers L7 and PEP19 have been shown to label rod and cone bipolar cells. ^68,69 Thus, it would be pertinent to examine whether Cbln1 is more strongly upregulated at later time points, when rod bipolar cells in TG mice display obvious morphologic and functional alterations. ¹⁶  

Three genes, all concerned with inflammation and autoimmunity, are markedly downregulated in TG mice at 9 months. The most affected is Ccl22 (10-fold downregulation), which encodes the secreted chemokine CCL22 (also known as MDC). Both Ccl22 and its receptor Ccr4 are naturally expressed at very low levels in mouse RPE and retina (“Genevestigator”; see above). CCL22 is involved in the migration of various types of leukocytes to inflammation sites ⁷⁰ and has also a direct proinflammatory role in certain pathologic conditions (postoperative proliferative vitreoretinopathy ⁷¹ ). The finding that CCL22 is expressed at similar levels in both juvenile and old WT animals (present data) further suggests a possible involvement in retinal homeostasis. Its absence from TG retinas starting at 9 months indirectly supports a location in rod outer segments since rods are the only retinal elements that are dying or displaying severely truncated outer segments ¹⁴ at this time point. 

Substantial downregulation also occurs for Ccr3 (encoding the receptor for the chemokine C–C motif ligand 11, eotaxin) and Il18rap (encoding the interleukin 18 receptor accessory protein). Although best known to promote eosinophil trafficking, a recent report ⁷² showed that CCR3 is present in choroidal endothelial cells of patients with AMD and that CCR3 blockade is strongly effective in reducing laser-induced choroidal neovascularization in mouse. The fact that vascular changes in TG retinas do not affect the choroidal plexus ¹⁶ might be related to the downregulation of Ccr3. As previously evidenced in various brain structures, ^73,74 Il18rap is expressed at very low levels in eyecup preparations. The reason why this gene is downregulated in TG mouse at 9 months is unknown. 

The Nf1 gene encodes the protein neurofibromin, a negative regulator of the Ras signaling pathway. Conditional mutant mice lacking this gene either in neurons or in astrocytes display reactive astrocytic hyperplasia characterized by enlarged somata, thickened processes, and increased expression of GFAP. ^75,76 Downregulation of Nf1 might lead progressively to the hypertrophy of the GFAP-positive astrocytes observed in the ultimate stage of the TG degenerating retina. ¹⁶  

The last three genes (Rpn1, Serp1, and Atf6β) undergo much weaker downregulation. With regard to their known functions, ^{77 –79} downregulation of these genes may lead to altered protein quality and to the induction of ER stress-response genes (ERSRGs) (such as Grp78/Bip/Hspa5). There was, however, no upregulation of this target in our preparation. This result suggests that the downregulation of these genes reflects the loss of rod photoreceptors rather than specific gene repression. 

Finally, the finding that three genes with potential importance in photoreceptor survival displayed strong (≥1.8-fold) upregulation borderline with significance (P = 0.058–0.0672) was intriguing. Additional experiments focusing on the transcription factor Nfκb1 indeed showed a concomitant upregulation (1.4-fold) of retinal NF-κB protein (p50/105) in Western blots, and a small expansion of NF-κB–labeled processes originating from the ganglion cell layer (data not shown). Of note, there is evidence that NF-κB is activated in retinal diseases including rod-induced dystrophy ^80,81 and is located preferentially in inner retina neurons. ⁸² Variability between biological samples ¹⁶ may explain why the difference in Nfκb1 expression levels between TG and WT was not found to be statistically significant. The same reasoning may apply to Adcyap1r1 and Lif, which encode respectively for the high affinity pituitary adenylate cyclase–activating polypeptide (PACAP) receptor (PAC-1) and the leukemia inhibitory factor (LIF). Future studies will examine these markers more closely for the following reasons: first, docosahexaenoic acid inhibits the activation of Toll-like receptors 1 to 6 and NF-κB in vitro ^83,84 ; second, both PACAP (which distributes throughout the retina and RPE) and LIF (mostly secreted by Müller cells) have neuroprotective properties that prevent photoreceptor degeneration ^{85 –87} ; and third, LIF stimulates IL-6 production in Müller cells through an intraretinal signaling process characterized by an upregulation of GFAP and FGF2 and, therefore, may be an early target to act on. 

Conclusions

Photoreceptor degeneration in the ELOVL4/TG1–2 mouse model of Stargardt-like disease starts at approximately 2 months and is completed beyond 18 months. At a mid-degeneration stage, approximately half of the rod population has disappeared. Unexpectedly, no sign of apoptosis is observed at this time point. Moreover, increases in expression levels occur in only a few genes involved in cell survival and/or homeostasis. With one exception, genes implicated in unfolded protein response and inflammation exhibit weak downregulation that is, for the most part, likely related to the actual loss of photoreceptor cells. The present data contrast with those obtained either with acute retinal injuries or rapidly degenerating retinas. As in multiple retinopathies, both induced and inherited, there is a very precocious upregulation of FGF2 in photoreceptors and GFAP in Müller cells, whereas the anatomic integrity of the retina is still preserved. Treatment applied within this time window would theoretically be most successful in preventing or delaying photoreceptor death. Future investigations will examine which factors induce this early upregulation of Fgf2 and whether antenatal dietary manipulation with omega-3 fatty acids and antioxidants might delay the progression of retinal degeneration in the ELOVL4 mouse retina. 

Footnotes

 Supported in part by Canadian Institutes of Health Research Grants 151145 and 192321; Alberta Innovation Health Services Establishment Grant 200700584; Canadian National Institute for the Blind; Olive Young Foundation; The Lena McLaughlin Foundation (Mona and Rod McLennan); Alberta Heritage Foundation for Medical Research Senior Scholar (YS); and Barbara Tuck/MacPhee Family Vision Research Award in Macular Degeneration (YS).

Footnotes

 Disclosure: S. Kuny, None; F. Gaillard, None; Y. Sauvé, None

The authors thank Xuejun Sun and Geraldine Barron at the Cross Cancer Institute Cell Imaging Facility for technical assistance in the use of the facility's laser scanning microscope (Zeiss LSM710).