Received 2 January 2008 | Accepted 19 November 2008 | Published 26 November 2008
1Institute for Research in Ophthalmology, Sion, Switzerland; 2Department of Ophthalmology, University of Lausanne, Lausanne, Switzerland; 3Department of Perinatology, Institute for Developmental Research, Aichi Human Service Center, Kasugai, Japan; 4EPFL-Ecole Polytechnique Fédérale, Lausanne, Switzerland
Correspondence to: Dr. Pascal Escher, IRO-Institut de Recherche en Ophtalmologie, Grand-Champsec 64 CH-1950, Sion, Switzerland; Phone: ++41 (0)27 205 79 02; FAX: ++41 (0)27 205 79 01; email: firstname.lastname@example.org
Purpose: An increased mRNA expression of the genes coding for the extracellular matrix proteins neuroglycan C (NGC), interphotoreceptor matrix proteoglycan 2 (IMPG2), and CD44 antigen (CD44) has been observed during retinal degeneration in mice with a targeted disruption of the Rpe65 gene (Rpe65−/− mouse). To validate these data, we analyzed this differential expression in more detail by characterizing retinal NGC mRNA isoform and protein expression during disease progression.
Methods: Retinas from C57/Bl6 wild-type and Rpe65−/− mice, ranging 2 to 18 months of age, were used. NGC, IMPG2, and CD44 mRNA expression was assessed by oligonucleotide microarray, quantitative PCR, and in situ hybridization. Retinal NGC protein expression was analyzed by western blot and immunohistochemistry.
Results: As measured by quantitative PCR, mRNA expression of NGC and CD44 was induced by about 2 fold to 3 fold at all time points in Rpe65−/− retinas, whereas initially 4 fold elevated IMPG2 mRNA levels progressively declined. NGC and IMPG2 mRNAs were expressed in the ganglion cell layer, the inner nuclear layer, and at the outer limiting membrane. NGC mRNA was also detected in retinal pigment epithelium cells (RPE), where its mRNA expression was not induced during retinal degeneration. NGC-I was the major isoform detected in the retina and the RPE, whereas NGC-III was barely detected and NGC-II could not be assessed. NGC protein expression was at its highest levels on the apical membrane of the RPE. NGC protein levels were induced in retinas from 2- and 4-month-old Rpe65−/− mice, and an increased amount of the activity-cleaved NGC ectodomain containing an epidermal growth factor (EGF)-like domain was detected.
Conclusions: During retinal degeneration in Rpe65−/− mice, NGC expression is induced in the neural retina, but not in the RPE, where NGC is expressed at highest levels.
Leber congenital amaurosis (LCA) is a genetically heterogeneous retinal dystrophy with prenatal onset. A subset of patients carries mutations in the retinal pigment epithelium protein of 65 kDa (RPE65) gene (LCA2; OMIM #204100) [1,2]. RPE65 is the iron (II)-dependent isomerohydrolase essential for the generation of the photopigment 11-cis retinal from all-trans-retinyl ester in the retinoid visual cycle [3-6]. In the RPE of mice with a targeted disruption of the Rpe65 gene (Rpe65−/− mice), no 11-cis retinal has been synthesized, and excessive accumulation of all-trans retinyl esters has been observed . This enzymatic defect in the RPE was found to result in profound effects in the underlying photoreceptors. Cone photoreceptor degeneration was found to be complete within the first postnatal weeks in Rpe65−/− mice, whereas rod photoreceptor degeneration progressed slowly [8,9]. Early cone loss was reflected by a rapid decrease in cone-specific gene expression [10,11].
The gene coding for the interphotoreceptor matrix (IPM) proteoglycan 2 (IMPG2) was induced in retinas of Rpe65−/− mice [11,12]. The IPM is a specialized extracellular matrix of fundamental importance to vision, e.g., in trafficking of retinoids and other metabolites between photoreceptors and the RPE, and in retinal adhesion or in photoreceptor outer segment recognition for phagocytosis . It has been proposed that proteoglycans containing hyaluronic acid-binding motifs, e.g., IMPG2, IMPG1, and CD44, a cell surface adhesion molecule specifically localized in the Müller cell microvilli that oppose the IPM , stabilize a scaffold of hyaluronic acid in the IPM . Interestingly, increased CD44 mRNA levels have also been observed in Rpe65−/− retinas .
Additionally, expression of the transmembrane neuronal proteoglycan with chondroitin sulfate (NGC: neuroglycan C; also called CALEB: chicken acidic leucine-rich EGF-like domain containing brain protein; CSPG5: chondroitin sulfate proteoglycan 5) has also been induced in Rpe65−/− retinas [11,16,17]. Chondroitin sulfate side chains become attached to the NGC core protein in the developing rat cerebellum and retina, but not the adult ones [18,19]. NGC has therefore been described as a part-time proteoglycan. In the central nervous system, NGC has been found to be associated with both glial and neuronal surfaces . In the retina, NGC is highly expressed on the axons of the nerve fiber layer and the inner plexiform layer at early postnatal stages (between P0 and P14), when active dendrite branching and conventional synapses between amacrine cells and ganglion cells can be observed in the inner retinal layers . At late postnatal and adult stages (between P14 and P42), when synapse formation and dendrite branching are almost complete, NGC expression was found to be reduced . NGC was localized to basal infoldings at P7 and to microvillis of the apical surface in the adult retina (P42) , indicating that high NGC expression in the RPE is differentially regulated during development. Neuronal depolarization of chick retinal cells in culture was shown to facilitate the processing of full-length NGC into a truncated transmembrane form and an ectodomain . This activity-dependent ectodomain shedding exposed the EGF-like domain, located in the C-terminus of the ectodomain of NGC . Interestingly, a recombinant ectodomain promoted neurite outgrowth from rat neocortical neurons in culture  and mediated dendritic tree and spine complexity in vivo .
The aim of this study was to validate and characterize the NGC expression during retinal degeneration in Rpe65-/- mice. Additionally, we assessed the expression of IMPG2 and CD44.
All experiments performed in this study were in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Veterinary Service of the State of Valais (Switzerland). C57BL/6 mice (RCC, Basel, Switzerland) and Rpe65−/− mice  were kept in a 12 h:12 h light-dark cycle with unlimited access to food and water until they were used for experiments.
Mice were killed by cervical dislocation. Eyes were enucleated, immobilized with 0.2 mm Austerlitz insect pins (Fine Science Tools, Heidelberg, Germany) on a Sylgard 184-filled cell culture dish (Dow Corning, Midland, MI), and covered with 1X PBS (phosphate-buffered saline: 154 mM NaCl, 1 mM KH2PO4 , 3 mM Na2HPO4 heptahydrate). Under a Leica 16MZF microscope (Leica Microsystems, Heerbrugg, Switzerland), the eyeball was sectioned below the ora serrata to remove cornea, lens, iris, and other attached tissues. The retina was then removed by cutting the optic nerve. For quantitative PCR experiments, pure RPE cells were obtained by trypsin-digestion from the posterior eyecup . Briefly, the posterior eyecup comprised of RPE, choriocapillaris, and sclera was incubated in 0.2% trypsin (Invitrogen, Basel, Switzerland) for 1 h at 37 °C in a 5% CO2 atmosphere. Then, RPE cells were gently peeled off with forceps. For RT–PCR experiments, the sclera was dissociated from the RPE and attached choroidal tissue through homogenization with 18 gauge Sterican needles (Braun, Melsungen, Germany) in TRIzol (Invitrogen). Dissociated RPE with attached choroid was separated from the sclera by centrifugation at 1000x g and stored at −80 °C. Total RNA from the different dissected eye tissues was prepared according to manufacturer’s instructions, with prolonged centrifugation times to increase RNA recovery.
The oligonucleotide microarray was previously described in detail . Briefly, three retinas were pooled for each time point. Total RNA (1 μg) was used to generate doublestranded cDNA used as a template for biotinylated cRNA synthesis using Affymetrix GeneChip Expression 3′-Amplification Kit for IVT Labeling Kit (Affymetrix, Santa Clara, CA). Next, 20 μg of target cRNA were fragmented and hybridized on Affymetrix Mouse Genome 430 2.0 GeneChips, and washed chips were scanned on an Affymetrix GeneChip Scanner 3000 using the GCOS software (Affymetrix). Data normalization was performed using the Robust Multi-Array Analysis algorithm (RMA) as implemented in the GeneSpring 7.2 software (Agilent Technologies, Santa Clara, CA). Triplicates were performed for each condition studied. The intensity files corresponding to our raw data are deposited in NCBI Gene Expression Omnibus (GEO) database.
Total RNA (2 μg) were used for reverse transcription (StrataScript™; Stratagene, La Jolla, CA). One-tenth of the reaction was used for subsequent PCR experiments. After an initial denaturation of 3 min at 94 °C, a 30 cycle PCR amplification was performed as follows: 30 s at 94 °C, 30 s at 60 °C, and 45 s at 72 °C. Amplification was followed by a final elongation of 5 min at 72 °C (Taq PCR Master Mix ; Qiagen, Hilden, Germany). Expression of the ribosomal protein L8 (RL8) was used as internal standard . PCR products were analyzed on 2% agarose gels. Quantitative PCR was performed on an Mx3000p sequence analyzer using Brilliant®SYBR®Green qPCR Master Mix I (Stratagene). Experimental data points with cycle threshold (Ct) values above 30 were not considered for data analysis. Primers are listed in Table 1.
Eyes were enucleated, rinsed in diethyl-pyrocarbonate (DEPC)-treated 1X PBS and fixed for 2 h with 4% paraformaldehyde-1X PBS-DEPC at 4 °C. After cryoprotection by immersion in 30% sucrose-1X PBS-DEPC, overnight at 4 °C, eyes were embedded in freezing compound (30% albumin/3% gelatine in DEPC-treated 1X PBS). Eyes were sectioned at −21 °C on a Leica CM1900 cryostat, and 12 μm sections were recovered on SuperFrost®Plus microscope slides (Menzel Gläser, Braunschweig, Germany) pretreated with Vectabond (Vector Laboratories, Burlingame, CA). PCR fragments of NGC and IMPG2 cDNAs, amplified with primers shown in Table 1, were subcloned into the pGEM®-T Easy Vector (Promega, Madison, WI). Plasmids were linearized with XbaI or NcoI for probe synthesis. Digoxygenin (DIG)-labeled sense and antisense probes were tested by immunodot blotting and equal amounts of probe used in the experiment . In situ hybridizations were performed as previously described, including a carbamoylation step in active 0.1% DEPC-1X PBS at a hybridization temperature of 48 °C .
Eyes were enucleated and fixed in 4% paraformaldehyde-1X PBS for 2 h at 4 °C. After cryoprotection by immersion in 30% sucrose-1X PBS overnight at 4 °C, eyes were embedded in freezing compound (30% albumin/3% gelatine in 1X PBS). For immunohistochemistry, 12 μm cryosections were collected on Superfrost®Plus glass slides (Menzel,). Sections were dried at room temperature for at least 1 h, quickly hydrated with 1X PBS, and blocked for 1 h in 1X PBS containing 2% goat serum and 0.2% Triton X-100. A rabbit polyclonal serum raised against the extracellular domain of rat NGC was diluted at 1:500 in blocking solution and incubated overnight at 4 °C . Sections were then briefly rinsed twice with blocking solution and washed once for 5 min. Secondary antibodies conjugated to Alexa Fluor 594 (Molecular Probes, Invitrogen) were diluted at 1:1,000 and incubated for 1 h at room temperature in the dark. Sections were rinsed briefly twice in 1X PBS then washed three times for 5 min. Parts of sections were stained with DAPI to visualize nuclei. To stain cone photoreceptor outer segments, we used 20 μg/ml fluorescein-conjugated peanut agglutinin (Sigma-Fluka, Buchs, Switzerland). Slides were washed three times for 5 min in 1X PBS, before mounting in Cityfluor (Cityfluor Ltd., London,UK).
On ice, four mouse retinas were homogenized with a plastic pestle in 200 μl of a buffer containing 100 mM NaCl, 50 mM Tris pH 7.5, 1 mM EDTA, 0.1% Triton X-100, 10 mM NaF, and freshly added proteinase inhibitors (Complete; Roche, Basel, Switzerland). Protein extracts were stored at −80 °C. For western blotting, 20 μg of protein extracts were separated on a 5% stacking gel and a 6% resolving gel. Proteins were transferred onto a polyvinlyidenfluoride (PVDF) membrane (Westran® Clear Signal; GE Healthcare, Piscataway, NJ). Membranes were blocked for 1 h in TBS-Tween containing 5% nonfat dry milk. The rabbit polyclonal anti-NGC antibody  and the rabbit polyclonal anti-α-tubulin (H-300) antibody (Santa Cruz Biotechnology, Santa Cruz, CA) were diluted 1:5,000. The secondary ECL™ donkey anti-rabbit IgG horseradish peroxidase-conjugated antibody was diluted 1:25,000 (GE Healthcare). Proteins were detected by chemiluminescence (ECL Plus Western Blotting Detection System, GE Healthcare) on a Hyperfilm™ ECL (GE Healthcare). Signals were quantified with Gelquant software Version 2.7.0 (DNR Imaging Systems Ltd., Jerusalem, Israel).
Statistical analysis was performed on normalized cycle threshold values for quantitative PCR and on raw densitometric data for western blot analysis. Data were analyzed by a two-way ANOVA, using genotype and age factors (Graphpad Prism 4.0.2; GraphPad Software Inc., La Jolla, CA).
In an initial gene expression profiling experiment using Affymetrix® oligonucleotide microarrays, 534 genes were differentially expressed in Rpe65−/− mouse retinas after two, four, and six months of disease progression . We extended the time-course to up to 18 months (Figure 1, left panels). The genes coding for the reported IPM proteins IMPG2 and CD44 were induced by over 2 fold during disease progression. Additionally, we observed an up to 1.7 fold induction of NGC mRNA expression in Rpe65−/− retinas. We validated the oligonucleotide microarray data by quantitative PCR (Figure 1, right panels). At all time points, NGC mRNA expression was induced by over 2 fold in Rpe65−/− retinas versus wild-type levels (p<0.01). Additionally, there was a significant decrease in NGC mRNA expression over time in both wild-type and Rpe65−/− retinas (p<0.01), consistent with previous observations in developing and adult retinas . IMPG2 mRNA levels were induced in Rpe65−/− retinas by 4.2 fold at two months, and decreased progressively to less than 2 fold at 18 months. The mRNA levels for CD44 were induced in Rpe65−/− retinas by 2.9 fold at two months and remained elevated during disease progression up to 18 months.
To date, three different NGC isoforms have been reported in mice . They were generated through alternative exon usage in 5′. Exon 1 present in NGC-I can be replaced by an intronic sequence (exon 1’) and yield isoform NGC-II (Figure 2A). Additionally, the presence of the alternatively spliced exon 5 results in isoform NGC-III. All three NGC isoforms were detected in adult mouse brain and eye by isoform-specific PCR (Figure 2B). NGC-I levels were highest in the RPE, the retina, and the brain; they were lower in the lens and the cornea. For NGC-II, the size of the PCR amplification products from mouse genomic DNA and cDNA were identical. NGC-II was barely detected in brain, retina and RPE, but a strong signal was present with mouse genomic DNA. NGC-III was detected in a similar pattern than NGC-I. IMPG2 was selectively expressed in the retina, whereas CD44 was detected in all analyzed tissues. Genomic fragments were also amplified on mouse genomic DNA with primers for RL8 and IMPG2. In contrast to what had been observed for NGC-II, these signals were weaker than those amplified from cDNA templates, thus suggesting low NGC-II mRNA expression levels. Furthermore, contamination of the cDNA templates by genomic DNA could be excluded by the absence of amplification products corresponding to the RL8 genomic fragments.
To assess isoform-specific NGC mRNA expression, we performed quantitative PCR experiments on retinas and RPE cells of 2-, 6- and 12-month-old wild-type and Rpe65−/− mice (Figure 3). NGC-I expression was over 2 fold induced in Rpe65−/− retinas at all time points. NGC-I transcripts were around 5 fold to 6 fold more abundant in the RPE than in the retina, but expression levels were not increased in Rpe65−/− mice. NGC-II levels were comparable between wild-type and Rpe65−/− mice, but at least 16.9 old lower than those of NGC-I, with cycle threshold values beyond our threshold of confidence (Ct>34; see Methods; data not shown). In wild-type retinas, NGC-III mRNA levels were about 10 fold lower than NGC-I levels. They were increased in Rpe65−/− retinas, to a similar extent than that observed for NGC-I (p<0.01). NGC-III mRNA levels were low in the RPE, but became significantly increased over age (p<0.05).
We analyzed NGC mRNA expression by in situ hybridization (Figure 4) to exclude the possibility that the observed increase in NGC mRNA expression in Rpe65−/− retinas was due to de novo expression in retinal cell populations that do not express NGC in wild-type mice. NGC mRNA was highly expressed in the ganglion cell layer, in the inner nuclear layer, around the outer limiting membrane, and in the RPE (Figure 4A,G). It was also detected at low levels in the inner plexiform layer and the outer nuclear layer (Figure 4A). We did not observe ectopic NGC mRNA expression during disease progression, not even in 6-month-old Rpe65−/− mice, where the outer retina was disorganized  (Figure 4C-F).
We also examined mRNA expression of IMPG2 during disease progression (Figure 5). In wild-type retinas, the mRNA expression pattern of IMPG2 was comparable to that of NGC (Figure 5A). In 2-month-old Rpe65−/− retinas, increased IMPG2 mRNA expression was observed at the outer limiting membrane (Figure 5E). At six months, IMPG2 mRNA expression was decreased at the outer limiting membrane and the inner nuclear layer in Rpe65−/− retinas (Figure 5F), and, to some extent, in wild-type retinas (Figure 5D).
To test whether the observed increase in NGC mRNA expression correlated with an increase in NGC protein levels, we performed western blot analysis on total retinal protein extracts of 2-, 4- and 6-month-old wild-type and Rpe65−/− mice (Figure 6). An antisera raised against the extracellular part of human NGC detected a protein as a smear from about 120 kDa upwards, corresponding to the NGC core protein of 120 kDa with multiple posttranslational modifications [16,26,27] (Figure 6A). In retinal protein extracts of 2-month-old mice, and to a lesser extent in 4-month-old mice, a marked increase in the activity-cleaved 75 kDa-ectodomain of NGC was observed (Figure 6A; data not shown). NGC protein levels, i.e., the sum of full-length NGC and the ectodomain, were over 2 fold induced in retinas of Rpe65−/− mice at 2 and 4 months of age, but no significant difference was detected at 6 months (Figure 6B).
The retinal NGC expression pattern detected by immunohistochemistry with the antisera raised against the entire extracellular domain of NGC  was comparable to that detected with an antisera against a partial ectodomain [19,28]. NGC was enriched in all neurite-containing retinal layers of adult mouse retina (Figure 7A,B). NGC was expressed in the nerve fiber layer, the inner plexiform layer, and the outer plexiform layer, but was absent in the cell bodies of ganglion cells and in the inner nuclear layer. In the outer neural retina, NGC staining was intense around the outer limiting membrane and was associated with longitudinal structures in the outer nuclear layer (Figure 7A,B). Consistent with the mRNA expression, the highest levels of NGC expression were detected in the RPE (Figure 7C). Counterstaining with a cone-specific lectin showed that NGC immunoreactivity was highest at the apical membranes of RPE cells, surrounding the outer segments of the photoreceptors (Figure 7C).
In the present work, we showed that NGC, IMPG2, and CD44 mRNA expression was induced during retinal degeneration in Rpe65−/− mice. The amplitude of fold-inductions were similar to those observed for proteoglycans and CD44 in other mouse models of retinal degeneration, namely rd1 (rodless retina), rds (retinal degeneration slow, rd2), and rhodopsin knockout (Rho−/−) mice [29-31]. Upregulation of extracellular matrix proteins expressed in the IPM might therefore be a general mechanism observed in retinal degeneration.
The induction of CD44 gene expression had been proposed to be an attempt of the Müller cells to strengthen the IPM, thereby counteracting the degenerative process . Elevated IMPG2 mRNA levels early in retinal degeneration might also be circumstantial evidence of the importance of chondroitin sulfate proteoglycans in the IPM to maintain functional photoreceptors. Indeed, both the inhibition of chondroitin sulfate proteoglycan synthesis by intravitreal injections of p-nitrophenyl-β-D-xylopyranoside and the impairment of β-glucuronidase-mediated lysosomal degradation of chondroitin sulphate, resulted in photoreceptor degeneration [32,33]. Notably, at 18 months, when the photoreceptors have almost completely disappeared in the Rpe65−/− mice, IMPG2 mRNA levels were down to levels observed in age-matched wild-type retinas. Consistently, the decrease in IMPG2 mRNA expression appeared to start from the outer retina toward the inner retina, as suggested by in situ hybridization on 6-month-old retinas.
In the present work, we identified NGC as an additional chondroitin sulfate proteoglycan expressed in the IPM during retinal degeneration in Rpe65−/− mice. NGC-I was the most highly expressed isoform in the retina and in the RPE. Additionally, mainly NGC-I mRNA expression was induced during retinal degeneration in Rpe65−/− mice. The low NGC-II mRNA expression in brain, retina, and RPE could not be assessed by quantitative PCR. It has to be mentioned at this point, that the presence of a NGC-II protein in the central nervous system has not been assessed to date . The isoform NGC-III was present at low levels both in the retina and the RPE. Taken together, our results define NGC-I as the major isoform in the retina and the RPE.
At the protein level, we observed a significant induction of NGC expression in retinas as old as four months. Most interestingly, we observed an increase in the shedded ectodomain of NGC in 2- and 4-month-old retinas. This EGF-like-domain-containing ectodomain is sufficient to promote neurite outgrowth from rat neocortical neurons in culture . Additionally, the full-length NGC is a critical modulator of dendritic branching and spine formation both in cultured primary neurons and in the mouse cortex [17,22]. The induction of NGC expression in neurite-containing retinal layers could therefore be an attempt to maintain proper synaptic transmission in the degenerating retina. Recently, tissue inhibitor of metalloproteinases 2 (TIMP-2) and TIMP-3 have been identified in vitro as inhibitors of NGC ectodomain shedding . It will be interesting to assess the in vivo activity of TIMP-2 and -3 in RPE65-/- retinas and test any physiological effect on ectodomain shedding.
In the adult eye, NGC was most abundantly expressed at the apical membrane of the RPE. However, its transcript levels were not differentially regulated during retinal degeneration, suggesting the presence of tissue-specific regulatory elements in the NGC promoter. Constitutive expression of NGC at the apical membrane of RPE cells might be necessary to modulate the physiologic interactions between the RPE and the photoreceptors, e.g. the phagocytosis of photoreceptor outer segments in the adult retina . The availability of NGC−/− mice will allow a detailed analysis of the importance of NGC in maintaining the retinal neuronal network during retinal degeneration in Rpe65−/− mice .
We thank Nathalie Voirol for excellent technical assistance. We also thank Dr. T. M. Redmond for the Rpe65−/− mice and Dr. Bertrand Léger for help with the gel quantification program. These data were presented as a poster at the ARVO 2008 meeting. This work was supported by a grant from Téléthon Action Suisse (to P.E.).