Altered Expression of Metallothionein-I and -II and Their Receptor Megalin in Inherited Photoreceptor Degeneration

Abstract

Purpose.: To examine in rodent models of retinitis pigmentosa (RP) the expression of the neuroprotectants metallothionein-I and -II and of megalin, an endocytic receptor that mediates their transport into neurons.

Methods.: Gene and protein expression were analyzed in retinas of rd1 and rds mice and in those of RCS (Royal College of Surgeons) rats of various ages. Glial cell markers (cellular retinaldehyde binding protein, CRALBP; glial fibrillary acidic protein, GFAP; CD11b; and isolectin B4) were used to establish the identity of the cells.

Results.: Metallothionein-I and -II gene expression increased with age in normal and degenerating retinas and was significantly greater in the latter. Protein expression, corresponding to metallothionein-I+II, was first observed in rd1 mice in Müller cells at postnatal day (P)12 and in rds mice at P16, coinciding with the onset of GFAP expression in these cells. In RCS rats, the same distribution was observed, but not until P32, long after the onset of GFAP expression. Metallothionein-I+II was observed also in a small number of microglial cells. Megalin was expressed in the nerve fiber layer and in the region of the inner and outer segments in normal animals, but expression in the outer retina was lost with age in degenerating retinas.

Conclusions.: Induction of metallothionein-I and -II occurs in the RP models studied and correlates with glial activation. The progressive loss of megalin suggests that transport of metallothionein-I+II into the degenerating photoreceptors (from e.g., Müller cells), could be impaired, potentially limiting the actions of these metallothioneins.

Development and maintenance of the retina require that sufficient levels of survival factors be continuously synthesized and secreted. Predictably, even higher levels are necessary after mutations or injury, and a number of studies have shown that acute and chronic retinal cell damage triggers distinct endogenous protective mechanisms. In retinitis pigmentosa (RP), a progressive and irreversible loss of photoreceptor cells occurs as a result of mutations in primarily rod-specific genes (http://www.sph.uth.tmc.edu/Retnet/ RetNet is provided in the public domain by the University of Texas Houston Health Science Center, Houston, TX). Over time, cone cells also die, leading eventually to total loss of vision. ^1,2 In the course of the disease, prosurvival signaling cascades are activated in the dysfunctional photoreceptor cells themselves and in support cells. Numerous endogenous protective molecules are upregulated, such as growth factors, cytokines, and antioxidants, some of which have also been shown to extend, at least temporarily, the lifespan of photoreceptors when applied exogenously. ^3–7  

In the present study, we have examined the expression of metallothionein-I and -II in three rodent models of RP. Metallothioneins constitute a class of low-molecular-mass (6–7 kDa), cysteine-rich proteins and exist in four major isoforms. Metallothionein-I and -II are ubiquitously expressed in most tissues, whereas metallothionein-III and -IV are rather restricted to a few specific tissues. Metallothionein-I and -II are structurally and functionally analogous and are often referred to as a single entity (metallothionein-I+II in the present study). In the rodent CNS, they are regulated and induced together by various pathogenic factors, including metals, hormones, proinflammatory cytokines, and reactive oxygen species (ROS). ^8–15  

Because of the high abundance of thiol groups, metallothioneins bind and release both essential (e.g., zinc) and toxic metals, serving thereby as intracellular regulating factors of metal ions during physiological and pathologic conditions. ^9,14–16 The redox properties of metallothionein, in addition, enable its own oxidation even by mild oxidants, allowing zinc to be transferred from metallothionein to other proteins to subserve numerous cellular functions. Via the metal thiolates, metallothioneins also exchange zinc with ROS that are scavenged during cellular oxidative stress. ^9,14,16–19  

Metallothionein-I and -II are expressed in the retina and in the retinal pigment epithelium (RPE), ^20–25 and significantly higher levels have been observed after intraocular application of N-methyl-d-aspartate (NMDA), ²⁶ phototoxic insult, ²⁷ mechanical damage, ²⁸ and hypoxic preconditioning. ²⁹ Cell loss is increased after NMDA application in metallothionein-I+II-deficient mice, ²⁶ and a correlation has been suggested between low levels of metallothionein in retinal pigment epithelial cells and macular degeneration. ^22,30–32 Accordingly, induction of metallothionein in the RPE directly, by plasmid transfection, and indirectly, by reduction of the adaptor protein p66Shc, has been shown to contribute to protect these cells against oxidative damage. ^31,32  

More recently, a report showed that metallothionein-I+II can also function in a paracrine manner, being secreted by astrocytes and thus regulating survival and regeneration of retinal ganglion cell axons. ^33,34 This effect is mediated through interaction with megalin, a member of the low-density lipoprotein receptor (LDLR) receptor family, which binds and takes up, not only lipoproteins, but also a range of other ligands, including metallothionein-I+II. ^35–37 In this article, we examine the expression patterns of megalin and of metallothionein-I+II in normal and degenerating retinas. 

Materials and Methods

Animals

All experiments were approved by the local committee for animal experimentation and ethics. Handling of animals was in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. The animals were kept on a 12-hour light–dark cycle, with no limitation of food or water. 

The examined animals included rd1 mice, rds mice, and wild-type (wt) mice (own colonies, homozygous, and C3H/HeA background), as well as pink-eyed, tan-hooded Royal College of Surgeon rats (RCS; rdy/rdy; own colony), and congenic wt RCS-rdy ⁺ rats (kindly provided by Olaf Strauss, Freie Universität, Berlin, Germany). Wild-type Sprague-Dawley (SD) rats were purchased from B&K Universal (Uppsala, Sweden) and Scanbur (Stockholm, Sweden). 

In the rd1 mouse, a mutation in the β subunit of the rod cGMP-phosphodiesterase gene (Pde6b) leads to a fast degeneration of rods during the first 3 weeks after birth. ³⁸ Homozygous retinal degeneration slow (rds) mice lack rod and cone outer segments because of a null mutation in the Prph2 gene (Prph2^Rd2 ) encoding peripherin, and photoreceptors are lost over a period of several months. ³⁹ In RCS rats, a mutation in the Mertk gene expressed in retinal pigment epithelial cells leads to the accumulation of photoreceptor outer segment debris in the subretinal space. ⁴⁰ Although the primary defect is not in a photoreceptor-specific gene, photoreceptors die within the first 3 months after birth. 

Microarray: Sample Isolation Procedures

Neural retinas from wt control and rd1 mice (number of replicates: wt/rd1), ⁴¹ were dissected at postnatal days (P) 5 (2/2), 6 (2/2), 7 (3/3), 8 (3/3), 9 (3/3), 11 (3/3), 12 (3/3), 13 (3/3), 15 (3/3), 21 (2/2), 28 (2/2), and 35 (2/2), resulting in 24 experimental conditions (12 time points × 2 strains). Total RNA was purified by sedimentation through cesium chloride. ⁴² Double-stranded cDNA was synthesized from 5 μg total RNA (Superscript Choice System; Invitrogen, Carlsbad, CA). The cDNA was then transcribed in vitro with an RNA transcript labeling kit (ENZO Diagnostics, Farmingdale, NY), to form biotin-labeled cRNA. 

GeneChip Hybridization and Scan

The labeled RNA was hybridized to mouse gene microarrays (MG 430 2.0 GeneChips; Affymetrix, Santa Clara, CA) using standard protocols. ⁴³ Quality control reports were generated as published elsewhere. ^44,45 The data were uploaded into Retinobase. ⁴⁶  

q-RT-PCR

Retinas of three to five animals per age group were isolated through a cut in the cornea, snap-frozen separately in liquid nitrogen, and stored at −80°C until further processing. The following ages were examined (the number of specimen per age appears in parentheses): rd1 mice at P7 (4), 11 (4), 14 (5), 21 (4), 28 (5), 60 (4), 120 (3) (total n = 29); rds mice at P7 (3), 14 (3), 17 (3), 21 (4), 28 (3), 60 (3), 120 (4), 280 (4) (total n = 27); wt mice at P7 (4), 11 (4), 14 (4), 17 (3), 21 (4), 28 (4), 60 (5), 120 (4), 280 (3) (total n = 35); RCS rats at P8 (3), 15 (3), 22 (4) 28 (4), 35 (4), 43 (4), 60 (4) (total n = 26); and SD rats at P8 (3), 15 (3), 22 (4), 28 (4), 35 (3), 43 (3), 60 (3) (total n = 23). 

After homogenization of the retinas, RNA was isolated (RNeasy MiniKit; Qiagen, Hilden, Germany), including a DNase treatment step to remove any contamination of genomic DNA, according to the manufacturer's instructions (RNase-Free DNase Set; Qiagen). RNA concentration was spectrophotometrically measured and its purity confirmed with 260-nm/280-nm absorbance ratios. RNA was reverse transcribed with a kit (QuantiTect Reverse Transcription Kit; Qiagen). 

Equal amounts of cDNA were applied for PCR amplification in triplicate in a thermocycler (LightCycler; Roche, Mannheim, Germany), using SYBR green master-mix (SYBR Green JumpStart Taq ReadyMix for qPCR Capillary Formulation; Sigma-Aldrich, Stockholm, Sweden) with a total reaction volume of 10 μL in each glass capillary (LightCycler Capillaries; Roche). The primers indicated in Table 1 were used in a final concentration of 0.25 μM. PCR conditions were set to 95°C for 10 minutes' initialization, followed by 45 cycles of 5 seconds' denaturation at 95°C, 8 seconds' annealing at 62°C, and 4 to 12 seconds' elongation at 72°C. Fluorescence from the SYBR green that binds to double-stranded DNA was measured at the end of each extension period. A calibrator sample, produced by mixing samples of different genotypes and ages from either mouse or rat tissue, provided a constant calibration point for all samples within and between runs. A software program (LightCycler Relative Quantification Software, ver. 1.01; Roche) automatically calculated ratios between calibrator-normalized target and reference. To verify the absence of genomic DNA contamination, control samples were run in which no reverse transcriptase was included. Also, controls without cDNA but only water were included in some experiments. Specificity of the PCR template was verified by melting-curve analysis. Subsequent gel electrophoresis assured amplification of only one PCR product of the expected size. A spreadsheet (Excel; Microsoft, Redmond, WA) was used for further data analysis and graphical visualization. Statistical evaluation was performed online (two-sample t-tests, F-test, one-way ANOVA, Tukey's HSD procedure for independent samples, http://faculty.vassar.edu/lowry/VassarStats.html). 

Table 1.

 

Primers Used to Amplify Mouse and Rat Metallothionein-I and -II, and β-Actin

Gene	Forward (5′–3′)	Reverse (5′–3′)	Size (bp)
β-Actin 47	CAACGGCTCCGGCATGTGC	CTCTTGCTCTGGGCCTCG	153
Mouse metallothionein-I 47	GAATGGACCCCAACTGCTC	GCAGCAGCTCTTCTTGCAG	104
Mouse metallothionein-II 47	TGTACTTCCTGCAAGAAAAGCTG	ACTTGTCGGAAGCCTCTTTG	94
Rat metallothionein-I 48	GCTGTGTCTGCAAAGGTGC	ATTTACACCTGAGGGCAGCA	82
Rat metallothionein-II (adapted from Ref. 47 )	GAATGGACCCCAACTGCTC	GCATTTGCAGTTCTTGCAG	94

Tissue Staining

Eyes were quickly enucleated and fixed with Bouin's solution (Sigma-Aldrich, St. Louis, MO) at 4°C overnight and thereafter washed several times with ethanol of increasing concentrations. After a last dehydration step in xylene, the eyes were embedded in paraffin. Sections (4–5 μm) were stained with hematoxylin and eosin. 

Some eyes were fixed with 4% paraformaldehyde in Sørensen's buffer for 2 hours at 4°C. The tissue was rinsed several times with the same buffer and thereafter cryoprotected by increasing concentrations of sucrose in the buffer. The eyes were embedded in an albumin-gelatin medium and frozen. Twelve-micrometer cryostat sections were collected on gelatin/chrome alum-coated glass slides and air-dried before storage at −20°C. 

Immunohistochemistry

Cryosections from one to seven animals of the following ages were used in the immunohistochemical studies: rd1 mice, P2 to P150 (total n = 53); rds mice, P2 to P274 (total n = 29); wt mice, P2 to P150 (total n = 37); RCS rats, P2 to P449 (total n = 50); RCS-rdy ⁺, P13 to P202 (total n = 4); and SD rats embryonic day (E)19 to P60 (total n = 15). 

The sections were preincubated for 60 to 90 minutes at room temperature in PBS-TBN: phosphate-buffered saline (PBS) containing 0.25% Triton X-100 (T), 1% bovine serum albumin (BSA), and 2% to 5% goat and/or donkey normal serum (N). Incubation with primary antibodies was performed overnight at 4°C and included an antibody that recognizes both metallothionein-I+II (monoclonal mouse anti-metallothionein E9, 1:50; Dako, Glostrup, Denmark); a polyclonal rabbit anti-glial fibrillary acidic protein (GFAP, 1:1500; Dako); or a polyclonal rabbit anti-megalin (1:50; Santa Cruz Biotechnology Inc., Santa Cruz, CA), all diluted in PBS-TBN. For colocalization studies, the metallothionein-I+II antibody was applied in combination with rabbit anti-cellular retinaldehyde binding protein (CRALBP, 1:5000 in PBS-TBN; kind gift from John C. Saari, University of Washington, Seattle, WA), monoclonal rat anti-mouse CD11b (1:75 in Tris-buffered saline [TBS] containing TBN; R&D Systems, Abingdon, UK), Alexa Fluor 594-conjugated isolectin B4 (IB4, 1:50; Molecular Probes, Eugene, OR) in buffer containing 300 mM NaCl, 100 μM CaCl₂, and 10 mM HEPES (pH 7.5), or tomato lectin (1:50 in PBS-TBN; Sigma-Aldrich). 

The sections were washed three times with PBS or TBS and incubated for 90 minutes with the corresponding secondary antibody at 1:200: biotinylated rabbit anti-rat (Vector Laboratories, Burlingame, CA) followed by streptavidin Cy3 (Jackson ImmunoResearch Laboratories, West Chester, PA); Texas red sulfonyl chloride-conjugated donkey anti-mouse (Jackson ImmunoResearch Laboratories); Alexa fluorescein isothiocyanate- or Alexa Fluor-conjugated goat anti-mouse or anti-rabbit (Molecular Probes). 

The specificity of the labeling was verified by preadsorbing the metallothionein-I+II antiserum with purified rabbit metallothionein protein (1:5 molar ratio for 4 hours; Sigma-Aldrich) before application to the sections. A secondary antibody control was also performed on some sections. 

TUNEL Assay

Dying cells were detected with a terminal deoxynucleotidyl transferase dUTP nick end- labeling (TUNEL) assay (TMR red In Situ Cell Death Detection Kit; Roche). Briefly, the enzyme solution was diluted 1:9 and the labeling solution 1:4 in PBS. The two components were mixed 1:4.44 immediately before application to the sections for 45 minutes at 37°C. The reaction was stopped by several washes with cold PBS before mounting. 

Oxidative Stress and Damage Assay

Oxidatively damaged DNA was detected by incubating retinal sections for 1 hour at room temperature with Texas red- or Alexa Fluor 488-conjugated avidin (10 μg/mL in PBS-TB, Molecular Probes Inc.). ^49,50  

In Situ Proximity Ligation Assay

The proximity ligation assay (PLA; Olink AB, Uppsala, Sweden) was applied to visualize the co-localization of metallothionein-I+II and megalin. In this assay, a signal is produced only when the oligonucleotide-labeled secondary antibodies (PLA probes) that have been applied bind to the corresponding primary antibodies (in the present case, metallothionein-I+II and megalin) in close proximity. ⁵¹ The PLA was performed according to the manufacturer's protocol, with the reagents provided. All incubation steps were performed in a humidity chamber at 37°C, except for the incubation with primary antibodies, which was done at 4°C. Briefly, the sections were blocked for 30 minutes before incubation with both primary antibodies (each 1:50) overnight. The slides were washed three times for 5 minutes each time with TBS-T before incubating with the secondary anti-mouse and anti-rabbit antibodies conjugated to unique oligonucleotides for 2 hours. A hybridization and a ligation step followed for 15 minutes each, with brief and careful washes between. The rolling-circle DNA amplification was applied for 90 minutes, and the product was labeled with a complementary DNA linker conjugated to a Tex613 fluorophore for 1 hour. The slides were washed with decreasing concentrations of SSC buffer (2×, 1×, 0.2×, 0.02×) and finally with 70% ethanol before being air-dried and mounted with the provided antifade medium. 

Other stained sections were mounted with antifade medium (Vectashield; Vector Laboratories) and examined with a fluorescence microscope (Axiophot; Carl Zeiss Meditec, Inc., Oberkochen, Germany). Images were taken with a digital camera and accompanying software (Axiovision 4.2; Carl Zeiss Meditec). 

Results

Gene Expression

A microarray analysis performed on samples from wt mouse retinas (P5–P35) showed a modest upregulation particularly of metallothionein-I with age, noted from P11. In rd1 mouse retinas, increased expression was observed from P12 to P13 and onward, peaking at ∼P21 for both genes (Fig. 1A). 

Figure 1.

 

Metallothionein-I and -II gene expression by microarray analysis (A) and q-RT-PCR (B–D). (A) Expression profiles of wt and degenerating homozygous rd1 retinas of corresponding ages. Bars represent mean ± SD. (B–D) q-RT-PCR analysis of metallothionein-I and -II expression at different ages in normal animals (wt mice, SD rats), in rd1 and rds mice, and in RCS rats. The bars represent the mean ± SEM of ratios created by comparing, at different ages, the expression of metallothionein-I and -II with that of β-actin in control and in degenerating retinas. Ratios of a calibrator sample were set to 1. Statistically significant differences between control and degenerating retinas for each age: *P < 0.05; **P < 0.01; ***P < 0.001. Differences between ages for each genotype are given in the text.

Gene induction was also verified by q-RT-PCR in retinas of different ages (Figs. 1B–D). This analysis confirmed that in wt animals, metallothionein-I and -II levels increase slightly with age. Significantly higher levels of metallothionein-I were noted in P21 and older mice than in P7 to P11 animals (P < 0.01); for metallothionein-II, a significant difference was seen between P7 and P11 and in animals P28 and older (P < 0.05). Levels of both metallothionein-I and -II were significantly higher in P43 and older normal rats than in younger animals (P8–P15; P < 0.05). 

The analysis showed also that in the two mouse models, metallothionein-I and -II gene levels increased relatively rapidly, whereas in RCS rats, an upregulation was not observed until weeks after the onset of photoreceptor degeneration, suggesting that cell death per se is not what triggered metallothionein-I and -II. In rd1 mouse retinas, significantly higher levels were found between P14 and P60, but not in older animals (Fig. 1B). In rds mice, altered expression was found from P14 (metallothionein-II) and P17 (metallothionein-I; Fig. 1C). At P280, the last time point examined, gene levels were still several times higher in rds retinas than in their wt counterparts (Fig. 1C). In RCS rats, an upregulation of metallothionein-I and -II was first observed in the fourth week of age (Fig. 1D). 

It should be noted that the method used to isolate the retina for gene analysis, although ensuring a quick dissection, does not control for the amount of RPE contamination in the sample, and therefore, the increases observed may not correspond only to stimulated expression in the retina. The immunohistochemical analysis did not allow us to conclusively establish whether metallothionein-I+II protein levels changed in the epithelium, but clearly indicated that they increase substantially with age in the degenerating neural retina. 

Avidin, TUNEL, GFAP, and Metallothionein-I+II

Wild-Type Mice.

In wt mouse retinas, no avidin staining was noted in the outer nuclear layer (ONL) at any of the time points examined. A small number of TUNEL-positive cells were seen in the ONL in the younger ages examined (P7 and P13), but none at P21 or onward. GFAP labeling was found along the vitreal margin of the retina and weakly in radial Müller cell processes in the far periphery (not shown). 

Immunolocalization of metallothionein-I+II resulted in positive staining in the RPE, in the inner plexiform layer (IPL), in sporadic cells with a morphology characteristic of microglia, and in retinal vessels (Fig. 2F). After preadsorption of the antiserum, labeling persisted in vessels. The latter were stained also after incubation of sections with secondary antibodies alone, so a specific staining could not be verified nor excluded. 

Figure 2.

 

Localization of various markers in wt (A, F) and rd1 (B–E, G–J) mouse retinas: All markers were detected in the rd1 retina at P12 to P13, when about half of the photoreceptor cells remain (E). Specific metallothionein-I+II immunoreactivity was observed in the RPE (F, I, J) and in microglia in the IPL (F, H, arrows). In rd1 retinas, labeling was also present in Müller cells as well as microglia (arrows) located in the ONL: (G) P12, central retina; (H) P13, midperiphery; (I) P17, midperiphery; (J) P21, far periphery. Labeling of vessels (F, *). HE, hematoxylin-eosin; MT, metallothionein. Scale bars, 20 μm.

rd1 Mice.

The number of avidin- and TUNEL-stained cells detected in the ONL was higher in rd1 retinas than in wt controls at P8 to P10, mainly in the central areas (Table 2). By P12 and P13, positive cells were found in all eccentricities (Figs. 2B, 2C) and at P14, the number of photoreceptor cell rows was reduced to less than half in the midperiphery (Fig. 2E) compared with age-matched wt controls (Fig. 2A). GFAP staining of radial Müller cell processes was noted throughout the retina from P11 onward (Table 2, Fig. 2D). 

Table 2.

 

Ages at Which the Various Markers Were First Detected

	Avidin	TUNEL	GFAP	Metallothionein-I+II
rd1	P8–P10	P8–P10	P11–P12	P11–P12
rds	P14–P16	P12–P14	P16–P18	P16–P18
RCS	P12–P14	P12–P14	P21–P25	P32–P35

 Avidin and TUNEL were detected in the ONL and GFAP and metallothionein-I+II in Müller cells and processes. The time periods indicated reflect individual differences and do not necessarily correspond to the time point when the markers were uniformly distributed throughout the retina.

Metallothionein-I+II-positive staining was observed also in the rd1 mouse in the RPE at all ages examined. Additional metallothionein-I+II expression was noted in Müller cell bodies and processes in the central retina, near the optic nerve head at P11 to P12 (Table 2, Fig. 2G). Cellular profiles resembling microglia were also found in the inner retina between P7 and P11 and in the ONL between P11 and P15 (Fig. 2H). Co-labeling with microglial cell markers, confirmed the identity of some of these metallothionein-I+II-positive cells (see Figs. 5G–I). 

Between P12 and P17, an increasing number of metallothionein-I+II-labeled Müller cells was observed toward the periphery (Fig. 2I). Staining was observed also in the innermost retina and could therefore correspond to Müller cell endfeet and/or astrocytes. 

By P21, the strongest signal was observed in Müller cell bodies and processes in the far periphery (Fig. 2J) and more weakly and in fewer numbers in the midperiphery and central retina. At later ages (P46, P150), only a few labeled Müller cells could still be observed, mainly in the far periphery and close to the optic nerve head (not shown). 

rds Mice.

Avidin-stained cells were observed in the ONL in rds mice from P14 to P16 and TUNEL-positive cells from P12 to P14 (Table 2). The number of stained cells declined between the peak at P16 to P18 (Figs. 3B, 3C) and at P270, at which point three to five rows of photoreceptor cells were still found (Fig. 3A). Uniform GFAP labeling of Müller cell radial processes throughout the retina was observed also at P18 and onward (Fig. 3D). 

Figure 3.

 

Localization of various markers in rds mouse (A–G) and RCS rat (H–N) retinas. At P270 in rds mice, about half of the photoreceptor cells remained (A), although all markers analyzed were already detected throughout the rds retina in the first postnatal weeks (B–E). Metallothionein-I+II staining was observed in the RPE and in Müller cell bodies and processes (E–F). In older animals, staining in Müller cells was restricted to the cell bodies (G). In RCS rats, about half of the ONL was lost by the first month of age (H). Avidin and TUNEL staining were observed relatively early (I, J), followed by GFAP (K) and metallothionein-I+II (L, M). The latter was initially observed in sporadic Müller cell bodies and processes (L) and in a large number of cells within a few days (M). Occasionally, staining was observed in the subretinal space (arrows; L, M). Metallothionein-I+II staining was observed also in older animals in a small number of Müller cells (N). HE, hematoxylin-eosin; MT, metallothionein; SRS, subretinal space. Scale bars, 20 μm.

Figure 4.

 

Localization of megalin immunoreactivity in normal and degenerating retinas. In the young normal mouse retina (A), megalin immunoreactivity was most prominent in the subretinal space and becomes more concentrated over the photoreceptor outer segments with age (B). In older animals, the most proximal processes of Müller cells were also labeled (B). In rd1 and rds mice, staining in the outer retina was initially weaker and, at older ages, was absent (C–F). In the latter, Müller cell processes were also labeled (D, F). In normal rat retinas (rdy ⁺), megalin expression was also detected over the photoreceptor inner and outer segments (G). In RCS rats, labeling in the outer retina was lost with age (H, I). OS, outer segments; IS, inner segments. Scale bars, 20 μm.

Metallothionein-I+II expression was observed in the RPE in rds mouse retinas at all ages examined. In addition, labeled Müller cells were noted in rds mouse retinas, but not until P16 (Table 2). Yet, by P18, expression could already be observed throughout the whole retina in Müller cell bodies and processes (Fig. 3E). This distribution was still observed in older animals (Fig. 3F), although in 9-month-old animals, labeling was restricted to the cell bodies only (Fig. 3G). A few microglial cells appeared also labeled in the younger ages (not shown). 

Normal Rats and RCS Rats.

Retinas obtained from SD and from congenic RCS-rdy ⁺ rats processed for the avidin and TUNEL assays showed no specific labeling of photoreceptor cells. GFAP staining was limited also in these specimens to the innermost retina (not shown). Avidin- and TUNEL-stained cells were observed in the ONL in RCS rats from P12 onward (Table 2, Figs. 3I, 3J). GFAP labeling of Müller cell radial processes was observed from ∼P21 onward (Table 2, Fig. 3K). 

Metallothionein-I+II expression was found in control rat retinas in the RPE and in vessels, as seen with normal mouse retinas (not shown). In RCS rats, labeling of Müller cells and processes was also observed, but not until P32 (Table 2). At this age, less than half of the photoreceptor layer remains (Fig. 3H). Within a few days, staining was seen throughout the retina with labeled processes extending beyond the outer limiting membrane, into the subretinal space (Figs. 3L, 3M). In older animals (P145, P449), expression of metallothionein-I+II was still observed but only in a few sporadic cells, with accumulation of labeling also in the subretinal space in some places (Fig. 3N). 

Megalin Expression

In younger normal mouse retinas (P6–P40), megalin staining revealed labeling in the subretinal space and in the nerve fiber layer (NFL). By P13, expression in the outer retina was still more prominent over the photoreceptor inner segments (IS; not shown), whereas at P29 and P40, strong staining was noted also over the outer segments (OS; Fig. 4A). In older animals, staining was strongest over the OS and was in addition observed in astrocytes and/or Müller cell endfeet and their ascending processes (Fig. 4B). 

In rd1 mice, staining of photoreceptor segments was initially detected, but was significantly reduced by P15 and absent at P25 (Fig. 4C). Strong staining was noted also in the inner retina, but at earlier ages than in normal retinas (Fig. 4C), and at P46 labeling of proximal Müller cell processes was already very prominent (Fig. 4D). In rds mouse retinas, labeling of IS was still detectable at P14, accompanied by distinct labeling in the innermost retina (Fig. 4E). At later ages, staining in the outer retina was no longer visible, whereas sporadic Müller cell processes appeared labeled (Fig. 4F). 

In young normal rat retinas (SD and rdy ⁺), megalin expression was found in the subretinal space, in the NFL, and in the plexiform layers (not shown), but decreased in the latter with age (Fig. 4G). As opposed to normal mouse retinas, staining of Müller cell processes was not observed in older animals up to P202. In retinas of RCS rats, megalin labeling was initially undistinguishable from that in normal rats. With age, staining over the photoreceptor segments disappeared completely (Figs. 4H, 4I). 

Co-localization

Most cell bodies and processes expressing metallothionein-I+II within the retina were also CRALBP positive (Figs. 5A–C). Co-labeling with the two markers was also observed in the RPE in all retinas analyzed (Figs. 5D–F). 

Figure 5.

 

Colocalization of various markers (A–I) and in situ PLA assay (J–N). (A) Metallothionein-I+II staining, rds mouse, P21; (B) CRALBP staining, rds mouse, P21; (C) merged image showing metallothionein-I+II accumulation in Müller cells and processes; (D) metallothionein-I+II staining, rd1 mouse, P15; (E) CRALBP staining, rd1 mouse, P15; (F) merged image showing metallothionein-I+II accumulation in the RPE; (G) metallothionein-I+II staining, rd1 mouse, P15; (H) IB4 staining, rd1 mouse, P15; (I) merged image showing metallothionein-I+II accumulation in microglial cells in the ONL (arrows); accumulation of IB4 in vessels co-localizes with unspecific metallothionein-I+II staining (*); (J–N) In situ PLA assay including DAPI nuclear staining (blue). Dots represent putative sites of megalin/metallothionein-I+II interaction (red): (J) wt mouse, P13: signal in the inner retina and in the subretinal space, over the photoreceptor inner and outer segment region; (K) rd1 mouse, P13: signal in the nerve fiber layer and reduced signal in the subretinal space; (L) rd1 mouse, P29: reduction of signal both in the inner and outer retina; (M) rds mouse, P18: signal in the inner retina and in a thin band in the subretinal space; (N) rds mouse, P110: signal in the inner retina and diffusely distributed in the subretinal space. MT, metallothionein; CRALBP, cellular retinaldehyde binding protein; IB4, isolectin B4; SRS, subretinal space. Scale bars, 20 μm.

Processing of mouse retinas with CD11b, isolectin IB4, or tomato lectin resulted in labeling of vessels (Figs. 5G–I). In addition, co-staining with these markers showed that metallothionein-I+II labeled structures in the ONL of rd1 and rds mouse retinas corresponded to microglial cells (Figs. 5G–I). Metallothionein-I+II labeling of cells resembling microglial cells was also noted in the IPL in normal (Fig. 2F) and degenerating (Fig. 2H) mouse retinas. 

Co-localization of metallothionein-I+II and megalin was assessed in retinal samples using the in situ PLA assay. A distinct and consistent signal was observed in normal mouse and rat retinas in the subretinal space, over the photoreceptor IS and OS and in the ganglion cell layer (GCL) and NFL at the ages examined (P13–40; Fig. 5J). This distribution was observed in all three models of degeneration in younger ages (Figs. 5K, 5M), but in older animals, signal in the outer retina was absent or reduced to a narrow band next to the RPE (Figs. 5L, 5N). 

Discussion

The present study confirmed previous reports showing that metallothioneins are expressed in mouse and rat retinas and pigment epithelium, ^{20,24,26–29,32} and showed in addition: (1) that adult expression levels of metallothionein-I and -II are not reached in normal retinas until around P21 or later; (2) that metallothionein-I+II are induced in glial cells in the three models of hereditary photoreceptor degeneration studied; and (3) that the receptor megalin is expressed not only in the inner retina, as previously shown, ³⁴ but also in the outer retina, and that this latter expression is lost in degenerating retinas. 

A small but significant increase in metallothionein-I and -II gene expression levels was observed during the first postnatal weeks both in mice and rats, which could correlate with the structural and functional maturation of the retina that takes place during this period. Expression of these metallothioneins is tightly regulated by levels of, for example, growth factors and cytokines, many of which are notably active during early postnatal retinal development. Increases in expression could also reflect an increased availability of free zinc. Metallothioneins are induced not only to protect cells against potentially toxic high zinc levels, but seem to be expressed also in association with increases in physiological zinc. Chelatable zinc and zinc transporters are observed in association with glutamatergic synapses ⁵² and co-release of vesicular zinc with glutamate has been shown to occur from photoreceptor terminals at the level of the outer plexiform layer (OPL). ^53,54 In addition, high levels of zinc have been noted around the photoreceptor IS in light-adapted retina. ^54,55 Full maturation of both these retinal regions occurs around the second postnatal week in a center-to-periphery gradient, ⁵⁶ thus correlating with the increasing levels of metallothionein-I and -II detected, and in agreement with observations made in the brain. ⁵⁷ However, no differences were noted in the present study in the expression of metallothionein-I+II in developing retinas. There is the possibility that small increases in protein levels escaped detection with the antibody used. Alternatively, the increase could correspond to expression in retinal vessels, which also develop in rodents during the same period, ⁵⁸ as it has been shown that endothelial and smooth muscle cells also express metallothionein. ⁵⁹ The specificity of metallothionein-I+II labeling in the retinal vessels structures could not be verified in the present study and needs therefore to be examined further. 

A marked increase in both gene and protein expression was seen, however, in the three models of photoreceptor degeneration. The microarray and q-RT-PCR analyses showed that levels of metallothionein-I and -II are significantly upregulated in rd1 mice at P12, and that protein expression can also be observed at this stage in the central retina in these mice. The increase in expression followed the center-to-periphery pattern of progression normally observed in this model, but increased protein expression was still seen in retinas of P46 and P150 animals, when only a few degenerating cones remained. ^60,61 The upregulation of metallothionein-I+II appears to correlate with glial activation rather than with the onset of photoreceptor cell death as such, as the latter was detectable by TUNEL staining as early as P9. Further, metallothionein-I+II expression was induced mainly in Müller cells and coincided with the onset of GFAP expression, one of the earlier indicators of activation of these cells. ⁶² Such a distribution corresponds to what has been observed after neuronal damage in various models of CNS injury, where it is seen that metallothionein-I+II accumulate mainly in activated astrocytes, the main source of these metallothioneins in the brain (see recent reviews in Refs. 19 and 34). 

The accumulation of metallothionein-I+II observed here in Müller cells does not, however, fully correspond to observations made with a model of light-induced damage. ²⁷ In the latter, an upregulation of metallothionein-I+II was noted in the RPE and in the plexiform layers immediately after a 7-hour exposure to bright light, in the ONL and GCL after 8 hours, and only in the inner retina 28 hours after exposure, ²⁷ with no obvious expression in the Müller cells. Bright light exposure causes massive, immediate damage to photoreceptors, which involves, at some point, elevated intracellular calcium levels and an accumulation of reactive oxygen species, which contribute to oxidative damage. ⁶³ Most photoreceptors die relatively fast in rd1 mice, compared with those in rds mice or RCS rats. In rd1 retinas, impaired cGMP-phosphodiesterase activity, ³⁸ supposedly also leads to deregulated Ca²⁺ homeostasis, as suggested by observations that activation of calpain is increased in these retinas ^64,65 and that application of a Ca²⁺ blocker confers protection. ⁶⁶ Others and we in this study have found evidence that the rd1 mouse photoreceptors also undergo oxidative damage. ^50,61 Nonetheless, no accumulation of metallothionein-I+II was seen in the present study in photoreceptors in rd1 mice, as reported to occur after light damage. 

As mentioned earlier, metallothionein-I+II are primarily of glial origin, and accumulation in photoreceptors would therefore require that the metallothioneins be transferred to the latter. Transport of glial-derived metallothionein into retinal ganglion cells has recently been shown to occur, where it was found to modulate axonal regeneration. ³⁴ It is possible, thus, that the events initiated in rd1 mouse photoreceptors are still not sufficient to induce significant metallothionein synthesis and accumulation in these cells or that they actually die before detectable amounts of metallothionein can be found. 

However, as found in the present study, metallothionein-I+II also did not accumulate in photoreceptors of homozygous rds mice. In these animals, a mutation of the gene encoding peripherin leads to the absence of rod photoreceptor disc membranes and the development of abnormal cone outer segments. ^39,67 As a result, the normal compartmentalization of proteins involved in phototransduction does not occur, altering expression of at least some genes. ⁶⁸ The inner segments appear to develop normally but are immediately adjacent to the microvilli of the RPE. ³⁹ We found the first TUNEL-stained cells in rds retinas at P12 and P14, with the largest number seen at ∼ P16 and P18, which agrees with previous reports. ^69,70 Despite this early peak of cell death, complete loss of photoreceptors was seen only at ∼12 months of age. ³⁹ Yet, no metallothionein-I+II was detected in photoreceptors in this slow-progressing model, but was seen to correlate rather with some stage of glial activation, as indicated by concomitant accumulation of GFAP in the Müller cells. 

In RCS rat retinas, overexpression of metallothionein-I+II was similarly seen in Müller cells. Yet, it was not detected until ∼10 days after the onset of GFAP accumulation in these cells. In RCS rats, phagocytosis of shed outer segments is not performed due to a mutation in the Mertk gene expressed by retinal epithelial cells. ⁴⁰ As a consequence, membrane debris accumulate in the subretinal space, eventually leading to secondary photoreceptor cell death, which was first detectable by TUNEL staining at ∼P12 in the present study. We found that upregulation of mRNA levels of metallothionein-I and -II was also delayed in RCS rat retinas, indicating that the late expression of metallothionein-I+II is not due to slow protein synthesis. 

The present study thus shows that metallothionein-I and -II are induced in the three models of degeneration. The upregulation did not coincide with the onset of photoreceptor cell loss, but rather with the widespread accumulation of GFAP in Müller cells, at least in rd1 and rds mice. In RCS rats, it was considerably delayed in relation to expression of GFAP. Accumulation of the GFAP is normally associated with the process of Müller glia activation, ^62,71 although it may not necessarily reflect the time when activation is initiated. We showed, for instance, in previous studies that glial-derived molecules accumulate in RCS rat retinas long before GFAP upregulation can be detected and even before signs of photoreceptor cell death. ⁷² Further, although GFAP accumulation occurred in Müller cells in the three models, previous studies have shown, for instance, that the downregulation of inwardly rectifying K⁺ (Kir) currents displayed by Müller cells and which normally accompanies fast neuronal degeneration is not observed in RCS rats or rds mice. ^73,74 Upregulation of metallothionein-I+II, as other features of Müller cell activation, may therefore not correlate always with the same specific events triggered in these cells. 

Furthermore, like other glial-derived factors, metallothionein-I+II levels can be stimulated without neuronal cell death. It is therefore possible that upregulation is not a response to cell death, but to changes in one or several factors accompanying or even preceding cell death, where the type and degree of change determine the onset of metallothionein-I+II expression. Previous studies have shown that the levels of free glutamate are increased as photoreceptors degenerate and that uptake and metabolism of glutamate by Müller cells is impaired, at least in some cases. ^75,76 As mentioned, zinc is co-released with glutamate in several neuronal structures, including photoreceptor terminals. ^77,78 Detectable levels of zinc have in fact been found, not only in the terminal region, but throughout the rod photoreceptor cell. ⁷⁷ It could thus be speculated that zinc leaking from dysfunctional/dying photoreceptors is one of the initial triggers of metallothionein-I and -II expression. Zinc transporters (ZnT) are localized throughout the retina, including the photoreceptor inner segment region and the apical processes of Müller cells. ^79–81 An upregulation of ZnT was not seen by microarray analysis in rd1 mouse retinas in young ages (P11–P15, P21), when some layers of photoreceptor cells remain (not shown). Yet, at older ages (P28, P35), no reduction in expression was observed, despite the almost complete loss of the photoreceptor cell layer, suggesting that there may indeed be a relative increase over time in the number of zinc transporters expressed by Müller cells and/or other cells in the inner portions of the retina. This would correlate with our observation that increased metallothionein-I+II expression was still observed in retinas of rd1 mice and RCS rats long after most photoreceptors had been lost. Moreover, it is possible that accumulation of other metals, in addition to zinc, may induce increases in metallothionein-I+II expression. In RCS rats, the concentrations of iron and copper are also increased in the outer retina. ^82,83 Although it is not clear how these increases are produced, their accumulation in the subretinal space could contribute to the triggering of metallothionein-I+II upregulation in Müller cells. 

A critical question is also whether the increases noted in endogenous metallothionein-I+II levels actually contribute to prolonged photoreceptor survival in the course of degeneration and how protection would be mediated. The endocytic receptor megalin has been recently shown to mediate the internalization of exogenous and glial-derived metallothionein-I+II into neurons, including retinal ganglion cells. ³⁴ The present study confirmed that megalin is expressed in normal retinas in the NFL and showed that it is also expressed in the IS and OS region of the retina at all ages analyzed. This distribution correlates well with the sites of metallothionein-I+II expression and the concept that the latter mediates its functions, at least in part, through interactions with megalin. ^33,34 Of note, the pattern of megalin expression changed with age in the normal retina, and accumulation was eventually observed also in Müller cells, at least in mouse retinas. As megalin mediates the transport of not only metallothioneins but of several other ligands, ^35,36 this could explain its localization in Müller cells in the normal retina, even in the absence of metallothionein-I+II. 

The expression of megalin observed in the outer retina in normal mice and rats was lost very early in degenerating retinas (confirmed by a lack of signal in this region using the PLA assay) and was probably due to the loss of the outer and inner segments as photoreceptors degenerate. In the degenerating mouse retinas, megalin expression was instead seen in the processes of Müller cells at younger ages than in normal animals. This expression did not correlate with the increases in metallothionein-I+II expression in Müller cells and may instead be a consequence of the loss of megalin expression in the subretinal space region. Cubulin, another multiligand endocytic receptor is found to be co-expressed with megalin in epithelial cells where they mediate the reabsorption of retinol-binding proteins, for example. ⁸⁴ Defects of chloride channels/transporters have been shown to cause retinal degeneration, ^85,86 possibly involving alterations in megalin and cubilin function. Metallothionein-I+II could thus be transported by either megalin or cubilin receptors in the outer retina. ⁸⁷ However, the activity of cubilin has been shown to depend on the presence and activity of megalin. ⁸⁸ It appears thus that if metallothionein-I+II were to exert a protective effect on photoreceptors, it would not involve megalin- or cubilin-mediated transport of metallothioneins in the outer retina, which could also explain why we found no metallothionein-I+II labeling in the photoreceptors. 

Inability to produce endogenous metallothionein-I+II has been shown to exacerbate ganglion cell loss after intraocular injections of N-methyl-d-aspartate, ²⁶ but not the loss of photoreceptors induced by hyperbaric oxygen exposure, ⁸⁹ suggesting that the dependence on metallothionein for survival varies with the type of insult and/or the type of cell affected. Moreover, it is possible that increases in metallothionein-I+II levels occur, not to protect photoreceptors, but to support the various events mediated by glial cells, secondary to photoreceptor degeneration and/or to support the glial cells themselves. The activity and function of Müller cells is regulated by several factors, including cytokines such as IL-6, acting through signal transducer and activator of transcription 3 (STAT3). ⁹⁰ The latter targets several genes, including metallothionein-I and -II, ^91,92 suggesting that upregulation of these metallothioneins is both a cause and an effect of glial activation. The same may apply to RPE and microglial cells, which we found also to express metallothionein-I+II and which react to damage and injury by upregulating various cytokines. ^93,94  

In summary, although the initial triggers of photoreceptor dysfunction differ in the three models studied, common processes are eventually activated, including an upregulation of metallothionein-I+II. The onset of their expression did not coincide with that of cell death (nor of glial activation in RCS rats), suggesting that stimulation of these metallothioneins is controlled, at least in part, by different events in the three diseases. As to the role of metallothionein-I+II in photoreceptor degeneration, one might argue that cell loss progresses in the three models studied despite upregulation of metallothioneins and of several other neuroprotective factors. However, whether endogenous metallothioneins actually protects photoreceptors would have to be verified in genetic experiments. 

Footnotes

 Supported by European Union Grants LSHG-CT-2005-512036 and MEST-CT-2005-020235, Foundation Fighting Blindness, Swedish Medical Research Council Grant 12209, Crown Princess Margareta's Committee for the Blind, Stiftelse för Synskadade i f.d. Malmöhus Län, Crafoordska Stiftelsen, Torsten och Ragnar Söderbergs Stiftelser, and Thorsten och Elsa Segerfalks Stiftelse (M-TP); Institut National de Santé et de Recherche Médicale, Agence Nationale pour la Recherche, and the European Commission (TL); and Kerstan-Stiftung (KAW).

Footnotes

 Disclosure: K.A. Wunderlich, None; T. Leveillard, None; M. Penkowa, None; E. Zrenner, None; M.-T. Perez, None

The authors thank Markus Thiersch, Marijana Samardzija (Department of Ophthalmology, University of Zurich, Switzerland), and Kenneth Beri Ploug (Department of Neurology and Danish Headache Center, Glostrup, Denmark) for help with the q-RT-PCR protocols; Hodan Abdalle for technical support; and Karin Arnér and Birgitta Klefbohm for assistance with the animal colonies.