The X‐linked juvenile retinoschisis protein retinoschisin is a novel regulator of mitogen‐activated protein kinase signalling and apoptosis in the retina

Abstract X‐linked juvenile retinoschisis (XLRS) is a hereditary retinal dystrophy in young males, caused by mutations in the RS1 gene. The function of the encoded protein, termed retinoschisin, and the molecular mechanisms underlying XLRS pathogenesis are still unresolved, although a direct interaction partner of the secreted retinoschisin, the retinal Na/K‐ATPase, was recently identified. Earlier gene expression studies in retinoschisin‐deficient (Rs1h −/Y) mice provided a first indication of pathological up‐regulation of mitogen‐activated protein (MAP) kinase signalling in disease pathogenesis. To further investigate the role for retinoschisin in MAP kinase regulation, we exposed Y‐79 cells and murine Rs1h −/Y retinae to recombinant retinoschisin and the XLRS‐associated mutant RS1‐C59S. Although normal retinoschisin stably bound to retinal cells, RS1‐C59S exhibited a strongly reduced binding affinity. Simultaneously, exposure to normal retinoschisin significantly reduced phosphorylation of C‐RAF and MAP kinases ERK1/2 in Y‐79 cells and murine Rs1h −/Y retinae. Expression of MAP kinase target genes C‐FOS and EGR1 was also down‐regulated in both model systems. Finally, retinoschisin treatment decreased pro‐apoptotic BAX‐2 transcript levels in Y‐79 cells and Rs1h −/Y retinae. Upon retinoschisin treatment, these cells showed increased resistance against apoptosis, reflected by decreased caspase‐3 activity (in Y‐79 cells) and increased photoreceptor survival (in Rs1h −/Y retinal explants). RS1‐C59S did not influence C‐RAF or ERK1/2 activation, C‐FOS or EGR1 expression, or apoptosis. Our data imply that retinoschisin is a novel regulator of MAP kinase signalling and exerts an anti‐apoptotic effect on retinal cells. We therefore discuss that disturbances of MAP kinase signalling by retinoschisin deficiency could be an initial step in XLRS pathogenesis.


Introduction
Pathogenic alterations affecting the RS1 gene on chromosome Xp22.1 have been shown to cause XLRS (OMIM #312700) [1], a macular degeneration disorder in young males with a prevalence of approximately 1:5000 to 1:20,000 [2]. Disorganization of retinal layers and distinct abnormalities in the electroretinogram (ERG) are hallmarks of the disease. Specifically, a characteristic splitting of retinal layers, presenting as a bilateral foveal schisis, is found at an early stage of the disease and results in cystic degeneration of the central retina [3][4][5][6]. Additionally, defects in signal transmission from photoreceptor to bipolar cells as visualized by ERG recordings are observed and reveal a characteristic reduction in the b-wave amplitude, whereas the a-wave remains almost unaffected [4,7]. Comparable pathological features are also evident in XLRS mice, generated via a targeted disruption of the murine orthologue of RS1, the Rs1h gene [8][9][10]. Due to the close resemblance of the retinal phenotype in Rs1h knockout mice and XLRS patients, the retinoschisin-deficient mouse represents an excellent disease model widely used in experimental studies addressing the mechanisms of XLRS pathology but also novel treatment approaches [11][12][13][14][15][16].
The RS1 gene is organized into six exons and encodes a 224amino acid (aa) precursor protein [1]. It is specifically expressed in the retina by photoreceptor and bipolar cells, as well as in pinealocytes of the pineal gland [1,17,18]. During protein synthesis, a 23aa signal sequence is cleaved to produce a 201-aa mature polypeptide which is secreted from photoreceptors and bipolar cells as a homooctamer held together by intermolecular disulphide bonds between aa 223 and aa 59 [19][20][21][22]. So far, over 190 unique XLRS-associated sequence variants in RS1 have been reported (Leiden Open Variation Database, http://grenada.lumc.nl/LOVD2/eye/home.php?select_db= RS1, accessed May 2016). Functional assessment of a subset of these variants demonstrated that the vast majority of mutations result in a complete loss of the functional protein [4].
Despite intensive research, the precise molecular function of retinoschisin remains unresolved. Searching for retinoschisin interaction partners, Molday et al. [14] identified the retina-specific Na/K-ATPase composed of the two subunits ATP1A3 (a3) and ATP1B2 (b2). Subsequently, our group confirmed the Na/K-ATPase to be required for anchoring retinoschisin to plasma membranes [23]. The Na/K-ATPase is a plasma membrane spanning ion pump, responsible for maintaining the cellular membrane potential by transporting Na + and K + ions across the plasma membrane against their electrochemical gradient [24,25]. Despite this essential task, Na/K-ATPases also mediate intercellular adhesion [26][27][28] and induce activation of intracellular signalling pathways upon binding of glycoside hormones such as ouabain [25,[29][30][31][32][33][34]. Members of the FXYD family, a class of Na/K-ATPase-binding proteins [35,36], were reported to be important regulators of the Na/K-ATPase, modulating its pump activity and mediation of intercellular adhesion [37][38][39]. Similar to FXYD proteins, one could consider retinoschisin to exert a role as a modulator of Na/K-ATPase activity.
A genomewide expression analysis of the Rs1h-deficient (Rs1h À/Y ) murine retina first indicated an increased activation of the ERK pathway in early XLRS pathogenesis, prior to apoptotic photoreceptor degeneration [40]. The ERK pathway is one of the four major MAP kinase pathways [41,42] known to play a crucial role in fundamental developmental and physiological processes such as apoptosis, neuroprotection, neuronal development and adhesion [41,[43][44][45][46][47][48][49][50]. It is tempting to speculate that misregulation of MAP kinase signalling caused by retinoschisin deficiency could be an initial step in XLRS pathogenesis. However, aberrant MAP kinase activation could also be a secondary event, caused by alterations of the cellular/retinal homeostasis in the XLRS disease process.
In this study, we examined whether retinoschisin binding to retinal membranes directly modulates MAP kinase signalling. Our findings in cultured Y-79 cells and in retinal explants of Rs1h À/Y mice demonstrate that the addition of recombinant retinoschisin, but not recombinant mutant retinoschisin, significantly down-regulates MAP kinase signalling, as well as protects against apoptosis. We conclude that retinoschisin deficiency could be a trigger for disease pathogenesis by a defective control of MAP kinase signalling and apoptosis in the retina.

Animal models
The Rs1h À/Y mouse was generated as described earlier [9] and kept on a C57BL/6 background. Mice were housed under specific pathogen-free barrier conditions at the Central Animal Facility of the University of Regensburg and maintained under conditions established by the institution for their use, in strict compliance with NIH guidelines. Mice were sacrificed 10 or 18 days after birth by decapitation or cervical dislocation after inhalation of carbon dioxide, respectively.

RNA analysis
RNA was isolated from cell lines using the Qiagen RNeasy Mini Kit (Qiagen, Venlo, the Netherlands). RNA from murine retinae and cultured retinal explants was isolated using the PureLink TM RNA Micro Kit (Invitrogen), according to the manufacturers' protocols. One microgram of total RNA was transcribed into cDNA using RevertAid M-MuLV Reverse Transcriptase (Fermentas, St Leon-Rot, Germany) and poly(dT) primers according to the manufacturer's instructions. Semiquantitative RT-PCR was performed as described by [23], with primers given in Table S1. Quantitative real-time RT-PCR was performed and analysed as published [51] with primers given in Table S1.

Expression cloning
The coding sequence of non-mutant retinoschisin (NM_000330.3) was amplified from cDNA of retinal tissue using oligonucleotide primers containing a EcoRI restriction site at the 5 0 end and a XhoI restriction site at the 3 0 end of the RS1 coding sequence (for primer sequences, see Table S1). The coding sequence of the XLRS-associated RS1 mutant RS1-C59S (NM_000330.3 (RS1):c.175T>A [p.Cys59Ser] http://grenada.lumc.nl/LOVD2/eye/home.php?select_db=RS1) was generated by sitedirected mutagenesis on the retinoschisin coding sequence (primer sequences shown in Table S1). For purification, the two RS1 variants were each tagged with an N-terminal Myc tag, following the leader sequence (after aa 23). This peptide insertion into the full-length RS1 coding sequence was performed by fusing the N-terminal part of both RS1 coding sequences from positions 1 to 69 (aa  to the N-terminal half of the Myc tag sequence (for primers, see Table S1). The Cterminal part of both RS1 coding sequences (aa 70 to stop codon) was fused to the C-terminal half of the Myc tag sequence (for primers, see Table S1). C-and N-terminal RS1 fragments were ligated via a BclI restriction site which was introduced into the Myc tag, and inserted into the pCDNA3.1 TM expression vector (Thermo Fisher Scientific).

Expression and purification of recombinant RS1 variants
Expression constructs were transfected into Hek293 cells using calcium phosphate transfection as described by [53]. About 7 hrs after transfection, the culture medium was replaced by Opti-MEM â containing 100 U/ml penicillin/streptomycin (Life Technologies) and cells were cultured for additional 48 hrs.
Myc-tagged retinoschisin and RS1-C59S were isolated from cultivation media by immunoprecipitation using Pierce TM Anti-c-Myc Agarose (Thermo Fisher Scientific) according to the manufacturer's instructions. Concentrations of purified proteins were determined using the Bio-Rad DC TM Protein Assay Kit (Bio-Rad Laboratories).
For use as a treatment control, Hek293 cells were transfected with empty pCDNA3.1 TM expression vector, and cultivation medium of these cells was subjected to purification procedure exactly like medium from cells transfected with RS1 variants.
Purity of purified Myc-tagged RS1 proteins and control eluate was verified via silver staining, Coomassie Blue staining and Western blot analysis using antibodies against the Myc tag and against retinoschisin (Fig. S1).

Binding of RS1 protein variants to membranes
Retinoschisin binding to adherent cell lines (BV-2, ARPE and Hek293) as well as to murine retinal membranes (P10) was assessed as described by [23], but with a prolonged incubation time of 1 hr.
For comparing binding affinity of retinoschisin and RS1-C59S to Y-79 cells and Rs1h À/Y murine retinal membranes, 6 lg of purified retinoschisin or RS1-C59S was added to 5 ml cultivation medium and incubated for 10, 30 and 60 min. Subsequent steps were performed as above.
For localization of bound recombinant RS1 variants on Rs1h À/Y murine retinae, Rs1h À/Y murine retinal explants were incubated with 1 lg RS1, RS1-C59S or control eluate, as described in signalling experiments. After 30 min. of incubation, the retinal tissue was washed with PBS once before immunolabelling of retinal cryosections was performed.

Analysis of signalling pathways in retinal explants and Y-79 cells
Y-79 cells were grown to a concentration of 4-5 9 10 5 cells/ml in 10 ml medium. The experiment was started by adding 1 lg purified retinoschisin or RS1-C59S, or equal volume of control eluate. After 10 or 30 min. of incubation at 37°C, cells were harvested by centrifugation. For Western blotting, cells were resuspended in 200 ll of pre-cooled PBS with Phos-STOP TM phosphatase inhibitor (Sigma-Aldrich) and lysed by sonication (10 sec., 40% intensity). For RNA isolation, cells were washed once with pre-cooled PBS before they were subjected to RNA isolation.
Eyes from Rs1h À/Y mice at post-natal day 10 were enucleated and retinal explants were dissected as described by [54]. Retinae were incubated in 800 ll DMEM/Ham's F12 containing 10% FCS, 100 U/ml penicillin/ streptomycin, 2 mM L-glutamine and 2 lg/ml insulin (Thermo Fisher Scientific). One microgram purified retinoschisin or RS1-C59S or equal volumes of control eluate were added. After 10 or 30 min. of incubation at 37°C, retinal explants were removed from medium and transferred to 200 ll of pre-cooled PBS containing PhosSTOP TM phosphatase inhibitor. For Western blot analysis, retinal explants were sonicated for 10 sec. at 40% intensity. For RNA isolation, retinal explants were immediately transferred into lysis buffer (PureLink TM RNA Micro Kit; Invitrogen).

Analysis of caspase-3 activity in Y-79 cells
About 2 9 10 6 cells/well were seeded onto poly-L-lysine-coated 24-well plates. Cells were allowed to adhere overnight before 0.1 lg of purified retinoschisin or RS1-C59S, or equal volumes of control eluate were added to 1 ml medium per well. After 1 hr, the culture medium was changed to 1 ml RPMI (containing RS1 variants or control eluate as before) and 0.2 mM H 2 O 2 to induce apoptosis or 0 mM H 2 O 2 as control. After 2 hrs, the medium was replaced by 1 ml RPMI containing RS1 variants or control eluate as before. Cells were allowed to recover for 18 hrs before they were subjected to a caspase-3 activity test using the EnzChek â Caspase-3 Assay Kit #2 by Thermo Fisher Scientific according to the manufacturer's directions.

Analysis of photoreceptor degeneration in retinal explants
Eyes were enucleated from mice at post-natal day 18 and retinae were dissected as described before [55]. Five retinae each were subjected to treatment with retinoschisin, RS1-C59S and control protein. Retinal explants were transferred into pre-warmed medium (DMEM/Ham's F12 containing 10% FCS, 100 U/ml antibiotic-antimytotic, 2 mM L-glutamine and 2 lg/ml insulin, all from Life Technologies) immediately after preparation, rinsed once in pre-warmed medium and then transferred onto Track Etch Membrane Filters (Whatman plc, Maidstone, UK) in 35mm tissue culture dishes containing 3 ml of medium to which 1 lg of purified retinoschisin, RS1-C59S or control eluate had been added. The retinal explants on the filters were covered in a small droplet of medium. Cultivation was carried out under sterile conditions in a 37°C incubator with a 5% CO 2 environment. Medium was replaced every 36 hrs and total cultivation time was 1 week. Subsequently, the retinal explants were fixed, cryopreserved and cut into 10-lm sections for subsequent histological analyses as described before. After PNA staining, cones were counted in each two different sections of the same retina, in 200nm-wide regions to the left and the right side of the optic nerve. Rods were stained with anti-Rho-1D4 antibody. Staining signals of each two different sections of the same retina, in 200-nm-wide regions to the left and the right side of the optic nerve, were quantified using ImageJ.

Results
Increased MAP kinase signalling in murine Rs1h À/Y retinae A study by Gehrig et al. [40] previously indicated up-regulated MAP kinase activity in early retinal development of Rs1h À/Y mice. The authors found in the murine Rs1h À/Y retinae increased phosphorylation of extracellular-signal-regulated kinases 1 and 2 (Erk1/2), as well as up-regulated expression of Egr1 (early growth response protein 1), a prominent target gene of activated MAP kinases [56].
To first verify these results, we analysed phosphorylation of Erk1/2 as well as Egr1 expression in retinae of wild-type and Rs1h À/Y mice, 7, 10 and 14 days after birth (P7, P10 and P14). Furthermore, we investigated phosphorylation of c-Raf, a central constituent of the ERK pathway, the activation of which precedes and is required for Erk1/2 phosphorylation [57], in P7, P10 and P14 retinae. As an additional MAP kinase target gene, we assessed expression of the FBJ murine osteosarcoma viral oncogene homologue gene (c-Fos), which is expressed in response to transient and sustained ERK signalling [56,[58][59][60][61].
Western blot analyses showed an increase in c-Raf and Erk1/2 phosphorylation in Rs1h À/Y retinae compared with wild-type retina ( Fig. 1A). C-Raf phosphorylation levels in Rs1h À/Y retinae increased to around 150-200%, whereas Erk1/2 phosphorylation levels rose to around 250-300% above normal (Fig. 1B). These differences were obtained in all stages (P7, P10 and P14), and each were statistically significant (P < 0.05, except for c-Raf at P10; Fig. 1A and B) Retinoschisin deficiency had no influence on levels of total c-Raf and Erk1/2, independent of the post-natal stages.
Expression of MAP kinase target genes C-Fos and Egr1 was increased in Rs1h À/Y retinae, at all developmental stages; c-Fos expression levels in Rs1h À/Y retinae were between 150% and 250%, and Egr1 expression levels between 175% and 300% compared with wild-type retinae. Differences in expression were statistically significant between wild-type and Rs1h À/Y retinae (P < 0.05).

Binding of retinoschisin to different retinal cell types
Searching for an in vitro model system applicable for analysing the influence of retinoschisin on MAP kinase signalling, we tested different retinal cells including murine microglial cell line BV-2, human RPE-derived cell line ARPE-19, the human retinoblastoma cell lines Y-79 and Weri-Rb1 for their capacity to bind retinoschisin ( Fig. 2A). In these cells, we also investigated endogenous expression of the retinal Na/K-ATPase subunits ATP1A3 and ATP1B2, required for anchoring retinoschisin to retinal plasma membranes [23] (Fig. 2B and C).
Semiquantitative RT-PCR (Fig. 2B) and Western blot analyses (Fig. 2C) revealed endogenous expression of ATP1A3 and ATP1B2 only in Y-79 and Weri-Rb1. Weri-Rb1 cells also weakly expressed retinoschisin ( Fig. 2B and C). In all further analyses, Y-79 cells were used to assess effects of externally added retinoschisin on intracellular signalling. Cys59Ser] is one of the rare XLRS variants which are not subjected to co-or post-translational degradation, but instead are translated and secreted from cells, although not as a stable octamer but instead as a dimer [21,22]. For purification, retinoschisin (normal and mutant) was fused to an N-terminal Myc tag, which did not influence secretion, oligomerization or binding capacities of the resulting protein ( Fig. 2D and E).  Y-79 cells and Rs1h À/Y retinal explants exposed to recombinant retinoschisin for 10, 30 and 60 min. stably bound the externally added retinoschisin, even after only 10 min. of incubation (Fig. 2E). Notably, RS1-C59S exhibited a strongly reduced binding affinity to Y-79 cells and Rs1h À/Y retinal explants (Fig. 2E).
Immunohistochemical stainings of murine Rs1h À/Y retinal explants after treatment with retinoschisin for 30 min. (Fig. 2F) confirmed the binding of externally added recombinant retinoschisin. Recombinant retinoschisin colocalized with the endogenously expressed retinal Na/K-ATPase of the murine retina at the inner segments of photoreceptor cells. This is in agreement with the localization of retinoschisin in wild-type retinae [23]. In contrast to the known retinoschisin localization, no recombinant retinoschisin was detected in the plexiform layers of the retinal explants, which could possibly be explained by limited diffusion of the externally added retinoschisin through the retinal layers. No retinoschisin staining was observed in immunohistochemical analyses of Rs1h À/Y retinal explants treated with RS1-C59S or control protein (Fig. 2F).

Extracellular retinoschisin modulates ERK 1/2 signalling in Y-79 cells and Rs1h À/Y murine retinal explants
To assess the capacity of retinoschisin to directly modulate intracellular ERK1/2 signalling, we investigated an influence of extracellularly added recombinant retinoschisin (normal and mutant) on phosphorylation of C-RAF and ERK1/2 in Y-79 cells (Fig. 3) and Rs1h À/Y murine retinal explants (Fig. 4). As control, we applied protein purified from supernatant of mock vectortransfected cells.
In Y-79 cells, we observed a down-regulation of phosphorylated C-RAF (77.6 AE 9.7%) after 10 min. of treatment with recombinant retinoschisin (Fig. 3A). In contrast, RS1-C59S failed to inhibit C-RAF phosphorylation (Fig. 3A). The differences in phosphorylated C-RAF levels between retinoschisin treatment and control or RS1-C59S treatment were statistically significant (P < 0.05). The effect of retinoschisin on C-RAF phosphorylation was still evident after 30 min. of treatment with retinoschisin, where C-RAF phosphorylation was reduced to 78.4 AE 17.0% (Fig. 3B).
In contrast to its effect on C-RAF in Y-79 cells, retinoschisin treatment failed to show a significant decrease in ERK1/2 phosphorylation after 10 min. of incubation (Fig. 3A). Thirty minutes of retinoschisin treatment (Fig. 3B), however, reduced ERK activation to around 69.6 AE 5.3% in Y-79 cells (P < 0.05 compared with control protein or RS1-C59S). No alterations in total C-RAF and total ERK1/2 protein levels were detected, excluding an effect of retinoschisin on expression or stability of the two proteins (Fig. 3).
MAP kinase signalling in murine Rs1h À/Y retinal explants was similarly affected by retinoschisin treatment (Fig. 4). After 10 min. of incubation with retinoschisin, phosphorylated c-Raf levels were decreased to 73.4 AE 16.3% (Fig. 4A). RS1-C59S treatment had no effect on c-Raf phosphorylation. The differences in phosphorylated c-Raf levels were statistically highly significant Fig. 1 Influence of retinoschisin deficiency on MAP kinase signalling in the murine retina. C-Raf and Erk1/2 phosphorylation in murine wild-type and Rs1h À/Y retinae, harvested at post-natal days 7, 10 and 14 (P7, P10 and P14). Retinal lysates were subjected to Western blot analyses with antibodies against phosphorylated c-Raf (pRaf), total c-Raf (Raf), phosphorylated Erk1 and Erk2 (pErk1/2), total Erk1 and Erk2 (Erk1/2), as well as ActB as a control (A). Densitometric quantification (B) was performed with immunoblots from three independent sample sets. Signals for pErk1/2, Erk1/2, pRaf and Raf were normalized against ActB and calibrated against signals for wild-type retinae. Data represent the mean + S.D. (C) C-Fos and Egr1 expression in murine wild-type and Rs1h À/Y retinae harvested at post-natal days 7, 10 and 14 (P7, P10 and P14). mRNA expression of C-Fos and Egr1 was determined via quantitative real-time RT-PCR. Five independent sample sets were analysed. Results were normalized to Hprt transcript levels and calibrated with the control. The mean + S.D. for the three (immunoblot analyses) or five (quantitative RT-PCR) independent sample sets is given. Asterisks mark statistically significant (*P < 0.05) and highly significant (**P < 0.01) differences.

773
(P < 0.01) between control and retinoschisin treatment and statistically significant (P < 0.05) between retinoschisin and RS1-C59S treatment. After 30 min. of incubation, the reduction in c-Raf phosphorylation by retinoschisin treatment was still observable (Fig. 4B): Retinoschisin decreased c-Raf phosphorylation to 81.4 AE 14.1%, compared with the control. Erk1/2 phosphorylation was not affected after 10 min. (Fig. 4A), but only after 30 min. of treatment with recombinant retinoschisin (Fig. 4B). In contrast to control or RS1-C59S treatment, incubation with retinoschisin resulted in a clear reduction in phosphorylated Erk1/2 (68.7 AE 18.2% compared with control, Fig. 4B). Differences in phosphorylated Erk1/2 levels were statistically highly significant (P < 0.01) when compared between control and retinoschisin treatment and statistically significant (P < 0.05) between retinoschisin and RS1-C59S treatment. The different treatments caused no changes in total c-Raf and total Erk1/2 levels in the retinal explants (Fig. 4).
Finally, we addressed retinoschisin-dependent photoreceptor survival by following cone and rod degeneration in murine Rs1h À/ Y retinal explants [63]. Photoreceptor cell death in Rs1h À/Y mice is triggered by apoptotic events initiated around 14 days after birth (P14, [63]). We isolated Rs1h À/Y retinae 18 days after birth (P18) and incubated them in medium containing retinoschisin, RS1-C59S or control protein. One week of cultivation resulted in a strong degeneration of retinal explants, shown by a markedly decreased thickness of the central retina and a significant reduction in photoreceptor cells (Fig. 6C-F). More specifically, compared with untreated retinae, 1 week of cultivation reduced the number of cones to around 25% in control and RS1-C59S-treated explants (25.6 AE 8.5% for control and 24.8 AE 8.9% for RS1-C59S treatment ( Fig. 6C and D). Notably, in explants treated with  Results were normalized to HPRT transcript levels and calibrated with the control. The mean + S.D. for the five independent experiments is given. Asterisks mark statistically significant (*P < 0.05) and highly significant (**P < 0.01) differences. (B) Retinoschisin-dependent activation of caspase-3 in Y-79 cells subjected to oxidative stress. Y-79 cells, exposed to retinoschisin, RS1-C59S or control protein were treated with 0.2 mM H 2 O 2 for 2 hrs. About 18 hrs later, apoptosis was assayed by following caspase-3-specific proteolytic activity. Data represent the mean + S.D. of six independent experiments. Asterisks mark statistically highly significant differences (**P < 0.01). (C-F) Retinoschisin-dependent photoreceptor degeneration in murine Rs1h À/Y retinal explants. Retinal explants harvested 18 days after birth (P18) were cultured for 1 week in medium containing retinoschisin, RS1-C59S or control protein (purified from supernatant of empty expression vector-transfected cells). After washing and embedding, cryosections of these explants were subjected to staining for nuclei, cones and rods. OS, outer segments; IS, inner segments; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer. DAPI staining shows the nuclei of the different retinal layers. (C) Alexa488-conjugated peanut agglutinin (PNA) staining was applied to visualize cones. (D) The total number of cones per analyzed section was counted after staining with PNA. (E) Anti-Rho-1D4 antibody staining was applied to visualize rod specific Rhodopsin. (F) Rhodopsin signals per analyzed section were measured using ImageJ (imagej.nih.gov). Data represent the mean + SD. Asterisks mark statistically highly significant differences (**P < 0.01). retinoschisin, the cone number was decreased to only about 50% (50.3 AE 7.3% compared with untreated retinae), with a statistically highly significant difference to control and RS1-C59S-treated explants (P < 0.01). Investigations on rod degeneration revealed similar results (Fig. 6E and F). After 1 week of cultivation, rod signals were reduced to around 24.8 AE 9.3% for control treated and to 28.9 AE 4.7% for RS1-C59S-treated explants. Treatment with retinoschisin lead to a rod signal decrease of only 59.3 AE 7.6% compared with untreated retinae, with statistically highly significant differences to control and RS1-C59S-treated explants (P < 0.01).

Discussion
In this study, we investigated the role of retinoschisin in the regulation of intracellular MAP kinase signalling. Firstly, our experiments confirmed strongly increased MAP kinase signalling in early retinal development of Rs1h À/Y mice. Secondly, we demonstrated that retinoschisin binding directly decreased phosphorylation of C-RAF and MAP kinases ERK1 and ERK2, as well as expression of the MAP kinase target genes C-FOS and EGR1 in a retinal (Y-79) cell line and in murine Rs1h À/Y retinal explants. Thirdly, our data suggest a protective effect of retinoschisin against apoptotic cell death in Y-79 cells and Rs1h À/Y retinal explants. As a stringent control, the XLRS mutant RS1-C59S was deficient in binding to retinal membranes, and failed to reveal regulation on MAP kinase signalling or effects on apoptosis. Together, our results demonstrate that retinoschisin is a novel regulator of intracellular signalling and protects retinal cells from apoptosis. We suggest that aberrant MAP kinase signalling due to retinoschisin deficiency could be an initial trigger in XLRS pathogenesis.
In recent years, the importance of MAP kinase signalling in retinal development and homeostasis has attracted increasing attention [67][68][69]. Not surprisingly, several retinal dystrophies such as age-related macular degeneration [70][71][72], diabetic retinopathy [73] or retinitis pigmentosa [74,75] were linked to malfunctioning MAP kinase pathways. Aberrant MAP kinase signalling was also observed during early retinal development in the XLRS mouse model [40]. Our study verified the earlier observations from Gehrig et al. [40] by showing increased activation of central constituents of the ERK pathway, c-Raf and Erk1/2 [57], in Rs1h À/Y retinae of 7-, 10-and 14-day-old mice. Furthermore, we showed up-regulation of prominent target genes of MAP kinase signalling, namely c-Fos and Egr1 [56], indicating an early and sustained alteration in MAP kinase signalling in disease development of Rs1h À/Y mice.
To assess whether retinoschisin has the capacity to directly modulate MAP kinase signalling, we investigated the effect of recombinant retinoschisin on activation of the ERK pathway in two retinal model systems; the human retinoblastoma cell line Y-79 and Rs1h À/Y murine retinal explants, both capable to bind extracellularly added retinoschisin due to an endogenous expression of the NA/K-ATPase subunits a3 and b2. The addition of recombinant retinoschisin had an immediate and significant influence on MAP kinase signalling in these two model systems, reflected by decreased C-RAF and ERK1/2 phosphorylation. C-RAF activation (10 min.) occurred before ERK1/2 phosphorylation (30 min. after addition of retinoschisin), in agreement with the established sequence of C-RAF and ERK1/2 activation in the ERK signalling cascade [57,76]. Subsequently, retinoschisin treatment also induced down-regulation of C-FOS and EGR1 expression in Y79 cells and Rs1h À/Y murine retinal explants. These results establish retinoschisin as an important regulator of the MAP kinase pathway in retinal cells.
The contribution of MAP kinase signalling to various disease processes can be explained by its key role in the regulation of complex physiological processes such as apoptosis, adhesion, proliferation, differentiation or development [41,44,49]. For instance, several studies showed a pro-apoptotic effect of ERK activation specifically connected to neuronal cells, for example in neurodegenerative disease processes [77][78][79][80]. Of note, a characteristic increase in ERK1/2 activation with an effect size similar to our results has been described for early disease stages of Alzheimer's disease with 25% less ERK1/2 activation in temporal cortex of healthy individuals compared with patients [81], or of ocular ischaemic syndrome where 29% less ERK1 and 21% less ERK2 activation in murine retinae of control mice were found when compared to a mouse model of ocular ischaemic syndrome [82]. Additionally, comparably small alterations in MAP kinase signalling, related to cellular survival, were found in natural killer cells of chronic fatigue syndrome [83] and in lymphocytes of patients with Alzheimer's and Parkinson's disease [84].
Consistently, we demonstrate a protective influence of retinoschisin against apoptosis: Transcript levels of the pro-apoptotic BAX protein [85,86] were down-regulated in Y-79 cells and Rs1h À/Y retinae exposed to recombinant retinoschisin. Furthermore, in Y-79 cells subjected to oxidative stress, caspase-3 activity, a marker for the induction of apoptosis [65], was significantly decreased by retinoschisin. Similarly, apoptosisinduced cone and rod degeneration [63] in murine Rs1h À/Y retinal explants was strongly reduced in the presence of recombinant retinoschisin. Further studies are required to verify the direct contribution of increased MAP kinase activation to photoreceptor apoptosis in XLRS pathogenesis. Nevertheless, considering the current state of knowledge on the pathological role of MAP kinase activation in neurodegeneration [77][78][79][80][81][82], we speculate that increased MAP kinase signalling due to retinoschisin deficiency can induce or contribute to XLRS-associated neurodegenerative processes in humans, and apoptotic photoreceptor degeneration in the XLRS mouse model [40].
Our investigations included studies on the functionality of the XLRS-associated retinoschisin mutant, RS1-C59S. Unlike most RS1 mutants, RS1-C59S is translated and secreted from cells, but with defective oligomerization [21,22]. The functional consequences of this structural alteration have not been elucidated, so far. Here, we show that RS1-C59S cannot bind to retinal membranes and can thus not fulfil its function as a regulator of intracellular signalling.
The present data do not allow elucidation of how extracellular retinoschisin binding affects intracellular MAP kinase signalling. Previous analysis identified the retinal Na/K-ATPase as the specific binding partner for retinoschisin on retinal membranes [14,23]. Several groups reported that in addition to their function as an ion pump [87,88], Na/K-ATPases are important regulators of intracellular MAP kinase signalling [31,[89][90][91], although the exact mechanism of signal transduction from Na/K-ATPases to the MAP cascade is under debate   [30,[92][93][94][95][96]. It would thus be conceivable that retinoschisin modulates the capacity of the Na/K-ATPase to regulate intracellular signalling. A disruption of this retinoschisin-Na/K-ATPase signalosome complex by retinoschisin deficiency could therefore result in defective MAP kinase regulation by the Na/K-ATPase.
Taken together, we provide evidence that retinoschisin is a novel regulator of MAP kinase signalling in the retina with the capacity to protect cells against apoptotic cell death. We suggest that disturbances of intracellular MAP kinase signalling by retinoschisin deficiency might be one of the initial steps in XLRS pathology. Thus, our data could provide a novel basis for considerations to therapeutic treatments for this progressive and currently untreatable disease.