Degeneration of Photoreceptor Cells in Arylsulfatase G-Deficient Mice

Abstract

Purpose: Retinal degeneration is a common feature of several lysosomal storage disorders, including the mucopolysaccharidoses, a group of metabolic disorders that is characterized by widespread accumulation of glycosaminoglycans due to lysosomal enzyme dysfunction. We used a new mouse model of mucopolysaccharidosis IIIE to study the effect of Arylsulfatase G (ARSG) deficiency on retina integrity.

Methods: The retina of Arsg knockout mice aged 1 to 24 months was studied by immunohistochemistry and Western blot analysis. Electron microscopic analyses were performed on retinas from 15- and 22-month-old animals. Photoreceptor and microglia cell numbers and retina thickness were determined to quantitatively characterize retinal degeneration in ARSG-deficient mice.

Results: Arsg knockout mice showed a progressive degeneration of photoreceptor cells starting between 1 and 6 months of age, resulting in the loss of more than 50% of photoreceptor cells in 24-month-old mice. Photoreceptor loss was accompanied by reactive astrogliosis, reactive microgliosis that was evident in the outer but not inner retina, and elevated expression levels of some lysosomal proteins. Electron microscopic analyses of retinas revealed no evidence for the presence of storage vacuoles. Of note, expression of ARSG protein in wild-type mice was detectable only in the RPE which, however, appeared morphologically unaffected in knockout mice at the electron microscopic level.

Conclusions: To our knowledge, this is the first study demonstrating that ARSG deficiency results in progressive photoreceptor degeneration and dysregulation of various lysosomal proteins.

The lysosomal degradative pathway of sulfated glycosaminoglycans (GAG) comprises a sophisticated hydrolytic network of highly specific glycosidases and sulfatases for complete degradation of these complex polysaccharides to sulfate and monosaccharides.^1–5 For each sulfate residue in different positions of the sugar moiety, distinct sulfatases are indispensable for their desulfation, which in turn is a prerequisite for glycosidic hydrolysis. Pathogenic mutations in genes coding for these hydrolytic enzymes lead to impaired degradation of GAGs and as a consequence to an accumulation of the corresponding substrates in lysosomes, a clinical situation described as lysosomal storage disorder (LSD). Disorders resulting from impaired lysosomal degradation of sulfated GAGs (heparan sulfate [HS], dermatan sulfate, chondroitin sulfate, and keratan sulfate) are summarized as mucopolysaccharidoses (MPSs).^1–6 One subgroup of the MPSs is Sanfilippo syndrome (MPS type III), which exclusively affects the degradation of HS. Mutations in genes coding for four different enzymes, needed for the removal of sulfated glucosamine residues of HS, are known to cause MPS III subtypes in humans, including N-sulfoglucosamine sulfohydrolase (encoded by SGSH; MPS IIIA), N-α-acetylglucosaminidase (encoded by NAGLU; MPS IIIB), heparan-α-glucosaminide N-acetyltransferase (encoded by HGSNAT; MPS IIIC), and N-acetylglucosamine-6-sulfatase (encoded by GNS; MPS IIID).^3,7 Taking all pathogenic mutations in these four genes together, MPS III is the most frequently occurring type of MPS with a reported prevalence in different populations of 0.28 to 4.1 per 100,000 births.⁷ We have shown recently that a fifth enzyme is critical for complete degradation of HS glucosamine residues when sulfated in the C3 position of glucosamine: Arylsulfatase G (ARSG), also termed N-sulfoglucosamine-3-O-sulfatase.^6,8 We have generated Arsg knockout (KO) mice, and have demonstrated accumulation of HS in different organ systems including liver, kidney, and brain.⁶ Due to its assigned role in the degradation of HS and the resulting Sanfilippo syndrome-like pathological alterations, we tentatively assigned this MPS type as MPS IIIE. Compared to mouse models of the other MPS III subtypes,^9–11 Arsg KO mice presented with a milder phenotype and a later onset of the disease, with Purkinje cell degeneration in the cerebellum as the major neurological phenotype.¹² Severe ataxia and Purkinje cell degeneration also was observed in an American Staffordshire Terrier dog pedigree that lacks functional ARSG due to a point mutation in the Arsg gene.¹³ This canine model was assigned as a model of neuronal ceroid lipofuscinosis (NCL) because of the large amounts of accumulated lipofuscin in neurons. Human patients carrying pathogenic mutations in ARSG have not been identified until now. 

Mucopolysaccharidoses are multisystemic disorders affecting most cell types of the body. However, the impaired cellular clearance of GAGs and the resulting lysosomal dysfunction is of particular detrimental significance for postmitotic cells, such as neurons, as reflected by the profound neurological symptoms of most MPS patients. In MPS III patients, neurodegeneration is the key clinical feature, ultimately leading to premature death.^3,4,7 The four human MPS III subtypes are similar clinically, with typical symptoms, including developmental delay, mild coarse facial features, progressive loss of mental and motor functions, and epileptic seizures. Cortical atrophy and ventricular enlargement are common findings. Patients usually die at the end of the second or beginning of the third decade of life, often due to respiratory insufficiencies.^{2–4,7,14,15} 

In addition to the brain, the retina is affected to a significant extent in two prominent groups of LSDs: retinal degeneration is a characteristic feature of several NCLs^16–22 and also is seen frequently in MPSs. Ocular involvement was reported in patients and animal models of the majority of MPS subtypes, including MPS III.^9,23–29 Patients with MPS III typically present with progressive photoreceptor loss closely resembling that occurring in retinitis pigmentosa, whereas the ganglion cells and optic nerve usually are unaffected.^{23–25,28–32} 

In this study, we describe degenerative changes in the retina of ARSG-deficient mice, a new mouse model of MPS type III. Progressive photoreceptor loss in the mouse starts between 1 and 6 months of age and is accompanied by reactive astrogliosis and microgliosis, and a dysregulation of several lysosomal proteins. We define retinal degeneration as an important and early onset pathological feature of ARSG deficiency. 

Materials and Methods

Animals

Arylsulfatase G knockout (Arsg KO) mice were generated as described previously.⁶ Mice were maintained on a mixed C57BL/6J 129/Ola genetic background and housed according to the institutional guidelines of the University Bielefeld, with ad libitum access to food and water. Genotyping of mice was performed as described.⁶ Retina tissue from CLN7 mutant mice³³ was used to control the specificity of p62 immunostainings. In all experiments, age-matched C57BL/6J 129/Ola wild-type mice served as a control. All animal work was approved by the local Animal Care Committee and was in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Light and Electron Microscopy

For light and electron microscopic analyses, 15- and 22-month-old Arsg KO and age-matched wild-type mice were deeply anesthetized and transcardially perfused with PBS followed by perfusion with 6% glutaraldehyde (Merck, Darmstadt, Germany) in phosphate buffer. Eyes were enucleated, the lenses were removed, and the bulbs were post-fixed with 2% osmium tetroxide, dehydrated, and embedded in Araldite. Semithin sections were stained with toluidine blue. Ultrathin sections were stained with uranyl acetate and lead citrate, and examined with a Zeiss EM900 electron microscope (Zeiss, Jena, Germany) equipped with a Megaview III digital camera (Albert Tröndle, Moorenweis, Germany). 

Immunohistochemistry of Retina Sections

Immunohistochemical analyses were performed on retinas from 1-, 6-, 12-, 19-, and 24-month-old Arsg KO and age-matched wild-type mice. Animals were killed and eyes were quickly removed and fixed overnight in PBS (pH 7.4) containing 4% paraformaldehyde (PA). After dehydration in an ascending series of sucrose, eyes were frozen in Tissue-Tek (Sakura Finetek, Zouterwoude, The Netherlands) and serially sectioned with a cryostat at a thickness of 25 μm. Central (i.e., in the plane of the optic disc) retina sections were first blocked in PBS containing 0.1% bovine serum albumin (BSA) and 0.3% Triton X-100 (both from Sigma-Aldrich Corp., St. Louis, MO, USA) for 1 hour and then incubated with primary antibodies (see Table) overnight at room temperature. After washing with PBS, sections were incubated with Cy2- or Cy3-conjugated secondary antibodies (Jackson ImmunoResearch, West Grove, PA, USA) for 4 hours, stained with 4′,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich Corp.) and mounted onto slides. For detection of mannose 6-phosphate (M6P)–containing proteins, retina sections were incubated with the myc-tagged single-chain antibody fragment scFv M6P-1³⁴ followed by polyclonal rabbit anti-myc antibodies (Sigma-Aldrich Corp.) and Cy3-conjugated donkey anti-rabbit antibodies. To visualize cones, sections were stained with biotinylated peanut agglutinin (BPA; Vector Laboratories, Burlingame, CA, USA) overnight at room temperature, followed by Cy3-conjugated Streptavidin (Jackson ImmunoResearch) and DAPI. In all experiments sections from Arsg KO and age-matched wild-type mice were processed in parallel and under identical conditions. At least 6 animals were analyzed for each antigen, developmental age, and genotype. 

Table

 

Primary Antibodies Used for Immunohistochemistry

 

For immunohistochemical analyses of ARSG expression in RPE cells, the melanin pigment was bleached.³⁵ Sections were incubated in 0.05% potassium permanganate (Merck) for 25 minutes, washed with PBS, incubated for 5 minutes in 0.5% oxalic acid (Carl Roth GmbH, Karlsruhe, Germany) and washed again with PBS before they were incubated with anti-ARSG antibodies. To analyze flat-mounted RPE, eyes were fixed in 4% PA. The retina was removed, and the RPE with attached choroidea blocked in PBS containing 0.1% BSA and 0.3% Triton X-100 and incubated with antibodies to RPE65, OTX2, or ZO-1. Before incubation with anti-ARSG or anti-RPE65 antibodies, melanin pigment was bleached as described above. Primary antibodies were detected with Cy2- or Cy3-conjugated secondary antibodies, stained with DAPI and mounted onto slides. 

For each antigen, sections from Arsg KO and age-matched wild-type mice were analyzed in parallel and with the same microscope settings using an Olympus FV 1000 confocal microscope (Olympus, Hamburg, Germany). 

Photoreceptor Counts and Retina and Outer Nuclear Layer Thickness

To quantify the loss of photoreceptor cells in Arsg KO mice, central retina sections from mutant and age-matched wild-type mice were stained with anti-recoverin antibodies and DAPI. A merged confocal image of the entire retina section was prepared using Photoshop CS6 software (Adobe Systems, Inc., San Jose, CA, USA), and photoreceptor nuclei were counted at three defined positions corresponding to 25%, 50%, and 75% of the distance between the optic disc and the peripheral margin of the nasal and temporal retina, respectively. Each area defined for photoreceptor counts covered the outer nuclear layer over a length of 220 μm.³⁸ Statistical analyses of data were performed with the Student's t-test using GraphPad software (GraphPad Software, La Jolla, CA, USA). 

The thickness of the retina and the outer nuclear layer (i.e., photoreceptor cell bodies and inner and outer photoreceptor segments) was measured in central retina sections at nine equidistant positions between the optic disc and periphery of the nasal and temporal retinal halves, respectively. Numbers of Iba1- and CD68-positive cells with a clearly visible DAPI-positive nucleus were determined in the inner retina (i.e., nerve fiber layer, ganglion cell layer, inner plexiform layer, and inner nuclear layer) and outer retina (i.e., outer plexiform layer, photoreceptor cell bodies, and inner and outer photoreceptor segments) of Arsg KO and wild-type mice aged 1 to 24 months. The area of the inner and outer retina was measured using Photoshop CS6 software, and the density of positive cells was calculated. Statistical analyses of data were performed with the 2-way ANOVA test followed by a Bonferroni post hoc test using GraphPad software. 

To determine the density of retinal ganglion cells (RGCs), eyes of 19-month-old Arsg KO and age-matched wild-type mice (n = 5 for each genotype) were fixed in 4% PA and retinas were flat-mounted on nitrocellulose membranes (Sartorius AG, Göttingen, Germany) as described.³⁹ After blocking in PBS containing 0.1% BSA and 1% Triton X-100, retinas were incubated with polyclonal goat anti-Brn-3a antibodies overnight at room temperature. Primary antibodies were detected with Cy3-conjugated secondary antibodies, and flat-mounted retinas were stained with DAPI and mounted onto slides. Five images were taken from the center to the periphery of the superior, inferior, nasal, and temporal retinal quadrants, covering a total area of approximately 1.9 mm². All Brn-3a–positive RGCs visible on these images were counted using Adobe Photoshop CS6 software, and the density of RGCs per mm² retinal area was calculated. Statistical analysis of data was performed using Student's t-test. 

Western Blotting

Tissue lysates were prepared by homogenization of whole retinas in 15 volumes (wt/vol) of ice-cold ×1 TBS containing 1% Triton-X-100 and protease inhibitors. After incubation on ice for 30 minutes followed by centrifugation at 13,000g, the supernatant was used and protein concentration determined by Bicinchoninic acid assay (Thermo Fisher Scientific, Inc., Schwerte, Germany). For Western blots, 20 μg of the retina lysates were loaded on SDS-PAGE gels and blotted on nitrocellulose membranes with a semi-dry blotting apparatus. For Saposin D blotting, membranes were heated up to 100°C for 5 minutes immediately after blotting. The same antibodies as indicated for immunohistochemistry were used at the following dilutions: Saposin D 1:500, Lamp2 1:250, and Cathepsin D 1:1000. Gapdh (1:1000; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) was used as a loading control. After incubation with the primary antibodies overnight, blots were probed with horseradish peroxidase coupled secondary antibodies and subsequently detected by enhanced chemiluminescence reagent. For quantification, densitometry was performed using ImageJ software (National Institutes of Health [NIH], Bethesda, MD, USA). Wild-type and age-matched Arsg KO mice at 1 and 19 months of age were used (n = 6 for each genotype). Statistical analysis of data was performed using Student's t-test. 

Determination of β-Hexosaminidase Activity

The specific activity of the lysosomal marker enzyme β-hexosaminidase was determined as described previously.⁴⁰ In brief, approximately 5 μg retina lysates from Arsg KO and age-matched wild-type mice (as prepared for Western blots) were incubated with the artificial substrate ρ–nitrophenyl-N-acetyl-β-D-glucosaminide in 0.2 M citric acid buffer pH 4.6 for 16 hours at 37°C. After addition of 0.4 M glycine/NaOH (pH 10.4) and centrifugation at 13,000g, absorption was measured at 405 nm. The activity was normalized to protein concentration of each sample to calculate the specific activity. 

Results

Progressive Degeneration of Rod Photoreceptor Cells in ARSG-Deficient Mice

To characterize the retinal phenotype of mice deficient in ARSG,^6,12 we stained central retina sections from Arsg KO and age-matched wild-type mice with antibodies to glial fibrillary acidic protein (GFAP). In 1-month-old mutants (Fig. 1Ab) and wild-type mice (Fig. 1Aa), expression of GFAP was restricted to retinal astrocytes located at the vitreal margin of the retinas. A similar pattern of GFAP expression was observed in the retina of wild-type mice aged 6 to 24 months (Figs. 1Ac, 1Ae, 1Ag). In 6- (Fig. 1Ad), 12- (Fig. 1Af), and 24 (Fig. 1Ah)–month-old Arsg KO mice, in comparison, expression of GFAP was elevated in retinal astrocytes and became additionally detectable in Müller cells. Reactive astrogliosis in Arsg KO was accompanied by a progressive thinning of the outer nuclear layer. Analyses of retina sections from 1-month-old Arsg KO and age-matched wild-type mice stained with anti-recoverin antibodies to label photoreceptor cells revealed a similar thickness of the outer nuclear layer of both genotypes (compare Figs. 1Ai, Aj). In 6- (Fig. 1Al), 12- (Fig. 1An) and 24- (Fig. 1Ap)–month-old Arsg KO mice, the thickness of the photoreceptor layer decreased with increasing age of the animals. 

Figure 1

 

Expression of GFAP and recoverin and thickness of the retina and the outer nuclear layer of Arsg KO and wild-type mice at different developmental ages. In 1-month-old wild-type (Aa) and Arsg KO mice (Ab), expression of GFAP was restricted to astrocytes located at the vitreal margin of the retina. A similar pattern of GFAP expression was observed in 6- (Ac), 12- (Ae), and 24 (Ag)–month-old wild-type retinas. In 6- (Ad), 12- (Af), and 24 (Ah)–month-old mutants, in comparison, expression of GFAP was strongly elevated in retinal astrocytes and was additionally detectable in Müller cells. Immunostainings with anti-recoverin antibodies revealed a similar thickness of the outer nuclear layer (onl) in 1-month-old wild-type (Ai) and age-matched Arsg KO mice (Aj). In 6- (Al), 12- (An,) and 24 (Ap)–month-old mutants, the thickness of the outer nuclear layer decreased significantly when compared to age-matched wild-type mice ([Ak], [Am], [Ao], respectively). The thickness of the retina and the outer nuclear layer (i.e., photoreceptor cell bodies and inner and outer photoreceptor segments) was measured in central retina sections stained with anti-recoverin antibodies and DAPI (B) at nine equidistant positions (indicated with white lines in [B]) between the optic nerve head (ONH) and the periphery of the nasal and temporal retinal half, respectively. Analyses revealed a similar thickness of the retina and the outer nuclear layer in 1-month-old Arsg KO (gray squares in [Ca] and [Cb], respectively) and age-matched wild-type mice (black circles in [Ca] and [Cb], respectively). In 6-month-old mutants, the thickness of the retina and outer nuclear layer (green triangles in [Ca] and [Cb], respectively) was significantly decreased at all retina positions when compared to 1-month-old mutants. Retina and outer nuclear layer thickness was further decreased in 24-month-old mutants (red triangles in [Ca] and [Cb], respectively) when compared to 6-month-old Arsg KO mice. Each symbol represents the mean value (±SEM) of 6 animals. *P < 0.05; **P < 0.01; ***P < 0.001 according to the 2-way ANOVA followed by a Bonferroni post hoc test. gcl, ganglion cell layer; inl: inner nuclear layer; ipl, inner plexiform layer; M, month; Rec, recoverin. Scale bars: for (Aa–Ah), 50 μm; in (B) 200 μm.

Measurements of the retina thickness at 9 equidistant positions between the optic disc and periphery of the nasal and temporal retina, respectively (Fig. 1B), revealed similar values for 1-month-old wild-type and age-matched Arsg KO mice at all retinal positions analyzed (Fig. 1Ca). In 6-month-old mutants, in comparison, retina thickness was significantly decreased at all retinal positions compared to 1-month-old Arsg KO mice (Fig. 1Ca). Retina thickness was further decreased in 24-month-old mutants compared to 6-month-old Arsg KO mice (Fig. 1Ca), in line with the immunohistochemical data. Similarly, we observed no significant differences in the thickness of the outer nuclear layer between 1-month-old Arsg KO and wild-type mice, but a progressive thinning of this layer in older Arsg KO mice (Fig. 1Cb). The thickness of the inner retina, in contrast, was similar in mutant and wild-type mice at all ages analyzed (Supplementary Fig. S1), indicating that the retinal dystrophy in Arsg KO mice is mainly or exclusively due to a progressive loss of photoreceptor cells. In line with these results, we found similar densities of PKCα-positive bipolar cells in 24-month-old animals of both genotypes (Supplementary Fig. S1). Furthermore, the density of RGCs in 18-month-old animals did not differ significantly between both genotypes, with 3688 ± 165 RGCs/mm² retina area (mean ± SEM) in wild-type mice and 3932 ± 180 RGCs/mm² retina area in Arsg KO mutants (n = 5 for each genotype; Supplementary Fig. S1). 

To further quantify the loss of photoreceptor cells in the mutant, photoreceptor cell nuclei were counted in three areas located at defined positions of the nasal and temporal retina, respectively, each covering the outer nuclear layer over a length of 220 μm. In 1-month-old animals, we found similar numbers of photoreceptor cells in wild-type mice (427.1 ± 3.5 photoreceptor cells/area; mean ± SEM; n = 6) and Arsg KO mutants (434.3 ± 9.5; n = 6; Fig. 2A). In older mutants, the number of photoreceptor cells decreased significantly with increasing age of the animals. In 6-, 12-, and 24-month-old mutants, we found 302.1 ± 7.6, 258.0 ± 8.3, and 179.5 ± 6.6 photoreceptor cells/area, respectively (Fig. 2A). Retinas from wild-type mice analyzed for comparison contained 424.8 ± 5.3, 404.4 ± 4.7, and 404.2 ± 5.6 photoreceptor cells/area at the age of 6, 12 and 24 months, respectively (Fig. 2A). To analyze whether the retinal dystrophy is the result of a progressive loss of rods or cones or both photoreceptor cell types, retina sections were labeled with anti-rhodopsin antibodies or peanut agglutinin. These experiments revealed the presence of similar numbers of rods and cones with normal inner and outer segments in 1-month-old wild-type (Figs. 2Ba, 2Be) and mutant mice (Figs. 2Bb, 2Bf). In 24-month-old ARSG-deficient mice (Figs. 2Bd, 2Bh), the length of the outer segments of rods and cones was shortened when compared to age-matched wild-type mice (Figs. 2Bc, 2Bg). Furthermore, we found a similar density of cones in aged Arsg KO and wild-type mice (Figs. 2Bh, 2Bg, respectively). 

Figure 2

 

Photoreceptor numbers and analyses of rod and cone photoreceptor cells in Arsg KO and wild-type mice at different developmental ages. Photoreceptor counts in 1-month-old Arsg KO (black bars in [A]) and wild-type mice (gray bars in [A]) revealed similar cell numbers in both genotypes. In 6-, 12-, and 24-month-old animals, photoreceptor numbers were significantly lower in Arsg KO mice when compared to age-matched wild-type mice, and decreased significantly with increasing age of the mutant. In 1-month-old wild-type (Ba, Be) and Arsg KO mice (Bb, Bf), the morphology and density of rod (Ba, Bb) and cone (Be, Bf) photoreceptor cells was similar in both genotypes. In 24-month-old Arsg KO mice, outer segments (os) of rod (Bd) and cone (Bh) photoreceptor cells were reduced in length when compared to age-matched wild-type mice ([Bc] for rods, [Bg] for cones). Note the similar density of cones in mutant (Bh) and wild-type (Bg) retinas at this age. All sections were stained with DAPI. Each bar in (A) represents the mean value (± SEM) from 6 animals. **P < 0.01; ***P < 0.001 according to Student's t-test. n.s., not significant; mo, month; onl, outer nuclear layer; Rho, rhodopsin. Scale bar: for (Ba–Bh), 50 μm.

Accumulation of Activated Microglial Cells in the Outer Retina of ARSG-Deficient Mice

Retina sections from mutant and wild-type mice of different developmental ages were stained with antibodies to Iba1 and CD68 to study reactive microgliosis in Arsg KO mice. In 1-month-old animals, ramified Iba1-positive cells were found in the ganglion cell layer, inner plexiform layer, inner nuclear layer, and outer plexiform layer, with no obvious differences in cell density or cell morphology between both genotypes (compare Figs. 3Aa, 3Ab). In older Arsg KO mice, Iba1-positive cells with a rounded amoeboid-like morphology became additionally detectable in the subretinal space of Arsg KO retinas (Fig. 3Ad). CD68-positive activated microglia/macrophages were essentially absent from wild-type retinas (for a 24-month-old wild-type retina, see Fig. 3Ae) and from 1-month-old Arsg KO retinas. In comparison, CD68-positive cells with an amoeboid morphology were found frequently in 6-, 12-, and 24-month-old mutant retinas where they were localized mainly in the subretinal space (Fig. 3Af). Quantitative analyses revealed a similar density of Iba1-positive cells in the inner retina (defined as ganglion cell layer, inner plexiform layer, and inner nuclear layer) of Arsg KO mutants and wild-type mice aged 1 to 24 months, and in the outer retina (defined as outer plexiform layer, photoreceptor cell bodies, and inner and outer photoreceptor segments) of 1-month-old wild-type and Arsg KO mice (Fig. 3Ba). In the outer retina of older Arsg KO mice, however, the number of Iba1-positive cells increased significantly with increasing age of the mutants (Fig. 3Ba). Cells positive for CD68 essentially were absent from wild-type retinas, and only rarely were observed in the inner retina of the ARSG-deficient mice (Fig. 3Bb). In the outer retina of the mutant, in contrast, CD68-positive cells were found frequently in 6-month-old animals, and their number was significantly increased in 24-month-old Arsg KO mice (Fig. 3Bb). 

Figure 3

 

Distribution and density of Iba1-positive and CD68-positive cells in the retina of Arsg KO and wild-type mice. Analyses of 1-month-old animals revealed a similar distribution and density of Iba1-positive cells in wild-type (Aa) and Arsg KO mice (Ab). In 24-month-old Arsg KO retinas (Ad), the number of Iba1-positive cells was significantly increased when compared to age-matched wild-type retinas (Ac), and positive cells were now additionally detectable between the outer nuclear layer (onl) and the RPE. CD68-positive cells were absent from retinas of 24-month-old wild-type mice (Ae) but numerous in retinas of age-matched Arsg KO mice where they were located mainly in the subretinal space (Af). Quantitative analyses revealed similar numbers of Iba1-positive cells in the inner retina (i.e., nerve fiber layer, ganglion cell layer, inner plexiform layer, and inner nuclear layer) of Arsg KO (black bars in [Ba]) and wild-type mice (gray bars in [Ba]) aged 1 to 24 months, and in the outer retina (i.e., outer plexiform layer, photoreceptor cell bodies, and inner and outer photoreceptor segments) of 1-month-old Arsg KO and wild-type mice (Ba). In the outer retina of 6-, 12- and 24-month-old mutants, the number of Iba1-positive cells increased with increasing age of the animals, and was significantly higher than in age-matched wild-type mice (Ba). Cells positive for CD68 were essentially absent from wild-type retinas at all developmental ages analyzed, and were observed only occasionally in the inner retina of Arsg KO mice aged 1 to 24 months and in the outer retina of 1-month-old mutants (Bb). In the outer retina of older Arsg KO mice, the number of CD68-positive cells increased with increasing age of the animals (Bb). Each bar represents the mean value (±SEM) of 6 animals. ***P < 0.001 according to a 2-way ANOVA followed by a Bonferroni post hoc test. Scale bar: for Aa–Af), 50 μm.

Expression of ARSG is Restricted to RPE Cells

Immunohistochemical analyses of the expression pattern of ARSG were performed on sections of adult wild-type retinas. The melanin pigment in RPE cells was bleached before the immunostainings to exclude quenching of the immunofluorescence. Double immunostainings revealed expression of ARSG in RPE65-positive RPE cells of wild-type retinas (Figs. 4Aa–Af). In retina sections from Arsg KO mice that were processed in parallel as a negative control, RPE cells were ARSG-negative as expected (Figs. 4Ag–Al). The weak fluorescence associated with photoreceptor outer segments, the outer and inner plexiform layer, and the ganglion cell layer of wild-type retinas (Fig. 4Ab) also was observed in Arsg KO retinas (Fig. 4Ah), and, thus, likely represents unspecific background labeling. Immunostainings of flat-mount preparations of the RPE confirmed expression of ARSG in RPE cells (compare Figs. 4Ba, 4Be), and additionally revealed similar expression levels and expression patterns of RPE65, OTX2, and ZO-1 in wild-type and ARSG-deficient retinas (compare Figs. 4Bb and 4Bf, 4Bc and 4Bg, and 4Bd and 4Bh, respectively). Given that RPE cells were the only retinal cell type with detectable expression levels of ARSG, the RPE of aged Arsg KO mice was analyzed by conventional light microscopy (Fig. 5A) and electron microscopy (Fig. 5B) for lysosomal storage, and morphologic or lysosomal abnormalities. These experiments revealed no obvious morphologic alterations of RPE cells in 15- or 22-month-old ARSG-deficient mice when compared to age-matched wild-type mice (Fig. 5). Macrophages were observed frequently in the subretinal space of mutant retinas (Figs. 5Bb–Bd), but not in the subretinal space of wild-type retinas (Fig. 5Ba). 

Figure 4

 

Expression of ARSG in retinal sections and flat-mounted RPE. In adult wild-type mice, ARSG-immunoreactivity (Ab, Ae) colocalized with RPE65 (Aa, Ad) in RPE cells. In comparison, RPE cells in Arsg KO retinas were ARSG-negative as expected (Ah, Ak). The weak fluorescence of photoreceptor outer segments, outer plexiform layer, inner plexiform layer, and ganglion cell layer in wild-type retinas stained with anti-ARSG antibodies (Ab) also was evident in Arsg KO retinas (Ah), and, thus, likely represents unspecific background labelling. (Ac, Af, Ai, Al) Phase contrast photomicrographs of (Aa, Ab), (Ad, Ae), (Ag, Ah), and (Aj, Ak), respectively, to demonstrate complete bleaching of the melanin pigment in RPE cells. (Ad, Ae, Aj, Ak) Higher magnifications of the RPE shown in (Aa), (Ab), (Ag), and (Ah), respectively. Immunohistochemical analyses of flat-mounted RPE confirmed expression of ARSG in RPE cells (Ba, Be), and revealed similar expression levels and expression patterns of RPE65 (Bb, Bf), OTX2 (Bc, Bg), and ZO-1 (Bd, Bh) in wild-type ([Bb, Bc, Bd], respectively) and ARSG-deficient retinas ([Bf, Bg, Bh], respectively). Sections and flat-mounts were stained with DAPI to label cell nuclei. OTX2, Orthodenticle homeobox 2; RPE65, RPE-specific 65 kDa protein; ZO-1, Zonula Occludens 1. Scale bars: for (Aa–Ac) and (Ag–Ai), 50 μm; for (Ad–Af) and (Aj–Al), 20 μm; for (Ba–Bh), 50 μm.

Figure 5

 

Light and electron microscopic analysis of ARSG-deficient retinas. Analyses of semi-thin sections from 15-month-old wild-type (Aa) and Arsg KO (Ab) retinas revealed a dystrophic photoreceptor layer and a morphologically intact RPE in the mutant. Electron microscopy confirmed the presence of an intact RPE in 22-month-old Arsg KO mice (Bb). Activated microglia cells with an amoeboid-like morphology were observed in the subretinal space of ARSG-deficient mice (asterisks in [Bb, Bc, Bd]), but not of wild-type mice (Ba). (Bd) is a higher magnification of the microglia cell shown in (Bb). Scale bar: for (Aa, Ab), 20 μm; for (Ba, Bb), 5 μm; for (Bc, Bd), 1 μm.

Dysregulation of Lysosomal Proteins in Aged ARSG-Deficient Retinas

To study the impact of ARSG-deficiency on the expression of lysosomal proteins, we studied the expression pattern of lysosomal enzymes containing the M6P recognition marker and the expression pattern of the lysosomal markers lysosomal-associated membrane protein-1 (Lamp1) and lysosomal-associated membrane protein-2 (Lamp2), the lysosomal protease cathepsin D (Ctsd) and Saposin D in retinas of 1- and 24-month-old Arsg KO mice and age-matched wild-type mice (Fig. 6). 

Figure 6

 

Expression of lysosomal proteins in the retina of Arsg KO and age-matched wild-type mice. The distribution and expression levels of M6P, Lamp1, Lamp2, Ctsd, and Saposin D were similar in 1-month-old Arsg KO and wild-type retinas (compare [a, b], [e, f], [i, j], [m, n], and [q, r], respectively). In 24-month-old animals, expression of M6P, Ctsd, and Saposin D was significantly increased in Arsg KO retinas when compared to age-matched wild-type retinas (compare [c, d], [o, p], and [s, t], respectively). Expression levels of Lamp1 and Lamp2, in comparison, were not significantly different between both genotypes at this age (compare [g, h], and [k, l], respectively). Note the accumulation of lysosomal proteins in macrophages (labeled with white arrows in [h, l, p, t]) located between the outer nuclear layer (onl) and the RPE of 24-month-old mutant mice. All sections were stained with DAPI to label cell nuclei. Scale bar: for (a–t), 50 μm.

No detectable differences in the expression pattern and expression level of M6P and the different lysosomal proteins were observed between 1-month-old Arsg KO retinas and age-matched wild-type retinas (Fig. 6; compare Figs. 6a, 6b for M6P; 6e, 6f for Lamp1; 6i, 6j for Lamp2; 6m, 6n for Ctsd; and 6q, 6r for Saposin D). In 24-month-old animals, in comparison, expression levels of M6P were significantly increased in Arsg KO retinas when compared to wild-type retinas, particularly in the ganglion cell layer (compare Figs. 6c, 6d). Expression of Lamp1 (compare Figs. 6g, 6h) and Lamp2 (compare Figs. 6k, 6l) was only slightly elevated in 24-month-old mutants, whereas immunoreactivity for Ctsd (compare Figs. 6o, 6p) and Saposin D (compare Figs. 6s, 6t) was strongly increased in Arsg KO retinas when compared to age-matched wild-type retinas. Upregulation of Ctsd was particularly evident in the ganglion cell layer, inner nuclear layer, and outer nuclear layer (Fig. 6p), whereas elevated expression of Saposin D was detected mainly in the ganglion cell layer (Fig. 6t). Furthermore, Lamp1, Lamp2, Ctsd, and Saposin D were accumulated strongly in phagocytotic microglial cells located in the subretinal space of 24-month-old Arsg KO retinas (Figs. 6h, 6l, 6p, 6t). Finally, we also analyzed expression of the autophagy adapter protein p62. While p62-immunoreactivity was absent from retina sections of wild-type and ARSG-deficient mice, it was readily detectable in all layers of CLN7 mutant retinas that were processed in parallel as a positive control (Supplementary Fig. S2). 

Western blot analyses of retinas from 1-month-old wild-type and Arsg KO mice revealed no significant differences in expression levels of Lamp2, Cathepsin D, or Saposin D (Fig. 7A), in agreement with the immunhistochemical data. In 19-month-old mutants, expression levels of Lamp2 were slightly but not significantly increased, whereas expression levels of Cathepsin D and Saposin D were strongly elevated in mutant retinas (Fig. 7A). A particular strong increase in expression levels was observed for the proteolytically processed forms of Cathepsin D (Fig. 7A). Finally, we have determined the specific activity of the lysosomal marker enzyme β-hexosaminidase. While the specific β-hexosaminidase activity was comparable between wild-type and Arsg KO mice in 1-month-old animals (Fig. 7B), it was significantly increased by a factor of approximately 2.5 in 19-month-old Arsg KO animals when compared to wild-type controls (Fig. 7B). 

Figure 7

 

Immunoblot analysis of lysosomal proteins and β-hexosaminidase activity in retinas of Arsg KO and age-matched wild-type mice. Western blotting of retina lysates from 1-month-old wild-type and Arsg KO animals with antibodies against Lamp2, Cathepsin D, and Saposin D showed no differences in expression levels between genotypes (A). In 19-month-old retina, in comparison, levels of Saposin D and of proteolytically processed forms of Cathepsin D were significantly increased in the mutant when compared to the wild-type, whereas levels of Lamp2 were similar in both genotypes (A). Gapdh was used as a loading control (A). Determination of the specific activity of the lysosomal enzyme β-hexosaminidase revealed a similar specific activity in 1-month-old wild-type and Arsg KO retinas (B). In 19-month-old animals, in comparison, β-hexosaminidase activity was increased approximately 2.5-fold in the mutant when compared to the wild-type retina. *P < 0.05; ***P < 0.001 according to the Student's t-test. Gapdh, glyceraldehyde-3-phosphate dehydrogenase; dc; double chain; hc: heavy chain; lc: light chain; SapD: Saposin D; sc: single chain.

Discussion

A recent analysis of the Arsg KO mouse provided further insight into the degradative pathway of HS and the endogenous substrate of ARSG, 3-O sulfated glucosamine. Moreover, we depicted the in vivo relevance of the substrate and its degradation by ARSG through manifestation of lysosomal storage in several tissues, including the liver, kidney, and brain in the absence of ARSG, and identified Purkinje cell degeneration and ataxia as prominent neurologic symptoms in Arsg KO mice.^6,12 In the present study, we extended the phenotypic characterization of the Arsg KO mouse, tentatively assigned as a mouse model of MPS IIIE,⁶ to an early onset degeneration of photoreceptor cells starting between 1 and 6 months of age, when neuronal loss in the brain is not yet detectable. Retinal degeneration was accompanied by reactive astrogliosis, the appearance of phagocytic microglia/macrophages in the outer retina, and elevated expression of several lysosomal proteins. Thus, loss of photoreceptor cells is among the earliest phenotypic manifestations of ARSG deficiency in the central nervous system. 

An intriguing question raised by our observations is the actual cause of the progressive photoreceptor degeneration in the Arsg KO mouse. Of interest in this context, we found that expression of ARSG in the adult murine retina is confined to the RPE. Retinal pigment epithelial cells perform multiple functions that are vital for normal photoreceptor cell function and photoreceptor cell survival, and loss or dysfunction of RPE cells results in photoreceptor degeneration.⁴¹ However, immunohistochemical and electron microscopic analyses of ARSG-deficient retinas did not reveal obvious pathological alterations of the RPE, such as RPE atrophy or typical storage vacuoles as they have been observed in the kidney of Arsg KO mice and the RPE of animal models or patients of other MPS variants and MPS III subtypes.^{6,9,24,26,30,32,42} Although the specific cause of the progressive photoreceptor cell loss in Arsg KO mouse has, thus, to be elucidated, it is tempting to speculate that heparan sulfate fragments and oligosaccharides released into the extracellular matrix might interfere with the proper function of the RPE⁷ or that a subtle lysosomal dysfunction causes functional alterations of RPE cells, ultimately resulting in photoreceptor cell death. Of note, the RPE is known to significantly contribute to the synthesis and degradation of all major mucopolysaccharides in the interphotoreceptor matrix.⁴³ However, we cannot exclude the possibility of a low level expression of ARSG in retinal cell types other than RPE cells, and, thus, a direct impact of ARSG deficiency on photoreceptor cells. 

Significant thinning of the outer nuclear layer and preferential accumulation of activated microglia cells/macrophages in the outer retina suggests that neurodegeneration in the retina of Arsg KO mice is mainly or exclusively confined to the photoreceptor cell layer. Analysis of the outer nuclear layer revealed essentially normal numbers of cone photoreceptor cells in Arsg KO mice as old as 24 months, demonstrating that rod photoreceptor cells comprise the retinal cell type that is affected primarily in the absence of ARSG. Evidence for a significant loss of other retinal cell types than rod photoreceptors was not observed in mutant mice at this age, as indicated by the normal numbers of cone photoreceptor cells, bipolar cells, and RGCs, and the normal thickness of the inner retina. Of note, retinal degeneration displays the typical features of a rod-cone dystrophy also in other MPS subtypes, in animal models and in patients.^{26,28,29,44–46} 

While immunohistochemical analyses revealed detectable levels of ARSG expression in RPE cells only, elevated levels of M6P, Saposin D, and Cathepsin D were observed mainly in the inner retina, particularly in RGCs. The expression of most lysosomal proteins is coordinated and regulated by the transcription factor EB (TFEB).⁴⁷ Therefore, RGCs and other retinal cell types might respond to extrinsic pathological stimuli by upregulating TFEB-mediated lysosomal biogenesis or react to intrinsic alterations in the endolysosomal/lysosomal system, such as a subtle accumulation of 3-O sulfated heparan sulfate. The latter hypothesis implies weak expression of ARSG below the detection level of our immunohistochemical analyses in retinal cell types other than RPE cells. 

A characteristic feature of degenerating Purkinje cells in the cerebellum of Arsg KO mice was the presence of large intracellular aggregates that were immunoreactive for ubiquitin and p62, an autophagy adapter protein.¹² We hypothesized that these aggregates were the result of an impaired clearance of damaged lysosomes by autophagy. In the retina, however, similar p62-positive aggregates were not observed, indicating that neuronal cell death in the cerebellum and retina follows different mechanisms. 

Mucopolysaccharidoses IIIC is caused by mutations in HGSNAT, the gene encoding heparan-α-glucosaminide N-acetyltransferase, and retinal degeneration is among the typical symptoms of MPS IIIC patients.^24,31 Interestingly, a recent study identified novel mutations in HGSNAT in six patients that presented with retinitis pigmentosa but without any other clinical symptoms normally associated with MPS IIIC, such as neurological deterioration or visceral manifestations.²⁹ The mutations led to significantly reduced HGSNAT activities in these patients, ranging slightly above the level of typical MPS IIIC patients but considerably below the level of healthy subjects. As none of the patients manifested additional extraocular symptoms, the authors concluded that tissues usually affected by mutations in HGSNAT (especially the brain) express sufficient levels of residual enzymatic activity, but that the retina requires higher levels of HGSNAT activity to maintain normal structure and function.²⁹ Basically similar findings have been reported recently for another lysosomal storage disorder, CLN7 disease.²⁹ Similar to MPS IIIC, progressive photoreceptor loss is a typical feature of CLN7 disease in patients⁴⁸ and in a mouse model of this condition.³³ Using genome-wide linkage analysis and exome sequencing, the study identified compound heterozygous variants in MFSD8, the gene affected in CLN7 disease, in two families presenting with macular dystrophy with central cone involvement.⁴⁹ Characteristic neurologic symptoms normally associated with CLN7 disease, including mental regression, motor impairment or seizures were, however, not observed in these patients. Because both families carried a severe heterozygous mutation in combination with a missense mutation predicted to have a mild effect on the protein, it was proposed that there was sufficient residual activity of MFSD8 in all tissues of the patients, except in the retina.⁴⁹ Together, these studies point to an already high susceptibility of the retina to subtle changes in the lysosomal system, and, thus, might provide an explanation for the frequent involvement of the retina in various LSDs and for the progressive photoreceptor loss in Arsg KO mice despite the absence of detectable lysosomal storage. 

American Staffordshire Terrier dogs suffering from ataxia have been shown to carry a point mutation in Arsg in a triplet coding for an amino acid in the vicinity of the catalytic domain of the protein.¹³ Evidence was presented that this missense mutation resulted in a significant reduction of ARSG activity.¹³ In close analogy to Arsg KO mice,^6,12 affected dogs showed marked Purkinje cell degeneration in the cerebellum.¹³ However, different from our observations in the mouse model, retinal degeneration was not observed in this canine model of MPS IIIE.¹³ Similar to photoreceptor cells, Purkinje cells are affected in various LSDs,^40,50–55 indicating that this neuronal cell type also is highly susceptible to lysosomal dysfunction. Thus, it is tempting to speculate that the species-specific differences in the phenotypic expression of ARSG dysfunction are related to the residual enzyme activity in dogs which is sufficient to maintain photoreceptor cells but not Purkinje cells as opposed to the complete absence of ARSG in the Arsg KO mouse where both nerve cell types are affected. Alternatively, these findings may reflect species-specific differences in the functional relevance of ARSG for photoreceptor cell integrity. 

In conclusion, the present study demonstrates an early onset retinal degeneration in Arsg KO mice that in many aspects resembles the retinal dystrophy observed in other MPS III types and other LSDs. We suggest that the retinal phenotype of ARSG-deficient mice described in the present study might be helpful in the identification of possible human subjects suffering from MPS IIIE caused by mutations in ARSG. 

Acknowledgments

The authors thank Konrad Sandhoff for the gift of the Saposin D antibody, Thomas Braulke for providing the Cathepsin D and M6P antibodies, and Stephan Storch for the CLN7 mutant mice. The authors also thank Elke Becker, Sabine Helbing, Stefanie Schlichting, and Dagmar Niemeier for the excellent technical support, and Marion Knufinke and Christiane Grebe for animal care. 

Supported by the Deutsche Forschungsgemeinschaft Grant DI 575/6 (TD). 

Disclosure: K. Kruszewski, None; R. Lüllmann-Rauch, None; T. Dierks, None; U. Bartsch, None; M. Damme, None