Lack of Delta-Sarcoglycan (Sgcd) Results in Retinal Degeneration

Age-related macular degeneration (AMD) is the leading cause of central vision loss and severe blindness among the elderly population. Recently, we reported on the association of the SGCD gene (encoding for δ-sarcoglycan) polymorphisms with AMD. However, the functional consequence of Sgcd alterations in retinal degeneration is not known. Herein, we characterized changes in the retina of the Sgcd knocked-out mouse (KO, Sgcd−/−). At baseline, we analyzed the retina structure of three-month-old wild-type (WT, Sgcd+/+) and Sgcd−/− mice by hematoxylin and eosin (H&E) staining, assessed the Sgcd–protein complex (α-, β-, γ-, and ε-sarcoglycan, and sarcospan) by immunofluorescence (IF) and Western blot (WB), and performed electroretinography. Compared to the WT, Sgcd−/− mice are five times more likely to have retinal ruptures. Additionally, all the retinal layers are significantly thinner, more so in the inner plexiform layer (IPL). In addition, the number of nuclei in the KO versus the WT is ever so slightly increased. WT mice express Sgcd-protein partners in specific retinal layers, and as expected, KO mice have decreased or no protein expression, with a significant increase in the α subunit. At three months of age, there were no significant differences in the scotopic electroretinographic responses, regarding both a- and b-waves. According to our data, Sgcd−/− has a phenotype that is compatible with retinal degeneration.

knocked-out mouse (KO) compared to the WT ( Figure 1A, arrows). These observations were not associated with the slide-processing technique, since these results were replicated multiple times by experienced technicians. Instead, these results are in support of a frailer tissue. On higher magnification, there is a marked decrease in retinal thickness in the Sgcd −/− group (Panel B). To further dissect these observations, experienced technicians quantified each layer. Moreover, we corroborated these measures with an automated method of image processing in Mathlab (Mathworks, Natick, MA, USA) and built an algorithm to quantify any solution of continuity (See Methods, Section 4.3). The structural characterization of both groups is further described in Table 1. technicians. Instead, these results are in support of a frailer tissue. On higher magnification, there is a marked decrease in retinal thickness in the Sgcd −/− group (Panel B). To further dissect these observations, experienced technicians quantified each layer. Moreover, we corroborated these measures with an automated method of image processing in Mathlab (Mathworks, Natick, MA, USA) and built an algorithm to quantify any solution of continuity (See Methods, Section 4.3). The structural characterization of both groups is further described in Table 1.   3 13.0 ± 5.0 0.0 ± 3.0 <0.0001 Overall, there is a significant reduction in retinal thickness. On average, KO mice have ~100 µm less than wild type (p < 0.0001). This difference holds on different magnitudes for all retinal layers except for the ganglion cell layer (GCL) ( Table 1). The lack of statistical significance in GCL is likely related to its monocellular layer structure or the difficulty in measuring it precisely with an operator or an automated method. Moreover, our results suggest that the retina of the KO mice is frail since they had a median number of retinal ruptures of 13, while the WT mice had no quantifiable solution of continuity in the H&E slides ( Figure 1 Panel A, Table 1).  Overall, there is a significant reduction in retinal thickness. On average, KO mice have~100 µm less than wild type (p < 0.0001). This difference holds on different magnitudes for all retinal layers except for the ganglion cell layer (GCL) ( Table 1). The lack of statistical significance in GCL is likely related to its monocellular layer structure or the difficulty in measuring it precisely with an operator or an automated method. Moreover, our results suggest that the retina of the KO mice is frail since they had a median number of retinal ruptures of 13, while the WT mice had no quantifiable solution of continuity in the H&E slides ( Figure 1 Panel A, Table 1).
Moreover, to examine the thinned retina of the KO mice, we estimated the outer-and inner-nuclear cell layer densities (per 100 µm 2 ) using our automated method. Based on the number of data collected, there are significant differences in cell density in both layers between groups. The mean differences by genotype are displayed in Table 1.

Sarcoglycan Complex Expression and Localization by Immunofluorescence among Three-Month-Old Mice
To characterize the impact of Sgcd protein's absence in the KO mice, we began by examining the distribution and expression of the sarcoglycan-sarcospan (Sg-Sspn) protein complex composed by, α-, β-, γ-, δ-, ε-, and ζ-Sg and Sspn [14,15]. Here, we only studied the first five, since ζ-Sg was the last to be discovered, and we had limited access to antibodies. Excluding α-Sg, the rest of the sarcoglycans were significantly under-expressed in the KO compared to the WT, as demonstrated by immunofluorescence assays (Figure 2A). We observed the distribution of the Sg in the WT murine retina localized to the photoreceptor (PR), outer and inner plexiform, and ganglion cell layers in retinal slides (Figure 2 arrows). Our positive findings in the PR layer could indicate expression in the PR cells itself or, most likely, their positivity in the outer limiting membrane (Müller glial cells). So far, we are limited to indicate the approximate region of Sg-Sspn expression in the murine retina. Regardless, the absence of Sgcd leads to a significant decrease in the Sg-Sspn complex. Unlike the other Sgs, α-Sg is overtly expressed in the KO compared to the WT. This subunit is localized to the outer plexiform and possibly GCL ( Figure 2). Regarding Sspn, the KO seems to have a downwards expression, but our measurements were not significantly different from the WT. Moreover, to examine the thinned retina of the KO mice, we estimated the outer-and innernuclear cell layer densities (per 100 µm 2 ) using our automated method. Based on the number of data collected, there are significant differences in cell density in both layers between groups. The mean differences by genotype are displayed in Table 1.

Sarcoglycan Complex Expression and Localization by Immunofluorescence among Three-Month-Old Mice
To characterize the impact of Sgcd protein's absence in the KO mice, we began by examining the distribution and expression of the sarcoglycan-sarcospan (Sg-Sspn) protein complex composed by, α-, β-, γ-, δ-, ε-, and ζ-Sg and Sspn [14,15]. Here, we only studied the first five, since ζ-Sg was the last to be discovered, and we had limited access to antibodies. Excluding α-Sg, the rest of the sarcoglycans were significantly under-expressed in the KO compared to the WT, as demonstrated by immunofluorescence assays (Figure 2A). We observed the distribution of the Sg in the WT murine retina localized to the photoreceptor (PR), outer and inner plexiform, and ganglion cell layers in retinal slides (Figure 2 arrows). Our positive findings in the PR layer could indicate expression in the PR cells itself or, most likely, their positivity in the outer limiting membrane (Müller glial cells). So far, we are limited to indicate the approximate region of Sg-Sspn expression in the murine retina. Regardless, the absence of Sgcd leads to a significant decrease in the Sg-Sspn complex. Unlike the other Sgs, α-Sg is overtly expressed in the KO compared to the WT. This subunit is localized to the outer plexiform and possibly GCL ( Figure 2). Regarding Sspn, the KO seems to have a downwards expression, but our measurements were not significantly different from the WT.

Changes in Sarcoglycan Complex Protein Expression in Dissected Retinas of Three-Month-Old Mice
To further support our findings of the Sg-Sspn complex expression in retinal slides by immunofluorescence (IF), we extracted retinas from WT and KO mice and performed Western blot to analyze the protein levels of the complex. We used skeletal muscle lysates from both WT and KO mice as a positive control for the experiment. The Sgs and Sspn are expressed in the WT mouse retina. In support of our immunofluorescence results, the Sgcd protein knock-out leads to a down-regulation of all subunits in the protein complex except for α-Sg ( Figure 3A right panel and Figure 3B). Similarly, Sspn was not significantly different among genotypes. The uncut scans of our Western blot images are presented in the supplementary figures S1-S7.

Changes in Sarcoglycan Complex Protein Expression in Dissected Retinas of Three-Month-Old Mice
To further support our findings of the Sg-Sspn complex expression in retinal slides by immunofluorescence (IF), we extracted retinas from WT and KO mice and performed Western blot to analyze the protein levels of the complex. We used skeletal muscle lysates from both WT and KO mice as a positive control for the experiment. The Sgs and Sspn are expressed in the WT mouse retina. In support of our immunofluorescence results, the Sgcd protein knock-out leads to a down-regulation of all subunits in the protein complex except for α-Sg (Figure 3A right panel and Figure 3B). Similarly, Sspn was not significantly different among genotypes. The uncut scans of our Western blot images are presented in the Supplementary Figures S1-S7.
The α-Sg/Gapdh median normalized expression is significantly higher in the Sgcd −/− mice compared to the control (~0.14 units of difference, p = 0.0286, Table 2). The rest of the Sgs show a lower level in the Sgcd −/− mice, with the most evident change observed in γ-Sg (−0.7787, p = 0.0286). A detailed description of the normalized protein expressions of the Sg-Sspn complex is tabulated in Table 2. Sspn was not significantly different between genotype groups. to analyze the protein levels of the complex. We used skeletal muscle lysates from both WT and KO mice as a positive control for the experiment. The Sgs and Sspn are expressed in the WT mouse retina. In support of our immunofluorescence results, the Sgcd protein knock-out leads to a down-regulation of all subunits in the protein complex except for α-Sg ( Figure 3A right panel and Figure 3B). Similarly, Sspn was not significantly different among genotypes. The uncut scans of our Western blot images are presented in the supplementary figures S1-S7.

Effect of Sgcd Knock-Out in Retinal Function by Electroretinography among Three-Month-Old Mice
Finally, to assess the significance of Sgcd loss on retinal function, we performed scotopic electroretinography (ERG) assessments in both groups at different light intensities (Figure 4). We evaluated both genotypes at three months of age. At this age, no significant differences were evident in retinal function in a-, b-, and c-waves (Figure 4).   The α-Sg/Gapdh median normalized expression is significantly higher in the Sgcd −/− mice compared to the control (~0.14 units of difference, p = 0.0286, Table 2). The rest of the Sgs show a lower level in the Sgcd −/− mice, with the most evident change observed in γ-Sg (−0.7787, p = 0.0286). A detailed description of the normalized protein expressions of the Sg-Sspn complex is tabulated in Table 2. Sspn was not significantly different between genotype groups.

Effect of Sgcd Knock-Out in Retinal Function by Electroretinography among Three-Month-Old Mice
Finally, to assess the significance of Sgcd loss on retinal function, we performed scotopic electroretinography (ERG) assessments in both groups at different light intensities (Figure 4). We evaluated both genotypes at three months of age. At this age, no significant differences were evident in retinal function in a-, b-, and c-waves (Figure 4).

Discussion
Age-related macular degeneration is the leading cause of irreversible blindness in developed countries. AMD prevalence is rising due to population aging and a lack of effective treatment [1,16]. We recently described a possible association between the SGCD gene (rs931798 polymorphism) with increased odds of geographic atrophy AMD and a haplotype configuration with lowered odds of

Discussion
Age-related macular degeneration is the leading cause of irreversible blindness in developed countries. AMD prevalence is rising due to population aging and a lack of effective treatment [1,16]. We recently described a possible association between the SGCD gene (rs931798 polymorphism) with increased odds of geographic atrophy AMD and a haplotype configuration with lowered odds of disease (GATT: rs970476, rs931798, rs140617, rs140616) [8]. Thus, to investigate the contribution of this gene to the pathogenesis of AMD, here we analyzed a KO mouse and based on our findings proposed it as a model of retinal degeneration. Our results showed a statistically significant reduction in the retinal thickness of Sgcd −/− compared to the Sgcd +/+ . Interestingly, this retinal thinning is similar to the clinical findings of geographic AMD (GA/AMD) and diabetic retinopathy cases [17][18][19]. We speculate that δ-Sg s absence leads to a loss of structural proteins or the dysregulation of signaling pathways that could lead to a thinner retina. This assumption is based on our earlier finding of SGCD s polymorphisms associated with GA/AMD. [8] A new bioinformatic report described Sgcd s involvement in tissue regeneration, developmental growth, cell proliferation, and differentiation [20]. All these functions could explain our finding of a thinner frailer retina. However, further studies are warranted. Likewise, there was a significant increase in retinal frailty and outer and inner nuclear layer nuclear densities. Regarding the protein expression of the Sg-Sspn complex, we observed a significant downwards expression of β-, γ-, and ε-Sg in the KO. These findings in these subunits are comparable to the skeletal and cardiac muscle expression in the Sgcd −/− mouse [21,22]. We are first to report these protein alterations in the retina for this mouse.
The Sg-Sspn complex expression is partially independent of dystrophin in the mouse retina [12]. Based on its relationship and cellular membrane location, this complex might provide retinal stability through the binding of the cytoskeleton with the extracellular matrix [12,23]. Consequently, a decrease in the expression of the complex's subunits could be associated with higher retinal frailty, as observed here. Interestingly, we did not detect a significant change in Sspn retinal expression. This finding is different from the effect of knocking out Sgcd in skeletal and cardiac muscle, where it leads to a decrease or absence of Sspn [24]. Sspn gene deletion does not modify the expression of the Sg complex, which implies that its expression and assemblage are independent, and our results are in support of this hypothesis [24]. We observed by Western blot and immunofluorescence an increased α-Sg expression in the retina of the Sgcd −/− mouse. This is the first report of α-Sg over-expression in the murine retina. Our results suggest that this protein could play a role in the morphological findings when Sgcd is absent. α-Sg might compensate for a decrease in ε-Sg expression, since αand ε-Sg share an amino acid identity of almost 50% [22]. Hence, our findings foster the hypothesis that these proteins might constitute different complexes in the mouse retina. However, future studies should test this assumption.
Finally, we did not find evidence of any difference in retinal function, as measured by ERG, between the Sgcd KO and the WT. We possibly under-detected any functional difference owing to the early age of our mice. Commonly AMD (characterized by retinal degeneration) is diagnosed around age 65, and its prevalence increases with age. However, clinically, in early AMD phenotypes, there is an anatomical thinning of retinal layers [17,18,25]. It remains unclear if, in earlier stages of the disease, there is a measurable impact in retinal function. Another limitation of our study is the broad assessment of the Sg-Sspn protein distribution in retinal slides. So far, our positive results in the photoreceptor layer could indicate that some proteins of the complex could be either expressed in this cell or Müller glial cells. However, we feel that our approach with two complementary methods (IF and WB) is a first step in the complete assessment of the role of the Sg-Sspn in the murine retina. Future studies could comprehensively study the molecular underpinnings surrounding retinal degeneration in the Sgcd −/− mouse.

Study Design and Ethical Approval
We performed an animal study of 42 C57BL/6J mice, half wild-type (WT, Sgcd +/+ , n = 21) and  [26] and the Code of Practice for the Housing and Care of Animals Bred, Supplied or Used in Scientific Purposes [27].

DNA Extraction
To genotype our mice, we began by extracting genomic DNA by tailing. Approximately 3 mm of the tip of the tail was cut under local anesthesia and collected in 1.5-mL tubes. Then, we immediately washed three times all the tissues with phosphate-buffered saline (PBS) 1X and dehydrated with 90 µL of 5 mM of NaCl solution. Shortly after, the tail was crushed with an aluminum shank. Then, we performed cellular digestion by adding 37.5 µL of 10% Sodium Dodecyl Sulfate (SDS) and 307.5 µL of saturated NaCl. Tubes were mixed using a vortex and centrifuged for 10 min at 4 • C and 14,000 rpm. The supernatant was recovered in new 1.5-mL collection tubes, and DNA was precipitated with 1 mL of 4 • C absolute ethanol. Afterward, we centrifuged the DNA for 1 min at full speed and removed the supernatant. We washed the pellet with 70% ethanol and centrifuged for 10 min at full speed. The supernatant was decanted for allowing DNA to dry for 5 min. Finally, genomic DNA was resuspended in 20 µL of DNAse free water. DNA concentrations and purity were quantified using a Multiskan™ GO Spectrophotometer (Thermo Fisher Scientific Inc., Wilmington, DE, USA). After quantification, the DNA was diluted to a concentration of 150 ng/µL. Finally, to evaluate DNA integrity, we prepared a 1% agarose (Thermo Fisher Scientific Inc., Wilmington, DE, USA) gel stained with 0.5 µg/mL ethidium bromide.

Morphological Analysis
To characterize the effect of Sgcd on retinal structure, we performed a morphological analysis by firstly measuring the retinal thickness (4.3.1), then evaluating for retinal frailty (4.3.2), and finally estimating the nuclei density in the outer (ONL) and inner nuclear layers (INL) (4.3.3). We began by enucleating our groups following the ARVO statement and peer-reviewed protocols [26,28]. Both eyes from each mouse were fixed in 4% paraformaldehyde overnight at 4 • C. Then, all tissues were dehydrated and embedded in paraffin. We performed serial transversal sections of 2.5 µm, taking the optic nerve as a reference. Half of the slides were stained with hematoxylin and eosin (H&E) for aim 4.3 and the rest for immunofluorescence (4.4).

Retinal Thickness
We quantified the full retinal thickness and each layer by light microscopy at ×10 magnification using the AxioVison Rel. 4.8 software (Carl Zeiss Inc., Thornwood, NY, USA). On average, we obtained three images and three measurements of each unit of analysis (eye). Then, we measured, per standard protocol, 500 µm to the right and left of the optic nerve.

Frailty Index
We previously observed that Sgcd −/− mice tissues, especially neural, were frail compared to wild type [29]. We here aimed to quantify if there were any differences in the light microscopy of full-retinal segments. We began by assessing H&E stained slides, which was done by experienced blinded technicians, taken at ×5 in MatLab (Mathworks, Natick, MA, USA). We visualized each image using a script of Matlab (Mathworks, Natick, MA, USA), where we selected and extracted all retinal layers. Then, we transformed these photographs from the RGB color model to the HSV color model. Then, we obtained a grayscale image, representing the luminance of the RGB image, allowing us to highlight the regions of interest [30,31]. Considering the shape and orientation of the retina in the eye, we applied image-processing techniques to estimate a curve that encodes the shape and direction of the analyzed layer of the retina. Following this curve, we can analyze the changes in the intensity of the pixels along the given direction, where the discontinuities will be those pixels where its intensity difference with its neighbors is greater than a threshold. Finally, we quantified the number of discontinuities to describe the grade of the frailty of this layer.

Nuclei Density
To quantify ONL and INL nuclear density, we analyzed H&E slides taken at 40× magnification in MATLAB (Mathworks, Natick, MA, USA). We began by visualizing each image firstly on the script. Then, we transformed each photograph from RGB to HSV format to obtain a grayscale image similar to 4.3.2. Afterward, we selected and extracted from the original image the area between the ONL and INL. These images were binarized to cluster pixels into two categories, (1) layer nuclei equal to zero and (2) cytoplasm and extracellular matrix equal to 255, in the grayscale. Then, we applied a segmentation algorithm to extract the outer plexiform layer to obtain two images corresponding to the nuclear layers. Finally, we quantified, using the method of Wang et al. [32], in each layer all the pixels belonging to the nuclei. We examined four images per eye taken around 500 µm of the right and left of the optic nerve.

Electroretinographic (ERG) Assessments
To assess the role of Sgcd in retinal function, we performed ERG assessments in both groups. Mice were initially maintained in a dark room overnight. On the experimentation day, we then anesthetized with 1 µL/g body weight intraperitoneally of a combination of 70% ketamine (1000 mg/10 mL) and 30% xylazine (2%). The pupils were dilated with 0.5% tropicamide and 0.5% phenylephrine. We recorded flash ERG responses from both eyes by a silver chloride ring electrode placed on the cornea. Two reference electrodes were positioned subcutaneously near the eyes. All procedures were performed under dim red light. The light stimulation included a 1-ms flash with an intensity of 0.3, 0.6, 0.9 and 1.2 log cd.s/m 2 (PS33 Plus PhotoStimulator, GRASS Technologies, Warwick, RI, USA). The bandpass was set at 3 to 300 Hz (P511AC Amplifier, GRASS Technologies). We averaged 16 responses, as previously described in a work of ours [33]. We analyzed our data and calculated A and B-wave amplitudes and latency in MatLab (Mathworks, Natick, MA, USA) following a standardized approach. [34]

Statistical Analysis
To begin our approach, we verified the normal distribution of our data by plotting and Kolmogorov-Smirnov tests. Most of our data was normally distributed; hence, when the sample size was greater than 30 units of observation, we applied parametric methods to describe the effect of Sgcd or their lack of on several anatomic and electrophysiologic measures, e.g., means and standard deviations, Student's t-test, ANOVA with Bonferroni-corrected p-values. However, when the sample size was <30 or the data was not normally distributed nor adequately transformed, we applied nonparametric statistics for all our inferences, e.g., medians and interquartile range, Wilcoxon two-sample test, and two-way Friedman's ANOVA with Bonferroni-corrected pairwise comparisons.