Changes in Starburst Amacrine Cells in Mice with Diabetic Retinopathy

Background : Neurodegenerative diseases, such as diabetic retinopathy (DR) and glaucoma, induce retinal neuron loss. Acetylcholine-containing cholinergic neurons, known as starburst amacrine cells (SACs), play critical roles in the generation of precise neuronal activity in the retina and are located in the inner nuclear layer (INL, conventional) and ganglion cell layer (GCL, displaced). Methods : This study investigated the loss of and morphological changes in SACs in the retinas of streptozotocin (STZ)-induced diabetic and insulin-deficient C57BL/6-Tg(pH1-siRNA insulin /CMV-hIDE)/Korl (IDCK) mice. SACs were immunocytochemically localized with anti-choline acetyltransferase (ChAT) antibody, and ChAT-labeled cells in the INL and GCL in the control and experimental groups were counted along the central vertical meridian in the whole-mounted retina using conventional fluorescent or confocal microscopes. Results : ChAT-immunoreactive (IR) neurons in STZ-induced diabetic mouse retina decreased by 8.34% at 4–6 weeks and by 14.89% at 42 weeks compared with the control group. Localized ChAT-IR neuron counts in the retinas of 20-week-old IDCK mice were 16.80% lower than those of age-matched control mice. Cell body deformation and aggregation were detected in the retinas of mice with DR. Single-cell injection experiments revealed the loss and deformation of dendritic branches in ChAT-IR neurons in DR. All ChAT-IR neurons expressed the calcium-binding protein calretinin, whereas no ChAT-IR neuron colocalized with calbindin-D28K or parvalbumin. Conclusions : Our results revealed that the neurodegenerative effects of the loss and deformation of ChAT-IR neurons can provide a reference for future study of this disease.


Introduction
Diabetic retinopathy (DR) is a major complication of diabetes that can lead to progressive vision loss and blindness [1][2][3][4]. Approximately one-third of individuals with diabetes develop DR, the treatment of which requires multiple options due to its complex and multifactorial pathogenesis [3,5]. Hyperglycemia, which induces superoxide production and oxidative stress, plays a central role in the development of vascular alterations [6,7]. Diabetic microvascular complications typically include neuropathy, nephropathy, and retinopathy, whereas macrovascular complications include cerebrovascular, cardiovascular, and peripheral complications [8,9]. DR is also a strong predictor of microvascular and macrovascular complications [8,10,11].
DR has been classified into two types: nonproliferative diabetic retinopathy (NPDR) and proliferative diabetic retinopathy (PDR). NPDR occurs in the first stage of DR and is characterized by altered vascular permeability, basement membrane thickening, pericytes loss, and acellular capillary formation. NPDR can be further divided into three types based on its progression: mild NPDR, moderate NPDR, and severe NPDR. PDR is an advanced stage of DR and is characterized by neovascular complications and may lead to retinal detachment. In any stage of DR, diabetic macular oedema may develop due to the accumulation of exudative fluid in the macula [1,3,12,13].
A growing body of research have reported that neurodegeneration plays a significant role in the development of DR [6,14]. DR is a highly specific neurovascular complication accompanied by progressive damage to retinal neurons, including such as photoreceptors, horizontal, bipolar, amacrine, and retinal ganglion cells (RGCs), and glial cells, such as Müller cells, astrocytes, and microglia [2,15]. Previous studies have shown that the loss of retinal cells is associated with increased apoptosis, decreased thickness, and cell density reduction [16][17][18][19][20][21]. Neurodegeneration in DR occurs in early stages even before vascular lesions can be detected [3,22,23]. However, there is still a debate over whether microvasculopathy causes neurodegeneration or vice versa or if they are mutually independent [3,22,24,25]. Streptozotocin (STZ), initially isolated from Streptomyces achromogenes, is a naturally occurring chemical agent that is particularly toxic to the insulin-producing beta cells in the pancreatic islet and is the most prominent diabetogenic chemical compound in diabetes research [26]. Many studies have used STZ-induced diabetic animal models, especially mouse and rat models of DR [16][17][18]20,[27][28][29][30]. For example, periodic progression of cellular and vascular lesions in STZ-induced DR has been reported [27]. In mice with STZ-induced diabetes, the number of cells in the ganglion cell layer (GCL) was reduced [16,18]. In addition, in rats with STZ-induced diabetes, retinal cell apoptosis was increased, whereas the total retinal thickness was decreased [17,20,28]. The relationship between vascular damage and neuronal changes, such as apoptosis of retinal neurons, has also been investigated using mice with STZ-induced diabetes [29]. Basement membrane thickening of arterial and venous capillaries has been observed in the retina of rats with STZ-induced diabetes [30].
Starburst amacrine cells (SACs) release two neurotransmitters, the excitatory neurotransmitter acetylcholine and the inhibitory neurotransmitter gamma-aminobutyric acid (GABA). SACs are also called cholinergic amacrine cells because they are the only retinal cells that release acetylcholine [31][32][33][34]. SACs are distributed regularly in the retina [34][35][36][37]. SACs in the GCL are called displaced SACs or OFF-SACs, and those in the inner nuclear layer (INL) are called conventional SACs or ON-SACs. SACs play several important roles, including a major role in the detection of motion images in the retina [38,39].
The death of several retinal cell types with the progression of DR and the loss of ChAT-IR neurons in the retinas of 24-week-old Ins2 AKITA diabetic mice has been demonstrated [17,49]. However, there are no reports on the loss and changes in SACs in the retinas of STZ-induced diabetic mice or transgenic diabetic models in the early or late stages of DR. Hence, we aimed to investigate the loss of and morphological changes in SACs in the early (4-6 weeks) and late stages (42 weeks) of DR in mice with STZ-induced diabetes. Moreover, we examined 20-week-old insulindeficient C57BL/6-Tg(pH1-siRNA insulin /CMV-hIDE)/Korl (IDCK) mouse retinas and compared the results with previously reported findings in Ins2 AKITA diabetic mice. In addition, we investigated whether ChAT-IR neurons in the mouse retina express CB, CR, or PV. The findings are expected to provide information about the loss of and morphological changes in SACs in diabetic mouse retina and clarify the expressions of CB, CR, and PV in ChAT-IR neurons in the mouse retina.

Animals
All experiments involving animals were approved by the Animal Care and Use Committee of Kyungpook National University (permission no. 2020-0158). Animals were group housed under a 12-h light:12-h dark cycle until used for studies. Temperature and humidity levels in animal housing facilities ranged from 23°C to 26°C and from 45% to 65%, respectively.

STZ-Induced Diabetic Mice and Experimental Design
Adult mice (C57BL/6J, 8-10 weeks old, weighing 20-30 g) were examined. After 4 h of fasting, we measured their weight and blood glucose levels and randomly categorized the animals into experimental and control groups. Experimental mice were intraperitoneally administered 6 mg/ml STZ dissolved in 50 mM sodium citrate buffer (pH 4.5), and control animals were intraperitoneally administered 50 mM sodium citrate buffer (pH 4.5) in reference to standard protocol of previous study [27]. We calculated the amount of injection solution at ratio of 40 mg/kg of weight of each mouse and injected via intraperitoneal injection using 1 mL syringe. The aforementioned process was repeated for 5 days. After injection, we checked the weight and fasting blood glucose level of each mouse every 5 days. Blood was obtained from the tail vein. Two weeks after the injection, mice with blood glucose levels >250 mg/dL were considered diabetic (Table 1).

IDCK Mice
IDCK mice were obtained from the National Institute of Food and Drug Safety Evaluation (Cheongju, Korea). IDCK mice with blood glucose levels >240 mg/dL were used as experimental mice, and age-matched C57BL/6N mice with blood glucose levels <200 mg/dL were used as control mice (

Fluorescence Immunocytochemistry
Retinal whole mounts were processed free-floating in small vials at room temperature with gentle agitation. With three rinses in ice-cold 0.1 M PB between each step, the retinas were processed as follows: (1) freezing and thawing three times for better antibody penetration; (2) preincubation in 1% sodium borohydride for 20-30 min; (3) incubation in 0.1 M PB with 4% normal serum from the host of secondary antibody with 0.5% Triton X-100 for 1 day; (4) incubation in the primary antiserum in blocking solution for 3 days; (5) incubation in the secondary antiserum in blocking solution for 1 day; and (6) staining of the nuclei of the retinas with DAPI (1:1000). The primary antibodies used in this study were rabbit anti-CB . For CB, CR, and PV, fluorescein isothiocyanate (FITC)-conjugated anti-rabbit IgG (Jackson ImmunoResearch Inc., Baltimore, PA, USA) secondary antibody was used. To identify ChAT, FITCconjugated goat anti-rabbit IgG secondary antibody (Millipore) and Cy3-conjugated donkey anti-goat IgG (Jackson ImmunoResearch Inc.) secondary antibodies were used.

Single-Cell Injection
The procedures for single-cell injection following immunocytochemistry have been described in our previous reports [51][52][53]. Immunocytochemistry before the single-cell injection procedure was performed without Triton X-100 because this detergent can influence the diffusion of the dye in the neuronal membrane. The fixed tissues were incubated in a 1:200 dilution of goat anti-ChAT (Millipore) primary antibody in 0.1 M PB for 2 h at room temperature and then incubated in a 1:50 dilution of FITC-conjugated donkey anti-goat IgG secondary antibody (Jackson ImmunoResearch Inc.) in 0.1 M PB for 2 h at room temperature after three rinses with 0.1 M PB. The dish containing the immunolabeled retinal tissue was placed on a microscope stage, and FITC-labeled cholinergic neurons were viewed under a Zeiss 40× Plan Achroplan (NA 0.80) water immersion lens and Zeiss filter set 09 (excitation, 450-490 nm; emission, 515 nm). Single ChAT-labeled cells in the GCL were randomly selected. The cells were filled by iontophoretically passing a positive current of 5-20 nA through a micropipette containing a 1% solution of the lipophilic dye DiI (Molecular Probes, Eugene, OR, USA). DiI-filled cells were viewed using Zeiss filter set 20 (excitation, 540-552 nm; emission, 575-640 nm), and the optimal filling time was usually 15-30 min. We only injected cells in the mid-periphery of the retina (1.0-1.7 mm peripheral to the optic disc) because the dendritic field size varies with retinal eccentricity. After several cells in the retina were filled, the tissue was fixed for 2 h in 4% paraformaldehyde. The tissue was washed three times in 0.1 M PB, coverslipped with Vectashield (Vector Laboratories, Burlingame, CA, USA), and sealed with nail polish.

Microscopy and Quantitative Analysis
We counted ChAT-IR neurons manually in wholemounted retinas from the center to the periphery. As the density of retinal neurons changes with eccentricity, we carefully chose counting areas by positioning at the same retinal regions between control and experimental mice. We counted cells in fourteen sample areas at 300 µm intervals throughout the dorsoventral meridians of the retina with 200 × 200 µm 2 fields from three retina each of control and experimental mice. Cell density was expressed as the number of ChAT-IR neurons/mm 2 . The ratios were evaluated statistically using paired student t-test between control and experimental mice. ChAT-IR neuron distribution and doublelabeling percentage were examined using a Zeiss Axioplan microscope (Carl Zeiss Meditec, Inc., Jena, Germany). Fluorescence images were photographed with a Zeiss Axioplan microscope using a 40× objective or Zeiss LSM800 laserscanning confocal microscope (Carl Zeiss Meditec, Inc.) using a 20× or 40× objective.
The density of ChAT-IR neurons was reduced in the GCL and INL in STZ-induced diabetic mice at 4-6 and 42 weeks and in IDCK mice at 20 weeks compared with that in control mice (Figs. 1,2,    2 shows the result of the paired student t-test in the graph. Statistically significant differences in mean density per retina between experimental and control groups in STZinduced diabetic mice at 4-6 weeks and 42 weeks and in 20-week-old IDCK mice were identified (p < 0.05) (Fig. 2,  Supplementary Table 4). At 4-6 weeks, mean ChAT-IR neuron densities were 998 ± 34 cells/mm 2 in the GCL and 1179 ± 59 cells/mm 2 in the INL in the three retinas in STZinduced diabetic mice, whereas these values were 1076 ± 15 cells/mm 2 in the GCL and 1288 ± 54 cells/mm 2 in the INL in control mice (Fig. 2). At 42 weeks, mean ChAT-IR neuron densities were 907 ± 40 cells/mm 2 in the GCL and 1052 ± 9 cells/mm 2 in the INL in the three retinas in STZinduced diabetic mice, whereas these values were 1076 ± 27 cells/mm 2 in the GCL and 1226 ± 40 cells/mm 2 in the INL in control mice (Fig. 2). In 20-week-old IDCK mice, mean ChAT-IR neuron densities were 795 ± 38 cells/mm 2 in the GCL and 1046 ± 2 cells/mm 2 in the INL in the three retinas, while these values were 1008 ± 32 cells/mm 2 in the GCL and 1205 ± 36 cells/mm 2 in the INL in control mice (Fig. 2).
ChAT-IR neuron distributions along the dorsoventral axis of the retina in experimental and control mice at 300 µm intervals to the optic nerve head are shown in Fig. 3. The graph shows that the pattern of the numerical distribution of ChAT-IR neurons corresponds to that of typical ChAT-IR neurons in the mouse retina described in our previous study [19]. The graph also shows that the pattern of the numerical distribution of ChAT-IR neurons of the experimental groups in the present study is quite symmetrical. The center-to-periphery gradient along the dorsoventral axis is approximately two for both INL and GCL in the control and experimental groups.

Morphological Changes in ChAT-IR Neurons in DR
Abnormal cell aggregation and cell body deformation in the retina were observed in mice with STZ-induced diabetes (Fig. 4). Fig. 4A,F show the normal cells in GCL and INL, respectively, in C57BL/6J mouse retina. Cells with round/oval cell bodies were appeared to be arranged with adequate distances. They appeared to form independently spaced retinal mosaics. In early-stage (4-6 weeks) and late-stage DR (42 weeks), some ChAT-IR neurons in the GCL (filled arrowheads in Fig. 4C-E) and INL (filled arrowhead in Fig. 4H,J) became distorted. Some other retinal cells (empty arrowheads in Fig. 4C-E) stained with DAPI were also distorted. Abnormal ChAT-IR neuron aggregation was induced by DR in the GCL (arrows in Fig. 4B,D) and INL (arrows in Fig. 4G,I), but this phenomenon was rarely observed in normal retinas. Fig. 5 presents the dendritic morphological differences between normal (Fig. 5A) and damaged (Fig. 5B-H) ChAT-IR neurons from the retinas of control and experimental mice. The distinctive dendritic morphology of SACs has been described in many previous studies [54][55][56][57]. SACs are known to possess four or five primary dendrites that extend radially from the cell body, and dendritic arbors with many branches generally form circular dendritic fields. The dendritic arbors are concentrically and regularly distributed around the soma, and there is almost no overlap of dendritic arbors. The dendritic field has been categorized into three distinct annular zones, namely proximal, intermediate, and distal, which progress radially outward from the soma. The dendrites of the proximal and intermediate areas are relatively thin, and they taper with progression radially outward. The distal area is characterized by terminal varicosities and boutons [55]. Compared   with the dendrites of normal ChAT-IR neurons, those of damaged ChAT-IR neurons in DR start to lose their regular pattern. Fig. 5B,C show the dendritic branches that lost their dendrites at 4 and 6 weeks. Fig. 5D reveals that some of the dendritic fields were clearly altered, and the dendritic branches were entangled and overlapped in space at 6 weeks. Fig. 5E,F reveal significant reductions in dendritic branches in ChAT-IR neurons at 42 weeks. The dendritic arbors varied considerably, and the deprivation of dendritic branches in the distal and intermediate areas was  Fig. 5G,H).

Colocalization of ChAT and CBPs
We double-labeled ChAT with CB, CR, or PV in the mouse retina (Fig. 6). Fig. 6 shows cells expressing ChAT (Cy3), CB (FITC), CR (FITC), and PV (FITC) in both in GCL (Fig. 6A-I) and in INL (Fig. 6J-R). To confirm colocalization, we superimposed the images which were photographed at same location and same focal plane. All ChAT-IR neurons both in GCL (Fig. 6F) and INL (Fig. 6O) contained CR. No ChAT-IR neurons were colocalized with CB (Fig. 6C,L) or with PV (Fig. 6I,R).

Discussion
In this study, we observed changes in the density of ChAT-IR neurons in the retinas of mice with STZ-induced diabetes and IDCK mice. We also detected abnormal cell aggregation, cell body deformation, and morphological changes in dendrites caused by DR. All ChAT-IR neurons expressed CR but not CB and PV.
Amacrine cells represent the most diverse neuronal type in the retina, and approximately 30-60 different types of amacrine cells have been reported in the retina [32,58]. Of these cell types, SACs are one of the most intensively studied interneurons in the retina [59,60]. SACs can be categorized into two subpopulations based on their location (INL or GCL). In the present study, ChAT-IR neurons in retina of mouse with STZ-induced diabetes decreased total 8.34% at 4-6 weeks, and total 14.89% at 42 weeks, respectively. ChAT-IR neurons decreased total 16.80% in retina of 20-week-old IDCK mouse compared to age-matched con-trol mouse retina. Decreased ChAT-IR neuron counts were detected in the early and late phases of DR in STZ-induced diabetic mice and in IDCK mice. In addition, the loss of these cells was greater in the late phase than in the early phase. Furthermore, the loss of ChAT-IR neurons in STZinduced diabetic mice was slightly higher in the GCL than in the INL at 42 weeks, whereas almost no difference was detected at 4-6 weeks. The loss of ChAT-IR neurons in IDCK mice was also greater in the GCL in the present study. However, the loss of ChAT-IR neurons in 24-weekold Ins2 AKITA mice was greater in the INL in a previous study [17]. The functional significance of these subtle differences remains unknown. Other retinal amacrine cells, dopaminergic amacrine cells, which modulate various key visual processes, were decreased by 16.3% in the retinas of 24-week-old Ins2 AKITA mice compared with the control findings [17]. A previous study reported an obvious reduction in RGC counts in DR. In the retinas of STZ-induced diabetes at 10-14 weeks after the first injection, cell counts in the GCL were 20%-25% lower than the control counts [18]. In mice with STZ-induced diabetes, Brn3a-IR RGC counts in the retina were decreased by 7% at 6 weeks after the first STZ injection and those of NeuN-IR RGCs were decreased by 10% compared with the findings in control mice. In the retinas of STZ-treated mice, Brn3a-IR RGC counts in the retina were 15% lower at 12 weeks after the first injection and those of NeuN-IR RGCs were 22% lower than the control values [16]. Previous studies have also documented the apoptotic death of photoreceptors, including increased apoptosis in the outer nuclear layer and thickness reduction in the outer plexiform layer, in STZ-induced diabetic rat retina [20]. The loss of RGCs and reduced thickness of the inner plexiform layer and INL were detected in the retinas of rats with STZ-induced diabetes [21]. The present and previous data indicate that the loss of retinal cells, including amacrine and ganglion cells, has relevance in the progression of DR.
Previous studies have noted retinal cell loss, decreased retinal thickness, and increased apoptosis in mice with STZ-induced diabetes [18]. Increased apoptosis has been demonstrated with TUNEL and caspase-3 staining in the retinas of STZ-induced diabetic rats, Ins2 AKITA mice, and KKAY mice with type 2 diabetes [17,19]. The apoptogenic environment induced by diabetes in the human retina has been detected using apoptotic markers [61]. Also, apoptosis of RGCs in glaucoma has been shown to cause blindness [62]. Previous studies have documented that the number of retinal cells is reduced in DR and various other diseases. For example, RGC body loss and reduced SAC counts have been reported in glaucoma [63]. The number of cells in the GCL was also reduced in ischemia [64], and in the retina in Alzheimer's disease [65,66]. Thus, reduction in retinal cell counts is an apparently common symptom of diverse progressive neurodegenerative and pathologic disorders.
Abnormal aggregation and distorted cell bodies were found in ChAT-IR neurons in DR in the present study. These changes have been reported in CB-, CR-, and PV-IR neurons in the retinas of ischemic rabbits [67]. Abnormal aggregation has also been detected in some CR-IR neurons of the entorhinal cortex in Alzheimer's disease [68]. Hence, abnormal cell aggregation might represent a symptom of neurodegenerative diseases. In the present study, the length of dendrites and the number of branch points in the dendrites of ChAT-IR neurons were decreased in DR. In accordance with the present findings in the retina, the dendritic branching and spine density of neurons have been reported to be reduced in the parietal cortex of rats with diabetes [69]. The number of branch points and the total dendritic length of hippocampal CA3 pyramidal neurons were reduced in diabetic rats [70]. The total dendritic length and spine density of the pyramidal neurons of the prefrontal cortex, occipital cortex, and hippocampus were also lower in rats with diabetes [71]. Therefore, the pathological condition of DR might affect the alterations in somatic and dendritic structural elements. The reduced number of branches indicates lowered connectivity, suggesting reduced activity in the neurons.
In the present study, although abnormal aggregation, distorted cell bodies, dendritic branch loss, and SAC death were apparent in DR progression, the mechanism underlying these changes is yet to be determined. Previous studies have shown that amacrine cells die via apoptosis in DR [49]. Cell shrinkage and pyknosis are characteristic features of apoptosis and change of cytoskeleton occurs during apoptosis [72]. Thus, the morphological changes and death of SACs in the present study may be due to apoptosis. There is also a possibility that the morphological changes of SACs are due to losses of structural support of other retinal cell types. In DR, various other amacrine cell types, including dopaminergic and AII amacrine cells, and supporting glial cells were lost [2]. However, more studies are necessary to fully understand the mechanisms underlying changes and death of SACs in DR in the present study.
SACs in the retina secrete both GABA and acetylcholine. In many brain areas of mammalian species, GABAergic interneurons can be classified into subpopulations based on their expression of certain CBPs, such as CB, CR, and PV [73][74][75]. In the present study, all ChAT-IR neurons were expressed only in CR. Recently, transcriptomic analysis using high-throughput single-cell RNA sequencing has profiled 63 types of amacrine cells in mouse retina including SAC. They assessed the expression of ChAT including some other molecular markers in SAC but have not assessed the expression of CR to characterize the SAC [32]. It will be necessary to observe whether the CR expression in SAC agrees with single-cell genomics transcription analysis in the future. CBP expression in ChAT-IR neurons greatly differed among species [35,[44][45][46][47][48]76,77]. In some animals, ChAT-IR neurons expressed only one of the CB [35], CR [47,77], or PV [48], whereas ChAT-IR neurons expressed two or three of CB, CR, or PV in the other animals [45,76]. The reason for various expression profiles among animals is unclear. Members of EF-hand family has different structure within their EF-hand motifs. And this structural variety provides different Ca 2+ affinities and kinetics that affect spatiotemporal aspects of Ca 2+ signals. Considering diverse calcium affinities, binding kinetics, and CBP buffering capacities, different physiological environments may require SACs to have different CBPs for calcium signaling pathways [42,[78][79][80]. In addition, differential protein segregation may reflect subtle functional protein segregation in retinal visual processing and species differences [41][42][43]81].
Although its function is unknown, CR is a major CBP in the central visual system and has been used to label discrete neuron populations with distinctive morphology and electrophysiology in mice [52,82,83]. It has been suggested that CR is important in calcium buffering and transport similar to other CBPs, e.g., CB and PV [40,41,84,85]. CR is involved in sharpening the timing of action potentials and is associated with many biological processes, such as cell proliferation, differentiation, and cell death [85,86]. A recent study reported that CR may play a significant role in promoting synaptic efficacy during high-rate activity [87]. As SACs are important bridging cells along with directionselective RGCs [88,89], it will be interesting to examine if CR is related to sharpening action potential timings and promoting synaptic efficacy in motion pathways.
The functional aspects of SACs in the DR have not been thoroughly investigated. Previous studies have observed that ChAT-IR neurons constituted approximately 3% of all amacrine cells in the INL and approximately 20% of all amacrine cells in the GCL, comprising one of the largest populations of amacrine cells in the retina [36,37,59,90]. This finding implies that SACs have key roles in retinal function. In DR, the loss of SACs may cause the decrease in the optokinetic response [91]. In particular, SACs have been extensively studied as the key elements of the mechanism underlying direction selectivity, an essential neural computation that occurs in the retina for detecting a moving object [34,92,93]. Direction selectivity involves anatomically symmetric cholinergic and asymmetric GABAergic synaptic connectivity from SACs [34,87]. Thus, the loss of SACs and deprivation of dendritic branches caused by DR can interfere with direction-selective circuits in the retina. In the central visual system, diabetes disrupts functional connectivity between the primary visual cortex and higher visual regions [94]. Furthermore, diabetes causes brain atrophy, which results in a lower gray matter volume in the occipital lobe [95][96][97]. Therefore, these data indicate that diabetes disrupts the accuracy of visual function. However, more detailed studies are needed for understanding the impact of diabetes on vision in the retina and other visual areas.

Conclusions
Retinal cell death seems to be an evident clue in the progression of DR. Our results showed that the number of SACs, both in the GCL and INL, in DR mouse decreased in a time-dependent manner. Abnormal aggregation and distorted cell bodies were also found in SACs in the present study. Moreover, some of the dendritic arbors varied considerably revealing reduced and entangled dendritic branches. SACs in mouse retina expressed CR but not CB and PV. The findings of the present study are expected to contribute to a better understanding of changes of SACs in DR and may guide the development of diagnostic and therapeutic strategies for DR in the future.

Availability of Data and Materials
All data generated or analyzed during this study are included in this published article.

Author Contributions
JRS and CJJ designed the research study. JRS and MJL performed the research. JRS analyzed the data. JRS, MJL, and CJJ wrote the manuscript. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript.

Ethics Approval and Consent to Participate
All experiments involving animals were approved by the Animal Care and Use Committee of Kyungpook National University (permission no. 2020-0158).