The rod synapse in aging wildtype and Dscaml1 mutant mice

The retina is an intricately organized neural tissue built on cone and rod pathways for color and night vision. Genetic mutations that disrupt the proper function of the rod circuit contribute to blinding diseases including retinitis pigmentosa and congenital stationary night blindness (CSNB). Down Syndrome cell adhesion molecule like 1 (Dscaml1) is expressed by rods, rod bipolar cells (RBCs), and sub-populations of amacrine cells, and has been linked to a middle age onset of CSNB in humans. However, how Dscaml1 contributes to this visual deficit remains unexplored. Here, we probed Dscaml1’s role in the maintenance of the rod-to-RBC synapse using a loss of function mouse model. We used immunohistochemistry to investigate the anatomical formation and maintenance of the rod-to-RBC synapse in the young, adult, and aging retina. We generated 3D reconstructions, using serial electron micrographs, of rod spherules and RBCs to measure the number of invaginating neurites, RBC dendritic tip number, and RBC mitochondrial morphology. We find that while rod-to-RBC synapses form and are maintained, similar to wildtype, that there is an increase in the number of invaginating neurites in rod spherules, a reduction in RBC dendritic tips, and reduced mitochondrial volume and complexity in the Dscaml1 mutant retina compared to controls. We also observed precocious sprouting of RBC dendrites into the outer nuclear layer (ONL) of the Dscaml1 mutant retina compared to controls. These results contribute to our knowledge of Dscaml1’s role in rod circuit development and maintenance and give additional insight into possible genetic therapy targets for blinding diseases and disorders like CSNB.


Introduction
The neural retina is composed of two connected circuits: one for color vision, and the other for vision in dim light, or night vision.Some of the common causes of night vision loss (nyctalopia), including vitamin A deficiencies and cataracts, are readily treatable.Nyctalopia resulting from genetic mutations, including those that cause retinitis pigmentosa (RP) and congenital stationary night blindness (CSNB), have been historically more difficult to treat but are current targets for successful gene therapy approaches to restore the target gene function and prevent blindness [1][2][3][4][5][6][7][8].Examples include X-linked CSNB and Leber congenital amaurosis (LCA), caused by mutations in NYX and GUCY2D (respectively), where patients suffer a decreased ability to see in dimly lit environments that is usually present at birth or develops during adolescence [9][10][11].The success of gene therapy approaches in treating genetic causes of vision loss has highlighted the need to identify and better understand the activity of genes linked to blinding diseases in order to expand the pool of candidates for gene therapy trials.
Dscaml1 has been identified as a candidate gene in humans for a middle age of onset of CSNB [12].Dscaml1 is expressed by and required for normal organization of several cell types in the rod pathway including rod bipolar cells (RBCs) and All amacrine cells.However, ERG studies, gross morphology and synaptic markers suggest that the rod circuit remains functional in a Dscaml1 loss of function mouse model [13].Studies in mice have shown that wildtype cells and synapses of the rod circuit undergo significant age-related changes including denervation at the rod synapse and ectopic sprouting of horizontal and bipolar cell dendrites [14][15][16].We hypothesize that the middle age of onset of CSNB associated with a mutation of Dscaml1 could reflect increased sensitivity to normal aging of the rod circuit, and here test if signs of premature or exaggerated aging are observed in Dscaml1 mutant mice.We therefore used a Dscaml1 loss of function mouse model to test the role of Dscaml1 in establishing precise circuitry of the rod pathway and in maintaining the rod circuit in the aging mouse.
Dscam and Dscaml1 are neural cell adhesion molecules in the immunoglobulin superfamily [17].Dscams work alongside other cell adhesion molecules to pattern the retina, which is largely organized by differential expression of transcription factors and the adhesion molecules these regulate, as opposed to the neural activity that plays a prominent role in organizing neurons in other regions of the brain [18][19][20][21].Mammalian Dscams play important roles in the spatial arrangement of neural soma and dendrites in bipolar, amacrine and retinal ganglion cells, and also function as a pro-growth factor in the axon pathfinding of retinal ganglion cells into the brain [22][23][24][25].Dscaml1 is expressed by rods, RBCs and sub-populations of amacrine cells, including AII amacrine cells, which convey visual information from RBCs to retinal ganglion cells.These cells make up the core of the rod pathway, which initiates when rods detect photons and send visual information to RBCs at the rod-to-RBC synapse.While Dscaml1 regulates dendritic arborization and soma spacing of RBCs and AII amacrine cells, the rod-to-RBC synapse itself is histologically and functionally intact in the Dscaml1 loss of function mouse model, as measured by immunohistochemistry, electron microscopy and ERG [13].
Here we build on previous work by utilizing a loss of function allele of Dscaml1 to measure the requirement of Dscaml1 for maintenance of the rod-to-rod bipolar cell synapse.We used immunohistochemistry to investigate the development and maintenance of this synapse.We observed similar patterns of retinal maturation in the wildtype and Dscaml1 mutant retina.RBC dendrite sprouting into the outer nuclear layer (ONL), a correlation of aging and synaptic dysfunction, was increased at earlier ages in the Dscaml1 mutant retina compared to wildtype, but similar at advanced ages.Electron micrograph studies revealed an increase in the number of neurite tips invaginating into Dscaml1 mutant rods, a decrease in the number of Dscaml1 RBC dendritic tips and an increase in the number of RBC dendrites that did not contact rod spherules.In summary, we report that anatomical changes were observed comparing organization of RBC dendrites of the wildtype and Dscaml1 mutant retina, alongside subtle signs of precocious aging.

Animals/Housing
Mice were housed in the University of Idaho Laboratory Animal and Research Facility (LARF).Mice were fed ad libitum and on a 12-hour light:dark cycle.Dscaml1 mutant mice were derived from ES cell line CC0772 [13].Dscaml1 mutant mice carry a LacZ gene trap in the third exon of Dscaml1 that intercepts and truncates the Dscaml1 transcript early in the Nterminus of the protein [13].Bax null mice were obtained from the Jackson Laboratory [26].Mice were maintained on a mixed genetic background including C57BL/6J and 129/SvJ.Wild type litter mate controls were used for EM studies and in Fig 1 .In Fig 2 we utilized both Dscaml1 mutant heterozygous litter mate controls and inbred C57BL/6J mice wild type controls, the latter because our aging wild type mice succumbed before the time of study.The mice are broken out into separate pools and no difference was detected comparing litter mate heterozygous controls and C57BL/6J controls.Wild type C57BL/6 mice at 12 and 18-month of age were obtained from the NIH aged rodent center and were from a C57BL/6JN genetic background, which are C57BL/6J mice acquired, bred and periodically refreshed by the NIH from stock acquired from The Jackson Laboratory.These mice differ from the NIH strain C57BL/ 6N, which carries the rd8 retinal degeneration mutation.Younger wild type mice were bred in house derived from C57BL/6J mice acquired from the Jackson Laboratory.All protocols were approved by the University of Idaho Institutional Care and Use Committee.

Tissue preparation
Mice taken for immunohistochemistry studies were anesthetized with tri-bromoethanol, and transcardially perfused with 1X phosphate buffered saline (PBS) (10X PBS: distilled H 2 O, 80g NaCl, 2.0g KCL, 26.8g Na 2 HPO 4 7H 2 O, 2.4g KH 2 PO 4 , pH 7.4).Eyes were removed and hemisected in 1XPBS.The posterior eye cup was fixed in 4% paraformaldehyde (PFA) for thirty minutes at room temperature followed by three 10-minute washes in 1XPBS.Fixed tissue was sunk in 30% sucrose overnight followed by a 50/50 mixture of 30% sucrose and optimal cutting temperature (OCT; Sakura1 Finetek Inc.) and frozen in liquid nitrogen.Eye cups were sectioned on a Leica GM 1510 S cryostat in 10 μm sections and stored in at -20˚C for immunohistochemistry.All procedures were approved by the University of Idaho Animal Care and Use Committee.Mice were taken for study at post-natal days (P)14, P21, P35, 3 months, 8 months, 12 months, and 18-months.

Serial election micrographs
Serial block-face electron micrograph (EM) images were generated by Renovo neural EM services (Cleveland, Ohio) using a Zeiss Sigma VP scanning electron microscope.Renovo protocols were followed for tissue processing.Two wildtype, two Bax null, and three Dscaml1 mutant samples from two mice were used to generate electron micrograph volumes.Mice were taken for study at three months of age.Mice were perfused with cacodylate buffer and retinas were fixed for four days in cacodylate buffer before processing.Retina pieces were collected mid-range between central and peripheral retina.Sections were imaged at resolutions between 6.5-7 nm across images with 60-65 nm steps in between slices, depending on the individual data set.

3D electron micrograph reconstruction rendering/Data collection
The National Institutes of Health software, FIJI/ImageJ, was used for developing serial images into 3D stacks of retina tissue, using TrakEM plugin software as previously described [26].Individual rod spherules, RBCs, and mitochondria were traced on individual area lists to enable isolated analysis.Data sets include wildtype, Bax null, and Dscaml1 mutant retinas.For rod spherules, the number of invaginations were counted for each observed invagination of single neurites from interneurons (bipolar and horizontal cells).For RBC dendrite tip data collection, entire RBCs were traced on a single area list and each dendrite tip was then manually followed until it either terminated or contacted/invaginated into a rod spherule.The number of RBC dendrite tips that contacted a rod spherule was divided by the total number of dendrite tips to determine the percentage of tips that contacted rod spherules for comparison across genotypes.Mitochondria data was collected in the same manner as RBC data.Individual whole mitochondria were manually traced on individual area lists within already identified and traced RBCs.If mitochondria were not able to be traced completely due to going outside of the image stack, or were more than fifty percent contained in the axon, they were not included in the data.

Imaging/Analysis
Image z-stacks of retina sections for rod-to-RBC synapse analysis were obtained from a Nikon spinning disc microscope at the University of Idaho Imaging Core.Image adjustments were made uniformly across whole images.For displaced synapse data collection, all synapses colabelled with ribeye and dystroglycan antibodies were counted by hand, in a single in focus zplane, including one image on either side of z-stack to accommodate for tissue not lying flat.Synapses were determined to be displaced if they were inset beyond one rod spherule into the outer nuclear layer.The number of displaced synapses was divided by the number of total synapses to determine the percent of total synapses that are displaced for each image and data presented are the means.

Statistical analysis
Kruskal Wallis and Dunn's post hoc test was used for comparisons of displaced synapses across ages and comparing genotypes within ages.One-way ANOVA and Tukey Kramer post hoc tests were used for comparisons of total dendrite number and mitochondrial volume.Kruskal Wallis and Dunn's post hoc tests were used for comparison of the percent of dendrites that terminate at a spherule, mitochondrial complexity, and for percent of total mitochondria that fall over a certain volume across genotypes.

Preserved integrity of the rod-to-RBC synapse in Dscaml1 mutant retina in the young, adult, and aging retina
We first investigated the integrity of the rod-to-RBC synapse from post-natal development to advanced age of the retina (18-months) in wildtype and Dscaml1 mutant mice.Different species have different age-related sensitivities to aging of the rod circuit.For example, humans exhibit loss of rods and RBCs with aging [27][28][29][30].Denervation of the rod synapses and sprouting of RBC and horizontal cell neurites is observed in both mouse and human [15,16,31], while neither of these changes is observed in marmoset [32].Several factors contributing to rod denervation associated with aging have been identified including increases in AMPK signaling and microglial activation [33,34].We hypothesized that loss of Dscaml1 could increase sensitivity of the rod circuit to aging.We therefore imaged integrity of the rod synapse and associated cells, RBC dendrite sprouting and glial activation in wildtype and dscaml1 mutant mice at time point between P14 and 18-months.
We labeled the rod-to-RBC synapse by staining retina sections with antibodies to ribeye and dystroglycan, markers of the ribbon synapse and perisynaptic space, respectively (Fig 1A) [35,36].We imaged sections in the central retina approximately halfway between the optic nerve and peripheral retina.Synapses formed and were anatomically intact as evidenced by the presence of the ribbon protein ribeye and synaptic cleft protein dystroglycan localized to the dendritic tips of RBCs, identified with PKC staining by postnatal day 14 in both wildtype and Dscaml1 mutant retinas (not shown).Pairing was observed at RBC dendritic tips at all ages measured with no qualitative differences detected comparing wildtype and Dscaml1 mutant retina between postnatal days 14 to 18-months of age (P35, 240, and 365 shown) (Fig 1A -1F).Mu ¨ller glia and microglia play important roles in responding to damage in the retina and their activation and remodeling before during and after degeneration of rods [37][38][39].No differences were observed in Mu ¨ller glia, labeled with glutamine synthase, or microglia, labeled with antibodies to Iba1 (Fig 1G -1L), with occasional microglial projections into the outer nuclear layer (ONL) observed in both wildtype and Dscaml1 mutant retina sections (Fig 1G arrowhead).

Precocious increase in displaced rod-to-RBC synapses in the Dscaml1 mutant retina
Previous studies have demonstrated an increase in the number of rod-to-RBC synapses in the outer nuclear layer during the aging of the neural retina [14][15][16].The increase in the number of displaced synapses has also been observed in models of synaptic dysfunction and involves sprouting of RBC and horizontal cell neurites into the ONL to form synapses with denervated rod cells [16,[40][41][42][43].We measured the number of displaced synapses in Dscaml1 homozygous mutant, Dscaml1 heterozygous mutant and wildtype retinas at P14, P35, 3 months, and 18-months of age compared this to the number of synapses in the outer plexiform layer (Fig 2A -2F).We observed an increase in the number of displaced synapses at 18-months of age in all genotypes compared to P14, P35, and 3 months.The percent of displaced synapses increases from 0.167 percent of all synapses (P14) to 2.67 percent (18-months) in wildtype, from 0.23 percent (P14) to 3.75 percent (18-months) in heterozygous mutant mice, and from 0.40 percent (P14) to 2.21 percent (18-months) in the Dscaml1 mutant retina (p < 0.001 for all) (Fig 2G).We did not observe a significant difference in the number of displaced synapses comparing wildtype, heterozygous mutant, and Dscaml1 mutant retinas at early and late stages (P14: p = 0.22, P35: p = 0.39, 18-months: p = 0.35).We did observe an increase in the percent of displaced synapses in the Dscaml1 mutant retina compared to wildtype and heterozygous mutant controls at 3 months of age (3 months: p = 0.002 and p = 0.001, respectively) (Fig 2G).

Dscaml1 mutant rod spherules contain an increased number of invaginating neurites
The rod spherules, or axon terminals of rod photoreceptors, are the synaptic junction for rods, horizontal cell axon tips, and RBC and OFF bipolar cell dendritic tips [44][45][46][47][48][49][50].We generated 3-D reconstructions of rod spherule electron micrographs to study the ultrastructure of Dscaml1 mutant and control synapses.We included Bax null retinas as a control for RBC cell density.Bax null mice have an increase in the number of RBCs, but not rods, like that reported in the Dscaml1 mutant retina [13,51,52].After reconstructing the rod spherules, we compared the number of invaginations into rod spherules in wildtype, Bax null, and Dscaml1 mutant retinas (Fig 3A-3C).Wildtype and Bax null spherules had no difference in the number of invaginations with an average of 4.0 and 4.1 invaginations per spherule, respectfully.Dscaml1 spherules had an average of 5.29 invaginations per spherule compared to wildtype and Bax null retinas (p < 0.001 for both) (Fig 3D).

Intermingling of RBC dendritic arbors in the wildtype and Dscaml1 mutant retina
RBC dendrites do not tile, and the dendritic arbors of adjacent cells normally overlap with each other in the mouse retina [53,54].We reconstructed multiple RBCs to compare the RBC dendritic arbors in wildtype, Bax null and Dscaml1 mutant retina (Fig 4A -4C).We utilized reconstructed RBCs with adjacent cell soma to compare overlap of dendritic arbors in the wildtype, Bax null, and Dscaml1 mutant retina.RBC dendritic arbors in the wildtype, Bax null, and Dscaml1 mutant RBCs interdigitate extensively, with distinct arbors of Dscaml1 mutant RBC dendrites observed, compared to extensive fasciculation observed in other Dscam or Dscaml1 deficient retinal cell types (Fig 4D -4E).

Dscaml1 mutant RBCs have a reduced number of dendritic tips and a reduced percentage of tips that terminate at rod spherules
We next counted the number of dendritic tips of individual RBC dendritic arbors in wildtype, Bax null, and Dscaml1 mutant retinas (Fig 5A -5C).We simultaneously counted the number of dendritic tips that terminated within the rod spherule.We observed several examples of a single rod being innervated by multiple RBC dendrite tips originating from the same cell in the Dscaml1 mutant retina, something we did not observe in other genotypes (Fig 5D).We found a reduction in the number of total RBC dendritic tips in both the Bax null and Dscaml1 mutant retina compared to wildtype, and no difference between the number of dendritic tips formed by Dscaml1 mutant RBCs when compared to the number of Bax null RBC dendritic tips (WT:Bax, p = 0.019; WT:Dscaml1, p < 0.001; Bax:Dscaml1, p = 0.37) (Fig 5E).We divided the number of dendritic tips that invaginated into rod spherules by the total number of dendritic tips to compare the percentage of total dendritic tips that contacted rod spherules across the wildtype, Bax null, and Dscaml1 mutant retina.We traced wildtype and Bax null RBC dendritic tips to the rod spherule 86.9% and 91.0%(respectfully) but were able to trace only 63.5% of Dscaml1 mutant dendritic tips to the rod spherule (WT:Bax, p = 0.350; WT:Dscaml1, p = 0.010; Bax:Dscaml1, p < 0.001) (Fig 5F).

Mitochondrial volume and complexity are reduced in Dscaml1 mutant retina compared to controls
We measured the size, volume, and complexity of mitochondria in RBCs.Dscam genes regulate developmental cell death through Bax-dependent mitochondrial developmental cell death, and we were curious if there were ultrastructural differences at the mitochondria [13,23,55].We compared individual volumes of mitochondria within RBCs of wildtype, Bax null, and Dscaml1 mutant retinas (Fig 6A -6C).We observed a reduced average mitochondrion volume in Dscaml1 mutant RBCs when compared to both wildtype and Bax null RBCs (p = 0.002 and p < 0.001, respectfully) (Fig 6D).We also noticed that the percentage of larger mitochondria was reduced in the Dscaml1 mutant retina compared to controls (Fig 6E).Mitochondria with volumes greater than 1.00 μm 3 contributed to 42.8% and 44.5% of the total mitochondrial volume within wildtype and Bax null RBCs (respectfully), compared to only 19.3% of total mitochondrial volume in Dscaml1 mutant RBCs (WT:Bax, p = 0.910; WT:Dscaml1, p = 0.043; Bax: Dscaml1, p = 0.031) (Fig 6E).Increased mitochondrial function can be correlated with increased complexity and fusion of the mitochondria within a cell [56,57].We therefore also calculated the complexity of each mitochondrion using a surface area to volume ratio that allowed us to determine how far a mitochondrion strayed from a perfect sphere with a ratio of 1:1.We found that mitochondria in both Bax null and Dscaml1 mutant retinas had reduced complexity when compared to wildtype mitochondria (p = 0.006 and p < 0.001) (respectfully), with Dscaml1 mutant mitochondria showing even less complexity than Bax null RBC mitochondria (p < 0.001) (Fig 6F).

Discussion
In this study, we measured anatomical formation and maintenance of the rod synapse in wildtype and Dscaml1 mutant mice.A mutation in Dscaml1 has been linked to middle age onset of congenital stationary night blindness.Dscaml1 is expressed in cells of the rod circuit of the mouse retina, consistent with this finding.We used a loss of function mouse model to measure formation and anatomical connectivity of the synapse during the mouse life span to test if Dscaml1 is required to maintain the rod synapse.We identified changes in connectivity, aging and organization of the synapse.
The rod circuit is susceptible to different degrees of age-related degeneration and changes in different species.Loss of rods, RBCs and sprouting of RBC dendrites associated with aging have been reported in humans, while other animals, such as marmosets do not appear to undergo age related changes to the rod circuit, or even zebrafish, which continuously produce rods during the lifespan [27][28][29][30][31][32]58].Mice represent an intermediate between these examples with evidence of rod denervation and RBC dendrite sprouting but not significant cell loss.We observed a similar gross organization pattern to the wildtype in the aging Dscaml1 mutant mouse retina.RBC and horizontal cell neurite sprouting into the outer nuclear layer is associated with both aging in the mouse retina and synaptic dysfunction [15,16,31].We did not observe excess sprouting and synapse formation during early developmental stages or advanced age (1 year and 18-months) but did observe an increase in sprouting at an intermediate time point in the Dscaml1 mutant retina, suggesting that normal processes observed in the aging retina may be advanced in the absence of Dscaml1.An increase in RBC sprouting has been observed in mutant mouse models lacking ribbon and other synaptic proteins, and while Dscaml1 is not required for the formation of functional synapses [13,14,16,[40][41][42][43], its absence did result in precocious sprouting in this study, suggesting a role at the rod synapse.
Next, we measured the ultrastructure of the rod synapse and bipolar cell dendritic fields using serial electron micrograph reconstructions.We found that the number of invaginations was increased in Dscaml1 mutant retinas compared to controls (Fig 3).We observed overlap of RBC dendritic arbors in all genotypes, consistent with the known lack of RBC dendritic arbor tiling.We did not observe significant clumping or fasciculation of RBC dendrites with each other like has been observed for other cell types, but this observation may have been limited by the number of cells traced per field and the density of RBCs.Cone bipolar cells exhibit a form of plasticity after loss of Dscam that involves sprouting of dendrites [24].We observed a reduced number of dendrite tips in the Dscaml1 mutant retina and an increase in the number of tips that do not make anatomical connections with rod spherules, which would be consistent with dendritic sprouting observed in OFF bipolar cells in the Dscam mutant retina.In the wildtype retina, RBCs occasionally make synaptic connections at cone pedicles.Compared to other studies, we observed a higher percentage of wildtype RBC dendrite tips that did not terminate at the rod spherules and speculate that this may be due to the slice thickness of the serial electron micrograph stacks used in this study compared to other studies in which less tissue was removed between serial image acquisition.We also observed an increase in the number of invaginating neurites in rod spherules, despite a decrease in the number of dendritic tips per RBC, compared to wildtype and Bax null retinas.Contacts between cone bipolar cells and rod spherules have been observed in other mutant genetic backgrounds and may be occurring here also, but thorough tracing of stacks will be required to determine if additional invaginations observed at rod spherules, despite the decrease in RBC tips, can be explained by other cell types contacting RBCs in the Dscaml1 mutant retina [59].Different Dscam1 isoforms are known to interact weakly in Drosophila and related sidekick proteins also interact weakly with each other [25,60,61].Similar weak interactions between Dscam and Dscaml1 could contribute to rod-cone pathway specificity.Additional volume generation and tracing can be performed to test this.We also observed several instances in which RBC dendrite tips branched immediately before both tips invaginated into the same rod spherule in the Dscaml1 mutant retina (Fig 5D ) but not in other genotypes, suggesting an iso-neuronal role for Dscaml1 in RBC, similar to iso-neuronal avoidance mediated by Drosophila [62,63].Other studies have reported a change to the rod ribbon ultrastructure in aging mice [64].EM studies at more advanced ages could help to determine if similar changes occur in Dscaml1 mutant mice.
We also measured mitochondria in wildtype, Bax null, and Dscaml1 mutant mouse RBCs and observed a decrease in both the individual volume and complexity of the mitochondria in Dscaml1 mutant RBCs compared to controls.Mitochondria in neurons are involved in many intracellular processes beyond ATP production and are essential to proper signal transduction through nuclear gene regulation, metabolism, synaptic vesicles, reactive oxygen species production, and ion regulation [65][66][67][68][69]. Mitochondria accomplish this through their ability to undergo morphological changes: fusion and fission [70].Larger mitochondria increase ATP yields and calcium homeostasis, while smaller mitochondria are involved in intracellular transportation, observed at apoptosis initiation, and often are targeted to undergo mitophagy to improve cellular stability [56,57,[71][72][73][74][75].However, fragmentation and decline in mitochondrial quality has also been associated with mitochondrial dysfunction, contributing to the severity of neurological diseases, disorders, and aging [76][77][78][79][80][81].These findings demonstrate that intracellular factors may contribute to individual RBC function that is not detectable by a measure of bulk output like electroretinogram, especially in cases where there is an increase in RBC number that can disguise ERG results.However, it is unclear if changes in the mitochondria are primary or secondary to changes in dendritic organization in RBCs.A further limitation is that while a large number of mitochondria were reconstructed, these were sampled from a small number of RBCs.Additional tracing will establish the degree of variability of mitochondria within RBCs in addition to across genotypes.
In conclusion, we surveyed the rod-to-RBC synapse in Dscaml1 mutant mouse retina from postnatal development through aging and observed that synapses are formed and maintained like what is observed in wildtype mouse retina.We measured the number of neurite invaginations at the rod spherule and found an increase in the number of invaginations in Dscaml1 mutant retina compared to controls.We quantified the number of dendritic tips in wildtype, Bax null, and Dscaml1 mutant retina.Both the total dendritic tip number and percent of dendritic tips that contacted spherules was reduced in Dscaml1 mutant retina compared to controls.Finally, we measured individual mitochondrial volume and complexity in RBCs of wildtype, Bax null, and Dscaml1 mutant retina and report a decrease in both the mitochondrial volume and complexity when compared to controls.The rod circuit differs significantly in mouse and human.Rods represent the vast majority of photoreceptors in mice but have a sparse distribution in humans.It is possible that changes shown here are sufficient to further degrade the minimal human rod pathway.Mutations observed in humans are missense compared to the loss of function mutation used here and this is a known limitation of this study.However, these results contribute knowledge of Dscaml1's role in rod circuit development and maintenance and give insight into possible genetic therapy targets for blinding diseases and disorders like CSNB.

Fig 1 .
Fig 1.The rod-to-RBC synapse across the mouse lifespan.A) Representative image of rod-to-RBC synapses formed in the wildtype retina at P35. B) Rod-to-RBC synapses were maintained in the wildtype retina at 8 months.C) Rod-to-RBC synapses remain intact in the wildtype retina at one year.D) Representative image of synapses between rod and RBCs in the Dscaml1 mutant retina at P35. E) Rod-to-RBC synapses were maintained in the Dscaml1 mutant retina at 8 months.F) Rod-to-RBC synapses remain intact in the Dscaml1 mutant retina at one year.No differences were detected comparing wildtype and Dscaml1 mutant retinas across all the ages examined (P14, P21, P35, 3 months, 8 months, 12 months, 18-months).G) Representative image of Mu ¨ller glia and microglia in the wildtype retina at P35. H) Representative image of Mu ¨ller glia and microglia in the wildtype retina at 8 months.I) Representative image of Mu ¨ller glia and microglia in the wildtype retina at one year demonstrating occasional microglia projecting into the ONL (arrowhead).J) Representative image of Mu ¨ller glia and microglia in the Dscaml1 mutant retina at P35. K) Mu ¨ller glia and microglia in representative image of the Dscaml1 mutant retina at 8 months.L) Representative image of Mu ¨ller glia and microglia in the Dscaml1 mutant retina at one year.No differences were observed comparing wildtype and Dscaml1 mutant retinas.Arrowheads (A-F) point to ribeye horseshoe.Arrows (A-F) point to dystroglycan puncta.PKC = rod bipolar cells, dystroglycan = synaptic cleft, ribeye = presynaptic, GS = Mu ¨ller glia, Iba1 = microglia.Abbreviations: RBC = rod bipolar cell, ONL = outer nuclear layer, OPL = outer plexiform layer, INL = inner nuclear layer.Sample size = 12 to 20 images, across at least 3 mice per genotype at all time points.Scale bar = 20 μm (A-F) and 50 μm (G-L).https://doi.org/10.1371/journal.pone.0290257.g001

Fig 2 .
Fig 2. Displaced rod-to-RBC synapses during development and aging.A) Representative image of rod-to-RBC synapses in the wildtype mouse retina at P14. Synapses are restricted to the OPL.B) Rod-to-RBC synapses in the Dscaml1 mutant retina are also restricted to the OPL at P14. C) Rod-to-RBC synapses in the wildtype mouse retina at 3 months.Synapses are mostly located within the OPL.D) An increase in the number of rod synapses was observed in the ONL of the Dscaml1 mutant retina at 3 months of age.E) An increase in the number of rod synapses was observed in the ONL of the wildtype retina at 18-months.F) An increase in the number of displaced rod synapses was also observed in the Dscaml1 mutant retina at 18-months.G) No significant difference in the number of displaced synapses was detected comparing the wildtype, heterozygous mutants, and Dscaml1 mutant retina at P14, P35, and 18-months (P14, p = 0.22; P35, p = 0.39; 18-months, p = 0.35).An increase in displaced synapses was detected at 3 months of age comparing the homozygous Dscaml1 mutant retina to heterozygous mutant and wild type retina (p = 0.001 and p = 0.002, respectively).PKC = RBCs, dystroglycan = postsynaptic, ribeye = presynaptic.Abbreviations: RBC = rod bipolar cell, PKC = protein kinase C, ONL = outer nuclear layer, OPL = outer plexiform layer, INL = inner nuclear layer, WT = wildtype, DSL = Dscaml1 mutant.Sample sizes = between 10 and 15 images, collected from 3 mice per genotype at all time points, except at 18-month heterozygous mutant, which only had 6 images across 2 mice.Total synapses counted: wildtype = 14,417 synapses, littermate control = 9,907 synapses, Dscaml1 mutant = 10,879 synapses.Error bars = standard deviation.Scale bar = 10 μm.https://doi.org/10.1371/journal.pone.0290257.g002

Fig 4 .
Fig 4. Dendritic overlap of 3D reconstructed rod bipolar cells.A) Representative 3D reconstructions of three neighboring RBCs' spatial arrangement in wildtype retina.B) Representative 3D reconstructions of three neighboring RBCs' spatial arrangement in the Bax null retina.C) Representative 3D reconstructions of three neighboring RBCs' spatial arrangement in the Dscaml1 mutant retina.D) 3D RBC reconstruction comparing enface dendritic area with neighboring RBC in wildtype retina.E) 3D RBC reconstruction comparing enface dendritic area with neighboring RBC in Bax null retina.F) 3D RBC reconstruction comparing enface dendritic area with neighboring RBC in Dscaml1 mutant retina.Scale bars = 5 μm.https://doi.org/10.1371/journal.pone.0290257.g004

Fig 6 .
Fig 6. 3D reduced mitochondria size and complexity in the Dscaml1 mutant retina.A) 3D reconstruction of mitochondria in RBC of the wildtype retina.B) 3D reconstruction of mitochondria in RBC of the Bax null retina.C) 3D reconstruction of mitochondria in RBC of the Dscaml1 mutant retina.D) A decrease in the average mitochondrial volume was observed in the Dscaml1 mutant RBC soma and dendrites when compared to both wildtype and Bax null retinas.(p = 0.002 and p < 0.001, respectfully).E) A decrease in the percent of total mitochondrial volume that mitochondria over 1.00 um 3 make up in Dscaml1 mutant retina was observed compared to wildtype and Bax null retinas (p = 0.043 and p = 0.031, respectfully).F) A decrease in mitochondrial complexity was observed in both the Bax null and Dscaml1 mutant RBC when compared to wildtype retina (p = 0.006 and p < 0.001), with a larger decrease in complexity in Dscaml1 mutant compared to Bax null retina.(p < 0.001).Each color represents individual mitochondria.Abbreviations: WT = wildtype.Asterisks indicate significance.Mitochondria sample sizes (for all data) = WT: 234, across 3 RBC; Bax: 235, across 3 RBC; Dscaml1: 684, across 6 RBC.Error bars = standard error of the mean.Scale bar = 5μm.https://doi.org/10.1371/journal.pone.0290257.g006