The absence of BBSome function decreases synaptogenesis and causes ectopic synapse formation in the retina

Photoreceptors possess ribbon synapses distinct from the conventional synapses in the brain. Little is known about the function of the BBSome, a complex integral in ciliary and intracellular trafficking, in ribbon synaptic formation. We performed immunohistochemistry using retinas from Bardet-Biedl Syndrome (BBS) mouse models and found that BBS mutant animals have significantly fewer ribbon synapses in the outer plexiform layer and increased ectopic synapses in the outer nuclear layer compared to controls. Many ectopic synapses in BBS mutant retinas are associated with horizontal cell axonal processes that aberrantly intrude into the outer nuclear layer. To determine whether this horizontal cell phenotype is a consequence of retinal degeneration, we examined this phenotype in mice with photoreceptor-specific inactivation of the BBSome induced by Cre recombinase driven by the rhodopsin promoter. At three months of age, despite retinal degeneration, Bbs8floxed/floxed; Rho-Cre+ mice lack the aberrant intrusion of horizontal cell processes. At 6 months, some horizontal cell processes intrude into the outer nuclear layer in Bbs8floxed/floxed; Rho-Cre+ mice, but the phenotype does not recapitulate the phenotypic severity observed in young congenital BBS mutant mice. Therefore, the lack of BBSome function negatively impacts retinal synaptogenesis, and causes horizontal cell defects in a potentially cell-autonomous fashion.

synaptic connections 19 . The role of the BBSome in synaptogenesis and in modulating neuron-neuron contact is a growing area of research that merits further exploration.
Photoreceptors are neurons in the retina that have evolved enlarged ciliary structures, the outer segments, which specialize in light detection by displaying an enormous amount of opsin protein on their neatly stacked membranous discs. The formation of the photoreceptor outer segments requires ciliary genes including those genes that encode BBSome components 20,21 . Loss of BBSome function in Lztfl1 (Bbs17) mutant mice causes outer segment malformation, as well as leads to the mislocalization of more than a hundred proteins into the photoreceptor outer segment 22 . Therefore, the BBSome is not merely required for ciliary trafficking in photoreceptors but is also indispensable for the proper structural formation, or morphogenesis, of the outer segments. In human patients and mouse models with mutated components of the BBSome, loss of BBSome function causes retinal degeneration, leading to blindness 13,20,21,23 .
In addition to having specialized cilia to support phototransduction, photoreceptors form synaptic structures called ribbon synapses, which are distinct from synapses in the brain. Due to their extraordinary requirement for transducing signals with speed and graded precision, ribbon synapses are uniquely designed for a large number of release-ready vesicles to dock at the active zone 24,25 . This efficient signal transduction between photoreceptors, bipolar and horizontal cells facilitates rapid and sensitive perception of visual stimuli. Anchored to the presynaptic membrane by the protein Bassoon (BSN) 26 , ribbon synapses contain ribbons, unique cytomatrix structures that provide substrate for the docking and tethering of a large number of vesicles proximal to the synaptic active zone. One of the major protein constituents of the synaptic ribbon is RIBEYE, a protein that contains an A domain that is specific to ribbon synapses and capable of forming detergent-insoluble aggregates, and a B domain that is identical to the transcriptional repressor CtBP2 24 . RIBEYE and CtBP2 are transcribed from the same gene, with CtBP2 showing wide expression in many tissues and RIBEYE expression restricted to a few tissues. The unique ability of RIBEYE to form ribbons is conferred by its aggregate-forming A domain, an evolutionary innovation in vertebrates 24 . The deletion of RIBEYE abolished all synaptic ribbons in mice and reduces the quantity of docked vesicles 27 . The deletion of BSN, on the other hand, causes free-floating ribbons 26 . It is not clear whether loss of BBSome function impacts the ability of the photoreceptors to elaborate ribbon synapses. In the study reported here, we investigate the formation of ribbon synapses among photoreceptors, bipolar and horizontal cells in Bbs2 −/− , Bbs4 −/− , Bbs7 −/− , Bbs8 −/− and Bbs1 M390R/M390R mice to elucidate the role of the BBSome in synaptogenesis. Our results indicate that absence of BBSome function negatively impacts photoreceptor synaptogenesis and causes aberrant positioning of synaptic contacts between neurons.

Methods
Animal models and ethics statement. This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. All animals were handled according to approved Institutional Animal Care and Use Committee (IACUC) protocol #8072147 of the University of Iowa. Animals were housed according to IACUC recommendations. Both male and female mice were used in this study, and no sex differences were observed with regards to the phenotypes reported in this study. Methods of euthanasia used were carbon dioxide asphyxiation followed by cervical dislocation, or anesthesia induced by ketamine/xylazine followed by transcardial perfusion of 10% formalin. Humane endpoints were strictly observed, and every effort was made to minimize suffering. Bbs2, Bbs4, Bbs7, Bbs8 knockout, and Bbs1 M390R/M390R knockin mouse models used in this study were published previously 13,20,23,28,29 .
Mice carrying the Bbs8 floxed alleles, as well as Cre recombinase driven by the rhodopsin promoter (Rho-Cre) were generated from breeding mice carrying Bbs8 floxed alleles 20 and the Rho-Cre mouse line 30 . The genotyping for the presence or absence of the Rho-Cre was performed using GoTaq (Promega, Madison, WI) following the manufacturer's recommendations. Primers used for genotyping Rho-Cre were described elsewhere 30 and are listed as follows: F-Rho-Cre, TCAGTGCCTGGAGTTGCGCTGTGG; R-Rho-Cre, CTTAAAGGCCAGGGCCTGCTTGGC.
Immunohistochemistry. Mice were anesthetized by intraperitoneal injection of a ketamine and xylazine mixture as previously described and transcardiac perfusion was performed using 10% formalin (approximately 4% formaldehyde) at 1.0 mL/min for a total volume of 1.25 mL/g body weight 20 . Eyes were enucleated, and a small puncture was created through the lens using a 26-G syringe. Eyes were then embedded in Tissue-Tek O.C.T. compound (VWR, Batavia, IL) and frozen in a 2-methybutane bath chilled with liquid nitrogen. Eyes were sectioned using a Cryostat microtome at thickness of 10 microns and stored at −80 °C for further use.
For immunohistochemistry, sections were permeabilized with 0.3% Triton X-100 in PBS for 10 minutes at room temperature, blocked in a blocking buffer containing 5% BSA, 5% Normal Goat Serum, and 0.05% Triton X-100 in PBS for one hour at room temperature, and incubated with primary antibodies at 4 °C overnight. The next day, slides were washed three times with PBS followed by incubation with secondary antibodies or streptavidin-Alexa Fluor 568 conjugate at room temperature for 1 hour. After another round of washing, slides were mounted with VectaShield anti-fade mounting medium with DAPI (Vector Laboratories, Burlingame, CA). Images were taken using a fluorescence microscope. Minimal image processing including contrast enhancement was performed using ImageJ.
with fixative as described above were processed for H and E histology and visualized under a light microscope. Images were acquired at 40X magnification. Thicknesses of retinal layers were quantified by an observer masked to genotypes using ImageJ. For each animal, two different images were analyzed, and the values were averaged for that animal. Thicknesses of retinal layers of wild type or heterozygous control were compared to those of knockout littermates using two-tailed t test with the assumption of equal variance. For performing statistical comparisons between multiple groups, one-way ANOVA with post-hoc Tukey's test was performed. Error bars in graphs as well as range values denoted in descriptions represent the average ± standard error of the mean.
Quantification of ribbon synapses. For the quantification of ribbon synapses in the outer plexiform layer, immunohistochemical experiments were performed in pairs containing one control and its knockout littermate. Images were acquired at 100X plus an additional 1.6X magnification and RIBEYE-positive, horse-shoe shaped synapses in the outer plexiform layer were quantified by an independent observer masked to genotypes. At least 4 different images were analyzed per animal to determine the synaptic density of that animal. Following quantification, the density of ribbon synapses per 100 μm in knockout animals is compared to that of their control littermates using two-tailed t test for paired data analysis in Microsoft Excel. Error bars in graphs as well as range values denoted in text descriptions represent the average value ± standard error of the mean.
For the quantification of ectopic synapses in the outer nuclear layer, the total number of RIBEYE-positive puncta in the outer nuclear layer was counted in three different images acquired at 40X with an additional 1.6X magnification per animal. The number of ectopic synapses per 100 μm width of the retina for that animal was derived. Then, the density of ectopic synapses in control and mutant animals were compared using two-tailed t test for paired data analysis in Microsoft Excel. Error bars in graphs as well as range values denoted in descriptions represent the average value ± standard error of the mean.
Quantification of photoreceptor nuclei. For the quantification of photoreceptor nuclei, DAPI-positive nuclei located in the outer nuclear layer were counted by an observer masked to genotypes using 40X images of the retina for each animal. The number of photoreceptor nuclei in 100 μm width was determined and used for the normalization of ribbon synaptic number.

Results
Bbs2 −/− mice have a reduced density of ribbon synapses in the outer plexiform layer. In order to study ribbon synapses in the outer plexiform layer in BBS retinas, we probed control and Bbs2 −/− retinas with an anti-CtBP2/RIBEYE antibody. This antibody is frequently used and has been validated to visualize ribbon synapses [31][32][33] . To capture the formation of ribbon synapses in the retina before major photoreceptor loss, we performed immunohistochemical staining in 15-16 day old mice (postnatal day 15-16, P15-16). At this age, retinal degeneration in Bbs2 −/− retinas is mild and the thickness of Bbs2 −/− retinas is 89% of that of their wild type and heterozygous control littermates (Control, 68.6 ± 1.9 μm, n = 7; Bbs2 −/− , 61.2 ± 0.6 μm, n = 6; p = 0.006; Fig. 1a,b). We labeled cone outer segments and pedicles with PNA to visualize the position of outer segments and pedicles. At P15-16, wild type or heterozygous control mice have numerous ribbon synapses neatly confined to the outer plexiform layer (Fig. 1c). In Bbs2 −/− retinas, the number of ribbon synapses appear greatly reduced, and some ribbon synapses are observed within the outer nuclear layer normally devoid of ribbon synapses (Fig. 1d). To determine whether there is a reduced synaptic density in Bbs2 −/− retinas, the number of horseshoe shaped ribbon synapses is quantified and compared to their control littermates. Quantification shows that Bbs2 −/− retinas have a significantly reduced number of ribbon synapses compared to their wild type or heterozygous control littermates (Control, 71.1 ± 1.1 synapses, n = 6; Bbs2 −/− , 32.1 ± 2.2 synapses, n = 6; p = 1.84 × 10 −5 ; Fig. 1e). To account for any photoreceptor cell loss at this stage, the number of ribbon synapses per 100 μm was normalized to the number of photoreceptor nuclei in the outer nuclear layer per 100 μm. After normalization, Bbs2 −/− retinas have a reduced number of ribbon synapses compared to their wild type or heterozygous control littermates (Control, 0.13 ± 0.02, n = 6; Bbs2 −/− , 0.08 ± 0.01, n = 6; p = 0.010; Fig. 1f). In addition, Bbs2 −/− mice have an increased number of ectopic ribbon synapses in the outer nuclear layer (Control, 0.3 ± 0.2, n = 5; Bbs2 −/− , 3.2 ± 0.3, n = 5; p = 4.45 × 10 −5 ). These results show that Bbs2 −/− retinas have a reduced ribbon synaptic density that is not accounted for by photoreceptor cell loss. In addition, absence of BBS2 causes some ribbon synapses to ectopically mislocalize to the outer nuclear layer.
In order to discern whether Bbs2 −/− retinas develop comparable synaptic density to their control littermates as the retina approaches maturation around the third postnatal week, we examined retinas from 21-day old (P21) mice. At this age, the retinal degeneration in Bbs2 −/− mice is more advanced compare to that at P15. The thickness of the outer nuclear layers of Bbs2 −/− mice is approximately 80% of their wild type or heterozygous control littermates (Control, 67.5 ± 4.1 μm, n = 3; Bbs2 −/− , 54.1 ± 2.1 μm, n = 3; p = 0.043; Fig. 2a,b). The number of ribbon synapses in the outer plexiform layer of Bbs2 −/− retinas is approximately 72% of that of their control littermates (Control, 67.1 ± 4.4 synapses, n = 5; Bbs2 −/− , 48.2 ± 4.9 synapses per 100 μm, n = 5; p = 6.1 × 10 −5 ; Fig. 2c-e). It is worth noting that the ribbon synaptic count per 100 μm in the outer plexiform layer in control animals is comparable at P15-16 versus at P21, supporting the notion that the development of ribbon synapses is completed by the end of the second postnatal week in mice 34 . After normalizing to photoreceptor nuclei count, we still observe 12% fewer ribbon synapses in Bbs2 −/− retinas compared to their control littermates (Control, 0.116 ± 0.007, n = 5; Bbs2 −/− , 0.102 ± 0.010, n = 5; p = 0.038; Fig. 2f). Therefore, Bbs2 −/− mice have a reduced ribbon synaptic density by the end of retinal synaptogenesis at P15-16, and this defect persists even when the eye reaches maturation at P21.

Absence of BBSome function interferes with normal synaptogenesis.
We also probed the retinas of Bbs4 −/− and Bbs1 M390R/M390R mutant mice at P15-P16 with the anti-CtBP2/RIBEYE antibody and biotinylated-PNA to determine whether these phenotypes are observed in other mouse models with mutated components of the BBSome. Similar to the observation in Bbs2 −/− and Bbs7 −/− retinas, the density of ribbon synapses in the outer plexiform layer in Bbs4 −/− and Bbs1 M390R/M390R retinas appears greatly reduced compared to their control littermates, and some synapses are mislocalized (Fig. 5). Since all of the mutant mouse models of the BBSome we examined share the phenotype of reduced and ectopic ribbon synapse formation, we conclude that lack of BBSome function interferes with normal synaptogenesis, reduces ribbon synaptic density in the outer plexiform layer in the retina, and causes ectopic localization of ribbon synapses. Some ectopic ribbon synapses proximal to the outer plexiform layer are associated with intruding bipolar cell processes. Ectopic ribbon synapses could be caused by aberrant process extension www.nature.com/scientificreports www.nature.com/scientificreports/ of bipolar and/or horizontal cells into the outer nuclear layer. In order to determine whether intruding bipolar cell processes contribute to the ectopic ribbon synapse phenotype, we probed P21 Bbs2 −/− and Bbs7 −/− retinas with an anti-PKCα antibody along with the anti-CtBP2/RIBEYE antibody to visualize the morphology of bipolar cells and ribbon synapses. In both control and knockout retinas, the anti-PKCα antibody recognizes numerous bipolar cell bodies situated in the inner nuclear layer possessing long, vertical processes extending towards the ganglion cells (Fig. 6). In control retinas, bipolar cell dendrites appear as knob-like protrusions enveloped by the photoreceptor ribbon synapses, and these bipolar-photoreceptor contacts are orderly confined to the outer plexiform layer. In Bbs2 −/− and Bbs7 −/− retinas, however, some bipolar cell processes decorated with ribbon synapses intrude into the outer nuclear layer (Fig. 6, arrows). However, none of the aberrant bipolar cell processes intrude www.nature.com/scientificreports www.nature.com/scientificreports/ beyond a third of the thickness of the outer nuclear layer, and there are ectopic ribbon synapses observed beyond this distance within the outer nuclear layer that do not appear to be associated with these bipolar cell processes (Fig. 6, arrowheads). Therefore, aberrant extension of bipolar cell processes may account for some ectopic synapses located more proximal to the outer plexiform layer in BBS mutant retinas. Those ectopic synapses observed distant from the outer plexiform layer are not accounted for by intruding bipolar cell processes. ectopic ribbon synapses in Bbs2 −/− , Bbs7 −/− , and Bbs8 −/− mice are associated with invading axonal processes of horizontal cells. Furthermore, we probed P21 control and Bbs2 −/− retinas with an anti-calbindin antibody together with the anti-CtBP2/RIBEYE antibody in order to visualize the morphology of horizontal cells and ribbon synapses. In wild type or heterozygous control animals, horizontal cell bodies are found in the upper portion of the inner nuclear layer, and the outer nuclear layer is devoid of any intruding horizontal cell processes (Fig. 7a). It is worth noting that some amacrine cells and ganglion cells also stain positive for calbindin as previously observed 36 . In contrast to their littermate controls, numerous calbindin-positive processes are observed within the outer nuclear layer in Bbs2 −/− retinas (Fig. 7b-d). These processes are likely horizontal cell processes, since they are contiguous with horizontal cell bodies. Some of these processes extend across the entire outer nuclear layer, reaching the outer limiting membrane (Fig. 7c,d). Occasionally, it appears that an entire horizontal cell has migrated into the outer nuclear layer, although the great majority of the horizontal cell bodies are still confined to their normal positions in the inner nuclear layer. These aberrant horizontal cell processes are decorated with ribbon synapses in Bbs2 −/− retinas (Fig. 7c,d). Ectopic ribbon synapses observed at greater than one third of the thickness of the outer nuclear layer, beyond the reach of the intruding bipolar cell processes, are associated with the processes of these horizontal cells. Similar observations are made using P21 Bbs7 −/− retinas (Fig. 8) and Bbs8 −/− retinas (Fig. 9), suggesting that this phenomenon is not specific to loss of BBS2, but due to the loss of BBSome function in these mice. Together, these results suggest that horizontal cell processes invade the outer nuclear layer in BBS mutant mice, sometimes spanning the entire thickness of the outer nuclear layer. These intruding processes are frequently associated with ectopic ribbon synapses and may contribute to the formation of the ectopic ribbon synapses in the outer nuclear layer.
In order to determine whether these intruding processes are axonal processes of horizontal cells, we probed the P21 control and Bbs2 −/− retinal sections with an anti-neurofilament antibody targeting the neurofilament medium chain subunit along with the anti-calbindin antibody. Neurofilaments are intermediate filaments that are enriched in axons 37 . Axonal terminals of horizontal cells are neurofilament-positive 38 . In control retinas, the anti-neurofilament antibody recognizes filamentous structures in the outer plexiform layer that are also immunoreactive to the anti-calbindin antibody (Fig. 10, upper panels). These neurofilament-positive processes of horizontal cells are confined to the outer plexiform layer. In Bbs2 −/− retinas, calbindin-positive horizontal cell processes that invade into the outer nuclear layer are also positive for neurofilament (Fig. 10, lower panels). Therefore, in BBS mutant mice, those processes that aberrantly intrude into the outer nuclear layer are axonal processes of horizontal cells. www.nature.com/scientificreports www.nature.com/scientificreports/

Evidence for a cell-autonomous role of the BBSome in horizontal cells.
Since the intrusion of bipolar and/or horizontal cell processes into the outer nuclear layer is also observed in non-BBS mutant mouse models that have retinal degeneration, it is not clear whether it is a mere consequence of photoreceptor degeneration or whether BBS genes play a role in regulating the process extension of inner neurons such as horizontal cells. Therefore, using a mouse line where the expression of CRE recombinase is driven by the rhodopsin promoter 30 , we generated Bbs8 floxed/floxed ; Rho-Cre + mice, in which BBS8 proteins are specifically deleted in photoreceptor cells, to investigate whether retinal degeneration triggered by the loss of BBSome function in photoreceptors alone causes the intrusion of horizontal cell processes into the outer nuclear layer. We have previously shown that the absence of BBS8 in the eye causes the loss of the complete BBSome complex and BBSome function 20 . In Bbs8 floxed/ floxed ; Rho-Cre + mice, we observe retinal degeneration over time. At 6-7 months of age, when 80% of photoreceptors are lost in Bbs8 congenital knockout mice (Bbs8 −/− ), the thickness of the outer nuclear layer in Bbs8 floxed/ floxed ; Rho-Cre + mice is reduced by 55% compared to that in Bbs8 w/floxed ; Rho-Cre − mice (Fig. 11). The thickness of the outer nuclear layer in 6-7-month-old Bbs8 w/w or Bbs8 w/floxed ; Rho-Cre + mice with at least one copy of the wild type Bbs8 gene is not statistically different from that in Bbs8 w/floxed ; Rho-Cre − mice (p = 0.999), indicating that the presence of CRE did not cause adverse effects in photoreceptor cells (Fig. 11).
In 6-month-old Bbs8 floxed/floxed ; Rho-Cre + mice with more than 50% photoreceptor loss, we do observe some intrusion of horizontal cells processes into the outer nuclear layer in Bbs8 floxed/floxed ; Rho-Cre + mice (Fig. 14). However, the horizontal cell phenotype in 6 months old Bbs8 floxed/floxed ; Rho-Cre + mice appears mild even though the state of degeneration in these animals is much more advanced than in congenital BBS mutant mice at P21. Therefore, even though retinal degeneration does trigger intrusion of horizontal cell processes into the outer nuclear layer over time, the severity of the phenotype observed in young congenital BBS mutant mice is not www.nature.com/scientificreports www.nature.com/scientificreports/ recapitulated in Bbs8 floxed/floxed ; Rho-Cre + mice at any age, even when compared to Bbs8 floxed/floxed ; Rho-Cre + mice at a much more advanced stage of retinal degeneration.
The phenotypic severity of the horizontal cell phenotype observed in young congenital BBS mutant mice is not fully recapitulated by the inactivation of BBSome function in photoreceptors alone. Therefore, the horizontal cell phenotype is not solely a consequence of retinal degeneration. Instead, these results indicate that the aberrant intrusion of horizontal cell processes is due to the loss of a novel cell-autonomous role of the BBSome in horizontal cells, in addition to the effects of retinal degeneration.

Discussion
The absence of BBSome function significantly impairs the formation of photoreceptor outer segments during photoreceptor terminal differentiation 20 . In addition, having a dysfunctional BBSome causes proteins normally excluded from the photoreceptor outer segment to inundate the outer segment 22 . Since the photoreceptor outer segment is a modified cilium, it is not surprising that the BBSome is integral for protein trafficking in the outer segment. However, whether the absence of BBSome function affects synaptogenesis was an unexplored question. In this study, we found that the lack of BBSome function both decreases ribbon synaptic density and alters synaptic positioning in the retina, causing ectopically localized ribbon synapses within the outer nuclear layer. We found that both bipolar and horizontal cells aberrantly extend processes into the outer nuclear layer in BBS mutant retinas at very early stages of degeneration. The aberrant intrusion of horizontal cell processes is especially severe, reaching far beyond the aberrant bipolar cell processes. In 21-day old BBS mutant retinas where degeneration is still mild, some horizontal cell processes are observed spanning the entire thickness of the outer www.nature.com/scientificreports www.nature.com/scientificreports/ nuclear layer, and these intruding processes are decorated with ectopic ribbon synapses located deep within the outer nuclear layer.
One explanation for the intrusion of aberrant horizontal cell processes is that it is a consequence of retinal degeneration. Examples of ectopic synapses can be found in mouse models with mutated components of the ribbon synapse. In Nob2 mutant mice in which one of the subunit of the L-type calcium channel is non-functional, ectopic ribbon synapses are observed in the outer nuclear layer and are proposed to result from the retraction of photoreceptor axons 39 . In these mice, retinal bipolar cells and not horizontal cells constitute the main component of these ectopic synaptic contacts, although both cells types extend aberrant processes into the outer nuclear layer. In Bassoon mutant mice, numerous ectopic synapses form within the outer nuclear layer due to the aberrant sprouting of their horizontal and bipolar cells processes, seen as a part of the remodeling process 40 . In this mouse model, the ectopically localized synapses found closer to the outer plexiform layer are accounted for by axon retraction of the photoreceptor cells, but those found distant from the outer plexiform layer are thought to be the consequence of de novo formation of ectopic synapses as many of these distant synapses lack contact with bipolar cell processes 40 . Notably, in this study, aberrant intrusion of horizontal cell processes is not observed in Bbs8 floxed/floxed ; Rho-Cre + mice up to 3 months of age despite retinal degeneration in these mice. Even at up to 6 months, when greater than 50% of photoreceptor cells are lost in Bbs8 floxed/floxed ; Rho-Cre + mice, the intrusion of horizontal cell processes induced by retinal degeneration is relatively mild compared to the www.nature.com/scientificreports www.nature.com/scientificreports/ phenotype observed in young congenital BBS mutant mice. These observations point to the possibility that the BBSome plays a cell-autonomous role in horizontal cells, and that the phenotype is not solely caused by retinal degeneration. In congenital BBS mutant mice, two overlapping phenomena likely contribute to the intrusion of horizontal cell processes into the outer nuclear layer. First, during retinal synaptogenesis, the axonal pathfinding of horizontal cells is negatively impacted due to the lack of BBSome function. Second, photoreceptor loss over time exacerbates the intrusion of aberrant horizontal cell processes as a part of the retinal remodeling process. In retinas with photoreceptor-specific deletion of BBS8 by Rho-Cre, only the second phenomenon occurs, but not the first. Therefore, the phenotype is only observed in Bbs8 floxed/floxed ; Rho-Cre + mice with advanced stages of retinal degeneration.
It is worth noting that retinal degeneration in Bbs8 floxed/floxed ; Rho-Cre + mice is slower than in Bbs8 −/− congenital knockout mice. There are several explanations underlying the different rates of retinal degeneration in congenital Bbs8 −/− mice and Bbs8 floxed/floxed ; Rho-Cre + mice. First, in a reporter line consisting of a β-galactosidase gene controlled by a floxed transcription stop signal, the deletion of the floxed site is not complete until about P18 in this rhodopsin-Cre mouse line 30 , whereas BBS8 is congenitally absent in Bbs8 −/− retinas. Therefore, the www.nature.com/scientificreports www.nature.com/scientificreports/ initial presence of BBS8 in photoreceptors during the second-third postnatal week may allow the development of normal photoreceptor outer segments and delay the onset of retinal degeneration. Second, cones are not affected by genetic recombination induced by rhodopsin promoter driven CRE in this mouse line 30 . Therefore, the delay in cone degeneration in Bbs8 floxed/floxed ; Rho-Cre + mice may have caused their slower overall retinal degeneration. Lastly, the BBSome may be required in cell types other than photoreceptor cells in the retina, and therefore the inactivation of the BBSome in photoreceptor cells alone may not fully recapitulate the severity of retinal degeneration caused by the global loss of BBSome function in congenital knockout mouse lines. Investigating whether the BBSome is required in cell types other than photoreceptors has important implications for the design of gene therapies targeting blindness in BBS. www.nature.com/scientificreports www.nature.com/scientificreports/ Thus far, retinal degeneration in BBS is attributed to the impaired survival of the photoreceptors, whereas the inner retina is considered to be relatively intact even as degeneration progresses. As a result, gene therapy efforts in BBS are concentrated on rescuing photoreceptor cells alone, frequently employing gene vectors containing photoreceptor-specific promoters such as rhodopsin to drive exogenous gene expression. This report offers evidence of potential roles of the BBSome in other cell types in the retina. Whether the lack of BBSome function in non-photoreceptor cells in the retina merely causes a cellular dysfunction without contributing to photoreceptor degeneration or the dysfunction exacerbates photoreceptor degeneration is an important question to the optimal design of gene therapies.
On the other hand, it is unclear how loss of BBSome function impairs ribbon synapse formation in photoreceptor cells. Recent evidence suggests that BBS proteins are present at neuronal synapses in the brain 19 . Loss of BBS4, a member of the BBSome complex, causes reduced dendritic lengths as well as reduced dendritic spine count in neurons 19 . Based on these pieces of evidence, a potential role of the BBSome complex in trafficking receptors in dendritic spines has been proposed. The BBSome is also required for axonal targeting in olfactory www.nature.com/scientificreports www.nature.com/scientificreports/ neurons 18 . Therefore, the BBSome could potentially play a direct role at the photoreceptor synapse. However, the decreased retinal synaptogenesis in BBS mutant mice could also be a consequence of having reduced quantities of proteins important for synaptogenesis due to the massive protein mislocalization to the photoreceptor outer segment that is known to occur 22 . Alternatively, a global alteration in transcription caused by altered ciliary signaling in photoreceptors could also indirectly impact synaptogenesis. Therefore, whether the BBSome is directly involved in functions at the photoreceptor synapse remains unclear and merits further investigation.
Lastly, how the guidance of axonal processes of horizontal cells is affected in the absence of BBSome function is unknown. Beyond ciliary signaling, a growing body of work has demonstrated the role of the BBSome in regulating receptor localization and signaling on the neuronal plasma membrane in a cilia-independent manner 15 . These works contribute to an emerging picture containing both classical ciliary functions and novel, non-ciliary BBSome functions in neurons. Wandering axons have been observed in the olfactory bulb of mice lacking BBS8, a member of the BBSome 18 , as well as in olfactory sensory neurons lacking IFT88, a member of the IFT complex B, which plays a role in anterograde trafficking 41 . In this report, we propose that the intruding, or wandering horizontal cell axons in the outer nuclear layer could be due to cell-autonomous roles of the BBSome in these inner neurons of the retina. The connection between neuronal axons and their intended targets requires a series of steps involving signaling and pathfinding using receptors on the membrane, and many of these receptors are still unknown. It is possible that the BBSome is required for trafficking certain receptors to neuronal axons important for sensory input in pathfinding, and the lack of such receptors may cause their axons to be mistargeted, leading to aberrant synaptic positioning. This work adds to the possibility that the BBSome can be required, in certain cases, for interpreting attractive or inhibitory signals for neurons to establish synaptic contacts with their intended target in order to achieve proper synaptic positioning. This work paves the way for understanding the novel role of the BBSome in synaptogenesis and axonal targeting in neurons.

Data availability
All data generated or analyzed during this study are included in this manuscript.