The intracellular domain of Sema6A is essential for development of the zebrafish retina Free

During embryonic development, neurons are born and migrate to their proper location using extracellular guidance molecules. Eye development relies heavily on guidance molecules for cells to find their correct location and form synaptic connections to ensure a proper structure–function relationship. Semaphorin (Sema) proteins comprise a six-membered family of guidance molecules that are crucial for many aspects of development, such as migration, adhesion, proliferation, differentiation and transcription (Alto and Terman, 2017). They are also known to play roles in eye, heart, bone and brain development (Giacobini and Prevot, 2013; Jongbloets and Jeroen Pasterkamp, 2014; Kang and Kumanogoh, 2013; Neufeld et al., 2012; Pasterkamp, 2012; Roth et al., 2009).

Semaphorins have traditionally been shown to bind to plexin (Plxn) receptors on adjacent cells (trans interactions) to initiate ‘forward signaling’. This interaction classically regulates integrins and cytoskeletal dynamics to stop migration of the Plxn-expressing cell (Barberis et al., 2005; Boettner and Van Aelst, 2009). Additionally, Sema–Plxn binding on the same cell (cis interactions) has variable effects on forward signaling. The cis interaction of Sema6A–PlxnA2 (Renaud et al., 2008), Sema6A–PlxnA4 (Haklai-Topper et al., 2010), and Sema6B–PlxnA2 (Andermatt et al., 2014) are inhibitory, whereas Sema1–Plx1 in C. elegans leads to activation of the plexin (Mizumoto and Shen, 2013). Some transmembrane semaphorins can signal bi-directionally; they not only act as a ligand in forward signaling but also as a receptor in ‘reverse signaling’. Sema1a, has been studied extensively in its role of bi-directional signaling in invertebrates (Battistini and Tamagnone, 2016), as it plays a role in synapse formation (Godenschwege et al., 2002), neuronal axon pathfinding (Cho et al., 2012; Hsieh et al., 2014; Jeong et al., 2012; Yu et al., 1998, 2010) and visual system R-cell development (Cafferty et al., 2006). In mammals, Sema6D reverse signaling, utilizing PlxnA1 as a ligand, results in Abl kinase binding to the intracellular domain which initiates a cascade important for cardiac development (Toyofuku et al., 2004). Sema6A and Sema6B have also been shown to use reverse signaling in vitro (Eckhardt et al., 1997; Perez-Branguli et al., 2016); however, the phenotypic consequences of this signaling have yet to be determined.

Previously, we have shown that Sema6A and PlxnA2 play important roles in eye development. Loss of Sema6A or PlxnA2 in zebrafish leads to a loss of eye vesicle cohesion, decreases in retinal proliferation, defects in retinal lamination and changes in gene transcription (Ebert et al., 2014; Emerson et al., 2017; St. Clair et al., 2018). What remains unclear is which of these phenotypes rely on forward signaling, reverse signaling or a combination of both. In this paper, we investigated the role of the Sema6A intracellular domain in retinal development by utilizing a truncated construct (Sema6A-ΔC) in rescue experiments. In Sema6A morphants, the Sema6A-ΔC construct mostly rescued eye size and proliferation defects; however, it led to a new phenotype of acellular regions, a decrease in the number and end feet strength of Müller glia and an increase in cell death. These data suggest a specific role for the intracellular domain of Sema6A in zebrafish retinal development.

To determine whether the intracellular domain of Sema6A is necessary for zebrafish retinal development, we utilized DNA constructs encoding either full-length Sema6A (Sema6A-FL) or a version lacking the intracellular domain (Sema6A-ΔC) (Fig. 1) (Perez-Branguli et al., 2016). We verified cell surface expression of both constructs by performing immunocytochemistry on transfected mouse neuroblastoma NIE-115 cells. Both the Sema6A-FL and the Sema6A-ΔC constructs led to cell surface expression, and hence lack of the intracellular domain does not significantly alter Sema6A trafficking to the cell membrane (Fig. S1). We utilized mRNA synthesized from these DNA constructs in the following zebrafish rescue experiments to investigate the role of the Sema6A intracellular domain in vertebrate eye development.

We previously published a role for Sema6A in early eye development (Ebert et al., 2014). Loss of Sema6A led to smaller eye fields due to a decrease in proliferation and a loss of optic vesicle integrity. Importantly, these phenotypes could be rescued by injection of mRNA encoding human full-length Sema6A. The Sema6A morpholino used has been previously validated by a second non-overlapping morpholino, co-injecting with a p53 morpholino, rescue with human Sema6A-FL mRNA and quantification of knockdown efficiency using RT-PCR (Ebert et al., 2014; Emerson et al., 2017). To determine whether these phenotypes are due to forward and/or reverse signaling, we rescued Sema6A morphants with either Sema6A-FL or Sema6A-ΔC. Importantly, overexpression of Sema6A-FL or Sema6A-ΔC mRNA alone does not lead to any observable phenotypes that would be caused by mRNA toxicity or excess Sema6A in development (Fig. S2). Interestingly, we observed rescue of eye size and eye field proliferation using phospho-histone H3 immunohistochemistry (pHH3) at 18 h post fertilization (hpf) with both constructs (Fig. 2A–D,M,N). Importantly, controls without the primary pHH3 antibody showed no background staining (Fig. S3). Normalizing pHH3+ cell count to eye area showed the same pattern of less proliferation in Sema6A morphants and this was rescued by both Sema6A-FL and Sema6A-ΔC constructs (morphants P=0.004, FL rescue P=0.014, ΔC rescue P=0.013). Owing to known Sema6A expression in the brain, we observed a decrease in brain proliferation and size; however, we focused our study on eye development. We next tested whether the intracellular domain of Sema6A was important for eye size later in development. At 48 and 72 hpf, Sema6A-FL and Sema6A-ΔC rescue eye sizes were not significantly different from each other and showed significant rescue when compared to uninjected controls (Fig. 2E–L,O,P). These data suggest the early eye phenotype observed with loss of Sema6A is caused by forward signaling through PlxnA2 and does not require the intracellular domain of Sema6A.

We previously observed retinal lamination defects with loss of Sema6A or PlxnA2 (Ebert et al., 2014). To determine if the intracellular domain of Sema6A is required for proper lamination, we examined transverse retinal sections stained with hematoxylin and eosin (H&E) at 48 and 72 hpf. Surprisingly, although the Sema6A morphant showed loss of layering as previously observed (Fig. 3B,B′,F,F′), we did not observe layering defects in either Sema6A rescue conditions (Fig. 3C–D′,G–H′) when compared to control retinas (Fig. 3A,A′,E,E′). However, at 48 hpf, when compared to control and Sema6A-FL rescue retinas (Fig. 3A,C), Sema6A morphant and Sema6A-ΔC rescue retinas appear to be detached from the basal surface adjacent to the lens (Fig. 3B,D). We observed a similar phenotype with loss of PlxnA2 (Fig. S4). Additionally, we observed significantly more acellular regions in the Sema6A morphant and Sema6A-ΔC rescue retinas compared to non-injected control and Sema6A-FL rescue retinas (Fig. 3I). The Sema6A-ΔC rescue retinas had the largest acellular regions of all treatments (Fig. 3J). Together, these data suggest that the Sema6A intracellular domain is not required for proper layering but is required for retinal structural development and/or maintenance.

Retinal neurons are born at the apical surface of the retina starting at 28 hpf and must migrate basally to their final location for retinal circuitry to be properly established by 72 hpf (Amini et al., 2018). Owing to the lamination defects observed with loss of Sema6A, it is possible that neurons do not end up in their proper location by the end of this window of development. To address this hypothesis, we utilized the Spectrum of Fates (SoFa1) transgenic line, which labels retinal neurons with three fluorescent markers (CFP to label photoreceptor cells and bipolar cells, GFP to label retinal ganglion cells, and RFP to label amacrine and horizontal cells). We assessed gross neuronal positioning at 48 hpf and 72 hpf. Transverse retinal sections identified no aberrant differentiation or neuronal positioning in the Sema6A-FL rescue (Fig. 4C,G) compared to non-injected controls (Fig. 4A,E). In areas outside the acellular regions, the Sema6A-ΔC rescue retina also showed normal differentiation and neuronal positioning (Fig. 4D,H). This contrasted with what was seen in Sema6A morphant retinas, which had more undifferentiated cells and aberrant neuronal positioning (Fig. 4B,F). We next investigated whether there were retinal cell layer-specific defects by performing cell counts for the three main cellular layers. We first counted the number of cells in the entire retina and found no difference between the Sema6A-FL rescue and the Sema6A-ΔC rescue (Fig. 5A). Quantification of the retinal ganglion cell (RGC), inner nuclear (INL) and outer nuclear (ONL) cell layers also showed no difference between the two rescue conditions (Fig. 5B–D). These experiments suggest that the intracellular domain of Sema6A is not required for retinal neuronal differentiation and final positioning within the retina.

It has been previously reported that Müller glia are required to maintain retinal structural integrity (Benfey et al., 2022; Tworig and Feller, 2022), and when their differentiation is pharmacologically inhibited by inhibiting Notch signaling, the retina loses integrity near the basal surface (MacDonald et al., 2015). This phenotype is strikingly similar to what was observed with a loss of Sema6A (Fig. 3). To address the hypothesis that the intracellular domain of Sema6A regulates Müller glia development, we utilized the gfap:GFP transgenic line (Fig. 6A). Transverse retinal sections at 72 hpf demonstrated a significant decrease in the number of Müller glia in the Sema6A morphant retinas (Fig. 6B,E) compared to non-injected controls (Fig. 6A,E). This decrease was fully rescued in the Sema6A-FL retinas (Fig. 6C,E) but only partially rescued in the Sema6A-ΔC retinas (Fig. 6D,E). Additionally, we observed a significant thinning of the Müller glia end feet along the basal lamina in the Sema6A morphant and the Sema6A-ΔC rescue retinas (Fig. 6A′–D′,F). Interestingly, this phenotype is also observed with the loss of PlxnA2 (Fig. S4). These data suggest that the Sema6A intracellular domain, via its interaction with PlxnA2, contributes to proper Müller glia number in the retina and to proper Müller glia end feet in the basal lamina.

A crucial function of Müller glia is homeostasis of the retina and metabolic support of retinal neurons (Reichenbach and Bringmann, 2013). A decrease in the strength of glial connections could lead to glial and neuronal cell death. We investigated this hypothesis by immunolabeling for cleaved caspase3 (cCasp3), a marker for apoptosis, at 72 hpf. The non-injected control and Sema6A-FL rescue retinas had similarly low levels of cCasp3 staining as expected (Fig. 7A,C,E). We observed an increase in cCasp3 labeling in the Sema6A morphant and Sema6A-ΔC rescue retinas (Fig. 7B,D,E). This increase in cell death could be a combination of glial and neuronal cell death. Importantly, no primary antibody negative controls for cCasp3 showed no background staining (Fig. S2). This suggests that the Sema6A intracellular domain may play a protective role against cell death in the retina.

We previously reported that a loss of Sema6A leads to smaller eyes due to lack of eye vesicle integrity, decreased proliferation and increased cell death (Ebert et al., 2014). Additionally, we show here, the overall morphology of the retina is disorganized and shows impaired lamination. The goal of this study was to investigate a potential role for the intracellular domain of Sema6A in these early eye phenotypes. We approached this question by knocking down endogenous Sema6A and rescuing with either the full-length Sema6A or Sema6A lacking the intracellular domain. We report here that the intracellular domain of Sema6A is not necessary for early eye development or retinal neuronal positioning, as Sema6A-ΔC rescues eye size and proliferation defects from 18–72 hpf and retinal lamination at 72 hpf. This suggests that forward signaling through PlxnA2 is the predominant mechanism regulating these processes. However, retinal sections of Sema6A morphants and Sema6A-ΔC rescues revealed that there were acellular regions between 48 hpf and 72 hpf suggesting a lack of retinal integrity. Interestingly, we also observed a defect in Müller glia cell numbers and an increase in retinal cell death (Fig. 8). These data suggest a role for reverse signaling in development to prevent these phenotypes. Although semaphorin reverse signaling has been shown in other systems, our findings are the first to show a role for Sema6A reverse signaling in an in vivo vertebrate system.

During development, Müller glia are essential for retinal structure and health. Pharmacological inhibition of Notch signaling leads to depletion of Müller glia leading to defects in retinal integrity (MacDonald et al., 2015). These phenotypes have a striking resemblance to the phenotypes observed with the loss of Sema6A or just the intracellular domain. Owing to the decrease in Müller glia number and the thinning of the basal Müller glia end feet in the Sema6A-ΔC rescue, we hypothesize that Sema6A reverse signaling contributes to proper Müller glia development. This could be through differentiation or glia process migration and/or adhesion. However, Sema6A forward signaling is still important in this system as Sema6A-FL partially rescued the number of Müller glia cells. Future work investigating the signaling mechanism of Sema6A reverse signaling will be needed to fully delineate its role in Müller glia development.

This brings into question the potential Sema6A reverse signaling mechanisms responsible for the observed phenotypes. One possibility is that the intracellular domain is required to initiate an intracellular signaling cascade. It was previously shown that the intracellular domain of Sema1 plays a role in signal transduction by binding to Rho GTPases and that Sema6A binds Abl kinase to regulate cell behavior (Jeong et al., 2012; 2017; Perez-Branguli et al., 2016). One potential role of the intracellular domain could be the transduction of an anti-apoptotic signal, as without this domain, we observed increased retinal cell death. Investigating potential post-translational modifications on the Sema6A intracellular domain and identification of its interacting proteins will be an important first step to begin to elucidate the functional Sema6A reverse signaling cascades. Additionally, we know that cis and trans interactions for semaphorins and plexins play a role in their signal transduction, and the role of the Sema6A intracellular domain in these interactions is currently unknown. It is known that Sema6A reverse signaling can have cell autonomous and non-autonomous functions (Kerjan et al., 2005; Perez-Branguli et al., 2016); however, using a morpholino in this system, we cannot decipher autonomous versus non-autonomous signaling. Further investigation will be needed to determine what Sema6A reverse signaling cascades are initiated and are crucial for eye development.

A second possibility is that the intracellular domain regulates Sema6A internalization or release from the membrane to regulate signal transduction timing. We have previously demonstrated that Sema6A can be released from the membrane naturally and in a PKC-induced manner (St. Clair et al., 2019). Additionally, removing the intracellular domain in vitro leads to more release of the Sema6A ectodomain. This has implications of reducing reverse signaling while driving forward signaling. Again, further work will lend additional insight into the implications of this process during eye development.

Finally, intracellular domain signaling could be affecting gene transcription within the cell. We previously identified gene transcriptional changes that occur when Sema6A expression is knocked down (Emerson et al., 2017). There is a significant upregulation of genes involved in migration and proliferation, meaning Sema6A is likely inhibiting these genes during normal development. The transcriptional changes we observe with loss of Sema6A can be compared to the changes observed with loss of PlxnA2 to identify reverse signaling-specific genes. Identification of what genes are regulated specifically following Sema6A activation or inhibition would bring interesting insight into this process.

To conclude, we provide evidence to support a role for the intracellular domain of Sema6A in zebrafish eye development, given that lack of the intracellular domain leads to defects in retinal integrity, Müller glia development and an increase in cell death. Future work will determine the signaling mechanisms driven by the intracellular domain of Sema6A in eye development. This work will lend additional insight into the roles semaphorin signaling plays in many developmental processes and tissues.

Immunocytochemistry was performed on NIE-115 mouse neuroblastoma cells expressing Sema6A-FL and Sema6A-ΔC constructs (Perez-Branguli et al., 2016). Cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal bovine serum (Hyclone, Logan, UT, USA), 5% cosmic calf serum (Hyclone), and 50 U/ml penicillin and 50 µg/ml streptomycin (CellGro Technologies, Lincoln, NE, USA) at 37°C in 5% atmospheric CO₂. Cells were grown to 50–70% of confluence on glass coverslips, washed with PBS and put in Gibco^TM Opti-MEM without serum. Cells were transfected with Lipofectamine 2000 (Thermo Fisher Scientific, Waltham, MA, USA) in a ratio of 1:2 DNA to lipofectamine. At 24 h post transfection, cells were fixed with 3.7% paraformaldehyde (PFA) for 10 min and blocked with 1.5% BSA in PBS with rocking for 1 h at room temperature. Cells were incubated with primary antibody (anti-Myc 9E10 hybridoma conditioned medium, diluted 1:2 in block; ATCC, Manassas, VA, USA) rocking at room temperature for 1 h. Cells were washed four times for 5 min each in PBS plus 0.01% Triton X-100 (PBST) and incubated in secondary antibody (anti-mouse-IgG conjugated Alexa Fluor 555, 1:10,000; Cell Signaling Technology, cat. #4409) at 4°C for 45 min. Cells were washed four times for 5 min each with PBST and mounted on glass slides with Vectashield Hardset with DAPI (Vector Labs, cat. #H-1500, Burlingame, CA, USA). Cells were imaged on a Nikon Eclipse Ti inverted confocal microscope at 40× magnification.

Zebrafish transgenic lines used were: AB, Tg(rx3:GFP), which labels retinal progenitor cells and was used to quantify eye field area (Rembold et al., 2006), Spectrum of Fates (SoFa1), which includes Tg(atoh7:gapRFP), which predominantly labels retinal ganglion cells, Tg(ptf1a:cytGFP), to label amacrine and horizontal cells, Tg(crx:gapCFP), to label photoreceptors and bipolar cells and which was used to assess retinal neuronal positioning (Almeida et al., 2014), and Tg(gfap:GFP), which labels Müller glia and was used to quantify cell number and end feet strength (Bernardos and Raymond, 2006). Embryos were raised under standard conditions at 28.5 or 25°C and staged as previously described (Kimmel et al., 1995). Phenylthiourea was added at 0.003% at 24 hpf to prevent pigment formation. All procedures were approved by the University of Vermont Institutional Animal Care and Use Committee (IACUC PROTO201900024) and the University of Vermont Institutional Biosafety Committee (IBC REG201900054).

Zebrafish embryos were injected at the 1–2-cell stage using an Eppendorf Femtojet 4i microinjector. A splice-blocking sema6A antisense morpholino (5′-TGCTGATATCCTGCACTCACCTCAC-3′) and splice-blocking PlxnA2 morpholino (5′-AAAAGCGATGTCTTTCTCACCTTCC-3′) were injected at a final concentration of 4 ng per embryo. We previously validated these morpholinos by RT-PCR, a second site morpholino, rescue experiments and off-target effects were addressed by co-injecting p53 morpholino (Ebert et al., 2014; Emerson et al., 2017). Capped and tailed mRNA was made from plasmids containing mouse Sema6A-FL and mouse Sema6A-ΔC. Plasmids were linearized using Not1 for Sema6A-FL and Sma1 for Sema6A-ΔC. Capped and tailed mRNA was made using am mMESSAGE mMACHINE kit (Invitrogen, cat. #AM1344, Austin, TX, USA) and poly(A) tailing kit (Invitrogen, cat. #AM1350). mRNA was injected at a final concentration of 600 pg per embryo.

Embryos were raised to the desired stage and fixed in 4% PFA overnight at 4°C. Embryos were then washed in PBS, dehydrated in 100% ethanol for 30 min at room temperature and embedded using the JB-4 Embedding kit as per the manufacturer's protocol (Polysciences, Inc., Warrington, PA, USA). Embryos were sectioned at 7 μm, unless otherwise noted in the figure legend, using a Leica RM2265 microtome. Sections were mounted on glass slides and stained with H&E (Thermo Fisher Scientific, Waltham, MA, USA).

Embryos were raised to the desired stage and fixed in 4% PFA for 2 h at room temperature. Embryos were then washed three times for 15 min each in PBST. Embryos at 24 hpf and older were permeabilized with ice-cold acetone for 7 min. Embryos were rocked overnight at 4°C in primary antibody in 10% goat serum in PBST [anti-phospho-histone H3 (pHH3, 1:1000; Cell Signaling Technology, cat. #3377S) or anti-cleaved-caspase 3 (cCasp3, 1:500; Cell Signaling Technology, cat. #9661S)]. Embryos were washed 6 times for 15 min with PBST and incubated for 2 h at room temperature on secondary antibody diluted1:1000 in 10% goat serum in PBST, (anti-rabbit Alexa Fluor 555 or 488; Cell Signaling Technology, cat. #4413S and #4412S). Embryos were then washed six times for 15 min each time in PBST before imaging or further processing.

Images were taken on an Olympus IX71 inverted microscope using a SPOT insight camera. Eye area, cell counts and cell location were assessed in FIJI software (Schindelin et al., 2012). H&E-stained sections were imaged at 20× magnification. 18 hpf Tg:rx3:GFP embryos labeled with anti-pHH3 to assess cell proliferation were mounted in 4% methyl cellulose and imaged at 10× magnification. Eye area was measured using the ellipse tool in FIJI and the number of pHH3-positive cells in one eye field were counted using the cell count tool. Embryos labeled with cCasp3 to assess cell death were embedded, sectioned at 20 μm and imaged using a 20× objective. The number of cCasp3 puncta were counted in one eye field using the cell count tool in FIJI then normalized to the eye area. SoFa1 and Tg:gfap:GFP embryos were fixed at 48 and 72 hpf, embedded, sectioned at 7 μm and imaged at 20× magnification. The number of Muller glia nuclei were counted using the cell count tool. The end feet strength was measured by taking the average intensity of three end feet locations (top, middle and bottom of lens) and dividing it by the average intensity of three adjacent nuclei per section. Additionally, we quantified the total fluorescence intensity of the end feet area divided by the total florescence intensity of the nuclei area. Both quantification methods yielded the same results. Representative images for figures were taken on a Nikon Eclipse Ti inverted confocal microscope and processed for brightness and contrast in GIMP.

All statistics and graphs were done using R software and packages (https://www.r-project.org/). Sample size was calculated with a power of 0.8, effect size of 0.5 and significance level of 0.5 for 4 groups to yield at least 12 fish per group. Ordinary one-way ANOVA with a Tukey's multiple comparisons test was used to compare between groups. All box and whiskers plots show the first and third quartiles represented by the box, with the median being the middle line. The whiskers extend from the quartile ranges to 1.5 times the interquartile range from the Q1 and Q3 boundaries. The box and whisker plots are overlaid by individual points representing one fish per point. P-values: ns>0.05, *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001.

We would like to acknowledge Paula Deming for her guidance and use of antibodies during preliminary studies. The Vermont Integrative Genomics Core at UVM for use of their sequencing facility to verify plasmids. We thank Cody Smith (Department of Biological Sciences, University of Notre Dame) for gifting us the gfap:GFP transgenic zebrafish. Sema6A-FL and Sema6A-ΔC plasmids were kindly gifted from the Mitchell Lab (Smurfit Institute of Genetics and Institute of Neuroscience, Trinity College Dublin).

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.261469.reviewer-comments.pdf