Decision letter: Strip1 regulates retinal ganglion cell survival by suppressing Jun-mediated apoptosis to promote retinal neural circuit formation

Abstract

Article Figures and data Abstract Editor's evaluation eLife digest Introduction Results Discussion Materials and methods Appendix 1 Data availability References Decision letter Author response Article and author information Metrics Abstract In the vertebrate retina, an interplay between retinal ganglion cells (RGCs), amacrine (AC), and bipolar (BP) cells establishes a synaptic layer called the inner plexiform layer (IPL). This circuit conveys signals from photoreceptors to visual centers in the brain. However, the molecular mechanisms involved in its development remain poorly understood. Striatin-interacting protein 1 (Strip1) is a core component of the striatin-interacting phosphatases and kinases (STRIPAK) complex, and it has shown emerging roles in embryonic morphogenesis. Here, we uncover the importance of Strip1 in inner retina development. Using zebrafish, we show that loss of Strip1 causes defects in IPL formation. In strip1 mutants, RGCs undergo dramatic cell death shortly after birth. AC and BP cells subsequently invade the degenerating RGC layer, leading to a disorganized IPL. Mechanistically, zebrafish Strip1 interacts with its STRIPAK partner, Striatin 3 (Strn3), and both show overlapping functions in RGC survival. Furthermore, loss of Strip1 or Strn3 leads to activation of the proapoptotic marker, Jun, within RGCs, and Jun knockdown rescues RGC survival in strip1 mutants. In addition to its function in RGC maintenance, Strip1 is required for RGC dendritic patterning, which likely contributes to proper IPL formation. Taken together, we propose that a series of Strip1-mediated regulatory events coordinates inner retinal circuit formation by maintaining RGCs during development, which ensures proper positioning and neurite patterning of inner retinal neurons. Editor's evaluation The results provide mechanistic insight into Strip1 and Striatin-interacting phosphatase and kinase (STRIPAK) complex function at the cellular and molecular level in the developing retina. They show that a primary function of Strip1 and the larger STRIPAK complex in retinal ganglion cells is to promote survival by suppressing Jun-mediated apoptosis. Reviewers were most interested to know whether Jun-mediated, pro-apoptotic signaling occurs due to connectivity defects or if it is connectivity-independent, and the authors have recognized the difficulty in addressing this point, and conclude that it is unlikely that failure of connectivity in the inner plexiform layer is the cause of retinal ganglion cell death. https://doi.org/10.7554/eLife.74650.sa0 Decision letter eLife's review process eLife digest The back of the eye is lined with an intricate tissue known as the retina, which consists of carefully stacked neurons connecting to each other in well-defined ‘synaptic’ layers. Near the surface, photoreceptors cells detect changes in light levels, before passing this information through the inner plexiform layer to retinal ganglion cells (or RGCs) below. These neurons will then relay the visual signals to the brain. Despite the importance of this inner retinal circuit, little is known about how it is created as an organism develops. As a response, Ahmed et al. sought to identify which genes are essential to establish the inner retinal circuit, and how their absence affects retinal structure. To do this, they introduced random errors in the genetic code of zebrafish and visualised the resulting retinal circuits in these fast-growing, translucent fish. Initial screening studies found fish with mutations in a gene encoding a protein called Strip1 had irregular layering of the inner retina. Further imaging experiments to pinpoint the individual neurons affected showed that in zebrafish without Strip1, RGCs died in the first few days of development. Consequently, other neurons moved into the RGC layer to replace the lost cells, leading to layering defects. Ahmed et al. concluded that Strip1 promotes RGC survival and thereby coordinates proper positioning of neurons in the inner retina. In summary, these findings help to understand how the inner retina is wired; they could also shed light on the way other layered structures are established in the nervous system. Moreover, this study paves the way for future research investigating Strip1 as a potential therapeutic target to slow down the death of RGCs in conditions such as glaucoma. Introduction The retina is a highly organized neural circuit that comprises six major classes of neurons, assembled into three cellular layers with two synaptic or plexiform layers between them. This beautiful layered architecture is commonly referred to as ‘retinal lamination’ (Avanesov and Malicki, 2010; D’Orazi et al., 2014; Dowling, 1987). Lamination is conserved among vertebrates and is critical for processing visual information (Baden et al., 2020; Nassi and Callaway, 2009). During development, neurogenesis, cell migration, and neurite patterning are spatially and temporally coordinated to form retinal lamination. Any defect in these events can disrupt retinal wiring and consequently compromise visual function (Amini et al., 2017). However, molecular mechanisms that govern retinal neural circuit formation are not fully understood. The retinal neural circuit processes visual signals through two synaptic neuropils (Figure 1A). At the apical side, the outer plexiform layer (OPL) harbors synapses that transmit input from photoreceptors (PRs) in the outer nuclear layer (ONL) to bipolar (BP) and horizontal cells (HCs) in the inner nuclear layer (INL). At the basal side of the retina, the inner plexiform layer (IPL) is densely packed with synaptic connections formed between BPs and amacrine cells (ACs) in the INL, and retinal ganglion cells (RGCs) in the ganglion cell layer (GCL). The retina contains one type of glial cells called Müller glia (MGs), which span the apicobasal axis of the retina (Hoon et al., 2014; Huberman et al., 2010). Figure 1 with 2 supplements see all Download asset Open asset Striatin-interacting protein 1 (Strip1) is essential for inner retinal neural circuit development. (A) Zebrafish retinal neural circuit showing retinal neurons and synaptic layers. GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; RGCs, retinal ganglion cells; ACs, amacrine cells; BPs, bipolar cells; HCs, horizontal cells; PRs, photoreceptors; MGs, Müller glia. (B) Morphology of wild-type and rw147 embryos at 4 dpf. Dotted lines demarcate the eye. An arrowhead indicates abnormal lower jaw. An asterisk indicates heart edema. (C) Wild-type and rw147 mutant retinas at 4 dpf. Red and yellow arrowheads indicate the IPL and OPL, respectively. (D) A missense mutation occurs in strip1 gene of rw147 mutants leading to replacement of Leu195 with arginine. (E) Wild-type retinas labeled with anti-Strip1 antibody (upper panels) and anti-Strip1 plus Strip1-blocking peptide as a negative control (lower panels). Nuclei are stained with Hoechst. Scale bar, 50 μm. Right panels show higher magnification of outlined areas. Scale bar, 10 μm. (F) Whole-mount labeling of 3-dpf wild-type and strip1rw147 mutant retinas with anti-acetylated α-tubulin antibody. Bottom panels show higher magnification of outlined areas. Scale bar, 50 μm. (G) Projection images of single RGCs at 2 dpf expressing ath5:Gal4VP16; UAS:MYFP in wild-type and strip1rw147 mutants. Scale bar, 10 μm. (H) Projection images of single ACs at 3 dpf expressing ptf1a:GFP in wild-type and strip1rw147 mutants. Scale bar, 10 μm. (I) Projection images of single BPs at 3 dpf expressing nyx:Gal4VP16; UAS:MYFP in wild-type and strip1rw147 mutants. Scale bar, 10 μm. RGCs are the first-born retinal neurons, which extend their axons to exit the eye cup and innervate visual centers in the brain (D’Souza and Lang, 2020, Robles et al., 2014). In mouse and zebrafish models, when RGCs are absent or exhibit defects in axon projections, vision is compromised (Kay et al., 2001; Moshiri et al., 2008; Rick et al., 2000). Therefore, RGCs are indispensable for vision. RGC degeneration is often a secondary defect in optic neuropathies and one of the leading causes of blindness worldwide. Thus, tremendous efforts are being dedicated to deciphering signaling pathways involved in RGC death (Almasieh et al., 2012; Maes, 2017; Munemasa and Kitaoka, 2012). Striatin interacting protein 1 (Strip1) is a recently identified protein with emerging functions in neuronal development. It was first described as one of the core components of the striatin-interacting phosphatases and kinases (STRIPAK) complex (Goudreault et al., 2009). The STRIPAK complex is an evolutionarily conserved supramolecular complex with diverse functions in cell proliferation, migration, vesicular transport, cardiac development, and cancer progression (He et al., 2010; Hwang and Pallas, 2014; La marca, 2019; Madsen et al., 2015; Neisch et al., 2017; Shi et al., 2016). In addition, several STRIPAK components participate in dendritic development, axonal transport, and synapse assembly (Chen et al., 2012; Li et al., 2018; Schulte et al., 2010). In Drosophila, Strip (a homolog of mammalian Strip1/2) is essential for axon elongation by regulating early endosome trafficking and microtubule stabilization (Sakuma et al., 2014; Sakuma et al., 2015). In addition, Strip, together with other STRIPAK members, modulates synaptic bouton development and prevents ectopic retina formation (Neal et al., 2020; Sakuma et al., 2016). On the other hand, loss of mouse Strip1 causes early mesoderm migration defects leading to embryonic lethality (Bazzi et al., 2017; Zhang et al., 2021). Thus, the role of Strip1 in the vertebrate nervous system is largely unknown. Here, we report an essential role for Strip1 in neural circuit formation of zebrafish retina. In zebrafish strip1 mutants, retinal lamination, especially IPL formation, is disrupted. Loss of Strip1 causes RGC death shortly after birth. Cells in the INL subsequently infiltrate the degenerating GCL, leading to a disorganized IPL. Strip1 cell autonomously promotes RGC survival; however, it is not required in INL cells for IPL formation. Therefore, Strip1-mediated RGC maintenance is required to establish the IPL. Mechanistically, we identified Striatin 3 (Strn3) as a Strip1-interacting partner. Both Strip1 and Strn3 show overlapping functions in RGC survival through suppression of the Jun-mediated apoptotic pathway. We also found that Strip1 is cell autonomously required for RGC dendritic patterning, which likely promotes interaction between RGCs and ACs for IPL formation. Collectively, we demonstrate that Strip1 is crucial for RGC survival during development and thereby coordinates proper wiring of the inner retina. Results Strip1 is essential for inner retinal neural circuit development To understand mechanisms of retinal neural circuit formation, we screened zebrafish retinal lamination-defective mutants (Masai et al., 2003) and identified the rw147 mutant. At 4 days post-fertilization (dpf), rw147 mutant embryos have small eyes, lower jaw atrophy, and cardiac edema (Figure 1B). rw147 mutants also show defects in retinal lamination, in which retinal layers, especially in the inner retina, fluctuate in a wave-like pattern (Figure 1C). The rw147 mutation is lethal by 6 dpf due to cardiac edema. Mapping of the rw147 mutation revealed a missense mutation in exon 7 of the strip1 gene of the rw147 mutant genome (Figure 1D). Next, we performed CRISPR-Cas9-mediated mutagenesis to generate a 10-base deletion mutant, strip1crisprΔ10 (Figure 1—figure supplement 1A). strip1crisprΔ10 mutants show similar morphology and retinal lamination defects to those of strip1rw147 mutants (Figure 1—figure supplement 1B–D). Likewise, knockdown of Strip1 using translation-blocking morpholinos (MO-strip1) phenocopied strip1rw147 mutants (Figure 1—figure supplement 1E, F). We verified the specificity of MO-strip1 using a custom-made zebrafish Strip1 antibody that fails to detect a 93 kDa protein band corresponding to zebrafish Strip1 in the morphants (Figure 1—figure supplement 1G). Furthermore, we generated transgenic lines that express wild-type and rw147 mutant forms of zebrafish Strip1 protein under the control of the heat shock promotor, Tg[hsp:WT.Strip1-GFP] and Tg[hsp:Mut.Strip1-GFP], respectively. Wild-type Strip1, but not the mutant form, rescued the retinal defects of strip1rw147 (Figure 1—figure supplement 1H, I). Taken together, the strip1 mutation is the cause of retinal lamination defects. Next, we examined Strip1 expression in wild-type retinas by labeling with the zebrafish Strip1 antibody. Strip1 was expressed in RGCs and ACs at 2 dpf (Figure 1E). In situ hybridization shows that strip1 mRNA is maternally and zygotically expressed and by 2 dpf, expression becomes restricted to the eyes, optic tectum, and heart (Figure 1—figure supplement 2A, B). Like Strip1 protein, strip1 mRNA was expressed in RGCs and ACs (Figure 1—figure supplement 2C). To visualize retinal neuropils, we performed whole-mount staining of the retina with anti-acetylated α-tubulin antibody. In wild-type retinas, IPL and OPL are evident at 3 dpf. In contrast, IPL shows abnormal morphology, whereas OPL is relatively normal in strip1rw147 mutants (Figure 1F). We tracked IPL development using Bodipy TR stain. In wild-type retinas, a rudimentary IPL was formed as early as 52 hr post-fertilization (hpf); however, it was less defined in strip1rw147 mutants (Figure 1—figure supplement 2D). At 62 hpf, mutants exhibited a wave-like IPL. This temporal profile coincides with development of RGCs and ACs. Next, we visualized neurite morphology of RGCs, ACs, and BPs by transiently expressing fluorescent proteins under control of ath5 (Masai et al., 2003), ptf1a (Jusuf and Harris, 2009), and nyx promoters (Schroeter et al., 2006), respectively. In wild-type siblings, RGCs and ACs normally extend their dendrites toward the IPL; however, strip1rw147 mutants show randomly directed dendritic patterns of RGCs and ACs (Figure 1G, H). In wild-type siblings, BPs normally extend their axons and dendrites toward IPL and OPL, respectively; however, BPs of strip1rw147 mutants show misrouted axons and abnormal dendritic branching (Figure 1I). Thus, Strip1 is required for IPL formation and correct neurite patterning of RGCs, ACs, and BPs. RGCs are reduced and INL cells infiltrate the GCL in strip1 mutants To examine how the IPL is disrupted in strip1 mutants, we combined strip1rw147 mutants with two transgenic lines, Tg[ath5:GFP; ptf1a:mCherry-CAAX], to visualize RGCs and ACs. In Tg[ath5:GFP], GFP is expressed strongly in RGCs and weakly in ACs and PRs under control of the ath5 enhancer (Masai et al., 2003). In Tg[ptf1a:mCherry-CAAX], membrane-targeted mCherry is expressed in ACs and HCs under control of ptf1a promoter (Jusuf and Harris, 2009). Live imaging of 3-dpf retinas revealed that RGCs are severely reduced in strip1rw147 mutants (Figure 2A). Since we observe a slight reduction in total retinal area of strip1rw147 mutants at 3 dpf (Figure 2—figure supplement 1A, B), we quantified RGC area compared to total retinal area and found that mutant ath5+ RGCs are reduced, reaching only 7.45% ± 2.88% of total retinal area, compared to 24.18% ± 1.48% in wild-type siblings (Figure 2B). On the other hand, there was no significant change in the number of ptf1a+ ACs between strip1rw147 mutants and wild-type siblings (Figure 2A, C). However, ptf1a+ ACs abnormally extended their dendrites to form an irregularly patterned IPL (Figure 2A, asterisks). In wild-type retinas, the majority of ACs reside in the INL, except displaced ACs (Jusuf and Harris, 2009). However, in strip1rw147 mutants, a significant fraction of ptf1a+ ACs were abnormally located in the GCL (Figure 2A, arrowheads in bottom panels, and Figure 2D). Such abnormal positioning of ACs is correlated with the severity of IPL defects (Figure 2—figure supplement 1C). This phenotype is reminiscent of the ath5 mutant, lakritz, in which RGCs fail to undergo neurogenesis, leading to infiltration of ACs into the GCL and transient IPL formation defects (Kay et al., 2001; Kay et al., 2004). We confirmed similar IPL defects in ath5 morphant retinas at 3 dpf (Figure 2—figure supplement 1D), albeit weaker than those of strip1rw147 mutants. Figure 2 with 2 supplements see all Download asset Open asset Retinal ganglion cells (RGCs) are reduced and INL cells infiltrate the GCL in strip1 mutants. (A) Confocal sections of wild-type and strip1rw147 mutant retinas combined with the transgenic line Tg[ath5:GFP; ptf1a:mCherry-CAAX] to label RGCs and amacrine cells (ACs). Middle panels represent higher magnification. Lower panels show the magenta channel. Arrowheads indicate abnormal positioning of ptf1a+ ACs in the GCL. Asterisks show AC dendritic patterning defects. INL, inner nuclear layer; GCL, retinal ganglion cell layer. Scale bars, 50 μm (upper panels) and 10 μm (middle and lower panels). (B) Percentage of ath5+ area relative to total retinal area. Student’s t-test with Welch’s correction, n ≥ 4. (C) AC numbers per unified retinal area (8500 μm2). Student’s t-test with Welch’s correction, n ≥ 3. (D) Distribution of ACs (GCL or INL) per unified retinal area (8500 μm2). Two-way analysis of variance (ANOVA) with the Tukey multiple comparison test, n ≥ 3. (E) Wild-type and strip1rw147 mutant retinas at 3 dpf labeled with anti-Pax6 antibody which strongly labels ACs. Arrows indicate strong Pax6+ cells that infiltrate the GCL. Nuclei are stained with TOPRO3. Scale bar, 50 μm. (F) The number of strong Pax6+ cells per retina. Student’s t-test with Welch’s correction, n = 5. (G) Percentage of strong Pax6+ cells (GCL+ or INL+) to the total number of strong Pax6+ cells. Two-way ANOVA with the Tukey multiple comparison test, n = 5. (H) Wild-type and strip1rw147 mutant retinas at 3 dpf labeled with anti-Prox1 antibody. Arrows indicate Prox1+ cells that infiltrate the GCL. Nuclei are stained with TOPRO3. Scale bar, 50 μm. (I) The number of Prox1+ cells per retina. Student’s t-test with Welch’s correction, n = 5. (J) Percentage of Prox1+ cells (GCL+ or INL+) to the total number of Prox1+ cells. Two-way ANOVA with the Tukey multiple comparison test, n = 5. For all graphs, data are represented as means ± standard deviation (SD). ns, not significant, *p < 0.05, **p < 0.01, and ****p < 0.0001. Figure 2—source data 1 Data for Figure 2BCDFGIJ. https://cdn.elifesciences.org/articles/74650/elife-74650-fig2-data1-v1.xlsx Download elife-74650-fig2-data1-v1.xlsx Next, we visualized ACs using anti-Pax6 antibody, which strongly labels ACs and weakly labels RGCs (Macdonald and Wilson, 1997). In wild-type siblings, most strong Pax6+ cells were in the INL, and only 9.84% ± 4.13% were in the GCL (Figure 2E, G). However, in strip1rw147 mutants, a significant percentage of Pax6+ cells (44.26% ± 17.8%) was in the GCL (Figure 2E, G). The total number of Pax6+ cells did not differ between wild-type siblings and strip1rw147 mutants (Figure 2F). We confirmed the abnormal positioning of ACs in the GCL using anti-parvalbumin, which labels subsets of ACs in the INL, together with displaced ACs in the GCL (Maurer et al., 2010; Figure 2—figure supplement 1E-G). Next, we visualized BPs using anti-Prox1 antibody, which labels BPs and HCs (Jusuf and Harris, 2009). In wild-type, 100% of Prox1+ cells were in the INL (Figure 2H, J). However, 10.6% ± 6.26% of Prox1+ cells were abnormally located in the GCL in strip1rw147 mutants (Figure 2H, J). The total number of Prox1+ cells did not differ between wild-type siblings and strip1rw147 mutants (Figure 2I). We performed labeling of double-cone and rod PRs using zpr1 and zpr3 antibodies, respectively (Nishiwaki et al., 2008). Apart from occasional mildly disrupted areas, the PR cell layer was grossly intact, with no positioning defects (Figure 2—figure supplement 2A, B). MG and proliferating cells at the ciliary marginal zone were visualized using anti-glutamine synthetase (GS) (Peterson et al., 2001) and anti-PCNA antibodies (Raymond et al., 2006), respectively. Both cell types showed grossly normal positioning in strip1rw147 mutants (Figure 2—figure supplement 2C, D). Thus, in the absence of Strip1, INL cells abnormally infiltrate the GCL and seem to replace the reduced RGCs. Strip1 cell autonomously promotes RGC survival In zebrafish, RGC genesis starts in the ventronasal retina at 25 hpf, spreads into the entire retina by 36 hpf and is completed by 48 hpf (Avanesov and Malicki, 2010; Hu and Easter, 1999). Reduction of RGCs in strip1 mutants could be due to compromised RGC genesis or RGC death after birth. To clarify which, we examined RGC genesis by monitoring ath5:GFP expression, and apoptosis by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL). In strip1rw147 mutants, RGCs are normally produced at 36 hpf; however, apoptosis occurred in the GCL at 48 hpf (Figure 3A). The number of apoptotic cells in GCL reached its highest level at 60 hpf, and apoptotic cells were eliminated by 96 hpf (Figure 3A, B). Accordingly, RGC population was significantly lower in strip1rw147 mutants than in wild-type siblings at 60 hpf and progressively reduced by 96 hpf (Figure 3C). In contrast, other retinal layers of strip1rw147 mutants showed slightly, but not significantly increased apoptosis at 72 hpf (Figure 3—figure supplement 1A), suggesting a specific function of Strip1 in RGC survival. In addition, despite the reduction in ath5:GFP+ area, the total presumptive GCL area, which was defined by retinal area between the lens and the IPL, was unchanged in strip1rw147 mutants throughout the stages (Figure 3—figure supplement 1B), suggesting that infiltrating INL cells replace the lost RGCs. We confirmed RGC death in strip1crisprΔ10 mutants (Figure 3—figure supplement 1C, D). Interestingly, we observed apoptosis in the optic tectum of strip1rw147 mutants (Figure 3—figure supplement 1E, F), suggesting a common Strip1-dependent survival mechanism in the optic tectum. RGCs are the only retinal neurons which project their axons to the optic tectum. In strip1rw147 mutants, RGC axons appeared to exit from the eye cup and formed an optic chiasm at 3 dpf (Figure 3—figure supplement 1G). However, consistent with the reduction of RGCs, the optic nerve was thinner in strip1rw147 mutants than in wild-type siblings and showed elongation defects toward the optic tectum. Figure 3 with 2 supplements see all Download asset Open asset Strip1 cell autonomously promotes retinal ganglion cell (RGC) survival. (A) Transferase dUTP nick end labeling (TUNEL) of wild-type and strip1rw147 mutant retinas carrying the transgene Tg[ath5:GFP] to label RGCs. Nuclei are stained with Hoechst. Scale bar, 50 μm. (B) The number of TUNEL+ cells in ganglion cell layer (GCL). Two-way analysis of variance (ANOVA) with the Tukey multiple comparison test, n ≥ 3. (C) Percentage of ath5+ area relative to total retinal area. Two-way ANOVA with the Tukey multiple comparison test, n ≥ 3. (D) Cell transplantation design to evaluate the cell autonomy of Strip1 in RGC survival. Donor embryos from a strip1rw147 mutant background are labeled with dextran rhodamine and transplanted into host wild-type embryos. Hosts that show transplanted retinal columns at 60 hpf were subjected to TUNEL. (E) 60-hpf host retinas stained with TUNEL FL to visualize apoptotic cells in wild type to wild type (upper panel) or strip1rw147 mutant to wild type (lower panel). Arrows indicate the presence of apoptotic donor cells. Scale bar, 10 μm. (F) Percentage of TUNEL+ donor RGCs relative to total donor RGCs. Mann–Whitney U-test, n = 4. For all graphs, data are represented as means ± SD. *p < 0.05, **p < 0.01, and ****p < 0.0001. Figure 3—source data 1 Data for Figure 3B,C,F. https://cdn.elifesciences.org/articles/74650/elife-74650-fig3-data1-v1.xlsx Download elife-74650-fig3-data1-v1.xlsx To determine whether Strip1 cell autonomously promotes RGC survival, we conducted cell transplantation from strip1rw147 mutant donor cells into wild-type host embryos at the blastula stage. TUNEL of transplanted retinas at 60 hpf revealed that strip1rw147 mutant donor RGCs underwent apoptosis in wild-type host retinas (Figure 3D–F). To address whether Strip1 is also required for RGC neurite development, we repeated the same experiment using mutant donors carrying the transgene ath5:GFP, to examine RGC neurite patterns (Figure 3—figure supplement 2A). Visualization was performed at 57–58 hpf, when wild-type RGCs exhibit apically projected dendrites, while mutant RGCs had not yet undergone complete degeneration. As expected, wild-type transplanted RGCs display uniform dendritic patterns projecting toward the nascent IPL (Figure 3—figure supplement 2B). We can also observe several RGCs projecting their axons basally (arrowheads, Figure 3—figure supplement 2B). However, the majority of mutant RGCs transplanted in wild-type retina show irregular neurite projections, apically directed processes (presumably dendrites) do not project to a uniform layer and show distant abnormal branching (asterisks, Figure 3—figure supplement 2C). We also observe defects in basally directed neurites (probably axons), like bifurcation and misrouting (arrowheads, Figure 3—figure supplement 2C). Taken together, Strip1 is cell autonomously required for survival and neurite morphogenesis of RGCs. RGC death triggers abnormal positioning of ACs, leading to IPL disruption ACs are proposed to be the main cell type responsible for IPL formation (Godinho et al., 2005; Huberman et al., 2010). To clarify how RGC death influences infiltration of ACs into GCL and IPL disruption, we performed time-lapse imaging of wild-type and strip1rw147 mutant retinas combined with the transgenic line Tg[ath5:GFP; ptf1a:mCherry-CAAX]. At 48 hpf, there were no apparent differences in position or morphology of RGCs and ACs between wild-type siblings and strip1rw147 mutants (Figure 4A and Figure 4—videos 1; 2). In strip1rw147 mutants at 52 hpf, RGCs started to disappear, creating an empty spot in the GCL (Figure 4A, asterisks). However, ACs were still located in the INL. At 55 hpf, a rudimentary IPL was observed in the central retina of both wild-type siblings and strip1rw147 mutants. At 59 hpf, ACs started to invade the empty spaces in the GCL (Figure 4A, arrowheads). Infiltration of ACs into the GCL was more prominent at 62 hpf, resulting in a fluctuating IPL. Thus, loss of RGCs triggers infiltration of ACs into the GCL in strip1rw147 mutants. Figure 4 with 3 supplements see all Download asset Open asset Retinal ganglion cell (RGC) death triggers abnormal positioning of amacrine cells (ACs) leading to inner plexiform layer (IPL) disruption. (A) Time-lapse imaging of wild-type and strip1rw147 mutant retinas combined with the transgenic line Tg[ath5:GFP; ptf1a:mCherry-CAAX] to track ACs and RGCs during IPL formation. Asterisks denote empty areas in the ganglion cell layer (GCL). Arrowheads represent infiltration of ACs into empty spaces in the GCL. Panels on the right show higher magnification of outlined areas. Scale bar, 50 μm. (B) Cell transplantation design to evaluate the cell autonomy of Strip1 in AC-mediated IPL formation. Donor embryos are from intercross of strip1rw147 heterozygous fish combined with Tg[ptf1a:mCherry-CAAX] to label ACs. Host embryos are generated by nontransgenic intercross of strip1rw147 heterozygous fish. Donor cells are labeled with dextran Alexa-488 and transplanted into host embryos to make chimeric host retinas with donor-derived retinal columns. (C) Confocal images of four combinations of transplantation outcomes: wild type to wild type, wild type to mutant, mutant to wild type, and mutant to mutant. Arrowheads indicate abnormal positioning of ACs in basal side of IPL. Scale bar, 20 μm. (D) Percentage of ACs (either at the apical or the basal side of the IPL) relative to the total number of ACs within a transplanted column. Two-way analysis of variance (ANOVA) with the Tukey multiple comparison test, n ≥ 4. Data are represented as means ± standard deviation (SD). **p < 0.01 and ****p < 0.0001. Figure 4—source data 1 Data for Figure 4D. https://cdn.elifesciences.org/articles/74650/elife-74650-fig4-data1-v1.xlsx Download elife-74650-fig4-data1-v1.xlsx To examine whether Strip1 is required in ACs for IPL formation, we performed cell transplantation using donor embryos carrying the transgene Tg[ptf1a:mCherry-CAAX] (Figure 4B). When mutant ACs were transplanted into wild-type host retinas, most donor ACs were normally positioned in the INL and extended dendrites toward the IPL, as in the case of wild-type donor ACs transplanted into a wild-type host retina (Figure 4C, D). Occasionally, three ACs extended two dendritic trees instead of 1 among 73 transplanted ACs; however, such dendritic misprojection did not perturb IPL formation (Figure 4—figure supplement 1A). On the other hand, as with mutant donor ACs transplanted into mutant host retinas, when wild-type donor ACs were transplanted to mutant host retinas, they showed irregular neurite projection with many somas abnormally located toward the basal side, resulting in IPL formation defects (Figure 4C, D and Figure 4—figure supplement 1A). These data suggest a non-cell autonomous function of Strip1 in ACs for IPL formation. Similarly, we conducted cell transplantation to assess the role of Strip1 in BPs. Mutant donor BPs labeled with the transgene Tg[xfz43] (Zhao et al., 2009) projected axons normally toward the IPL in wild-type host retinas, in the same fashion as wild-type donor BPs (Figure 4—figure supplement 1B–D). Few t