Introduction

Unlike mammalian counterparts, retinal Müller glia (MG) of aquatic and amphibian species can enter injury-induced regeneration program(Goldman, 2014; Lahne et al., 2020b; Wan and Goldman, 2016), which unfolds a set of temporal events from glial reactivation, glial proliferation, new neuron generation, and circuit integration(Abraham et al., 2024; Powell et al., 2016). The initial glial reactivation is highly conversed across species, from lower vertebrates to mammals, with the characteristic glial expression of ascl1α, lin28a, sox2, and mycb/h and increased morphological complexity(Gorsuch et al., 2017; Jorstad et al., 2017; Lee et al., 2024; Ramachandran et al., 2010). Diverse signaling pathways have been identified to participate in this initial glial reactivation(Campbell et al., 2021; Wan and Goldman, 2017).

In contrast, our knowledge of the mechanism underlying injury-induced glial cell-cycle re-entry, a critical step for neuron regeneration specific to aquatic and amphibian species but not mammals(Lenkowski and Raymond, 2014; Vihtelic and Hyde, 2000; Wu et al., 2001), remains limited. Even in the case of an enforced proliferation by over-expressing cell-cycle regulators, MG can undergo gliogenesis rather than neurogenesis in the rat retina(Close et al., 2005; Close et al., 2006; Nishino et al., 2023). The mechanistic understanding of MG cell-cycle re-entry in the injured zebrafish retina will provide critical knowledge for inspiring new strategies for in-vivo re-programming MG to repair the human retina.

The intrinsic and environmental factors contribute to injury-induced MG proliferation(Gao et al., 2021; Lahne et al., 2020b; Xiao et al., 2023). Recent efforts in single-cell RNA-sequencing (scRNA-Seq) have identified ascl1α, clcf1/crlf1a(Boyd et al., 2023), and mycb/mych(Lee et al., 2024) as key intrinsic drivers of injury-induced MG proliferation in the zebrafish retina. Also, the elegant study of MG injury responses across multiple species, including zebrafish, chicken, and mice, revealed the hmga1/yap1 signaling network as a key regulator of MG reprogramming and neurogenesis in the regenerating retina(Hoang et al., 2020). Interestingly, previous studies further reported that post-injury MG in the zebrafish are capable of re-acquiring the regeneration program that largely recapitulates the embryonic retinal developmental program(Celotto et al., 2023; Hoang et al., 2020; Lahne et al., 2020a; Lyu et al., 2023).

In terms of extrinsic cues, two sources, those from damaged neurons and injury-recruited microglia, are considered indispensable for injury-induced MG proliferation(Leach et al., 2021; White et al., 2017). Previous studies have shown that signals derived from various retinal neuron injury models trigger MG regeneration, each exerting distinct effects on MG reprogramming(Lyu et al., 2023). Meanwhile, accumulative evidence increasingly appreciates the role of inflammatory signals derived from injury-recruited microglia, such as TNFα(Conner et al., 2014; Nelson et al., 2013), cytokines IL-1β, and IL-10(Lu and Hyde, 2024). Notably, inflammatory responses are intrinsically associated with redox signaling, which is involved in the regeneration processes in various non-neuronal tissues(Breus and Dickmeis, 2021; Jaeschke, 2000; Zhang et al., 2024). However, the involvement of the redox signaling resulted from microglia-derived inflammatory signals in MG cell-cycle re-entry following retina damage remains to be clarified. Nitric oxide (NO), a redox signal, has participated in regenerating various tissues in zebrafish, including the heart(Rochon et al., 2020; Yu et al., 2024), spinal cord(Bradley et al., 2010), and fin(Matrone et al., 2021). Endogenous NO is derived from three forms of NO synthase (NOS) in mammalians: neuronal NOS (nNOS), endothelial NOS (eNOS), and inducible NOS (iNOS). The zebrafish comprises Nos1, a form of neuronal Nos, and Nos2a/2b, two inducible Nos.

In this study, we performed single-cell RNA sequencing to characterize the post-injury MG states in the zebrafish retina following the specific ablation of green/red cone. This comprehensive analysis led us to define a set of previously unknown cxcl18b positive transitional states as an essential route of MG cell-cycle re-entry in response to the injury, and the inflammatory reactions of recruited microglia were indispensable for the induction of these transitional states. Intriguingly, we found these cxcl18b-defined transitional states exhibited gene patterns shared by cell states in the CMZ, favoring the idea that cxcl18b-defined MG transitional states might represent the developmental program conserved by neuronal regeneration beyond the embryonic development. A remarkable phenotype of enriched redox features in these gene patterns suggested the importance of redox signaling. The subsequent screening of redox-related genes revealed the essential role of nitric oxide, produced by Nos2b, in triggering MG cell-cycle re-entry after the retinal injury. Notably, we developed a sophisticated viral-based approach to achieve the cxcl18b-defined MG transitional state-specific knockout nos2b successfully and verified the requirement of transitional state-specific NO signaling in injury-induced MG proliferation. These findings provide novel cellular and molecular insights into this species-specific post-injury MG cell-cycle re-entry process, with potential implications for the development of regenerative medicine strategies.

Results

Single-cell transcriptome analysis reveals the landscape of injury-induced Müller glia states

Zebrafish Müller glia can respond to retina injury by the cell cycle re-entry, a critical step evolutionarily absent from their mammalian counterparts but essential for neuron regeneration(Goldman, 2014; Powell et al., 2016). We created a zebrafish retina injury model by crossing Tg(opn1lws2: nfsb-mCherry)^uom3 (referred to as Tg(lws2: nfsb-mCherry)) with Tg(mpeg1: GFP) fish, in which the bacterial nitroreductase (NTR) enzyme was specifically expressed in G/R cone. We selectively ablated the G/R cone starting at 5 days post-fertilization (dpf) by a subsequent 120 hours of metronidazole (MTZ) exposure(Curado et al., 2007; Curado et al., 2008) (Figure 1A and Figure 1-figure supplement 1A). G/R cone became significantly reduced in number since 48 hours post-injury (hpi) and was mostly depleted at 120 hpi (Figure 1B). Meanwhile, a number of microglia (marked by Tg(mpeg1: GFP)) migrated to the outer nuclear layer (ONL) as early as 48 hpi, peaked at 72 hpi, and began to reduce in number at 96 hpi and onward (Figure 1C). Notably, in response to G/R cone ablation, the proliferative MG population increased starting at 48 hpi, peaked at 72 hpi, and began to decline since 96 hpi and forward (Figure 1D). In light of the result that the number of proliferative MG peaked at 72 hpi, we focused on the 72-hpi time point for the further exploration of MG proliferative behaviors following G/R cone ablation (Figure 1E).

Figure 1.

By revisiting previously obtained scRNA-seq data of MG enriched from Tg(lws2: nfsb-mCherry) crossed with Tg(gfap: EGFP) and Tg(her4.1: dRFP) retina before and after G/R cone ablation at 72 hpi(Krylov et al., 2023), we selected 5,932 and 3,999 MG cells and their derived progenies from the uninjured and 72-hpi retina, respectively (Details in Methods; Figure 1-figure supplement 1B). By clustering these cells, we identified 13 clusters, 8 out of which (Clusters 2, 3, 5, 6, 9, 10, 11, and 12) with an increased proportion in response to the G/R cone ablation (Figure 1-figure supplement 1C). Subsequently, we re-clustered cells of these 8 clusters (695 cells of uninjured retinae and 3,477 cells of 72-hpi retinae), aggregating into 10 new clusters. After the quality control procedure, we did not consider Clusters 7/8/9 due to their small populations with ribosomal, dendritic, and doublet features, resulting in 7 clusters for further analysis. (Details in Methods; Figure 1F)

We performed the pseudo-time trajectory analysis to reveal the progression of these MG clusters following the cone ablation (Figure 1G). Cluster 4 cells were highly expressing genes related to mature MG (glula, slc1a2b, apoeb, and rlbp1a)(Bernardos and Raymond, 2006; Raymond et al., 2006; Thummel et al., 2008; Yurco and Cameron, 2005), the quiescent state (cx43)(Dermietzel et al., 2000; Janssen-Bienhold et al., 1998) and major MG population marker (fgf24)(Krylov et al., 2023) (Figure 1-figure supplement 1D). Furthermore, Cluster 4 began to express s100α10b and gfap, reactive state makers(Celotto et al., 2023; Hoang et al., 2020), in the injured retinae, but not in the uninjured retina (Figure 1-figure supplement 1D). Thus, we set Cluster 4 as the most original MG states.

To examine the transition of these 7 MG clusters, the pseudo-time trajectory showed that the main developmental branch consisted of Clusters 4/1/2 and then became divergent into two sub-branches, including Cluster 0 and 5/3/6 (Figure 1G). In contrast to the sub-branch of Cluster 0, the sub-branch of Clusters 5/3/6 was highly expressing proliferative cell markers (pcna, mki67, and mcm2). Within the latter sub-branch, while Cluster 3 and 6 with the highest levels of proliferative cell markers, Cluster 6 began to express neuronal differentiation factors (otx5, crx, and pde6gb)(Abalo et al., 2020; Asaoka et al., 2014; Shen and Raymond, 2004) (Figure 1-figure supplement 1D). Thus, we identified 6 major post-injury MG states, from the most original states (Cluster 4) to three transitional states (Cluster 1/2/5), to finally two proliferative states (Cluster 3/6).

Remarkably, chemokine (C-X-C motif) ligand 18b (cxcl18b), an inflammatory chemokine, was uniquely expressed in three transitional states but largely absent from the most original Cluster 4 and two proliferative Cluster 3/6. Specifically, while the first cxcl18b⁺ transitional state (Cluster 1) was expressing cx43, a maker for MG quiescence(Dermietzel et al., 2000; Janssen-Bienhold et al., 1998), the last transitional state (Cluster 5) began to show a weak induction of ascl1α, a master regulator of injury-induced MG reprogramming(Ramachandran et al., 2010). Immediately following this last transitional state, Cluster 3 started with high ascl1α expression and entered the proliferative state with the expression of pcna and mik67 (Figure 1H). Our analysis highlighted a new set of cxcl18b-defined MG transitional states preceding ascl1α induction, bridging MG from the most original quiescence state to injury-induced proliferation.

Clonal analysis revealing injury-induced MG proliferation via inflammation inducted cxcl18b-defined transitional states

To directly verify the presence of cxcl18b-defined MG transitional states, we first examined the temporal relationship of cxcl18b expression and MG proliferation after the cone ablation using in situ hybridization combined with immunostaining of either BLBP (an MG maker) or PCNA (a proliferative cell maker) (Figure 2A and Figure 2-figure supplement 1A). The result showed that as early as 24 hpi, the number of cxcl18b⁺ MG was rapidly peaked with no emergence of proliferative MG (cell number: 11 ± 4, n=10 in cxcl18b⁺ MG; mean ± SEM), and then cxcl18b⁺ MG continued declining in number over time and reached the lowest level since 96 hpi (1 ± 1, n=7; mean ± SEM; Figure 2B). In contrast, the number of proliferative MG (PCNA⁺) peaked at 72 hpi and decreased to the lowest level at 120 hpi (9 ± 2, n=11 in 72-hpi retina; 1 ± 1, n=5 in 120-hpi retina; mean ± SEM; Figure 2B). Note that cxcl18b⁺ MG was mostly proliferative at 72 hpi (Figure 2-figure supplement 1B).

Figure 2.

To further verify the temporal expression of cxcl18b in MG following the cone ablation, we created a new transgenic reporter Tg(cxcl18b: GFP) by cloning a 3k-bp-long cis-element of 5’UTR with GFP, allowing real-time monitoring of injury-induced cxcl18b expression in vivo (Figure 2C and Figure 2-figure supplement 1C-D). Combining this line with PCNA immunostaining, we confirmed the remarkable increase of cxcl18b expression (GFP⁺) at 48 hpi (0 ± 1, n=4 in uninjured retinas vs 11 ± 4, n=7 in 48-hpi; mean ± SEM; Figure 2D and E). Due to the prolonged stay of GFP protein, we could also observe some GFP⁺ MG has become proliferative (PCNA positive) (Figure 2D). The result of proliferative MG as a subpopulation of cxcl18b⁺ MG led to an outstanding question as to whether cxcl18b-defined MG transitional states represented an essential route to injury-induced proliferation.

To address it, we created a new transgene Tg(cxcl18b: Cre-vmhc: mCherry:: ef1α: loxP-DsRed-loxP-EGFP:: lws2: nfsb-mCherry) to perform the clonal analysis of cxcl18b-expressing MG after the G/R cone ablation (Details in Method; Figure 2-figure supplement 1E). After ablating the G/R cone by MTZ treatment at 6 dpf for 3 consecutive days, we found that the numbers of cxcl18b⁺ MG and PCNA⁺ MG were significantly increased at 72 hpi (Figure 2F). Further analysis showed that all PCNA⁺ MG were cxcl18b⁺, indicating the cxcl18b⁺ MG were the ones who could eventually enter into the cell cycle (13 ± 2 cells for PCNA⁺ vs 13 ± 2 cells PCNA⁺ and cxcl18b⁺ MG; percentage of PCNA⁺ and cxcl18b⁺ vs PCNA⁺ = 93 ± 2%; n=14; p=0.19; Mean ± SEM; Figure 2G). Further analysis showed that at 120 hpi, the time point that the MG proliferation has largely creased, GFP⁺ MG with the lineage history of injury-induced cxcl18b expression constituted about 44% of GS⁺ MG at the central retina, indicating that only about half of MG could enter cxcl18b⁺ transitional states following the cone ablation (cell number of clones: 14 ± 3 in GS⁺ and Cre⁺ MG; 28 ± 4 in GS⁺ and Cre^- MG, n=14; p<0.05; mean ± SEM; Figure 2I). Together, our clonal analysis demonstrated that proliferative MG mostly originated from cxcl18b⁺ MG transitional states, and 44% central MG could become cxcl18b positive. To investigate whether cxcl18b was required for MG proliferation following G/R cone ablation, we employed CRISPR-Cas9-mediated gene disruption, using two sgRNAs targeting cxcl18b (Figure 2-figure supplement 1F and G). We found that cxcl18b knockout did not reduce MG proliferation after G/R cone ablation at 72 hpi (13 ± 3, n=11 in WT; 11 ± 3, n=7 in scramble sgRNA-injected; 13 ± 3, n=7 in cxcl18b sgRNA-injected; mean ± SEM), suggesting that cxcl18b per se does not regulate MG proliferation directly (Figure 2-figure supplement 1H and I). This led us to wonder about the induction of cxcl18b-defined MG transitional states.

As an inflammatory chemokine, cxcl18b serves as a reliable marker of inflammation and regulates neutrophil recruitment to injury sites(Goumenaki et al., 2024; Torraca et al., 2017). Inflammation has been previously shown to be critical for inducing regenerative responses in adult zebrafish, where it promotes reactive microglia/macrophages and Müller glial proliferation in the retina(Iribarne and Hyde, 2022; Kyritsis et al., 2012). Notably, suppressing the immune response using dexamethasone (Dex) in zebrafish retina reduced microglial reactivation and significantly decreased the number of proliferative MG(Silva et al., 2020; Zhang et al., 2020). In our study, we identified the cxcl18b-defined transitional states as the essential routing for MG proliferation after G/R cone ablation. These results prompted us to investigate whether the inflammatory responses mediated by recruited microglia are indispensable for the formation of these cxcl18b-defined transitional states.

To address this, we examined cxcl18b expression using Tg(cxcl18b: GFP) after inhibiting inflammation using Dex(Iribarne and Hyde, 2022), and observed a significant reduction in the number of cxcl18b⁺ MG (GFP⁺ cells) at 72 hpi (11 ± 4, n=10 in Dex treated retina vs 19 ± 5, n=6 in DMSO treatment; and 18 ± 3, n=14 in the 72-hpi retina; mean ± SEM) (Figure 2K). Consistent with earlier reports, we observed that Dex treatment inhibited the migration of microglia (indicated by Tg(mpeg1: GFP); 16 ± 4, n=14 in Dex treated retina vs 27 ± 5, n=7 in DMSO treatment; and 28 ± 4, n=9 in the 72-hpi retina; mean ± SEM) to the ONL and significantly reduced the number of proliferative MG (PCNA⁺; 8 ± 2, n=14 in Dex treated retina vs 17 ± 2, n=7 in DMSO treatment; and 14 ± 3, n=9 in the 72-hpi retina; mean ± SEM) at 72 hpi after G/R cone ablation (Figure 2L and M and Figure 2-figure supplement 1J). These findings strongly suggested that injury-induced, microglia-mediated inflammation is critical for activating the cxcl18b-defined transitional states that drive MG proliferation.

cxcl18b-defined MG transitional states recapitulate molecular features of retinal stem cells in the ciliary marginal zone

Interestingly, we observed the cxcl18b expression in the CMZ after the cone ablation besides its high expression in the MG (Figure 2A). We were then curious about the cxcl18b expression in the developmental retina as well as in the CMZ without the injury. Notably, in situ results showed that cxcl18b was largely absent from the central region but presented in the peripheral region of 30-hpf retinae, whereas it was highly expressed in the most peripheral region of the CMZ, where fabp11a and col15α1b, two putative makers for post-embryonic retinal stem cells (RSCs) are located(Gonzalez-Nunez et al., 2010; Raymond et al., 2006) (Figure 3A-B and E-F, and Figure 3-figure supplement 1A). The transgenic line of Tg(cxcl18b: GFP) also showed a robust cxcl18b expression in the CMZ (Figure 3-figure supplement 1D). Consistently, our scRNAseq data of CMZ cells also confirmed the co-expression of cxcl18b, fabp11a, and col15α1b (Figure 3C-D and G-H, and Figure 3-figure supplement 1B). Furthermore, cluster 1 MG, at the earlier stage of transitional states, has the highest cxcl18b expression with col15α1b expression (Figure 1F and Figure 3-figure supplement 1C). All these results suggested that cxcl18b-defined transitional states, at least to some extent, represent the developmental state of retinal stem cells in the CMZ, but not that of embryonic retinal progenitors.

Figure 3.

Nos2b is required for MG entry into the proliferation via NO signaling

Previous studies have demonstrated the involvement of redox signaling in cell regeneration processes in various tissues across species(Han et al., 2014; Hunter et al., 2018; Matrone et al., 2021; Yoo et al., 2012). All these evidences led us to directly test the roles of redox genes in serving as the molecular mechanism underlying injury-induced MG proliferation. Thus, we first examined the expression levels of a comprehensive list of redox genes in cxcl18b-defined MG transitional states (Clusters 1/2/5) in our scRNA-seq data (Figure 1H) and screened the influence of 18 genes from each major category of redox signaling on injury-induced MG proliferation using CRISPR-Cas9-mediated gene disruption (Figure 4-figure supplement 1A). We focused on the nitric oxide (NO) signaling pathway, targeting three genes encoding nitric oxide synthases (Nos): neuronal Nos (nos1) and two inducible forms (nos2a and nos2b), as well as the gene encoding S-nitrosoglutathione reductase (gsnor or adh5), which modulates reactive NO signaling (Figure 4A and Figure 4-figure supplement 1B). The consequence and efficiency of gene disruption were verified by DNA sequencing (Figure 4-figure supplement 1C). Notably, the disruption of nos2b resulted in a significant reduction of PCNA⁺ MG at 72 hpi (number of PCNA⁺ clones: 6±2, n=22 in nos2b-disrupted vs 11±3, n=7 in scramble sgRNA-injected; Mean ± SEM) (Figure 4B). Noted that nos gene disruption did not significantly alter microglia recruitment or G/R cone ablation at 72 hpi, suggesting that the influence of NO on injury-induced MG proliferation was not via inflammatory reactions of recruited microglia or injury degree (Figure 4-figure supplement 1D-F).

Figure 4.

To further examine the function of Nos2b via NO, we employed various Nos inhibitors (L-NG-nitro arginine methyl ester, L_NAME; L-NG-Monomethyl Arginine, L_NMMA; 1400W) and NO scavengers (carboxy-PTIO, C-PTIO)(Goldstein et al., 2003; Hong et al., 2012; Moore et al., 1990; Rees et al., 1990). We performed the intravitreal injection of the drugs (PBS as control) into the zebrafish eye from 2 days before cell ablation until 72 hpi (Figure 4C). Notably, the NO scavenger C-PTIO mostly suppressed MG proliferation, indicating the involvement of NO (cell number of PCNA⁺ MG: 3±2, n=27 after blocking NO by C-PTIO vs 14±4, n=16 in PBS-injected retina; Mean ± SEM) (Figure 4D and E). Moreover, 1400W (an inhibitor specific to inducible Nos, 7±2, n=10; Mean ± SEM) and L_NAME (a broad inhibitor to all three Nos forms, 8±2, n=12; Mean ± SEM) could also significantly reduce the number of proliferative MG after the ablation, whereas L_NMMA (the inhibitor specific to neuronal Nos) did not influence MG proliferation (11±3, n=12; Mean ± SEM) (Figure 4E). Taken together, these results highlight the critical role of NO signaling in regulating injury-induced MG proliferation.

We further generated nitric oxide pathway mutant zebrafish (nos1, nos2a, nos2b, and gsnor) to investigate the role of NO in MG proliferation following G/R cone ablation. Utilizing CRISPR/Cas9-mediated gene disruption, we successfully screened out nos1, nos2a, nos2b, and gsnor mutants, characterized by deletions of 133-bp, 13-bp, 220-bp, and 11-bp coding sequences, respectively (see Methods for details, Figure 4-figure supplement 2A and B). Consistent with the gene disruption experiment above, in both heterozygous and homozygous, we observed a significant reduction in proliferative (PCNA⁺) MG at 72 hpi following G/R cone ablation in nos2b mutant (3±1, n=20 in nos2b^+/- mutants and 6±2, n=20 in nos2b^-/-, vs 11±2, n=14 in WT; Mean ± SEM), while no influenced across the nos mutants (nos1, nos2a) and gsnor mutants (Figures 4F-4I).

Interestingly, the reduction in proliferative MG was more pronounced in nos2b heterozygous mutants (nos2b^+/-) than in homozygous mutants (nos2b^-/-) (nos2b^+/-vs nos2b^-/-; p<0.001; Mean ± SEM) (Figure 4J). We observed that cone ablation induced by injury was not significant in nos2b mutants, indicating that the observed differences in MG proliferation regulated by NO in these mutants are independent of the injury (45±8 photoreceptor cells remain, n=24 in WT; vs 49±12, n=20 in nos2b^+/- mutants and 46±9, n=20 in nos2b^-/-; Mean ± SEM) (Figure 4K). This unexpected result suggests a concentration-dependent effect of NO on proliferative MG. Specifically, compared to homozygous mutants, heterozygous mutants with intermediate NO levels more effectively suppressed MG proliferation, whereas wild-type animals with higher NO levels promoted MG proliferation. This concentration-response pattern highlights the role of NO as a regulator, rather than a mediator, of injury-induced MG proliferation.

Specific nos2b expression in cxcl18b-defined transitional states MG after G/R cone ablation

To further examine whether the expression of nos2b in cxcl18b⁺ MG, we used the newly generated transgene fish line in our study Tg(cxcl18b: GFP), enabling us to monitor cxcl18b⁺ MG after the G/R cone ablation in a real-time manner. By crossing different fish lines, we fluorescently sorted out three post-injury MG populations (72-hpi MG, 72-hpi PCNA⁺ MG, and 72-hpi cxcl18b⁺ MG) and three control groups (uninjured MG, 72-hpi retinal cells other than MG and 72-hpi G/R cone) (Figure 5A-C). The real-time quantitative PCR (RT-qPCR) analysis showed that compared to the three control groups, the expression of nos2b was significantly higher in cxcl18b⁺ MG than in 72-hpi MG and 72-hpi PCNA⁺ MG (relative expression of nos2b: 89±32, repeats n=7 in cxcl18b⁺ MG; 14±17, repeats n=5 in 72-hpi MG; and 1±1, repeats n=6 in PCNA⁺ MG; Mean ± SEM) (Figure 5D and Figure 5-figure supplement 1D-F). Together with the fact that cxcl18b⁺ MG contained PCNA⁺ and PCNA^- populations in Tg(cxcl18b: GFP) (Figure 2D), our result indicated that nos2b was specifically expressed in cxcl18b⁺ non-proliferative MG, which was in agreement with the scRNA-seq result of the specific expression of cxcl18b in three transitional states (Clusters 1, 2, and 5; Figure 1H). In situ hybridization using an HCR molecular probe in the Tg(lws2: nfsb-mCherry x cxcl18b: GFP) fish line also suggests that injury-induced nos2b expression was specific in the cxcl18b-defined transitional state MG (Figure 5-figure supplement 1A-C). Thus, our analysis reveals that nos2b was specifically expressed in cxcl18b⁺ transitional MG states, particularly in non-proliferative cells.

Figure 5.

NO produced by nos2b determines cxcl18b⁺ MG entry for proliferation

To further explore the role of cxcl18b⁺ MG-specific nos2b in regulating the entry of MG proliferation after the G/R cone ablation, we developed a sophisticated clonal analysis of cxcl18b⁺ MG with genetic disruption of nos2b using CRISPR-Cas9 method. To achieve glial type-specific gene manipulation, we employed an adenovirus strain that specifically infects radial glia in zebrafish(Jia et al., 2019; Liu et al., 2021), and confirmed that it could faithfully mark MG in the zebrafish retina (Figure 6A and B). For the clonal analysis, we performed the intravitreal injection of two viruses packaged with elements of cxcl18b: gal4 and UAS: Cas9-T2A-Cre-u6: sgRNA (nos2b) into the eye of Tg(lws2: nfsb-mCherry,ef1α: loxP-DsRed-loxP-EGFP) at 5 dpf, virus packaged with element of UAS: Cas9-T2A-Cre-u6: empty as the control (Figure 6C-D and Figure 6-figure supplement 1A). One day after the virus infection, the fish was treated with MTZ for 3 consecutive days to ablate the G/R cone (Figure 6E).

Figure 6.

At 72 hpi, we collected the clones (GFP⁺) derived from cxcl18b⁺ MG and analyzed their proliferative property (Figure 6F and G). The analysis of virus-infected clones (GFP⁺) cell cycle re-entering (PCNA⁺) after G/R cone ablation revealed that only ∼23% of GFP⁺ MG clones remained PCNA⁺ in the nos2b knockout group, compared to ∼75% in the control group (Figure 6H). Note that the efficiency of nos2b sgRNAs was confirmed in terms of mutation types and knockout efficiency by sequencing (Figure 6-figure supplement 1B-D).

Taken together, these findings suggested that it is the NO produced by Nos2b within cxcl18b-defined transitional state MG that specifically drives MG from the transitional state into proliferation following injury, highlighting a pivotal mechanism underlying injury-induced regenerative processes.

NO decreased Notch activity that is responsible for injury-induced MG proliferation

Previous studies have shown that ascl1α and Notch signaling are essential for MG proliferation in the injured zebrafish retina(Conner et al., 2014; Wan et al., 2012). Regarding Notch signaling, a high Notch3 expression is reported to maintain MG quiescence. In response to the injury, notch3 expression is downregulated, but notch1a is necessary for the continued proliferation of the progenitors(Campbell et al., 2022). In consistent, we observed that notch3 and hey (the Notch downstream target) were highly expressed in uninjured MG clusters and became reduced from the most original states (Cluster4) to the proliferative states (Cluster 3 and 6), whereas notch1a/1b and ascl1α were prominently expressed in the late stage of cxcl18b⁺ transitional states (clusters 5) and the proliferative states (Cluster 3; Figure 1H and 7A). Interestingly, upstream regulators of Notch signaling activation, such as fgf8a, fgf8b(Wan and Goldman, 2017), and tgfb3(Lee et al., 2020), were predominantly expressed in clusters 4 and 1, preceding the expression of cxcl18b (Figure 7A). These results led us to wonder whether NO regulated cxcl18b-defined transitional state MG cell cycle re-entry via the Notch signaling pathway.

Figure 7.

To examine the influence of NO signaling blockage on Notch activity dynamics following the cone ablation, we employed a reporter line Tg(Tp1bglob: EGFP) (referred to as Tg(Tp1:EGFP)), in which EGFP is driven by the TP1 element, the direct target of the intracellular domain of Notch receptors (NICD) that is generated upon Notch activation(Parsons et al., 2009; Quillien et al., 2014). We treated fish with the NO scavenger C-PTIO and MTZ, starting at 5 dpf for 5 continuous days and at 7 dpf for 3 continuous days, respectively, followed by immunostaining for EGFP and PCNA (Figure 7B). Interestingly, Tp1: EGFP⁺ MG were significantly reduced at all three injury time points (cell number of EGFP⁺ clones: 34±3, n=9 in 24-hpi retina; 29±3, n=4 in 48-hpi retina; 33±4, n=6 in 72-hpi retina vs 42±1, n=4 in the uninjured retina; Mean ± SEM), demonstrating a decrease in Notch activity following G/R cone ablation (Figure 7C). Notably, this reduction in Notch activation was further rescued by NO blocking using C-PTIO (cell number of EGFP⁺ clones: 42±1, n=6 in C-PTIO treated retina at 24 hpi; 42±2, n=5 at 48 hpi; 44±2, n=7 at 72hpi; Mean ± SEM), suggesting that NO modulates Notch signaling (Figure 7C). Meanwhile, C-PTIO treatment significantly reduced the number of proliferative MG (marked by PCNA) (Figure 7D). These findings indicated that injury-induced NO suppresses Notch signaling activation, which potentially drives MG to exit quiescence and enter proliferation. Together, our results highlight that NO signaling drove the injury-induced cxcl18b-defined transitional state MG to enter into proliferation, which may be mediated by Notch pathway regulation.

Discussion

Our study provides the single-cell transcriptomic landscape of Müller glia (MG) state progression following the cone ablation. Combined with clonal analysis, we identified a previously unreported cxcl18b-defined transitional MG state as the essential routing for MG cell cycle re-entry. It led to our further genetic analysis revealing the concentration-dependent regulatory mechanism of nitric oxide derived from nos2b underlying injury-induced MG proliferation. Furthermore, NO signaling accounted for a decreased Notch activity, which has been previously reported to be essential for MG proliferation after the injury. Finally, cell state-specific gene disruption revealed that cxcl18b-defined MG transitional state-specific nos2b is required for injury-induced MG proliferation. Thus, our study provides the novel redox-related mechanism underlying MG proliferation in response to cone ablation in the zebrafish retina, opening up exciting possibilities for future research and potential therapeutic interventions.

Identification of cxcl18b-defined MG transitional states following the cone ablation

Combined scRNA-seq analysis and clonal analysis, we identified a previously unreported transitional MG state, marked by the expression of cxcl18b, as the essential routing for MG to re-enter the cell cycle following the retina injury (Figure 1H, 2F, and 2G). To our knowledge, it is the first transitional state verified by in vivo clonal analysis to show a faithful prediction of injury-induced MG proliferation. Notably, this cxcl18b induction in MG depends on microglial recruitment in response to cone ablation, suggesting this transitional state is an MG response to the signals derived from the inflammatory reaction (Figure 2K). The underlying mechanism of this crosstalk is crucial to be addressed. Interestingly, the cxcl18b-containing gene module was also expressed in the CMZ, a region crucial for adult retina neurogenesis for a lifetime (Figure 3A-D). However, its expression is mainly absent from the central regions of the developing retina (Figure 3F). It suggests that the cxcl18b-defined transitional state might represent a developmental state used by constitutive neuron generation programs and injury-induced neuron regeneration programs beyond embryonic development. Furthermore, the cxcl18b-defined transitional state exhibits robust redox-related characteristics, such as the expression of sod1, sod2, and catalase (Figure 4-figure supplement 1A). It led to the critical discovery of this study, which showed the essential role of nos2b in regulating injury-induced MG proliferation. Unfortunately, our preliminary effects on gene disruption using CRISPR/Cas9 in F0 founders did not observe a significant reduction in the number of proliferative MG after cxcl18b disruption (Figure 2-figure supplement 1I). However, a recent study reported the essential role of cxcl18b in heart regeneration using mutant fish, providing a mechanism of cxcl18b as innate immune signaling in injury-induced tissue regeneration(Goumenaki et al., 2024). Their results raised concern about the efficiency of cxcl18b disruption in our system. It is essential to use mutant fish to re-examine the role of cxcl18b in injury-induced MG proliferation in the future. Also, we cannot rule out the possibility that other co-factors are involved in the action of cxcl18b in MG regeneration, which is another critical issue that needs to be solved in the future.

The possible mechanism of NO signaling underlying injury-induced MG proliferation

Our study, for the first time, demonstrated an essential role of NO signaling in regulating MG proliferation after the cone ablation. However, we still need to understand more about the underlying mechanism. There are two well-characterized molecular events responsible for MG proliferation following the retina injury: a decreased Notch activity and an increased ascl1α expression(Fausett et al., 2008; Jorstad et al., 2017; Ramachandran et al., 2010). Previous studies have shown that Notch3 is responsible for this decreased Notch activity, leading to increased Ascl1a through the de-depression mechanism(Campbell et al., 2021; Sahu et al., 2021). Interestingly, unlike notch3, notch1a has been reported to stimulate MG proliferation(Campbell et al., 2021; Campbell et al., 2022). Our scRNA-seq analysis also showed that as MG progressed into the proliferative states after the cone ablation, notch3 expression gradually declined (Figure 7A). In contrast, ascl1α, notch1a, and notch1b expressions were up-regulated and peaked at the proliferative states (Figure 7A). Thus, both previous studies and our current analysis support the idea that the transcriptional regulation of Notch expression accounts for the decreased Notch activity after the injury. Intriguingly, NO has been reported to activate the Notch1 signaling cascade by promoting the release and accumulation of the Notch1 intracellular domain (NICD) through nitrification reactions, subsequently enhancing tumorigenesis and stem-like features in various cellular systems(Charles et al., 2010; López-Juárez et al., 2017; Villegas et al., 2018). It raises the possibility that NO signaling regulates injury-induced MG proliferation through the post-translational modification of Notch3. Previous studies have demonstrated two significant forms of NO-mediated post-modification: S-nitrosylation on cysteines and nitration on tyrosine. Our preliminary analysis showed that all 11 cysteine residues within the putative γ-secretase-dependent cleavage sites are conserved between Notch1a and Notch3, while notable differences were observed at four tyrosine residues. It leads to an outstanding question of whether NO regulated injury-induced MG proliferation by decreasing Notch activity via tyrosine nitration of Notch3, which is worthwhile to elucidate.

The production of NO signaling in MG following retina damage

Previous studies reported that cxcl18b is a reliable inflammatory marker(Goumenaki et al., 2024; Torraca et al., 2017) and different inflammation responses modulate MG proliferation in the damaged zebrafish retina(Iribarne and Hyde, 2022). The induction of cxcl18b may represent the inflammatory responses of MG after the cone ablation, pointing out the potential link between the inflammatory response and the emergence of NO signaling in MG. Previous studies have demonstrated that iNOS is induced in various tissues by proinflammatory cytokines(Förstermann and Sessa, 2012; Pacher et al., 2007). One of the approaches to test the role of inflammatory responses is to manipulate the levels of inflammatory responses in MG to see the NO production and MG proliferative behaviors. Also, previous studies appreciate the essential role of electrical activity in tissue regeneration(Levin, 2009; Qin et al., 2023); in particular, calcium signaling has been shown to regulate various molecular pathways for liver regeneration, including the hepatocyte growth factor-Met-tyrosine kinase (HGF-Met) transduction pathway(Bedi et al., 2024) and the epidermal growth factor receptor (EGFR) signaling(Kimura et al., 2023). It is interesting to speculate that the abnormal electrical activity of MG in the injured retina may result in an elevated level of intracellular calcium, which activates calmodulin and induces the conformation change of NOS to NO production(Hanson et al., 2018; Jones et al., 2007). A similar mechanism has been proposed in the long-term potentiation of excitatory post-synaptic structure(Grover and Teyler, 1990; Kawamoto et al., 2012; Park, 2018) as well as in the glutamate neurotoxicity model(Ashpole et al., 2013; Ashpole et al., 2012). Thus, the production of NO derived from Nos may be the product of the interplay between the inflammatory responses and the electrical activity in MG after the retina damage.

Limitations of this study

A few limitations of this study should be acknowledged: 1. The current study’s description of the landscape of MG transitional states is based on single-cell transcriptomic data obtained at a single time point (72 hpi) after the cone ablation, which may not provide a complete picture of the state transition of post-injury MG. Future scRNA-seq analysis of MG at multiple post-injury time points is necessary to clarify this issue. 2. While we demonstrated a critical role of NO in injury-induced MG proliferation, the potential contribution of microglia-derived NO was not directly examined. 3. We did not perform direct measurements of NO levels specifically within cxcl18b-defined MG cells, leaving open the question of localized NO signaling.

STAR methods

Key Resources Table

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Chang Chen (changchen@moon.ibp.ac.cn).

Materials availability

All unique reagents generated in this study will be made available from the contact. We may require a completed MTA if there is potential for commercial application.

Data and code availability

This paper does not report new raw scRNA-seq data and original code.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental Model and Subject Details

Single-cell RNA sequencing data analysis

In this study, single-cell RNA sequencing (scRNA-seq) raw data of Müller glia enriched after G/R cone ablation at 72 hpi are from(Krylov et al., 2023), embryonic RPCs at 24 hpf and postembryonic retinal stem cells (RSCs) at 14 dpf are from(Xu et al., 2020). We re-processed the single-cell FASTQ sequencing reads (Novogene) and converted to digital gene expression matrices using the Cell Ranger software (version 3.1.0) provided by 10X Genomics after mapping to the zebrafish GRCz11 (Ensembl release-96) genome assembly. An average of 47,545 mean reads per cell with 1,343 median genes per cell in no ablation control, 63,177 mean reads per cell with 1,556 median genes per cell in G/R cone ablation (lws2_72 hpi), 61,772 mean reads per cell with 2,474 median genes per cell in the embryonic RPCs at 24 hpf, and 49,243 mean reads per cell with 1,181 median genes per cell in the postembryonic retinal stem cells (RSCs) at 14 dpf were obtained.

We then used the cellranger aggr (version 6.0, https://support.10xgenomics.com/single-cell-gene-expression/software/pipelines/latest/using/aggregate) to aggregate the cellranger counts from the two datasets and normalize them to the same sequencing depth. The feature barcode matrix was recalculated and analyzed on the combined data. This resulted in a dataset of 19,706 aggregated cells, with 100% of the no ablation control reads retained and 71.4% of the lws2_3 dpi reads retained. The aggregated matrix for the no ablation control and lws2_72 hpi samples was loaded into the Seurat R package (version 4.3.0, https://satijalab.org/seurat/). Cells with gene expression of more than 200 and less than 4000, lower than 5% mitochondrial content were filtered for further analysis. The filtered data were normalized, scaled, and clustered using principal component analysis (PCA) with a significance threshold of p < 0.001 (FindClusters, resolution = 0.5), and a UMAP was computed using scanpy.tl. umap, resulting in 20 distinct clusters. MG clusters were identified by high expression of marker genes including rlbp1a, fabp7a, slc1a2b, glula, glulb, gfap, and her4.1. Proliferative MG clusters were identified from proliferative cell clusters by high expression of proliferative cell markers including pcna, mki67, gfap, her4.1, and low expression of the CMZ markers including fabp11a, col15α1b, fabp7b, rx2. We further analyzed the MG and proliferative MG clusters to obtain 13 clusters, including no ablation control 5,932 cells and G/R cone ablation 72 hpi 3,999 cells in the UMAP plot (Figure 1-figure supplement 1B). We compared the proportion of cell numbers in each cluster before and after injury and chose clusters with the high proportion of cell numbers in G/R cone ablation 72 hpi (clusters 2, 3, 5, 6, 9, 10, 11, 12, and 13) to further clustered into 10 clusters (Figure 1-figure supplement 1C).

Single-cell RNA sequencing data of embryonic RPCs at 24 hpf and postembryonic RCSs at 14 dpf were generated into 10 and 11 clusters, respectively. Each cluster is identified by the cluster-specific marker gene and different development stage markers including fabp11a, col15α1b, cxcl18b, her4.2, npm1a, her9, fabp7a, dla, atoh7, vsx1, mafba, otx5, and rem1 (Figure 3A, 3C and Figure 3-figure supplement 1A and B).

Pseudo-time trajectory analysis

After the UMAP cluster analysis of the single-cell data, trajectory analysis was performed to investigate the pseudo-time transcriptomic change of these ten clusters using the ‘monocle 2’ R package, which revealed three distinct states (Figure 1G).

Zebrafish husbandry and transgenic fishlines

All experimental zebrafish embryos, larvae, and adults were produced, grown, and maintained at 28°C according to standard protocols. Embryos and larvae were kept in embryo medium (E3; 5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂·2H₂O, 0.33 mM MgSO₄•7 H₂O, 1.3 × 10⁻⁵ % w/v methylene blue in RO water) at 28.5 °C, under a 14:10 light: dark cycle. Animal procedures performed in this study were approved by the Animal Use Committee of the Institute of Neuroscience, Chinese Academy of Sciences (NA-069-2023).

Published fish lines used in this study include the following: AB (Wild type, WT) and transgenic lines (Tg): Tg(opn1lw2: NTR-mCherry)^uom3 (ZDB-ALT-201012-2)(Wang et al., 2020) in this study named Tg(lws2: nfsb-mCherry), Tg(gfap: EGFP)^mi2001 (ZDB-FISH-150901–29307)(Bernardos and Raymond, 2006), Tg(her4.1: dRFP) (ZDB-TGCONSTRCT-070612–2)(Yeo et al., 2007), Tg(mpeg1: GFP) (ZDB-TGCONSTRCT-170801–5)(Ellett et al., 2011), Tg(pcna: GFP) (ZDB-LAB-070129-2)(Xu et al., 2020), Tg(ef1α: loxP-DsRed-loxP-EGFP)(Hans et al., 2009), Tg(Tp1bglob: EGFP)(Yu and He, 2019).

Newly generated transgenic fishline contained: Tg(cxcl18b: GFP), Tg(cxcl18b: Cre-vmhc: mCherry:: ef1α: loxP-DsRed-loxP-EGFP:: lws2: nfsb-mCherry), and mutants nos1^+/-, nos2a^+/-, nos2b^+/-, gsnor^+/-, nos1^-/-, nos2a^-/-, nos2b^-/-, gsnor^-/-.

Generation of Tg(cxcl18b: GFP)

The plasmid of pTol2-cxcl18b: GFP (10 ng/ μl) and tol2 mRNA (50 ng/ μl) were co-injected into a wild-type embryo at the one-cell stage. Zebrafish embryos with green fluorescent were grown at the E3 medium according to standard protocol and selected as the founder (F0). The adult F0 fish crossed with the wild-type selected the fish with GFP⁺ offspring as F1. In this stable transgenic line, cxcl18b⁺ cells expressed the fluorescent protein GFP and the full name of this line is Tg(cxcl18b: GFP).

Generation of Tg(cxcl18b: Cre-vmhc: mCherry:: ef1α: loxP-DsRed-loxP-EGFP:: lws2: nfsb-mCherry)

The plasmid of pTol2-cxcl18b: Cre-vmhc: mCherry (10 ng/ μl) and tol2 mRNA (50 ng/ μl) were co-injected into the embryo obtained from Tg(lws2: nfsb-mCherry) cross Tg(ef1α: loxP-DsRed-loxP-EGFP) at the one-cell stage. Zebrafish embryos were maintained at E3 medium according to standard protocol. Larvae with mCherry expression in the heart and eyes were selected and grown as the founder (F0) for further construction. The adult founder was crossed with wild-type (AB), considering the expression of cxcl18b during zebrafish development, larvae with mCherry and GFP double positive were selected and grown as the F1. In this zebrafish line, EGFP permanently marks all cells that have expressed cxcl18b, as well as their entire lineage of progeny. The full name of this line is Tg(cxcl18b: Cre-vmhc: mCherry:: ef1α: loxP-DsRed-loxP-EGFP:: lws2: nfsb-mCherry).

Generation of mutant fish lines

The CRISPR/Cas9 system was employed to efficiently and precisely generate mutant zebrafish lines(Wu et al., 2018). Two sgRNAs targeting gene coding sequences involved in nitric oxide metabolism signaling (nitric oxide synthase, Nos; and S-nitrosoglutathione reductase, GSNOR) were designed based on the zebrafish GRCz11 genome assembly.

Primers were designed approximately 200 bp from the sgRNA target sites for genotyping (sgRNAs and genotyping primers are listed in Supplementary Table S1). We mixed the two sgRNAs (100 ng/μl) for each target gene with Cas9 protein (400 ng/μl) and injected them into transgenic fishline Tg(lws2: nfsb-mCherry) embryos at the one-cell stage. Adult fish were crossed with wild-type, and offspring displaying mCherry expression in the eyes at 5 dpf were selected for further genotyping. Mutants were confirmed by sequencing, and the fish with open reading frame (ORF) shift were collected as heterozygous fish for further studies.

Plasmid construction

We cloned the cxcl18b regulatory element (3048 bp upstream of the start codon including the 5’UTR) as the cxcl18b promoter from the zebrafish genomic DNA (Figure 2C). The primers used for amplification were 5’-GCATTTGTCTCCTCATGCATTGACTAC-3’ (forward primer) and 5’-TTGCTGCAAACTATATGTAGGAAATGCTG-3’ (reversed primer). For constructing the plasmids pTol2-cxcl18b: GFP, pTol2-cxcl18b: Cre-vmhc: mCherry, and pTol2-cxcl18b: gal4FF, each DNA elements cassette was inserted into the pDestTol2pA2 vector(Kwan et al., 2007). The plasmid of pUAS: Cas9T2AGFP; U6: sgRNA1; U6: sgRNA2 was kindly provided by Professor Filippo Del Bene.(Di Donato et al., 2016). To prepare for plasmid pTol2-UAS: Cas9-T2A-Cre-U6-empty and pTol2-UAS: Cas9-T2A-Cre-U6: nos2b sgRNA1; U6: nos2b sgRNA2, the same two nos2b sgRNAs used to construct nos2b mutant fish were inserted into pUAS: Cas9T2AGFP; U6: sgRNA1; U6: sgRNA2 plasmid, with the GFP element replaced by Cre using ClonExpressMultiS One Step Cloning Kit (Vazyme, C113-02). For the plasmid construction of the MG-specific knockout system based on recombinant adenoviral vectors, regulatory elements were inserted into adenoviral vectors pAdc68-S according to the standard protocol outlined by Jia et al.(Jia et al., 2019). The final plasmids generated included, pAdC68XY2-E1-CMV-GFP pAdC68XY2-E1-10xUAS: Cas9-2a-Cre-U6: sgRNA1(nos2b); U6: sgRNA2(nos2b), pAdC68XY2-E1-cxcl18b: gal4FF and pAdC68XY2-E1-10xUAS: Cas9-2a-Cre-U6: empty as the control. Plasmid construction primer sequences are listed in Supplementary Table S2.

Metronidazole treatment

Zebrafish larvae were exposed to a 10 mM metronidazole (MTZ) solution (Sigma-Aldrich, M3761-100G) in standard fish water. For Tg(lws2: nfsb-mCherry) larvae processed for immunohistochemical analysis, MTZ treatment was initiated at 6 dpf to ablate green and red cones. Larvae were maintained in the MTZ solution at densities of fewer than 50 larvae per 50 mL petri dish and kept at 28.5°C. The MTZ solution was refreshed every 24 hours to ensure continuous cone ablation until fixation in 4% PFA, while control larvae were kept in standard fish water.

Gene disruption via CRISPR/Cas9 system

We used the CRISPR/Cas9 system for efficient gene disruption. Two sgRNAs were designed to target the coding sequences of genes involved in the nitric oxide signaling pathway (including nos1, nos2a, nos2b, gsnor) and cxcl18b. The sgRNAs were in vitro transcribed and purified using the LiCl precipitation approach (MEGAscript T7 Transcription Kit, Invitrogen, AM1334). A mixture of the two sgRNAs (in total 200 ng/μl) with Cas9 protein (400 ng/μl, Novoprotein, E365-01A) were co-injected into the embryo of Tg(lws2: nfsb-mCherry) at the one-cell stage. The efficiency of gene knockout for each sgRNA was validated, as presented in Figure 4-figure supplement 1C.

Pharmacological treatment with NO scavenger and dexamethasone or Nos Inhibitors intravitreal injection

Zebrafish larvae at 6 dpf were anesthetized in 0.04% MS222 (Sigma, A5040) for 30 to 45 seconds. Inhibitor solutions of 2 μL were prepared as follows: 10 mM N(ω)-nitro-L-arginine methyl ester (L-NAME, Sigma-Aldrich, N5751-1G), a broad Nos inhibitor, diluted in sterile PBS from a 100 mM stock; 10 mM N(ω)-Methyl-L-arginine acetate salt (L-NMMA, Sigma-Aldrich, M7033-5MG), a specific inhibitor of nNos, diluted in sterile PBS from a 20 mM stock; and 200 nM 1400W dihydrochloride (SW1400, MedChemExpress, HY-18730), a specific inhibitor of iNos, diluted in sterile PBS from a 39.97 mM stock. PBS was used as the control. Each solution was loaded into glass capillaries prepared using a micropipette puller (Narishige, PC-10) and connected to a microinjector (Applied Scientific Instrumentation, MPPI-3). Intravitreal injections of the inhibitor solutions were administered to the zebrafish eyes starting 2 days before G/R cone ablation and continued until 72 hpi (Figure 4C).

Additionally, zebrafish were pre-treated with 10 mM of the NO scavenger, Phenyl-4,4,5,5-tetramethyl imidazoline-1-oxyl 3-oxide (C-PTIO, Sigma-Aldrich, P5084-25MG), diluted in standard water (from a 100 mM stock), or dexamethasone (Dex, Sigma-Aldrich, D1756) which diluted in DMSO, starting 2 days before G/R cone ablation and continuing until 72 hpi (Figure 2J and 4C).

The real-time quantitative PCR after FACS-sorted MG

We crossed Tg(lws2: nfsb-mCherry) with Tg(pcna: GFP), Tg(her4.1: dRFP), Tg(gfap: EGFP), and Tg(cxcl18b: GFP) to collect retina cell. the retinas were dissected and the cells dissociated for sorting and enrichment of signal-positive cells using fluorescence-activated cell sorting (FACS, Beckman, MoFlo XDP). After 72 hours post G/R cone ablation, GFP⁺/RFP⁺ cells were collected from the retinas of Tg(lws2: nfsb-mCherry x pcna: GFP) as the proliferative MG group. For the uninjured MG group, GFP⁺/RFP⁺ cells were collected from Tg(lws2: nfsb-mCherry x gfap: EGFP x her4.1: dRFP) uninjured retinas, while GFP⁺/RFP⁺ cells from injured retinas represented the 72-hpi MG group, while GFP^-/RFP^- cells were enriched as the other retinal cell types group in the 72 hpi retinas. Additionally, from Tg(lws2: nfsb-mCherry x cxcl18b: GFP) retinas, GFP⁺/RFP⁺ cells were collected as the cxcl18b⁺ MG group and single RFP⁺ cells as the G/R cone group at 72 hpi (Figure 5A).

Interesting cells were collected into a 1.5-ml tube containing lysis buffer (20 mg/ml PK in TE buffer) and total RNA was isolated using this lysis buffer. cDNA synthesis was conducted by adding RT mix contains: 200 U Superscript II reverse transcriptase (Invitrogen, 18064-014), 1 × First-strand buffer (Invitrogen, 18064-014), 5 mM DTT (Invitrogen, 18064-014), 20 U Recombinant RNase inhibitor (Clontech, 2313A), 6 mM MgCl2 (Sigma, M8266), 1 μM TSO(Picelli et al., 2013); 8% PEG8000 (Sigma, P1458). PCR amplification was performed as previously described(Picelli et al., 2013).

Real-time quantitative PCR (RT-qPCR) reactions were carried out using TB Green Premix Ex Taq (Takara, RR420A) on a LightCycler 480 II real-time PCR detection system (Roche). The qPCR primers are listed in Supplementary Table S3. The ΔΔCt method was used to determine the relative expression of mRNAs in different group retinae cells and normalized to actin mRNA levels. Each group comparison was performed using a two-way ANOVA followed by Tukey’s HSD test. Error bars represented SEM. ****p<0.0001, ***p<0.001; **p<0.01; *p<0.05; ns, p>0.05.

Construction and injection of adenoviral-based MG-specific knockout system in the zebrafish retina

The type of AdC68 used in our study is a replication-deficient chimpanzee adenovirus. The plasmids used for viral packaging were described above, and the amplification and purification of recombinant chimpanzee adenovirus followed the standard protocol(Jia et al., 2019; Liu et al., 2021). We generated four adenoviruses for this study, including AdC68XY2-E1-CMV-GFP pAdC68XY2-E1-10xUAS: Cas9-2a-Cre-U6: sgRNA1(nos2b); U6: sgRNA2(nos2b) (referred to as UAS: Cas9-2a-Cre-sgnos2b, infectious titer: 2.00 x 10¹³ ifu/ml), pAdC68XY2-E1-cxcl18b: gal4FF (referred to as cxcl18b: gal4, infectious titer: 1.90 x 10¹³ ifu/ml) and pAdC68XY2-E1-10xUAS: Cas9-2a-Cre-U6: empty (referred to as UAS: Cas9-2a-Cre, infectious titer: 1.75 x 10¹³ ifu/ml) as a control.

All adenoviruses were diluted in sterile PBS to a final infectious titer: 1.00 x 10¹³ ifu/ml. The UAS: Cas9-2a-Cre-sgnos2b and cxcl18b: gal4 adenoviruses were mixed at a 1:1 ratio for the MG-specific knockout system, while the UAS: Cas9-2a-Cre with cxcl18b: gal4 adenoviruses were similarly mixed as a control. To achieve the cxcl18b⁺ MG cell-specific knockout nos2b, we performed the intravitreal injection of the adenovirus mixtures into the eyes of Tg(ef1α: loxP-DsRed-loxP-EGFP x lws2: nfsb-mCherry) at 5 dpf. One day post-injection, the fish were treated with MTZ for 3 consecutive days to ablate the G/R cone (Figure 6E).

Tissue preparation and immunostaining

Zebrafish were fixed in 4% paraformaldehyde (PFA, Electron Microscopy Services, 157–8) overnight, then cryoprotected in 30% sucrose for 6 hours, flash-frozen, and cryosectioned at a thickness of 14 μm. Immunostaining was performed following the protocol described by Tang et al.(Tang et al., 2017). The primary antibody including mouse anti-PCNA (Abcam, ab29) at a 1:500 dilution, as well as mouse anti-GS (glutamine synthase, BD Transduction Laboratories, 610518), rabbit anti-BLBP (Abcam, ab32423), rabbit anti-DsRed (Clontech, 632496), rabbit anti-TaqGFP (Proteintech, 50430-2-AP), and chicken anti-GFP (Abcam, ab13970) each at a 1:1000 dilution. Secondary antibodies conjugated to Alexa Fluor 488, 594, or 647 (Jackson ImmunoResearch Laboratories Inc.) were used at a 1:1000 dilution. Primary antibodies were incubated overnight at 4°C, while Alexa Fluor secondary antibodies were incubated at room temperature for 2 hours. DAPI staining was performed according to the standard protocol.

In situ hybridization

In this study, three digoxigenin (DIG)-labeled RNA probes targeting endogenous cxcl18b, fabp11a, and col15α1b were synthesized using the MEGAscript™ T7 High Yield Transcription Kit (Invitrogen, AM1334) and the DIG RNA Labeling Kit (Roche, 11277073910), following the manufacturer’s instructions. The cDNA of each gene was amplified by PCR using the following primers: cxcl18b -F: 5’-ATGGCATTCACACCCAAAGCG-3’; cxcl18b-R: 5’-TAATACGACTCACTATAGGGATTGGCCCTGCTGTTTTTGTG-3’; fabp11a-F: 5’-GTTGGAAACCGGACCAAACC-3’; fabp11a-R: 5’-TAATACGACTCACTATAGGGACGGCTCGTTGAGCTTGAAT-3’; col15a1b -F: 5’-CCTCAATGGAGGTCCTAAAGGT-3’; col15α1b -R: 5’-TAATACGACTCACTATAGGGACCAGCTTCTGAGACCAAGC-3’.

Following the in situ hybridization protocol described by Tang et al.(Tang et al., 2017), fresh zebrafish retinal sections were incubated overnight in the HybEZ system (Advanced Cell Diagnostics, 310013) at 65°C with 200 ng of probe for each slide. The next day, slides were sequentially washed in 5x SSC buffer and incubated overnight at 4°C with an anti-DIG-POD antibody (Roche, 11093274910) diluted 1:500 in TNB buffer (TN buffer with 0.5% blocking reagent; Roche). On the third day, the signal was detected using the TSA™ Plus Cyanine 3 (PerkinElmer, NEL744001KT) or Cyanine 5 (PerkinElmer, NEL745001KT)/Fluorescein System.

Imaging

Images were taken using an inverted confocal microscope system (FV1200, Olympus) confocal microscope using 40 × (silicon oil, 1.05 NA), or 60 × (silicon oil, 1.3 NA) objectives.

Quantifications and statistical analysis

All quantification and visualization were performed with FV31S-SW 2.3.1.163 Viewer (Olympus), and ImageJ. Z intensity projection was used to process the Z-stack images acquired from 14-μm-thick sections of zebrafish retina for statistical analysis. For cell counting, PCNA⁺ MG and their lineage in this study were defined as one proliferative MG, cxcl18b⁺ MG and their lineage were counted as one cxcl18b⁺ MG.

To perform the statistical analysis, p values were calculated with GraphPad Prism 8 (or Microsoft Excel). The unpaired, non-parametric Wilcoxon test was applied for comparison of two groups. The one-way ANOVA, followed by Tukey’s HSD test was applied for comparison of different groups with one treatment. The two-way ANOVA followed by Tukey’s HSD test was applied for comparison of four groups with two treatments. Error bars represent SEM. ****p<0.0001, ***p<0.001; **p<0.01; *p<0.05; ns, p>0.05.

Supplementary materials

Figure 1-figure supplement 1.

Figure 2-figure supplement 1.

Figure 3-figure supplement 1.

Figure 4-figure supplement 1.

Figure 4-figure supplement 2.

Figure 5-figure supplement 1.

Figure 6-figure supplement 1.

Supplementary Table S1.

Supplementary Table S2.

Supplementary Table S3.

Acknowledgements

This research was supported by grants from the National Key R&D Program of China (2022YFA1303000 to C.C.), the Chinese Academy of Sciences Strategic Priority Research Program B grants (XDB39000000 to C.C.), National Natural Science Foundation of China-Key program project (Grant No. 91849203 to C.C.), the Creative Research Groups of the National Natural Science Foundation of China (32321003 to J.H.), STI2030-Major Projects (2021ZD0204500 to J.H.), the National Key Research and Development Program of China (2020YFA0112700 to J.H.), National Natural Science Foundation of China (32471029 to J.H.).

Additional information

Author Contributions

A.Y., design and experimental investigation, visualization, data analysis, methodology, writing-original draft, writing-review and editing. S.Y., M.D., and D.Z., investigation, methodology. J.H. and C.C., conceptualization, experimental design, data analysis, supervision, funding acquisition, writing-original draft, project administration, writing-review and editing.