Immune regulation of neuronal stem cells in the medaka retina

ABSTRACT

Stem cell populations in tissues require precise regulation of their number and quality to maintain proper organ growth. Among the various regulatory mechanisms, immune cells are emerging to directly regulate stem cell populations. The medaka retinal stem cell (RSC) niche, a model for lifelong neurogenic growth, provides a system for studying immune-stem cell interactions. We investigate how microglia, which are resident macrophages of the central nervous system, regulate the RSC niche. We identify that bona fide RSCs express the chemokine Ccl25b, while its cognate receptor, Ccr9a, is expressed in microglia. These microglia form a surveillance ring adjacent to the RSC niche and actively phagocytose RSCs. Interference with microglia by deletion of spi1b reveals that microglia absence leads to increased numbers of ccl25b-positive RSCs and results in morphological defects of the retina. Targeted mutation of ccl25b specifically affects microglia mobility under injury conditions; however, we did not observe any morphological defects, indicating that Ccl25b-Ccr9a signaling is not essential for stem cell maintenance. Overall, our data show that, under homeostatic conditions, the individual RSCs, which are essential for proper eye development, are actively phagocytosed by immune surveillance.

Highlighted Article:  Regulation of retinal stem cells by microglia to maintain the stem cell niche and coordinate homeostatic tissue growth in the fish retina.

INTRODUCTION

Stem cell reservoirs are essential for growth, homeostasis and tissue regeneration in multicellular organisms (Hollyfield, 1971; Johns, 1977; Johns and Easter, 1977). These long-lived cells continuously produce new stem and progenitor cells while resisting cell death (Soteriou and Fuchs, 2018). Their quantity and quality must be precisely regulated to prevent tissue overgrowth and cellular deterioration.

The retinal stem cell (RSC) niche of the teleost medaka (Oryzias latipes) provides a well-established model for studying neuronal stem cells (Centanin et al., 2011) due to the short generation time of medaka, the ease with which they can be genetically manipulated and their vast landscape of available transgenic reporter lines. Located in the ciliary marginal zone (CMZ) in a ring around the distal edge of the optic cup, multipotent stem cells sustain lifelong retinal growth (Centanin et al., 2014; Reinhardt et al., 2015; Soteriou and Fuchs, 2018). A population of neuroretinal progenitor cells, located between the stem cell ring and the stratified retina, contains transiently amplifying cells that contribute to both the inner and outer nuclear layers (Tsingos et al., 2019). This distinctive architecture allows reliable identification of multipotent stem cells through lineage-tracing experiments (Centanin et al., 2014, 2011).

Recent evidence suggests that immune cells, particularly tissue-resident macrophages and microglia, extend beyond their classical immune functions to contribute to tissue homeostasis, stem cell regulation and regeneration (Cunningham et al., 2013; de Almeida et al., 2023; Nath et al., 2024; Ratnayake et al., 2021). In the central nervous system (CNS), microglia are unique among retinal cells as they originate outside the RSC niche, entering the CNS at embryonic stages (E11.5 in mice, 25 h post fertilization in zebrafish) (Herbomel et al., 2001; Santos et al., 2008).

Traditionally, microglia are known primarily for efferocytosis – the removal of apoptotic cells (Ravichandran, 2011) – and thus contribute to shaping developing tissues. Recently, a ‘groom or doom’ interaction was reported for the hematopoietic stem cell niche of zebrafish and mice during embryogenesis: local macrophages were observed to undergo lengthy interactions with stem cells preceding cell division. These interactions often resulted in either the partial, or complete, phagocytosis of stem cells that did not display markers for apoptosis (Wattrus et al., 2022).

Building on our understanding of the RSC niche organization, we addressed if and how microglia regulate retinal stem cell numbers and contribute to coordinated retinal growth in medaka. Using a newly established retinal stem cell reporter line, we characterize the interactions between RSCs expressing the chemokine ccl25b and microglia expressing its cognate receptor ccr9a.

Unlike the previously established role of Ccl25b-Ccr9a signaling in the recruitment of lymphocytes in the intestine in mammals and fish (Aghaallaei et al., 2022; Bajoghli, 2013; Uehara et al., 2002), we report a previously unreported function of this chemokine-receptor interaction in the retina. Furthermore, we examine how the complete absence of microglia affects retinal development and growth, providing insights into immune cell-mediated regulation of the retinal stem cell pool.

RESULTS

ccl25b chemokine and ccr9a receptor expression in the medaka retina

RSCs are specifically labelled in a stable transgenic line where a ccl25b regulatory element drives GFP reporter expression (ccl25b::GFP). This line was initially generated to follow the dynamics of cells expressing ccl25b, a chemokine that is known to be involved in the recruitment of lymphocytes in the intestine (Aghaallaei et al., 2022; Campbell and Butcher, 2002; Sigmundsdottir and Butcher, 2008) and thymus (Baubak Bajoghli et al., 2009). We detected the expression of ccl25b chemokine at the periphery of the ciliary marginal zone (CMZ), partially overlapping with previously described rx2-expressing retinal stem and progenitor cells (Fig. 1A) (Reinhardt et al., 2015). The endogenous expression of ccl25b transcripts in the RSC zone was validated by in situ hybridization chain reaction (HCR) (Fig. S1B). To test whether ccl25b-expressing cells indeed represent multipotent retinal stem cells, we used GaudíRSG, an established Cre-LoxP based lineage tracing tool (Centanin et al., 2014). Here, Cre induced recombination will lead to the expression of GFP after the Cre mediated excision of a default DsRed/Stop cassette. We employed the Ccl25b regulatory element mentioned above (Aghaallaei et al., 2022) to specifically drive the expression of a hormone-inducible Cre recombinase in ccl25b-expressing cells and established a stable transgenic line in the background of the GaudíRSG line. We generated a tamoxifen-inducible ccl25b::Cre^ERT2,GaudíRSG transgenic line and clonally activated color switches in individual ccl25b-positive cells, specifically labelling those and all of their descendants. The long-term lineage of ccl25b-expressing cells was analyzed in fish 6 weeks post-Cre-mediated recombination at hatching stage; GFP-expressing clones originating from ccl25b-expressing cells in the CMZ were analyzed via whole-mount immunostaining. Lineage analysis revealed that ccl25b-positive cells established induced arched continuous stripes, iArCoSs (Centanin et al., 2014), which are clonal strings originating in the stem cell niche of the CMZ differentiated into all neuroretinal cell types and thus represent neuroretinal stem cells (Fig. 1B). Early induction for 3 days post hatching was also evaluated to validate the clonal capacity of these ccl25b-expressing cells in a short period. As expected, few GFP-positive cells were observed in the CMZ for such a short induction time wherein the clone partially overlaps with early retinal progenitor cells expressing rx2 and eventually produces late progenitors (Fig. S1E,E′). The proliferative capacity of ccl25b-expressing RSCs was evaluated with BrdU labelling, showing that these cells are slowly proliferating as they did not incorporate BrdU when exposed for 1 day unlike the adjacent fast-cycling progenitor cells, which show BrdU incorporation (Fig. S1D,D′). ccl25b-based iArCoSs were exclusively detected in the neuroretina. This validates ccl25b-expressing cells as bona fide multipotent neuro-retinal stem cells residing in the peripheral tip of the CMZ.

Fig. 1.

Characterization of ccl25b as a retinal stem cell marker and ccr9a as marker for microglia in the retina. (A) Transverse section of a retina derived from a ccl25b::GFP reporter hatchling. GFP-positive cells (open arrowheads) are located in the most peripheral cells of the ciliary marginal zone (CMZ) and the Rx2-positive (magenta) domain, cell nuclei are stained with DAPI (gray). Scale bar: 20 μm. (B) Cre/loxP-mediated lineage tracing of ccl25b-positive cells. Genetic construct, induction scheme and whole-mount ccl25b::Cre^ERT2 X GaudíRSG retina, where recombination (red to green) was induced at hatching stage and analyzed 6 weeks later. GFP-positive clones (iArCoS) (yellow arrowheads) can be detected in the neural retina (NR) and the ciliary body (CB). Scale bar: 200 μm. The GFP-positive iArCoS is shown in more detail on the right. There are GFP⁺ cells in all neuroretinal layers (white arrowheads) in ccl25b::Cre^ERT2 × GaudíRSG retinae. Scale bar: 20 μm. (C) Transverse section of a ccr9a::GFP hatchling retina. GFP-positive cells are present in proximity to the CMZ region and in the inner nuclear layer (INL) region. Lcp1 staining overlaps with ccr9a::GFP-positive cells (white arrowheads). Cell nuclei are stained with DAPI (gray). Scale bar: 20 μm. (D) Co-expression of the macrophage reporter cxcr3a::GFP with ccr9a::mCherry in embryonic stage retinae (stage 28), as observed by in vivo confocal imaging. White arrowheads indicate co-expressing cells. Yellow dashed lines indicate the retina. Scale bars: 20 μm.

We complemented our analysis by probing for the presence and localization of cells expressing ccr9a, the cognate receptor of ccl25b in the retina of young fish at the hatching stage. We used an established transgenic line, ccr9a::GFP (Bajoghli et al., 2015), and detected GFP expressing cells around the CMZ and in the inner nuclear layer (INL) region of the retina (Fig. 1C). The endogenous expression of ccr9a in the retina was confirmed by HCR in situ staining (Fig. S1C).

ccr9a-expressing cells were characterized as immune cells using immunostaining for lymphocyte cytosolic protein 1 (Lcp1) (Jones et al., 1998). Among all Lcp1-positive cells analyzed (n=316 cells in 56 sections from 4 fish), 97% co-expressed the chemokine receptor ccr9a (Fig. 1C). ccr9a-positive cells tagged with mCherry fluorophore were further validated as immune cells by the co-expression of cxcr3a::GFP, a well-established macrophage-specific marker (Aghaallaei et al., 2010) (Fig. 1D). Under homeostatic conditions, the only immune cells known to colonize the immune-privileged retina are primordial macrophages, which give rise to microglia (Herbomel et al., 1999, 2001).

Retinal stem cells are phagocytosed by tissue resident microglia

To address if there are microglia-stem cell interaction in the retina, we first carried out a detailed analysis of the spatial arrangement of retinal stem cells (expressing ccl25b) and microglia expressing the cognate receptor ccr9a or cxcr3a in the retina of freshly hatched medaka larvae. We carefully analyzed 3D reconstructions of retinae of double-positive ccl25b::H2B-RFP; ccr9a::GFP reporter fish. While RSCs resided in a ring in the very periphery of the CMZ (open arrowheads in Fig. 2A, 3D schematic), microglia formed a concentric double ring at the proximal edge of the CMZ close to inner plexiform layer (IPL) and outer plexiform layer (OPL). Each ring was composed of evenly spaced ccr9a- or cxcr3a-positive microglia contacting each other (Figs 2A and 3D schematic, Movie 1).

Fig. 2.

ccl25b-expressing retinal stem cells are phagocytosed by ccr9a⁺ microglia. (A) (Left) Transverse section of a ccl25b::H2B-RFP;ccr9a::GFP reporter retina of a hatchling. ccl25b-driven H2B-RFP expression (open arrowhead) is located in the peripheral ciliary marginal zone (CMZ) (dashed line). ccr9a-driven GFP expression can be detected at the edge of the proximal end of the CMZ, as well as in the inner nuclear layer (INL) of the retina (asterisk). Scale bar: 20 μm. (Right) 3D schematic representation of a retina illustrating the distribution of ccl25b-expressing retinal stem cells (RSCs; magenta) and ccr9a-expressing microglia cells (green). (B) Retinal stem cell region showing the interaction between H2B-RFP-expressing ccl25b⁺ (magenta) and GFP-expressing cxcr3a-positive (green) cells, detected by antibody staining against RFP and GFP in whole-mount retinae. Arrowhead indicates a phagocytic cup in the process of forming. Scale bars: 10 μm. (B′) Schematic of the formation of a phagocytic cup by microglia to engulf RSC. (C) Whole-mount staining showing the CMZ and retinal layers expressing H2B-RFP in ccl25b-positive RSCs and GFP expression in ccr9a⁺ microglia cells. Open white arrowheads indicate ccr9a-expressing cells enclosing H2B-RFP fractions (magenta) within phagosomes. Solid white arrowheads indicate ccr9a⁺ microglia without H2B-RFP fractions. Scale bar: 20 μm. (D) Time lapse of a ccl25b::H2B-RFP (magenta) and ccr9a::GFP (green) retina depicting microglia-transporting phagosomes within the CMZ (open arrowheads) and a static immune cell with a phagosome (asterisks). Scale bar: 10 μm.

Fig. 3.

The chemokine Ccl25b is necessary for activity of ccr9a-positive immune cells in response to injury. (A) Representative images of phagosomes in wild-type and ccl25b^−/− retinae of ccl25b:H2B-RFP reporter fish. Scale bars: 10 μm. (A′) Comparison of the percentage of ccr9a-positive immune cells with H2B-RFP-positive phagosomes in wild-type (n=8) and ccl25b^−/− (n=6) retinae (**P=0.004662; Mann–Whitney U-test). (B) Quantification of ccl25b⁺ retinal stem cells in wild-type (n=11) and ccl25b^−/− (n=17) (P=0.4515; Mann–Whitney U-test). (C) Representative tracks with temporal color-coding representing immune cell migration in ccr9a::GFP hatchling retinae in wild-type (left; n=44 cells from five fish) and ccl25b^−/− (right; n=47 cells from seven fish). Scale bars: 20 μm. (D) Quantification of ccr9a-positive microglia migration distance in the retinae of ccl25b^−/− versus wild-type hatchlings (***P=0.0008512; Mann–Whitney U-test). (E) Quantification of average migration velocity of microglia in the retinae of wild type compared to ccl25b^−/− hatchlings (***P=0.0001817; Mann–Whitney U-test). (F) Analysis of mean square displacement (MSD) as a function of time from microglia cell tracking data. MSD of ccl25b^−/− compared to MSD of microglia cells in wild-type retinae. Paler shading indicates the standard error of the mean (s.e.m.).

An important feature of immune cells, particularly macrophages, is their capacity to phagocytose debris, damaged cells and foreign particles (Arandjelovic and Ravichandran, 2015; Nakayama, 2018). To address whether microglia phagocytose ccl25b-positive stem cells, we took advantage of the extended stability of red-fluorescent-protein (RFP)-tagged-H2B in the phagosomal compartment (Villani et al., 2019). We used a double transgenic line specifically labelling the nuclei of RSCs with histone-coupled RFP (ccl25b::H2B-RFP) and the microglia with GFP (ccr9a::GFP) (Fig. 2A). We found that GFP-labelled microglia migrate towards the RSC niche and come into direct contact with the ccl25b-positive RSCs forming a phagocytic cup (Movie 3, Fig. 2B,B′). Microglia eventually contained RFP-positive phagosomes (Fig. 2C) and quantification of the phagocytosis of stem cells was carried out by counting the number of microglia containing stem cell-derived RFP positive phagosomes. At hatching stage, we detected an active uptake by phagocytosis of stem cells in 61% of the ccr9a-positive microglia (Fig. 3A′).

We investigated the dynamics of interactions between RSCs and microglia, as well as their phagocytic activity using two-photon in vivo imaging at hatching stage. This analysis revealed microglia actively transporting stem cell-derived, RFP-positive, phagosomes within the ciliary marginal zone (CMZ) (Fig. 2D, Movie 2). We observed microglia containing H2B-RFP-positive nuclei within their phagosomes, indicative of the direct uptake of stem cell nuclei (Fig. 2D, asterisk). We next addressed whether the phagocytosed stem cells represent cells being cleared through classical efferocytosis of apoptotic cells. Strikingly, TUNEL staining did not detect any apoptotic cells among RSCs (Fig. S2C). This indicates that the detected prominent phagocytosis of RSCs is independent of apoptosis.

Activity of microglia is dependent on ccl25b chemokine

The RSC specific expression of the chemokine ligand ccl25b, combined with our finding of ccr9a-positive microglia-mediated stem cell phagocytosis, suggests that the ligand provides localized guidance cues that direct the activity of ccr9a-positive microglia. To address the role of Ccl25b in the microglia mediated phagocytosis of RSCs, we generated a ccl25b^−/− mutant allele by targeted mutagenesis using CRISPR/Cas9. This allowed us to test if the absence of Ccl25b affected the interaction between RSCs and immune cells. Targeted mutagenesis was designed to delete the majority of the ccl25b gene and the mutant transcript contained only 22% of the wild-type coding sequence. The mutated gene was lacking the CC motif crucial for folding and function of CC class chemokines (Thomas et al., 2015) (Fig. S2A).

The effect of loss of ccl25b functionality on phagocytosis, was assessed in ccl25b^−/− mutants in the background of the ccl25b::H2B-RFP, ccr9a::GFP double reporter line. We examined the number of immune cells containing RSC-derived phagosomes in homeostatic conditions. We found that, in wild-type fish, RSC-derived phagosomes were present in 61% of ccr9a-positive cells (n=201/331 cells in eight retinae). In ccl25b^−/− mutants, the number of ccr9a-positive microglia containing RSC-derived H2B-RFP-positive phagosomes was reduced to 36% (Fig. 3A′) (n=113/317 cells in six retinae). Moreover, there was an increase in the number of ccr9a-positive microglia observed in the ccl25b^−/− fish. Strikingly, the reduction in phagocytosis and the increase in the number of microglia in the absence of the Ccl25b chemokine did not affect the number of RSCs expressing ccl25b::H2B-RFP (Fig. S2B, Fig. 3B). The comparison of transverse sections of 3-week post-hatch retinae between wild type and ccl25b^−/− showed no gross morphological differences (Fig. S2D,D′).

Since chemokines are known to have a key function in acute injury responses, by attracting immune cells towards the site of inflammation (Sokol and Luster, 2015), we addressed whether Ccl25b is required to direct the migration of ccr9a-positive microglia within the retina to the stem cell domain upon injury. We used two-photon-mediated RSC ablation to inflict a wound to the stem cell niche in the retina at hatching stage, as has been described previously (Lust and Wittbrodt, 2018). We recorded the response of immune cells to the injury over 4 h, with a temporal resolution of 2 min and analyzed their speed and directionality. In wild-type retinae, ccr9a-positive microglia migrated rapidly to the injured area (Fig. 3C-F) while in ccl25b^−/− retinae, the migratory behavior of ccr9a-positive microglia was significantly reduced and lacked directionality. In general, they displayed lower average migration velocity and travelled shorter total distances in the absence of Ccl25b (Fig. 3D-F). These results indicate the crucial role of the Ccl25b-Ccr9a signaling axis in giving direction and specificity to microglia under inflammatory conditions and impacting on their ability to phagocytose stem cells without affecting the overall stem cell number under homeostatic condition.

Absence of microglia affects retinal stem cell numbers

Because phagocytosis was only modestly reduced in the absence of Ccl25b, we next asked whether microglia directly control RSC homeostasis. To eliminate microglia, we targeted two transcription factors required for their differentiation: spi1b and irf8. While Spi1b is required for the development of both, lymphoid and myeloid lineages (Rhodes et al., 2005; Roh-Johnson et al., 2017; Scott et al., 1994), Irf8 is specifically involved during macrophage differentiation (Shiau et al., 2015).

Using CRISPR/Cas9-mediated mutagenesis, we generated a stable spi1b^−/− mutant line and established and analyzed irf8 crispants. The spi1b mutant carries a 6 bp insertion and 10 bp deletion in the 5th exon of spi1b, resulting in a premature stop codon (Fig. S3A,A′), truncating the C-terminus of the Spi1b protein. Loss of Spi1b resulted in the complete absence of macrophages, as validated by Sudan Black and Neutral Red staining. In spi1b^−/− homozygous mutants, Sudan Black-stained blue-black cells were absent from the tail fin (Fig. S3B,B′). Neutral Red staining, which specifically labels lysosome-rich cells such as macrophages, was observed in the tail fin of wild-type but not the spi1b^−/− fish (Fig. S3C,C′). Additionally, in situ hybridization with a macrophage-specific marker, cxcr3a, labelled macrophages in the brain, gills, intestine and skin of control wild-type hatchlings but produced no signal in spi1b^−/− fish (Fig. S3D,D′). These results validate the complete absence of macrophages upon loss of Spi1b function. The loss of spi1b is lethal at juvenile stages (3 weeks post hatching) and the overall morphology of the retina is disrupted by this stage (Fig. S3F).

To determine whether microglia regulate the RSC niche in the CMZ, we introduced the ccl25b::H2B-RFP reporter into the spi1b^+/− mutant background. Loss of microglia resulted in a pronounced expansion of the RSC population already at the hatching stage, reflected by an increased width of the ccl25b-positive CMZ (Fig. 4A,A′). Quantification across the entire CMZ revealed a 34.3% increase in RSC number in the absence of microglia (spi1b^−/−, n=13 retinae; wild-type siblings, n=51 retinae) (Fig. 4A″).

Fig. 4.

Ccl25b retinal stem cells are affected in absence of microglia. (A,A′) Sagittal view of retinae of ccl25b::H2B-RFP reporter hatchlings in wild-type and spi1b^−/− backgrounds. H2B-RFP was detected by antibody staining of whole retina (magenta). Scale bars: 50 μm. (A″) Comparison of ccl25b-expressing retinal stem cells in wild-type hatchings (n=51 fish) and spi1b^−/− hatchlings (n=13 fish) (***P=0.0000908; Mann–Whitney U-test). (B,B′) Sagittal view of retinae of wild-type and irf8 crispant ccl25b::H2B-RFP reporter hatchlings. H2B-RFP was detected by antibody staining of whole retina (magenta). Scale bars: 50 μm. (B″) Comparison of ccl25b-expressing retinal stem cells in wild-type hatchings (n=16 fish, 23 retinae) and irf8 crispant hatchlings (n=10 fish, 19 retinae) (*P=0.02878; Mann–Whitney U-test).

To distinguish microglia-specific effects from the potentially broader immune defects caused by the loss of spi1b, we independently depleted microglia by CRISPR/Cas9-mediated targeting of the microglia-specific transcription factor irf8. irf8 crispants phenocopied the spi1b mutants and displayed an 18.1% increase in the number of RSCs (irf8 crispant, n=19 retinae; wild-type siblings, n=23 retinae) (Fig. 4B″). This demonstrates that the expansion of the stem cell pool is a direct consequence of the loss of microglia.

The enlarged RSC compartment persisted throughout larval development until the death of the spi1b^−/− mutants, presented here in an exemplary section of the CMZ in a young fish 5 days post-hatching (Fig. 5A,A′). At the tissue level, mutant larvae displayed smaller, frequently asymmetric retinae, while the overall body length was only mildly affected (Fig. 5B, Fig. S3E).

Fig. 5.

The sizes of the ciliary marginal zone and retina are affected in the absence of microglia. (A,A′) Transverse section of wild-type and spi1b^−/− retina stained with DAPI (gray) at 5 days post-hatching. The ciliary marginal zone (CMZ) region is outlined and magnified on the right, showing nuclei organization. Scale bars: 50 μm. (B) Dorsal view of wild-type (top) and spi1b^−/− (bottom) hatchlings to visualize differences in the size of retina. Scale bars: 1 mm. (B′) Quantification of size of the retina by area normalized to the length of the fish to determine any differences between wild-type (n=5), spi1b^+/− (n=20) and spi1b^−/− (n=15). The retina sizes are grouped into small and big retina for every hatchling (**P=0.008900154 for wild-type small retina versus spi1b^−/− small retina, *P=0.013824911 for spi1b^+/− small retina versus spi1b^−/− small retina; Dunn's test with Holm-Bonferroni correction).

Thus, despite the expansion of the retinal stem cell pool, retinal growth is impaired in the absence of microglia, indicating that microglial uptake regulate retinal stem cell numbers to ensure proper tissue architecture under homeostatic growth conditions.

DISCUSSION

Our findings identify microglia as key regulators of retinal stem cell (RSC) homeostasis in the medaka retina. The life-long activity of the RSC niche, marked by the expression of ccl25b, requires continuous extrinsic regulation by microglia through an apoptosis-independent mechanism. Loss of microglial surveillance in spi1b^−/− mutants and irf8 crispants results in pronounced expansion of the RSC population and consequently an enlarged stem cell niche (CMZ). Surprisingly, this increase in stem cell number is accompanied by a reduction in overall growth during post-embryonic development. These opposing phenotypes suggest that stem cell abundance alone is not sufficient to sustain coordinated tissue growth. Instead, excessive RSC density may impair the balance between self-renewal and differentiation ultimately compromising retinal growth. We therefore hypothesize that microglia maintain an optimal RSC population size that preserves the functional output of the stem cell niche. Consistent with this notion, the uneven expansion of RSCs around the mutant CMZ leads to anisotropic retinal growth, demonstrating that local regulation of stem cell abundance directly translates into altered tissue morphogenesis within the same individual.

Although spi1b broadly controls the differentiation of myeloid and lymphoid lineages, whereas irf8 is more specifically required for macrophage and microglial differentiation, disruption of either gene resulted in a comparable expansion of the RSC population. Since the molecular mechanisms governing microglial differentiation are heterogenous and can vary between different developmental stages, we focused a large part of the analysis on our established stable spi1b mutant line. This strategy also avoided reliance on a single microglial differentiation pathway that may affect only a subset of microglia and therefore provides greater confidence that the observed phenotype reflects the consequences of robust and sustained microglial depletion rather than defects in a specific microglial subpopulation.

While the Ccl25-Ccr9 signaling pathway has previously been associated with antigen presentation in the intestine and T cell development in the thymus (Aghaallaei et al., 2022; Baubak Bajoghli et al., 2009), our observations demonstrate an additional function. We show that microglia, an innate immune cell that has not been previously linked to this pathway, express the ccr9a receptor and are involved in direct stem cell interaction. This highlights the tissue-specific role of Ccl25-Ccr9 signaling and its ability to influence both myeloid and lymphoid cell types differently.

Loss of Ccl25b specifically impairs microglial migration towards the site of injury in the CMZ and disrupts homeostatic phagocytosis of retinal stem cells. Although chemokines are known to facilitate immune cell homing during development (Wurbel et al., 2007, 2001), the absence of ccl25b in mutant retinae led to an increase in the number of ccr9a microglia present at the interface between CMZ and central retina, indicating that the presence of Ccl25b chemokine ensures proper number of ccr9a microglia in the retina. While microglia actively express the ccl25b receptor ccr9a, and an inflammatory response after injury is clearly affected without Ccl25b, this pathway is not essential for regulating stem cell numbers. Even though there is an increase in retinal microglia, the reduction in phagocytic capacity argues for the existence of an alternative mechanism of stem cell-microglia interaction for homeostatic stem cell maintenance.

While the knowledge about interaction between immune and stem cells has been rather limited, a similar regulatory mechanism by immune cells has recently been described for the hematopoietic stem cell niche (HSC) in zebrafish (Wattrus et al., 2022). The underlying mechanisms functioning in neuronal stem cells and microglia interaction in the medaka retina remain to be explored.

The teleost retina exhibits continuous, highly organized growth that is dependent on tightly regulated stem cell activity; we demonstrate that microglia maintain this balance by phagocytosing RSCs. A fundamental question is whether this regulation serves as a quality control mechanism where specific cells display markers targeting them for phagocytosis, or if the process is stochastic, based on physical constraints within the CMZ similar to other developmental processes where cell elimination occurs through mechanical or signaling mechanisms (Huels et al., 2018; Levayer et al., 2016). This newly discovered role of the immune system opens exciting possibilities for understanding its involvement in the transition to and maintenance of multicellularity during development, providing an excellent foundation for exploring key mechanisms in stem cell niche regulation.

MATERIALS AND METHODS

Fish husbandry and transgenic lines

Medaka (Oryzias latipes) were kept as closed stocks at Heidelberg University. All experimental procedures and husbandry were performed in accordance with the German animal welfare law and approved by the local government (Tierschutzgesetz §11, Abs. 1, Nr. 1, husbandry permit AZ 35–9185.64/BH and line generation permit AZ 359185.81/G-145-15). Fish were maintained in a constant recirculating system at 28°C on a 14 h light/10 h dark cycle. The following previously described stocks and transgenic lines were used for this study: wild-type Cab, rx2::H2B-RFP (Reinhardt et al., 2015), ccr9a::GFP (Bajoghli et al., 2015), ccl25b::GFP, ccr9a:mCherry, ccl25b::H2B-RFP (Aghaallaei et al., 2022), GaudíRSG (Centanin et al., 2014) and cxcr3a::GFP (Aghaallaei et al., 2010).

The following lines were generated for this study: ccl25b^−/−, ccl25b::Cre^ERT2 and spi1b^−/−. Transgenic lines were created by microinjection with Meganuclease (I-SceI) in medaka embryos at the one-cell stage, as previously described (Thermes et al., 2002).

CRISPR/Cas9-mediated mutation of the ccl25b, spi1b and irf8 locus

For generating truncated variants of ccl25b and spi1b, we used CRISPR/Cas9-mediated mutation. The medaka ccl25b gene (Gene ID: ENSORLG00000028881, Exon ID: ENSORLE00000270681) comprises six exons. We selected two sgRNAs targeting the CC motif in exon 4. The following target sites were used (PAM in brackets): GCGCCAATCCAGAGGACCCG(TGG) and CTTGGCTACGTGAGAGAACT(CGG). For mutation of Spi1b, we used one sgRNA targeting exon 5 of the spi1b gene: GAGGCCCTCGCACACCGCTGGG (Gene ID: ENSORLG00000006090, Exon ID: ENSORLE00000068479). For creating irf8 crispants (Gene ID: ENSORLG00000018216.2, Exon ID: id73007.1) a sgRNA targeting exon 2 of the irf8 gene was used. All sgRNAs were designed using the CCTop prediction tool (Stemmer et al., 2015). Cloning of sgRNA templates and in vitro transcription was performed as described previously. The pCS2+Cas9 plasmid was linearized using NotI and the mRNA was transcribed in vitro using the mMessage mMachine SP6 kit (ThermoFisher Scientific, AM1340). The CRISPR generated edits were mapped and analyzed using geneious software version 2019.2.3 (www.geneious.com).

One-cell stage medaka were injected with 150 ng/μl of Cas9 mRNA and 15 ng/μl of the sgRNAs. Injected embryos were maintained at 28°C in embryo rearing medium (ERM; 17 mM NaCl 40 mM KCl, 0.27 mM CaCl₂, 0.66 mM MgSO₄ and 17 mM Hepes). One day post injection (dpi), embryos were screened for survival.

Genotyping and mutant selection

For genotyping, either injected F0 embryos or fin clips of juvenile or adult fish were lysed in DNA extraction buffer [0.4 M Tris/HCl (pH 8.0), 0.15 M NaCl, 0.1% SDS, 5 mM EDTA (pH 8.0) and 1 mg/ml proteinase K] at 60°C overnight. Proteinase K was inactivated at 95°C for 10 min and the solution was diluted 1:2 with H₂O.

The PCR amplification was performed with Q5 polymerase (NEB) according to the manufacturer's protocol using 30 PCR cycles. The resulting PCR product was analyzed on a 1% agarose gel. F0 animals were raised to adulthood and outcrossed to wild-type Cab fish. F1 fish were then raised to adulthood, finclipped and genotyped. Mutants carrying the same allele were incrossed to obtain homozygous mutants.

To identify mutations in the locus, fin clip DNA of F1 fish was analyzed as described above. Genotyping of ccl25b mutants revealed a 333 bp deletion in the ccl25b locus, completely eliminating exon 4, including the CC motif along with two-thirds of exon 5 (Fig. S2). Transcript analysis of wild-type and ccl25b^−/− fish showed that the resulting truncated transcript contained only 22% of the wild-type coding sequence. We selected spi1b^−/− mutants carry a 6 bp insertion and 10 bp deletion in exon 5, which leads to a frame shift and premature stop codon. In case of irf8 crispants, fin clip DNA of F0 hatchlings was used and sent for sequencing.

Tamoxifen induction of Cre/lox recombination in ccl25b-expressing cells

For Cre^ERT2-induced recombination of ccl25b-expressing cells, ccl25b::Cre^ERT2×GaudiRSG (Centanin et al., 2014) hatchlings (0 day post hatch) were treated with 5 μM tamoxifen solution (Sigma-Aldrich) in 1× ERM overnight. Fish were then grown for 3 days or 6 weeks before they were sacrificed for whole-mount immunostaining of retinae, as described below.

Immunohistochemistry on whole-mount retinae

ccl25b::Cre^ERT2×GaudiRSG fish grown for 6 weeks after induction of recombination were euthanized with 20× Tricaine [stock: 4 g/l triciane and 10 g/l Na₂HPO₄·2 H₂O (pH 7-7.5)], fixed overnight in 4% PFA, and treated and stained as described by Perez Saturnino et al. (2018).

To detect GFP after Cre-induced recombination in ccl25b-expressing cells, we used chicken anti-eGFP (1:500, Life Technologies, A10262), donkey anti-chicken Alexa Flour 488 (1:750, Jackson, 703-485-155) and DAPI (2 mg/ml) (1:500). Retinae of cxcr3a::GFP;ccl25b::H2B-RFP reporter hatchlings were treated and stained as described by Perez Saturnino et al. (2018), except bleaching time was reduced to 1 h.

For the detection of GFP, we used chicken anti-eGFP (1:500, Life Technologies, A10262) and donkey anti-chicken Alexa Flour 488 (1:750, Jackson, 703-485-155). For detection of RFP, we used rabbit anti-RFP (1:500, Clontech 632496) as a primary antibody and goat anti-rabbit DyLight549 (1:750, Jackson, 112-505-144) or goat anti-rabbit Alexa Fluor 647 (1:250, Thermo LifeTech, A21245) as a secondary antibody. To visualize nuclei organization, retinae of spi1b^+/− and spi1b^−/− hatchlings were stained using DAPI (1:500) at 5 days post-hatching.

BrdU incorporation assay

To follow cell division in ccl25b-positive cells, we incubated a ccl25b::GFP reporter hatchling in 2.5 mM BrdU diluted in 1× Embryo Rearing Medium (ERM) for 1 day at 28°C. After removal of BrdU, fish were euthanized, fixed and treated for cryo-sectioning.

Immunohistochemistry on cryosections

To obtain cellular resolution of the CMZ and surrounding cell layers, immunohistochemistry was performed on hatchlings, as described previously (Lust and Wittbrodt, 2018). For the detection of GFP in ccr9a:GFP, we used chicken anti-eGFP (Life Technologies, A10262) and donkey anti-chicken Alexa Flour 488 (Jackson, 703-485-155). For the detection of BrdU, cryosections were fixed in 4% PFA (paraformaldehyde; pH 7) for 30 min after the incubation with secondary antibody and DAPI. After fixation, sections were washed three times for 10 min each with 1× 0.05% Tween 20 in 1× PBS (PTW). The sections were then treated with 2 N HCl, 0.5% Triton X in 1× phosphate-buffered saline (PBS) for 60 min at 37°C for antigen retrieval, washed again three times for 10 min each with 1× PTW and then treated with Borax/PTW for 15 min for pH recovery. After washing three times for 10 min each, sections were blocked with 10% normal goat serum (NGS) in 1× PTW for 2 h, then washed twice for 5 min each with 1× PTW before applying the primary antibody [1:100 or 1:200 BrdU (Abcam, ab6326) in 1% NGS] overnight at 4°C. The next day, the sections were washed six times for 10 min with 1× PTW before secondary antibody was applied [1:750 anti-rat DyLight488 (Jackson, 112-485-143) in 1% NGS] for 2 h at 37°C and sections were washed again with 1× PTW before mounting as described previously (Lust and Wittbrodt, 2018).

In situ hybridization chain reaction

Hybridization chain reaction (HCR) probe sets for ccl25b, rx2 and ccr9a were designed against unique mRNA sequences, based on the Japanese medaka HdrR ASM223467v1 transcriptome. Probes (50 pmol) were ordered as IDT Oligo pools. Alexa-488- and Alexa-594-conjugated hairpins were purchased from Molecular Instruments, together with the buffers. The CUBIC-based clearing method was used to improve probe penetration (Susaki and Ueda, 2016). Fixed and washed medaka hatchlings were incubated in 50% CUBIC-L/R1a (mixed 1:1 and then diluted to 50% in dH₂O) for 30 min at room temperature while shaking and moved to a 100% CUBIC-L/R1a solution for another 30 min incubation at room temperature on a shaker. Afterwards, the embryos were washed four times for 15 min each with 1× PBS. HCR was performed according to the Molecular Instruments instructions. Probe concentration was increased to 4 pmol. Embryos were incubated with DAPI (1:500 in 5× 0.1% Tween 20 in saline sodium citrate SSCT) for 10 min at room temperature, transferred to a MatTek glass-bottom dish and equilibrated in CUBIC refractive index matching solution CUBIC R+(N) until transparent.

Imaging was performed using the 63× objective on Leica SP8 confocal microscope. Image processing was carried out using Fiji. Bright outliers (of 1 pixel radius) were removed to denoise the image. Plugin “Subtract background” was used to correct background (rolling ball diameter=75 pixels). A maximum projection of 6-10 z-slices (1 μm z-step) was used.

TUNEL staining

To measure the amount of apoptotic cells, we made use of In situ cell death detection kit, TMR red (Roche, 12156792910). To stain cryosections, 50 μl TUNEL reaction mix was added per slide and incubated at 37°C for 1 h. Slides were washed twice with 1×PTW and mounted in 60% glycerol. For whole-mount staining, retinae was permeabilized using 0.1% Triton X-100, 0.1% sodium citrate for 30 min at room temperature. Samples were washed twice with 1×PTW and incubated with the TUNEL reaction mix at 37°C for 1 h in the dark. After washing twice with 1×PTW, they were mounted for imaging.

In vivo imaging and laser induced injury

For in vivo imaging, transgenic Cab embryos were kept in 5×1-phenyl-2-thiourea [PTU (0.0165% m/v), Sigma] in 1× ERM from 1 dpf until imaging to block pigmentation. For embryonic stage imaging, stage 34 embryos were hatched using lab-made hatching enzyme. Live hatchlings were anaesthetized in 1× Tricaine diluted in 1× ERM and mounted in glass bottomed Petri dishes (MaTek) in 1% low melting point agarose.

In vivo imaging and laser ablations were performed on Leica TCS SP8 and SP5 microscopes equipped with a Spectra Physics Mai Tai HP DeepSee Ti:Sapphire laser, which has a tuneable wavelength range from 690-1040 nm, and Leica Hybrid Detectors. The retinal injury was induced using a 14× zoom in combination with the high-energy 2-photon laser, precisely tuned to 880 nm. Subsequent follow-up imaging for GFP detection was conducted using the identical laser wavelength of 880 nm and a 40× long distance objective. Imaging of double reporter fish was performed on a Leica DIVE microscope.

Sudan Black staining

For visualizing macrophages in medaka hatchlings, we used Sudan Black staining, as described by Aghaallaei et al. (2022). Medaka embryos were fixed in 4% PFA overnight, washed with PTW and incubated in Sudan Black staining reagent (Sigma Aldrich, 380B-1KT) for 1 h in the dark. After staining, embryos were washed three times with 70% EtOH and three times with PTW. Pigments were bleached using 3% H₂O₂+0.5% KOH for 15 min and embryos were washed again three times with PTW before they were counterstained with Hematoxylin staining solution for 1 h. Whole hatchlings stained with Sudan Black were captured using a Nikon SMZ18 Stereomicroscope fitted with a Nikon DS-Ri1 camera.

Neutral Red staining

Medaka embryos from the spi1b mutant line were grown until hatchling stage (1 dph) in 1× ERM at 28°C. Hatchlings were stained live with 2.5 μg/ml of Neutral Red solution (Sigma Aldrich, 72210) prepared in 1× ERM for 6 h at 28°C. After staining, hatchlings were rinsed with 1× ERM and anesthetized using 2× Tricaine solution. Hatchlings were mounted in 3% methylcellulose and imaged using Nikon Eclipse 80i microscope. This staining protocol was established as described by Wei et al. (2024).

In situ hybridization

For visualizing macrophages in medaka hatchlings whole-mount in situ hybridization was performed using digoxygenin-labelled cxcr3a probe transcribed from the cDNA of the second exon of the gene (ENSORLG00000013459.2). Medaka hatchlings from spi1b mutant line were fixed in 4% PFA overnight, washed with PTW and permeabilized with Proteinase K for 10 min. Whole-mount in situ hybridization using NBT/BCIP solution was carried out as described previously (Loosli et al., 1998). After staining, hatchlings were washed three times with PTW and mounted on microscopic slides with 60% glycerol in PBS. Images were acquired using Nikon Eclipse 80i microscope.

Immunohistochemistry imaging

All cryosections and whole-mount samples were imaged using an inverted confocal laser scanning microscope Leica TCS SP8. The microscope was equipped with ACS APO objective lenses (10×/0.3 dry, 20×/0.75 multi-immersion and 63×/1.3 glycerol) and featured laser lines at 405 nm, 488 nm, 532 nm and 638 nm. The system was configured with two PMT detectors. For whole-mount sample imaging, retinae were mounted in MatTek dishes using 1% low melting agarose in 1× PTW or clearing solution (Zhu et al., 2019) [20% (wt/vol) urea (Sigma–Aldrich, V900119), 30% (wt/vol) D-sorbitol (Sigma–Aldrich, V900390) and 5% (wt/vol) glycerol (Sigma–Aldrich, V900122) dissolved in DMSO (Sigma Aldrich, V900090)], with the ciliary marginal zone (CMZ) positioned so it was touching the bottom of the dish. After the agarose solidified, the dish was covered with 1× PTW. No solution was added on top of the clearing solution.

Image acquisition of hatchlings

To measure the size of the retina and length of the spi1b wild-type and mutant hatchlings, samples were fixed in 4% PFA in PTW and mounted in 1.5% agarose plates with customized wells. Images were acquired using a Nikon SMZ18 microscope with DS-Ri1 camera.

Image processing and statistical analysis

Images were processed using the Fiji image processing software. Statistical analysis and graphical representation of the data were performed using Rstudio and Prism software package (GraphPad) (https://www.r-project.org/). Tracking of immune cells was carried out using MtrackJ plugin in Fiji. Average velocity, total migration distance and mean square displacement (MSD) as a function of time were calculated for each immune cell. The MSD at time point t was calculated as the squared distance between immune cell position at time point t and at t=0. Box and whisker plots show median (line), 25th and 75th percentiles (box limits) and spread of data to lowest and highest values (whiskers). Cell counting was carried out using the cell counter plugin in Fiji. Dunn's test and the Mann–Whitney test were performed to determine the statistical significances. P<0.05 was considered significant and P-values are given in the figure legends. Sample size (n) is mentioned in figure legends. The experimental groups were allocated randomly, and no anonymization was carried out during allocation. The figures in this study were created using the software Affinity Designer 2, version 2.5.5 (2024), Serif.

Peer review history

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.205198.reviewer-comments.pdf