A glial cell-derived pathway directs regenerating optic nerve axons toward the optic chiasm

ABSTRACT

After optic nerve injury, several retinal ganglion cell (RGC) intrinsic signaling pathways have been shown to enhance RGC survival and RGC axonal growth. In contrast, few extrinsic cues have been identified that guide regenerating RGC axons toward and across the optic chiasm. Here, we use live-cell imaging in larval zebrafish to show that regrowing RGC axons initiate growth toward the midline and extend along a trajectory similar to their original projection. From a candidate genetic screen, we identify the glycosyltransferase Lh3 (also referred to as Plod3) to be required to direct regrowing RGC axons toward the midline during active regeneration. Moreover, we show that transgenic lh3 expression in sox10⁺ presumptive olig2⁺ oligodendrocytes located near the optic chiasm restores directed axonal growth in lh3 mutants. Finally, we find that mutants in collagen 18a1 (col18a1), a putative Lh3 substrate, display RGC axonal misguidance phenotypes similar to lh3 mutants, suggesting that lh3 may act through col18a1 during regeneration. Combined, these data identify lh3 as part of a glial derived molecular pathway that is critical for guiding in vivo regenerating RGC axons toward and across the optic chiasm.

Highlighted Article: Live-cell imaging and genetic screening in larval zebrafish reveal that Lh3 is required to direct regrowing retinal ganglion cell axons toward the optic chiasm during the early stages of optic nerve regeneration.

INTRODUCTION

As a component of the central nervous system, retinal ganglion cell (RGC) axons within the mammalian optic nerve display a poor regenerative capacity. Multiple factors contribute to the minimal regeneration in mammals, including massive injury-induced RGC death and overall limited axonal regrowth (Villegas-Pérez et al., 1993; Berkelaar et al., 1994; Goldberg et al., 2002). Several studies have identified a number of RGC intrinsic signaling pathways that enhance RGC survival and increase long-range axonal growth after injury. For example, deletion of the transcriptional repressor KLF4 in RGCs increases axonal regrowth (Moore et al., 2009). Similarly, activation of inflammatory responses within the retina through injection of Zymosan and cAMP enhances cell survival and axonal regrowth (Leon et al., 2000; Goldberg et al., 2002; Yin et al., 2009). Moreover, deleting intrinsic factors in RGCs, such as PTEN and TSC1, which function as negative regulators of the mTOR pathway, improves both RGC cell survival and axonal growth (Park et al., 2008; Smith et al., 2009; Norsworthy et al., 2017). However, enhancing RGC axonal growth by manipulating RGC intrinsic programs frequently results in axonal misguidance as axons project inappropriately from the optic tract before and at the optic chiasm during the initial stages of regeneration (Kurimoto et al., 2010; Luo et al., 2013; Pernet et al., 2013a,b; Berry et al., 2019). This demonstrates a need to identify molecular and cellular mechanisms that guide RGC axons as they regrow toward and across the optic chiasm during regeneration.

Following a central nervous system (CNS) injury in mammals, many non-neuronal cells create an extracellular environment that is largely inhibitory to axonal regrowth. The cellular and molecular barriers that form within a lesion site have long been attributed to an influx of inflammatory cells and an upregulation of inhibitory molecules that are not conducive for robust regeneration (Yiu and He, 2006; Fitch and Silver, 2008). Additional studies in mice have since revealed roles for microglia and oligodendrocytes in functional recovery and remyelination from CNS injury (Wang et al., 2020; Mendonça et al., 2021; Brennan et al., 2022). Yet due to the dominance of factors inhibiting axonal regrowth, the cellular and molecular mechanisms that guide regrowing RGC axons following axonal injury are largely unknown. Recently, spontaneous CNS regeneration has been described in two mammalian species, the naked mole rat and the spiny mouse, demonstrating that pro-regenerative environments that support robust CNS regeneration are not unique to the CNS of non-mammalian vertebrates (Park et al., 2017; Nogueira-Rodrigues et al., 2022). Therefore, model organisms that exhibit spontaneous robust CNS axonal regeneration, without an overwhelmingly inhibitory extracellular environment, provide unique opportunities to identify mechanisms that promote CNS regeneration.

We recently developed a rapid and robust optic nerve transection assay in larval zebrafish to take full advantage of its regenerative capabilities and optical transparency to study functional optic nerve regeneration in vivo (Harvey et al., 2019, 2023). At just 5 days post fertilization (dpf), larval zebrafish already possess a functional visual system driving several visually guided behaviors, such as prey capture (Brockerhoff et al., 1995; Easter and Nicola, 1996; Gahtan et al., 2005). Following complete transections of the RGC axons in 5 dpf larvae, we observe axonal degeneration, subsequent initiation of regrowth and reinnervation of the optic tecta by 72 h post transection (hpt) (Harvey et al., 2019). We have also shown that RGC axonal regeneration in larval zebrafish occurs independently of RGC death or neurogenesis, and that visual function is restored by 8 days post-transection (Harvey et al., 2019). Using this optic nerve transection assay, we performed a candidate genetic screen and identify Lh3, a collagen-modifying glycosyltransferase whose gene name is referred to as lh3 or also plod3, to be crucial for directing axonal regrowth toward the optic chiasm during optic nerve regeneration. We show that following optic nerve transection, lh3 expression is upregulated near the site of axonal injury and that Lh3 functions during the time when active regeneration occurs, specifically during the early stages of regeneration as RGC axons are growing toward the midline. We also show that Lh3 expression in oligodendrocytes is required to direct early regrowing RGC axons across the midline along their original path. Furthermore, we show that mutants in collagen 18a1 (col18a1), a putative Lh3 substrate, display RGC axonal misguidance phenotypes similar to lh3 mutants, indicating at a glial-derived molecular pathway in which Lh3 acts through Collagen 18a1 to direct RGC axons. Combined, our results demonstrate a critical pro-regenerative in vivo role for oligodendrocytes during early optic nerve regeneration to direct axon regrowth toward and across the midline.

RESULTS

Lh3 is required for growth across midline during regeneration

To identify molecular mechanisms that guide regrowing RGC axons across the optic chiasm during optic nerve regeneration, we performed a candidate genetic screen consisting of 35 genes that are expressed in the visual system or have known functions in axon growth and guidance. Specifically, we used a sharpened tungsten needle to transect the optic nerves of genetic mutants and their siblings at 5 dpf and assessed at 72 hpt whether RGC axons had regrown across the optic chiasm to re-innervate the optic tectum. One gene of particular interest to us was lh3, which encodes a multidomain glycosyltransferase that post-translationally modifies collagen proteins (Heikkinen et al., 2000; Wang et al., 2002a). We have previously shown that Lh3 is required in Schwann cells to promote target-specific regeneration of dorsal motor nerve axons in the peripheral nervous system (Isaacman-Beck et al., 2015). We therefore asked whether Lh3 plays a similar role in optic nerve regeneration.

RGC axon projections in lh3 mutants cross the midline and project to the optic tecta in a manner indistinguishable to their siblings (Zeller et al., 2002). However, we took advantage of a previously validated transgene Tg(hsp70:lh3-myc), which expresses a Myc-tagged Lh3 under the control of the hsp70 heat-shock-activated promoter (Isaacman-Beck et al., 2015) to rescue various morphological deficits that would affect the survivability of the larvae for the duration of the regeneration experiments (Zeller and Granato, 1999; Taler et al., 2019) (Fig. 1A). At 5 dpf, uninjured RGC axons in both lh3 siblings and mutants exited the eye and projected across the optic chiasm to terminate in the optic tecta (Fig. 1B-E). Following optic nerve transection, axons in lh3 siblings grew across the midline and reinnervated the optic tecta (Fig. 1F-H,L). In contrast, transected optic nerve axons in lh3 mutants failed to cross the midline and instead projected along aberrant trajectories, often along anterior and posterior paths, or also dorsally or ventrally (Fig. 1I-L, magenta arrowheads). These results reflect a functional role for Lh3 during optic nerve regeneration in directing RGC axons towards and across the midline at the optic chiasm.

Fig. 1.

Lh3 is required for growth across the midline during regeneration. (A) Timeline of a transection experiment with zebrafish lh3 larvae (hpf, hours post-fertilization; dpf, days post-fertilization; hpt, hours post-transection). (B-E) Before injury, RGC axons labeled using Tg(isl2b:GFP) in siblings (B,C) and conditional lh3 mutants (D,E) cross the midline and innervate contralateral tecta. (F-H) In lh3 siblings at 72 hpt, RGC axons grow across the midline and re-innervate contralateral optic tecta, schematized in H. (I-K) In contrast, RGC axons in lh3 mutants turn away from the midline, project aberrantly (magenta arrowheads) and do not reinnervate contralateral optic tecta, schematized in K. Dashed lines indicate the outline of the eyes. Scale bars: 50 µm. Representative images of chiasms and tecta for each timepoint are of the same fixed larva, whereas across timepoints are different larvae. (L) Quantification of the percentage of nerves at 72 hpt that grew across the midline to re-innervate the optic tectum, did not grow across the midline and instead grew along aberrant trajectories, or showed stalled axonal growth. Numbers indicate total number of transected nerves. ****P<0.0001 (Fisher's exact test).

Lh3 expression is upregulated in the optic chiasm during regeneration

We next asked whether Lh3 is required during the process of optic nerve regeneration. For this, we took advantage of the conditional Tg(hsp70:lh3-myc) transgene (Isaacman-Beck et al., 2015). We induced Lh3 expression in 5 dpf larvae 4-5 h before performing the optic nerve transections, and then assessed axonal regrowth across the optic chiasm and re-innervation of the optic tecta at 72 hpt (Fig. 2A). RGC axons in lh3 siblings that did or did not receive a regeneration heat shock regrew across the midline and innervated the optic tecta (Fig. 2B,D,F). In contrast, lh3 mutants without a heat shock displayed aberrant axonal growth, while heat shock-induced Lh3 expression during regeneration in lh3 mutants restored optic nerve axonal regrowth across the midline (Fig. 2C,E,F). These data demonstrate that Lh3 acts during the process of optic nerve regeneration.

Fig. 2.

lh3 expression is upregulated in the optic chiasm during early regeneration. (A) Timeline of induced Lh3 expression with Tg(hsp70:lh3-myc) during zebrafish development as well as regeneration (hpf, hours post-fertilization; dpf, days post-fertilization; hpt, hours post-transection). (B,C) At 72 hpt in lh3 siblings with no heat-shock induction, RGC axons cross the midline to innervate the optic tecta, while RGC axons in lh3 mutants turn away from the midline and project aberrantly (C, magenta arrowheads). (D,E) In lh3 siblings and mutants that received a heat shock, RGC axons project across the midline (E, white arrowheads) and innervate the optic tecta. Dashed lines indicate the outline of the eyes. Scale bar: 50 µm. (F) Quantification of the percentage of nerves at 72 hpt that grew across the midline to re-innervate the optic tectum, did not grow across the midline and instead grew along aberrant trajectories, or showed stalled axonal growth in Tg(isl2b:GFP) larvae with and without a regeneration heat shock. Numbers indicate total number of transected nerves. ****P<0.0001, *P=0.0279; n.s., P=0.4804 (Fisher's exact test). (G-I) lh3 mRNA expression in Tg(isl2b:GFP) larvae pre-transection and at 2-6 hpt. (G) Outlined areas of the retina and the optic chiasm are shown in H and I, respectively. Scale bars: 50 µm in G,I; 20 µm in H. Axons in the chiasm region at 2-6 hpt have not yet fully degenerated following transection. Dashed lines indicate the outline of the eyes in I. Representative images across timepoints are of different larvae.

We then sought to determine in which cell-type Lh3 was expressed during optic nerve regeneration. We used in situ hybridization chain reaction to detect lh3 mRNA expression before and after optic nerve transections. Before transections, we failed to detect lh3 mRNA expression in retinas nor in the region of the optic chiasm (Fig. 2G-I; n=17 larvae). Similarly, after transection, we failed to detect lh3 mRNA expression in RGC neurons in the retina (Fig. 2G,H; n=41 larvae). In contrast, we detected upregulated lh3 mRNA expression around the transected proximal stump of RGC axons and near the midline during early regeneration between 2 and 6 hpt (Fig. 2G,I; n=30/41 larvae), demonstrating that lh3 mRNA is upregulated following optic nerve transection in the region of the optic chiasm.

Lh3 directs regrowing RGC axonal growth toward their original trajectory

Following optic nerve transections in wild-type larvae, we have previously observed new RGC axons emerging from the proximal stump of transected optic nerves by 24 hpt and exhibiting robust growth across the optic chiasm by 48 hpt (Harvey et al., 2019). To determine whether Lh3 continuously directs regenerating RGC axons once they emerge from the optic nerve stump, we used live cell imaging and examined the dynamics of regenerating RGC axons as they approach and cross the optic chiasm, which has not previously been reported. For simplicity, we removed one eye from 4 dpf larvae to only visualize the regrowth from a single retina per lh3 sibling and mutant larva (Fig. 3A, Fig. S1). Following transections on 5 dpf lh3 sibling and mutant larvae, we observed that the region of axons distal to the transection site underwent Wallerian degeneration, as has been observed for other nerves in zebrafish (Fig. 3B-D,G-I, Movies 1 and 2, brackets) (Waller, 1851; Martin et al., 2010; Rosenberg et al., 2012; Bremer et al., 2019). In lh3 siblings, some axons emerged from the proximal stump of RGC axons in a range of directions, often extending and retracting rapidly (Fig. 3B,C, Movie 1). By 16 hpt, we observed that a small group of axon fascicles grew toward the midline and rapidly extended across the midline (Fig. 3D,E, Movie 1, white arrowheads). Over the next few hours, additional fascicles of axons were observed following the growth of the axons that had crossed the midline (Fig. 3F, Movie 1), revealing in vivo axonal dynamics during the early stages of optic nerve regeneration. Axons in lh3 mutants emerged at similar timepoints to lh3 siblings from the proximal stump (Fig. 3G,H, Movie 2). However, unlike the lh3 siblings, axonal growth did not persist across the midline and axons often grew anteriorly and posteriorly, or dorsally and ventrally around the eye (Fig. 3I-K, Movie 2, magenta arrowheads), strongly suggesting that Lh3 directs the initial regrowth of RGC axons toward the midline during the early process of optic nerve regeneration.

Fig. 3.

lh3 directs the initial regrowing RGC axons. (A) Schematic of zebrafish larvae expressing Tg(isl2b:GFP) to label RGCs that were used for in vivo imaging. Larvae have one eye removed 1 day before timelapse imaging. The region outlined is shown in B-K. (B,G) Transected RGC axons in lh3 sibling (B) and lh3 mutant (G) larvae separate from the distal region of the optic nerve that will eventually degenerate (indicated by brackets in B-D,G-I). (C) In lh3 siblings, RGC axons initiate growth out from the proximal stump (white arrowhead). (D-F) A small group of axons (white arrowhead) in lh3 siblings projects directly toward the midline along the original path of the projection (D) and then subsequently additional axonal growth follows (E, white arrowhead, F) (n=7/10 nerves). (H-K) In lh3 mutants, axons grow out from the proximal stump (H), but persist along aberrant trajectories (I-K, magenta arrowheads; n=6/12 nerves). See also Fig. S1. Dashed lines indicate the outline of the eyes. Scale bar: 25 µm.

We next sought to identify the mechanisms by which Lh3 directs regrowing RGC axons toward and across the optic chiasm. We first asked whether Lh3 is critical for the initiation of axonal growth extending from the proximal stump of transected RGC axons, and consequently whether optic nerve regrowth is delayed in lh3 mutants. We found no significant difference in the timing of initiation of axonal regrowth post transection (Fig. 4A,B). To further define the trajectories that RGC axons take as they grow across the optic chiasm, we measured the angle between distinct axon fascicles and the original projection path before transection (Fig. 4C; see Materials and Methods for more details). In optic nerves in lh3 siblings and mutants that regrew across the midline, we observed that while some fascicles strayed from the pre-injury path, most fascicles grew very close to their original projection path (Fig. 4D; Table S1). Nerves in lh3 siblings that failed to grow toward the midline grew along a mostly single aberrant trajectory (Fig. 4D; Table S1). In contrast, in lh3 mutants we found that the aberrant paths of axon projections in nerves that failed to cross the midline displayed a greater range of angles (comparing lh3 siblings ‘Tectal Innervation Across Midline’ to lh3 mutants ‘Aberrant Axonal Growth,’ P=0.0005, F-test to compare variances; Fig. 4D; Table S1). These data strongly suggest that lh3 is required to direct RGC axons along their regenerative path, which is very similar to their pre-transection path, i.e. towards and across the optic chiasm.

Fig. 4.

lh3 is required to direct regrowing RGC axons along their original path. (A) Representative frames from live timelapses of lh3 sibling (top) and lh3 mutant (bottom) zebrafish larvae expressing Tg(isl2b:GFP) to label RGCs showing the initial axonal growth that emerges from the injured proximal stump of axons. hpt, hours post-transection. (B) Comparison of the time when axon fascicles first emerge from the proximal stump in lh3 siblings and mutants. Time was determined by the first frame when axon fascicles were observed extending from the proximal stump that extended for at least one other consecutive frame. Data are represented as mean±s.d.; n.s., P>0.05 (unpaired t-test). (C) Representative image of a regenerating nerve in a lh3 sibling at 20 hpt and the mark-up to measure the angle of growing fascicles from the injured stump of axons. The dashed magenta line marks the parasphenoid cartilage. See Materials and Methods for more details on quantifications. (D) Quantification at 20 hpt of the absolute angle of axon fascicles from the end of a projection to the path of the original axonal projection (Tectal Innervation Across Midline: sibling, n=6 nerves; mutants, n=3 nerves; Aberrant Axonal Growth: sibling, n=3 nerves; mutants, n=6 nerves). Individual dots are measured fascicle absolute angles, lines represent means. n.s., P>0.05; *P<0.05 (one-way ANOVA). See Table S1 for individual angles.

Glial expression of Lh3 restores RGC axonal growth across the midline

Given that lh3 mRNA expression was upregulated in area of the optic chiasm (Fig. 2I), we hypothesized that the relevant cell type crucial for lh3-mediated optic nerve regeneration would be located in the region of the optic chiasm. Transcriptomics studies in mice have found lh3 to be expressed most greatly in microglia (Zhang et al., 2014). Moreover, macrophages and microglia have been shown to clear debris from injured axons and can promote axonal growth and remyelination during CNS regeneration (Rosenberg et al., 2012; Wang et al., 2020; Brennan et al., 2022). Therefore, we first focused on macrophages and microglia as a potential source of Lh3 activity during optic nerve regeneration. In untransected larvae, we observed sporadic mpeg⁺ (mpeg1.1⁺) cells near the optic chiasm (Fig. 5A). At 24 hpt, we observed a greater concentration of mpeg⁺ cells surrounding the proximal stump of RGC axons and the new axonal growth (Fig. 5A; n=11 larvae). We next used a transgenic rescue approach to test whether expression of lh3 in macrophages and/or microglia is sufficient to restore axonal regrowth across the optic chiasm in lh3 mutants. We found that Lh3 expression from a Tg(mpeg:lh3-myc) transgene failed to rescue the lh3 mutant phenotype, where nerves in lh3 mutants mostly displayed aberrant growth similar to lh3 mutants without the Tg(mpeg:lh3-myc) transgene (Fig. 5B). Since RGC axons regrew very closely to their original path (Fig. 4D), we next sought to examine oligodendrocytes associated with the optic nerve (Brösamle and Halpern, 2002). Using both a Tg(olig2:dsred) transgenic line labeling oligodendrocyte lineage cells (Fig. 5C, n=5 untransected larvae, n=21 transected larvae) as well as in situ hybridization chain reactions for mbpa mRNA in myelinating oligodendrocytes (Fig. 5D, n=6 untransected larvae, n=5 transected larvae), we observed oligodendrocytes in the optic chiasm and associated with RGC axons before and after transection (Kucenas et al., 2008; Ackerman and Monk, 2016). We then tested whether expression of lh3 in oligodendrocytes would rescue axonal regrowth across the optic chiasm in lh3 mutants. Sox10, like olig2, is a transcription factor that mediates oligodendrocyte lineage specificity and is maintained in all oligodendrocyte lineage cells (Bergles and Richardson, 2015; Ackerman and Monk, 2016). We found that lh3 expression from a Tg(sox10:lh3-mkate) transgene restored RGC regrowth across the midline in otherwise mutant lh3 larvae (Fig. 5E). Furthermore, we detected lh3 mRNA expression that overlaps with transgenic labeling of oligodendrocyte cells, demonstrating that lh3 is upregulated in oligodendrocytes surrounding the proximal stump of transected RGC axons and along an axonal projection path across the midline during optic nerve regeneration (Fig. 5F,G, Movie 3; n=15/21). Altogether, these findings demonstrate that lh3 expression in oligodendrocytes near the proximal stump of transected RGC axons directs early regrowing RGC axons toward the midline along their original path to reinnervate the optic tecta.

Fig. 5.

Lh3 in sox10-expressing glia restores RGC axonal growth across the midline. (A) In wild-type zebrafish larvae expressing Tg(isl2b:GFP) to label RGCs, macrophages and microglia labeled by Tg(mpeg1:Gal4FF);Tg(UAS:nfsB-mCherry) are observed in the area of the optic chiasm before transection. At 24 h post-transection (hpt), an influx of mpeg⁺ cells is seen. (B) Quantification of the percentage of nerves at 72 hpt that grew across the midline to re-innervate the optic tectum, did not grow across the midline and instead grew along aberrant trajectories, or showed stalled axonal growth with and without the Tg(mpeg:lh3-myc) transgene. Numbers indicate total number of transected nerves. ****P<0.0001; n.s., P>0.05 (Fisher's exact test). (C,D) Oligodendrocyte lineage cells labeled by Tg(olig2:dsred) as well as mbpa mRNA expression in myelinating oligodendrocytes are also seen in the chiasm before and after transection, near the injured proximal stump of RGC axons or along the degenerating original projection. (E) Quantification of regeneration with and without Tg(sox10:lh3-mkate) transgene. Numbers indicate total number of transected nerves. ****P<0.0001; n.s., P>0.05 (Fisher's exact test). (F) lh3 mRNA expression at 4-6 hpt is upregulated and overlaps with oligodendrocytes labeled by Tg(olig2:dsred). Scale bar: 50 μm (bar in F applies to A,C,D,F). (G) Additional images of individual oligodendrocytes (arrowheads) labeled by Tg(olig2:dsred) co-expressing lh3 mRNA (see also Movie 3). Scale bar: 10 μm. Representative images across timepoints are of different larvae. Axons in the chiasm region at 4-6 hpt have not yet fully degenerated following transection. Dashed lines indicate the outline of the eyes.

Col18a1 is also critical for directing regenerating RGC axons

We further sought to identify the relevant Lh3 substrate during optic nerve regeneration. Given the well-documented role of Lh3 in post-translationally modifying collagen proteins (Heikkinen et al., 2000; Wang et al., 2002a), we focused on collagen substrates. As part of our initial candidate genetic screen, we examined col4a5, which is required for proper RGC axon targeting of tectal layers and also functions to scaffold guidance cues during both the development of the visual system and during motor nerve regeneration (Xiao and Baier, 2007; Xiao et al., 2011; Isaacman-Beck et al., 2015; Murphy et al., 2022). We also tested col7a1, which has been shown to be expressed in the eye (Wullink et al., 2018). We ultimately found that RGC axon regeneration toward and across the midline is unaffected in col4a5 and col7a1 mutants (Fig. S2). We then examined col18a1, a collagen associated with Knobloch syndrome in humans that is characterized by eye abnormalities (Fukai et al., 2002; Suzuki et al., 2002). Col18a1 has several putative Lh3 sites within X-Lys-Gly- motifs and has been shown in vitro to bind to Lh3 (Risteli et al., 2004; Salo and Myllyharju, 2021). First, we used in situ hybridization chain reaction to detect col18a1 mRNA expression and observed an upregulation of col18a1 mRNA at 12 hpt in the optic chiasm, similar to the expression pattern of lh3 following optic nerve transection (Fig. 6A; n=13/14 larvae). We also detected col18a1 mRNA expression that overlaps with transgenic labeling of oligodendrocyte cells, which demonstrates that col18a1 is also upregulated in oligodendrocytes surrounding the proximal stump of transected RGC axons and along the axonal projection path across the midline during optic nerve regeneration (Fig. 6B, Movie 4). We then tested col18a1 mutants in our optic nerve transection assay. Prior to transection, RGC axonal projections were indistinguishable between col18a1 siblings and mutants (Fig. 6C-F). In contrast, at 72 hpt we observed regrowing RGC axons projecting away from the CNS midline, similar to what we observed in lh3 mutants where regenerating RGC axons grew along aberrant paths (Fig. 6G-K). This demonstrates that, during regeneration, col18a1 is critical for directing regenerating RGC axons toward the midline and suggests that glial-derived Lh3 acts, as least in part, through Col18a1 during optic nerve regeneration.

Fig. 6.

The Lh3 candidate substrate Col18a1 is upregulated after transection and directs RGC axons across the midline during regeneration. (A) col18a1 mRNA expression pre-transection and at 12 h post-transection (hpt) in zebrafish larvae expressing Tg(isl2b:GFP) to label RGCs. Axons in the chiasm region at 12 hpt have not yet fully degenerated following transection. Scale bar: 50 μm. (B) Additional images of individual oligodendrocytes labeled by Tg(olig2:dsred) co-expressing lh3 mRNA (see also Movie 4). Scale bar: 10 μm. (C-F) Before injury, RGC axons labeled by Tg(isl2b:GFP) in siblings (C,D) and col18a1 mutants (E,F) cross the midline and innervate contralateral tecta. (G-J) In contrast to col18a1 siblings at 72 hpt (G,H), RGC axons in col18a1 mutants turn away from the midline, project aberrantly (magenta arrowheads) and do not re-innervate contralateral optic tecta (I,J). Dashed lines indicate the outline of the eyes. Scale bar: 50 µm. (K) Quantification of the percentage of nerves at 72 hpt that grew across the midline to re-innervate the optic tectum, did not grow across the midline and instead grew along aberrant trajectories or showed stalled axonal growth. Numbers indicate total number of transected nerves. ****P<0.0001 (Fisher's exact test). Representative images of chiasms and tecta for each timepoint are of the same fixed larva, whereas across timepoints are of different larvae.

DISCUSSION

Regenerating axons often encounter multiple choice points along their paths that they must navigate properly to robustly reinnervate their original targets. Previous studies have shown that PTEN inactivation in RGC neurons strongly promotes RGC regrowth following an optic nerve injury, but that these axons display axonal misguidance at the optic chiasm, proving the optic chiasm to be one such choice point (Kurimoto et al., 2010; Luo et al., 2013; Pernet et al., 2013a,b; Berry et al., 2019). Here, we identify a mechanism critical for guiding RGC axons toward and across the optic chiasm during optic nerve regeneration. We find that lh3 is required during regeneration to direct regenerating RGC axons toward and across the midline. We find that, following optic nerve transection, expression of lh3 and its presumptive substrate col18a1 is upregulated in oligodendrocytes along the regenerative path near the optic chiasm. Using live cell imaging, we demonstrate that lh3 is required to direct regrowing RGC axons along a regenerative trajectory very similar to their original path. Together, these results provide compelling evidence for an RGC extrinsic molecular pathway that guides optic axons during regeneration.

Live cell imaging reveals axonal dynamics of early optic nerve regeneration

To date, most optic nerve regeneration studies in adult rodents and fish monitor RGC axonal regrowth with static time point analyses (Horder, 1971; Murray, 1976; Pernet et al., 2013b; Yin and Benowitz, 2018). In contrast, peripheral nervous system regeneration, as well as spinal cord regeneration, have been visualized in live larval zebrafish using in vivo imaging studies (Bhatt et al., 2004; Rosenberg et al., 2012; Ceci et al., 2014; Lewis and Kucenas, 2014; Isaacman-Beck et al., 2015; Bremer et al., 2019). To visualize RGC axonal dynamics during optic nerve regeneration, we previously established an optic nerve transection assay in larval zebrafish (Harvey et al., 2019, 2023). Here, we performed live-cell imaging of RGC axons initiating regrowth and regenerating across the optic chiasm in vivo. These timelapses revealed two intriguing insights into mechanisms that may mediate robust CNS regeneration in zebrafish. First, we observed that axons distal to the transection site underwent Wallerian degeneration (Fig. 3B-D,G-I, Movies 1 and 2, brackets) (Waller, 1851). While a hallmark of the mammalian CNS is a slower rate of axonal fragmentation following axonal injury, here we observe complete degeneration of the distal axons by about 15 hpt (Vargas and Barres, 2007). While we did not examine whether persistence of degenerating debris would repel axonal growth, we did observe new axons reaching the midline around the same time axonal debris was cleared from that region (Fig. 3D). Axonal debris and associated myelin proteins are known to inhibit axonal regeneration (Vargas and Barres, 2007). But here, we found that Wallerian degeneration did not appear to be delayed in lh3 mutants, and we therefore would not hypothesize that axonal debris contributes to the misguidance of RGC axons in lh3 mutants.

Second, our live-cell imaging revealed that a small group of axon fascicles exhibited strongly directed growth toward the midline by 16 hpt and subsequently extended across rapidly (Fig. 3D,E; Movie 1, white arrowhead). Additional fascicles of axons continued to grow and follow the growth of the axons that initially crossed the midline (Fig. 3F, Movie 1). These axons grew along a path that almost completely overlapped with their original path (Fig. 4D), and strongly suggests that there are factors expressed from cells associated with that projection that direct regrowing RGC axons during optic nerve regeneration.

A previously unappreciated role for oligodendrocytes in axon guidance during axon regeneration

To determine the relevant cell type through which lh3 is functioning during optic nerve regeneration, we used cell-type specific transgenic rescue. Based on the lh3 and col18a1 mRNA expression in the optic chiasm (Figs 2G-I and 6A), we hypothesized that the relevant cell type would be one found in that region during the initial stages of regeneration. We observed macrophages and microglia dynamically infiltrating the chiasm and surrounding regenerating axonal growth (Fig. 5A). Tg(mpeg:lh3-myc) expression failed to rescue aberrant axonal growth in lh3 mutants, suggesting that lh3 expression needs to be more specifically spatially regulated (Fig. 5D). We therefore examined oligodendrocytes and observed oligodendrocyte lineage cells present in the optic chiasm, and we detected mbpa mRNA representing myelinating oligodendrocytes along the degenerating axons (Fig. 5B,C). We found that lh3 expression from a Tg(sox10:lh3-mkate) transgene restored RGC regrowth across the midline in lh3 mutants (Fig. 5E). However, the transgenic tools we employed here did not allow us to distinguish whether oligodendrocyte precursor cells or mature oligodendrocytes are the relevant cell type for lh3 and col18a1, as sox10 and olig2 are both transcription factors that are expressed in all oligodendrocyte lineage cells (Bergles and Richardson, 2015; Ackerman and Monk, 2016). Previous studies have implicated oligodendrocytes as playing roles in remyelination (Wang et al., 2020; Mendonça et al., 2021) and axon repulsion or inhibition via myelin-associated glycoprotein (McKerracher et al., 1994; Mukhopadhyay et al., 1994), NogoA (Chen et al., 2000; GrandPré et al., 2000), oligodendrocyte-myelin glycoprotein (Wang et al., 2002b) and semaphorin 4D (Moreau-Fauvarque et al., 2003). From our results, we propose a mechanism in which Lh3 and Col18a1 expressed in oligodendrocytes mediate axon attraction to direct regenerating RGC axons along their original path, suggesting a previously unappreciated role for oligodendrocytes in CNS regeneration.

A molecular mechanism for lh3 and col18a1 in specifying a regenerative path

We identified lh3 to be required to direct regenerating RGC axons toward and across the optic chiasm (Fig. 1). We then determined that col18a1 is also critical for directing RGC axons during optic nerve regeneration, unlike col4a5, found in basement membranes, or fibrillar col7a1, which demonstrates a specific role for col18a1 in optic nerve regeneration (Fig. 6, Fig. S2). But the question remains: what are the mechanisms by which Lh3 and Col18a1 target RGC axons to their original path? Using in situ hybridizations, we observed col18a1 expression surrounding the injured axon stumps but also expressed somewhat broadly in the chiasm region (Fig. 6B). We therefore cannot exclude a possible mechanism involving RGC axons directly interacting with Col18a1 in the extracellular matrix through integrins (Elango et al., 2022). Transcriptomic analysis of RGCs at different stages of regeneration in adult zebrafish shows that several integrins are upregulated as axons grow toward and across the midline (Dhara et al., 2019). Therefore, the spatial and temporal specificity of Lh3 and Col18a1 expression could establish the regenerative path for regrowing RGC axons independently of any other guidance cues.

But in addition to playing structural roles, there is growing evidence showing a crucial role for extracellular matrix proteins to bind and regulate the spatial distribution of axon guidance molecules in the extracellular space. For decades, glycosaminoglycans, such as heparan sulfate proteoglycans, have been known to regulate the extracellular localization of Slit repulsive guidance cues to affect axon guidance in development (Wang and Denburg, 1992; Johnson et al., 2004; Smart et al., 2011). More specifically, col4a5 has been shown to be crucial during development of the retinotectal projection for localizing Slit1 expression in the superficial layers of the optic tectum, which mediates lamina-specific axon targeting (Xiao and Baier, 2007; Xiao et al., 2011). During motor nerve regeneration, spatially restricted col4a5 and lh3 are required in a subset of Schwann cells to possibly create a Slit1a gradient that directs dorsal nerve axons to their original projection paths (Isaacman-Beck et al., 2015; Murphy et al., 2022). In this study, our results describe a similar mechanism where lh3 is required to direct regrowing RGC axons toward the midline along their original path, but here, we found that, in optic nerve regeneration, Lh3 potentially acts through Col18a1, not Col4a5. What are possible guidance cues that Col18a1 could be spatially regulating? Col18a1 has three isoforms and within the long isoform is a Frizzled-like domain that has been shown to interact with Wnt molecules (Quélard et al., 2008). Moreover, Wnt signaling has been implicated in both promoting as well as repelling axonal growth in spinal cord and optic nerve regeneration (Yin et al., 2008; Hollis and Zou, 2012; Yam and Charron, 2013; Strand et al., 2016; Patel et al., 2017; Wehner et al., 2017). Therefore, we propose another potential mechanism in which Lh3 is required for Col18a1 expression that possibly spatially restricts Wnts in the extracellular space along the original RGC axonal projection to direct regenerating RGC axons during optic nerve regeneration. Altogether, our studies demonstrate that live-cell imaging of optic nerve regeneration in larval zebrafish allows the unprecedented visualization of axonal growth dynamics and potential interactions with other cell types in the optic chiasm. Our data here reveal a glial-derived molecular pathway that directs regenerating RGC axons across the optic chiasm during optic nerve regeneration, furthering our understanding of the cellular and molecular mechanisms that promote robust functional optic nerve regeneration.

MATERIALS AND METHODS

Fish lines and maintenance

All zebrafish (Danio rerio) work was performed in compliance with the University of Pennsylvania Institutional Animal Care and Use Committee regulations and raised as previously described (Mullins et al., 1994). All transgenic lines were maintained in the Tupfel long fin genetic background except the col18a1 line, which were maintain in the AB background. The following transgenic lines and mutants were used: lh3^TV2O5 (Schneider and Granato, 2006), col18a1 (Isaacman-Beck et al., 2015), col4a5^s510 (Xiao and Baier, 2007), col7a1^sa3473, Tg(hsp:lh30-myc) (Isaacman-Beck et al., 2015), Tg(sox10:lh3-mkate) (Isaacman-Beck et al., 2015), Tg(isl2b:GFP) (Pittman et al., 2008), Tg(olig2:dsred) (Kucenas et al., 2008), Tg(mpeg1:Gal4FF)^gl25 (Ellett et al., 2011) and Tg(UAS:nfsB-mCherry)^c264 (Davison et al., 2007). Adult zebrafish and embryos were raised at 28°C on a 14-h:10-h light:dark cycle or at 29°C for larvae in PTU E3, see the section ‘Transection assay’.

To generate Tg(mpeg1.1:lh3-myc), Gateway cloning was employed to combine lh3-myc into a pDest vector with the mpeg1.1 promoter and Tol2 transposon sites. Tol2 transgenesis was performed as previously described (Suster et al., 2009) into one-cell stage embryos from lh3 heterozygous incrosses. Transgenic lines were identified by genotyping for the transgene with forward primer ACAGTCTCTTGCGTCATCAAAACC and reverse primer TGATCACCAGCAGCTCATTG.

Transection assay

Optic nerve transection assays were performed as previously described (Harvey et al., 2019, 2023). Briefly, to inhibit melanocyte pigmentation, larvae were raised in phenylthiourea (PTU, 0.2 mM in E3 medium) in the dark at 29°C beginning at 1 dpf. For live-cell imaging and distal transections, larvae were anesthetized in PTU E3 plus 0.0053% tricaine then mounted in 2.5% low-melt agarose (SeaPlaque, Lonza) prepared with PTU E3 plus 0.016% tricaine ventral upwards on a glass microscopy slide. One eye was then completely removed using forceps at 4 dpf. Larvae were recovered in Ringer's solution and then kept in PTU E3 at 29°C until transected. At 5 dpf, larvae were anesthetized, mounted in agarose and then optic nerve transections were carried out on an Olympus SZX16 fluorescent microscope or a dissecting microscope with a NightSea Stereo Microscope Fluorescence Adapter (RB-GO only wavelength set, bandpass filter) using a sharpened tungsten needle (Fine Science Tools; tip diameter, 0.001 mm; rod diameter, 0.125 mm). Optic nerves were transected at the region of the nerve distal to it exiting the eye, yet proximal to the optic chiasm. Following injuries, larvae were removed from the agarose, allowed to recover in Ringer's solution with 0.2 mM PTU, then returned to 0.2 mM PTU in E3 medium at 29°C. Larvae were inspected for transection efficiency at 16-18 hpt, except for larvae that were fixed at early post-transection timepoints, and only larvae with complete optic nerve transections with no visible intact axons remaining from the eye to the tectum were kept until fixation at later designated timepoints.

Heat shock-induced lh3 rescue

To induce expression of Lh3 from the Tg(hsp:lh3-myc) transgene during development, embryos were developmentally staged in the late afternoon and no more than six embryos were placed in each tube in PCR strip-tubes. Tubes were incubated in a thermal cycler at 28°C until embryos reached 11.66 hpf, then they were heated to 38°C for 10 min and held at 28°C until they could be removed from the tubes.

To induce expression of Lh3 from the Tg(hsp:lh3-myc) transgene during regeneration, one larva at 5 dpf homozygous for the Tg(hsp:lh3-myc) transgene was placed in each tube in PCR strip-tubes. Tubes were incubated in a thermal cycler at 38°C for 30 min then held at 29°C until they could be removed from the tubes. Larvae were transected 4-5 h later.

Immunostaining

Larvae were stained using methods modified from those previously described (Randlett et al., 2015). Briefly, larvae were fixed in 4% PFA in PBS overnight at 4°C. Larvae were washed in PBS+0.25% Triton (PBT), incubated in 150 mM Tris-HCl (pH 9.0) for 15 min at 70°C, then washed in PBT. Larvae were permeabilized in 0.05% Trypsin-EDTA for 5 min on ice (Marquart et al., 2017), washed in PBT, and blocked in PBT containing 1% bovine serum albumin (BSA), 2% normal goat serum (NGS) and 1% dimethyl sulfoxide (DMSO), then incubated with primary and secondary antibodies overnight at 4°C in PBT containing 1% BSA and 1% DMSO. Stained larvae were stored and mounted in Vectashield (Vector Laboratories) for imaging using confocal microscopy. For static timepoint analysis of lh3 and col18a1 mutants, primary antibodies used were mouse anti-GFP (JL-8, 1:200, Takara Bio, AB_10013427). Secondary antibodies used were goat anti-mouse Alexa 488 (1:500, Molecular Probes, AB_2534088). For static timepoint analysis of macrophages/microglia, primary antibodies used were mouse anti-GFP (JL-8, 1:200, Takara Bio, AB_10013427) and rabbit anti-RFP (1:200, Abcam AB_945213), and secondary antibodies used were goat anti-mouse Alexa 488 (1:500, Molecular Probes, AB_2534088) and goat anti-rabbit Alexa 594 (1:500, Molecular Probes, AB_2534079). For static timepoint analysis of oligodendrocytes, primary antibodies used were rabbit anti-GFP (1:500, Molecular Probes, AB_221569) and mouse anti-dsred (1:500, BD Pharmingen, AB_394264), and secondary antibodies used were goat anti-rabbit 488 (1:500, Molecular Probes, AB_2576217) and goat anti-mouse Alexa 594 (1:500, Molecular Probes, AB_2534091).

Fluorescent in situ hybridization with hybridization chain reaction

lh3 (plod3), col18a1 and mbpa mRNA expression was detected using fluorescent in situ hybridization with hybridization chain reaction (HCR) (Molecular Instruments) (Choi et al., 2018). HCR probes, buffers and hairpins were purchased from Molecular Instruments. Tg(isl2b:GFP) larvae were fixed at 5 dpf with 4% paraformaldehyde in PBS overnight at 4°C. No antibody staining was required together with in situ HCR. The staining was performed as previously described (Shainer et al., 2023).

Static confocal imaging, processing and analysis

Larvae were imaged on a Zeiss 880 confocal microscope using a 20× objective. Airyscan super-resolution settings were used to obtain Figs 5G and 6B. Image stacks of optic chiasms or tecta were compressed into maximum intensity projections. All images were adjusted for brightness and contrast in Adobe Photoshop CS5, and color assigned using Fiji.

Live cell imaging, processing and analysis

Larvae were anesthetized in PTU E3 with 0.0053% tricaine then mounted in 2.5% low-melt agarose (SeaPlaque, Lonza) prepared with PTU E3 plus 0.016% tricaine ventral upwards on a glass microscopy slide. Then, a region of the jaw, including the Meckel's cartilage, palatoquadrate and the most distal parts of the basihyal and ceratohyal structures, as well as one eye, were carefully removed using forceps at 4 dpf. Care was taken so as not to damage the heart. Larvae recovered in Ringer's solution and then were kept in PTU E3 at 29°C until transected. At 5 dpf, larvae were anesthetized again, mounted in agarose and then optic nerve transections were performed. Larvae recovered in Ringer's solution and then were kept in PTU E3 at 29°C until imaged for live-cell imaging. For live imaging, larvae were mounted ventral side upwards in 35 mm glass-bottomed Petri dishes in 1.5% low-melt agarose with 0.016% tricaine. Larvae were imaged using a Leica SP8 confocal microscope with the resonant scanner on, equipped with a petri dish heater and heated water perfusion, so fresh E3 water flowed through the petri dish and the temperature was kept constant at around 28°C during the imaging session. Larvae were imaged from the ventral side of the brain, which was not covered by any jaw tissue. Image stacks were acquired every 10-15 min with a 20×1.0 NA lens. All images were adjusted for brightness and contrast, and color assigned using Fiji.

To determine the time of growth initiation extending from the proximal stump of transected RGC axons, we recorded in which frame axonal extension longer than 1 μm occurred that continued to extend for another subsequent frame.

To measure the angle of the original path before transection to the end of distinct axon fascicles, we first rotated the images so that parasphenoid cartilage was precisely aligned vertically against a straight line drawn in Fiji. A circle with a radius of 80 μm centered where RGC axons exit the eye was drawn within Fiji using a Make Oval Macro in timelapse image sequences. Axons and fascicles terminating farther than the 80 μm region of interest at 20 hpt were used for measuring. First the angle of the original path to the parasphenoid was determined using frames before Wallerian degeneration was complete. The angles of the axons and fascicles to the parasphenoid were then determined, and the difference was calculated and graphed.

Statistical analysis

Data were imported into GraphPad Prism for analysis. Statistical analysis was performed as indicated throughout the text.

Peer review history

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.205048.reviewer-comments.pdf