Circadian clocka regulates the zebrafish central neuronal Mauthner-cell axon regeneration via phosphodiesterase family pde4a

Abstract

Axon regeneration is the key to repairing spinal cord injury. Circadian rhythm plays regulatory roles in axonal regeneration. The endogenous molecular clock serves as the molecular basis for the generation and precise maintenance of circadian rhythms. CLOCK is one of the most core transcription factors in the endogenous molecular clock. However, the role of CLOCK in axon regeneration has so far remained elusive. Therefore, this study explored how the circadian gene clocka regulates the zebrafish central neuronal Mauthner cells (M-cells) axon regeneration. Using the M-cell axonal regeneration model, we found that M-cell axons exhibited circadian-phase-dependent regeneration after two-photon laser ablation, while in the clocka mutant, the circadian-phase-dependent regeneration of M-cells was deprived and their regenerative ability was inhibited. Subsequently, a combination of single-cell electroporation, single-cell capture, transcriptome sequencing, and Rolipram treatment demonstrated that clocka is required for M-cell axonal regeneration and regulates axonal regeneration through phosphodiesterase pde4a-cAMP axis. Together, these findings demonstrate that endogenous molecular clock gene clocka regulates zebrafish central M-cell axonal regeneration via pde4a-cAMP pathway and provide new insights into rhythmic regulation for central nerve repair.

Introduction

Spinal cord injury (SCI) causes permanent neurological disability due to limited axonal regeneration^1,2. In recent years, a few reports have revealed that the circadian rhythm plays regulatory roles in axonal regeneration^3,4. Circadian rhythms are approximately 24-hour physiological and behavioral cycles in organisms that regulate numerous life processes^5,6. The endogenous molecular clock serves as the molecular basis for the generation and precise maintenance of these rhythms^7,8. The core mechanism of the mammalian circadian clock relies on a transcriptional-translational feedback loop formed by the positive core regulatory components CLOCK-BMAL1 and the negative core regulatory components PER-CRY. This cyclic process drives and maintains the endogenous molecular clock, generating autonomous circadian oscillations. Previous studies have demonstrated that dorsal root ganglia (DRG) neurons possess an intrinsic molecular clock machinery and exhibit time-of-day dependent axonal regeneration after injury^3,4. Mechanistically, they mainly discussed the role of BMAL1 deficiency in the DRG neuron regeneration process^3,4. However, the regulatory roles of other circadian-regulated genes in axon regeneration and their downstream mechanisms are largely unknown. CLOCK, as another core component of the molecular clock, it has been reported to play a non-classical role in stabilizing heterochromatin, promoting tissue and cartilage regeneration⁹. Although the clocka gene has been documented as critical for maintaining circadian rhythm in zebrafish^10–12, its role in regulating axonal regeneration and the underlying mechanisms that can support axonal regeneration after neuronal injury remain elusive.

Zebrafish, as a diurnal animal, have become an ideal model for studying the circadian clock because of the conserved circadian rhythm shared with mammals^11,13–15. Mauthner cells (M-cells), a pair of central neurons located in the hindbrain of zebrafish, are characterized by their stereotyped soma morphology and a crossed axon descending in the spinal cord^16–18, and exhibit strong regenerative ability after axonal injury^18–20. Several studies have investigated the intrinsic mechanisms underlying central nervous system (CNS) neuronal repair following injury using the M-cell regeneration model after axotomy^21–24. Unlike mammalian sciatic nerve injury models, where thousands of mixed axons regenerate as a population, the M-cells exhibit precise somato-axonal matching, wherein regenerative phenotypes demonstrate strict one-to-one correspondence to their somata. This enables genetic manipulation at the single-cell level, allowing for more precise investigation of intrinsic factors influencing axonal regeneration. Moreover, as M-cells exclusively innervate the fast escape circuit in zebrafish^16,17,19,25, distinct axonal regeneration phenotypes directly determine variance in locomotor functional recovery. These intrinsic properties overcome cellular heterogeneity challenges inherent to conventional CNS axon regeneration models, thus providing an ideal model for exploring endogenous regulators of axonal regeneration.

In this study, we first analyzed single-cell transcriptomes associated with M-cell regenerative competence²⁶, revealing constitutive expression of the full complement of core circadian clock genes within M-cells. Notably, positive regulators potentially facilitate axonal regrowth. Consequently, we investigated whether M-cell axonal regeneration is governed by the endogenous circadian clocka. Using confocal in vivo imaging, we demonstrated that M-cells displayed time-of-day dependent axonal regeneration after two-photon laser axotomy over 24 hours. However, such circadian-phase-dependent axon regeneration was abolished in clocka mutants, concomitant with significantly impaired regenerative ability. By rescuing clocka in M-cells of mutants via single-cell electroporation, M-cell axonal regeneration was partially restored, which further demonstrated that clocka was involved in axonal regeneration. Mechanistically, previous studies have shown phosphodiesterase 4 (PDE4) family members are involved in regulating axonal regeneration after SCI^27–29, we found that clocka mutation eliminated oscillatory expression of pde4a and elevated its expression levels. Combining single-cell overexpression and shRNA-mediated knockdown of pde4a with in vivo single-cell capture assays, we demonstrated that it was the elevated pde4a in clocka mutant M-cells that inhibited the axonal regenerative capacity. Finally, behavioral analyses revealed that clocka mutant larvae exhibited slower recovery of motor function after axonal injury. Furthermore, treatment with the Pde4 inhibitor Rolipram rescued M-cell axonal regeneration in the mutants and subsequently promoted the recovery of their motor function.

Together, our study demonstrated that M-cells displayed a circadian-phase-dependence in axonal regeneration. The circadian gene clocka participated in regulating regeneration through the PDE4a–cAMP–CREB axis. These new findings suggest that circadian rhythms play an important role in CNS axon regeneration, and provide new insights for understanding how circadian disruption harms regeneration and designing clock-dependent treatment strategies after SCI.

Results

M-cells display time-of-day dependent axonal regeneration after laser ablation

We first aimed to determine whether M-cells possess an endogenous molecular clock and examine its potential role in axonal regeneration. Using Tg(Tol 056: EGFP) transgenic strain zebrafish with specific M-cells labeling^20,30, we performed two-photon laser axotomy at 6 days post-fertilization (dpf). Axonal regeneration was evaluated at 2 days post-injury (dpi), with cells categorized into Regeneration (Re) and Non-Regeneration (NRe) groups based on regrowth phenotypes. In vivo capture of M-cell somata followed by single-cell transcriptome sequencing (scRNA-seq) was conducted for each group (Fig. 1A). We previously obtained single-cell transcriptome sequencing data related to M-cell axonal regenerative ability²⁶. ScRNA-seq analyses revealed that firstly, all the core circadian genes were generally expressed in M-cells (Fig. 1B). We further verified their roles during axonal regeneration. Results indicated that at 2 dpi, the expression of positive regulators bmal1b and clocka significantly increased in the Re group compared to the NRe group. Correspondingly, negative core regulating genes per1b and cry2 significantly elevated expression in the NRe group (Fig. 1C). Additionally, the expression of bdnf, bhlhe40, and mycbp2 in the Re group was also higher than that in the NRe group (Fig. 1D). Among these, bdnf—a well-established neurotrophic factor protecting neuronal survival and promoting axonal regeneration after injury—shows expression closely linked to the circadian clock^31,32. The basic helix-loop-helix family member bhlhe40 (also known as dec1), can be activated by CLOCK and BMAL1 to initiate negative feedback^33,34. Furthermore, the E3 ubiquitin ligase mycbp2 has been reported to degrade Rev-erbα thereby regulating circadian rhythms³⁵. These results collectively suggest that circadian rhythm is a potential regulator of M-cell axonal regeneration, and positive feedback regulation may promote axonal regeneration. Therefore, we hypothesized that the endogenous molecular clock may influence the regenerative capacity of M-cell axons based on the timing of injury occurrence. Thus, we performed two-photon laser axotomy on 6 dpf larvae M-cell axons directly above the cloaca every 4 hours at Zeitgeber times (ZTs) 1, 5, 9, 13, 17, and 21 (i.e., hours after light onset in a 14 h light/10 h dark environment), corresponding to 9 a.m., 1 p.m., 5 p.m., 9 p.m., 1 a.m., and 5 a.m.), and observed axonal regeneration two days later at the same time (Fig. 1E, F). Confocal imaging results showed that M-cell axonal regeneration exhibited certain time-of-day dependence during the day, among which axons injured at ZT9 had the longest regeneration length while ZT21 had the shortest regeneration length, compared with other time points (Fig. 1G, H). Moreover, statistical results showed that the average regenerative capacity of M-cell axons was overall higher during the light phase than during the dark phase (Fig. 1I). However, age is one of the factors affecting axonal regeneration. We found that there was no significant difference in axonal regeneration capacity even between the most age-disparate groups, ZT1 and ZT21 (Fig. 1H). To directly examine the effect of age differences on axonal regeneration, we performed two-photon laser axotomy on 5 dpf and 7 dpf larvae M-cell axons at ZT1, and observed axonal regeneration two days later at the same time (Additional file: Supplementary Fig. S1A). Imaging results showed that there was no significant difference in the length of axonal regeneration after injury among larvae at 5 dpf, 6 dpf, and 7 dpf (Additional file: Supplementary Fig. S1B). These results excluded the influence of a 20-hour age difference on M-cell axonal regeneration.

Fig. 1. Circadian rhythm is a regulator of M-cell axonal regeneration.

A Timeline for Mauthner single-cell axonal regeneration transcriptome sequencing. By Figdraw. B Basal expression of core circadian genes in M-cells. The horizontal axis represents transcripts per million (TPM). Line at the mean. C Statistical graphs of the expression of positive regulators bmal1b, clocka, and negative core regulators per1b and cry2 in the Re group compared to NRe group. D Statistical graphs of the expression of circadian rhythm related genes bdnf, bhlhe40, mycbp2 elevated in Re group compared to NRe group (Mann‒Whitney test, *p < 0.05). E Timeline for two-photon laser axotomy on M-cells in vivo. By Figdraw. F Representative image of an injured M-cell axon. The asterisk indicates the injury site, which is above the cloacal pore. G Representative images of M-cell axonal regeneration at ZT9 and ZT21 after two-photon laser ablation. The asterisks indicate the injury site and the arrows indicate the regenerating axon terminals. Scale bars, 100 μm. (H) Statistical graphs of M-cell axonal regeneration length from the injury site during 24 hours (One-way ANOVA, ZT1 versus ZT5: p = 0.9991, ZT1 versus ZT9: p = 0.9541, ZT1 versus ZT13: p > 0.9999, ZT1 versus ZT17: p = 0.9995, ZT1 versus ZT21: p = 0.2796, ZT5 versus ZT9: p = 0.8196, ZT5 versus ZT13: p = 0.9976, ZT5 versus ZT17: p > 0.9999, ZT5 versus ZT21: p = 0.4912, ZT9 versus ZT13: p = 0.9697, ZT9 versus ZT17: p = 0.8399, ZT9 versus ZT21: p = 0.0420, ZT13 versus ZT17: p = 0.9985, ZT13 versus ZT21: p = 0.2421, ZT17 versus ZT21: p = 0.4649, *p < 0.05, ns, not significant, n = 12 biologically independent larvae for each group). Bar graphs represent the mean ± SEM (ZT1: 488.4 ± 37.02 μm, ZT5: 474.7 ± 24.64 μm, ZT9: 520.4 ± 27.83 μm, ZT13: 491.4 ± 21.44 μm, ZT17: 476.2 ± 20.36 μm, ZT21: 408.9 ± 22.55 μm). (I) Statistical graphs of M-cell axonal regeneration length from the injury site during light and dark phases (student’s t-test, *p < 0.05, n = 48 biologically independent larvae for daytime group and n = 24 for nighttime group). Bar graphs represent the mean ± SEM.

Taken together, these results indicated that the intrinsic circadian clock regulated M-cell axonal regeneration and demonstrated that M-cells displayed circadian-phase-dependent axonal regeneration after two-photon laser ablation.

Generation of clocka mutant zebrafish

The clock gene, which is only present as a single copy in humans, is present in two copies in zebrafish. Since scRNA-seq results indicated higher baseline expression levels of clocka in M-cells, we then focused on its role in axonal regeneration. Firstly, we constructed clocka mutant zebrafish using CRISPR-Cas9 technology. Using the online design tool CRISPOR (http://crispor.tefor.net/)³⁶, we designed targeting sgRNA sequences on the third exon within the core coding sequence of clocka (Fig. 2A). After in vitro synthesis, we mixed them with Cas9 protein and microinjected into the single-cell stage embryos for genome editing (Fig. 2B). Genomic sequencing results showed a 38-base insertion in third exon of F2 zebrafish, causing premature termination of translation (Fig. 2C). To ensure the knockout of clocka, we designed specific primers upstream and downstream of the target site in the third exon of clocka. Primer pair 1 was located in the intronic regions flanking the third exon. Therefore, amplified product fragment 1 encompassed the entire edited region (Additional file: Supplementary Fig. S2A). In contrast, the reverse primer of primer pair 2 spanned the mutation site within the targeted region, enabling normal amplification in the wild-type, whereas mutations in the targeted site led to failed amplification (Additional file: Supplementary Fig. S2B). Using these primers, we performed a PCR assay to amplify the genomic DNA extracted from both wild-type and clocka mutant zebrafish. Subsequently, the PCR products were analyzed via agarose gel electrophoresis. Image results clearly demonstrated that the amplified Fragment 1 in the clocka mutant was approximately 40 bp longer than that of the WT/AB. While Fragment 2 was successfully amplified in the WT/AB, it failed to amplify in the clocka mutant due to mutations in the target region (Fig. 2D). Furthermore, analysis using the bioinformatics tool (https://smart.embl.de/smart/change_mode.cgi) revealed a premature stop codon in the clocka mutant, resulting in a truncated protein of only 68 amino acids and indicating the loss of the entire key functional domain HLH-PAS (Fig. 2E). These results demonstrated that we successfully established a homozygous clocka mutant zebrafish line.

Fig. 2. Generation of clocka mutant zebrafish.

A Schematic of the Cas9-sgRNA target site in the third exon of clocka. The target site was highlighted in the red box. In the mutant sequence alignment, inserted nucleotides were denoted in red font, and the deleted nucleotides were represented by red dashed lines. (B) Schematic of microinjection sgRNA-Cas9 complex. By Figdraw. C Sanger sequencing results of WT/AB and clocka mutant zebrafish lines. The mutant sequencing result showed a 38-basepair insertion. D Agarose gel electrophoresis analysis of PCR fragments confirmed clocka knockout in mutant zebrafish. The amplified Fragment 1 in the clocka mutant was approximately 40 bp longer than that of the WT/AB. While the Fragment 2 was successfully amplified in the WT/AB, it failed to amplify in the clocka mutant due to mutations in the target region. E Bioinformatics analysis (https://smart.embl.de/smart/change_mode.cgi) indicated that the mutation affects the HLH-PAS-PAS functional domain in Clocka protein. The clocka mutants translated only 68 normal amino acids, whereas WT/AB fish expressed 892 amino acids. F Statistical graphs of qRT-PCR analysis showed the mRNA expression of core circadian genes bmal1b, per1b, per2, and cry1ba during 24 hours. The genes were expressed relative to β-actin. The rhythmic mRNA expression was analyzed by the JTK-CYCLE method. Total RNA was extracted from 6 dpf WT/AB and clocka mutant larvae at 4-hour intervals for 24 hours under normal light-dark condition (two-way ANOVA, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3 biologically independent samples for each group, each sample contained 50 larvae). Bar graphs represent the mean ± SEM. G Statistical graphs of the amplitudes of core circadian genes bmal1b, per1b, per2, and cry1ba (student’s t-test, ****p < 0.0001, n = 3 biologically independent samples for each group, each sample contained 50 larvae). H The locomotor activity of WT/AB and clocka mutant larvae was tested beginning at 6 dpf 9 a.m. under normal light-dark condition during 48 hours (n = 8 biologically independent larvae for each group). I Statistical graphs of locomotor activity of WT/AB and clocka mutants at J)daytime and K nighttime (student’s t-test, ***p < 0.001, ns, not significant, n = 8 biologically independent larvae for each group). Bar graphs represent the mean ± SEM.

We further verified the impact of the clocka mutation on circadian rhythms in zebrafish larvae and additionally demonstrated the successful establishment of the clocka mutant. Previous studies have shown that clocka mutation causes disruption of the circadian clock in zebrafish larvae¹², and our results once again proved this finding. We first detected the expression of core circadian rhythm genes in clocka mutant zebrafish by qRT-PCR. We sampled every four hours, and the results showed that compared with the control group, clocka mutants exhibited significant differences in the expression levels of core clock regulatory genes per1b, per2, cry1ba, and bmal1b at certain time points within 24 hours, with the expression levels of per1b and cry1ba both decreasing within a day (Fig. 2F). Further analysis by JTK-CYCLE method indicated that the amplitudes of oscillatory expression for all four clock genes were significantly reduced in mutants (Fig. 2G), and the oscillation period of bmal1b was shortened. In addition to altering clock gene expression patterns, circadian disruption also affects normal sleep-wake cycles^13–15. Therefore, we analyzed free-swimming behavior of clocka mutants over 48 hours (Fig. 2H). Results revealed significantly reduced activity in clocka mutants compared to WT/AB (Fig. 2I). Notably, clocka mutants showed weakened light sensitivity, with significantly decreased daytime activity but no significant difference in nighttime movement (Fig. 2J, K). Together, these results indicate the successful establishment of clocka mutants and validate clocka plays roles in zebrafish larval circadian rhythms.

clocka regulates M-cell axonal regeneration in vivo

To investigate the effect of clocka mutation on axonal regeneration in M-cells, we crossed Tg(T056: EGFP) strain with clocka mutants to obtain Tg(T056: EGFP); clocka^-/- zebrafish, achieving in vivo visualization of mutant M-cells (Fig. 3A). Next, we examined body length, M-cell soma size, and axon length from the cloaca at 6 dpf in Tg(T056: EGFP); clocka^-/-. Results showed no significant differences in body length, soma area, or axon length between clocka mutants and controls (Additional file: Supplementary Fig. S3). These data demonstrated that the clocka mutant had no detectable impact on the morphological development of the M-cell axon itself. Afterwards, we performed two-photon laser axotomy on M-cell axons dorsal to the cloaca in larvae at ZT9 and ZT21, respectively, and conducted imaging 2 dpi to quantify regeneration length (Fig. 3B). In vivo confocal imaging results showed that clocka mutant larvae exhibited a reduction in the length of M-cell axonal regeneration, indicating that the regenerative capacity of the clocka mutant is significantly lower than that of controls. It is worth noting that no significant difference in regeneration length was observed between ZT9-injured and ZT21-injured M-cell axons in clocka mutants (Fig. 3C, D), indicating that the knockout of the molecular clock eliminated the circadian-phase-dependent regeneration of M-cell axons, further suggesting that the regenerative capacity of M-cells is regulated by endogenous molecular clocks.

Fig. 3. clocka mutation attenuates M-cell axonal regeneration and its recovery of motor function.

A Hybridization of the transgenic line: Tg(Tol 056: EGFP) and clocka mutants were crossed for two consecutive generations to obtain Tg(Tol 056: EGFP); clocka ^-/- lines. By Figdraw. B Timeline for two-photon laser axotomy in vivo. By Figdraw. C Representative images of WT/AB and clocka mutants M-cell axonal regeneration at ZT9 and ZT21. The asterisks indicate the injury site and the arrows indicate the regenerating axon terminals. Scale bars, 100 μm. D Statistical graphs of WT/AB and clocka mutants M-cell axonal regeneration length from the injury site at ZT9 and ZT21 (two-way ANOVA, *p < 0.05, **p < 0.01, ****p < 0.0001, n = 12 biologically independent larvae for each group). Bar graphs represent the mean ± SEM. E Timeline for C-start response behavioral test. By Figdraw. F Representative images of the original orientation and maximal turning angle position of the WT/AB and clocka mutant larvae in the uninjured (θ, α) and injured (θ’, α’) groups. Red lines indicate the heading direction. Scale bars, 1 mm. G Statistical graphs of maximal angle for WT/AB and clocka mutants C-start response (two-way ANOVA, ***p < 0.001, ****p < 0.0001, ns, not significant, n = 9 biologically independent larvae for each group). Bar graphs represent the mean ± SEM. (H) A series of representative images of movement trajectory from WT/AB and clocka mutant larvae in the uninjured and injured groups. The asterisks indicate the maximal turning angle position. Scale bars, 1 mm. I Statistical graphs of time to maximal angle for WT/AB and clocka mutants (two-way ANOVA, **p < 0.01, ****p < 0.0001, ns, not significant, n = 9 biologically independent larvae for each group). Bar graphs represent the mean ± SEM.

M-cells innervate zebrafish fast escape behavior, being one of the rare examples where single cells determine a behavior^16,25. Since the escape behavior resembles the letter C, it is called the C-start response escape behavior. Therefore, we used high-speed photography to analyze C-start response in clocka mutants after axonal injury to investigate locomotor functional recovery (Fig. 3E). Results demonstrated that compared to controls, clocka mutants had significantly reduced maximal turning angle of C-start response at 2 dpi (Fig. 3F, G), and time to reach maximal turning angle was significantly prolonged (Fig. 3H, I).

To further validate the role of clocka in M-cell axonal regeneration, we performed single-cell-level specific rescue of clocka via single-cell electroporation to observe its effect on axonal regeneration. Here, we used a dual-plasmid expression system of CMV-GAL4 and UAS-clocka-mCherry (Fig. 4A). First, we verified the effectiveness of clocka overexpression plasmids through microinjection: by co-injecting dual plasmids into the single-cell stage embryos (Additional file: Supplementary Fig. S4A), we observed expression of the reporter gene mCherry red fluorescence in the larval brain at 4 dpf (Additional file: Supplementary Fig. S4B). Results showed distinct punctate red fluorescence in the brain, and qRT-PCR results confirmed successful clocka overexpression (Additional file: Supplementary Fig. S4C). After validating plasmid efficacy, we co-transfected the endotoxin-free dual-plasmid system into unilateral M-cell somata of 4 dpf larvae using single-cell electroporation. Axons successfully expressing red fluorescence were selected 12 h post-electroporation. Subsequently, we performed two-photon laser ablation dorsal to the cloaca on mCherry-positive axons at 6 dpf and observed axonal regeneration at 8 dpf by confocal imaging in vivo (Fig. 4B, C). In vivo confocal imaging results demonstrated that although single-cell complementation of clocka enhanced the regenerative capacity of M-cell axons in clocka mutants, the regenerated axon length remained significantly shorter than that of WT controls (Fig. 4D, E). Furthermore, we wondered whether clocka rescue could restore the circadian-phase-dependence of axonal regeneration. To test this, we performed two-photon laser ablation on successfully rescued axons in clocka mutants at ZT9 and ZT21 of 6 dpf and observed axonal regeneration at the corresponding time points on 8 dpf via in vivo confocal imaging (Additional file: Supplementary Fig. S4D). However, there is no significant difference in the regeneration length of clocka-rescued M-cell axons between ZT9 and ZT21 (Additional file: Supplementary Fig. S4E).

Fig. 4. Single-cell complementation of clocka rescues M-cell axonal regeneration.

A Construction of the clocka expression system. Plasmids expressing only mCherry served as the control vector. By Figdraw. B Timeline for single-cell electroporation, axotomy, and regeneration imaging. By Figdraw. C Representative confocal images of positive expression (red) in Mauthner single-cell electroporation. D Representative images of single-cell clocka rescue and empty vector in M-cell axonal regeneration of clocka mutants. The asterisks indicate the injury site and the arrows indicate the regenerating axon terminals. Scale bar, 50 μm. E Statistical graphs of M-cell axonal regeneration length after single-cell clocka complementation (One-way ANOVA, *p < 0.05, **p < 0.01, n = 7 biologically independent larvae for complementation).

The above results indicated that clocka regulates axonal regeneration in M-cells. Mutation of clocka significantly impaired post-injury axonal regeneration and the recovery of M-cell function. Single-cell restoration of clocka partially rescued the axonal regenerative capacity of mutant M-cells but failed to restore their circadian-phase-dependent regeneration.

clocka regulates genes associated with the second messenger metabolic process

To determine how clocka affects axonal regeneration and its potential downstream mechanisms, high-throughput RNA sequencing was performed using 6 dpf clocka mutants and WT/AB zebrafish. Volcano plot analysis revealed 709 significantly upregulated and 939 significantly downregulated differentially expressed genes in clocka mutants compared to controls (Fig. 5A). Furthermore, GO enrichment analysis demonstrated significant changes in genes related to intracellular signal transduction, cyclic compound binding, and purine nucleotide binding in clocka mutants (Fig. 5B) and revealed several significantly enriched biological processes. The top categories included: immune system process, regulation of cyclic nucleotide metabolic process, oxidoreductase activity, second-messenger-mediated signaling, circadian rhythm, regulation of hydrolase activity, and response to cAMP (Fig. 5C). KEGG analysis similarly identified enrichment of related signaling pathways (Additional file: Supplementary Fig. S5A). Among these, previous studies have reported that immune and oxidoreductase-related processes, as well as circadian rhythm-related processes, are significantly altered in clocka mutants^10,12, which further supports the credibility of our sequencing results. The remaining top-ranked biological processes were all associated with cyclic purine nucleotide metabolism. Critically, second messengers cAMP and cGMP play essential roles in neurogenesis, growth cone navigation, and axonal outgrowth^37,38. In particular, cAMP has been reported to promote axonal regeneration and motor functional recovery after M-cell injury in zebrafish¹⁸. Therefore, we hypothesized that the clocka mutation might affect axonal regeneration through cAMP metabolism. Previous studies have shown that the cAMP hydrolysis-specific phosphodiesterase PDE4 family is the main subtype that functions in the CNS and is involved in regulating axonal regeneration processes^27–29,39. These findings are consistent with our sequencing data. Thus, we reconfirmed the sequencing results by qRT-PCR and found that the mRNA levels of axon regeneration-related pde4a, pde4ba, and pde4d were all significantly increased (Fig. 5D). To further verify the expression of pde4 family in M-cells, we analyzed pde4 family gene expression in M-cells using scRNA-seq. Analysis showed that all subtypes of pde4 were expressed in M-cells (Additional file: Supplementary Fig. S5B). In addition, previous studies have shown that activation of the CREB pathway is a key step in cAMP regulation of axonal regeneration^40,41. Western blot results also showed that the activity of the clocka mutant Creb signaling pathway was significantly reduced (Fig. 5E, F). Together, these results suggested that clocka mutation may regulate axonal regeneration through the Pde4-cAMP-Creb axis in zebrafish.

Fig. 5. clocka regulates genes associated with the second messenger metabolic process.

A Volcano plot shows downregulated and upregulated genes in clocka mutant zebrafish. B GO enrichment analysis of DEGs categorized the genes into specific functional groups. C The top-ranked biological processes enriched in clocka mutants. The horizontal axis represents -lg P-value. D Statistical graphs of qRT-PCR reconfirmation of pde4 family genes for WT/AB and clocka mutants. The genes were expressed relative to β-actin. Total RNA was extracted from 6 dpf WT/AB and clocka mutant larvae (student’s t-test, **p < 0.01, ***p < 0.001, ns, not significant, n = 3 biologically independent samples for each group, each sample contained 50 larvae). Bar graphs represent the mean ± SEM. E Validation of p-Creb protein expression in clocka mutants compared with that of the WT/AB. F Quantification of p-Creb expression in (D). The p-Creb were expressed relative to Gapdh. The protein expressions were quantified by Image J software (student’s t-test, **p < 0.01, n = 3 biologically independent samples for each group, each sample contained 50 larvae).

clocka mutation eliminates the rhythmic expression of pde4a and upregulates its expression

The phosphodiesterase Pde4 family has been identified as a therapeutic target for SCI, but little is known about its rhythmic expression. We first analyzed 24-hour transcriptome sequencing data of 6 dpf zebrafish larvae. Results showed that among axon regeneration-related subtypes in the pde4 family, only pde4a exhibited oscillatory expression (Fig. 6A and Additional file: Supplementary Fig. S5C). Therefore, we confirmed the analysis through qRT-PCR. JTK-CYCLE analysis indicated pde4a mRNA indeed displayed significant oscillations within 24 hours. Subsequently, we examined pde4a expression in clocka mutants. qRT-PCR results demonstrated that compared to controls, pde4a expression lost its oscillation in mutants and was significantly upregulated throughout the day (Fig. 6B). Moreover, clocka plasmid was injected into the embryo to rescue the expression of clocka in mutants, qRT-PCR analysis showed it could significantly downregulate the level of pde4a in clocka mutants (Additional file: Supplementary Fig. S5D). These results indicated that pde4a could be regulated by clocka in zebrafish.

Fig. 6. clocka mutation suppresses M-cell axonal regeneration by upregulating pde4a.

A Statistical graphs of oscillatory expression of pde4a analyzed by larval 24-hour RNA-seq. The rhythmicity was analyzed by the JTK-CYCLE method (p = 0.0053). B qRT-PCR validation of pde4a mRNA expression during 24 hours for 6 dpf clocka mutants (red) and WT/AB (black) larvae. The genes were expressed relative to β-actin. The rhythmic mRNA expression was analyzed by the JTK-CYCLE method (WT/AB, p = 0.00019, clocka mutants, p > 0.05). Total RNA was extracted from 6 dpf larvae at 4-hour intervals for 24 hours under normal light-dark conditions (two-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n = 3 biologically independent samples for each group, each sample contained 50 larvae). C Schematic of in vivo single-cell capture on M-cell somata. By Figdraw. D Representative images of in vivo single-cell capture on M-cell somata. E Statistical graphs of qRT-PCR analysis of WT/AB and clocka mutants single M-cell level pde4a mRNA expression relative to β-actin. (Student’s t-test, **p < 0.01, n = 10 biologically independent samples for clocka mutants and n = 7 biologically independent samples for WT/AB. Each sample is derived from a M-cell soma). F Construction of the pde4a expression system. Plasmid expressing only mCherry served as the control. By Figdraw. (G) Timeline for single-cell electroporation, axotomy, and regeneration imaging. By Figdraw. H Representative images of M-cell axonal regeneration after single-cell overexpression pde4a in WT/AB. The asterisks indicate the injury site, and the arrows indicate the regenerating axon terminals. Scale bar, 50 μm. I Statistical graphs of WT/AB axonal regeneration length after single-cell overexpression pde4a (student’s t-test, ***p < 0.001, n = 7 biologically independent larvae for overexpression pde4a and n = 6 biologically independent larvae for control vector). J Representative images of M-cell axonal regeneration after single-cell knockdown pde4a in WT/AB and clocka mutants. The asterisks indicate the injury site, and the arrows indicate the regenerating axon terminals. Scale bar, 100 μm. K Statistical graphs of M-cell axonal regeneration length after single-cell knockdown pde4a in WT/AB and clocka mutants (two-way ANOVA, **p < 0.01, ***p < 0.001, n = 6 biologically independent larvae for each group). All the bar graphs represent the mean ± SEM.

We previously developed a protocol similar to patch-clamp that enables in vivo capture of high-quality single soma samples from M-cells²⁶. To further confirm pde4a expression levels in M-cells of clocka mutants, we performed in vivo single-cell capture on M-cells somata of clocka mutants at 2 dpi using this protocol (Fig. 6C). Imaging results showed the originally swollen and bright soma was completely aspirated by the specially designed electrode pipette to capture (Fig. 6D). Captured single-cell samples underwent lysis followed by in vitro reverse transcription. The captured single-cell samples were lysed and subjected to in vitro reverse transcription, followed by cDNA integrity testing. Afterwards, strict contamination screening was conducted on samples with qualified cDNA integrity (Additional file: Supplementary Fig. S5E). Cell-type purity analysis was conducted by comparing the proportion of the M-cell marker gene nefmb with the glial marker gene mbp and the red blood cell marker gene hbae3. Ultimately, samples with a M-cell marker gene expression ratio of over 97% were retained (Additional file: Supplementary Fig. S5F). Finally, we detected pde4a mRNA levels in quality-controlled single-cell samples. qRT-PCR results demonstrated significantly elevated pde4a expression in clocka mutants M-cells during axonal regeneration (Fig. 6E). It is worth noting that the expression of pde4d also increased in M-cells, but there was no significant difference in the expression level of pde4ba (Additional file: Supplementary Fig. S5G). These results further indicated that the clocka mutation leads to an increase in the expression of pde4a in M-cells.

clocka mutation suppresses M-cell axonal regeneration by upregulating pde4a

Given that pde4a is the only oscillatory expression of subtype associated with axonal regeneration, we further validated the impact of pde4a on M-cell axonal regeneration. Similarly, we constructed pde4a overexpression plasmids in vitro and verified their effective expression via microinjection (Fig. 6F and Additional file: Supplementary Fig. S6A, B). Using single-cell electroporation, we co-transfected them with CMV-GAL4 plasmid into unilateral M-cell soma of 4 dpf larvae. Then, we performed two-photon laser axotomy on electroporation-positive M-cell axons and proceeded to image the regenerated length at 2 dpi (Fig. 6G). Imaging results showed that overexpressing pde4a significantly decreased axon regeneration length in wild-type M-cells (Fig. 6H, I). While in clocka mutants, further overexpressing pde4a significantly reduced axon regeneration success rate. Compared to overexpressing an empty vector, overexpressing pde4a resulted in two-thirds of axons failed to regenerate successfully within two days post-injury (Additional file: Supplementary Fig. S6C–E). To further pinpoint the restriction of pde4a on axonal regeneration, we used shRNA to specifically knockdown pde4a expression in M-cells^22,42. To test shRNA knockdown efficiency, we injected the miR-shRNA expressing plasmid into the animal pole of single-cell zebrafish embryos via microinjection. We selected larvae with red fluorescence in the brain under fluorescence microscopy four days later, extracted total RNA, and detected pde4a expression by qRT-PCR. Results showed shRNA2 significantly reduced pde4a mRNA expression (Additional file: Supplementary Fig. S6F–H). Then, we transfected it into 4 dpf larvae M-cells via single-cell electroporation. Two days later, we performed two-photon laser axotomy on mCherry-positive axons and quantified regeneration lengths at 8 dpf. In vivo imaging results demonstrated that shRNA significantly improves the axonal regeneration capacity of M-cells, while clocka knockout weakened this promoting effect (Fig. 6J, K). In summary, these results showed that pde4a expression negatively regulates M-cell axonal regeneration. Furthermore, knocking down pde4a can partially rescue the M-cell axonal regeneration deficiency of clocka mutant, indicating that clocka regulates M-cell axonal regeneration through pde4a.

Pde4 inhibitor Rolipram rescues clocka mutants M-cell axonal regeneration and promotes its recovery of motor function

To further validate the regulatory role of the clocka-pde4a axis on M-cell axonal regeneration, we treated zebrafish with Rolipram^43,44, a specific inhibitor of the Pde4 family. Previous studies have shown Rolipram crosses the blood-brain barrier and promotes axonal regeneration and functional recovery after SCI in rats both in vitro and in vivo by inhibiting PDE4 activity to reduce cAMP degradation²⁷. We first performed two-photon laser ablation on the 6 dpf larvae M-cell axons, and immediately treated it with different concentrations of Rolipram after injury. Two days after treatment, the regeneration length was quantified (Fig. 7A). In vivo imaging results showed both 30 μM and 50 μM Rolipram significantly promoted post-injury axon regeneration compared to vehicle control, but 30 μM Rolipram had the best therapeutic effect (Additional file: Supplementary Fig. S7A–C), so this concentration was selected for subsequent experiments. Therefore, we performed two-photon laser ablation at ZT9 and ZT21 on 6 dpf WT/AB and clocka mutants M-cells, immediately administered 30 μM Rolipram via immersion post-injury, and quantified regeneration lengths at corresponding times two days later. In vivo imaging demonstrated Rolipram treatment significantly enhanced M-cell axonal regeneration capacity and rescued regenerative ability in clocka mutants. However, it is worth noting that Rolipram treatment eliminated the circadian-phase-dependent regeneration of M-cell axons and did not restore this circadian dependence in clocka mutants (Fig. 7B–E).

Fig. 7. Pde4 inhibitor Rolipram rescues clocka mutants M-cell axonal regeneration deficiency and promotes its recovery of motor function.

A Timeline for Rolipram treatment. Larvae were immediately exposed to Rolipram for 48 hours after laser ablation to M-cells until imaging. By Figdraw. B Representative images of WT/AB M-cell axonal regeneration at ZT9 and ZT21 after 30 μM Rolipram treatment. The asterisks indicate the injury site and the arrows indicate the regenerating axon terminals. Scale bars, 100 μm. C Statistical graphs of Rolipram-treated WT/AB axonal regeneration length from the injury site (two-way ANOVA, **p < 0.01, ****p < 0.0001, ns, not significant, n = 9 biologically independent larvae for each group). D Representative images of clocka mutants M-cell axonal regeneration at ZT9 and ZT21 after 30 μM Rolipram treatment. The asterisks indicate the injury site and the arrows indicate the regenerating axon terminals. Scale bars, 100 μm. E Statistical graphs of Rolipram treated the clocka mutant axonal regeneration length from the injury site (two-way ANOVA, ****p < 0.0001, ns, not significant, n = 9 biologically independent larvae for each group). F Timeline for Rolipram-treated clocka mutants C-start response behavioral test. By Figdraw. G Representative images of the original orientation and maximal turning angle position of Rolipram-treated clocka mutants in the uninjured (θ, α) and injured (θ’, α’) groups. Red lines indicate the heading direction. Scale bars, 1 mm. H Statistical graphs of maximal angle for Rolipram-treated clock mutants C-start response (two-way ANOVA, ***p < 0.001, ns, not significant, n = 9 biologically independent larvae for each group). I A series of representative images of movement trajectory from Rolipram treated clocka mutants in the uninjured and injured groups. The asterisks indicate the maximal turning angle position. Scale bars, 1 mm. J Statistical graphs of time to maximal angle for Rolipram-treated clocka mutants (two-way ANOVA, *p < 0.05, ns, not significant, n = 9 biologically independent larvae for each group). All the bar graphs represent the mean ± SEM.

We similarly assessed locomotor functional recovery after M-cell axon ablation in Rolipram-treated clocka mutants. We performed two-photon laser axotomy on 6 dpf clocka mutants M-cells, immediately administered 30 μM Rolipram via immersion post-injury, and detected C-start response behavior using high-speed photography at 2 dpi (Fig. 7F). Results showed rolipram treatment significantly increased the maximal turning angle during C-start response in clocka mutants (Fig. 7G, H) and shortened the time to reach maximal turning angle (Fig. 7I, J). These results further demonstrated that clocka regulates M-cell axonal regeneration through pde4a.

Discussion

Our work revealed the role of circadian gene clocka in M-cell axonal regeneration in vivo. Multiple findings indicated that endogenous molecular clocks are involved in regulating the regeneration process of various tissues in the body, including the skin^45,46, intestine⁴⁷, and hematopoietic system⁴⁸. Recently, there have been studies on endogenous circadian rhythms regulating axonal regeneration^3,4, which are limited to peripheral neurons. This is mainly due to mammalian CNS axonal limited intrinsic regenerative capacity and injury-induced alterations in the external microenvironment⁴⁹, resulting in the loss of axonal regenerative ability. In contrast, CNS axons in fish are permissive for regeneration^18,49,50, which provides the possibility of exploring the regulation of CNS axonal regeneration by endogenous circadian rhythms and offers new insights into overcoming the inherent limitations of mammalian CNS regeneration. Previous studies identified circadian regulation as the most significantly enriched pathway in transcriptomic datasets associated with DRG sensory neuron regenerative competence³, confirming their intrinsic molecular clock and time-of-day dependent regenerative capacity. In this study, M-cell single-cell RNA-seq similarly revealed expression of all core circadian genes and demonstrated time-of-day-dependent axonal regeneration phenotypes, thereby expanding our understanding of circadian-controlled neural repair to CNS motor neurons.

Research on the regulation of axonal regeneration by specific circadian genes has only focused on the effect of Bmal1^3,4. Thus, understanding the regulatory roles of other clock genes in axonal regeneration is also of great significance. Studies have shown that CLOCK is involved in regulating tissue and cartilage regeneration in mice⁹. Furthermore, knocking out bdnf can weaken the oscillatory expression of zebrafish clocka³¹, suggesting a close relationship between clocka and axonal regeneration. This study extends the findings of Bmal1-mediated time-of-day dependent axon regeneration in mouse peripheral neurons to the regeneration of zebrafish central motor neuronal M-cell axons. We unsurprisingly identified some notable differences. Firstly, De Virgiliis et al. demonstrated that mouse DRG sensory axon regenerative capacity was weakest at ZT8 (3 p.m.) and peaked at ZT20 (3 a.m.), with independent work confirming enhanced nighttime regeneration compared to daytime^3,4. Intriguingly, this pattern is diametrically opposed to our findings in M-cells, where regeneration peaks at ZT9 (5 p.m.) and reaches its nadir at ZT21 (5 a.m.). This divergence might stem from the distinct diurnal/nocturnal habits of mice versus zebrafish⁵¹, or it may be caused by inherent regeneration differences between central axons and peripheral axons, which requires further investigation. Secondly, our results demonstrated that single-cell restoration of clocka in mutants failed to restore their circadian-phase-dependent regeneration. We attributed this discrepancy primarily to differences in experimental paradigms. The previous rescue experiments were conducted in cultured mouse DRG neurons in vitro, which inherently removed neurons from their native cellular environment. This represented an idealized experimental setup free from interference by other cells. In contrast, our clocka rescue was performed in vivo and in situ, where M-cell axonal regeneration occurred within a much more complex microenvironment, making the restoration of rhythmic regeneration more challenging. Thirdly, in contrast to our finding that clocka mutation significantly impairs zebrafish M-cell axonal regeneration, Halawani D et al. demonstrated that Bmal1 deletion enhances axon regeneration via Tet3-mediated epigenetic regulation⁴. We proposed that this discrepancy primarily stemmed from functional divergence between Bmal1 in mice and clocka in zebrafish. In mice, Bmal1 was considered the only non-redundant core circadian gene, suggesting it might occupy a higher hierarchical position within the molecular clock network. In zebrafish, however, circadian gene families exhibited greater functional redundancy, resulting in a more complex regulatory network. Furthermore, as demonstrated in that study, Bmal1 regulates DRG neuron axonal regeneration through an epigenetic mechanism involving the restriction of Tet3 expression and 5hmC modifications. Given the well-established epigenetic differences in injury and regeneration between the CNS and the PNS^52,53, the injury-induced epigenetic responses might have differed substantially, ultimately leading to distinct regulatory outcomes. More broadly, interspecies variations also contributed to the observed differences in regenerative phenotypes. Notably, the mutation of clocka eliminates circadian-phase-dependent regeneration and impairs overall regenerative capacity in M-cell axons, similar to the phenotype observed in Bmal1^fl/fl mouse DRG neurons³. The single-cell restoration of clocka partially rescued the axonal regenerative capacity of mutant M-cells. We considered that the primary reason was that the single-cell rescue of M-cells was performed in a global clocka knockout background, and thus could not rescue the effects of the clocka mutation in other cells, such as glial cells and inflammatory cells, on the systemic or local microenvironment. In addition, clocka knockout might have caused irreversible changes in downstream pathways, such as epigenetic modifications and potential developmental defects. Even with later restoration of clocka, it was too late to compensate for these functional deficits. However, the fact that single-cell restoration alone could partially restore axonal regeneration further demonstrated the importance of intrinsic clocka in M-cell axonal regeneration. Taken together, our results once again demonstrated that axonal regeneration is a complex process regulated by circadian rhythms. This knowledge also provides novel insights into rhythmic regulation of axonal regeneration.

As a unique model for CNS axonal regeneration, the M-cell axon exhibits strong regeneration after two-photon laser ablation. Two-photon laser axotomy can induce very point-precise injury to only one side of the M-cell axon without damaging other tissues outside the axon^19,20, so the inflammatory response after damage is very minor and limited, which is consistent with our observation (Additional file: Supplementary Fig. S8A, B). In addition, we examined whether macrophages at the injury site displayed time-dependent recruitment. We performed imaging of macrophages recruited to the M-cell injury site at ZT9 and ZT21 in 2 dpi larvae, respectively. The imaging results showed no significant difference in recruitment between the two time points (Additional file: Supplementary Fig. S8C, D). Therefore, we did not explore the effect of inflammatory response on regenerative rhythm, but focused on the impact of endogenous molecular clock on M-cell axonal regeneration. However, the injury microenvironment also contains many other molecules, such as extracellular matrix components and inflammatory factors. Due to limitations of the model, we are currently unable to precisely sample the injury site. Therefore, we cannot fully demonstrate whether the injury microenvironment is influenced by circadian rhythm.

In this study, CRISPR-Cas9 technology was used to generate homozygous clocka mutant strains of zebrafish. Consistent with previously reported clocka mutants¹², clocka mutation did not lead to early survival defects and morphological developmental abnormalities in zebrafish larvae. Further in vivo imaging results also showed that clocka knockout did not affect the morphology and development of M-cells. Unlike the previous results of increasing neutrophil migration but not eliminating migration rhythmicity in the clocka mutant¹², the mutation of clocka not only weakens the regenerative ability of M-cell axons but also abolishes their circadian-phase-dependent regeneration. This may relate to varying cell-type dependence on the endogenous molecular clock. Axonal regeneration is a complex process strictly dependent on the precise expression program of endogenous genes. The mutation of clocka disrupts the core rhythm, preventing the relevant regeneration programs to be activated normally, and inevitably leading to the loss of circadian-phase-dependent optimization in axonal regeneration.

Previous studies have confirmed the important role of cAMP in the regeneration of M-cell axonal regeneration¹⁸. The cAMP hydrolysis-specific PDE4 family-specific inhibitor Rolipram can promote axonal regeneration after rats SCI in vitro and in vivo²⁷ by inhibiting the expression of the PDE4A, PDE4B, and PDE4D subfamilies²⁸. Here, based on previous findings and our results, we hypothesized that the clocka mutation may regulate M-cell axonal regeneration through the Pde4 family. By analyzing the high-throughput transcriptome sequencing results and qRT-PCR validation of larvae during 24 hours, we were surprised to find that only the pde4a exhibited oscillatory expression throughout the day, and its expression pattern aligned with the daily fluctuations in the regenerative capacity of M-cell axons. However, both of these oscillations were eliminated in clocka mutants. In addition, our results indicated that the expression level of pde4a in M-cells is negatively correlated with their regenerative ability. Knocking down pde4a only partially compensated for the regeneration deficiency of clocka mutants M-cells, indicating that pde4a is downstream of clocka and suggesting the possibility of other downstream pathways, which needs further investigation. An intriguing finding from our qRT-PCR analysis was the divergent regulation of pde4 family members following injury. While pde4a, pde4ba, and pde4d were upregulated, pde4cb was significantly downregulated. This pattern likely reflects the functional specialization within the PDE4 family⁵⁴. In a previous study on the role of the PDE4 family in thyroid carcinoma (THCA), the results showed that PDE4C was significantly upregulated in THCA tissues, while other PDE4 subtypes were significantly downregulated⁵⁵. This finding aligns with our results and suggests that PDE4C and other PDE4 subtypes may have different expression regulatory mechanisms. Furthermore, a previous study has demonstrated that PDE4C expression is positively correlated with TP53 expression and TP53 signaling pathway activity through overexpression and knockdown experiments⁵⁶, while TP53 alone was sufficient to promote sensory axon and optic nerve regeneration in vivo^57,58. Therefore, in addition to causing significant upregulation of pde4a, pde4b, and pde4d, which are known to be negatively associated with regeneration, the clocka mutation may further impair M-cell axonal regeneration by suppressing the non-canonical Tp53 pathway through significantly reduced pde4cb expression. Notably, the oscillatory expression of the Pde family is not isolated. Previous studies have shown that pde6c and pde6h also exhibit oscillatory expression in larvae⁵⁹. However, since the Pde6 family is mainly expressed in the visual system, this article did not delve into them. Moreover, the PDE4 family has been reported to play a central role in various diseases, such as neurodegenerative diseases⁶⁰, skin diseases^61,62, and autoimmune diseases^62,63. These results hint that the circadian clock may also play roles in these diseases.

Our study demonstrated that Rolipram treatment promoted the regeneration of M-cell axons, which is consistent with previous results obtained in rodents^27–29. It is worth noting that we found that although drug therapy can enhance the regenerative ability of M-cell axons, it also deprives time-of-day dependent axon regeneration. This may be due to the sustained increase in downstream signaling molecule cAMP levels during treatment with Rolipram, thereby overriding the circadian fine-tuning of axonal regeneration. In clocka mutants, although Rolipram can rescue the deficiency of axonal regeneration ability, it cannot restore the rhythm of regeneration, further suggesting that cAMP levels may be the key to generate and maintain axonal regeneration chronobiology. Once the endogenous circadian regulation within neurons is impaired, time-of-day-dependent regeneration becomes irrecoverable. Critically, these results demonstrate that while Rolipram enhances axonal regeneration, it does so at the expense of spontaneous time-of-day dependent regeneration. In particular, these results were similar to those from our single-cell rescued clocka of M-cell axons, in which later interventions restored baseline regenerative capacity but did not restore circadian-phase-dependent regeneration, suggesting the challenge of achieving precision in neural repair. This loss of intrinsic circadian regulation may lead to unforeseen neurological risks, further highlighting the harm of rhythm disorders to axonal regeneration and meriting further investigation.

This study, however, also bears some limitations. Firstly, although we identified and validated circadian regulation of M-cell axonal regeneration with demonstrable time-of-day dependence, the inherent technical limitations of this model, including the challenge of capturing single cell in vivo and the inability to establish in vitro culture, hinder us from getting 24-hour single-cell transcriptome sequencing data on M-cell axonal regeneration in the time dimension. Secondly, a zebrafish strain with conditional knockout of clocka in M-cells will be generated to investigate its specific role in axon regeneration. Finally, although we provide direct evidence for the regulation of axonal regeneration by clocka through pde4a, we have not elucidated the precise interaction mechanism between clocka and pde4a, which requires further exploration in the future.

Taken together, we found that the endogenous molecular clock is an important factor in regulating the axonal regeneration of zebrafish central motor neuronal M-cells. Our study provides direct evidence for the regulation of axonal regeneration and recovery of motor function by the circadian clock core regulatory gene clocka through pde4a (Fig. 8). This study aims to better understand CNS axonal regeneration and provide new insights into the search and development of novel targets for circadian clock based on neural repair.

Fig. 8. Overview of clocka regulates M-cell axonal regeneration via pde4a.

clocka mutation elevated the expression level of pde4a in M-cells, thereby weakening the ability of axonal regeneration and the recovery of motor function. Single-cell complementation and Pde4 family inhibitor could rescue the M-cell axonal regenerative ability in clocka mutants. By Figdraw.

Methods

Zebrafish husbandry and animal procedures

The zebrafish strains used in this study included wild-type (WT)/AB, Tg(Tol 056: EGFP) (also known as Et(T2KHG)zf206, obtained from RIKEN, Japan)³⁰, which labels M-cells with EGFP for in vivo two-photon laser ablation, and Tg(mpeg1: mCherry), which labels macrophages with mCherry in vivo. Adult zebrafish were maintained in a fish housing system with a constant temperature of 28.5°C and a 14/10 light/dark (14-h light and 10-h dark) cycle. After natural spawning, embryos were collected in culture medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂, 0.33 mM MgSO₄, and 0.1% methylene blue, pH 7.0) and incubated in a 28.5 °C constant temperature chamber. From 1 dpf to 2 dpf, 0.003% N-phenylthiourea (PTU, Sigma‒Aldrich, USA) was added to the culture medium to inhibit pigmentation. Larvae used for experiments were euthanized by hypothermal shock⁶⁴, fully immersed in the ice-water mixture for more than 60 minutes, and then transferred to sodium hypochlorite solution (500 mg/L). The University of Science and Technology of China (USTC) Animal Resources Center and the University Animal Care and Use Committee provided the rules and regulations that all experiments were implemented. All procedures were performed in accordance with the ARRIVE 2.0 guidelines and approved by the Committee on the Ethics of Animal Experiments of the USTC (permit no. USTCACUC1103013).

Genome editing

CRISPR-Cas9 technology was used for genome editing for the generation of clocka mutants in zebrafish. We designed a targeting sgRNA sequence, GCCGGGCAACACCCGCAAGA, on the third exon within the core coding sequence of clocka by using the online design tool CRISPOR (http://crispor.tefor.net/)³⁶. Next, an in vitro transcription template containing the T7 promoter, target sequence, and scaffold sequence was obtained through PCR amplification. Then, MEGAshortscript™ T7 Transcription Kit (AM1354, Thermo Fisher, USA) was used to generate sgRNA using this template. The sgRNA-Cas9 ribonucleoprotein complex was formed by combining sgRNA (300 ng/μL) with Cas9 protein (300 ng/μL) (PC1350, Inovogen, China) at 37 °C for 10 minutes. Then, the ribonucleoprotein complex was microinjected into single-cell embryos of wild-type zebrafish to generate clocka mutant chimeras. Afterwards, homozygous clocka mutant zebrafish were produced through hybridization.

Identification of clocka mutants

We performed a PCR assay by Phanta Flash Master Mix (P510, Vazyme, China) to amplify the genomic DNA extracted from both wild-type and clocka mutant zebrafish by using specific primers upstream and downstream of the target site in the third exon of clocka. Subsequently, the PCR products were analyzed via 2% agarose gel electrophoresis. Then, the gel was imaged with a gel imaging system (GenoSens 2100, Clinx, China). All primers used are listed in the Additional file: Supplementary Table S1.

Plasmid construction

Every overexpression system consisted of two plasmids, CMV-Gal4 and UAS-mCherry. UAS-clocka-mCherry and UAS-pde4a-mCherry were constructed in vitro. The coding sequence (CDS) of the target gene was amplified from a zebrafish cDNA library using Phanta Flash Master Mix (P510, Vazyme, China). The amplified CDS was ligated to an empty UAS-mCherry plasmid digested with EcoRI restriction enzyme (AI11964A, Takara, Japan) by using a homologous recombination kit (CU201, Transgene, France).

To knock down the expression of pde4a, we designed shRNAs employing the primary miR-30 backbone based on the shRNA design tool (https://www.genscript.com). All primers used are listed in the Additional file: Supplementary Table S1.

Single-cell electroporation in vivo

Larvae at 4 dpf were anesthetized with 160 mg/L ethyl 3-aminobenzoic methanesulfonate (MS-222, Sigma‒Aldrich, USA) and immobilized in 1% low-melting agarose (Sangon, China) on a specially made agar plate. A specially-designed micropipette tip pulled by SUTTER P-97 (Sutter Instrument, USA) was filled with endotoxin-free plasmids and moved to the side of the M-cell soma under the microscope by motorized crossed roller bearing micromanipulator (MX7600, Siskiyou, USA). Subsequently, a sinusoidal voltage amplified by a power amplifier (ATA-304C, Aigtek, China) was applied to the glass electrode fixed on the needle holder (Axoporator 800 A, Molecular Devices, USA) to deliver the plasmid, whose concentration is 300 ng/μL each, into the soma. Positive axons were screened 12 h post-electroporation under a fluorescence microscope and prepared for subsequent laser ablation.

Two-photon laser axotomy

Before laser ablation in vivo, larvae at 6 dpf were anesthetized with MS222 and immobilized in 1% low-melting agarose. A laser scanning confocal microscope (LSM980, Carl Zeiss, Germany) equipped with a two-photon laser was used at a wavelength of 800 nm and an intensity of 18–30% to precisely ablate the M-cell axon above the cloacal cavity under a 25 × oil dipping objective lens.

In vivo imaging

MS222-anesthetized larvae were embedded in 1% low-melting agarose and imaged for M-cell axons and somata by using a 40 × water dipping objective lens on a confocal microscope (FV1000, Olympus, Japan). Z-stack images were captured every 3 μm. For axonal regeneration, larvae were imaged at 2 dpi. The length of axonal regeneration was defined as the straight-line distance from the site of axonal injury to the tip of the longest regenerating axonal branch.

In vivo single-cell capture assay

The single-cell in vivo capture technology recently developed in our laboratory can obtain high-quality single-cell samples of larval M-cells. The feasibility analysis and detailed process of the single-cell capture assay are as described previously²⁶. Firstly, larvae at 2 dpi were anesthetized with MS222 and immobilized in 2% low-melting agarose in a chamber. Next, a specially-designed micropipette was used under 60 × water dipping objective lens to completely extract the ablated axonal soma under negative pressure. Afterwards, the captured single-cell samples were lysed and reverse transcribed in vitro, and the cDNA integrity of the samples was detected by the bioanalyzer system (Agilent 2100, Agilent, USA). Then, samples with qualified cDNA integrity underwent strict contamination analysis by using qRT-PCR (Quantitative Real-time PCR). The contamination analysis was conducted by comparing the proportion of the M-cell marker gene nefmb with the glial marker gene mbp and the red blood cell marker gene hbae3. Lastly, samples with an M-cell marker gene expression ratio of over 97% were retained.

Free-swimming behavior assay

To detect the motion rhythm of larvae, their free-swimming ability was assessed using the Viewpoint system (ZebraBox, Viewpoint, France) at 28.5 °C during 48 hours. Behavioral test started at 9 a.m., and the 6 dpf larvae were placed in the system 30 minutes before the test to adapt to the environment. The test procedure was set as normal photoperiod (14 h light/10 h dark) and maintained at 28.5 °C. An automatic video tracking system was used to monitor the movement of larvae for 2 consecutive days and record it using Zebralab 3.11 software (Viewpoint, France).

C-start response behavior assay

C-start response behavior of larvae was recorded by a high-speed camera (M230, Revealer, China) with a frame rate of 1000 fps. 2 dpi larvae were placed in a petri dish with 28.5 °C culture water and moved to the platform with appropriate light around to adapt to the environment in advance. C-start response was triggered by computer-controlled speakers emitting sound stimuli, and the entire movement process was recorded by camera software (Revealer, China), including the maximal turning angle and the time to reach the maximal turning angle. The maximal turning angle is defined as the angle at which the larval midline rotates before and after escape behavior.

Transcriptome sequencing analysis

Total RNA was extracted from clocka mutants and WT/AB zebrafish at 6 dpf; each sample included 50 larvae and underwent three biological replicates. Total mRNA was enriched using Oligo (dT) beads, then cleaved into short fragments using lysis buffer, and reverse transcribed into cDNA using the NEB Next Ultra RNA Library Illumina Preparation Kit (Cat. No. #7530, New England Biolabs, USA). The cDNA library was sequenced by Sangon Biotech (Shanghai) Co., Ltd. on the Illumina sequencing platform. Using the reference genome as the reference sequence, HISAT2 was employed to align the quality-controlled sequencing reads to the reference genome. The alignment results were then statistically analyzed using RSeQC. Subsequent differential expression analysis was performed using DESeq. The differentially expressed genes (DEGs) were identified with the thresholds set as: qValue < 0.05 and absolute value of FoldChange > 2. The Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of DEGs were conducted using the clusterProfiler.

The single-cell transcriptome sequencing process and results related to M-cell axonal regeneration were derived from the previous description²⁶.

RNA extraction and qRT‒PCR

Total RNA was extracted from 50 larvae using TRIzol reagent (Takara, Japan). Afterwards, samples were ultrasonically shattered in an EP tube, and total RNA was extracted by adding chloroform. Subsequently, 1 μg of total RNA was reverse transcribed into cDNA using HiScript II qRT SuperMix II (R223, Vazyme, China). qRT-PCR application was performed using ChamQ Universal SYBR qPCR Master Mix (Q711, Vazyme, China) in a 10 μL system on a real-time quantification system (Light Cycler 96, Roche, Switzerland). The mRNA expression levels of target genes were normalized to those of housekeeping gene β-actin using the 2^–ΔΔCt method to calculate relative expression level. Each sample was detected with three technical replicates, and the average of three technical replicates represents the expression level of each gene. All primers used are listed in the Additional file: Supplementary Table S1.

Protein extraction and western blotting

WT/AB and clocka mutant larvae at 6 dpf were collected and lysed with RIPA (Radio Immunoprecipitation Assay Lysis) buffer supplemented with a protease inhibitor and phosphatase inhibitor (Sangon, China). The lysates were centrifuged, and the collected supernatant was kept on ice. The concentration of each protein sample was assessed on a microplate reader using the BCA Protein Assay Kit (Beyotime, China). Samples were boiled for 4 minutes and run on a 10% SDS-PAGE gel along with 5× loading buffer and then transferred to a PVDF membrane. After incubation in 5% BSA and TBST for 1 hour at room temperature, the membranes were washed once with TBST and incubated with antibodies against p-Creb (1:1000, Abmart, China) or Gapdh (1:2000, HuaAn, China). Subsequently, the membranes were incubated with secondary goat anti-rabbit antibodies (1:5000, Proteintech, China) for 1 hour at room temperature. The blots were washed with TBST three times and visualized by the enhanced chemiluminescence (ECL) system (Thermo Fisher, USA). The densities of the bands were quantified by ImageJ software (National Institutes of Health, USA) and normalized to protein Gapdh.

Pharmacological treatment

Continuous immersion treatment was performed on larvae using the PDE4 inhibitor Rolipram (abs817932, Absin, China). Larvae at 6 dpf were immediately exposed to 10, 30, or 50 μM Rolipram for 48 hours after two-photon laser ablation to M-cells until in vivo imaging. The control group was treated with an equal concentration of vehicle DMSO (D12345, Thermo Fisher, USA).

Statistical analysis

Graphic drawing, data statistics, and statistical significance were analyzed with GraphPad Prism 9 (San Diego, USA), Adobe Photoshop 2023, and Figdraw (Home for Researchers, China). Unpaired two-tailed Student’s t-tests, one-way analyses of variance (ANOVAs), two-way ANOVAs, Mann‒Whitney test were used in this study. Detailed statistical data, such as sample size and methods, were explained in the figure legends. The rhythmicity of the data was evaluated by JTK-CYCLE analysis. The analysis of single-cell transcriptome sequencing data from M-cells was as described previously. Larvae were randomly assigned to different groups. All experiments were repeated at least three times. A threshold level of significance was set at p < 0.05, and statistical significances were marked as follows: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, and ns, not significant (p > 0.05). Error bar represented the mean ± standard error of the mean (SEM).

Author contributions

The conception and design of this article, as well as the manuscript draft and revision, were completed by Zhao ZA, Song Z, and Hu B. Zhao ZA was a major contributor in writing the manuscript, acquisition of data, and data analysis. Han AL performed an in vivo single-cell capture assay. Fan DG, Li KQ, and Zhou JH also participated in the acquisition of data. All authors read and approved the final manuscript.

Data availability

The reference genome sequences for zebrafish (Danio rerio) used in this study were obtained from the GRCz12tu Primary Assembly. The Mauthner single-cell axonal regeneration transcriptome dataset used in this study is not publicly available due to the source article not yet being published, but is available from the corresponding author on reasonable request. Other data generated during this study are included in this published article and its supplementary information files. Further information can be revealed if desired by a reasonable request to the corresponding author.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41536-026-00488-5.