Stat3 has a different role in axon growth during development than it does in axon regeneration after injury

As a pleiotropic factor, signal transducer and activator of transcription 3 (STAT3) has been implicated in both neural development and regeneration. Here, we demonstrate that Stat3 plays distinct roles in PLL axon embryonic growth and regeneration using zebra�sh posterior lateral line (PLL) axons. Mutation of stat3 rather than Jak/Stat inhibition resulted in PLL axons truncation during embryonic growth. We found that Stat3 was co-localized with mitochondria in PLL axon and the ATPase activity and mitochondrial membrane potential was decreased in stat3 mutant and mitochondria agonist partially rescued PLL axon growth in stat3 mutant indicating Stat3 regulated PLL axon growth through mitochondrial Stat3 function. By contrast, Jak/Stat signaling inhibitors retarded PLL axon regeneration and Schwann cell migration. Stat3 promotes PLL axon regeneration mainly through regulating Schwann cell migration via Jak/Stat signaling. We provide a new mechanism of Stat3 in axon growth and regeneration and suggest that Stat3 is a promising therapeutic target for neural regeneration.


Introduction
Axon extension is a critical process for both embryonic neural development and regeneration after injury.
The extension of axons requires both intrinsic and extrinsic factors [1].During axon extension, signi cant energy is required for forming the abundant dynamic actin laments in growth cones, microtubule-based axonal transport of organelles and proteins, and synthesis of protein and phospholipids used for membrane expansion.Oxidative phosphorylation (OXPHOS) in mitochondria is the main source of ATP synthesis.Recent investigations have shown a role for mitochondria in axon extension [2][3].Besides providing energy, axon regeneration requires additional processes such as axon debris clearance, scar gap bridging, and guiding the migration of axons.All of these processes require glia cells wrapping around the regenerating axons [4].
Signal transducer and activator of transcription 3 (STAT3) is a conserved protein with multiple essential functions [5].The most widely studied function of Stat3 is in mediating the signaling from extracellular cytokines and growth factors.Stat3 is phosphorylated by Janus kinase (JAK) and other tyrosine kinases at a tyrosine in the SH2 domain, after which it dimerizes and translocates to the nucleus to activate genes involved in in ammation, cell proliferation, and/or survival [6].Recently, Stat3 was also found to bind to complex I and II of ETC, and ATP synthase in the inner mitochondrial membrane and to promote the electron transport chain (ETC) and ATP synthesis [7][8][9][10].Some studies provide evidence that Stat3 is involved in axon extension through Jak/Stat signaling.
Deletion of suppressor of cytokine signaling 3 (SOCS3), an inhibitor for Jak, promotes robust regeneration of injured optic nerve axons in mice [11].In contrast, over-expression of socs3 attenuates optic nerve regeneration in zebra sh [12].Nerve regeneration is also regulated by the activation of the Jak/Stat pathway in Schwann cells [13][14].However, other studies have shown that mitochondrial Stat3 is essential for axon extension.Mitochondrial Stat3 promotes neurite outgrowth in PC12 cells and cultured dorsal root ganglion (DRG) neurons [15][16].In mouse retinal ganglion cells (RGC), fusing Stat3 with a mitochondria targeting sequence (MTS) promotes further enhancement in axon regeneration compared to the constitutively active form of Stat3 [17].However, the details of Stat3 in axon development and regeneration remain unclear.
The lateral line is a speci c sensory system present in sh and amphibian.The posterior lateral line (PLL) ganglion is located posterior to the otic vesicle and relays information from the lateral line sensory organs, known as neuromasts, to the medial octavolateralis nucleus (MON) in the hindbrain.During embryonic development, the axons of the PLL nerves extend from the ganglion to the tip of tail over the course of approximately 20 h [18].In addition, the PLL axons can regenerate spontaneously after neuroaxotomy [19].This makes the PLL an ideal model system for axonal growth studies in vivo [20].Stat3 is expressed speci cally in lateral line, including the neuromasts and lateral line ganglia [21][22].In this study, we generated a zebra sh stat3 mutant and showed its PLL axons failed to extend properly.We identi ed a cell-autonomous role of Stat3 in PLL axon growth through a PLL neuron speci c rescue assay.Unexpectedly, we found that mitochondrial Stat3 rather than Jak/Stat signaling was required for PLL embryonic axonal growth.In contrast, Jak/Stat signaling was involved in PLL axon regeneration probably through promoting Schwann cell migration.Thus, Stat3 plays distinct roles in PLL axon embryonic growth and regeneration.

Zebra sh strains and husbandry
The AB, TgBAC(neurod1:EGFP) [23] and Tg(mbp:EGFP) [24] transgenic lines used in this study were purchased from the China Zebra sh Resource Center.Adult zebra sh were maintained at 28.5°C on a 14 h:10 h light/dark cycle, fed with artemia twice per day.Embryos were collected by natural mating with a male to female ratio of 1:1, raised at 28.5°C in blue water and staged according to hours-or days-post fertilization (hpf or dpf) as described previously [25].All procedures for zebra sh were approved by the Animal Ethics Committee of Shanghai Ocean University.

Whole-mount in situ hybridization
The templates for anti-sense RNA probe synthesis were produced by PCR from embryonic cDNA.All primers for probe synthesis are listed in supplementary information Table S1.The T7 promoter sequence was added to the 5'-end of anti-sense primers.The RNA probes were synthesized using T7 RNA polymerase with DIG RNA Labeling Mix (Roche).The larvae were xed in 4% formaldehyde at 4°C overnight, dehydrated with 100% methanol and stored at -20°C.Samples were rehydrated into PBST (1xPBS with 0.1% tween-20) gradually, bleached with 10% H 2 O 2 , permeabilized with 10 µg/µL proteinase K for 10 min, then post-xed in 4% PFA for 20 min.Hybridization was carried out at 70°C overnight, followed by post-hybridization washing, blocking with 2% bovine serum albumin (BSA) and 2% sheep serum, incubation with anti-DIG antibody (Roche), and stained with NBT/BCIP solution (Roche).

Generation of zebra sh stat3 mutant line
Zebra sh stat3 mutants were generated by Clustered Regularly Interspaced Short Palindromic repeats (CRISPR)/ CRISPR-Associated Protein 9 (Cas9) according to the published protocol [26].In brief, a single guide RNA (sgRNA) targeting to stat3 exon 3 (5′-GGTGCTGCTTGATGCGCCGCAGG-3′) was designed using the Zebra sh Genomics track in UCSC Genome Browser [27].The target-speci c oligo (supplementary information Table S1) including a T7 promoter and the universal oligo with the crRNA-tracRNA sequence were used for producing the template for sgRNA synthesis.The sgRNA and capped Cas9 mRNA were synthesized by in vitro transcription.Approximately 1.4 nL of sgRNA (about 50 pg) and capped Cas9 mRNA (about 300 pg) was co-injected into wild-type embryos at the one-cell stage.The genomic DNA was extracted from each injected embryo using alkaline lysis.The insertion and deletion (In-Del) e ciency was evaluated by uorescent PCR using the stat3 genotyping primer set (supplementary information Table S1), followed with capillary electrophoresis [28].The embryos with high In-Del e ciency were established as founder sh and raised to adulthood.The F1 were obtained by crossing the founder with AB, n-clipped and genotyped as described above.A pairwise heterozygous F1 with selected mutation was incrossed to get the stat3 mutant.

qPCR
The embryos derived from a stat3 heterozygous mutant incross were genotyped by PCR analysis of genomic DNA extracted from embryonic n clips at 48 hpf and separated according their genotypes.
Total RNAs of 20 wild-type or stat3 mutants was extracted using TRIzol reagent (Invitrogen).cDNA of wild-type and stat3 mutants were generated by the ProtoScript First Strand cDNA Synthesis Kit (NEB) using random primers.qPCR was performed with the Luna Universal qPCR Master Mix (NEB) on an ABI PRISM 7000 Real-Time PCR System (Applied Biosystems, USA).The eef1a1l1 gene was used as endogenous control.The qPCR primers for eef1a1l1, atf3, c-jun, klf6a, klf7a, socs3a, socs3b, tub1a, tub1b, tub1c, and thymosin gene refer to previous study [12,29].The qPCR primers for stat3 and il6 are listed in supplementary information Table S1.All reactions were performed in technical triplicates.
Relative expression level of each gene was calculated using 2 −∆∆Ct method and normalized to the mean of control group.The results represent biological replicates, including the standard error of the mean.

Immuno uorescence and TMRE staining
The larvae were xed in freshly-made formaldehyde as described above.After three washes with PBST for 20 min each, samples were incubated in blocking solution (2% BSA and 2% normal goat serum in PBST) for 1 h, and in mouse acetylated-tubulin antibody (1:1000 dilution, Sigma-Aldrich) at 4°C overnight.After three washes with PBST for 20 min each, samples were incubated in goat anti-mouse IgG conjugated with Alexa Fluor 488 (1:200, Invitrogen) for 2 h at room temperature, then three washes with PBST.Samples were visualized using a Zeiss Axio Observer uorescent microscope (Carl Zeiss), the genomic DNA of each xed sample was extracted with the FastPure FFPE DNA Isolation Kit (Vazyme) and the genotype was determined as above.
Zebra sh larvae at 3 dpf were incubated in 2.5 µM mitochondria speci c dye tetramethylrhodamine ethyl ester (TMRE, diluted with Holtfreter's buffer) for 1 h in the dark.After washing with Holtfreter's buffer twice, the larvae were immobilized in 2% low-melting-point agarose (LMP, Fisher Scienti c) and visualized using the 568 nm excitation channels.

Generation the constructs
All constructs used in this study were generated by seamless cloning strategy using the ClonExpress II One Step Cloning Kit (Vazyme).The Tol2 vector backbone were ampli ed from pHuC:farnesylated-TdTomato [30], the full length of the coding sequence (CDS) of stat3, socs3a, and cox8a was ampli ed from zebra sh embryonic cDNA, and 5 kb-neurod promoter in zebra sh was ampli ed from zebra sh genomic DNA as described [31].Firstly, the pHuC:TdTomato-stat3 was generated by fusing stat3 CDS downstream of tdTomato in pTol2-HuC:farnesylated-tdTomato.pNeurod:tdTomato-Stat3 was constructed by replacing the HuC promoter in pHuC:tdTomato-Stat3 with zebra sh 5 kb-neurod promoter.pNeurod:tdTomato-Cox8a were generated by replacing stat3 CDS in pNeurod:tdTomato-Stat3 with zebra sh Cox8a CDS.These constructs were used to express stat3 or display the mitochondria in PLL axon.
The plasmids for mRNA synthesis were modi ed from pcDNA3.1.tdTomato was cloned into the multiple cloning site of pcDNA3.1 as a control.The CDS of socs3a and stat3 was fused downstream of tdTomato, respectively.Constructs for expressing mutated Stat3 Y708F and Stat3 EE435-436AA were generated by modifying pcDNA3.1-Stat3using Mut Express II Fast Mutagenesis Kit V2 (Vazyme).Similarly, constructs for expressing Stat3 tagged with nuclear localization sequence (NLS) and mitochondrial targeting sequence (MTS) were generated by amplifying the whole pcDNA3.1-Stat3with primers containing NLS or MTS coding sequence at their 5'-end and re-ligating.All the constructs were veri ed by sequencing.

Transmission electron microscopy (TEM)
The wild-type and stat3 mutant zebra sh was euthanized with tricaine methanesulfonate (MS222, Sigma) and xed using 2.5% glutaraldehyde (diluted in PBS) at 4°C overnight.After rinsing in 0.1 M sodium cacodylate buffer, the samples were post-xed in 1% osmium tetroxide for one hour, and then rinsed in 0.1 M sodium cacodylate buffer again.Larvae were then dehydrated with a graded series of ethanol, in ltrated with and embedded in epoxy resin, and polymerized in a 70°C oven for 48 h.After identifying areas of interest, ultra-thin sections of 100 nm thickness for TEM analysis were then cut using an ultramicrotome and stained with 33% methanolic uranyl acetate for 15 minutes and lead citrate (Electron Microscopy Sciences) for 7 minutes.Digital electron micrographs were collected using a HITACHI H-7700.

mRNA and plasmid injection and quanti cation of axon growth
The capped mRNA of Tol2 transposase, TdTomato, Socs3a, wild-type, mutant, and tagged Stat3 were synthesized with linearized plasmids using T7 mMessage mMachine kit (Invitrogen).For overexpression essays, one nanogram of the capped mRNA was injected into the cytoplasm of each embryo derived from stat3 heterozygous mutant incross, with TgBAC(neurod1:EGFP) background, respectively.To get the transient transgene, the mixture of Tol2 plasmid with Tol2 transposase mRNA was used for injection.
Following injections, the PLL nerves of each larva were visualized and quanti ed according the position of PLL axon terminal at 2 dpf.And each larvae injected was genotyped as above.

Drug treatments
In this study, inhibitors of Jak/Stat signaling, mitochondrial respiratory chain, and mitochondrial boosters were used to treat zebra sh embryos.All compounds were dissolved in DMSO and stored at -80°C in small aliquots.Respectively, the embryos of TgBAC(neurod1:EGFP) were treated with 80 µM Pyridone 6 (MCE, HY-14435), 400 µM S3I-201 (MCE, HY-15146), 5 µM Stattic (MCE, HY-13818), and 50 nM rotenone (MCE, HY-B1756) diluted with Holtfreter's buffer from 24 to 48 hpf.Equal volume of DMSO was used as control.The length of PLL axons of the treated larvae was quanti ed based on the somite position.The progeny of stat3 heterozygous mutant incrosses using the TgBAC(neurod1:EGFP) background were treated with 50 µM DCHC (Sigma, D5569) from 24 to 48 hpf.The percentage of PLL axons reaching the tip of tail was calculated.PLL axons of TgBAC(neurod1:EGFP) or Tg(mbp:EGFP) background were amputated between the 3rd and 8th somites using electric neurectomy methods at 3 dpf as described previously [20].Then, the larvae were treated with the Jak/Stat inhibitor, Pyridone 6 (80 µM) for 2 days during PLL axon regeneration.The PLL nerves were imaged using the 488 nm excitation channels and the length of PLL axons were quanti ed as above.10.ATP synthase activity ATP synthase activity of wild-type and stat3 mutant embryos was measured using the ATP Synthase Speci c Activity Microplate Assay Kit (Abcam) following the manufacturer's protocol.Brie y, 25 embryos of wild-type and stat3 mutants were homogenized in 200 µL of homogenization buffer.After freezing and thawing, the homogenates were centrifuged at 16,000 rpm and the solubilized fraction was collected.
After adjusting the protein concentration to 5.5 µg/µL, the sample was extracted in Detergent Buffer on ice for 30 min followed by centrifugation at 16,000 rpm for 20 min.Fifty µL of supernatant was loaded in each well of a pre-coated microplate and kept at 4°C overnight.Solution 1 was used as blank reference.Each well was rinsed twice with Solution 1 and incubated with Lipid Mix for 45 min.Reagent Mix was added to the reaction and ATP synthase activity was measured at OD340 at 1 min intervals for 1-2 h.For quantifying the mount of ATP synthase, each well was loaded with Solution A and incubated at room temperature for 1 h.After washing with Solution 1 twice, the wells were loaded with Solution B and incubated at room temperature for 1 h.Development Solution was added to each well, and the amount of ATP synthase was measured at OD405 at 1 min intervals for 30 min.

Axonal mitochondria localization analysis of Stat3
To display the transport of Stat3 and mitochondria in PLL axons, Neurod:TDT-Stat3 and Neurod:EGFP-Cox8 plasmids together with Tol2 transposase mRNA, were co-injected in the one-cell stage embryos of AB.The embryos expressing EGFP and tdTomato in a single PLL ganglion cell were mounted in 1.5% low melting point agarose, anesthetized with 0.02% tricaine on a glass bottom plate and imaged with 488 nm and 568 nm excitation channels in sequence at 72 hpf.Photographs were collected at fastest speed for 1 min.The mitochondrial transport in PLL axons was analyzed using kymograph analysis in the ImageJ FIJI.Kymographs were generated from each imaging session.

PLL nerve amputation
Wild-type or stat3 mutant larvae with the TgBAC(neurod1:EGFP) and Tg(mbp:EGFP) transgenes were anesthetized in MS-222, and the PLL axons were amputated between the 3rd and 8th somites with an eye scalpel under a dissecting microscope.The larvae after axonectomy were recovered in Holtfreter's buffer and imaged under uorescent inverted microscope.

Statistical analysis
Statistical analysis and graphs were performed using R with the packages "rstatix" and "ggpubr".Normality of the data was tested using Shapiro-Wilk test.Two groups comparisons were analyzed using Student's t-test or Wilcoxon Rank Sum Test as described in the gure legends.For multiple groups comparisons, differences between treatment groups and their matched controls were tested, and adjusted p-value were calculated using the False Discovery Rate (FDR, Benjamini & Hochberg) controlling method.Differences of proportions for categorical variables were tested using Fisher's exact test.For all comparisons, p or adjusted p < 0.05 was considered as statistically signi cant.

Results
1. stat3 is required for PLL axon growth in the zebra sh embryo It has been shown in previous studies that stat3 is enriched in the cranial ganglion and lateral line system during the larval stages [21][22].We examined the expression of stat3 during lateral line axon growth and found that stat3 was enriched in anterior lateral line (ALL) and PLL at 32 hpf, disappearing after 48 hpf (Fig. 1A).The dynamic expression of stat3 in lateral line ganglion suggests that it is probably involved with the early formation of the lateral line neurons.
To study the role of stat3 in PLL ganglion development, we knocked-out stat3 in zebra sh using CRISPR/Cas9 and created a zebra sh stat3 mutant line with 5 bp-deletion resulting in a premature stop codon.The putative truncated protein has an incomplete N-terminal domain and is missing the coiledcoil, DNA-binding, SH2 (Src Homology 2), and transcription activation domains (TAD) compared to wildtype Stat3 (Fig. 1B).qPCR results indicated that wild-type the expression of stat3 is signi cantly decreased in the stat3 homozygousmutants, which suggested nonsense-mediated mRNA decay was induced by the mutation (Fig. 1C).We also found stat3 mutants had curved bodies at approximately 20 dpf (data not shown) as described previously [22].These results indicated that stat3 was successfully knocked-out in the mutant line.
To determine the effect of a stat3 mutation on the PLL ganglion, we labeled axons using anti-acetylatedtubulin antibodies.By 48 hpf, the PLL axons reached the caudal neuromasts in wild-type.However, the PLL axons in stat3 mutants only migrated approximately 70% of the distance of wild-type siblings (Fig. 1D).We crossed TgBAC(neurod1:EGFP)transgenic zebra sh, whichlabels both the cell body and axon of lateral line ganglion with enhanced green uorescent protein (EGFP) [23], with stat3 mutation to enable tracing PLL axon growth using time-lapse imaging.The initiation of PLL axons occurred at approximately 20 hpf for both the stat3 mutant and wild-type siblings.Growth speed didn't show signi cant differences by 20-38 hpf, however, the pioneer growth cone of PLL axons stopped migrating and got thinner in stat3 mutants at about the 20 th -22 nd somite while the PLL axons reached to caudal neuromasts in wild-type (N=7) (Fig. 1E-F; Video S1 and S2).We also checked the PLL axons in stat3 mutants after 2 dpf.Although the PLL axons in stat3 mutants became thinner during migration, they didn't show obvious degeneration or retraction even by 17 dpf (Supplementary Fig. S1).
To elucidate whether axon truncation in stat3mutantswas due to loss of neurons innervating the caudal neuromasts speci cally, we counted the cell bodies in PLL ganglion in the TgBAC(neurod1:EGFP) transgenics at 30 hpf.The PLL ganglion in stat3 mutantcontained similar average number of neurons (22.0 ± 0.97) compared to wild-type (22.3±0.63)(Supplementary Fig. S2).The stat3 mutation does not affect the number of cell bodies in the PLL ganglion.
In addition, we counted the terminal branches under TgBAC(neurod1:EGFP) transgenic background and the stat3 mutants displayed fewer terminal branches in PLL axons innervating the trunk neuromasts, however, it didn't show the swollen axon terminals seen in the jip3 nl7 mutant described previously (Fig. 1G-H) [31].These data suggest that the underlying mechanism of PLL axon truncation in stat3 mutants is mechanistically different from the jip3 nl7 mutants.
The axons of the dorsal lateral line (DLL) also project from the PLL ganglion.We found the length of DLL axons in stat3 mutant was also signi cant shorter than wild-type, and the branch of DLL in stat3 mutant was fewer than wild-type (Supplementary Fig. S3A).Blocking Jak/Stat signaling with morpholinos targeting stat3 and jak1 affected path nding and axon branching in spinal nerves [32].However, in stat3 mutants, neither the ventral nor the dorsal projections of the spinal nerve were signi cantly changed (Supplementary Fig. S3B).There is no obvious expression of stat3 in spinal nerves so this would be an expected result.

Stat3 promotes PLL axon growth in a cell-autonomous manner
To illustrate whether Stat3 is involved in Schwann cells differentiation during embryonic development, we analyzed the expression of sox10, a marker of neural crest,using in situ hybridization in the stat3 mutants before PLL axons stopped migrating at 30 hpf.In both stat3 mutants and wild-type, sox10 was expressed in the anterior region of the horizontal myosyptum (Fig. 2A), which indicated that Schwann cell precursors were not affected by the stat3 mutation.We crossed the stat3 mutation into the Tg(mbp:EGFP) transgenic line whose Schwann cells are labeled with EGFP (Jung et al., 2010) and analyzed mature Schwann cells in the stat3 mutants.At 3 dpf, the Schwann cells in stat3 mutants formed normally at the anterior part of the PLL, however they were sparse at the posterior part of PLL (Fig. 2B).We analyzed the ultrastructure of PLL neural bers using transmission electron microscopy between the third and sixth somites at 72 hpf.PLL axons in both the wild-type and stat3 mutants were correctly myelinated.The thickness of the myelin showed no obvious difference between mutants and wild-types.However, the PLL axons in stat3 mutants were thinner than wild-type(Fig.2C-D), which con rmed the observations in the TgBAC(neurod1:EGFP) transgenic.The decrease in myelination in stat3 mutant PLL is likely the result of PLL axons failing to extend properly.We asked whether stat3 was involved in PLL axon growth in a cell-autonomous manner or cell-nonautonomous manner.We used the construct encoding tdTomato tagged Stat3 driven by the 5 kb-neurod to express stat3 in some neurons of PLL ganglion (Neurod-Stat3).The construct only containing the reporter gene, tdTomato was used as a control.Either the Neurod-Stat3 or a tdTomato construct was injected into the offspring of a stat3;TgBAC(neurod1:EGFP) heterozygous carrier incross.Each larva was imaged to determine the growth of the PLL axons and then genotyped.EGFP or tdTomato uorescence was used to show the PLL axon phenotype and neurons with stat3 overexpression, respectively (Fig. 2E).The labeled PLL axons were divided into rostral, middle or caudal according to which neuromast they innervated.In wild-types injected with Neurod-Tomato and Neurod-Stat3, similarly labeled PLL axons reached caudal neuromasts.This indicated that extra exogenous Stat3 did not affect PLL axon growth.In stat3 mutants injected with Neurod-Tomato, no PLL axons reached caudal neuromasts (0/19), however, Neurod-Stat3 extended the PLL axons to caudal signi cantly (6/20), Fisher's exact test p=0.0316(Fig. 2F).Fig. 2G shows the representative stat3 mutant larvae injected with the rescue constructs.On the side with or without Neurod-tdTomato labeling, the PLL axons remained truncated.On the side with Neurod-Stat3 labeling, some PLL axons reached the caudal neuromasts (Fig. 2G).These results demonstrate that Stat3 is required autonomously for PLL axon extension.
A Jak inhibitor and a dominant negative Stat3 were also used as an additional approach to inhibit Jak/Stat signaling and to con rm pharmacological results.We injected zebra sh Socs3a (an inhibitor of Stat3) mRNA into fertilized egg and found the PLL axons could still reach caudal neuromasts at 48 hpf.The substitution of double glutamic acids in the Stat3 DNA binding domain to alanines (EE434-435AA) or tyrosine in the SH2 domain to phenylalanine (Y705F) leads to a dominant-negative effect on Stat3 function [32,36].However, both the injection of Stat3 Y708F and Stat3 EE435-436AA mRNA did not affect PLL axon growth (N=8) (Fig. 3C-D).All these results suggested Jak/Stat signaling is not essential for zebra sh embryonic PLL axon growth.
We further injected wild-type or modi ed stat3 mRNA into stat3 mutant zebra sh embryos to investigate which function of Stat3 is necessary for PLL axons growth.The growth of PLL axons in stat3 mutants could be rescued by injecting wild-type stat3 mRNA.However, a Stat3 containing a strong nuclear localization signal (NLS-Stat3) did not rescue PLL axon growth in stat3 mutants.The modi ed Stat3 Y708F and Stat3 EE435-436AA also failed to rescue axonal growth (N=8) (Fig. 3E-F).This indicated that nuclear Stat3 is not su cient for PLL axon growth, but the SH2 and DNA binding domains are still required for its function in PLL axon growth.

Mitochondrial Stat3 regulates embryonic PLL axon growth by promoting ATP synthesis
As reported, a small amount of Stat3 localizes to the mitochondria and enhances electron transport chain (ETC) function and increase ATP synthesis in mammalian cell lines [9].We asked whether mitochondrial Stat3 was involved in PLL axon growth in zebra sh.We tested whether mitochondrially associated Stat3 had a role in PLL axon outgrowth.Cox8 is a subunit of complex IV in the mitochondrial respiratory chain.We co-expressed Stat3 tagged with tdTomato (Stat3-tdTomato) and Cox8a tagged with EGFP (Cox8a-EGFP) in PLL neurons driven by a 5kb-neurod promoter, and performed sequential imaging of axons at 3 dpf.Fig. 4A-B and Video S3 displays a representative result of co-localization of Stat3 and a mitochondrion through time-lapse imaging and kymographs analysis.
To investigate the function of stat3 in zebra sh mitochondria, we measured the ATPase activity of the whole embryos at 24 hpf and 48 hpf, respectively.Notably, the ATPase activity of stat3 mutants was signi cantly lower than wild-type at 24 hpf when the PLL axons initiate growth (N=3).At 48 hpf, after the PLL axon migration nished, the level of ATPase activity in stat3 mutant was still reduced (N=3) (Fig. 4C).Thus, Stat3 is required for mitochondria to maintain ATPase activity during zebra sh embryonic and larval stages.To determine whether the mitochondrial activity in PLL axons was also affected by the stat3 mutation, the larvae were stained with Tetramethylrhodamine ethyl ester (TMRE) at 48 hpf, which is used to quantify the mitochondrial membrane potential in the PLL ganglion in vivo [37].The membrane potential of mitochondria in PLL axons in stat3 mutant was signi cantly decreased compared to wildtype (Fig. 4D).Thus, Stat3 is involved in the regulation of intermembrane potential of mitochondria in PLL axons.
To con rm the role of mitochondrial activity on PLL axon growth, we treated the zebra sh larvae with 50 μM rotenone, an inhibitor of complex I of the mitochondrial respiratory chain, during PLL axon growth (24 hpf-48 hpf).The PLL axon growth was obviously inhibited by rotenone (Fig. 4E).This result was similar to the phenotype in stat3 mutants.In addition, we treated the progeny of a stat3 heterozygous mutant incross with DCHC, a promoter of mitochondria activity, and found that exposure to DCHC at 40 μM or 80 μM increased the proportion of complete PLL axons (82.3% and 83.9% vs 72.5%) (Fig. 4F; N=120).These results suggest that Stat3 regulation of PLL axon extension is dependent on mitochondria activity.
In mice, Stat3 fused with a mitochondrial targeting sequence (MTS-Stat3) promotes axon regeneration in RGC [17].We injected MTS-Stat3 mRNA in stat3 mutant embryos.MTS-Stat3 mRNA rescued the PLL axon growth in stat3 mutants with most of the PLL axons reaching to tail, whereas the tdTomato mRNA couldn't promote PLL axon growth in stat3 mutants (Fig. 4G-H; N=8).Combined with result of NLS-Stat3 mRNA above, we concluded that the mitochondrial Stat3 is su cient to promote PLL axon growth in zebra sh.

Jak/Stat signaling is necessary for axon regeneration
We also analyzed PLL axon regeneration in wild-type and stat3 mutants.The axons were ablated between the fourth and sixth somites at 48 hpf, when the PLL axons completed their embryonic growth.The regenerated axon in both wild-type and stat3 mutants extended through the ablation site 12 hours post-lesion (hpl).At 48 hpl, PLL axons in wild-type were nearly completely regenerated, but PLL axons in stat3 mutant failed to regenerate to their original length (Fig. 5A).We quanti ed the length of PLL axons at 20 hpl, 30 hpl, 48 hpl, and 72 hpl and found that PLL axons in stat3 mutants at the longest timepoint only regenerated to 30.5-39.9% of their original length, and were signi cantly lower in length than wildtype at every time point (Fig. 5B).
To investigate whether Jak/Stat signaling is involved in PLL axon regeneration, we ablated the PLL axons of the wild-type embryos at 48 hpf, then incubated the embryos with Jak/Stat signaling inhibitors.The PLL axons in larvae treated with Pyridone 6 and S3I-201 only regenerated to approximately 65.6% or 63.5% of their original length by 48 hpn, respectively (N=6) (Fig. 5C-D).We also injected modi ed stat3 mRNA into embryos to generate a dominant-negative effect on Jak/Stat signaling.At 48 hpl, PLL axons in wild-type larvae injected with Stat3 Y708F or Socs3a mRNA only regenerated 65.8% or 56.1%, respectively (N=6) (Fig. 5E-F).Together, it is suggested that Jak/Stat signaling is required for PLL axon regeneration suggesting there is a separate role for Stat3 in axon regeneration that is not part of normal developmental outgrowth of the PLL axons.
To investigate whether mitochondrial Stat3 was involved in PLL axon regeneration, we injected wild-type Stat3 or MTS-Stat3 mRNA in stat3 mutants to trigger PLL axons extending to the caudal neuromasts and then ablated the PLL axons at 48 hpf.At 48 hpl, the PLL axons of larvae injected with Stat3 could regenerated nearly completely (97.9%),but the larvae injected with MTS-Stat3 only regenerated approximately 50%, no signi cant difference from the control (tdTomato) mRNA (N=8) (Fig. 5G-H).Taken together, the above results indicate that Jak/Stat signaling is required for PLL axon regeneration, but mitochondrial Stat3 is not su cient to promote PLL axon regeneration.6. Jak/Stat3 is required for Schwann cell migration during PLL axon regeneration During regeneration, Schwann cells migrate to the ablation site and guide the regrowth of axons [4,19].We examined whether Stat3 regulated Schwann cells migration during PLL axon regeneration using the Tg(mbp:EGFP) sh line.In wild-type, Schwann cells near distal regions of PLL axons dispersed at approximately 12 hpl, and migrated back to the distal region at about 48 hpl.In stat3 mutants, the Schwann cells dispersed normally at 12 hpl, but they could not bridge the gap and migrate back to the distal regions of PLL axons (Fig. 6A).To determine whether Jak/Stat signaling was involved in Schwann cells migration, we treated the larvae with Jak/Stat signaling inhibitor Pyridone 6.Although the Schwann cells migrated across the ablation gap, Schwann cell migration was signi cantly inhibited by Pyridone 6 at 48 hpl.The length of Schwann cell surrounding the PLL axons was shorter in larvae treated with 160 μM Pyridone 6 than larvae treated with 80 μM (Fig. 6B-C).The inhibition effect of Pyridone 6 on Schwann cell migration was in a dose-dependent manner.Taken together, Jak/Stat3 signaling also plays an important role in Schwann cell-mediated PLL axon regeneration.

Discussion
In this study, we generated a zebra sh stat3 mutant line using CRISPR/Cas9 and found stat3 plays distinct roles in PLL axon growth at the embryonic and regenerative stage.The mitochondrial Stat3 is required for embryonic axon growth through promoting ATP synthesis.However, Stat3 is involved in PLL axon regeneration mainly through promoting Schwann cells migration by Jak/Stat signaling (Fig. 6D).Our ndings show a new mechanism of Stat3 in axon growth and regeneration and suggest that Stat3 is a promising therapeutic target for axon regeneration.
As a broadly used transcription factor, Stat3 has been found to be essential for embryonic development.
Targeted disruption of stat3 in mice leads to embryonic lethality, which is explained as a consequence of the failure in establishing metabolic exchange between the embryo and maternal blood [38].However, no obvious morphological phenotype was identi ed in zebra sh stat3 mutant embryos in previous studies.Maternal and zygotic stat3 mutants exhibited transient and mild cell cycle increases during zebra sh embryogenesis [22].In this study, we found stat3 was expressed in the PLL ganglion dynamically, and identi ed a PLL axons truncation phenotype in zebra sh stat3 mutant embryos.This nding expands our understanding of the function of stat3 gene in vertebrate embryonic development.Axon extension in Drosophila embryos also requires stat92E [39].It suggests that the Stat family, speci cally Stat3, plays critical and conserved role in embryonic axon growth in insects and vertebrates.As reported, blocking stat3 with antisense morpholinos results in a CaP axon path nding defect [32].However, we didn't detect an obvious phenotype in spinal axons in stat3 mutant.The phenotype of spinal axons might be an offtarget effect of stat3 morpholino in zebra sh embryonic development.Liu and colleagues found that zebra sh mutants lacking maternal and zygotic Stat3 expression do not exhibit abnormal convergence movements during gastrulation as shown in stat3 morpholino studies [22].Another explanation is genetic compensation from other Stat genes could be occurring in the CRISPR generated allele [40].Stat3 and Stat5 proteins can bind to the same regulatory loci and exhibit functional redundancy to some extent [41].Further studies will be necessary to determine if other members of the Stat family can contribute to embryonic axon extension.
Mitochondria play important roles in axon growth [2][3].As reported, a fraction of Stat3 protein localizes in mitochondria and increases the e ciency oxidative phosphorylation [9].The mitochondrially localized Stat3 has been shown to be important in both cancer cells and heart muscles [7,9].Axon growth is an energy consuming process.We showed that Stat3 co-localized with the moving mitochondria in the PLL axons.The ATPase activity and ETC decreased signi cantly in zebra sh stat3 mutants during PLL ganglion projecting axons.stat3 is co-expressed with mtDNA transcription genes in the proliferation tissues of zebra sh and stat3 knock-out impairs normal mitochondrial transcription and cell proliferation [42].Rotenone, a speci c inhibitor of mitochondrial complex I, has been used to inhibit neurogenesis in rats [43].Rotenone treatment mimicked the PLL axon growth defect in stat3 mutants.As reported, MTS-Stat3 promotes RGC axon regeneration in mice and induces neurite outgrowth in the PC12 cell line [16][17].Mitochondrially-targeted Stat3 could rescue PLL axon growth in zebra sh stat3 mutant but the nuclear-localized Stat3 could not.We speculate mitochondrial Stat3 plays an important role in energy consuming cells, including differentiating neurons.Mitochondrial Stat3 is also involved in mitochondrial Ca 2+ , reactive oxygen species (ROS) production, and mitochondrial gene expression [8].It will be important to establish if other aspects of mitochondrial function are as important as ATP production for axon outgrowth.Phosphorylation at S727 is critical for Stat3 translocation into the mitochondria [16][17].
In this study, we found Stat3 Y708F and Stat3 EE435-436AA could not rescue PLL axon growth in stat3 mutants.EE434-435 is required for DNA binding, and Y708 and S727 are required for transcription activation of Stat3 [42,44].How this relates to the speci c functions in the mitochondria to enable axon outgrowth needs to be explored further.
Although many regeneration processes recapitulate embryonic development, some genes exhibit distinct roles during embryonic development or regeneration.Jak/Stat signaling has been shown to be important for promoting tissue regeneration [45].It has been shown essential for optic nerve regeneration in mouse and zebra sh, and spinal nerve regeneration in gecko [3,12,46].The regeneration of most peripheral axons is linked to Schwann cell migration which forms cords guiding the injured axon to target the original location correctly [4,20,47].We found that stat3 mutations or Jak/Stat signaling inhibitors hampered PLL axon regeneration and Schwann cell migration.Although we didn't examine the localization of Jak/Stat in PLL Schwann cells, previous studies have shown the target genes of Jak/Stat are highly expressed in glia cells during axon regenerating [13,46].Since the Schwann cell migration defect is earlier and more severe than the PLL axon regeneration defect in stat3 mutant, it is tempting to speculate that Jak/Stat signaling promotes PLL axon regeneration through Schwann cell migration.
Inactivating the Jak/Stat speci cally in Schwann cells during PLL axon regeneration will give a more convincing evidence.In contrast, several lines of evidence indicate that Jak/Stat signaling is dispensable for normal PLL embryonic axon growth.Inhibiting Jak/Stat signaling with small molecules, Socs3a, or a dominant mutant form of Stat3 does not affect PLL axon embryonic growth.Nuclear localized Stat3 (NLS-Stat3) didn't rescue the axon phenotype of stat3 mutants.In line with our results, blocking Jak/Stat signaling inhibits DRG (dorsal root ganglion) axon regeneration but not axon embryonic extension [48].Therefore, Stat3 perhaps plays distinct roles in embryonic and regenerative axon extension.
In conclusion, zebra sh stat3 mutant shows defect in both PLL axon embryonic growth and regeneration.Mitochondrial Stat3, rather than nuclear Stat3 promotes PLL axon growth through ATP synthesis in a cellautonomous manner.In contrast, Jak/Stat signaling is required for PLL axon regeneration probably through promoting Schwann cell migration.Our results demonstrate a distinct role of Stat3 in axon embryonic growth and axon regeneration.