Coordinated NADPH oxidase/hydrogen peroxide functions regulate cutaneous sensory axon de- and regeneration

Abstract

Tissue wounding induces cutaneous sensory axon regeneration via hydrogen peroxide (H₂O₂) that is produced by the epithelial NADPH oxidase, Duox1. Sciatic nerve injury instead induces axon regeneration through neuronal uptake of the NADPH oxidase, Nox2, from macrophages. We therefore reasoned that the tissue environment in which axons are damaged stimulates distinct regenerative mechanisms. Here, we show that cutaneous axon regeneration induced by tissue wounding depends on both neuronal and keratinocyte-specific mechanisms involving H₂O₂ signaling. Genetic depletion of H₂O₂ in sensory neurons abolishes axon regeneration, whereas keratinocyte-specific H₂O₂ depletion promotes axonal repulsion, a phenotype mirrored in duox1 mutants. Intriguingly, cyba mutants, deficient in the essential Nox subunit, p22Phox, retain limited axon regenerative capacity but display delayed Wallerian degeneration and axonal fusion, observed so far only in invertebrates. We further show that keratinocyte-specific oxidation of the epidermal growth factor receptor (EGFR) at a conserved cysteine thiol (C797) serves as an attractive cue for regenerating axons, leading to EGFR-dependent localized epidermal matrix remodeling via the matrix-metalloproteinase, MMP-13. Therefore, wound-induced cutaneous axon de- and regeneration depend on the coordinated functions of NADPH oxidases mediating distinct processes following injury.

Keywords: EGFR; Wallerian degeneration; axon fusion; axon regeneration; hydrogen peroxide.

Fig. 1.

Cysteine oxidation is essential for cutaneous sensory axon regeneration at 3 dpf. (A) Model for thiol oxidation reactions and targeting of sulfenic acid with dimedone or alkyne-tagged DYn-2. (B) Click chemistry with DYn-2 detects sulfenylated protein gradient (red) at 30 mpi, but not in injured DYn-2⁽⁻⁾ or uninjured DYn-2⁽⁺⁾ animals. (C and D) Time-lapse imaging of Tg(krt4:HyPer) fish for 60 min following amputation reveals peak oxidation at ∼15 to 30 mpa, whereas oxidation is largely prevented in DYn-2–treated animals (n = 10 animals per group, two-way ANOVA: P < 0.0001 for vehicle vs. DYn-2). (E) Mean gradient width from wound is greatest with HyPer detection compared to DYn-2 and dimedone staining (n = 7, 8, and 4 animals, respectively). (F–H) Time-lapse imaging of sensory neurons mosaically expressing CREST3:Gal4VP16_14xUAS_GFP shows impaired axon regeneration with dimedone (F and G) compared to vehicle controls (H). (I) Quantification shows reduced regeneration also in the presence of the antioxidant, PHG (n = 10 animals per group). One-way ANOVA and Tukey’s (E) or Bonferroni’s (I) posttest was used. Significance above columns compares to vehicle group.

Fig. 2.

Thiol oxidation in cutaneous axon branches near the wound. (A) HyPer fluorescence in Rohon-Beard neuron innervating the caudal fin. White dotted circle indicates the site of skin wounding. (B) Intensity plots show HyPer fluorescence (unoxidized [Upper] and oxidized [Lower]) in cutaneous axon branches of neuron shown in A at 5, 25, and 60 mpi. White-dotted circle indicates the wound site. White arrows point to small HyPer puncta in branches near the wound that become oxidized. Note the presence of stable oxidized HyPer fluorescence in the soma (yellow arrowheads) and primary branch (small red circle). (C) HyPer ratios in axonal puncta near (<150 µm) and distant from (>150 µm) the wound. The oxidation peak in puncta near the wound is detected ∼35 mpi but is absent in distant branches and branches near the wound in duox1^(−/−) (n = 3 animals per group). (D) HyPer detection upon H₂O₂ addition to uninjured zebrafish within ∼ to 5 min (n = 6 animals). (E) Schematic representation of Duox-dependent H₂O₂ secretion and cytoplasmic uptake into keratinocytes and cutaneous axons.

Fig. 3.

Neuronal and keratinocyte specific H₂O₂ depletion induces distinct axon regeneration phenotypes. (A and B) Overexpression of gpx1a in sensory neurons abolishes axon regeneration, confirmed upon 12-h tracking (colored lines). (B). Quantification of A (control: n = 4; CREST3:gpx1a: n = 5 animals). (C) Coexpression of krt4:Gal4VP16_tdTomato_5xUAS with CREST3:LexA_lexAop_GFP (Upper) shows axon regeneration distant from the wound (white arrow points to branch that regenerates upon contact with tdTomato⁺ keratinocyte). (Lower) Axon regeneration does not occur and branches retract when in contact with keratinocytes expressing krt4:Gal4VP16_tdTomato_5xUAS_gpx1a (yellow circle; yellow arrow points to degenerated axon branches, white arrow points to distant regenerating branch). (D) Axon track length is similar between injured branches in the controls and distant branches, while branches in contact with tdTomato_gpx1⁺ keratinocytes retract (left to right: n = 6, 5, and 5 animals). Statistical comparisons use ANOVA and Sidak’s multiple comparisons test. Significance above columns compares to the control column. Bracket indicates alternative comparison. (E) Schematic of observations shown in C and D.

Fig. 4.

Phenotypic differences between duox1^(−/−) and cyba^(−/−) mutants. (A) Time-lapse imaging of a 3dpf duox1^(−/−) mutant shows largely retraction of axon branches at the amputation wound, as indicated by 12-h colored tracks/arrows. (B) A cyba^(−/−) mutant retains some regenerative capacity, as indicated by 12-h tracks. (C) Cutaneous axons retract in duox1^(−/−) mutants and morphants, whereas cyba^(−/−) mutants show limited growth, although significantly reduced compared with amputated wild-type fish (left to right: n = 3, 12, 5, 10, and 5 animals, one-way ANOVA, all comparisons are significantly different [P < 0.0001] except where indicated otherwise); amp, amputation; MO, morpholino; uninj. uninjured. (D) HyPer oxidation measured in axon branches near the amputation wound of control and cyba^(−/−) mutants reveals similar oxidation increase following injury [n = 3 control, n = 5 cyba^(−/−) fish]. (E) Wallerian degeneration onset in cyba^(−/−) mutant at ∼6 hpa compared with wild-type onset at ∼1.75 hpa (arrows). (F) Fragmentation onset occurs within 2 hpa in ∼80% of wild-type and >2 hpa in ∼30% of cyba^(−/−) fish, whereas ∼40% cyba^(−/−) fish show no fragmentation (wild-type: n = 5; cyba^(−/−): n = 4, two-way ANOVA, wild-type vs. cyba^−/−: P = 0.002) (G) Severed axon branches fuse in a cyba^(−/−) mutant.

Fig. 5.

Keratinocyte-specific oxidation of EGFR-Cys797 promotes axon regeneration. (A) The transcription inhibitor, DRB, and the translation inhibitor, cycloheximide, but not actinomycin D and 6-MP, impair axon regeneration in 3-dpf zebrafish. (B) Ingenuity Upstream Pathway analysis of RNA-seq data to determine H₂O₂ transcriptional profile in 4-dpf zebrafish shows H₂O₂ and EGF as common upstream regulators of transcriptionally up-regulated (red) and down-regulated (green) genes. (C) Antagonistic EGFR functions regulate axon regeneration during (1 dpf) and after (3 dpf) skin innervation by Rohon-Beard neurons. (D) Reduced membrane localization of zebrafish EGFR(C797A)-GFP following transfection into human HaCaT keratinocytes compared with wtEGFR-GFP. (E) Membrane:cytoplasmic GFP ratio is reduced in cells expressing EGFR(C797A) (five cells and two biological replicates, each). (F) Detection of EGFR sulfenylation in HEK001 human keratinocytes via Western blotting following click chemistry shows a 1.5-fold increase upon 30-min stimulation with H₂O₂. GPX serves as positive control. (G, Upper) Zoomed view of red square shown in Lower, Left image: Axon branches retract around sites of tdTomato_EGFR(C797A) expressing keratinocytes (yellow arrows), whereas axons regenerate in areas devoid of tdTomato_EGFR(C797A)⁺ keratinocytes (white arrows). (Lower) Transmitted light overlay of Upper image in low magnification to depict the injury site. Retracting (yellow) and growing axons (white) are overlaid as 12-h tracks in the right image. (H) Quantification of G (from left: n = 5, 5, and 6 animals). (I) Axon branch number is reduced at 6 and 12 h postinjury in regions where keratinocytes express krt4:Gal4VP16_tdTomato_ EGFR(C797A) compared with animals in which keratinocytes express krt4:Gal4VP16_tdTomato. (J) EGFR inhibition (n = 34 animals) reduces HyPer oxidation compared with control animals (n = 33). Statistical comparisons were used as follows: (A, C, H, and J) ANOVA and Tukey’s multiple-comparisons test; (E) Student’s t test; (I) two-way ANOVA and Sidak’s multiple-comparisons test.

Fig. 6.

EGFR promotes cutaneous axon regeneration via MMP-13 dependent ECM remodeling. (A) Correlation between qPCR and RNA-seq dataset shows increased egfra and mmp9, and mmp13a expression (red dots), among others, following H₂O₂ treatment. (B) String-DB shows MMP-13 interacting with the EGFR. (C) DB04760 but not GM6001 (MMP-1/2/3/9 inhibitor) impairs axon regeneration following injury, whereas selective MMP-9 inhibition increases regeneration (left to right: n = 13, 4, 8, 7, 6, and 4 animals). (D) Time-lapse sequence showing impaired axon regeneration with DB04760, with 12-h colored tracks depicting axonal retraction (Movie S18). (E) qPCR shows increased mmp13a expression following 3-h H₂O₂ treatment. mmp13a expression is abolished at 3 hpa upon treatment with the antioxidant, NAC, and EGFR inhibitor (n = 3 biological replicates with 10 pooled 3-dpf zebrafish each). Statistical comparison with uninjured control group is shown. (F) Transmission electron microscopy at 3 hpa shows large gaps between keratinocytes (periderm = green, basal layer = yellow, axons = blue) at sites harboring axon branches. Gaps are absent in uninjured epidermis and following treatment with DB04760. (G) Model shows dual H₂O₂ functions in keratinocytes leading to EGFR oxidation and MMP-13 activation mediating axon growth within the epidermis. Loss of neuron-intrinsic p22Phox slows Wallerian degeneration and promotes axonal fusion mediated. Statistical comparisons in C and E used one-way ANOVA and Bonferroni’s posttest.