DLK-dependent axonal mitochondrial fission drives degeneration after axotomy

Abstract

Currently there are no effective treatments for an array of neurodegenerative disorders to a large part because cell-based models fail to recapitulate disease. Here we develop a reproducible human iPSC-based model where laser axotomy causes retrograde axon degeneration leading to neuronal cell death. Time-lapse confocal imaging revealed that damage triggers an apoptotic wave of mitochondrial fission proceeding from the site of injury to the soma. We demonstrate that this apoptotic wave is locally initiated in the axon by dual leucine zipper kinase (DLK). We find that mitochondrial fission and resultant cell death are entirely dependent on phosphorylation of dynamin related protein 1 (DRP1) downstream of DLK, revealing a mechanism by which DLK can drive apoptosis. Importantly, we show that CRISPR mediated Drp1 depletion protects mouse retinal ganglion neurons from degeneration after optic nerve crush. Our results provide a platform for studying degeneration of human neurons, pinpoint key early events in damage related neural death and provide potential focus for therapeutic intervention.

Fig. 1. Axotomy triggers a wave of DRP1-dependent mitochondrial fission, axon degeneration and cell death.

a Schematic representation of laser axotomy. Created in BioRender. Le Pichon, C. E. (2022) BioRender.com/z21b395. b Representative images of WT and SARM1 KO axons expressing cytoplasmic mApple (red) pre- and 4 h post-axotomy (PA). Scalebar 100 µm. c Axon degeneration index (ADI) in WT and SARM1 axons 4 h post-axotomy. n = 38. Two-way ANOVA (p = 0.0086 **, not significant, ns). d Representative images of mitoGFP (green) in axons pre- and 5 min post-axotomy (PA). Scalebar 20 µm. e Representative time-lapse images of mitochondria undergoing fission after axon injury. Scalebar 2 µm (valid for all 5 examples). f Schematic. g Quantification of mitochondrial length pre- (green) and 5 min post-axotomy (gray). n = 29. Two-way ANOVA (10 min post-axotomy, 0–100 µm vs 200–300 µm p = 0.0021; vs 400–500 µm p < 0.0001. 20 min post-axotomy, 0–100 vs 200–300 µm p = 0.0019). h Axonal mitochondrial particles in control and DRP1 knockdown (KD) neurons. n = 26 WT, n = 25 DRP1 KD. Two-way ANOVA (p ≤ 0.05 *). i Axonal mitochondrial particles 3 min post-axotomy in DMSO and P110 treated neurons. n = 27 DMSO n = 26 P110. Two-way ANOVA (p = 0.0467 *). j Mitochondrial particles 0–100, 200–300 and 400–500 µm from axotomy. n = 23. One-way ANOVA (*p ≤ 0.05, 0–100 µm vs 200–300 µm, # p ≤ 0.05, 0–100 µm vs 400–500 µm). k Representative images of axotomy-induced mitochondrial fission wave. mitoGFP (green). Scalebar 25 µm. l Axon degeneration index (ADI) 0–100, 200–300 and 400–500 µm from axotomy. n = 14. One-way ANOVA (*p ≤ 0.05, 0–100 vs 200–300 µm, ^## p ≤ 0.01, ^### p ≤ 0.005, 0–100 vs 400–500 µm). m Representative images of retrograde axon degeneration 0, 10-, 30- and 120-min post-axotomy. Cyto-mApple (red). Scalebar 25 µm. n Representative images of axotomy-induced neuron death, pre- and 12-h post axotomy. Arrowhead indicates axotomized neuron; the other neuron is uninjured. Scalebar 40 µm. o Quantification of axotomy-induced neuron death. Percentage of dead neurons 12 h post-axotomy. n = 30. Unpaired, two-tailed, t test (p = 0.004 **). Graphical representations show mean ± SEM. Dashed box in (a, f) indicates field of view. Asterisks * in (a, b, d) mark site of axotomy. All experiments include axotomies from N = 3 independent differentiations. Bonferroni correction applied to all ANOVAs. Schematics in a and f made using Biorender.

Fig. 2. DLK regulates axotomy-induced mitochondrial fission.

a Representative images of DLK colocalization at the site of mitochondrial fission after axotomy. DLK-GFP (green) and mitochondria (mitoRFP, magenta). Scalebar = 10 µm. b Percentage mitochondrial fission events where DLK is localized at the site of fission. N = 3 independent differentiations, 74 fission events from 33 axotomized neurons. c Representative images of WT and DLK KO neurons transduced with mitoGFP (green) pre (Pre) and 5 min post axotomy (PA). * Marks the site of axotomy. Scalebar = 25 µm. d Normalized number of mitochondrial particles post axotomy in WT (black) and DLK KO (green) neurons. N = 3 independent differentiations, N = 29 WT and N = 30 DLK KO axotomized neurons (Two-way ANOVA, Bonferroni correction (p < 0.0001 ****). e Normalized number of mitochondrial particles post axotomy in DMSO (black) and GNE3511 (green)-treated neurons. N = 3 independent differentiations, N = 23 DMSO and N = 25 GNE3511 axotomized neurons. (Two-way ANOVA, Bonferroni correction, p < 0.0001 ****). f Normalized number of mitochondrial particles post axotomy in DLK KO neurons transduced with WT DLK-GFP (green), DLK-C127S-GFP (black) or DLK-S302A-GFP (purple). N = 3 independent differentiations, N = 31 DLK WT and N = 27 DLK-C127S and N = 23 DLK-S302A axotomized neurons. Two-way ANOVA, Bonferroni correction (p < 0.0001 ****, WT DLK vs DLK C127S, p < 0.0001 ####, WT DLK vs S302A). g Representative images of neurons transduced with mitoGFP (green) treated with DMSO and JNKi pre (Pre) and 5 min post axotomy (PA). * Marks the site of axotomy. Scalebar = 25 µm. h Normalized number of mitochondrial particles post axotomy in DMSO (black) and JNKi (green)-treated neurons. N = 3 independent differentiations, N = 32 DMSO and N = 31 JNKi axotomized neurons. (Two-way ANOVA, Bonferroni correction, p < 0.0001 ****). Graphical representations show mean ± SEM.

Fig. 3. Activation of the DLK pathway by overexpression causes increased DRP1 phosphorylation.

a Representative western blots of HEK-293A cells transfected for 24 h with WT, C127S and S302A DLK-GFP. Immunoblot for pS616-DRP1, pS637-DRP1, total DRP1, DLK, pS63-cJun and loading control ß actin. b Quantification of pS63-cJun/ ß actin levels after expression of WT, C127S and S302A DLK-GFP in HEK-293A cells. N = 3 independent transfections. One-way ANOVA, Bonferroni correction (DLK WT vs DLK C127S p = 0.0335 *, DLK WT vs DLK C127S p = 0.0373 *). c Quantification of pS616-DRP1/total DRP1 levels after expression of WT, C127S and S302A DLK-GFP in HEK-293A cells. N = 3 independent transfections. One-way ANOVA, Bonferroni correction (DLK WT vs DLK C127S p = 0.0103 *, DLK WT vs DLK C127S p = 0.0105 *). d Quantification of total DRP1/ ß actin levels after expression of WT, C127S and S302A DLK-GFP in HEK-293A cells. N = 3 independent transfections. One-way ANOVA, Bonferroni correction (not significant, ns). e Quantification of pS637 DRP1/ ß actin levels after expression of WT, C127S and S302A DLK-GFP in HEK-293A cells. N = 3 independent transfections. One-way ANOVA, Bonferroni correction (not significant, ns). f Phospho-mass spectrometry. pS616-DRP1 peptide / total DRP1 peptide levels from purified GST-DRP1 in HEK-293A cells expressing GST-DRP1, GST-DRP1 and DLK-GFP or GST-DRP1 and DLK-S302A-GFP. N = 4 independent transfections. One-way ANOVA, Bonferroni correction (p = 0.0046 **). g Representative images of HEK-293A cells expressing WT, C127S and S302A DLK-GFP (green) and mitoTracker (orange) for 24 h. Scalebar = 10 µm. h Quantification of mitochondrial morphology in HEK-293A cells transfected with WT, C127S and S302A DLK-GFP. N = 3 individual transfections. Two-way ANOVA, Bonferroni correction (WT vs C127S p ≤ 0.0001 ****, WT vs S302A p ≤ 0.0001 ****, C127S vs S302A not significant, ns). i Representative western blots HEK-293A cells expressing dCAS9 transduced with control and DRP1 gRNAs. Immunoblot for total DRP1 and loading control ß actin. Replicated twice with similar results. j Representative images of Control and DRP1 KD HEK-293A cells expressing DLK-HALO for 24 h. NLS-GFP (green), DLK-HALO (white), mitoTracker (orange). Scalebar = 20 µm. k Quantification of Control and DRP1 KD HEK-293A cells mitochondrial morphology. Cells transfected with DLK-HALO. N = 3 individual transfections. (Two-way ANOVA, Bonferroni correction (p = 0.0024 **) All graphical representations show mean ± SEM.

Fig. 4. Axotomy causes DLK-dependent phosphorylation of DRP1.

a Schematic representation of i3Neuron center plating and axotomy for protein harvesting after injury. b Representative western blots of WT and DLK KO neuron cell bodies untreated (UT), 2 and 4 h post axotomy. Immunoblot for p-S616 DRP1, total DRP1 and loading control ß actin. c Quantification of pDRP1 (S616)/ total DRP1 levels in WT and DLK KO neuron cell bodies untreated (UT), 2 and 4 h post axotomy. Results normalized to UT. N = 3 independent differentiations. One-way ANOVA. No significant changes observed. d Representative Western blots of WT and DLK KO neuron axons untreated (UT), 2 and 4 h post axotomy. Immunoblot for p-S616 DRP1, total DRP1 and loading control ß actin. e Quantification of pDRP1 (S616)/ total DRP1 levels in WT and DLK KO neuron axons 0, 2 and 4 h post axotomy. Results normalized to UT. N = 5 independent differentiations. Two-way ANOVA. Bonferroni correction (p = 0.0058 **). f Representative images of WT and DLK KO axons stained for ßIII tubulin (red) and pDRP1 S616 (white) untreated (UT), 30 min and 1 h post axotomy. Scalebar 25 µm. g Quantification of pDRP1 S616 fluorescence in WT and DLK KO neuron axons after axotomy. Results normalized to untreated axons. N = 3 independent differentiations. Two-way ANOVA, Bonferroni correction (UT vs 30 min p = 0.0156*, UT vs 1 h p = 0.0185*). h Schematic representation of optic nerve crush injury model. Bracket and * indicate the proximal portion of the injured nerve harvested for Western blotting. i Representative Western blots of proximal portion of ipsilateral (I) injured and contralateral (C) optic nerves 24 h after optic nerve crush. Immunoblot for p-S616 DRP1, total DRP1 and loading control ß actin. j Quantification of pDRP1 (S616)/ total DRP1 levels after optic nerve crush. Results normalized to contralateral side (C). N = 4 samples. Unpaired, two-tailed, t test (p = 0.0346 *). k Quantification of total DRP1/ ß actin levels after optic nerve crush. Results normalized to contralateral side (C). N = 4 samples. Unpaired, two-tailed t test. not significant, ns. All graphical representations show mean ± SEM.

Fig. 5. Blocking DRP1 or DLK delays axon degeneration.

a Schematic representation of i3Neuron laser axotomy. Dashed box indicates the field of view. b Representative images of WT and DLK KO neuron axons transduced with cyto mApple (red) proximal to the site of injury 0,4, 8 and 24 h post axotomy. Scalebar 25 µm. c Quantification of axon degeneration index (ADI) in WT and DLK KO neurons 0, 4, 8 and 24 h post axotomy. N = 3 independent differentiations, N = 33 WT and N = 35 DLK KO axotomized axons. Two-way ANOVA, Bonferroni correction (p ≤ 0.005 ***, p ≤ 0.001 ****). d Schematic representation of i3Neuron laser axotomy. Dashed box indicates field of view. e Representative images of Control and DRP1 KD neuron axons transduced with cyto mApple (red) proximal to the site of injury 0, 4, 8 and 24 h post axotomy. Scalebar 25 µm. f Quantification of axon degeneration index (ADI) in WT and DRP1 KD neurons 0, 4, 8 and 24 h post axotomy. N = 3 independent differentiations, N = 35 control gRNA and N = 34 DRP1 gRNA axotomized axons. Two-way ANOVA, Bonferroni correction (p ≤ 0.005 ***, p ≤ 0.001 ****). g Schematic representation of i3Neuron laser axotomy. Dashed box indicates field of view. h Representative images of WT and DLK KO neuron cell bodies transduced with cyto mApple (red) pre and 18 h post axotomy (PA). Scalebar 40 µm. i Percentage degenerated WT and DLK KO 18 post axotomy. N = 3 independent differentiations. Unpaired, two-tailed, t test (p = 0.0129 *). j Schematic representation of i3Neuron laser axotomy. Dashed box indicates field of view. k Representative images of Control and DRP1 KD neuron cell bodies transduced with cyto mApple (red) pre and 18 h post axotomy (PA). Scalebar 40 µm. l Percentage degenerated Control and DRP1 KD 18 post axotomy. Results are represented as mean ± SEM. N = 3 independent differentiations. Unpaired, two-tailed, t test (p = 0.0174 *). All graphical representations show mean ± SEM.

Fig. 6. DRP1 mediates cell death after ONC.

a Schematic representation of RGC survival following optic nerve crush (ONC). b Western blots of Control and DRP1 gRNA transduced optic nerved. Immunoblot for DRP1 and loading control ß actin. c Quantification of DRP1 knockdown by Western blot. N = 4 samples. d Representative images of cleaved caspase 3 positive cells in saCas9-Control and saCas9-DRP1 gRNA transduced RGCs 3 days after ONC. Scalebar 25 µm. e Average cleaved caspase 3 positive cells per section in saCas9-Control, saCas9-DRP1 gRNA and EGFP transduced RGCs after nerve crush. N = 6 Control gRNA, n = 4 DRP1 gRNA and n = 6 EGFP animals per condition. One-way ANOVA, Bonferroni correction. (Control gRNA vs EGFP, not significant, ns. Control gRNA vs DRP1 gRNA p = 0.025 *. DRP1 gRNA vs EGFP p = 0.0093 **). f Representative images of retinas 7 days post ONC and immunostained for RGC marker RBPMS (magenta). Images acquired at similar areas and the same distance from the optic nerve head (ONH). Scalebar 100 µm. g Quantification of percentage RBPMS-positive RGCs in the retinas of control and DRP1 gRNA mice at 7DPI (normalized to contralateral condition). (N = 8 mice per condition. Unpaired, two-tailed, t test. p < 0.0001 ****). h Representative isodensity maps display the topographical survival of RBPMS+RGCs at 7 days post-injury. DRP1 gRNA treatment delays ONC-induced RGC degeneration across the retina. Color scale for isodensity maps ranges from 0 (purple) to 3600 (red) RGCs/mm². All graphical representations show mean ± SEM.

Fig. 7. BAX regulates cell death after axotomy.

a Representative images of WT neurons treated with DMSO or GNE3511 pre- and 1 h post-axotomy (PA) expressing mitoGFP (green) and A₁-BAX/B₁-DRP1 complexes (red). Scale bar = 5 µm. b Quantification of A₁-BAX/B₁-DRP1 fluorescence fold change in the mitochondria of DMSO or GNE3511 treated neurons 1-h post-axotomy. N = 3 independent differentiations, N = 35 DMSO and N = 34 GNE3511 axotomized neurons. Two-way ANOVA, Bonferroni correction. (p = 0.0168 *). c Quantification of A₁-BAX/B₁- WT DRP1 and A₁-BAX/B₁- DRP1 S616A fluorescence fold change in the mitochondria 1-h post-axotomy. N = 3 independent differentiations, N = 37 DRP1 WT and N = 38 DRP1 S616 axotomized neurons. Two-way ANOVA, Bonferroni correction. ( = 0.0192 *). d Representative images of DMSO and BAXi treated neuron cell bodies transduced with cyto mApple (red) pre and 18 h post axotomy (PA). Scalebar 40 µm. e Percent degenerated neurons 18 h post axotomy treated with either DMSO or BAXi. N = 3 independent differentiations. Unpaired, two-tailed, t test. (p = 0.0118 *). f Quantification of axon degeneration index (ADI) of DMSO and BAX-A-treated neurons 24 h after axotomy. N = 3 independent differentiations, total of 46 axons. One-way ANOVA, Bonferroni correction (p ≤ 0.0001 ****, not significant, ns). g Illustration of microfluidic devices used to separate axons from somas allowing for the local treatment of i3Neuron axons. h Representative images of WT axons separated using microfluidic chambers treated with DMSO, BAX-A or BAX-A + GNE3511 for 24 h. Scalebar 25 µm. i Quantification of axon degeneration index (ADI) of isolated axons treated with DMSO BAX-A and BAX-A + GNE3511 for 24 h. N = 3 independent differentiations, 10 images per differentiation. One-way ANOVA, Bonferroni correction (p ≤ 0.0001 ****, not significant, ns). All graphical representations show mean ± SEM.