Roles of Kdm6a and Kdm6b in Regulation of Mammalian Neural Regeneration

Abstract

Epigenetic regulation of neuronal transcriptomic landscape is emerging to be a key coordinator of mammalian neural regeneration. The roles of two histone 3 lysine 27 (H3K27) demethylases, Kdm6a/b, in controlling neuroprotection and axon regeneration are investigated here. Deleting either Kdm6a or Kdm6b leads to enhanced sensory axon regeneration in the peripheral nervous system (PNS), whereas in the central nervous system (CNS), only deleting Kdm6a in retinal ganglion cells (RGCs) significantly enhances optic nerve regeneration. Moreover, both Kdm6a and Kdm6b function to regulate RGC survival but with different mechanisms. Mechanistically, Kdm6a regulates RGC regeneration via distinct pathway from that of Pten, and co-deleting Kdm6a and Pten results in long distance optic nerve regeneration passing the optic chiasm. In addition, RNA-seq profiling reveals that Kdm6a deletion switches the RGC transcriptomics into a developmental-like state and suppresses several known repressors of neural regeneration. Klf4 is identified as a direct downstream target of Kdm6a-H3K27me3 signaling in both sensory neurons and RGCs to regulate axon regeneration. These findings not only reveal different roles of Kdm6a and Kdm6b in regulation of neural regeneration and their underlying mechanisms, but also identify Kdm6a-mediated histone demethylation signaling as a novel epigenetic target for supporting CNS neural regeneration.

Keywords: Kdm6a; Kdm6b; Klf4; epigenetic regulation; histone methylation; neuroprotection; optic nerve regeneration; sensory axon regeneration.

Figure 1

Knocking down Kdm6a or Kdm6b in adult mouse sensory neurons markedly enhanced sensory axon regeneration in vivo. a) Time course of expression of Kdm6a and Kdm6b transcripts in ipsilateral L4/L5 DRG from naive mice and 1, 3, and 7 days post‐sciatic nerve crush by real‐time PCR analysis (one way ANOVA followed by Tukey's multiple comparison test, Kdm6a: p = 0.0001, n = 5 independent experiments for each condition; Kdm6b: p = 0.0165, n = 7 independent experiments for each condition). b) Immunostaining of DRG sections showing significantly reduced level of Kdm6a in adult mouse sensory neurons in vivo after SNI. The 4th column shows enlarged images of areas indicated by the dashed white boxes in the 2nd column. Images were stained for neuronal marker Tuj1 (red) and Kdm6a (green). Scale bar, 100 µm for the first 3 columns and 30 µm for the 4th column. c) Quantification of relative fluorescence intensity of Kdm6a immunostaining shown in (b) (two tailed student's t tests, Naïve vs SNI 1d: p = 0.0076, Naïve vs SNI 3d: p = 0.0036; n = 3 independent experiments for each condition). d) Immunostaining of DRG sections showing significantly reduced level of Kdm6b in adult mouse sensory neurons in vivo at day 3 after SNI. The 4th column shows enlarged images of areas indicated by the dashed white boxes in the 2nd column. Images were stained for neuronal marker Tuj1 (red) and Kdm6b (green). Scale bar, 100 µm for the first 3 columns and 30 µm for the 4th column. e) Quantification of relative fluorescence intensity of Kdm6b immunostaining shown in (d) (two tailed student's t tests, Naïve vs SNI 3d: p = 0.0426; n = 3 independent experiments for each condition). f) Western blot image showing greatly reduced protein level of Kdm6a in DRG tissues 2 days after in vivo electroporation of siRNAs against Kdm6a (siKdm6a). A representative image from 2 independent experiments is shown. g) Western blot image showing greatly reduced protein level of Kdm6b in DRG tissues 2 days after in vivo electroporation of siRNAs against Kdm6b (siKdm6b). A representative image from 2 independent experiments is shown. h) Western blot image showing significantly increased H3K27me3 levels in DRG neurons 2 days after in vitro electroporation of siRNAs against Kdm6a/b (siKdm6a/b). A representative image from 2 independent experiments is shown. i) Schematics of in vivo electroporation and investigation of axon regeneration after knocking down Kdm6a or Kdm6b. L4 and L5 DRGs of wild type adult mice were electroporated in vivo with EGFP or EGFP + siRNAs against Kdm6a (siKdm6a) or Kdm6b (siKdm6b). To assess the promoting effects, axon regeneration analyses were performed 2 days post‐sciatic nerve crush, when sensory axon regeneration has not reached its optimal rate. j) Representative images of in vivo sensory axon regeneration in mice electroporated with either EGFP or EGFP+siKdm6a. The right column shows enlarged images of areas indicated by the dashed white boxes in the left column. The red dotted lines indicate the nerve crush sites. Yellow arrows indicate the distal ends of regenerating axons. Scale bar, 1 mm for the left column and 0.5 mm for the right column. k) Knocking down Kdm6a in sensory neurons significantly promoted sensory axon regeneration in vivo (two tailed student's t tests, p = 0.0152; EGFP; n = 8 mice, EGFP+siKdm6a; n = 10 mice). l) Representative images of in vivo sensory axon regeneration in mice electroporated with either EGFP or EGFP+siKdm6b. The right column shows enlarged images of areas indicated by the dashed white boxes in the left column. The red dotted lines indicate the nerve crush sites. Yellow arrows indicate the distal ends of regenerating axons. Scale bar, 1 mm for the left column and 0.5 mm for the right column. m) Knocking down Kdm6b in sensory neurons significantly promoted sensory axon regeneration in vivo (two tailed student's t tests, p = 0.0033; EGFP: n = 7 mice, EGFP+siKdm6b: n = 10 mice). *, **, ***p < 0.05, 0.01, 0.001, respectively. Data are represented as mean ± SEM.

Figure 2

Conditional deletion of H3K27 demethylase Kdm6a in mouse RGCs markedly promoted optic nerve regeneration and RGC survival. a) Representative images of Kdm6a immunostaining in retina sections showing unchanged level of Kdm6a in RGCs after ONC. Images were stained for neuronal marker Tuj1 (red) and Kdm6a (green). Scare bar, 50 µm. b) Enlarged images of areas indicated by the dashed white boxes in (a). Scale bar, 10 µm. c) Quantification of relative fluorescence intensity of Kdm6a immunostaining in (a) (one way ANOVA followed by Tukey's multiple comparison test, p = 0.2274; n = 3 independent experiments for each condition). d) Top: schematic time course of optic nerve regeneration model. Specifically, AAV2‐Cre was injected at day 1 and the optic nerve was crushed 2 weeks later. Optic nerve regeneration was analyzed 2 weeks after the nerve crush, and the fluorescence (Alexa 594) tagged CTB was injected 2 days before collecting the optic nerve. Bottom: representative 2D projected confocal images of whole mount cleared optic nerves of wild type, male Kdm6a knockout (Kdm6a KO/male), female Kdm6a knockout (Kdm6a KO/female) mice 2 weeks after ONC. The regenerating axons were labeled with CTB‐Alexa 594. Yellow arrows indicate the distal ends of regenerating axons. The red dotted lines indicate the nerve crush sites. The columns on the right display magnified images of the areas in white, dashed boxes on the left, showing axons at 0.25, 0.75, and 1.25 mm distal to the crush sites. Scale bar, 200 µm for the left column and 100 µm for the magnified images. e) Quantification of regenerating axons at different distances distal to the nerve crush site (0.25–1.50 mm) (one way ANOVA followed by Tukey's multiple comparisons test, p < 0.0001; wild type: n = 10 mice, Kdm6a KO/male: n = 7 mice, Kdm6a KO/female: n = 7). f) Quantification of the average length of the top 5 longest axons of each nerve in (d) (one way ANOVA followed by Tukey's multiple comparisons test, p = 0.0022; wild type: n = 10 mice, Kdm6a KO/male: n = 7 mice, Kdm6a KO/female: n = 7). g) Top: schematic time course of optic nerve regeneration model. Optic nerve regeneration was analyzed 2 weeks after the nerve crush, and the fluorescence (Alexa 594) tagged CTB was injected 2 days before collecting the optic nerve. Bottom: representative 2D projected confocal images of whole mount cleared optic nerves of Kdm6a mutant knockin (Kdm6a‐mut KI) and wild type mice 2 weeks after ONC. The regenerating axons were labeled with CTB‐Alexa 594. The red dotted lines indicate the nerve crush sites. The columns on the right display magnified images of the areas in white, dashed boxes on the left, showing axons at 0.25, 0.75, and 1.25 mm distal to the crush sites. Scale bar, 200 µm for the left column and 150 µm for the magnified images. h) Quantification of regenerating axons at different distances distal to the nerve crush site (0.25–1.75 mm) in Kdm6a‐mut KI and wild type mice after ONC. Bar graph showing robust optic nerve regeneration in Kdm6a‐mut KI mice (two tailed student's t tests, p < 0.05; Wild type: n = 5 mice, Kdm6a‐mut KI: n = 5 mice). i) Quantification of the average length of the top 5 longest axons of each nerve in (g) (two tailed student's t tests, p = 0.0087; Wild type: n = 5 mice, Kdm6a‐mut KI: n = 5 mice). j) Representative images of flat whole mount retina immunostained for neuronal marker Tuj1 (green), indicating increased RGC survival rates in either male Kdm6a knockout (Kdm6a KO/male) or female Kdm6a knockout (Kdm6a KO/female) mice compared with that in wild type mice. Scale bar, 50 µm. k) Quantification of RGC survival rates in wild type, male Kdm6a knockout or female Kdm6a knockout mice after ONC (one way ANOVA followed by Tukey's multiple comparisons test, p < 0.0001; wild type: n = 6 mice, Kdm6a KO/male: n = 6 mice, Kdm6a KO/female: n = 5). l) Representative images of flat whole mount retina immunostained for neuronal marker Tuj1 (green), indicating increased RGC survival rates in female Kdm6a knockout (Kdm6a KO/female) mice compared with that in wild type mice after NMDA induced injury. Scale bar, 50 µm. m) Quantification of RGC survival rate in (l) (two tailed student's t tests, p = 0.0485; n = 3 mice for each condition). *, **, ***p < 0.05, 0.01, 0.001, respectively. Data are represented as mean ± SEM. See also Figures S1–S3 (Supporting Information).

Figure 3

Knocking out H3K27 demethylase Kdm6b in mouse RGCs markedly enhanced neuronal survival, but poorly promoted axon regeneration. a) Representative images of Kdm6b immunostaining in retina sections showing increased level of Kdm6b in RGCs after ONC. Images were stained for neuronal marker Tuj1 (red) and Kdm6b (green). Scare bar, 50 µm. b) Enlarged images of areas indicated by the dashed white boxes in (a). Scale bar, 10 µm. c) Quantification of relative fluorescence intensity of Kdm6b immunostaining in (a) (one way ANOVA followed by Tukey's multiple comparison test, p = 0.0004; n = 3 independent experiments for each condition). d) Top: schematic time course of optic nerve regeneration model. Bottom: representative 2D projected confocal images of whole mount cleared optic nerves of wild type, Kdm6b knockout (Kdm6b KO), Kdm6a and Kdm6b double knockout (Kdm6a/b dKO) mice 2 weeks after ONC. The regenerating axons were labeled with CTB‐Alexa 594. Yellow arrows indicate the distal ends of regenerating axons. The red dotted lines indicate the nerve crush sites. The columns on the right display magnified images of the areas in white, dashed boxes on the left, showing axons at 0.25, 0.75, and 1.25 mm distal to the crush sites. Scale bar, 200 µm for the left column and 100 µm for the magnified images. e) Quantification of regenerating axons at different distances distal to the nerve crush site (0.25–1.50 mm) in wild type or Kdm6b knockout mice (Kdm6b KO) after ONC. Bar graph showing no difference in axon regeneration between wild type and Kdm6b KO (two tailed student's t tests, p > 0.05; wild type: n = 10 mice, Kdm6b KO: n = 6 mice). f) Quantification of regenerating axons at different distances distal to the nerve crush site (0.25–1.50 mm) in female Kdm6a knockout (Kdm6a KO/female) or female Kdm6a/Kdm6b double knockout mice (Kdm6a/b dKO/female) after ONC. Bar graph showing no additive/synergistic effects of Kdm6a/Kdm6b co‐deletion on optic nerve regeneration (two tailed student's t tests, p > 0.05; Kdm6a KO: n = 7 mice, Kdm6a/b dKO: n = 4 mice). g) Representative images of flat whole mount retina immunostained for neuronal marker Tuj1 (green), indicating increased RGC survival rates in either Kdm6b knockout (Kdm6b KO) or Kdm6a/Kdm6b double knockout (Kdm6a/b dKO) mice compared with that in wild type mice. Co‐deleting Kdm6a/Kdm6b resulted in additive/synergistic effects on RGC survival. Scale bar, 50 µm. h) Quantification of RGC survival rate in (g) (one way ANOVA followed by Tukey's multiple comparisons test, p < 0.0001; n = 3 mice for each condition). *, **, ***p < 0.05, 0.01, 0.001, respectively. Data are represented as mean ± SEM. See also Figures S2 and S3 (Supporting Information).

Figure 4

Pharmacological delayed inhibition of Kdm6a/b in mouse RGCs also promoted optic nerve regeneration and RGC survival. a) Confocal images of retinal sections from wild type mice injected with Kdm6a/b inhibitor GSK‐J4 or DMSO (as control) showing significantly increased level of H3K27me3 in RGCs by blockade of histone demethylases activity. The sections were stained for neuronal marker Tuj1 (red), which labeled RGCs, and H3K27me3 (green). Scale bar, 50 µm. b) Quantification of relative fluorescence intensity of H3K27me3 immunostaining shown in (a) (two tailed student's t tests, p = 0.0091; n = 3 mice for each condition). c) Top: schematic time course of optic nerve regeneration model. Specifically, GSK‐J4 or DMSO was injected once every 3 days for 4 times after ONC. Bottom: representative 2D projected confocal images of whole mount cleared optic nerves of wild type mice with intravitreal injection of GSK‐J4 or DMSO 2 weeks after ONC. The regenerating axons were labeled with CTB‐Alexa 594. The red dotted lines indicate the nerve crush sites. The columns on the right display magnified images of the areas in white, dashed boxes on the left, showing axons at 0.25, 0.50, and 0.75 mm distal to the crush sites. Scale bar, 200 µm for the left column and 130 µm for the magnified images. d) Quantification of regenerating axons at different distances distal to the nerve crush site (0.25–0.75 mm) in injected GSK‐J4 or DMSO mice after ONC. Bar graph showing robust optic nerve regeneration induced by GSK‐J4 (two tailed student's t tests, p < 0.0001; DMSO: n = 11 mice, GSK‐J4: n = 7 mice). e) Quantification of the average length of the top 5 longest axons of each nerve in (c) (two tailed student's t tests, p = 0.0004; DMSO: n = 8 mice, GSK‐J4: n = 4 mice). f) Representative images of flat whole mount retina immunostained for neuronal marker Tuj1 (green), indicating increased RGC survival rates in injected GSK‐J4 mice compared with that in mice with DMSO injection. Scale bar, 50 µm. g) Quantification of RGC survival rate in (f) (two tailed student's t tests, p = 0.0063; DMSO: n = 8 mice, GSK‐J4: n = 5 mice). **, ***p < 0.01, 0.001, respectively. Data are represented as mean ± SEM.

Figure 5

Synergistic promoting effects of Kdm6a and Pten double knockout on axon regeneration after ONC. a) Confocal images of retinal sections from Kdm6a^f/f mice injected with AAV2‐GFP or AAV2‐Cre showing no effect of Kdm6a single deletion on the percentage of p‐S6+ RGCs in the ganglion cell layer (GCL). The sections were stained for neuronal marker Tuj1 (red), which labeled RGCs, and p‐S6 (gray). Scale bar, 50 µm. b) Quantification of p‐S6+ RGCs in (a) (two tailed student's t tests, p = 0.8747; n = 3 mice for each condition). c) Confocal images of retinal sections from Kdm6a/Pten^f/f mice injected with AAV2‐GFP or AAV2‐Cre showing markedly increased percentage of p‐S6+ RGCs in the GCL via Cre‐mediated double deletion. The sections were stained for neuronal marker Tuj1 (red), which labeled RGCs, and p‐S6 (gray). Scale bar, 50 µm. d) Quantification of p‐S6+ RGCs in (c) (two tailed student's t tests, p = 0.0002; n = 3 mice for each condition). e) Confocal images of retinal sections from Kdm6a/Pten^f/f mice injected with AAV2‐GFP or AAV2‐Cre showing significantly increased level of H3K27me3 in RGCs via Cre‐mediated double deletion. The sections were stained for neuronal marker Tuj1 (red), which labeled RGCs, and H3K27me3 (gray). Scale bar, 50 µm. f) Quantification of relative fluorescence intensity of H3K27me3 immunostaining shown in (e) (two tailed student's t tests, p = 0.0252; n = 3 mice for each condition). g) Top: schematic time course of optic nerve regeneration model. Bottom: representative 2D projected confocal images of whole mount cleared optic nerves of Pten knockout (Pten KO) or Kdm6a/Pten double knockout (Kdm6a/Pten dKO) mice 2 weeks after ONC. The regenerating axons were labeled with CTB‐Alexa 594. Yellow arrows indicate the distal ends of regenerating axons. The red dotted lines indicate the nerve crush sites. The columns on the right display magnified images of the areas in white, dashed boxes on the left, showing axons at 1.00, 2.00, and 3.00 mm distal to the crush sites. Scale bar, 500 µm for the left column and 250 µm for the magnified images. h) Quantification of regenerating axons at different distances distal to the nerve crush site (0.25–3.00 mm) in Pten knockout (Pten KO) or Kdm6a/Pten double knockout mice (Kdm6a/Pten dKO) 2 weeks after ONC (two tailed student's t tests, p < 0.05; Pten KO: n = 4 mice, Kdm6a/Pten dKO: n = 6 mice). i) Quantification of the average length of the top 5 longest axons of each nerve in (g) (two tailed student's t tests, p = 0.0184; Pten KO: n = 4 mice, Kdm6a/Pten dKO: n = 6 mice). j) Top: schematic time course of optic nerve regeneration model. The optic nerve was crushed at the distal intraorbital site. Optic nerve regeneration was analyzed 4 weeks after the nerve crush. Bottom: Representative 2D projected confocal images of whole mount cleared optic chiasm areas of Pten knockout (Pten KO) or Kdm6a/Pten double knockout (Kdm6a/Pten dKO) mice 4 weeks after distal intraorbital ONC. The right column shows enlarged images of areas indicated by the dashed white boxes in the left column. The regenerating axons at the optic chiasm were labeled with CTB‐Alexa 594. Yellow arrows indicate the distal ends of regenerating axons. Blue and yellow dotted lines indicate the chiasm midline and the OCTZ, respectively. OX indicates the optic chiasm. Scale bar, 100 µm. k) Quantification of regenerating axons reaching the optic chiasm in (j) (two tailed student's t tests, p = 0.0125; Pten KO: n = 3 mice, Kdm6a/Pten dKO: n = 5 mice). n.s. p > 0.05; *, **, ***p < 0.05, 0.01, 0.001, respectively. Data are represented as mean ± SEM. See also Figure S4 (Supporting Information).

Figure 6

Kdm6a or Kdm6a/Pten knockout induced changes of axonal morphology. a) Representative images of optic nerves showing that Kdm6a deletion modified RGC axonal morphology 2 weeks after ONC. Neurite tracing of single axon showing straight trajectories (green lines) and U‐turns (red lines). Red and green arrows indicate retraction bulbs and growth cones, respectively. Scale bar, 200 µm. b) Representative images of growth cones and retraction bulbs found in different optic nerves. Scale bar, 25 µm. c) Quantification of the size of retraction bulbs and growth cones by tip/axon shaft ratio in (a) (one way ANOVA followed by Tukey's multiple comparisons test, p < 0.0001; n = 105, 78, and 124 axons pooled from 7 mice for wild type, 3 mice for Kdm6a KO/male, and 7 mice for Kdm6a KO/female, respectively). d) Quantification of U‐turn rate in (a) (one way ANOVA followed by Tukey's multiple comparisons test, p = 0.0007; wild type: n = 6 mice, Kdm6a KO/male: n = 6 mice, Kdm6a KO/female: n = 7; top 20 longest axons were analyzed for each mouse). e) Representative images of U‐turns and branches found in the distal nerve region. Neurite tracing showing that co‐deletion of Kdm6a/Pten in RGCs led to markedly decreased U‐turns and increased branches compared to Pten deletion alone. Green lines, red lines and blue lines indicate straight trajectories, U‐turns and branches, respectively. Scale bar, 200 µm. f) Quantification of U‐turn rate in (e) (two tailed student's t tests, p = 0.0013; Pten KO: n = 4 mice, Kdm6a/Pten dKO: n = 6 mice; top 20 longest axons were analyzed for each mouse). g) Quantification of branching index by tip number/axon number (the top 20 longest axons) ratio in (e) (two tailed student's t tests, p = 0.0088; Pten KO: n = 4 mice, Kdm6a/Pten dKO: n = 6 mice). h) Neurite tracing of regenerating axon into the optic chiasm showing complex growth patterns in Kdm6a/Pten double knockout mice. The bottom row shows enlarged images of areas indicated by the dashed white boxes in the top row. The regenerating axons at the optic chiasm were labeled with CTB‐Alexa 594. Yellow arrows indicate the distal ends of regenerating axons. Blue and yellow dotted lines indicate the chiasm midline and the OCTZ, respectively. Green lines, red lines and blue lines indicate straight trajectories, U‐turns and branches, respectively. OX indicates the optic chiasm. Scale bar, 200 µm. i) Distribution of regenerating axons reaching the distal optic nerve (Pre‐chiasmic) and entering the optic chiasm (Post‐chiasmic) in (h). The dashed lines showing quantification of post‐chiasmic axons projecting into the ipsilateral optic tract (Ipsi optic tract), contralateral optic tract (Contra optic tract), and contralateral optic nerve (Contra optic nerve) in 5 individual mice. N/A, not applicable. j) Quantification of U‐turn rate in (h) (n = 5 mice; all axons into the optic chiasm were analyzed for each mouse). k) Quantification of branching index by tip number/axon number (regenerating axons into the optic chiasm) ratio in (h) (n = 5 mice). *, **, ***p < 0.05, 0.01, 0.001, respectively. Data are represented as mean ± SEM. See also Figure S4 (Supporting Information).

Figure 7

Deleting Kdm6a switched the RGC transcriptomics into a developmental‐like state. a) PCA of transcriptomic profiles indicating the variation among all samples. Score plot shows a clear separation between P1, P14, P21, wild type (AAV2‐GFP) and Kdm6a KO (AAV2‐Cre). The first two dimensions (Dim) accounted for 70.7% of variance. Note that Dim 1 predominantly reflects differences between wild type and Kdm6a KO. b) Selected GO terms were generated separately for Dim 1 and 2 genes (adjusted p‐value < 0.05). c) Number of DEGs between wild type (AAV2‐GFP) and Kdm6a KO (AAV2‐Cre) at the threshold of absolute log₂ Fold change > 1 and adjusted p‐value < 0.05. d) Volcano plot showing differential gene expression between wild type (AAV2‐GFP) and Kdm6a KO (AAV2‐Cre) RGCs. Positive or negative log₂ Fold change indicates upregulation or downregulation in Kdm6a KO RGCs relative to wild type RGCs, respectively. Gray points (NS) indicate the genes with no significant change. Green points (Log₂ FC) indicate the genes with absolute log₂ Fold change > 1. Blue points (p.adjust) indicate the genes with adjusted p‐value < 0.05. Red points (p.adjust and log₂ FC) indicate that the genes are considered significantly different if absolute log₂ Fold change > 1 and adjusted p‐value < 0.05; vertical and horizontal reference lines at respective values. e) Selected GO terms were generated separately for genes upregulated or downregulated (adjusted p‐value < 0.05) by Kdm6a KO RGCs relative to wild type RGCs. See also Figures S5–S7 (Supporting Information).

Figure 8

Klf4 was identified as a downstream mediator of Kdm6a function in optic nerve regeneration. a) Heatmap of selected genes (absolute log₂ Fold change > 1) relevant to axon regeneration and neuroprotection. b) Sample genome tracks showing reduced expression levels of Klf4 (chromosome 4 (chr4): 55527143–55536984) in the purified RGCs of Kdm6a^f/f mice injected with AAV2‐Cre, compared to Kdm6a^f/f mice with AAV2‐GFP injection. y axis indicates normalized reads. c) Top: Schematic drawing of the 5 kilobase (kb) genomic regions (R1–R5) after the TSS of the Klf4 gene on chr4 were assayed in CUT&Tag‐qPCR experiment using the antibody against H3K27me3 in the purified RGCs. In the H3K27me3 CUT&Tag‐qPCR experiment, the genomic region R1–R5 were amplified from the purified RGCs to bind to H3K27me3. Bottom: CUT&Tag‐qPCR results showing that deletion of Kdm6a (Kdm6a KO) led to increased interaction with R1, R2, and R3 regions of the Klf4 gene (two tailed student's t tests, R1: p = 0.0402; R2: p = 0.0240; R3: p = 0.0378; n = 3 independent experiments for each condition). d) Top: schematic time course of optic nerve regeneration model. Bottom: representative 2D projected confocal images of whole mount cleared optic nerves of wild type, female Kdm6a knockout (Kdm6a KO), female Kdm6a knockout/Klf4 overexpression (Kdm6a KO/Klf4 OE) mice 2 weeks after ONC. The regenerating axons were labeled with CTB‐Alexa 594. The red dotted lines indicate the nerve crush sites. The columns on the right display magnified images of the areas in white, dashed boxes on the left, showing axons at 0.25, 0.50, and 1.00 mm distal to the crush sites. Scale bar, 200 µm for the left column and 150 µm for the magnified images. e) Quantification of regenerating axons at different distances distal to the nerve crush site (0.25–1.00 mm) (one way ANOVA followed by Tukey's multiple comparisons test, p < 0.0001; wild type: n = 6 mice, Kdm6a KO: n = 7 mice, Kdm6a KO/Klf4 OE: n = 7). f) Real‐time PCR results showing the expression of downregulated genes in RGCs from Kdm6a knockout mice (Kdm6a KO) after ONC (one sample t test, Rabgef1: p = 0.0019; Tnfsf13: p = 0.0095; Pias3: p = 0.0111; n = 3 independent experiments for each condition). *, **, ***p < 0.05, 0.01, 0.001, respectively. Data are represented as mean ± SEM. See also Figures S8 and S9 (Supporting Information).

Figure 9

Kdm6a and Kdm6b play different roles in regulation of PNS and CNS neural regeneration. Top panel: Kdm6a and Kdm6b act similarly in the PNS to regulate sensory axon regeneration. Sciatic nerve crush‐induced injury signal leads to spontaneous sensory axon regeneration of DRG neurons. During this process, the protein levels of histone demethylase Kdm6a and Kdm6a are reduced in the nucleus. In the meantime, the level of H3K27me3 is elevated, together with decreased expression level of the regeneration repressor gene Klf4, which is enriched in H3K27me3 mark. Knocking down either Kdm6a or Kdm6b before nerve injury mimics the injury signal, leading to elevated level of H3K27me3 and enhanced axon regeneration. Finally, overexpression of Klf4 in sensory neurons significantly blocked sensory axon regeneration. Bottom panel: Kdm6a and Kdm6b act differently in the CNS to regulate optic nerve regeneration. Unlike that of sensory neurons, after the optic nerve axonal injury the level of Kdm6a in either the cytoplasm or the nucleus of RGCs remains the same. Interestingly, Kdm6b is mainly located in the RGC cytoplasm and its level is slightly elevated after the injury. As a result, under control condition the H3K27me3 level remains low after the injury and axon regeneration repressors (Klf4 and other unknown factors) function to suppress axon regeneration. Knocking out Kdm6a results in elevated level of H3K27me3 and reduced level of Klf4, leading to enhanced optic nerve regeneration. In contrast, knocking out Kdm6b mainly located in the RGC cytoplasm has no effect on the H3K27me3 level. With Kdm6a still present in the nucleus, the expression of axon growth repressors (Klf4, etc.) leads to failed axon regeneration. In addition to failed axon regeneration, optic nerve injury leads to significant neuronal cell death. Knocking out either Kdm6a or Kdm6b resulted in enhanced RGC survival with elusive underlying mechanism. The dotted lines indicate pathways lacking direct experimental support in this study. Schematic was created with BioRender (www.biorender.com).