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.

Introduction

Axon regeneration in the mammalian central nervous system (CNS) has been a long standing and highly challenging issue in the biomedical research field. A major reason for the failed CNS axon regeneration is that mature mammalian CNS neurons permanently lose their intrinsic ability to support axon growth during maturation.^[1–3] Recent studies have identified many genes that could be manipulated to significantly boost the intrinsic axon regeneration ability of mature CNS neurons, including Pten, Klf4/6/7/9, Gsk3b, Socs3, c-Myc, B-Raf, and Lin28.^[4–12] However, the molecular mechanisms, especially at the transcriptional and chromatin regulation level, by which these genes act coordinately to regulate axon regeneration are still rudimentary and fragmented. Thus, understanding the intrinsic transcriptomic networks and gene expression profile specifically governing axon growth during development and regeneration is of great importance in the neural regeneration field.

Epigenetic regulation acts to control chromatin structure, gene transcription, translation and the subsequent cellular state,^[13] making it an ideal regulatory mechanism for coordinating gene expression during axon regeneration. For instance, several studies, including ours,^[11,14–19] have illustrated the important roles of microRNAs in regulation of peripheral nervous system (PNS) axon regeneration. In addition, DNA methylation has also been shown to be involved in axon regeneration.^[20–22] Dynamic histone acetylation could control chromatin accessibility and gene transcription of regeneration-associated genes (RAGs) to support the regenerative capacity.^[23–26] Moreover, our study^[16] has provided in vivo evidence that microRNA-138 and histone deacetylase Sirt1 form a mutual negative feedback loop to control sensory axon regeneration, suggesting crosstalk between the two epigenetic pathways. Furthermore, a recent study^[27] using multiomics sequencing characterized the unique chromatin signature and associated transcription profile during spontaneous sensory axon regeneration, highlighting the key roles of histone acetylation in supporting axon regeneration. Although these previous studies provided clear evidence about the involvement of epigenetic signaling in neural regeneration, to our knowledge, to date very few studies^[28,29] have demonstrated that manipulation of a single epigenetic regulator alone is able to induce long-distance CNS axon regeneration.

Tri-methylation of histone 3 lysine 27 (H3K27me3) functions to induce chromatin compaction and therefore silences selected gene transcription. Our recent studies investigated the roles of H3K27 methyltransferase Ezh2 in neural development and regeneration, revealing Ezh2 as a major factor shaping neural structure. Specifically, we showed that in postmitotic neurons Ezh2 acted to control multiple steps of neuronal morphogenesis during development, thereby leading to cognitive consequences in adult mice.^[30] Our latest study^[28] showed that Ezh2 acted to support long-distance axon regeneration in both PNS and CNS via coordinating gene transcription in multiple regenerative pathways. To date, three H2K27 demethylases have been identified, including Lysine (K)-specific demethylase 6a encoded by Kdm6a (or Utx) located on the X chromosome, Kdm6b encoded by Kdm6b (or Jmjd3), and Kdm6c encoded by Kdm6c (or Uty) located on the Y chromosome. They act specifically to erase trimethyl groups of H3K27me3, thereby generating transcriptionally permissive chromatin.^[31,32] As the paralog of Kdm6a, Kdm6c has been shown to have much reduced enzymatic activity.^[31,33,34] The roles and mechanisms of both Kdm6a and Kdm6b have been extensively studied for their involvement in development,^[31,32,35] differentiation,^[36,37] cell plasticity,^[38] aging,^[39–42] cell reprogramming,^[43,44] inflammatory response,^[45,46] tumorigenesis,^[47,48] and neurodegeneration.^[49,50] Previous studies have shown that Kdm6a and Kdm6b could play either similar or distinct roles in different biological processes. For instance, in mouse spinal motor neurons Kdm6b but not Kdm6a functions to regulate subtype diversification during development.^[51] In contrast, Kdm6a but not Kdm6b is able to act as a direct sensor for cellular oxygen level and subsequent changes in chromatin and cell fate.^[52] In the nervous system, our recent study^[53] demonstrated that Kdm6a functioned to regulate neuronal dendritic development and various cognitive functions through demethylation of H3K27me3. In addition, previous studies suggested that Kdm6b served as a crucial regulator of neuronal maturation and survival.^[54,55] There is early evidence that PNS axotomy can trigger dynamic changes of histone methylation in sensory neurons.^[26] However, to date, the functional role of histone methylation in axon regeneration is still under intensive investigation.

Here, we report that deleting Kdm6a or Kdm6b significantly promotes sensory axon regeneration in vivo. Importantly, deleting Kdm6a, but not Kdm6b, in retinal ganglion cells (RGCs) markedly enhanced optic nerve regeneration. Mechanistically, our data demonstrated that Kdm6a acted via a Pten/PI3K independent pathway to enhance optic nerve regeneration. Moreover, by analyzing and comparing transcriptomic changes of purified RGCs during development/maturation to that of Kdm6a knockout, we found that deleting Kdm6a reactivated developmental-like growth programs via modified chromatin structure and transcriptomics in RGCs. Moreover, we identified several known axon growth repressor genes as downstream targets suppressed by the Kdm6a-H3K27me3 signaling. In particular, we provided strong evidence that the transcription factor Klf4 was directly regulated by H3K27me3, and overexpressing Klf4 in either adult sensory neurons or RGCs significantly blocked axon regeneration in vivo. In addition, deletion of either Kdm6a or Kdm6b markedly increased RGC survival, and co-deleting Kdm6a/b exhibited additive effect on RGC protection, indicating their distinct underlying mechanisms. Collectively, our findings revealed that Kdm6a played key roles in repressing either PNS or CNS axon regeneration, whereas Kdm6b was not involved in regulating CNS axon regeneration. Moreover, both demethylases were able to regulate CNS neuronal survival after injury, but with non-overlapping mechanisms. Finally, we provided evidence that deleting Kdm6a functioned to enhance mammalian CNS neural regeneration by reshaping the neuronal transcriptomic landscape to a developmental-like state. Our study not only characterized the novel but non-overlapping roles of Kdm6a and Kdm6b in regulation of mammalian neural regeneration, but also discovered promising new molecular targets for promoting neuroprotection and axon regeneration.

Results

Kdm6a and Kdm6b Are Negative Regulators of Spontaneous Sensory Axon Regeneration

To characterize the roles of demethylases Kdm6a/b in axon injury and regeneration, we first examined how they acted in sensory neurons in the dorsal root ganglion (DRG) during spontaneous axon regeneration. Time-course analyses by real-time PCR showed that the mRNA levels of Kdm6a were gradually downregulated after sciatic nerve injury (SNI), reaching its minimum at day 7 post-SNI (Figure 1a). Similarly, the levels of Kdm6b were also significantly reduced 1-day post-SNI but recovered back at day 3 and 7 (Figure 1a). We then performed immunostaining analyses to further examine the protein levels of Kdm6a or Kdm6b in sensory neurons in vivo. Confocal imaging of DRG sections showed that Kdm6a and Kdm6b were presented in both the nucleus and cytoplasm of sensory neurons before SNI. After SNI, the protein levels of Kdm6a in both the nuclei and cytoplasm of sensory neurons were significantly reduced 1 and 3 days later (Figure 1b,c). Kdm6b staining demonstrated different change patterns after SNI with its protein level only reduced in both the nucleus and cytoplasm 1 day after SNI. In day 3 post SNI, the level of Kdm6b was restored in the cytoplasm but kept low in the nucleus (Figure 1d,e). As histone demethylases, both proteins function in the nucleus. Thus, these data suggested that Kdm6a/b might act to suppress sensory axon regeneration. We therefore tested if downregulation of Kdm6a or Kdm6b in sensory neurons could enhance axon regeneration in vivo. We used our well-established in vivo electroporation technique^[56] to co-transfect sensory neurons with a group of 4 different siRNAs against Kdm6a or Kdm6b and the EGFP plasmid. According to our previous study, the transfection efficiency of fluorescence labeled siRNAs and EGFP via in vivo electroporation is over 96%,^[11] and 5–10%,^[56] respectively. Therefore, we considered all EGFP positive neurons as siRNA positive. Western blot results showed that siRNAs significantly reduced the protein level of Kdm6a or Kdm6b in vivo (Figure 1f,g). Furthermore, we found that knocking down Kdm6a/b resulted in a significant elevation of H3K27me3 levels in DRG neurons (Figure 1h). To assess the functional roles of Kdm6a or Kdm6b in sensory axon regeneration, we performed SNI 2 days after electroporation and analyzed axon regeneration 2 days post-SNI when spontaneous sensory axon regeneration was just starting (Figure 1i). The results showed that knocking down Kdm6a or Kdm6b in sensory neurons markedly enhanced axon regeneration in vivo (Figure 1j–m). Together, these results demonstrated clearly that demethylases Kdm6a/b function similarly as important repressors of sensory axon regeneration.

[IMAGE OMITTED. SEE PDF]

Kdm6a Knockout Enhances Optic Nerve Regeneration and RGC Survival

Since the role of Kdm6a in sensory axon regeneration has been confirmed, we explored how Kdm6a regulated CNS neural regeneration using the optic nerve regeneration model. To examine the mRNA levels of Kdm6a in RGCs, we enriched the Thy1.2 positive RGCs by immunomagnetic cell separation (MACS) and the positive rate of Thy1.2 in positive fraction was 80.8 ± 4.43% based on flow cytometric analysis (Figure S1a,b, Supporting Information). Unlike that of sensory neurons, real-time PCR analysis showed that the mRNA levels of Kdm6a in RGCs were not changed by optic nerve crush (ONC) (Figure S1c, Supporting Information). We next performed immunostaining analysis to examine the protein levels of Kdm6a in RGCs after ONC. The results showed that Kdm6a was highly enriched in the nuclei of RGCs and its level was unchanged after ONC (Figure 2a–c), confirming the real-time PCR results. By using an available single-cell RNA-seq (scRNA-seq) dataset of adult mouse RGCs,^[57] we further confirmed that the expression levels of Kdm6a in RGCs were unchanged at different time points after ONC (Figure S2a,b, Supporting Information). The unchanged levels of Kdm6a in RGCs upon ONC are in stark contrast to that occurred in sensory neurons upon SNI, suggesting that Kdm6a in RGCs act to suppress optic nerve regeneration.

[IMAGE OMITTED. SEE PDF]

We therefore obtained the Kdm6a floxed mice to examine its role in optic nerve regeneration. Because the Kdm6a gene is located on the X chromosome, the female mice have both alleles of Kdm6a floxed (Kdm6a^f/f), whereas the male mice have only allele floxed with the kdm6c (Uty) gene on the Y chromosome (Kdm6a^f/y). Kdm6a was knocked out by intravitreal injection of adeno-associated virus (AAV2) vector encoding the Cre recombinase. The transduction rate of AAV2-Cre in RGCs was 88.8 ± 7.56% (n = 2 retinas), evaluated by co-immunostaining the whole mount retina with antibodies against Tuj1 and Cre (Figure S3a, Supporting Information). The mRNA level of Kdm6a in the purified RGCs from the Kdm6a^f/f/AAV2-Cre mice was markedly downregulated compared with that from the control Kdm6a^f/f/AAV2-GFP mice (Figure S3b, Supporting Information). In consistent, immunostaining analysis 2 weeks after viral infection showed significantly reduced level of Kdm6a in RGCs in AAV2-Cre injected retinas (Figure S3c,d, Supporting Information). As expected, deleting the demethylase Kdm6a resulted in significantly elevated level of H3K27me3 in RGCs (Figure S3e,f, Supporting Information).

To determine how knocking out Kdm6a in RGCs affects RGC survival and optic nerve regeneration, we performed ONC two weeks after the virus injection in Kdm6a^f/f/AAV2-Cre (female), Kdm6a^f/y/AAV2-Cre (male) or the control mice (wild type littermate mice injected with AAV2-Cre). Two weeks after ONC, RGC axons were labeled anterogradely by intravitreal injection of the Alexa-594-conjugated cholera toxin β (CTB) and optic nerve regeneration was assessed (Figure 2d). Fluorescence 3D images of tissue-cleared optic nerves were acquired with confocal microscopy as described in our previous studies.^[11,28,58] Wild type mice showed minimal optic nerve regeneration passing the crush sites, whereas knocking out Kdm6a in female mice significantly promoted optic nerve regeneration (Figure 2d–f). Knocking out Kdm6a in male mice also induced enhanced optic nerve regeneration, but to a much lesser degree (Figure 2d–f), likely due to the presence of the Kdm6c gene on the Y chromosome with reduced demethylase activity.^[59] In addition, we obtained Kdm6a enzyme-dead knockin mice (Kdm6a-mut KI) which possess the H1146A and E1148A point mutations in exon 24.^[60] To investigate the role of Kdm6a enzymatic activity in optic nerve regeneration, we first performed ONC in either Kdm6a-mut KI mice or the control mice (wild type). Two weeks later, optic nerve regeneration was analyzed by quantifying the number of CTB-Alexa 594 labeled axons (Figure 2g). The number and length of regenerating axons in Kdm6a-mut KI mice were enhanced significantly compared to those in wild type mice, providing further evidence that the histone demethylase activity of Kdm6a functions to repress optic nerve regeneration (Figure 2g–i). Next, we investigated whether Kdm6a affected the RGC survival with two widely used in vivo cell death models, including the ONC and the NMDA receptor-mediated excitotoxicity, a shared pathological pathway of neurodegenerative diseases. The results showed that knocking out Kdm6a in either female or male mice significantly protected RGCs from cell death to the same extent (Figure 2j,k). These results indicated that Kdm6c on the Y chromosome was not involved in ONC-induced RGC death, different from that of axon regeneration. In addition, deleting Kdm6a also markedly increased RGC survival after NMDA treatment (Figure 2l,m), suggesting its wide range of neuroprotection after different types of neural injuries.

Knocking Out Kdm6b Enhances RGC Survival, but Poorly Promotes Axon Regeneration

We next assessed the effects of deleting Kdm6b in adult RGCs on optic nerve regeneration. The expression pattern of Kdm6b within RGCs was firstly analyzed at different time points after ONC. The dot plot and violin plot obtained from scRNA-seq dataset^[57] showed that the mRNA level of Kdm6b was increased upon ONC (Figure S2a,c, Supporting Information). By performing immunofluorescence staining, we further confirmed the upregulation of Kdm6b in RGCs after ONC (Figure 3a–c). In contrast to Kdm6a, the staining results showed that Kdm6b was mainly localized in the cytoplasm of RGCs before and after ONC, suggesting that it might not be the major histone demethylase in RGCs. Indeed, when Kdm6b in RGCs was knocked out with AAV2-Cre in Kdm6b^f/f mice (Figure S3g–i, Supporting Information), the level of H3K27me3 in RGCs was actually slightly decreased but without statistical significance (Figure S3j,k, Supporting Information). When optic nerve regeneration was quantified, knocking out Kdm6b did not significantly increase the number of regenerating axons at different distances from the crush site (Figure 3d,e). Moreover, when both Kdm6a and Kdm6b were knocked out, there was no additive promoting effect on optic nerve regeneration (Figures 2d and 3d,f) compared to Kdm6a knockout alone, further confirming that Kdm6b was not involved in regulating optic nerve regeneration. When RGC survival was examined, deleting Kdm6b significantly enhanced RGC survival, and double knocking out Kdm6a/b showed additive effect on RGC survival (Figure 3g,h). Together, these findings provided clear evidence that Kdm6b did not regulate optic nerve regeneration, in contrast to that of Kdm6a or that observed in sensory neurons. However, Kdm6b is able to regulate RGC survival but with different underlying mechanisms from that of Kdm6a.

[IMAGE OMITTED. SEE PDF]

Pharmacological Inhibition of Kdm6a/b in RGCs after ONC Still Promotes Optic Nerve Regeneration and RGC Survival

To explore the potential of translational application, we tested whether a delayed inhibition of Kdm6a in RGCs after the injury could also promote optic nerve regeneration. We used GSK-J4, a small molecule catalytic site inhibitor selective for the H3K27me3-specific demethylase subfamily (Kdm6a/b).^[61] GSK-J4 was injected intravitreally every 3 days for 2 weeks after the ONC. The immunostaining results showed that GSK-J4 treatment significantly increased H3K27me3 levels in RGCs, indicating the inhibitory effect on Kdm6a/b (Figure 4a,b). When axon regeneration was examined, the results showed that in control mice DMSO had little effect on optic nerve regeneration, whereas significant regeneration was observed in mice with GSK-J4 treatment (Figure 4c,d). The average lengths of the top 5 longest regenerating axons in mice injected with GSK-J4 were markedly longer than those in control mice with DMSO injection (Figure 4e). Delayed pharmacological inhibition of Kdm6a also promoted RGC survival (Figure 4f,g). Collectively, these results indicated that inhibiting the demethylase activity of Kdm6a after neural injury was able to effectively enhance CNS axon regeneration and neuronal survival.

[IMAGE OMITTED. SEE PDF]

Kdm6a Knockout Enhances Optic Nerve Regeneration via Distinct Pathways from That of Pten Deletion

Activation of the mammalian target of rapamycin (mTOR) signaling in adult RGCs is one of the major pathways for promoting optic nerve axon regeneration downstream of several genes, such as Pten, Lin28, Osteopontin, and Akt/GSK3, etc.^[62] Thus, we investigated if Kdm6a deletion also enhanced optic nerve regeneration through the same pathway. Immunostaining of retina sections showed that at 3 days after injury, deleting Kdm6a alone in RGCs had no effect on the percentage of phospho-S6 (p-S6) positive RGCs (Figure 5a,b), indicating that the mTOR pathway was not activated. In contrast, Kdm6a/Pten co-deletion dramatically increased the percentage of p-S6+ RGCs (Figure 5c,d), as well as the level of H3K27me3 in RGCs (Figure 5e,f). These results indicated that Kdm6a knockout led to novel downstream signaling pathways distinct from that of Pten deletion and mTOR activation. We therefore examined the effects of co-deleting Kdm6a and Pten in adult RGCs on optic nerve regeneration. AAV2-Cre were injected into the vitreous body of Pten^f/f or Pten^f/f/Kdm6a^f/f mice to delete Pten or Pten/Kdm6a. Real-time PCR analysis showed that the mRNA levels of both Kdm6a and Pten were significantly reduced in Pten/Kdm6a double knockout mice (Figure S4a, Supporting Information). Optic nerve regeneration was examined 2 weeks after ONC. The results showed that concomitant deletion of Kdm6a and Pten in adult RGCs triggered faster optic nerve regeneration and more regenerative axons, compared with Pten deletion alone (Figure 5g–i). At 2 weeks after ONC, some long regenerative axons almost reached the region proximal to the optic chiasm (Figure 5g). Furthermore, the additive effects of Kdm6a/Pten double knockout were persistent to 6 weeks after ONC (Figure S4b–d, Supporting Information). In Pten single knockout mice almost no regenerating axons could be identified in the proximal end of the optic chiasm, whereas some regenerating axons in Kdm6a/Pten double knockout mice reached and crossed the optic nerve-chiasm transition zone (OCTZ) (Figure S4b, Supporting Information). To evaluate whether regenerating axons entered the optic chiasm and continued to grow into the optic tracts, we performed a distal intra-orbital ONC that shortened the distance axons needed to grow to reach the optic chiasm (Figure 5j). At 4 weeks after the injury, in Kdm6a/Pten double knockout mice many regenerating axons entered the chiasm, whereas in significant contrast almost no axons in Pten knockout mice reached the chiasm (Figure 5j,k). Together, these results demonstrated that co-deletion of Kdm6a and Pten could result in an additive effect on promoting long-distance axon regeneration.

[IMAGE OMITTED. SEE PDF]

Regenerating Axons Show Complex Growth Patterns in Kdm6a or Kdm6a/Pten Knockout Mice

Tissue clearing and 3D imaging allowed us to perform detailed analyses of axonal tip morphology and axon trajectory at the single axon level, revealing potential cellular mechanisms associated with axon regeneration. As described in our previous study,^[58] we first quantified the sizes of the distal axon ends (growth cones) of each regenerating axon. We found that Kdm6a deletion significantly reduced the sizes of the distal axon ends, compared to wild type (Figure 6a–c). Specifically, in wild type mice, the majority of weakly regenerated short axons had bulbous structures at the distal tips. About 20% of the axons formed the bigger terminal swellings (tip/shaft ratio > 4) defined as retraction bulb-like structures, which are the characteristic structures of non-regenerating axons (Figure 6c). In contrast, most regenerating axons in male or female mice with Kdm6a deletion showed slim growth cones (Figure 6a–c), leading to enhanced axon extension. Furthermore, detailed analysis of axon trajectory showed that in either wild type or Kdm6a knockout mice, within the optic nerve regenerating axons often had a meandering path and many of them made U-turns (Figure 6a,d), indicating the inhibitory nature of the mature visual system. Interestingly, deleting Kdm6a seemed to enhance the U-turn rate, likely due to increased number of regenerating axons.

[IMAGE OMITTED. SEE PDF]

Based on whole tissue 3D imaging, we also tracked the trajectories of individual RGC axons in the distal region of the optic nerve (Figure 6e). Within the distal optic nerve, many regenerating RGC axons induced by Pten deletion alone showed wandering trajectories with frequent U-turns (Figure 6f), similar to that of Kdm6a knockout alone (Figure 6d). However, the U-turn rate was markedly reduced when Kdm6a and Pten were both knocked out (Figure 6f and Figure S4e, Supporting Information). Moreover, we noticed that some regenerating RGC axons extended multiple branches from the axonal shafts or at the axon ends (Figure 6e). Quantification of axonal branching showed that compared to Pten deletion alone, the combination of Kdm6a and Pten deletion led to increased formation of axon branches (Figure 6g), indicating enhanced axon growth capacity.

To better examine how regenerating axons behaved at the optic chiasm, we analyzed morphological characteristics of regenerating axons passing the optic chiasm in the distal intra-orbital ONC mice with Kdm6a/Pten co-deletion (Figure 6h). We observed that in the mice with Kdm6a/Pten co-deletion, more than 60% of regenerating axons stopped growing or turned back at the entry area of the chiasm, indicating the optic chiasm as a significant inhibitory barrier (Figure 6h,i). For regenerating axons entering the chiasm most of them were found in the ipsilateral optic tract (Figure 6h,i), indicating axon misguidance. A few regenerating axons crossed the midline into the contralateral optic tract or extended into the opposite uninjured optic nerve (Figure 6h,i). We also assessed growth patterns of RGC axons entering the chiasm in Kdm6a/Pten double knockout mice (Figure 6h). The results showed that within the optic chiasm, ≈20% of axons made sharp turns, and axonal branches were observed in different regions (Figure 6h,j,k). Notably, RGC axons entering the chiasm showed reduced U-turns and axon branching (Figure 6j,k) compared with those within optic nerve (Figure 6f,g), suggesting spatial differences of the environment between the two regions.

Kdm6a Deletion in RGCs Triggers Activation of Developmental-Like Transcriptomic Programs

To further gain mechanistic insights into promoting effect of Kdm6a on axon regeneration, we performed RNA sequencing (RNA-seq) of the enriched RGCs across different developmental stages (postnatal day 1, 14, and 21) and at 3 days after injury of adult (P42) wild type (AAV2-GFP) or Kdm6a knockout mice (AAV2-Cre). Dissociated RGCs were labeled with Thy1.2 antibody, subjected to fluorescence-activated cell sorting (FACS) for RGC enrichment, and identified by labeling for Tuj1 (Figure S5, Supporting Information). Two RNA-seq datasets (13 bulk RNA-seq libraries) were generated from FACS-purified RGCs (Figure S6a, Supporting Information). It is widely believed that during neuronal maturation the changes in cellular states at the transcriptional level underlie their reduced ability to support regeneration. Young neurons have strong axon growth ability to form neural circuits, whereas mature neurons lose the ability to grow and change to support stable synaptic function.^[63] Significant differentially expressed genes (DEGs) were first recognized by comparing the gene expression profiles between P1 and P14 RGCs, but modest DEGs between P14 and P21 RGCs (Figure S6b,c,e, Supporting Information). These results indicated that RGCs changed their cellular states from young to mature between P1 and P14. Gene enrichment ontology (GO) analysis showed that downregulated DEGs in P14 RGCs compared to P1 RGCs were associated with axonogenesis, developmental cell growth and neuron projection guidance, supporting that axon growth ability reduced during neuronal maturation (Figure S6d,f, Supporting Information). We further compared the transcriptomic profiles of injured adult control or Kdm6a knockout RGCs (P42) to RGC maturational datasets. Hierarchical clustering analysis (HCA) revealed that the profiles were grouped into two main clusters with similar gene expression patterns (Figure S7a, Supporting Information). The first cluster included P1, P14, and AAV2-Cre (Kdm6a knockout P42 RGCs), while the second cluster involved P21 and AAV2-GFP (control P42 RGCs). In addition, the transcriptomic profiles of injured Kdm6a knockout RGCs (AAV2-Cre) were more similar to the P1 and P14 than to wild type control RGCs (AAV2-GFP) and P21 RGCs. Furthermore, principal component analysis (PCA), the linear dimensionality reduction method, was adopted for precisely screening the significant components and finding potential clusters across all samples. The results showed that the first principal component (Dim 1) contributes the most (60.6%) to explaining the variance (Figure 7a). Kdm6a knockout (AAV2-Cre) clustered close to P1 along Dim 1 compared to wild type (AAV2-GFP), suggesting the strong commonalities of Kdm6a knockout and developmental stage RGCs (Figure 7a). Next, we identified genes with significant contributions in the PCA clustering, especially relevant to Dim 1. GO enrichment analysis was conducted to reveal the biological roles of genes involved in Dim 1 and 2. As shown in Figure 7b, Dim 1 explained the most variance in the transcriptomic pattern of genes involved in postsynaptic organization, dendrite development, neuron projection organization, neuron apoptotic process, etc.

[IMAGE OMITTED. SEE PDF]

By calculating the z-score of genes contributed to the top 20 enriched GO terms, it showed distinct heterogeneity between wild type and other groups, but significant homogeneity between Kdm6a deletion and P14 (Figure S7b, Supporting Information). Meanwhile, double hierarchical clustered heatmap showed that many of these genes with top contribution in Dim 1 were associated with activation of developmental and growth programs. For instance, several genes, including Ehmt2,^[64] Brsk1,^[65] Eef1a2,^[66] and Lrfn4,^[67] have previously been shown to induce axon initiation, formation, sprouting, and outgrowth (Figure S7c, Supporting Information). Thus, this finding suggests that knocking out Kdm6a in adult injured RGC triggers reactivation of developmental-like growth programs, leading to an immature cellular state.

To further identify gene regulatory networks that control the changes mediated by Kdm6a deletion, we performed a comparative analysis that profiled transcription in enriched RGCs. In total, 574 common DEGs were identified (425 significantly upregulated and 149 downregulated) at the thresholds of |log₂ FC| > 1 (FC, fold change) and adjusted p‑value of < 0.05 (Figure 7c,d). Furthermore, GO enrichment analysis indicated that the majority of genes downregulated by Kdm6a deletion were associated with sensory perception, sensory system development, retina homeostasis and response to light stimulus (Figure 7e), most of which related to mature neuronal functions. On the other hand, the highly enriched GO terms of up-regulated DEGs were mainly related to regulation of virus, response to cytokine and lymphocyte activation (Figure 7e) that were mostly non-neuronal genes. Several significantly altered genes were involved in axon regeneration (Figure 8a). For example, Klf4,^[68] Myc,^[9] and Thbs1^[69] have been shown to play crucial roles in controlling axon regeneration. Additionally, Camk2 reactivation can protect RGCs in multiple injury/disease models.^[70] Taken together, these results suggested that Kdm6a deletion promoted optic nerve axon regeneration via activating developmental-like growth programs and repressing mature neuron enriched genes.

[IMAGE OMITTED. SEE PDF]

Klf4 Acts Downstream of Kdm6a to Regulate Axon Regeneration

H3K27me3 is well known to suppress gene expression. Here we showed that in sensory neurons upon peripheral axotomy the decreased level of Kdm6a/b and the subsequent elevated level of H3K27me3 acted to support the robust spontaneous sensory axon regeneration. Similarly, deletion of Kdm6a in RGCs resulted in an elevated level of H3K27me3, which in turn enhanced optic nerve regeneration. Therefore, it is very likely that the genes suppressed by H3K27me3 during axon regeneration function to repress axon regeneration.^[28] In cultured RGCs, several Klf family proteins have been shown to suppress axon growth, including Klf1, 2, 4, 5, 9, 13, 14, 15, 16, among which Klf4 is well-known to repress axon regeneration in vivo.^[4] When Klf4 expression was examined in RGCs after Kdm6a knockout, the results showed that it was markedly downregulated (Figure 8a,b). To investigate if Klf4 was directly targeted by Kdm6a knockout-induced H3K27me3 elevation, we performed CUT&Tag with real-time PCR (CUT&Tag-qPCR) analysis in enriched RGCs using anti-H3K27me3 antibody for selected regions at 0 to 5000 bp after the transcription start site (TSS) of the Klf4 gene on chromosome 4 (Figure 8c). The results revealed that comparing with wild type RGCs, in RGCs lacking Kdm6a the interactions between H3K27me3 and 4 specific regions of the Klf4 gene (R1-4) were significantly increased, especially at R1, R2, and R3 regions (Figure 8c), indicating that in RGCs Klf4 expression was suppressed by Kdm6a-mediated H3K27me3 elevation. More importantly, when Klf4 was overexpressed in RGCs, it markedly impeded optic nerve regeneration induced by Kdm6a knockout in vivo (Figure 8d,e). Together, these results provided strong evidence to support the hypothesis that Klf4 might act as a downstream mediator of Kdm6a-H3K27me3 signaling in RGCs to regulate optic nerve regeneration. In contrast, deleting Kdm6b in RGCs had no effect on the interactions between H3K27me3 and specific regions of the Klf4 gene (Figure S8, Supporting Information), consistent with the failure of Kdm6b knockout to promote optic nerve regeneration. In addition to Klf4, by examining downregulated genes in RGCs from Kdm6a knockout mice (Figure 8a) with real-time PCR analysis, several other axon growth repressors were identified as potential downstream targets suppressed by the Kdm6a-H3K27me3 signaling, including Rabgef1, Tnfsf13, and Pias3 (Figure 8f).

Because Klf4 has also been shown to inhibit sensory axon regeneration in vivo,^[71] we thus investigated if H3K27me3 could directly target Klf4 and suppress its expression in sensory neurons during axon regeneration. By using an available scRNA-seq dataset of adult mouse DRG neurons,^[72] we first examined the expression levels of all Klf family members (Klf1-18) in adult sensory neurons after SNI (Figure S9a, Supporting Information). The sequencing data suggested that at day 3 post-SNI, Klf3, 4, 5, 7, 8, 9, 11, 12, 13, 14, 16 were reduced, whereas Klf1, 2, 6, 10, 15, 17 were elevated. Additionally, Klf18 was not expressed in adult sensory neurons. We then reprofiled the expression levels of all Klf family members in sensory neurons after in vivo axotomy by real-time PCR and detected transcripts for 17 (Klf1-17) (Figure S9b, Supporting Information). We found that the mRNA levels of Klf4 and Klf5 were significantly downregulated in adult sensory neurons during peripheral nerve injury-induced axon regeneration (Figure S9b, Supporting Information), consistent with their roles in repressing axon growth. Furthermore, chromatin immunoprecipitation with PCR (ChIP-PCR) results showed that similar to that in RGCs, H3K27me3 interacted with 2 specific regions (R2 and R3) of the Klf4 gene, indicating that H3K27me3 directly binds to the promoter region of Klf4 in adult sensory neurons (Figure S9c, Supporting Information). Importantly, by using in vivo electroporation, we found that overexpressing Klf4 in adult sensory neurons significantly blocked axon regeneration in vivo (Figure S9d,e, Supporting Information), supporting its functional roles in repressing axon regeneration. Together, with Kdm6a/b reduction and H3K27me3 elevation in regenerating sensory neurons, these results suggest that Klf4 might also act downstream of Kdm6a/b-H3K27me3 signaling in DRG neurons to control sensory axon regeneration.

Discussion

Recent progresses in axon regeneration research have shown that significant changes in gene transcription underlie the regenerative capacity of mammalian PNS neurons, and silence of such genetic programs is responsible for the diminished intrinsic axon regeneration ability of mammalian CNS neurons. Our study provided clear and strong evidence that X chromosome encoded histone deacetylase Kdm6a and its paralog Kdm6b, act as regeneration repressors to regulate spontaneous sensory axon regeneration in vivo. More importantly, conditional Kdm6a deletion in RGCs resulted in robust optic nerve regeneration and increased RGC survival via reshaping RGC chromatin and transcriptomic landscape back to a developmental-like state. Moreover, Klf4 was identified as a downstream target of the Kdm6a-H3K27me3 signaling and overexpression of Klf4 resulted in impaired axon regeneration. Furthermore, several additional genes known to suppress axon growth have also been identified as potential targets regulated by Kdm6a. These results suggest that in addition to many regeneration enhancing genes, the expression of axon growth repressors also play critical roles in silencing the intrinsic axon regeneration ability of mature mammalian neurons. Because H3K27me3 functions to repress gene expression, it might be a key epigenetic factor for suppressing the expression of regeneration repressor genes in mature neurons. Collectively, our study offers a novel strategy for identifying novel repressors of axon regeneration, expanding the pool of genes that can be manipulated to promote CNS axon regeneration in future studies.

To our surprise, knocking out Kdm6b in RGCs had little effect on optic nerve regeneration after ONC. Despite up to 84% of sequence similarity in the JmjC domain between Kdm6a and Kdm6b,^[34] our results suggested that Kdm6b might not be the major H3K27me3 demethylase in RGCs. Indeed, accumulating evidence showed that Kdm6a and Kdm6b in many cases played different roles in cellular reprogramming,^[43,44] neural commitment,^[73] skeletal formation,^[74] embryonic development,^[75,76] muscle regeneration,^[77] lifespan extension,^[78] plasma cell formation,^[79] and tumorigenesis.^[80,81] The distinct functional roles of Kdm6a and Kdm6b might be associated with their demethylase-dependent or -independent activities, relative expression levels, different post-translational modifications or complex formation. In our study, knocking out Kdm6a or knocking in Kdm6a mutant without demethylase activity enhanced optic nerve regeneration to the similar extent, indicating that the demethylase activity of Kdm6a was responsible for repressing axon regeneration. Based on single-cell RNA-seq results from recent study,^[57] the expression level of Kdm6b is actually higher than that of Kdm6a in RGCs. Interestingly, our immunostaining results showed that Kdm6b was mainly located in the cytoplasm of RGCs, which might explain its lack of histone demethylase activity in RGCs. Indeed, evidence from other studies has shown that in addition to regulating histone methylation in the nucleus, Kdm6b might also control the methylation of non-histone proteins in both the nucleus and the cytoplasm. Moreover, the demethylase independent functions of Kdm6b have also been discovered. For instance, Kdm6b reduces the efficiency and kinetics of somatic cell reprogramming via upregulating INK4a/Arf expression in a demethylase activity-dependent manner or targeting PHF20 for ubiquitination and degradation through recruiting the E3 ubiquitin ligase Trim26 in a demethylase activity-independent manner, with the latter having a predominant role.^[82] Furthermore, Kdm6b has been shown to interact directly with p53, controlling its trafficking and subcellular distribution. This functional interaction regulates the retention or translocation of p53 to the nucleus, thereby modulating a number of biological processes, including neural stem cell differentiation,^[83] gliomagenesis,^[84] and neuronal apoptosis.^[85] Therefore, in our study it is very likely that Kdm6b located in the cytoplasm regulated RGC survival via an H3K27 demethylation independent manner.

Unlike that of axon regeneration, deletion of either Kdm6a or Kdm6b markedly protected RGCs from apoptosis, but via different underlying mechanisms. Thus, our study demonstrated that Kdm6a and Kdm6b played different roles with different mechanisms in regulation of optic nerve regeneration and neuroprotection, respectively (Figure 9). It is notable that multiple studies have demonstrated that Kdm6a and Kdm6b have both shared and distinct target genes.^[81,86,87] The precise mechanism by which this differential target recognition is determined remains unknown. Nevertheless, it is conceivable that the involvement of their N-terminal region in distinct complexes or subcellular localizations contributes to this distinctive specificity. A recent study showed that Kdm6a downregulation conferred neuroprotective effects via the H3K27me3-mediated NOTCH2/Abcb1a axis in vitro and in vivo.^[88] Similar studies of Kdm6b showed that its depletion also prevented neuronal apoptosis by a dual mechanism, with distinct downstream genes. The inhibition of Kdm6b has been demonstrated to significantly alleviate neuronal apoptosis by downregulating matrix metalloprotease (MMP) in cooperation with NF-κB,^[89] as well as Bax/Caspase-3 in a p53-dependent manner.^[85] Strikingly, our study further emphasized that Kdm6a/b exerted different roles by a distribution-dependent mechanism, despite the high degree of sequence similarity in their catalytic domains. Future studies are necessary to dissect out the molecular mechanisms underlying their different roles in controlling CNS neuroprotection and axon regeneration.

[IMAGE OMITTED. SEE PDF]

Although many genes have been identified to induce optic nerve regeneration, we have limited knowledge about how neuronal transcriptomics landscape is reshaped during regeneration via epigenetic modification. By performing and analyzing RNA-seq of enriched RGCs at different developmental/maturation stages and between wild type and Kdm6a knockout mice, we found that deleting Kdm6a in mature RGCs changed their transcriptomic landscape back to a developmental-like state. As a result, mature RGCs regained their intrinsic ability to support axon regeneration. In our latest study,^[30] we found that H3K27 methyltransferase Ezh2 in post-mitotic neurons functioned to repress dendrite development during development. Together with this study, it is emerging that H3K27me3 functions as a major epigenetic regulator of neuronal morphogenesis, promoting axon growth but suppressing dendrite development. Importantly, epigenetic regulation, such as histone modifications, can orchestrate large-scale changes in gene expression to regulate biological functions. The Kdm6a signaling revealed in this study are well positioned to reshape the gene expression landscape of mature CNS neurons to enhance axon regeneration.

In our study, we also provided evidence that Kdm6a regulated optic nerve regeneration via distinct mechanism from that of Pten, independent of mTOR activation. Moreover, combining Kdm6a and Pten double deletion resulted in long-distance axon regeneration passing the optic chiasm. 3D imaging of tissue-cleared optic nerve showed that regenerating axons induced by double deletion have more streamlined trajectory and increased branching, indicating enhanced growth ability. At the border of the optic chiasm, more than half axons either stopped or turned back, highlighting the inhibitory nature of the chiasm microenvironment. For axons entering the chiasm, many of them projected into the ipsilateral optic tract, indicating the lack of proper axon guidance. Therefore, for long-distance regenerating axons to re-innervating their original targets for functional recovery, modifying the microenvironment of the chiasm and/or the guidance sensing receptors in the regenerating axons would be major near-future research focus in the field.

Lastly, Kdm6a has been shown to escape from X inactivation and be expressed in the inactivated X chromosome. As a result, the protein level of Kdm6a is higher in female than that in male mice in most cells, including neurons. A recent study^[90] showed that Kdm6a contributed to the resistance of neurodegenerative insults of neurons in female mice. Interestingly, based on previous epidemiology studies, it is known that women are more likely to be affected by glaucomatous disease, especially for angle-closure glaucoma (ACG), for which nearly 70% are women.^[91,92] Our study raised the possibility that a higher level of Kdm6a in female RGCs might increase the susceptibility of female animals to retinal injury.

Experimental Section

Animals

All experiments involving animals were performed in accordance with the animal protocol approved by the Institutional Animal Care and Use Committee of the Johns Hopkins University (MO20M235), Chinese Academy of Sciences (IOZ-IACUC-2020-079) and Zhejiang University (SRRSH2025-0004). 8- to 12-week old adult female CF-1 mice (weighing from 30 to 35 g) were purchased from Charles River Laboratories and housed in the University Animal Facility. The female Kdm6a^f/f mutant mice (Kdm6atm1c(EUCOMM)Jae) possessing loxP sites flanking exon 3 of the Kdm6a (lysine (K)-specific demethylase 6A) gene, the Kdm6b^f/f mutant mice (B6.Cg-Kdm6btm1.1Rbo/J) possessing loxP sites flanking exons 14–20 of the Kdm6b (lysine (K)-specific demethylase 6B) gene and the Pten^f/f mutant mice (B6.129S4-Ptentm1Hwu/J) possessing loxP sites flanking exon 5 of the Pten (phosphatase and tensin homolog) gene were purchased from The Jackson Laboratory. Kdm6a enzyme-dead knockin mice possessing the H1146A and E1148A point mutations in exon 24, as a gift, were obtained from Dr. Kai Ge at the National Institutes of Health.

Immunostaining, Fluorescence Microscopy, and Image Analysis

The DRG or retina cryostat sections were collected and immunostained with the antibody against H3K27me3 (1:1000), Kdm6a (1:500), Kdm6b (1:500), or p-S6 (1:500) each co-immunostained with the neuronal marker Tuj1 (1:1000) antibody. To quantify the fluorescence intensity of H3K27me3, Kdm6a, Kdm6b, or p-S6 in sensory neurons, only Tuj1 positive cells were selected for measurement. The mean fluorescence intensity of each sensory neuron in DRG or RGC in retina section was measured using the ImageJ software. The percentage of p-S6 positive RGCs was calculated by dividing the number of p-S6/Tuj1 double positive cells by the number of Tuj1 positive cells. For DRG, 3 independent DRG tissues in each condition were used in each condition. For each DRG, 3 sections were selected and 30 sensory neurons in each section were selected for measurement. For retina, 3 independent retinal tissues in each condition were used. For each retina, 3–8 sections were selected and 15–20 RGCs in each section were selected for measurement. The “n” in the figures indicates the number of independent experiments.

In Vivo Electroporation of Adult DRG Neurons and Quantification of Axon Regeneration

The in vivo electroporation of adult mouse DRGs was performed as described previously.^[56] Briefly, under anesthesia induced by ketamine (100 mg kg⁻¹) and xylazine (10 mg kg⁻¹), a small dorsolateral laminectomy was performed to expose the L4 and L5 DRGs. EGFP plasmid or EGFP plus siRNA oligos or expression plasmid (1–1.5 µL per ganglion) were injected into the DRGs using pulled glass capillaries and a Picospritzer II (Parker Ins.; pressure, 30 psi; duration: 8 ms). Immediately after injection, electroporation was performed by applying five pulses of current (35 V, 10ms, 950-ms interval) using a custom- made tweezer-like electrode powered by the Electro Square Porator ECM830 (BTX Genetronics). The mice were allowed to recover after closing the wound. Two days after the electroporation, the sciatic nerves were crushed with fine forceps and the crushed sites were marked with nylon epineural sutures (size 10-0). Two days later, the mice were perfused with ice-cold 4% PFA in sodium phosphate buffer (pH 7.4). The whole nerve segment was then dissected out and further fixed in 4% PFA overnight at 4 °C. Before whole-mount flattening, it was confirmed that the place of epineural suture matched the injury site, and experiments were included in the analysis only when the crush site was clearly identifiable.

For quantification of in vivo axon regeneration, the fluorescence images of the whole mount nerves were first obtained. All identifiable EGFP-labeled axons in the sciatic nerve were then manually traced from the crush site to the distal growth cone to measure the length of axon regeneration. Only nerves with at least 15 identifiable axons were measured, and the “n” in the figures indicates the number of mice.

Quantitative Real Time Polymerase Chain Reaction (Real-Time PCR)

Total RNA was isolated with the TRizol Reagent (Thermo Fisher Scientific), then reverse transcribed by using the M-MLV reverse transcriptase (Roche Applied Science). To quantify the mRNA levels with real-time PCR, aliquots of single-stranded cDNA were amplified with gene-specific primers and Power SYBR Green PCR Master Mix (Invitrogen) using the CFX96 real-time PCR detection system (Bio-Rad). Specific primers used in this study are listed in Table S1 (Supporting Information). The PCR reactions contained 20–40 ng of cDNA, Universal Master Mix (Invitrogen), and 200 nm of forward and reverse primers in a final reaction volume of 20 µL. The ratio of different samples was calculated by the built-in data analysis software of the CFX96 real-time PCR system.

Intravitreal Injection, Optic Nerve Crush, and RGC Axon Anterograde Labeling

The mice were anaesthetized with intra-peritoneal injection of a mixture of ketamine (100 mg kg⁻¹) and xylazine (10 mg kg⁻¹). 1 µL of AAV2 viruses (Titers > 1 × 10¹³ vg mL⁻¹), GSK-J4 (300 mm), 1.5 µL NMDA (20 mm) for NMDA-induced retinal neurotoxicity were injected into the vitreous body of 6- to 8-week old mutant or wild type mice by glass micropipette connected to a Picospritzer II (Parker Inc.) (pressure: 15 psi; duration: 6 ms). Two weeks after intravitreal injection, the right optic nerve was exposed intraorbitally and crushed with Dumont #5 forceps for 2 s ≈1 mm behind the optic disc.

To label RGC axons in the optic nerve by anterograde labeling, 1.5 µL of CTB conjugated with fluorescence Alexa-594 (2 µg µL⁻¹, Invitrogen) was injected into the vitreous 2 days before sacrificing the animals. Sacrificed animals were fixed by perfusion with 20 mL of 0.1 m phosphate buffered saline (PBS), followed with 40 mL of 4% paraformaldehyde (PFA) at 5 mL min⁻¹. Retina and optic nerve segments were dissected out and post-fixed in 4% PFA overnight at 4 °C. The fixed optic nerves were rinsed with cold PBS three times and stored in cold PBS until further use.

Tissue Dehydration and Clearing

Optic nerves were first dehydrated in increasing concentrations of tetrahydrofuran (THF, 50%, 70%, 80%, 100%, and 100% for 20 min each) in amber glass bottles as described in a previous publication.^[93] Incubations were carried out on an orbital shaker at room temperature. Optic nerves were then transferred into benzyl alcohol/benzyl benzoate (BA:BB, 1:2 in volume; Sigma) clearing solution for 20 min. During the whole procedure optic nerves were protected from light to reduce the loss of CTB fluorescent signal.

Imaging and Quantification of RGC Axon Regeneration in the Optic Nerve

The whole-mount cleared optic nerves were imaged using a Zeiss LSM 710 confocal microscope controlled by ZEN software. A 20× objective was used to acquire image stacks with 2-µm z spacing. The Motor-driven XY scanning stage with tiling function was used to scan the optic nerve with 15% overlap in the X Y dimensions.

For quantification of optic nerve axon regeneration, every 4 consecutive images were projected to generate a series of Z-projection images of 8-µm optical sections. In each Z-projected image, the number of CTB-labeled fibers was counted at 250-µm intervals distal to the crush site till the place where no fluorescence signals were visible. The numbers at each 250-µm interval were summed over all Z-projection images.

For quantification of the average lengths of the top 5 longest regenerating axons, all acquired images were merged together using ZEN software to generate the maximum intensity Z-projection images. Then the lengths of the top 5 longest axons were manually measured from the ends to the crush site with ImageJ software (NIH).

Analyses of Axonal Tip Morphology and Axon Trajectory

The analysis of axonal tip morphology was performed as described previously.^[94] For quantification of the size of the distal axon ends, the tips of the top 20 longest regenerating axons were identified in the Z-projection images of each nerve. Using ImageJ software (NIH), the maximal width of the axonal tips was measured and was presented as a ratio with respect to the diameter of the adjacent axon shaft (tip/axon shaft ratio). The axon ends with tip/axon shaft ratio > 4 were defined as retraction bulbs, while the others were characterized as growth cones.

For quantification of the U-turn rate, in the Z-projection images the trajectories of the top 20 longest axons within optic nerve or all axons entering the optic chiasm were tracked near the axonal ends. The axonal trajectory that made a turning with the angle between the direction of the axonal tip and the longitudinal axis of the optic nerve greater than 90° was defined as a U-turn. The U-turn rate of each nerve was presented by U-turn number/axon number.

For quantification of the branching index, in the Z-projection images the axonal tips number was counted from the selected site (with 20 regenerating axons passing or the optic nerve-chiasm transition zone) till the place where no tips were visible. The branching index of each nerve was calculated by axonal tips number/selected axon number (20 axons for optic nerve or all axons for the optic chiasm).

Quantification of RGC Transduction and Survival Rates

Retinas were dissected out for whole-mount preparations. Retinal whole-mounts were blocked in the staining buffer containing 2% BSA and 0.5% Triton X-100 in PBS for 1 h before incubation with primary antibodies. Primary antibodies used: Tuj1 (1:500), Cre (1:500). Retinas were incubated with primary antibodies overnight at 4 °C and washed four times for 10 min each in PBS containing 0.5% Triton X-100. For secondary antibodies, Alexa Fluor-488/647-conjugated antibodies were used. Confocal images were acquired using a Zeiss LSM 710 confocal microscope (20× objective). Images were organized and analyzed using ZEN and ImageJ software. For RGC survival quantification, whole-mount retinal tissues were immunostained with Tuj1 antibody to label the survival RGCs. 15–20 fields were randomly sampled from the peripheral regions of each retina. The survival rate was calculated by measuring the ratio of Tuj1 positive cell number in retina with ONC to that in control retina without ONC. To quantify RGC transduction efficiency by AAV2-Cre, whole-mount retinal tissues were double-immunostained with Tuj1 and Cre antibodies. 8–12 fields were randomly sampled from the peripheral regions of each retina. The transduction efficiency was calculated by counting and calculating the ratio of Cre/Tuj1 double-positive cell number to the number of Tuj1 positive cells.

Western Blot

Total protein was extracted with the RIPA buffer containing protease inhibitor cocktail and phosphatase inhibitor cocktail. 15 µg of each protein sample was separated by a 4–12% gradient SDS-PAGE gel and electrotransferred onto polyvinylidene fluoride membranes. Nonspecific bands were blocked with 5% non-fat milk in 1× TBS containing 0.1% tween-20 (TBST) for 1 h at room temperature. Membranes were incubated with the primary antibodies of Kdm6a (1:1000), Kdm6b (1:1000), H3K27me3 (1:1000), β-actin (1:10000) and Histone 3 (1:1000) as internal control overnight at 4 °C and followed by HRP-conjugated secondary antibodies with a concentration of 1:2000 for 1 h at room temperature. Membranes were washed with TBST for 15 min three times after each antibody incubation. The immunocomplex bands were visualized by the ECL kit (GE Healthcare).

Purification of RGCs

Retinas were dissected, incubated with 20 units mL⁻¹ papain and 0.005% DNase (Worthington) for 10 min at 37 °C, then dissociated into a single-cell suspension in pre-sort buffer (BD FACS). Retinal cells were filtered through a 40-µm cell strainer (BD Falcon), then collected by centrifugation at 500 g for 5 min. Supernatant was removed and immediately pelleted cells were resuspended in 400 µL of pre-sort buffer. Cell suspension was incubated with Fc block antibodies for 5 min at 4 °C and then stained with Thy1.2-FITC antibody and IgG2a isotype antibody (Thermo fisher) as a negative control for 30 min at 4 °C, respectively. Retinal cells were pelleted by centrifugation at 500 g for 5 min. Supernatant was discarded and pelleted cells were washed one time with 500 µL of pre-sort buffer. Cell suspension was incubated with DAPI for 5 min before FACS sorting, and then sorted on a BD FACSAria Fusion using a 100-µm nozzle. Sorted cells were immediately subjected to subsequent experiments.

For RGC enrichment by immunomagnetic cell separation (MACS), cell suspension was incubated with Fc block antibodies for 5 min at 4 °C and then labeled with anti-mouse Thy1.2 magnetic particles (BD IMag) for 30 min at 4 °C. Tubes were placed on the cell separation magnet at room temperature for 8 min (BD IMag), then the supernatant was aspirated off carefully for leaving unlabeled cells. Labeled cells were resuspended in 1× BD IMag buffer and the tubes were returned to the magnet for another 4 min. The separation was repeated twice to increase the purity of the positive fraction. After the final wash step, purified cells were immediately used for further analyses.

Library Preparation and RNA Sequencing

A total amount of 2 µg RNA per sample was used as input material for the RNA sample preparations. Sequencing libraries were generated using NEBNext Ultra RNA Library Prep Kit for Illumina (#E7530L, NEB, USA) following the manufacturer's recommendations and index codes were added to attribute sequences to each sample. Briefly, mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. Fragmentation was carried out using divalent cations under elevated temperature in NEBNext First Strand Synthesis Reaction Buffer (5×). First strand cDNA was synthesized using random hexamer primer and RNase H. Second strand cDNA synthesis was subsequently performed using buffer, dNTPs, DNA polymerase I, and RNase H. The library fragments were purified with QiaQuick PCR kits and elution with EB buffer, then terminal repair, A-tailing and adapter added were implemented. The aimed products were retrieved and PCR was performed, then the library was completed.

RNA concentration of library was measured using Qubit RNA Assay Kit in Qubit 3.0 to preliminary quantify and then dilute to 1 ng µL⁻¹. Insert size was assessed using the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA), and accurate quantification of the qualified insert size was achieved by using the StepOne Plus real-time PCR system (Library valid concentration > 10 nm). The clustering of the index-coded samples was performed on a cBot cluster generation system using HiSeq PE Cluster Kit v4-cBot-HS (Illumina) according to the manufacturer's instructions. After cluster generation, the libraries were sequenced on an Illumina platform and 150 bp paired-end reads were generated.

Bulk RNA-seq Data Analysis

Using bulk RNA-seq, gene expression profiles of AAV-Cre (Kdm6a KO), AAV-GFP (wild type), and postnatal day (P) 1, P14, P21 samples we measured, with at least two biological replicates for each condition or timepoint. The bulk RNA-seq data were obtained from Gene Expression Omnibus (GEO: GSE244243). All raw sequencing reads were quality-checked with FastQC (v0.11.9), and leading bases were trimmed from the reads with Trimmomatic (v0.39) as appropriate. Quality-checked reads were mapped to the GRCm38/mm10 mouse reference genome with splice-aware aligner HISAT2 (v2.2.1). FeatureCounts (Subread v2.0.3) was used to estimate the expression counts of the genes.

For differential gene expression analysis, DESeq2 (v1.34.0) was used to perform normalization and differential expression tests on the raw read counts for each gene. DEGs were defined using DESeq2 with the adjusted p-value < 0.05 and |log₂(Fold Change)| > 1. The GO enrichment analysis was performed by clusterProfiler (v4.2.2). In order to integrate and analyze Kdm6a knockout data and optic nerve development data, the TPM (transcripts per million reads) matrix for genes and transcripts calculated by RSEM (v1.3.3) were used. The ComBat in sva (v3.42.0) was also used to minimize the technical effect of different data. The corrected values were subjected to PCA and hierarchical clustering.

Single-Cell RNA-Seq Data Analysis

The data used to analyze the time course of transcriptomic changes in single RGCs following ONC was downloaded from Gene Expression Omnibus (GEO: GSE137398).^[57] Then, the expression matrices were integrated by Seurat (v4.1.0) with default parameters. The changes of average expression levels of Kdm6a and Kdm6b in all RGCs were mainly analyzed at different time points after ONC, as shown in dot plot and violin plot. The data used to analyze the expression levels of all Klf family members (Klf1-18) in adult sensory neurons after SNI was downloaded from Gene Expression Omnibus (GEO: GSE154659).^[72] Dot plot showed the changes of average expression levels of all Klf family members in all sensory neurons at day 3 after SNI.

CUT&Tag

≈1 × 10⁵ RGCs purified by coupling Thy 1.2 beads were processed by centrifugation, and were washed with Wash Buffer (20 mm HEPES, pH 7.5; 150 mm NaCl; 0.5 mm Spermidine; 1× Protease inhibitor cocktail; 0.05% DMSO-Digitonin). The supernatant was removed and the bead-bound cells were resuspended with 50 µL Dig-wash Buffer (20 mm HEPES, pH 7.5; 150 mm NaCl; 0.5 mm Spermidine; 1× Protease inhibitor cocktail; 0.05% DMSO-Digitonin) containing 2 mm EDTA and 1% BSA and a 1: 50 dilution of the primary antibody. Primary antibody incubation was performed on a rotating platform for 2 h at room temperature. The primary antibody was removed by transient centrifugation and placing the tube on the magnet stand to clear and pulling off all the liquid. The appropriate second antibody was diluted 1: 50 in 50 µL Dig Wash Buffer and bead-bound cells were incubated for 1 h at room temperature. Cells were washed using the magnet stand 2 times for 5 min in 200 µL Dig-Wash buffer to remove unbound antibodies. A 1: 200 dilution of pG-Tn5 adapter complex (≈0.04 µm) was prepared in Dig-300 Buffer (0.05% DMSO-Digitonin, 20 mm HEPES, pH 7.5, 300 mm NaCl, 0.5 mm Spermidine, 1× Protease inhibitor cocktail). After removing the liquid on the magnet stand, 100 µL was added to the cells with gentle vortexing, which was incubated with pG-Tn5 at RT for 1 h. Cells were washed 2 times for 5 min in 200 µL Dig-300 Buffer to remove unbound pG-Tn5 protein. Next, cells were resuspended in 300 µL Tagmentation Buffer (10 mm MgCl 2 in Dig-300 Buffer) and incubated at 37 °C for 1 h. To stop tagmentation, 2.25 µL of 0.5 m EDTA, 2.75 µL of 10% SDS and 0.5 µL of 20 mg mL⁻¹ Proteinase K was added to 300 µL of sample, which was incubated overnight at 37 °C. To extract the DNA, 300 µL PCI (Phenol: Chloroform: Isoamylol = 25: 24: 1) were added to each sample with vortexing, quickly spun 5 s, 16 000 g for 15 min at room temperature. Absorbed the upper water phase to a new 1.5 mL tube and added 300 chloroform, 16 000 g for 15 min at room temperature. Absorbed the upper water phase to a new 1.5 mL tube and added 700 µL 100% ethyl alcohol, 16 000 g for 15 min at 4 °C. Pulled off all the liquid and added 1 mL 100% ethyl alcohol 16 000 g for 1 min at 4 °C. Carefully discarded all the liquid and aired the tube 5 min, 25 µL 1× TE (10 mm Tris-HCl, pH 8.0) containing 1 mm EDTA were added to dissolve the DNA. To amplify libraries, 24 µL DNA was mixed with 5 µL a uniquely index i7 and i5 primer; Using a different index for each sample. A volume of 16 µL PCR Master mix was added and mixed. The sample was placed in a Thermocycler with a heated lid using the following cycling conditions: 72 °C for 3 min; 98 °C for 30 s; 17 cycles of 98 °C for 10 s and 60 °C for 30 s and 72 °C for 30 s; final extension at 72 °C for 5 min and hold at 4 °C. Post-PCR clean-up was performed by adding 1.1 × volume of DNA clean beads (Nanjing, Vazyme), and libraries were incubated with beads for 15 min at RT, washed twice gently in 80% ethanol, and eluted in 20 µL sterile water.

Chromatin Immunoprecipitation (ChIP) Assay

ChIP assay of DRG tissue was conducted in accordance with the protocol provided by the Active Motif Company, with a few modifications. Briefly, 6–8 naïve or 10–15 axotomized L4 and L5 DRGs were collected and homogenized with 1% formaldehyde (Sigma-Aldrich) for 20 min. The homogenized tissue was washed with cold PBS, suspended with 500 µL cold cell lysis buffer (5 mm PIPES, pH 8.0, 85 mm KCl, 0.5% NP40, and 1× complete proteinase inhibitor), and then incubated on ice for 5 min. The lysates were centrifuged at 3000 rpm for 5 min, and the pellets were resuspended in 800 µL of SDS Lysis Buffer (Millipore). After 20-min incubation on ice, lysates were sonicated (6 pulses, 10 s each at a power output of 40%, with 1-min incubations on ice in between each pulse) to shear the genomic DNA into 200 and 1000 base pair fragments. To verify the size of the sheared chromatin (average size ≈500–600 bp), 5 µL aliquots of the lysates were treated with 1 µL of Proteinase A (20 mg mL⁻¹) for 20 min at 50 °C and the sample was analyzed using a 1.5% agarose gel.

To perform the immuno-precipitation, the sonicated cell supernatant was mixed with protein G Magnetic Beads, ChIP buffer 1, protease inhibitor cocktail, water and 5 µg of normal rabbit IgG or rabbit anti-H3K27me3 antibody in 1.7 mL tubes. The chromatin was rotated overnight at 4 °C. Next day the beads were washed one time with 800 µL ChIP Buffer 1 and two times with 800 µL ChIP buffer 2. After the final wash, the supernatant was removed as much as possible without disturbing the beads. The washed beads were treated with the elution buffer, followed by reverse cross-linking buffer. Formaldehyde-induced protein-DNA crosslinking was heat reversed by incubating protein-DNA complex at 65 °C overnight. The obtained DNA was then treated with 2 µL proteinase K for 1 h, followed by 2 µL proteinase K stop solution. Briefly centrifugation was performed to collect the purified DNA fragments, which could be used immediately in PCR or stored at −20 °C.

Statistics

Statistics were analyzed using GraphPad Prism 7.0 and the significance was set at a value of p < 0.05. Data are presented as mean ± SEM. For two group comparisons, if the data were normalized to the control condition and presented as percentage of control, one sample t test was used to determine the statistical significance. Otherwise, the regular two tailed student's t test was used. For comparison between multiple groups, the one-way ANOVA followed by Tukey's multiple comparison test was used to determine the statistical significance. The “n” indicates the number of independent experiments unless specifically stated otherwise.

Acknowledgements

The study was supported by grants (to F.-Q.Z.) from NIH (R01NS085176, R01EY027347, R01EY030883, R01EY031779), the Craig H. Neilsen Foundation (259450), the BrightFocus Foundation (G2017037), the Pioneer and Leading Goose R&D Program of Zhejiang Province, China (2024C03028), and the Leading Innovation and Entrepreneurship Team Program of Zhejiang Province, China (2023R01005). C.-M.L. was supported by the National Science Foundation of China (No. 81571212). X.-W.W. was supported by NIH (R00EY031742).

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

S.-G.Y., C.-P.L., X.-W.W., and T.H. contributed equally to this work. S.-G.Y., C.-M.L., and F.-Q.Z. initiated the projects and designed the experiments. Y.-C.H. designed the experiments and revised the manuscript. S.-G.Y., C.-P.L., and X.-W.W. performed most of experiments and analyzed data. T.H. performed most of the bioinformatics analyses. C.Q., Q.L., L.-R.Z., S.-Y.Z., C.-Y.D., R.N., and S. performed experiments and analyzed the data. S.-G.Y. and F.-Q.Z. wrote the manuscript with inputs from all the authors.

Data Availability Statement

To review GEO accession GSE244243: Go to Enter token qdifigiunvmzryv into the box.

References

Z. He, Y. Jin, Neuron 2016, 90, 437.

C. Qian, F. Q. Zhou, J. Mol. Cell Biol. 2020, 12, 798.

S. G. Yang, C. P. Li, X. Q. Peng, Z. Q. Teng, C. M. Liu, F. Q. Zhou, Front. Cell. Neurosci. 2020, 14, 119.

D. L. Moore, M. G. Blackmore, Y. Hu, K. H. Kaestner, J. L. Bixby, V. P. Lemmon, J. L. Goldberg, Science 2009, 326 , 298.

F. Blandini, E. Bazzini, F. Marino, F. Saporiti, M. T. Armentero, C. Pacchetti, R. Zangaglia, E. Martignoni, S. Lecchini, G. Nappi, M. Cosentino, Clin. Neuropharmacol. 2009, 32, 133.

P. D. Smith, F. Sun, K. K. Park, B. Cai, C. Wang, K. Kuwako, I. Martinez‐Carrasco, L. Connolly, Z. He, Neuron 2009, 64, 617.

F. Sun, K. K. Park, S. Belin, D. Wang, T. Lu, G. Chen, K. Zhang, C. Yeung, G. Feng, B. A. Yankner, Z. He, Nature 2011, 480, 372.

K. J. O'Donovan, K. Ma, H. Guo, C. Wang, F. Sun, S. B. Han, H. Kim, J. K. Wong, J. Charron, H. Zou, Y. J. Son, Z. He, J. Zhong, J. Exp. Med. 2014, 211, 801.

S. Belin, H. Nawabi, C. Wang, S. Tang, A. Latremoliere, P. Warren, H. Schorle, C. Uncu, C. J. Woolf, Z. He, J. A. Steen, Neuron 2015, 86, 1000.

F. M. Nathan, Y. Ohtake, S. Wang, X. Jiang, A. Sami, H. Guo, F. Q. Zhou, S. Li, Mol. Ther. 2020, 28, 1902.

X. W. Wang, Q. Li, C. M. Liu, P. A. Hall, J. J. Jiang, C. D. Katchis, S. Kang, B. C. Dong, S. Li, F. Q. Zhou, Cell Rep. 2018, 24, 2540.

Y. Zhang, P. R. Williams, A. Jacobi, C. Wang, A. Goel, A. A. Hirano, N. C. Brecha, D. Kerschensteiner, Z. He, Neuron 2019, 103, 39.

J. Minarovits, F. Banati, K. Szenthe, H. H. Niller, Adv. Exp. Med. Biol. 2016, 879, 1.

I. T. Strickland, L. Richards, F. E. Holmes, D. Wynick, J. B. Uney, L. F. Wong, PLoS One 2011, 6, e23423.

J. J. Jiang, C. M. Liu, B. Y. Zhang, X. W. Wang, M. Zhang, Saijilafu, S. R. Zhang, P. Hall, Y. W. Hu, F. Q. Zhou, Cell Death Dis. 2015, 6, e1865.

C. M. Liu, R. Y. Wang, Saijilafu, Z. X. Jiao, B. Y. Zhang, F. Q. Zhou, Genes Dev. 2013, 27, 1473.

S. Li, R. Zhang, Y. Yuan, S. Yi, Q. Chen, L. Gong, J. Liu, F. Ding, Z. Cao, X. Gu, Mol. Neurobiol. 2017, 54, 4379.

D. Motti, J. K. Lerch, M. C. Danzi, J. H. Gans, F. Kuo, T. I. Slepak, J. L. Bixby, V. P. Lemmon, FEBS Lett. 2017, 591, 2091.

Y. Song, K. M. Ori‐McKenney, Y. Zheng, C. Han, L. Y. Jan, Y. N. Jan, Genes Dev. 2012, 26, 1612.

Y. L. Weng, R. An, J. Cassin, J. Joseph, R. Mi, C. Wang, C. Zhong, S. G. Jin, G. P. Pfeifer, A. Bellacosa, X. Dong, A. Hoke, Z. He, H. Song, G. L. Ming, Neuron 2017, 94, 337.

B. J. Iskandar, E. Rizk, B. Meier, N. Hariharan, T. Bottiglieri, R. H. Finnell, D. F. Jarrard, R. V. Banerjee, J. H. Skene, A. Nelson, N. Patel, C. Gherasim, K. Simon, T. D. Cook, K. J. Hogan, J. Clin. Invest. 2010, 120, 1603.

Y. E. Loh, A. Koemeter‐Cox, M. J. Finelli, L. Shen, R. H. Friedel, H. Zou, Epigenetics 2017, 12, 77.

P. Gaub, Y. Joshi, A. Wuttke, U. Naumann, S. Schnichels, P. Heiduschka, S. Di Giovanni, Brain 2011, 134, 2134.

Y. Cho, R. Sloutsky, K. M. Naegle, V. Cavalli, Cell 2013, 155, 894.

M. J. Finelli, J. K. Wong, H. Zou, J. Neurosci. 2013, 33, 19664.

R. Puttagunta, A. Tedeschi, M. G. Soria, A. Hervera, R. Lindner, K. I. Rathore, P. Gaub, Y. Joshi, T. Nguyen, A. Schmandke, C. J. Laskowski, A. L. Boutillier, F. Bradke, S. Di Giovanni, Nat. Commun. 2014, 5, 3527.

I. Palmisano, M. C. Danzi, T. H. Hutson, L. Zhou, E. McLachlan, E. Serger, K. Shkura, P. K. Srivastava, A. Hervera, N. O. Neill, T. Liu, H. Dhrif, Z. Wang, M. Kubat, S. Wuchty, M. Merkenschlager, L. Levi, E. Elliott, J. L. Bixby, V. P. Lemmon, S. Di Giovanni, Nat. Neurosci. 2019, 22, 1913.

X. W. Wang, S. G. Yang, M. W. Hu, R. Y. Wang, C. Zhang, A. R. Kosanam, A. J. Ochuba, J. J. Jiang, X. Luo, Y. Guan, J. Qian, C. M. Liu, F. Q. Zhou, J. Clin. Invest. 2024, 134, e163145.

W. L. Tai, K. S. Cho, E. Kriukov, A. Ashok, X. Wang, A. Monavarfeshani, W. Yan, Y. Li, T. Guan, J. R. Sanes, P. Baranov, D. F. Chen, bioRxiv 2023.

M. Zhang, Y. Zhang, Q. Xu, J. Crawford, C. Qian, G. H. Wang, J. Qian, X. Z. Dong, M. V. Pletnikov, C. M. Liu, F. Q. Zhou, Neurosci. Bull. 2023, 39, 1512.

F. Lan, P. E. Bayliss, J. L. Rinn, J. R. Whetstine, J. K. Wang, S. Chen, S. Iwase, R. Alpatov, I. Issaeva, E. Canaani, T. M. Roberts, H. Y. Chang, Y. Shi, Nature 2007, 449, 689.

K. Agger, P. A. Cloos, J. Christensen, D. Pasini, S. Rose, J. Rappsilber, I. Issaeva, E. Canaani, A. E. Salcini, K. Helin, Nature 2007, 449, 731.

K. B. Shpargel, T. Sengoku, S. Yokoyama, T. Magnuson, PLoS Genet. 2012, 8, e1002964.

S. Hong, Y. W. Cho, L. R. Yu, H. Yu, T. D. Veenstra, K. Ge, Proc. Natl. Acad. Sci. USA 2007, 104, 18439.

G. G. Welstead, M. P. Creyghton, S. Bilodeau, A. W. Cheng, S. Markoulaki, R. A. Young, R. Jaenisch, Proc. Natl. Acad. Sci. USA 2012, 109, 13004.

X. Yang, B. Xu, B. Mulvey, M. Evans, S. Jordan, Y. D. Wang, V. Pagala, J. Peng, Y. Fan, A. Patel, J. C. Peng, Nat. Neurosci. 2019, 22, 362.

Y. Shan, Y. Zhang, Y. Zhao, T. Wang, J. Zhang, J. Yao, N. Ma, Z. Liang, W. Huang, K. Huang, T. Zhang, Z. Su, Q. Chen, Y. Zhu, C. Wu, T. Zhou, W. Sun, Y. Wei, C. Zhang, C. Li, S. Su, B. Liao, M. Zhong, X. Zhong, J. Nie, D. Pei, G. Pan, Nat. Commun. 2020, 11, 382.

N. Pediconi, D. Salerno, L. Lupacchini, A. Angrisani, G. Peruzzi, E. De Smaele, M. Levrero, L. Belloni, Cell Death Dis. 2019, 10, 518.

E. Cuyas, S. Verdura, L. Llorach‐Pares, S. Fernandez‐Arroyo, F. Luciano‐Mateo, N. Cabre, J. Stursa, L. Werner, B. Martin‐Castillo, B. Viollet, J. Neuzil, J. Joven, A. Nonell‐Canals, M. Sanchez‐Martinez, J. A. Menendez, Aging Cell 2018, 17, e12772.

C. Jin, J. Li, C. D. Green, X. Yu, X. Tang, D. Han, B. Xian, D. Wang, X. Huang, X. Cao, Z. Yan, L. Hou, J. Liu, N. Shukeir, P. Khaitovich, C. D. Chen, H. Zhang, T. Jenuwein, J. D. Han, Cell Metab. 2011, 14, 161.

T. Y. Lin, Y. C. Cheng, H. C. Yang, W. C. Lin, C. C. Wang, P. L. Lai, S. Y. Shieh, Oncogene 2012, 31, 3287.

K. Agger, P. A. Cloos, L. Rudkjaer, K. Williams, G. Andersen, J. Christensen, K. Helin, Genes Dev. 2009, 23, 1171.

A. A. Mansour, O. Gafni, L. Weinberger, A. Zviran, M. Ayyash, Y. Rais, V. Krupalnik, M. Zerbib, D. Amann‐Zalcenstein, I. Maza, S. Geula, S. Viukov, L. Holtzman, A. Pribluda, E. Canaani, S. Horn‐Saban, I. Amit, N. Novershtern, J. H. Hanna, Nature 2012, 488, 409.

W. Zhao, Q. Li, S. Ayers, Y. Gu, Z. Shi, Q. Zhu, Y. Chen, H. Y. Wang, R. F. Wang, Cell 2013, 152, 1037.

X. Li, Q. Zhang, Q. Shi, Y. Liu, K. Zhao, Q. Shen, Y. Shi, X. Liu, C. Wang, N. Li, Y. Ma, X. Cao, J. Autoimmun. 2017, 80, 85.

A. Salminen, K. Kaarniranta, M. Hiltunen, A. Kauppinen, J. Mol. Med. (Berl.) 2014, 92, 1035.

G. van Haaften, G. L. Dalgliesh, H. Davies, L. Chen, G. Bignell, C. Greenman, S. Edkins, C. Hardy, S. O'Meara, J. Teague, A. Butler, J. Hinton, C. Latimer, J. Andrews, S. Barthorpe, D. Beare, G. Buck, P. J. Campbell, J. Cole, S. Forbes, M. Jia, D. Jones, C. Y. Kok, C. Leroy, M. L. Lin, D. J. McBride, M. Maddison, S. Maquire, K. McLay, A. Menzies, et al., Nat. Genet. 2009, 41, 521.

K. Yamamoto, K. Tateishi, Y. Kudo, T. Sato, S. Yamamoto, K. Miyabayashi, K. Matsusaka, Y. Asaoka, H. Ijichi, Y. Hirata, M. Otsuka, Y. Nakai, H. Isayama, T. Ikenoue, M. Kurokawa, M. Fukayama, N. Kokudo, M. Omata, K. Koike, Carcinogenesis 2014, 35, 2404.

Y. Tang, T. Li, J. Li, J. Yang, H. Liu, X. J. Zhang, W. Le, Cell Death Differ. 2014, 21, 369.

W. Song, N. Zsindely, A. Farago, J. L. Marsh, L. Bodai, Hum. Mol. Genet. 2018, 27, 649.

W. Wang, H. Cho, J. W. Lee, S. K. Lee, Nat. Commun. 2022, 13, 958.

A. A. Chakraborty, T. Laukka, M. Myllykoski, A. E. Ringel, M. A. Booker, M. Y. Tolstorukov, Y. J. Meng, S. R. Meier, R. B. Jennings, A. L. Creech, Z. T. Herbert, S. K. McBrayer, B. A. Olenchock, J. D. Jaffe, M. C. Haigis, R. Beroukhim, S. Signoretti, P. Koivunen, W. G. Kaelin Jr., Science 2019, 363, 1217.

Q. Y. Tang, S. F. Zhang, S. K. Dai, C. Liu, Y. Y. Wang, H. Z. Du, Z. Q. Teng, C. M. Liu, Stem Cell Rep. 2020, 15, 439.

R. Wijayatunge, F. Liu, K. B. Shpargel, N. J. Wayne, U. Chan, J. V. Boua, T. Magnuson, A. E. West, Mol. Cell. Neurosci. 2018, 87, 4.

R. Wijayatunge, L. F. Chen, Y. M. Cha, A. S. Zannas, C. L. Frank, A. E. West, Mol. Cell. Neurosci. 2014, 61, 187.

Saijilafu, E. M. Hur, F. Q. Zhou, Nat. Commun. 2011, 2, 543.

N. M. Tran, K. Shekhar, I. E. Whitney, A. Jacobi, I. Benhar, G. Hong, W. Yan, X. Adiconis, M. E. Arnold, J. M. Lee, J. Z. Levin, D. Lin, C. Wang, C. M. Lieber, A. Regev, Z. He, J. R. Sanes, Neuron 2019, 104, 1039.

X. W. Wang, S. G. Yang, C. Zhang, M. W. Hu, J. Qian, J. J. Ma, Y. Zhang, B. B. Yang, Y. L. Weng, G. L. Ming, A. R. Kosanam, Saijilafu, F. Q. Zhou, Cell Rep. 2020, 31, 107537.

L. J. Walport, R. J. Hopkinson, M. Vollmar, S. K. Madden, C. Gileadi, U. Oppermann, C. J. Schofield, C. Johansson, J. Biol. Chem. 2014, 289, 18302.

C. Wang, J. E. Lee, Y. W. Cho, Y. Xiao, Q. Jin, C. Liu, K. Ge, Proc. Natl. Acad. Sci. USA 2012, 109, 15324.

L. Kruidenier, C. W. Chung, Z. Cheng, J. Liddle, K. Che, G. Joberty, M. Bantscheff, C. Bountra, A. Bridges, H. Diallo, D. Eberhard, S. Hutchinson, E. Jones, R. Katso, M. Leveridge, P. K. Mander, J. Mosley, C. Ramirez‐Molina, P. Rowland, C. J. Schofield, R. J. Sheppard, J. E. Smith, C. Swales, R. Tanner, P. Thomas, A. Tumber, G. Drewes, U. Oppermann, D. J. Patel, K. Lee, et al., Nature 2012, 488, 404.

K. K. Park, K. Liu, Y. Hu, P. D. Smith, C. Wang, B. Cai, B. Xu, L. Connolly, I. Kramvis, M. Sahin, Z. He, Science 2008, 322, 963.

S. G. Yang, X. W. Wang, C. Qian, F. Q. Zhou, Prog. Neurobiol. 2022, 214, 102284.

C. Wilson, L. E. Giono, V. Rozes‐Salvador, A. Fiszbein, A. R. Kornblihtt, A. Caceres, Cell Rep. 2020, 31, 107639.

A. P. Barnes, B. N. Lilley, Y. A. Pan, L. J. Plummer, A. W. Powell, A. N. Raines, J. R. Sanes, F. Polleux, Cell 2007, 129, 549.

D. Romaus‐Sanjurjo, J. M. Saikia, H. J. Kim, K. M. Tsai, G. Q. Le, B. Zheng, Cell Death Discovery 2022, 8, 390.

P. Y. Wang, G. K. Seabold, R. J. Wenthold, Mol. Cell. Neurosci. 2008, 39, 83.

S. Qin, Y. Zou, C. L. Zhang, Nat. Commun. 2013, 4, 2633.

E. R. Bray, B. J. Yungher, K. Levay, M. Ribeiro, G. Dvoryanchikov, A. C. Ayupe, K. Thakor, V. Marks, M. Randolph, M. C. Danzi, T. M. Schmidt, N. Chaudhari, V. P. Lemmon, S. Hattar, K. K. Park, Neuron 2019, 103, 642.

X. Guo, J. Zhou, C. Starr, E. J. Mohns, Y. Li, E. P. Chen, Y. Yoon, C. P. Kellner, K. Tanaka, H. Wang, W. Liu, L. R. Pasquale, J. B. Demb, M. C. Crair, B. Chen, Cell 2021, 184, 4299.

J. H. Xu, X. Z. Qin, H. N. Zhang, Y. X. Ma, S. B. Qi, H. C. Zhang, J. J. Ma, X. Y. Fu, J. L. Xie, Saijilafu, Neural Regener. Res. 2021, 16, 166.

W. Renthal, I. Tochitsky, L. T. Yang, Y. C. Cheng, E. Li, R. Kawaguchi, D. H. Geschwind, C. J. Woolf, Neuron 2020, 108, 128.

T. Burgold, F. Spreafico, F. De Santa, M. G. Totaro, E. Prosperini, G. Natoli, G. Testa, PLoS One 2008, 3, e3034.

C. Naruse, S. Shibata, M. Tamura, T. Kawaguchi, K. Abe, K. Sugihara, T. Kato, T. Nishiuchi, S. Wakana, M. Ikawa, M. Asano, FASEB J. 2017, 31, 2252.

L. Yang, L. S. Song, X. F. Liu, Q. Xia, L. G. Bai, L. Gao, G. Q. Gao, Y. Wang, Z. Y. Wei, C. L. Bai, G. P. Li, Sci. Rep. 2016, 6, 26711.

M. Shahhoseini, Z. Taghizadeh, M. Hatami, H. Baharvand, Biochem. Cell Biol. 2013, 91, 116.

K. Nakka, S. Hachmer, Z. Mokhtari, R. Kovac, H. Bandukwala, C. Bernard, Y. Li, G. Xie, C. Liu, M. Fallahi, L. A. Megeney, J. Gondin, B. Chazaud, M. Brand, X. Zha, K. Ge, F. J. Dilworth, Science 2022, 377, 666.

A. R. R. Guillermo, K. Chocian, G. Gavriilidis, J. Vandamme, A. E. Salcini, J. Mellor, A. Woollard, BMC Biol. 2021, 19, 59.

A. K. Kania, M. J. Price, L. E. George‐Alexander, D. G. Patterson, S. L. Hicks, C. D. Scharer, J. M. Boss, J. Immunol. 2022, 208, 1873.

P. Ntziachristos, A. Tsirigos, G. G. Welstead, T. Trimarchi, S. Bakogianni, L. Xu, E. Loizou, L. Holmfeldt, A. Strikoudis, B. King, J. Mullenders, J. Becksfort, J. Nedjic, E. Paietta, M. S. Tallman, J. M. Rowe, G. Tonon, T. Satoh, L. Kruidenier, R. Prinjha, S. Akira, P. Van Vlierberghe, A. A. Ferrando, R. Jaenisch, C. G. Mullighan, I. Aifantis, Nature 2014, 514, 513.

J. Van der Meulen, V. Sanghvi, K. Mavrakis, K. Durinck, F. Fang, F. Matthijssens, P. Rondou, M. Rosen, T. Pieters, P. Vandenberghe, E. Delabesse, T. Lammens, B. De Moerloose, B. Menten, N. Van Roy, B. Verhasselt, B. Poppe, Y. Benoit, T. Taghon, A. M. Melnick, F. Speleman, H. G. Wendel, P. Van Vlierberghe, Blood 2015, 125, 13.

W. Zhao, Q. T. Li, S. Ayers, Y. F. Gu, Z. Shi, Q. Y. Zhu, Y. D. Chen, H. Y. Wang, R. F. Wang, Cell 2013, 152, 1037.

S. Solá, J. M. Xavier, D. M. Santos, M. M. Aranha, A. L. Morgado, K. Jepsen, C. M. P. Rodrigues, PLoS One 2011, 6.

C. I. Ene, L. Edwards, G. Riddick, M. Baysan, K. Woolard, S. Kotliarova, C. Lai, G. Belova, M. Cam, J. Walling, M. Zhou, H. Stevenson, H. S. Kim, K. Killian, T. Veenstra, R. Bailey, H. Song, W. Zhang, H. A. Fine, PLoS One 2012, 7, e51407.

H. N. Zhang, J. Wang, J. Huang, T. Y. Shi, X. Ma, X. X. Luo, X. Li, M. K. Li, Exp. Neurol. 2018, 308, 132.

F. Lan, P. E. Bayliss, J. L. Rinn, J. R. Whetstine, J. K. Wang, S. Z. Chen, S. Iwase, R. Alpatov, I. Issaeva, E. Canaani, T. M. Roberts, H. Y. Chang, Y. Shi, Nature 2007, 449, 689.

P. Ntziachristos, A. Tsirigos, G. G. Welstead, T. Trimarchi, S. Bakogianni, L. Y. Xu, E. Loizou, L. Holmfeldt, A. Strikoudis, B. King, J. Mullenders, J. Becksfort, J. Nedjic, E. Paietta, M. S. Tallman, J. M. Rowe, G. Tonon, T. Satoh, L. Kruidenier, R. Prinjha, S. Akira, P. Van Vlierberghe, A. A. Ferrando, R. Jaenisch, C. G. Mullighan, I. Aifantis, Nature 2014, 514, 513.

C. Z. Gao, F. Yin, R. Li, Q. Ruan, C. Y. Meng, K. C. Zhao, Q. S. Zhu, Oxid. Med. Cell. Longevity 2021, 2021, 2580619.

J. Y. Lee, W. H. Na, H. Y. Choi, K. H. Lee, B. G. Ju, T. Y. Yune, Neurobiol. Dis. 2016, 95, 66.

E. J. Davis, L. Broestl, S. Abdulai‐Saiku, K. Worden, L. W. Bonham, E. Minones‐Moyano, A. J. Moreno, D. Wang, K. Chang, G. Williams, B. I. Garay, I. Lobach, N. Devidze, D. Kim, C. Anderson‐Bergman, G. Q. Yu, C. C. White, J. A. Harris, B. L. Miller, D. A. Bennett, A. P. Arnold, P. L. De Jager, J. J. Palop, B. Panning, J. S. Yokoyama, L. Mucke, D. B. Dubal, Sci. Transl. Med. 2020, 12, 5677.

H. A. Quigley, A. T. Broman, Br. J. Ophthalmol. 2006, 90, 262.

S. L. Chiu, C. L. Chu, C. H. Muo, C. L. Chen, S. J. Lan, J. Glaucoma 2016, 25, e514.

X. Luo, Y. Salgueiro, S. R. Beckerman, V. P. Lemmon, P. Tsoulfas, K. K. Park, Exp. Neurol. 2013, 247, 653.

A. Erturk, F. Hellal, J. Enes, F. Bradke, J. Neurosci. 2007, 27, 9169.