Promoting axon regeneration by inhibiting RNA N6-methyladenosine demethylase ALKBH5

Although a previous study demonstrated that conditional knockout of the METTL3 or YTHDF1 impaired axonal regeneration in mice (Weng et al., 2018), whether other key proteins involved in m⁶A modification could modulate axonal regeneration remains unknown. To answer this question, we analyzed the expression profile of methyltransferases, demethylases, and m⁶A reader proteins in the lumbar 4 and 5 (L4-5) DRG following sciatic nerve crush (SNC) injury in rats (Figure 1A), which is a classic animal model used for axon injury repair study. QRT-PCR analysis showed that the RNA levels of the genes involved in RNA m⁶A modification were not significantly changed following SNC. Given that Mettl3, Wtap, Fto, Alkbh5, Ythdc1, and Ythdf1-3 were expressed in DRG with relatively high abundance (Figure 1B), we selected these genes to perform RNA interference (RNAi)-mediated functional screening through an in vitro neurite regrowth assay (Figure 1—figure supplement 1A–D). We found that the neurite outgrowth induced by cell replating was significantly enhanced by silencing Alkbh5 or Ythdf3 (Figure 1C and D). ALKBH5 knockdown exhibited the most dominant phenotype and was chosen for subsequent research. To investigate the potential role of ALKBH5 in axonal regeneration of DRG neurons after SNC, we performed immunofluorescent (IF) staining of rat DRG sections using specific antibodies against ALKBH5. The results showed that ALKBH5 protein was predominantly distributed in the cytoplasm of the soma of DRG neuron and was downregulated following SNC (Figure 1E and F). Approximately 51.9% of neurofilament-200 (NF200, a marker for medium/large DRG neurons and myelinated A-fibers) positive neurons were labeled with ALKBH5, constituting 18.5% and 30.5% of calcitonin gene-related peptide (CGRP, a marker for small peptidergic DRG neurons) and isolectin B4 (IB4, a marker for small non-peptidergic DRG neurons)-positive neurons, respectively (Figure 1—figure supplement 1E and F). Western blot further validated the markedly reduced expression of ALKBH5 protein in the DRGs following SNC (Figure 1G and H). These results suggest that ALKBH5 may play an important role in nerve injury repair after SNC.

As mentioned above, reduced ALKBH5 expression in DRG neurons promotes neurite outgrowth, leading us to investigate whether ALKBH5 overexpression inhibits axon regeneration and whether this regulation is achieved via its RNA demethylase activity. To this end, we overexpressed wild type ALKBH5 (wt-ALKBH5) or ALKBH5 with a mutation at location 205 (H205A, mut-ALKBH5), in which there was no RNA demethylase activity (Feng et al., 2021), in primary DRG neurons using adeno-associated virus (AAV) 2/8. Western blotting analysis showed increased ALKBH5 expression in both wt-ALKBH5 and mut-ALKBH5 groups (Figure 2A and B). The primary DRG neurons that were infected with AAV and expressed EGFP, wt-ALKBH5, or mut-ALKBH5 were cultured for 7 d, and then replated for another 16–18 hr (Figure 2C). To examine the neurite regrowth, the replated neurons were immunostained using an antibody against Tuj1 (Figure 2D). The results showed that the axon length was significantly decreased in the wt-ALKBH5 group, but not in the mut-ALKBH5 group, compared to that in the control group (Figure 2E).

To confirm this result in vivo, rat DRGs were infected with AAVs that expressed EGFP, wt-ALKBH5, or mut-ALKBH5 through an intrathecal injection (Figure 2—figure supplement 1A and B). Fourteen days later, the sciatic nerves were crushed, and regenerated axons were labeled by SCG10 following another 3 d (Figure 2F and G). A regeneration index was calculated by normalizing the average SCG10 intensity at distances away from the crush site to the SCG10 intensity at the crush site. The result indicated that ALKBH5 overexpression in sensory neurons reduced axonal regeneration past the crush site, while mutant ALKBH5 had no significant effect (Figure 2H). The maximum axon length was also dramatically attenuated in the wt-ALKBH5 group but was not significantly changed in the mut-ALKBH5 group compared to that in the control group (Figure 2I). These data indicate that ALKBH5 inhibits axonal regeneration in an m⁶A-dependent manner.

To further validate the role of ALKBH5 in injury-induced axonal regeneration in the PNS, we knocked down ALKBH5 in primary DRG neurons using AAV2/8 expressing shRNA against Alkbh5 or nonsense. Then we conducted the in vitro neurite regrowth assay described above. ALKBH5 downregulation in DRG neurons was verified using western blot (Figure 3A and B). The results showed that ALKBH5 deficiency either in the Alkbh5-shRNA1 (KD1) or Alkbh5-shRNA2 (KD2) group significantly increased the axon length compared to that in the control shRNA (NC) group (Figure 3C–E). Next, we examined whether ALKBH5 inhibition in sensory neurons could promote axon outgrowth over inhibitory substrates chondroitin sulfate proteoglycans (CSPGs), which are prevalent inhibitors of axonal regeneration. The data showed that ALKBH5 knockdown significantly enhanced axon outgrowth in the presence of inhibitory substrates (Figure 3—figure supplement 1A and B).We next assessed the in vivo role of ALKBH5 deficiency in axonal regeneration of adult DRG neurons via an intrathecal AAV2/8 injection (Figure 3F, Figure 3—figure supplement 2A and B). The results showed that the extension of SCG10-positive axons and the maximum axon length of the sciatic nerve were significantly increased following ALKBH5 knockdown compared to the control group after SNC (Figure 3G–I).

Regarding the potential clinic use of selective ALKBH5 inhibitors (SAI) in axon regeneration promotion, we tried to analyze the effect of two SAI (Z56957173 [Sabnis, 2021] and Z52453295 [Takahashi et al., 2022]) on axon regeneration of DRG neurons. We first performed the CCK-8 assay to determine the effect of the SAI on cell viability to exclude the possible cell toxicity. The results showed that no more than 5 μM Z56957173 or no more than 20 μM Z52453295 has no toxicity to DRG neurons (Figure 3—figure supplement 3A and B). Next, we chose two dose of the two SAI respectively to perform the in vitro neurite regrowth assay and observed that 0.5 μM Z56957173 presented a significant promotion effect on axon growth. Meanwhile, the significantly increased axon length was also observed in the neurons treated with 10 μM Z52453295 (Figure 3—figure supplement 3C and D). Furthermore, we examined the in vivo effect of Z52453295 on sciatic nerve regeneration through intrathecal injection with a dose course. The results showed that Z52453295 significantly increased the length of the maximum regenerated axon at 6.25 and 25 mM after SNC (Figure 3—figure supplement 3E and F). These results indicated that ALKBH5 inhibition by SAI promoted axon regeneration after nerve injury and suggested the therapeutic potential of modulating ALKBH5 function with SAI in nerve injury repair.

To identify the target mRNA of ALKBH5-mediated demethylation for axonal regeneration control, we first examined the differential gene expression profiles between control and ALKBH5-knocked down DRG neurons by RNA-Seq and performed the Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis. Interestingly, the results showed that quite a few genes were enriched in the metabolism pathway after ALKBH5 inhibition (Figure 4A). The expression of these genes was validated using qRT-PCR, which showed that several metabolism-related genes, including Aldh3b1, Galns, Ndufa11, Lpin2, Pold1, Cox8a, and St6gal1, were significantly downregulated following ALKBH5 knockdown (Figure 4B). Next, we examined the potential m⁶A modification in these genes (p<0.01) by MeRIP-qPCR analysis and found that the m⁶A enrichment of Aldh3b1, Lpin2, and Galns was increased on day 3 following SNC (Figure 4C). RNAi-mediated functional screening was performed to explore the role of these genes in neurite regrowth, which showed that interfering the expression of Lpin2 significantly increased the neurite length of primary DRG neurons (Figure 4D and E, Figure 4—figure supplement 1), suggesting that Lpin2 is a target of ALKBH5 during the axonal regeneration of DRG neurons.

To confirm this, we investigated the change of m⁶A level in Lpin2 precursor (pre)- or mature mRNA in vivo. MeRIP-qPCR analysis showed that the m⁶A level in Lpin2 mature mRNA, but not in pre-mRNA, was significantly increased on day 3 post SNC compared to that in intact animals (Figure 4C, Figure 5—figure supplement 1A), which showed an opposite trend compared to that of ALKBH5 (Figure 1E–H). These results suggest that ALKBH5 downregulation contributes to the increase in m⁶A in Lpin2 mature mRNA. Furthermore, MeRIP-qPCR assay showed that ALKBH5 knockdown increased the m⁶A enrichment of Lpin2 mature mRNA, whereas wt-ALKBH5, but not mut-ALKBH5, decreased the m⁶A level of Lpin2 mature mRNA (Figure 5A and B). Next, we examined the expression levels of pre- and mature Lpin2 mRNA in DRG neurons with ALKBH5 knockdown or overexpression. ALKBH5 knockdown reduced the expression of Lpin2 mature mRNA (Figure 5C), but not the Lpin2 pre-mRNA (Figure 5—figure supplement 1B), whereas overexpression of wt-ALKBH5, but not of mut-ALKBH5, increased the level of Lpin2 mature mRNA (Figure 5D), but not that of Lpin2 pre-mRNA (Figure 5—figure supplement 1C). The western blot assay showed that ALKBH5 knockdown downregulated the LPIN2 protein level (Figure 5E and F), whereas overexpression of wt-ALKBH5, but not of mut-ALKBH5, increased the LPIN2 protein expression (Figure 5G and H). These data suggest that Lpin2 mRNA is a target of ALKBH5-mediated demethylation.

Previous reports have indicated that m⁶A modification usually regulates RNA metabolism by impacting RNA stability, translation, or nuclear export (Frye et al., 2018; Shi et al., 2019; Zaccara et al., 2019). To explore the mechanism by which Lpin2 is regulated through ALKBH5-induced m⁶A demethylation, we first examined RNA nuclear export by separating the isolated nuclear and cytoplasmic RNAs. The results showed no significant difference in the nuclear/cytoplasmic localization of both pre- and mature mRNAs of Lpin2 mRNA between the control and ALKBH5-manipulated DRG neurons (Figure 5—figure supplement 2), indicating that ALKBH5 has no impact on nuclear/cytoplasmic localization of Lpin2 pre- and mature mRNA. Next, we detected RNA stability by treating DRG neurons with the transcription inhibitor actinomycin D (Act-D). We found that the half-life of Lpin2 mature mRNA in ALKBH5-knockdown DRG neurons was significantly shorter than that in control cells (Figure 5I), although there was no significant difference in the level of remaining Lpin2 pre-mRNA (Figure 5—figure supplement 1D). Furthermore, the half-life of Lpin2 mature mRNA was upregulated in neurons with overexpression of wt-ALKBH5, but not of mut-ALKBH5, compared to that in the neurons with EGFP overexpression (Figure 5J). In contrast, there were no significant changes in the Lpin2 pre mRNA (Figure 5—figure supplement 1E). Previous studies have shown that ALKBH5 regulates mRNA stability through m⁶A sites located in the 3' UTR region near the stop codon (Tang et al., 2018). In the present work, by using SRAMP (Zhou et al., 2016), we predicted the specific m⁶A modification sites of Lpin2 and identified the GGACA motif with the highest prediction score in the 3' UTR region near the stop codon as the candidate site involved in Lpin2 mRNA stability regulation (Figure 5—figure supplement 3). To determine the confident changes of m⁶A levels in putative targets of Lpin2, we employed a sensitive and reliable site-specific method SELECT-m⁶A to detect m⁶A in Lpin2 (Liu et al., 2019; Wang et al., 2020c; Xu et al., 2021). With the probe targeting GGACA, we found that the m⁶A level at motif GGACA in Lpin2 3' UTR region was significantly increased in ALKBH5 knockdown DRG neurons compared with NC. Moreover, overexpression of wt-ALKBH5, but not mut-ALKBH5, reduced the m⁶A level at motif GGACA (Figure 5K and L). These findings strongly support the role of ALKBH5 in reducing the m⁶A level at the m⁶A motif GGACA located in the 3' UTR of Lpin2 mRNA. Then, we performed a luciferase assay to further elucidate the molecular mechanism underlying m⁶A-mediated regulation of Lpin2 mRNA stability. The results showed that ALKBH5 knockdown decreased the activity of the luciferase vector containing the 3′UTR of Lpin2 mRNA, while overexpression of wt-ALKBH5, but not of mut-ALKBH5, increased the luciferase activity. Mutation of the luciferase reporter at the potential m⁶A site (A to C) almost completely reinstated the luciferase activity when ALKBH5 was knocked down or overexpressed (Figure 5M and N). These data indicate that ALKBH5 upregulates the level of Lpin2 mature mRNA by increasing its stability through m⁶A in the 3' UTR.

LPIN2, a phosphatidic acid phosphatase enzyme, plays a central role in the penultimate step of the glycerol phosphate pathway and catalyzes the conversion of phosphatidic acid (PA) to diacylglycerol (DG) in coordination with LPIN1 (Donkor et al., 2007). Recent reports have indicated the pivotal function of LPIN1 in retina axonal regeneration (Yang et al., 2020); however, the role of LPIN2 in sciatic nerve regeneration remains unknown. We found decreased expression of Lpin2 mRNA (Figure 6A) and protein following SNC (Figure 6B and C), indicating that Lpin2 is an injury-reduced gene. To further explore the direct role of LPIN2 in axonal regeneration, we first examined the function of LPIN2 in neurite outgrowth in vitro and observed that the axon length in DRG neurons with LPIN2 overexpressed was significantly decreased compared to that in the control group (Figure 6D–F). We also determined the contribution of LPIN2 to axonal regeneration in vivo (Figure 6—figure supplement 1A and B). The results showed that LPIN2 overexpression reduced the axonal regeneration index (Figure 6G and H) and maximum length of the regenerated sciatic nerve (Figure 6I), suggesting that LPIN2 impairs sciatic nerve regeneration following SNC. It is worth noting that previous studies have indicated that sciatic nerve regeneration can be hindered by an excess of lipid droplets containing triglycerides. Interfering with LPIN1, a key enzyme in the glycerol phosphate pathway that converts phosphatidic acid to diglyceride, has been shown to enhance axon regeneration (Yang et al., 2020). To further investigate the link between lipid droplets and LPIN2 in axon regeneration, we examined DRG neurons with overexpressed LPIN2 and found that these neurons had an increased presence of visible lipid droplets containing triglycerides, unlike the control group (Figure 6J and K). Therefore, our findings suggest that LPIN2 plays an important role in regulating axon regeneration by modulating lipid metabolism.

Next, we investigated whether the forced expression of LPIN2 in DRG neurons can reverse the axonal regeneration phenotypes induced by ALKBH5 deficiency. To this end, we overexpressed LPIN2 in primary DRG neurons with ALKBH5 knockdown using AAV2/8 and performed the neurite regrowth assay in vitro. The results showed that LPIN2 could diminish the promoting effect of ALKBH5 deficiency on neurite outgrowth (Figure 7A and B). Furthermore, LPIN2 largely restored the increased regeneration index and length of the maximum regenerated axon in vivo in ALKBH5-deficient animals following SNC (Figure 7C–E). Taken together, ALKBH5-deficiency induces axonal regeneration through LPIN2 following SNC.

To investigate the role of ALKBH5 in CNS nerve injury repair, we explored ONC injury, which is an important experimental model to investigate CNS axonal regeneration and repair. In contrast to the expression in the DRG, no significant change in ALKBH5 expression was observed after ONC (Figure 8A and B). AAV2/2 with EGFP expression were injected into the vitreous body between the lens and the retina of the eyes in mice. Two weeks later, we observed successful infection of approximately 70.85 ± 8.85% RGCs (Figure 8—figure supplement 1A and B). To explore the role of ALKBH5 in RGC survival and axonal regeneration, AAV2/2 containing nonsense or Alkbh5 shRNA were intravitreally injected, and ONC injury was performed 2 wk later. After another 12 d, the Alexa Fluor 555-conjugated Cholera toxin b subunit (CTB) was injected into the vitreous body to label the regenerating axons (Figure 8C). By quantifying the numbers of Tuj1-positive cells in the retina, we found that ALKBH5 knockdown increased the RGC survival rates compared to the control group following ONC (Figure 8D and E). Moreover, the number of regenerated axons was increased in the ALKBH5 knockdown group compared to the control group, and the maximum length of axons crossing the lesion site in ALKBH5 knockdown mice was up to 1.2 mm at 2 wk after ONC (Figure 8F and G). These data indicate that inhibition of ALKBH5 promotes the survival and axonal regeneration of RGCs in the CNS.

Specific-pathogen-free degree male Sprague–Dawley (SD) rats (180–220 g) and male C57BL/6J mice (18–22 g) were provided by the Experiment Animal Center of Nantong University. All of the animals were handled according to protocol approved by the Institutional Animal Care and Use Committees of Nantong University (approval ID: S20210303-011).

SNC was performed as previously reported (Wang et al., 2020a). In brief, 12 SD rats were randomly divided into four groups. A 2 cm incision was made in the skin at the left thigh perpendicular to the femur following the intraperitoneal injection of 40 mg/kg sodium pentobarbital. The muscle tissue was bluntly dissected to expose the sciatic nerves, which were then crushed 1 cm proximal to the bifurcation of the tibial and fibular nerves using fine forceps three times at 54 N force (F31024-13, RWD, Shenzhen, China). The incision was sutured after surgery. The L4-5 DRGs were collected at days 0, 1, 3, and 7 following SNC. For ONC, the mice were anesthetized by intraperitoneal injection of 12.5 mg/mL tribromoethanol (20 mL/kg body weight), an incision was made on the conjunctiva, and the optic nerve was crushed by jeweler’s forceps (F11020-11, RWD) for 2 s at 1–2 mm behind the optic disk. The retina was collected at days 0, and 3 or 14 following ONC.

The cDNA samples were prepared using a Prime-Script RT reagent Kit (TaKaRa, Dalian, China) according to the manufacturer’s instructions, and qRT-PCR was performed using SYBR Premix Ex Taq (TaKaRa) on an ABI system (Applied Biosystems, Foster City, CA) according to the standard protocols. The primers shown in Supplementary file 1 were used to validate the candidate genes, and glyceraldehyde 3-phosphate dehydrogenase (Gapdh) was used as the internal reference.

For intrathecal injection, adult rats were anesthetized and shaved to expose the skin around the lumbar region. A total of 10 μL of virus solution was injected into the cerebrospinal fluid between vertebrae L4 and L5 using a 25 μL Hamilton syringe. After injection, the needle was left in place for an additional 2 min to allow the fluid to diffuse. For intravitreal injection, adult mice were anesthetized and 2 μL of AAV was injected into the eyes using a 10 μL Hamilton syringe. After injection, the animals were left to recover for 2 wk to ensure substantial viral expression before the following surgical procedures.

Sciatic nerves were crushed as mentioned above following intrathecal injection with the indicated AAV for 14 d. Animals were perfused with cold 4% paraformaldehyde (PFA) (Sigma-Aldrich, St Louis, MO), and the sciatic nerves were collected after 3 d post-SNC. The sciatic nerves were immersed in 4% PFA overnight before being transferred to 30% sucrose (Sigma-Aldrich) in a phosphate buffer (Hyclone, Logan, UT) for cryoprotection. The sciatic nerves were fixed and immune-stained with anti-SCG10 antibody (NBP1-49461, 1:500; Novus Biologicals, Littleton, CO). Regenerated axons were measured and quantified using ImageJ software.

DRG neurons were separated and maintained in vitro as previously reported (Cheng et al., 2008). In detail, DRGs were harvested and transferred to Hibernate-A (Gibco BRL USA, Grand Island, NY) and washed twice with phosphate buffered saline (PBS; Hyclone). After removing the connective tissue, DRGs were dissociated in a sterile manner and incubated with 0.25% trypsin (Gibco) for 10 min with intervals of trituration, followed by 0.3% collagenase type I (Roche Diagnostics, Basel, Switzerland) for 90 min at 37°C. The cells were centrifuged and purified using 15% bovine serum albumin (BSA; Sigma-Aldrich). The cell suspension was filtered through a 70 μm nylon mesh cell strainer (BD Pharmingen, San Diego, CA) to remove tissue debris, before plating in a cell culture plate coated with poly-l-lysine-coated in Neurobasal medium (Gibco) with 2% B27 supplement (Gibco) and 1% GlutaMax (Thermo Fisher Scientific, Waltham, MA).

DRG neurons were stabilized and infected with AAV (OBIO, Shanghai, China). After 14 hr of AAV infection, the medium was changed and cultured for 7 d for subsequent experiments. SiRNA transfection was performed using Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA). The scrambled control, Cy3 labeled control or target siRNA (RiboBio, Guangzhou, China) was incubated with the siRNA transfection reagent for 15 min at room temperature according to the manufacturer’s instructions. Then, the transfection mixture was added to the DRG neurons and plated on a cell culture plate. After overnight incubation, the medium was replaced and cultured for 48 hr for subsequent experiments. DRG neurons were replated on poly-D-lysine (Sigma-Aldrich)- or CSPG (5 μg/mL, Sigma-Aldrich)-treated coverslips and incubated for 16–18 hr after infection with AAV containing control and target genes or target gene-specific shRNAs for 7 d, or transfection with the target siRNA for 48 hr. The interfering sites of the target genes are presented in. The AAVs were packaged by OBiO Biotechnology Co., Ltd. (Shanghai, China), and the siRNA fragments were synthesized by RiboBio Biotechnology Co., Ltd. (Guangzhou, China). Supplementary file 2

The L4-5 DRG samples and retinas were collected, fixed with 4% PFA overnight at 4°C, and cryoprotected in 30% sucrose until use. Sections were cut and washed twice with PBS, before pre-treating with 0.3% PBST for 30 min at room temperature (25℃). After incubation with a blocking buffer for 60 min at room temperature, the sections were incubated with the primary antibody at 4°C overnight and then with Alexa Fluor-conjugated secondary antibody (Invitrogen). Images were obtained using a Zeiss Axio Imager M2 microscope. The exposure time and gain were maintained at constant levels between the conditions for each fluorescence channel.

DRG neurons were replated following treat with 0.025% trypsin (Gibco) for 10 min, then cultured on the slide for 16–18 hr. The cells were washed twice with PBS, fixed with 4% PFA in PBS for 15 min at room temperature, and immune-stained with anti-Tuj1 antibody (R&D, Minneapolis, MN). The neurite length was measured and quantified using ImageJ software.

The total RNA was extracted using TRIzol reagent (Invitrogen), following the manufacturer’s instructions, and the purity and concentration of total RNA were measured. RNA-seq analysis was performed by Shanghai Biotechnology Corporation. RNA-seq data have been deposited in SRA database under accession number PRJNA914071. The differential expression profiles of mRNAs were determined using bioinformatics analysis, as previously reported (Yu et al., 2012). KEGG pathway enrichment analyses were performed to elucidate the signaling pathways associated with the differentially expressed genes.

Cytoplasmic and nuclear RNAs from DRG neurons were isolated using the PARIS Kit (Life Technologies, Carlsbad, CA) following the manufacturer’s protocols. The expression levels of the target gene in the cytoplasm and nucleus were measured by qRT-PCR. SYBR Green Mix (TaKaRa) was used for quantitative PCR, with the validated primers listed in. Supplementary file 1

Total RNA was extracted from NC, KD1, KD2 AAV or control, wt-ALKBH5, and mut-ALKBH5 AAV transfected adult DRG neurons or the L4-5 DRGs at varying times after SNC using TRIzol reagent (Invitrogen). The Seq-StarTM poly (A) mRNA Isolation Kit (Arraystar, Rockville, MD) was used to obtain the complete mRNA. Then, RNA (2 μg) was incubated with m⁶A antibody (202003, Synaptic Systems, Gottigen, Germany) or negative control IgG (ab172730, Abcam, Cambridge, UK) overnight at 4°C, before conducting immunoprecipitation based on the instructions of the M-280 Sheep anti-Rabbit IgG Dynabeads (11203D, Invitrogen). The mRNA with m⁶A enrichment was assayed using RT-qPCR. MeRIP-qPCR was performed to measure the m⁶A levels of the target gene in DRG neurons.

The Cell Counting Kit-8 (Beyotime Biotechnology, Shanghai, China) was used to examine cell viability according to the manufacturer’s instructions. Briefly, 10 μL of CCK-8 solution was added to 100 μL of neuronal culture medium in each well of a 96-well plate and incubated for 2 hr at 37°C. The absorbance was measured by the BioTtek Synergy2 system (BioTek Instruments, Bad Friedrichshall, Germany) at 450 nm.

The single-site m⁶A methylation of Lpin2 was detected by the SELECT m⁶A site identification kit (Epibiotek, Guangzhou, China) according to the manufacturer’s instructions. In brief, the relative ratio of Lpin2 RNA levels between the control group (control or NC) and the treatment group (wt-ALKBH5, mut-ALKBH5 or KD1, KD2) was obtained according to real-time quantitative PCR, and an equivalent amount of Lpin2 RNA was used for reverse transcription. Next, the single-base ligation was performed by using SELECT DNA polymerase and SELECT ligase. Finally, real-time quantitative PCR was conducted using the special primers designed for the Lpin2 m⁶A site. Up probe sequence: tagccagtaccgtagtgcgtgAGTGCCAGCTTCGGGGACTCTG; down probe sequence: 5phos/CCCTGTTCTGGAAAGCAGGTTCCTcagaggctgagtcgctgcat. Forward primer: TACAGATGAAGACCCAGGAG; reverse primer: TGAGTGGTGGCTTAGGAA.

To measure the Lpin2 and pre-Lpin2 mRNA stability in DRG neurons infected with different AAVs, 5 μg/mL actinomycin D (MCE, Monmouth Junction, NJ) was added to cells following AAV infection on day 7. RNA was extracted using TRIzol regent after incubation at the indicated times (0, 2, 4, and 8 hr). The stability of Lpin2 and Lpin2 pre-mRNA was examined using qRT-PCR.

To detect the lipid droplets in cultured DRG neurons following LPIN2 overexpression, cells were fixed in 4% PFA for 30 min and permeabilized with 0.3% PBST for 30 min at room temperature. After blocking, the Tuj1 antibody was applied in blocking buffer and incubated at 4°C overnight. Cells were washed thrice with PBS and incubated with Alexa Fluor-conjugated secondary antibody at room temperature for 2 hr. Finally, coverslips were incubated with 200 nM BODIPY (Sigma-Aldrich) in blocking buffer for 30 min before mounting. Images were obtained using a Zeiss Axio Imager M2 microscope. The exposure time and gain were maintained at constant levels between conditions for each fluorescence channel.

The 3′ UTR sequence of Lpin2 and the mutations of the 3′ UTR sequence of Lpin2 (GGACA to GGCCA) were constructed into the pmirGLO vector. The indicated mutations were generated by direct DNA synthesis (Miaoling Biology, Wuhan, China). The luciferase reporter assay was performed 48 hr after co-transfection of the reporter vectors with NC, KD1, and KD2 or control, wt-ALKBH5, and mut-ALKBH5 expression vectors into HEK-293T cells. Renilla luciferase reporter was used as an internal control, and the relative luciferase activity was normalized to Renilla luciferase activity measured by a dual-luciferase reporter assay system (Promega, Madison, WI). SRAMP (http://www.cuilab.cn/sramp/↗) was used to predict the m⁶A modification site of the target gene. HEK-293T cells were purchased from ATCC (CRL-3216) and were tested negative for mycoplasma using the MycoAlert PLUS mycoplasma detection kit (LT07-705; Lonza, Basel, Switzerland).

Protein extracts were prepared from primary cultured DRG neurons. The cultured DRG neurons were washed twice with PBS and lysed with RIPA buffer (Thermo Fisher Scientific) with protease inhibitor (Roche Diagnostics) and phosphatase inhibitor (Roche Diagnostics) at 4°C for 30 min to extract proteins. The protein concentration was determined using the BCA protein assay kit (Thermo Fisher Scientific). Equal quantities of protein were electrophoresed on 10% SDS-PAGE and transferred onto a nitrocellulose membrane (Roche Diagnostics). The membranes were incubated with the primary antibody overnight at 4°C after blocking with 5% milk dissolved in TBST buffer for another 2 hr at room temperature (25℃). The membranes were washed with TBST and incubated with horseradish peroxidase-conjugated secondary antibody for 1.5 hr at room temperature, followed by chemiluminescent detection after incubation with ECL substrate (Thermo Fisher Scientific). The blots were probed with the candidate antibodies shown in Key Resources Table. ImageJ software was used to quantify the results of the western blot.

Mice were subjected to ONC following intravitreal injection with 2 μL of ALKBH5 knockdown AAV2 for 14 d, as previously described (Zhang et al., 2019). Mice with obvious eye inflammation or shrinkage were sacrificed and excluded from further experiments. To analyze the optic nerve regenerating axons, the optic nerves were anterogradely labeled with 2 μL CTB-555 (1 μg/μL, Invitrogen) 12 d after injury. The fixed optic nerves were dehydrated in incremental concentrations of tetrahydrofuran (THF; 50%, 80%, 100%, and 100%, %v/v in distilled water, 20 min each; Sigma-Aldrich) in amber glass bottles on an orbital shaker at room temperature. Then, the nerves were incubated with benzyl alcohol/benzyl benzoate (BABB, 1:2 in volume, Sigma-Aldrich) clearing solution for 20 min. The nerves were protected from light throughout the whole process to reduce photo bleaching of the fluorescence. The number of CTB-labeled regenerated axons was measured at different distances from the crush site.

The eyeballs with the ONC injury were collected and fixed with 4% PFA overnight, before dissecting the whole retina and dehydrating with 30% sucrose in phosphate buffer for 2 d. The whole retina was sectioned using a cryostat (10 μm), and the serially collected retina sections were stained with the Tuj1 antibody, before capturing the Tuj1-positive RGCs and quantifying by ImageJ software.

The numbers of independent animals are presented in the ‘Materials and methods’ and ‘Results’ sections or indicated in the figure legends. All analyses were performed while blinded to the treatment group. The data were analyzed with GraphPad Prism 8 using unpaired, two-tailed Student’s t-test or ANOVA followed by a Bonferroni’s, Tukey’s, or Dunnett’s post hoc test. p-values<0.05 were considered statistically significant. All quantitative data are expressed as mean ± standard deviation (SD).