Silencing the enhancer of zeste homologue 2, Ezh2, represses axon regeneration of dorsal root ganglion neurons.

Recovery from injury to the peripheral nervous system is different from that of the central nervous system in that it can lead to gene reprogramming that can induce the expression of a series of regeneration-associated genes. This eventually leads to axonal regeneration of injured neurons. Although some regeneration-related genes have been identified, the regulatory network underlying axon regeneration remains largely unknown. To explore the regulator of axon regeneration, we performed RNA sequencing of lumbar L4 and L5 dorsal root ganglion (DRG) neurons at different time points (0, 3, 6, 12 hours, 1, 3 and 7 days) after rat sciatic nerve crush. The isolation of neurons was carried out by laser capture microscopy combined with NeuN immunofluorescence staining. We found 1228 differentially expressed genes in the injured sciatic nerve tissue. The hub genes within these differentially expressed genes include Atf3, Jun, Myc, Ngf, Fgf2, Ezh2, Gfap and Il6. We verified that the expression of the enhancer of zeste homologue 2 gene (Ezh2) was up-regulated in DRG neurons after injury, and this up-regulation differed between large- and small-sized dorsal root ganglion neurons. To investigate whether the up-regulation of Ezh2 impacts axonal regeneration, we silenced Ezh2 with siRNA in cultured DRG neurons and found that the growth of the newborn axons was repressed. In our investigation into the regulatory network of Ezh2 by interpretive phenomenal analysis, we found some regulators of Ezh2 (including Erk, Il6 and Hif1a) and targets (including Atf3, Cdkn1a and Smad1). Our findings suggest that Ezh2, as a nerve regeneration-related gene, participates in the repair of the injured DRG neurons, and knocking down the Ezh2 in vitro inhibits the axonal growth of DRG neurons. All the experimental procedures approved by the Administration Committee of Experimental Animals of Jiangsu Province of China (approval No. S20191201-201) on March 21, 2019.

Chinese Library Classification No. R459.9; R363; R364

Introduction

Although the peripheral nervous system has some ability of self-regeneration and repair after injury, the effect of such repair is not ideal (Gu et al., 2020; Yang et al., 2020; Li et al., 2021). At present, the main clinical treatments are direct nerve repair, nerve transplantation, nerve transfer, fibrin glue, nerve conduit, cell therapy, drugs and electrical nerve stimulation (Hussain et al., 2020). Dorsal root ganglions (DRGs) are a group of cell bodies responsible for transmitting sensory information from receptors such as heat receptors, nociceptors, proprioceptors and chemoreceptors to the central nervous system (CNS) in response to a variety of symptoms, such as tingling, burning, numbness or severe pain (Pope et al., 2013). In severe cases, sensory and motor dysfunction and even lifelong disability may occur (Noble et al., 1998). Many studies have shown that macrophages (Zigmond and Echevarria, 2019) and Schwann cells (Poplawski et al., 2018) in the peripheral nervous system play an important role in axonal regeneration during the process of DRG injury repair and that various transcription factors, including ATF3 and Smad1, that promote axon regeneration are part of this process (Zou et al., 2009; Gey et al., 2016).

Enhancer of Zeste Homolog 2 (Ezh2) is a methyltransferase that is enthusiastically discussed in the field of cancer as it can either promote carcinogenesis and metastasis or inhibit tumorigenesis. Thus, it has become a target gene during the therapeutic process of many cancers (Chu et al., 2020; Kleer et al., 2003; Mallen-St Clair et al., 2012; Zingg et al., 2015; Gan et al., 2018; Krill et al., 2020). It has also been studied in other fields such as angiogenesis (Tsou et al., 2019), neurodevelopment (Wever et al., 2018) and bone formation (Dudakovic et al., 2018). However, there have been few studies in the peripheral nervous system. There is evidence that knocking down of Ezh2 affects Schwann cell myelin formation when Schwann cells are co-cultured with DRG neurons (Heinen et al., 2012).

To investigate the expression and function of Ezh2 in DRG neurons after sciatic nerve injury in a rat model, we examined DRG sections and isolated neurons at seven time periods after sciatic nerve injury and performed RNA-seq and immunofluorescence analysis.

Materials and Methods

Ethics statement

All the experimental procedures involving animals were ethically approved by the Administration Committee of Experimental Animals of Jiangsu Province of China (approval No. S20191201-201) on March 21, 2019 and were used in accordance with Institutional Animal Care guidelines of Nantong University.

Modeling and single cell capture

We used 150 adult Sprague-awley rats weighing 180–220 g in this study. Among them, 56 rats were used for scRNA-seq, 56 rats were used for the in vivo experiment (RNA sequencing; RNA extraction; immunofluorescence and immunocytochemistry) and 38 rats were used for the in vitro experiment (primer DRG neuron cultures). First, the rats were anesthetized intraperitoneally by an injection of a mixture of 85 mg/kg trichloroacetaldehyde monohydrate (RichJoint, Shanghai, China), 42 mg/kg magnesium sulfate (Xilong Scientific, Guangzhou, China) and 17 mg/kg sodium pentobarbital (Sigma-Aldrich, St. Louis, MO, USA). Under anesthesia, the sciatic nerve was exposed and hemostatic forceps clipped the sciatic nerve for 30 seconds to form a 3-mm transparent section, similar to the crush procedure described by Kalender et al. (2009). Finally, the wound was sutured and wiped with iodophor to avoid wound infection. Normal uninjured rats were used as the control group (0 hour).

After the model was constructed, seven rats were anesthetized as above and sacrificed by neck dislocation at each of the times 0, 3, 6, 12 hours, 1, 3 and 7 days. The L4-L5 DRG was removed and then embedded in O.C.T. compound (Sakura, Torrance, CA, USA) for the frozen sectioning at 12 μm. The sections were dried for 10 minutes in the box containing silica desiccant and then stained with NeuN direct labeled antibody (1:100; Abcam, Cambridge, MA, USA) for 10 minutes. After staining, the sections were rinsed with 0.01 M PBS (Vazym, Nanjing, China) three times, then with diethyl pyrocarbonate (DEPC) water, and dried for 20 minutes. Finally, the DRG neurons were isolated by laser microdissection (Leica, Solms, Germany), during which we cut the target cells close to the boundary of the target cells.

RNA sequencing and data analysis

At 0, 3, 6, 12 hours, 1, 3 and 7 days after modeling, the DRG neurons were collected using laser capture microscopy (LCM) and then dropped into the cell lysis buffer. RNA extraction, reverse transcription, library construction and sequencing were performed according to the manufacturer's protocol (Gene Denovo Biotechnology Co., Guangzhou, China). Briefly, after extracting total RNA from each sample using TRIzol reagent kit (Invitrogen, Carlsbad, CA, USA), Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA) was employed for RNA quality control. The samples with high quality were then subjected to mRNA enrichment by Oligo (dT) beads and fragmented with buffers (Gene Denovo Biotechnology Co., Ltd., Guangzhou, China).

Random primers were used in the reverse transcription from RNA to cDNA, the second-strand cDNA was synthesized by DNA polymerase I, RNase H, dNTP and buffers. Before sequencing, the cDNA fragments were purified with QIAquick PCR extraction kit (Qiagen, Venlo, Netherlands), end-repaired, poly(A) added and finally ligated to Illumina sequencing adapters. The sequencing was carried out using Illumina HiSeq2500 (Gene Denovo Biotechnology Co.).

The raw data from the sequencing machines were further filtered by using filtering of clean reads by Fastp (version 0.18.0), removing ribosome RNA by Bowtie2 (version 2.2.8) (http://bowtie-bio.sourceforge.net/bowtie2), genome mapping using HISAT2.2.4 (http://www.ccb.jhu.edu/software/hisat) and quantification of gene abundance using StringTie (http://ccb.jhu.edu/software/stringtie/). The final measure of gene expression was the fragment per kilobase of transcript per million mapped reads (FPKM) value of each gene in each sample. The expression of differentially expressed genes was analyzed with DESeq2 software between the two different groups. The genes with the parameters of P-value ≤ 0.05 and |log2FC|≥1 were considered as differentially expressed genes (DEGs).

The network of DEGs interaction comes from the String database (https://string-db.org/cgi/input.pl?sessionId=vsDE728IjCfB&input_page_show_search=on). We chose Rattus norvegicus for “Organism” and uploaded the list of DEGs to the website. After the successful load, we hid “disconnected nodes” in the network in the settings before exploring the picture and tables.

RNA extraction

Seven rats were anesthetized then sacrificed by neck dislocation at 0, 3, 6, 12 hours, 1, 3, and 7 days after modeling. The L4–L5 DRG from each rat was taken out and frozen with liquid nitrogen. The tissue/cells were cut up and fully lysed with TRIzol (Vazyme, Guangzhou, China). The samples were then extracted with chloroform on ice for 30 minutes, and the tissue/cells were centrifuged for 30 minutes at 12,000 × g. After addition of isopropanol, the collected supernatant was incubated at –80°C overnight. The next day, the mixture was centrifuged at 12,000 × g for 30 minutes, washed with 75% ethanol, dried, dissolved in DEPC water, and then purified with an organic solvent. The purity and concentration of RNA, dissolved in DEPC, were determined by NanoDrop2000/2000c (Thermo, MA, USA).

Quantitative real-time polymerase chain reaction

After each RNA sample was extracted and its concentration measured, the volume of RNA solution needed for a 1-μg template was calculated and each RNA was reverse transcribed to cDNA by a 2720 Thermal cycler (Applied Biosystems®, UK) with a reverse transcription kit (Vazyme). The quantitative PCR system was prepared with primers, cDNA, and RT-qPCR reverse transcription reagent (Vazyme). The cDNA was diluted three times and then mixed with primer (10 μM, Genscript, Nanjing, China), ChamQ SYBR qPCR Master Mix (High ROX Premixed) (Vazyme) and ddH2O. Finally, the mixture was put into a LightCycler 96 (Roche, Switzerland) for amplification, the temperature gradient of amplification was as follows: 95°C, 30 seconds, 1 cycle; 95°C, 10 seconds, 60°C, 30 seconds, 40 cycles; 95°C, 15 seconds, 60°C, 60 seconds, 95°C, 15 seconds, 1 cycle. The quantitative method we used was 2–ΔΔCt. The relative expression level was calculated using the comparative 2–ΔΔCt method. The primers we used in our study are shown in .

Immunofluorescence and immunocytochemistry

The cells were fixed with 4% PFA for 30 minutes, then the cells/tissue sections were washed with PBS for 5 minutes three times at room temperature. The frozen tissue sections were treated with formaldehyde at 0, 3, 6, 12 hours, 1, 3 and 7 days after modeling, then blocked with blocking buffer (Beyotime, Shanghai, China) for 1 hour. Next, the tissue sections were treated with primary antibodies [Ezh2, 1:200, rabbit monoclonal antibody (CST, USA); NF160/200, 1:500, mouse Monoclonal antibody (Sigma-Aldrich); IB4, 1:200, FITC conjugate (Sigma-Aldrich); TuJ1, 1:1 000, mouse monoclonal antibody (Sigma-Aldrich); NeuN, 1: 300, rabbit monoclonal antibody] at 4°C overnight. Eighteen hours later, tissue sections were incubated with secondary antibodies [Cy3-conjugated goat anti-rabbit, 1:500 (Sigma); 488-conjugated goat anti-mouse, 1:500 (Sigma-Aldrich)] for 1.5 hours at room temperature, and then mounted on microscope slides. Finally, an antifade mounting medium with 4,6-diamino-2-phenyl indole (DAPI) solution (Beyotime) was used to seal the slice. Samples were visualized and images captured using an optical and epi-fluorescence microscope (Axio Imager M2, Carl Zeiss Microscopy GmbH, Jena, Germany). All assays were performed in triplicate.

Primer DRG neuron cultures

DRG neurons were primary cultured according to a previous study (Christie et al., 2010). Briefly, the rats were anesthetized by injection of a mixture of 85 mg/kg trichloroacetaldehyde monohydrate, 42 mg/kg magnesium sulfate, and 17 mg/kg sodium pentobarbital. The hair on the back of the rats was removed, then the rats were sacrificed by neck dislocation. The dorsal spinal cord of each rat was quickly separated and the lamina of the spinal cord was opened to facilitate the removal of DRG from the cone hole. The DRG was initially placed in 1% penicillin/streptomycin (Gibco, Carlsbad, CA, USA) in PBS solution on ice. The collected DRGs were washed twice with PBS, treated with 0.5% collagenase (Sigma) for 1.5 hours, and then transferred to 0.25% trypsin (Gibco) for digestion for 25 minutes. When the cell particles were homogeneous, the digestion was terminated with 10% FBS (Gibco) in PBS. The cell particles were washed in 15% BSA PBS and the purified cell precipitate was resuspended by Neurobasal A (Gibco) with 2% B27 (Gibco), 1% penicillin/streptomycin and 1% L-glutamine (Sigma) before being filtered through a sieve (70 μm) into the centrifuge tube for standby. The collected cells were cultured on PLL-coated plates and cultured in an incubator at 37°C in 95% O2/5% CO2. Twelve hours later, the culture medium was refreshed with DRG medium containing 10 μM cytosine arabinoside (Cytosine, C21867, Sigma). Neurites from the neurons were observed by their immunoreaction with NeuN 1 day after the cells were planted. The culture medium was refreshed every 3 days.

Small interfering RNA (siRNA) transfection

Primer DRG neurons were transfected with siRNA immediately after planting. The siRNA (RiboBio, China) was diluted to 10 μM with RNase-free water, the mixture of RNAiMax and siRNA transfection was prepared with opti-medium (Gibco) according to the instructions of RNAiMax reagent (Invitrogen). The mixture was incubated at room temperature for 10 minutes and then added to the cell culture medium. The cells were left in an incubator at 37°C in 95% O2/5% CO2 for 16 hours. The cells were cultured with complete culture medium without 1% penicillin/streptomycin. The siRNA used in our study is shown in .

Fluorescence in situ hybridization

The frozen sections were acquired at 0, 3, 6, 12 hours, 1, 3 and 7 days after modeling, as described above, and the slices used for fluorescence in situ hybridization (FISH) were kept free of contamination by RNase. All reagents used before tissue slicing were RNase-free, including 4% PFA and dehydrated sucrose solution. Tissue sections were washed twice with 4% DEPC-PFA at room temperature for 5 minutes each, washed twice with PBS at room temperature for 5 minutes each, incubated with 5% DEPC-protein kinase at room temperature for 30 minutes, washed with PBS twice at room temperature for 5 minutes each, fixed with 4% DEPC-PFA at room temperature for 10 minutes, washed with PBS twice at room temperature for 5 minutes each, treated with 0.1 M TEA at room temperature for 10 minutes, washed with PBS twice at room temperature for 5 minutes each. After incubation in pre-hybridizing solution (RiboBio) at 42°C for 2 hours, tissue sections were hybridized with an Ezh2 probe (Sangon Biotech, Shanghai, China) containing cy5 in the 5′ terminal by hybridizing solution (RiboBio) at 42°C for 12 hours. Then, they were washed with 0.1% Tween-20 4× SSC at 42°C for 10 minutes three times, washed with 2× SSC at 42°C for 10 minutes, with 1× SSC at 42°C for 10 minutes, and with SSC at room temperature for 10 minutes. Finally, tissue sections were washed with PBS for 10 minutes and sealed with antifade mounting medium with DAPI solution (Beyotime). The probe we used is Ezh2 [5′-CAU UGA CAA AAC UUU UCA CAA AAA UUU UGU GCU AUC ACA CAA GGG CAC GAA CUG UCA CAA GGC UGC-3′ (66 bp)].

Statistical analysis

All data were processed using GraphPad prism 8 software (GraphPad Software Inc., San Diego, CA, USA) and are expressed as the mean ± SEM. Repeated measures analysis of variance, one-way analysis of variance followed by the least significant difference post hoc test were used for comparison between groups. A level of P < 0.05 was considered statistically significant.

Results

The importance of Ezh2 in peripheral nerve injury (PNI) is revealed by scRNA-seq

The DRG cells, harvested after establishing the crush model, were subjected to RNA sequencing by Smart-seq2. The sequence data underwent quality control to obtain clean data. The workflow is shown in . Although the number of single DRG neurons obtained by LCM was 100 to 200 at each time point, only 100 repeats at each time point are needed to undergo scRNA-seq2 sequencing.

Single cell sequencing of rat dorsal root ganglion neurons.

(A) The flow chart shows rat single dorsal root ganglion neuron sequencing. (B) The collection of all differentially expressed genes comparing with the control group (0 h) at 3, 6, 12 hours, 1, 3 and 7 days. The red dots represent up-regulated differentially expressed genes, the green dots represent down-regulated differentially expressed genes, and the gray dots represent no significant change in differentially expressed genes. The standard of the differentially expressed gene was |log2FC| ≥ 1 and P-value ≤ 0.05. X axis is log2 (FC), and Y axis is –log10 (P value). (C) The gene difference between 1, 3 and 7 days vs. the control group (0 hours) was shown by a Wayne's chart. (D) On the left side is the network map of common differentially expressed genes between 1, 3 and 7 days vs. the control group, and on the right side is the enlarged map of the core part of the map on the left side.

We obtained the differentially expressed genes (DEGs) by comparing the expression of genes in the injury groups (3, 6, 12 hours, 1, 3 and 7 days) with the normal group (0 hour). A total of 1228 genes showed different expression after injury (|log2FC| ≥ 1 and P-value ≤ 0.05), as shown in , with red representing up-regulated genes, green representing down-regulated genes and gray representing genes with no significant differences (). According to a previous report (Wang et al., 2020), axonal regenerative repair started from 3 days after PNI. From 1 day after the injury, genetic reprogramming was induced intensely and lasted until the 7th day (Renthal et al., 2020). Therefore, the genes with altered expression in each of these three groups may have an important influence on axon regeneration. We found 189 common DEGs by intersection of 391 DEGs at 1 day (0 hour vs. 1 day), 795 DEGs at 3 days (0 hour vs. 3 days) and 771 DEGs at 7 days (0 hour vs. 7 days) ().

The interaction between these 189 DEGs was explored with String database. After filtering out the molecules that had no relationship with others, the remained relationships are shown in . The molecules in the center of the network present more relationships with others, indicating they may have more influence on this network. The core molecules include a number of cytokines or growth factors, such as Il6, Ngf, Fgf2, Gfap, Cd44, and Ccl2. and transcription factors or regulators such as Atf3, Jun, Myc, Crem, Socs3, and Ezh2.

Ezh2 expression increases in injured DRG neurons and has a difference in large- and small-sized DRG neurons

To investigate the role of Ezh2 in the repair of PNI, we first verified the expression of Ezh2 in DRG neurons. Our scRNA-seq results show that Ezh2 expression consistently increased in L4–5 DRG neurons (). Our RT-qPCR, FISH and immunofluorescence experiments showed that Ezh2 expression was consistently increased in injured DRG neurons (Figure ).

Expression of Ezh2 in dorsal root ganglion neurons.

(A) Single cell sequencing statistics showed the expression of Ezh2 in dorsal root ganglion neurons after sciatic nerve crush injury. (B) The expression of Ezh2 in dorsal root ganglion neurons after sciatic nerve crush was verified by quantitative reverse transcription-polymerase chain reaction. (C) The quantification of immunostaining of []D, the vertical axis is mean gray value. (D) The changes of Ezh2 expression in DRG neurons after sciatic nerve crush were verified by immunofluorescence. Red is the Ezh2 marker, green is NeuN neuron specific marker, and blue is the 4,6-diamino-2-phenylindole (DAPI) nuclear marker. In the first row the bar is 250 μm, and in the subsequent rows the bar is 50 μm. The arrow indicates Ezh2 co-labeled with NeuN. (E) Ezh2 probe was used to detect the changing of Ezh2 in dorsal root ganglion neurons after sciatic nerve crush injury. Red represents Ezh2, blue is the DAPI. The experiment was repeated three times. **P < 0.01, ***P < 0.001 (mean ± SEM, n = 3, repeated measures analysis of variance). d: Day; FPKM: fragments per kilobase of exon model per million mapped fragments; Gapdh: glyceraldehyde-3-phosphate dehydrogenase; h: hour.

In DRG neurons immunofluorescence experiments, we found that the expression of Ezh2 differed in different sizes of DRG neurons. Hu et al. (2016) reported that the heterogeneity of DRG neurons resulted in different sizes of neurons with different sensory functions. Therefore, we labeled DRG neurons by IB4 (small) and NF160/200 (large) (Usoskin et al., 2015). We found that the expression of Ezh2 also increased in large- and small-sized neurons after injury. Moreover, the expression of Ezh2 was different between large- and small-sized neurons, and there were many NF160/200-positive neurons (). In conclusion, the expression of Ezh2 during the repair of PNI was continuously increased, and the expression of Ezh2 in NF160/200-positive large-sized neurons was high.

Expression of Ezh2 in NF160/200- and IB4-positive dorsal root ganglion neurons after sciatic nerve injury.

(A) The expression of Ezh2 in IB4-positive dorsal root ganglion neurons: red is the specific marker of Ezh2, green is the specific marker of IB4, blue is DAPI nuclear marker. Scale bar: 50 μm. (B) The gray value statistical results of Ezh2 expression in IB4-positive dorsal root ganglion neurons. (C) The expression changes of Ezh2 in NF160/200-positive dorsal root ganglion neurons: red is the specific marker of Ezh2, green is the specific marker of NF160/200, and blue is DAPI nuclear marker. Scale bar: 50 μm. (D) The gray value statistical results of Ezh2 expression in NF160/200-positive dorsal root ganglion neurons. (E) The gray value statistical results of the expression difference between NF160/200 and IB4. *P < 0.05, **P < 0.01, ***P < 0.001 (mean ± SEM, n = 3, repeated measures analysis of variance). The arrows represent Ezh2-IB4 or Ezh2-NF160/200 co-labeled DRG neurons. DAPI: 4,6-Diamino-2-phenyl indole.

The growth of DRG neuron neurites is inhibited by knockdown of Ezh2 in vitro

We have shown above that the expression of Ezh2 in DRG was consistently increased after PNI. To explore the function of Ezh2 in DRG neurons, we cultured DRG neurons in vitro. First, we transfected neurons with siRNA of Ezh2 and screened the knockdown efficiency of siRNA on Ezh2 expression (), then we stained the neurite of DRG neurons with TuJ1. The staining results indicate that the lengths of neuron neurites in the three groups transfected were significantly shorter than the control group (), including the mean length and the longest length (Figure and ). These results suggest that the knockdown of Ezh2 expression can inhibit the axonal growth of DRG neurons. Counts of the distribution of neurite lengths showed that the percentage of neurite lengths < 400 μm was significantly higher in Ezh2-silenced DRGs than that in the normal control group, whereas the percentage of neurite length > 400 μm was lower in Ezh2-silenced DRGs than that in the normal control group (). This indicated that the neurite growth was significantly inhibited after silencing Ezh2. Finally, we examined the axon growth statistics of DRG neurons after Ezh2 knockdown in large-sized neurons (surface area > 800 μm2) and small-sized neurons (surface area < 800 μm2), which revealed that the neurite growth was significantly suppressed in both (). All these results indicate that Ezh2 is important for DRG neuron regeneration.

Functional verification of Ezh2 in cultured dorsal root ganglion neurons.

(A) The knockdown efficiency of Ezh2 by small interfere RNA (siRNA). (B) After siRNA transfection, the axons of the experimental and control groups were stained with TuJ1. Scale bar: 100 μm. (C–E) Statistical analysis of the mean and maximum neurite lengths (μm) of dorsal root ganglion neurons in each group after siRNA transfection, and the distribution of neurite lengths in each group. (F) Statistical analysis of the mean neurite length of large-sized neurons (surface area > 800 μm2) and small-sized neurons (surface area < 800 μm2) after siRNA transfection. *P < 0.05, **P < 0.01, ***P < 0.001 (mean ± SEM, n = 3, one-way analysis of variance followed by the least significant difference post hoc test). NC: Normal control.

Predicting the regulatory network of Ezh2 involved in axonal growth

Ezh2 is very important for DRG neuron regeneration. Inhibition of Ezh2 expression inhibits the growth of DRG axons. However, it is not clear how Ezh2 works or what the possible mechanism and reaction network are. To further search for the regulatory mechanism of Ezh2 involved in axonal growth, we used Ingenuity Pathway Analysis (IPA) to intersect these molecules with our 1-day vs. control differential genes. We found 28 up-regulated regulators and 13 down-regulated effectors of Ezh2 (). The different shapes in the graph represent different molecule types. Among the many molecule types, upstream factors such as CASR and MYC have been found to play a role in neuronal axon growth. We hope that the possible upstream and downstream factors of Ezh2 predicted by us in will provide the basis for the follow-up studies on PNI repair.

Ingenuity pathway analysis predicts the upstream and downstream interaction factors of .

On the left is the upstream effector of Ezh2 in dorsal root ganglion neurons, which is the differentially expressed gene 1 day vs. 0 hour after sciatic nerve injury. On the right is the downstream effector. Each arrow indicates an interaction. Pink molecules are up-regulated and green molecules are down-regulated on day 1 vs. controls.

Discussion

Analysis of experimental results

After 189 common DEGs were input into the String database, we acquired a molecule interaction net. The functions of most of these molecules, such as Atf3, in neuronal injury or axon regeneration have been reported. The DEG deficient mice show suppression of axons regeneration and function recovery after injury (Renthal et al., 2020). Ngf is the normal factor that promotes neuron survival and axon regeneration both in vivo and in vitro (Oppenheim, 1991). Il6 and its soluble receptor support the survival of sensory neurons (Thier et al., 1999). Among these core factors, we found that the expression of Ezh2 in DRG neurons tends to increase after PNI, and interacts with Myc (Shin et al., 2021; Whitesides, 2006), Il-6 (Fischer, 2017), Cdkn1a (Utada et al., 2005) and Egfr (Stupack et al., 2020) and together these factors regulate axon regeneration. Intrathecal injection of an inhibitor of Ezh2 can alleviate the onset and maintenance of mechanical and thermal hyperalgesia induced by partial sciatic nerve ligation in rats (Yadav and Weng, 2017). It has also been reported that the inhibition of Ezh2 in DRG neurons under co-culture conditions with Schwann cells leads to disturbance of myelin formation (Heinen et al., 2012). However, there has been no research into the function of Ezh2 when the sciatic nerve was injured. We found that Ezh2 expression is up-regulated after neuronal damage and Ezh2 is at the core of the regulatory network of axonal injury. We demonstrated by RT-qPCR and immunofluorescence staining that there is an upward trend in Ezh2 expression after sciatic nerve injury. However, we found that inhibition of Ezh2 expression impeded the axonal growth of DRG neurons. Our results showed that Ezh2 was differentially expressed in different types of neurons. DRG cells are heterogeneous because they contain different classes of primary sensory neurons, such as pain-associated nociceptive neurons, non-nociceptive temperature sensing and mechanosensory neurons (Haberberger et al., 2020). The heterogeneous populations of sensory neurons in the DRG are subdivided into subclasses based on morphological, neurochemical, electrophysiological and transcriptional characteristics (Platika et al., 1985; Wood et al., 1990; Rugiero and Wood, 2009). There are different subclasses that respond to different physiological and pathological conditions (Zemel et al., 2018). To verify the difference of Ezh2 expression between large- and small-sized neurons, we chose the classical marker NF160/200 for large-sized neurons and IB4 for small-sized neurons to co-localize with our target molecule Ezh2 (Usoskin et al., 2015). The results showed that Ezh2 expression was up-regulated in NF160/200-positive DRG neurons after sciatic nerve injury, and there was a difference between NF160/200- and IB4-positive DRG neurons. It has been reported that NF160/200 can be subdivided into C8 subtypes of DRG neurons, which are typical mechanoreceptor neurons (Li et al., 2016). It has been shown that mechanical stimulation can increase the axon growth of large-sized mechanoreceptor neurons and is accompanied up-regulation of the expression of Atf3 (Kampanis et al., 2020). Our in vitro experiments showed that the longest neurite length of large-sized neurons was shorter after interference with Ezh2. Therefore, whether Ezh2 can mediate the repair of DRG injury through mechanoreceptor neurons and how Ezh2 plays a role in it will need further investigation.

Study limitations

The increasing research in single-cell genomics, bioinformatics and computation has brought great benefits to research neuroscience, which not only enables further cellular subtype analysis of neurons and discovery of new neuronal subclasses but also provides a fuller understanding of the molecular expression patterns specific to different types of neurons (Tran et al., 2019). Therefore, we hoped to use scRNA-seq to find breakthrough points in nerve repair. We obtained single cell samples by LCM and chose smart-seq2 to perform deep sequencing of the transcriptome. The advantage of LCM is the ability to obtain cells in situ without going through enzymatic digestions. In the process of isolating single cells most sequencing methods rely on the enzymatic disassembly of protein junctions between cells that has the disadvantage that it can reduce the vitality of the cells and may also alter the molecular expression pattern of the cells (Nichterwitz et al., 2016). A flaw with our use of LCM was the 12-μm thickness of our frozen sections, resulting in the microtissue containing multiple cells, including a few glial cells surrounding a neuron. To reduce the interference from non-neuronal cells, the laser cutting was done close to the cell borders of neurons and later we removed some single cell samples with high glial cell markers. We determined by immunofluorescence staining that the target molecules were mainly expressed in neurons.

Conclusion and prospects

In addition to its role in gene epigenetic silencing (Cao and Zhang, 2004), Ezh2 can also bind to many important transcription factors to turn on gene expression (Jung et al., 2013). Among the molecules predicted by our IPA to act downstream of Ezh2 were those reported as transcription factors relating to PNI repair. For example, Atf3 promotes axonal regeneration by regulating regeneration-related gene clusters (Gey et al., 2016). Jun has been reported to mediate axonal regeneration by promoting nerve germination, lymphocyte recruitment, and microglia activation (Raivich et al., 2004). Smad1 has been reported to inhibit axonal regeneration in DRG neurons after disrupted expression (Zou et al., 2009). Thus, we speculate that Ezh2 may act in conjunction with these transcription factors to affect the regeneration of DRG neurons. In addition to acting in conjunction with transcription factors, other effectors of Ezh2 in DRG neurons need to be explored. Among our predicted downstream factors, we have found that inhibition of Bcl2l11 and Ptgs2 can inhibit neuronal apoptosis (Xiao et al., 2019; Li et al., 2020) and that they are down-regulated in expression after nerve injury.

Therefore, whether Ezh2 can enhance neuron survival after axonal injury through those molecules and thus synergistically promote axonal injury repair should also be investigated further.

There are so many predicted upstream regulators of Ezh2, for example, CASR was reported to promote axon and dendrite growth in developing neurons (Vizard et al., 2015; Markworth et al., 2019). MYC has also been reported to play a facilitating role in neuronal axon regeneration after PNI or optic nerve injury (Belin et al., 2015; Shin et al., 2021). In addition, downstream of Ezh2, we found various types of molecules including transcription factors, cytokines and enzymes, among which transcription factors ATF3 and SMAD1 have been reported to promote axonal injury repair (Zou et al., 2009; Gey et al., 2016). Some other molecules, including CDKN1A (Tomita et al., 2006), CCl2 (Zigmond and Echevarria, 2019) and SOCS3 (Gallaheret al., 2018), have been reported to participate in axonal damage repair. Among them, we identified a post-injury immediate-early gene Myc, a molecule that is classically up-regulated in expression after PNI and whose overexpression promotes axonal regeneration of the injured optic nerve (Belin et al., 2015). It has been reported that MYC was able to bind to the promoters of Jun, Atf3, and Sparr1 of regeneration associated genes (RAGs) after pre-injury of peripheral nerve conditions and that MYC was up-regulated earlier than RAGs (Shin et al., 2021). In addition, it has been found that MYC can directly bind to the promoter of Ezh2 to turn on the transcription of Ezh2 (Koh et al., 2011). Ezh2 can also be recruited to the promoter of a gene MYC dependent, and the two combine to jointly regulate gene expression (Zhao et al., 2013). Our data showed that Myc expression is also up-regulated in DRG neurons after sciatic nerve injury, and it is likely that Ezh2 is regulated by Myc and thus its expression is up-regulated. For the molecules which are downstream of Ezh2, a previous study found that Klf4 was a highly methylated state after 7 days of peripheral nerve injury treatment (Shin et al., 2020), and the knockdown of Klf4 in turn promoted the regeneration of optic nerve axons. Thus, future investigations should address how Ezh2 works in promoting axon regeneration. In addition, future studies are needed to determine whether Ezh2 turns on the expression of downstream RAGs by directly binding transcription factors such as ATF3, JUN, and SMAD1, turns off the expression of regeneration repressors by methylation, or whether both approaches work together.

Table 1

Primer information

Gene	Sequence	Product size (bp)
Ezh2	Forward: 5’-CTT TGA TTA CAG ATA CAG CCA GG-3’	23
	Reverse: 5’-AGA ATC AGT TGG TGA TGT TCT GT-3’	23
GAPDH	Forward: 5’-ACG CCA GTA GAC TCC ACG ACA T-3’	22
	Reverse: 5’-CAA CGG GAA ACC CAT CAC CA-3’	20

GAPDH: Glyceraldehyde-3-phosphate dehydrogenase.

Table 2

Primer information of small interfering RNA (siRNA) transfection

Gene	Sequence	Product size (bp)
siRNA-r-Ezh2_001	5’-GGA AAG TGT ATG ACA AAT A-3’	19
siRNA-r-Ezh2_002	5’-GGT AAA TGC TCT TGG TCA A-3’	19
siRNA-r-Ezh2_003	5’-GAG GAA GAC TTC CGA ATA A-3’	19
Negative control-	5’-GGC UCU AGA AAA GCC UAU GC dTdT-3’	44
siRNA	3’-dTdT CC GAG AUC UUU UCG GAU ACG-3’