IGFBP2–Ceramide Pathway Mediates Divergent Myelin Breakdown in the PNS and CNS Following Injury

Abstract

Following injury, the peripheral nervous system (PNS) exhibits remarkable regenerative capacity, whereas the central nervous system (CNS) has limited regenerative potential. This difference is partially attributed to distinct postinjury myelin breakdown. However, the underlying mechanisms driving this disparity remain unclear. By comparing the expression profiles of injured peripheral and central nerves in adult male and female C57BL/6J mice, we identified insulin-like growth factor-binding protein 2 (IGFBP2) as a key regulator that determines the differences in myelin breakdown between the injured PNS and CNS. Schwann cell-derived IGFBP2 in the injured PNS promotes myelin breakdown and facilitates axonal regeneration. Furthermore, through lipidomics, we identify ceramide, a sphingolipid regulated by ceramide synthase 6 in injured nerves, as playing a critical role in IGFBP2-mediated myelin breakdown. Conversely, minimal IGFBP2 expression is observed in the injured CNS, contributing to the limited myelin breakdown and axon regeneration in injured CNS. These findings provide insights into the divergent regenerative potential of the PNS and CNS and unveil IGFBP2 and ceramide as promising targets for promoting CNS regeneration after injury.

Significance Statement

Our research sheds light on the contrasting regenerative capacities of the peripheral nervous system (PNS) and central nervous system (CNS) after injury. Understanding why the PNS exhibits robust regeneration, while the CNS does not revolutionize treatments for neurological injuries and diseases. We discovered that Schwann cell-derived insulin-like growth factor-binding protein 2 plays a crucial role in promoting myelin breakdown and axon regeneration in the PNS. Moreover, our findings highlight the involvement of ceramide, a lipid molecule, in this process. Identifying these key players not only deepens our understanding of nerve regeneration but also unveils potential targets for therapeutic interventions aimed at enhancing CNS regeneration postinjury.

Introduction

The central nervous system (CNS) and the peripheral nervous system (PNS) exhibit distinct regenerative capacities following injury (Huebner and Strittmatter, 2009). While the PNS possesses a remarkable ability to regenerate, the CNS has a limited capacity for regeneration (Nagappan et al., 2020). Numerous factors contribute to the differential regenerative responses of the CNS and PNS following injury (Huebner and Strittmatter, 2009), with one key factor being the distinct postinjury degradation of myelin, which harbors molecules inhibitory to axon growth. In the injured PNS, myelin clearance occurs rapidly and plays a critical role in facilitating axon regeneration. This process relies heavily on the timely recruitment and accumulation of immune cells, particularly macrophages and neutrophils (Lindborg et al., 2017; Msheik et al., 2022). The infiltration of macrophages is tightly regulated by various signaling pathways (Talsma et al., 2022; Huang et al., 2024), which together help to establish a permissive environment for neuronal repair. In contrast, myelin clearance in the injured CNS is significantly delayed (Filbin, 2003; He and Koprivica, 2004; Vargas and Barres, 2007; Huebner and Strittmatter, 2009). This discrepancy is thought to stem, in part, from differences in immune cell recruitment and function, which contribute to the divergent regeneration outcomes observed between the PNS and CNS (Gaudet et al., 2011; Miron and Franklin, 2014). However, the underlying mechanisms driving this disparity remain to be elucidated.

In addition to immune factors, intrinsic differences in myelin formation and composition also appear to play a role. The lipid composition of myelin in the CNS and PNS is broadly similar. However, specific lipid species show regional enrichment: CNS myelin is enriched in glycolipids such as galactosylceramide, whereas PNS myelin contains higher levels of sphingomyelin and phosphatidylcholine (Poitelon et al., 2020). Furthermore, protein composition also differs between CNS and PNS myelin: myelin protein zero and periaxin are predominantly expressed in PNS myelin, while myelin oligodendrocyte glycoprotein and claudin-11 are specific to CNS myelin (García-García et al., 2024). These chemical and molecular differences may contribute to PNS-/CNS-specific myelin degradation mechanisms.

Myelin is enriched in lipids, including phospholipids, cholesterol, and sphingolipids, which are crucial for maintaining the structural integrity and functionality of the myelin sheath (Schmitt et al., 2015; Poitelon et al., 2020). Research suggests a close link between lipid stability and myelin degradation (Barnes-Vélez et al., 2022). For instance, studies have shown that mechanical injury can induce lipid peroxidation in the myelin sheath, leading to its disruption (Sundaram et al., 2023). Additionally, certain proteins, like mixed lineage kinase domain-like protein, have been identified to bind to myelin lipids and promote myelin degradation (Ying et al., 2018).

Insulin-like growth factor-binding protein 2 (IGFBP2) is an evolutionarily conserved protein that plays a multifaceted role in regulating metabolism, including lipid metabolism (Russo et al., 2015; Yau et al., 2015a,b; Shin et al., 2017; Mu et al., 2019; Boughanem et al., 2021). Studies have implicated IGFBP2 in various physiological processes, including adipocyte differentiation (Yau et al., 2015b; Wang et al., 2017) and glucose homeostasis (Wheatcroft and Kearney, 2009; Boughanem et al., 2021; Faramia et al., 2021). IGFBP2 is shown to provide neuroprotection in hypoxic ischemia and spinal cord injury possibly by promoting neuronal survival (Fletcher et al., 2013; Wang et al., 2024). Moreover, IGFBP2 can enhance neurogenesis and improve cognitive function (Khan et al., 2019). Despite its well-established roles in lipid metabolism and neuroprotection, the potential involvement of IGFBP2 in regulating myelin degradation following nerve injury remains unexplored.

In this study, we demonstrate that Schwann cell-derived IGFBP2 is a key regulator of myelin breakdown following peripheral nerve injuries (PNI). In the PNS, IGFBP2 upregulation promotes myelin breakdown, facilitating axonal regeneration. Conversely, minimal IGFBP2 expression is observed in the CNS, contributing to the limited regenerative capacity observed after CNS injury. Furthermore, our data show that ceramide, a sphingolipid class, plays a critical role in the IGFBP2-mediated myelin breakdown. These findings not only provide insights into the divergent regenerative potential of the PNS and CNS but also unveil IGFBP2 and ceramide as promising targets for promoting CNS regeneration after injury.

Materials and Methods

Animals

All experiments used age- and sex-matched C57BL/6 mice (8–12 weeks old) of both sexes: Igfbp2^fl/fl mice (strain number, T013255) from GemPharmatech, Cx3cr1-Cre/ER^T (strain number, C001032) from Cyagen Biosciences, and Plp1-Cre/ER^T mice (stock number, 005975) from Jackson Laboratory. All experimental animals were bred and housed at the Laboratory Animal Center of China Agricultural University under specific pathogen-free conditions. Mice were kept at a stable room temperature of 22 ± 1°C with a 12 h light/dark cycle. All animal procedures were approved by the Institutional Animal Care and Use Committee of China Agricultural University and complied with the national guidelines issued by the Ministry of Health of China. For Cre recombinase induction, tamoxifen was administered via intraperitoneal injection at 1 mg per mouse daily for 5 consecutive days. The primers used for genotyping are as follows: Igfbp2, forward 5′-ATCCATCCATCACTTAACAGGCC-3′, reverse 5′-TTAAGGGTCATGGTCCATCCTG-3′; Plp, forward 5′-AGGTGGACCTGATCATGGAG-3′, reverse 5′-ATACCGGAGATCATGCAAGC-3′; Cx3cr1, forward 5′-GACATTTGCCTTGCTGGAC-3′, reverse 5′-GCAGGGAAATCTGATGCAAG-3′; and Camk2a, forward 5′-GCTAAACATGCTTCATCGTCGG-3′, reverse 5′-GATCTCCGGTATTGAAACTCCAGC-3′. Both male and female mice (8–12 weeks old) were used in all experiments.

Cell lines

HEK293T-AAV cells (female human origin; Cell Biolabs) were cultured in Dulbecco's Modified Eagle Medium (DMEM; Thermo Fisher Scientific, catalog #11965092) supplemented with 10% fetal bovine serum (FBS; Biological Industries, catalog #04-007-1A) and 100 U/ml penicillin–streptomycin (Solarbio, catalog #P8420/S8290). Cells were maintained under standard conditions at 37°C in a humidified atmosphere containing 5% CO₂. For plasmid transfection, polyethylenimine (PEI; linear, MW 25,000; Polysciences, catalog #24765-1) was used at a DNA-to-PEI mass ratio of 1:2.5. The final concentration of PEI used was 1 mg/ml.

Primary dorsal root ganglion culture

The spinal cords were obtained from 8- to 12-week-old male and female mice and were placed in a 10 cm diameter culture dish containing 10 ml L15 medium (Thermo Fisher Scientific, catalog #41300070). Using tissue scissors and forceps, dorsal root ganglion (DRG) mass was carefully collected from the foramen. Subsequently, DRG mass was transferred to a 1.5 ml centrifuge tube, and collagenase (10 mg/ml, Sigma-Aldrich, catalog #C0130) and trypsin-EDTA (0.25%, Solarbio, catalog #T8150) were added. The mixture was then incubated at 37°C for 45 min with gentle inversion of the tube every 15 min to facilitate thorough digestion. The mixture was carefully transferred into a 15 ml centrifuge tube containing 15% BSA, with the mixture layered on top and the BSA at the bottom, followed by centrifugation at room temperature at 850 rpm for 20 min. After centrifugation, the supernatant and BSA layers were carefully removed, leaving the DRG at the bottom of the tube. The DRG was then gently resuspended and transferred to a 24-well plate precoated with poly-l-lysine (PLL, 5 mg/ml, Solarbio, catalog #P2100) and laminin (10 mg/ml, Sigma-Aldrich, catalog #CC095-M). The culture medium of DRG was Neurobasal (Thermo Fisher Scientific, catalog #21103049) containing B-27 (50×, Invitrogen, catalog #17504044), l-glutamine (100×, Sigma-Aldrich, catalog #G7513), and β-NGF (100 ng/ml, MedChemExpress, catalog #HY-P7661).

EAE model

The 8–12-week-old female mice were immunized subcutaneously on Days 0 and 7 with 100 μg of MOG 35–55 (MEVGWYRSPFSRVVHLYRNGK, Scilight Peptide, catalog #C213425) emulsified in CFA (Sigma-Aldrich, catalog #F5506) which was supplemented with 3 mg/ml of Mycobacterium tuberculosis (BDDS, catalog #231141) and injected intraperitoneally on Day 1 with 200 ng pertussis toxin (List Biological Laboratories, catalog #180) to induce EAE. Clinical assessment of EAE was performed daily; scoring was as follows: 0, no disease; 1, decreased tail tone; 2, hindlimb weakness or partial paralysis; 3, complete hindlimb paralysis; 4, front and hindlimb paralysis; and 5, moribund state.

Cuprizone model

Wild-type (WT) mice were fed with a diet containing 0.2% cuprizone (Sigma-Aldrich, catalog #C9012), and a new diet was changed every 3 d to ensure the freshness of the feed. In the following 5 weeks, the state of the mice is continuously observed. If there is any fighting phenomenon, the cage mouse experiment should be terminated and the mice should be killed, so as to ensure that all experimental individuals in the subsequent study are in a resting, peaceful, and normal growth state during the whole period of the experiment. After 5 weeks, the experimental individuals who met the criteria were executed, and the brain tissue was removed for follow-up study. The diet was changed to normal, after 1 week, collecting the brain tissue for the next research.

Immunofluorescence

Sciatic nerves (SNs) were embedded in a Tissue-Tek cryomold, and 6 μm sections were cut and collected on precoated slides. Sections were blocked with SuperBlock blocking buffer (Thermo Fisher Scientific, catalog #37515), for 1 h at room temperature. Tissues were incubated with primary antibodies overnight at 4°C. Sections were then washed with PBS and incubated with secondary antibodies for 1 h at room temperature. Slides were covered with antifade reagent.

Primary antibodies used for immunofluorescence are as follows: IGFBP2 (Abcam, catalog #ab188200, RRID: AB_2938998, 1:100), Neurofilament (Abcam, catalog #ab8135, RRID: AB_992703, 1:2,000), GAP43 (Cell Signaling Technology, catalog #8945S, RRID: AB_10860076, 1:400), MBP (Novus, catalog #NBP2-50035, RRID: AB_2938664, 1:2000), and S100 (Abcam, catalog #ab76749, RRID: AB_1566703, 1:100). Secondary antibodies used for immunofluorescence are as follows: donkey anti-mouse, Alexa Fluor 488 (Thermo Fisher Scientific, catalog# A-21202, RRID: AB_141607, 1:1,000); donkey anti-mouse, Alexa Fluor 555 (Thermo Fisher Scientific, catalog #A-31570, RRID: AB_2536180, 1: 1,000); donkey anti-rabbit, Alexa Fluor 488 (Thermo Fisher Scientific, catalog #A-21206, RRID: AB_141708, 1: 1,000); donkey anti-rabbit, Alexa Fluor 555 (Thermo Fisher Scientific, catalog #A-31572, RRID: AB_162543, 1: 1,000), and donkey anti-chicken, Alexa Fluor 555 (Thermo Fisher Scientific, catalog #A-21437, RRID: AB_2535858, 1:1,000).

Western blot

Protein samples were electrophoresed on SDS polyacrylamide gels and then transferred onto nitrocellulose membranes. Membranes were blocked with skim milk (5% in PBST) at room temperature for 1 h. Membranes were then incubated with primary antibodies overnight at 4°C, followed by incubation with secondary antibodies for 1 h at room temperature. Proteins were visualized using Western Lightening Plus ECL (Thermo Fisher Scientific, catalog #34580). GAPDH or β-actin was used as a loading control.

Primary antibodies used for Western blot are as follows: GAPDH (Proteintech, catalog #60004-1-Ig, RRID: AB_2107436, 1:3000), β-actin (Proteintech, catalog #66009-1-Ig, RRID: AB_2687938, 1:1000), GFP (Thermo Fisher Scientific, catalog #MA1-952, RRID: AB_889471, 1:1000), IGFBP2 (Abcam, catalog # ab188200, RRID:AB_2938998, 1:1,000), MAG (Proteintech, catalog #14386-1-AP, RRID: AB_2878051, 1:2000), NF-H/NF200 (Proteintech, catalog #60331-1-Ig, RRID: AB_2881440, 1:5000), and MBP (Novus, catalog #NBP2-50035, RRID: AB_2938664, 1:8,000). Secondary antibodies used for Western blot are as follows: peroxidase AffiniPure donkey anti-chicken IgY (H + L; Jackson ImmunoResearch Laboratories, catalog #703-035-155, RRID: AB_10015283, 1:20,000), peroxidase AffiniPure donkey anti-mouse IgG (H + L; Jackson ImmunoResearch Laboratories, catalog #703-035-150; RRID: AB_2340770, 1: 20,000), and peroxidase AffiniPure donkey anti-rabbit IgG (H + L; Jackson ImmunoResearch Laboratories, catalog #711-035-152, RRID: AB_10015282, 1: 20,000).

LFB staining

Cryosections were rehydrated in PBS for 5 min, dehydrated in 95% ethanol for 5 min, and incubated overnight at 56°C in 0.1% luxol fast blue (LFB, Sigma-Aldrich, catalog #S3382). Subsequently, the sections were washed in 95% ethanol and distilled water, followed by differentiation in 0.05% lithium carbonate for 30 s, rinsed in 70% ethanol and distilled water. Finally, the sections were stained with hematoxylin.

AAV packaging

A total of 20 dishes (15 cm²) of HEK293T-AAV cells were transfected with pAAV-MCS (16 μg/dish), together with pHelper (20 μg/dish) and pAAV-RC (24 μg/dish, carrying AAV-PHP.eB) using PEI (150 μl/dish). Post-transfection, cells were harvested at 72 h and resuspended in 1× gradient buffer. Cell suspension was subjected to freeze–thaw treatment (−80 and 55°C) and triturated through a 23 gauge syringe needle to aid cell lysis. This process was repeated an additional three times. To enhance cell lysis, benzonase (Sigma-Aldrich, catalog #E1014) was added at 1 ml per 5 ml of cell suspension, followed by incubation at 37°C for 1 h. After removing cell debris through centrifugation at 3,000 × g for 15 min, the supernatants containing the virus were purified through density gradient ultracentrifugation to a titer of 1 × 10¹³ viral genomes per milliliter (vg/ml).

Inhibitor injection

The SN of mice was cut-injured and intraperitoneal injection of fumonisin B1 (FB1, 8 mg/kg, ddH₂O, MCE, catalog #HY-N6719), or acid ceramidase-IN-1 (ACIN-1, 25 mg/kg, DMSO, MCE, catalog #HY-141866) was started on the day after the injury, lasting for 3 d. The optic nerve (ON) was crush-injured in mice. ACIN-1 (25 mg/kg) was injected intraperitoneally on the day after the injury was completed and once every other day. Samples were collected 7 d after the injury.

Surgical procedure

For SN cut or crush injury, 8–12-week-old male and female mice were anesthetized with narcolan (0.2 ml/10 g, Labseer, catalog #SR-RC9052). The SN was exposed at its mid-thigh level and was cut with fine surgical scissors or crushed for 40 s using forceps. Subsequently, the surgical site was sutured, and the mice were placed on a 37°C heating pad to aid recovery. After waiting for the corresponding injury time point, the nerves were dissected directly for Western blot analysis. Alternatively, they were dissected, postfixed with 4% paraformaldehyde, and cryoprotected with 20% sucrose for subsequent use in immunofluorescence analysis.

The intrasciatic nerve injection procedure was conducted according to the following steps: after inducing deep anesthesia, the SN in mice was exposed. Subsequently, 8 × 10¹⁰ vg virus particles were injected directly into the SN using a glass capillary and a microsyringe pump at a rate of 0.3 μl/min. Post injections, the surgical site was sutured, and the mice were placed on a 37°C heating pad to aid recovery.

For ON crush and intraocular injections. After inducing deep anesthesia, 8 × 10¹⁰ vg virus particles were injected directly into the vitreous body using a glass capillary and a microsyringe pump at a rate of 0.3 μl/min. Four weeks after vitreous body injection, using angled forceps, the ON was crushed 2 mm behind the nerve head for 5 s. Four weeks after ON injury, 3 μl cholera toxin subunit B (CTB)–Alexa Fluor 555 (Thermo Fisher Scientific, catalog #C34776) was injected into the vitreous body in the same way.

Ex vivo cultured nerve injury model and siRNA transfection. SNs or ON of WT, male and female mice were dissected and placed in a 35 mm diameter culture dish containing 2 ml L15 medium. Following the removal of the epineurium using forceps under a dissecting stereo microscope, each nerve was individually placed in separate wells of a 24-well plate. Each well contained 500 μl of DMEM supplemented with 5% FBS and 100 units/ml penicillin/streptomycin. Subsequently, the plates were incubated at 37°C with 5% CO₂ for 3 d.

Following the same procedure as nerves cultured ex vivo, after placing in separate wells of a 24-well plate, they were transfected with 10 nM siRNAs targeting different genes and nontargeting negative control for 72 h using lipofectamine RNAiMAX transfection reagent. After 3 d, use forceps to transfer the nerves to 2 ml centrifuge tubes for RNA extraction experiments, Western blot, or store them at −80°C.

RNA extraction, cDNA synthesis, and real-time qPCR

Extraction of total RNA was using Trizol Reagent of RNAsimple Total RNA Kit (TIANGEN, catalog #DP419). RNA was quantified by Nanodrop, with 1 μg used for cDNA synthesis with FastKing gDNA Dispelling RT SuperMix (TIANGEN, catalog #KR118). Quantitative PCR was run with 0.1 μg cDNA input in a 20 ml reaction using ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech, catalog #Q711) using standard cycling parameters (95°C for 30 s, 40 cycles of 95°C for 10 s, and 60°C for 30 s), followed by standard dissociation (95°C for 15 s at 1.6°C s⁻¹, 60°C for 60 s at 1.6°C s⁻¹, 95°C for 15 s at 0.075°C s⁻¹). ΔΔCt was calculated with the housekeeper gene GAPDH as the control. The primers used for real-time qPCR are as follows: Gapdh, forward 5′-AGGTCGGTGTGAACGGATTTG-3′, reverse 5′-TGTAGACCATGTAGTTGAGGTCA-3′; 18s, forward 5′-GACAGGATTGACAGATTGATAG-3′, reverse 5′-CCAGAGTCTCGTTCGTTA-3′; Igfbp2, forward 5′-TAGCTCGCTGCACTACCGTT-3′, reverse 5′-CAGCAGCAAGAGCAACGAC-3′; Kbtbd11, forward 5′-TGGAGACGCAGTGGGATGT-3′, reverse 5′-CCAAGCTCGTCCAAAGACTCG-3′; Ptafr, forward 5′-AACGAGGGCGACTGGATTCTA-3′, reverse 5′-CTACTGCCTGGTAGCGGTTA-3′; Tgfa, forward 5′-CACTCTGGGTACGTGGGTG-3′, reverse 5′-CACAGGTGATAATGAGGACAGC-3′; Cyb5r2, forward 5′-TACCCGCTGCCCTTGATTG-3′, reverse 5′-TGGACATAGTTACCGACAGGAAG-3′; Map3k11, forward 5′-CAACAGCAGCTTATGCCAATCC-3′, reverse 5′-TTCCTCAGGGCTAGTTCGTCC-3′; Zeb1, forward 5′-GCTGGCAAGACAACGTGAAAG-3′, reverse 5′-GCCTCAGGATAAATGACGGC-3′; Mthfr, forward 5′-GGCAGCGAGAGTTCCAAGG-3′, reverse 5′-CAGGGAGAACCACTTGTCACC-3′; Wwc1, forward 5′-TGCTGAGGGAAACCAAAGCC-3′, reverse 5′-CTGGACCATAGGTCGGAGTG-3′; Slc22a4, forward 5′-TGGTATGTCAGTCGTGTTCCT-3′; reverse 5′-AGCCCCATCGCAGAGAAGT-3′; Akr1b3, forward 5′-AGGCCGTGAAAGTTGCTATTG-3′, reverse 5′-ATGCTCTTGTCATGGAACGTG-3′; Kcnj8, forward 5′-AAGAGCATCATCCCGGAGGA-3′, reverse 5′-ATGTTCTTGTGTGCCAGGTTG-3′; P4ha2, forward 5′-CACCTCCATTGGGCACATGA-3′, reverse 5′-GCTCTTAATCTTGGCGAGCTT-3′; Synpo, forward 5′-CCTGCCCGTAACTTCCGTG-3′, reverse 5′-GAGCGGCGGTAGGGAAAAG-3′; Rhoq, forward 5′-GTACGTGCCCACTGTCTTCG-3′, reverse 5′-GGCCTCAGACGATCATAGTCTT-3′; Ly9, forward 5′-TCAGGGATGCTAGGGGGTTC-3′, reverse 5′-TTCGCTGACTTTGAGTCTGCC-3′; Myo1f, forward 5′-CTTTCACTGGCAGAGTCACAA-3′, reverse 5′-ATGAAGCGTTTGCGGAGGTT-3′; Rnf128, forward 5′-ACAAGGAAATTGGCCCTGATG-3′, reverse 5′-CCTGTGTTCTAAAAGCCACGG-3′; Ifi30, forward 5′-CCTGGTCTCCGATCCTACCAT-3′, reverse 5′-TTGCAGGTGGTTGTGCCTT-3′; Tnfaip8l2, forward 5′-TCAGCTCAAAGAGTCTGGCAC-3′, reverse 5′-GGTAAAGCTCGTCTAGCACCTC-3′; Clec4a3, forward 5′-CTTCACTTCAACTGACTTGGTGG-3′, reverse 5′-TCACTGCTAGGCTCACCTTTG-3′; Prss12, forward 5′-GGCAGACCTTGGTGCTTCTAT-3′, reverse 5′-CACTCGACCTTCATGCCCA-3′; Col13a1, forward 5′-GGAGCACCTGGACTAGACG-3′, reverse 5′-GCCTTGGACTGGTAAGCCAT-3′; Pdgfb, forward 5′-CATCCGCTCCTTTGATGATCTT-3′, reverse 5′-GTGCTCGGGTCATGTTCAAGT-3′; Chst11, forward 5′-GATGAACAGAATTTGCCGGATGG-3′, reverse 5′-GTTCCTCCGCATGACTGGG-3′; Apobec1, forward 5′-TGTAGCTGTTGATCCCACTCT-3′, reverse 5′-CTGGTGTTTTGGCTCGTGT-3′; Mustn1, forward 5′-GTCTAAGACATACCAGGTCATGC-3′, reverse 5′-GCGGCTGAATACAGATGGGG-3′; Fads3, forward 5′-GAGTCATCTGTGGAGTAT-3′, reverse 5′-CTATACTTCTTCCTGATTGG-3′; Fscn1, forward 5′-CACCCACCTCATTCCTTT-3′, reverse 5′-AGGAAATGTTACTGTTGAAAGG-3′; Plppr4, forward 5′-TCAGGAGAGCCGTCAGATTCGTT-3′, reverse 5′-CTGCCACTGTTGATGACTGTAAGGT-3′; Alb, forward 5′-AACCAGGCGACTATCTCCAGCAA-3′, reverse 5′-CAGCAGCAATGGCAGGCAGAT-3′; Apol9a, forward 5′-CCATAGCAGACAGAGGCAGACTTC-3′, reverse 5′-AGAGACCAAGGAGGCTCAGGATG-3′; Olfr78, forward 5′-ATGCCACCTTCCTGCTTATTG-3′, reverse 5′-GCTCCGCTCTGTTCTCACTA-3′; Ptgs2, forward 5′-GAGTCTGGAACATTGTGAAC-3′, reverse 5′-GTAGTAGGAGAGGTTGGAGA-3′; Lep, forward 5′-TCAAGCAGTGCCTATCCAGAA-3′, reverse 5′-GAATGAAGTCCAAGCCAGTGA-3′; Lpl, forward 5′-GCTCCATTCATCTCTTCATTG-3′, reverse 5′-TACATCTTGCTGCTTCTCTT-3′; and Ankrd26, forward 5′-AGCCACTGTGAACTACTTAC-3′, reverse 5′-TCCTCTTCTCTTCCTCTTGT-3′.

RNA sequencing

RNA from SNs or ONs of WT, Plp/Cre ER^T; Igfbp2^fl/fl mice was prepared for RNA-seq (two biological repeats for each group). The management approaches for in vivo and ex vivo nerve injuries are detailed above, Surgical procedure. Total RNA was extracted from SNs or ONs using RNAsimple Total RNA Kit and was quantified by Nanodrop. RNA-seq experiments and analysis were performed by Berry Genomics.

LC–MS lipidomic

Lipid extraction from intact and 3-d-injured SNs of WT and Plp/Cre ER^T; Igfbp2^fl/fl mice. Afterward, supernatant was carefully transferred to sample vials for the LC–MS/MS analysis. Lipidomic experiments and analysis were performed by Majorbio.

Electron microscopy

Fresh isolated SNs were primary fixed with 1% paraformaldehyde and 2.5% glutaraldehyde, followed by secondary fixing with 1% OsO₄. Samples were then dehydrated through an ascending acetone series (15–100% in seven steps). The samples were then embedded in Spurr resin and sectioned to 60–80 nm. Sections were visualized using transmission electron microscope.

Mouse behavioral assessments

To evaluate motor function, the SN functional recovery assay was performed. Specifically, toe spread (TS, defined as the distance between the first and fifth toes) and print length (PL, measured from the tip of the third toe to the rearmost point of the footprint in contact with the ground surface) were assessed. TSF (toe spread factor) was calculated with the following formula: (injured TS − intact TS) / intact TS. PLF (print length factor) was calculated with the following formula: (injured PL − intact PL) / intact PL. Sciatic functional index (SFI) was calculated with the following formula: 118.9 × TSF-51.2 × PLF-7.5.

To eliminate the influence of sex on mouse behavior, male mice were exclusively used for mouse behavioral assessments. Additionally, the experimenter was blinded from the genotypes while performing behavioral testing and analyzing behavioral data.

Quantification and statistical analysis

The statistical details of our study can be found in the method details and figure legends. Statistical analyses were performed with the ImageJ and GraphPad Prism 9 software. Differences between means were assessed by an unpaired, two-tailed t test with unequal variance or one-way ANOVA with post hoc Tukey's analysis or two-way ANOVA with post hoc Tukey's analysis. The number of animals per group or independent replicates, the statistical test used for comparison, and the statistical significance (p value) are indicated in the figure legends. Values are expressed as means ± SD or means ± SEM, with differences considered significant at p < 0.05; n.s., no significance; *p < 0.05; **p < 0.01; ***p < 0.0005; and ****p < 0.0001.

Results

Differential gene expression profiles following ex vivo injury

PNI exhibits remarkable regenerative capacity compared with CNS injuries (Varadarajan et al., 2022). This difference is partially attributed to the efficient removal of myelin, a process crucial for axonal regeneration in PNI (Varadarajan et al., 2022; Yuan et al., 2022). To elucidate the underlying mechanisms responsible for this distinct response, we employed RNA sequencing (RNA-seq) to compare the gene expression profiles of ex vivo injured SNs and ONs, which serve as simplified models of PNS and CNS tissues in an isolated setting, eliminating macrophage-derived confounding signals and allowing specific exploration of intrinsic pathways regulating myelin breakdown (Ying et al., 2018; Babetto, 2020; Sundaram et al., 2023), respectively (Fig. 1A). Although the ex vivo models do not fully replicate the complexity of the in vivo injury microenvironment, they allow for the investigation of tissue-intrinsic molecular changes, thereby facilitating the identification of candidate regulators for subsequent in vivo validation.

Figure 1.

Differential gene expression profiles in injured sciatic and ONs. A, Experimental process of RNA-seq. SNs and ONs of WT mice were cultured ex vivo for 1 and 3 d. B, C, Volcano of the differential expressed genes of the injured SNs (B) and ONs (C) compared with intact nerves (sham control). The x-axis is log2 [fold change (FC)], and the y-axis is −log10 (p value). Upregulated genes and downregulated genes are indicated in light red [|log2(fold change) | >1, p < 0.05], genes with no significant difference are marked in gray [|log2(fold change) | <1, p < 0.05], genes with |log2(fold change) | <1 and p > 0.05 are denoted in blue, and genes with |log2(fold change) | >1 and p > 0.05 are labeled in green. D, Cluster analysis model diagram of DEGs. M1 and M2 represent the upregulated expression level with injury time, M3 and M4 represent the downregulated expression level with injury time. E, The Venn diagram of opposite expression trend between in vitro cultured SNs and ONs. The yellow and blue histograms depict the sizes of corresponding gene sets, while the matrix section with black dots indicate the data source represented by the blue histogram; connected dots represent the number of intersecting genes between the gene sets corresponding to the heights of the blue histogram; the intersection of gene sets with opposite expression trends in the SNs (black) and ONs (green). F, Heatmap of the DEGs in the Venn diagram (E). Excluding three predicted genes.

RNA-seq analysis revealed distinct gene expression patterns between injured and intact nerves (sham control) in both the sciatic and ON models (Fig. 1B,C). A significant number of genes displayed altered expression compared with intact nerves in both injured models. Differentially expressed genes (DEGs) were classified into four clusters (Fig. 1D). We hypothesized that the distinct myelin breakdown profiles observed in the PNS and CNS following injury are caused by differential gene expression patterns in these two tissues. A total of 16 genes were identified to be significantly upregulated in injured SNs compared with injured ONs (Fig. 1E; SNs-M1/ONs-M3, SNs-M2/ONs-M4). These genes, collectively referred to as “SNs-upregulated genes,” may be involved in regulating efficient myelin breakdown in the injured SN, as well as other nerve injury response processes. Conversely, 16 genes were found to be significantly downregulated in injured SNs compared with injured ONs (Fig. 1E; SNs-M3/ONs-M1, SNs-M4/ONs-M2). These “ON-upregulated genes” may play a role in the limited myelin breakdown observed in the ONs after injury. Among these 32 genes exhibiting opposing expression patterns, we excluded 3 computational predicted genes that were only predicted by computational analysis without experimental validation or known biological function and focused our investigation on the remaining 29 genes (Fig. 1F).

To further investigate the 29 candidate genes, we performed quantitative real-time PCR (qPCR) analysis (Fig. S1A) and Gene Ontology (GO) analysis (Fig. S1B) on ex vivo cultured nerves. We then focused on a subset of nine genes for further analysis (Fig. S1A). These nine genes exhibited a specific upregulation in ex vivo injury model of SNs compared with ex vivo injury model of ONs. This targeted selection aimed to prioritize genes potentially involved in the processes unique to PNI.

To assess the functional roles of the nine candidate genes, we employed siRNA-mediated knockdown in ex vivo cultured SNs. Western blot analysis was performed to quantify the protein levels of myelin markers, myelin basic protein (MBP) and myelin-associated glycoprotein (MAG), following injury and siRNA treatment (Fig. S1C–K). As expected, both MBP and MAG levels decreased in injured SNs compared with intact controls (Fig. S1C,F,I, first three lanes). Interestingly, knockdown of four specific genes among the nine candidates resulted in a significant increase in MBP and MAG protein levels compared with control siRNA treatment (Fig. S1C–E). These findings suggest that these four genes may play a role in regulating myelin breakdown following PNI. Additionally, immunofluorescence staining with MBP antibody was employed to visualize myelin integrity in ex vivo cultured SNs following siRNA treatment (Fig. S1L). Quantification of intact myelin sheaths from the immunofluorescence images corroborated the Western blot findings (Fig. S1M).

Among the four genes investigated, Igfbp2 emerged as the most promising candidate. Knockdown of Igfbp2 resulted in the most pronounced increase in both MBP and MAG protein levels (Fig. S1C–E) and the best-preserved myelin integrity (Fig. S1L,M) compared with control siRNA treatment.

SN injury induces IGFBP2 expression in Schwann cells

To investigate the cellular localization and expression profile of IGFBP2 following nerve injury, we employed immunofluorescence staining and qPCR analysis (Fig. 2). In intact nerves, MBP displayed robust and uniform staining throughout the nerve fibers, indicating intact myelin sheaths. Notably, IGFBP2 staining was minimal in these sections. Following injury (3 and 5 d), a significant decrease in MBP staining intensity was observed, suggesting a myelin breakdown (Fig. 2A,B). Conversely, IGFBP2 staining intensity increased in the injured nerve fibers compared with intact controls (Fig. 2A,B). The expression level of S100, a Schwann cell marker, was increased after SN injury, indicating an expansion of the Schwann cell population (Fig. 2C).

Figure 2.

IGFBP2 upregulation in Schwann cells of injured peripheral nerves. A, B, Immunofluorescence staining of cross sections of intact, 3 or 5 d cut-injured SNs from WT mice was conducted using antibodies against MBP (red, myelin sheath marker), IGFBP2 (green). Cell nuclei (blue, nuclear marker). Scale bars, 10 μm in A and 2 μm in B. C, Following crush-injured SNs for 3 or 5 d in WT mice, immunofluorescent staining was performed to assess the expression of S100 (red, Schwann cell marker) and IGFBP2 (green) protein on longitudinal nerve sections. White asterisks indicate the site of nerve injury. Scale bar, 100 μm. D, Relative mRNA expression levels of Igfbp2 in 3 d cut SNs and 3 d crushed ONs from WT mice by qPCR. Mean ± SD (n = 3 independent experiments). **p < 0.001; ****p < 0.0001; two-way ANOVA. E, Time course analysis of protein levels by Western blot in intact and cut-injured SNs from WT mice. β-Actin was used as a loading control. F, Time course analysis of protein levels by Western blot in intact and crush-injured ONs from WT mice. β-Actin was used as a loading control. G, In the EAE model, cross sections of the spinal cord was coimmunostained with antibodies against MBP and IGFBP2. Scale bar, 500 μm. The white-bordered areas outline regions of demyelination. All tissues were harvested at postmodeling Day 22, strictly adhering to the inclusion criterion of a clinical severity score ≥3. H, I, In the cuprizone model, immunofluorescent staining of corpus callosum cross sections was conducted using antibodies against MBP and IGFBP2. Scale bars, 500 μm (H) and 200 μm in I. Samples from animals fed a cuprizone diet for 5 weeks were designated as CPZ-5W, while those subjected to a subsequent 2 week recovery period on a standard diet were labeled CPZ-5 + 2W.

This spatial and temporal pattern of IGFBP2 expression suggests a potential role in the postinjury response. To quantify the mRNA expression of Igfbp2, we performed qPCR from in vivo injured SNs (3 d postinjury) and ONs (3 d postinjury) from WT mice. Igfbp2 expression was significantly upregulated in injured SNs and downregulated in injured ONs (Fig. 2D). We employed the Western blot to assess IGFBP2 protein levels in SNs and ONs following injury. In SNs, IGFBP2 protein levels progressively increased over time (from intact to 3 or 5 d postinjury), mirroring the immunofluorescence staining results (Fig. 2E). Conversely, IGFBP2 protein levels remained unchanged in injured ONs (Fig. 2F).

To determine if IGFBP2 expression is specific to PNI, we examined its presence in two established models of CNS demyelination models: experimental autoimmune encephalomyelitis (EAE) and cuprizone-induced demyelination. MBP staining confirmed demyelination in both models. Immunofluorescence staining revealed minimal IGFBP2 expression in spinal cord lesions from the EAE model (Fig. 2G) and corpus callosum lesions from the cuprizone model (Fig. 2H,I). These findings suggest that IGFBP2 upregulation might be a specific response to PNI.

Schwann cell-specific knock-out of Igfbp2 slows down axotomy-induced myelin breakdown

To investigate the functional role of Schwann cell-derived IGFBP2 in myelin breakdown and subsequent regeneration following PNI in vivo, we utilized Schwann cell-specific Igfbp2 knock-out (Igfbp2-ScKO) mice generated through a Cre-loxP recombination system (Fig. 3A). Igfbp2-ScKO mice (Plp/CreER^T; Igfbp2^fl/fl) were compared with heterozygous (Plp/CreER^T; Igfbp2^fl/+) littermates and WT after SN injury.

Figure 3.

Schwann cell-specific knock-out of Igfbp2 impairs myelin breakdown and peripheral nerve regeneration. A, F1 generation mice were obtained by crossing Plp/CreER^T genotype mice with Igfbp2^fl/fl mice, resulting in Plp/CreER^T; Igfbp2^fl/+ mice; Plp/CreER^T; Igfbp2^fl/fl mice were obtained by self-fertilization of Plp/CreER^T; Igfbp2^fl/+ mice. B, Myelin-associated proteins levels examined by Western blot in intact, 3 or 5 d cut-injured SNs from WT, Plp/CreER^T; Igfbp2^fl/+ and Plp/CreER^T; Igfbp2^fl/fl mice. GAPDH was used as a loading control. C, Immunofluorescent staining of cross sections of intact and cut-injured SNs was carried out using antibodies against MBP, IGFBP2. Cell nuclei (DAPI, blue). Scale bar, 10 μm. D, Representative transmission electron micrographs of intact, 3 or 5 d cut-injured SNs from WT, Plp/CreER^T; Igfbp2^fl/+ and Plp/CreER^T; Igfbp2^fl/fl mice. Top, Low magnification; scale bar, 5 μm. Bottom, High magnification of the boxed regions at the top; scale bars, 1 μm. E, Western blot analysis was conducted to assess the protein expression levels in the 5 d cut-injured SNs from Plp/Cre ER^T; Igfbp2^fl/fl mice rescued by AAV-CMV-Igfbp2 injection. The expression levels of IGFBP2 were assessed using a GFP-specific antibody. GAPDH was used as a loading control. F, Immunofluorescence staining about cross sections of 3 or 5 d cut-injured SNs from Plp/Cre ER^T; Igfbp2^fl/fl mice rescued by AAV-CMV-Igfbp2 injection. Scale bar, 10 μm. G, Following crush-injured SNs for 14 or 28 d in WT and Plp/CreER^T; Igfbp2^fl/fl mice, immunofluorescent staining was performed to assess the expression of GAP43 (green, neuronal regeneration marker) protein on longitudinal nerve sections. White asterisks indicate the site of nerve injury. Scale bar, 500 μm. H, Representative images of hindlimb footprints under different conditions: normal footprint (uninjured), WT mice at 1 and 4 weeks after SN crush injury, and Plp/CreER^T; Igfbp2^fl/fl mice at 4 weeks postinjury. Scale bar, 0.5 cm. I, The SN functional index was measured in WT, Plp/CreER^T; Igfbp2^fl/fl mice following SN crush injury. Mean ± SEM (n = 6 mice from three independent experiments); ****p < 0.0001; two-way ANOVA.

Western blot analysis (Fig. 3B; Fig. S2A,B) revealed that Igfbp2-ScKO mice displayed higher levels of MBP and MAG in injured nerves compared with WT and heterozygous mice at all time points (3 and 5 d postinjury). This suggests preservation of myelin sheaths in Igfbp2-ScKO mice following injury. Immunofluorescence staining (Fig. 3C) corroborated this finding, showing a significantly higher degree of intact myelin sheaths in Igfbp2-ScKO mice compared with WT mice (Fig. S2C). Importantly, intact SNs from WT and Igfbp2-ScKO mice showed no significant differences in intact myelin sheath number (Fig. 3C), indicating that IGFBP2 does not affect myelin under physiological conditions and that its role in myelin degradation requires injury as a triggering signal. Analysis of longitudinal nerve sections using immunostaining (Fig. S2D,E) and Luxol fast blue staining (Fig. S2F,G) mirrored the results obtained from cross sections.

Similarly, transmission electron microscopy (Fig. 3D) confirmed a greater preservation of myelin ultrastructure in Igfbp2-ScKO mice (Fig. S2H). These findings collectively demonstrate that Schwann cell-derived IGFBP2 deficiency hinders myelin breakdown after PNI. To investigate other cell-type–specific role of Igfbp2 in myelin breakdown, we employed conditional knock-out mice with Cre recombinase expression driven by either Camk2a (for neuron-specific deletion) or Cx3cr1 (for macrophage-specific deletion) promoters. Neither neuronal nor macrophage-specific deletion of Igfbp2 had a measurable impact on myelin breakdown after PNI, as assessed by myelin sheath integrity (Fig. S3A,B,D,E) and protein levels (Fig. S3C,F).

To validate further the critical role of Schwann cell-derived IGFBP2 in promoting myelin breakdown after PNI, we performed a rescue experiment in Igfbp2-ScKO mice. AAV-CMV-Igfbp2 vectors were injected into the SNs of Igfbp2-ScKO mice to restore IGFBP2 expression (Fig. 3E). Western blot analysis confirmed successful restoration of IGFBP2 protein levels in injured nerves from Igfbp2-ScKO mice following AAV-CMV-Igfbp2 injection compared with control AAV-CMV-Gfp injection (Fig. 3E; Fig. S2I,J). Immunofluorescence staining revealed a significant decrease in MBP staining in AAV-CMV-Igfbp2–injected Igfbp2-ScKO mice compared with control AAV-injected mice at both 3 and 5 d postinjured nerves (Fig. 3F). This suggests a restoration of efficient myelin breakdown upon IGFBP2 re-expression. Quantification of intact myelin sheaths from the immunofluorescence images corroborated this observation, demonstrating a reduction in myelin integrity following AAV-CMV-Igfbp2 treatment (Fig. S2K).

Since the myelin clearance is a prerequisite for effective nerve regeneration, we next assessed the impact of IGFBP2 on axon regeneration; we evaluated GAP43, a marker for axon regeneration, in injured nerves from WT and Igfbp2-ScKO mice (Fig. 3G). Igfbp2-ScKO mice displayed a significantly lower mean fluorescence intensity of GAP43 in the distal SNs compared with WT mice (Fig. S2L), suggesting limited axonal regeneration following injury in the absence of Schwann cell-derived IGFBP2.

To further investigate the impact of IGFBP2 on nerve regeneration, we assessed the footprints and SFI in WT and Igfbp2-ScKO mice following SN injury. In WT mice, footprints before injury reflect normal SN function—hindlimbs are strong during locomotion, and all five toes are fully extended. One week after SN crush, WT mice display classic foot drop: weakened hindlimb strength causes the entire sole of the hindpaw to drag against the ground leading to an increase in PLF and a reduction in TSF. After 4 weeks of recovery, WT footprints resemble those prior to injury, with restored toe extension and gait. In contrast, Igfbp2-ScKO mice exhibit persistent deficits even after 4 weeks postinjury, characterized by continued foot dragging (increased PLF) and incomplete toe extension (reduced TSF; Fig. 3H). Prior to surgery, WT and Igfbp2-ScKO exhibited SFI values close to 0, indicating normal SN function (Baptista et al., 2007; Santos et al., 2014). SFI dropped equally in both mice after surgery. Over the subsequent 4 week recovery period, SFI values for WT mice showed a significantly faster and more extensive increase compared with those for Igfbp2-ScKO mice (Fig. 3I). These results indicate that Schwann cell-specific Igfbp2 deficiency slows down functional recovery after PNI.

Schwann cell-specific knock-out of Igfbp2 alters lipid metabolism in injured SNs

To gain further insights into the mechanisms by which IGFBP2 regulates myelin breakdown, we performed RNA-seq analysis of SNs from Schwann cell-specific Igfbp2 knock-out mice and WT mice at 3 d postinjury (Fig. 4A). Interestingly, this analysis revealed a significant downregulation of genes associated with lipid metabolism pathways in Igfbp2-ScKO mice compared with WT controls (Fig. S4A). We then employed GO analysis to categorize the DEGs in the comparison between WT and Igfbp2-ScKO injured nerves (Fig. 4B) as well as between injured and intact SNs in WT mice (Fig. 4C). This analysis revealed a marked upregulation of genes involved in lipid metabolism pathways following peripheral nerve injury (Fig. 4C), which was abolished in Schwann cell-specific Igfbp2 knock-out mice (Fig. 4B).

Figure 4.

Schwann cell-specific knock-out of Igfbp2 alters gene expression and lipid metabolism after peripheral nerve injury. A, The volcano plot depicts DEGs in the 3 d cut-injured SNs of WT and Plp/CreER^T; Igfbp2^fl/fl mice. Upregulated genes and downregulated genes are indicated in light red (|log2 fold change| >1, p < 0.05), genes with no significant difference are marked in gray (|log2 fold change| <1, p < 0.05), genes with |log2 fold change| <1 and p > 0.05 are denoted in blue, and genes with |log2 fold change| >1 and p > 0.05 are labeled in green. Points specifically annotated with gene names represent genes associated with lipid metabolism. B, C, GO analysis of DEGs of WT versus Plp/CreER^T; Igfbp2^fl/fl 3 d cut-injured SNs (B), intact versus 3 d in vitro cultured SNs from WT mice (C). The black box shows the pathways involved in lipid metabolism. D, GO analysis of DEGs of diabetic neuropathy patients’ sural nerve versus normal patients’ sural nerve. E, The volcano plot illustrates relevant fatty acids differentially expressed in the 3 d cut-injured SNs of WT and Plp/CreER^T; Igfbp2^fl/fl mice. Green represents downregulated fatty acids (log2FoldChange < −1, p < 0.05), red represents upregulated fatty acids (log2FoldChange > 1, p < 0.05), and gray represents fatty acids with no significant difference in expression (|log2FoldChange| ≤1, p < 0.05). F, Classification of differentially expressed fatty acids.

To validate the RNA-seq data, we selected a subset of lipid metabolism-related genes from the DEGs and assessed their mRNA expression levels by qPCR (Fig. S4B). The qPCR results confirmed the downregulation of these genes in Igfbp2-ScKO mice compared with WT mice following SNs injury. Intriguingly, our analysis of RNA-seq data from patients with diabetic neuropathy, a condition often associated with progressive myelin damage (Cermenati et al., 2012; Guo et al., 2022), revealed a similar activation of multiple lipid metabolism pathways in nerves (Fig. 4D). This observation suggests that the induction of lipid metabolism is an important response to PNI, potentially serving as a prerequisite for myelin breakdown and subsequent axonal regeneration.

Building upon the findings of altered lipid metabolism gene expression in Igfbp2-ScKO mice, we performed lipidomics to investigate the specific lipid metabolite profile in injured SNs. Liquid chromatography coupled with tandem mass spectrometry (LC–MS/MS) analysis was performed on lipid extracts isolated from injured SNs of WT and Igfbp2-ScKO mice. A volcano plot depicted the differentially expressed fatty acids between the two groups (Fig. 4E). This analysis revealed a significant alteration in the levels of various fatty acids following PNI in Igfbp2-ScKO mice compared with WT mice. The differentially expressed fatty acids were categorized based on their chemical structure (Fig. 4F). Heatmap analysis was then employed to visualize the expression patterns of the top three enriched classes: triglyceride (Fig. S4C), cholesterol (Fig. S4D), and ceramide (Fig. S4E). Notably, the heatmaps revealed distinct expression profiles for these lipid classes which were significantly decreased in Igfbp2-ScKO mice compared with WT mice, suggesting a potential influence of Schwann cell-derived IGFBP2 on their abundance after PNI.

We also assessed changes in lipid metabolism in WT mice following PNI (Fig. S5A). This analysis revealed a significant alteration in the lipid profile following PNI (Fig. S5B). Further classification of the differentially expressed fatty acids also identified changes in triglycerides, cholesterol, and ceramides (Fig. S5C). Heatmap analysis (Fig. S5D–F) demonstrated a clear upregulation of triglycerides, cholesterol, and ceramides levels. These findings further indicate the induction of lipid metabolism pathways as a crucial response to PNI.

Cers6-mediated ceramide synthesis modulates myelin breakdown in Igfbp2-ScKO mice following PNI

Since triglycerides, cholesterol, and ceramide levels were significantly reduced in Igfbp2-ScKO mice following PNI (Fig. 4F), we investigated whether supplementation of these three types of lipids could influence the observed phenotype. Igfbp2-ScKO mice with SNs injury were administered triglycerides (Fig. S6A–C), cholesterol (Fig. S6D–F), and ceramide or a control vehicle after injury (Fig. 5A). Myelin integrity was then assessed using various techniques. Western blot analysis (Fig. 5B; Fig. S6G,H) revealed that ceramide-treated Igfbp2-ScKO mice displayed lower levels of MAG and MBP compared with vehicle-treated controls, suggesting improved myelin breakdown. This finding was corroborated by immunofluorescence staining (Fig. 5C,D) and transmission electron microscopy (Fig. 5E,F), which both demonstrated a significantly greater degree of myelin breakdown in ceramide-treated Igfbp2-ScKO mice compared with controls. However, supplementation with triglycerides and cholesterol did not have a significant effect on the rate of myelin breakdown after PNI in Igfbp2-ScKO mice (Fig. S6).

Figure 5.

Schwann cell-specific knock-out of Igfbp2 impairs ceramide-mediated myelin breakdown after peripheral nerve injury. A, Experimental process of ceramide rescue in the Plp/CreER^T; Igfbp2^fl/fl mice. The control group received equivalent injections of vehicle. B, Myelin-associated proteins levels examined by western blot in 3 d cut-injured SNs after ceramide rescue from Plp/CreER^T; Igfbp2^fl/fl mice. GAPDH was used as a loading control. C, Immunofluorescent staining of cross sections of 3 d cut-injured SNs after ceramide rescue from Plp/CreER^T; Igfbp2^fl/fl mice was carried out using antibodies against MBP, IGFBP2. Cell nuclei (DAPI, blue). Scale bar, 10 μm. D, Quantification of intact myelin sheath of C. Mean ± SEM (n = 4 images from three independent experiments); **p < 0.01; ****p < 0.0001; one-way ANOVA. E, Representative transmission electron micrographs of 3 d cut-injured SNs after ceramide rescue from Plp/CreER^T; Igfbp2^fl/fl mice. Top, Low magnification; scale bar, 5 μm. Bottom, High magnification of the boxed regions at the top; scale bars, 2 μm. F, Quantification of intact myelin sheath of (E). Mean ± SEM (n = 3 mice from three independent experiments); **p < 0.01; ****p < 0.0001; one-way ANOVA. G, Myelin-associated proteins levels examined by Western blot in 3 d cut-injured SNs after FB1 injection in WT mice and ACIN-1 injection in Plp/CreER^T; Igfbp2^fl/fl mice. GAPDH was used as a loading control. H, Immunofluorescent staining of cross sections of 3 d cut-injured SNs after FB1 injection in WT mice and ACIN-1 injection in Plp/CreER^T; Igfbp2^fl/fl mice was carried out using antibodies against MBP, IGFBP2. Cell nuclei (DAPI, blue). Scale bar, 10 μm. I, Quantification of intact myelin sheath of H. Mean ± SEM (n = 4 images from three independent experiments). ns, p > 0.05; **p < 0.01; ***p < 0.0005; ****p < 0.0001; one-way ANOVA. J, Representative transmission electron micrographs of 3 d cut-injured SNs after FB1 injection in WT mice and ACIN-1 injection in Plp/CreER^T; Igfbp2^fl/fl mice. Top, Low magnification; scale bar, 5 μm. Bottom, High magnification of the boxed regions at the top; scale bars, 2 μm. K, Quantification of intact myelin sheath of J. Mean ± SEM (n = 3 mice from three independent experiments). ns, p > 0.05; ***p < 0.0005; ****p < 0.0001; one-way ANOVA.

We then focused on exploring the role of ceramide in myelin breakdown by pharmacological interventions. WT mice with SN injury were treated with fumonisin B₁ (FB₁), a ceramide synthase inhibitor to decrease the ceramide level (Shaheen et al., 2016), while Igfbp2-ScKO mice were treated with ACIN-1, a ceramidase activator to increase the ceramide level (Di Martino et al., 2020). Similar to Igfbp2-ScKO mice, FB₁ treatment in WT mice resulted in increased levels of MAG and MBP (Fig. 5G; Fig. S6I,J), enhanced myelin sheath preservation as revealed by immunofluorescence staining (Fig. 5H,I), and electron microscopy (Fig. 5J,K). Conversely, ACIN-1 treatment in Igfbp2-ScKO mice significantly enhanced myelin breakdown phenotype compared with vehicle-treated controls (Fig. 5G–K).

To investigate how IGFBP2 regulates ceramide, we analyzed the expression of ceramide synthase enzymes in Igfbp2-ScKO mice. Among ceramide synthase isoforms (ceramide synthase 1–6, Cers1–Cers6; Xie et al., 2023), the expression of Cers6 was markedly upregulated in the injured SNs of WT mice, but the upregulation was significantly decreased in Igfbp2-ScKO mice (Fig. 6A). To determine whether overexpressing Cers6 in Igfbp2-ScKO mice could restore ceramide levels and rescue the delayed myelin degradation phenotype caused by Igfbp2 deficiency, we injected adeno-associated virus (AAV) carrying Cers6 in the SN prior to the injury. Results showed that CERS6 overexpression significantly increased ceramide levels in KO mice, as revealed by lipidomics analysis (Fig. 6B). Western blot demonstrated a marked reduction in myelin-associated proteins MAG and MBP upon CERS6 overexpression in injured SNs of Igfbp2-ScKO mice, indicating an enhanced myelin breakdown (Fig. 6C–E). Immunofluorescence further confirmed that the delayed myelin degradation observed in Igfbp2-ScKO mice was rescued by CERS6 overexpression (Fig. 6F,G). These findings establish that Cers6 mediates the effects of IGFBP2 on the ceramide level and myelin degradation.

Figure 6.

CERS6 overexpression restores ceramide levels and myelin degradation in Igfbp2-ScKO mice. A, Relative mRNA expression levels of Igfbp2 and ceramide synthetase 1–6 in 3 d cut SNs from WT and Plp/CreER^T; Igfbp2^fl/fl mice by qPCR. Mean ± SEM (n = 3 independent experiments). ns, p ≥ 0.05, **p < 0.01; ***p < 0.0005; ****p < 0.0001; two-way ANOVA. B, Heatmap of the different expression levels of ceramides by lipid metabolism in intact and 3 d cut SNs from WT, Plp/CreER^T; Igfbp2^fl/fl mice and Plp/Cre ER^T; Igfbp2^fl/fl mice rescued by AAV-CMV-Cers6 injection. C, Myelin-associated protein levels examined by Western blot in 3 d cut-injured SNs after CERS6 rescue from Plp/CreER^T; Igfbp2^fl/fl mice. The expression levels of CERS6 were assessed using a GFP-specific antibody. GAPDH was used as a loading control. D, Grayscale analysis of MAG in C. Mean ± SD (n = 3 images from three independent experiments). ns, p ≥ 0.05; *p < 0.05; **p < 0.01; one-way ANOVA. E, Grayscale analysis of MBP in C. Mean ± SD (n = 3 images from three independent experiments). ns, p ≥ 0.05; *p < 0.05; ***p < 0.0005; one-way ANOVA. F, Immunofluorescent staining of cross sections of 3 d cut-injured SNs after CERS6 rescue from Plp/CreER^T; Igfbp2^fl/fl mice was carried out using antibodies against MBP, cell nuclei (DAPI, blue). Scale bar, 10 μm. G, Quantification of intact myelin sheath of (G). Mean ± SD (n = 4 images from three independent experiments); ns, p ≥ 0.05; ****p < 0.0001; one-way ANOVA.

IGFBP2 and ceramides promote myelin breakdown and ON regeneration

Having established the role of Schwann cell-derived IGFBP2 in PNI, we next investigated the potential function of IGFBP2 in the CNS using a model of ONs crush injury. Given that the expression of IGFBP2 remains constant following ON injury, we utilized AAV carrying IGFBP2 (AAV-CMV-Igfbp2) and administered it via injection into the vitreous body of WT mice to achieve ectopic IGFBP2 expression (Fig. 7A). Western blot analysis confirmed increased degradation of MAG and MBP in the ONs of AAV-CMV-Igfbp2–injected mice compared with controls (Fig. 7B; Fig. S7A,B). Dissociated DRG neurons were cultured in vitro on myelin fragments from injured ONs with or without prior AAV-Igfbp2 injection (Fig. 7C). While DRG neurons are widely used in axon–myelin interaction studies due to their robust outgrowth, it should be noted that they are peripheral neurons and do not fully represent CNS neuronal behavior. Nevertheless, we observed a significant increase in axonal branching (Fig. 7D) and length (Fig. 7E) in DRG neurons cultured on myelin from AAV-Igfbp2–injected mice compared with controls. These findings suggest that IGFBP2 overexpression in the CNS significantly attenuates myelin-mediated inhibition of axonal regeneration.

Figure 7.

Activating the IGFBP2–ceramide pathway promotes CNS regeneration after ON injury. A, Experimental process of AAV-CMV-Igfbp2 injection in vitreous body in the WT mice. B, Western blot analysis was conducted to assess the protein expression levels in the ONs of WT mice. GAPDH was used as a loading control. C, Dissociated DRG were cultured in vitro using myelin fragments from 7 d crush-injured and 7 d crush-injured after AAV-Igfbp2 virus injection. DRGs were immunostained with antibody against Neurofilament. Scale bar, 100 μm. D, E, Statistical analysis of axonal branching (D) and length (E) in differentially treated DRG cultures. Mean ± SEM (n = 10 images from three independent experiments); *p < 0.05; **p < 0.01; ****p < 0.0001; one-way ANOVA. F, Representative transmission electron micrographs of the same condition as B. Top, Low magnification; scale bar, 5 μm. Bottom, High magnification of the boxed regions at the top; scale bars, 1 μm. G, Quantification of intact myelin sheath of F. Mean ± SD (n = 3 nerves from three independent experiments); ns, p ≥ 0.05; ****p < 0.0001; one-way ANOVA. H, Western blot analysis was conducted to assess the protein expression levels in the 7 d crush-injured ONs after inhibitors injection. β-Actin was used as a loading control. I, Representative transmission electron micrographs of 7 d crush-injured ONs after inhibitor injection. Top, Low magnification; scale bar, 5 μm. Bottom, High magnification of the boxed regions at the top; scale bars, 2 μm. J, Quantification of intact myelin sheath of I. Mean ± SD (n = 3 nerves from three independent experiments); ns, p ≥ 0.05; ****p < 0.0001; one-way ANOVA. K, Longitudinal sections of crush 4 week injured ONs with injecting CTB into the vitreous, after 28 d post AAV-Igfbp2 virus injection. Scale bar, 100 μm. L, The mean fluorescence intensity of CTB along the entire length of the imaged ONs was quantified, which were encompassing the entire longitudinal axis of the ONs. Mean ± SD (n = 3 nerves from three independent experiments), **p < 0.01; t test.

To examine myelin breakdown in injured ONs in vivo, we examined the myelin ultrastructure in injured ONs with or without prior AAV-Igfbp2 injection, transmission electron microscopy (Fig. 7F), and quantification of intact myelin sheaths (Fig. 7G) revealed significant decreased myelin integrity in AAV-Igfbp2–injected mice. Since ceramide supplementation improved myelin breakdown in Igfbp2-ScKO mice (Fig. 5), we investigated if modulating the ceramide pathway in the CNS could influence myelin breakdown. ONs with crush injury were treated with inhibitors targeting ceramide breakdown to increase the level of ceramide (Fig. 7H; Fig. S7C,D). Analysis of myelin integrity through electron microscopy (Fig. 7I,J) revealed significantly increased myelin breakdown in inhibitor-treated mice.

To evaluate ON regeneration in vivo, CTB was injected into the vitreous body of mice with ON crush injury that had previously received AAV-CMV-Igfbp2 or control injections. Quantification of CTB fluorescence, a marker for axonal regeneration, revealed a significant increase in ON regeneration in the distal ONs of mice with IGFBP2 overexpression compared with controls (Fig. 7K,L). The mean fluorescence intensity of CTB was measured across the entire ON.

Discussion

Our findings demonstrate the critical role of IGFBP2 in regulating myelin breakdown following PNI. In contrast to the PNS, CNS exhibits minimal IGFBP2 expression, potentially contributing to its limited myelin breakdown and axon regenerative capacity. These observations align with previous reports highlighting the differential response of the PNS and CNS to injury. Our data further suggest that ceramide, a class of sphingolipids, acts downstream of IGFBP2 to mediate myelin breakdown.

Exploring the contrasting expression patterns of IGFBP2 in the CNS and PNS postinjury presents an intriguing avenue for future investigation. IGFBP2 expression is tightly regulated by various factors and signaling pathways (Wei et al., 2021). During fasting, peroxisome proliferator-activated receptor directly targets the IGFBP2 promoter, while in tumor growth contexts, hypoxia-inducible factor 1 (HIF-1) and the tumor suppressor gene p53 modulate IGFBP2 expression (Clemmons et al., 1991; Feldser et al., 1999, p. 2; Grimberg et al., 2006). Additionally, CD147, highly expressed in tumors, regulates IGFBP2 expression along with the PTEN/PI3K/AKT signaling pathway, contributing to malignant melanoma cell apoptosis (Zhao et al., 2018). These regulatory mechanisms highlight IGFBP2's diverse roles in physiological processes and diseases (Wei et al., 2021). Understanding the underlying mechanisms responsible for the differential expression of IGFBP2 in the CNS and PNS could provide valuable insights into the divergent regenerative capacities of these two systems. Following injury in CNS and PNS, axonal regeneration is suppressed not only by differences in myelin degradation but also by additional inhibitory mechanisms, including glial scar formation and deleterious inflammatory cascades (F. Li et al., 2020; Tom et al., 2004). Another contributing factor to the distinct regenerative capacities between the PNS and CNS lies in the differences in axon degeneration mechanisms. Wallerian degeneration, which rapidly clears damaged axonal and myelin debris in the PNS, facilitates the timely recruitment of Schwann cells and macrophages essential for repair. In contrast, Wallerian degeneration in the CNS is slower and often incomplete, resulting in prolonged presence of inhibitory debris and delayed clearance, which impedes regeneration (Geisler et al., 2019; Coleman and Höke, 2020). These fundamental differences in axon breakdown kinetics may further contribute to the limited regenerative capacity observed in the CNS. Whether IGFBP2, beyond its demonstrated regulatory effects on myelin clearance, modulates these secondary inhibitory pathways remains unexplored and warrants further investigation.

Further investigation is required to elucidate the precise mechanisms by which IGFBP2 regulates ceramide metabolism in the PNS. One possibility is that IGFBP2 signaling directly modulates the activity of ceramide synthases. Alternatively, IGFBP2 may act indirectly through other signaling pathways, such as those involving neutral sphingomyelinase (nSMase; Jana et al., 2009). nSMase catalyzes the hydrolysis of sphingomyelin, another sphingolipid, into ceramide (Jana and Pahan, 2007). Interestingly, nSMase expression is increased during CNS demyelination, suggesting the potential role of this enzyme in the contrasting responses of the PNS and CNS to injury.

In this study, we found that Igfbp2 may influence PNS regeneration by regulating Cers6-mediated ceramide synthesis. One possible mechanism is that IGFBP2, potentially through modulation of the IGF signaling pathway, alters the transcription or stability of Cers6 (T. Li et al., 2020; Boughanem et al., 2021). The elevated ceramide levels may then promote myelin clearance (Gomez-Sanchez et al., 2015). However, these mechanisms remain to be fully elucidated and need further investigation in future studies.

IGFBP2 emerges as a multifaceted player in various physiological processes, from neural development to metabolic regulation (Yau et al., 2015a). In neurodevelopmental disorders, aberrant BMP signaling in astrocytes increases IGFBP2 expression, negatively impacting neuronal development (Caldwell et al., 2022). Conversely, after peripheral nerve injury, adipocyte-derived factors like leptin signal to Schwann cells, promoting nerve regeneration (Hedbacker et al., 2010). IGFBP2's therapeutic potential extends to diabetes management, as its expression correlates with leptin's antidiabetic effects (Hedbacker et al., 2010). In the brain, IGFBP2 not only influences growth and development but also regulates neuronal plasticity, impacting higher cognitive functions such as spatial learning and memory (Khan et al., 2019). Additionally, IGFBP2 promotes neural stem cell proliferation and maintenance, with its deficiency impairing cell cycle progression and Notch pathway gene expression in neural stem cells (Shen et al., 2019). Together, IGFBP2 acts as a link between neural development, metabolic regulation, and adipocyte signaling.

Notably, the role of IGFBP2 appears highly context-dependent. In models of spinal cord injury and hypoxic–ischemic brain injury, upregulated IGFBP2 expression has been shown to reduce neuronal loss and promote axonal regeneration (Fletcher et al., 2013). In contrast, in autoimmune demyelination models such as EAE, IGFBP2 has been associated with enhanced inflammation and demyelination, possibly via activation of NF-κB signaling (Zhang et al., 2023). The diverse functions highlight a complex regulatory network in which IGFBP2 may exert both protective and detrimental effects depending on the cell type, tissue context, and disease model.

The mechanism proposed in our current study—that IGFBP2 promotes myelin degradation through modulation of ceramide metabolism—represents a pathway distinct from previously reported proinflammatory roles. This mechanistic difference may reflect the specific involvement of Schwann cells and lipid metabolic pathways in the PNS, as opposed to astrocyte- or immune-mediated inflammatory responses in the CNS. Further comparative studies are necessary to reconcile these differing roles and clarify the full spectrum of IGFBP2's molecular actions.

Lipids play a crucial role in the development, function, and pathology of the nervous system (Chrast et al., 2011; McKerracher and Rosen, 2015; Schmitt et al., 2015). They are essential components of myelin and are involved in various cellular processes like signal transduction, energy storage, and membrane integrity (Poitelon et al., 2020; Barnes-Vélez et al., 2022). Ceramide, a sphingolipid, plays a crucial role in various aspects of neuronal function and health (Jana et al., 2009; Podbielska et al., 2023). The levels of ceramide and metabolism are intricately linked to neurological disorders (Posse de Chaves, 2006; van Echten-Deckert and Herget, 2006). Elevated levels of ceramide have been observed in the blood plasma of individuals with depression, while disruptions in ceramide metabolism are associated with schizophrenia and Parkinson's disease (Young and Geyer, 2015; Xing et al., 2016; Brodowicz et al., 2018). Additionally, abnormalities in ceramide metabolism are evident in the early stages of Alzheimer's disease, with increased activity of acid sphingomyelinase contributing to memory impairment (Cutler et al., 2004; He et al., 2010; Brodowicz et al., 2018). Moreover, ceramide affects axonal branching and stability to influence neuronal growth and survival (Schwarz et al., 1995). Furthermore, sphingolipids are implicated in myelin formation and maintenance, with sphingomyelin inducing various cellular behaviors depending on cell type and receptor interactions (Dasgupta and Ray, 2019). In the PNS, mTORC1 signaling pathway plays a pivotal role in controlling myelination through regulation of lipid biosynthesis via the SREBP pathway. Disruption of mTORC1 activity results in delayed myelination, abnormal lipid composition, and reduced nerve conduction velocity (Norrmén et al., 2014; Barnes-Vélez et al., 2022). In cerebrovascular diseases, elevated ceramide levels are closely linked to conditions such as stroke and cerebral small vessel disease (Yuan et al., 2023). Modulating sphingolipid metabolism, either through altering sphingomyelinase activity or targeting key enzymes like serine palmitoyltransferase, emerges as a novel therapeutic approach for preventing or treating cerebrovascular damage-related disorders (Dasgupta and Ray, 2019). Building on the established roles of ceramide in neuronal function and myelin integrity, recent studies further elucidate the specific contributions of CERS6 to inflammatory and neurodegenerative processes. Exacerbated EAE in CERS6 knock-out mice and elevated C16-ceramide levels in MS patients link CERS6 to inflammation regulation via TNF-α and nitric oxide. Its inactivation also disrupts sphingolipid metabolism and causes behavioral abnormalities, highlighting its therapeutic potential for neuroinflammatory and neurodegenerative diseases (Ebel et al., 2013).

In conclusion, our study reveals the critical role of the IGFBP2–ceramide pathway in myelin breakdown following PNI. These findings provide new insights into the differential regenerative potential of the PNS and CNS and offer promising therapeutic targets for promoting CNS regeneration after injury.