Abstract
Neurodegenerative diseases of both the central and peripheral nervous system are characterized by selective neuronal vulnerability, i.e., pathology that affects particular types of neurons. While much of this cell type selectivity may be driven by intrinsic differences among the neuron subpopulations, neuron-extrinsic mechanisms such as the selective malfunction of glial support cells may also play a role. Recently, we identified a population of Schwann cells (SCs) expressing Adamtsl1, Cldn14, and Pmp2 (a.k.a. PMP2+ SCs) that preferentially myelinate large-caliber motor axons. PMP2+ SCs are decreased in both amyotrophic lateral sclerosis (ALS) model mice and ALS patient nerves. Thus, PMP2+ SC dysfunction could contribute to motor-selective neuropathies. We engineered a tamoxifen-inducible Pmp2-CreERT2 mouse and expressed diphtheria toxin in PMP2+ SCs to assess the consequences of ablating this SC subtype in male and female mice. Loss of PMP2+ SCs led to significant loss of large-caliber motor axons with concomitant behavioral, electrophysiological, and ultrastructural defects. Subsequent withdrawal of tamoxifen restored both PMP2+ SCs and large-caliber motor axons and improved behavioral and electrophysiological readouts. Together, our findings highlight that the survival of large-caliber motor axons relies on PMP2+ SCs, demonstrating that malfunction of a specific SC subtype can lead to selective neuronal vulnerability.
Significance Statement
A hallmark of neurodegenerative disease is the differential vulnerability of neuron subtypes. While differences between neurons explain some differential sensitivity of neuronal subtypes, neuron-extrinsic mechanisms likely also contribute to selective neuronal vulnerability. Building on the recent identification of genetically distinct subtypes of myelinating Schwann cells (SCs), we test the hypothesis that SC subtypes support distinct classes of peripheral axons. To examine this, we ablated the PMP2+ subclass of myelinating SCs and found preferential loss of large-caliber motor axons. These findings demonstrate that disrupting a specific SC subtype results in selective axonal vulnerability, highlighting the importance of considering neuron-extrinsic mechanisms when dissecting selective neuronal vulnerability in neurodegenerative disorders.
Introduction
Selective vulnerability of neuronal subtypes is a hallmark of many neurodegenerative diseases (Saxena and Caroni, 2011; Mattsson et al., 2016; Fu et al., 2018; Ragagnin et al., 2019). For example, Alzheimer's disease selectively targets pyramidal neurons, Parkinson's disease selectively disrupts dopaminergic neurons, Huntington's disease features selective decay of medium spiny neurons, and amyotrophic lateral sclerosis (ALS) leads to selective atrophy of upper and lower motor neurons (Saxena and Caroni, 2011; Mattsson et al., 2016; Fu et al., 2018; Ragagnin et al., 2019). Similarly, neurons in the peripheral nervous system are selectively sensitive to different pathologies, giving rise to sensory, motor, or mixed neuropathies (Lehmann et al., 2020). Much research has been conducted seeking to explain this phenomenon by identifying intrinsic differences between neuron types, but the governing mechanisms remain poorly understood (Saxena and Caroni, 2011; Mattsson et al., 2016; Fu et al., 2018; Ragagnin et al., 2019). Notably, neurons do not function in isolation, and in the peripheral nervous system, axons are intimately associated with Schwann cells (SCs), which nourish and physically support them (Jessen et al., 2015a; Jessen and Mirsky, 2019). Hence, it is plausible that differential neuronal susceptibility may partially arise from neuron-extrinsic mechanisms. For example, SC-specific loss of either the broadly expressed tumor suppressor liver kinase B1 (Beirowski et al., 2014) or insulin signaling (Hackett et al., 2020) triggers selective degeneration of sensory axons. Similarly, motor neuropathies arise from SC-specific disruption of lactate metabolism (Bloom et al., 2022; Deck et al., 2022), and mixed neuropathies are caused by the SC-specific disturbance of nicotinamide adenine dinucleotide (NAD+) metabolism (Sasaki et al., 2018), fatty acid synthesis (Montani et al., 2018), and mitochondrial function (Viader et al., 2011; Fünfschilling et al., 2012). These findings collectively highlight that distinct axon types have specific metabolic needs and that the differential susceptibility of neurons to peripheral pathologies could also involve the contribution of their diverse supporting glia. However, this has been difficult to study due to limited knowledge of SC subtypes.
Recently, we performed cellular profiling of mouse peripheral nerves using single nuclei RNAseq and discovered several subpopulations of myelinating and nonmyelinating SCs (Yim et al., 2022). One of the myelinating SC subtypes, defined by expression of Adamtsl1, Cldn14, and Pmp2 (hereafter referred to as PMP2+ SCs), forms thick myelin sheaths preferentially around large-caliber axons. The PMP2+ SCs selectively myelinate motor axons and are, therefore, abundant in the predominately motor femoral nerve and rare in the primarily sensory sural nerve. Moreover, we observe a significant decrease in PMP2+ SC numbers in the peripheral nerves of both ALS model mice and ALS patients. Altogether, these findings suggest that malfunction of this important SC subpopulation could contribute to motor axon degeneration.
Identification of the PMP2+ SC subtype enables testing the glia-centric hypothesis that the loss of a distinct SC subpopulation drives the development of a particular neuropathy. Here, we hypothesize that malfunction of the motor axon-selective PMP2+ SCs will result in the loss or dysfunction of a specific group of motor axons with concomitant motor behavioral defects. To test this, we generated a novel mouse strain that contains a tamoxifen-inducible Cre–3′P2A cassette (CreERT2–3′P2A) under the Pmp2 promoter, granting us genetic control over PMP2+ SCs. We made use of this line to selectively ablate PMP2+ SCs by driving diphtheria toxin subunit A (DTA) expression. Loss of PMP2+ SCs induces severe neuropathy with preferential loss of large-caliber motor axons and behavioral, electrophysiological, and ultrastructural defects. Ablation of PMP2+ SCs stimulates regenerative processes in the peripheral nerves that ultimately lead to the restoration of PMP2+ SCs and large motor axons, as well as improvement in behavioral and electrophysiological readouts. These findings demonstrate that the survival of a specific group of motor axons relies on the proper functioning of the PMP2+ SCs and strongly supports the hypothesis that differential neuronal susceptibility can arise from neuron-extrinsic mechanisms, such as inherent differences among distinct SC types.
Materials and Methods
Animals
All experiments were performed in accordance with the protocols of the Institutional Animal Care and Use Committee of Washington University in St. Louis and the guidelines for the Care and Use of Laboratory Animals of the National Institutes of Health. Mice were housed on a 12 h light/dark cycle with less than five mice per cage and with water and food available at all times. Male and female mice were used for all experiments.
We engineered the Pmp2-CreERT2 mouse line with the Genome Engineering & Stem Cell (GESC) Center at Washington University in St. Louis by knocking in a tamoxifen-inducible Cre cassette (CreERT2) following the ATG codon of the Pmp2 gene. The construct also contains a 3′P2A sequence to avoid disrupting PMP2 protein expression (Fig. 1A).
Pmp2-CreERT2 mouse line selectively targets PMP2+ SCs. A, A tamoxifen-inducible Cre cassette (CreERT2) was knocked in following the ATG codon of Pmp2 with a 3′P2A sequence to preserve endogenous PMP2 protein expression from the locus. B, Quantification of tdTomato+ cells in the femoral (n = 3) and sural (n = 3) nerves of the Pmp2-CreERT2 mouse. From the left, The percentage of tdTomato+ cells that express PMP2, the percentage of PMP2+ cells that express tdTomato, the percentage of axons myelinated by PMP2+ SCs, the percentage of axons myelinated by PMP2+/tdTomato+ SCs. C, tdTomato and PMP expression patterns in femoral and sural nerves from Pmp2-CreERT2(+):ROSA26-tdTomato(+) mice.
The tdTomato mice [B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J; strain number, 007909; Jackson Laboratory) were crossed to the Pmp2-CreERT2 mice to generate Pmp2-CreERT2(+/−); tdTomato (+/−) progeny that were then used to validate targeting of the desired cell population with the Pmp2-CreERT2 mice. Similarly, DTA mice [B6.129P2-Gt(ROSA)26Sortm1(DTA)Lky/J; strain number, 009669; Jackson Laboratory) were crossed to the Pmp2-CreERT2 mice to generate Pmp2-CreERT2(−/−); DTA(+/−) (control) and Pmp2-CreERT2(+/−); DTA(+/−) (experimental) progeny. When the progeny mice reached 6–8 weeks of age, both groups were injected with 100 mg/kg tamoxifen dissolved in corn oil. The injections were performed intraperitoneally on Tuesday and Friday (semiweekly) for the course of 12 weeks. After 12 weeks, mice were either killed or left alive for another 12 weeks and then killed.
Inverted screen test
One mouse at a time was placed on a wire mesh screen. The screen then was inverted 180° so that the mouse was hanging upside down, and the latency to fall was recorded. If the mouse was still hanging on the screen at 120 s, they were taken off the screen, and 120 s was recorded. Each mouse was tested three times with 5 min break periods in between. The average of the three trials was taken.
Nerve electrophysiology
Compound muscle action potentials (CMAPs) were acquired using a Viking Quest electromyography device (Nicolet) as previously described (Beirowski et al., 2011). Briefly, mice were anesthetized, and then a stimulating electrode was placed in the sciatic notch or ankle, and a recording electrode was placed in the foot. Supramaximal stimulation was used.
Nerve structural analysis
Femoral and sural nerves were processed as previously described (Geisler et al., 2016). Briefly, nerves were fixed in 3% glutaraldehyde in 0.5 ml PBS overnight at 4°C, washed, and stained with 1% osmium tetroxide (Sigma-Aldrich) overnight at 4°C. Nerves were washed and dehydrated in a serial gradient of 50–100% ethanol. Nerves were then incubated in 50% propylene oxide/50% ethanol and then 100% propylene oxide. Following that, nerves were incubated in Araldite resin/propylene oxide solutions in 50:50, 70:30, and 90:10 ratios for 24 h and subsequently embedded in 100% Araldite resin solution (Araldite, DDSA, DMP30; 12:9:1; Electron Microscopy Sciences) and baked at 60°C overnight.
Resin-embedded nerves were sectioned at 400–600 nm using a Leica EM UC7 Ultramicrotome and placed on microscopy slides. Slides were stained with 1% toluidine blue solution (1% toluidine blue, 2% borax), washed with acetone and xylene, and then mounted in Cytoseal XTL (Thermo Fisher Scientific). Whole nerves were imaged with a Nikon 80i inverted microscope and MetaMorph software at 20× magnification, and three random fields were obtained using a 100× oil-immersion lens. Axon diameter and g-ratio calculations were performed with MyelTracer (Kaiser et al., 2021).
For transmission electron microscopy, femoral nerve sections were cut by the Washington University Core for Cellular Imaging (WUCCI) and placed on grids for imaging. A JEOL JEM-1400 Plus transmission electron microscope was used for imaging, and 6–10 images were taken per sample.
Immunohistochemistry and imaging
Mouse nerve samples were harvested and immediately fixed in 4% PFA for 1 h at room temperature, transferred to 30% sucrose overnight, and embedded into optimal cutting temperature compound. Eight micron sections were cut and collected onto slides. Next, the slides were fixed with ice-cold acetone for 10 min at −20°C. After air-drying the slides for ∼10 min, they were washed with PBS three three times (5 min each) and blocked in 10% normal goat or donkey serum for 1 h at room temperature. Next, the primary antibodies were suspended in the blocking buffer to stain the slides overnight at 4°C. The following antibodies were used: PMP2 (Proteintech, 12717-1-AP, dilution 1:100), myelin basic protein (MBP; Millipore Sigma, MAB386, dilution 1:100), β-tubulin (β-TUB; Aves Labs, TUJ, dilution 1:100), choline acetyltransferase (ChAT; Millipore Sigma, AB144P, dilution 1:100), cJUN (Cell Signaling Technology, 9165, dilution 1:500), Iba1 (Fujifilm, 019-19741, dilution 1:400), and CD68 (Bio-Rad, MCA1957GA, dilution 1:200). On the next day, the slides were washed three times with 0.03% Triton X-100 in PBS (5 min each) and stained for 1 h at room temperature with appropriate secondary antibodies bought from Invitrogen. Next, the slides were washed three times with 0.03% Triton X-100 in PBS (5 min each) and covered with coverslips using HardSet Antifade Mounting Medium (VECTASHIELD, H-1500-10). Imaging was performed with Nikon 80i inverted microscope and MetaMorph software, and quantification was performed using ImageJ (Schneider et al., 2012).
Statistical analysis
Data are reported as mean ± 95% confidence interval. Statistics were calculated with the aid of Statannotations (Charlier et al., 2023) and Pingouin (Vallat, 2018). For the between-group comparisons, one-way ANOVA, two-way ANOVA, or unpaired t tests were used. For the within-group comparisons, a paired t test was used. Two-tailed significance tests were used with p < 0.05 considered statistically significant. n represents the number of animals used with the exception of the nerve structural analysis, where n represents the number of quantified axons. For all studies, n is reported in the figure legend. All data visualizations and final figures were made with Seaborn (Waskom, 2021).
Results
Generation of the Pmp2-CreERT2 mouse line
We identified a subtype of myelinating SCs with a distinct gene expression pattern defined by the expression of Pmp2, Cldn14, and Adamtsl1. This PMP2+ SC subset primarily myelinates motor axons of large caliber (Yim et al., 2022). To probe the function of this SC subset, we first sought to gain genetic control over the PMP2+ SCs. To accomplish this, we knocked in a tamoxifen-inducible Cre–3′P2A cassette (CreERT2-3′P2A) at the Pmp2 gene to engineer a transgenic Pmp2-CreERT2 mouse without disrupting PMP2 protein expression (Fig. 1A). To assess the expression of Cre in the Pmp2-CreERT2 mouse, we crossed it to ROSA26-tdTomato Cre reporter mice (Madisen et al., 2010). When the progeny reached 6–8 weeks of age, we administered tamoxifen intraperitoneally twice a week for 12 weeks to activate Cre and excise a floxed-STOP codon leading to the expression of fluorescent tdTomato in Cre+ cells. This regimen was chosen to match the functional studies described below. To test whether the Pmp2-CreERT2 line expresses in bona fide Pmp2-expressing cells, we harvested and analyzed the femoral and sural nerves—two nerves with widely disparate axon compositions. In the femoral nerve, ∼60% of axons are motor axons compared with only ∼12% of axons in the mainly sensory sural nerve (Yim et al., 2022). Additionally, we previously demonstrated that Pmp2+ SCs preferentially associate with motor axons as Pmp2+ SCs myelinate 44.4% of femoral axons and only 5.5% of sural axons (Yim et al., 2022).
We stained for PMP2 protein and assessed the fraction of tdTomato+ cells that coexpress PMP2. In the femoral nerve, 95% of tdTomato+ cells express PMP2 protein, demonstrating excellent overlap between the Cre line and the endogenous protein (Fig. 1B,C). In the sural nerve, we observed colocalization of PMP2 protein in 76% of tdTomato-expressing cells. Conversely, 55 and 35% of femoral and sural PMP2+ SCs, respectively, coexpress tdTomato (Fig. 1B,C). The divergences we detected between PMP2 and tdTomato could indicate Cre expression in some non-PMP2+ cells and vice versa, but since tdTomato is a soluble protein that fills the cell, whereas PMP2 protein selectively associates with myelin, much of the discrepancy may arise from nerve cross sections that include tdTomato+ cell bodies without their associated myelin and the relative absence of soluble tdTomato in densely packed myelin (Fig. 1B,C). In total, 34% of axons contact tdTomato+ cells in the femoral and 11% of axons in the sural nerve (Fig. 1B,C). Taken together, these findings demonstrate that the Pmp2-CreERT2 mouse line expresses tamoxifen-inducible Cre in a large subset of PMP2+ SCs, thereby enabling the study of these cells in the peripheral nerve.
DTA-mediated death of PMP2+ SCs leads to motor axon loss
To address the role of PMP2+ SCs, we crossed the Pmp2-CreERT2 mice to DTA mice (Voehringer et al., 2008) to induce Cre-dependent–selective cell death of PMP2+ cells. Prior studies that investigated SCs by driving DTA in all myelinating cells using Plp-CreERT suggested that long-term treatment with tamoxifen is required to observe demyelination of peripheral nerves (Traka et al., 2010, 2016). We therefore treated 6–8-week-old Pmp2-CreERT2(+):DTA(+) mice and Pmp2-CreERT2(−):DTA(+) controls with tamoxifen twice a week for 12 weeks. We harvested femoral and sural nerves to perform morphological analysis and assess the efficacy of PMP2+ SC ablation. We stained nerve sections for PMP2 protein, the SC marker MBP, the axonal marker β-TUB, and the motor axon marker ChAT. We found that the number of the PMP2+ SCs in the Pmp2-CreERT2(+):DTA(+) femoral nerve was reduced by 55% when compared with the Pmp2-CreERT2(−):DTA(+) femoral nerve (p < 0.001; Fig. 2A,B), which closely matches the percentage of PMP2+ SCs that expressed tdTomato in the validation experiment (55%; Fig. 1B). Similarly, we observed a 52% reduction in the number of the PMP2+ SCs in the Pmp2-CreERT2(+):DTA(+) sural nerve (p < 0.001; Fig. 2A,C), which is slightly more than the percentage of PMP2+ SCs that expressed tdTomato in the validation experiment (35%; Fig. 1B). We then assessed the consequences of PMP2+ SC ablation, finding a significant decrease in the numbers of myelinating SCs (41% drop; p < 0.001), total axons (35% drop; p < 0.001), and motor axons (48% drop; p < 0.001) in the Pmp2-CreERT2(+):DTA(+) femoral nerve (Fig. 2A,B). In the Pmp2-CreERT2(+):DTA(+) sural nerve, there were also trends toward fewer myelinating SCs (10% drop; p > 0.05) and axons (8% drop; p > 0.05) and significantly fewer motor axons (33% drop; p < 0.01), consistent with the tendency of PMP2+ SCs to myelinate motor axons (Fig. 2A,C). Notably, the total axon loss observed across the femoral and sural nerves (35 and 8%; Fig. 2B,C) is in excellent agreement with the percentage of axons that were in contact with tdTomato+ cells (34 and 10%; Fig. 1B). These findings indicate that PMP2+ SCs are required for the survival of the axons they myelinate, and this effect is most pronounced in the motor axon-rich femoral nerve where PMP2+ SCs are most abundant.
Ablation of PMP2+ SCs leads to motor axon loss. A, Representative images of femoral and sural nerves from tamoxifen-treated Pmp2-CreERT2(+):DTA(+) and Pmp2-CreERT2(−):DTA(+) mice stained for PMP2, MBP, β-TUB, and ChAT to visualize PMP2+ SCs, myelinating SCs, all axons, and motor axons, respectively. B,C, Quantification of PMP2+ cells, myelinating SCs (MBP+), axons (β-TUB+), and motor axons (ChAT+) in the femoral (B) and sural (C) nerves of the Pmp2-CreERT2(−):DTA(+) (n = 6, gray) and Pmp2-CreERT2(+):DTA(+) (n = 9, red) mice after 12 weeks of tamoxifen treatment. All data are presented as mean ± 95% confidence interval. Statistical significance determined by a Student's unpaired two-tailed t test. Not significant (ns); **p < 0.01; ****p < 0.0001.
Large axons are selectively sensitive to PMP2+ SC ablation
PMP2+ SCs preferentially ensheath large axons (Yim et al., 2022). Having observed robust degeneration of axons in the femoral nerve upon PMP2+ SC ablation, we asked whether there is a selective loss of large-caliber axons. We performed morphometric analysis on femoral and sural nerves to determine the distributions of axonal diameters in the surviving axons. In the control femoral nerve, we observed axons ranging from 1 to 13 µm in diameter (Fig. 3A,B). However, in PMP2+ cell-ablated femoral nerves, axons between 7 and 9 µm in diameter were reduced in number, and those larger than 9 µm were nearly absent (Fig. 3A,B). We also found that in the PMP2+ cell-ablated samples, the larger axons had unusually thin myelin sheaths (Fig. 3C). This suggests that myelin is being lost from the large-caliber axons and/or that large-caliber axons are regenerating and being remyelinated, which leads to thinner myelin (Franklin and Goldman, 2015). In contrast, there is no significant difference in axon diameter distributions or myelin thickness (g-ratio) between PMP2+ cell-ablated sural nerves and their control counterparts (Fig. 3A,D,E). However, even in basal conditions, there are few axons larger than 7 µm in the sural nerve (Fig. 3E), which correlates with the rarity of the PMP2+ SCs in the sural nerve. These results demonstrate the selective requirement of PMP2+ SCs for large axon survival.
Large axons are selectively sensitive to PMP2+ SC ablation. A, Representative images of femoral and sural nerves from tamoxifen-treated Pmp2-CreERT2(−):DTA(+) and Pmp2-CreERT2(+):DTA(+) mice stained with toluidine blue. B,D, Histograms presenting axon diameter distributions of femoral (B) and sural (D) nerves of the Pmp2-CreERT2(−):DTA(+) (n axons = 862, gray) and the Pmp2-CreERT2(+):DTA(+) (n axons = 984, red) mice. Axons larger than 7 µm are selectively depleted from the femoral nerve of the Pmp2-CreERT2(+):DTA(+) mice (red). C,E, A scatterplot of the g-ratio, a metric of myelin thickness (axon diameter / axon + myelin diameter) as a function of axon diameter for femoral (C) and sural (E) nerves of the Pmp2-CreERT2(−):DTA(+) (n axons = 862, gray) and the Pmp2-CreERT2(+):DTA(+) (n axons = 984, red) mice. The larger axons in the Pmp2-CreERT2(+):DTA(+) (red) have unusually thin myelin sheaths, indicating myelin decomposition from the large-caliber axons.
PMP2+ SC loss induces demyelination
To determine the effects of PMP2+ SC ablation on motor behavior, we used the inverted screen assay to test the strength of Pmp2-CreERT2(+):DTA(+) mice treated with tamoxifen beginning from 6 to 8 weeks of age. While control mice maintained function, experimental mice showed a gradual decrease in the amount of time they could hang on to the wire mesh, plateauing between 8 and 12 weeks of treatment (Fig. 4A). At this time, we performed electrophysiological analysis of motor nerve function by assaying CMAPs. Firstly, we found that while the electrical impulse in the Pmp2-CreERT2(−):DTA(+) mice arrived at the muscle in a discrete waveform, the impulse triggered in the Pmp2-CreERT2(+):DTA(+) mice arrived at the muscle as a temporally dispersed complex waveform which is indicative of segmental demyelination (Fig. 4B,C). Secondly, the latency was increased and nerve conduction velocity decreased in the Pmp2-CreERT2(+):DTA(+) mice, demonstrating slowed action potential propagation (Fig. 4D,E,F). Finally, the CMAP amplitude was reduced in the Pmp2-CreERT2(+):DTA(+) mice, suggesting a decreased number of properly firing axons (Fig. 4G,H). Each of these nerve conduction abnormalities is typical for demyelinating diseases (Van Asseldonk et al., 2003; Susuki, 2010; Ahn et al., 2018), demonstrating that ablating PMP2+ SCs induces a demyelinating neuropathy.
PMP2+ SC loss induces demyelination. A, Average latency time to fall from an inverted screen (max. 120 s) for Pmp2-CreERT2(−):DTA(+) (n = 18, gray) and Pmp2-CreERT2(+):DTA(+) (n = 22, red) mice. B,C, Representative traces of CMAP recordings from the ankle (top trace) and sciatic notch (bottom trace) stimulation of Pmp2-CreERT2(−):DTA(+) (B) and Pmp2-CreERT2(+):DTA(+) (C) mice after 12 weeks of tamoxifen. D–H, CMAP latency (D,E), motor nerve conduction velocity (F), and CMAP amplitude (G,H) of the Pmp2-CreERT2(−):DTA(+) (n = 18, gray) and Pmp2-CreERT2(+):DTA(+) (n = 22, red) mice measured after stimulation at the ankle (D,F,G) and sciatic notch (E,F,H) after 12 weeks of tamoxifen. All data are presented as mean ± 95% confidence interval. Statistical significance determined by a two-way ANOVA (A) and Student's unpaired two-tailed t test (D–H). Not significant (ns); ****p < 0.0001.
Repair SCs and activated macrophages respond to PMP2+ SC ablation
Upon peripheral nerve damage, SCs transition from homeostatic to repair SCs, losing their differentiated features while they downregulate myelin-related genes and upregulate genes that promote axon survival and regeneration (Arthur-Farraj et al., 2012; Jessen et al., 2015a; Gomez-Sanchez et al., 2017; Jessen and Mirsky, 2019). In addition, repair SCs release cytokines to attract macrophages to the injury site to help remove axonal and myelin debris (Martini et al., 2008; Rotshenker, 2011; Cattin et al., 2015; Jessen et al., 2015a; Jessen and Mirsky, 2019). To explore whether the ablation of the PMP2+ SCs leads to changes typical for nerve injury, we stained femoral and sural nerve sections for c-Jun, a transcription factor that regulates SC dedifferentiation, IBA1, a pan-macrophage marker, and the glycoprotein CD68, expressed in activated macrophages. In femoral nerves from tamoxifen-treated Pmp2-CreERT2(+):DTA(+) mice, repair SCs, total macrophages, and activated macrophages increased 16.2-fold, 8.1-fold, and 8.9-fold, respectively, compared with the controls (Fig. 5A,B). Similarly, we observed elevated numbers of repair SCs (10.6-fold), total macrophages (4.5-fold), and activated macrophages (3.6-fold) in the sural nerve following PMP2 SC ablation (Fig. 5A,C). Note that while the fold change for all markers was similar in the femoral and sural nerves, the absolute change was much greater in the femoral nerve likely because axon loss was much greater there than that in the sural nerve (Fig. 5B,C). These findings show that PMP2+ SC ablation induces a robust injury response from surrounding SCs and macrophages and that this response is more pronounced in the primarily motor femoral nerve.
Repair SCs and activated macrophages respond to PMP2+ SC ablation. A, Representative images of femoral and sural nerves from tamoxifen-treated Pmp2-CreERT2(+):DTA(+) and Pmp2-CreERT2(−):DTA(+) mice stained for cJUN, CD68, and IBA1 to visualize repair SCs, activated macrophages, and total macrophages, respectively. B,C, Quantification of repair SCs (cJUN+), activated macrophages (CD68+), and total macrophages (IBA1+) in the femoral (B) and sural (C) nerves of the Pmp2-CreERT2(−):DTA(+) (n = 6, gray) and Pmp2-CreERT2(+):DTA(+) (n = 9, red) mice after 12 weeks of tamoxifen treatment. All data are presented as mean ± 95% confidence interval. Statistical significance determined by a Student's unpaired two-tailed t test. Not significant (ns); *p < 0.05; **p < 0.01; ****p < 0.0001.
Ultrastructural consequences of PMP2+ SC ablation
We performed ultrastructural analysis on femoral nerves from control and experimental animals following 12 weeks of tamoxifen treatment to ablate PMP2+ SCs. Control nerves display thickly myelinated large-caliber axons, thinly myelinated smaller caliber axons, and unmyelinated axons within Remak bundles (Fig. 6A, arrowhead). In experimental nerves, thickly myelinated large- (Fig. 6B, 1) and small-caliber axons and Remak bundles (Fig. 6B, arrowhead) are still observed, but numerous abnormalities occur that are consistent with a demyelinating hypertrophic neuropathy (Fig. 6B–H) with superimposed axonal degeneration. Upon PMP2+ SC ablation, we observed a variety of striking ultrastructural alterations. Thinly myelinated large-caliber axons show multilayered concentric proliferation of SC processes designated “onion bulbs” (Fig. 6B, 2), whereas thickly myelinated axons (Fig. 6B, 1) typically lack such processes. Onion bulbs are indicative of rounds of demyelination and remyelination. Fewer compact Remak bundles are observed in experimental animals. Instead, unmyelinated axons form loose collections of axons/SC processes, with individual axons surrounded by delicate SC processes (“singletons”; Fig. 6B, 3). In other cases, bundles of collagen are surrounded by SC processes to form “collagen pockets.” These abnormalities do not necessarily imply that PMP2+ SC ablation directly influences nonmyelinating SCs, as they are nonspecific changes that occur routinely in human neuropathies characterized by demyelination and/or axonal degeneration (Pestronk et al., 2023). Numerous examples of macrophages containing debris are present within experimental nerves (Fig. 6C, arrow). A large axon is observed, completely demyelinated but still intact (Fig. 6D) with the myelin debris confined to its surrounding SC(s). In a later stage of that phagocytic process, a naked axon is observed without surrounding myelin debris (Fig. 6E, 1), the debris is removed by endoneurial macrophages, possibly the adjacent one (Fig. 6E, 3). The underlying pathological process appears to preferentially involve large myelinated axons, mainly sparing small myelinated axons (Fig. 6E, 4). Thinly myelinated axons are observed which appear to be regenerating their myelin sheaths (Fig. 6E, 2). A very large axon is completely demyelinated, and myelin debris from the demyelination process has been removed—note the adjacent surrounding supernumerary SC processes (Fig. 6F). Residual myelin debris following axonal degeneration, i.e., myelin within a SC, absent an axon, and regenerating axon clusters in which the axon has degenerated and regenerated are also present (Fig. 6G,H). These ultrastructural abnormalities are consistent with our light-level analysis and support the conclusion that PMP2+ SC ablation preferentially leads to demyelination and loss/regeneration of large-caliber axons.
PMP2+ SC ablation leads to demyelination of large-caliber axons. A,B, Representative electron microscopy images of femoral nerves from (A) tamoxifen-treated control Pmp2-CreERT2(−):DTA(+) and (B) experimental Pmp2-CreERT2(+):DTA(+) mice. Arrowheads point to Remak bundles with normal morphology. Arrows point to (1) a thickly myelinated large-caliber axon, (2) an onion bulb, (3) a singleton, and (4) a collagen pocket (scale bar, 6 µm). (C–H) Further pathological features of PMP2+ SC ablation in femoral nerves. C, The arrow points to a macrophage containing phagocytosed debris (scale bar, 2 µm). D, A SC containing phagocyted debris (scale bar, 1 µm). E, Arrows point to (1) a demyelinated large-caliber axon, (2) a thinly myelinated large-caliber axon, (3) a macrophage containing phagocyted debris, and (4) a fully myelinated small-caliber axon (scale bar, 2 µm). F, Demyelinated large-caliber axon (scale bar, 500 nm). G, Regenerating cluster of axons (scale bar, 6 µm). H, Degenerated myelinated axon within a SC (scale bar, 1 µm).
PMP2+ SCs and motor axons regenerate after PMP2+ SC ablation
Repair SCs and macrophages are primary architects of peripheral nerve regeneration, creating an environment that promotes axon regrowth and guides axons back to their target tissue (Cattin et al., 2015; Jessen et al., 2015a; Jessen and Mirsky, 2019; Stierli et al., 2019). After axons regenerate, they instruct repair SCs to redifferentiate into mature myelinating or nonmyelinating SCs (Jessen and Mirsky, 2019). Having observed that the PMP2+ SC loss causes a robust increase in repair SC and macrophage numbers, we wondered whether these cells promote effective axon regeneration. Therefore, after 12 weeks of tamoxifen, we aged experimental and control mice for an additional 12 weeks in the absence of tamoxifen treatment. We then stained femoral nerves for PMP2, MBP, β-TUB, and ChAT to evaluate axon and myelin regeneration. We observed robust evidence of regeneration and repair, with similar numbers of PMP2+ SCs, myelinating SCs, total axons, and motor axons in the femoral nerves of the aged Pmp2-CreERT2(+):DTA(+) mice and age-matched CreERT2(−):DTA(+) control mice (Fig. 7A,B). While the number of PMP2+ SC is restored to WT levels, the staining for PMP2 protein is noticeably fainter in the treated animals, likely reflecting the thinner myelin ensheathing regenerating axons. The axon diameter distributions were also very similar between control and experimental femoral nerves (Fig. 8A,B). However, the myelin thickness (g-ratio) of large axons in experimental nerves was decreased compared with controls (Fig. 8C), a hallmark of regenerated axons (Franklin and Goldman, 2015).
PMP2+ SCs and motor axons regenerate after PMP2+ SC ablation. A, Representative images of femoral nerves from Pmp2-CreERT2(+):DTA(+) and Pmp2-CreERT2(−):DTA(+) mice immediately after 12 weeks of tamoxifen or after 12 additional weeks of recovery, stained for PMP2, MBP, β-TUB, and ChAT to visualize PMP2+ SCs, myelinating SCs, all axons, and motor axons, respectively. B, Quantification of PMP2+ cells, myelinating SCs (MBP+), axons (β-TUB+), and motor axons (ChAT+) in the femoral nerves of Pmp2-CreERT2(−):DTA(+) mice after 12 weeks of tamoxifen followed by 12 weeks of recovery (n = 7, gray), Pmp2-CreERT2(+):DTA(+) mice after 12 weeks of tamoxifen (n = 9, red), and Pmp2-CreERT2(+):DTA(+) mice after 12 weeks of tamoxifen and 12 weeks of recovery (n = 7, blue). All data are presented as mean ± 95% confidence interval. Statistical significance determined by one-way ANOVA. Not significant (ns); ****p < 0.0001.
Large axons are replenished after PMP2+ SC ablation. A, Representative images of femoral nerves harvested from tamoxifen-treated Pmp2-CreERT2(+):DTA(+) and Pmp2-CreERT2(−):DTA(+) mice immediately after 12 weeks of tamoxifen or after an additional 12 weeks of recovery stained with toluidine blue. B, Histogram presenting axon diameter distribution in the femoral nerves of the Pmp2-CreERT2(−):DTA(+) (n axons = 893, gray) and the Pmp2-CreERT2(+):DTA(+) mice (n axons = 989, blue) after 12 weeks of tamoxifen and 12 weeks of recovery. Axons larger than 7 µm regenerated in the femoral nerve of the Pmp2-CreERT2(+):DTA(+) mice (blue). C, Scatterplot of the g-ratio, a metric of myelin thickness (axon diameter / axon + myelin diameter) as a function of axon diameter for the femoral nerves of the Pmp2-CreERT2(−):DTA(+) (n axons = 893, gray) and the Pmp2-CreERT2(+):DTA(+) (n axons = 989, blue) mice after 12 weeks of tamoxifen and 12 weeks of recovery as a function of axon diameter. The larger axons in the Pmp2-CreERT2(+):DTA(+) femoral nerves (blue) have thinner myelin sheaths, a hallmark of regenerated axons.
Having observed potent regeneration of PMP2+ SCs and motor axons, we next assessed behavioral and electrophysiological function. Indeed, the Pmp2-CreERT2(+):DTA(+) mice became stronger 12 weeks after the withdrawal of tamoxifen, returning to control levels as assayed by the inverted screen test (Fig. 9A). Similarly, CMAP recordings (Fig. 9B–H) dramatically improved to nearly control levels. The waveform at the muscle became synchronized (Fig. 9B,C), while the latency to fire decreased and nerve conduction velocity increased (Fig. 9D,E,F). Moreover, the CMAP amplitude in the aged Pmp2-CreERT2(+):DTA(+) mice returned to control levels (Fig. 9G,H), consistent with the anatomical findings of regenerated motor axons (Figs. 7A,B, 8A,B). Collectively, these results demonstrate that following 12 weeks without tamoxifen, the PMP2+ SCs and their associated axons robustly regenerate and reform an effective neuronal circuit.
Experimental mice recover from demyelination caused by PMP2+ SC ablation. A, Average latency time to fall from an inverted screen (max. 120 s) for Pmp2-CreERT2(−):DTA(+) (n = 13, gray) and Pmp2-CreERT2(+):DTA(+) (n = 22, red) mice after 12 weeks of tamoxifen and 12 weeks of recovery. B,C, Representative traces of CMAP recordings from the ankle (top trace) and sciatic notch (bottom trace) stimulation of the same leg of the same Pmp2-CreERT2(+):DTA(+) mouse collected after 12 weeks of tamoxifen (B) and an additional 12 weeks of recovery (C). D–H, CMAP latency (D,E), motor nerve conduction velocity (F), and CMAP amplitude (G,H) of the Pmp2-CreERT2(−):DTA(+) (n = 13, gray) and Pmp2-CreERT2(+):DTA(+) (n = 19, red) mice measured after stimulation at the ankle (D,F,G) and sciatic notch (E,F,H) after 12 weeks of tamoxifen and 12 weeks of recovery. Statistical significance determined by a two-way ANOVA (A) and Student's unpaired two-tailed t tests (D–H). Not significant (ns); *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.
Discussion
Peripheral neuropathies constitute a spectrum of diseases that can involve motor, sensory, or mixed symptoms. It is proposed that this phenotypic diversity results from the selective vulnerability of neuron subtypes to different pathologies (Lehmann et al., 2020). However, manipulating SC metabolic pathways can also induce the degeneration of particular axon subtypes (Viader et al., 2011; Fünfschilling et al., 2012; Beirowski et al., 2014; Montani et al., 2018; Sasaki et al., 2018; Hackett et al., 2020; Bloom et al., 2022; Deck et al., 2022), indicating that selective vulnerability can arise from neuron-extrinsic mechanisms as well. Having recently identified subtypes within the population of myelinating SCs, we tested whether a distinct SC subpopulation is required for the survival of a specific axon subpopulation. We addressed this glia-centric hypothesis by generating mice that express tamoxifen-inducible Cre in the PMP2+ subtype of myelinating SCs. Upon DTA-mediated ablation of PMP2+ SCs, these mice develop profound axon loss in the mostly motor femoral nerve but not in the primarily sensory sural nerve, manifesting as motor behavior and electrophysiological defects, as well as ultrastructural defects reminiscent of human hypertrophic peripheral neuropathies such as CMT1A. Large-caliber motor axons preferentially rely on PMP2+ SCs for their survival and proper function. To our knowledge, this is the first report of SC subtype-selective impairment resulting in selective axonal vulnerability. Ablation of PMP2+ SCs also robustly induces the emergence of repair SCs and activated macrophages, hallmarks of regeneration, in the heavily damaged femoral nerve. Of note, we also observe ultrastructural abnormalities of unmyelinated axons and their SC processes in response to ablation of PMP2+ SCs. This may reflect a secondary effect upon the unmyelinated axons caused by the demyelination of large axons with concomitant repair SC formation and neuroinflammation, or it may reflect an unappreciated role of PMP2+ SCs in the ensheathment of unmyelinated axons. When diphtheria toxin expression is stopped, PMP2+ SCs and large-caliber motor axons are eventually replenished, coinciding with improved behavior and electrophysiology. These findings strongly support the hypothesis that neuron-extrinsic mechanisms can drive selective neurological deficits and provide a foundation for the development of SC subtype-targeted therapies to treat peripheral neuropathies.
SCs along nerves are typically classified into just two basic categories: myelinating or nonmyelinating SCs (Jessen et al., 2015b). However, we and others recently identified two distinct subtypes of nonmyelinating SCs and multiple subtypes of myelinating SCs (Yim et al., 2022), showing that the PNS glial landscape is significantly more diverse than previously understood (Wolbert et al., 2020; Gerber et al., 2021). Furthermore, we demonstrated that PMP2+ myelinating SCs are preferentially associated with large-caliber motor axons (Yim et al., 2022). This suggested that selective neuronal vulnerability might also stem from the malfunction of distinct SC subtypes. Indeed, mutations in the PMP2 gene have been linked with Charcot–Marie–Tooth (CMT) disease, a relatively common progressive distal neuropathy (Bird, 2024; Scherer and Svaren, 2024). PMP2 belongs to a family of proteins that bind and transport fatty acids to different cellular compartments (Smathers and Petersen, 2011). The point mutation I43N, located in the highly conserved cytosolic fatty acid-binding domain of PMP2, was identified in a family with autosomal-dominant demyelinating CMT neuropathy, and transgenic mice bearing the I43N mutation develop aberrant myelin sheaths, motor behavior defects, and reduced motor nerve conduction velocity (Hong et al., 2016). This indicates that SCs expressing pathological PMP2-I43N protein fail to sufficiently support axons, leading to their malfunction, which is in line with our observations following PMP2+ SC ablation.
PMP2+ SCs are necessary for the survival of large-caliber motor axons, but what role does the axonal subtype itself play in determining the specific fates of associated SCs? Axonal expression of neuregulin1 type III (NRG1t3) determines myelinating SC fate (Nave and Salzer, 2006) and regulates myelin thickness (Michailov et al., 2004). PMP2+ SCs develop exceptionally thick myelin, and PMP2 expression is directly regulated by NRG1t3 (Belin et al., 2019; Scapin et al., 2019; Hong et al., 2024). However, NRG1t3 overexpression does not similarly increase the expression of major myelin genes, such as Pmp22, Mbp, and Egr2, indicating that NRG1t3 regulates PMP2 expression through a yet-unidentified transcription factor independent of EGR2 (Belin et al., 2019; Scapin et al., 2019). These results suggest that the development of the PMP2+ SC subset could be initiated after signaling downstream of NRG1t3 crosses a certain threshold, similarly to the role of NRG1t3 in myelinating SC fate determination. Future studies will address how signaling downstream of NRG1t3 and other factors regulates the differentiation of the PMP2+ SC subtype from the general myelinating SC population.
A remarkable characteristic of mature SCs is their ability to respond to injury by transdifferentiating into repair SCs that recruit and activate macrophages and orchestrate nerve regeneration (Jessen and Mirsky, 2016, 2019). We have shown that upon ablation of PMP2+ SCs, repair SCs and activated macrophages arise in the damaged nerves, after which PMP2+ SCs are replenished as large-caliber motor axons regenerate. Therefore, our Pmp2-CreERT2 mouse line is an excellent tool for studying repair SC plasticity. One question to address is the capacity of repair SCs to redifferentiate into myelinating and nonmyelinating SCs after nerve regeneration. During development, SCs adopt myelinating or nonmyelinating phenotypes determined by signals from axons (Jessen and Mirsky, 2019), but are repair SCs similarly plastic? Do those that contact large-caliber axons always develop myelin, and do those that contact small-caliber axons necessarily convert into nonmyelinating SCs? Lineage tracing experiments utilizing a tamoxifen-inducible Mpz-CreERT2 line to ablate all myelinating SCs in adult mice demonstrated that some formerly myelinating SCs assume a nonmyelinating phenotype after recovering from injury (Stierli et al., 2018). Furthermore, nearly all myelinating SCs had the capacity to convert into proliferating repair SCs, suggesting no need for a stem cell pool to populate injured nerves with repair SCs (Stierli et al., 2018). Our Pmp2-CreERT2 mouse could be used to explore these important phenomena with an even tighter focus: are all SC subtypes equally predisposed to become repair SCs, and can even thickly myelinating PMP2+ SCs redifferentiate back into any other subtype, or are they permanently restricted by their original fate? The Pmp2-CreERT2 mouse will allow us to address this important question of SC biology relevant to disease and regeneration.
Footnotes
We thank all members of the DiAntonio and Milbrandt labs for the rigorous discussion and kind support. We thank the Genome Engineering & Stem Cell Center (GESC) at Washington University in St. Louis for their assistance with CRISPR reagent design and validation. This work was supported by National Institutes of Health Grants R37NS065053 to A.D. and R01-NS105645 to J.M.
A.D. and J.M. are cofounders, scientific advisory board members, and shareholders of Disarm Therapeutics, a wholly owned subsidiary of Eli Lilly. A.J.B. is a consultant for Disarm Therapeutics.
- Correspondence should be addressed to Jeffrey Milbrandt at jmilbrandt{at}wustl.edu or Aaron DiAntonio at diantonio{at}wustl.edu.
