Schwann cells contribute to demyelinating diabetic neuropathy and nerve terminal structures in white adipose tissue

Summary

Peripheral neuropathy, which can include axonal degeneration and/or demyelination, impacts adipose tissues with obesity, diabetes, and aging. However, the presence of demyelinating neuropathy had not yet been explored in adipose. Both demyelinating neuropathies and axonopathies implicate Schwann cells (SCs), a glial support cell that myelinates axons and contributes to nerve regeneration after injury. We performed a comprehensive assessment of SCs and myelination patterns of subcutaneous white adipose tissue (scWAT) nerves, and changes across altered energy balance states. We found that mouse scWAT contains both myelinated and unmyelinated nerves and is populated by SCs, including SCs that were associated with synaptic vesicle-containing nerve terminals. BTBR ob/ob mice, a model of diabetic peripheral neuropathy, exhibited small fiber demyelinating neuropathy and alterations in SC marker gene expression in adipose that were similar to obese human adipose. These data indicate that adipose SCs regulate the plasticity of tissue nerves and become dysregulated in diabetes.

Graphical abstract

Highlights

• •Mouse scWAT is densely innervated by myelinated and unmyelinated nerves
• •Mouse scWAT is populated by multiple Schwann cells (SC) subtypes
• •SCs are located near synaptic vesicle-containing neuro-adipose nexus (NAN)
• •BTBR ob/ob scWAT has demyelinating neuropathy; downregulation of SC markers

Biological sciences; Cellular physiology; Cell biology; Cell

Introduction

Peripheral nerves in adipose tissue are demonstrably important for the regulation of tissue functions and whole-body metabolic health, as reviewed in the study by Blaszkiewicz et al.¹ Bidirectional neural communication between adipose and brain is achieved by afferent sensory nerves that communicate from adipose to the brain via the dorsal root ganglia and release neuropeptides to adipose tissue when activated, as well as sympathetic nerves that communicate in an efferent manner from brain to adipose and release norepinephrine to the tissue. Only recently have the diversity of nerve fibers in adipose tissue, their patterns of innervation, and identification of nerve terminals or junctions in the parenchyma versus the vasculature been explored.^2,3 Many functional properties of adipose innervation, including the importance of axonal myelination, remain understudied.

Peripheral neuropathy, or the dying-back of peripheral axons innervating tissues and organs, blunts neural communication with the brain, leading to significant health complications including neuropathic pain.^4,5,6 We previously demonstrated that aging leads to peripheral neuropathy in subcutaneous white adipose tissue (scWAT),⁷ and that in obese and diabetic mice homozygous for the spontaneous Lep^ob mutation (BTBR ob/ob),⁸ the peripheral neuropathy also extended to metabolically relevant tissues like adipose and muscle.⁷ Deletion of brain-derived neurotrophic factor (BDNF), a neurotrophic factor implicated in human obesity, from myeloid-lineage immune cells also caused a “genetic denervation” of adipose tissue^9,10 that was accompanied by an altered metabolic phenotype. From these prior studies, it was clear that peripheral neuropathy negatively impacts adipose tissue, and that adipose neuropathy has implications for whole-body metabolic health. Experimental adipose denervation studies have similarly demonstrated the importance of adipose innervation by both sensory and sympathetic nerves for several decades now.¹

Broadly, peripheral neuropathies are categorized as either demyelinating neuropathy, involving the progressive breakdown of the myelin sheath, axonopathy which is characterized by the loss of the entire axonal structure, or a combination. Peripheral neuropathies are prevalent in patients with metabolic diseases, particularly diabetes—the number one cause of neuropathy. Diabetic distal symmetric polyneuropathy is length dependent and involves severe axonopathy of long axons (sensory, motor, and sympathetic) beginning in the extremities, typically in a “stocking and glove” pattern, as reviewed in the study by Feldman et al.⁴ Diabetic peripheral neuropathy (DPN) is a small fiber neuropathy that leads to allodynia and hyperalgesia. From our prior work, small fiber innervation in scWAT may be particularly important for metabolic control.^9,10 Although DPN has been studied extensively, the mechanisms that initiate nerve loss, potentially involving demyelination in a tissue-dependent manner, are still unclear, and little research has specifically investigated how neuropathy impacts metabolically relevant tissues such as adipose. Loss of adipose innervation with diabetes may exacerbate an already metabolically unhealthy state and promoting adipose re-innervation may present a therapeutic approach to restore metabolic control.

Mechanisms of both demyelinating neuropathy and axonopathy implicate the impairment or dysregulation of Schwann cells (SCs). SCs are enwrapping glia that support the development and physiology of peripheral nerves, similar to the role oligodendrocytes carry out in the brain. During development, SCs associate with axons of differing diameters, resulting in two distinct mature phenotypes: myelinating SCs (mSCs) which enwrap a single large axon to form the myelin sheath, thereby enhancing electrical conductance by providing insulation for the axon’s ionic flow; and non-myelinating SCs (nmSCs) that are associated with multiple small-diameter axons (Remak bundles), such as sensory axons and axons of the autonomic nervous system, as reviewed in the study by Jessen et al.¹¹

scRNAseq studies in adipose have identified populations of SCs in the stromal vascular fraction (SVF)^12,13,14 and SCs have been visualized in visceral white adipose tissue (WAT)¹⁵ and brown adipose tissue (BAT)³ by transmission electron microscopy (TEM). Most recently, neural crest-derived SCs were isolated from nerve bundles in mouse and human scWAT for use in neurogenic therapies.¹⁶ Despite these recent studies, the contributions of adipose tissue SCs to nerve function and tissue function remain unclear. This study aimed to better understand the presence and distribution of SCs within scWAT and investigate SC phenotypic changes in response to metabolic interventions. We provide the first data on adipose-resident SCs, their association with adipose nerve structures including nerve terminals, and data on SC plasticity and their roles in adipose-related pathologies, including diabetic demyelinating neuropathy. We report the presence of a heterogeneous myelination pattern among scWAT axons, including the first description of demyelinating neuropathy in the adipose of diabetic BTBR ob/ob mouse models. Finally, changes in SC gene expression in adipose from human obese patients emphasize the clinical relevance of these findings.

Results

Patterns of nerve myelination in scWAT

We used C57BL/6-Tg(Uchl1-EGFP)G1Phoz/J mice which endogenously express a pan-neuronal GFP reporter (henceforth referred to as PGP9.5-EGFP^+/−) to investigate the relative proportions of myelinated and unmyelinated nerves in inguinal scWAT (Figures 1A–1C). To label the myelin sheath of adipose-resident peripheral nerves, we used antibodies against two myelin-specific proteins, MPZ that is more commonly found in the peripheral nervous system (PNS), and MBP that is more commonly found in the CNS but also localized to PNS myelin.¹⁷ Co-expression of PGP9.5-EGFP^+/− with either MPZ or MBP was interpreted as a myelinated fiber and the absence of either marker on a PGP9.5-EGFP^+/− nerve was interpreted as an unmyelinated fiber. Each whole-adipose tissue depot was systematically visually scanned for qualitative analyses, and representative images for each tissue were captured of nerve bundles, nerves innervating blood vessels, and small fiber nerves in the parenchyma. Blood vessels were identified by structural morphology/autofluorescence, based on previous work in our lab.²

Figure 1.

Immunofluorescence in subcutaneous white adipose tissue (scWAT) reveals a heterogenous mix of myelinated and unmyelinated axons

Intact inguinal scWAT depots were excised from PGP9.5-EGFP^+/− reporter mice and immunolabeled for myelin with MPZ and MBP. Representative images are displayed of nerve bundles (top-down and cross-sectional), nerves interacting with blood vessels, and those in the parenchyma. Each marker was immunostained and analyzed individually (A and B) and co-labeled (C) to investigate co-expression. Structures co-labeled with PGP9.5-EGFP^+/− (green) and either of the myelinating markers (red) were identified as myelinated nerves (A and B). Overlap of MPZ (red) and MBP (blue) was used to determine extent of MPZ and MBP co-expression (C). White boxes show selected regions digitally enlarged to aid visualization. Images were captured on Stellaris 5 confocal microscope. See Figure S1 for single-color channels of each image. Scale bars from left to right: (A) 48.2 μm, 23.3 μm, 61.5 μm, 96.9 μm, (B) 61.5 μm, 23.3 μm, 61.5 μm, 61.5 μm, (C) 61.3 μm, 23.3 61.3 μm, 61.3 μm.

All large nerve bundles (>25 μm diameter when viewed from above) in scWAT presented with myelination, as indicated by MPZ/MBP labeling (Figures 1A and 1B, Nerve Bundle). TEM images also revealed the presence of large myelinated axons (Figure 3A). The large nerve bundles displayed numerous branching points that eventually led to individual myelinated fibers that branched further into thinly or unmyelinated axonal nerve endings. The myelin sheath along each axon is non-uniform, and myelin sheaths could not be traced along each individual axon in the field of view. Digital cross-sectioning of these nerve bundles provided a clear view of individual axons within the bundle, and the MPZ+ and MBP+ myelin sheaths around them (Figures 1A and 1B, cross-section). In this view, it was evident that MPZ and MBP labeling were specific to the myelin sheath. Light scatter and a potential inability for antibodies to penetrate the full diameter of nerve bundles resulted in the superficial aspect (closest to the light source) to appear brighter than the nerve fibers distal from the light source and some visual obscurity by the overlying nerves within the bundle. Myelin was present throughout the depth of each bundle, but localization of MPZ+ and MBP+ staining to individual fibers was most obvious superficially on surface axons.

Figure 3.

Transmission electron microscopy (TEM) of scWAT nerve bundle confirms presence of myelinated and unmyelinated axons

A large nerve bundle traversing into the inguinal scWAT depot was excised and processed for imaging by TEM. Myelinating Schwann cell (mSC) forming a myelin sheath around an axon within the nerve bundle (A, interior of dashed line). Axons lacking myelin sheaths observed bundled together by non-myelinating Schwann cell to form a Remak bundle (B, interior of dashed line). A = axon, M = myelin, Nu = Schwann cell nucleus. Scale bars: (A) 900 nm, (B) 600 nm.

Nerves innervating the adipose vasculature followed consistent patterning to what has been described previously²; small unmyelinated fibers contacted vessel walls and myelinated fibers traversed in parallel with the vessel. Here, we noted surprising variation between the markers for myelin sheaths (Figures 1A and 1B, neurovascular). MPZ consistently labeled one or two myelinated fibers running parallel to most blood vessels, excluding capillaries, and the small fibers connecting with the vessel wall did not co-express MPZ, supporting previous findings² that the nerves directly innervating tissue vasculature are unmyelinated sympathetic nerves (Figure 1A, neurovascular). By contrast, MBP staining of neurovascular fibers was less consistent. MBP was observed to label the thicker nerve fibers running parallel to vessels as well as many of the smaller fibers contacting the vessel wall (Figure 1B, neurovascular). Similar to the neurovascular labeling, MPZ was expressed only on a subset of nerve fibers in the parenchyma with the majority being MPZ- (Figure 1A, Parenchymal). MBP labeled parenchymal fibers found alongside adipocytes although MPZ appeared to label more parenchymal fibers. (Figure 1B, Parenchymal).

Because MPZ and MBP had slightly different patterns of labeling, it was necessary to co-stain with these antibodies to gauge the extent of their overlap (Figure 1C). Co-labeling MPZ with MBP demonstrated complete co-expression in the myelin sheaths around nerve fibers within bundles (Figure 1C, nerve bundle, cross-section). Co-labeling of MBP with MPZ showed that labeling around vasculature (Figure 1C, Neurovascular) and the parenchyma (Figure 1C, Parenchymal) was consistent for both markers.

Distinguishing sympathetic and sensory nerve myelination in scWAT

Given scant prior reports of the myelination status for the sensory and sympathetic innervation in WAT, we assessed myelin presence on sympathetic nerves that were labeled for tyrosine hydroxylase (TH), the rate-limiting enzyme for catecholamine (eg: norepinephrine) synthesis and a canonical marker for sympathetic nerves, as these nerves comprise the majority of scWAT innervation.¹⁸ We also labeled sensory nerves for the sensory-specific neuropeptide calcitonin gene-related peptide (CGRP). Of note, recent findings reveal that up to 40% of adipose sensory nerves are TH+,¹⁹ making it difficult to clearly delineate between sympathetic and sensory identity. In addition, this binary distinction between sensory and sympathetic may underrepresent the overall diversity of adipose nerves since sensory nerves can release a variety of neuropeptides, sympathetic nerves can co-release modulatory neuropeptides with norepinephrine, and non-peptidergic sensory fibers are only beginning to be described in scWAT.²

Intact inguinal scWAT depots (including the dorsolumbar portion) were excised from PGP9.5-EGFP^+/− mice and co-stained for TH and MBP. TH+ axons comprised only a subset of the axons within each myelinated nerve bundle (Figure 2A, nerve bundle, cross-section), but this confirmed that the bundles are indeed mixed nerves. Neurovascular axons were almost entirely TH+, excluding rarely observed myelinated nerves running in parallel with blood vessels, which were TH-/MBP+ (Figure 2A, neurovascular). These neurovascular TH+ axons may include some of the reported TH+ sensory fibers, since sensory innervation is needed for vascular control. We observed that the majority of TH+ fibers co-expressed MBP around vasculature and in the parenchyma (Figure 2A, parenchymal).

Figure 2.

Sympathetic and sensory nerve myelination in scWAT

To investigate the status of sympathetic and sensory nerve myelination, intact inguinal scWAT depots were excised from PGP9.5-EGFP^+/− (green) reporter mice and co-labeled for MBP (blue) and either the sympathetic nerve marker TH (red) (A) or the sensory nerve marker CGRP (red) (B). Representative images of nerve bundles, vascular innervation, and parenchymal innervation are displayed. White boxes were digitally enlarged to aid in visualization. Red and blue overlap identified myelinated sympathetic (A) and sensory nerves (B). Axons in the parenchyma demonstrate nuclei (DAPI, gray) embedded in both sympathetic and sensory nerves (C). Images were captured on Stellaris 5 confocal microscope. See Figure S2 for single-color channels of each image. Scale bars from left to right: (A) 61.5 μm, 23.3 μm, 61.5 μm, 61.5 μm, 61.5 μm, (B) 61.5 μm, 23.3 82 μm, 61.5 μm, 61.5 μm, (C) 15.1 μm, 25.3 μm, 22.5 μm.

PGP9.5-EGFP^+/− nerve bundles co-stained for CGRP and MBP revealed punctate CGRP+ labeling across the length of a subset of the axons within myelinated nerve bundles (Figure 2B, nerve bundle). Digital cross-sections supported this finding, although due to the punctate nature of the staining (likely because the vesicles containing CGRP were not uniformly distributed along the axon), it is possible that digital cross-sections underrepresented the total number of CGRP+ axons in each bundle (Figure 2B, cross-section). Neurovascular imaging of CGRP+ nerves showed that the majority of CGRP+ axons outside of nerve bundles were closely aligned with blood vessels (Figure 2B, neurovascular), again supporting sensory innervation of vessels. This was not surprising given CGRP’s known role in vasodilation.^20,21 Small axons in contact with the vessel wall were largely CGRP-/MBP-, and only rarely were CGRP+/MBP+ (Figure 2B, neurovascular). Most parenchymal nerves were CGRP-/MBP-, with infrequent CGRP+/MBP-, CGRP+/MBP+, and CGRP-/MBP+ axons also observed across the depots (Figure 2B, Parenchymal). Parenchymal nerve fibers presented with what appeared to be cell bodies residing between axons and forming gaps between adjacent fibers (Figure 2C), which were presumed to be SCs or neuroimmune cells as we have observed in separate co-labeling experiments, since neuronal cell bodies are in the dorsal root ganglia and not in the tissue itself. DAPI-labeled nuclei were found embedded in both myelinated (MBP+) and unmyelinated (MBP) axons, regardless of CGRP or TH expression (Figure 2C). Combined, these data exemplify the heterogeneity of nerve fibers within scWAT, including myelinated and unmyelinated sensory and sympathetic axons.

SCs in scWAT

To confirm that the nuclei residing between adjacent nerve fibers and around scWAT axons were SCs, we first confirmed the presence of mSCs and nmSCs in large nerve bundles innervating ing-scWAT using TEM, which positively identified myelinated axons (Figure 3A) and non-myelinated Remak bundles (Figure 3B). Next, we labeled whole inguinal scWAT from PGP9.5-EGFP^+/− reporter mice for the SC lineage-determining transcription factor SOX10,^22,23 which was used as a pan-SC marker. DAPI+ nuclei residing between adjacent fibers and along the parenchymal nerves were found to be SOX10+ (Figure 4A), thus confirming their SC identity. A thorough analysis of the whole tissue revealed that SOX10+ cells were prominently distributed throughout the tissue (Figure 4B), with the majority localized to nerve bundles. SOX10+ cells contributed to most of the DAPI-labeled nuclei within and around each nerve bundle (Figure 4B, nerve bundle, cross-section). The remainder of the nuclei associated with each bundle was speculated to be neuroimmune cells, based on our prior data.⁹ The second greatest proportion of SOX10+ cells were observed in contact with the nerves innervating tissue vasculature (Figure 4B, neurovascular), with relatively few SOX10+ cells observed in the parenchyma around adipocytes (Figure 4B, parenchymal). Importantly, the SOX10 antibody used in this manuscript also demonstrated labeling of what appeared to be mammillary ducts and/or lymphatic vessels across the depot (Figure S3A). Fortunately, SCs and ducts/vessels were easily distinguished from one another due to the distinct morphology of cells versus vessels (Figure S3B).

Figure 4.

Schwann cell (SC) bodies are embedded in heterogenous nerves within scWAT

Intact inguinal scWAT depots were excised from PGP9.5-EGFP^+/− (green) reporter mice and co-stained to label SCs (SOX10, red) and nuclei (DAPI, gray) (A and B). High magnification image of SOX10-positive nuclei embedded in small parenchymal nerve fibers (A). Representative images of SOX10-positive nuclei distribution throughout scWAT (B). SC localization in relation to myelinated nerves stained with MBP (C). Overlap of red and gray identified SCs (A–C). Overlap of green and blue identified myelinated nerves (C). Images were captured on Stellaris 5 confocal microscope. See Figure S3 for single-color channels of each image. Scale bars from left to right / top to bottom: (A) 31.4 μm, 23.2 μm, (B) 82 μm, 23.3 μm, 61.2 μm, 49.2 μm, (C) 50 μm, 20 μm, 50 μm, 61.5 μm.

Although SOX10 staining cannot differentiate between mSCs and nmSCs, we observed that the greatest proportion of SOX10+ cells were distributed in structures that we had previously observed to have significant myelination.² To confirm this, we co-stained scWAT from PGP9.5-EGFP^+/− reporter mice for SOX10 and MBP (Figure 4C). As expected, most SOX10+ cells were associated with MBP+ nerves, indicating they were mSCs. There were also SOX10+ cells associated with unmyelinated nerves in the parenchyma, but these were less common (Figure 4C).

The neuro-adipose nexus (NAN) is preceded by SCs

Following the recent discovery of putative nerve terminals in scWAT, a structure where axons terminate by wrapping around individual adipocytes which we termed the “neuro-adipose nexus” (NAN),² we sought to learn more about these structures by labeling various nerve, synaptic, and SC markers (Figure 5). Post-synaptic labeling has proven to be challenging in adipose, given the poor selectivity of antibodies for neurotransmitter receptors. Whole inguinal scWAT depots were excised from C57BL/6J mice and immunostained for sympathetic nerves (TH). Tissue autofluorescence was captured intentionally to visualize adipocyte boundaries and tissue structure along with TH labeling (Figure 5A), as is often exploited in tissue clearing experiments. Images were also captured with TH labeling that did not include tissue autofluorescence (Figure S4A), as a control. NANs were labeled clearly by TH and were visualized in clusters (2–4 adipocytes) as well as individually (Figures 5A and S4A). These images revealed that branching axons terminate on the cell surface of adipocytes (Figure 5A), which we can see.² Axons forming each NAN were pearled by varicosities, similar to those observed in autonomic neuroeffector junctions²⁴ (Figure 5A). Varicose axons were found throughout scWAT, with NANs characteristically displaying the greatest number of varicose axons (Figure 5B). To investigate whether synaptic transmission/neurotransmitter release could be occurring at these nexuses, we immunostained whole PGP9.5-EGFP^+/− inguinal scWAT depots for the pre-synaptic vesicle organization protein, Synapsin I²⁵ (Figure 5C), and for the membrane glycoprotein SV2, which is localized to secretory vesicles²⁶ (Figure 5D). NANs were labeled by both of these characteristic pre-synaptic markers, which provided further evidence that these terminal structures are points of communication between the peripheral nervous system and single adipocytes. This is likely a site of neurotransmitter release, potentially norepinephrine, as NANs are largely labeled by TH (Figures 5A and 5B). However, the actual release of neurotransmitters at NANs has not yet been confirmed. Unexpectedly, we found that NANs are also labeled by the sensory marker CGRP (Figure 5E) which suggests that they may be innervated by the population of TH+ dorsal root ganglion neurons that comprise 40% of sensory innervation in scWAT.¹⁹ This unfortunately provides little incite to their function as TH+ DRG neurons innervate numerous target organs and tissues and display distinct functional characteristics specific to the target.²⁷ Because of this, it is difficult to clearly define the function of NANs.

Figure 5.

The neuro-adipose nexus (NAN) is densely pearled by synaptic vesicle containing axonal varicosities and is often immediately preceded by SCs

Intact inguinal scWAT depots were excised from C57BL/6J mice and labeled for TH which revealed NANs throughout the tissue parenchyma terminating on single adipocytes or small clusters of adipocytes (A). TH labeling of NANs was merged with single Z-planes of purposefully captured tissue autofluorescence to display adipocyte boundaries in relation to the spread of axons in each NAN; displayed as inverted monochrome images (A). TH+ axons are black overlying white adipocytes with gray extracellular matrix separating each cell (A). NANs were characteristically densely populated by axonal varicosities when compared to other parenchymal axons within the same tissue; images displayed with Glow LUT (B). Co-labeling of PGP9.5-EGFP^+/− (green) reporter mice demonstrated that NAN axonal varicosities contain the synaptic vesicle organizing protein, SYN1 (red) (C), as well as synaptic vesicles (SV2, red) (D). NANs were also labeled by CGRP (red). Boundaries between adjacent adipocytes visualized using autofluorescence (white) (E). Varicose axons were observed surrounding unlabeled putative SCs (gaps between adjacent fibers; white arrows) in the parenchyma (F) and preceding many NANs (C). SC identity was confirmed by SOX10 labeling; overlap of SOX10 (red) with DAPI (blue) within a nerve (green) leading to NAN (G). White boxes were digitally magnified to aid in visualization. All images were captured on a Stellaris 5 confocal microscope. (H) Schematic of NAN structure and comparison to neuromuscular junction (NMJ). Lightning deconvolution was utilized to resolve axonal varicosities (A–C, E). See Figure S4 for single-color channels of each image. Scale bars from left to right / top to bottom: (A) 40 μm, 15 μm, 30 μm, 20 μm, 15 μm, 15 μm, 40 μm, (B) 8.1 μm, (C) 53.8 μm, (D) 29.1 μm, (E) 48.2 μm, (F) 16.6 μm, 15.4, (G) 60 μm.

Varicose TH+ axons were also visualized surrounding putative nerve-embedded SCs, as indicated by unstained gaps between adjacent fibers (Figure 5F, white arrows), similar to those previously shown to contain nuclei (Figure 2C) and confirmed to be SCs (Figure 4A). These same nerve-embedded SCs were frequently observed preceding NANs (Figure 5G, white arrows and Figure S4, white arrows). Co-staining of PGP9.5-EGFP^+/− inguinal scWAT revealed that these gaps between adjacent fibers were indeed SOX10+ SCs. To better characterize the NAN SCs, we stained whole inguinal scWAT from C57BL/6J mice for neural cell adhesion molecule (NCAM), a marker for nmSCs²⁸ (Figure S4E). However, we found that NCAM was not specific to nmSCs in scWAT and labeled all parenchymal nerves indiscriminately (Figure S4F), including myelinated fibers (Figure S4G).

Several similarities were observed between the NAN and other peripheral nerve termini, the well-characterized neuromuscular junction (NMJ) (Figures 5H and S4H). For both termini, an axon branches from peripheral nerve bundles toward specific cellular targets, to form morphologically distinct terminal connections. The axons leading to NMJs are myelinated,²⁹ whereas the more distal pre-synaptic terminal is unmyelinated.^28,30 mSCs are found in/on the myelinated region of the NMJ axon, and nmSCs (important for maintaining synapses³⁰) are found at the terminal, which are called terminal SCs (tSCs). In adipose, the NAN similarly branches from myelinated nerves with the terminal itself being unmyelinated, and like the NMJ, the NAN is immediately preceded by SCs. However, SCs have yet to be identified at the terminal NAN in adipose. Despite many parallels to NMJ structure, the presence of terminal nerve structures within adipose is much more infrequent than that of muscle. Typically, all muscle fibers are innervated by an NMJ, but very few adipocytes form a NAN.

Fluorescence-activated cell sorting of SVF indicates the presence of two distinct SC populations

To determine both the presence and relative quantity of nmSC and mSC sub-populations in scWAT, O4 and O4/p75NTR were used as their respective canonical markers, as previous literature indicated the use of O4 and p75NTR expression to positively select for SCs in rat sciatic nerve.³¹ CD45 was used to negatively select for immune cells, after FSC/SSC gating on live singlet cells. O4 was used to positively select for mSCs, while O4 and p75NTR together were used to positively select for nmSCs.³² The population of mSCs in scWAT was significantly higher compared to the nmSC population (Figure 6A). This was consistent with our immunofluorescence labeling of SOX10+ SCs being associated primarily with myelinated structures in scWAT (Figure 7C). Since exercise has been shown to increase scWAT innervation,⁷ we aimed to investigate whether this intervention affects the relative distribution of SC populations in scWAT. Adult male BL6 mice were exercised for 7 days (caged with unrestricted running wheel access) or maintained sedentary (caged with locked running wheel). We observed no difference between exercised and sedentary SC populations in the scWAT of the two groups (Figure 6B). However, for both groups, the amount of mSCs trended higher compared to nmSCs (Figure 6B). Conversely, we have also previously shown that aging decreased innervation in scWAT,⁷ so we also examined whether aging alters SCs populations in scWAT. Following fluorescence-activated cell sorting (FACS) of scWAT SVF from male BL6 mice across different age groups (4 months, 8 months, and 15 months), we again observed a higher amount of mSCs compared to nmSCs at 4 months (p = 0.0011) and 8 months (p = 0.0017), but at the age of 15 months there was a loss of statistical significance (Figure 6C), possibly indicative of a relative decrease in myelinating SCs. Limitations in the FACS approach may preclude detailed phenotyping of SC subtypes in the tissue, including cell size exclusion and “stickiness” of myelin to the adipocyte fraction. Future experiments utilized a magnetic bead sorting approach that avoided cell size exclusion.

Figure 6.

Fluorescence-activated cell sorting (FACS) and quantification of SCs in scWAT

Stromal vascular fraction (SVF) was isolated from inguinal scWAT depots excised from C57BL/6J mice and sorted into two distinct SC populations: CD45-O4+p75-, myelinating SCs (mSCs); and CD45-O4+p75+, non-myelinating SCs (nmSCs). mSCs and nmSCs were quantified as percent of live cells. Basal: 21-week-old male mice (N = 4) (A). Exercise: Male mice aged between 29 and 68 weeks were either given continuous access to run (Exercise, N = 5) or had the running wheel locked in place (Sedentary, N = 5) (B). Aging: Male mice at three ages (4-months, N = 5; 8-months, N = 5; 15-months, N = 5) (C). Statistics: unpaired Student’s t-test (A) and one-way ANOVA with multiple comparisons (B–C). Error bars are SEMs. P-values are as shown or are not provided when not significant (n.s.).

Figure 7.

Overview of data collection for metabolic models

Changes in SC phenotypes were assessed through RT-qPCR, cell sorting, and imaging in multiple energy balance. Positive energy balance states include diet-induced obesity and leptin-deficient BTBR ob/ob mice. Negative energy balance states include voluntary exercise and cold-exposure cohorts. Aged mice from 4 months to 75 weeks were also assessed.

Changes to SC phenotype with altered energy balance status

Adipose tissue innervation is highly plastic and imbalances in energy intake and energy expenditure have correlative impacts on tissue total innervation patterns and sympathetic nerve activity,^7,33,34 but the impacts of changing energy balance on adipose-resident SCs have not yet been examined despite the known presence of SCs in the tissue and their prominence in recent single-cell datasets. We investigated gene expression changes in scWAT with obesity, aging, cold exposure, and exercise (Figure 7) using six SC-specific markers: pan-SC markers including Sox10,^22,23 oligodendrocyte marker 4 (O4),³⁵ and neurotrophin receptor p75 (p75)^35,36; myelin-specific markers Mpz and Krox20^37,38; and the repair SC transcription factor c-Jun.³⁹ Because the associated changes in innervation status are likely to impact the total number of SCs, we also normalized gene expression to the pan-SC gene Sox10 to investigate changes not confounded by changes in total SC number (Figure 8).

Figure 8.

SC gene expression in scWAT with changing metabolic status

Relative gene expression measured by qPCR, normalized to total SCs (Sox10), and represented as fold change in ΔΔCt value (A–F). Gene expression of scWAT from BL6 mice that were either housed at 30°C, thermoneutrality (TN, N = 6); at 25°C, room temperature (RT, N = 6); or at 5°C (Cold, N = 6) (A). Fold change is normalized to RT group. Gene expression of scWAT from BL6 mice that were exercised for 7 days or remained sedentary (B). Mice were either given continuous access to running wheels (Exercise, N = 6) or had the running wheel locked in place (Sedentary, N = 4) (B). Gene expression of scWAT from BL6 mice aged 15 (Young, N = 5) and 75 weeks (Old, N = 4) (C). Gene expression of scWAT excised from BTBR ob/ob (MUT, N = 4) and wild-type littermates (WT, N = 5) (D). Gene expression of scWAT excised from BL6 mice fed either a 58% high-fat diet (HFD, N = 5) or chow (Control, N = 5) (E). Gene expression of scWAT biopsies excised from lean (BMI >30) and obese (BMI <25) human donors (F). Linear regression of Mpz and Sox10 fold change normalized to housekeeper gene Ppia (G). Statistics: one-way ANOVA with multiple comparisons (A) and unpaired Student’s t-test (B–F) and linear regression goodness of fit measured by R-squared and the significance of slope determined by F-test (G). Error bars are SEMs. P-values are as shown or are not provided when not significant (n.s.). See Figure S5 for all genes normalized to the housekeeper gene Ppia.

Cold exposure is a common means to increase sympathetic drive within adipose tissue and results in the tissue taking on a more metabolically favorable, energy expending phenotype with browning, increased sympathetic innervation, and non-shivering thermogenesis.³³ Male BL6 mice (N = 6) were cold exposed at 5°C for 3 days and scWAT gene expression of SC markers was compared with mice housed at thermoneutrality (30°C, a temperature at which mice use no thermogenesis) (N = 6) or at room temperature (25°C, a temperature at which mice experience mild cold stimulation) (N = 6). Cold-exposed mice exhibited a downregulation in Krox20 compared to mice housed at RT (p = 0.0157) (Figure 8A).

Exercise promotes adipose tissue lipolysis through increased sympathetic drive and has a demonstrated role in neuroplasticity,⁴⁰ similar to cold exposure. Accordingly, we compared mice with access to voluntary exercise (running wheel cages; N = 6) for 7 days, with locked wheel-caged sedentary littermates (N = 4) as control. Mice with access to running wheels displayed no significant differences in SC markers (Figure 8B) compared to sedentary controls (N = 4).

Aging correlates with an increased prevalence of age-related neuropathy in the skin,⁴¹ the underlying muscle,³⁰ and in scWAT.⁷ We investigated the effects of aging on SC gene expression in scWAT by comparing male BL6 mice at 15 weeks old (N = 5) to mice at 75 weeks old (N = 4) (Figure 8C). Aged mice exhibited an upregulation in Krox20 compared to the young controls (p = 0.0491) (Figure 8C).

BTBR ob/ob (MUT) mice are leptin deficient (exhibiting an obese, diabetic, and neuropathic phenotype), with reduced innervation of scWAT.^7,42 We compared relative gene expression of SC markers in MUT mice (N = 4) to wild-type BTBR (WT) littermate controls (N = 5) (Figure 8D). MUT mice had a significant upregulation of Krox20 (p = 0.0147), O4 (p = 0.0051), and c-Jun (p = 0.0063) (Figure 8D).

To investigate whether a diet-induced obesity model would result in similar changes seen in the adipose SCs of obese BTBR ob/ob MUT mice, we measured relative gene expression in BL6 mice fed a high-fat diet (HFD) for 19 weeks (N = 5) compared to chow-fed controls (N = 5) (Figure 8E). Interestingly, no significant changes were observed in SC marker gene expression. These mice also showed a less severe obese and neuropathic phenotype (Figure S5) than what has been reported for BTBR ob/ob mice,⁷ as well as increased differences in fasting blood glucose.⁸ We observed no differences in baseline fasting blood glucose level during a glucose tolerance test between HFD-fed and chow controls (Figure S5). These phenotypic differences may contribute to the differences we see in SC gene profiles of these two obesity models. However, scWAT biopsies from male and female lean (N = 11) and obese (N = 12) human donors did display an upregulation in p75NTR (p = 0.002) and a trending increase in Krox20 with obesity, similar to what was observed in the BTBR ob/ob mice (Figures 8F and S5).

In separate analyses, we normalized gene expression to the housekeeper gene Ppia to investigate what changes were occurring relative to the whole tissue (Figure S5). We found that cold exposure resulted in downregulation of Krox20 when compared to TN (p = 0.0184) (Figure S5A). There were no changes in gene expression with exercise or aging (Figures S5B and S5C). Prominently, BTBR ob/ob MUT showed downregulation of Sox10 (p = 0.0076), Mpz (p = 0.0089), and p75 (p = 0.0004), as well as upregulation of Krox20 (p = 0.0123) (Figure S5D). A linear regression was performed looking at the relative fold change between Sox10 and Mpz gene expression in BTBR ob/ob scWAT which found a strong correlation between the downregulation of Mpz with Sox10 in scWAT (r² = 0.8034, p = 0.0011) (Figure 8G). Diet-induced obesity in mice displayed no changes (Figure S5H), but obese human scWAT showed an increase in Krox20 gene expression (p = 0.0456) (Figure S5I).

When taking all datasets into consideration, it becomes evident that pro-myelinating Krox20 gene expression in scWAT is impacted by different metabolic states. Krox20 is downregulated in scWAT under an energy expending metabolic state (cold exposure) and significantly upregulated in unfavorable metabolic states (aging and obesity). Importantly, these changes are consistent regardless of the reference gene.

BTBR ob/ob mice are neuropathic and show demyelination of small nerve fibers

BTBR ob/ob mice displayed the most changes in SC gene expression (Figures 8D and S5D), and as such, we wanted to see if these changes were reflected in inguinal scWAT innervation and demyelination. Male and female BTBR ob/ob (MUT, N = 4) and BTBR +/+ wild-type littermates (WT, N = 3) were aged to at least 12 weeks old (when they show an obese and neuropathic phenotype⁷). At the time of tissue collection, MUT mice had significantly greater body weight (p = 0.0280), inguinal scWAT weight (p = 0.0013), and subcutaneous adiposity (p=<0.0001) (Figures S6A–S6C). Protein expression of whole inguinal scWAT lysates showed a decrease in sympathetic nerve activity (TH, p = 0.0020) in MUT (Figure 9A) when compared to WT, as previously shown to correlate with a decrease in total innervation.⁷ MUT mice also displayed a decrease in total SCs (SOX10) (p = 0.0452) (Figure 9B) and a decrease in total myelination (MPZ) (p = 0.0092) (Figure 9C). Isolated SCs from MUT and WT BTBR show no difference in expression of neurotrophic factors; nerve growth factor (Ngf), Bdnf, and neurotrophin-3 (Ntf3) (Figure S6D).

Figure 9.

BTBR ob/ob mice present with a loss of small fiber sympathetic innervation in scWAT, accompanied by small fiber demyelination

Male and female BTBR ob/ob (MUT, N = 4) and wild-type littermates (WT, N = 3). Western blot protein analysis of TH (A), SOX10 (B), and MPZ (C) in inguinal scWAT. A consistent nerve bundle entering scWAT depot was excised from each mouse, cross-sectioned, and stained with toluidine blue (D). G-ratio was measured for all myelinated axons in a nerve bundle and averaged for each mouse (E). G-ratios were pooled for each group (WT, 377; MUT, 981) and plotted against axon diameter (F). Linear regression values plotted in table below with slopes and Y-intercepts compared (F). Whole mount immunostaining of sympathetic nerves in axillary scWAT was captured as Z-stacks (10 μm step size) with a 10X objective, tiled 5 × 5 (22.85 mm² area), and maximum intensity projected (G). Neurite density measured as fluorescence area in each 22.85 mm² field of view (H). Whole mount immunostaining of inguinal scWAT nerves (TUBB3, green) and myelin (MPZ, red) demonstrating myelinated nerve bundles (I) and demyelination of small nerve fibers in MUT mice (J). Demyelination shown progressing along the length of comparable small fiber nerves overlying the subiliac lymph node (SiLN) in scWAT (K). Numbers 1–4 correspond to adjacent images digitally magnified to aid visualization (K). Brightfield images were captured on a Zeiss Axioskop microscope (D) and fluorescence images were captured on a Stellaris 5 confocal microscope (G,I–K). Lightning deconvolution (I–J). Statistics: unpaired Student’s t-test (A–C,E,H) and a linear regression (F). Error bars are SEMs. P-values are as shown. Scale bars from left to right: (D) 100 μm, 25 μm, (G) 1 mm, (I) 30.7 μm, (J) 13.5 μm, (K) 50 μm.

To gauge the scale at which demyelination was occurring, we started by consistently excising the same nerve bundle (Figure S6E) entering the inguinal scWAT depot and analyzed nerve fiber myelination status by G-ratio. We observed no differences in the average G-ratio between WT and MUT mice (Figure 9E). There was also no change in G-ratio (axon diameter/diameter of axon + myelin) compared to axon diameter between both groups, as determined by nearly identical linear regressions (Y-intercepts and slopes not significantly different) between WT and MUT mice (Figure 9F). With apparently no change in large bundle myelination, we turned our focus to the small fibers innervating the tissue parenchyma and vasculature.

Whole axillary scWAT depots from BTBR ob/ob mice were excised (WT N = 3; MUT N = 4) and immunostained for TH, which revealed a loss of small-diameter TH+ axons in the tissue parenchyma (Figure 9G), confirmed by quantifying neurite density (p = 0.036) (Figure 9H). A similar decrease in scWAT innervation was observed in C57BL/6J ob/ob mice.⁴³ To investigate if the observed reduction in Mpz gene (Figure 8D) and MPZ protein (Figure 9C) was due to an overall reduction in tissue innervation, or axonopathy (Figures 9G and 9H), or a demyelination of the existing small nerve fibers, we co-stained inguinal scWAT from BTBR ob/ob mice (WT N = 3; MUT N = 4) for the pan-neuronal marker beta III tubulin (TUBB3) and the myelin marker MPZ. Immunostaining of tissue-resident nerve bundles was consistent with our previous data, with no noticeable differences in bundle myelination (Figure 9I). However, many of the small fibers displayed a deterioration of the myelin sheath with the associated nerves taking on irregular shapes and a punctate appearance (Figure 9J). Because we could not differentiate between unmyelinated nerves and those that had become completely demyelinated (though nerve irregularity was an indicator), we felt that we could not make definitive claims as to the ratio at which demyelination was occurring throughout each tissue. Nerves also did not show uniform demyelination along their length (Figure 9K), adding an additional layer of complexity in the neuropathy phenotype. Regardless, a demyelinating neuropathy appears to be a character of the BTBR ob/ob MUT adipose tissues.

Discussion

Despite extensive research on SC development and injury responses within large nerve bundles such as the sciatic nerve, far less is known about tissue-resident SCs and how unique local environmental and metabolic cues may influence SC function. This is a significant gap in knowledge, given that chronic inflammatory demyelinating polyneuropathy (CIDP) is associated with metabolic conditions such as diabetes,^44,45 and the fact that demyelinating neuropathies may also be important in aging-related or idiopathic cases. Prior work has demonstrated the presence of SCs in brown adipose tissue,^3,13,46 and scRNAseq studies have also reported the presence of SCs in WAT.^12,14 Regardless, it stands out that no studies have directly assessed the contributions of myelinated nerves and SCs to adipose tissue physiology. We sought to begin to fill this gap in understanding by investigating the tissue-resident SCs present in scWAT, how they contribute to myelinated axons in the tissue, and how myelination or SC phenotype may shift with metabolic state.

Our data provide conclusive evidence for myelinated and unmyelinated nerve subtypes within adipose, carried into scWAT by shared/mixed peripheral nerve bundles. Within these myelinated nerve bundles, we observed co-localized expression of both MPZ and MBP, demonstrating a role for both myelin proteins in adipose nerves. Of note, MBP+ myelination was observed in both CGRP+ and TH+ nerve fibers (Figure 2). This is contrary to statements in the available literature which postulate that post-ganglionic TH+ sympathetic nerves are largely unmyelinated.⁴⁷ Limitations in imaging or other experimental approaches may have led to this prior conclusion. By contrast, previous work in rats had identified thinly myelinated sympathetic axons in the superior cervical ganglia and paravertebral chain ganglia;⁴⁸ and more recently, it was reported that MBP+ myelination may be protective against sympathetic denervation of the left ventricle in rhesus macaques.⁴⁹ Whereas these studies identified selected populations of myelinated sympathetic nerves, we observed a nearly complete overlap between TH and MBP within scWAT and found that the majority of TH+ nerve fibers are at least thinly myelinated in the mouse scWAT tissue environment. This observation may be partially explained by the recent discovery that over 40% of sensory nerves in scWAT are TH+,¹⁹ highlighting the need for more in-depth characterization of adipose nerve heterogeneity and more specific sympathetic nerve markers.

By contrast, we find that CGRP+ nerve fibers show greater heterogeneity in myelination. MBP and CGRP mark both thick and thin nerve fibers entwined in nerve bundles together, further exemplifying the heterogenous nature of nerve bundles in adipose tissue. This makes sense, given that sensory nerve fibers in the skin are also highly diverse, both functionally and characteristically. They can be either myelinated or unmyelinated, and when present, the myelin thickness corresponds to axon diameter and depends on several factors including stimuli detected by that nerve type, whether the skin is glabrous or hairy, and even the dermal layer the nerve resides within.⁵⁰ Given our initial observations here, similar diversity in scWAT sensory fibers also exists and warrants further exploration and a defined nomenclature.

Although it has been shown that TH+ nerve fibers may have a sensory function, TH is a canonical marker of sympathetic nerve activation. The exact role of sympathetic myelination within adipose tissue is beyond the scope of this study but may strengthen the bidirectional communication between the CNS and adipose depots by allowing faster neural conduction, given myelin’s function as insulation for ionic movement in axons. Axonal diameter and myelin sheath thickness are directly related to the speed of signal conductance and resulting physiological functions.⁵¹ Adipose nerves in the parenchyma also displayed heterogeneity in axon thickness and myelination and may signify diversity in SNS functions in the tissue, an important area of emphasis for future work.

We were limited to assessing sympathetic and sensory myelination exclusively with anti-MBP labeling due to antibody host species cross-reactivity. This poses a significant caveat of these assessments, as we did observe less MPZ+ labeling around small neurovascular and parenchymal fibers (Figure 1), which tended to be sympathetic. As MPZ is the primary myelinating protein of the PNS, future studies utilizing an MPZ fluorescent reporter mouse line would be important for clarifying this matter, and these are now underway in our laboratory. Our TEM images also underscored the presence of myelinated and unmyelinated axons in adipose, however.

The presence of nerve termini in scWAT (NANs), unique from the majority of innervation observed across the tissue architecture, is intriguing but ultimately provides more questions than it does answers. It is still unclear if a true synapse is being formed or what the function is of these terminal endings that form connections with a subset of mature adipocytes. Many of the parenchymal nerves in scWAT are varicose, indicating en passant nerve product release, and it was shown in BAT that sympathetic axonal varicosities can make direct contact with adipocytes.³ We have now shown that NANs themselves also house synaptic vesicles in axonal varicosities and likely function as a pre-synaptic terminal releasing nerve products (neurotransmitters or neuropeptides, depending on nerve type) onto effector adipocytes. Release of norepinephrine from sympathetic nerves activates β3-adrenergic receptors on white adipocytes, leading to WAT thermogenesis and WAT “browning”⁵² and thus perhaps these NAN-associated adipocytes are primed for the browning process. We have shown that NANs are comprised of TH+ nerve endings, implicating their involvement in the regulation of norepinephrine release and scWAT thermogenesis, but they may also represent TH+ sensory nerve endings.

A true post-synaptic junction on adipocytes has yet to be observed or described, despite the presence of post-synaptic proteins expressed in adipose tissue, such as PSD95.⁷ This may instead be similar to other autonomic neuroeffector junctions which are characterized by varicose nerve fibers that release neurotransmitters onto effector cells that do not contain a post-synaptic specialization, but do have neurotransmitter receptors on their cell membrane.²⁴ As mentioned, many of the nerves in scWAT parenchyma were varicose suggesting that they may be largely releasing neurotransmitter and neuropeptide along the length of their axons (“en passant”), diffusing onto nearby adipocytes. NANs would accordingly serve a specialized function requiring more targeted synaptic release versus the diffuse release that occurs en passant, or may be sensory nerve endings responding to cell size or other acute signals in an interoceptive manner that has yet to be explored. The distinguishing characteristics between an adipocyte which forms a nexus and one that does not are still a mystery, but is supported by the vast literature describing numerous subtypes of scWAT mature adipocytes. The mechanisms that induce nexus formation also remain to be determined.

By characterizing the glia associated with NANs, such as the SCs that appear similar to the terminal SCs observed at the NMJ, we hoped to further develop our understanding of NANs as well as the important functions SCs serve that are unique to adipose tissue. NCAM labeling indicated that these NAN terminals are unmyelinated, but this was undermined by the NCAM antibody’s apparent lack of binding specificity within a tissue environment. The presence of SCs immediately preceding the NAN suggests the alternative that the axons leading to each NAN are myelinated, as we have demonstrated that scWAT SCs are mostly myelinating and tend to be localized to myelinated structures. Regardless, the consistency with which SCs are localized to NANs hints to their importance for maintaining and/or eliciting these connections. By drawing comparisons with the NMJ (a well-described peripheral nerve terminal), we hoped to tease out potential functional similarities shared between the supporting glia at each terminal. Terminal SCs (tSCs) are crucial for maintaining NMJs, so we hypothesized that tSCs would be present at the terminal junction of the NAN as well. However, the terminal SCs of the NAN were different than the NMJ, in that they were only observed in the preceding axon leading to the NAN and not at the terminal itself. These SCs still may be functioning as tSCs in synapse maintenance and nerve repair, but potentially the long-term maintenance provided by abundant tSCs as required by the NMJ may not be required by NANs. This could suggest that NANs are highly plastic and do not form life-long connections.

The frequency of nerve myelination within adipose tissue, and presence of both mSCs and nmSCs subtypes (Figure 4), demonstrates a role for SCs in adipose nerve maintenance and function and makes them a potential target of dysregulation during adipose neuropathy. We have shown that the scWAT of BTBR ob/ob MUT mice became neuropathic, with observed demyelination and fragmentation of small fibers in the tissue. In addition, these obese mice also underwent the most changes in SC-related genes across changing energy balance states, specifically exhibiting upregulation of Krox20 (Figure 8). Krox20 has been identified as the main regulator of SC myelination during the pro-myelination phase, and plays a key role in interacting with Sox10 during the formation of nodes of Ranvier. Krox20 also acts as a transcription factor for the production of MPZ,^35,53,54 thus emphasizing Krox20’s importance in myelination of axons.^37,55,56

We also observed a significant upregulation of Krox20 in BL6 mice with age and in humans with obesity (Figure 8). Interestingly, in each instance, Krox20 increased without a correlative increase in Mpz expression (Figure 8). We hypothesize that upregulation of Krox20 may be a compensatory response of adipose mSCs to attempt to increase myelin production in neuropathic environments. Additionally, an increase in Krox20 expression would be expected if mature SCs were transdifferentiating into repair SCs⁵⁷ in response to obesity or age-related neuropathy. The observed decrease in Krox20 expression may then indicate an impaired ability to produce repair SCs in neuropathic states. Further research is needed to identify environmental signals that may be preventing functional myelination in these models despite robust Krox20 activation, or whether Krox20 may be functioning in other non-standard ways in the tissue. Additional WAT-specific SC markers may also exist and will be revealed by new single-cell transcriptomic data.

A hallmark feature of SCs is their ability to shift toward a repair phenotype in response to injury, releasing neurotrophic factors such as BDNF. Previous studies have demonstrated the decreased ability of aged mice to regenerate nerves post-injury when compared to younger animals, as a result of decreased c-Jun expression.³⁹ Diseases such as CIDP are characterized by a loss of SC plasticity.⁵⁸ Our qPCR of c-Jun in scWAT revealed almost no difference between young and old mice while displaying upregulation in obese BTBR ob/ob mice (Figure 8). In order to promote Wallerian degeneration, c-Jun is an inhibitor of myelination,⁵⁹ and recent studies found that SC mitochondrial dysfunction may drive c-Jun expression that contributed to demyelination.⁶⁰ The upregulation of c-Jun in obese BTBR ob/ob mice may be contributing to the demyelination of small fibers despite high Krox20, or alternatively, as a means for repairing the neuropathy that has already occurred.

We have previously shown that obese BTBR ob/ob mice display neuropathy in scWAT^7,42 and demonstrated a specific decline in scWAT neurovascular innervation.^7,42 Here, we have provided evidence that this neuropathy likely begins with the small parenchymal nerve fibers and is associated with deteriorating myelin sheaths, fitting with the clinical observations of small fiber neuropathy in diabetic human patients. It is still unclear if demyelination is causing the axonopathy. Interestingly, in BTBR ob/ob mice, we observed deterioration of myelin sheaths labeled with MPZ as well as a decrease in small sympathetic fibers. This suggests that there is axonopathy of the sympathetic fibers that may be independent from the demyelination. Whether or not the neuropathy affects both sensory and sympathetic nerves equally is also unclear. The downregulation of Mpz correlated with a downregulation of Sox10 and decreased SOX10 protein expression, but whether this is a loss of just mSCs, or both mSCs and nmSCs is unclear. The drastic reduction of small fiber nerves in inguinal scWAT concordant with demyelination of those that remain—thus rendering them functionally compromised—is evidence for a causative role in the decreased browning potential and decreased non-shivering thermogenesis that has been observed in obese mice previously.^7,61 As such, developing methods to maintain nerve fiber myelination in obese and diabetic states provides an intriguing avenue for combatting energy expenditure imbalance in humans. One could imagine that therapies capable of preventing or reducing the loss of adipose tissue resident SCs, or promoting their local function as repair SCs, may protect from nerve loss and demyelination, ultimately contributing to healthier metabolism, energy balance, and thermogenesis.

In conclusion, we have now characterized SCs in scWAT, their association with myelinated and unmyelinated nerves, and localization to synaptic vesicle-containing terminal nerve structures or NANs. Most importantly, we have provided evidence that obesity/diabetes-related adipose neuropathy is a small fiber demyelination, with loss of small fiber innervation density and a decrease in SC markers—together illustrating the importance of SCs in the maintenance of adipose tissue peripheral nerves and thereby healthy metabolism.

Limitations of the study

Analyses of tissue SC subtypes are challenging, given the large size and “stickiness” of SCs. Prior studies have demonstrated the presence of adipose SCs in single-cell transcriptomics data, and here, we validate the presence of SC subtypes in adipose by FACS and MACS, but the accurate quantification of these cell types may be hindered by methodologies that isolate the cells from the tissue. While we observe demyelination of adipose nerves in diabetic mice, this may have occurred in response to axonopathy, and we did not capture earlier timepoints when changes to SC numbers and phenotypes could have contributed to the onset of neuropathy. Finally, while we observe changes to SC markers across altered energy balance states, we have not yet correlated these with changes to SC behaviors in the tissue that may have direct impacts on nerve function.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
MPZ	Abcam	Cat#ab31851; RRID:AB_2144668
SOX10	Abcam	Cat#ab227680; RRID:AB_2927464
CGRP	EMD Millipore	Cat#PC205L; RRID:AB_2068524
TH	EMD Millipore	Cat#AB152; RRID:AB_390204
NCAM	Millipore	Cat#AB5032; RRID:AB_2291692
SV2	Developmental Studies Hybridoma Bank	Cat#SV2; RRID:AB_2315387
2H3	Developmental Studies Hybridoma Bank	Cat#2H3; RRID:AB_531793
TUBB3 conjugated to Alexa Fluor 488	Abcam	Cat#ab195879
MBP conjugated to Alexa Fluor 555	Cell Signaling	Cat#84987
Anti-CD45-BV421	Biolegend	Cat#103133; RRID:AB_10899570
Anti-O4-APC	Miltenyi Biotec	Cat#130-096-670; RRID:AB_2847907
Anti-p75-VioBright FITC	Miltenyi Biotec	Cat#130-110-115; RRID:AB_2656844
Anti-Pref1	Cell Signaling	Cat#2069; RRID:AB_2092685
β-Tubulin	Cell Signaling	Cat#2146BC; RRID:AB_2210545
Cyclophilin B	Abcam	Cat#16045; RRID:AB_443295
Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 555	Thermofisher	Cat#A-21428
Goat anti-Rabbit IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor Plus 594	Thermofisher	Cat#A32740
Goat anti-Rabbit IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor Plus 647	Thermofisher	Cat#A-32733
Goat anti-Mouse IgG1 Cross-Adsorbed Secondary Antibody, Alexa Fluor 488	Thermofisher	Cat#A-21121
Anti-rabbit HRP-linked secondary antibody	Cell Signaling	Cat#7074; RRID:AB_2099233
Chemicals, peptides, and recombinant proteins
Paraformaldehyde	Sigma-Aldrich	Cat#P6148
Glutaraldehyde solution	Sigma-Aldrich	Cat#G5882
Osmium tetroxide	Ted Pella	Cat#18456
Uranyl acetate	Ted Pella	Cat#19481
Ethanol	Decon Labs	Cat#2716
Acetone	Millipore Sigma	Cat#AX0115P-1
Toluidine Blue	Electron Microscopy Sciences	Cat#22050
EMBed 812 resin	Electron Microscopy Sciences	Cat#14900
Dodecenyl Succinic Anhydride	Electron Microscopy Sciences	Cat#1370
N-Methylaniline	Electron Microscopy Sciences	Cat#19000
Benzyl-N,N-dimethylamine	Electron Microscopy Sciences	Cat#11400
10X PBS Solution	Teknova	Cat#P0496
50% Dextrose	Hospira	Cat#0409664802
Bovine Serum Albumin	Sigma-Aldrich	Cat#A4503
Typogen Black/Sudan Black	Sigma-Aldrich	Cat#199664
Triton X-100	Bio-Rad Laboratories	Cat#1610407
DAPI, dilactate	Sigma-Aldrich	Cat#D9564
EMS Glycerol Mounting Medium With DABCO	Electron Microscopy Sciences	Cat#17989-5
Red Blood Cell Lysing Buffer Hybri-Max	Sigma	Cat#R7757
Dulbecco’s Modified Eagle’s Medium (High Glucose)	Sigma	Cat#D5671
Collagenase A	Sigma-Aldrich	Cat#10103586001
Dispase II	Sigma	Cat#D4693
0.25% Trypsin-EDTA (1X)	Gibco	Cat#25200-056
TRIzol Reagent	Life Technologies	Cat#15596026
SYBR Green	Bio-Rad	Cat#1725271
Sso Advanced Universal Probes Supermix	Bio-Rad	Cat#12001950
Critical commercial assays
Zymo DirectZol RNA extraction kit	Zymo	Cat#R2052
RNeasy Micro Kit	Qiagen	Cat#74004
High-Capacity Synthesis Kit	Applied Biosystems	Cat#4368813
Experimental models: Organisms/strains
Mouse: C57BL/6J	The Jackson Laboratory	Cat#000664
Mouse: PGP9.5+/-: C57BL/6-Tg(Uchl1-EGFP) G1Phoz/J	The Jackson Laboratory	Cat#022476
Mouse: BTBR.Cg-Lepob/WiscJ The Jackson Laboratory Cat#004824	The Jackson Laboratory	Cat#004824
Oligonucleotides
Primers for c-Jun (mouse), see Table S1	This paper	N/A
Primers for Krox20(mouse), see Table S1	This paper	N/A
Primers for p75ntr (mouse), see Table S1	This paper	N/A
Primers for Mpz (mouse), see Table S1	This paper	N/A
Primers for O4 (mouse), see Table S1	This paper	N/A
Primers for Ppia (mouse), see Table S1	This paper	N/A
Primers for c-Jun (human), see Table S1	This paper	N/A
Primers for Krox20(human), see Table S1	This paper	N/A
Primers for p75ntr (human), see Table S1	This paper	N/A
Primers for Mpz (human), see Table S1	This paper	N/A
Primers for O4 (human), see Table S1	This paper	N/A
Primers for Ppia (human), see Table S1	This paper	N/A
Probe for Ppia (qMmuCEP0043519)	Biorad	Cat#10031228
Probe for Ngf (qMmuCIP0042317)	Biorad	Cat#12001950
Probe for Bdnf (qMmuCEP0058759)	Biorad	Cat#12001950
Probe for Ntf3 (qMmuCEP0042141)	Biorad	Cat#12001950
Software and algorithms
FIJI		https://imagej.net/software/fiji/downloads
GraphPad Prism 9	GraphPad	https://www.graphpad.com/
LAS X Leica Microsystems	Leica Microsystems	https://www.leica-microsystems.com/products/microscope-software/p/leica-las-x-ls/
Nikon Elements Advanced Research Package	Nikon	https://www.microscope.healthcare.nikon.com/products/software/nis-elements/nis-elements-advanced-research
Microsoft PowerPoint 2019	Microsoft	https://www.microsoft.com/en-us/download/office.aspx
Other
Touch Test Sensory Evaluator, Kit of 20	Stoellting Co.	Cat#58011
OneTouch UltraMini (Glucometer)	LifeScan	Cat#5388591301
OneTouch Ultra Test Strips	LifeScan	Cat#LFS20244Z
75 × 51 mm Glass Slide, 1.2 mm Thick	Electron Microscopy Sciences	Cat#71862-01
Cover glass, 48 × 60mm, 1.5 thick	Brain Research Laboratories	Cat#4860-1.5D
CFX384 real-time PCR detection system	Bio-Rad	Cat# 1855484
gentleMACS Octo Dissociator with Heaters	Miltenyi Biotec	Cat#130-096-427
Large Cell Column	Miltenyi Biotec	Cat#130-042-202
MiniMACS Separator	Miltenyi Biotec	Cat#130-042-102
Anti-Cd45 microbeads	Miltenyi Biotec	Cat#130-052-301; RRID:AB_2877061
Anti-Rabbit IgG microbeads	Miltenyi Biotec	Cat#130-048-602; RRID:AB_244362
Anti-O4 microbeads	Miltenyi Biotec	Cat#130-042-202

Resource availability

Lead contact

The datasets generated during and/or analyzed during the current study are available from the corresponding author and lead contact Kristy Townsend (kristy.townsend@osumc.edu) upon reasonable request.

Materials availability

No applicable resources or reagents were generated or analyzed during the current study.

Experimental model and subject details

Mice

All mice were obtained from The Jackson Laboratory (Bar Harbor, ME) and/or bred at our mouse facilities at University of Maine and The Ohio State University. Animals were housed in a temperature-controlled environment, kept on a 12 h light-dark cycle, and allowed access to food and water ad libitum (unless stated otherwise for a particular study). For all studies animals were euthanized using CO₂ followed by cervical dislocation. All procedures were performed in compliance with the National Institute of Health Guide for the Care and Use of Laboratory Animals and were approved by an Institutional Animal Care and Use Committee.

PGP9.5-EGFP^+/- reporter mice

Male and female PGP9.5-EGFP^+/- (C57BL/6-Tg(Uchl1-EGFP)G1Phoz/J, Stock # 022476) pan-neuronal reporter mice were used for microscopy experiments to investigate adipose innervation. Mice were housed 2-5 a cage, fed a standard chow diet and were aged 10-26 weeks prior to tissue collection.

Sedentary/exercised mice:

Adult (12-15 week old) C57BL/6J (Stock #000664) male mice, age and body weight matched, were assigned to either sedentary or exercised groups. Animals were single housed in running wheel cages that allowed ad libitum access to running for a period of 7 days. Wheel running was monitored using odometers (magnets were placed on the outer arm of the running wheels with the sensor attached to the inside of the cage). Sedentary (control) animals were single caged with locked running wheels for the same period.

BTBR ob/ob mice

BTBR +/+ (WT) and BTBR ob/ob (MUT) mice (BTBR.Cg-Lep^ob/WiscJ, Stock # 004824) were fed a standard chow diet, housed two to a cage, and aged at least 12 weeks, when they exhibit a strong phenotype (including obesity and diabetes), for imaging, RT-qPCR, and Western Blot experiments. For MACS isolation of SCs, mice were aged to 10 weeks.

Cold exposure experiments:

Adult (8 week-old) male C57BL/6J (Stock #000664) mice were housed two to a cage, and either maintained at room temperature (RT), continuously cold exposed (at 5ºC), or kept at thermoneutral temperature (30ºC) for 3 days with ad libitum access to food and water. All cold exposure experiments occurred in a diurnal incubator with 12 h light/dark cycle and humidity control (Caron, Marietta, OH, USA).

Diet-induced obesity mice

Adult (11 week old) male C57BL/6J (Stock #000664) mice, housed 2-3 to a cage, were fed either chow control (LabDiet® 5LG6) or a 58% high fat diet (HFD) (Research Diets Cat #D12330) ad libitum for 19 weeks. Animals were euthanized with CO₂ and inguinal scWAT was collected and frozen before being processed for qPCR.

Aged mice

C57BL/6J (Stock #000664) male mice were housed 2-4 in a cage and aged to 15 weeks (young) and 75 weeks (aged). Tissues where snap-frozen in liquid nitrogen upon collection for gene expression analysis using RT-qPCR. For ageing related FACS studies, C57BL/6J male mice were aged to 4 months, 8 months, and 15 months.

Human tissues

Human abdominal scWAT were obtained from 23 individuals, 11 were healthy lean (BMI ≤ 25) and 12 were healthy obese (BMI >30) individuals who participated in a cross-sectional study at the Translational Research Institute at AdventHealth. The study protocol was approved by the AdventHealth Institutional Review Board and was in line with the Declaration of Helsinki. Samples were collected from informed and consenting male (lean N = 2, 18-23 years old; obese N=3, 35-64 years old) and female (lean N=9, 23-49 years old; obese N=9, 25-55 years old) subjects during elective surgery. Abdominal scWAT was obtained under local anesthesia [18]. Tissue was immediately snap-frozen in liquid nitrogen and then processed for RNA extraction. RNA was used to measure gene expression by real-time qPCR. BMI = kg/m².

Method details

Von Frey neuropathy testing

Von Frey testing and analysis was performed for Diet-induced obesity cohorts using the methods described in Blaszkiewicz et al.⁷ Mice were placed on a grid platform, followed by a 20-minute acclimation period. Five filaments, corresponding to 4, 2, 1, 0.4, 0.02 –grams of force (g), were individually applied to the hind paw. Paw removal or licking was considered a positive response, while no licking was considered a negative response. Each filament strength test was performed in 5 cycles.

Glucose tolerance test (GTT)

Insulin sensitivity was measured for Diet-induced obesity cohorts after 15 weeks of HFD via a glucose tolerance test. Mice were fasted overnight (16h) and a small tail cut was made to measure baseline fasting blood glucose (mg/dL). Following baseline measurements, mice received an intraperitoneal injection of 50% dextrose (0.001 mL/g of BW), and blood glucose measurements were taken at 15-, 30-, 60-, and 120-minutes post-injection.

Whole mount immunofluorescence (IF)

Whole inguinal and axillary scWAT depots were carefully removed to remain fully intact and fixed in 2% PFA (Sigma, Cat#P6148) for 16hrs (or 24hrs for obese BTBR ob/ob) at 4°C. Tissues were Z-depth reduced and subsequently blocked in 2.5% BSA / 1% Triton X-100 / 1X PBS for 24hrs at 4°C and incubated in 0.1% Typogen Black (Sigma, Cat#199664) to reduce autofluorescence. BTBR ob/ob tissues did not receive autofluorescence quenching. Tissues were incubated in primary antibody solution for 48hrs at 4°C, rinsed in 1X PBS, and incubated in secondary antibody solution for 24hrs at 4°C. These steps were repeated for additional co-labeling. When desired, tissues were incubated in 100 ng/mL DAPI (Sigma-Aldrich, Cat#D9564) for 1hr at room temperature to label nuclei. Finally, tissues were mounted on glass slides and imaged. For additional details see² and the accompanying peer-reviewed protocol.⁶²

Confocal microscopy

Confocal micrographs were captured on a Leica Stellaris 5, laser scanning confocal microscope using LASX software. Fluorescent labels were excited with either a diode 405 nm laser (DAPI, Autofluorescence) or a white light laser: EGFP (499 nm), Alexa Fluor 555 (553 nm), Alexa Fluor Plus 594 (590 nm), Alexa Fluor Plus 647 (653 nm). Emission spectra were tuned specifically for each fluorophore or groups of fluorophores to reduce and eliminate crosstalk. Multiple channels were scanned sequentially, and all channels were line averaged 3 times. Photons were detected with Power HyD S detectors. Objectives included: HC PL APO 10x/0.40 CS2, HC PL APO 40x/1.30 OIL CS2, and HC PL APO 63x/1.40 OIL CS2. PinholeAiry 1.00 AU. Confocal zoom was applied to further increase magnification when necessary. Entire tissues were visually scanned, and representative images were captured at 2048 × 2048 pixel resolution as Z-stacks (1-6 μm step size) that were maximum intensity projected. LUTs were adjusted to improve structure visualization. Digital cross-sections were captured by utilizing the XZY scan function. One iteration of a 5-kernel median noise filter was applied. Image processing performed in Leica LASX software.

Deconvolution

In specific instances (stated in figure legends) images were acquired using Lightning (Leica) 3D deconvolution. Z-stack images were captured at Nyquist lateral and axial resolutions at 63X objective magnification with a PinholeAiry of 0.50 AU with varying confocal zooms applied. Adaptive deconvolution was calculated for a 1.4429 refractive index. Deconvolved z-stacks were maximum intensity projected and LUTs were adjusted to improve structure visualization. Image processing performed in Leica LASX software.

Epifluorescence microscopy

Epifluorescence micrographs were captured on a Nikon Eclipse E400 epifluorescence microscope using a Hamamatsu ORCA-Flash4.0 V2 Digital CMOS monochrome camera. Alexa Fluor 555 fluorophores were excited using a Cy3 filter cube. Objectives used: Nikon CFI Plan Fluor 20x/0.50 and Nikon CFI Plan Fluor 40x/0.75. Images were captured utilizing the extended depth of field (EDF) function; LUTs were adjusted to improve structural visualization. Post processing was performed in Nikon Elements BR software.

Neuromuscular junction immunostaining

Medial gastrocnemius muscle was fixed in 2% PFA for 2hrs and teased apart. Tissue was blocked in 2.5% BSA / 1% Triton X-100 / 1X PBS for 24hrs at 4°C. Tissues were incubated in a series of primary and secondary antibody solutions each lasting 24hrs and performed at 4°C on a rotator. Tissues were washed in 1X PBS between incubations. Primary antibodies SV2 and 2H3 were incubated simultaneously with α-bungarotoxin (BTX) conjugated to Alexa Fluor 555 (1 mg/mL, Thermofisher, Cat#B35451) to label pre- and post-synapse. This was followed by the secondary antibody goat anti-mouse IgG1 Alexa Fluor 488, and then the primary antibody MPZ followed by its secondary antibody goat anti-rabbit IgG Alexa Fluor 647 Plus. Finally, the tissues were incubated in 100 ng/mL for 1hr at room temperature to label nuclei and mounted on glass slides for imaging.

Antibodies used for immunostaining

Primary antibodies

MPZ (1:250, Abcam, Cat#ab31851); SOX10 (1:200, Abcam, Cat#ab227680); CGRP (1:200, EMD Millipore, Cat#PC205L); TH (1:250, EMD Millipore, Cat#AB152); NCAM (1:250, Millipore, Cat#AB5032); SV2 (1:250, Developmental Studies Hybridoma Bank, Cat#SV2); 2H3 (1:500, Developmental Studies Hybridoma Bank, Cat#2H3); SYN1 (1:200, Cell Signalling, Cat#5297); TUBB3 conjugated to Alexa Fluor 488 (1:200, Abcam, Cat#ab195879); MBP conjugated to Alexa Fluor 555 (1:500, Cell Signaling, Cat#84987).

Secondary antibodies

Goat anti-Rabbit IgG (H+L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 555 (1:500, Thermofisher, Cat#A-21428); Goat anti-Rabbit IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor Plus 594 (1:1000, Thermofisher, Cat#A32740); Goat anti-Rabbit IgG (H+L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor Plus 647 (1:250, Thermofisher, Cat#A-32733), Goat anti-Mouse IgG1 Cross-Adsorbed Secondary Antibody, Alexa Fluor 488 (1:500, Thermofisher, Cat#A-21121).

Transmission electron microscopy (TEM)

Samples were dissected and fixed in 2% PFA / 2.5% glutaraldehyde mixture. Samples were postfixed with 1% osmium tetroxide and then en bloc stained with 1% aqueous uranyl acetate, dehydrated in a graded series of ethanol, and embedded in Eponate 12 epoxy resin (Ted Pella Inc., Redding, CA). Ultrathin sections were cut with a Leica EM UC7 ultramicrotome (Leica microsystems Inc., Deerfield, IL) and collected on copper grids. Images were acquired with an FEI Technai G2 Spirit transmission electron microscope (Thermo Fisher Scientific, Waltham, MA) operating at 80kV, and a Macrofire (Optronics, Inc., Chelmsford, MA) digital camera and AMT image capture software.

Adipose stromal vascular fraction (SVF) dissociation

SVF from bilateral whole inguinal adipose depots was isolated as previously described [12]. Inguinal scWAT was quickly harvested and minced in DMEM (high glucose, serum free, pre-warmed in a 37ºC water bath) containing 2 mg/mL of Collagenase A (Sigma-Aldrich, Cat#10103586001) at a volume of 20 mL/g of tissue. Collagenase was added to warmed media immediately before tissue dissection. Inguinal scWAT was pooled bilaterally from each animal for FACS analysis of exercise and ageing cohorts. For FACS analysis at basal state, bilateral depots from 4 animals were pooled together after the dissociation step. Minced tissue in dissociation media was placed in 50 mL conical tube and transferred to a shaking warm water bath (90 rotations/min at 37ºC). Dispersion of cells was furthered via gentle vortexing and trituration using Pasteur Pipettes at various bores until full dissociation was achieved and floating adipocytes were visible. Samples were filtered through 100 μM cell strainers and rinsed with DMEM then centrifuged at 500 g for 10 min to separate adipocytes and SVF pellet. SVF pellet was incubated with 500 μl of red blood cell lysis buffer for 2 min on ice. Lysis was stopped by the addition of 2 mL of DMEM containing 5% FBS. Cells were centrifuged at 500 g for 5 min at 4°C and resuspended in 100 μL of FACS buffer (1X PBS with FBS and EDTA) for cell sorting.

Fluorescence-activated cell sorting (FACS)

For antibody labeling, 2 μL FACS block (BSA and FBS in 1X PBS) was added to SVF resuspended in FACS buffer (as described above) and then left to sit for 15-20 min on ice, after which 500 μL FACS buffer was added, and samples were spun at 1800 rpm for 5 min. Samples were resuspended in 100 μL FACS buffer with conjugated antibodies. Antibodies used included; Anti-CD45-BV421 (Biolegend, Cat. # 103133 (clone 30-F11)), Anti-O4-APC (Miltenyi Biotec, Clone O4), Anti-p75-VioBright FITC (Miltenyi Biotec, Clone REA648). Cells were washed 1-2 times by centrifugation at 1500 rpm for 5 min and then resuspended in 300 μL of FACS buffer. UltraComp eBeads (Invitrogen # 01-2222-42) were used for compensation controls. DAPI exclusion was used for viability. Sorting was performed on a BD™ FACS Aria II™ cell sorter or BD™ Influx equipped with a 100uM nozzle to accommodate Schwann cell size. Cells were gated on FSC/SSC, live cells and CD45 to exclude immune cells. Cells representing Schwann cell populations (CD45-p75+, CD45-p75+O4+, or CD45-O4-), were sorted into 400 μL of Trizol (Zymo, Irvine, CA, USA; Cat. #R2050-1-200). FACS performed sorts reported in this paper were performed at The Jackson Laboratory in Bar Harbor, ME and at The Nationwide Children’s Hospital Flow Cytometry Core Facility in Columbus, OH.

Magnetic-activated cell sorting (MACS)

From N=5 BTBR ob/ob and N=5 BTBR wild type mice, bilateral inguinal subcutaneous white adipose depots were pooled for each animal. Tissues were placed in an enzyme digest containing 2mg/mL collagenase A (Sigma-Aldrich, Cat#10103586001), 2mg/mL dispase II (Sigmam Cat#D4693) , and 0.05% Trypsin (Gibco, Cat#25200056). Tissues were dissociated for 30 minutes at 37°C using a gentleMACS Octo Dissociator with Heaters (Miltenyi, Cat#130096427). Cells were passed through a 100μm filter and spun down to pellet stromal vascular fraction (SVF). Cells were resuspended in buffer containing anti-CD45 microbeads (Miltenyi, Cat#130052301) and incubated for 15 minutes at 4°C. Cells were washed and incubated for 10 minutes at room temperature with anti-Pref1 antibody (Cell Signaling, Cat#2069) followed by incubation with anti-Rabbit IgG microbeads (Miltenyi, Cat#130048602) for 15 minutes at 4°C. Cells were passed through Large cell columns (Miltenyi, Cat# 130042202) in a magnetic field and unlabeled flow-through (Cd45-, Pref-1-) was collected. Collected cells were incubated with anti-O4 microbeads (Miltenyi, Cat#130096670). Cells were passed through a Large cell column (Miltenyi, Cat# 130042202) in a magnetic field and O4+ Schwann cells were collected from column in 300 μl of Trizol.

RNA extraction and real-time quantitative PCR (qPCR)

Zymo DirectZol RNA extraction kit (Zymo, Irvine, CA, USA; Cat. #R2052) was used for RNA extraction from whole tissue and the RNeasy Micro Kit (Qiagen, Cat. # 74004) was used to isolate RNA from sorted cells. RNA yield was determined using a Nanodrop and cDNA was synthesized using High-Capacity Synthesis Kit (Applied Biosystems, Foster City, CA, USA; Cat. #4368813). Real-time quantitative polymerase chain reaction was performed using SYBR Green (Bio-Rad, Cat#1725271) on a CFX384 real-time PCR detection system (Bio-Rad, Hercules, CA, USA). Gene expression was normalized to housekeeper gene Ppia and Sox10 for analysis. Primers used for qPCR are listed in Table S1.

RNA extraction and cDNA synthesis from MACS isolated Schwann cells is described above. To detect gene expression changes in neurotrophic factors (Ngf, Bdnf, Ntf3). Real-time quantitative polymerase chain reaction was performed using Sso Advanced Universal Probes Supermix (Bio-Rad, Cat#1725281) on a CFX384 real-time PCR detection system (Bio-Rad, Hercules, CA, USA). Neurotrophic factors were measured with FAM-labeled Taqman PCR probes (Biorad Cat#12001950); Ngf (qMmuCIP0042317), Bdnf (qMmuCEP0058759), and Ntf3 (qMmuCEP0042141). Each sample was run in duplicate and multiplexed with HEX-labeled probe for housekeeper gene Ppia (Biorad, Cat#10031228, qMmuCEP0043519).

Western blot (WB)

Protein expression was measured by western blotting analysis of scWAT lysates. Whole adipose depots were homogenized in RIPA buffer with protease inhibitors in a Bullet Blender. A Bradford assay was performed to measure total protein from which equal concentrations of protein lysates were prepared in Laemmli buffer using 1X PBS as diluent. 30 μg of protein were loaded per lane of a 10% polyacrylamide gel, and following gel running, proteins were transferred to PVDF membrane and incubated with 10% Roche Blocking Reagent for 1 hr at room temperature prior to antibody incubation. The membrane was bisected horizontally at 37 kDa to avoid the need for stripping proteins later. Primary antibodies included: TH (62 kDa, 1:250, EMD Millipore, Cat#AB152); SOX10 (49 kDa, 1:400, Abcam, Cat#ab155279); MPZ (25 kDa, 1:1000, Abcam, Cat#ab31851); β-Tubulin (55 kDa, 1:1000, Cell Signaling Technology, Cat#2146BC); Cyclophilin B (21 kDa, 1:40,000, Abcam, Cat#16045). Primary antibodies were incubated overnight at 4°C on a rotator with gentle agitation. Membranes were rinsed with 1X TBS-T and then incubated in anti-rabbit HRP-linked secondary antibody (1:3000, Cell Signaling Technology, Cat#7074) for 1hr a room temperature. Blots were visualized with enhanced chemiluminescence (ECL; Pierce) on a Syngene G:BOX. TH and SOX10 expression were normalized to Cyclophilin B, and MPZ was normalized to β-Tubulin and quantified by densitometry in Fiji [19].

BTBR ob/ob scWAT nerve bundle processing.

A prominent nerve bundle traversing into the inguinal scWAT depot was excised from each mouse and samples were fixed in a 2% PFA / 2.5% glutaraldehyde mixture, postfixed with 1% osmium tetroxide, and then en bloc stained with 1% aqueous uranyl acetate. The samples were dehydrated in a graded series of ethanol and embedded in Eponate 12 epoxy resin (Ted Pella Inc., Redding, CA). One-micron thick sections were cut with a Leica EM UC7 ultramicrotome (Leica microsystems Inc., Deerfield, IL) and stained with toluidine blue. Images of nerve bundle cross-secitons were acquired with a Zeiss Axioskop microscope (Carl Zeiss Microscopy, LLC, White Plains, NY) using Zeiss Achroplan 20X/0.45 Ph2 and Zeiss Plan-Apochromat 63X/1.4 Oil Ph3 objective lenses. Images were captured with a Nuance multispectral imaging camera (PerkinElmer, Waltham, MA).

BTBR ob/ob scWAT whole depot imaging and quantification

Intact whole axillary scWAT depots were excised and immunostained for TH. Alexa fluor plus 594 was excited at 590 nm, and emitted photons were detected for 600-700 nm. Laser intensity (10%) and detector gain (8%) remained constant for all images. The entirety of each whole tissue was visually scanned at 10X objective magnification (1.00 confocal zoom) and a representative 5x5 tiled (22.85 mm²) area was captured as a series of Z-stacks (10 μm step size) extending through the full thickness of each tissue (120-230 μm) and were maximum intensity projected. Images were processed for quantification in Fiji [20] by first applying background subtraction (50 pixel rolling ball radius). Next, a threshold (30-255) was applied to further remove autofluorescence. Area of remaining pixels was measured.

Quantification and statistical analysis

For all animal experiments, mice were body weight matched and then randomized to experimental or control groups to mitigate differences in starting body weights. ROUT outlier test (Q=1%) was performed on raw data to identify and remove statistical outliers within data sets. Von Frey was analyzed by multiple t-tests. GTT testing was analyzed by a two-way ANOVA with Tukey’s multiple comparison test for mixed models. FACS data was analyzed by two-tailed Student’s t-test or two-way ANOVA with Tukey’s multiple comparison test for mixed models. qPCR data was analyzed for each gene individually by either unpaired two-tailed Student’s t-test or one-way ANOVA with multiple comparisons. Linear regression analysis was performed with Goodness of Fit measured by R-squared and the significance of slope determined by F-test, when comparing two linear regressions statistical difference between slopes and Y-intercepts were analyzed. Western blot was analyzed by unpaired Student’s t-test. G-ratio was analyzed by unpaired Student’s t-test. Neurite density was analyzed as unpaired Student’s t-test. All error bars are SEMs. Statistical calculations for determining significance were calculated in GraphPad Prism software (La Jolla, CA, USA). For all figures statistically significant p-values are displayed on each graph. n. s. = not significant.

Additional resources

Protocols used for this publication can be found on Protocols.io at https://doi.org/10.17504/protocols.io.brs2m6ge.

Published: February 13, 2023

Data and code availability

• •All data reported in this paper will be shared by the lead contact upon request.
• •No original code was generated for this paper.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.