Authors: Jake W. Willows, Magdalena Blaszkiewicz, Kristy L. Townsend
Categories: Article
Source: Comprehensive Physiology
Doi: 10.1002/cphy.c220030
Authors: Jake W. Willows, Magdalena Blaszkiewicz, Kristy L. Townsend
The sympathetic nervous system (SNS) is a crucial arm of the peripheral nervous system (PNS) and includes catecholaminergic neurons that release norepinephrine (NE) onto numerous effector tissues and organs in the body. SNS innervation of both white (WAT) and brown adipose tissue (BAT) is clearly essential for proper tissue function and metabolic control, as decades of surgical, chemical, and genetic denervation studies have demonstrated. Despite our vast knowledge about adipose sympathetic innervation, especially in the context of cold-stimulated browning and thermogenesis that are under SNS control, newer data now provide a nuanced view of the SNS supply to adipose, including its regulation by local neuroimmune cells and neurotrophic factors, the co-release of modulatory neuropeptides along with NE, the importance of local SNS drive to adipose versus systemic increases in circulating catecholamines, and the long-overlooked interplay between adipose sympathetic and sensory nerves. This article brings a modern view to the regulation of sympathetic innervation patterns in WAT and BAT, how to image and quantify the nerve supply, contributions of adipose SNS to tissue functions, and how adipose tissue nerves respond to tissue remodeling and plasticity with changing energy demands.
The sympathetic nervous system (SNS) carries out numerous systemic functions in the body and is widely recognized as being responsible for the “fight or flight” response via the sympathetic adrenomedullary system (231). In brief, the hypothalamus (Ht) is stimulated by the amygdala when there is perceived danger. This signal is then sent from the Ht to the adrenal gland via preganglionic sympathetic fibers that form synapses onto chromaffin cells in the adrenal medulla. Synaptic release of the neurotransmitter acetylcholine causes the chromaffin cells to secrete epinephrine and norepinephrine (NE) into circulation, which acts systemically on numerous effector organs/tissues to elicit the acute fight or flight response (248). This well-studied SNS response overshadows the myriad other roles of the SNS, including in maintaining energy homeostasis, even though the average human spends only a small fraction of their lifetime under sufficient duress to elicit the fight or flight SNS mechanisms. While the parasympathetic nervous system (PSNS) is colloquially responsible for the “rest and digest” response—a counterbalance to the “fight or flight” response—it is actually the SNS that responds to dietary nutrients and nutrient storage by mediating numerous aspects of energy balance. Such SNS metabolic functions include regulating basal cardiac output, maintaining body temperature, and mobilizing nutrients. These functions are maintained by widespread SNS innervation of tissues and organs, and controlled release of SNS neurotransmitters and neuropeptides, including NE, neuropeptide Y (NPY), and adenosine triphosphate (ATP) (43).
The brain region responsible for integrating energy balance signals and triggering an outflow of sympathetic drive is the hypothalamus (Ht), which acts in conjunction with numerous extrahypothalamic brain regions via neuromodulatory pathways, including via the Raphe Nuclei, Locus Coeruleus, and brainstem regions (i.e., the nucleus of the solitary tract) (170, 178, 183, 252). The central nervous system (CNS, or brain and spine) regulates appetite and energy expenditure and maintains energy balance (energy input/energy output) in large part through direct communication with the body’s lipid-storing adipose tissue (252). SNS drive to adipose via peripheral nerve communication is coordinated by hypothalamic signals and can regulate adipose tissue lipolysis, adipogenesis, browning [development of inducible brown adipocytes in white adipose tissue (WAT) depots], and thermogenesis, which are discussed in greater detail below. In addition to the CNS receiving numerous circulating signals such as endocrine hormones (leptin, insulin, adiponectin) and nutrients such as free fatty acids (FFAs) released from adipose stores, the neural communication between adipose tissue and the CNS is bidirectional in nature, with a circuit comprised of efferent sympathetic output from the CNS to adipose, and afferent sensory input from the adipose tissue back to the brain. This bidirectional circuit intersects at co-labeled cell bodies in the spinal cord (short-feedback loop) and in the brain (long-feedback loop) that have been revealed by neural tracing studies (223, 224). While the focus of this article is on the SNS, it should be noted that sensory nerves also constitute a large percentage of the nerve density in adipose tissue, and have recently been demonstrated to be definitively important for regulation of sympathetic drive (267, 275), as had been indicated in prior studies (223, 224). How sensory nerves regulate sympathetic nerve activity (SNA) in adipose is still unclear. Prior pharmacological studies found that sensory nerves facilitated an increase in SNA (76, 198, 236) while recent studies using genetic approaches have found that sensory nerve activity may dampen adipose SNS activity (133, 267).
SNA is the electrophysiological signal of stimulated nerves as measured by electrophysiology or neurotransmitter release. Sympathetic tone is maintained by the basal SNA present in an organism/tissue which can be modified through active stimulation of SNA (sympathetic drive). Elevated sympathetic drive and SNA result in increased SNS outflow to tissues such as adipose and facilitates necessary metabolic processes. The regulation of lipolysis, adipogenesis, browning, and thermogenesis by the SNS in WAT and brown adipose tissue (BAT) of humans and rodents has been reviewed numerous times previously (19, 21, 29, 52, 143, 183). Here, we provide a comprehensive update and synthesis of the available literature, including a fundamental overview needed to appreciate relevant findings, such as the most recent developments regarding SNS innervation of adipose.
In rodents, WAT exists both subcutaneously and viscerally, in discreet, mostly bilateral depots (60, 290). The prominent subcutaneous WAT (scWAT) depots are the inguinal scWAT (ing-scWAT; above the lower legs), axillary scWAT(ax-scWAT behind the front legs), and interscapular scWAT (i-scWAT; found above the interscapular BAT) depots. The visceral depots include the mesenteric WAT (mesWAT; along the intestines), epididymal/perigonadal WAT (pgWAT), perirenal WAT (prWAT), retroperitoneal WAT (rpWAT; behind the kidneys against the peritoneum), perivascular adipose tissue (PVAT; along blood vessels), epicardial adipose tissue (EAT; incased by visceral pericardium lined directly adjacent to myocardium), and pericardial/thoracic adipose tissue (PAT, surrounds parietal pericardium) [although, EAT and PAT are often considered the same depot in mice (13, 90, 263, 290)]. BAT is deposited primarily into two bilateral interscapular lobes (iBAT), although classical BAT is also localized to other lesser-described regions throughout the body (290). Some WAT depots, like ing-scWAT, and prWAT are also more prone to browning, and browning patterns (such as in response to cold stimulation) can be sex specific, as female pgWAT is more responsive to browning cues than male pgWAT (142). PVAT shares characteristics of both WAT and BAT and each depot of PVAT lands differently on the WAT-to-BAT spectrum (119).
Human scWAT is distributed in amorphous depots and is primarily distributed abdominally, and gluteofemorally. Human intermuscular adipose tissue (IMAT) is distinct from scWAT and visceral adipose and appears to play a role in regulating muscular insulin sensitivity (225). Human intra-abdominal visceral adipose tissue is organized into many of the same depots as rodents; mesWAT, retroWAT, EAT, PAT, and pgWAT, with the additional omental WAT (oWAT) found in the omentum (59). Human classical BAT is primarily deposited in the upper torso cervically and supraclavicularly (151), but most of the human brown adipocytes are the beige/brite type that appears in white fat depots. In humans, there is evidence suggesting that PAT exhibits signs of browning, while EAT appears purely white (90).
It is also worth noting that there are several other adipocyte anatomical niches that include bone marrow adipose tissue (BMAT; or “yellow adipose”), adipose within the dermal skin layer (dWAT), around the mammary glands (“pink adipose”), and within the epineurium of nerve bundles (156, 186, 217). Each of these may receive distinct neural inputs or cross talk with nerves in unique manners.
Leptin is an adipose-derived hormone and adipokine that circulates at levels roughly similar to adipose mass and is responsible for mediating lipolysis (breakdown of stored lipids into FFAs) through hypothalamic actions that result in the stimulation of the SNS (288). Leptin is synthesized in adipocytes where its production and secretion into surrounding vasculature are regulated by hormonal signals that are dependent on fat mass and food intake (150, 288), thus leptin circulates at levels roughly equivalent to adipose mass. Once in circulation, leptin eventually crosses the blood-brain barrier (BBB; or reaches hypothalamic neurons without much of a barrier such as those near the median eminence, a circumventricular organ) and is recognized by leptin receptors (LepRs) in the arcuate nucleus of the Ht (i.e., central leptin-signaling) (94, 265). LepRs on hypothalamic nuclei, such as anorexigenic pro-opiomelanocortin (POMC) neurons and orexigenic agouti-related peptide (AgRP) neurons, become activated and trigger melanocortin signaling and downstream release of brain-derived neurotrophic factor (BDNF) (265). BDNF is one signal that ultimately increases SNS drive-through actions in the hypothalamus, but BDNF also has peripheral actions and can increase nerve fiber density in WAT and BAT directly, together serving to stimulate lipolysis (265, 288). While LepRs are predominately located within the CNS, peripheral organs including adipose tissue also express LepRs (i.e., peripheral leptin-signaling) which function mostly independent from the nervous system (26, 33, 169).
Upon CNS stimulation by circulating leptin, Norepinephrine (NE) is released from sympathetic nerve endings which acts on beta-adrenergic receptors (β-ARs) on adipocytes to facilitate the downstream intracellular signaling cascades that are responsible for lipolytic enzyme activities (8, 114). It should be noted that lipolysis is not always a beneficial calorie-dispensing process—the FFAs that are produced by the action of lipolytic enzymes like hormone-sensitive lipase (HSL) must also be utilized appropriately (e.g., oxidized in mitochondrial metabolism) to avoid high circulating FFAs that can be lipotoxic. Indeed, chronically high levels of circulating FFAs due to obesity will become toxic to nonadipose tissues (83, 255), and impaired storage of FFAs in triglycerides of adipocyte lipid droplets leads to ectopic lipid accumulation in other tissues and organs like heart, muscle, and liver, which becomes pathophysiological. This poses a problem for adipose resident nerves which reside in a lipid-laden tissue but may lack the required protection from lipotoxicity. A high lipid, lipotoxic environment is thought to enhance the development of peripheral neuropathy with obesity and diabetes (202). For a comprehensive review of SNS-mediated lipolysis, see previous coverage (19, 21).
The SNS is also responsible for regulating fat cell number in WAT and BAT through the proliferation and differentiation of adipocyte progenitor cells and preadipocytes into mature adipocytes (35, 96). NE released from sympathetic nerve terminals acts on β1-ARs on progenitor cells and preadipocytes to inhibit proliferation and differentiation in WAT (228). Studies that employed chemical denervation of WAT sympathetic nerves using localized injections of the catecholaminergic-specific neurotoxin 6-hydroxydopamine (6-OHDA) reported an increase in hyperplasia in the tissue, with an over 400% increase in bromodeoxyuridine (BrdU) incorporation in studies using Siberian hamster (92). Knock-out of the neurotrophic factor Bdnf in myeloid lineage cells also resulted in scWAT hyperplasia along with a genetic denervation of the tissue (30), further cementing the SNS as a negative regulator of adipocyte proliferation. Conversely, it has been found that increased sympathetic output to adipose following cold exposure induces proliferation of brown adipocytes in BAT through activation of β1-ARs (40, 95, 97). The corresponding signaling pathways downstream of β1-ARs leading to adipogenesis have not yet been fully described, but include PGC1a, PPARs/CEBPs, and EBF2 (4). For a comprehensive review of SNS-mediated adipogenesis, see (4, 20).
BAT dissipates energy through nonshivering thermogenesis, a process mediated by the unique expression of uncoupling protein 1 (UCP1) in the inner mitochondrial membrane of brown adipocytes, which facilitates re-entry of protons into the mitochondrial matrix, thereby uncoupling protons from the electron transport chain to produce heat at the cost of ATP (255). BAT thermogenesis is β3-AR signaling dependent (293) and requires SNS innervation for UCP1 activation (46). β3-AR activation by NE on brown adipocytes causes a cAMP-PKA signaling cascade that results in the upregulation of UCP1 and mitochondrial biogenesis (245). However, surgically denervated iBAT can maintain some UCP1 expression despite “whitening” of the tissue (254), likely due to a compensatory increase in circulating NE and fatty acids (47). Cyclin-dependent kinase 4 (CDK4) expression in the ventromedial Ht (VMH) is one pathway that negatively regulates thermogenesis via controlling SNS drive. CDK4 knock-out (KO) in the VMH resulted in increased sympathetic innervation to the tissue that led to increased cold tolerance, higher BAT mitochondrial content, and decreased lipid content (53). Similarly, actions of bone morphogenetic protein 7 (BMP7) in the Ht also increased SNA in BAT of rats (130, 253).
While BAT provides the majority of thermogenic output, a proportion of cells in the ing-scWAT depot are able to phenotypically switch into an energy-expending phenotype as either “beige” adipocytes (de novo brown adipogenesis) or paucilocular cells [potentially via SNS-induced trans-differentiation (61)]; both as a result of increased SNS drive (64). The SNS can induce WAT “browning” through the activation of β3-ARs on white adipocytes by release of NE (16). While WAT browning is regulated by numerous inter-organ molecular signaling pathways in healthy and diseased states, the SNS and NE release is the driving factor behind most physiological browning (266). WAT browning is characterized by an increase in multilocularity, UCP1 expression, and mitochondrial biogenesis, and is only maintained by sustained NE release. In the absence of SNS drive, the beige adipocytes will pivot back to the default energy-storing phenotype (“whitening”). Interestingly, beige adipocytes will retain some epigenetic and transcriptional hallmarks of cold-activated browning, even after rewarming (218). This provides evidence for longer-lasting beneficial metabolic outcomes following bouts of increased SNS drive, such as repeated cold exposure. Mice are even able to survive cold challenge following SNS denervation of iBAT, partially due to a compensatory browning of ing-scWAT (47). Thermogenesis in BAT and browning of ing-scWAT have become the hallmark indicators for SNS drive in these tissues.
As a brief aside, it is worth mentioning that leptin is not the only adipokine secreted following β-AR signaling. Another recent example is adipose-secreted signaling protein (Adissp) which is secreted from brown adipocytes following β3-AR activation to facilitate WAT browning and thermogenesis (55). While Adissp is secreted following β-AR signaling, it acts on white adipocytes independently from β-ARs (55). Little work has been done to investigate the link between adipokine release in response to SNA or whether other adipokines like Adissp can facilitate adipose tissue nerve activity or outgrowth similarly to leptin.
Human BAT, while occupying proportionally less volume relative to rodent BAT, is activated by cold exposure as identified by PET/CT imaging, UCP1 expression, and increased energy expenditure (67, 152, 190, 284, 296). Single nucleotide RNA sequencing (snRNAseq) of human adipocytes identified a subpopulation of visceral adipocytes (hAd6-expressing) that correlated with increased ambient RNA (freely floating transcripts captured during snRNA-seq processing) expression of pan-neuronal markers Uchl1 and Tubb3, as well as the thermogenic genes Ebf22 and Pgc1a. Thus, hAd6 may represent a subpopulation of thermogenic visceral adipocytes in humans that receive increased neural input (81, 131). The ability/extent to which human WAT can undergo browning in response to stimuli similar to mouse models is not yet clear (116), but human WAT does not brown in response to exercise like mouse WAT does (171).
The heterogeneity of adipocytes is an evolving area of research, and we don’t yet know if nerve products control the development of these different cell types in a depot, if innervation patterns differ specifically because of signals emanating from beige adipocytes that require more dense sympathetic terminals, or if specific nerve ending structures (such as the Neuro-adipose Nexus—discussed later) form only around subsets of mature adipocytes.
Taken together, sustained SNS drive facilitates a systemic transition of adipose tissue from energy storing to energy expending through the accumulation of thermogenic brown adipocytes, UCP1 activation, decreased proliferation of white adipocytes, browning of scWAT, and increased lipolysis to provide the required energy (Figure 1).
Although the gaps in knowledge regarding sympathetic innervation patterns and neural circuitry to the brain have narrowed in recent years for both BAT and scWAT, the knowledge gap about neuroanatomy is still quite significant for other adipose depots. In summarizing the neural tracing studies available to date, we will cover the SNS neural circuitry to WAT and BAT, starting in the CNS with the hypothalamus, out through the spinal cord, sympathetic chain ganglia (SChG), and to the varicose axons that form neuroeffector junctions directly onto adipocytes.
The SNS is organized to provide functionally specific autonomic output to organs as needed. This remains true for adipose tissue (177). Differential sympathetic drive to WAT and BAT, distinct from elevations in circulating catecholamines, has been identified in response to multiple lipolytic stimuli (cold exposure, glucoprivation, and fasting) (38, 238). Several research groups have traced the sympathetic nerve circuitry between the brain and adipose depots. However, interpreting these data is not straightforward, as results vary quite substantially between methods, models, and studies. Seminal experiments performed in Siberian hamster by the Bartness lab, a pioneer in the understanding of adipose sympathetic innervation, utilized fluorescently labeled neurotropic pseudorabies virus (PRV), a retrograde trans-synaptic tracer injected directly into adipose depots. These studies revealed that BAT, ing-scWAT, and mesWAT innervation could all be traced back poly-synaptically to the Ht (15, 192), further confirming this brain region as the command center for energy expenditure via regulating sympathetic drive to adipose.
Studies in rats found that iBAT postganglionic innervation could be traced primarily to the stellate ganglion (273). Recent PRV tracing of mouse iBAT found that postganglionic fibers project from stellate-T5 of the SChG via dorsal rami T1-T5 (127), and preganglionic sympathetic fibers were traced to T2-T6 of the spinal cord (93). Cholera toxin subunit B (CTB) retrograde tracing, an alternative to PRV, also traced the SNS innervation of BAT to the Stellate ganglia (18). Differences may be due to technique or method, and as newer methods using modern viruses are utilized the contradictory literature will likely be settled. Of note, surgically denervating all but one intercostal nerve in rats (performed separately for all four iBAT nerve bundles) found that each nerve bundle contributed differentially to the thermogenic potential of iBAT, and with substantial variation between each rat (91). Together, these studies indicated that there is site-specific control of the nerve supply to each depot, or regions of each depot, and that inter-individual differences that may be due to developmental conditions could also impact innervation patterns (Figure 2).
For scWAT nerve tracing studies, PRV tracing in Siberian hamsters revealed that postganglionic sympathetic nerves project from T13-L3 SChG (primarily T13) (286), while another study showed tracing from T1-T3 SChG (191), and a study in rats found that ing-scWAT innervation was traced back to T13-L1 SChG (273). In mice, ing-scWAT postganglionic sympathetic nerves were traced to T12-L1 SChG, and preganglionic nerves were traced to T7-T11 of the spinal cord, with the majority (46%) at T11 (126). A few studies using CTB retrograde labeling in ing-scWAT instead observed that the majority of SNS fibers were traced to the celiac ganglia, which is primarily responsible for innervating abdominal organs. Importantly, this was the extent of the tracing performed and an effort was not made to continue the tracing of the neural circuit from the celiac ganglion to the brain (49, 134). Others utilizing PRV-tracing found that the celiac ganglion does not significantly contribute to ing-scWAT innervation (93) (Figure 2).
PRV tracing in Siberian hamsters revealed that pgWAT sympathetic innervation was traced to T13-L2 of the SChG (primarily T13) (286). Although ing-scWAT and pgWAT sympathetic fibers were both traced back to similar ganglia in Siberian hamsters, the innervation was largely nonoverlapping, supporting differential innervation of each depot (286). Additionally, PRV tracing in mice has recently identified sympathetic input that traced back to the prevertebral aorticorenal ganglion (51). Relatively few studies have been performed tracing SNS innervation to pgWAT, perhaps because this depot is not a good surrogate for human WAT, and it is still unclear at what level of the spinal cord the preganglionic fibers exit. Additionally, the other scWAT depots (axillary) have also not been specifically traced. The neuroanatomy of adipose depots is summarized in (Figure 2).
Simultaneous retrograde tracing from two depots via dual injection of fluorescently distinct PRV tracers from iBAT and ing-scWAT of rats demonstrated that each depot was innervated by dedicated axonal projections with approximately 10% to 15% of the axons branching collaterally to both depots (273), which may explain the compensatory browning of ing-scWAT that is observed when iBAT is denervated (47). Dual PRV labeling in Siberian hamsters traced ing-scWAT and mesWAT neurons to the CNS where 20% to 55% of labeled neurons were doubly labeled in the brain, with less overlap in the spinal cord and SChG between the two circuitries (192). The greater percentage of collateral nerves between ing-scWAT and mesWAT is likely attributed to their shared lipolytic functions. Dual PRV tracing in mice demonstrated that a proportion of BMAT and ing-scWAT sympathetic innervation traced to the same neurons in the CNS (270). Dual tracing also found shared CNS input between pgWAT and the liver in mice, ing-scWAT and the left gastrocnemius muscle in rats, and BAT and the left gastrocnemius muscle in rats (77, 240). These findings support the presence of command neurons in the CNS, that regulate adipose functions similarly across multiple depots.
Together these studies underscore the regional control exerted on discrete adipose depots by different SNS projections, independent of any circulating catecholamines that may be released from the adrenal gland (Figure 2).
Sympathetic nerves in human adipose tissues have not yet been traced. While it is possible that similar levels of the spinal cord and SChG innervate respective human adipose depots, the absence of a T13 vertebrae in humans suggests that there would be some degree of reorganization, at least for visceral and subcutaneous depots. The larger body size and differences in thermoregulation across evolution may also have impacted this neuroanatomy.
The density and distribution of SNS nerve endings in adipose differ significantly between each depot, with prominent differences reported between BAT, ing-scWAT, and pgWAT (Figure 2). Relative SNS fiber density was compared histologically between BAT and ing-scWAT, concluding that SNS fibers innervated BAT more densely than ing-scWAT in mouse (127, 157). More neurons in the SChG innervate ing-scWAT than mesWAT (192), and histologically ing-scWAT is more densely innervated than pgWAT (157). While PVAT has been shown to be innervated by both sympathetic and sensory nerves (1, 41), PVAT SNS density has not been directly compared with other BAT or WAT depots, and due to the variation in PVAT between vessels, a definitive statement about PVAT SNS density may not be justifiable. Furthermore, given the “snapshot” of innervation status across all histological samples, it is unclear if nerve remodeling in response to various stimuli happens on a timescale that could impact total tissue innervation over shorter periods, such as days or weeks.
Taken together, these findings uncover a hierarchy in which thermogenic BAT receives relatively greater SNS input than scWAT, which due to its thermogenic potential, is relatively more innervated than visceral WAT. Importantly, it is unclear if input from the SChG is directly proportional to the density of arborizations within each depot. A single neuronal projection leaving the SChG may innervate a single cell, or it may branch and innervate several distinct tissue regions. While density of innervation may be a good indication of how active the sympathetic drive is to that tissue, there can still be variation in activity of single axons, and how much neurotransmitter or neuropeptide is released from a single nerve terminal or en passant varicosities. Additionally, postsynaptic receptor density on target cells will influence the response.
Furthermore, there are distinct patterning differences in SNS fibers between scWAT, BAT, and pgWAT depots. BAT and pgWAT are mostly homogenously innervated across the depot (57, 127). However, SNS innervation of ing-scWAT has striking regional differences. Ing-scWAT can be subdivided into three distinct anatomical the posterior/inguinal, a center region surrounding the subiliac lymph node (SiLN), and the anterior/dorsolumbar region (28, 57). Whole tissue observations revealed that the SNS density was greatest in the posterior half of the tissue and lateral to the SiLN (28, 57). The SiLN itself has also been found to be densely innervated by sympathetic (27) and sensory nerves (124). Ing-scWAT browning is localized primarily to morphologically distinct lobules and is associated with increased SNS density relative to surrounding tissue (17, 57, 75, 126). These browning-prone lobules have been termed beige islands/islets and vary in size and shape to reflect the size of the depot they reside in, but do not vary in number (15-21 lobules per fat pad) (75). With data suggesting that segmented clusters of beige islets are molecularly distinct, they will be an exciting avenue for continued investigation (127). The proximity of browning to SNS terminals makes sense given the need for NE in the browning process and for activation of UCP1-mediated thermogenesis.
Sympathetic nerve terminals (or any nerve terminal/junction/synapse) in WAT and BAT have been historically underexamined until recently. The first description of sympathetic neuro-adipose connections was made in 2015, describing a role for the SNS in facilitating lipolysis (288). This neuro-adipose connection was imaged through intravital microscopy using a sympathetic nerve-reporter mouse, which revealed a cluster of sympathetic fibers running in parallel and terminating at an adipocyte. This terminal presented as large patches of fluorescent labeling near the adipocyte, which the authors described as “bouton-like structures” (288). While the authors quantified from eight separate micrographs, the one image provided of this putative neuro-adipose connection was likely an innervated blood vessel that was severed during microdissection, since the supposed terminal junctions shown are not visibly connected to the incoming nerves and similar bouton-esque fluorescent splotches can be seen elsewhere in the image, not associated with nerves. This newer interpretation would explain why in the 7 years since this observation, these very large bouton-like structures have not again been visualized, even though microscopy techniques have greatly improved.
Instead, a large body of evidence has formed in favor of en passant release of neurotransmitter along sympathetic axonal varicosities in adipose. Transmission electron microscopy (TEM) of mouse BAT demonstrated direct contact of axonal varicosities with adipocytes (127). This is reminiscent of an autonomic neuroeffector junction; the junction where autonomic nerve varicosities release neurotransmitter onto tissue, such as smooth muscle (42). TEM of sympathetic fibers in rat mesWAT found that they, unsurprisingly, contain synaptic vesicles across their length (239). Our own imaging has also revealed varicose sympathetic axons in ing-scWAT (275). It has not yet been demonstrated that sympathetic varicosities come into direct contact with PVAT adipocytes (9), but the release of neurotransmitters en passant in PVAT is accepted (41).
While varicose sympathetic fibers have been observed to densely innervate both WAT (276) and BAT (127), the recent discovery of specialized nerve terminals in scWAT has also emerged via observation of a structure we call the neuroadipose nexus (NAN) (276). The NAN is characterized by an adipocyte, or small cluster of 2 to 4 adipocytes, that are webbed by sympathetic fibers stemming from one or more incoming nerves (Figure 3) (275). NANs are typically located near the tissue surface boundaries (Figure 3A). NANs are a dense, but discreet, varicose innervation of single adipocytes that are distinct from the surrounding parenchymal innervation (Figures 3A-3B). Not only are the adipocytes within a NAN in contact with a greater number of nerve fibers, but each fiber is densely varicose, and these varicosities contain synaptic vesicles (275). NANs also give rise to definitive terminal endings in addition to varicose junctions (Figure 3C). NANs have so far only been visualized in mouse scWAT, but a hand-drawn microscopy observation by Alexandre Dogiel in 1898 showed an adipocyte in PAT ensnared by varicose axons that appear structurally similar to a NAN (29), providing credence to NANs existing in PAT and potentially other WAT depots.
NANs are labeled by classical markers of both sympathetic (tyrosine hydroxylase; TH) and sensory (calcitonin gene-related peptide; CGRP) nerves, and as of now, it is unclear if NANs are formed by afferent, efferent, or both fiber types (275), especially given new insights that revealed 40% of TH^+^ nerves in adipose are sensory and not sympathetic (267). The fact that such a high proportion of TH^+^ fibers in adipose are indeed sensory/peptidergic raises the important issue about all prior TH-marker data in the adipose literature to date—it is likely that results were confounded by interpreting the data with a sympathetic-nerve lens when the data could have come from sensory nerves as well.
Upward of 90% of adipocytes make contact with parenchymal sympathetic fibers in healthy mice (134), with many being varicose and likely releasing neurotransmitter and/or neuropeptide en passant (275). It remains unclear why only a small proportion of these cells require specialized nerve-adipose connections, or if these NAN structures are highly plastic in response to changing tissue environment and metabolic demands but are only captured in static fixed-tissue imaging. The first evidence of NAN neuroplasticity was found with aging (278). A two-fold increase in the number of NANs in ing-scWAT was observed between mice 20 and 60 weeks old that were independent of changes in whole tissue SNS innervation. The number of NANs returned to the basal numbers by 100 weeks old (278). At this time, it is thought that NANs serve a specialized function that requires targeted synaptic release of nerve products, or that these may be sensory nerves serving an interoceptive role on a single adipocyte. The distinguishing characteristics between a NAN-forming adipocyte, and one that does not, is unclear, and so too are the mechanisms driving their formation or remodeling.
It has been proposed that BAT contains a greater number of sympathetic varicosities than WAT (126, 127, 183). While this could be true, as of now, no direct comparisons have been made for the number of sympathetic varicosities of an axon innervating white, beige, or brown adipocytes. Recent studies making such claims (126, 127) do so by analyzing TH-labeled nerve puncta in 5 μm thick paraffin-embedded sections using automated spot detection with a 3 μm diameter cut-off. The authors interpreted puncta as axonal varicosities, but the results may be confounded by the thin tissue sections. By this method, the authors calculated relative SNS fiber density (183). Drawing conclusions on varicosity density should take into consideration the number of varicosities per length of axon in addition to total fiber density, and even then, the activity of each axon is not measurable—just total innervation and presynaptic terminals. On this point, it is also worth noting that in scWAT, parallel sympathetic varicosities can be separated by less than 250 nm (Figure 3D) which is below the resolution limit of standard confocal microscopy. Super-resolution techniques such as Stimulated Emission Depletion (STED) microscopy, transmission electron microscopy (TEM), or deconvolution (as demonstrated in Figure 3) are required to resolve separate varicosities in such instances.
Adipose tissue innervation is as heterogenous as the organ itself. While the most well-studied nerves in adipose tissue have been catecholamine-releasing sympathetic nerves, sensory nerves have also been identified and characterized, albeit with less scrutiny (92, 235, 275, 276). That is, until recently when Wang et al. provided unequivocal evidence of the importance of sensory innervation in adipose by conducting adeno associate virus (AAV)-mediated ganglia ablation of adipose sensory nerves (267). Multiple nerve products have been detected in adipose, including sensory and sympathetic neuropeptides, lending to an emerging granularity in the classification of nerve subtypes in the adipose tissues.
TH is the rate-limiting enzyme in catecholamine synthesis (70), including NE, which regulates numerous adipose functions (as described above) and provides robust labeling of sympathetic nerves by immunohistochemical or genetic approaches. However, a substantial subset of dorsal root ganglion (DRG) sensory neurons also express TH (39, 140, 211), which is frequently ignored in the adipose literature. TH-expressing DRG neurons comprise 15% of mouse DRG neurons, innervate cutaneous and visceral tissues, and have been identified in DRGs that also innervate WAT and BAT (39). Therefore, although DRGs are considered sensory ganglia, there is a mix of neuronal cell types. It was recently shown that 40% of DRG nerve fibers in ing-scWAT were TH-positive (267) and it is likely that this has skewed previous comparisons of sympathetic versus sensory innervation in adipose and brings into question results from all previous TH-based sympathetic nerve observations. Moving forward, TH-positive sensory nerves should be considered when interpreting data on adipose innervation, until ruled out and accounted for experimentally.
CGRP, Substance P, Nav1.8, advillin, and transient receptor potential (TRP) channels/receptors have all been used to identify sensory nerves in adipose (71, 174, 276). While there is strong evidence of sensory and sympathetic innervation of adipose, to date there is no convincing evidence that parasympathetic innervation exists in adipose tissue (100), and uncertainty remains regarding parasympathetic innervation of BAT (21) and PVAT (41). Instead, the sensory nerves which appear to inhibit SNS drive to adipose (267) may be acting as a brake, instead of the classical role of PSNS nerve activity counteracting SNS drive. In ing-scWAT, there are nerve bundles containing mixed nerve types that express both the sympathetic nerve marker TH and sensory nerve markers Nav1.8, advillin, and CGRP (275, 276). These mixed nerve bundles are predominately restricted to the large nerves consistently present in the inguinal depot, which we have previous described anatomically as adipose “transverse nerves” (28, 276). There is also no evidence for vagal innervation of WAT (the vagus nerve, or 10th cranial nerve, is a mixed nerve containing afferent and efferent axons) (101), an important observation since there is vagal innervation of other metabolically relevant tissues and organs, such as the gut. However, since the time that vagal and parasympathetic innervation was last assessed in WAT, adipose imaging techniques and antibody availability/specificity have vastly improved, therefore a secondary investigation into its existence might be worth pursuing. Unlike WAT, there is evidence that BAT receives vagal afferent innervation which upon stimulation can inhibit BAT SNA and thermogenesis (165).
There has been some debate that the majority of sympathetic innervation within the scWAT is restricted to vasculature. Although adipose tissue vasculature (both blood and lymphatic) is highly innervated by both sympathetic (TH^+^) and sensory nerves (CGRP^+^), parenchymal innervation independent of the vasculature is clearly evident and is heterogeneously distributed in the adipose tissue with regions of high neurite density (28, 30, 184). However, parenchymal innervation may appear minimal compared to dense vascular innervation depending on the metabolic status of the tissue.
Studies of adipose nerve fiber myelination have so far been performed exclusively in scWAT (275, 276). Whole mount staining with pan-neuronal marker β3-tubulin and myelination marker myelin protein zero (MPZ), as well as Luxol Fast Blue myelin staining, revealed that nerve bundles ≥25 μm in diameter were predominantly myelinated (276). TH^+^ fibers enter scWAT via these large nerve bundles consisting of mostly myelinated fibers with relatively fewer unmyelinated Remak bundles (275). Cross-sectional images of these nerve bundles show that TH^+^ fibers are co-labeled by myelin basic protein (MBP), suggesting that they are myelinated themselves. These large nerves traverse through or nearby the SiLN region, although it is not yet clear if most large and myelinated nerves are sensory or sympathetic. Within the parenchyma, thinly myelinated and unmyelinated TH^+^ nerves dominate and exhibit high plasticity with cold exposure (49, 184, 275). Many parenchymal TH^+^ fibers were labeled by MBP, but MPZ labeled a more discreet population of fibers, particularly with less co-labeling observed in the parenchymal and neurovascular nerve fibers compared to MBP. This detail is crucial, as these fibers are primarily TH^+^ (275, 276). Besides confirming that some TH^+^ fibers are myelinated in ing-scWAT, there is still some uncertainty as to the exact extent of myelination of TH^+^ fibers. TEM of BAT sympathetic innervation noted that fibers contacting brown adipocytes could be wrapped by Schwann cells (SCs), but it is unclear if these were myelinating or nonmyelinating SCs (127). Myelinating and nonmyelinating SCs have been identified in ing-scWAT, with a greater proportion being myelinating even with changing energy balance (275). SCs are also consistently located close to TH^+^ NANs but whether these SCs are myelinating has yet to be confirmed and is an ongoing area of investigation (275).
Seminal studies by the Bartness group and their contemporaries clearly demonstrated the necessity of innervation of iBAT and ing-scWAT in brain-adipose communication and normal metabolic function [as reviewed in (29)]. The main findings of these landmark studies are summarized in Table 1. Surgical denervation of iBAT resulted in impaired thermogenic function, decreased energy expenditure with concomitant increase in body fat mass, and the “whitening” of iBAT (78, 254). Surgical denervation of ing-scWAT is more challenging than surgical denervation of iBAT due to the numerous nerve bundles innervating this depot and entering the tissue at multiple locations. Studies using surgical denervation of ing-scWAT resulted in increased ing-scWAT depot mass (287) and white adipocyte proliferation and differentiation, in both rats and Siberian hamsters (Table 1).
Surgical denervation often indiscriminately eliminates both sympathetic and sensory nerves which travel together in large nerve bundles and can disrupt the blood supply to the tissue, therefore subsequent denervation studies employed chemical agents that specifically target nerve subtypes, to gain a deeper understanding of the distinct roles they play in adipose tissue. The catecholaminergic-specific neurotoxin 6-OHDA has been used as a peripheral nerve-denervating agent since the 1960s (210, 251). It was first used in iBAT denervation studies in the 1970s, when it was demonstrated as effective at eliminating sympathetic innervation of adipocytes and vasculature (251). Guanethidine, another sympathetic nerve toxin, was shown to be highly effective (near complete elimination of NE); however, 6-OHDA has proven to be the more consistent drug of choice in the adipose field (260) and has been repeatedly shown to be effective at sympathectomy in various adipose depots (113, 221, 251). Effectiveness of sympathetic denervation is often presented as a decrease in norepinephrine turnover (NETO) of NE content at the tissue level. Of importance, the best way to measure sympathetic activity in a tissue is directly through electrophysiological measurements of nerve activity, but measuring NETO, an assay that is used to determine NE content per gram of adipose tissue to show effectiveness of sympathetic denervation, is an accepted surrogate marker. In this sense, total NE content within the whole adipose tissue is seen as a measure of sympathetic innervation and not sympathetic activity. TH itself as a marker also represents nerve activity, and not just fiber density per se (70).
When comparing results from chemical and surgical sympathectomy of ing-scWAT in Siberian hamsters, effects on fat mass appear to occur significantly later than the effects on NE content (72, 92, 287); however, there have been inconsistencies in these findings (Table 1). The authors attributed a lack of change in fat mass in certain studies due to a short postdenervation period (10 days) at assessment versus longer studies (10–11 weeks) (72, 92). This brings up an interesting chemical sympathectomy by 6-OHDA is reversible, as nerves grow back following the initial immediate loss of sympathetic innervation (NE content in 6-OHDA injected and contralateral ing-scWAT is significantly decreased by 24 h postinjection) (113). This is a unique feature of the peripheral nervous the readiness for nerve regeneration in response to injury or denervation, which is lost in disease states like peripheral neuropathy (112). Although it is unclear when reinnervation is completed following 6-ODHA sympathetic denervation in Siberian hamsters, in rats the reinnervation period is complete three weeks after chemical denervation with 6-OHDA (14). In mice, it has been reported that after 4 weeks following 6-OHDA unilateral ing-scWAT denervation, the amount of NE per mg of tissue was only 40% lower than sham controls, as compared to 60% lower at 24 h after injection, suggesting reinnervation is underway after about a month (113). Microscopy advances over the past several years have permitted visualization of total innervation in adipose tissues which can be applied to the monitoring of reinnervation following denervation, similar to how it has been applied to cold/thermoneutral or the cold-rewarming paradigms (27, 28, 49). Unlike measures of NE content, reinnervation studies using microscopy can also address patterns of reinnervation, a form of nerve remodeling that may be distinct from total neurite density in the tissue.
Early studies of adipose denervation employed a unilateral denervation model with the contralateral depot serving as an in-animal control. However, it is now clear that sympathetic denervation of a single adipose depot produces compensatory effects not only in the contralateral depot but also across other adipose tissues, see Table 1 (113, 221). Aside from direct effects on adipocytes, chemical denervation of sympathetic nerves in pgWAT and iBAT resulted in increased expression of the proinflammatory cytokine TNF-a, as well as the potent monocyte/macrophage recruitment cytokine MCP-1 (in BAT only), demonstrating a role of the SNS in regulating inflammatory responses in adipose (247). These studies highlight the systemic effects of sympathetic denervation of even a single adipose depot, something that needs to be considered more carefully in future denervation studies.
More recent studies have employed sympathetic chemical denervation of adipose depots as a tool to investigate which mechanisms in adipose tissue function and disease etiology are mediated by SNS signaling. Huang et al. used 6-OHDA-mediated denervation to demonstrate that “phosphatase and tensin homologue deleted on chromosome ten” (PTEN)-related adipose remodeling in mouse ing-scWAT is dependent on SNS signaling in a “adipose PTEN-leptin-sympathetic nervous system” feedback loop (125). Earlier work strongly suggested that denervation of a single depot can result in compensation by other depots (113). Cao et al. clearly demonstrated that 6-OHDA-mediated denervation of iBAT resulted in compensatory upregulation of UCP1 and browning in ing-scWAT, and that denervation of ing-scWAT prevented cold-induced thermogenesis and browning of ing-scWAT (46). Newer genetic-based denervation approaches have been employed to investigate the role of SNS in adipose (30, 207). Using a Cre/diphtheria toxin receptor (DTR) system, local denervation of sympathetic nerves can be achieved via injection of brain-sparing diphtheria toxin to the adipose depot in a TH-cre mouse crossed to a Cre-inducible diphtheria toxin receptor (Rosa26-LSL-DTR) mouse (207, 288). This approach was elegantly used to demonstrate that leptin stimulates lipolysis in ing-scWAT via SNA (288). “Genetic denervation” of ing-scWAT has also been accomplished by deletion of BDNF from myeloid lineage immune cells (LysMCre driver) and a subset of myeloid cells that can function as neuroimmune cells (Cx3cr1CreER) (27, 30). These genetic denervation studies showed that myeloid cell-derived BDNF is necessary to maintain healthy innervation in ing-scWAT (27, 30), and genetic denervation of the tissue resulted in dysregulated whole-body energy balance and metabolism. Finally, recent advances in optogenetics and AAV-based designer receptors exclusively activated by designer drugs (DREADD) technology (as discussed below) now permit manipulation of adipose nerve function (stimulation/inhibition) without changing the total nerve supply. These studies will bring much-needed insights regarding the etiology of adipose nerve dysfunction prior to the occurrence of nerve fiber loss [such as with adipose neuropathy; (28)] and allow for localized nerve targeting to investigate contributions to adipose tissue function and health.
Sensory denervation of adipose tissue can also influence adipose sympathetic innervation. In Siberian hamsters, bilateral ing-scWAT sensory denervation with the chemical agent capsaicin decreased NETO in both ing-scWAT and iBAT and led to decreased UCP1 expression in iBAT when animals were exposed to cold for 24 h (193). Interestingly, NETO in mesWAT was also increased, with no effect on pgWAT or rpWAT, yet fat mass was decreased in pgWAT and rpWAT of sensory denervated animals (193). These studies clearly underscored that sensory-SNS cross talk exists across adipose tissue depots, yet few studies have focused on the role of sensory innervation in adipose tissue since then. Similar to the unreliable results of guanethidine for sympathetic denervation, capsaicin-mediated denervation can be unreliable from one batch of drug formulation to another, and has a high, yet underreported, mortality rate. In our own studies, we have noted mortality rates as high as 70% in capsaicin-mediated ing-scWAT denervation. An alternative to capsaicin-mediated sensory denervation is the highly potent capsaicin analog resiniferatoxin (RTX), which has successfully been used to denervate perirenal adipose tissue (155), as well as scWAT by our laboratory. Sensory denervation was also recently achieved by AAV-mediated DRG ablation, which demonstrated through more sophisticated methods what numerous early sensory denervation studies had that sensory innervation also regulates tissue lipolysis. This AAV-mediated adipose sensory ablation approach revealed changes to tissue lipolytic status upon AAV-mediated sensory denervation of ing-scWAT, as well as blunted browning, and impacts on SNA (267). While this study did not investigate central pathways in the hypothalamus, the mechanism by which sensory nerves impact sympathetic drive, nor the mediators of the sensory nerve lipolytic activities, we know from other data that the sensory neuropeptide CGRP is capable of lipolytic activity (68, 168, 264) and may be contributing to tissue lipolysis in addition to NE. The disregard for adipose tissue sensory innervation and the contribution of these nerves (nonpeptidergic and peptidergic subtypes, releasing numerous neuropeptides) in prior work is glaring, but with recent investigations into the role of adipose sensory nerves (267), hopefully, the research landscape will be updated in this regard.
The catecholamines epinephrine and norepinephrine are released by sympathetic nerves and act on adrenergic receptors (ARs; which are also noradrenergic receptors for norepinephrine/noradrenaline) to regulate many physiological processes in the body (115). Originally divided into two groups, type α- and β-ARs, there are currently nine known ARs subdivided across three major three α1-ARs (α1A, α1B, and α1D), three α2-ARs (α2A/D, α2B, and α2c), and three β-ARs (β1, β2, and β3) (115). Catecholamines have a pronounced lipolytic effect on adipocytes by causing the de-esterification of triglycerides stored in the lipid droplets of adipocytes into FFAs that can be oxidized for fuel (7). While both epinephrine and NE are released from sympathetic nerve terminals and bind to ARs, in adipose NE is the main driving force for lipolysis (2). Indeed, in vitro studies on adipocytes in culture regularly utilize NE or its mimetics (Table 2) to stimulate cellular function as a surrogate for SNA. In both rodents and humans, NE binds to β-ARs on adipocytes which begins signaling cascades that result in the transcription of lipolytic and thermogenic gene programs (Figure 4). SNS-mediated NE release in adipose tissues happens in response to energy stress such as fasting, or environmental triggers such as cold exposure, and the transcriptional response in adipose to NE signaling is almost immediate given the speed of neural conduction and neurotransmitter vesicle release (118).
Evidence for a third, adipocyte-specific βAR receptor came through pharmacological studies by Arch et al. in 1984 (5), and by 1989 β3-AR was characterized as the unique AR (82) only found in BAT and WAT (185). With the cloning of a rat homolog of the β3-AR (107), molecular studies soon followed and pharmacological targets of the receptor were developed. What made the discovery of β3-AR so exciting for obesity research was that NE signaling on β3-AR appeared to regulate lipolysis (5) and energy expending thermogenesis in BAT (269) and held promise as obesity therapeutics. Several studies utilizing mouse KO models were performed to further investigate the role of β3-AR in adipose tissue. Whole-body KO of β3-AR resulted in a moderate increase in adiposity that was more pronounced in female mice, however, no deficits in adaptive thermogenesis in BAT were observed, suggesting that β3-AR signaling in BAT may not be required for thermogenesis (242). This study also demonstrated that KO of β3-AR increased gene expression of β1-AR (but not β2-AR) in both white and BAT, suggesting a compensatory mechanism by which at least some of the physiological effects of β3-AR signaling are achieved to try to maintain regular metabolic function (242). It was also demonstrated that the effects (increased energy expenditure, fatty acid oxidation, insulin release, and glucose uptake) of the drug CL 316,243 are exclusively mediated by β3-AR signaling (242). While it appeared that β3-AR was dispensable in BAT thermogenesis, that was not the case for WAT. Following cold exposure, WAT of β3-AR KO mice did not express UCP1 or typical multilocular browning associated with thermogenesis induction of WAT (135). Further understanding of the differences in β3-AR signaling in BAT versus WAT was sought through tissue-specific re-expression studies and CL 316,243 treatments. Re-expression of β3-AR only in BAT of otherwise β3-AR KO mice was unable to rescue CL 316,243 mediated effects on insulin release or food intake and only minimally restored energy expenditure (111). However, re-expression of β3-AR in both BAT and WAT fully restored CL 316,243 mediated physiological effects (111). These studies suggested that β3-AR mediated SNS signaling in both WAT and BAT work in concert to maintain energy balance.
While single KO of β1 or β2-ARs or double KO of both β1/β2-AR only resulted in minimal metabolic perturbations related to cardiac function and vascular tone, but no effects in adipose tissue (58, 219, 220), triple KO of β1/β2/β3-ARs resulted in increased fat mass, phenotypically white characteristics of BAT such as decreased UCP1 expression, and a comparative inability to maintain core body temperature with cold exposure (11). These studies suggested that the phenotypes described in single βAR KO models need to be considered under the caveat of compensatory βARs signaling.
It was clear that β-ARs on adipose were required for healthy adipose function, and these studies provided a potential therapeutic target for obesity. Although β3-AR expression in rodent adipose is much higher than in human adipose, downregulation of β3-AR in adipocytes has been linked to catecholamine resistance in both mice and humans with obesity (258). However, this enthusiasm for a new drug target was soon replaced with controversy regarding the physiological roles of β3-AR in human adipose. Although it was demonstrated that β3-AR did exist in human adipose (108), the low mRNA expression levels in human compared to rodent, combined with lower contributions to lipolysis as compared to the β1 and β2-AR isoforms (249), raised concerns over whether β3-AR was a relevant player in human adipose tissue function. Furthermore, as adult humans lack an abundance of classical BAT, the effectiveness of a β3-AR targeted therapy became questionable. However, recent evidence that human brown and beige adipocytes express β3-AR which can be pharmacologically targeted to stimulate lipolysis and thermogenesis has revitalized interest (54).
Along with this renewed interest in human adipose β3-AR, new studies are yet again challenging its importance in regulating BAT thermogenesis. It has been proposed that β1-AR is the main regulator of lipolysis and UCP1 induction in human BAT due to β1-AR mRNA comprising the large majority of β-AR mRNA in human BAT biopsies (216). Another recent study argued that β2-AR and not β3-AR activation is the main regulator of human BAT activity, based again on mRNA expression in human BAT biopsies and evidence that previous studies reporting β3-AR-stimulated thermogenesis in human BAT had used the β3-AR agonist mirabegron at a dose high enough to stimulate all three subclasses of β-ARs (31). While it seems that the β3-AR controversy may not end soon, it is worth revisiting the significance of NE signaling via β1-, β2-ARs. In mice, NE stimulation of β1 and β2-ARs in WAT was shown to suppress hyperplasia (228), a finding consistent with SNS denervation of WAT leading to hyperplasia. Considering that humans, and especially those suffering from obesity, lack sufficient active BAT mass, perhaps new investigation into adrenergic signaling in WAT is again warranted.
NE is not the only SNS nerve product that regulates adipose tissue function. Within sympathetic nerves, NE can be co-localized with the neuropeptide NPY and/or the purine nucleotide ATP (243). NPY and ATP are co-released with NE peripherally, and receptors for NPY, ATP, and adenosine (a degradation product of ATP) have all been identified in WAT and BAT (257, 291).
NPY functions in opposition to the effects of NE in adipose, driving increased adipogenesis and reduced lipolysis through activation of its receptor NPY2R (147). The modulatory impact of NPY is dependent on the amount of NPY co-released with NE, which is dependent on nerve stimulation intensity and action potential frequency (161). Outside of the SNS, NPY was reported to be released by mature adipocytes (145) and immune cells (237), but this has not been validated by additional studies. In iBAT, the vasculature and not adipocytes are innervated by NPY^+^ axonal varicosities (45, 127). In WAT there is evidence for parenchymal innervation by NPY^+^ fibers (99), indicating that NPY may influence WAT and BAT differentially.
SNS drive also increases ATP release from the nerves, which influences shivering thermogenesis in contrast to NE-stimulated nonshivering thermogenesis by acting on purinergic P2 receptors (146). However, the role of ATP co-transmission with NE in adipose tissue should not be over-shadowed, and the neuromodulatory role of ATP is largely dependent on which receptor is being activated, as a number of P2 receptors (ionotropic and metabotropic) have been identified in adipose tissue (111). Purinergic signaling in adipose tissue has been implicated in adipogenesis, lipid metabolism, glucose uptake, and leptin production (257). BAT sympathetic nerves were found to express vesicular nucleotide transporter (VNUT), the transporter for ATP, supporting that ATP is co-released from sympathetic nerves in BAT (215). It was also found that ATP co-transmission upregulated ATP synthase activity in BAT to support NE-induced thermogenesis (281). Like NPY, ATP may not be solely released from sympathetic nerves. It was shown in at least one study that white adipocytes can release ATP via the pannexin-1 pore in response to SNS stimulation to facilitate lipolysis (256).
Adenosine is a product of ATP breakdown and acts on a separate family of purinergic P1 receptors in WAT and BAT (257). Although adenosine is created from the degradation of ATP released from sympathetic nerves within adipose tissue, it was also found that the treatment of BAT with NE alone was sufficient for stimulating the release of adenosine from brown adipocytes (102). Adenosine acting on A2A (one of the four P1 receptors for adenosine) increased thermogenesis in mouse BAT and induced browning in ing-scWAT (102). Furthermore, adenosine is rapidly deaminated by adenosine deaminase to form the nucleoside inosine (272). Inosine is secreted by apoptotic brown adipocytes to upregulate the thermogenic output of neighboring cells upon cell death (194). Inosine also acts on adenosine receptors to increase BAT thermogenesis and ing-scWAT browning (194, 274). Importantly, brown adipocytes were shown to be the primary source of inosine in mouse BAT (194), again indicating that nerves may not be the only source of these signaling molecules.
Several microscopy techniques have been developed and implemented to image and map adipose tissue sympathetic innervation or total innervation. Visualization of sympathetic fibers in adipose has primarily relied on fluorescence labeling of TH via immunostaining (27, 49, 57, 134, 206, 276) or via TH-cre mice crossed with endogenous fluorescent reporters (48, 127, 288). Current tissue processing methods for fluorescent labeling of adipose nerves can be categorized as one of the (i) embedded tissue sliced into thin cross-sections, (ii) optically cleared tissue imaged in cubes, (iii) optically cleared whole tissue, (iv) and Z-depth reduced whole mount tissues (Figure 5).
Seminal discoveries were made analyzing adipose tissue SNS through adipose tissue cross-sections [reviewed previously (29)] despite how analyzing cross-sections limits observations of true neurite density versus branching. Tissue cross-sections reduce the dense innervation of adipose tissue into discreet puncta, eliminating the ability to follow nerves on their path through the tissue. Through thin sections, arborizations are obscured and total tissue innervation can only be loosely inferred. While this method provides little structural information, it can still prove useful for providing comparisons of relative SNS density of representative tissue areas (53, 183, 289). The benefit of this technique is that the axial thickness of the sample is thin enough that widefield epifluorescence microscopy is typically suitable for many qualitative and quantitative applications. SNS fiber density can be quantified by dividing the fluorescence area by the area of the field of view to obtain relative fiber density. “Relative” is a key term here, as due to the heterogenous distribution of SNS fibers in adipose (primarily reported from ing-scWAT), a two-dimensional (2D) representative image does little to convey the true scope and complexity of the SNS innervation across the entire depot.
A vastly improved but still technically limited technique is the use of optical clearing in conjunction with sectioning small cubes of tissue for three-dimensional analysis. One such method was developed and described for use on 2 mm^3^ sections of 90% glycerol-cleared BAT, scWAT, and pgWAT. Tissues were mounted in “homemade” well-slides and imaged by confocal microscopy. Representative three-dimensional (3D) optically sectioned regions were captured for each cleared tissue cube (157). Such a technique, again, provided only relative SNS density but allows for observation of structural details such as the relative number of arborizations.
Currently, the most utilized technique for analyzing SNS fiber density utilizes optical clearing of entire adipose depots for imaging on light sheet microscopes (49, 57). Similar approaches have also been used with human adipose biopsies (127, 206). This was a revolutionary development for analyzing the SNS in adipose as researchers were now able to map out innervation patterns across entire depots, and effectively measure total SNS fiber density. Intact depots are quite large and light sheet microscopes fitted with low magnifying optics provide a quick, whole tissue assessment that leaves tissue structure mostly unchanged, aside from the well-known shrinking, expansion, and stiffening that can be associated with many clearing techniques (232).
Tissue size poses several problems for imaging. Large tissue sometimes needs to be bifurcated to fit in the sample holder (57) and thick tissues require large working distances to image the full axial thickness, which prevents higher magnifying optics from being used. Additionally, obese adipose tends to be fibrotic or full of pockets of lipofuscin, making the tissue even more autofluorescent and obscuring nerve fluorescent signals. Thus, these are common limitations for imaging obese mouse depots, compounded by the high volume of antibody needed to cover larger tissues. Frequently, 1X-4X objective magnifications have been achievable for larger or thicker samples, but low magnification optics typically come hand-in-hand with low numerical apertures which thereby limits spatial resolution and can easily obscure fine axons (i.e., small fiber innervation). For these reasons, researchers additionally image whole tissues on confocal microscopes and capture representative images as optically sectioned cubes (~0.3 mm^3^) at higher magnifications to measure SNS fiber density. This is effectively the same as how SNS fiber density is quantified in the cleared 2 mm^3^ sections.
We developed an alternative clearing-free whole depot processing technique in which tissue is gently compressed to reduce axial thickness; or “Z-depth reduction.” While this can potentially alter some aspects of tissue morphology, the total SNS fiber density remains intact (276). This technique overcomes some of the drawbacks of clearing techniques, including reduced axial thickness which allows for decreased working distances, higher magnification objectives, and higher resolutions. Using a confocal microscope, an intact adipose depot can be imaged in its entirety at low magnifications with a seamless transition to high magnification (up to 63× objective magnification—which can be further increased by applying confocal zoom). The addition of deconvolution provides unsurpassed structural resolution for intact whole adipose tissue imaging (275). Most importantly, this means that an entire adipose depot can be imaged at resolutions sufficient for visualizing single sympathetic parenchymal axons across the entire tissue and quantification is not limited to only representative images/areas (276). Protocols for each of the above methods have been made thin section (25), cleared cubed sections (157), cleared whole tissue (56), and Z-depth reduction of uncleared whole tissue (277).
Quantification of SNS fiber density varies between each method of tissue processing and imaging modality used and can be categorized as either 2D analysis, 3D analysis, or Z-maximum intensity (Z-max) analysis (Figure 5). Images captured from paraffin-embedded thin sections or single Z-planes by confocal both generate 2D data. 2D image sets are ideal for measuring colocalization but not for measuring SNS fiber density. However, a relative value of SNS fiber density can be obtained by dividing the fluorescence area (TH-labeled sympathetic fibers) by the area of the field of view (127). 2D analysis can be easily performed with open-source software (e.g., Fiji/Image J). Measuring neurite branch length in 2D images by tracing software (via Image J plugins or more expensive Imaris software) can be a waste of time, as sectioned nerves have drastically truncated neurite length. This is because the entire length of an axon cannot be captured within a single Z-plane and the portions of the axon exceeding the Z-plane will be “cut off” from the final image and not included in length measurements. However, tracing SNS fibers in 2D sections may be advisable if tissue autofluorescence is so bright that threshold-based analysis of fluorescence area was impossible. For such instances, relative SNS fiber density can be measured by combining the total length of all traced neurites/puncta and dividing this length by the area of the field of view (289). 2D tracing can be performed manually in Fiji, or with semi-automated plug-ins such as NeuriteJ (also freely available via open-source programming).
3D datasets obtained from confocal or light sheet microscopes, while more informative, can be far more labor intensive to quantify. Although researchers now can measure intact adipose depot SNS density, most studies have continued to only quantify a handful of representative cubic areas of each tissue sample (49, 65, 265), since imaging an entire tissue at a magnification sufficient to trace individual sympathetic axons at axial and lateral resolutions required for 3D skeletonization, while technically feasible, is not currently a viable approach. This is largely due to the exorbitant time it would take to image a whole tissue at sufficiently high spatial resolutions as well as the massive data file apropos of such granular imaging.
When quantifying SNS fiber density from representative 3D images, sympathetic fibers can be traced to create a 3D skeleton of SNS neurite branch length. This length, while still truncated when imaging cubed optical sections, can be used to calculate relative SNS fiber density in a 3D space. This can be calculated by the ratio of total neurite length divided by regional volume (265). Tracing nerves in 3D is often performed with high-end image analysis platforms such as Imaris (Oxford Instruments) (48, 56, 57, 65) or Amira (Thermofisher) (157). As a free alternative, the SNT plug-in for Fiji can also perform 3D neurite tracing.
Researchers can reduce the dimensionality of their samples by Z-max projecting 3D acquired image stacks into 2D. The total innervation of the tissue is captured, but the axial neurite branch length is not quantified to allow mapping of entire tissue lateral innervation. This is the preferred choice of quantification for imaging approaches that somewhat obscure axial detail, such as the Z-depth reduction method. Analyzing by this approach was found to be quantifiably identical to measuring neurite density without Z-max projection but required less time and computing power (276). Additionally, analyzing Z-max images can be implemented when researchers do not have access to high-end image analysis software as SNS fiber density can be measured in Z-max data by standard 2D approaches (fluorescence area, or 2D fiber tracing) (27, 276, 278). It should be noted that reducing dimensionality can lead to misinterpretation of structural data and that analyzing nerve branching and intersection points should be carried out with caution. There is little difference between two intersecting nerves versus two overlapping nerves when viewed as a Z-max image. Such measurements should be conducted in 3D when possible. Regardless, 2D structural tracing can be performed in Fiji or with AngioTool (157).
Another important aspect of quantifying adipose innervation is normalization, whether through image analysis quantification or molecular methodologies such as tissue western blot or ELISA. Normalizing to tissue mass or adipocyte size may make sense in terms of relative density of innervation, but numerous scenarios can represent an overall relative reduction in the tissue’s nerve (i) if the tissue/cell size goes up and innervation markers go down; (ii) tissue/cell size stays the same and innervation markers go down; or (iii) tissue/cell size goes up and innervation stays the same.
Neuroplasticity is the nervous system’s ability to grow and strengthen pertinent connections, create new patterns of connection, and prune those that are unnecessary. This is largely carried out through directed neurite outgrowth, the formation of new synaptic connections, and the strengthening of preexisting connections through numerous nervous system developmental signaling pathways (230), such as the pro-growth and pro-survival signals elicited by neurotrophic factors. The adipose depot is subject to changing size and structure throughout adulthood and in different energetic situations, and with this frequent tissue remodeling, there is also SNS restructuring in the tissue. Recently, several studies have detailed mechanisms driving neuroplasticity in adipose tissue by various neurotrophic factors (Figure 6). While several studies have defined roles for neurotrophic factors in specific cell types of adipose, most of these proteins are expressed by numerous cell types, as has been confirmed by the recent glut of adipose tissue single-cell sequencing studies. Thus whole tissue deletion studies (versus cell-type specific) and/or potential compensation by other tissue cell types need to be considered when interpreting data.
While Calsyntenin-3 is involved with synaptogenesis in the CNS, Calsyntenin-3β (Clstn3β) is a distinct form of the gene highly expressed brown and beige adipocytes which facilitates adipocyte multilocularity and lipid utilization (272, 289). When globally ablated (Clstn3β KO), Clstn3β deficiency led to impaired thermogenesis and decreased SNS fiber density in BAT, whereas adipose-specific transgenic overexpression of Clstn3β had the opposite effect in BAT. Proteomic analysis of BAT from Clstn3β KO mice showed that the calcium-binding protein S100b was strongly downregulated (289). A global S100b KO mimicked the effects of the Clstn3β KO, and also prevented positive effects of transgenic overexpression of Clstn3β, providing evidence that Clstn3β was dependent on S100b. Clstn3β was found to promote the secretion of S100b which functions as a growth factor for the sympathetic neurites in BAT (289). Because the Clstn3β-S100b axis is seemingly specific for brown and beige adipocytes, the authors propose this as an explanation for the foundational abundance of SNS innervation in BAT relative to WAT (289).
Genetically knocking out tropomyosin receptor kinase A (TrkA), the receptor for nerve growth factor (NGF), in sympathetic nerves decreased SNS fiber density in ing-scWAT with a resulting decrease in browning (134). NGF neutralization in vivo, via an NGF-neutralizing antibody, resulted in decreased browning and β-AR activation. β-ARs were required for NGF production by white adipocytes (49). Put concisely, the release of sympathetic neurotransmitters onto white/beige adipocytes activates β-ARs which stimulate NGF production and secretion from adipocytes. NGF acts on TrkA receptors on the surrounding sympathetic nerves driving increased nerve outgrowth as evidenced by the axon outgrowth marker STMN2 and increased SNS fiber density (49). This study utilized a KO of all three β-ARs so it is still unclear if all, or just one subtype, are required. Additionally, an endothelial TrkA isoform was also found to facilitate neurovascular alignment in WAT and BAT in response to PEP3, an NGF mimetic that targets the vascular endothelium of BAT and beige scWAT. Mice with endothelium-specific TrkA KO had reduced cold tolerance and demonstrated decreased nerve-vessel overlap and UCP1 expression in BAT and scWAT (10, 69). Total SNS innervation was not assessed in this model. In a related study, TrkA KO from sympathetic nerves decreased SNS density in the WAT parenchyma but left arterial innervation intact (134).
Neurotrophin-3 (NT3) is highly expressed in both WAT and BAT where it is released from blood vessels and facilitates nonshivering thermogenesis by maintaining dense sympathetic innervation to the tissue. Overexpression of NT3 in adipocytes is correlated to an increase in UCP1 expression. NT3 binds to TrkC receptors in the SChG and drives SNS outgrowth in adipose tissue, which facilitates browning, UCP1 expression, and lipolysis (34, 65). While TrkC is not highly expressed on sympathetic axons, NT3 can additionally bind to TrkA and TrkB receptors with low affinity (117). However, NT3-TrkA/B signaling has not been thoroughly investigated in adipose tissue. Furthermore, NT3-TrkC signaling may play a role outside of neurite outgrowth as there is some evidence that TrkC receptors may be expressed directly on adipocytes and function in response to NT3 activation to facilitate lipolysis (34).
Local overexpression of vascular endothelial growth factor A (VEGF-A) in scWAT and BAT stimulated blood vessel formation, increased sympathetic innervation, and upregulated β3-AR with functional increases in lipolysis, browning, and thermogenesis (294). The authors proposed that VEGF-A may directly interact with sympathetic fibers to increase neurite outgrowth, but this has yet to be shown experimentally in adipose, despite other studies demonstrating a role for VEGF as a neurotrophic factor as well as an angiogenic factor (136). Human endothelial cells (ECs) were found to secret BDNF to guide axonal outgrowth, providing additional support that angiogenesis and neurogenesis function in tandem in peripheral tissues (109), which likely includes adipose. In our own observations, we see blood vessels highly innervated by nerves in adipose, nerve bundles with their own capillary supply, and nerves and blood vessels that appear to run alongside each other (276), but it remains to be determined if angiogenesis is a requirement to precede neurite outgrowth or if axon extension is a signal for increased angiogenesis—or both.
As mentioned previously, BDNF plays a crucial role in regulating adipose SNS density in part via the leptin-BDNF axis in the CNS (265). Additionally, BDNF and its receptor TrkB are expressed in WAT and BAT, including TrkB expression on ing-scWAT-resident nerves (27). Adipocyte-specific knock-outs of Bdnf and Ntrk2 (the gene for TrkB) found that only Ntrk2 expression went down in KO mice, concluding that BDNF was not produced by adipocytes (188). Indeed, we have demonstrated that BDNF is expressed in stromal vascular cells in adipose tissue, and not by mature adipocytes (30), and BDNF may also be expressed by the nerve itself (172). Knocking out BDNF in myeloid lineage cells decreased TH and UCP1 expression in ing-scWAT following cold exposure, providing evidence that immune cells are at least one of the primary sources of BDNF in WAT (30). Myeloid-lineage BDNF KO mice presented with decreased SNS nerve fiber density, particularly around the SiLN, and a worse phenotype in response to a high-fat diet (27, 30). These data provided new evidence for the importance of BDNF in energy balance outside of the brain, of particular relevance given that single nucleotide polymorphisms (SNPs) in the human BDNF gene have been associated with obesity (3, 122, 180).
Other immune cells have been implicated in providing neurotrophic support in adipose. For example, γδ T cells release IL17 which binds to IL17 receptor C (IL17-RC) on adipocytes, thereby driving expression TGFβ1 which promotes sympathetic innervation of BAT. Ablating either γδ T cells or IL17-RC in BAT decreased tissue SNS fiber density and resulted in impaired lipolysis and thermogenesis. This was rescued by forced expression of TGFβ1, which acted as a neurotrophic factor (123). Eosinophils can infiltrate ing-scWAT in response to cold exposure and release NGF, which acts on TrkA receptors on tissue-resident sympathetic nerves to drive nerve outgrowth. Eosinophil-specific deletion of Ngf was found to blunt the effect that cold exposure had on increased SNS fiber density in ing-scWAT (173). M2-like macrophages are also home to ing-scWAT following cold exposure and secrete Slit guidance ligand 3 (SLIT3) which binds to roundabout guidance receptor 1 (ROBO1) on sympathetic nerves resulting in increased presynaptic NE release to promote thermogenesis. While SLIT3 is not functioning as a neurotrophic factor to increase SNS density, there was still the same effect of a net increase of NE release in the tissue (268).
SNA in adipose tissue is commonly stimulated systemically through cold exposure, exercise, and caloric restriction. While these methods reliably increase sympathetic innervation and nerve activity in adipose, there are whole-body effects that can confound observations. This includes an increase in circulating factors that can drive lipolysis and thermogenesis independent of the nerve activity in adipose depots. Tissue-specific targeted approaches have been developed to address this and can be implemented in specific adipose depots unilaterally, leaving the contralateral depot as a control or source of measurement of compensatory changes. Tissue-specific SNS stimulation approaches include various pharmacological, genetic, and electrophysiological methods.
Exposing mice to cold temperatures (4°C) has been a common approach to stimulate nonshivering thermogenesis and adipose SNA in BAT and WAT. Thermoreceptors on cutaneous sensory nerves are activated by a drop in ambient temperature which is relayed to the CNS, resulting in a coordinated increase in SNA in WAT and BAT. This precipitates an accumulation of brown and beige adipocytes to increase nonshivering thermogenesis, with a concordant increase in lipolysis and release of fatty acids to provide the necessary energy supply. Sustained cold exposure increases SNS activity as demonstrated by increased NETO in WAT and BAT of Siberian hamsters (38). Cold exposure also increases sympathetic nerve outgrowth through a coordinated release of neurotrophic and neurostimulating factors, many of which are released by cold-induced immune cells (27, 30, 173, 268). Cold exposure increases TH expression (a marker of SNA) in BAT more so than scWAT of male and female mice, with no observed sex differences. Interestingly, axillary (ax)-scWAT displayed significantly greater TH expression than ing-scWAT following cold exposure (28). This could be due to the proximity of the ax-scWAT depot to iBAT, as compared to ing-scWAT. Due to the ease of administering cold exposure treatments and the profound systemic effects it has, cold exposure had been the standard method of stimulating the SNS in rodent adipose tissue for decades and is a viable method for the recruitment and activation of human BAT (66, 259)
Exercise induces adipose depot-specific effects in rodents. Exercise increases sympathetic drive to WAT to mobilize lipids and induces the browning of ing-scWAT, with mixed evidence for the activation of BAT (66, 151). Chronic exercise has been shown to increase circulating BDNF and VEGF, thus increasing neurogenesis and synaptogenesis systemically (80), both of which have been shown to regulate SNS fiber density in adipose (27, 30, 294). We have shown that exercise increases innervation in the ing-scWAT (28); a finding that has since been independently replicated (197). Specifically, chronic exercise-induced browning of ing-scWAT has been observed (196) and is likely in part due to increased SNS density/activity within the tissue (28). Exercise was found to increase TH protein expression in ing-scWAT of 12-week-old C57BL/6J mice, but this effect was blunted by 65 weeks of age. Exercise had no effect on TH expression in BAT, further supporting that the increase in SNS drive is primarily to mobilize lipids (28, 196). There is also evidence that the increased sympathetic drive may be most prominent in males, as female mice did not display ing-scWAT browning (196). However, human BAT in healthy male subjects was not activated by 24 weeks of supervised exercise (171), and there is little evidence to support exercised-induced WAT browning in humans (74, 181, 204).
Diet can also play an important role in regulating adipose tissue SNS activity. Acute high caloric intake activates BAT thermogenesis through β-AR activation, presumably to burn excess fatty acids (11), but chronic high-caloric intake can result in obesity which is associated with decreased innervation and thermogenic potential of scWAT (28). On the other hand, chronic low-calorie diet can increase sympathetic drive to lipolytic WAT depots (99) as a means of mobilizing lipid fuels. For example, caloric restriction drives increased SNA in adipose tissue to upregulate the mobilization of lipids (74, 238) and food deprivation increased NETO in lipolytic WAT, but not in thermogenic BAT in Siberian hamsters (38). Interestingly, it was found that caloric restriction increased sympathetic innervation of rpWAT, but not in pgWAT in rats (99). It has since been pointed out that diet-induced changes in SNS drive are not only depot specific, but temporally-regulated. SNS drive in pgWAT was highest early in caloric restriction (3 days) and decreased as the diet was maintained. By day twelve of caloric restriction the SNS drive to pgWAT returned to baseline and the drive to scWAT was significantly increased. Again, SNS drive to BAT was unchanged by caloric restriction, but contrary to previous observations, increased SNS drive to rpWAT was not observed throughout caloric restriction (238).
Increasing sympathetic drive systemically via cold exposure, exercise, or by caloric restriction causes changes in the differential sympathetic drive to WAT and BAT. By performing systemic interventions, we learn more about adipose tissue as it functions as an interconnected organ (98) in response to whole-body metabolic demand and communication with other tissues/organs. However, to parse the functions of adipose tissue innervation not confounded by whole-body interventions, more targeted approaches must be employed.
Numerous pharmacological approaches have been developed that simulate sympathetic nerve activation by acting as β-AR agonists. CL-316,243 is highly selective agonist for β3-ARs that has been used extensively to target WAT and BAT (120, 138, 158). Isoproterenol, a pan-AR agonist, has also been used to induce browning (138, 175). Some other pharmaceuticals acting synergistically with SNS input include Rosiglitazone, a PPARγ agonist that induces browning (138, 203), and triiodothyronine which stimulates thermogenesis and BAT hyperplasia (138, 160, 283). While these drugs function to mimic sympathetic activity (sympathomimetics), they do not alter SNA or neural plasticity. Therefore, all effects are independent from basal tissue innervation and sympathetic tone. While some pharmacological agents can be delivered directly to adipose, most are given systemically and therefore mimic the global sympathetic drive characterized by adrenal release of catecholamines and not the specific SNS outflow to adipose that occurs physiologically.
A handful of genetic approaches have been developed to target SNS stimulation directly to adipose tissues. Channelrodopsin-2 (ChR2) is a blue light (~470 nm) sensitive cation channel that can be genetically inserted into the mouse genome and when stimulated by blue light, causes depolarization of nerves and the release of neurotransmitter from the presynaptic terminal/neuroeffector junctions. This is commonly achieved in vivo through positioning a light-emitting optic fiber into the target organ (50), although a wireless implantable LED has also been developed specifically for use in optogenetic stimulation of adipose tissues (246). Generation of ChR2-expressing sympathetic nerves in adipose has been achieved by crossing TH-cre mice with Rosa26-LSL-ChR2-YFP mice, or alternatively, through viral vector gene delivery. Adeno-associated virus (AAV) is a nonpathogenic parvovirus that is commonly used as a vector for gene delivery, and AAV containing the human synapsin (hSyn) promoter is neuron-specific (23, 195). Injection of a retrograde AAV6-hSynChR2(H134R)-EYFP directly into mouse adipose was used to specifically transfect sympathetic nerves in iBAT to express ChR2. Alternatively, a Cre-dependent AAV vector injected into TH-cre mice could also provide sympathetic nerve-specific transfection (106). Optogenetic stimulation of SNS fibers induced lipolysis in ing-scWAT (288) and promoted thermogenesis in iBAT (132, 163). A recent development and important consideration for light-stimulating techniques is the finding that white and brown adipocytes express encephalopsin/opsin3, a blue light-sensitive photoreceptor. Opsin3 activation by blue light stimulation of adipocytes causes increased lipolysis and thermogenesis, which are absent in Opsin3 KO mice (189, 226). Both ChR2 and Opsin3 display peak activation when excited by 470 nm wavelength light (187, 241). As such, wild-type mice should be included in all optogenetic studies that use blue light to target adipose tissue. It is also important to note that optogenetic stimulation of the SNS in peripheral tissues, including adipose, has not been thoroughly validated by electrophysiology and that the determination of effectiveness, especially in adipose tissue, has been primarily based on phenotypic changes such as increased heat expenditure, UCP1 expression, and changes in fat mass (132, 163, 288).
DREADDs have become a common chemogenetic approach for exciting or inhibiting neural activity. DREADDs are specialized GPCRs that are inserted into the genome by mouse transgenic approaches or via AAV vectors, and which are only activated by a synthetic ligand; most commonly clozapine-N-oxide (CNO) (222). Very little work thus far has utilized DREADDs in adipose tissue to excite or inhibit sympathetic nerves. A recent study utilized adipose-specific Gq-coupled DREADD mice activation by CNO via intraperitoneal injection to decrease the lipolysis caused by type 2 diabetes (144). Prior to that, the function of free fatty acid receptor 2 (FFA2)-activation in pgWAT was investigated using FFA2-DREADD mice (32). Like with AAVs, there are now Cre-dependent DREADD mice that can be used to help target DREADDs to sympathetic nerves by crossing with TH-cre mice (295). With AAV-based gene therapies becoming FDA-approved and seeing clinical use to target the CNS in human patients (148, 162), the development and application of gene therapies (AAV or DREADD) to target adipose tissue and tissue nerve supply are becoming more and more important. However, as with optogenetics, thorough validation (by electrophysiology) of chemogenetic approaches to stimulate or inhibit the SNS in adipose tissue is still needed. Additionally, the “designer drugs” used are not as innate or specific as once thought and must be controlled for within each experimental design (104, 166). Optogenetic approaches can cause phototoxicity (167), and both opto- and chemogenetic approaches can also cause excitotoxicity (24, 208), rebound activity (73, 154), and the potential for desensitization from prolonged activation (62, 73).
Finally, electrical stimulation of adipose tissue nerves was found to be a viable approach for driving thermogenesis in BAT, as demonstrated by hook electrode stimulation of the cervical afferent neural input to BAT in mice (159) and by electrical field stimulation of the dorsal skin in rats (129). Vagus nerve stimulation was used to confirm vagal input to BAT and determined that the vagal afferent innervation of BAT could function to inhibit SNA and thermogenesis in rats (165, 179). Interestingly, with electrical stimulation of neural input to BAT, the TH protein content was also increased as measured by immunostaining (159). It is unknown if nerve activity alone drives tissue innervation, although literature in other tissues suggests this could be a mechanism (103, 205).
SNA can be directly assessed by measuring NETO, a method used to gauge NE concentrations by competitively inhibiting TH with alpha methyl-para-tyrosine (AMPT) injections, and is the preferred option for measuring NETO in rodents, though there is the alternative of using [H^3^]NE (19, 262, 285). The rate of NE decline is directly proportional to the starting concentration of NE within the tissue as measured by high-performance liquid chromatography (HPLC) (121, 260). Alternatively, SNA can be assessed by direct nerve recording (262). Electrophysiology recordings from the nerve bundles innervating adipose depots have been performed in WAT (201) and BAT (164, 165, 176, 199) of rodents but must be rigorously controlled and is not easily achieved in small animals like laboratory mice due to the small size and the difficulty to control for breathing. There are also several caveats with this method. Because adipose nerves traverse in mixed bundles, researchers cannot be certain that they are only recording from sympathetic fibers. Multiple nerve bundles from separate levels of the SChG innervate each depot with unequal contribution from each bundle, which varies between subjects (91).
Recent developments in measuring NE activity, such as the insertion of genetically encoded GPCR activation-based fluorescent NE (GRABNE) sensors (88) and NE optical tracers (79) will likely prove to be valuable tools for analyzing SNS activity in adipose moving forward. It is also important to point out that increased NE does not equate to increased innervation which needs to be measured by neurite density quantification (Figure 5) or by western blot for neurite and nerve terminal markers (28), and innervation does not indicate increased nerve activity which would be demonstrated by TH or NE levels.
Type II diabetes is the leading cause for peripheral neuropathy, or the loss of tissue innervation. Leptin-deficient mice homozygous for the spontaneous Lep^ob^ (ob/ob) mutation, and LepR-deficient mice homozygous for Lepr^db^ (db/db) are prone to obesity and diabetes. Ob/ob mice display reduced TH expression in scWAT and BAT, as well as decreased expression of GAP43 (neurite-outgrowth) and PSD95 (postsynaptic density) in ing-scWAT (28, 275). Db/db mice also have decreased SNS fiber density (265). The reduced SNS fiber density in WAT and BAT of ob/ob mice was mitigated by chronic leptin treatment (265). It was shown that Bdnf expression increased with obesity in mouse adipose tissue, potentially to compensate for the observed down-regulation of the receptor Ntrk2 (188). However, an increase in a gene’s mRNA level does not guarantee an increase in its protein levels, as posttranscriptional, translational, and protein degradation regulation, all play a role in maintaining protein abundance. On the other hand, protein expression of BDNF was shown to be reduced in WAT with obesity (137). This fits with the data on leptin acting on the BDNF axis, but it is unclear exactly how leptin promotes tissue innervation and through which cell types, including potential actions on the nerve itself. Leptin injected into pgWAT was found to stimulate afferent sensory nerves and facilitate increased SNA as measured by electrophysiology, suggesting that leptin can act directly on peripheral nerves in addition to the CNS (198). Protein expression of the pan-neuronal marker PGP9.5 decreased in correlation with increasing age and BMI in humans (28). There is emerging evidence that obesity-related neuropathy in scWAT may be demyelinating in nature (275), similar to what has been reported for diabetic polyneuropathy in peripheral tissues (85, 233).
The other leading cause for peripheral neuropathy is aging (37, 261). Aged C57BL/6J mouse ing-scWAT and BAT had decreased pan-neuronal PGP9.5 expression, with nerves visibly receding away from vasculature in ing-scWAT. Interestingly, there was no significant change in TH expression with advanced age, which may suggest a loss of sensory innervation rather than sympathetic nerves, and a distinct etiology from obesity-related adipose neuropathy (28), or simply a lack of change in SNS nerve activity. Gene expression of NTF3 (which encodes the neurotrophin NT-3) was downregulated with age in human retroWAT (34). Predisposition to age-related peripheral neuropathy seems to be in part genetically influenced, as genetically diverse HET3 mice demonstrate a blunted neuropathy phenotype compared to inbred C57BL/6J mice (278).
Due to the extensive physiological roles of the SNS across numerous tissues and organs, many drugs aimed at modulating SNS activity have been developed. Many types of sympathomimetics (SNS agonists) and sympatholytics (SNS antagonists) exist and can be either direct acting (acting on α- and/or β-ARs), indirect acting by modulating NE release/availability, or a combination of both mechanisms. Although used in herbal medicine for thousands of years (e.g., ginseng) (280), the first sympathomimetic drugs were not synthesized until the early 1900s (214). Most therapeutics aimed at obesity seek to increase sympathetic drive to adipose tissue to stimulate lipolysis and increase energy expenditure, thereby decreasing fat mass. Therefore, for a time sympathomimetic drugs became quite popular as weight loss remedies. However, historically, pharmacological studies targeting the SNS for obesity treatment have performed poorly in clinical trials, most often due to the off-target stimulation of ARs, specifically cardiac β1- and β2-ARs, leading to adverse heart effects. The use of sympathomimetics for obesity treatment is extensively reviewed elsewhere (128, 182), thus, we only discuss a few of these compounds that were investigated for their potential to modulate adipocyte metabolism.
Phenethylamine compounds and their derivatives (i.e., amphetamine, diethylpropion), phenylpropanolamine, and phentermine are indirect sympathomimetics that increase the release of NE. Although they may be used to treat a variety of ailments, from attention deficit hyperactivity disorder (amphetamine) to sinus/chest congestion (phenylpropanolamine), they were also once widely prescribed as weight loss drugs due to their appetite-suppressive effect on the CNS (149). Peripheral effects of these drugs include increased thermogenesis and lipolysis through increased catecholamine release, but they also produced adverse cardiovascular effects, such as increased heart rate and hypertension, especially in high doses and in patients pre-disposed to cardiac problems (including a large population of patients suffering from obesity). In 2000, the release of a longitudinal stroke study linking phenylpropanolamine use to increased risk of hemorrhagic stroke (141) led the Food and Drug Administration (FDA) to ban the marketing of phenylpropanolamine as a weight loss drug and subsequently banning it all together. Following this, the use of these drugs for weight loss, and the use of stimulants in general, was much more controlled and discouraged (244).
In the 1990s, a combination drug therapy that included fenfluramine and phentermine, better known as fen-phen, became widely popular. Unlike phentermine, fenfluramine increases extracellular serotonin (5-HT) levels which results in appetite suppression (229). As a combination drug, fen-phen was more effective at long-term weight management and appetite control and initially had reported fewer adverse cardiac effects compared to long-term phentermine use (271). Thanks to pervasive marketing fen-phen became a weight-loss sensation, and the combination drug became widely prescribed despite not being FDA-approved for long-term use. The excitement was short lived, however, when an alarming percentage of patients (mostly women) taking the drug developed heart valve damage and pulmonary hypertension (63), which led to market withdrawal of the drug in 1997. Phentermine remains the most widely prescribed weight loss drug in the United States, followed by diethylpropion (250), but only for short-term use and in combination with behavioral therapies (nutritional changes and exercise) (139).
Sibutramine was a combination of serotonin and NE reuptake inhibitor, initially intended to be marketed as an antidepressant, that incidentally showed dose-related weight loss effects in clinical trials (200). As with amphetamines and phentermine, sibutramine’s weight loss effects were mediated by a greater availability of NE in the periphery. Yet sibutramine was considered a promising replacement for amphetamine drugs since it did not appear to have the same risk for dependency and abuse, despite showing some “small” impacts on heart rate and blood pressure (200), and was marketed in the United States and abroad for nearly a decade (128). However, in 2010, following the Sibutramine Cardiovascular OUTcomes (SCOUT) trial, the FDA recommended against the use of sibutramine due to unacceptable adverse cardiovascular risks (86). The rise and fall of sibutramine as an obesity panacea underscored the need for a selective sympathomimetic that would only target adipose tissue and avoid undesirable off-target effects that could increase health risks in patients with obesity.
Mirabegron, another sympathomimetic approved by the FDA in 2012 to treat overactive bladder (36), has recently been reevaluated as a potential treatment for obesity after it was shown to activate β3-AR in human brown and beige adipocytes, thereby increasing lipolytic and thermogenic activity (22, 54). Mirabegron showed promise as an anti-obesity drug as it appeared to be a selective agonist of β3-AR (36), and therefore had the potential to minimize off-target adverse effects by not acting on β1-, and β2-ARs in the heart and other tissues. The daily dose of mirabegron is 25 to 50 mg for the treatment of overactive bladder (36); however, at the high end of that therapeutic dose (50 mg) mirabegron does not stimulate human BAT oxidative metabolism (31). Only the maximum allowable oral dose of 200 mg elicited lipolysis and increased BAT blood flow and BAT oxidative metabolism (22). At this dose, mirabegron can produce adverse cardiovascular effects (22), and it has recently been shown to stimulate all 3 β-ARs at the maximal dose (31). The same study also argued that human BAT thermogenesis is driven by β2-AR, using a β2-AR knockdown model to demonstrate that human BAT thermogenesis is impaired under such conditions (31). Although it would appear that the lipolytic and thermogenic effects of mirabegron in BAT came from cross-reactivity with other β-ARs, if the main goal of treating obesity is reestablished energy balance, there may be hope yet for mirabegron as an anti-obesity drug. At the therapeutic dose (50 mg) which produces no adverse cardiac events, energy expenditure increased by 12% over basal levels (31), which is not much less than the 17% increase in energy expenditure at the max allowable dose (200 mg). Taken together, the use of pharmaceutical agents to target adrenergic signaling in adipose is promising as a means to drive increased energy expenditure in the promotion of metabolic health but requires further refinement.
Perhaps targeting the SNS may not be advantageous in obesity treatment. Aside from the continued controversy regarding which β-ARs are physiologically relevant in human adipose, the question of increased lipolysis remains. Circulating FFAs released from adipose lipolysis have been linked to coronary arterial disease (282) and sustained NE release has been linked to increased systemic inflammation (213), in addition to known risks of lipotoxicity if the released FFAs are not safely stored, excreted, or oxidized. Instead, approaches that improve adipose-brain neural cross talk to promote tissue plasticity and healthy function may be preferable as a physiological approach; however, the impressive ability of the nervous system to counter-balance increased energy expenditure (at any body weight or health status) by increasing appetite, will consistently be a therapeutic challenge. There is promise in glucagon-like peptide-1 receptor agonists (GLP-1RAs), a class of antidiabetic drugs recently used to treat obesity. GLP-1RAs are effective at reducing food intake, maintaining satiety, and improving glucose control and have low adverse effects (6). Ongoing clinical trials will determine whether GLP-1RAs are effective at long-term weight management or if they impact SNS activity or innervation patterns.
It is important to remember that the innervation of adipose tissue is not restricted to vasculature and adipocytes. The adipose organ houses a wide variety of cell types, each with their own needs for innervation and neural communication with the PNS (Figure 7). The innervation of these support cells in the adipose stromal-vascular fraction (SVF) can also have consequences for adipocyte function, nerve cross talk, and the overall health of the adipose tissue, which is discussed in this section.
The current breadth of neuroimmune regulation in WAT has recently been reviewed by Meng co-workers (212), therefore, it will only be briefly discussed here, and illustrated in Figures 7A-7B. Neuroimmune cells are closely associated with nerves in many tissues, including adipose, and engage in bidirectional cross talk with the nervous system to maintain tissue function and homeostasis. Notably, most neuroimmune monocyte/macrophages identified thus far that interact with sympathetic nerves in adipose express the cell surface marker Cx3rc1, a receptor once thought to only exist on microglia (27, 44, 209). These cells, alternatively called nerve-associated macrophages (NAMs) (44) or sympathetic-associated macrophages (SAMs) are present in both BAT and WAT and have been shown to degrade NE (209). In obesity, SAMs are recruited to adipose tissue and presumably contribute to impaired energy balance by degrading too much NE and limiting its energy-expending functions. Our own research has identified a myeloid-lineage monocyte/macrophage marked by Ly6c+Ccr2+Cx3cr1+ that infiltrates scWAT after cold exposure (27, 30). These cold-induced neuroimmune cells (CINCs) exhibit a genetic profile associated with axonal remodeling, neurogenesis, synaptogenesis, and regulation of neuronal survival (27) and may act opposite to SAMs by promoting healthy innervation of adipose, especially in response to energy-expending stimuli like cold. Eosinophils, another myeloid cell type, have recently been shown to regulate axonal plasticity in WAT via NGF release, and they are found in close proximity to nerves, similar to NAMs and SAMs (173). Separately, deletion of NGF from eosinophils and eosinophil depletion resulted in reduction of cold-induced axonal outgrowth and browning in inguinal scWAT (173). Eosinophils are also found in close association with sympathetic nerves in various adipose tissues of humans, including omental and subcutaneous (173), where they may play a similar role. Eosinophils are recruited to adipose tissue via group 2 innate lymphoid cells (ILC2s) that release IL5 in response to IL33, which is released by adipose stromal cells following sympathetic signaling by NE (173).
Cross talk between nerves and immune cells in adipose is bidirectional. Immune cells highly express ARs (105, 110), and sympathetic nerves signal to immune cells via various nerve products including catecholamines, in what is termed a “neuroimmune synapse.” In pgWAT and BAT, sympathetic innervation has been implicated as a regulator of inflammation via the β2-AR-PKA axis (247). Mast cells (MCs) have been implicated in regulating browning in human adipose tissue via histamine and IL4 release (89), and degranulation of MCs is stimulated by the autonomic nervous system (153); however, sympathetic signaling on MCs in adipose tissue has not been directly shown. Contradictory to these findings, inhibition of MC degranulation in adipose of mice promoted browning in scWAT (292). Serotonin released by MCs inhibits scWAT browning, while MC-derived histamine and IL4 promote scWAT browning. MC activating is controlled by the autonomic nervous system, while SNS signaling inhibits MC degranulation parasympathetic signaling enhances MC degranulation (153).
Schwann cells, although not quite immune cells, provide support to adipose innervation by secreting BDNF (Figure 7B) (275).
Besides immune cells, the adipose SVF includes mesenchymal cells (MSCs), endothelial precursor cells (EPCs), ECs, smooth muscle cells, lymphocytes, pericytes, and preadipocytes, and over time it is becoming clear that the SNS can interact with this entire milieu of cells. This may be via direct contact, the formation of specialized nerve terminal structures in the SVF, or en passant release of nerve products from axonal varicosities. It is likely that all cell types residing in adipose maintain uni- or bidirectional communication with the CNS via peripheral innervation including the SNS, and SNS modulation of one cell type may affect whole tissue physiology.
Cardoso et al. recently identified “neuro-mesenchymal units” that regulate ILC2 function in pgWAT. This study used multiple mouse models as well as 6-OHDA sympathetic denervation experiments to demonstrate how sympathetic nerves signaling on adipose MSCs via the β2-AR control expression of glial-derived neurotrophic factor (GDNF) and the activity of ILC2s in pgWAT (51). 6-OHDA sympathetic denervation was shown to reduce GDNF expression in pgWAT, specifically in Pdgfra^+^ MSCs, while the sympathomimetic clenbuterol increased GDNF expression (51). The authors also demonstrated that GDNF regulated ILC2 function via the GDNF co-receptor RET and that this ILC2-GDNF signaling controlled adipose homeostasis and obesity (51). Even though SNS denervation/activation altered ILC2 cytokine production, the authors excluded direct SNS communication with ILC2s in pgWAT by demonstrating that ILC2 function was unaffected by in a Il7ra-Cre X Adrb2fl/fl mouse model, which deleted β2-adrenergic receptor in lymphoid cells (51). However, lymphoid cells express multiple catecholamine receptors along with β2-AR, including β1-, α1-, and α2-AR (227), so it is, therefore, uncertain whether SNS signaling was maintained in this Il7ra-Cre x Adrb2fl/fl mouse model via compensatory signaling through another catecholamine receptors.
In fact, SNS innervation of lymph nodes (LNs) would suggest that direct SNS signaling on lymphocytes occurs. All LNs are surrounded by adipose tissue, and major drainage LNs reside within discrete adipose depots, such as the SiLN in ing-scWAT—a site of much nerve remodeling following cold exposure (184). Therefore, SNS innervation, or lack thereof, to these adipose lymphoid organs may have direct effects on the health and function of the adipose tissue and may be the link between obesity and lymphedema. Felten et al., 1984 demonstrated that the popliteal and mesenteric LNs are highly innervated by the SNS, with nerve fibers appearing to “synapse” on immune cells inside the LNs (87) (Figure 7C). Although all LNs are surrounded by adipose tissue, it was not until recently that autonomic innervation of LNs and lymphatic vasculature in ing-scWAT was shown (12, 30, 124). Autonomic innervation of lymphatic structures appears to be involved in regulating lymphatic flow (12), again linking to obesity-induced adipose edema. Both sensory (124) and sympathetic (87) nerves penetrate the LN capsid (Figure 7C, right panel). While sympathetic nerves have been shown to terminate within the LN amongst lymphocytes (87); there has been strong evidence that bidirectional cross talk occurs between sensory nerves within LNs, with lymph node-resident cells, including lymphocytes, stromal cells, and lymphatic endothelial cells (124). Within the past few years, it has become clear that lymphatic structures within adipose tissue are highly innervated and serve as a nexus of neuroimmune cell cross talk. Surprisingly, unlike in other tissues, the initial lymphatics as well as the lymph node are innervated in adipose (Figure 7D).
Blood vasculature is highly innervated by the SNS (Figures 7C and 8), and for a long time, it was believed that adipose innervation was just innervation of the vasculature within the adipose depot (279). While vasculature in both depots displays a similar degree of innervation, pgWAT lacks the widespread parenchymal innervation evident in ing-scWAT (27, 57). Although we now know that individual adipocytes receive just as much direct innervation as blood vessels, teasing apart whole organ effects from aberrant innervation of vasculature will be a challenge for future studies.
Parenchymal and neurovascular innervation can be differentiated and quantified separately using some of the processing and imaging techniques in conjunction with vascular labeling, as discussed previously. Firstly, the same principles discussed for measuring neurite density are at play (Figure 5). Fluorescence labeling of adipose blood vessels can be achieved with fluorophore-conjugated Isolectin-IB4 (IB4) which has been used extensively in adipose tissue with and without optical clearing (29, 48, 276, 278). Here we have provided labeling with TH and Isolectin-IB4 in whole mount ing-scWAT (Figure 8) with the protocol made available (277). Alternatively, antibodies against CD31/PECAM-1 (48) or endomucin (157) have also been used in adipose, which may be necessary since IB4 can also stain some sensory nerves (84).
Following labeling, neurovascular innervation can be differentiated from parenchymal innervation by dense small fiber arborizations that contact blood vessel walls which branch from larger nerves running parallel to the vessel (276) (Figure 8). Parenchymal innervation would accordingly be the small nerves not directly associated with vasculature, excluding large nerve bundles (Figure 8). While capillaries rarely receive input from the SNS (234), many of the small nerves in the adipose tissue parenchyma traverse alongside capillaries (Figure 8). This is due to limited interstitial space between adjacent adipocytes which funnels nerves and capillaries into the same tunnels/pathways between adipocytes. Even NANs often stem from nerves traversing along capillaries shown here (Figure 8), and previously (276). This does pose a problem for neurovascular quantification as some of the methods described below cannot differentiate between nerves that are innervating arterioles, and those that are simply traversing with capillaries.
Relative neurovascular innervation can be quantified in thin tissue sections by counting nerve puncta in a predetermined area around blood vessels, or by measuring the fluorescence area in a region of interest (ROI) around the blood vessel of choice and normalizing to the area of the ROI. If 3D datasets are acquired the neurovascular innervation can be traced by hand separately from parenchymal innervation and used to generate relative density values as described previously. This is a time-consuming approach but has the potential for being the most accurate. Alternatively, the ratio of relative nerve area to relative vascular area can provide a gross analysis of the changes in total innervation across an entire tissue relative to its vascularity (278). This method lacks the granularity of 3D tracing, and albeit informative, does not differentiate neurovascular innervation from parenchymal innervation. It may be tempting to analyze colocalization to provide this needed detail but keep in mind that the nerves and blood vessels are separate structures and, if imaged at appropriate resolutions, should not colocalize in 3D space. However, by Z-max projecting 3D data, the nerves contacting the vessel above and below are digitally superimposed into the same 2D space as the blood vessels which will yield a “pseudo-colocalization” or “overlap” coefficient (278). The addition of a binary dilation of the blood vessel image by a pixel or two will also cause the nerves contacting the vessel at the sides to be included into this overlap if desired. The Mander’s colocalization coefficients will describe the amount of nerves that overlap with blood vessels, as well as the amount of blood vessels that overlap with nerves in the Z-max image (278). This is a useful technique for quantifying changes in relative neurovascular innervation separately from parenchymal innervation and vice versa (278). But again, this is not a true colocalization analysis as colocalization was induced by Z-max projecting the data and must be reported as such. This method is also prone to overestimating the extent of overlap in larger Z-stacks. This is because as the Z-axis increases in size so too does the likelihood for a nerve and blood vessel separated in space to appear to overlap once converted to 2D. Setting Z-stack thickness just above and below the blood vessel of interest is recommended when possible.
In conclusion, adipose tissue nerves are essential for tissue function and whole-body metabolic regulation, and more is understood about adipose sympathetic nerves than sensory nerves, despite the new finding that traditional markers for the SNS (such as TH) also mark a large subset of sensory nerves. Data from SNS denervation studies in adipose or SNS mimetic treatments have clearly demonstrated the importance of these nerve products (including NE and modulatory neuropeptides like ATP and NPY) for the function of cells including adipocytes, vasculature, and stromal vascular cells. Neuroimmune cross talk, local production of neurotrophic factors, and other signals that can be stimulated by physiological means (including exercise and cold stimulation) play an important role in regulating the amount of tissue innervation, as well as nerve activity levels.
Despite these substantial advancements in understanding, numerous major gaps remain, as