The metabolic and functional roles of sensory nerves in adipose tissues

Abstract

Homeostatic regulation of adipose tissue is critical for the maintenance of energy balance and whole-body metabolism. The peripheral nervous system provides bidirectional neural communication between the brain and adipose tissue, thereby providing homeostatic control. Most research on adipose innervation and nerve functions has been limited to the sympathetic nerves and their neurotransmitter norepinephrine. In recent years, more work has focused on adipose sensory nerves, but the contributions of subsets of sensory nerves to metabolism and the specific roles contributed by sensory neuropeptides are still understudied. Advances in imaging of adipose innervation and newer tissue denervation techniques have confirmed that sensory nerves contribute to the regulation of adipose functions, including lipolysis and browning. Here, we summarize the historical and latest findings on the regulation, function and plasticity of adipose tissue sensory nerves that contribute to metabolically important processes such as lipolysis, vascular control and sympathetic axis cross-talk.

The body’s energy balance homeostasis is controlled by the central nervous system (CNS) and peripheral nervous system (PNS). White adipose tissue (WAT) and brown adipose tissue (BAT) play crucial roles in storing and expending energy, respectively, and both are innervated by the sensory and sympathetic nerves of the PNS (Figs. 1 and 2). While sensory nerves that innervate tissues and organs are typically interoceptive or somatosensory and therefore inform the CNS of input from the internal and external environments through afferent signalling, the sympathetic nerves allow efferent communication from the CNS to the periphery to exert effects on tissue and organ functions. Over the years, multiple lines of evidence from studies that have ablated adipose innervation by chemical, surgical or genetic denervation techniques (Fig. 3) have confirmed that adipose tissue nerves are critical for metabolic regulation^1–9.

Fig. 1 |. Overview of adipose tissue innervation by sensory and sympathetic nerves and associated release of nerve products (neuropeptides and neurotransmitters).

a, Cartoon of a typical pseudo-unipolar sensory neuron as it innervates adipose tissues. b, Sensory–sympathetic loops between the adipose tissues and the CNS. Sensory nerves (red arrows) stereotypically partake in afferent signalling (as electrical action potentials) from adipose tissue to the CNS via the DRG, and simultaneously release sensory neuropeptides in adipose tissues. Sympathetic nerves (blue arrows) partake in efferent signalling from the CNS to adipose tissues via the sympathetic chain ganglia and release products such as norepinephrine (NE).

Fig. 2 |. Anatomical distribution of sensory afferent cell bodies across levels of the DRG in rodents (Siberian hamster, mouse and rat) and corresponding peripheral tissue and organ sensory innervation projections.

The DRG houses cell bodies of peripheral sensory nerves that innervate organs and tissues throughout the body (peri-gonadal white adipose tissue (pgWAT) or epididymal WAT (eWAT), peri-renal white adipose tissue (prWAT), BAT and ing-scWAT). For simplicity, the schematic shows only unilateral and not bilateral innervation of organs by sensory afferents. Direction of the arrows shows the direction of sensory neuropeptide release into peripheral organs. DRG levels: C, cervical; T, thoracic; L, lumbar; S, sacral.

Fig. 3 |. Commonly adopted methods to study adipose innervation in rodents, including adipose denervation techniques.

a, Techniques used to visualize tissue innervation include light-sheet microscopy, whole-mouse or tissue clearing, whole-mount immunofluorescence labelling, intravital imaging and two-photon imaging. b, Methods of achieving sensory denervation of adipose tissues in rodents include chemical denervation, surgical denervation, viral denervation and genetic denervation. 6-OHDA, 6-hydroxydopamine; IB4-Saporin, isolectin B4-Saporin; RTX, resiniferatoxin.

Sensory nerves in adipose tissues release neuropeptides such as calcitonin gene-related peptide (CGRP) following activation, whereas sympathetic nerves in adipose tissues release neurotransmitters, most prominently norepinephrine (Fig. 1). For decades, sympathetic nerve activity in adipose tissues has been extensively studied because of its clear involvement in mediating negative energy balance states, such as cold tolerance and non-shivering thermogenesis. Additionally, norepinephrine triggers lipolysis, or the breakdown of stored triglycerides into free fatty acids and glycerol, and can promote the development of inducible/recruitable brown adipocytes in WAT. In the mitochondria-rich brown adipocytes, norepinephrine also elicits uncoupling protein 1 (UCP1) expression and activity to stimulate non-shivering thermogenesis.

Sensory nerves may also contribute to these processes, but their functions in adipose tissues remain vastly understudied. More evidence for the roles of sensory versus sympathetic nerves in metabolic processes is important, particularly in conditions such as obesity, diabetes and ageing, where adipose tissues can experience neuropathy or loss of innervation, which may affect adipose sympathetic, sensory or both nerve types¹⁰. Here we provide an up-to-date summary of the state of knowledge regarding adipose tissue sensory nerve functions in adipose tissue and metabolic homeostasis, taking into account the historical research literature on this topic.

Structural and functional overview of adipose sensory nerves

Sensory nerves innervate peripheral tissues and organs, contributing to various functions such as inflammation regulation, pain sensation, thermoregulation, metabolic homeostasis and energy balance^11,12. Most sensory nerves have pseudo-unipolar axons that extend in two directions from the cell body, towards the peripheral tissues and towards the CNS (Fig. 1a). The cell bodies of sensory neurons are housed in the dorsal root ganglia (DRG) outside the spine, along with associated glial and neuroimmune cells. Different ganglia levels (including cervical (C), thoracic (T), lumbar (L) and sacral (S); Fig. 2) provide sensory innervation to peripheral organs and tissues along the anterior to posterior axis. While many tissues and organs receive additional vagal sensory innervation as well as these ‘spinal’ afferents, current knowledge indicates that the vagus nerve does not innervate adipose tissues. This anatomical organization of sensory neurons across different DRG levels is largely conserved across different species, as evidenced by multiple viral tracing studies that tracked adipose innervation back to the DRG cell bodies and/or up to the CNS control centres^13,14. Information from peripheral tissues is transmitted to the CNS as sensory nerve action potentials (electrical activity) across multiple synapses between neurons that form a line of communication from the tissue to the brain. This sensory nerve activation is accompanied by neuropeptide synthesis (in the DRG cell body) and release of neuropeptide from dense core vesicles stored in the nerve endings that terminate in the peripheral tissues. Although the best-studied sensory neuropeptides are CGRP and substance P, other sensory neuropeptides are also functional in the PNS (Fig. 1b) in multiple nerve subtypes.

While CGRP and substance P are mainly expressed in sensory ganglia, sympathetic ganglia innervating adipose tissues may also express these neuropeptides, particularly during altered energy balance states¹⁵. Specifically, cold exposure (8 °C for 1 week) boosted sensory neuropeptide (CGRP and substance P) gene expression in the T13 and L1 sympathetic ganglia which innervate the inguinal subcutaneous WAT (ing-scWAT). Others have reported that sensory ganglia innervating adipose tissues express tyrosine hydroxylase (a protein typically associated with sympathetic nerves, and which increases in expression following neural activation as the rate-limiting enzyme for catecholamine production)³. Similarly, in other organs and tissues around the body, secretomotor sympathetic neurons express CGRP and release acetylcholine, a neurotransmitter that is abundantly expressed in the parasympathetic nervous system (which is probably absent in adipose tissues)¹⁶. Last, sensory axons that innervate adipose tissues also express tyrosine hydroxylase; around 40% of sensory nerves in mouse ing-scWAT were immunopositive for tyrosine hydroxylase in a recent study by Wang et al.³. This discovery challenges previous assumptions regarding tyrosine hydroxylase as solely a sympathetic nerve marker in adipose studies, and suggests a potential role in sensory nerves as well³. Collectively, these findings highlight both the heterogeneity of PNS nerves innervating peripheral tissues and the plasticity of adipose-innervating nerves.

Sensory innervation in adipose tissues

Mapping sensory nerves innervating adipose tissues

Viral tracing of neuronal circuits is a common technique used to map the connections between tissues and the CNS, often using fluorescently tagged recombinant viruses that selectively infect and label specific groups or subtypes of neurons. Historically, sensory nerve projections from WAT and BAT were mapped to the hypothalamus (and other brain regions) via the DRG using viral tracing studies in rodent models^17,18. Sensory projections in C4–C8/T1–T2 DRG were traced from BAT in Siberian hamsters¹⁷ and sensory neurons in T13/L1–L3 DRG were traced from ing-scWAT of rats¹³ and mice³. Visceral WAT depots such as epididymal WAT (eWAT) in Siberian hamsters and perirenal WAT in rats were traced back to T10–T13/L1–L2 DRG¹⁹ and T13/L1 DRG²⁰, respectively (Fig. 2). Sensory innervation of other visceral depots such as mesenteric WAT and retroperitoneal WAT remains unmapped. Lastly, sensory nerves project from WAT and BAT in a multi-synaptic manner back to the brainstem and hypothalamus, specifically to nuclei that regulate sympathetic outflow to adipose depots, thus pointing to a neural feedback loop between BAT/WAT and the CNS, and the potential for ‘command neurons’ in the brain that may coordinate tissue cross-talk between the sensory and sympathetic systems¹⁸.

Receptors on sensory nerves innervating adipose tissues

Sensory nerve receptors can be classified into five groups: mechanoreceptors, thermoreceptors, nociceptors, electromagnetic receptors and chemoreceptors²¹. Transient receptor potential (TRP) channels, which encompass receptors from all five groups, are extensively studied in sensory nerves, and some work on TRPs has been undertaken in adipose tissues. TRPs are classified as follows: TRPC (canonical), TRPV (vanilloid), TRPM (melastatin), TRPML (mucolipin), TRPP (polycystin), TRPA (ankyrin) and TPRN (NomPC) subfamilies. Of these, TRPC3, TRPC6 (ref. 22), TRPM2 (ref. 23), TRPM8 (ref. 24) and TRPV1–TRPV4 (ref. 25) are expressed on neurons across the PNS and the CNS, and among those, TRPV1 has been substantially investigated in a metabolic context due to its sensitivity to capsaicin, the pungent compound in hot chilli peppers. Capsaicin was previously explored as a potential anti-obesity therapy and is now used as a high-dose neurotoxin in adipose sensory denervation studies²⁶. TRPV1 expression is restricted to peptidergic sensory neurons expressing CGRP and substance P²⁷, making sensory denervation with high-dose capsaicin exclusive to peptidergic sensory nerves. However, caution is needed when interpreting studies involving TRPV1, as it is not exclusively expressed in sensory nerves in adipose tissues, but also in a subset of pre-adipocytes that may be stimulated following TRPV1 agonist administration to adipose tissue^28,29.

Other receptors (orexin receptor 2, pituitary adenylate cyclase 1 receptor and vasoactive intestinal peptide receptor 1) were recently identified by quantitative PCR on sensory nerves innervating rodent BAT and WAT^30,31, and may act as auto-receptors.

TRPV1 and adipose metabolism

TRPV1 ligands in adipose tissues.

Research on adipose–neural communication by sensory nerves has gained momentum since the discovery of bioactive lipid-based compounds that bind TRPV1 receptors in adipose tissues. The postulated TRPV1 ligands in adipose include the n-6 and n-3 polyunsaturated fatty acid compounds arachidonic acid and eicosapentaenoic acid, which are naturally released lipids from the WAT cellular membrane^32,33. Sensory nerve activation by eicosapentaenoic acid and arachidonic acid was evidenced by immediate electrophysiological activation of isolated sensory nerve afferents after arachidonic acid and eicosapentaenoic acid were injected directly into the ing-scWAT depot of Siberian hamsters¹⁹. This suggested that adipose sensory nerves can detect and relay changes in local lipid composition in adipose to the brain. Hence, because dietary changes can alter the stored fatty acid composition in adipose triglyceride-rich lipid droplets, as well as the membrane polyunsaturated fatty acids and resulting production of oxylipin signalling molecules known to act on TRP channels^34–36, it is probable that adipose sensory nerves can sense and communicate changes in the local lipid milieu to the brain.

Additionally, local lipid products released during sympathetic nervous system (SNS) activity (such as norepinephrine-induced lipolysis that releases fatty acids from stored triglyceride) may subsequently stimulate sensory afferents in WAT, which could then communicate adipose tissue lipid fuel status back to the CNS. However, this conceptual model has never been directly or systematically validated. Furthermore, due to the multicellular effects of these lipids in the tissue, more rigorous studies are needed to clarify their impact on sensory nerves directly.

TRPV1 and whole-body energy balance.

TRPV1 activation has been consistently demonstrated to have anti-obesogenic effects in both humans and rodents. Systemic TRPV1 activation by low-dose dietary capsaicin, and its non-pungent analogues such as capsinoids, were linked to reduced adiposity, enhanced energy expenditure from BAT thermogenesis, and lower food intake in healthy humans, making TRPV1 an appealing potential therapeutic target for obesity^26,37–39. Similarly, dietary supplementation with low-dose capsaicin protected wild-type (WT) mice from diet-induced obesity when fed a high-fat diet (HFD), but this effect was not seen in Trpv1^−/− mice. WT obese mice treated with dietary capsaicin exhibited an enhanced appearance of ‘beige’ multilocular adipocytes in WAT with increased UCP1 expression⁴⁰. By contrast, 12-week-old Trpv1^−/− mice fed a HFD had pronounced adiposity, higher energy intake and increased leptin and insulin resistance compared to control mice⁴¹. These findings collectively support the notion that systemic TRPV1 activation promotes energy expenditure and protects against diet-induced obesity.

A more recent study examined the impacts of TRPV1 on UCP1-induced thermogenesis in BAT. Mice with a whole-body double-knockout of TRPV1 and UCP1 (Trpv1^−/−Ucp1^−/−) had lower BAT mass, reduced oxygen consumption and heat production, and reduced BAT-specific mitochondrial respiration⁴². These phenotypes were blunted in Ucp1^−/− mice that did express TRPV1, indicating TRPV1 is essential for adequate UCP1-induced thermogenesis in BAT^42–44. Although these findings aligned with the pro-thermogenic effects of TRPV1 activation, whether these outcomes were directly linked to sensory nerve or preadipocyte-specific TRPV1 loss has not been clarified. Additionally, none of these studies specifically investigated TRPV1 activity on sensory nerves innervating BAT or WAT.

Only one rodent study has challenged the established anti-obesity and pro-thermogenic effects of TRPV1 activation⁴⁵, finding that Trpv1^−/− mice gained less weight than control animals, despite similar energy intake when fed a HFD. Furthermore, cold exposure did not decrease core body temperature in these Trpv1^−/− mice, unlike in the WT mice. This suggested that TRPV1 may impact adaptation to cold exposure⁴⁵. However, it is important to note that the HFD used in the study had a lower fat content (11%), which may not accurately represent a true HFD-induced metabolic challenge that can reach 60% kcal from fat in some studies. Possible differences in Trpv1^−/− mouse strains between this study⁴⁵ and those previously discussed^26,37–41 could explain the differences in outcomes. Additionally, indirect effects on TRPV1 activity should be considered, as in vitro studies using human embryonic kidney stem cells demonstrated TRPV1 is inhibited following cold-induced TRMP8 activation (cold-sensing receptor)⁴⁶. Thus, the decrease in core body temperature in WT but not Trpv1^−/− mice could be a consequence of enhanced systemic TRPM8 activation after cold exposure in the absence of TRPV1.

Other sensory nerve ligands, receptors and signalling systems in adipose tissues

Adipokine-induced activation of sensory nerves in adipose tissues.

Communication along the adipose–neural axis involves adipokines, particularly adipocyte-secreted leptin. Recent work has demonstrated that sensory neurons innervating WAT have the ability to bind leptin^47,48, challenging the traditional belief that leptin signalling in the brain occurs exclusively by leptin travelling from WAT in the bloodstream and acting in an endocrine manner. In rats, unilateral leptin administration into eWAT increased electrophysiological activity of sensory afferents within the depot (and in sympathetic bundles in the contralateral depot) in a dose-dependent manner⁴⁷. Similarly, leptin administration to ing-scWAT of Siberian hamsters enhanced cFOS expression (a neural activation marker) in T12–L1 DRG, and amplified the nerve spike rate of sensory afferents in ing-scWAT⁴⁸. Together, these findings support the concept of leptin-induced neural activation of sensory nerves in WAT, followed by subsequent SNS activation via CNS control. This may be an additional fuel-sensing system for the brain to gauge fuel availability in WAT, because leptin release is typically directly proportional to fat mass.

Given the clear importance of sensory nerves in WAT, there has been a surge in studies attempting to identify other bioactive compounds (endogenously produced, or pharmacological) that may activate these nerves. However, further mechanistic investigations are necessary to fully comprehend the interoceptive systems in adipose tissues, and what role other adipokines may play in stimulating adipose sensory nerves.

Other factors activating sensory nerves in adipose tissues.

Besides leptin, several other ligands such as growth factors⁴⁹ and lipids⁵⁰ have also been proposed as stimuli being ‘sensed’ via interoceptive means by the sensory nerves in adipose tissue⁵¹. Among these, WAT sensory afferents are thought to primarily respond to and undergo activation through binding lipolytic products generated by the sympathetic drive to WAT¹⁴. This was evidenced through electrophysiological sensory activation in WAT measured after systemic delivery of a glucoprivation agent (2-deoxy-d-glucose) to Siberian hamsters¹⁴; however, a mechanistic explanation was not offered.

Free fatty acids and glycerol, as well as other byproducts of norepinephrine-induced lipolysis such as prostaglandin E2, have also been postulated as candidates that may trigger sensory nerve activation in WAT⁵². For instance, to assess whether prostaglandin E2 elicits sensory nerve activation, vagal nerve fibres isolated from rats underwent electrophysiological activation after ex vivo treatment with prostaglandin E2 (ref. 53). Similarly, DRG neurons treated with prostaglandin E2 in vitro underwent an increase in cyclic adenosine monophosphate, a signalling pathway otherwise associated with SNS-induced lipolytic activation⁵³. Both findings indicated that prostaglandin E2 promotes neuronal activity in sensory neurons, although this has not been tested in adipose tissue specifically. However, owing to the plausibility of prostaglandin E2 acting on receptors not specific to sensory nerves, replicating these findings in vivo has been challenging. Lastly, like capsaicin and leptin, bradykinin, adenosine and adenosine triphosphate are also sensory nerve agonists that are capable of eliciting electrophysiological activation in sympathetic efferent nerves innervating ing-scWAT depots in rats⁵⁴, suggesting sensory–sympathetic cross-talk in adipose.

Metabolic functions of sensory nerves in adipose tissue

Sensory nerves in adipose tissue with metabolic diseases

Hypertension and the role of adipose sensory nerves.

Early viral tracing studies identified hypothalamic regions that receive sensory input from, and drive sympathetic output to, adipose tissues^2,14,55. These hypothalamic neurons have been implicated in the sympathetic control of cardiovascular and renal responses, including blood pressure regulation^56–59. Intriguingly, both unilateral⁶⁰ and bilateral⁶¹ sensory nerve activation by low-dose capsaicin microinjections to ing-scWAT of rats enhanced renal sympathetic nerve activity and mean arterial pressure in the kidneys, suggesting involvement of WAT-specific sensory afferents in modulating whole-body blood pressure⁶⁰. Mechanistically, sensory activation in WAT led to sympathetic-induced renin angiotensin aldosterone system activation, as evidenced by elevated plasma renin, norepinephrine and angiotensin II levels⁶¹.

In a recent study on male mice, researchers compared the impact of sensory nerve activation (by low-dose capsaicin) in ing-scWAT and eWAT on blood pressure changes. Sensory stimulation in eWAT, but not ing-scWAT, enhanced mean arterial pressure in the injected tissue⁶². These differences in mean arterial pressure elicited depending on the WAT depot targeted and may have been attributed to variations in CGRP activity between the depots. Nevertheless, these findings emphasized the involvement of WAT sensory nerves in blood pressure regulation. Because obesity is associated with adipose neuropathy¹⁰ and hypertension, the bidirectional adipose–neural signalling disruption may be the cause of obesity-induced elevations in blood pressure, which can present as upregulated sympathetic tone to the kidneys, an organ directly involved in blood pressure regulation^63,64.

Diabetes mellitus, obesity, ageing and the role of adipose sensory nerves.

Diabetic peripheral neuropathy, characterized by loss or damage of sensory, sympathetic and motor innervation in peripheral tissues and organs, has been extensively studied in the skin, heart and reproductive organs^65–68. However, we demonstrated that obesity, diabetes and ageing also present with ‘adipose neuropathy’ in ing-scWAT in mice, and in omental WAT and scWAT of obese or aged women¹⁰. Leptin-deficient BTBR ob/ob mutant mice, a model for obesity or diabetes and peripheral neuropathy, exhibited reduced levels of the pan-neuronal protein PGP9.5 in ing-scWAT at 12 weeks of age, which worsened with ageing. Similarly, loss of sympathetic innervation was demonstrated by reduced tyrosine hydroxylase protein in both rodent and human fat pads. However, changes in CGRP content were not measured, and the possibility that a subset of these tyrosine hydroxylase-expressing fibres were sensory cannot be ruled out.

Furthermore, age-related neuropathy in adipose, skin and muscle differs by genetic strain of mouse and by biological sex⁶⁹, indicating that multiple factors impact the survival of adipose nerves. One factor that may impact neural plasticity and survival is the association with neuroimmune cells in the tissue, which can express neurotrophic factors. We found that loss of neurotrophic factors from myeloid lineage cells (LysM-Cre or adult-inducible Cx3CR1-CreERT2) resulted in loss of adipose innervation^70,71. Our more recent work has revealed a demyelinating neuropathy and changes to adipose Schwann cells in the WAT of obese and diabetic animals⁷². We have found both sensory and sympathetic axons in adipose tissue with myelin sheaths, so it is unclear whether both nerve types are impacted by demyelinating neuropathy. These findings also underscore the importance of the tissue-specific environment, because demyelination is not a feature of diabetic neuropathy in the skin.

The anatomical location of sensory neurons in the ganglia, outside the protection of the spinal cord, makes them more susceptible to damage by circulating factors because they lack protection from the blood–brain barrier (unlike motor neurons, located within the more protected spinal cord). Thus, diabetic neuropathy—a condition characterized by higher circulating glucose (driving nerve glucotoxicity) and lipids (driving nerve lipotoxicity)—can predominantly affect sensory nerves⁷³.

Indeed, dyslipidaemia has been recognized as a notable factor contributing to the onset and progression of diabetic neuropathy^74–77. Low-density lipoproteins (LDLs) are lipid-based carrier proteins responsible for transporting cholesterol throughout the body. LDLs are susceptible to oxidation in the presence of reactive oxygen species and form oxidized LDLs. Sensory nerve neuropathy has been associated with the presence of oxidized LDLs⁷⁸, which bind to the lectin-like oxidized low-density lipoprotein receptor 1 (LOX1), a transmembrane receptor for cellular uptake of oxidized LDL⁷⁹. Studies using HFD-fed mice showed increased oxidized LDL deposition in the DRG and functional deficits in sensory nerves, as demonstrated by increased hind-paw latency in response to heat stimulus and reduced nerve conduction velocity. These effects were observed after 12 weeks of HFD feeding, before the onset of glucose intolerance⁸⁰. In vitro experiments using rat DRGs treated with high glucose or oxidized LDLs (to mimic increased LOX1 receptor activation during diabetes) demonstrated an increase in oxidative stress only with oxidized LDL treatment, but not with high-glucose treatment⁸⁰. Collectively, these findings indicated that circulating oxidized LDLs accumulate in the DRG during HFD feeding, leading to oxidative damage within sensory neurons via LOX1 receptor activation. Diabetic mice also had higher levels of circulating hydroxyoctadecadienoic acids⁸⁰, known sensory nerve TPRV1 ligands⁸¹ that can be produced from the adipocyte membrane during polyunsaturated fatty acid breakdown for eicosanoid synthesis, which may also contribute to lipotoxicity and thereby precipitate neuropathy. Given the complexity of lipids stored and produced within adipose tissue, there may be both neuroprotective and neurotoxic subtypes that need to be categorized and studied more closely.

An ongoing debate: positive versus negative feedback between sensory and sympathetic nerves in adipose tissues

The presence of robust sensory and sympathetic, but not parasympathetic, innervation in adipose tissues is largely indisputable. However, the functional relationship between sensory and sympathetic nerves in adipose tissues remains debated. Available data provide opposing evidence for sensory nerves either promoting or inhibiting sympathetic nerve activation in adipose tissues. Much of this disagreement may result from differences in experimental approaches adopted to better understand sensory nerve function. For example, the data supporting a positive feedback loop in adipose tissue, which entails sensory afferent activation promoting activation of sympathetic efferent nerves, are mainly based on earlier studies that used surgical and chemical denervation techniques^{9,18,31,47,54,82} (Figs. 3 and 4). Conversely, data supporting the negative feedback loop, wherein sensory nerves are postulated to inhibit sympathetic activation in the tissue, rest on more recent studies utilizing genetic, viral and pharmacological (non-TRPV1 targeting) denervation techniques^31,83 (Figs. 3 and 4). Each of these methods has caveats (Box 1) that impact careful interpretation of the resulting data, and more experimental details of these studies are discussed below.

Fig. 4 |. Evidence for positive or negative feedback loops between sensory and sympathetic nerves in adipose tissue.

a, The concept of a positive feedback loop between sensory and sympathetic nerves in adipose tissues is supported by pharmacological agonism or ablation (using high-dose capsaicin) of sensory nerves in ing-scWAT, which measured increases in serum norepinephrine, mean arterial pressure (MAP), norepinephrine turnover (NETO) at room temperature (RT) in BAT, electrophysiological activation of sympathetic nerves in WAT and renal sympathetic nerve activity in the kidneys. b, The concept of a negative feedback loop between sensory and sympathetic nerves in adipose tissue is supported by genetic and viral denervation approaches and is evidenced by the following observations: vasoconstriction with reduced tail temperature at room temperature after systemic genetic ablation of CGRP (Advillin-Cre crossed with Calca-lox-GFP-lox-DTR), increased BAT and core body temperature, and enhanced lipid utilization in BAT after sensory ablation by high-dose capsaicin or systemic genetic CGRP ablation, reduced tyrosine hydroxylase (TH) levels in BAT after orexin receptor agonism in BAT by [Ala11, d-Leu15] OxB, and increased lipid utilization in ing-scWAT after unilateral or bilateral viral sensory denervation with mCherry-flex-DTA injected into T13/L1 DRG with ROOT-Cre injected into ing-scWAT.

Box 1. Caveats of commonly adopted chemical denervation techniques of sensory nerves in adipose tissues.

While early studies by the Bartness group laid the groundwork for sensory innervation and sensory nerve involvement in adipose tissue homeostasis^{9,19,55,96,97}, their techniques of surgical and chemical denervation also had limitations that affect modern interpretations of the findings. High-dose capsaicin has been historically used as a sensory denervation agent owing to its affinity for TPRV1 on sensory nerves. However, high-dose capsaicin may fail to achieve total sensory denervation, as detectable levels of CGRP are still evident in the tissue after treatment¹⁸. Capsaicin-induced sensory denervation is also reversible, making it a poor tool for studying chronic effects of sensory denervation because the nerves can regenerate over time⁸⁶. High-dose capsaicin also presents a risk of lethality¹⁰⁸. Rodents display severe neurotoxicity symptoms (convulsions, tremors) within a few minutes of capsaicin administration, with lethality occurring within 4–26 min after treatment¹⁰⁸. Our laboratory has noted up to a 70% mortality in mice after high-dose intra-adipose capsaicin treatment (following previously published protocols¹⁰⁹). While under-reported in the field, some proposed explanations for high-dose capsaicin-induced lethality are poor thermoregulatory capacity due to sensory nerve excitotoxicity, accumulation in the CNS (after intravenous injection in rats), damage to internal organs from toxin exposure and respiratory dysregulation^108,110. Thus, current approaches to achieving sensory denervation have been largely limited to whole-body Trpv1^−/− rodent models (rather than tissue or nerve-specific TRPV1 deletion), thereby limiting the interpretability of the data from these studies¹¹¹. Several other chemical and viral agents have been developed as alternatives to circumvent the toxicity and lethality associated with capsaicin. Recently, a marked reduction in CGRP immunoreactivity was noted in rat perirenal WAT after resiniferatoxin administration²⁰, highlighting its potential as an alternative to capsaicin as a sensory denervation agent for use in adipose denervation studies. Whether resiniferatoxin-induced sensory denervation is reversible has been a subject of debate (reviewed in ref. 112), with the consensus leaning towards permanent ablation due to the nonreversible binding to TRP channels^113,114, although this has never been examined in adipose tissues specifically.

Metabolic and functional roles of sensory nerves in brown adipose tissue

Sensory and sympathetic nerve cross-talk in brown adipose tissue and effects on thermogenesis.

BAT is best recognized for its role in non-shivering thermogenesis, an energy-expending process characterized by high mitochondrial UCP1 activity (heat dissipation uncoupled from ATP production) following cold stimulation. Thermogenesis, however, is commonly stimulated by skin sensory nerve activity following sensation of cold temperature, and subsequent sympathetic nerve activation in BAT leading to norepinephrine release, which stimulates UCP1 in BAT and beige adipocytes (the UCP1-positive, mitochondria-rich brown adipocytes that develop in WAT, such as with cold acclimation)⁸⁴. Despite being overlooked historically, there is increasing recognition of the involvement of sensory neuropeptides in this thermoregulatory process.

Early studies in rats demonstrated a link between thermoregulation and sensory nerve function⁸⁵. Rats lacking sensory nerve function from systemic pretreatment with fractionated capsaicin had lower body temperature at both ambient (22 °C) and cold (4 °C) temperatures⁸⁵. This impairment in thermoregulation alluded to the possible involvement of sensory nerves in BAT thermogenesis, potentially independent of sympathetic nerve stimulation of UCP1 and browning processes.

Soon after, targeted sensory denervation of BAT in rats⁸⁶ and Siberian hamsters¹⁸ by intra-BAT high-dose capsaicin led to reduced UCP1 protein levels, lower adipocyte and mitochondria counts, and impaired thermoregulation (as measured by reduced core body temperature). However, in studies using the Siberian hamster, animals were acclimatized for 3 weeks after BAT sensory denervation before any recordings were performed, which may have allowed sensory reinnervation and recovery of impaired thermoregulation¹⁸ (a unique feature of the PNS is that, unlike in the CNS, typically the damaged peripheral nerve axons can readily regrow and reinnervate the tissue^87,88). Alternatively, some observed changes could have resulted from initial capsaicin-induced sensory nerve activation, followed by denervation after prolonged capsaicin exposure. The reduced BAT and core body temperature persisted only for the first 12 h of cold exposure, indicating an ability of the denervated animals to adjust to the stimulus. Nevertheless, failure to appropriately thermoregulate whole-body and BAT temperatures signified the thermoregulatory function of sensory nerves in BAT as an additional contributor beyond sympathetic nerves¹⁸.

If and how sensory afferents in BAT coordinate the SNS-induced thermogenic response has been a long-standing question in the field. Studies in rats demonstrated that BAT sensory denervation by high-dose capsaicin prolonged norepinephrine-induced thermogenesis after intravenous norepinephrine infusion and local warming of BAT (with warming pads placed over interscapular BAT, a condition that would not normally promote thermogenesis) for 40 min, suggesting that sensory nerves in BAT inhibit this process. Additionally, CGRP infusion (mimicking sensory activation) for 30 min in BAT lowered a norepinephrine-induced increase in thermogenesis⁸³. This suggests sensory afferents may inhibit sympathetic nerve-induced thermogenesis during high local/environmental temperatures, demonstrating the presence of a protective, negative feedback loop. Conversely, it is possible that sensory nerves promote sympathetic nerve activation to elicit norepinephrine-induced thermogenesis in BAT during cold exposure, thus presenting a positive feedback loop; however, this was not tested experimentally. Lastly, all thermogenesis measurements in this study were based on local norepinephrine infusion to test responses, and sympathetic nerve activity (through electrophysiological measurements or norepinephrine turnover) was not directly measured.

Circulating factors may also act to coordinate sensory–SNS cross-talk in adipose tissue. One such factor is orexin, a hypothalamic sensory neuropeptide that binds orexin receptor type 2 (OX2R) in the hypothalamus to inhibit thermogenesis⁸⁹. However, orexin also circulates⁹⁰ and can interact with OX2Rs expressed on peripheral tissues³¹, although these specific effects in the periphery are not well understood. Studies have demonstrated that intact sensory innervation is necessary for OX2R expression in mouse BAT, because BAT OX2R expression was observed after targeted sympathetic denervation with 6-hydroxydopamine, but not following surgical denervation of BAT that would have reduced both sensory and sympathetic innervation³¹. Pharmacological OX2R agonism in BAT attenuated UCP1 expression and thermogenic capacity, reduced both food intake and energy expenditure, and promoted BAT ‘whitening’ (loss of a thermogenic histological phenotype) in mice³¹. Chronic OX2R agonism lowered sympathetic innervation in BAT, as noted by reduced tyrosine hydroxylase immunopositivity (however, this was interpreted as reduced sympathetic activity by the authors). Chronic OX2R agonism in BAT also reduced proopiomelanocortin immunopositivity in known hypothalamic regions receiving sensory input and driving sympathetic output to BAT³¹. These findings suggested that circulating sensory neuropeptides may modulate sympathetic nerve activity in adipose depots. Because no direct measurements of sympathetic nerve activity were performed in BAT, these data do not definitively support a negative feedback loop between sensory and sympathetic nerves in BAT.

In summary, sensory nerves in BAT are implicated in thermoregulation and interact with the SNS, but the relationship between sensory and sympathetic nerves in adipose tissue is complex and involves feedback loops and unclear mechanisms of cross-talk. The communication between the sensory and sympathetic systems could be occurring at multiple levels: (1) at the tissue level, such as direct inhibition of sympathetic nerves by CGRP or other neuropeptides that may inhibit norepinephrine release; (2) at the level of the ganglia, where there can be mixed sensory and sympathetic cell bodies; or (3) at the level of the brain, where sensory afferent activity leads to sympathetic efferent stimulation via hypothalamic or control regions and/or command neurons.

Whether sensory nerve activity in BAT is required for metabolic processes beyond thermogenesis has been sparsely investigated. Early rodent studies reported reduced UCP1 expression in both BAT and browning-prone regions of WAT after complete surgical denervation of BAT⁹¹. However, complete surgical resection ablates mixed nerve bundles containing sensory and sympathetic axons and therefore does not clarify which nerves are responsible for the observed loss of UCP1 expression.

To assess the role of sensory nerves specifically in maintaining BAT health, Himms-Hagen et al. denervated BAT sensory nerves in rats using high-dose capsaicin injections⁹². This decreased UCP1 expression, lowered BAT mass and reduced both mitochondrial number and activity 1 week after treatment. BAT atrophy was evident 24 h after treatment and lasted up to 2 weeks⁹². Therefore, effects beyond UCP1 inhibition appear to be contributing. There was likely sensory reinnervation of the BAT at later time points, given that 28 d following capsaicin injections there was an increase in mitochondrial count back to baseline levels. A longer 52-d study similarly showed no pathophysiological differences between BAT depots with and without intact sensory innervation⁹², probably owing to nerve regrowth by that time point, although this was not directly measured.

To investigate the chronic metabolic effects of BAT sensory denervation, the same study was replicated to examine whole-body metabolic outcomes over 8 months⁹³. Rats with sensory-denervated BAT exhibited higher body mass, reduced resting metabolic rate, decreased oxidative capacity, lower total mitochondria count and decreased UCP1 levels in BAT at 3.5 months after denervation. These rats also had higher eWAT mass 3.5 months after BAT sensory denervation, indicating that sensory nerves in BAT may mediate BAT–WAT cross-talk and thereby promote compensatory changes in WAT in response to a loss of BAT thermogenic capacity⁹³. All of these physiological and metabolic parameters returned to baseline levels 8 months after BAT sensory ablation. However, since the authors originally observed some recovery of BAT phenotypes at 28 d after capsaicin⁹², it is worth questioning whether metabolic changes seen 3.5 months after BAT ablation occurred in part as a result of ageing, which may have been exacerbated by prior loss of BAT sensory inputs⁹⁴. These studies highlight the involvement of BAT sensory nerves in maintaining metabolic functions, and influencing neural communication between different adipose tissue depots. Further research is needed to gain a better understanding of sensory nerve functions in BAT at the level of neuropeptide actions and influence on BAT sympathetic nerves.

Metabolic and functional roles of sensory nerves in white adipose tissue

Sensory and sympathetic nerve cross-talk in white adipose tissue and effects on lipolysis.

The need for intact sensory innervation for lipolytic regulation in WAT was initially demonstrated in studies using sensory nerve chemical denervation of ing-scWAT in Siberian hamsters with high-dose capsaicin⁹. Sensory denervation resulted in adipocyte hypertrophy and decreased CGRP immunopositivity in ing-scWAT, with no effects on tyrosine hydroxylase levels, suggesting a possible role of sensory neuropeptides in regulating lipolysis owing to the large lipid droplet sizes following denervation. These findings challenged the conventional understanding that only sympathetic nerve products stimulate lipolysis. An alternative interpretation is that the adipocyte hypertrophy observed was due to sensory neuropeptide loss in WAT, which indirectly inhibited lipolysis by removing the brake on sympathetic nerve activity to the tissue.

Indeed, some studies in rodents have examined whether sensory nerves in WAT can modulate sympathetic drive to BAT^9,54 and other peripheral organs⁵⁴. Targeted bilateral ing-scWAT sensory denervation in Siberian hamster using high-dose capsaicin reduced norepinephrine turnover, a direct measure of sympathetic output in a tissue, in both ing-scWAT and BAT 24 h after cold exposure⁹, thus demonstrating that sensory nerve activation in ing-scWAT promotes sympathetic activity not only within the same depot, but also in BAT^9,54. This also suggested the presence of a positive feedback loop between sensory and sympathetic nerves in adipose tissues. Moreover, ing-scWAT sensory denervation increased norepinephrine turnover in mesenteric WAT, but not in eWAT or retroperitoneal WAT. It also decreased UCP1 expression in BAT, but had no substantial effects on food intake, blood glucose or fat mass⁹. Collectively, these findings indicated that the influence of sensory nerves on sympathetic nerves is depot specific. Similarly, ing-scWAT sensory nerve stimulation by low-dose capsaicin enhanced both renal sympathetic nerve activity and mean arterial pressure in kidneys of rats⁵⁴.

A more recent study provided a new take on the cross-talk between adipose sensory and sympathetic nerves. To achieve sensory denervation in ing-scWAT, Wang et al.³ used a combination of viral tracing and denervation in WAT. Mice were first injected in T13/L1 DRG with a fluorescently tagged (mCherry) adeno-associated virus that also contained a Cre-driven gene (diphtheria toxin subunit A) to specifically ablate the injected cells. ROOT-Cre, a cre-containing adeno-associated virus capable of retrograde transport, was then injected into ing-scWAT (either unilaterally or bilaterally) to ablate sensory projections in ing-scWAT by inducing diphtheria toxin subunit A expression in T13/L1 DRG³. Sensory denervation elevated de novo lipogenic markers such as fatty acid synthase, acetyl-coA carboxylase beta, and carbohydrate-responsive element-binding protein gene expression, as observed though RNA-sequencing analysis and confirmed by quantitative PCR with reverse transcription after both unilateral and bilateral sensory ablation of ing-scWAT, thus increasing fat mass. In unilaterally denervated animals, norepinephrine levels were not reduced in the sensory-denervated depot, but norepinephrine and norepinephrine turnover, as surrogates of sympathetic nerve activity, were not measured in the contralateral depot with intact sensory innervation of the unilateral denervation model, where compensatory changes are likely to occur⁹⁵. Regardless, the authors stressed the presence of a negative feedback loop between ing-scWAT sensory and sympathetic nerves based on enhanced de novo lipogenic gene expression in sensory-denervated ing-scWAT (and because these were blunted following bilateral 6-hydroxydopamine-mediated sympathetic denervation in ing-scWAT), both of which are indirect measurements of sympathetic nerve activation³. These indirect measurements of sympathetic nerve activation suggested a potential negative feedback loop between sensory and sympathetic nerves in scWAT, a relatively new school of thought as compared to prior evidence. This could largely be methodology driven, as this study demonstrated successful ablation of sensory nerves in adipose tissue using a new viral denervation method, a departure from commonly used chemical denervation approaches delivered into the adipose depot^{9,19,55,96,97}. While it is possible that sensory nerves communicate directly with sympathetic nerves within the adipose depot via neuropeptide signalling, it is more likely that disruptions in hypothalamic integration of incoming afferent and outgoing efferent signals to and from the adipose⁹⁸, respectively, explain the observed outcomes. Alternatively, CGRP itself could directly activate pro-lipolytic pathways as an alternative to norepinephrine, as has been reported⁹⁹.

Despite the useful information provided by the ROOT-Cre technique, it is imbued with several caveats. First, although this technique ensures specificity down to the level of the adipose depot, like capsaicin it did not produce complete sensory ablation in the injected DRGs³. Additionally, sensory afferents in ing-scWAT have been previously traced back to T12–L1 in mouse¹⁰⁰ (and T13–L3 in Siberian hamsters¹³). In this study, however, only T12–L1 sensory neurons were targeted, leaving a substantial subset of sensory afferents intact in ing-scWAT. Consequently, sensory neurons innervating ing-scWAT in T12, L2 and L3 DRGs may have compensated for ablated sensory neurons in T13–L1 DRG by increasing branching or activity. Therefore, functional differences in the surviving sensory neurons within the T13–L1 DRG from those that were successfully denervated, could have influenced the observed outcomes. Tissue-level neuropeptides were not measured as a readout for potential compensation.

Another difference between this and earlier studies is that viral ablation of the adipose-innervating DRG typically eliminates the entire neuron from the cell body to the axon terminals. By contrast, chemical ablation by direct injections into the adipose depot targets only the sensory nerve endings (axonal endings) within the injected depot and may be more transient in nature, with axonal regeneration more likely over time. The conclusions from Wang et al.³, which provided compelling evidence for the importance of sensory nerves in adipose functions, are based partly on unilateral sensory denervation (although they also performed bilateral denervation in ing-scWAT). However, it is known that unilateral denervation can promote compensation in the contralateral depot⁹⁵, a phenomenon which was not accounted for in all experiments in this study.

Taken together, the roles of sensory neuropeptides in functions such as lipolysis, traditionally attributed to norepinephrine, warrants further investigation, including the contributions of the various neurotransmitters and neuropeptides acting in adipose depots both individually and synergistically from diverse subsets of neuronal axons innervating adipose tissues.

Sensory nerves in white adipose tissue physiology, browning, vascular control and compensatory effects in other depots.

Sensory nerve activity in WAT and the resulting effects on whole-body metabolism and energy balance remain unclear, largely owing to the dearth of studies on this topic to date. In one study assessing the importance of sensory nerves in WAT on browning, both unilateral and bilateral sensory ablation in mouse ing-scWAT by ROOT-Cre-AAV upregulated expression of thermogenic genes (Ucp1, very-long-chain fatty acid protein 3 (Elovl3), and cell death-inducing DNA fragmentation factor alpha-like effector (Cidea)) in the sensory-denervated depot, concomitant with enhanced browning and thereby demonstrating scWAT sensory nerve involvement in thermogenesis³.

Furthermore, to demonstrate the role of CGRP (a major sensory neuropeptide) in adipose and whole-body metabolism, one study systemically ablated CGRP from all advillin-expressing sensory neurons in mice⁸. CGRP loss enhanced mitochondrial respiration in BAT and increased isoproterenol-induced free glycerol release in WAT. CGRP-deficient mice also exhibited increased energy expenditure and were protected against diet-induced obesity—findings that aligned with the conclusions of the more recent viral-mediated sensory denervation, which suggested that sensory innervation in WAT reduced lipolytic and thermogenic potential³. Collectively, these data indicate that sensory nerves, particularly through CGRP action, may inhibit energy-expending processes in adipose tissues, but this may be a physiological context-dependent effect. Given the clearly distinct roles of depot-specific innervation and inter-depot cross-talk, a depot-specific CGRP ablation approach, or inhibition of sensory neuropeptide actions in adipose or adipocytes, would be more informative.

Moreover, global CGRP deletion⁸ (to model sensory denervation) does not clarify adipose-specific CGRP functions. Given that CGRP is critical for vascular homeostasis in other central and peripheral tissues¹⁰¹, systemic CGRP deletion would disrupt other tissue functions in the body. As such, any outcomes observed in adipose tissues after global CGRP loss are influenced by changes to other peripheral and central tissues that otherwise express CGRP. We cannot assume that CGRP produces similar outcomes in all subcutaneous WAT and visceral WAT depots, as each adipose depot has distinct neurochemical innervation and characteristics, with the ing-scWAT depot comparably more densely innervated and prone to browning than visceral scWAT¹⁰², and visceral WAT having higher lipolytic potential than scWAT depots¹⁰³. Thus, it is appealing to speculate that the demand for CGRP and its functional contributions are similarly unique to each depot.

Sensory nerve activity in ing-scWAT also plays a role in the regulation of glucose^3,104 and blood pressure¹⁰⁵, in addition to influencing whole-body metabolism. Mice lacking sensory nerve activity in ing-scWAT (through viral sensory denervation) were resistant to HFD-induced glucose intolerance³ but were more susceptible to systemic insulin resistance during normal chow feeding¹⁰⁴. These findings indicate that sensory nerves in scWAT are necessary for appropriate systemic glucose and insulin regulation at basal states and may undergo dysregulation during diet-induced obesity.

CGRP⁺ sensory nerves innervating the perivascular adipose tissue (PVAT) are involved in blood pressure regulation¹⁰⁶ as well. To investigate the impact of sensory nerves in PVAT on the vasculature, electric field stimulation of isolated mesenteric arterial beds (which are densely innervated by CGRP⁺ nerves¹⁰⁷) has served as an informative technique. This ex vivo preparation maintains tissue integrity and the nerve endings, closely resembling physiological conditions. Electric field stimulation of perfused mesenteric arterial beds of rats, with and without excised PVAT, increased vascular relaxation in PVAT⁺ mesenteric arterial beds, and enhanced leptin release from PVAT⁺ beds, suggesting that intact PVAT is required for appropriate vasorelaxation. This outcome is probably due to CGRP (or other sensory neuropeptide) release from sensory nerve endings in excised tissues. In related studies, ex vivo capsaicin treatment of PVAT⁺ mesenteric arteries increased CGRP release, which promoted vasodilation in the isolated mesenteric arterial bed¹⁰⁶. Thus, adipose-innervating sensory nerves release neuropeptides that modulate vascular tone and thereby contribute to blood pressure regulation. Whether or not CGRP directly mediates changes in adipose blood vessel function remains unknown, including in physiologically relevant situations, such as contributing to heat dissipation during thermogenic states. Additionally, whether other sensory neuropeptides contribute to vascular regulation in adipose tissue is currently unclear.

Conclusions and open questions in the field

To summarize, the neurometabolic regulation of adipose tissue is, in large part, influenced by sensory nerves and their neuropeptides that are released into the tissue. However, several unanswered questions continue to riddle the field and emphasize the vast unknowns that remain regarding adipose tissue sensory innervation and sensory nerve functions in the tissue that may have broader effects on systemic metabolism.

First, the plasticity of adipose nerves is not well understood, including how the tissue’s total innervation, axonal subtypes, branching patterns and release of neuropeptides respond to changing energy balance states. Axon outgrowth across adipose depots, or degeneration of axon nerve terminals, and the effects of these on target cells, remain very unclear. While we previously reported reduced total and sympathetic innervation in WAT and BAT with obesity/diabetes or ageing, changes in sensory innervation across neuropathic states have not yet been explored¹⁰. Because diabetic neuropathy is mainly a sensory neuropathy in tissues such as the skin, sensory denervation of adipose may be a key aspect of neuropathic pathophysiology.

Second, the neurometabolic regulation of adipose tissue is increasingly appreciated to be influenced by sensory nerves and their secreted neuropeptides, such as CGRP. Although CGRP⁺ sensory nerves in adipose tissue have been identified, the functions of CGRP in adipose tissue and the roles of other sensory neuropeptides remain a major gap in knowledge. It is also currently unclear what the relative contributions are between sensory neuropeptides and sympathetic nerve products, such as norepinephrine (although, sympathetic nerves also release other understudied nerve products like ATP and neuropeptide Y). The current data indicate that there may be overlapping or redundant roles between norepinephrine and sensory neuropeptides, such as in lipolytic or thermogenic stimulation. Sensory neuropeptides are notoriously difficult to study as they are synthesized distally from the organ of interest in the DRG and are too small and transiently expressed to be reliably detected by commonly used approaches to study protein level changes (western blot, proteomics). Subsequently, adipose-specific roles of sensory neuropeptides have received limited attention in the research community, despite the clear presence in the tissue of numerous sensory neuropeptides as revealed by immunostaining studies and ELISA assays. Modern neuroscience techniques that are currently used in the brain to stimulate or inhibit subsets of nerves, or to investigate specific nerve product signalling pathways, would be a boon to adipose tissue sensory nerve research. Single-cell transcriptomic data will help reveal cellular targets of these sensory neuropeptides, and spatial transcriptomics at nerve terminals in the tissue may also assist in a better understanding of how the sensory nerves function across metabolic states. Importantly, non-peptidergic sensory axons marked by P2X purinoceptor 3 also exist in adipose tissue, and their function is unexplored.

Third, the directionality of the cross-talk between the adipose sensory and sympathetic nervous systems remains unclear, with interpretations differing by model species and methodological technique used to alter sensory innervation or activity in adipose tissues. These studies become even more confounded by compensatory changes to innervation and nerve activity in non-targeted depots during denervation studies, as it is abundantly clear that nerves readily compensate through neurite branching or changes in nerve activity in the same and distal depots following denervation.

Fourth, a clear understanding of what signals in the adipose tissue are being sensed and communicated back to the brain (via adipose interoception, somatosensation or nociception) is lacking for adipose tissue, compared to other tissues such as skin or gut. While some endogenously produced hormones and lipids have been postulated as candidates, the field is far from defining the exact identity and mechanisms of adipose tissue ‘interoception’, or as the National Institutes of Health (NIH) Blueprint project defines it: the sensing and processing of signals from inside the body. As most sensory nerves in adipose are Na_v1.8 positive, suggesting they are nociceptors, which by definition sense noxious stimuli, it is possible that sensory nerves are sensing inflammatory signals or lipotoxic products as a defence mechanism to maintain healthy adipose tissue, but this is not certain.

Fifth, the nerve terminal structures formed by adipose sensory nerves are currently unclear. The neuro-adipose nexus, or axonal endings wrapping around adipocytes (where synaptic and vesicle markers are colocalized), may be one sensory nerve junction in the tissue¹⁰². Whether these terminals are mechanosensing, chemosensing or some other sensory modality remains to be determined.

In conclusion, adipose tissue sensory nerves are now more appreciated as important contributors to the neural control of adipose tissue functions and overall metabolic health, but numerous gaps in knowledge remain. The tools used to modify sensory nerve activity in specific depots, and the compensatory nature of nerve ablation studies, make the investigations more challenging. However, it is clearly advantageous to continue to nail down mechanisms by which sensory nerves function in adipose tissue, their heterogeneous diversity and the actions of their neuropeptide products on various cell types in adipose tissues, which may have important implications for metabolic diseases including obesity and metabolic syndrome.