The Afferent Function of Adipose Innervation

Abstract

Adipose tissue innervation is critical for regulating metabolic and energy homeostasis. While the sympathetic efferent innervation of fat is well characterized, the role of sensory or afferent innervation remains less explored. This article reviews previous work on adipose innervation and recent advances in the study of sensory innervation of adipose tissues. We discuss key open questions, including the physiological implications of adipose afferents in homeostasis as well as potential cross talk with sympathetic neurons, the immune system, and hormonal pathways. We also outline the general technical challenges of studying dorsal root ganglia innervating fat, along with emerging technologies that may overcome these barriers. Finally, we highlight areas for further research to deepen our understanding of the afferent function of adipose innervation.

Introduction

In the peripheral nervous system (PNS), the term efferent typically refers to signals that convey output from the central nervous system (CNS) to the peripheral organs (for example, from motor nerves to the skeletal muscle). Conversely, afferent refers to pathways that transmit information from the peripheral tissues to the CNS. In general terms, the brain receives ascending sensory information from the afferents and controls peripheral functions through efferents. By this convention, for example, the sympathetic/parasympathetic nervous systems, although not made up of motor nerves, are generally considered to be the efferent part of the PNS, as they exert control rather than transmit information to the brain.

In mammals, the major afferent pathways are somatosensory and vagal afferents. Somatosensory afferents, with their soma in the dorsal root ganglia (DRG), primarily innervate the skin and muscle; they transmit pain, temperature, mechanical, and other stimuli to the spinal cord. Vagal afferents, with their soma in the nodose ganglia, mostly innervate many internal organs. Traditionally, the role of interoception, that is, the sense of internal organs, is thought to be mediated largely by the vagal afferents (1–7). However, the involvement of DRG and somatosensory pathways in interoception is being increasingly recognized across many internal organs (8–10).

In addition to bridging the brain to internal organs, efferent signals also connect adipose tissue to the CNS. Adipose tissues are heavily innervated. The sympathetic component of this innervation, which was first identified by histofluorescence in the 1960s (11), plays a crucial role in modulating adipose activity, as summarized by Bartness et al. (12,13). Recent technical advances have revealed extensive details about the architecture and mechanism of sympathetic innervation in fat. Tissue clearing and three-dimensional imaging showed morphological and anatomical features of sympathetic nerves. AdipoClear, for example, has been used to demonstrate regional variation in nerve distribution as well as neurite growth during development (14,15). Studies with transgenic mice and molecular tools have shown that there is bidirectional regulation between adipocytes and sympathetic neurons (16–18). It is now well accepted that sympathetic neurons substantially innervate thermogenic fat, controlling thermogenesis, lipolysis, and lipogenesis (19–21). In parallel, adipose biology has been rapidly advancing. Extensive research has established the molecular characteristics, cellular origins, and functions of brown and beige adipocytes, and various adipokines, such as adiponectin and leptin, have been shown to exert regulation on other peripheral organs and the brain (22,23). Collectively, these findings depict a situation whereby CNS regulation of adipose activity is mediated by rapid and anatomically defined sympathetic-adrenergic (efferent) actions, while feedback from fat is mediated by circulating and hormonal mechanisms.

In addition to the sympathetic innervation, there have also been reports of afferent or sensory innervation of adipose tissues (13). Sensory-specific neuropeptides have been detected (although very sparsely) in white adipose tissues (WAT) (e.g., perigonadal WAT and inguinal WAT [iWAT]) and in interscapular brown adipose tissue through immunostaining (24–26). Neuroanatomical tracing suggested the presence of sensory fibers in WAT (27,28). Further evidence for the role of sensory innervation in adipose tissue came from viral tract tracing studies that suggested potential interaction pathways between sympathetic and sensory neurons (29) and implicated the involvement of higher brain regions (i.e., dorsal motor nucleus and paraventricular hypothalamic nucleus) in fat regulation (28).

Despite these elegant studies, however, the significance of sensory innervation has been overshadowed for a few potential reasons. First, early anatomical studies presented inconsistent findings across different rodent models. For example, retrograde tracing studies from iWAT to DRG have yielded dramatically different results: True Blue in rats identified 2–20 neurons per DRG (27), whereas herpes simplex virus in hamsters showed more than 80 neurons per 20-μm DRG section (28), but a later study reported around 20 neurons in each ganglion (29). A more recent report using fluorescent cholera toxin subunit B in mice found <5 DRG neurons labeled per animal (30), calling into question the significance of sensory innervation in mice, especially compared with the prominent role of the sympathetic counterpart. Second, sensory nerve denervation resulted in mild or indirect effects. Injection of high-dose capsaicin, which supposedly kills sensory afferents, into local fat had no effect on the injected depot but affected cell growth and activity in distant depots (26,31,32). These findings have led to the notion that the roles of sensory nerves in fat may be less significant and indirect compared with those of sympathetic pathways.

Recent Technology Developments and Discoveries

Recently, we began to revisit the role of afferent signaling in fat, first by examining the limitations of conventional methodologies used in studying fat innervation. First, there is a general lack of a robust and specific molecular marker for sensory neurons. DRG neurons are known to be molecularly heterogenous (33), and pan-sensory markers used in transgenic lines, such as Pirt or Advillin (34,35), do not exclusively identify sensory neurons over sympathetic neurons. SNS (also known as Nav1.8 or Scn10a-Cre) (36,37) are also used as pan-sensory labeling lines, but their specificity (vs. sympathetic neurons) and coverage in adipose-innervating DRG remain to be thoroughly characterized (35,38). CGRP, despite being a prevalent marker for staining sensory neurons in adipose tissues, covers less than half of all DRG neuron somas (33,39,40), and its representation among all axons is indeterminate. Thus, reliable quantification of adipose afferents remains challenging. A second limitation related to the lack of a molecular marker is that experiments using conventional tools for manipulating adipose afferents can be easily confounded by anatomical and pharmacological limitations. For instance, surgical denervation failed to separate afferent versus efferent nerves, while chemical denervation with capsaicin can only target a subset of DRG neurons (30–40%). These denervation methods (which affect nerve terminals), while useful for exploring axonal plasticity within fat, can lead to misconstrued functional interpretations due to nerve regeneration and timing of postdenervation analysis (41–43). Furthermore, capsaicin-mediated denervation can be confounded by other cell types expressing TRPV1 (44–46), potential nerve regeneration at the chosen time of examination (42), and nonspecific effects of capsaicin at high doses (47). Third, traditional histological approaches are limited to samples of thin sections of adipose tissues and thus cannot trace the origin of the nerves. Although newer adipose-clearing methods are enabling a paradigm-shifting understanding of the nerve structures in fat depots, it is still difficult to use three-dimensional imaging methods to trace the neuronal soma of fat innervation (i.e., to identify the originating ganglia), as this requires a much larger clearing/imaging volume beyond a single fat pad, which is challenging for these methods.

To overcome these limitations and assess the role of sensory nerves in fat, we developed a set of new technologies: 1) a tissue-clearing method, called HYBRiD (for hydrogel-based reinforcement of DISCO), that expands imaging capacity to a whole mouse, which allowed us to trace full-length sensory projections from the soma (located in DRG) to axonal terminals (located in fat) without physical sectioning and reconstruction; 2) a surgical preparation to specifically deliver adeno-associated virus (AAV) vectors into DRGs that allows long-term AAV-mediated gene expression; and 3) a retrograde AAV vector that specifically labels DRG neurons innervating fat in an input-output defined manner (Fig. 1). By applying these new tools, we found that mouse WAT (iWAT and perigonadal WAT) receives robust DRG afferent innervation. Functionally, this sensory innervation serves as a counteractive mechanism on sympathetic functions in regulating thermogenesis and lipogenesis in iWAT (35).

Figure 1.

Peripheral innervation and strategy for manipulating sensory neurons that innervate a specific organ. (Top) Afferent and efferent innervation of internal organs, focusing on the somatosensory neurons (blue) and sympathetic neurons (magenta). Vagal afferents/efferents, parasympathetic efferents, and enteric neurons are not shown. (Bottom) Strategy for manipulating sensory neurons that innervate a target organ, using fat as an example. An AAV-encoding Cre-dependent construct is administrated into DRG, whereas retrograde AAV-expressing Cre recombinase is delivered into targeted organs to achieve organ specificity.

This recent work made a few interesting observations. First, sensory afferents have a stronger presence in WAT than in BAT. Previous work found a positive correlation between sympathetic parenchymal fiber density and depot thermogenic capacity, favoring BAT over WAT (12,14). These new results suggest a reverse correlation between sensory fiber density and thermogenic capacity, consistent with the proposed counterregulatory role of sensory nerve and sympathetic actions. Moreover, tyrosine hydroxylase (TH), which has been widely used as a sympathetic marker, not only marks a subset of DRG sensory neurons (48) but also is directly quantified to be expressed in 40% of smaller sensory afferents in the adipose tissue, suggesting that using TH to quantify sympathetic innervation could be substantially confounded by sensory fibers.

Thus, a comprehensive reexamination of WAT innervation is needed, along with a recalibration of adipose innervation profiles across different depots. For example, it has been suggested that obesity, insulin resistance, and aging are associated with reduced sympathetic activity/innervation in thermogenic fat due to changes in many neurotrophic factors (17,18,49–54). It is worth reexamining the sensory versus sympathetic changes under different metabolic conditions to gain a refined understanding of how these factors affect innervation.

The Physiological Role of Afferent Innervation

Our recent work began to reveal that sensory afferents could dampen sympathetic activity in fat, but it remains unclear how these counteractive pathways work together to control adipose function under physiological conditions. Although sensory neurons can also function as efferents through release of neuropeptides in peripheral terminals to regulate local environments (55,56), here we only focus our discussion on the potential role based on the afferent aspect.

The foremost question is what information is being sensed and transmitted by the afferent nerves. Having a dedicated sensory pathway to relay adipose information through an ascending spinal cord pathway (a typical somatosensory route to transmit pain, hot, cold, or mechanical sensations from skin) is indicative of several key differences compared with circulating signals.

First, a dedicated sensory pathway is faster than sending information via circulating signals and can potentially report acute changes in real time (on a second-to-minute scale, in theory). For example, can the brain gauge how much fat is being taken out of or put into adipose tissue (as a general term, as we currently do not know the specific biochemical process being sensed) during lipolysis and postmeal fat storage? Both physical and chemical signals can be detected by the sensory terminals to allow the brain to get real-time control over the rate of both processes. Furthermore, local temperature within the fat pad could change upon exposure to cold (especially for subcutaneous fat [57]) or with increased thermogenesis, changes in volume and mechanical properties of fat tissue (adipocytes as well as the mesenchymal structures) with obesity or starvation, or the release of lipid metabolites during lipolysis. It does not have to be one of these, but multiple signals from these processes can be monitored by multiple unimodal sensory neurons or multimodal sensory neurons (58). In recent years, the molecular identities of several sensory modalities have been revealed, including TRPV1, PIEZO2, and other genetic markers from single-cell RNA sequencing and screens (33,59,60), opening many possibilities to combine genetic and circuit approaches to specifically test what and how adipose information is being sensed by afferent neurons.

Along these lines, a fast-sensing channel also suggests that the brain can potentially react to adipose changes rapidly, in contrast to the typical metabolic responses that happen on a much longer time scale for the more common homeostatic control (regulated by hormones). It has been suggested that adipose thermogenesis promotes satiation and suppresses feeding (61). Conversely, there is strong evidence that high-calorie intake activates thermogenic pathways in fat (62,63). Together, these studies suggest there may be an afferent feedback loop from adipose tissue that plays a role in these rapid, cross-system regulations, and it will be informative to test this idea in future studies.

A second unique feature of afferent pathways is that by having hardwired routes, spatial information about adipose metabolism is preserved, allowing the brain to “know” the anatomical source of the signal, whereas it is less likely for the brain to discern where circulating hormonal signals are coming from (in terms of their location of origin). How much does spatial/depot specificity contribute to the metabolic differences between depots? For example, does it affect the sensitivity of subcutaneous and visceral fat to insulin signaling or β-adrenergic hormones? Can the brain differentially regulate different aspects of adipose biology (lipolysis, lipogenesis, and adipogenesis) in different fat depots based on specific afferent feedback? Are these episodes of feedback mediated by different DRG subtypes? Such questions are particularly important in disease contexts where the heterogeneity of the depot is known to have a big contribution to pathology. Another key question is the role this location and depot-specific information plays in response to changes in adiposity (i.e., during development or via systemic weight loss or surgical removal). These conditions will likely result in distinct changes in afferent feedback, and understanding their contribution to maintaining fat loss could be significant. We believe the framework and approaches established by Wang et al. (35). can be deployed to help answer these newer questions about the temporal and spatial specificity of adipose regulation, particularly neuronal control of adipose metabolism.

Potential Cross Talk Among Sensory Nerves, Sympathetic Nerves, and the Immune System

By studying the molecular signature of the whole fat pad, we found that the most prominent phenotypic changes after afferent ablation in inguinal fat are on the upregulation of many canonical sympathetic nerve–controlled programs, such as thermogenesis and de novo lipogenesis. This phenotype can be largely abolished by additional sympathetic ablation, suggesting that the mechanism is dependent on sympathetic signaling, although this does not fully exclude the possibility of direct sensory involvement in these pathways. Moreover, we note that we only examined fat gene expression after several weeks of sensory ablation. Although the observed directionality of sympathetic activity differs from a previous report using local capsaicin denervation in hamsters (32), it aligns with findings from several genetic mouse models, such as whole-body knockout of CGRP (64) and ablation of CGRP-expressing sensory neurons (65), both of which showed elevated expression of β3-adrenergic receptor and enhanced sympathetic activity in fat tissue. Thus, although our results suggest sensory-sympathetic interactions, the mechanisms of this interaction, be they direct or indirect, remain to be determined, which we discuss below.

In addition to our findings, multiple previous examples of potential cross talk between the sympathetic and sensory nervous systems were reported. Even though parasympathetic signals typically antagonize sympathetic signals, there are reports of sensory afferents regulating sympathetic impulses, such as in blood pressure regulation and thermoregulation (66–68). The sensory-mediated inhibition of sympathetic output can happen at multiple levels in the CNS in the brain (through the canonical dorsomedial hypothalamus–rostral raphe pallidus nucleus/paraventricular hypothalamic nucleus–rostral ventrolateral medulla pathway [69]), at the spinal level (70), ganglion level (through nerve terminal sprouting) (71), or in the tissues (72); however, where such cross talk happens in the case of adipose sensory afferent signals and sympathetic efferent signals remains to be determined. A longer, brain-dependent loop has been proposed (by Bartness et al. [32]), but such a long loop would lose depot specificity and likely work through systemic increase of sympathetic tone, which is contradictory to our observation that such a loop has spatial and depot specificity.

The cross talk between the sensory and sympathetic systems could also be indirect. Besides adipocytes and nerve endings, there are diverse cell populations in the adipose tissue, such as immune cells and vasculature. In particular, immune cells have been shown to have a significant role in regulating thermogenic pathways and sympathetic activity in fat (73,74). Recently, the sensory system has been widely reported to regulate interactions between peripheral immune functions and the brain (75,76). Specifically, the lymph node in inguinal fat is also innervated by sensory nerves (8), and it would be interesting to test the role of immune components in the sensory regulation of adipose functions. It is also worth noting that besides the sensing (afferent) function, it is well recognized that sensory terminals can secrete neural peptides such as CGRP or substance P, many of which have been proposed to have functions in adipose tissues (64,65,77) and the local microenvironment (76,78).

Potential Interactions With Hormonal Pathways

It is conceivable that the brain receives both “wired” (sensory afferent) and “wireless” (circulating) signals from fat in an integrated and coordinated fashion. It is thought that homeostatic signals are primarily received and processed by the hypothalamus. Leptin and leptin receptor (LepR) neurons in the arcuate and lateral hypothalamus are well studied in this regard (79–81). It would be interesting to determine if afferent signals from fat also converge onto the same populations or brain regions for coordinated homeostatic control. In the PNS, it has been reported that DRG neurons express LepR and can respond to leptin signaling in an ex vivo setting (82), suggesting the cross talk between hormonal and afferent pathways can occur as early as the spinal cord level, although whether this cross talk happens in vivo with endogenous leptin is unclear. Another recent article reported that the major component of the insulin signaling pathway, mTORC2, regulates sensory innervation (83). It remains to be determined if, where, and how other adipokines interact with afferent pathways at the peripheral or at the central level.

Challenges and Opportunities

As mentioned earlier, the technical barriers to study DRG, especially in an organ-specific manner, have hindered the study of afferent functions in adipose tissues. Although recent advances in multiple domains, such as tissue clearing and imaging (14,84,85) as well as viral and genetic engineering, have begun to open new directions in this area, several major challenges remain. First, the anatomical complexity of DRG makes them difficult to study with typical brain circuit approaches. The somatosensory system has a large number of pairs of ganglia (∼30–31 pairs of DRG for mice) and has “multiple-to-multiple” projections to peripheral tissues (i.e., each organ receives innervation from multiple levels, and each ganglion innervates multiple organs). The DRG is deeply buried in the bone and lacks stereotypical coordinate targeting, such that it requires manual and fine surgical manipulation. This embedded anatomy also makes it hard to access for in vivo electrical physiology or optical imaging methods, and long-term recording of DRG activity in awake or freely moving animals is much more challenging compared with that of the brain or trigeminal nerve due to severe motion artifacts. Second, molecularly and functionally, sensory neurons are more heterogeneous than sympathetic neurons. The latter primarily consist of adrenergic nerve terminals for release of norepinephrine, whereas sensory counterparts are known to be a mixture of multiple modalities, making it extremely hard to use a single genetic marker to achieve coverage and specificity. In addition, multiple established sensory Cre lines have leaky expression in sympathetic neurons (34,35). While such leakage is not a major concern in typical somatosensory studies focusing on the skin, adopting these Cre drivers to study adipose afferent function can be uniquely challenging due to the significant confounding contribution from the sympathetic system. Third, current approaches for DRG manipulation (genetic or chemical ablation) generally lack temporal resolution and reversibility. Incorporating modern neuroscience techniques such as optogenetics and chemogenetics would better reveal the dynamic nature of afferent regulation. However, adopting these CNS-optimized tools to the PNS needs thorough validation in addition to phenotypical readouts. This is an important caveat that has not been adequately addressed in the literature.

Nonetheless, we are optimistic that the emerging tools, especially those adapted from brain technologies, will continue to transform the approaches used to study peripheral innervation. Several recent reviews (86,87) highlight the rapid advances and revived enthusiasm toward understanding the communication between fat and the brain. Better optics, sensors, and sample preparation are making long-term multiphoton imaging of spinal cord and DRG feasible (88–90). Ultrasensitive opsins are enabling noninvasive optogenetics to reach deeper organs (91,92). It is conceivable that soon we will be able to have better optical recording and manipulation of fat afferents in a specific and robust manner. Meanwhile, major efforts and breakthroughs have led to the generation of a large set of genetic tools for better combinatorial labeling of sensory cell types based on newer single-cell RNA-sequencing transcriptome data (33,93), enabling highly specific genetic dissection of different afferents and their functions. We anticipate these Cre lines can be combined with ROOT-based target specificity to enable unprecedented access to organ-specific, cell type–defined study of afferent function in fat and beyond.

Funding Statement

This work was supported by the National Institutes of Health Director’s New Innovator Award (DP2DK128800), National Center for Complementary and Integrative Health grant R01AT012051, and National Institute of Diabetes and Digestive and Kidney Diseases grant K01DK114165.