Neuro-Immunity Controls Obesity-Induced Pain

The prevalence of obesity skyrocketed over the past decades to become a significant public health problem. Obesity is recognized as a low-grade inflammatory disease and is linked with several comorbidities such as diabetes, circulatory disease, common neurodegenerative diseases, as well as chronic pain. Adipocytes are a major neuroendocrine organ that continually, and systemically, releases pro-inflammatory factors. While the exact mechanisms driving obesity-induced pain remain poorly defined, nociceptor hypersensitivity may result from the systemic state of inflammation characteristic of obesity as well as weight surplus-induced mechanical stress. Obesity and pain also share various genetic mutations, lifestyle risk factors, and metabolic pathways. For instance, fat pads are often found hyper-innervated and rich in immune cell types of multiple origins. These immunocytes release cytokines, amplifying nociceptor function, which, in turn, via locally released neuropeptides, sustain immunocytes’ function. Here, we posit that along with mechanical stress stemming from extra weight, the local neuro-immune interplay occurring within the fat pads maintains the state of chronic low-grade inflammation and heightens sensory hypersensitivity. Overall, stopping such harmful neuro-immune crosstalk may constitute a novel pathway to prevent obesity-associated comorbidities, including neuronal hypersensitivity.

painkillers (Thomazeau et al., 2014). Inversely, patients suffering from chronic pain are also more likely to become obese (Stone and Broderick, 2012). Being highly prone to develop osteoarthritis (OA), they chiefly report musculoskeletal pain (King et al., 2013). Finally, obesity-induced pain varies with the patients' sex, diet, and body fat distribution. While significant advances were made in the understanding of the molecular mechanisms driving persistent pain, this knowledge has yet to translate into effective therapies. Thus, the management of chronic pain still constitutes a significant unmet clinical need (Yekkirala et al., 2017). The following section will explore the molecular mechanism and the relative contribution of (i) mechanical stress; (ii) chronic low-grade inflammation; and (iii) neuro-immunity to obesity-induced pain; and whether targeting such mechanism may constitute a therapeutic strategy to alleviate sensory hypersensitivity.

ORGANIZATION OF THE NERVOUS SYSTEM AND PAIN SENSATION
The nervous system is designed to quickly detect stimuli and direct avoidance behavior (Basbaum et al., 2009). In the periphery, autonomic neurons monitor and regulate organ functions (involuntary) while the somatic (voluntary) system provides sensitivity and motor control. In addition to these, nociceptors are a peripheral population of unmyelinated (Cfibers) or thinly myelinated (Aδ fibers) neurons specialized in sensing potentially damaging stimuli (pressure, temperature, chemical). These signals are transduced in electrical impulses, which are integrated centrally. In turn, effector signals are produced and transmitted to motor neurons activating muscle contraction (Albright et al., 2000) and occasionally tuning immune responses.
Upon activating nociceptor, the host generally experienced pain, which is defined as an unpleasant sensory and emotional experience (IASP, 2018). Pain is either acute, serving a protective physiological response to a harmful stimulus, or chronic, when physically debilitating and lasting over 3 months (IASP, 2018). Pain can also be stratified as (i) nociceptive, associated with the detection of potentially harmful tissue-damaging stimuli (Woolf and Ma, 2007); (ii) neuropathic, triggered by an injury/disease to the somatosensory nervous system (Costigan et al., 2009); (iii) dysfunctional, when associated to a disease state of the nociceptive nervous system (i.e., fibromyalgia) (Nagakura, 2015); or (iv) inflammatory, associated with tissue damage and active inflammation (Pinho-Ribeiro et al., 2017).
Inflammatory pain is characterized by the influx of immunocyte-producing cytokines, chemokines, and growth factors. These mediators typically bind to G protein-coupled receptors (GPCRs) and/or tyrosine kinase receptors expressed by nociceptor terminals and lead to intracellular kinases activation. Subsequently, these kinases decrease the activation threshold of ion channel transducer [i.e., transient receptor potential vanilloid-1 (TRPV1); transient receptor potential cation channel, subfamily A, member 1 (TRPA1)] and voltage-gated sodium channels (Na V ) (i.e., Na V 1.7, Na V 1.8, and Na V 1.9) (Caterina, 2000;Kerr et al., 2001;Nassar et al., 2004;Amaya et al., 2006;Bautista et al., 2006) or increased their membrane expression (Julius, 2013). Consequently, these effects sensitized nociceptors, which results in hypersensitivity to non-noxious stimuli, a situation termed allodynia (Woolf et al., 1992;Woolf and Ma, 2007). Besides, the nerve terminal "sensitized" state intensifies the response to painful trigger (termed hyperalgesia). Such hypersensitivity focuses the host on the injury site and leads to alternative behavioral (i.e., gait change). These effects are typically limited to the inflammatory sites, while systemic hypersensitivities result from central sensitization Rajchgot et al., 2019); see detailed review by Woolf et al. (1992) and Woolf and Ma (2007).

INFLAMMATION
Characterized by its pain, heat, redness, and edema (Larsen and Henson, 1983), inflammation suggests an immune-mediated phenomenon, but each of these characteristics can also stem from neuronal activation. Typically, noxious stimuli-induced action potential travels to the brain to initiate sensation (Talbot et al., 2016b;Azimi et al., 2017). When they reach the sensory neuron branch points, these electrical signals are also transmitted antidromically, back to the peripheral terminals, to initiate neurogenic inflammation via the local release of neuropeptides (Richardson and Vasko, 2002;Chiu et al., 2012). Subsequently, these peptides act on the endothelium to generate redness and heat (secondary to vasodilation) and edema (extravazation due to enhanced capillary permeability) (Foreman, 1987;Weidner et al., 2000).

MECHANICAL STRESS
Sustained overloading on the musculoskeletal structure of the lower back, hip, and knee joints prompts the development of OA in obese patients (King et al., 2013). Obesity-related OA is multifactorial and involves direct joint damage as well as genetic, biological, and metabolic factors (Davis et al., 1988). For instance, increased mechanical load changes the chondrocyte mechanotransducer signaling (Haudenschild et al., 2008), promotes cytokine secretion (IL-1β, IL-6, and TNFα) and matrix metalloproteinases release, and contributes to establishing a pro-oxidative microenvironment (Stannus et al., 2010). These prompts the degradation of type II collagen, joint extracellular matrix, and hyaluronic acid fragmentation. These factors caused (i) imbalance between deterioration and repair of the cartilage; (ii) chondrocyte apoptosis; (iii) reduced synoviocyte fluid viscosity; and (iv) increased joint friction, all of which changed the patients' posture and gait, reducing their mobility and increasing pain scores (Jordan et al., 2003). While OA affects 16% of adults, it is present in 23% of overweight and 31% of obese patients (Vincent et al., 2012). As such, obese patients are 35% and 11% more at risk of developing knee and hip OA, respectively ( Table 2; Jiang et al., 2012). Knee radiograph showed smaller width of the medial and lateral joint space in obese patients (Çimen et al., 2004), resulting in a 35% greater risk to undergo knee arthroplasty (Bourne et al., 2007).
These effects may be partly explained by the production of adipose tissue (AT)-secreted adipokines (leptin and adiponectin) (de Boer et al., 2012). These adipokines, whose circulating and cartilage levels correlate with BMI, activate their cognate receptors on the surface of chondrocytes, increasing the production of matrix metalloproteinases, nitric oxide (NO), and cytokines (IL-1β, IL-6, IL-8, TNF-α) (Figenschau et al., 2001). In turn, these mediators heighten synovial inflammation and pain hypersensitivity by driving fibroblast proliferation and immune cell infiltration (Francin et al., 2014).

OBESITY, A LOW-GRADE INFLAMMATION DISEASE
White adipose tissue (WAT) comprised pre-adipocytes, adipocytes, endothelial cells, and immunocytes and has a primary role in energy storage. Adipocytes originate from mesenchymal stem cells (Qian et al., 2010), and their production is mainly controlled by epigenetic regulators, growth factors, and CCAAT-enhancer-binding proteins (or C/EBPs) and peroxisome proliferator-activated receptor γ (PPARγ) transcriptional regulators (Wu et al., 1999). Increased mitochondrial metabolism and biogenesis results in reactive oxygen species production, which initiate adipocyte differentiation in a mammalian target of rapamycin complex 1 (mTORC1)-dependent manner (Tormos et al., 2011). White adipose tissue content strikingly increases FIGURE 1 | Neuro-immune crosstalk controls obesity-induced pain. Lean individual's adipose tissue is sparsely innervated and comprises few adipocytes and anti-inflammatory immunocytes. The accumulation of fat leads to the rupture of adipocytes and the secretion of adipokines. These mediators increase the chemotaxis of immune cells, enhancing the level of pro-inflammatory cytokines. By acting on their cognate receptors present on sensory nerves, these cytokines sensitize nociceptor neurons by increasing the expression and phosphorylation of NaV and TRP channels. Upon sensitization, the sensory neurons secrete neuropeptides, further polarizing the fat pad's immune cells. QX-314 silences nociceptor neuron's electrical activity, while CNCB2 prevents their release of neuropeptides. By stopping neuro-immune crosstalk, these treatments would help resolve fat pad inflammation and blunt pain hypersensitivity.
It is increasingly recognized that immunological factors drive obesity induction, a phenomenon present in the WAT as well as pancreas, liver, and intestines (Osborn and Olefsky, 2012;Jin et al., 2013;Winer et al., 2016). Thus, lean AT is mainly composed of M2 macrophages, ILC2s, eosinophils, regulatory T cells (Tregs), and T H 2 cells, while the obese fat pad is dominated by neutrophils, ILC1, M1 macrophages, and cytotoxic T cells. Similarly, the fat stromal vascular fraction of lean mice is composed of anti-inflammatory immune cells such as M2 macrophages (Lumeng et al., 2007a), Treg (Feuerer et al., 2009), and ILC2 . In contrast, M1 macrophages represent 10% and 40-50% of obese mice and patient stromal fractions, respectively (O'Rourke et al., 2012;Blaszczak et al., 2019). Such alternative and inflammatory composition support insulin resistance and maintain lowgrade inflammation.
Adipocytes release the monocyte chemoattractant protein-1 (MCP-1), which attracts C-C chemokine receptor type 2 (CCR2)expressing monocytes and favors their differentiation into M1 macrophages (Arner et al., 2012). Once in the WAT, macrophages form "crown-shaped structures" around dead adipocytes (Cinti et al., 2005), producing TNF-α, IL-1β, and IL-6 (Lumeng et al., 2007a). Therefore, M1 macrophages constitute one of the main source of cytokines in obese-WAT (Lumeng et al., 2007b) and, through PPARγ production, stimulate adipogenesis (Rosen et al., 2000). In addition, adipocytes tend to rupture due to their limited expansion capacity observed during obesity. The massive apoptosis of these cells drastically increased the levels of cytokines within the fat pad microenvironment and leads to the chemotaxis of M1 macrophages (Lumeng et al., 2007b).
Along with macrophages, the numbers of circulating monocyte and neutrophil increased during obesity (Poitou et al., 2011). Given that monocytes originate from hematopoietic stem cells (HSCs) and that obese patients have increased circulating HSCs progenitors, it was posited that HSCs might give rise to leukocyte influx (Bellows et al., 2011). Thus, HSCs enhance macrophage generation, via the myeloid differentiation primary response 88 (MyD88), a protein known to serve as an intermediate between extracellular danger signals sensed by TLR and the activation of the transcription factor nuclear kappa β (NF-κB) (Singer et al., 2014). These macrophages then go on to accumulate within the dorsal root ganglion (DRG) of high-fat diet (HFD)-fed mice (Song et al., 2017).
Diet-induced obesity CD4 + T lymphocytes were found to have biased T cell receptor (TCR) repertoires, suggesting an antigen-specific expansion. Glucose homeostasis typically becomes dysregulated in diet-induced obesity, when numbers of IFN-γ-secreting T H 1 cells overwhelm the non-expending pools of T H 2 (CD4 + GATA-3 + ) and Tregs (CD4 + CD25 + Foxp3 + ). Through T H 2 cell increases, CD4 + T cell transfer into HFD Rag1 −/− animals reversed weight increase and insulin resistance. Transient CD3 depletion also restores the T H 1/Treg balance and reverses HFD-induced insulin resistance, suggesting an upstream role for CD4 + T cell in controlling obesity-associated metabolic abnormalities (Winer et al., 2009).
Adipose tissue macrophages (ATM) initiate WAT inflammation. However, recent data suggest that other tissueresident innate immune cells, such as ILCs, also are major contributors (Yang et al., 2016). ILC2 number decreased in obese mice epididymis and human subcutaneous WAT (Brestoff et al., 2015). Being resident in lean AT, ILC2s maintain a T H 2-like status of AT , in which ILC2-produced IL-5 and IL-13 promote the beiging of WAT Brestoff et al., 2015;Hashiguchi et al., 2015). These effects are present in Rag1 null mice and are, therefore, independent of tissue-resident M2 macrophages (Hams et al., 2013;Molofsky et al., 2013). Thus, IL-33-driven WAT biogenesis suggests another role for an ILC2-inducing cytokine in regulating obesity (Brestoff et al., 2015;Lee et al., 2015). In effect, IL-33 null mice gain more weight than their wild-type counterpart and have reduced frequency of ILC2s. This phenomenon is also present in IL-33 KO mice fed a normal diet (Brestoff et al., 2015). Exogenous IL-33 rescued WAT ILC2s number and M2 macrophage (Brestoff et al., 2015).
High-fat diet -exposed mice have enhanced IFN-γ levels (Wensveen et al., 2015). The depletion of natural killer (NK) cells decreased HFD-induced insulin resistance and M1 macrophage levels but stopped the onset of obesity. Inversely, the adoptive transfer of splenic NK cells into IFN-γ null animals restores HFD-mediated insulin resistance (Wensveen et al., 2015). Tissueresident ILC1 may directly promote obesity-induced insulin resistance without the influence of natural killer T (NKT) or T cells (O'Sullivan et al., 2016). IL-12 activated ILC1 lead IFNγ production and subsequent polarization of M1 macrophages (O'Sullivan et al., 2016). ILC1-derived IFN-γ balanced out the effect of IL-33 mediated ILC2 activation within visceral AT (Oboki et al., 2010), providing an in situ negative regulators of ILC2 anti-obesity effects.
Overall, the influx and polarization of immunocytes by WATderived mediators enhance fat accumulation, speed up joint damages, and may directly sensitize nociceptor, as discussed in the next section. Blocking the inflammatory component of obesity may, therefore, constitute a potential therapeutic avenue to stop obesity progression and its comorbidities.

INNERVATION OF THE WAT
Under the control of a complex set of humoral and neural factors (Ismael et al., 2008;Dias et al., 2010;Talbot et al., , 2016aEl Midaoui et al., 2015), WAT-released mediators tuned the host energy status as well as the number and phenotype of immune, vascular, and structural cells (Ouchi et al., 2011). Along with blood-derived factors, adipocyte size, lipid mobilization, and paracrine secretion are controlled by sensory nerve terminals . Thus, fat pad sensory innervation is increased in obesity Vaughan et al., 2014). In addition, evidences suggest a dual, yet segregated, sympathetic and parasympathetic innervation of WAT (Kreier et al., 2002). We refer the reader's attention to the work of Bartness and colleagues for more information on WAT innervation ( Bartness and Bamshad, 1998). Overall, neurons control WAT production of cytokines and immune influx, making fat innervation a central component player in obesity-induced low-grade inflammation.

AUTONOMIC NERVOUS SYSTEM IN OBESITY
Sympathetic neurons control catabolic functions (Migliorini et al., 1997) via neuropeptide Y (NPY) suppression of lipolysis and promotion of angiogenesis and heighten adipocyte differentiation (Kuo et al., 2007). It is worth noting that adipocytes and macrophage-produced cytokines increase sympathetic flow, while excessive cytokine levels, such as during severe inflammation, have the opposite effect (Pongratz and Straub, 2014).
Parasympathetic vagal neurons that innervate the fat pad have anabolic functions helping tune insulin-mediated glucose and free fatty acid uptake and help promote lipid accumulation (Bartness, 2002). Conversely, lipid accumulation further increases their anabolic functions (Bartness, 2002). In doing so, norepinephrine controls triacylglycerol lipolysis, NO production, and tissue remodeling (Nguyen et al., 2018). Fat pad denervation decreased transcript expression of resistin and leptin, without impacting the levels of adiponectin. It, therefore, supports an anabolic role for WAT-parasympathetic neurons (Bartness, 2002;Kreier et al., 2002).
Harnessing such inflammatory reflex using bioelectronic devices (Chavan et al., 2017), non-invasive vagus neurons activate blunt inflammation in arthritic patients (Koopman et al., 2016) and mice with experimental inflammatory bowel disease (Ji et al., 2014). As such, parasympathetic neuron activation constitutes an innovative approach to tone down systemic inflammation, obesity progression, and obesity-associated comorbidities, such as pain hypersensitivity.
The effects of sensory neurons on immunity seem to vary between inflammatory context (T H 1/T H 17 vs T H 2), the neuron subpopulation implicated, as well as the nature of the peptides being secreted (Foster et al., 2015;Azimi et al., 2016Azimi et al., , 2017Talbot et al., 2016b). Typically, substance P (SP) promotes T cell activity and increased dendritic cell (DC) recruitment and recognition of non-self-antigens (Calvo et al., 1992;Siebenhaar et al., 2007). Calcitonin gene-related peptide (CGRP) have the inverse action, stopping T cell proliferation and reducing DC migration to LN (Mikami et al., 2011).
Within the WAT microenvironment, locally released neuropeptides (CGRP, SP) increase immune influx and polarization, heightening WAT inflammation and nociceptor sensitivity (Foster and Bartness, 2006). From the data obtained in mouse models of allergic inflammation, one would imagine that CGRP produced by fat pad-innervating sensory neurons would block the function of ILC2s cells, unbalancing the type 1/type 2 immunocyte ratio within the WAT. By favoring T H 1-mediated immunity, sensory neurons would enhance the influx of pro-inflammatory immune cells such as IL-1β and TNFα-secreting M1 macrophages and IFN-γ-producing ILC1. On the one hand, these cytokines would sensitize nociceptor neurons TRP and Na V channels, triggering pain hypersensitivity; on the other hand, the neurons would locally release more neuropeptides to further imbalance the fat pad local immunity (hypothesized integrated system in Figure 1). Blocking the neuro-immune interplay in such a context would have a twofold, yet synergistic, effect: (i) directly decreased obesity-induced pain trigger by inflammatory cytokines, and (ii) decreased chronic low-grade inflammation. We devised two translational approaches to do this.

TARGETING NEURO-IMMUNE CROSSTALK
First, we modified an efficient pain and itch neuron-blocking strategy to locally silence tumor-innervating nociceptors (Roberson et al., 2013). This strategy uses large pore TRP channels as a specific drug delivery device to transport charged local anesthetic (such as QX-314) into nociceptor neurons to stop Na + currents. In the context of inflammation, as found in the fat pad micro-environment, TRP channels open, allowing QX-314 (263 Da) to enter these neurons, producing a specific and durable electrical silencing (Binshtok et al., 2007). Of note, QX-314 did not impact immune cell function , confirming its selectivity for inflammation-activated nociceptors (Talbot et al., , 2020Foster et al., 2017). This strategy offers three major potential advantages: (1) high specificity (the effect is limited only to sensory neurons that express activated large pore channels), (2) long-lasting activity, and (3) limited side effects; the charge on QX-314 would limit diffusion through lipid membranes and redistribution outside of the respiratory epithelium.
Second, Bean and colleagues devised novel charged N-type Ca 2+ channel blockers, including an NCE termed CNCB2. The latter induced a prolonged pain blockade and was more potent than its neutral analog at inhibiting nociceptor release of CGRP and acted at lower concentrations to stop the neurogenic inflammation component of asthma. Such cationic molecules are therefore suited to treat pain by stopping potential action generation in nociceptive neurons and reducing inflammation by blocking pro-inflammatory neuropeptide release (Lee et al., 2019).

CONCLUDING REMARKS
Peripheral sensitization is a major contributor to inflammatory pain (Albright et al., 2000;Woolf and Ma, 2007;Ji et al., 2014). Because several sensitizing mediators are released simultaneously during inflammation, stopping one of these mediators is likely to have a limited impact. Silencing sensitized neurons or shared downstream signaling pathways should therefore have larger and provide more prolonged pain relief. Among the others, QX-314, CNCB2, or activation of parasympathetic neurons using bioelectronic medicine, may constitute such broadly acting strategy in reversing the neuro-immune component of obesityinduced inflammation and pain.

AUTHOR CONTRIBUTIONS
All authors wrote the manuscript.