The five mitochondrial respiratory chain complexes (complex I–V), known as the mitochondrial electron transport chain (mETC), are located in the inner membrane. These series of complexes transfer electrons from nicotinamide adenine dinucleotide (NADH) or flavin adenine dinucleotide (FADH), formed in the tricarboxylic acid cycle (TCA) in the mitochondrial matrix, to oxygen. This electron transfer is needed to create a transmembrane electrochemical gradient by pumping protons across the membrane. The flow of protons back into the matrix through complex V allows ATP production (45, 122, 123). OxPhos is responsible for 90% of the ATP consumed by neurons. Its importance is highlighted by deficits in OxPhos, which reduces dendritic synaptic plasticity in neurons, leading to neuronal injury or even cell death (39, 124). Exposure of neurons to inflammatory cytokines, such as TNF and/or IFNγ, depolarizes the mitochondrial membrane potential, impairs OxPhos and ATP production (125, 126). Conversely, exposure of sensory neurons to IL-17, IL-1α or IL-1β increased mitochondrial respiration to support neurite outgrowth (127, 128). Thus, inflammation affects mitochondrial respiration and neuronal function directly. Importantly, interfering with mitochondrial respiration in neurons affects sensory processing and pain. For example, intrathecal administration of complex I or III inhibitors induced mechanical allodynia in naïve mice (129). Furthermore, complex I deficiency in humans is associated with muscle pain (111). In addition, a deficit in mitochondrial respiration may also indirectly contribute to pain development. Genetic disruption of mETC complex IV activity in mouse DRG neurons increased the adenosine diphosphate (ADP)/ATP ratio due to impaired ATP production. The relative increase in ADP induced mechanical and thermal hypersensitivity through activation of purinergic receptor P2Y1 expressed on sensory neurons afferents (49).

The oxygen consumption rate (OCR), a measure of mitochondrial respiration, is reduced in lumbar DRG neurons at the peak of transient carrageenan-induced inflammatory pain, but is increased again when pain had resolved (25). In CFA-induced persistent inflammatory pain, mass spectrometry analysis identified that several proteins involved in mETC are reduced, suggesting that mitochondrial respiration in the DRG is affected in persistent inflammatory pain (66). In the lumbar spinal cord, OxPhos is reduced in rats with CFA-induced pain, or in mice after chronic constriction induced nerve injury (65), indicating that deficits in mitochondrial respiration may contribute to persistent inflammatory pain and neuropathic pain, respectively. Dichloroacetate (DCA), which activates pyruvate dehydrogenase through inhibiting pyruvate dehydrogenase kinase, boosts the TCA cycle and OxPhos (130). The reversal in mitochondrial respiration deficit induced by DCA in the spinal cord reduced pain-associated behaviors in rats and mice after CFA-induced inflammatory pain and chronic construction induced nerve injury (65). In contrast, when rotenone, a mETC complex I inhibitor that decreases OxPhos, is injected at the site of inflammation, it decreased CFA-induced mechanical hypersensitivity (66). Even though these findings seem contradictory, they may be explained by differential effects of modulating mitochondrial respiration at peripheral nerves and at the spinal cord/DRG. Indeed, preliminary data shows that intrathecal administration of complex III inhibitor myxothiazol into the spinal cord/DRG inhibited hyperalgesic priming, a form of latent nociceptor plasticity, but not when myxothiazol was injected locally in the inflamed paw (64). Moreover, intrathecal injection of rotenone (complex I inhibitor) or antimycin (complex III inhibitor) in naive mice generated persistent pain (129), whilst rotenone decreased CFA-pain when injected into the inflamed hind paw (66). To what extent these differences are merely a difference between non-inflammatory and inflammatory conditions remains to be determined. Moreover these injections do not target only neurons, but also various other cells such as skin cells, immune cells, glia, and others. Thus, differences may depend on which cells are targeted.

There is also evidence that a single inflammatory agent changes mitochondrial activity in sensory neurons. As example, the prototypic inflammatory prostaglandin E₂ (PGE₂) increased the TCA cycle and OxPhos through an EPAC2 dependent pathway, contributing to acute inflammatory pain (63). Another indication that mitochondrial respiration may contribute to inflammatory pain is that expression of ATPSc-KMT (also known as FAM173B), a mitochondrial methyltransferase that promotes OxPhos and mtROS production in neurons (131), is increased in DRG neurons of mice with chronic inflammatory pain. Knockdown of ATPSc-KMT in the lumbar DRG during the established chronic inflammatory pain attenuated hyperalgesia. Conversely, expression of ATPSc-KMT in sensory neurons of mice with transient inflammatory pain caused persistent pain (46). These findings may be relevant to rheumatic pain patients because a genetic polymorphism downstream of the ATPSc-KMT gene is linked to joint-specific chronic widespread pain (132). Moreover, ATPSc-KMT expression is increased in the subchondral bone of OA patients and is negatively correlated with pressure pain sensitivity (104). The cells in which ATPs-KMT expression is increased remain to be determined. Nerve cells may be possible candidates, because innervation is increased in the subchondral bone of OA patients (32), but nerve cells are still much less abundant than joint cells in this case.

Hyperalgesic priming triggered by peripheral inflammation affects the plasticity of sensory neurons, by augmenting mRNA translation and causing a switch from cyclic adenosine monophosphate (cAMP) to protein kinase C ε (PKCε-dependent signaling, following a subsequent inflammatory stimulus that acts on G-protein-coupled receptors (GPCRs) (30, 133). Hyperalgesic priming requires the activity of mETC complexes and various mitochondrial proteins are targets of PKCε-dependent signaling (134, 135), pointing to a role of mitochondria in hyperalgesic priming. Activated PKCε promotes the opening of mitochondrial ATP-dependent K⁺ channels (mitoKATP) and increases OxPhos (136, 137). Moreover, preliminary data suggest that hyperalgesic priming increased OxPhos in DRG sensory neurons (64).

Overall, inflammatory mediators alter mitochondrial respiration in neurons. Based on current evidence, some inflammatory agents promote mitochondrial respiration, whilst others reduce it. Possibly, the effect differs depending on the time point after the administration of the inflammatory agent (25). Nevertheless, current data show that both increased and decreased OxPhos result in pain. The question remains why both reduced and increased OxPhos may contribute to pain? Possibly, impaired OxPhos reduces ATP production, which could affect the stability of the membrane potential and disturb neuronal excitability (45, 138). In contrast, increased OxPhos may come at the costs of more production of mtROS as a byproduct of mETC, leading to hyperexcitability, as will be discussed in the next section. Further research will need to contemplate how exactly mitochondrial respiration is linked to sensory neuron function.

Mitochondria generate approximately 90% of cellular ROS (139). MtROS production damages mitochondria in a range of pathologies, including neuropathies (45), and is important in redox signaling (140). Superoxide (O₂⁻) is the proximal mtROS, mainly released at the mETC during OxPhos. Complex I and Complex III are the major sources of mtROS, and complex II to a lesser extent. Although the mETC is the main source of ROS, several other matrix proteins and complexes, like TCA cycle enzymes (e.g., pyruvate dehydrogenase), or some inner mitochondrial membrane proteins whose activity is partially dependent on mitochondrial membrane potential, also produce ROS.

Approximately 0.2%–2.0% of the oxygen consumed by mitochondria is reduced to O₂⁻, which is subsequently converted to other ROS, such as hydrogen peroxide (H₂O₂) and hydroxyl ions (OH⁻) (140, 141). An imbalance between mtROS production and its removal, either due to ROS overproduction and/or decreased antioxidants defense, causes oxidative stress (139, 142). Neurons, like all other mammalian cells, have antioxidant enzyme systems, such as superoxide dismutase, peroxidases or catalases, which scavenge mtROS when they are generated. Despite contributing to pathology, mtROS also act as signaling molecules, to ensure quality control and maintenance of functional cells, and regulate a variety of physiological processes, such neuronal differentiation, synaptic pruning, and neurotransmission (143, 144). A disturbed mETC is one of the major drivers of mtROS (139), but also decreased mitochondrial membrane potential, disrupted mitochondrial Ca²⁺ buffering, altered mitochondrial morphology, or cellular stress in general augment mtROS production, which disrupts the redox balance and causes oxidative stress that negatively affects neuron function (143, 145).

Neurons are more likely to suffer from mtROS-induced oxidative stress compared to other cell types, because of their large energy consumption, mainly supported through OxPhos. Moreover, neurons have a high content of unsaturated fatty acids and proteins that are vulnerable to oxidation (146). A disrupted redox balance has detrimental consequences for neuronal functioning (139, 141, 147). As an example, mutations in superoxide dismutase (Cu-Zn), also known as superoxide dismutase 1 (SOD1), cause motor neuron degeneration in amyotrophic lateral sclerosis patients (125, 148), of which 60% develop pain (149). Antimycin A (mETC complex III inhibitor) induces mtROS, which activates TRPA1 receptor and increases the excitability of sensory neurons (150–152). Similarly, treatment of spinal cord neurons with ROS donors augmented their excitability and intrathecal administration of these ROS donors induced mechanical hypersensitivity in rats (153). Increased ROS production in the spinal cord also promoted pain through reducing firing of spinal inhibitory neurons. In contrast, others have found that ROS increase the activity of a different subset of inhibitory neurons (154, 155). Although some studies have shown decreased activity, in general, ROS appears to increase neuronal excitability and promote pain.

Both inflammatory and neuropathic pain are associated with increased ROS production in the peripheral and central nervous system. In mice, intraplantar capsaicin administration induced mtROS in spinal cord neurons. Overexpression of the anti-oxidant mitochondrial manganese dependent superoxide dismutase (MnSOD), also called SOD2, prevented capsaicin-induced hyperalgesia, indicating requirement of mtROS for hyperalgesia development (69). Importantly, rare variants present in mETC genes that are major sites of mtROS formation, are associated with the severity of erosive RA (156). Accordingly, increased mtDNA mutations, often a consequence of mtROS, are found in RA and PsA patients (157). These mtDNA mutations affect mitochondrial function and further promote mtROS production (144, 158). Even though these mutations were identified in synoviocytes, similar mutations may occur in sensory neurons innervating chronically inflamed tissues. Indeed, inflammation triggers mtROS production causing mutations in mtDNA in neurons (146). A role for mtROS in pain is further substantiated by findings that overexpression of mitochondrial ATPSc-KMT induced mtROS formation in DRG neurons and prolonged inflammatory pain. The antioxidant phenyl-N-t-butylnitrone (PBN) or the mitochondria-targeted antioxidant mitoTEMPOL reversed the persistent pain induced by ATPSc-KMT overexpression (46, 64). Various other studies showed that antioxidants (e.g., resveratrol) reduced pain in rodent models of OA and neuropathic pain (159–162). Overall, these data indicate that ROS, likely mitochondrial derived, contributes to inflammatory pain.

There is even more evidence for a role of mtROS in neuropathic pain. In spinal dorsal horn neurons, mtROS is increased in rats 1 week after spinal nerve ligation (70). SOD2-like antioxidants reduced mechanical and heat hyperalgesia induced by spinal nerve ligation or chronic constriction injury in rats (163, 164). Chemotherapeutic agents, such as cisplatin and oxaliplatin, cause nerve damage and pain through mitochondrial dysfunction, due to increased mtROS production and decreased antioxidant protection (58, 71, 144, 158, 165, 166). Intraperitoneal administration of mitochondria specific (SS-31, TEMPOL, or a SOD2 mimetic) or unspecific antioxidants (PBN, pioglitazone) reduced ROS production and attenuated pain in chemotherapy-induced neuropathy models (45, 47, 48, 71), showing that mtROS production is a driver of neuronal dysfunction and pain.

Although direct evidence for a contribution of mtROS in the nervous system for pain in rheumatic disease is still lacking, findings in neuropathic pain and inflammatory pain models suggest that oxidative stress induced by mitochondrial dysfunction may contribute to pain by increasing neuronal excitability.

The endoplasmic reticulum and mitochondria are the two major intracellular Ca²⁺ storages (167). In neurons, mitochondria take up Ca²⁺ into the mitochondrial matrix through the mitochondrial Ca²⁺ uniporter (MCU) complex (55, 167–169). The MCU is Ca²⁺-sensitive and opens by elevated cytosolic Ca²⁺, allowing Ca²⁺ to flow into the matrix. Mitochondria release Ca²⁺ primarily by the Na⁺/Ca²⁺ exchanger (NCLX) (170). The maximal rate of release is much lower than the maximal uptake rate. Therefore, under conditions of continuous high cytosolic Ca²⁺, e.g., due to continuous firing of neurons, mitochondrial Ca²⁺ accumulates. Mitochondria have an enormous capacity to accumulate and store Ca²⁺. In resting neurons, total mitochondrial Ca²⁺ is approximately 100 μM and free mitochondrial Ca²⁺ only 0.1 μM (171). These concentrations steeply increase when neurons are active and mitochondria start accumulating Ca²⁺ (172). In extreme conditions, mitochondrial Ca²⁺ reaches a concentration of up to 1,500 μM (173). In case of mitochondrial Ca²⁺ overload, the mitochondrial permeability transition pore (mPTP) is formed, which releases Ca²⁺ and other molecules, such as Cytochrome C, into the cytosol, inducing apoptosis. This pore has been mainly studied under pathological conditions, and several core components like ATP synthase, cyclophilin D, and the adenine nucleotide translocators are thought to be involved (123, 167, 174).

Only a relatively small fraction of Ca²⁺ is handled by mitochondria during physiological cytosolic Ca²⁺ signals. Nevertheless, mitochondrial Ca²⁺ influx and efflux play a role in the spatiotemporal organization of the cytosolic Ca²⁺ signals (175), which regulates activity-dependent signaling and neuronal excitability in nociceptors, thus is important to prevent aberrant signaling and pain (176). A rise in mitochondrial matrix Ca²⁺ stimulates OxPhos-mediated ATP and mtROS production, regulates organelle dynamics and trafficking, and modulates neurotransmitter release, synaptic transmission, and excitability. Mitochondria are a major regulator of Ca²⁺ signaling at the first sensory synapse (39, 76, 174, 177, 178). Moreover, mitochondrial Ca²⁺ mediates signaling to the nucleus and the release of death signals, in case of very high Ca²⁺ (39, 75, 76, 174, 177, 179–181).

Inflammatory stimuli can lead to a rise in cytosolic Ca²⁺ in sensory neurons, which facilitates the release of neurotransmitters, excitability, and pain (39, 76, 177, 181). Studies showed that inflammatory cytokines increase spontaneous Ca²⁺ oscillations in organotypic spinal cord slices (182). Intraplantar carrageenan or CFA injection in rats increased intracellular Ca²⁺ in brain neurons or increased evoked Ca²⁺ transients in DRG neurons, respectively (72–74). These studies show that cytosolic Ca²⁺ is regulated by inflammation, yet direct evidence that mitochondrial Ca²⁺ is affected during inflammatory pain is lacking. Because mitochondrial and cytosolic Ca²⁺ are interdependent, the data may suggest a potential involvement of mitochondrial Ca²⁺ in inflammatory pain.

Nerve damage causes disturbed mitochondrial Ca²⁺ buffering (180). For example, spinal nerve ligation in rats reduced mitochondrial Ca²⁺ storage capacity in lumbar DRG neurons (75). Rats with chemotherapy-induced neuropathic pain have a decreased duration of depolarization-evoked Ca²⁺ transient, which is partially mediated by an augmented mitochondrial Ca²⁺ uptake and increased mitochondrial volume (78). In vitro studies showed that the chemotherapeutic drug paclitaxel induces the formation of mPTP and promotes the release of mitochondrial Ca²⁺, contributing to sensory neuron hyperexcitability and cell death (76). The chemotherapeutic drugs cisplatin and oxaliplatin increased cytosolic Ca²⁺ concentration and depolarization-evoked Ca²⁺ transients in cultured sensory neurons (71, 165). Taken together, different insults disturb distinctive aspects of mitochondrial Ca²⁺ buffering.

Does modulating mitochondrial Ca²⁺ buffering affect neuronal excitability and pain? The noxious heat-activated receptor TRPV1 conducts Ca²⁺ and Na⁺, producing a depolarizing receptor potential that activates nociceptors. Knockdown of NCLX in DRG neurons decreased mitochondrial Ca²⁺ release, reduced capsaicin-induced TRPV1 activation and neuronal firing (183), indicating that mitochondrial Ca²⁺ buffering is important for neurons' ability to respond to pungent reagents. Blocking MCU, with an intrathecal injection of a MCU inhibitor, prevents mitochondrial Ca²⁺ uptake in spinal cord neurons and diminished capsaicin-induced neuron hyperexcitability and pain in mice (184). Mechanistically, by regulating cytosolic and mitochondrial ionic transients, NLCX and MCU modulate Ca²⁺-dependent desensitization of TRPV1 channels, thereby controlling nociceptive signaling (183). Oxaliplatin increases cytosolic Ca²⁺ by increasing the expression of non-selective cation channel TRPV1 in DRG neurons, thus increasing sensory neurons' excitability. Silencing or blocking TRPV1 reduced both oxaliplatin and paclitaxel induced neuropathic pain (185–188), which is indirect evidence that mitochondrial Ca²⁺ buffering is also involved in chemotherapy-induced neuropathic pain. A role for MCU in mechanical allodynia has been found in mice with painful diabetic neuropathy. Selective knockdown of the MCU in Nav1.8-positive DRG neurons resolved mechanical allodynia in diabetic mice (79). Likewise, blocking mPTP through intraperitoneal injection of cyclosporine A reversed spared nerve injury-induced allodynia in rats, by reducing Cytochrome C release and loss in activity of spinal cord GABAergic inhibitory neurons (76, 77).

Mitochondrial Ca²⁺ buffering also plays a role in spinal synaptic plasticity (184). Inhibition of spinal mitochondrial Ca²⁺ uptake in mice, using different pharmacological strategies, blocked N-methyl D-aspartate (NMDA)-induced activation of downstream protein kinases that mediate spinal synaptic plasticity, and reduced induction of long term potentiation, a process that increases synaptic strength in the dorsal horn of the spinal cord and contributes to chronic pain. Importantly, inhibition of mitochondrial Ca²⁺ uptake in the spinal cord prevented animals from developing mechanical hyperalgesia in response to intrathecal NMDA or intradermal capsaicin injection (184).

Inflammasomes are intracellular multiproteic complexes, composed of a sensor protein that oligomerizes, in order to recruit caspase-1. MtROS production is a main trigger of nucleotide-binding oligomerization domain-like receptor pyrin domain containing 3 (NLRP3) inflammasome activation (189, 190). MtROS causes translocation of the inner mitochondrial membrane protein cardiolipin to the outer mitochondrial membrane, where cardiolipin serves as a docking place for caspase-1 and NLRP3. Subsequently, the adaptor apoptosis-associated speck-like protein containing a CARD (ASC) binds both NLRP3 and caspase-1, activating the inflammasome to initiate caspase-1-mediated cleavage of pro-IL-1β and pro-IL-18 into their active mature form (189). Mitochondria can also sustain NLRP3 inflammasome activation via generation of ATP (191) and disturbed Ca²⁺ signaling (189, 190). While NLRP3 inflammasome activation is classically thought to occur in immune cells, such as macrophages, it has also been detected in neurons (97, 192). Moreover, NLRP3 expression is increased in the DRG neurons of patients with neuropathic pain in comparison with controls with no pain (121). Neuronal NLRP3 inflammasome activation likely contributes to neuro-inflammation and neurodegeneration, as it has been detected in brain neurons in Parkinson's and Huntington's disease (193–195). It is not known whether NLRP3 inflammasome activation interferes directly with neuronal functioning.

In the context of rheumatic diseases, mitochondrial dysfunction (e.g., increased mtROS, disturbed OxPhos or Ca²⁺ uptake) may trigger NLRP3 inflammasome activation in the nervous system, affecting sensory processing. For example, in the CFA-induced chronic inflammatory pain, the compound muscone, which diminishes ROS production, prevented the loss of mitochondrial membrane potential and Ca²⁺ influx. Moreover, it blocked the CFA-induced increase in spinal cord NLRP3 and IL-1β expression and hyperalgesia (87, 196). Importantly, the downstream product of inflammasome activation, IL-1β, induces firing of nociceptors and mediates CFA-induced pain (189, 197, 198), suggesting a putative role for NLRP3 inflammasome activation in inflammatory pain.

Although studies about the role of NLRP3 inflammasome in the nervous system in inflammatory pain models are still scarce, various findings suggest the involvement of this pathway in other types of pain. Intraperitoneal or intrathecal administration of MCC950, a specific NLRP3 inflammasome inhibitor, alleviated mechanical allodynia, and decreased IL-1β and IL-18 release in the lumbar dorsal spinal cord in cancer-induced bone pain and oxaliplatin-induced neuropathy, respectively (97, 98). In the cancer-induced bone pain model, IL-1β and IL-18 were detected predominantly in neurons (97). Suppressing NLRP3 inflammasome activation with different microRNAs, reduced NLRP3 inflammasome activation in the sciatic nerve, spinal astrocytes and microglia, while concurrently attenuating mechanical allodynia after chronic constriction injury in mice and rats (92–96). The chemotherapeutic drug bortezomib, which disrupts mitochondrial Ca²⁺ buffering and promotes mtROS production in neurons, increased NLRP3 expression in the DRG (138). Silencing NLRP3 in the DRG prevented the development of bortezomib-induced neuropathic pain in mice and rats (99). Treatment with a ROS scavenger decreased paclitaxel-induced mechanical allodynia and reduced NLRP3 expression in DRG macrophages and caspase 1/IL-1β activation in lumbar DRG and sciatic nerve (100). These studies indicate that nerve damage and inflammation may lead to inflammasome activation in the nervous system, although the majority of these studies did not identify in which cells the inflammasome was activated. Given that nerve damage and inflammation promote mtROS production in neurons, it is possible that mtROS triggers inflammasome activation in neurons and contributes to pain.

Mitochondria are highly dynamic organelles and have several quality control pathways in order to maintain their integrity. A controlled balance between mitophagy and mitochondrial biogenesis guarantees optimal mitochondrial turnover. Mitochondrial dynamics involve continuous fission and fusion forming a dynamic network to maintain their content, morphology and quality (145, 199).

Mitochondrial biogenesis occurs close to the nucleus, since the majority of mitochondrial proteins are encoded by the nuclear genome. In neurons, mitochondria are mainly made in the soma, therefore mitochondria have to be transported to the dendrites and axonal terminals (230). Some studies identified that mitochondrial biogenesis also occurs to some extent in the axons, but the mechanisms are not completely clear (200). Motor proteins from the kinesin family mediate anterograde transport of new mitochondria from the soma to the axons, whilst motor proteins that form a dynein-dynactin complex mediate retrograde transport towards the soma (199, 230). Motor proteins interact with the microtubule network and mitochondrial Rho GTPases (MIROs) to regulate mitochondrial motility for both anterograde and retrograde transport (231).

Neuron-specific knockdown of Miro1 in mice depletes mitochondria in dendrites and diminishes neuron survival, highlighting the importance of mitochondrial motility for neuronal homeostasis (232). Impairing mitochondrial transport by decreasing the activity of motor proteins in cervical ganglion neurons diminished the number of axonal and dendritic mitochondria, which was associated with synaptic dysfunction that limited neurotransmission and neuroplasticity (231). Restoring the diminished local ATP pool that was affected by the impaired transport, reestablished these parameters, indicating that mitochondrial transport is essential to maintain sufficient ATP levels throughout neurons (233). Enhancing mitochondrial transport in adult murine DRG or cortical neurons by knockdown of syntaphilin, an anchoring protein that keeps mitochondria static, improved axonal regrowth and normalized cellular ATP/ADP ratio after axotomy (233, 234). Inflammation in the peripheral nervous system, induced by experimental autoimmune neuritis, reduced retro- and anterograde transport of mitochondria in nerves (126), indicating that inflammation reduces mitochondrial transport. Similarly, exposure of murine brain slices to lipopolysacharide (LPS) reduced retrograde transport (235). Thus, efficient mitochondrial transport is a requirement for effective energy production and synaptic transmission in neurons. Inflammation can reduce this mitochondrial transport through axons, thereby affecting the local ATP pool.

Histone deacetylase 6 (HDAC6) deacetylates the cytoskeletal protein α-tubulin, limiting mitochondrial transport and distribution in neurons. Systemic administration of HDAC6 inhibitors decreased inflammation-induced mechanical hypersensitivity in CFA- and collagen-induced arthritis models (236, 237), and neuropathic pain models (236, 238). Pharmacological inhibition of HDAC6 also decreased LPS-induced neuronal loss and pro-inflammatory cytokines production in the brain, demonstrating a neuroprotective effect of improving mitochondrial transport (239, 240). Cognitive impairment and neuro-inflammation (microgliosis, production of TNF and IL-1β) in the spinal cord/brain in neuropathic pain models are reversed by HDAC6 inhibition (236, 238). In line with the previous results, oral administration of HDAC6 inhibitors, Hdac6 knockout, or cell-specific deletion of Hdac6 in advillin-positive sensory neurons, promoted mitochondrial transport in sensory neurons, attenuated mechanical hyperalgesia, and spontaneous pain in chemotherapy-induced neuropathy (85, 86). HDAC6 inhibition reestablished OCR in lumbar DRG neurons and the tibial nerve of cisplatin-treated mice to the values of control animals (85, 86), indicating mitochondrial transport in neurons is essential to ensure sufficient OxPhos when damaged by neurotoxic agents. Currently, oral administration of ACY-1215 (Ricolinostat, HDAC6 inhibitor) is being assessed in patients suffering from painful diabetic peripheral neuropathy (clinical trial NCT03176472). A selective HDAC6 inhibitor reduced the inflammatory phenotype of immune cells from RA patients in vitro and pre-clinical studies show an analgesic effect of this approach in RA models (236, 237, 241). Thus facilitating mitochondrial transport through HDAC6 inhibitors may be a promising approach to relieve rheumatic pain.

In several models of rheumatic diseases and in rheumatic patients, mitochondrial respiration is reduced in cells of the joint. As an example, chondrocytes of rabbits and guinea pigs with experimental OA, have an impaired mitochondrial respiration, reduced intracellular ATP levels, and a rise in the lactate/pyruvate ratio (67, 68). These differences worsened during the progression of cartilage damage (68). Similarly, chondrocytes collected from OA patients have decreased activity of mETC complexes and reduced ATP levels compared to control patients, which correlated with an increase in the number of apoptotic chondrocytes (101–103). Apoptotic chondrocytes release substances that can activate receptors on sensory neurons and elicit pain (244). Increased formation of oxidative metabolites (e.g., H₂O₂) and decreased expression of endogenous antioxidant molecules (e.g., SOD2, peroxidase) are found in the knee (synovium and cartilage) in pre-clinical rheumatic disease models (112, 245), and in humans with OA (112–116), RA (51, 115, 117) and PsA (118). In RA, an imbalance between oxidants and antioxidants leads to mitochondrial oxidative stress in synoviocytes, which triggers activation of transcription factor NF-κB and subsequent release of inflammatory mediators, such as IL-8 or PGE₂ (246), that can activate receptors on sensory neurons to induce pain (247, 248). MtROS is important in this process, because mitoTEMPOL, a mitochondrial specific ROS scavenger, reduced the release of inflammatory mediators by human synoviocytes (246).

Oxidative stress can also induce apoptosis of chondrocytes via reduction in the expression of FOXO3, a transcription factor upstream of various antioxidant genes, such as SOD2 (248, 249). Intra-articular or oral treatment with antioxidants, such as amobarbital, N-acetylcysteine (NAC), resveratrol or methylene blue have analgesic effects in different species (mice, rat, porcine) with OA (250, 251) and CFA-induced joint inflammation (160). Similarly, a specific mitochondrial targeted-antioxidant, plastoquinonyl-decyl-triphenylphosphonium bromide (SkQ1), reduced bone destruction and cartilage damage in OA mice (252). Chondrocytes damage may lead to the release of neurotrophin NGF by immune cells, which activates its high-affinity receptor TrkA on neurons and induces expression of a variety of ion channels (e.g., TRPV1, voltage gated sodium channels) and pain (253, 254).

Increased ROS production is a feature of senescence cells, i.e., cells that have entered a non-proliferative state. Cellular senescence is accompanied by a distinct secretory phenotype, senescence-associated secretory phenotype (SASP), which includes a variety of secreted proteins, cytokines and chemokines that can drive chronic inflammation (255, 256). Recently, cellular senescence has been identified in nervous tissue in mice with long-lasting (>4 months) neuropathic pain. Intriguingly, p53-mediated senescence drove neuropathic pain specifically in male, but not female mice. Moreover, a mutation in the P53 gene is associated with chronic pain in men (256). In mice with experimental OA, the number of senescent cells in the cartilage is increased. Selective elimination of these cells, either genetically or pharmacologically by intra-articular administration of a senolytic, reduced pain-associated behavior and enhanced cartilage reconstruction (257). Similarly, blocking p53-mediated senescence or removing senescent cells reversed neuropathic pain, by reducing the expression of SASP effectors (e.g., p53, IL-1β, IL-6) (256, 258). These results indicate a potential role of mitochondria-driven cellular senescence in pain.

NLRP3 inflammasome activation has been linked to amplified inflammatory cytokine signaling in senescent cells (259, 260). The expression of NLRP3 inflammasome related molecules (NLRP3, ASC, caspase-1, IL-1β, IL-18) is increased in the synovial tissue of collagen-induced arthritis (CIA) mice (88, 192, 261) and MIA-induced OA rats (89, 90). Intraperitoneal administration of the specific NLRP3 inhibitor MCC950, or blocking IL-18, diminished synovial inflammation and cartilage damage in RA. Moreover, pharmacological inhibition of the inflammasome products IL-1β or IL-18 attenuates RA or neuropathic pain (262, 263). Additionally, administration of NLRP3 inflammasome inhibitors reversed NLRP3 activation in FLSs and relieved OA-induced pain (89, 90). Moreover, indirect regulation of the inflammasome, by downregulating microRNA miR-30b-5p in the joint, attenuated the increase in NLRP3, ASC, and cleaved caspase-1 in joints, and decreased cartilage damage and pain in OA rats (91). In the joint tissue of OA patients, miR-30b-5p is elevated compared to healthy controls and its expression correlates with disease severity and levels of IL-1β in the synovial fluid (91), suggesting a putative role of inflammasome activation in OA patients. However, it remains unclear if inflammasome activation also mediates pain in this rheumatic disease in humans, as pain-associated outcomes were not investigated in this study.

Interestingly, the expression of translocator protein (TSPO), found in the outer mitochondrial membrane, is increased in the synovial membrane of OA patients and its expression levels correlated with lower pain intensity (104). Although TSPO has been commonly used as a target for PET imaging to assess neuro-inflammation and microglia/macrophage activation (264), it is a mitochondrial protein that has also been linked to mitochondrial bioenergetics. Knockout of TSPO in human microglia cell lines reduced OxPhos and ATP production. In addition, TSPO ligands in in vivo models prevented mitochondrial membrane depolarization, mPTP formation and NLRP3 inflammasome activation (265, 266), suggesting TSPO activity may indirectly affect some mitochondrial functions. Thus, the TSPO correlation with lower pain intensity may potentially link to overall better mitochondrial performance.

Mitochondrial quality control mechanisms are essential to maintain cellular homeostasis and prevent the production of inflammatory cytokines. Intra-articular MIA administration, to induce experimental OA in rats, increased the expression of PINK1 and Parkin in knee cartilage. Likewise, treatment of human chondrocytes with MIA increases PINK1 expression in these cells. In OA patients, PINK1 expression in chondrocytes in damaged parts of joint cartilage is higher compared to healthy cartilage (44). These data point to exacerbated PINK-1-mediated mitophagy in chondrocytes, which may result in cartilage damage and pain. Several studies also described possible disturbances in mitochondrial biogenesis in chondrocytes of rodents or OA patients, since AMPK activity, SIRT-1, PGC-1α, TFAM, and NRF-2 expression are reduced compared to controls (81, 82). In CIA-induced arthritis, intraperitoneal administration of mitochondrial fission inhibitor mdivi-1 reduced arthritis scores and paw thickness in mice (120). Moreover, DRP1, a protein that promotes fission, is increased in synovial tissues and FLSs from RA patients and in primary human chondrocytes following MIA treatment (120). In RA, DRP1 expression correlates with disease severity (44). Finally, neurotrophins, such as NGF or BDNF, are released in the synovial fluid of RA, OA and AS patients and can induce mitochondrial fission in neurons (267–269). These data suggest that mitochondrial fission is enhanced in various rheumatic diseases and that inhibiting fission reduces arthritis. Whether changes in fission also contribute to pain remains to be explored.

In conclusion, several mitochondrial functions are clearly affected in the local tissue in rheumatic diseases. It is not fully understood how these mitochondrial alterations contribute to rheumatic pain. We propose that changes in mitochondrial functions in the site of injury may promote the release of inflammatory mediators that sensitize or activate the nervous system, contributing to pain.

Given that inflammation reduces mitochondrial respiration in sensory neurons and diminished OxPhos is associated with prolonged pain, boosting mitochondrial respiration may resolve pain. DCA boosts the TCA cycle and OxPhos. DCA reduced pain-associated behaviors in pre-clinical inflammatory pain models and showed a safe profile when tested in humans (65, 130, 279). DCA reduced lactate levels in endometriosis cells, and is currently being tested as treatment for endometriosis-associated pain (clinical trial NCT04046081) (279).

Altered mitochondrial respiration can lead to unfavorable cellular redox balance, e.g., NAD/NADH ratio. NAD⁺ is an essential cofactor in the regulation of mitochondrial health. A decline in NAD⁺ in various tissues and cells, including neurons, is associated with pathologies, such as neurodegenerative diseases, diabetic-induced and chemotherapy-induced neuropathic pain (280–283). Recently, preliminary data showed that peripheral inflammation reduces NAD⁺ levels in DRG of mice. Systemic or intrathecal administration of nicotinamide riboside, a precursor of NAD⁺, reversed CFA-induced inflammatory pain (64). Importantly, NAD⁺ supplementation with NAD⁺ precursors, such as nicotinamide riboside, mitigates oxidative stress and improves mitochondrial functions. More than 70 years ago, studies already showed that NAD⁺ supplementation reduces pain in RA patients (284, 285). More recently, oral NAD⁺ supplementation reversed diabetic and chemotherapy-induced neuropathic pain in pre-clinical models (281, 283). Other recent studies also showed that NAD⁺ supplementation has favorable outcomes on RA and SLE, but pain was not assessed in these studies (286). NAD⁺ supplementation also improved the global impact of OA, possibly through the increase of NAD⁺ in the synovial fluid and cartilage matrix, ensuring proper energy levels for cartilage repair. However, in this study NAD⁺ supplementation did not affect pain (287). Recent clinical trials in larger cohorts and with more extensive assessment exams have confirmed that oral supplementation with NAD⁺ precursors (including nicotinamide riboside) are safe and well-tolerated (288, 289). Future larger studies will have to show whether NAD⁺ supplementation may hold promise for treatment of inflammation and pain in patients with rheumatic disease.

Antioxidants are promising in reversing oxidative stress in diseases of the nervous system (290, 291). The antioxidant resveratrol is effective in managing rheumatoid arthritis by decreasing plasma inflammatory markers and joint swelling, when given in combination with common anti-rheumatic drugs (117). Similarly, resveratrol administered as adjuvant with the anti-rheumatic drug meloxicam decreased serum levels of inflammatory mediators (TNF, IL-1β, IL-6) and reduced pain by ∼70%, compared to OA patients that received meloxicam together with placebo (292, 293). Oral resveratrol also attenuated chronic musculoskeletal pain by 20% in postmenopausal women (294), indicating a potential analgesic effect of resveratrol in several pain conditions. Currently studies are further exploring resveratrol's analgesic effect in knee OA patients as combined (anti-inflammatory drugs and/or analgesics) and as mono-therapy (clinical trial NCT02905799) (295).

Pre-clinical studies show that targeting inflammasome-related proteins has analgesic effects (92–98). In a clinical trial, oral treatment with Dapansutrile, a specific NLRP3 inflammasome inhibitor, reduced joint pain 56%–68% and reduced joint and systemic inflammation in gout flare patients (296). Other drugs that target the inflammasome pathway, such as the caspase-1 inhibitor Pralnacasan, the anti-IL18 monoclonal antibody GSK1070806, and compounds targeting IL-1β or its receptor (e.g., anakinra and canakinumab), are being clinically evaluated or already approved to treat inflammatory diseases, including rheumatic disease. Anakinra and canakinumab reduce pain in RA and JIA, but they still need to be assessed in other rheumatic conditions. The ability of GSK1070806 to reduce pain has not been studied (clinical trial NCT03681067) (297–300).

Mitochondria are transferred between cells via extracellular vesicles, tunneling nanotubes, gap junctions or free mitochondrial ejection (42, 301–308). Intercellular transfer of healthy mitochondria can improve the energy production status of recipient cells and restore their viability. The release of damaged mitochondria might function as a help request to surrounding cells, triggering signaling pathways to restore homeostasis in the cell releasing damaged mitochondria (42, 301–304). Mitochondrial transfer has been observed in a growing list of pathologies that include brain injury (42, 309), neurodegenerative diseases (306, 310), cancer (303, 311, 312), and cardiomyopathy (304). Mitochondria can be transferred from human synovial mesenchymal stem cells (sMSCs) to Th17 cells, which reduces the production of the pro-inflammatory cytokine IL-17. Mitochondrial transfer between sMSCs and Th17 is impaired in RA patients (313). Likewise, sMSCs donate mitochondria to stressed articular chondrocytes and failure of this transfer has been hypothesized to impair healing of orthopedic tissues (314). Importantly, transfer of mitochondria from macrophages to DRG neurons is required to resolve inflammatory pain (25). These findings highlight that intercellular mitochondrial transfer may be an approach to reduce inflammatory pain.

Some research groups have explored the delivery of entire functional mitochondria into damaged cells, a technique called mitochondrial transplantation (315, 316). Pre-clinical studies show mitochondrial transplantation is neuroprotective (317, 318). Additionally, a small clinical trial with 5 patients showed that autologous mitochondrial transplantation (i.e., mitochondria from non-ischemic skeletal muscle of each patient were injected in the myocardium) improved myocardial function in patients with ischemia-reperfusion injury, without inducing short-term complications (319). In all these studies, “naked”/free mitochondria were directly injected in the tissue of interest. Classically, some free mitochondria components are thought to act as DAMPs, thus triggering an inflammatory response (317). However, “naked” mitochondria are found in the human blood, suggesting that healthy free mitochondria are not necessarily inflammatory (320). Nevertheless, to prevent any adverse effect of free mitochondria, packaging mitochondria into vesicles would prevent them from acting as DAMPs, whilst it would allow specific targeting strategies by adjusting lipid composition. Synaptosomes are lipidic membranous particles released at nerve terminals. Synaptosomes can be obtained by homogenization and gradient centrifugation of nervous tissue (synaptic terminals), and contain several synaptic vesicles and mitochondria. Synaptosomes are specifically taken up by neuronal cells, making them an ideal delivery system to target neurons (316).

In summary, targeting mitochondria shows some beneficial effects in multiple pathologies and could be a novel approach to treat chronic pain. The challenge is to develop a strategy to deliver a drug to mitochondria in specific cell types. Possibly, encapsulation of drugs/mitochondria into nanolipidic carriers may overcome these challenges (321–324) and allow the use of mitochondria/mitochondria targeting drugs to treat pain. As an example, in a pre-clinical study, extracellular vesicle containing mitochondria resolved persistent inflammatory pain, albeit transiently (25).