Osteoarthritis: Molecular causes and possible treatments

TRP channels are key players in pain perception, as they transduce noxious stimuli into pain signals at the joint. These channels, along with the regulation of NGF, represent promising targets for therapeutic intervention in OA pain.

The TRPV1 channel is the most extensively studied TRP channel. Several small molecules that act as TRPV1 antagonists have been identified and are expected to form the basis for a new class of pain-relieving drugs. Notable compounds include AMG517, MK2295, and AZD1386.^205,206 However, systemic administration of these antagonists in preclinical studies has been associated with adverse effects, such as hyperthermia or burn-like injuries, due to the loss of the sensation of noxious heat.^206,207 The mechanism underlying these side effects is believed to involve proton activation of TRPV1 (at least in rat models),²⁰⁶ where antagonists blocking proton activation lead to hyperthermia, whereas compounds enhancing proton sensitivity can cause hypothermia.²⁰⁸ In response, efforts have been made to develop TRPV1 antagonists that retain analgesic effects while avoiding interference with proton activation. The second generation of these antagonists has shown promising results in clinical trials. For example, SB-705489 (GlaxoSmithKline) reduced dental pain in rats and has completed phase 2 trials for dental pain, followed by a phase 2 rectal pain trial with no hyperthermia reported.^205,206 NEO6860 has demonstrated an analgesic effect without altering body temperature in knee OA patients,^162,209 and XEN-D0501, although causing a slight increase in body temperature in healthy individuals, has been approved by the FDA for treating erythromelalgia.^208,210 These findings suggest that TRPV1 antagonists could offer new insights into pain management for OA. However, caution is needed, as some TRPV1 antagonists may exhibit varying effects across species (e.g., JYL1421 causes hypothermia in rats but hyperthermia in dogs and cynomolgus monkeys),^206,208 underscoring the need for further research to clarify their therapeutic potential in OA pain relief.

Interestingly, preclinical studies have shown that TRPV1 agonists may also exert analgesic effects, a phenomenon that contrasts with the actions of TRPV1 antagonists. Topical or site-specific capsaicin injections, which are potent TRPV1 agonists derived from chili peppers, alleviate OA pain.^205,211 This paradox can be explained by prolonged or repeated exposure to capsaicin causing a large influx of calcium ions into nociceptors, activating calcineurin, a protein phosphatase that dephosphorylates TRPV1. This results in the desensitization of TRPV1-expressing nociceptors.^212,213 However, the clinical application of capsaicin is limited due to its initial pain-inducing effects at the administration site, rapid metabolism (with a half-life of 1.6 h), and side effects such as hypothermia, pruritus, erythema, and papules.^205,213

To overcome these limitations, efforts have focused on developing long-lasting, tolerable TRPV1 agonists that avoid the adverse effects of capsaicin. Resiniferatoxin (RTX), an ultrapotent capsaicin analog, can reversibly desensitize nociceptors for several months with a single low-dose injection and produces similar effects upon second administration. RTX may also provide permanent analgesia through irreversible ablation of nociceptors and is currently undergoing preclinical testing.^205,214,215 CNTX-4975, a synthetic form of trans-capsaicin, considerably reduced joint pain in OA patients after a single 1.0 mg intra-articular injection, with effects lasting up to 24 weeks. In a phase 2 multicenter, double-blind trial, treatment-emergent adverse events were comparable to those of the placebo.²¹² Zucapsaicin, the cis-isomer of capsaicin, was approved in 2010 as a topical analgesic for severe knee pain in OA patients and is associated with fewer side effects than traditional capsaicin.²¹⁶ Other TRPV1 agonists, including NE19550, MRD-652, and CPIPC, have shown promising analgesic effects in animal models,^205,213,216 but their clinical efficacy in human OA remains to be fully established.

The TRPA1 channel is commonly characterized as a nonselective cation channel that detects noxious cold, mechanical stimuli, and irritant chemicals.^217,218 While TRPA1 activation by noxious cold in healthy tissues is debatable, its role in cold hypersensitivity following tissue injury is well supported by multiple studies.¹⁶¹ In various OA rat models induced by complete Freund adjuvant (CFA), carrageenan, MIA, and monosodium urate, direct activation of TRPA1 by ROS, IL-1β, and TNF-α induces both mechanical and cold hypersensitivity. Notably, genetic deletion or pharmacological blockade of TRPA1 channels reverses allodynia without inducing the thermosensory or thermoregulatory disturbances typically observed with TRPV1 antagonism.^161,218 These findings highlight the potential of TRPA1 antagonists as novel analgesics for OA pain management. Several promising TRPA1 antagonists, including HC-030031, AP-18, and A-967079, have been shown to reduce mechanical and cold hypersensitivity in animal models of CFA-induced inflammatory pain.^217,219 However, a recent study reported conflicting results, in which systemic administration of HC-030031 did not alleviate ongoing pain in MIA-induced OA rats.²²⁰ This discrepancy may arise from the significant difference in TRPA1 expression between species, as TRPA1 is expressed at a much higher percentage (~80%) in human DRG neurons than in mouse DRG neurons, suggesting that the analgesic potency of TRPA1 antagonists may be underestimated in preclinical studies.^217,219 Further clinical research is needed to elucidate the therapeutic potential of TRPA1 antagonists in OA pain management.

However, some TRPA1 agonists function similarly to TRPV1 agonists, desensitizing TRPA1-expressing neurons to produce analgesia. For example, systemic administration of atractylodin (ATR), a TRPA1 agonist, leads to sustained activation of TRPA1 (>1 h), which mitigates nocifensive behaviors induced by another TRPA1 agonist, allyl isothiocyanate (AITC).²²¹ Additionally, intravenous injection of Ms 9a-1 markedly enhances TRPA1 responsiveness to agonists, thereby exerting an antinociceptive effect on CFA-induced inflammatory pain in rats.²²² The mechanisms underlying the pro- or antinociceptive effects of certain TRPA1 agonists remain unclear. One hypothesis is that while pronociceptive TRPA1 agonists reversibly modulate TRPA1, antinociceptive agonists may irreversibly bind to cysteine residues near the cation-passing gate of TRPA1, thereby preventing channel reopening.²¹⁹ Despite the analgesic potency of TRPA1 agonists, many challenges remain unresolved. For example, the administration of TRPA1 agonists may cause initial pain that is intolerable to patients, and TRPA1 agonists often suffer from issues such as poor chemical properties, low selectivity, or weak efficacy. Additional studies may provide further prospects for the use of TRPA1 agonists in OA pain treatment.

TRPM8 is recognized as a sensor for both innocuous cool (15–25 °C) and noxious cold (<15 °C) temperatures and is activated by compounds that elicit cool sensations, such as menthol, icilin, and eucalyptol.^223,224 In contrast to the well-defined mechanisms of TRPV1 and TRPA1 channels in nociception, TRPM8 appears to play a bidirectional role in the pain pathway but exhibits a distinct anti- or pronociceptive pattern in specific pain conditions.²²³ For example, in the CFA model of inflammatory pain, mechanical and heat pain are substantially reduced following menthol injection, whereas pretreatment with TRPM8 antagonists or genetic deletion of TRPM8 abolishes this analgesic effect.^223,225 Inflammatory conditions, including low pH and “inflammatory soup,” inhibit TRPM8 activity, and the mRNA expression of TRPM8 is reduced in human OA chondrocytes.²²⁶ These findings suggest that activation of TRPM8 may attenuate mechanical and heat allodynia in patients with inflammatory pain.²²³ Additionally, given that most nociceptors express TRPV1, the lack of coexpression of TRPM8 with TRPV1 suggests that TRPM8 does not interfere with the perception of mechanical and heat stimuli but blocks the transmission of evoked action potentials to second-order neurons.^223,227

Conversely, TRPM8 activation can lead to cold-induced pain following injury and exacerbate cold hyperalgesia in patients with neuropathies.²²³ The application of high concentrations (30–40% wt/vol) of menthol has been shown to cause cold pain, cold allodynia, and cold hyperalgesia in human trials.^228,229 These data suggest that patients with mechanical or heat pain as their primary symptom may be better treated with TRPM8 agonists at a low concentration, whereas individuals with cold-hyperalgesia may respond better to TRPM8 antagonists. Therefore, TRPM8 agonists are likely to be preferable therapeutic options for OA pain.

However, the use of TRPM8 agonists in OA pain treatment has rarely been reported. In a study involving 20 OA patients, the topical application of 3.5% menthol gel to OA knee joints resulted in no significant improvements in pain behavior compared with placebo.²³⁰ Additionally, coculture of menthol with OA chondrocytes increased the expression of MMPs and proinflammatory mediators in vitro.²²⁶ These findings suggest that further studies are needed to confirm the potential analgesic effects of TRPM8 agonists in treating OA pain.

TRPM3 is also a nonselective cation-permeable channel that is predominantly expressed on nociceptors, where it mediates nocifensive responses to the endogenous neurosteroid pregnenolone sulfate (PS) and noxious heat.^207,231,232 Intraplantar injection of PS in mice elicits pain behaviors that are absent in TRPM3-deficient mice, suggesting that TRPM3 activation is critical for initiating pain sensation.^207,232 Several studies have demonstrated that TRPM3 expression in DRG neurons increases under inflammatory conditions and that administration of TRPM3 antagonists in rodents abolishes TRPM3-mediated pain without affecting core temperature. Additionally, these antagonists reduce the responsiveness of TRPV1 and TRPA1 channels to their respective agonists, thereby enhancing the analgesic effects of TRPM3 blockade.^232,233 Although TRPM3 antagonism may attenuate heat sensation, the perception of noxious heat remains intact as long as the heat sensor, TRPV1, remains functional,²¹⁸ underscoring the potential of TRPM3 antagonists as promising analgesics for OA pain treatment.

However, research on TRPM3 antagonists as analgesics is still in its infancy. Plant-derived TRPM3 antagonists, such as isosakuranetin, ononetin, and primidone, suffer from suboptimal pharmacokinetic properties and lack target selectivity, hindering their therapeutic applicability in humans.²⁰⁷ Further research is needed to address these challenges, potentially by encapsulating these antagonists in appropriate delivery systems. Unlike other TRP channels, which are typically tetramers with a central cation-permeable pore,²³⁴ recent findings suggest that TRPM3 comprises an alternative pore. This alternative pore, synergistically activated with the central pore, intensifies pain sensation.^218,231 This novel mechanism has sparked interest in targeting the alternative pore of TRPM3, and the concurrent blockade of both pores may enhance analgesia, offering a new avenue for OA pain management.

Blockade of Na_v channels represents another promising strategy for alleviating OA pain, as these channels contribute to pain hypersensitivity in OA patients by initiating action potential firing and transmitting pain signals through Na⁺ influx in nociceptors.²³⁵ Current Na_v blockers used for pain management, such as local anesthetics (e.g., lidocaine), class I antiarrhythmics (e.g., mexiletine), antiepileptics (e.g., carbamazepine), and antidepressants (e.g., TCAs), have demonstrated significant efficacy in pain relief but also cause severe adverse effects, such as dizziness, sedation, convulsions, and cardiotoxicity, as these drugs nonselectively act on multiple Na_v subtypes related to motor and neurological functions,^236,237 highlighting the need for selective Na_v blockers to minimize off-target effects.

Na_v subtypes, including Na_v1.1 through Na_v1.9, are expressed in various tissues and play pivotal roles in multiple physiological functions, necessitating careful consideration of potential side effects before specific subtypes are targeted.²³⁷ Blockade of Na_v1.1, Na_v1.2, and Na_v1.3 channels, which are expressed primarily in the central nervous system (CNS), poses a risk of altered mental status or seizures.²³⁶ In contrast, the inhibition of Na_v1.4 and Na_v1.5 has been associated with muscular and cardiac channelopathies, respectively.²³⁸ Na_v1.6 is expressed mainly in motor neurons, and blockade of this isoform leads to paresthesia and weakness.²³⁹ Na_v1.9 is expressed in nociceptors and myenteric neurons in the gastrointestinal tract. Research on Na_v1.9-specific blockers has been limited due to the rarity of clinical cases involving Na_v1.9 mutations as well as challenges in developing Na_v1.9-expressing cell models, which could explain the lack of promising Na_v1.9-specific blockers despite clear evidence of Na_v1.9 participation in pain signaling.^236,240

The potential analgesic effects of Na_v channel blockers are often outweighed by their severe side effects. In contrast, specific inhibitors of Na_v1.7 and Na_v1.8 have shown the most promising results in pain treatment, with manageable adverse effects.²³⁶ Blockade of Na_v1.7 results in manageable anosmia, whereas inhibition of Na_v1.8 is associated with a risk of visual disturbances.^241,242 Studies in OA mouse models have revealed upregulated expression of Nav1.7 and Nav1.8 in nociceptors innervating the knee joint, which is correlated with the pain hypersensitivity observed in these models.^243,244 Several Nav1.7- and Nav1.8-specific blockers have been shown to attenuate secondary allodynia and improve motor function in murine OA models. For example, PF-04856264, a selective Na_v1.7 blocker, substantially alleviates OA pain and upregulates type II collagen levels following intra-articular injection in a murine meniscectomy model, with similar results observed in MIA-induced OA rat models through oral administration.²⁴⁵ ProTx II, a tarantula toxin that selectively blocks Na_v1.7, demonstrated no analgesic efficacy following intravenous or intrathecal injection but was found to upregulate the expression of type II collagen and aggrecan while reducing MMP13 and ADAMTS5 levels.^245–247 A recently designed peptide, PTx2-3127, which is based on ProTx II, has shown efficacy in chronic and thermal pain in rat models via intrathecal administration, positioning it as a prospective novel analgesic.²⁴⁸

However, intra-articular injection of A-803467, a potent and selective Na_v1.8 blocker, considerably reduced hindlimb disability and secondary allodynia in MIA-induced OA model rats.^247,249 While specific Na_v blockers have shown promising analgesic effects in animal models of skeletal degenerative diseases, the pharmacological and toxicological profiles of these blockers in humans remain largely unknown, necessitating further studies before their clinical application.

NGF promotes the expression of the TRPV1 channel, Na_v1.8 channel, CGRP, and SP, thereby enhancing pain sensation in OA patients. The inhibition of NGF signaling has emerged as a promising strategy for pain relief, with several clinical trials demonstrating that NGF antagonists outperform NSAIDs in terms of efficacy while avoiding severe side effects. These findings suggest that NGF is a prime target for the development of novel analgesics for OA pain management.^{168,169,173,250}

Several strategies have been employed to inhibit NGF signaling, including (i) agents that capture free NGF to prevent its interaction with its receptor TrkA, (ii) agents that directly occupy the NGF-binding site, and (iii) agents that block TrkA receptors.^168,172 NGF-capturing agents, such as humanized monoclonal antibodies, have shown remarkable specificity and potency in neutralizing NGF activity by directly binding to NGF.^168,169,172 Notable examples include tanezumab (RN624), fasinumab (REGN475), and fulranumab, all of which have proven effective in alleviating OA pain. For example, tanezumab demonstrated superior analgesic effects to NSAIDs in treating hip or knee OA, although it was associated with a higher incidence of adverse events in a phase III trial conducted by Pfizer and Eli Lilly in 2019.²⁵¹ Similarly, fasinumab and fulranumab considerably reduce pain and improve physical function in OA patients, with adverse events comparable to those of placebo in two phase II, randomized, double-blind, placebo-controlled trials.^252,253

Additionally, several molecules that occupy NGF-binding sites and disrupt the interaction between NGF and its receptors, including ALE0540, PD90780, Ro 08-2750, Y1036, and PQC 083, have been investigated.^172,173 Notably, systemic or spinal intrathecal administration of ALE0540 has demonstrated antiallodynic efficacy in animal models of neuropathic pain and thermally induced inflammatory pain.²⁵⁴ However, the analgesic effects of other molecules in this class remain to be fully elucidated.

The analgesic mechanisms of both NGF-capture agents and NGF-binding site antagonists rely on blocking NGF activity. However, this blockade may inadvertently disrupt homeostatic cellular responses mediated by NGF-p75^NTR interactions, potentially complicating long-term use.¹⁷³ Consequently, selective inhibitors of TrkA, which is a high-affinity NGF receptor, may offer greater clinical benefits. Nonetheless, TrkA is widely expressed in cholinergic neurons of the CNS, raising concerns about potential side effects in the CNS. To mitigate these risks, the use of TrkA inhibitors should ideally be restricted to peripheral applications.²⁵⁵

Most TrkA inhibitors target the ATP-binding site, a highly conserved motif shared among Trk family members, which may inadvertently inhibit other Trk receptors and exacerbate the CNS or other adverse effects.^168,255,256 To address this, numerous small molecules with high selectivity for TrkA have been developed by targeting less conserved ligand-binding regions. An exemplary compound in this category is PF-06273340, a potent TrkA inhibitor with low cytotoxicity. PF-06273340 inhibits TrkA by 98% at a concentration of 1 μM and can be effluxed by P-glycoprotein (P-gp), a transporter that prevents its entry into the brain, thereby reducing the risk of side effects in the CNS.^255,256 In preclinical models, PF-06273340 has shown efficacy in alleviating pain in an ultraviolet burn-induced hyperalgesia model of inflammatory pain, suggesting that it is a promising candidate for clinical application in OA pain management.²⁵⁶

Another promising TrkA inhibitor is AR786, which has good oral bioavailability and low CNS penetration. Chronic oral administration of AR786 effectively prevented pain in MIA-induced OA model rats.²⁵⁷ As research into TrkA-selective inhibitors continues to expand, these compounds hold considerable promise for addressing unmet medical needs in OA pain treatment.

Neurotransmitters play a critical role in nociceptive transmission. Targeting their transport or receptor activity can reduce nociceptive signaling and offer analgesic benefits. SP is one such neurotransmitter involved in pain transmission, and its receptor, NK1R, has emerged as a potential therapeutic target. Despite the success of various NK1R antagonists in animal models, human trials have not demonstrated significant analgesic effects.^183,258 This discrepancy may be due to species differences or the limited impact of blocking a single SP signaling pathway amid the broader network of neurotransmitters involved in nociception.²⁵⁸

In addition to SP, CGRP is another important neurotransmitter in nociception. This 37-amino acid peptide binds to an atypical receptor composed of CLR, RAMP1, and RCP, which is located on both nociceptors and second-order neurons.¹⁸¹ CGRP sensitizes nociceptors and transmits pain signals via its receptor, suggesting that blocking CGRP receptors could offer analgesic potential. While CGRP receptor antagonists are well established in migraine treatment, their role in OA pain remains less explored. However, two studies have shown that CGRP receptor antagonists considerably reverse nociceptor sensitization and pain-related behaviors in OA animal models. For example, the subcutaneous injection of BIBN4096BS in MIA-induced OA rats provided morphine-like analgesia within 4 h.²⁵⁹ Intra-arterial injection of CGRP 8-37 reversed joint nociceptor mechanosensitization in both MIA-induced and meniscectomy OA models.²⁶⁰ These studies confirmed that CGRP receptor antagonists block receptors on joint nociceptors to prevent sensitization. However, whether they also inhibit nociceptive transmission by acting on second-order neurons remains to be clarified.

Glutamate is another central neurotransmitter involved in nociceptive transmission. Pharmacological blockade of glutamate receptors, including iGluRs and mGluRs, has demonstrated analgesic effects in both preclinical and clinical studies.^261,262 The iGluR family includes N-methyl-D-aspartate (NMDA) receptors, α-amino-3-hydroxy-5-methylisoxazole-4-propionic acid (AMPA) receptors, kainate (KA) receptors, and δ (GluD) receptors. GluD receptors differ in that they require the intracellular ligand D-serine for activation rather than glutamate, and their role in nociception is still under investigation,²⁶³ so they are not covered here.

Excluding GluD, all iGluRs are ligand-gated ion channels permeable to Ca²⁺, Na⁺, and K⁺, causing depolarization of second-order neurons upon activation.²⁶⁴ NMDARs are heteromeric complexes typically containing the NR1 subunit and one or more NR2A-D subunits.²⁶⁵ While most subunits are ubiquitously expressed in the CNS, NR2B-containing NMDA receptors are more selectively distributed in the forebrain and superficial dorsal horn of the spinal cord.^265,266 Traditional nonselective NMDA antagonists—such as ketamine, memantine, and MK-801—act on all NMDA subunits,²⁶⁷ but they can cause adverse effects such as neurotoxicity, motor impairment, learning impairment, and psychotomimetic symptoms at or near the antinociceptive dosage.^268,269 To avoid these side effects, selective antagonists of NR2B-containing NMDA receptors represent a novel therapeutic option that might be devoid of the most prominent side effects of nonselective NMDA antagonists while retaining analgesic efficacy.

Ifenprodil is a potent and selective NR2B-containing NMDA receptor antagonist. Spinal administration of ifenprodil alleviates mechanical allodynia and neuropathic pain without cognitive or psychotomimetic side effects.^265,266 Other newly developed NR2B-containing NMDA receptor antagonists, including CP101606↗, Ro256981, and CI1041, have shown antinociceptive effects without CNS side effects.²⁶⁶ However, the pharmacological properties of these drugs have never been tested in animal models of OA pain; more preclinical data will be necessary to assess their feasibility in relieving OA pain.

AMPA receptors, another iGluR subtype, are composed of GluR-A-D (also known as GluR1-4).²⁷⁰ GluR-B has an arginine residue at the pore-forming segment, which introduces additional positive charges into the pore that inhibit Ca²⁺ influx,²⁷¹ increasing its antinociceptive properties and limiting its potential as a drug target. Unlike GluR-C and GluR-D, Ca²⁺-permeable AMPA receptors are predominantly composed of homomeric GluR-A subunits at second-order neurons of the spinal dorsal horn,²⁷⁰ and several animal studies have demonstrated that the activation of GluR-A-containing AMPA receptors leads to long-lasting nociceptive hypersensitivity,^270,272 making GluR-A the most attractive target among other subunits in analgesia. However, selective antagonists for GluR-A remain elusive. Nonselective AMPA receptor antagonists, such as NBQX, have yielded mixed results in OA pain models. For example, in antigen-induced arthritis (AIA) rats, intra-articular injection of NBQX, an AMPA/KA receptor antagonist, reduced pain-related behaviors within 2 days,²⁷³ but it had no analgesic effect on posttraumatic OA rats.²⁷⁴ More data are needed to demonstrate the effectiveness of AMPA receptor antagonists in analgesia and the role of GluR-A antagonists in OA pain treatment.

Kainate receptors, the final members of the iGluR family, consist of five subunits, including GluK1-5 (formerly known as Glu5-7, KA1, and KA2).²⁷⁵ In contrast to the restricted localization of the NMDA receptor and AMPA receptor on postsynaptic neurons, the kainate receptor is distributed on both presynaptic and postsynaptic neurons at the spinal dorsal horn.^275,276 Kainate receptors on presynaptic neurons (including nociceptors) mediate the release of glutamate, whereas those on postsynaptic neurons (including second-order neurons) contribute to nociceptive transmission by binding to glutamate.²⁷⁷ Among the kainate receptor subunits, GluK1 is predominantly expressed in nociceptors within the DRG, and it appears to play a crucial role in nociceptive transmission, as the kainate receptor exhibits an absolute requirement for GluK1, for which kainate receptor-mediated currents are not detected in GluK1 knockout DRG neurons.^277,278 Recent studies have extensively explored the efficacy of several GluK1 antagonists in analgesia, and some have shown promising results in preclinical trials. MSVIII-19 (8,9-dideoxy-neodysiherbaine) is a potent GluK1 antagonist. An elegant experiment conducted by Chang-Shen Qiu and colleagues revealed that intrathecally administered MSVIII-19 reduced thermal hypersensitivity in animal models of inflammatory pain and both thermal hyperalgesia and mechanical hypersensitivity in a chronic construction injury model of neuropathic pain.²⁷⁵LY467711↗ and LY525327↗ are two orally effective prodrugs that are metabolized into two selective GluK1 antagonists called LY458545↗ and LY457691↗ in vivo. Compared with morphine and ibuprofen, the oral administration of these two prodrugs has shown preferable analgesic efficacy in three classical pain models, including formalin-induced late-phase paw-licking pain, carrageenan-induced thermal hyperalgesia, and capsaicin-induced mechanical hyperalgesia.²⁷⁹

Finally, mGluRs, another critical family of glutamate receptors, also play crucial roles in the pain pathway. mGluRs can be divided into three groups on the basis of amino acid homology: group I (mGluR1 and mGluR5), group II (mGluR2 and mGluR3), and group III (mGluR4, mGluR6, mGluR7, and mGluR8).²⁸⁰ Like iGluRs, mGluRs are involved in transmitting pain signals, but the functions of different groups vary on the basis of their distributions. Group I mGluRs are expressed in both pre- and postsynaptic neurons and initiate action potential firing in second-order neurons when activated, thus contributing to pain signal transmission. In contrast, group II and III mGluRs are located predominantly at presynaptic terminals of nociceptors, where they act as autoreceptors to suppress nociceptor excitability upon activation.^281,282 Given their distinct roles, targeting specific subtypes of mGluRs, such as blocking group I mGluRs or activating group II/III mGluRs, offers promising therapeutic strategies for OA pain relief. Numerous studies have demonstrated the analgesic potential of group I mGluR antagonists and group II/III mGluR agonists. The spinal administration of group I mGluRs such as LY367385↗ and AIDA has been shown to reverse thermal hyperalgesia in rats with knee inflammation without affecting basal perception.²⁸³ Activating group II mGluRs with the agonist APDC blocks nociceptor hyperexcitability without affecting basal membrane excitability.²⁸¹ Other group II mGluR agonists (LY379268↗, LY354740↗, LY389795↗, and LCCG1) have demonstrated analgesic properties in many inflammatory pain models (capsaicin-treated and formalin-treated models),²⁸⁰ whereas group III mGluR agonists, such as ACPT-I and PHCCC, inhibit nociceptive behaviors in rats subjected to formalin tests and mechanical hyperalgesia in various models of inflammatory pain (carrageenan-treated and monoarthritic rats).²⁸² These findings underscore the importance of mGluRs in nociceptive transmission and highlight their potential as therapeutic targets for OA pain treatment.

IL-1 exerts a variety of negative effects on the pathogenesis of OA, inducing both local and systemic inflammation. It upregulates MMPs and ADAMTS, inhibits the synthesis of the proteoglycans aggrecan and type II collagen, and disrupts the balance between osteoblasts and osteoclasts. These effects collectively lead to tissue degradation, bone erosion, and osteophyte formation. Given its centrality in OA pathology, IL-1 signaling represents a promising therapeutic target for mitigating disease progression. A deeper understanding of IL-1 signaling mechanisms is essential for identifying potential therapeutic interventions. On the basis of the involvement of IL-1 signaling in OA progression, therapeutic strategies targeting IL-1 can be categorized into three main approaches: (i) agents that antagonize IL-1RI, (ii) agents that neutralize circulating IL-1, and (iii) agents that inhibit the maturation of IL-1.

IL-1Ra exerts anti-inflammatory effects by competitively antagonizing IL-1 at IL-1RI, thereby blocking downstream proinflammatory signaling without receptor activation. This mechanism suggests therapeutic potential for OA, a hypothesis supported by epidemiological evidence demonstrating an inverse correlation between serum IL-1Ra concentrations and the radiographic severity of knee OA.^284,285 Despite its anti-inflammatory properties, the therapeutic application of IL-1Ra in OA remains limited by its short intra-articular half-life and suboptimal therapeutic efficacy. Current research focuses on bioengineering strategies to increase IL-1Ra pharmacokinetics. For example, Khaled et al. engineered a PLGA microsphere encapsulating IL-1Ra, demonstrating that intra-articular administration preserved the inherent cartilage-protective bioactivity of IL-1Ra while substantially enhancing disease-modifying efficacy in a rat model of ACLT-induced OA.²⁸⁶ Similarly, Shikhar et al. designed an IL-1Ra through cationic carrier-mediated modification, enabling electrostatic targeting to the negatively charged aggrecan-glycosaminoglycan matrix within cartilage. While unmodified IL-1Ra failed to attenuate IL-1-induced cartilage degradation over 16 days, the modified IL-1Ra considerably suppressed induced glycosaminoglycan loss and nitrite release concomitant with improvements in chondrocyte metabolism and viability.²⁸⁷

Despite promising preclinical efficacy, IL-1Ra protein-based therapies remain limited by the need for repeated administration. Gene therapy has emerged as a strategic alternative to circumvent this challenge, offering sustained IL-1Ra expression and novel therapeutic insights for OA management.²⁸⁸ Deng et al. developed a polyplex nanomicelle-based delivery system for mRNAs encoding IL-1Ra. Compared with the control, a single intra-articular injection of this construct in a rat model of temporomandibular joint OA alleviated pain and suppressed OA progression for 28 days, as evidenced by reduced pain-related behaviors and attenuated cartilage degradation.²⁸⁹ Alan et al. employed helper-dependent adenovirus (HDAd)-mediated intra-articular gene therapy to deliver the IL-1Ra-encoding gene, effectively preventing pathological progression in murine and equine OA models.²⁹⁰ Frisbie et al. designed an adenoviral vector expressing equine IL-1Ra DNA (Ad-EqIL-1Ra), which was administered via intra-articular injection in equine OA models. This intervention reduced joint pain-related lameness and pathological changes while exerting protective effects against proteoglycan loss. Notably, therapeutic efficacy was observed even when treatment was initiated 14 days post-OA induction, indicating potential utility in addressing established joint trauma.²⁸⁸

While these preclinical advancements in IL-1Ra delivery optimization underscore its therapeutic promise, clinical translation necessitates confronting inherent pharmacokinetic challenges. The translational gap between preclinical efficacy and clinical applicability becomes evident in human trials, where even mechanistically validated strategies face hurdles such as short half-lives and suboptimal biodistribution. For example, anakinra, a recombinant IL-1Ra, has been approved by the FDA for treating rheumatoid arthritis (RA).^56,64,291 Clinical trials have demonstrated its safety and tolerability, with only rare cases of anaphylaxis. However, owing to its relatively short half-life of 4–6 h, daily subcutaneous injections are required to maintain therapeutic efficacy, which can lead to frequent injection-site reactions.^49,50,64 AMG108, a humanized monoclonal antibody that targets IL-1RI, is still in clinical development for OA and RA treatment.⁵⁶ Phase II trials have shown a favorable safety profile, with intravenous or subcutaneous administration, but the clinical efficacy has been limited, with only modest improvements observed in RA patients and no significant effects in OA patients to date.^292,293

Canakinumab is a humanized monoclonal antibody that selectively neutralizes IL-1β, with no effect on IL-1α-driven inflammation.^56,291 With a long average half-life of 26.1 days,⁵¹ canakinumab is administered every 4–8 weeks via subcutaneous injection. It is approved for treating cryopyrin-associated periodic syndrome (CAPS) and systemic-onset juvenile idiopathic arthritis.^49,56,64 Rilonacept, a dimeric fusion protein comprising the Fc region of human IgG1 and the extracellular IL-1 binding site of IL-1RI and IL-1RAcP, can neutralize circulating IL-1α, IL-1β, and IL-1Ra.^51,56 The safety profile of rilonacept is still inconclusive due to limited clinical data, but it appears comparable to that of anakinra.^50,56 It requires weekly subcutaneous injections and has been approved by the FDA for treating CAPS, recurrent pericarditis, and IL-1Ra deficiency (DIRA). However, indiscriminate neutralization of IL-1Ra may hinder the therapeutic efficacy of rilonacept.^56,291 To optimize clinical outcomes, the modulation of rilonacept to preferentially neutralize IL-1α and IL-1β while preserving the affinity of IL-1Ra may be beneficial. Recent studies by Yusha Wang et al. revealed a de novo variant, p.Lys131Glu (K131E), in IL-1R1 that reduces its affinity for IL-1Ra. They further designed a novel IL-1-neutralizing agent, rilabnacept, upon the modulation of rilonacept with the K131E mutation, which showed promise in preclinical models of IL-1-driven diseases.²⁹¹

In addition to the agents already in clinical use, several other neutralizing IL-1 agents are currently under investigation. These include bermekimab, a naturally occurring monoclonal antibody that neutralizes IL-1α; gevokizumab, a humanized monoclonal antibody that targets IL-1β; and lutikizumab, a dual-targeting antibody that blocks both IL-1α and IL-1β.⁵⁶ These emerging therapies offer promising alternatives or adjuncts to existing treatments, potentially enhancing the ability to modulate the IL-1-mediated inflammatory pathways in OA. As clinical development advances, the efficacy and safety profiles of these agents will be key in determining their future role in managing IL-1-mediated diseases, including OA.

VX-740 (also known as pralnacasan) is an orally active, reversible caspase-1 inhibitor developed in cooperation with Vertex Pharmaceuticals and Aventis.²⁹⁴ In murine OA models, high-dose VX-740 (50 mg/kg) substantially alleviated inflammation and reduced cartilage degradation and bone erosion indicators by 59% and 84%, respectively.²⁹⁵ The subsequent phase I clinical trials demonstrated favorable tolerability of VX-740 with 50% oral bioavailability. In phase II clinical trials involving 285 RA patients, VX-740 showed a trend toward symptom improvement but failed to reach statistical significance. Phase III trials for RA and OA treatment were discontinued because of the liver toxicity observed in long-term animal studies.^50,294 VX-765 (belnacasan), another reversible caspase-1 inhibitor developed by Vertex Pharmaceuticals, has greater potency than VX-740.⁵⁰ Phase I clinical trials have shown that VX-765 meets pharmacokinetic and pharmacodynamic objectives, while phase II trials for psoriasis were completed in 2005, although no further updates have been published.²⁹⁴ A recent study demonstrated that oral administration of VX-765 considerably alleviated inflammation and joint tissue damage in a murine OA model,²⁹⁶ suggesting its potential as a novel anti-inflammatory agent for OA treatment.

To summarize, currently FDA-approved anti-IL-1 drugs—such as anakinra, canakinumab, and rilonacept—are associated with common biologic-related side effects, including injection-site reactions, reversible neutropenia, and an increased risk of bacterial infections.⁴⁹ Despite encouraging preclinical and early clinical findings, monotherapy with IL-1-targeting agents (such as anakinra, canakinumab, rilonacept, AMG 108, and VX-740) has not yielded substantial clinical improvements in RA or OA patients, although some trends toward relief have been noted.⁴⁸ These findings suggest that exclusive inhibition of IL-1β may be insufficient for effective OA treatment. Combination therapies targeting IL-1β along with other anti-inflammatory agents may enhance the therapeutic efficacy of reducing OA-associated inflammation.

As discussed earlier regarding TNF-α-mediated signaling in OA pathogenesis, therapeutic strategies targeting TNF-α can be categorized into three main approaches: (i) neutralizing both soluble and transmembrane sTNF-α and tmTNF-α; (ii) antagonizing TNFR-1 and TNFR-2 receptors; and (iii) inhibiting TACE, which reduces sTNF-α production and may partially prevent TNFR-1-expressing (including chondrocytes) apoptosis induced by the binding of sTNF-α and TNFR-1. However, since tmTNF-α also binds TNFR-1 with high affinity and TACE also cleaves TNFRs into soluble forms that act as natural TNF-α antagonists,^297,298 TACE inhibition may yield minimal or even counterproductive effects. For example, marimastat, a TACE inhibitor, suppressed LPS-induced sTNF-α in mice but only slightly delayed LPS-induced lethality.²⁹⁸

Traditional TNF-α-targeting agents approved by the FDA include infliximab, etanercept, and adalimumab.^68,298 Infliximab is a chimeric monoclonal antibody composed of a murine TNF-α-binding site fused with the Fc region of human IgG1, enabling it to bind specifically to TNF-α and block its interaction with TNFRs.^68,299,300 When infliximab is administered intravenously with a half-life of 8–9.5 days, it is approved for treating Crohn’s disease (CD), ankylosing spondylitis (AS), psoriatic arthritis (PsA), psoriasis (PSO), and RA.^68,301 However, its murine component may elicit human anti-chimeric antibodies (HACAs), reducing therapeutic efficacy—an effect that can be ameliorated with the coadministration of methotrexate.³⁰¹ Adalimumab is a fully human IgG1 monoclonal antibody with a similar mechanism but lower immunogenicity.^68,298 Nevertheless, long-term use may still lead to the development of anti-adalimumab antibodies.³⁰¹ When administered subcutaneously with a half-life of 10–13 days, it is approved for PsA, PSO, AS, CD, juvenile idiopathic arthritis (JIA), uveitis, and RA.⁶⁸

Etanercept is a fusion protein that combines the extracellular region of TNFR-2 and the Fc region of IgG1. It is administered subcutaneously with a half-life of 3–3.5 days and is approved for PsA, PSO, AS, and RA.^68,299,301 All these agents contain an IgG1 Fc region, enabling them to induce lysis of tmTNF-α-producing cells (including chondrocytes) via complement-dependent cytotoxicity and antibody-dependent cell-mediated cytotoxicity, potentially exacerbating OA progression.³⁰²

This issue is addressed by certolizumab pegol, a recently approved TNF-α-targeting agent. It is a humanized monoclonal antibody lacking the Fc region and consists of a human IgG1 Fab fragment with a murine anti-TNF-α amino acid sequence covalently linked to a polyethylene glycol moiety.^68,299 This PEGylation enhances pharmacokinetics, yielding a subcutaneous half-life of ~14 days. In CD, PSO, PsA, AS, and RA, certolizumab pegol avoids Fc-mediated cytotoxicity.⁶⁸ However, similar to other TNF-α antagonists, it may still induce immunogenicity and antibody formation, limiting its long-term use. These agents also increase the risk of infections (particularly tuberculosis), malignancies, systemic lupus erythematosus-like syndromes, and demyelinating diseases.³⁰⁰ The latter two are thought to be associated with disrupted TNF-α-TNFR-2 signaling (discussed below).²⁹⁹

Recent studies underscore the role of TNFR-2 in TNF-α-associated autoimmunization. Approximately 30% of regulatory T (T_reg) cells—key mediators in the suppression of immune overactivation—express TNFR-2 but not TNFR-1.³⁰³ The binding of TNF-α and TNFR-2 initiates expansion, prolonged survival, and increased phenotypic stability of T_reg cells in humans, thus spurring research into the use of TNFR-2 agonists as potential therapies for autoimmune diseases.³⁰⁴ However, paradoxically, the proportion of T_reg cells increases in most autoimmune diseases, including RA, following TNF-α inhibition therapy. This phenomenon may be explained by certain anti-TNF-α agents, such as adalimumab, which conversely increase the expression of tmTNF-α on monocytes, thereby promoting the interaction between tmTNF-α and TNFR-2.³⁰⁵ The change in the proportion of T_reg cells caused by nonselective blockade of TNF-α-TNFR-2 signaling might partially account for the autoimmune-related adverse effects and infections observed with anti-TNF-α monoclonal antibody therapy.

However, the proinflammatory effects are predominantly mediated by TNFR-1, suggesting that selective antagonism of TNFR-1 may be a promising strategy for treating OA, as it preserves TNFR-2-mediated immunoregulation and potentially prevents the side effects typically associated with monoclonal antibody therapy.³⁰⁶ Several attempts have been made to inhibit TNFR-1, such as R1antTNF—a mutant form of human TNF-α that selectively blocks TNFR-1 by competing with TNF-α for the binding site—showing promise in OA inflammation treatment. However, its therapeutic effects are limited by its short half-life (~10 min) in murine plasma when it is administered intravenously. PEGylation of R1antTNF extended its half-life to 12 h without compromising bioactivity. Both prophylactic and therapeutic use of PEGylated R1antTNF showed comparable efficacy to etanercept in reducing the severity of arthritis in a murine CIA model without impairing the immune response to infections—an issue observed with etanercept.³⁰⁷

Atrosimab, a monoclonal antibody targeting TNFR-1, considerably reduces the development and severity of arthritis in murine models, showing effects comparable to those of infliximab, etanercept, and certolizumab pegol in both prophylactic and therapeutic settings.³⁰⁸ Other selective TNFR-1 antibodies (IZI-06.1, TNFR1-AlbudAb, Nb 70 and Nb 96), as well as small molecules (PMG, R1), have demonstrated the ability to prevent TNFR-1-induced apoptosis and cytokine production in certain cell types without compromising TNFR-2-mediated immune regulation, suggesting promising therapeutic strategies for OA.³⁰⁶ However, these agents still rely on a similar mechanism of blocking TNFR-1 by competing for the same TNF-α binding site, which may limit their therapeutic efficacy owing to the high affinity of both sTNF-α and tmTNF-α for TNFR-1.

Recent studies have also identified noncompetitive TNFR-1 inhibitors, such as DS41, DSA11, and Zafirlukast, which bind to distinct sites on TNFR-1 to stabilize its nonfunctional state, thereby blocking downstream signaling without interfering with TNFR-2 regulation.³⁰⁶ In conclusion, current pharmacological anti-TNF-α agents (infliximab, etanercept, adalimumab, and certolizumab pegol) have demonstrated substantial clinical efficacy in treating various forms of arthritis, particularly RA. With a deeper understanding of TNF-α signaling, selective TNFR-1 antagonists have been developed to minimize the side effects of traditional anti-TNF-α therapies while retaining their anti-inflammatory properties. These agents have shown promising therapeutic effects in preclinical arthritis models. However, their efficacy in OA treatment remains underexplored, and further research is needed to evaluate the potential of anti-TNF-α therapy in OA management.

The IL-6 pathway has emerged as a pivotal factor in the pathogenesis of various diseases, including OA. As such, targeting IL-6 represents a promising anti-inflammatory strategy for OA treatment. IL-6 has three key binding sites: site I, site II, and site III. Through site I, IL-6 binds to either mIL-6R or sIL-6R to form a membrane-bound or circulating complex, which subsequently interacts with dimerized gp130 via sites II and III, initiating classic or trans-signaling.³⁰⁹ As discussed in the previous section, this IL-6 pathway is a central focus of therapeutic intervention.

Therapeutic strategies targeting the IL-6 pathway can be summarized into several categories: (i) blockade of gp130, (ii) blockade of circulating IL-6, (iii) blockade of IL-6R, (iv) inhibition of ADAM10/ADAM17, (v) inhibition of sIL-6R transcription, and (vi) inhibition of intracellular participants such as JAKs and STAT3. While gp130 serves as a signaling receptor for multiple cytokines, its blockade could disrupt cytokine homeostasis. Animal studies have shown that gp130-deficient mice are nonviable,^309,310 making gp130 inhibition an undesirable approach.

Conventional therapies primarily target circulating IL-6 or IL-6R via monoclonal antibodies. For example, tocilizumab is a humanized anti-IL-6R monoclonal antibody created by grafting the complementarity-determining regions of a mouse anti-human IL-6R antibody into human IgG1κ.³¹¹ It blocks both mIL-6R and sIL-6R by competing for site I on IL-6R. Tocilizumab has shown efficacy in protecting against joint destruction in RA treatment and is approved for treating RA, juvenile idiopathic arthritis, and other IL-6-related diseases.³¹⁰

Other monoclonal antibodies targeting IL-6 (e.g., olokizumab, siltuximab, sirukumab, and clazakizumab) and IL-6R (e.g., sarilumab, satralizumab, and vobarilizumab) have also been shown to be effective in reducing joint inflammation and destruction.^310,311 However, an increased risk of bacterial infections is commonly reported,^83,310 likely due to the role of IL-6 in plasma-cell differentiation and antibody production.^310,311 Further studies suggest that this risk is associated primarily with classic-signaling^312–314, whereas trans-signaling drives catabolic activity in chondrocytes.³¹⁵ For example, infections are the most common adverse events in tocilizumab-treated hand OA patients, whereas the trans-signaling-specific antagonist olamkicept has shown minimal immune impact and preserved bone repair in a murine femur osteotomy model.^316,317 This distinction has fueled interest in trans-signaling inhibitors.

Olamkicept (also known as sgp130Fc), a humanized monoclonal antibody composed of two sgp130 proteins fused to the Fc region of human IgG, is designed to compete for sites II and III on IL-6 with membrane-bound gp130.^309,318 While this configuration theoretically blocks both classic and trans-signaling by neutralizing the IL-6-sIL-6R and IL-6-mIL-6R complexes, animal studies have shown that olamkicept selectively inhibits trans-signaling unless it is administered at high doses,^{309,310,313,314} suggesting that sgp130 may bind to IL-6-sIL-6R with increased affinity. Olamkicept has demonstrated promise in preclinical arthritis models, with intra-articular injections considerably reducing joint inflammation and cartilage damage in RA mice, supporting the potential of selective IL-6 trans-signaling inhibition for OA.³¹⁹

However, sgp130 is also involved in other cytokine pathways, such as the IL-11 pathway, raising concerns about disruption of cytokine homeostasis.³¹⁸ A recent study addressed this issue by developing cs-130Fc, a monoclonal antibody that combines the binding region of sgp130 (domains D1–D3), a nanobody (VHH6) that targets the IL-6-sIL-6R complex, and a human IgG Fc region. Cs-130Fc was shown to double the potency of olamkicept in inhibiting IL-6 trans-signaling while reducing IL-11 trans-signal inhibition by eightfold,^313,318 offering new insights into anti-inflammatory therapy for OA.

Targeting the mIL-6R-cleaving enzymes ADAM10 and ADAM17 or inhibiting the transcription of sIL-6R to prevent the formation of IL-6–sIL-6R complexes represents another alternative strategy for inhibiting trans-signaling. However, a recent study suggested that the downregulation of ADAM10 and ADAM17 promotes the differentiation of murine bone marrow cells and human blood cells into osteoclasts, which may exacerbate bone resorption in OA patients.³²⁰ In contrast, inhibiting sIL-6R transcription may be insufficient for blocking IL-6R trans-signaling, as only ~15% of sIL-6R production is derived from RNA transcription, and no available agents targeting this process have been reported.³¹³

Although targeting IL-6 trans-signaling to inhibit inflammation while preserving host defense against bacterial infections is generally considered the most promising approach for anti-IL-6 therapy, a recent study revealed that IL-6R antibodies, the trans-signaling inhibitor sgp130, and the cleavage enzyme inhibitor TAPI2 did not reduce the expression of IL-6 or inflammatory mediators (NO, PGE₂, and MMPs) in chondrocytes or increase chondrocyte viability in a late-stage human knee OA model. However, an intracellularly active gp130 inhibitor, SC144, substantially improved these conditions.³¹⁵ These findings suggest that agents targeting extracellular IL-6 signaling may have limited therapeutic effectiveness once intracellular signaling is established in late-stage OA.^83,315

Therefore, blocking intracellular signaling components, such as JAKs (especially JAK1, which predominantly activates IL-6) or STAT3, may be preferable options for advanced OA treatment.³²¹ Several JAK and STAT3 inhibitors have been developed, including pan-JAK inhibitors (tofacitinib and baricitinib), JAK1-selective inhibitors (filgotinib), TYK2-selective inhibitors (deucravacitinib), and STAT3 inhibitors (OPB-31121, OPB-51602).³¹⁰ However, the development of IL-6 intracellular inhibitors is associated with several challenges, including low cell membrane permeability,³⁰⁹ disruption of cytokine homeostasis due to the involvement of JAK and STATs in other cytokine pathways,³¹³ and the nonselective blockade of both classic and trans-signaling (as both intracellular mechanisms are currently understood to be identical).^309,310,312

Targeting global intracellular signaling could result in side effects associated with classic-signaling, such as increased susceptibility to bacterial infections. Recent studies have shown that the administration of STAT3 inhibitors is customarily accompanied by side effects of peripheral neuropathy, and only limited efficacy in improving cartilage lesions has been reported in OA mice.^76,310 In conclusion, the selective inhibition of IL-6 trans-signaling remains the most promising therapeutic approach for the prevention or early treatment of OA until novel intracellular signaling discriminating targets or refined dosage forms of JAK and STAT3 inhibitors with reduced side effects are developed, which may provide new insights for advanced OA treatment (Table 1).

Mitochondrial homeostasis is crucial for chondrocytes to generate sufficient cellular energy and maintain normal cellular function. However, studies have shown that mitochondrial biogenesis-related proteins, such as AMPK, SIRT1, PGC-1α, NRF1, NRF2, and TFAM, are downregulated in OA chondrocytes from both humans and mice.^97,322 In addition, decreased expression of mitochondrial respiratory complexes I–IV and ATP synthase (Complex V) results in insufficient energy production.³²³ The buildup of dysfunctional mitochondria also leads to excessive ROS production, contributing to chondrocyte damage and apoptosis.^101,324 Therefore, restoring mitochondrial biogenesis by upregulating biogenesis-related proteins or enhancing mitophagy through proteins such as PINK1, Parkin, TOM, and LC3 to remove damaged mitochondria is a promising therapeutic strategy for OA. Potential drugs targeting these pathways, categorized by their roles in mitochondrial biogenesis or mitophagy, are summarized in Table 3.

Lysosomes play a vital role in degrading cellular waste and damaged organelles through hydrolytic enzymes, thereby supporting ECM homeostasis. Lysosomal destabilization, which triggers chondrocyte apoptosis, is implicated in the pathogenesis of OA. A study conducted by Tao et al. revealed that the phagocytosis of hydroxyapatite crystallites by chondrocytes can disrupt lysosomal integrity.¹¹³ Consequently, therapeutic agents capable of mitigating cartilage calcification may represent promising drugs for slowing OA progression by stabilizing lysosomes. Ideal candidates would inhibit chondrocyte hypertrophic differentiation or interfere with Ca²⁺ and P_i binding during cartilage calcification without causing severe side effects.

Over the past few years, several therapies aimed at reducing cartilage calcification have been explored at the cellular and animal levels, offering potential treatments for OA. Recent developments in this area are summarized in Table 3. These therapies help maintain the chondrocyte phenotype by downregulating the expression of calcification-promoting genes and proteins, such as RUNX2 and type X collagen, or by reducing ACP formation in MVs, which may prevent crystal-induced LMP. However, caution is warranted, as some of these therapies have not been tested in OA models, and no relevant animal experiments have been conducted.

Furthermore, the use of cathepsin inhibitors to preserve BID integrity, along with functional molecules that alleviate LMP—such as heat shock protein 70 (Hsp70), annexin A7, endosomal sorting complex required for transport protein III (ESCRT-III), LAMP-1, and LAMP-2—has been proposed as an alternative treatment to prevent lysosome-induced chondrocyte apoptosis.^325,326 Current studies suggest that cathepsins B and D primarily cleave BID. Specific inhibitors have been developed for cathepsins B (such as CA-074 and CA-030) and D (e.g., pepstatin A), which are used to treat various diseases.^327,328 However, the efficacy of these inhibitors in reducing lysosome-induced chondrocyte apoptosis remains unconfirmed, as no published studies have specifically addressed this issue. Notably, inhibitors of cathepsins K and L have been widely studied for OA treatment because of the high expression of these enzymes in osteoclasts, where they are responsible for collagen degradation in bone and cartilage.^329,330 Additionally, Hsp70 is an upstream regulator of ASM involved in the normal metabolism of SM.^331,332 Annexin A7 and ESCRT-III are components of the plasma membrane repair system that facilitate the resealing of damaged lysosomal membranes by aggregating at the injury site.^326,332 Moreover, LAMP-1 and LAMP-2 constitute ~50% of lysosomal membrane proteins and are essential for maintaining lysosomal morphology.³³³ Thus, the exogenous administration of recombinant forms of these proteins or their agonists may help repair hydroxyapatite-induced LMP in chondrocytes. This approach is supported by a study showing that recombinant Hsp70 (rHsp70), combined with the Hsp70 inducer arimoclomol, corrected pathological lysosomal morphology in primary fibroblasts from patients with Niemann–Pick disease.³³⁴

The ER is a central organelle for the proper synthesis, maturation, and folding of proteins. Over one-third of proteins in eukaryotic cells are folded within the ER lumen.¹³⁸ This substantial workload makes the ER highly susceptible to environmental stimuli, often leading to errors and the accumulation of misfolded proteins, which induce ER stress. The UPR signaling pathway, which is activated in response to ER stress, helps restore ER homeostasis to some extent. However, if stress becomes irreversible, it triggers apoptosis. Upon elaboration of ER stress-associated mechanisms, two therapeutic strategies are under investigation for pharmacological intervention: one focuses on reducing or eliminating the aggregation of misfolded proteins in the ER lumen, whereas the other targets the modulation of stress sensor activity (PERK, IRE1α, and ATF6α) or the enzymes involved in the UPR.³³⁵

PERK and IRE1α are known to initiate apoptotic processes in cells. Emerging studies have tested inhibitors of these sensors in various ER stress-related diseases, including cancer, neurodegenerative disorders, and obesity. GSK2656157, a PERK inhibitor, effectively reduces cancer growth but is discontinued because of its toxicity.³³⁶ ISRIB, a small molecule that inhibits eIF2α phosphorylation, thereby reducing CHOP expression, has been found to enhance cognitive memory in brain-injured mice without severe side effects.³³⁷ Inhibitors of IRE1α, such as 4µ8c, MKC3946, and STF-083010, reversed hepatic steatosis in animal models.³³⁵ While these agents may help alleviate ER stress during OA progression, their efficacy in OA models remains untested.

Another therapeutic strategy for treating ER stress involves reducing the aggregation of misfolded proteins. Certain small molecules and proteins can alleviate this aggregation in the ER, offering potential benefits for OA. Chemical chaperones are small molecular compounds that reduce the free movement of proteins, which helps prevent the aggregation of misfolded proteins. They also increase the free energy between misfolded and native proteins, promoting proper folding to restore a more stable state.³³⁸ However, chemical chaperones are nonselective and often require high concentrations to be effective, which may cause toxicity and limit their application in vivo.^335,338,339 Nonetheless, two chemical chaperones, 4-phenylbutyric acid (4-PBA) and tauroursodeoxycholic acid (TUDCA), are FDA-approved for human use.³³⁹ Among these, 4-PBA reduced cartilage destruction in a murine meniscectomy model, whereas TUDCA decreased the expression of ER stress-related markers and increased type II collagen expression.^340,341

Pharmacological chaperones (PCs) are another type of small molecule designed to selectively bind to misfolded proteins and bridge noncovalent interactions lost or loosened during misfolding, thereby promoting proper folding.^342,343 Compared with chemical chaperones, PCs work at lower concentrations and are more selective, making them promising therapeutic agents for OA. Cellular molecular chaperones are proteins that prevent the aggregation of misfolded proteins by recognizing and binding to their hydrophobic regions.³⁴⁴ The most common group of molecular chaperones is heat shock proteins (Hsps), which are classified by their molecular weight into Hsp40s, Hsp60s, Hsp70s, Hsp90s, Hsp100s, and small Hsps. These proteins are also involved in macromolecular complex assembly, protein transport, degradation, and the refolding of misfolded proteins,³⁴⁵ making them promising candidates for treating ER stress. Some molecular chaperones tested in OA chondrocytes are summarized in Table 3.

Certain chemical compounds and natural products derived from plants, fruits, and bacteria significantly downregulate the expression of ER stress-induced apoptotic genes or proteins (such as BiP, IRE1α, PERK, eIF2α, ATF4, and CHOP), positioning them as potential candidates for OA treatment (Table 3). However, the exact mechanisms through which these compounds alleviate ER stress are not fully understood, and further research is needed to elucidate how they affect ER stress pathogenesis.

Mitochondria-targeting carriers have been extensively studied in recent years, and most of them share the common trait of possessing lipophilic and cationic properties that enable the selective targeting of normal mitochondria through electrostatic interactions.³⁵² Lipophilic cations such as triphenylphosphonium (TPP)-based cations, rhodamine 123 (Rh123), cyanine cations, and cationic peptides are usually conjugated synergistically with other functional ligands on the surface of nanomedicines such as nanoparticles, liposomes, and vesicles, in which therapeutic molecules are encapsulated. This approach enhances tissue/cell specificity while increasing their permeability and bioavailability, regulating their distributions in vivo, and managing their release rates with minimal side effects.^352–354 For example, Huang et al. engineered a PEG/TPP-modified nanoparticle loaded with a natural anti-inflammatory compound, rhein, for delivery to mitochondria in IL-1β-stimulated chondrocytes. These nanoparticles exhibited anti-inflammatory effects by inhibiting IL-1β, IL-6, TNF-α, and MMP-13; upregulated the expression of anabolic genes (Col2a1 and Acan); protected the ΔΨm; and inhibited chondrocyte apoptosis.³⁵⁵ Yu et al. developed the pH-sensitive and TPP-modified nanoparticle DOX-PLGA/CPT/PD on the basis of the acidic property of the tumor environment for the delivery of DOX to mitochondria in tumor cells. The DDS is composed of positively charged nanoparticles synthesized from poly-lactide-co-glycolic acid (PLGA) and a TPP-containing amphiphilic polymer, with a negatively charged PD shell on the outer layer. This formulation ensures electroneutrality at physiological pH, preventing rapid clearance and nonselective ionic interactions in vivo. In an acidic tumor environment, the PD shell detaches from the nanoparticle, releasing the TPP-modified nanoparticle through amino bond breakage, resulting in improved DOX uptake and increased mitochondrial DOX concentrations in tumor cells.³⁵⁰ Similarly, Yuan et al. developed a biodegradable silica nanoparticle conjugated with TPP and CPD on the surface to deliver macromolecules to the mitochondria. The results further demonstrated the successful delivery of the native protein _FLBSA and an anti-MTCO2 antibody to the mitochondria in HeLa cells.³⁵⁶ These strategies can also be adapted to deliver functional molecules, such as the mitochondrial outer membrane protein MDI (CG3249) and C17orf80, which promote mtDNA replication in the normal mitochondria of chondrocytes to treat OA.^357,358 However, developing pharmaceutical formulations tailored specifically for OA is essential to overcome challenges such as the rapid in vivo clearance of cationic ligand-modified nanomedicines and the limitations of lipophilic cations, which primarily localize to the mitochondrial surface rather than the matrix. Furthermore, nonselective ionic interactions between cationic ligands and anionic proteins in the bloodstream may induce unintended side effects.^350,352

The promising mitochondria-targeting properties of lipophilic cations have spurred significant interest in exploring their therapeutic potential in various mitochondria-related diseases. However, these strategies are unlikely to effectively target depolarized mitochondria with impaired ΔΨm in chondrocytes.³⁵³ Therefore, potential independent mitochondria-targeting approaches may offer more suitable alternatives for treating OA.

Szeto-Schiller 31 (SS-31), a short-chain peptide consisting of ~10 amino acids, targets the IMM independently of ΔΨm through interaction with cardiolipin, a unique phospholipid residing on the IMM.^354,359 Owing to its small size, ease of synthesis, water solubility, favorable safety profile (with injection site reactions as the predominant side effect and no fatal events reported), and relatively long half-life demonstrated in animal studies (~2 h in rats, dogs, and monkeys),³⁵⁴ SS-31 is an ideal candidate for targeting dysfunctional mitochondria in chondrocytes.

Furthermore, derivatives of gramicidin S, which are traditionally known as antibiotics because of their ability to disrupt bacterial membranes, are also being explored as mitochondria-targeting carriers because of the similarities between bacterial and mitochondrial membranes. XJB-5-131, a derivative of gramicidin S, has been shown to be inserted into the IMM and deliver a ROS scavenger (4-AT) without dependence on the ΔΨm, suggesting its potential for targeting depolarized mitochondria in chondrocytes.^354,360 However, no mitochondrion-targeted pharmaceutical formulations using SS-31 or gramicidin S derivatives have been established for OA treatment, although relevant insights can be gleaned from other studies.

For example, Zhang et al. developed a PEG-PLGA nanoparticle conjugated with SS-31 that encapsulated cyclosporine A (CsA@PLGA-PEG-SS31) for treating myocardial ischemia‒reperfusion. The results demonstrated enhanced mitochondria-targeting efficacy and desirable cardioprotective effects.³⁵⁹ Similarly, Kuang et al. designed SS-31-conjugated geranylgeranylacetone-loaded PLGA nanoparticles that exhibited specific accumulation in mitochondria and enhanced ΔΨm in hair cells.³⁶¹

Recently, Li et al. reported the development of mitochondrion-targeted polyethylene glycol-fluorocarbon (PEG_m-F_n) self-assembled micelles, which consist of a fluorinated core and a PEGylated shell. PEG_m balances the strong hydrophobicity of F_n to form stable micelles, which disengage upon reaching the cell membrane due to its “stealth property.” Once internalized, F_n acts as a drug carrier and facilitates mitochondrial accumulation through interaction with cardiolipin on the IMM. These micelles exhibit electrically neutral properties, effective cellular uptake, endosomal escape, and successful mitochondrial targeting across various cell types.³⁵² This strategy addresses both the challenges of cell membrane penetration and mitochondria targeting, positioning PEG_m-F_n self-assembled micelles as a promising approach for mitochondrial targeting in chondrocytes.

Strategies for accessing the mitochondrial matrix involve the intricate process of navigating the mitochondrial import machinery. While lipophilic cations and potential-independent mitochondria-targeting carriers are effective at delivering therapeutic cargo to mitochondria, their ability to reach the mitochondrial matrix, where therapeutic effects are exerted, is often limited. To overcome this limitation, modification of the therapeutic cargo is necessary.

Substances entering the mitochondria are tightly regulated by the mitochondrial import machinery, which includes various translocases, such as the TOM and translocase of the inner membrane (TIM) complexes.³⁵⁴ The TOM complex, an oligomeric structure equipped with multiple channels, is responsible for recognizing and translocating proteins across the OMM. In this process, TOM40, in collaboration with TOM20, which identifies protein presequences, catalyzes the initial import of proteins into the intermembrane space. TOM20, characterized by its ability to adopt multiple conformations and interaction modes, specifically recognizes the presequences of proteins destined for the mitochondrial matrix. Upon recognition, TOM20 facilitates the translocation of these proteins to the central receptor TOM22, thereby initiating their import into the intermembrane space.

Conversely, the translocation of proteins destined for the inner mitochondrial membrane is orchestrated by the TIM complex, which plays a crucial role in their passage across this barrier. The TIM23 complex, which serves as the channel core of the TIM machinery and is composed of TIM17, TIM21, and TIM50, is complemented by the motor protein PAM. This motor protein is encircled by the mitochondrial chaperone heat shock protein 70 (mtHsp70), which is instrumental in catalyzing ATP hydrolysis. This ATP hydrolysis powers the translocation of proteins into the mitochondrial matrix. The synergistic function of the TOM and TIM complexes is indispensable for the precise and efficient delivery of therapeutic agents to the mitochondrial matrix, ensuring the accurate transport of proteins to their designated locations within the mitochondria.^354,362,363

Most mitochondrial-targeted proteins possess an MTS, which is essential for their transport to the matrix. Over half of the MTSs are localized at the N-terminus of proteins and often adopt an amphiphilic helical structure, with one positively charged surface and one hydrophobic surface—features believed to be necessary for matrix translocation.^353,364 Conjugating MTSs with therapeutic genes or proteins in DDSs can facilitate their delivery to the mitochondrial matrix. However, proteins transported to the matrix by MTSs must typically be unfolded, which may compromise their therapeutic efficacy to some extent.³⁵⁴

The size and complexity of DDSs often limit the efficiency of therapeutic cargo delivery to mitochondria.^349,354 To address this issue, several innovative strategies have been proposed. Yamada et al. developed a mitochondrial fusogenic liposome (termed the MITO-Porter DDS) on the basis of the concept that mitochondria undergo fusion and fission to share biomolecules. This system directly delivers the encapsulated therapeutic cargo to mitochondria by fusing with the mitochondrial membrane, regardless of the physical properties of the cargo,³⁴⁹ making the MITO-Porter DDS a promising candidate for gene or protein delivery in OA treatment.

In recent years, biomimetic nanoparticles with hybrid membranes from cancer cells and mitochondria that encapsulate therapeutic molecules have been developed for cancer treatment by leveraging homotypic binding between cancer and mitochondrial membranes.³⁶⁵ Membrane-camouflaged nanoparticles combine the physicochemical benefits of synthetic materials with their biological features, enabling efficient mitochondrial delivery of therapeutic molecules by modulating the ΔΨm in target cells.³⁶⁶ A similar approach could be adapted for OA treatment. Deng et al. developed chondrocyte membrane-coated nanoparticles for the treatment of cartilage damage. They demonstrated that chondrocyte membranes exhibit inherent binding affinity to the ECM, promoting longer drug retention and facilitating homotypic binding between chondrocytes, which enhances nanoparticle internalization.³⁶⁷ These findings provide both theoretical and practical foundations for the development of biomimetic nanoparticles for chondrocyte mitochondrial targeting in OA treatment.