Abstract
Alterations in mitochondrial function are the linchpin in numerous disease states including in the development of chemotherapy-induced neuropathic pain (CIPN), a major dose-limiting toxicity of widely used chemotherapeutic cytotoxins. In CIPN, mitochondrial dysfunction is characterized by deficits in mitochondrial bioenergetics (e.g., decreased ATP production) that are thought to drive the degeneration of the peripheral nerve sensory axon terminal sensory arbors in the skin (the intraepidermal nerve fibers; IENFs) and induce abnormal spontaneous discharge in peripheral nerve sensory axons. Preserving mitochondrial function is anticipated to prevent CIPN. We have now discovered that the G-protein-coupled receptor, A3 adenosine receptor subtype (A3AR), is expressed on the mitochondrial outer membrane. Ex vivo application of a highly selective A3AR agonist, MRS5980, to saphenous nerve microfilaments harvested from male oxaliplatin-treated rats reversed the loss in ATP production underscoring mitoprotective effects resulting from A3AR activation on mitochondria. Moreover, in vivo administration of A3AR agonists to rats during oxaliplatin treatment was associated with reduced IENF loss and a lower incidence of spontaneous discharge in peripheral afferent axons. These effects are accompanied by improved mitochondrial ATP production in primary afferent sensory axons and overall inhibition of the development of neuropathic pain. These data identify for the first time mitochondrial A3AR and indicate that activation of A3AR protects mitochondrial function in primary afferent sensory axons against chemotherapy-induced neurotoxicity. Repurposing A3AR agonists that are already in clinical trials as anticancer agents as adjunct to chemotherapeutics will address a major unmet medical need for which there are no FDA-approved drugs.
Significance Statement
Alterations in mitochondrial function are central to the development of chemotherapy-induced neuropathy and ensuing pain syndromes. Here we show that the adenosine A3 receptor (A3AR) is expressed on mitochondria, and its activation can reverse deficits in mitochondrial ATP production in peripheral sensory nerves from animals with chemotherapy-induced neuropathic pain. Moreover, administration of A3AR agonists to animals receiving chemotherapy prevented the development of pain and pathophysiological signs of neuropathy in sensory nerves such as the loss of intraepidermal nerve fibers and increased spontaneous firing. These findings identify for the first time mitochondrial A3AR and show that activation of A3AR protects mitochondrial function in primary afferent sensory axons against chemotherapy-induced neurotoxicity.
Introduction
As many as 65% of cancer patients undergoing treatment with first-line chemotherapeutics will experience chemotherapy-induced peripheral neuropathy (CIPN; Seretny et al., 2014). CIPN originates from injury to peripheral sensory nerve axons in response to chemotherapeutic cytotoxins in the taxane, vinca alkaloid, platinum complex, and proteasome inhibitor classes (Was et al., 2022). Many patients with CIPN will present a neuropathic pain syndrome characterized by ongoing pain, hyperalgesia, and allodynia in the lower extremities and hands (Starobova and Vetter, 2017) that can last for months to years after chemotherapy and be severe enough to cause dose limitation or cessation of chemotherapy (Seretny et al., 2014). The current pharmacological treatments for CIPN are limited and address the symptoms, not the drug-induced pathology (Y. Li et al., 2021). Novel strategies alleviating symptoms and preventing or modifying the underlying pathology would be invaluable to patients receiving such chemotherapeutics in their treatment program.
Mitochondrial dysfunction is a common contributor to chemotherapy-induced neurotoxicities that affect peripheral and central nervous systems in animal models (Bennett et al., 2014; Was et al., 2022) and humans (Kober et al., 2018). Chemotherapy-induced mitochondrial dysfunction is marked by an increased prevalence of abnormal mitochondrial morphology (e.g., swollen, vacuolated, and disrupted cristae), persistent energy deficits (reduced ATP production), and increased nitro-oxidative stress in peripheral nerve sensory axons (PNSAs; Bennett et al., 2014; Doyle and Salvemini, 2021) and brain synaptosomes (Chiu et al., 2017). In CIPN, mitochondrial dysfunction has been proposed to drive the degeneration of the intraepidermal nerve fibers (IENF), comprising PNSA terminal arbors (Bennett et al., 2014), and to compromise sodium–potassium pump function, resulting in depolarization and abnormal spontaneous discharge in both A- and C-fibers (Xiao and Bennett, 2007, 2008). In animal studies, strategies providing long-term mitochondrial protection prevent the development of CIPN (Bennett et al., 2014; Doyle and Salvemini, 2021). Thus, clinically useful compounds possessing mitochondrial protective effects might be effective adjuncts to chemotherapeutics to prevent CIPN and allow more aggressive chemotherapeutic regimens.
Selective agonists for the G-protein-coupled receptor (GPCR) adenosine A3 receptor (A3AR) have profound effects on the pain symptomology in male and female rodent CIPN models (Janes et al., 2014; Wahlman et al., 2018; Stockstill et al., 2020) with the exception of bortezomib where only male rodents respond to A3AR agonists (Stockstill et al., 2020). A3AR agonists modulate central sensitization and neuroinflammation to attenuate CIPN pain behaviors in animals (Chen et al., 2012; Janes et al., 2014). Notably, animals administered A3AR agonists during chemotherapy remain asymptomatic for CIPN weeks after the cessation of chemotherapy (Janes et al., 2014). It is not known whether A3AR agonists preclude or modify the underlying peripheral nerve pathology. However, in a mouse model of chemotherapy-induced cognitive impairment, MRS5980 protected cognitive function and synaptic integrity in the brain with preservation of mitochondria function (Singh et al., 2022). How A3AR agonists preserve mitochondrial function remains elusive.
In subcellular imaging studies, mitochondrial ATP levels indicate the metabolic setting of a given cell (Depaoli et al., 2018). Several mitochondrial-associated GPCRs have been identified and found to regulate mitochondrial function, including respiration, calcium uptake, cytochrome c release and nitro-oxidative stress (Belous et al., 2006; Benard et al., 2012; Valenzuela et al., 2016; Wang et al., 2016; Suofu et al., 2017). Whether adenosine receptors are expressed in mitochondria is not known. However, the presence of mitochondrial A3AR could serve as a biosensor of intracellular adenosine to regulate mitochondrial function and ATP production. By acting on mitochondrial A3AR, an A3AR agonist could counter-regulate chemotherapy damage to mitochondria in the peripheral sensory afferent and preserve their bioenergetics, thus preventing the development of CIPN.
Using in vitro and ex vivo studies, we investigated whether A3AR is expressed in mitochondria and impacts mitochondrial ATP production in the context of CIPN. We then used in vivo CIPN models to investigate whether selective A3AR agonists are mitoprotective and by extension neuroprotective during chemotherapy to prevent the peripheral nerve pathology that drives the development of CIPN and pain.
Materials and Methods
Experimental animals
Male Sprague Dawley rats (200–220 g starting weight) from Envigo (Frederick, MD breeding colony) were housed 3–4 per cage. Male wild-type C57BL\6 mice from our transgenic breeding colonies at Saint Louis University were housed 5–6 per cage. Animals were housed in a controlled environment (12 h light/dark cycle) with food and water available ad libitum. Our previous work established that A3AR agonists are effective in male and female rodent models of paclitaxel and oxaliplatin-induced CIPN (Janes et al., 2014; Wahlman et al., 2018; Stockstill et al., 2020). In contrast, we reported that A3AR agonists were protective in male, but not female, models of bortezomib-induced CIPN (Stockstill et al., 2020). For these reasons, all studies were performed in males. All animals were randomly assigned to treatment groups, and experiments were conducted with the experimenters blinded to treatment conditions. All experiments were performed in accordance with the International Association for the Study of Pain and the National Institutes of Health guidelines on laboratory animal welfare with the approval of Saint Louis University and the University of California, San Diego Institutional Animal Care and Use Committees.
Super-resolution stimulated emission depletion microscopy of mitochondrial A3AR in cell culture
Rat astrocytes (CTXTNA2) or mouse microglia (BV2) were cultured to 60–75% confluency on glass coverslips (no. 1.5) coated with Cell-Tak (Corning) or poly-ʟ-lysine (MilliporeSigma) in DMEM with GlutaMAX (Invitrogen) supplemented with 5% heat-inactivated fetal bovine serum (MilliporeSigma), penicillin (100 U/ml, MilliporeSigma), and streptomycin (100 µg/ml; MilliporeSigma). Cells were fixed in 4% paraformaldehyde and permeabilized by incubating with 0.5% Tween 20 for 30 min. The fixed and permeabilized cells were blocked for 1 h (10% normal goat serum, 2% BSA, 0.2% Triton X-100 in PBS) and then stained overnight at 4°C with rabbit polyclonal antibody to A3AR (1:100; Bioss USA) and mouse monoclonal antibody to TOMM20 (2 µg/ml; Abcam) in 1% BSA in PBS and 0.1% Tween 20. The coverslips were washed several times in 1× PBS with 0.1% Tween 20 and incubated with goat anti-rabbit IgG conjugated to Oregon Green (1:100; Thermo Fisher Scientific) and goat anti-mouse conjugated to Alexa Fluor 592 (1:100; Invitrogen) in 1% BSA in PBS and 0.1% Tween 20 for 2 h in the dark. Coverslips were washed several times in 1× PBS with 0.1% Tween 20 and twice in 1× PBS before being mounted on glass slides with ProLong Gold Antifade mounting media without nuclear stains (Thermo Fisher Scientific). Single isolated cells were identified and imaged using a Leica SP8 TCS STED 3X microscope. Images are representative of three independent experiments.
Mitochondria isolation
Isolated mitochondria from normal rat spinal cord, sciatic nerve, saphenous nerve, and mouse kidney tissues were initially isolated as previously described (Mattiasson, 2004). Briefly, tissues were homogenized on ice in isolation buffer (in mM: 320 sucrose, 1 EGTA, 10 Tris, pH 7.3) in a Potter-Elvehjem tissue grinder. Crude live mitochondrial pellets were generated by removing debris, unbroken cells, and nuclei by centrifugation at 1,380 × g for 3 min at 4°C and then separating mitochondria in the resulting supernatant by centrifugation at 7,800 × g for 5 min at 4°C. The crude mitochondrial pellet was further purified in 8 ml of 19% Percoll in isolation buffer and centrifuging 11,200 × g for 10 min at 4°C. The pellets were washed in isolation buffer and centrifuged 7,800 × g for 5 min at 4°C. Due to the small number of mitochondria recovered in saphenous nerves, the mitochondrial pellet at this step was used for Western blot analyses. For all other tissues, the mitochondria were further enriched on a discontinuous OptiPrep gradient (15, 25, 30 and 35%; Sigma) and centrifuged at 100,000 × g for 3 h at 4°C according to manufacturer's protocols (OptiPrep Application Sheet S14, Axis-Shield). The fractions at the 25–30% interface contained the highly enriched mitochondrial fraction and were washed in equal volume of 0.25 M sucrose, 1 mM EDTA, 20 mM HEPES-NaOH, pH 7.4, and centrifuged at 30,000 × g for 10 min to remove the OptiPrep.
Mitochondria from wild-type C57BL\6 mouse cortical brain tissues were harvested according to previous methods (Kristian, 2010; Suofu et al., 2017). Briefly, whole mouse brains were harvested within 5 min of killing and pooled two per sample. In ice-cold isolation buffer (in mM: 225 sucrose, 75 mannitol, 1 EGTA, and 5 HEPES, pH 7.4), the hind and mid brain regions were removed, and the remaining cortical tissue was minced and homogenized with eight up and down strokes in a Potter-Elvehjem homogenizer with a Teflon pestle. Crude mitochondrial pellets were collected after differential centrifugation (1,300 × g for 3 min and 21,000 × g for 10 min at 4°C). Mitochondria were purified on a discontinuous Percoll gradient (15, 24 and 40% Percoll in isolation buffer) centrifuged for 8 min at 30,700 × g, 4°C. The mitochondria fraction was collected at the 24–40% Percoll interface, diluted in isolation buffer and centrifuged at 16,700 × g for 10 min at 4°C). The pellet was resuspended in isolation buffer + 10 mg/ml fatty acid free bovine serum albumin (A7030, MilliporeSigma) and centrifuged at 6,900 × g for 10 min at 4°C. The resulting pellet was resuspended in isolation buffer without EDTA for analysis.
Western blot
Mitochondrial pellets were resuspended in 1× Laemmli buffer and boiled for 5 min. Mitochondrial protein bands were resolved by SDS-PAGE using 4–20% TGX gels (Bio-Rad) and electrotransferred to nitrocellulose membrane. Membranes were blocked for 1 h with 5% low fat milk in 1× PBS + 0.05% Tween 20 (PBS-T) and probed overnight at 4°C with rabbit antibodies to A3AR (1:1,000; #bs-1224, Bioss USA), calreticulin (1:1,000, #12238, Cell Signaling Technology), calnexin (1:1,000; #ab75801, Abcam), or VDAC1 (1:1,000; #4661, Cell Signaling Technology). After washes in PBS-T, the membranes were labeled with goat anti-rabbit IgG conjugated to horseradish peroxidase (1:1,000; #7074, Cell Signaling Technology) for 1 h at RT. Protein bands were imaged using enhanced chemiluminescence (Clarity, Bio-Rad) or SuperSignal West Femto (Thermo Fisher Scientific) on a Chemidoc MP system (Bio-Rad).
PNGase F treatment
Aliquots of purified mitochondria from wild-type mouse brains were divided in two and treated with or without Remove-iT PNGase F (New England Biolabs) and chitin beads according to manufacturer's protocols.
Transmission electron microscopy
Pellets of purified mitochondria from rat spinal cord were fixed by immersion in 4% paraformaldehyde in 0.1 M cacodylate buffer overnight at 4°C. The pellet was dehydrated in a cold ethanol series, embedded in LR White and polymerized under UV light. Ultrathin cross sections (90 nm) were mounted on Formvar-coated nickel slot grids, treated (10 min) with 0.1 M sodium citrate, washed, and incubated in 3% sodium metaperiodate (10 min) with washes to expose antigenic sites. Following incubation (20 min) in 10% normal goat serum in 1% Triton X-100, the sections were incubated with anti-A3AR rabbit antibody (1:100, bs-1221, Bioss USA) in 1% Triton X-100 at 4°C for 12 h. Labeled sections were then incubated (30 min) with 25 nM anti-rabbit colloidal gold in 0.1% Triton X-100. Rinses in 10 M Tris-HCl/PBS were followed by a rinse (5 min) in 2% glutaraldehyde in dH2O, staining in 2% uranyl acetate, and a poststain in 1% osmium tetroxide. Sections were examined using Hitachi H-7500 transmission electron microscope. Digitized images were obtained and archived by an ORCA camera with IC-PCI frame-grabber and AMT 12-HR software.
Chemotherapy-induced neuropathic pain models
Paclitaxel (Parenta Pharmaceuticals) or its vehicle (Cremophor EL and 95% ethanol in 1:1 ratio, MilliporeSigma) were injected intraperitoneally on Days 0, 2, 4, and 6. Daily doses of paclitaxel were 2 mg/kg for a final cumulative dose of 8 mg/kg (Polomano et al., 2001). Oxaliplatin (Oncology Supply) or its vehicle (5% dextrose, MilliporeSigma) was injected intraperitoneally on 5 consecutive days (Days 0–4). Daily doses of oxaliplatin were 2 mg/kg for a final cumulative dose of 10 mg/kg (Xiao et al., 2012). Bortezomib (Selleck Chemicals) or its vehicle [5% Tween 80 and 5% ethanol (MilliporeSigma) in 0.9% sterile saline] were injected intraperitoneally for 5 consecutive days (Days 0–4). Daily doses of bortezomib were 0.2 mg/kg for cumulative dose of 1 mg/kg (Zheng et al., 2012).
Test compounds
IB-MECA (CF101, piclidenoson; Jacobson, 1998), an agonist with 50- to 100-fold specificity for rat A3AR (Fishman et al., 2012) over other adenosine receptor subtypes, was purchased from Cayman Chemical. IB-MECA has been clinically investigated for rheumatoid arthritis (Phase III, completed) and plaque psoriasis (Phase III, ongoing) with good safety profile (Fishman, 2022). The A3AR agonists MRS5698 (Tosh et al., 2015b), MRS5980 (Tosh et al., 2014), and MRS7154 (Tosh et al., 2015a) were produced by our group (K.A.J) and respectively have >3,000-fold (Tosh et al., 2015b), >1,000-fold (Tosh et al., 2014), and >2,000-fold (Tosh et al., 2015a) more selectivity for human A3AR versus other adenosine receptor subtypes. Using a prophylactic paradigm as previously described (Janes et al., 2014), daily doses of the selective A3AR agonists or their vehicle (0.01% dimethyl sulfoxide in phosphate-buffered saline, pH 7.4) were administered intraperitoneally (0.1 mg/kg/d) beginning at the start of chemotherapy (Day 0) and continued to Day 6 for paclitaxel-treated rats or Day 4 for oxaliplatin- or bortezomib-treated rats. On chemotherapy days, A3AR agonists or their vehicles were administered 15–20 min before the chemotherapy. MRS1523 (2 mg/kg/d; Janes et al., 2014), a selective A3AR antagonist (100- to 800-fold affinity for A3AR vs other adenosine receptor subtypes; A. H. Li et al., 1998), or its vehicle (0.01% dimethyl sulfoxide in phosphate-buffered saline, pH 7.4) was given intraperitoneally 15–20 min prior to A3AR agonists.
ATP assay
The saphenous nerves were excised and placed in ice-cold mitochondria preservation media for testing ATP production using a technique adapted for the saphenous nerve explant, as previously described (Zheng et al., 2011, 2012; Janes et al., 2013). Briefly, the saphenous nerves were minced and teased in preservation media into axonal microfilaments. The preservation media was replaced with room temperature respiration media and incubated with substrates for respiratory Complexes I and II (5.0 mM glutamate and 2.5 mM malate and 5.0 mM succinate, respectively). Samples of respiration media (100 µl) were taken from each explant before (baseline) and 5 min following stimulation with ADP (1.0 mM). For assays where explants were treated acutely ex vivo with A3AR agonist, MRS5980 (1 μM) was added for 15 min, and a sample of respiration media (100 µl) was taken before (baseline) and 5 min after the addition of ADP (1.0 mM). Samples were stored at −80°C until they could be assessed with a flash luciferin-luciferase assay (Promega Enliten ATP Assay; Promega) according to manufacturer's protocol. The ATP concentrations were normalized to citrate synthase activity (a mitochondrion-specific enzyme) in homogenates prepared from the explants using a commercial citrate synthase activity kit (Sigma). Data are expressed as ATP concentration over citrate synthase units (ATP nmol/ml/CSU) or as the difference in ATP nmol/ml/CSU before and after ADP stimulation.
Confocal Raman imaging of the A3AR agonist distribution in live cells
Chinese hamster ovary (CHO) and mouse BV2 microglial cells were grown on quartz slides in FluoroBrite DMEM (Life Technologies) and treated for 1 h with MRS5698 (0.2–0.25 µM) or MRS5980 (1 µM). Individual cells were located by bright-field imaging using a 63× (NA = 1.0) immersion objective on a WITec alpha300R microscope equipped with a 532 nm (green) excitation laser and a hyperspectral EMCCD camera. The system measures a complete Raman scattered vibrational spectrum at each pixel. In the image the resolution in the x–y plane is near the diffraction limit (250 nm). This technique is “label-free” in that it does not require adding fluorescent groups to visualize molecules. The whole-cell images were first filtered at the 2,950 cm−1 spectral peak to visualize single C–H bonds within all biomolecules of the cell; this provides a Raman image of all cellular compartments. Then the images were filtered at the signature spectral peak for the A3AR agonist. The triple C–C bond in MRS5698 and MRS5980 produces a peak in the silent region of the Raman spectra (2,227 cm−1 and 2,224 cm−1, respectively). In order to localize mitochondria, confocal images were filtered at the signature Raman spectral peak for cytochrome c (750 cm−1; Okada et al., 2012). When used at low concentrations (10–20 nM), DiOC6(3) labels the mitochondrial compartment (Cottet-Rousselle et al., 2011). DiOC6(3) (10 nM) was added to the media and its fluorescence intensity measured over the same x–y–z coordinates that were used as an alternative method to identify mitochondria.
IENF counts
IENFs were visualized immunocytochemically as described previously (Siau et al., 2006). Samples of plantar paw skin were harvested from rats treated with paclitaxel, oxaliplatin, bortezomib, or their vehicles at the approximate time of peak pain severity (Day 25 for paclitaxel and Days 34–35 for oxaliplatin and bortezomib) after the presence of allodynia and hyperalgesia was confirmed by behavioral testing. Sections were stained with anti-protein gene-product 9.5 antiserum (PGP9.5; Research Diagnostics) diluted 1:6,400. IENFs were counted by an observer blinded to the animal treatment group. Using a 40× objective, all ascending nerve fibers that were seen to cross into the epidermis were counted. One 30-mm-thick section, 8–10 mm long, was counted per rat. IENF counts are expressed as the number per centimeter of epidermal border (straight line approximation). No staining was present in sections processed without exposure to the primary antibody.
Spontaneous discharge
The procedure is described in detail in previous work (Xiao and Bennett, 2007, 2008). Briefly, recordings were made in normal control rats and rats with confirmed bortezomib-induced CIPN at the approximate time of peak pain hypersensitivity. Subcutaneous needle electrodes were inserted across the lateral surface of the ankle for stimulation of sensory afferent axons in the sural nerve. Microfilaments were dissected from the distal end of the transected nerve in the popliteal fossa and draped over a silver-wire hook electrode that was referenced to a needle electrode inserted in adjacent muscle. This arrangement records axonal activity originating in the periphery. By gradually increasing the stimulus voltage, we determined the number of individually identifiable axons in each microfilament. Axons were considered to be individually identifiable if they had a discreet and constant threshold, latency, and waveform. The incidence of individually identifiable axons with spontaneous discharge (a minimum of five impulses in 5 min) and the frequency of the discharge was noted, and the ratios of the number of axons with spontaneous discharge relative to the number of individually identifiable axons were calculated for A-fibers and C-fibers. Axonal conduction velocity was determined by measuring the distance between the stimulation site and the nerve transection. We did not differentiate between A-fibers with conduction velocities in the Aβ and Aδ ranges, because it is impossible to differentiate functional classes of A-fibers on this basis (Djouhri and Lawson, 2004). We avoided characterizing fiber responses to receptive field stimulation because this requires repeated application of stimuli that might sensitize nociceptors. Sensitized nociceptors might have an ongoing discharge indistinguishable from spontaneous discharge. It was nearly always easy to individually identify Aδ (2–30 m/s) and C-fiber (≤2.0 m/s) axons, but the microfilaments contained a relatively large number of Aβ axons with very similar thresholds and conduction velocities, which precluded individual identification of more than one to three of these fibers per microfilament.
Behavioral testing
Rats were assessed for mechanical hypersensitivity by a previously described method (Flatters and Bennett, 2006). Briefly, 4 and 15 g Von Frey filaments were applied to the plantar hindpaws five times per side with 1–2 min between successive stimuli and the percentage of stimuli that elicited a withdrawal reflex was recorded. Mechano-allodynia was indicated by significant increases in the number of responses to the 4 g filament. Mechano-hyperalgesia was indicated by significant increases in the number of responses to the 15 g filament. Cold allodynia was assessed by the acetone method as previously described (Xing et al., 2007). Briefly, a drop (0.05 ml) of acetone was placed on the center of the plantar hindpaw and the response graded as follows: 0, no response; 1, quick withdrawal or flick of the paw; 2, prolonged withdrawal/flicking; and 3, prolonged flicking and licking of ventral side of the paw. The accumulative score of three acetone applications was expressed as a percentage of the maximal score (100% = 18 out of 18).
Experimental design and statistical analysis
Experiment 1
Ex vivo ATP production: Male rats were treated daily for 5 d with intraperitoneal injections of oxaliplatin (n = 12) or vehicle (n = 13). On Day 27, the saphenous nerves were harvested and randomly assigned to the following ex vivo treatment groups: (1) in vivo vehicle to oxaliplatin and ex vivo vehicle to MRS5980 (n = 6); (2) in vivo vehicle to oxaliplatin and ex vivo MRS5980 (n = 7); (3) in vivo oxaliplatin + ex vivo vehicle to MRS5980 (n = 7), or (4) in vivo oxaliplatin and ex vivo MRS6980 (n = 5). Data were analyzed by two-tailed one-way ANOVA with Dunnett's comparisons to oxaliplatin + vehicle group.
Experiment 2
Saphenous nerve ATP production: Male rats were treated daily for 5 d with intraperitoneal injections of (1) vehicle to oxaliplatin and vehicle to MRS5698/IB-MECA (n = 6); (2) oxaliplatin and vehicle to MRS5698/IB-MECA (n = 6); (3) oxaliplatin + IB-MECA (n = 6), (4) oxaliplatin and MRS5698 (n = 5); (5) oxaliplatin + IB-MECA + MRS1523 (n = 5), or (6) oxaliplatin + MRS5698 + MRS1523 (n = 5). Saphenous nerves were harvested for ATP production assays on Day 35 post first injection. Data were analyzed by two-tailed one-way ANOVA with Dunnett's comparisons to oxaliplatin + vehicle (Groups 1–4) and to oxaliplatin + IB-MECA/MRS5698 (Groups 3–6).
Experiment 3
MRS5698 attenuation of IENF loss: Male rats (n = 10/group) were treated with (1) vehicles; (2) vehicle and MRS5698; (3) paclitaxel and vehicle; (4) paclitaxel and MRS5698; (5) oxaliplatin and vehicle; (6) oxaliplatin and MRS5698; (7) bortezomib and vehicle; or (8) bortezomib and MRS5698. Data were analyzed by two-tailed unpaired t test between chemotherapy/vehicle groups cotreated with vehicle and chemotherapy/vehicle groups cotreated with MRS5698.
Experiment 4
MRS7154 attenuation of IENF loss: Male rats (n = 10/group) were treated with (1) vehicles; (2) bortezomib and vehicle; or (3) bortezomib and MRS7154. Data were analyzed by two-tailed one-way ANOVA with Dunnett's comparisons to bortezomib + vehicle group.
Experiment 5
Spontaneous discharge: Male rats were treated with (1) vehicles, (2) bortezomib + vehicle, or (3) bortezomib + MRS7154. At peak hypersensitivity, spontaneous discharge for 197 A-fibers and 74 C-fibers were recorded. Data were analyzed by two-tailed one-way ANOVA with Dunnett's comparisons to bortezomib + vehicle.
Experiment 6
MRS5698 attenuation of pain behavior: Male rats (n = 10/group) were treated with (1) vehicles; (2) vehicle and MRS5698; (3) paclitaxel and vehicle; (4) paclitaxel and MRS5698; (5) oxaliplatin and vehicle; (6) oxaliplatin and MRS5698; (7) bortezomib and vehicle; or (8) bortezomib and MRS5698. Mechano-allodynia data were analyzed by two-tailed unpaired t test between chemotherapy/vehicle groups cotreated with vehicle and chemotherapy/vehicle groups cotreated with MRS5698. Cold allodynia data were analyzed by the Friedmann test with Dunn's comparisons (time comparisons) and Mann–Whitney U test (treatment comparisons).
Experiment 7
MRS7154 attenuation of IENF loss: Male rats (n = 10/group) were treated with (1) vehicles; (2) bortezomib and vehicle; or (3) bortezomib and MRS7154. Data were analyzed by two-tailed two-way ANOVA with Dunnett's comparisons to bortezomib + vehicle group.
Results
A3AR is expressed on mitochondria in vitro
To determine the potential for intracellular expression of A3AR along the pain pathway, we first utilized readily available established microglia and astrocyte cell lines. Unstimulated cultured mouse microglia (BV2; Fig. 1A) immunolabeled for A3AR and imaged by super-resolution stimulated emission depletion (STED) immunofluorescence microscopy revealed intracellular A3AR signal in addition to its expected expression on the plasma membrane. A substantial portion of the intracellular A3AR signal was contained in subcellular regions that were positive for the mitochondrial protein translocase of outer mitochondrial membrane 20 (TOMM20; Fig. 1B,C). The signal from A3AR immunolabeling was present in many, but not all, TOMM20-labeled mitochondrial regions, which could be due to relatively low A3AR expression in unstimulated cells and/or the complexity of elongated and branched mitochondria not captured in our imaging. A3AR signal was also found in TOMM20-immunolabeled regions of rat cortical astrocytes (CTXTNA2), at lower levels than microglial cultures, with little detection of A3AR expression at the plasma membrane (Fig. 1D,E).
In vitro expression of A3AR in mitochondria. A–C, STED microscopy of immunolabeled A3AR revealed the presence of A3AR (green) in the cell membrane and within the TOMM20-labeled (red) mitochondria of cultured mouse microglia (BV2; A). Higher magnification (B) and three-dimensional rendering of the z-stack images of a BV2 mitochondrion further demonstrates A3AR is embedded within the outer mitochondrial membrane. D, E, STED microscopy of immune-labeled A3AR (green) in rat cortical astrocytes (CTXTNA2) was much lower in the cell membrane but still present within TOMM20-labeled (red) mitochondria. Yellow box indicates regions of higher magnification. After STED, maximum resolutions of 52 and 107 nm were achieved for Oregon Green 488 (A3AR) and Alexa Fluor 592 (TOMM20), respectively.
A3AR is expressed on mitochondria in vivo
To validate the expression of A3AR on mitochondria in vivo, we isolated mitochondria subcellular fractions from normal rodent nervous system tissues. A3AR expression was detected by Western blot in subcellular fractions isolated from rat nervous tissue (spinal cord, sciatic nerve and saphenous nerve; Fig. 2A) and mouse cortical brain tissue (Fig. 2B). These A3AR-positive subcellular fractions contained high mitochondrial VDAC1 protein levels but no or very low expression of the endoplasmic reticulum proteins, calreticulin or calnexin. To validate the specificity of the A3AR signaling in mitochondrial subcellular fractions, mitochondria were isolated from A3AR knock-out (A3AR−/−) mouse tissues. These mouse colonies are no longer available and isolation of mitochondria from spinal cord and brain tissues from our cryopreserved archive did not yield mitochondria of sufficient quality and quantity. However, kidney tissue has been reported to express A3AR (Dorotea et al., 2018) and mitochondria of ample quality and quantity were isolated from wild-type and A3AR−/− cryopreserved kidney tissue. As shown in Figure 2C, A3AR was detected in purified mitochondrial fractions of wild-type kidney tissues, but not in kidney mitochondria from A3AR−/− mice.
A3AR is expressed in isolated rat and mouse mitochondria. Western blot analyses of isolated mitochondrial fractions from normal (A) rat spinal cord, sciatic nerves, and saphenous nerves and (B) mouse cortical brain tissues. C, Western blot analyses of kidney tissues from wild-type and A3AR−/− mice demonstrated the loss of mitochondrial A3AR expression in knock-out mice (only archival kidney tissue was available from A3AR−/− mice). D, PNGase F treatment of mitochondria purified from mouse brain reduced the level of higher molecular weight bands compared with the untreated samples. Purity of mitochondrial fractions were determined by high expression of the mitochondrial VDAC1 and no-to-low expression of endoplasmic reticulum proteins, calreticulin, or calnexin. Images are representative of at least 3–4 isolations/tissue. E, F, Representative transmission electron microscopy images of two mitochondrial pellets isolated from rat spinal cords immunolabeled with A3AR antibody with immunogold secondary antibodies (25 nm gold; arrowheads).
The predicted molecular mass of A3AR protein sequence is 36–37 kDa (Sayers et al., 2022). However, Western blots yielded bands at 42, 52, and 66 kDa when labeled for A3AR, and this varied depending on the tissue from which the mitochondria were harvested. These higher molecular mass bands are consistent with previous reports (Giannaccini et al., 2008). A3AR has three predicted N-glycosylation sites that could give rise to these higher molecular mass bands (Zhou et al., 1992). To test this, mitochondria purified from wild-type mouse brains were treated with PNGase F. The higher molecular weight bands (>37 kDa) demonstrated a lower signal intensity in PNGase F-treated mitochondria isolates than untreated samples. In contrast, the intensity of the predicted A3AR band 36–37 kDa band increased with PNGase F treatment (Fig. 2D). These results are consistent with the notion that the higher molecule weight bands contain glycosylated A3AR.
A3AR is expressed on the mitochondrial outer membrane
To understand where A3AR was expressed on mitochondria, purified mitochondria from normal rat spinal cord were immunogold labeled for A3AR and imaged by transmission electron microscopy. A3AR appear to reside in the outer membrane of mitochondria of isolated mitochondria (Fig. 2E,F).
Mitochondrial A3AR contributes to the maintenance of mitochondrial bioenergetics ex vivo
To determine a functional role for mitochondrial A3AR, we tested whether direct activation of the A3AR expressed on mitochondria would counter-regulate oxaliplatin-induced mitochondrial dysfunction and restore normal ATP production in ex vivo preparations of peripheral sensory afferent mitochondria. Axonal microfilaments minced and teased from the purely sensory saphenous nerves of rats exposed mitochondria to bath application of compounds while remaining in contact with cytoskeleton, protecting the dysfunctional mitochondria that are otherwise too fragile to undergo standard isolation procedures (Zheng et al., 2011).
The mitochondrial ATP production following ADP stimulation of saphenous nerve axonal microfilaments isolated from rats treated with oxaliplatin was significantly reduced compared with those from vehicle-treated animals (Fig. 3). This is consistent with our previous studies with similar sciatic nerve axonal microfilament explants from animals with CIPN and neuropathic pain due to paclitaxel, oxaliplatin, and bortezomib (Xiao and Bennett, 2012; Zheng et al., 2012). However, exposure to an ex vivo bath application of A3AR agonist MRS5980 15 min prior to ADP stimulation significantly improved ATP production in saphenous nerve axonal microfilament explants from oxaliplatin-treated animals (Fig. 3). There was no significant effect on ATP production when saphenous nerve axonal microfilament explants from vehicle-treated rats were exposed to MRS5980.
Ex vivo application of A3AR agonists attenuated mitochondrial ATP deficits in saphenous nerve microfilament preparations from rats treated with oxaliplatin. A, Treatment and assay timeline. B, Exposed mitochondria in nerve microfilament explants prepared from saphenous nerves harvested on Day 27 from rats treated with oxaliplatin (Ox + Veh; n = 7) demonstrated significantly reduced ATP production following ADP stimulation when compared with saphenous nerve microfilament explants from vehicle-treated rats (Veh + Veh; n = 6). The deficits in ATP production were reversed in saphenous nerve microfilament explants from oxaliplatin-treated rats that received a bath application of MRS5980 (Ox + MRS5980; 1 µM; n = 5) 15 min before ADP stimulation. Bath application of MRS5980 had no significant effect on ATP production in saphenous nerve microfilament explants from vehicle-treated rats (Veh + MRS5980; n = 7). Data are expressed as mean ± SEM and analyzed by two-tailed one-way Brown–Forsythe ANOVA with Dunnett's T3 comparisons. B, F*(3,12.34) = 11.01, p = 0.041; *p < 0.05 versus vehicle and †p < 0.05 versus Ox + Veh.
A3AR agonists accumulate intracellularly
For A3AR agonists to elicit such a direct response on mitochondrial ATP production in intact cells (in vivo or in vitro), they would have to gain access to the intracellular compartment and be in proximity to regions rich in mitochondria. To test this, live unstained cells treated with A3AR agonists were examined using confocal Raman microspectroscopy. The typical Raman spectrogram of live unstained cells exhibits spectral peaks around 2,800–3,100 cm−1 for the vibrational energy associated with stretching at excited C–H bonds in all biomolecules. The energy produced at these peaks when filtered can be used to visualize the entirety of unstained cells by confocal microscopy (Fig. 4A–D). Other natural biochemical bonds in cells (e.g., C=O, S=O, C–C) produce a variety of peaks below ∼1,800 cm−1 (Pezzotti, 2021). However, the triple bonds in the A3AR agonists MRS5698 (∼2,227 cm−1; Fig. 4A) and MRS5980 (∼2,224 cm−1; Fig. 4C) produce peaks within a so-called silent region (∼2,000–2,500 cm−1) of unstained cells, which lack natural molecules with triple bonds.
A3AR agonists accumulate in the intracellular space. A, C, Using Raman microspectroscopy, unstained live cells will produce spectral peaks around 2,800–3,100 cm−1 for the vibrational energy associated with stretching at excited C–H bonds in all biomolecules. Other biochemical bonds in cells (e.g., C=O, S=O, C–C) produce a variety of peaks below ∼1,800 cm−1. Unstained cells typically lack any spectral peaks within the silent region (∼2,000–2,500 cm−1) associated with triple bonds. However, the triple bond in the A3AR agonists MRS5698 and MRS5980 produces peaks within this silent region (MRS5698: ∼2,227 cm−1; A) and MRS5980 (∼2,224 cm−1; C). B, D, CHO cells (normal and A3AR overexpressing CHO-A3AR) were treated with MRS5698 (0.2 µM) or MRS5980 (1 µM) for 1 h and imaged. Using filters for total C–H stretch (∼2,950 cm−1), the entire cellular compartment of a CHO cell is visible. When filtered for A3AR agonists, the detectable Raman signatures for MRS5698 (red, B) and MRS5980 (red, D) are largely located in the intracellular space (orange in merged figure) and not at the plasma membrane. E, F, The intracellular signature of MRS5698 (red) in CHO cells and mouse BV2 microglia was found within regions containing high levels of mitochondria indicated by the Raman signature for mitochondrial cytochrome c (∼750 cm−1; blue, E) or mitochondrial DiOC6(3) staining (10 nM; green, F). The total cell was visualized by the C–H stretch (green; E) or by differential interference contrast (DIC; E, F).
When live wild-type and A3AR overexpressing CHO cells, a commonly used transgenic cell system, were treated with MRS5698 (Fig. 4B) or MRS5980 (Fig. 4D) for 1 h and imaged using filters at the Raman spectral peaks for MRS5698 or MRS5980, the A3AR agonists were found localized within the intracellular compartment. Further investigation revealed that the Raman signature for MRS5698 was localized in regions where the Raman signature for mitochondrial cytochrome c (Fig. 4E; Raman spectral peak = 750 cm−1) in MRS5698-treated CHO cells and mouse BV2 microglia was the greatest. To validate the mitochondrial localization of A3AR agonists, CHO cells and mouse BV2 microglia were treated with MRS5698 for 1 h, stained with mitochondrial marker 3,3′-dihexyloxacarbocyanine iodide [DiOC6(3), 10 nM], and imaged. The Raman signature of MRS5698 was localized in DiOC6(3) positive regions (Fig. 4F).
In vivo administration of A3AR agonists protects mitochondrial ATP production in peripheral sensory axons during oxaliplatin treatment
Rats were given IB-MECA or MRS5698 during oxaliplatin treatment to test whether in vivo administration of A3AR agonists prevented chemotherapy-induced deficits in ATP production within the peripheral sensory afferents. On Day 35, saphenous nerve axonal microfilament explants were generated from each group, and their level of ATP production was measured before and after ADP stimulation. There was very little ATP production in all groups before the addition of ADP and explants from all groups showed increased ATP production following ADP stimulation (Fig. 5). However, the levels of ADP-stimulated ATP production in explants from oxaliplatin-treated rats were significantly lower than those from vehicle-treated rats (Fig. 5). Yet, when rats received IB-MECA or MRS5698 during oxaliplatin, ATP production in their explants were significantly improved compared with oxaliplatin group (Fig. 5). Moreover, the beneficial effects of IB-MECA and MRS5698 on ATP production were blocked in explants from rats that received concurrent administration of a selective A3AR antagonist, MRS1523 (Fig. 5).
A3AR agonists attenuate oxaliplatin-induced deficits in mitochondrial ATP production in rat saphenous nerve microfilament preparations. A, Treatment and assay timeline. B, In vivo treatment with A3AR agonist during oxaliplatin regimen prevented mitochondrial ATP deficits in saphenous nerves. Baseline ATP production in saphenous nerve microfilament preparations harvested on Day 35 was low in all groups in the presence of substrates for Respiratory Complexes I and II prior to stimulation with ADP. The addition of ADP led to the expected increase in ATP production in the vehicle group. When compared with vehicle (n = 6), treatment of saphenous nerve microfilament isolated from rats with CIPN due to oxaliplatin (n = 6) had a significant reduction in stimulated ATP production. The loss in ATP production was attenuated in saphenous nerve microfilaments isolated from rats that received IB-MECA (0.1 mg/kg/d, i.p.; n = 6) or MRS5698 (0.1 mg/kg/d, i.p.; n = 5) during oxaliplatin treatment. The effects of A3AR agonists were blocked in animals treated with the A3AR antagonist, MRS1523 (2 mg/kg/d; n = 5/group). Data are expressed as mean ± SEM and analyzed by two-tailed two-way ANOVA with Dunnett's comparisons. B, F(5,54) = 5.2, p = 0.0006, η2p = 0.32; *p < 0.05 versus Vehicle; †p < 0.05 versus Ox + Veh and #p < 0.05 versus Ox + IB-MECA or MRS5698.
A3AR agonists prevent chemotherapy-induced neurodegeneration
As we have shown previously (Siau et al., 2006; Bennett et al., 2011; Xiao et al., 2012; Zheng et al., 2012), IENF loss in animals with neuropathic pain following treatment with paclitaxel, oxaliplatin, or bortezomib is significant compared with naive rats (Fig. 6A,B). However, the number of IENFs in the hindpaw tissue of rats treated with MRS5698 during chemotherapeutic exposure was significantly greater than those treated with the chemotherapeutic alone (Fig. 6A). Similar cotreatment with MRS7154 significantly reduced the IENF loss caused by bortezomib (Fig. 6B).
A3AR agonists reduced chemotherapy-induced IENF loss in rats. A, B, When compared with the naive control group, animals receiving paclitaxel, oxaliplatin, or bortezomib exhibited a significant loss of IENFs in the glabrous skin of the hindpaw (the site of mechanosensitivity testing). The IENF loss was attenuated in rats cotreated with MRS5698 (0.1 mg/kg/d, i.p.; A) or MRS7154 (0.1 mg/kg/d, i.p.; B). Results are expressed as mean ± SEM for n = 10/group and analyzed by (A) one-tailed, unpaired t test or (B) two-tailed, one-way ANOVA with Dunnett's comparisons; (A) Chemo + Veh versus Veh + Veh: paclitaxel, t(18) = 3.383, p = 0.0016, d = 1.5; oxaliplatin, t(18) = 4.079, p = 0.00035, d = 1.8; bortezomib, t(18) = 3.832, p = 0.00061, d = 1.7. Chemo/Veh + MRS5698 versus Chemo/Veh + Veh: vehicle, t(18) = 0.2085, p = 0.418, d = 0.094; paclitaxel, t(18) = 2.097, p = 0.025, d = 0.94; oxaliplatin, t(18) = 2.979, p = 0.0040, d = 1.3; bortezomib, t(18) = 1.859, p = 0.397, d = 0.83; (B) F(2,33) = 24.60, p = 2.89 × 10−7, η2p = 0.60. *p < 0.05 versus vehicle group and †p < 0.05 versus Chemo + Veh group.
A3AR agonists prevent chemotherapy-induced spontaneous discharge in primary afferent sensory axons
We have shown previously that rats with behaviorally confirmed CIPN due to paclitaxel and oxaliplatin acquire abnormal spontaneous discharge (Xiao and Bennett, 2008; Xiao et al., 2012), but there are no reports as to whether this also occurs with bortezomib. Here, we found that rats with confirmed bortezomib-induced CIPN and neuropathic pain also had a significantly increased incidence of spontaneous discharge in A-fibers and C-fibers (Fig. 7). Similar to our studies with paclitaxel and oxaliplatin (Xiao and Bennett, 2008; Xiao et al., 2012), spontaneous discharge in A-fibers and C-fibers was rare or absent in vehicle-treated rats. The spontaneous discharge in bortezomib-treated rats had an irregular pattern and low frequency (0.5–2.0 Hz) that was nearly identical to that reported that reported with paclitaxel or oxaliplatin treatment in rats (Xiao and Bennett, 2008; Xiao et al., 2012). This increase in spontaneous discharge in both A-fibers and C-fibers of animals treated with bortezomib was significantly reduced in rats coadministered with MRS7154 (Fig. 7).
A3AR agonists blocked the development of abnormal spontaneous discharge in primary sensory afferent A-fiber and C-fiber axons of rats with bortezomib-induced CIPN. Bortezomib treatment caused an increase in the incidence of abnormal spontaneous discharge in both A-fibers and C-fibers. This was significantly decreased in rats cotreated with MRS7154 (0.1 mg/kg/d, i.p.). Results are expressed as mean ± SEM for n = 197 A-fibers and n = 74 C-fibers and analyzed by two-tailed, one-way ANOVA with Dunnett's comparisons; F(2,15) = 13.38, p = 0.00046, η2p = 0.64. *p < 0.05 versus Veh + Veh group and †p < 0.05 versus Chemo + Veh group.
A3AR agonists prevent chemotherapy-induced neuropathic pain
Rats treated with paclitaxel or oxaliplatin (Fig. 8A) exhibited mechano-allodynia (Fig. 8B) and mechano-hyperalgesia (Fig. 8C) on Day 35 (D35) as expected. Systemic administration of MRS5698 (0.1 mg/kg/d) during chemotherapy (Fig. 8A) prevented the development of these neuropathic pain behaviors (Fig. 8B,C), consistent with our previous reports using IB-MECA (Janes et al., 2014). Bortezomib treatment also produced mechano-allodynia (Fig. 8B,D), mechano-hyperalgesia (Fig. 8C,E), and cold allodynia (Fig. 8F) with a time course like that found previously (Zheng et al., 2012). These pain behaviors were inhibited in rats given an A3AR agonist during bortezomib treatment (Fig. 8B–F).
A3AR agonists attenuated chemo-induced neuropathic pain in rats. A, Treatment and testing timeline. When compared with the control group, animals receiving paclitaxel, oxaliplatin, or bortezomib exhibited mechano-hypersensitivity to 4 g (B) and 15 g (C) Von Frey (VF) filaments at a time of peak pain severity on Day 35 (D35). Systemic administration of MRS5698 (0.1 mg/kg/d, i.p.) during treatment with chemotherapeutic agents prevented the development of mechano-allodynia (B) and mechano-hyperalgesia (C). D–F, The development of bortezomib-induced (D) mechano-allodynia, (E) mechano-hyperalgesia, and (F) cold allodynia was attenuated in rats that were cotreated with MRS7154 (0.1 mg/kg/d, i.p.). Results are expressed as mean ± SEM for n = 10/group. Data were analyzed by one-tailed (B, C) unpaired t test or (D–F) Friedmann test with Dunn's comparisons (time) and Mann–Whitney (treatment). B, Chemo + Veh versus Veh + Veh: paclitaxel, t(18) = 4.191, p = 0.0003, d = 1.9; oxaliplatin, t(18) = 4.465, p = 0.00015, d = 2.0; bortezomib, t(18) = 4.450, p = 0.00015, d = 2.0. Chemo/Veh + MRS5698 versus Chemo/Veh + Veh: vehicle, t(18) = 1.134, p = 0.1359, d = 0.51; paclitaxel, t(18) = 5.399, p = 1.98 × 10−5, d = 2.4; oxaliplatin, t(18) = 4.732, p = 8.31 × 10−5, d = 2.1; bortezomib, t(18) = 4.031, p = 0.0004, d = 1.8. C, Chemo + Veh versus Veh + Veh: paclitaxel, t(18) = 6.60, p = 1.7 × 10−6, d = 3.0; oxaliplatin, t(18) = 6.040, p = 5.2 × 10−6, d = 2.7; bortezomib, t(18) = 4.450, p = 0.00015, d = 2.9. Chemo/Veh + MRS5698 versus Chemo/Veh + Veh: vehicle, t(18) = 0.4174, p = 0.3407, d = 0.19; paclitaxel, t(18) = 6.442, p = 2.0 × 10−6, d = 2.9; oxaliplatin, t(18) = 5.366, p = 2.1 × 10−5, d = 2.4; bortezomib, t(18) = 6.483, p = 2.1 × 10−6, d = 2.7. D, F(5,110) = 11.36, p = 1.2 × 10−9, η2p = 0.55. E, F(5,110) = 8.135, p = 1.5 × 10−6, η2p = 0.41. F, Bort + Veh (time), H(6,12) = 41.03, p = 9.2 × 10−8, w = 8.2; Bort + MRS7154 (time), H(6,12) = 10.89, p = 0.054, w = 2.2. Bort + Veh versus Bort + MRS7154: Day 0, U = 10.89, p = 0.011; Day 7, U = 69, p = 0.8724; Day 14, U = 53, p = 0.2831; Day 21 U = 52.5, p = 0.2682; Day 28 U = 28, p = 0.0087; Day 35 U = 66.5, p = 0.7633. *p < 0.05 versus vehicle group, #p < 0.05 versus day 0 and †p < 0.05 versus Chemo + Veh group.
Discussion
Our findings identified the presence of A3AR on mitochondria that previously had been only associated with the plasma membrane and demonstrated their activation supports ATP production in ex vivo peripheral sensory afferent mitochondria compromised by chemotherapy. In CIPN, such persistent energy deficits are proposed to result in the IENF degeneration (Siau et al., 2006; Bennett et al., 2011; Xiao et al., 2012) and compromised sodium–potassium pump function in axons that lead to depolarization and abnormal spontaneous discharge (Xiao and Bennett, 2007, 2008). Dysfunction in peripheral sensory afferent mitochondria is also linked to a switch to glycolysis that can exacerbate neuronal bioenergetic deficits (Duggett et al., 2017) and produce by-products that potentiate voltage-gated sodium channels and trigger proton-sensitive ion channels, both of which would contribute to abnormal firing and neuronal hypersensitivity (Ludman and Melemedjian, 2019). As such, engagement of mitochondrial A3AR signaling and protecting ATP production during chemotherapy would be expected to prevent peripheral neuropathy from developing. Our in vivo experiments indeed demonstrated that administration of A3AR agonists during chemotherapy prevented deficits in ATP production from occurring in peripheral nerve sensory afferent mitochondria that was accompanied by reduced IENF loss, lower incidence of abnormal spontaneous discharge in primary sensory afferent A- and C-axons, and no development of pain. This is potentially transformative in that we now have a class of drugs that not only attenuates the neuroinflammatory events in the CNS responsible for the central sensitization associated with painful CIPN (Janes et al., 2014), but whose target receptor is located on mitochondria and prevents the causative peripheral neuropathy from developing in the first place. Current analgesics only relieve the pain once symptomatic CIPN is established (Hershman et al., 2014), at which point it is probable that substantial and long-lasting nerve injury has already occurred.
Several GPCRs have been identified within mitochondria and regulate mitochondrial function (Belous et al., 2006; Benard et al., 2012; Valenzuela et al., 2016; Wang et al., 2016; Suofu et al., 2017). The mitochondrial Gαs-coupled serotonin 5-HT4 receptor reduces mitochondrial calcium uptake, respiration, and opening of the mitochondrial permeability transition pore, which can lead to mitochondrial swelling, cytochrome c release, and apoptosis (Wang et al., 2016). The mitochondrial Gαi-coupled melatonin MT1 (Suofu et al., 2017) and cannabinoid CB1 (Benard et al., 2012) receptors reduce mitochondrial respiration and demonstrate agonist-induced mitoprotective effects via inhibition of adenylyl cyclase, reduced cytochrome c release, and reduced nitro-oxidative stress. Gαq-coupled receptors such as angiotensin AT1 (Valenzuela et al., 2016) and purine P2Y1 (Belous et al., 2006) and P2Y2 (Belous et al., 2006) receptors have also been identified in mitochondria. AT1 activity in mitochondria increases superoxide production and mitochondrial respiration (Valenzuela et al., 2016), whereas activation of mitochondrial P2Y1 and P2Y2 results in mitochondrial calcium uptake that may increase respiration and/or increase mitochondrial permeability transition pore opening (Belous et al., 2006).
The possibility of signaling through both intracellular adenosine receptors and P2Y receptors at the mitochondrial level is interesting mechanistically, because the corresponding agonists are present at significant concentrations in every cell. Sensing the local fluctuations of ATP and adenosine concentrations through these GPCRs would provide mitochondria with a rapid refined physiological control of their response to bioenergetic demands of cells, whereas severe or sustained imbalances in their signaling would contribute to pathological mitochondrial dysfunction. We have found that adenosine kinase (ADK) expression increases in the spinal cord following chemotherapy and its inhibition attenuated neuropathic pain in animal models of CIPN (Wahlman et al., 2018). Studies investigating adenosine signaling in peripheral neurons also indicate similar dysregulation of adenosine metabolism (Sawynok et al., 1998). Increased ADK activity lowers the intracellular adenosine concentrations by converting adenosine to adenosine monophosphate (Boison, 2013). This lowers intracellular adenosine levels driving extracellular adenosine down its gradient through equilibrative nucleoside transporters 1 and 2 (ENTs) to raise intracellular adenosine (Boison, 2013). A3AR on mitochondria may be positioned to sense such changes in adenosine levels and offer protection in two ways. First, sustained reductions in adenosine levels by increased ADK could prevent mitochondrial A3AR-dependent maintenance of mitochondrial ATP production, which would lead to mitochondrial dysfunction. Second, elevated intracellular adenosine levels also can be detrimental to mitochondrial function (Ko et al., 2020). In lung endothelial cells, sustained elevated intracellular adenosine has been shown to increase production of nitro-oxidative species in the mitochondria and lower basal and maximal, spare capacity, and ATP production (Ko et al., 2020). These effects were relieved by blocking ENT transport of adenosine into the cell (Ko et al., 2020). Mitochondrial A3AR may be positioned to protect mitochondria from increased flow of adenosine down its gradient when ADK activity is high.
The mechanisms within the mitochondria engaged by mitochondrial A3AR activation that protect ATP production remain elusive for now. As a Gαi-coupled receptor, A3AR signaling would be expected to reduce respiration similar to that reported for mitochondrial MT1 (Suofu et al., 2017) and cannabinoid CB1 (Benard et al., 2012) receptors. However, A3AR has been reported to also couple to Gαq and stimulate intracellular calcium mobilization (Borea et al., 2009). Moreover, despite indications of low expression of A3AR mRNA in DRG neurons from single-cell RNA-seq databases (Sharma et al., 2020; Bhuiyan et al., 2023), a recent electrophysiological study on DRG from animals with CIPN indicated that A3AR protein was expressed in neurons and functionally active through inhibiting N-type voltage-gated calcium channel activity (Coppi et al., 2019). Dysregulated calcium levels in peripheral nerve axons have been reported with chemotherapy (Siau and Bennett, 2006), and pharmacologically reducing intracellular or extracellular calcium has been found to attenuate hyperalgesia and allodynia provoked by paclitaxel or vincristine (Siau and Bennett, 2006). Mitochondria are major contributors to the clearance of intracellular calcium (Brini et al., 2014). Calcium uptake increases mitochondrial ATP production (Brini et al., 2014). Thus, Gαq signaling may be one mechanism by which mitochondria A3AR protect mitochondrial function. Our findings do not exclude possible contributions from activation of A3AR at the plasma membrane. Separating and identifying specific contributions from A3AR signaling from the membrane and A3AR at the mitochondrial as well as the mechanisms engaged within the mitochondria will require more investigation.
Our preclinical studies here suggest A3AR agonists when used as an adjunct to chemotherapy could prevent peripheral neuropathy and ensuing pain. However, protection of mitochondrial respiration would seem superficially counter-productive in cancer patients where retention of high mitochondrial respiration in tumor cells has been shown to be a significant factor in cancer progression (Zong et al., 2016). Our previously published in vitro data suggest that A3AR agonists will not interfere with the antitumor actions of chemotherapeutics in colon carcinoma, breast cancer, and multiple myeloma cell lines (Chen et al., 2012), while others have shown A3AR agonists exhibit beneficial antitumor function (Fishman et al., 2012). The derivative of IB-MECA, Cl-IB-MECA (CF102, Namodenoson), completed phase II clinical trials for hepatocellular carcinoma with promising results (Stemmer et al., 2021) and began phase III trials in 2023 (ClinicalTrials.gov, NCT05201404). Consequently, the use of A3AR agonists during chemotherapy could prove to be a viable strategy to prevent the development of CIPN and the resulting neuropathic pain condition. This would have a substantial impact on improving the health and quality of life in patients by obviating the need for dose reduction or discontinuation of life-saving chemotherapies. Ongoing studies within our research group are validating the effects of A3AR agonists on tumor growth and CIPN in tumor-bearing animal models. Future studies in females are also necessary. While the protective effects of mitochondrial A3AR signaling would be anticipated in females receiving oxaliplatin or paclitaxel based on our previous findings of no sex-dependent differences in the effects of A3AR agonists in animals treated with these chemotherapeutic agents (Stockstill et al., 2020), there is the potential for sex-dependent differences in mitochondrial A3AR signaling in individuals receiving bortezomib (Stockstill et al., 2020).
The ability of A3AR agonists to maintain mitochondrial function and prevent peripheral neuropathy may have broader application in other painful disorders where mitochondrial function is also impaired, such as diabetic, antiretroviral-induced, and nucleoside/nucleotide reverse transcriptase inhibitor-induced neuropathies (Bennett et al., 2014) and given the presence of mitochondrial A3AR in central nervous tissue and glia, to nonpainful neurodegenerative conditions where mitochondrial dysfunction has been implicated.
Footnotes
We thank our colleagues at Saint Louis University, St. Louis, MO, for the following: Grady Phillips for the electron microscopy processing and imaging, Zhoumou Chen for assistance in behavior testing and tissue harvesting, and Carrie Wahlman for obtaining saphenous nerves from oxaliplatin animals. This work was funded by the National Institutes of Health/National Cancer Institute (NIH R01 CA169519 and R01 CA230512 to D.S.) and the National Institute of Diabetes and Digestive and Kidney Diseases Intramural Research Program (Z01 DK031117-26 to K.A.J.).
D.S. and G.J.B. are co-founders of BioIntervene Inc., which has licensed the A3AR agonists used in this study.
- Correspondence should be addressed to Daniela Salvemini at daniela.salvemini{at}health.slu.edu.
