Macrophage-derived Exosomes Regulate Neuronal ACSL6 in Experimental Temporomandibular Disorder Pain

Abstract

Background/Aim

Temporomandibular disorders (TMD) are frequently accompanied by persistent pain, but the underlying mechanisms remain incompletely understood. Macrophage-neuron communication in the trigeminal ganglion (TG) may contribute to TMD pain. This study investigated whether macrophage-derived exosomes (EXOs) promote neuronal inflammation and pain by regulating acyl-CoA synthetase long-chain family member 6 (ACSL6).

Materials and Methods

A mouse model of TMD pain was established by intra-articular injection of complete Freund’s adjuvant (CFA) into the temporomandibular joint. Pain behavior, TG histopathology, inflammatory mediator expression, and ACSL6 levels were evaluated. The effects of clodronate liposome-mediated macrophage depletion and GW4869-mediated inhibition of EXO release were examined in vivo. In vitro, conditioned medium (CM) and EXOs derived from tumor necrosis factor-alpha-activated RAW264.7 cells were applied to primary TG neurons and PC12 cells. GW4869 treatment of RAW264.7 cells, ACSL6 overexpression, and docosahexaenoic acid (DHA) supplementation were used for further functional evaluation.

Results

CFA reduced the head withdrawal threshold and increased TG inflammation, macrophage activation, and expression of interleukin-6, cyclooxygenase-2, and substance P, while decreasing ACSL6 expression. Macrophage depletion and systemic GW4869 treatment alleviated mechanical hypersensitivity, reduced inflammatory mediator expression, and partially restored ACSL6 levels. Activated macrophage-derived CM and EXOs decreased neuronal ACSL6 expression and enhanced inflammatory responses. Inhibition of EXO release by GW4869 attenuated these effects in primary TG neurons and PC12 cells. ACSL6 overexpression and DHA supplementation independently reduced macrophage-induced neuronal inflammatory responses.

Conclusion

Macrophage-derived EXOs contribute to CFA-induced TMD pain and neuronal inflammation and are associated with reduced neuronal ACSL6 expression. ACSL6 may represent a functionally relevant downstream regulator, whereas DHA exerts an independent protective anti-inflammatory effect.

Introduction

Temporomandibular disorder (TMD) refer to a series of disorders associated with pain and functional limitations in the maxillofacial area, primarily caused by pathological changes in the masticatory muscles, temporomandibular joint, and surrounding tissues (1, 2). Epidemiological studies have reported a high prevalence of TMD across all age groups (2). Pain is the primary complaint among patients with TMD and can impair mandibular function, reduce masticatory efficiency, and negatively affect mood and motivation for daily activities (3). Accordingly, substantial efforts have been made worldwide to identify effective treatment strategies. Despite the use of conventional non-surgical treatments, robust evidence for durable analgesic benefits remains scarce, likely due to an incomplete understanding of the biological mechanisms underlying TMD pain.

The trigeminal ganglion (TG) is the primary cluster of sensory neuronal cell bodies in the trigeminal nerve pathway. It receives peripheral sensory input from the oral and maxillofacial regions and transmits this information to the central nervous system, thereby maintaining normal sensory function in the head and face (4). Previous studies have identified pathological alterations in the TG under painful conditions, including widened interneuronal spaces and disorganized nerve fibers (5). Under disease conditions, disruption of cytokine homeostasis in the nervous system can drive the dysregulated release of mediators from immune cells, neurons, and other cell types (6, 7). In a chronic constriction injury model of the infraorbital nerve, interleukin-6 (IL-6) expression in the TG was significantly upregulated (8, 9). In addition, cyclooxygenase-2 (COX-2), a key mediator of peripheral sensitization, shows expression levels that are positively correlated with pain scores in intervertebral disc tissues from patients with lumbar disc herniation (10). Moreover, under inflammatory stimulation, nociceptive neurons in the TG synthesize and release large amounts of substance P (SP), further enhancing pain signal transmission at the spinal cord level (11).

Within the TG microenvironment, resident immune cells are key contributors to the modulation of sensory responses triggered by exogenous signals (12, 13). Recent findings support the contributory role of sensory ganglion macrophages in somatic pain (14-16). In models of arthritis, peripheral nerve injury, and orofacial pain, significant increases in macrophage infiltration and accumulation within the ganglia have been observed (17-21). Dynamic interactions between macrophages and neurons constitute a central hub of neuroimmune regulation, and such cross-cellular communication has been documented in multiple diseases (22, 23).

Exosomes (EXOs) are extracellular vesicles released by various cell types, with diameters ranging from approximately 30 to 150 nm (24). They carry a variety of bioactive substances, and play an important role as mediators of intercellular communication. EXOs participate in the regulation of various physiological and pathological activities (25-27). Recent evidence further indicates that EXOs can modulate the local pathological microenvironment and influence disease progression or therapeutic responses. For example, Choi et al. reported that co-administration of Wharton’s jelly-derived mesenchymal stem cells and their EXOs exerted greater anti-fibrotic effects than either treatment alone in experimental pulmonary fibrosis, suggesting a modulatory role of EXOs within the diseased tissue microenvironment (28).

Alterations in neuronal lipid metabolism may also contribute to inflammatory neuronal dysfunction. ACSL6, a member of the long-chain acyl-CoA synthetase family, catalyzes the activation of long-chain fatty acids and is closely associated with neural lipid homeostasis. Previous studies have shown that ACSL6 is required for DHA enrichment and retention in the brain and promotes membrane DHA enrichment in neurons (29). Given that DHA is an abundant polyunsaturated fatty acid in neuronal membranes and a substrate for specialized pro-resolving mediators, ACSL6 may connect neuronal lipid remodeling with the control of inflammatory responses (30). Therefore, among the candidate genes identified by bioinformatic screening, ACSL6 was selected for further investigation because of both its differential expression and its biological relevance to neuronal lipid metabolism and DHA-related neuroprotection.

This study aimed to elucidate the role of macrophage-derived EXOs in the pathogenesis of TMD pain. First, the model of TMD pain was induced by intra-articular injection of complete Freund’s adjuvant (CFA), and pain-related behavioral changes were assessed in the mice. The expression of macrophage- and neuron-associated effector molecules in the TG was then examined. Macrophages were subsequently depleted to assess their regulatory role in neuronal function. In parallel, an in vitro cell-based experimental system was used to further analyze the effects of EXOs secreted by activated macrophages on neuronal function.

Materials and Methods

Animals. To minimize the variability in pain responses associated with the estrous cycle, 8-week-old male C57BL/6J mice supplied by Liaoning Changsheng Biotechnology Co., Ltd. (Liaoning, PR China) were used in this study. The animals were housed under strictly controlled environmental conditions in accordance with laboratory animal facility standards, with the room temperature maintained at 23-25°C, a 12 h light/12 h dark cycle, and relative humidity of approximately 50%. All animal experiments were performed under the relevant institutional guidelines for laboratory animal care and welfare, with ethical approval granted by the Animal Ethics Committee of Anhui Medical University (LLSC20221097).

CFA-induced TMD pain. 12 C57BL/6J mice were randomized into normal control (NC) and CFA groups (n=6 per group). Mice were anesthetized using the animal anesthesia protocol described previously (4). After isoflurane anesthesia, mice in the CFA group received bilateral intra-articular injections of 10 μl of CFA (1 mg/ml heat-inactivated Mycobacterium tuberculosis; Biosharp, Beijing, PR China) into the temporomandibular joint (TMJ), as described by Xu et al. (31). The NC group received bilateral intra-articular administration of isotonic saline.

Infiltrating macrophage depletion. To deplete the infiltrating macrophages, mice received local injections of 10 μl clodronate liposomes (CLs; Yeasen, Shanghai, PR China) into the TG via the infraorbital foramen, following the method described by Neubert (32). Phosphate-buffered saline (PBS) liposomes (Cont-CLs; Yeasen) were used as controls. The efficiency of macrophage depletion was subsequently validated.

Using the established TMJ inflammatory pain model, 18 mice were randomly assigned to 3 groups (n=6 per group): the CFA-Vehicle group received 10 μl sterile saline; the CFA-Cont-CLs group received blank liposomes at the same volume; and the CFA-CLs group received an equal volume of CLs. All injections were performed using the same intra-TG injection procedure.

Intraperitoneal administration of GW4869. GW4869 (MCE, Shanghai, PR China), a neutral sphingomyelinase 2 inhibitor commonly used to suppress EXO release, was used for EXO inhibition experiments. To inhibit extracellular vesicle release in vivo, mice received an intraperitoneal dose of GW4869 at 2.5 mg/kg (33). GW4869 was initially dissolved in dimethyl sulfoxide (DMSO) to prepare a stock solution and was subsequently diluted with sterile physiological saline immediately before administration. Mice assigned to the vehicle control groups were administered a matched volume of carrier solution formulated with an identical final DMSO concentration.

Using the established CFA-induced TMD pain model, mice were allocated at random among three experimental conditions, with six animals in each: NC, CFA, and CFA +GW4869. The first dose of GW4869 or vehicle was administered 1 h before CFA injection. Thereafter, GW4869 or vehicle was administered once daily until day 2 after CFA injection.

Intraperitoneal administration of docosahexaenoic acid (DHA). After establishment of the TMD model, mice in the DHA-treated group received intraperitoneal injection of DHA (300 mg/kg; MCE, China), with corn oil used as the vehicle. Vehicle-treated mice received an equal volume of corn oil. DHA was administered immediately after TMD modeling and once again 6 h before the Von Frey test on day 3. Intraperitoneal injection was performed through the lower abdominal quadrant after routine disinfection.

Measurement of head withdrawal threshold (HWT). To assess HWT, the experimenter placed the mouse on a palm, restricting its hind limbs while leaving its forelimbs and head free. A rigid probe attached to an electronic von Frey device was used to stimulate the skin over the TMJ of the mouse, and numerical values were recorded when the mice exhibited squeals, avoidance, scratching, or aggressive responses. Each mouse underwent 5 tests on each side with several seconds between consecutive trials. The average of the resulting 10 readings was used for analysis. Before CFA injection, HWT was first assessed on day 1 to obtain baseline data, and additional measurements were performed on days 1, 3, 5, 7, and 9 thereafter. To ensure consistency in behavioral response assessments, an investigator who was unaware of the group assignments conducted every behavioral assessment.

Hematoxylin and eosin (HE) staining. 3 days after model establishment, the TGs were dissected, rinsed with physiological saline, and immediately cryopreserved in liquid nitrogen or fixed in paraformaldehyde for subsequent analyses. The TG samples were fixed for 24 h at 4°C in 4% paraformaldehyde. Following fixation, the samples were dehydrated using a stepwise alcohol gradient, wax-embedded, and sectioned into 4-µm-thick slices. The sections were then dewaxed and stained with HE.

After staining, the sections were dehydrated through a xylene gradient and mounted with neutral resin. Finally, the sections were observed using an optical microscope (Leica, Wetzlar, Germany).

Immunofluorescence staining. Coverslip-cultured cells were treated with 4% paraformaldehyde for 15 min and rinsed 3 times using PBS. Blocking and membrane permeabilization were subsequently carried out with the same buffer. Next, NeuN primary antibody (1:100, Abcam, Cambridge, MA, USA) was applied. The samples were kept at 4°C overnight. After washing 3 times in PBS, fluorescently labeled secondary antibody (1:100, Proteintech, Wuhan, PR China) was added for 2 h at room temperature away from light. The coverslips were subsequently mounted using an antifade medium containing DAPI and examined with a fluorescence microscope (Motic, Xiamen, PR China).

RNA sequencing (RNA-seq) and data analysis. The GSE207126 (34) and GSE188922 (35) datasets were downloaded from the Gene Expression Omnibus (GEO) database. Gene-expression analysis was performed separately for each dataset using the limma package in R. Nominal P-values were adjusted for multiple testing using the Benjamini-Hochberg method to obtain false discovery rate (FDR) values. For exploratory candidate-gene screening, genes with an absolute fold change >4 and a nominal P-value <0.05 were retained. Candidate genes showing concordant changes in both datasets were identified by intersection analysis. The corresponding nominal P-values and FDR values are provided in Table I. Volcano plots were generated using the ggplot2 package. Shared upregulated and downregulated genes between the two GEO datasets were identified using Venn diagram analysis.

Table I. Statistical characteristics of shared candidate genes identified by exploratory analysis of GSE207126 and GSE188922.

Isolation and identification of EXOs. RAW264.7 cell culture supernatants were subjected to gradient centrifugation at 4°C using a high-speed centrifuge as described previously (36). The post-centrifugation pellet was collected as EXOs and resuspended in 100 μl PBS.

The isolated EXOs were subjected to ultrastructural observation using transmission electron microscopy (TEM; Thermo Talos L120C G2, Waltham, MA, USA), which revealed a typical cup-shaped morphology. Particle size distribution was analyzed using nanoparticle tracking analysis (NTA; Particle Metrix ZetaView, Inning am Ammersee, Germany). Briefly, EXO samples diluted in PBS were loaded into the chamber, and the particle size and concentration were recorded automatically.

Labeling and tracking of EXO uptake. EXOs were labeled with dioctadecyloxacarbocyanine perchlorate (DiO; Aladdin, Shanghai, PR China) according to the manufacturer’s instructions. The labeled EXOs were centrifuged and washed to remove excess dye and then incubated with PC12 cells and primary TG neurons for 24 h. After fixation, the cells were counterstained with DAPI and imaged using an inverted fluorescence microscope (Carl Zeiss, Oberkochen, Germany).

Enzyme-linked immunosorbent assay (ELISA). ELISA (Byabscience, Nanjing, PR China) was used to quantify tumor necrosis factor-alpha (TNF-α) in TG tissues and EXO fractions. The TG tissue was homogenized five times at 4°C and 70 Hz for 2 min. After centrifugation, the pellet was stored at -80°C for at least 1 h. The TG homogenate was thawed on ice and centrifuged to collect the supernatant. All ELISA procedures were performed in accordance with the manufacturer’s instructions.

Primary TG neuron culture. For each independent neuronal isolation, bilateral TGs from 4 mice were pooled to establish one primary neuronal culture. TG tissues were collected in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, Grand Island, NY, USA). After washing with PBS containing 2% penicillin/streptomycin (P/S), the TG tissues were minced into approximately 1 mm³ fragments and digested using type I collagenase (2 mg/ml) at 37°C for 20 min, followed by digestion with deoxyribonuclease I (DNase I; Thermo Fisher Scientific, Waltham, MA, USA) for 10 min.

The pellet obtained after centrifugation was suspended in TrypLE Select enzyme (Gibco) and incubated at 37°C for 15 min. Subsequently, the cells were centrifuged and redispersed in high-glucose DMEM containing 10% fetal bovine serum (FBS; Gibco) and 1% P/S, and allowed to rest for 2-3 min. The supernatant was collected and centrifuged again. The final pellet was resuspended in serum-free neuronal medium supplemented with 2% B27 (Gibco) and 1% P/S and seeded onto plates coated with 0.1 mg/ml poly-L-lysine. Half of the culture medium was replaced every 48 h using the same formulation.

Cell treatment. RAW264.7 cells were obtained from the laboratory cell stock. RAW264.7 cells were pretreated with 50 ng/ml TNF-α (PeproTech, Cranbury, NJ, USA) for 24 h. Subsequently, the cells were washed and cultured for an additional 24 h in a complete medium containing 10% EXO-depleted FBS (Gibco Life Technologies, Grand Island, NY, USA). Culture supernatants were collected for EXO isolation.

Control and TNF-α-treated RAW264.7 cells were cultured for 24 h in EXO-depleted medium with or without 10 μM GW4869. The conditioned media were collected and subjected to stepwise centrifugation to eliminate cells and cellular debris. Specifically, the media were centrifuged at 1,000 × g for 10 min, followed by a second centrifugation of the transferred supernatants at 10,000 × g for 15 min. The obtained supernatants were defined as NC-CM, NC-CM + GW4869, TNF-α-CM, and TNF-α-CM + GW4869.

Transfection of PC12 cells with plasmid. PC12 cells were purchased from Nuozhuo Biotechnology Co., Ltd. (Hefei, PR China). The pEGFP-C1-ACSL6 overexpression plasmid was supplied by Sangon Biotech (Shanghai, PR China) and transfected into PC12 cells in 6-well plates at approximately 60% confluence. For transfection, 2 μg plasmid DNA and 2 μl P3000 reagent (Thermo Fisher Scientific) were mixed with 125 μl Opti-MEM (Gibco), and 2 μl Lipofectamine 3000 reagent (Lip3000; Thermo Fisher Scientific) was separately mixed with 125 μl Opti-MEM.

The mixtures were combined and incubated at 37°C for 10 min, added to 2 mL RPMI-1640 medium (Gibco), mixed thoroughly, and applied to pre-cultured cells. At 12 h after transfection, the culture medium was changed to complete medium, and the cells were collected 24 h after transfection for subsequent experiments.

DHA treatment of neuronal cells. Primary TG neurons and PC12 cells were cultured under the conditions described above. Complete medium and TNF-α-CM were added at a ratio of 1:1, or 15 μg/ml of TNF-α-EXO was added to complete medium, and 50 μmol/l DHA was added to the experimental group.

Western blot. Tissue and cell proteins were prepared, and western blotting was performed following the protocol established in our previous study (37). Protein band intensity was evaluated using ImageJ software. The primary antibodies used were as follows: anti-Iba1 (1:1,000; HUABIO, Hangzhou, PR China), anti-ACSL6 (1:1,000; Abcam), anti-β-actin (1:3,000; Proteintech), anti-COX-2 (1:1000; MCE), anti-calnexin (1:1,000; Proteintech), anti-CD9 (1:1,000; Proteintech), anti-HSP70 (1:1,000; Proteintech), and anti-TSG101 (1:1,000; Proteintech). Horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (1:7,500; ZSGB, Beijing, PR China) and HRP-conjugated goat anti-mouse IgG (1:7,500; ZSGB) were used as secondary antibodies.

Quantitative real-time polymerase chain reaction (qRT-PCR). Total RNA extraction from tissues/cells and qRT-PCR were performed as previously described (37). The primer information is provided in Table II.

Table II. Primer sequences for the target genes.

Statistical analysis. Biological and technical replicates were defined according to the experimental unit. For in vivo experiments, each individual mouse was considered one biological replicate. For experiments using primary TG neurons, each independently prepared primary neuronal culture obtained from a separate neuronal isolation was considered one biological replicate. For RAW264.7 and PC12 cell experiments, independently initiated experiments were considered biological replicates. Multiple culture wells derived from the same cell preparation, as well as replicate measurements of the same sample, were considered technical replicates. Technical replicates were averaged to generate a single value for each biological replicate before statistical analysis. Statistical analyses were performed using biological replicates only.

Statistical analyses were performed using GraphPad Prism 10.0 (GraphPad Software, San Diego, CA, USA). Data are expressed as mean±standard deviation (SD). Differences between two groups were analyzed using an unpaired Student’s t-test, whereas experiments involving at least three groups were assessed by one-way analysis of variance. Behavioral outcomes were examined through a two-factor repeated-measures ANOVA. The Geisser-Greenhouse adjustment was applied, and subsequent pairwise comparisons were conducted with Šidák correction. Statistical significance was defined as p<0.05.

Results

Macrophage depletion and GW4869 treatment attenuate CFA-induced TMD pain and TG inflammation. Following CFA injection, the HWT in the TMJ region decreased on day 1, reached its lowest level on day 3, and remained reduced through day 7. By day 9, the HWT had gradually returned to a level similar to the baseline value (Figure 1A). Based on this temporal profile, day 3 after CFA injection was selected for subsequent experiments. Histological examination of TG tissue revealed widened intercellular spaces, neuronal shrinkage, and accumulation of surrounding satellite cells in the CFA group (Figure 1B). In addition, the expression levels of Iba1 were increased in the TG of CFA-treated mice (Figure 1C, D). The expression levels of IL-6 and COX-2 were also increased, and ELISA showed elevated TNF-α levels in TG tissue (Figure 1E-G).

Figure 1.

CFA injection induced TMD pain and neuronal injury. (A) HWT, expressed as a percentage of baseline, in the NC and CFA groups from days 1 to 9 after CFA injection, n=6 mice per group. (B) H&E staining was used to assess histopathological changes in TG tissues, bar=20 μm. (C) Relative mRNA expression of Iba1 in TG tissues, n=6 mice per group. (D) Relative protein expression of Iba1 in TG tissues was measured by western blotting, n=3 mice per group. (E, F) Relative mRNA expression levels of IL-6 and COX-2 in TG tissues, n=6 mice per group. (G) TNF-α levels in TG tissues, n=6 mice per group. Data are presented as the mean ± SD. *p<0.05, **p<0.01 and ***p<0.001 versus the NC group. HWT: Head withdrawal threshold; NC: normal control; CFA: complete Freund’s adjuvant; TG: trigeminal ganglion; H&E staining: hematoxylin and eosin staining; Iba1: ionized calcium-binding adaptor molecule 1; IL-6: interleukin-6; COX-2: cyclooxygenase-2; TNF-α: tumor necrosis factor-alpha.

To clarify the contribution of TG macrophages to CFA-induced TMD pain, CLs were locally injected into the TG during model establishment (Figure 2A, B). No significant difference in HWT was observed between the CFA-Vehicle and CFA-Cont-CLs groups. In contrast, HWT was significantly increased in the CFA-CLs group compared with both control groups (Figure 2C). Macrophage depletion was confirmed by the reduction in Iba1 mRNA and protein expression (Figure 2D, E). Concurrently, the expression levels of IL-6 and COX-2 were significantly decreased in the CFA-CLs group (Figure 2F, G).

Figure 2.

Macrophage depletion and GW4869 treatment attenuated CFA-induced TMD pain and TG inflammation. (A) Schematic representation of the injection needle insertion position within the TG. (B) The TG dissected and exposed after the injection. (C) HWT (percentage of baseline) on day 3 in the CFA-Vehicle, CFA-Cont-CLs, and CFA-CLs groups, n=6 mice per group. (D) Relative mRNA expression of Iba1 in TG tissue, n=6 mice per group. (E) Relative protein expression of Iba1 in TG tissue, n=6 mice per group. (F and G) Relative mRNA expression of IL-6 and COX-2 in TG tissue, n=6 mice per group. (H) HWT (percentage of baseline) on day 3 in the NC, CFA, and CFA + GW4869 groups, n=6 mice per group. (I) Relative mRNA expression levels of IL-6, COX-2, and SP in TG tissue from the NC, CFA, and CFA + GW4869 groups, n=6 mice per group. All data are presented as mean±SD. ns indicates no statistical significance. *p<0.05, **p<0.01, ***p<0.001 and ****p<0.0001 versus CFA group or CFA-Vehicle group. ^#p<0.05, ^##p<0.01, ^###p<0.001 and ^####p<0.0001 versus the CFA-Cont-CLs or NC group. HWT: Head withdrawal threshold; TG: trigeminal ganglion; CFA: complete Freund’s adjuvant; Vehicle: vehicle control; Cont-CLs: control liposomes; CLs: clodronate liposomes; NC: normal control; Iba1: ionized calcium-binding adaptor molecule 1; IL-6: interleukin-6; COX-2: cyclooxygenase-2; SP: substance P; GW4869: neutral sphingomyelinase 2 inhibitor used to suppress exosome release.

To further assess whether EXO-associated signaling contributed to CFA-induced pain in vivo, GW4869 was administered intraperitoneally to CFA-treated mice. On day 3, GW4869 treatment significantly increased HWT in CFA-injected mice relative to animals receiving CFA alone (Figure 2H). In addition, GW4869 treatment reduced the expression levels of IL-6, COX-2, and SP in TG tissue (Figure 2I).

Collectively, these findings indicate that TG macrophage activation and EXO-associated signaling contribute to CFA-induced mechanical hypersensitivity and inflammatory responses in the TG.

TNF-α-activated RAW264.7-derived EXOs promote neuronal inflammation. EXOs were isolated from the supernatant of RAW264.7 cells using ultracentrifugation and subsequently characterized using TEM. TEM confirmed the presence of vesicles with a characteristic cup-like shape and bilayer membrane (Figure 3A). Particle size analysis using NTA revealed that vesicles were mainly concentrated in the 50-150 nm range, matching the generally accepted size spectrum of EXOs (Figure 3B). Western blotting confirmed the presence of CD9, HSP70, and TSG101 in the isolated vesicles, whereas calnexin was absent (Figure 3C), confirming their EXO identity. Primary TG neurons exhibited typical neuronal morphology and were identified by NeuN immunofluorescence staining (Figure 3D, E). DiO-labelled EXOs were detected in the cytoplasm of both primary TG neurons and PC12 cells after co-culture, indicating that macrophage-derived EXOs were internalized by neuronal cells (Figure 3F).

Figure 3.

Isolation and identification of EXOs and primary TG neurons. (A) Transmission electron micrographs of EXO structure. Scale bar=100 nm. (B) Particle size of EXOs measured using NTA. (C) Protein levels of HSP70, TSG101, CD9, and calnexin on the EXO surface assessed using western blotting. (D) Brightfield microscopic image of cultured primary TG neurons. Scale bar=100 μm. (E) Immunofluorescence identification of TG neuronal cells. Scale bar=100 μm. (F) DiO-labeled EXOs (green) were cultured with TG neuronal and PC12 cells for 24 h. The nuclei of TG neuronal and PC12 cells were stained with DAPI (blue). Scale bar=25 μm. EXOs: Exosomes; NTA: nanoparticle tracking analysis; HSP70: heat-shock protein 70; TSG101: tumor susceptibility gene 101; TG: trigeminal ganglion; NeuN: neuronal nuclei; DAPI: 4′,6-diamidino-2-phenylindole.

Based on the increased TNF-α expression observed in CFA-treated TG tissue, RAW264.7 cells were stimulated with TNF-α to establish an activated macrophage model. TNF-α stimulation significantly increased the mRNA expression levels of IL-6 and IL-1β in RAW264.7 cells (Figure 4A, B). Conditioned medium and EXO were subsequently collected from activated macrophages and applied to primary TG neurons and PC12 cells (Figure 4C). Compared with NC-CM, TNF-α-CM significantly increased the expression levels of IL-6, COX-2, and SP in primary TG neurons (Figure 4D, E). Similar changes were observed after treatment with TNF-α-EXO (Figure 4F, G). Consistent results were also obtained in PC12 cells exposed to TNF-α-CM or TNF-α-EXO (Figure 4H-K).

Figure 4.

EXOs from TNF-α-activated RAW264.7 cells promote inflammation in neurons. (A and B) Relative mRNA expression levels of IL-6 and IL-1β in RAW264.7 cells, n=3 independent biological experiments. (C) RAW264.7 cells were stimulated with 50 ng/ml TNF-α for 24 h to activate RAW264.7 cells, and the cell culture supernatant was collected. One part of the collected supernatant was used to co-culture with mouse TG primary neurons and PC12 cells, and the other part used to extract EXOs by differential ultracentrifugation. The extracted EXOs and mouse TG primary neurons and PC12 cells were co-cultured. (D) Relative mRNA expression of SP, COX-2, and IL-6 in TG neuronal cells co-cultured with TNF-α-CM, n=3 independent neuronal isolations. (E) Relative protein expression of COX-2 in TG neuronal cells after co-culture with TNF-α-CM, n=3 independent neuronal isolations. (F) Relative mRNA expression of SP, COX-2, and IL-6 in TG neuronal cells after co-culture with TNF-α-EXO, n=3 independent neuronal isolations. (G) Relative protein expression of COX-2 in TG neuronal cells after co-culture with TNF-α-EXO, n=3 independent neuronal isolations. (H) Relative mRNA expression of SP, COX-2, and IL-6 in PC12 cells after co-culture with TNF-α-CM, n=3 independent biological experiments. (I) Relative protein expression of COX-2 in PC12 cells after co-culture with TNF-α-CM, n=3 independent biological experiments. (J) Relative mRNA expression of SP, COX-2, and IL-6 in PC12 cells after co-culture with TNF-α-EXO, n=3 independent biological experiments. (K) Relative protein expression of COX-2 in PC12 cells after co-culture with TNF-α-EXO, n=3 independent biological experiments. (L) Relative mRNA expression levels of SP, COX-2, and IL-6 in primary TG neurons treated with NC-CM, NC-CM + GW4869, TNF-α-CM, or TNF-α-CM + GW4869, n=3 independent neuronal isolations. (M) Relative protein expression of COX-2 in TG neuronal cells after co-culture with NC-CM, NC-CM + GW4869, TNF-α-CM, or TNF-α-CM + GW4869, n=3 independent neuronal isolations. (N) Relative mRNA expression levels of SP, COX-2, and IL-6 in PC12 cells treated with the four conditioned media, n=3 independent biological experiments. (O) Relative protein expression of COX-2 in PC12 cells after co-culture with NC-CM, NC-CM + GW4869, TNF-α-CM, or TNF-α-CM + GW4869, n=3 independent biological experiments. All data are presented as mean±SD. *p<0.05, **p<0.01, ***p<0.001 vs. TNF-α, TNF-α-CM, or TNF-α-EXO group. NC: Normal control; CFA: complete Freund’s adjuvant; EXOs: exosomes; TNF-α: tumor necrosis factor-alpha; CM: conditioned medium; TG: trigeminal ganglion; GW4869: neutral sphingomyelinase 2 inhibitor used to suppress EXO release; SP: substance P; COX-2: cyclooxygenase-2; IL-6: interleukin-6; IL-1β: interleukin-1 beta.

To further determine whether the inflammatory effects of activated macrophage-conditioned medium were associated with EXO release, RAW264.7 cells were treated with GW4869 under basal or TNF-α-stimulated conditions before conditioned medium collection. The resulting conditioned media were divided into four groups: NC-CM, NC-CM + GW4869, TNF-α-CM, and TNF-α-CM + GW4869. No significant differences in IL-6, COX-2, or SP expression were observed between the NC-CM and NC-CM + GW4869 groups. Compared with NC-CM, TNF-α-CM significantly increased the expression levels of IL-6, COX-2, and SP in primary TG neurons. These increases were significantly attenuated in the TNF-α-CM + GW4869 group compared with the TNF-α-CM group (Figure 4L-M). Similar changes were observed in PC12 cells, in which GW4869 treatment of TNF-α-stimulated RAW264.7 cells reduced the ability of the conditioned medium to induce IL-6, COX-2, and SP expression (Figure 4N-O).

These findings indicate that EXOs derived from TNF-α-activated macrophages can be internalized by neuronal cells and contribute to macrophage-induced neuronal inflammatory responses.

Activated macrophage-derived EXOs reduce neuronal ACSL6 expression. Exploratory bioinformatic analysis of the GSE188922 and GSE207126 datasets identified genes showing concordant expression changes in the TG following CFA treatment. Using the predefined screening criteria of an absolute fold change >4 and a nominal p<0.05, 15 commonly upregulated and 16 commonly downregulated candidate genes were identified (Figure 5A-D, Table I). The expression levels of 11 neurologically relevant genes were subsequently examined in TG tissue using qRT-PCR (Table III). Among these genes, the expression changes in Astn1, Lypd1, and ACSL6 were consistent with the RNA-seq results (Figure 5E). Based on the reported involvement of ACSL6 in neuronal lipid metabolism and inflammatory regulation, ACSL6 was selected for further investigation. Western blotting showed that ACSL6 protein expression was significantly decreased in the TG of CFA-treated mice compared with the NC group (Figure 5F).

Figure 5.

Exploratory transcriptomic screening and experimental validation of ACSL6 downregulation in the TG following CFA treatment. (A, B) Volcano plots of GSE188922 and GSE207126: The blue dots on the left represent mRNAs with significantly decreased expression, and the red dots on the right represent mRNAs with significantly increased expression in the CFA group. (C, D) Venn diagrams of the commonly upregulated (left) and downregulated (right) DEGs in GSE188922 and GSE207126. (E) Relative mRNA expression levels of Ank3, Anxa3, Gabra1, Slc25a22, Arpp21, Astn1, ACSL6, Frmd5, Grm5, lypd1, and ntng1 in TG tissue, n=6 mice per group. (F) Relative protein expression of ACSL6 in TG tissue, n=3 mice per group. *p<0.05, **p<0.01 vs. NC group. TG: Trigeminal ganglion; CFA: complete Freund’s adjuvant; NC: normal control.

Table III. Neurologically relevant candidate genes identified in GSE207126 and GSE188922.

To determine whether TG macrophages contributed to the regulation of ACSL6, macrophages were locally depleted using CLs. Compared with the CFA-Vehicle and CFA-Cont-CLs groups, the CFA-CLs group exhibited significantly increased ACSL6 expression at both the mRNA and protein levels (Figure 6A, B).

Figure 6.

Activated macrophage-derived EXOs reduced neuronal ACSL6 expression, whereas inhibition of EXO release partially restored ACSL6 expression. (A) Relative mRNA expression of ACSL6 in TG tissue from the CFA-Vehicle, CFA-Cont-CLs, and CFA-CLs groups, n=6 mice per group. (B) Relative protein expression of ACSL6 in TG tissue from the CFA-Vehicle, CFA-Cont-CLs, and CFA-CLs groups, n=3 mice per group. (C) Relative mRNA expression of ACSL6 in TG tissue from the NC, CFA, and CFA + GW4869 groups, n=6 mice per group. (D) Relative protein expression of ACSL6 in TG tissue from the NC, CFA, and CFA + GW4869 groups, n=3 mice per group. (E) Relative mRNA expression of ACSL6 in primary TG neurons after treatment with NC-CM or TNF-α-CM, n=3 independent neuronal isolations. (F) Relative protein expression of ACSL6 in primary TG neurons after treatment with NC-CM or TNF-α-CM, n=3 independent neuronal isolations. (G) Relative mRNA expression of ACSL6 in primary TG neurons after treatment with NC-EXO or TNF-α-EXO, n=3 independent neuronal isolations. (H) Relative protein expression of ACSL6 in primary TG neurons after treatment with NC-EXO or TNF-α-EXO, n=3 independent neuronal isolations. (I) Relative mRNA expression of ACSL6 in PC12 cells after treatment with NC-CM or TNF-α-CM, n=3 independent biological experiments. (J) Relative protein expression of ACSL6 in PC12 cells after treatment with NC-CM or TNF-α-CM, n=3 independent biological experiments. (K) Relative mRNA expression of ACSL6 in PC12 cells after treatment with NC-EXO or TNF-α-EXO, n=3 independent biological experiments. (L) Relative protein expression of ACSL6 in PC12 cells after treatment with NC-EXO or TNF-α-EXO, n=3 independent biological experiments. (M) Relative mRNA expression of ACSL6 in primary TG neurons treated with NC-CM, NC-CM + GW4869, TNF-α-CM, or TNF-α-CM + GW4869, n=3 independent neuronal isolations. (N) Relative protein expression of ACSL6 in primary TG neurons treated with NC-CM, NC-CM + GW4869, TNF-α-CM, or TNF-α-CM + GW4869, n=3 independent neuronal isolations. (O) Relative mRNA expression of ACSL6 in PC12 cells treated with NC-CM, NC-CM + GW4869, TNF-α-CM, or TNF-α-CM + GW4869, n=3 independent biological experiments. (P) Relative protein expression of ACSL6 in PC12 cells treated with NC-CM, NC-CM + GW4869, TNF-α-CM, or TNF-α-CM + GW4869, n=3 independent biological experiments. All data are presented as mean ± SD, ns indicates no statistical significance. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 vs. CFA-Vehicle, CFA, TNF-α-CM or TNF-α-EXO group. ^##p<0.01 and ^###p<0.001 vs. CFA-Cont-CLs group. ACSL6: Acyl-CoA synthetase long-chain family member 6; TG: trigeminal ganglion; CFA: complete Freund’s adjuvant; Vehicle: vehicle control; Cont-CLs: control liposomes; CLs: clodronate liposomes; NC: normal control; CM: conditioned medium; EXOs: exosomes; TNF-α: tumor necrosis factor-alpha; GW4869: neutral sphingomyelinase 2 inhibitor used to suppress EXO release.

The involvement of EXO-associated signaling in ACSL6 regulation was further examined in vivo. Compared with the CFA group, intraperitoneal administration of GW4869 significantly increased ACSL6 expression in TG tissue (Figure 6C, D). This finding was consistent with the restoration of ACSL6 observed following macrophage depletion.

In vitro, TNF-α-CM significantly reduced ACSL6 mRNA and protein expression in primary TG neurons compared with NC-CM (Figure 6E, F). TNF-α-EXOs produced similar suppressive effects on neuronal ACSL6 expression (Figure 6G, H). Consistent results were observed in PC12 cells treated with TNF-α-CM or TNF-α-EXO (Figure 6I-L).

To further determine whether the reduction in neuronal ACSL6 depended on EXO release from activated macrophages, conditioned media were collected from RAW264.7 cells under four conditions: NC-CM, NC-CM + GW4869, TNF-α-CM, and TNF-α-CM + GW4869. No significant difference in ACSL6 expression was observed between the NC-CM and NC-CM + GW4869 groups. In contrast, ACSL6 expression was significantly decreased in the TNF-α-CM group and was partially restored in the TNF-α-CM + GW4869 group in both primary TG neurons and PC12 cells (Figure 6M-P).

Together, these findings indicate that activated macrophages contribute to the downregulation of neuronal ACSL6 and that this effect is, at least in part, associated with EXO release.

Overexpression of ACSL6 reduces macrophage-induced inflammation in PC12 cells. To overexpress ACSL6, PC12 cells were transfected with an ACSL6 overexpression or an empty control plasmid (Figure 7A). At 24 h after transfection, a marked increase in ACSL6 levels was observed in the PC12 cells (Figure 7B, C). Furthermore, after ACSL6 overexpression, co-culture of PC12 cells with conditioned medium from activated macrophages for 24 h reduced the expression of IL-6, COX-2, and SP (Figure 7D, E). Similarly, co-culture of ACSL6-overexpressing PC12 cells with EXOs derived from activated macrophages for 24 h resulted in comparable downregulation of IL-6, COX-2, and SP (Figure 7F, G). These findings indicate that increased ACSL6 expression attenuates activated macrophage-induced inflammatory responses in neuronal cells.

Figure 7.

Overexpression of ACSL6 reduces macrophage-induced inflammation in PC12 cells. (A) PC12 cells transfected with ACSL6 overexpression plasmid (green). Scale bar=100 μm. (B) Relative mRNA expression of ACSL6 in ACSL6-overexpressing PC12 cells, n=3 independent biological experiments. (C) Relative protein expression of ACSL6 in ACSL6-overexpressing PC12 cells, n=3 independent biological experiments. (D) Relative mRNA expression of SP, COX-2, and IL-6 in ACSL6-overexpressing PC12 cells co-cultured with TNF-α-CM, n=3 independent biological experiments. (E) Relative protein expression of COX-2 in ACSL6-overexpressing PC12 cells co-cultured with TNF-α-CM, n=3 independent biological experiments. (F) Relative mRNA expression of SP, COX-2, and IL-6 in ACSL6-overexpressing PC12 cells co-cultured with TNF-α-EXO, n=3 independent biological experiments. (G) Relative protein expression of COX-2 in ACSL6-overexpressing PC12 cells co-cultured with TNF-α-EXO, n=3 independent biological experiments. All data are presented as mean ± SD, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 vs. NC, pEGFP-C1, pEGFP-C1+NC-CM, or pEGFP-C1+NC-EXO group. ^#p<0.05, ^##p<0.01, ^###p<0.001, ^####p<0.0001 vs. pEGFP-C1+TNF-CM or pEGFP-C1+TNF-EXO group. NC: Normal control; CFA: complete Freund’s adjuvant; EXOs: exosomes; TNF-α: tumor necrosis factor-alpha; CM: conditioned medium; SP: substance P; COX-2: cyclooxygenase-2; IL-6: interleukin-6.

DHA supplementation alleviates TMD pain and neuronal inflammation. Given the role of ACSL6 in DHA metabolism and the involvement of DHA in inflammatory pain regulation, DHA was administered intraperitoneally to mice with TMD inflammatory pain. Behavioral testing showed that the CFA + DHA group exhibited a significantly higher HWT in the TMJ region on day 3 than that in the untreated CFA group (Figure 8A). Concurrently, the CFA + DHA group exhibited lower mRNA expression of IL-6 and COX-2 (Figure 8B, C). In vitro experiments involved co-culturing PC12 and TG neuronal cells with either TNF-α-CM or TNF-α-EXOs, with DHA administered as an intervention for 24 h (Figure 8D). Treatment of TNF-α-CM-stimulated TG neurons with DHA reduced SP, COX-2, and IL-6 expression (Figure 8E, F), indicating that DHA supplementation suppressed the neuronal inflammatory responses induced by activated macrophages. Similarly, when DHA was added to the co-culture system of TNF-α-EXOs and TG neurons for 24 h, the corresponding inflammatory responses in the neurons were attenuated (Figure 8I, J). Consistent results were obtained in the PC12 cells. DHA treatment reduced SP, IL-6, and COX-2 expression in PC12 cells exposed to TNF-α-CM or TNF-α-EXOs (Figure 8G, H, K, L), further supporting the protective effect of DHA against macrophage-mediated neuronal inflammation.

Figure 8.

DHA supplementation restored the pain threshold and alleviated neuronal inflammation in mice with TMD pain. (A) HWT (percentage of the baseline) on day 3 in the CFA + PBS and CFA + DHA groups, n=6 mice per group. (B) Relative mRNA expression of IL-6 in TG tissue, n=6 mice per group. (C) Relative mRNA expression of COX-2 in TG tissue, n=6 mice per group. (D) RAW264.7 cells were stimulated with 50 ng/ml TNF-α for 24 h to induce activation, and the cell culture supernatant collected. A portion of the collected supernatant was used in co-culture with mouse TG primary neuronal cells and PC12 cells for 24 h in the presence of 50 μmol/l DHA. EXOs were extracted from the remainder using differential ultracentrifugation, and the extracted EXOs and mouse TG primary neurons were co-cultured with PC12 cells for 24 h with 50 μmol/l DHA. (E) Relative mRNA expression of SP, COX-2, and IL-6 in TG neuronal cells after co-culture with DHA and TNF-α-CM, n=3 independent neuronal isolations. (F) Relative protein expression of COX-2 in TG neuronal cells after co-culture with DHA and TNF-α-CM, n=3 independent neuronal isolations. (G) Relative mRNA expression of SP, COX-2, and IL-6 in PC12 cells after co-culture with DHA and TNF-α-CM, n=3 independent biological experiments. (H) Relative protein expression of COX-2 in PC12 cells after co-culture with DHA and TNF-α-CM, n=3 independent biological experiments. (I) Relative mRNA expression of SP, COX-2, and IL-6 in TG neuronal cells after co-culture with DHA and TNF-α-EXO, n=3 independent neuronal isolations. (J) Relative protein expression of COX-2 in TG neuronal cells after co-culture with DHA and TNF-α-EXO, n=3 independent neuronal isolations. (K) Relative mRNA expression of SP, COX-2, and IL-6 in PC12 cells after co-culture with DHA and TNF-α-EXO, n=3 independent biological experiments. (L) Relative protein expression of COX-2 in PC12 cells after co-culture with DHA and TNF-α-EXO, n=3 independent biological experiments. All data are presented as mean ± SD. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001 vs. CFA+PBS, NC-CM, or NC-EXO group. ^#p<0.05, ^##p<0.01, ^###p<0.001, ^####p<0.0001 vs. TNF-CM+PBS or TNF-EXO+PBS group. DHA: Docosahexaenoic acid; CFA: complete Freund’s adjuvant; PBS: phosphate-buffered saline; HWT: head withdrawal threshold; TG: trigeminal ganglion; CM: conditioned medium; EXOs: exosomes; TNF-α: tumor necrosis factor-alpha; SP: substance P; COX-2: cyclooxygenase-2; IL-6: interleukin-6.

Discussion

In this study, we established a CFA-induced model of TMD pain and observed macrophage activation and increased expression of inflammatory mediators in the TG. Selective depletion of TG macrophages alleviated mechanical hypersensitivity and suppressed local inflammatory responses. Pharmacological inhibition of EXO-associated signaling using GW4869 similarly increased the HWT and reduced IL-6, COX-2, and SP expression in CFA-treated mice. ACSL6 expression was decreased in the TG following CFA injection, whereas macrophage depletion and GW4869 treatment partially restored ACSL6 expression. In vitro, conditioned medium and EXOs derived from activated macrophages increased neuronal inflammatory mediator expression and reduced ACSL6 expression. Inhibition of EXO release from activated macrophages attenuated these changes in both primary TG neurons and PC12 cells. In addition, ACSL6 overexpression and DHA supplementation reduced macrophage-induced neuronal inflammatory responses. These findings support the involvement of macrophage-derived EXOs in the regulation of neuronal ACSL6 and inflammatory responses during TMD pain.

In this study, the injection of CFA resulted in a reduction in HWT on day 1, which persisted until day 7. The pain threshold reached its lowest level on day 3 after CFA injection. Based on these observations, day 3 after TMJ administration was selected for sample collection in subsequent experiments. This temporal pattern of pain behavior is consistent with the findings reported by Li et al. in a TMD pain model (38). Concurrent with the development of hyperalgesia, TG tissue exhibited morphological changes, including alterations in neuronal morphology and disorganization of nerve fibers. These findings are consistent with TG alterations reported in trigeminal neuralgia models (5), suggesting that local CFA injection induces neural injury-associated changes that may contribute to pain generation.

Elevated TNF-α levels were detected in TG tissue 3 days after CFA injection, prompting the use of TNF-α-activated macrophages in subsequent experiments. Compared with the saline-injected group, IL-6, COX-2, and SP expression in the ganglion was also increased. As a key mediator of early inflammatory responses, TNF-α plays a central role in peripheral sensitization in multiple orofacial pain models. Increased TNF-α expression in the TG has been reported in rat inferior alveolar nerve transection, mouse CCI-ION, and infraorbital nerve injury models (39-41). Collectively, these findings support the view that CFA-induced TMD pain involves both neuronal alterations and a coordinated inflammatory response mediated by key cytokines.

Macrophage activation plays a critical role in pain transmission. Tanaka et al. reported that skin macrophages maintain normal nociceptive perception by expressing nerve growth factor (42). Activated macrophages may regulate neurons or induce neuronal injury through complex signaling pathways, thereby promoting the release of pain mediators, such as prostaglandins and bradykinin, and exacerbating pain sensitization (43). The increase in TG Iba1 expression observed in this study supports the presence of local macrophage activation and is consistent with previous evidence from Villa et al. (44). Collectively, these findings support the important role of macrophages in orofacial inflammatory pain and are consistent with reports of neuroimmune activation in other neuropathic pain models. To clarify the functional role of TG macrophages in pain transmission, CLs were locally administered to the TG to deplete macrophages from this region. Preliminary results indicated that changes in Iba1 expression were not evident on day 1 after CL injection, whereas effective depletion of TG macrophages was achieved by day 3. By day 3, the pain threshold recovered, accompanied by reduced TG inflammation and decreased expression of IL-6, COX-2, and SP. Similar findings have been reported by Fujiwara et al., who showed that HWT recovered to preoperative levels by day 3 after CL treatment (21). In other models, CLs also inhibited the production of inflammatory mediators. For example, Song et al. reported reduced IL-6 secretion in plasma and bronchoalveolar lavage fluid after radiation injury in the CL-treated group (45), and Sakurai et al. showed that intravenous CLs suppressed monosodium iodoacetate-induced increases in COX-2 expression and alleviated knee pain responses (46). Overall, activated macrophages appear to participate in nociceptive modulation by affecting neuronal responsiveness and regulating the production of proinflammatory mediators.

In addition to macrophage depletion, systemic administration of GW4869 attenuated CFA-induced mechanical hypersensitivity and reduced IL-6, COX-2, and SP expression in the TG. GW4869 inhibits neutral sphingomyelinase-dependent ceramide production and is widely used to reduce the biogenesis and release of EXOs (47). Zhang et al. demonstrated that local administration of GW4869 into the medullary dorsal horn increased the pain threshold in rats with experimental pulpitis, indicating that inhibition of inflammatory EXO secretion alleviated orofacial pain and neuroinflammation (48). Similarly, Wang et al. reported that blocking EXO release using GW4869 mitigated spinal neuroinflammation and prevented the development of hyperalgesia (49). Consistent with these findings, the present results indicate that EXO-associated signaling contributes to the maintenance of inflammatory responses and mechanical hypersensitivity in CFA-induced TMD pain. However, because intraperitoneally administered GW4869 may affect EXO release and ceramide-associated processes in multiple cell types, the in vivo findings do not independently demonstrate that its analgesic effects were exclusively caused by inhibition of macrophage-derived EXOs.

To determine whether macrophages influence TG neurons and thereby affect pain processing, we first co-cultured neurons with the conditioned medium from activated macrophages. The results showed an increased expression of inflammatory mediators in neuronal cells. EXOs are bilayer lipid nanoparticles with unique biogenesis pathways and shared cargo components, and they represent a major mode of intercellular communication that is implicated in multiple diseases (50). A previous study reported that macrophage-nociceptor interactions in chronic pain are also mediated by EXOs (51). Therefore, we isolated EXOs from RAW264.7 cell culture supernatants and tested whether these particles could be directly taken up by neurons. The results showed that EXOs were successfully isolated and internalized by TG neurons. Compared with EXOs derived from non-activated macrophages, EXOs from activated macrophages significantly increased IL-6, COX-2, and SP expression in neuronal cells. Similar results were obtained for PC12 cells. These findings are consistent with those reported by Gugliandolo et al., in which activated macrophages regulate neuronal biological behavior (52).

To further examine the contribution of EXO release, RAW264.7 cells were treated with GW4869 before conditioned medium collection. More recently, inhibition of macrophage-derived EXO release using GW4869 was shown to attenuate microglial inflammatory activation and macrophage EXO-associated hippocampal inflammation and neuronal injury (53, 54). Consistent with these findings, NC-CM + GW4869 did not substantially alter neuronal inflammatory marker expression compared with NC-CM in the present study. In contrast, conditioned medium derived from GW4869-treated, TNF-α-stimulated macrophages induced lower expression of inflammatory mediators in both primary TG neurons and PC12 cells, supporting the involvement of EXO-mediated macrophage-neuron communication.

The dysregulation of ACSL6 in multiple diseases suggests that it may be a potential therapeutic target (55-57). Fernandez et al. reported that, compared with controls, ACSL6−/− mouse brains showed increased expression of immune response genes (29). Reduced ACSL6 expression is associated with impaired locomotor performance, memory deficits, and enhanced neuroinflammation in the aging brain (29). In addition, ACSL6 expression has been confirmed in mature neurons (29).

Macrophage depletion partially restored ACSL6 expression in the TG, suggesting that activated TG macrophages contribute to the suppression of neuronal ACSL6. Notably, intraperitoneal administration of GW4869 also increased ACSL6 expression in CFA-treated TG tissue. Although systemic GW4869 is not macrophage-specific, the consistent restoration of ACSL6 following either macrophage depletion or inhibition of EXO-associated signaling suggests that macrophage-neuron communication may participate in ACSL6 regulation in vivo.

In vitro, both conditioned medium and isolated EXOs derived from TNF-α-activated macrophages reduced ACSL6 expression in primary TG neurons and PC12 cells. Furthermore, ACSL6 expression was partially restored in the TNF-α-CM + GW4869 group compared with the TNF-α-CM group, whereas GW4869 treatment under basal conditions did not substantially alter ACSL6 expression. These findings provide additional evidence that the suppression of neuronal ACSL6 by activated macrophage-conditioned medium is, at least in part, dependent on EXO release. Together, the in vivo and in vitro findings support ACSL6 as a downstream neuronal molecule associated with activated macrophage-derived EXO signaling. However, the specific EXO cargo responsible for regulating ACSL6 remains to be identified. Accordingly, ACSL6 was selected for further functional analysis. ACSL6 overexpression attenuated the increases in IL-6, COX-2, and SP induced by activated macrophage-conditioned medium and activated macrophage-derived EXOs in PC12 cells. Together with the restoration of ACSL6 following macrophage depletion and GW4869 treatment, these findings suggest that ACSL6 is not merely an inflammation-associated marker but may represent a functionally relevant downstream component of macrophage-neuron communication. Nevertheless, ACSL6 overexpression was examined only in PC12 cells, and additional validation in primary TG neurons and in vivo models is required to establish its causal role in TMD pain.

ACSL6 plays an important role in the incorporation and retention of DHA in neural tissues. Studies using ACSL6 knockout mice have shown that neuronal ACSL6 deficiency reduces brain DHA levels and is accompanied by motor dysfunction, altered glutamate metabolism, astrocytosis, and microglial activation (58). Given the neuroprotective and anti-inflammatory properties of DHA, DHA supplementation was further evaluated as a potential intervention in the present model. Intraperitoneal DHA administration increased the HWT and reduced inflammatory mediator expression in CFA-treated mice. In vitro, DHA attenuated the inflammatory responses induced by activated macrophage-conditioned medium and TNF-α-EXOs in both primary TG neurons and PC12 cells. These findings support the anti-inflammatory and analgesic potential of DHA in TMD pain. However, because ACSL6 overexpression and DHA supplementation were investigated as separate interventions, the present results do not establish that DHA exerted its protective effects through ACSL6. Experiments examining the effects of DHA following ACSL6 knockdown or inhibition will be required to determine whether ACSL6 is necessary for DHA-mediated protection.

Study limitations. First, GW4869 inhibits neutral sphingomyelinase-associated vesicle release but is not selective for macrophages, particularly after systemic administration. Therefore, macrophage-specific inhibition or in vivo tracing of macrophage-derived EXOs will be required to confirm their cellular origin in the TG. Fluorescent labeling of exosome-associated proteins has previously enabled direct visualization of the in vivo distribution and cellular uptake of cell-derived exosomes, supporting the feasibility of such tracing approaches (59). Second, TNF-α stimulation does not fully reproduce the complex macrophage activation state induced by CFA in vivo, which may involve multiple inflammatory signals. Third, only male mice were used in the present study. Given that sex may influence pain sensitivity, inflammatory responses, and neuroimmune interactions, the present findings may not be fully generalizable to females. Future studies including both male and female animals will be necessary to determine whether the observed macrophage-EXO-neuronal responses exhibit sex-dependent differences. Fourth, the specific EXO cargo responsible for ACSL6 regulation was not identified. EXOs can contain regulatory molecules, including miRNAs and lncRNAs, whose abundance may change under pathological conditions (60). Comprehensive profiling and functional validation of macrophage-derived EXO cargo will therefore be required to identify the key molecules responsible for ACSL6 regulation. In addition, lipidomic analyses of DHA-related lipids were not performed, and the causal relationship between DHA and ACSL6 was not directly tested. ACSL6 loss-of-function experiments and targeted lipidomic analyses will be required to determine whether DHA-mediated protection depends on ACSL6 and to further clarify the underlying lipid-metabolic mechanisms.

Collectively, these findings support a role for macrophage activation and macrophage-derived EXOs in CFA-induced TMD pain. Our results suggest that EXOs contribute to neuronal inflammatory responses and are associated with reduced ACSL6 expression in TG neurons. Modulation of macrophage-neuron communication mediated by EXOs may therefore represent a potential therapeutic strategy for inflammatory pain associated with TMD.

Conclusion

Macrophage-derived EXOs promoted neuronal inflammation and were associated with reduced ACSL6 expression in CFA-induced temporomandibular pain. Inhibition of macrophages or EXO release alleviated pain and TG inflammation, whereas ACSL6 overexpression and DHA supplementation independently exerted protective effects. These findings highlight macrophage-neuron EXO signaling and neuronal ACSL6 as potential therapeutic targets for TMD pain.

Conflicts of Interest

The Authors declare no conflicts of interest.

Authors’ Contributions

Yali Zhang: Data curation, Software, Formal analysis, Visualization, Writing – original draft; Chenyu Fan: Conceptualization, Writing – original draft, Validation; Jingwen Wu: Investigation, Validation, Methodology; Shuo Mu: Methodology, Validation; Zhuo Chen: Methodology, Validation; Siyu Wu: Methodology; Xian Wu: Conceptualization, Supervision; Zhenguo Shen: Writing – review & editing, Project administration, Resources; Tian Xing: Conceptualization, Supervision, Funding acquisition.

Acknowledgements

We gratefully acknowledge the Gene Expression Omnibus database for providing the public datasets GSE188922 and GSE207126. We thank the investigators who generated and submitted these valuable datasets. We also acknowledge the Library and Archives of Anhui Medical University for providing access to literature retrieval resources. In addition, we sincerely thank Professor Yu Li and her team for their valuable assistance and support in this study. Flowcharts were created using FigDraw (www.figdraw.com).

Funding

This study was supported by the Scientific Research Project of the Anhui Educational Committee (No. 2025AHGXZK30949); the project of Anhui Province Key Laboratory of Cognition and Neuropsychiatric Disorders, The First Affiliated Hospital of Anhui Medical University (No. 2025KLCND04); and the “Fengyuan” cooperation project of Disciplinary Construction Project in School of Dentistry, Anhui Medical University (2023xkfyhz02).

Artificial Intelligence (AI) Disclosure

During the preparation of this manuscript, a large language model (ChatGPT, OpenAI) was used solely for language editing and stylistic improvements in select paragraphs. No sections involving the generation, analysis, or interpretation of research data were produced by generative AI. All scientific content was created and verified by the authors. Furthermore, no Figures or visual data were generated or modified using generative AI or machine learning-based image enhancement tools.