Warm Needle Acupuncture Attenuates Pain-Related Responses and Inflammation in Knee Osteoarthritis: Evidence for Involvement of TAK1/MAPK Signaling from Human Cartilage and Experimental Models

Abstract

Background

Warm needle acupuncture (WNA) can improve symptoms in knee osteoarthritis (KOA). However, the molecular changes linked to its effects remain unclear.

Methods

This study combined human cartilage proteomics, a rat model of destabilization of the medial meniscus (DMM), and ex vivo primary chondrocyte analysis. The final clinical analysis included 61 patients with advanced KOA (control, n = 31; WNA, n = 30). Cartilage and synovial fluid were collected after a 21-day preoperative intervention. Cartilage samples underwent four-dimensional data-independent acquisition (4D-DIA) proteomics, followed by hypothesis-driven analyses focusing on mitogen-activated protein kinase (MAPK)-related processes. Differentially expressed proteins (DEPs) were defined as |log₂FC| ≥ 1 and nominal P < 0.05. A DMM-induced KOA rat model involving 60 rats was used for in vivo validation. Mechanical pain hypersensitivity was assessed using the von Frey test, and cartilage pathology, MAPK signaling, and inflammatory cytokines were evaluated. Primary chondrocytes were isolated after in vivo treatment. Ex vivo analyses assessed transforming growth factor-β-activated kinase 1 (TAK1)/MAPK signaling and interleukin-6 (IL-6). Intra-articular interleukin-1β (IL-1β) was used as an additional inflammatory challenge.

Results

WNA was associated with lower immunoreactivity for p-TAK1, p-JNK, p-ERK, and p-p38 in human KOA cartilage. Synovial fluid IL-6 (P < 0.01), IL-1β (P < 0.05), and tumor necrosis factor-α (TNF-α) (P < 0.05) were also lower. Proteomics identified 475 DEPs. MAPK-focused analyses showed changes involving JNK-, ERK1/ERK2-, and p38 MAPK-related processes. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses also identified inflammatory pathways including MAPK, nuclear factor kappa B (NF-κB), interleukin-17 (IL-17), and TNF signaling. In DMM rats, mechanical paw withdrawal thresholds were reduced after model establishment. WNA progressively increased paw withdrawal thresholds during treatment, with significant improvement compared with the Model group after 14 and 21 days of intervention (both P < 0.001). WNA also reduced cartilage damage and OARSI scores (P < 0.001), accompanied by lower TAK1/MAPK signaling and inflammatory cytokine levels. Chondrocytes from WNA-treated DMM rats also showed lower TAK1/MAPK signaling and IL-6 expression. Intra-articular IL-1β attenuated several WNA-associated molecular changes.

Conclusion

WNA was associated with lower TAK1/MAPK signaling and reduced IL-1β, TNF-α, and IL-6 in KOA. Human cartilage proteomics also showed WNA-associated changes in MAPK-related processes. Together, these findings support the involvement of TAK1/MAPK-related inflammatory signaling in the response to WNA. Direct functional studies are needed to determine the causal roles of TAK1 and individual MAPK branches.

Trial Registration

International Traditional Medicine Clinical Trial Registry (ITMCTR), registration number: ITMCTR2025002585.

Introduction

Knee osteoarthritis (KOA) is a common degenerative joint disease. The 2021 Global Burden of Disease study estimated about 607 million people with osteoarthritis worldwide. KOA accounts for more than half of this burden.¹ Its prevalence rises sharply with age. KOA is therefore a major cause of pain and functional loss in older adults.² The disease involves progressive cartilage degeneration, synovial inflammation, and abnormal inflammatory signaling. Interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α) can increase pain sensitization and matrix metalloproteinase expression. These changes promote extracellular matrix breakdown and cartilage damage.³

Several inflammatory pathways contribute to KOA progression. The mitogen-activated protein kinase (MAPK) system is important in chondrocyte inflammation, stress responses, and matrix degradation.^4,5 Its major branches include c-Jun N-terminal kinase (JNK), p38 MAPK, and extracellular signal-regulated kinase (ERK). These pathways regulate inflammatory and catabolic responses in cartilage.^6,7 JNK activation, for example, promotes c-Jun phosphorylation and activator protein-1 (AP-1)-dependent transcription. This response can increase interleukin-6 (IL-6), matrix metalloproteinases (MMPs), cyclooxygenase-2 (COX-2), and inducible nitric oxide synthase (iNOS).⁸ IL-6 is also associated with KOA progression. It may contribute to cartilage catabolism and inflammation-related pain.^9–11 Although MMPs are important downstream effectors of extracellular matrix degradation in OA,¹² the present study primarily focused on inflammatory signaling and pain-related responses. IL-6 was therefore selected as a representative downstream inflammatory mediator because it is linked to both cartilage catabolism and pain and could be evaluated across the clinical, animal, and chondrocyte experiments. This focus does not exclude the contribution of MMP-mediated matrix degradation to OA progression or the potential effects of WNA on these processes.

Warm needle acupuncture (WNA) combines acupuncture with heat generated by moxibustion. Clinical studies and meta-analyses have reported improvements in pain and knee function in patients with KOA.^13–15 Experimental studies also suggest that acupuncture-related treatments can affect MAPK and nuclear factor kappa B (NF-κB) signaling. Moxibustion has been linked to changes in inflammatory cytokines and cartilage catabolic mediators.^16–19 These data support further study of WNA as an anti-inflammatory intervention. Still, molecular changes specific to WNA remain poorly defined. Most previous mechanistic work has used animal or cellular models. Direct molecular evidence from human diseased cartilage is limited.

We therefore used four-dimensional data-independent acquisition (4D-DIA) proteomic profiling to characterize molecular changes in human KOA cartilage after WNA. Based on our study hypothesis, subsequent enrichment and network analyses focused on MAPK-related processes. These analyses showed changes involving JNK, ERK1/ERK2, and p38 MAPK. We then selected transforming growth factor-β-activated kinase 1 (TAK1) as an upstream regulator for targeted analysis.

Materials and Methods

Participants, Eligibility, and Study Design

Patients with advanced KOA awaiting primary total knee arthroplasty (TKA) at Shanghai Guanghua Hospital of Integrated Traditional Chinese and Western Medicine were prospectively recruited and randomized to the study groups.

Eligibility required a diagnosis of KOA according to the 2021 Chinese guideline for the diagnosis and treatment of osteoarthritis,²⁰ Kellgren–Lawrence (KL) grade III or IV disease, an age of 65–75 years, planned primary TKA, and written informed consent. Exclusion criteria comprised major chronic conditions likely to affect treatment or outcome evaluation (including diabetes mellitus, poorly controlled hypertension, or heart failure); other knee disorders, particularly inflammatory or rheumatic diseases such as rheumatoid arthritis or gouty arthritis; active bleeding or infection around the knee; severe cardiovascular or cerebrovascular disease, severe hepatic or renal dysfunction, cognitive impairment, or psychiatric disorders; and inability to tolerate acupuncture or moxibustion.

Participants were withdrawn in the event of voluntary withdrawal, inadequate adherence preventing completion of study procedures, receipt of KOA-directed treatment outside the study protocol, a serious adverse event or severe complication, or failure to provide the required cartilage samples. Demographic and clinical characteristics recorded at baseline comprised age, sex, body mass index (BMI), KL grade, disease duration, comorbidities other than KOA, and concomitant medications for non-KOA conditions.

The protocol complied with the Declaration of Helsinki and was approved by the Ethics Committee of Shanghai Guanghua Hospital of Integrated Traditional Chinese and Western Medicine (Ethics No. 2025-K-97). Written informed consent was obtained before enrollment. The trial was registered with the International Traditional Medicine Clinical Trial Registry (ITMCTR; No. ITMCTR2025002585).

Sample Size Calculation

The sample size of the parent clinical trial was calculated using the Hospital for Special Surgery (HSS) knee score at 14 days after TKA as the primary efficacy outcome. In the absence of published data for comparable interventions, the parameters for the calculation were derived from a pilot study conducted at Shanghai Guanghua Hospital of Integrated Traditional Chinese and Western Medicine between March and June 2025. Thirty eligible participants in the pilot study were randomized equally to the WNA and control groups (n = 15 each) and received the same interventions subsequently used in the formal trial.

The pilot data yielded an estimated standard deviation of 4.47 points and an anticipated between-group difference of 3.9 points. Assuming a one-sided α of 0.025, 90% power, and equal allocation, 28 participants were required per group. After allowing for 20% attrition, the enrollment target was increased to 35 participants per group (70 participants in total).

The present study was conducted as a mechanistic analysis of the parent trial, using cartilage and synovial fluid specimens from enrolled participants together with subsequent experimental validation. Animal group sizes were selected with reference to previous DMM-induced OA studies employing comparable histological and molecular endpoints.^21–25 The numbers were chosen to provide sufficient biological replication while limiting animal use in accordance with the reduction principle and ARRIVE guidelines.²⁶

Randomization and Blinding

Participants were allocated 1:1 to the WNA or control group by simple randomization. An independent statistician generated the allocation sequence using SPSS version 26.0. Group assignments were concealed in sequentially numbered, opaque, sealed envelopes and revealed according to the order of enrollment.

Blinding of participants and acupuncturists was not feasible because of the nature of WNA. However, investigators responsible for clinical outcome assessment, sample collection, laboratory measurements, and data analysis were blinded to group allocation. Biological samples were labeled with anonymized codes, and personnel performing the laboratory measurements were unaware of group allocation during data acquisition and quantification.

Clinical Interventions

Both groups received standardized preoperative education and a 21-day knee exercise program delivered under a uniform protocol by trained investigators. Daily exercise records were maintained by participants and reviewed to monitor adherence. Following the intervention period, TKA was performed by the same surgical team using an identical operative procedure and prosthesis type, with tourniquet time restricted to ≤ 90 min. Postoperative management and early rehabilitation were standardized across groups. Thus, the preoperative WNA treatment constituted the principal difference between groups.

Control group. The exercise regimen comprised straight-leg raising, quadriceps strengthening, and seated knee flexion. Straight-leg raising was performed with the affected knee extended and the quadriceps contracted; the leg was elevated approximately 20–30 cm, held for 5 s, and then lowered with 5 s of relaxation (20 repetitions/set, three sets/day, adjusted to individual tolerance). Quadriceps strengthening consisted of alternating ankle dorsiflexion and plantar flexion while maintaining knee extension (200 repetitions/day). Seated knee flexion was performed to the individual’s tolerable limit of discomfort (20–30 repetitions/session, four sessions/day). This program was continued daily for 21 days before TKA.

WNA group. In addition to the same exercise regimen, participants underwent WNA once daily for 21 consecutive days. Treatment was administered by licensed physicians trained in the standardized protocol using sterile disposable acupuncture needles (0.25 mm × 40 mm; Wuxi Jiajian Medical Instrument Co., Ltd., China). With participants supine and the affected knee flexed, needles were placed at Dubi (ST35), Neixiyan (EX-LE4), Zusanli (ST36), Yanglingquan (GB34), Yinlingquan (SP9), Liangqiu (ST34), and Xuehai (SP10), located according to the Chinese national standard GB/T 12346–2006. ST35 and EX-LE4 were inserted obliquely upward at approximately 45° to a depth of 2 cm; the other points were needled vertically to approximately 2 cm. Even reinforcing–reducing manipulation was applied to elicit deqi.

Warm stimulation was delivered by attaching a 2-cm moxa segment (1.2 cm × 2 cm; Yueyang Aijiantang Biotechnology Co., Ltd., China) to the needle handles at ST35 and EX-LE4. Local temperature was monitored with a TCP-400B system (MEASURE FINE, China) and maintained at 46±1°C by adjusting the moxa-to-skin distance. A heat-insulating card protected the skin from falling ash. Two consecutive moxa segments were burned per session, providing approximately 30 min of treatment. A total of 21 daily WNA sessions were completed before TKA.

Clinical Sample Collection and Processing

Articular cartilage and synovial fluid were obtained during TKA following completion of the 21-day preoperative intervention. Evaluable cartilage specimens were available from 61 participants (control, n = 31; WNA, n = 30). Approximately 1×1 cm of cartilage was sampled from the most severely affected region of the medial tibial plateau. Within 5 min of excision, specimens were transferred on dry ice and rinsed with phosphate-buffered saline (PBS) for 10s. Tissue allocated to proteomic and molecular assays was processed, aliquoted, and maintained at −80°C until analysis, whereas samples for histological and immunohistochemical examination were fixed in 4% paraformaldehyde.

Synovial fluid was aspirated before opening the joint capsule and centrifuged at 3000 rpm for 10 min. The resulting supernatant was stored at −80°C. Samples showing hemolysis or providing insufficient volume were omitted from cytokine measurement, leaving 45 samples for ELISA (control, n = 22; WNA, n = 23).

Experimental Animals and DMM Model

Sixty specific-pathogen-free male Sprague–Dawley rats, aged 6 weeks and weighing 180 ± 20 g, were obtained from Spelford (Suzhou) Biotechnology Co., Ltd., China. Animals were maintained under a 12-h light/dark cycle at 18–22°C and 50–70% relative humidity, with unrestricted access to food and water. Following a 1-week acclimation period, the 60 rats were randomly allocated to two experimental cohorts. Randomization was performed using a computer-generated random sequence. Cohort 1 comprised 36 rats and was used for the principal in vivo experiment, whereas Cohort 2 comprised 24 rats and was used for the mechanistic experiment involving intra-articular IL-1β challenge.

For induction of DMM-associated KOA, the right knee was surgically exposed and the medial meniscotibial ligament (MMTL) was transected. Sham-operated rats underwent the same exposure of the skin and joint capsule without MMTL transection. WNA was initiated 8 weeks after surgery to evaluate treatment effects on established osteoarthritic changes rather than the immediate postoperative response.

At study completion, animals were deeply anesthetized with intraperitoneal Zoletil 50 (tiletamine–zolazepam, 50 mg/kg). After loss of the toe-pinch withdrawal reflex confirmed an adequate depth of anesthesia, euthanasia was performed by cervical dislocation. Animal procedures were approved by the Animal Ethics Committee of Shanghai Hospital of Traditional Chinese Medicine (Ethics No. 2024020) and complied with the Guidelines for Ethical Review of Laboratory Animal Welfare of the People’s Republic of China (GB/T 35892–2018).

Animal Grouping and Intervention Protocols

In Cohort 1, 36 rats were randomly allocated to the Sham, Model, and WNA groups (n = 12 per group). Sham-operated rats underwent joint exposure without medial meniscotibial ligament transection, whereas rats in the Model and WNA groups underwent DMM surgery. Beginning 8 weeks after surgery, rats in the WNA group received WNA at ST35 and EX-LE4 for 30 min once daily for 21 consecutive days, with the local skin temperature maintained at 46 ± 1°C. All 12 animals per group were included in behavioral assessment. Subsequent tissue-based analyses were performed using subsets of these animals, with six animals per group used for histological and ELISA analyses and cartilage samples from three separate animals per group used for Western blotting.

In Cohort 2, 24 rats were randomly allocated to the Sham+PBS, Model+PBS, WNA+PBS, and WNA+IL-1β groups (n = 6 per group). The Sham+PBS group underwent sham surgery, whereas the remaining three groups underwent DMM surgery. Beginning 8 weeks after surgery, rats in the WNA+PBS and WNA+IL-1β groups received the same WNA regimen as described for Cohort 1. For intra-articular interventions, the Sham+PBS, Model+PBS, and WNA+PBS groups received 50 μL PBS into the right knee on days 0, 7, 14, and 21, whereas the WNA+IL-1β group received 50 ng recombinant IL-1β (501-RL; R&D Systems, USA) in 50 μL PBS by the same route and on the same schedule. Animals were euthanized within 4 h after the final intervention, and knee tissues were harvested for subsequent analyses.

Assessment of Mechanical Paw Withdrawal Threshold

Mechanical sensitivity was quantified using calibrated von Frey filaments and the up–down method. Rats were individually acclimated to the testing chambers for at least 60 min; an additional 30 min was allowed when exploratory behavior persisted. The investigator remained quietly in the room for 5 min before testing.

Filaments of 0.4, 0.6, 1.0, 2.0, 4.0, 6.0, 8.0, and 15.0 g were applied perpendicularly to the plantar hind paw for approximately 3s. Paw withdrawal or licking was considered a positive response. Filament force was decreased after a positive response and increased after a negative response. Four additional stimulations were performed according to the up–down sequence after the first negative-to-positive response transition, and the 50% paw withdrawal threshold (PWT) was calculated.

PWT was measured immediately before WNA treatment after establishment of the DMM model and again after 7, 14, and 21 days of intervention. All behavioral assessments were conducted by an investigator blinded to group assignment.

Isolation and ex vivo Analysis of Primary Chondrocytes

For the mechanistic analysis, primary chondrocytes were isolated from rats in Cohort 2 after completion of the designated in vivo interventions. Articular cartilage was collected from the right knees, cut into approximately 1-mm³ fragments, and subjected to sequential enzymatic digestion, first with 0.25% trypsin at 37°C for 15–20 min and subsequently with 0.2% type II collagenase for 3 h.

After digestion, the cell suspension was passed through a 70-μm strainer and centrifuged at 1200 rpm for 10 min. The recovered cells were resuspended in complete Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum and 1% antibiotics and maintained at 37°C in humidified 5% CO₂.

Primary Chondrocyte Characterization

The phenotype of the isolated cells was verified by immunofluorescence detection of SOX9 and type II collagen. Following fixation, permeabilization, and blocking, cells were sequentially incubated with the appropriate primary and fluorescence-conjugated secondary antibodies. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI), and images were obtained by fluorescence microscopy.

4D-DIA Proteomic Analysis

Fifteen human cartilage samples from each group were randomly selected for 4D-DIA quantitative proteomics (WNA, n = 15; control, n = 15). Cartilage proteins were extracted and quantified by bicinchoninic acid assay, with protein quality assessed by SDS-PAGE. After reduction, alkylation, enzymatic digestion, and peptide desalting, samples were subjected to liquid chromatography–tandem mass spectrometry.

Data acquisition was performed using a timsTOF HT mass spectrometer (Bruker, Germany) coupled to a Vanquish Neo liquid chromatography system (Thermo Fisher Scientific, USA). DIA data were analyzed with DIA-NN against the reviewed human UniProt database. Trypsin was specified as the digestion enzyme, permitting up to two missed cleavages. Carbamidomethylation of cysteine was defined as a fixed modification, whereas methionine oxidation and protein N-terminal acetylation were treated as variable modifications. False discovery rates at both the peptide-spectrum match and protein levels were controlled at 1%.

Protein abundance was compared between the WNA and control groups using an unpaired t-test. Differentially expressed proteins (DEPs) were defined by |log₂FC| ≥ 1 and nominal P < 0.05. To assess the robustness of the differential expression results to multiple-testing correction, Benjamini–Hochberg-adjusted q values were additionally examined as a sensitivity analysis.

Functional Enrichment and Network Analyses

Functional annotation of the DEPs was performed using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. In accordance with the predefined study hypothesis, subsequent analyses focused on MAPK-related processes. Gene set enrichment analysis (GSEA) was performed using proteins ranked by log₂ fold change. MAPK-related gene sets involving JNK, p38 MAPK, and ERK1/ERK2 were selected for focused analysis and visualization. Proteins associated with the selected MAPK-related processes were further evaluated by category–protein network analysis and hierarchical clustering.

A category–protein network linking enriched MAPK-related processes with their corresponding DEPs was generated using the cnetplot function of the enrichplot R package (version 1.18.4). Protein color in the network reflected the corresponding log₂ fold change.

Hierarchical Clustering and Heatmap Analysis

Sixteen DEPs associated with the selected JNK-, p38 MAPK-, and ERK1/ERK2-related processes were subjected to hierarchical clustering and heatmap visualization. The 16 proteins were selected because they were differentially expressed proteins mapped to the predefined JNK-, p38 MAPK-, and ERK1/ERK2-related GO terms included in the MAPK-focused analysis. Protein abundance values were Z-score standardized before clustering, and the heatmap was used to display relative abundance patterns across individual WNA and control samples.

Histology and OARSI Assessment

Knee specimens were fixed, decalcified in 10% EDTA (pH 7.4), paraffin embedded, and sectioned at 5 μm. Hematoxylin and eosin staining was used for histopathological examination. Cartilage degeneration was graded using the Osteoarthritis Research Society International (OARSI) scoring system under comparable anatomical and imaging conditions across groups. Histological sections were coded before evaluation, and OARSI scoring was performed by an investigator blinded to group assignment.

Immunohistochemistry

Paraffin sections were processed for immunohistochemistry by deparaffinization, rehydration, antigen retrieval, and blocking, followed by incubation with antibodies against phospho-JNK1/2/3, phospho-p38 MAPK, phospho-ERK1/2, and phospho-TAK1 (1:200; Affinity Biosciences, China). Immunoreactivity was detected with the corresponding secondary antibodies and 3,3′-diaminobenzidine (DAB).

Images were acquired under standardized conditions and analyzed using ImageJ. Immunohistochemical staining intensity was quantified as the average optical density (AOD) of DAB-positive staining. AOD was calculated from the integrated optical density and the corresponding positive staining area. For each cartilage specimen, AOD measurements were obtained from three separate, non-overlapping fields within the defined cartilage region. The three field-level measurements were then averaged to generate one value representing each biological sample. Three independent cartilage specimens were included in each group (n = 3).

Western Blotting

Western blotting was used to assess protein expression in rat cartilage and ex vivo primary chondrocytes. Protein lysates were resolved by SDS-PAGE and transferred to membranes, which were subsequently blocked and incubated with the appropriate primary antibodies overnight at 4°C, followed by horseradish peroxidase-conjugated secondary antibodies. Primary and secondary antibodies were generally diluted 1:1000 and 1:10,000, respectively, and GAPDH was used as a loading control where applicable.

Three independent biological replicates were analyzed per group (n = 3). Cartilage replicates were obtained from separate animals, whereas each chondrocyte replicate originated from an independent donor animal. Band intensity was quantified and normalized to the relevant total protein or loading control.

For descriptive assessment of the extent to which IL-1β attenuated WNA-associated molecular changes, a reversal percentage was calculated from group means: [(W+IL-1β) − W]/(M − W) × 100, where M, W, and W+IL-1β denote the Model+PBS, WNA+PBS, and WNA+IL-1β groups, respectively. A value of 0% represents no reversal toward the Model level, whereas 100% represents restoration to the Model level. These percentages were used solely to describe reversal magnitude and were not analyzed as independent statistical outcomes.

Enzyme-Linked Immunosorbent Assay

IL-1β, IL-6, and TNF-α concentrations were determined using ELISA kits according to the manufacturers’ protocols. Human synovial fluid was analyzed in the clinical component, whereas cartilage homogenates were used for the animal experiments. Cytokine concentrations were derived from standard curves based on the measured absorbance values.

Statistical Analysis

Statistical analyses were performed using SPSS 26.0. Normality was assessed by the Shapiro–Wilk test. Normally distributed continuous data are presented as mean ± standard deviation (SD), while non-normally distributed data are summarized as median with interquartile range.

Between-group comparisons involving two groups were conducted with an independent-samples Student’s t-test for normally distributed data or a Mann–Whitney U-test for non-normally distributed data. Categorical variables were compared using the chi-square test. For three or more groups, one-way ANOVA followed by Tukey’s post hoc test was used when parametric assumptions were satisfied; otherwise, a Kruskal–Wallis test with Dunn’s multiple comparisons was performed. Changes in PWT over time were evaluated by two-way repeated-measures ANOVA, treating group as the between-subject factor and time as the within-subject factor. Where appropriate, subsequent comparisons were adjusted using Šídák’s method. Statistical significance was defined as a two-sided P < 0.05.

Results

Participant and Sample Flow

A total of 70 participants were enrolled and randomized in a 1:1 ratio to the control and WNA groups (n = 35 per group). Of these, 61 participants completed the preoperative intervention and provided evaluable cartilage samples for the present mechanistic analysis (control group, n = 31; WNA group, n = 30). The baseline characteristics of these 61 participants are presented in Table 1. Among them, synovial fluid samples from 45 participants were suitable for cytokine analysis (control group, n = 22; WNA group, n = 23). Synovial fluid samples with insufficient volume or hemolysis were excluded from the cytokine analysis.

Table 1.

Baseline Characteristics of Participants Included in the Mechanistic Analysis

Characteristics		Total Participants	Control Group	WNA Group	Test Statistic	P
Age (years)		72.0 (70.0, 73.0)	72.0 (70.0, 74.0)	71.0 (70.0, 73.0)	542.5	0.263
Sex, n (%)	Male	13 (21.3%)	6 (19.4%)	7 (23.3%)	0.004	0.947
	Female	48 (78.7%)	25 (80.6%)	23 (76.7%)
BMI (kg/m2)		26.84 ± 2.43	26.98 ± 2.60	26.70 ± 2.27	0.435	0.665
Disease duration (months)		84.0 (36.0, 180.0)	84.0 (42.0, 168.0)	78.0 (27.0, 180.0)	493.5	0.685
Affected side, n (%)	Unilateral	29 (47.5%)	13 (41.9%)	16 (53.3%)	0.403	0.526
	Bilateral	32 (52.5%)	18 (58.1%)	14 (46.7%)
KL grade, n (%)	Grade III	15 (24.6%)	5 (16.1%)	10 (33.3%)	1.594	0.207
	Grade IV	46 (75.4%)	26 (83.9%)	20 (66.7%)
Underlying comorbidities, n (%)	Absent	21 (34.4%)	12 (38.7%)	9 (30.0%)	0.199	0.655
	Present	40 (65.6%)	19 (61.3%)	21 (70.0%)

Notes: Data are presented as mean ± SD, median (IQR), or n (%), as appropriate. Between-group comparisons were performed using the independent-samples Student’s t-test for normally distributed continuous variables, the Mann–Whitney U-test for non-normally distributed continuous variables, and the chi-square test for categorical variables.

Abbreviations: BMI, body mass index; IQR, interquartile range; KL, Kellgren–Lawrence; SD, standard deviation; WNA, warm needle acupuncture.

Differential Proteomic Analysis Reveals WNA-Associated Alterations in MAPK-Related Biological Processes in Human KOA Cartilage

4D-DIA proteomic analysis of human knee cartilage identified 475 DEPs between the WNA and control groups using the predefined criteria of |log₂FC| ≥ 1 and nominal P < 0.05. Of these, 232 proteins were upregulated and 243 were downregulated in the WNA group (Figure 1A). In a sensitivity analysis applying Benjamini–Hochberg multiple-testing correction while retaining the same fold-change threshold, 370 proteins (186 upregulated and 184 downregulated) remained significant at q < 0.05. The distribution and magnitude of the DEPs identified in the primary analysis are shown in the volcano plot (Figure 1B). After normalization, protein abundance distributions were comparable across samples (Figure 1C). Principal component analysis (PCA) showed partial separation between the WNA and control groups (Figure 1D).

Figure 1.

4D-DIA proteomic analysis reveals WNA-associated alterations in MAPK-related biological processes in human KOA cartilage. (A) Bar graph showing the numbers of differentially expressed proteins (DEPs) between the WNA and control groups (n = 15 per group). A total of 475 DEPs were identified, including 232 upregulated and 243 downregulated proteins in the WNA group compared with the control group. (B) Volcano plot showing the distribution of DEPs between the WNA and control groups. Differential expression was defined as |log₂FC| ≥ 1 and nominal P < 0.05. (C) Box plots showing the distribution of normalized protein abundance across individual samples. (D) Principal component analysis (PCA) of cartilage proteomic profiles in the WNA and control groups. (E) Category–protein network showing the relationships between enriched MAPK-related biological processes and their associated DEPs. Protein color represents the corresponding log₂ fold change. (F) Gene set enrichment analysis (GSEA) plots of selected MAPK-related gene sets involving the JNK, ERK1/ERK2, and p38 MAPK cascades and their regulation. Proteins were ranked according to log₂ fold change between the WNA and control groups. (G) Hierarchical clustering heatmap of 16 proteins associated with selected MAPK-related biological processes, showing relative protein abundance patterns across individual samples from the WNA and control groups. Protein abundance values were standardized using Z-score transformation. (H) Gene Ontology (GO) enrichment analysis of DEPs, including biological process (BP), cellular component (CC), and molecular function (MF) categories. Bubble size represents protein count, and color represents the P value. (I) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis of DEPs, including inflammation-related pathways such as NF-κB, IL-17, TNF, cytokine–cytokine receptor interaction, and MAPK signaling. Bubble size represents protein count, and color represents the P value.

Abbreviations: BP, biological process; CC, cellular component; DEPs, differentially expressed proteins; ERK, extracellular signal-regulated kinase; FC, fold change; GO, Gene Ontology; GSEA, gene set enrichment analysis; JNK, c-Jun N-terminal kinase; KEGG, Kyoto Encyclopedia of Genes and Genomes; KOA, knee osteoarthritis; MAPK, mitogen-activated protein kinase; MF, molecular function; PCA, principal component analysis; WNA, warm needle acupuncture.

Functional enrichment and network analyses were then performed to examine the biological processes represented by these proteins. In the category–protein network, several DEPs, including TAOK1, TAOK3, TLR3, SMAD3, and RIPK1, were shared among biological processes related to the JNK, p38 MAPK, and ERK1/ERK2 cascades (Figure 1E). Hypothesis-driven GSEA of selected MAPK-related gene sets showed enrichment patterns involving the ERK1/ERK2, JNK, and p38 MAPK cascades and their regulation (Figure 1F).

Hierarchical clustering of 16 proteins associated with the selected MAPK-related biological processes showed differences in abundance patterns between the WNA and control groups, as well as variation among individual samples (Figure 1G). The proteins included MAP1LC3A, MAP3K20, SMAD3, TLR3, CTSH, EGF, FGF2, VEGFA, PRMT1, STK39, TAOK3, APP, SERPINF2, RIPK1, TAOK1, and CCL19.

GO analysis identified biological processes involving MAPK regulation, including positive regulation of the MAPK and JNK cascades, along with inflammation- and cellular-response-related processes (Figure 1H). KEGG analysis identified NF-κB, IL-17, TNF, cytokine–cytokine receptor interaction, and MAPK signaling among the enriched pathways (Figure 1I).

WNA is Associated with Reduced MAPK Pathway Activation and Inflammatory Cytokine Levels in Human KOA

Immunohistochemistry was used to assess the immunoreactivity of phosphorylated TAK1/MAPK-related proteins in human cartilage. Staining for p-TAK1, p-JNK, p-p38, and p-ERK was weaker in the WNA group than in the control group (Figure 2A). Quantitative analysis of the average optical density (AOD) showed lower p-TAK1 immunoreactivity in the WNA group (P < 0.01), and lower p-JNK, p-p38, and p-ERK immunoreactivity (all P < 0.001) (Figure 2B).

Figure 2.

WNA is associated with reduced TAK1/MAPK pathway activation in human knee cartilage and decreased inflammatory cytokine levels in synovial fluid. (A) Representative immunohistochemical (IHC) images of phosphorylated TAK1 (p-TAK1), JNK (p-JNK), p38 MAPK (p-p38), and ERK (p-ERK) in human knee cartilage from the control (C) and WNA (W) groups (n = 3 per group). Scale bar = 100 μm. (B) Quantitative analysis of the average optical density (AOD) showed lower p-TAK1 immunoreactivity in the WNA group (P < 0.01), and lower p-JNK, p-p38, and p-ERK immunoreactivity (all P < 0.001). (C) ELISA quantification of IL-1β, IL-6, and TNF-α levels in human synovial fluid (C, n = 22; W, n = 23). Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Synovial fluid cytokine concentrations were also lower after WNA treatment. Compared with the control group, IL-1β (P < 0.05), IL-6 (P < 0.01), and TNF-α (P < 0.05) were significantly reduced in the WNA group (Figure 2C).

WNA Attenuates Mechanical Pain Hypersensitivity and Cartilage Damage and is Associated with Reduced TAK1/MAPK Signaling and Inflammation in DMM Rats

Mechanical pain sensitivity was assessed by measuring the paw withdrawal threshold (PWT) before WNA treatment and after 7, 14, and 21 days of intervention (Figure 3A). Two-way repeated-measures ANOVA showed significant effects of group [F(2,33)=55.23, P < 0.001] and time [F(3,99)=7.33, P < 0.001], as well as a significant group × time interaction [F(6,99)=8.60, P < 0.001].

Figure 3.

WNA attenuates mechanical pain hypersensitivity and cartilage damage and is associated with reduced TAK1/MAPK signaling and inflammatory cytokine levels in DMM rats. (A) Paw withdrawal threshold (PWT) assessed using von Frey filaments before treatment and after 7, 14, and 21 days of intervention in the Sham (S), DMM model (M), and WNA-treated (W) groups (n = 12 per group). (B) Representative H&E-stained sections of knee articular cartilage from the Sham, DMM model, and WNA-treated groups. Scale bar = 200 μm. (C) OARSI scores for histopathological assessment of cartilage damage (n = 6 per group). (D) Representative Western blot images of phosphorylated and total TAK1, p38, JNK, and ERK in cartilage tissues. GAPDH was used as the loading control. (E) Quantitative analysis of the relative phosphorylation levels of TAK1, p38, JNK, and ERK (n = 3 biological replicates per group). Phosphorylation levels were normalized to the corresponding total protein levels. (F) ELISA quantification of IL-1β, IL-6, and TNF-α levels in cartilage tissues (n = 6 per group). For (A), ***P < 0.001 vs Sham and ###P < 0.001 vs Model. For (C), (E), and (F), statistical comparisons are indicated by brackets; *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

Šídák-adjusted multiple comparisons showed that PWT was significantly lower in the Model group than in the Sham group before treatment (P < 0.001), confirming mechanical hypersensitivity after DMM modeling. No significant difference was observed between the Model and WNA groups before treatment or after 7 days of intervention. In contrast, PWT was significantly higher in the WNA group than in the Model group after 14 and 21 days of treatment (both P < 0.001). These findings indicate that WNA progressively attenuated DMM-associated mechanical pain hypersensitivity, with a clear treatment effect emerging after approximately 2 weeks of intervention.

H&E staining showed preserved cartilage morphology and structural integrity in the sham group, whereas DMM rats exhibited surface irregularity, cartilage erosion, and disorganized chondrocytes. These changes were less pronounced in WNA-treated rats (Figure 3B). OARSI scores were higher in the model group than in the sham group (P < 0.001) and were lower after WNA treatment than in the model group (P < 0.001), although they remained higher than those in the sham group (P < 0.001) (Figure 3C).

Western blot analysis showed increased phosphorylation of TAK1 (P < 0.001), ERK (P < 0.001), JNK (P < 0.01), and p38 (P < 0.01) in the model group compared with the sham group (Figure 3D and E). Compared with the model group, WNA treatment reduced TAK1 and ERK phosphorylation (both P < 0.001), as well as JNK (P < 0.05) and p38 (P < 0.01) phosphorylation. JNK and p38 phosphorylation did not differ significantly between the WNA and sham groups. TAK1 phosphorylation remained higher in the WNA group than in the sham group (P < 0.001), whereas ERK phosphorylation did not differ significantly between these groups.

Cartilage concentrations of IL-1β, IL-6, and TNF-α were higher in the model group than in the sham group (all P < 0.001) (Figure 3F). WNA treatment reduced all three cytokines compared with the model group (all P < 0.001), although their levels remained higher than those in the sham group (all P < 0.001).

IL-1β Partially Attenuates WNA-Associated Reductions in TAK1/MAPK Signaling in Primary Chondrocytes

Primary chondrocytes were isolated from rats in the Sham+PBS (S), Model+PBS (M), WNA+PBS (W), and WNA+IL-1β (W+IL-1β) groups. Immunofluorescence staining showed positive expression of type II collagen and SOX9 in the isolated cells, supporting their chondrocyte phenotype (Figure 4A).

Figure 4.

IL-1β partially attenuates WNA-associated reductions in TAK1/MAPK signaling and IL-6 expression in primary chondrocytes. (A) Representative immunofluorescence images showing type II collagen and SOX9 expression in primary chondrocytes isolated from the Sham+PBS (S), Model+PBS (M), WNA+PBS (W), and WNA+IL-1β (W+IL-1β) groups. Type II collagen and SOX9 are shown in red, and nuclei are counterstained with DAPI (blue). Scale bar = 50 μm. (B) Representative Western blot images of phosphorylated and total TAK1, p38, JNK, and ERK, together with IL-6 expression, in primary chondrocytes from the four groups. GAPDH was used as the loading control. (C) Quantitative analysis of the relative phosphorylation levels of TAK1, ERK, JNK, and p38 and the relative expression level of IL-6 (n = 3 biological replicates per group). Phosphorylation levels were normalized to the corresponding total protein levels, and IL-6 expression was normalized to GAPDH. Data are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001; ns, not significant.

Abbreviations: S, Sham+PBS group; M, Model+PBS group; W, WNA+PBS group; W+IL-1β, WNA+IL-1β group; DAPI, 4′,6-diamidino-2-phenylindole; DMM, destabilization of the medial meniscus; ERK, extracellular signal-regulated kinase; IL, interleukin; JNK, c-Jun N-terminal kinase; MAPK, mitogen-activated protein kinase; TAK1, transforming growth factor-β-activated kinase 1; WNA, warm needle acupuncture.

Western blot analysis showed higher phosphorylation levels of TAK1 (P < 0.001), ERK (P < 0.001), JNK (P < 0.01), and p38 (P < 0.001), together with higher IL-6 expression (P < 0.001), in the Model+PBS group than in the Sham+PBS group (Figure 4B and C). Compared with the Model+PBS group, the WNA+PBS group showed lower phosphorylation of TAK1 (P < 0.01), ERK (P < 0.01), JNK (P < 0.05), and p38 (P < 0.001), as well as lower IL-6 expression (P < 0.01). Compared with the WNA+PBS group, the WNA+IL-1β group showed higher phosphorylation of TAK1 (P < 0.05), JNK (P < 0.05), and p38 (P < 0.01), together with increased IL-6 expression (P < 0.05). ERK phosphorylation also increased numerically, but the difference between the WNA+PBS and WNA+IL-1β groups was not significant. TAK1, JNK, and p38 phosphorylation and IL-6 expression did not differ significantly between the WNA+IL-1β and Model+PBS groups.

IL-1β attenuated the WNA-associated reductions to different degrees. Based on group mean values, the estimated reversal was 102.9% for TAK1 phosphorylation, 88.9% for JNK phosphorylation, 74.8% for p38 phosphorylation, 69.9% for ERK phosphorylation, and 58.1% for IL-6 expression. TAK1 phosphorylation in the WNA+IL-1β group returned to approximately the Model+PBS level, whereas recovery of the other readouts was incomplete. These percentages provide a descriptive measure of reversal magnitude and should be interpreted together with the corresponding between-group statistical comparisons.

Discussion

This study examined the effects of WNA on pain-related behavior and inflammatory molecular changes in KOA. We combined human cartilage proteomics with a DMM rat model and primary chondrocyte experiments. In DMM rats, WNA progressively increased the mechanical paw withdrawal threshold, with significant improvement becoming evident after 14 days of treatment. This behavioral improvement was accompanied by reduced cartilage damage, lower inflammatory mediator levels, and decreased TAK1/MAPK signaling. Based on the study hypothesis, subsequent proteomic analyses focused on MAPK-related processes and identified WNA-associated changes involving JNK, ERK1/ERK2, and p38 MAPK. Together, these findings link the molecular changes observed in human KOA cartilage with improvement in pain-related behavior in the experimental model.

Inflammation is an important part of OA pathogenesis rather than simply a result of cartilage damage.²⁷ Cytokines from cartilage, synovium, and other joint tissues form a local inflammatory network. This network contributes to abnormal chondrocyte activity and cartilage breakdown.^28,29 In our study, IL-1β, TNF-α, and IL-6 were lower after WNA. This pattern suggests a broader change in the inflammatory environment rather than an isolated cytokine response. IL-6 was selected as a representative downstream marker based on its relevance to KOA³⁰ and its response in our experimental system. The behavioral findings provide complementary evidence for the analgesic effects of WNA. Mechanical hypersensitivity was already present before treatment in DMM rats, as indicated by the reduced PWT in both the Model and WNA groups compared with the Sham group. Importantly, the Model and WNA groups did not differ before intervention, supporting comparable pain-related behavior at treatment initiation. No clear separation between these groups was observed after 7 days of treatment. In contrast, PWT increased markedly after 14 and 21 days of WNA. These findings suggest that improvement in mechanical pain hypersensitivity developed progressively rather than immediately during the 21-day intervention.

At the molecular level, human cartilage proteomics was used to characterize WNA-associated changes in MAPK-related signaling. The primary analysis identified 475 differentially expressed proteins using the predefined fold-change and nominal P-value criteria. Notably, 370 of these proteins remained significant after Benjamini–Hochberg correction at q < 0.05, indicating that most of the differential proteomic signals were retained after multiple-testing correction. MAPK-focused analyses showed changes involving JNK, ERK1/ERK2, and p38 MAPK. MAPK signaling regulates inflammatory and catabolic responses in OA cartilage.³¹ Synovial fluid from end-stage OA can also alter chondrocyte behavior through MAPK signaling,³² a finding particularly relevant to the present study because our cartilage samples were obtained from patients with advanced KOA. The proteomic analysis also showed enrichment of NF-κB, IL-17, TNF, and cytokine-related pathways. Interactions between MAPK and other inflammatory pathways have also been reported in OA chondrocytes.³³ Together, these findings suggest coordinated changes in MAPK-related and inflammatory processes associated with WNA.

JNK-related processes were particularly prominent within the MAPK-focused analysis. However, several proteins were shared among JNK-, ERK1/ERK2-, and p38-related processes. This overlap argues against a response involving JNK alone. The targeted experiments showed a similar pattern. Lower JNK phosphorylation was accompanied by lower ERK and p38 phosphorylation across the study systems. JNK therefore appears to be an important part of the response, but not an isolated regulator. The overall pattern is more consistent with coordinated changes across several MAPK branches.

TAK1 was examined because it connects inflammatory signals with several MAPK pathways. It acts upstream of multiple inflammatory responses and can regulate downstream MAPK activity.³⁴ TAK1 is also involved in cartilage homeostasis and OA-related changes.³⁵ Although other MAP3Ks, including ASK1 and members of the MEKK family, can also activate MAPK signaling under specific stress conditions,³⁶ TAK1 was prioritized because of its established role in coupling pro-inflammatory cytokine signaling to multiple MAPK branches relevant to OA. Thus, the selection of TAK1 was hypothesis-driven and was not intended to exclude possible contributions from other upstream MAP3Ks. WNA was associated with lower p-TAK1 immunoreactivity in human cartilage and reduced TAK1 phosphorylation in the experimental models, while JNK, ERK, and p38 showed broadly similar patterns. The repeated pattern across these systems supports an association between WNA and TAK1/MAPK signaling.

The IL-1β challenge tested whether this molecular pattern was sensitive to inflammatory stimulation. IL-1β attenuated the changes associated with WNA, but the degree of reversal differed among the markers. The estimated reversal was 102.9% for TAK1 phosphorylation and 88.9% for JNK phosphorylation. It was 74.8% for p38 and 69.9% for ERK phosphorylation. IL-6 expression showed a 58.1% reversal. These percentages were calculated from group means and are descriptive measures. TAK1 returned to approximately the Model level after IL-1β challenge. JNK and p38 showed less complete reversal. IL-6 showed a smaller response. ERK also increased numerically, but the between-group difference was not significant. Its reversal percentage should therefore be interpreted with caution. These results show that the measured signals did not respond equally to the same inflammatory challenge.

The different reversal levels also help explain the signaling pattern. TAK1, the MAPK branches, and IL-6 did not respond to IL-1β to the same extent. This finding is not consistent with a simple linear pathway. Instead, several connected inflammatory pathways are likely involved.³⁷ The proteomic findings support this interpretation, showing enrichment of NF-κB, IL-17, TNF, and cytokine-related pathways in addition to MAPK. Thus, the IL-1β experiment provides additional evidence for the involvement of TAK1/MAPK signaling in the effects of WNA.

The present findings also add molecular context to previous studies of WNA. Clinical evidence suggests that WNA can improve pain and function in OA. Other forms of physical stimulation applied at acupuncture points, such as low-level laser therapy, have also been reported to improve knee pain and physical function in patients with KOA.³⁸ However, the quality of the available evidence remains variable.¹⁵ WNA combines acupuncture with local heat generated by moxibustion.³⁹ Experimental studies have linked moxibustion with changes in inflammatory responses in KOA models.⁴⁰ Clinical reviews have also reported potential benefits of moxibustion for KOA.^41,42 These findings support the biological relevance of both needle and thermal stimulation. However, our results reflect the combined WNA intervention. They cannot separate the effects of needling from those of heat.

The use of human cartilage is an important feature of this study. Our samples were obtained from patients with KL Grade III–IV KOA undergoing TKA and therefore represent established, advanced human disease. In Western clinical practice, conservative management of KOA includes pharmacological treatment, exercise-based rehabilitation, and physical modalities aimed primarily at reducing pain and improving function. Among these approaches, high-intensity laser therapy (HILT) has been reported to provide benefits in pain relief and functional improvement in patients with KOA.⁴³ However, conservative treatments are primarily intended for symptom management and should not be considered substitutes for TKA in patients with advanced KOA who meet surgical indications.⁴⁴ Accordingly, the 21-day preoperative WNA intervention used in the present study should also be interpreted as an adjunctive intervention rather than an alternative to indicated joint replacement. Our findings show that measurable molecular changes can occur in advanced KOA cartilage following preoperative WNA, providing biological evidence for molecular changes associated with preoperative WNA in advanced KOA.

These findings should be interpreted in light of several limitations. First, we did not directly manipulate TAK1 or individual MAPK branches. The results therefore support pathway involvement but do not establish causality or a hierarchy among MAPK branches. Second, IL-1β is a broad inflammatory stimulus rather than a pathway-specific rescue.⁴⁵ The reversal experiment cannot show that WNA acts only through TAK1/MAPK signaling. Third, the DMM model and primary chondrocytes cannot fully reproduce the complex environment of human KOA.^46,47 Fourth, the clinical study lacked component-specific intervention controls. No predefined washout period was used for previous nonsteroidal anti-inflammatory drug (NSAID) treatment, so residual drug effects cannot be excluded. We also lacked pretreatment proteomic profiles and an independent validation cohort. Thus, the proteomic findings cannot yet be used as predictive biomarkers. Future studies should include direct pathway manipulation, component controls, longitudinal sampling, and independent validation.

Conclusion

This study integrates human cartilage proteomics, a DMM-induced rat model, and primary chondrocyte analyses to characterize molecular changes associated with WNA in KOA. WNA attenuated mechanical pain hypersensitivity in DMM rats and was associated with reduced cartilage damage, lower inflammatory mediator levels, and decreased TAK1/MAPK-related signaling across clinical and experimental systems. Human proteomic analysis also showed coordinated changes in MAPK-related processes. These findings support the involvement of TAK1/MAPK-related inflammatory signaling in the response to WNA but do not establish a direct causal relationship between pathway inhibition and analgesia. Direct manipulation of TAK1 and individual MAPK branches is needed to define their specific roles in the biological and analgesic effects of WNA.

Funding Statement

This study was supported by the National Natural Science Foundation of China (Grant number [82474619]) and the Shanghai Action Plan for Inheritance, Innovation and Development of Traditional Chinese Medicine (Grant number [ZY(2025-2027)-3-2-1]). These grants provided financial support for this research. The funders had no role in study design, data collection and analysis, interpretation of results, manuscript preparation, or the decision to publish.

AI Disclosure Statement

During the preparation of this manuscript, the authors used Google Gemini 2.5 Pro (July 2026) to translate the manuscript from Chinese into English. The authors reviewed and verified all translated content to ensure accuracy and take full responsibility for the final manuscript. The AI tool was not used for any scientific content generation.

Data Sharing Statement

The data that support the findings of this study are available from the corresponding author, Yiqun Mi, upon reasonable request.

Ethics Statement

The human study was conducted in accordance with the Declaration of Helsinki and was reviewed and approved by the Ethics Committee of Shanghai Guanghua Hospital of Integrated Traditional Chinese and Western Medicine (Ethics No. 2025-K-97). Written informed consent was obtained from all participants prior to enrollment.

The animal experiments were approved by the Animal Ethics Committee of Shanghai Hospital of Traditional Chinese Medicine (Ethics No. 2024020). All animal procedures were conducted in accordance with the Guidelines for Ethical Review of Laboratory Animal Welfare of the People’s Republic of China (GB/T 35892-2018).

Author Contributions

Bo Xu and Qin Su contributed equally to this work and share first authorship.

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors declare that they have no conflicts of interest.