Abstract
Background:
Osteoarthritis (OA) is a globally prevalent degenerative joint disorder that imposes significant socioeconomic burdens. While traditionally viewed as a localized “wear-and-tear” disease, emerging evidence supports a systemic pathogenesis involving the gut-joint axis. The oral-gut-joint pathway remains underexplored in OA pathophysiology.
Objective:
This study aimed to characterize oral and gut microbiota signatures in OA patients and elucidate their functional connections to cartilage degeneration through multiomics integration.
Methods:
We conducted a cross-sectional observational study involving 25 OA patients and 20 healthy controls. 16S rDNA gene amplicon sequencing was performed on fecal and oropharyngeal swab samples. Cartilage tissues were subjected to transcriptomic and proteomic analyses.
Results:
We identified distinct dysbiosis patterns in both the gut and oral microbiomes of OA patients. The α-diversity of the gut microbiota significantly increased (P<0.05) with enrichment of Ruminococcaceae and Subdoligranulum. Concurrently, the oral microbiota showed increased α-diversity and activation of the lipopolysaccharide biosynthesis pathway. We constructed two significant cross-omics correlation modules: one linking gut microbes (Lachnospiraceae and Muribaculaceae) to cartilage inflammatory genes (MAPK11, ITGB3, CD55 and ANGPT2) and extracellular matrix remodelling proteins and another connecting gut microbes (Helicobacter, Pseudomonas, and Phocea) with CXCL14 and GNGT2.
Conclusion:
Our study revealed the dysbiotic characteristics of the oral-gut microbiome and its complex associations with pathological changes in cartilage. These findings offer novel mechanistic insights and potential therapeutic targets for microbiota-based precision interventions in OA.
1 Introduction
Osteoarthritis (OA), a chronic degenerative joint disease characterized by synovitis, cartilage degradation, bone remodeling, and osteophyte formation, is a leading global cause of pain and disability (Martel-Pelletier et al., 2016). A 2020 Global Burden of Disease study reported 595 million OA cases (7.6% of the global population), a number projected to increase exponentially by 2050 (Hunter and Bierma-Zeinstra, 2019; G.O. Collaborators, 2023). Traditionally regarded as a localized “wear-and-tear” condition driven by mechanical stress and aging, OA is increasingly recognized as having a systemic pathogenesis. Many OA patients without significant joint injury exhibit persistent low-grade systemic inflammation (Robinson et al., 2016), along with metabolic dysfunction and the activity of adipose tissue-derived inflammatory mediators (Binvignat et al., 2024; Herrero-Beaumont et al., 2024). Understanding the metabolic drivers of systemic inflammation has become essential for developing more effective therapeutic strategies.
As the most metabolically complex organ in the human body, the gut microbiota generates bioactive metabolites capable of regulating host physiology across multiple systemic axes, such as the gut-bone pathways (Borody and Campbell, 2012; Dong et al., 2025). Clinical observations indicate that the microbial community differs significantly in patients with knee OA, showing reduced abundance of short-chain fatty acid producers (e.g., Faecalibacterium and Lachnospiraceae) and expansion of proinflammatory Enterobacteriaceae (Sánchez Romero et al., 2021; Ning et al., 2022). This dysbiosis not only alters the local immune microenvironment but also remotely reshapes joint immunity through circulating metabolites (Sun et al., 2025). For example, studies in animal models have demonstrated that gut microbiome dysbiosis may correlate with OA progression by inducing infrapatellar fat pad-synovial membrane inflammation, and that increased gut permeability and intestinal inflammation may precede the onset of arthritis (Hecquet et al., 2023; Liu et al., 2025). Specific gut commensals, such as Faecalibacterium prausnitzii, have been shown to alleviate inflammatory arthritis by modulating the production of IL-17 and short-chain fatty acids (Moon et al., 2023). Lipopolysaccharide (LPS) activates synovial macrophage M1 polarization through the TLR4-MyD88 pathway, creating a proinflammatory feedback loop (Huang et al., 2020; Liu et al., 2023). Butyrate promotes peripheral Treg differentiation to suppress inflammation, and secondary bile acids regulate chondrocyte survival through vitamin D receptor signaling (Boer et al., 2019; Liu et al., 2019).
The oral microbiota, another potential inflammatory driver, has been underexplored in OA research. As the second most microbially dense human site, the oral cavity continuously interacts with the gut microbiota (Catalan et al., 2024). A large cohort study revealed a significant bidirectional relationship between periodontitis and OA in more than 140,000 patients (Ma et al., 2022). Mechanistically, oral commensals such as Fusobacterium nucleatum can secrete outer membrane vesicles that alter gut immune homeostasis upon ingestion and may be linked to OA (Hong et al., 2023). Products of oral pathogens can enter the bloodstream and induce systemic inflammation (Yamazaki and Kamada, 2024). This “oral-gut-joint” axis suggests that studying the gut or oral microbiota in isolation may not fully capture the microbial signature of OA.
Based on emerging evidence for the oral-gut-joint axis, we hypothesized that OA patients exhibit distinct oral-gut dysbiosis functionally associated with cartilage molecular changes. This study aims to map the relationship between oral-gut microbiota dysbiosis and joint degeneration in OA patients through multiomics integration, offering new experimental evidence for the systemic mechanisms of OA.
2 Results
2.1 Study workflow and patient clinical information
In this study, 36 OA patients and 25 healthy controls (HC) were included, with 45 participants (25 OA patients and 20 matched HCs) ultimately included for multiomics analysis. OA patients were visited and recruited at Weifang People’s Hospital between July 1, 2025, and October 15, 2025, and HCs were community-recruited. The experimental flowchart is shown in Figure 1. Baseline demographic data for the participants are provided in Supplementary Materials S1. Potential confounding factors, including body mass index, smoking status and drinking status, were collected. Potential confounders were considered in downstream analyses where appropriate, or their impact was acknowledged as a limitation. There were no significant differences in age (OA: 66.6 ± 7.9 vs. HC: 67.5 ± 6.0 years; p = 0.653), sex distribution (OA: 52% female vs. HC: 40% female; p = 0.616) and body mass index (OA: 27.64 ± 2.8 vs. HC: 26.27 ± 2.25; p = 0.076) between the two groups, confirming their comparability. Clinical samples (including cartilage from inflammatory and noninflammatory regions in OA patients and blood, feces, and oral swabs from all participants) were used for subsequent gut/oral microbiota analysis and cartilage transcriptomics/proteomics.
Figure 1
2.2 Gut and oral microbiota dysbiosis in OA patients
To explore the microbial characteristics of OA patients, 16S rDNA gene sequencing was performed on fecal and oral samples from OA patients and HCs. α-diversity analysis revealed that compared with HCs, OA patients had significantly greater observed, Chao1, ACE, J and Shannon indices (p < 0.05), with no significant differences in Simpson (Figure 2A). These findings indicated a significant increase in species richness in the gut microbiota of OA patients. However, the concentrations of the dominant bacteria in the two groups were similar. With respect to the oral microbiota, all α-diversity indices were significantly greater in OA patients than in healthy individuals (Figure 2B). β-diversity, such as PCoA, clearly distinguished the gut and oral microbiota between OA patients and HCs (p = 0.001, p = 0.002) (Figures 2C, D).
Figure 2
At the community composition level, Venn diagram of the gut microbiota revealed586 operational taxonomic units (OTUs)shared between OA patients and HCs, with 737 unique to OA patients and 441 unique to HCs (Figure 3A). Similar patterns were observed for the oral microbiota, with 584 shared OTUs, 500 unique to OA patients, and 496 unique to HCs (Figure 3B). LEfSe analysis revealed differentially abundant taxa between the two groups. In the gut microbiota, OA patients were enriched in f_Ruminococcaceae, s_un_g_Subdoligranulum, g_Subdoligranulum, etc. While HCs had higher abundances of c_Negativicutes, o_Selenomonadales, s_un_g_Megamonas, etc. (Figure 3C). In the oral microbiota, OA patients were enriched in c:Bacteroidia, p_Bacteroidota and p_Fusobacteriota, etc. Whereas HCs were enriched in c_Bacilli, f_Streptococcaceae and g_Streptococcus, etc. (Figure 3D).
Figure 3
PICRUSt2-based functional prediction revealed significant differences in metabolic pathways of the gut and oral microbiota between OA patients and HCs at the KEGG pathway level. In terms of the gut microbiota, OA patients differed significantly from HCs in terms of pathways such as apoptosis (Figure 3E). In terms of oral microbiota, OA patients showed predicted enrichment of pathways such as LPS biosynthesis and polycyclic aromatic hydrocarbon degradation (Figure 3F).
2.3 Cartilage transcriptomic and proteomic profiles
We performed transcriptomic analysis and proteomic analysis on paired cartilage samples from inflammatory and noninflammatory regions from OA patients. We performed differential expression analysis on transcriptional data from inflamed and noninflamed cartilage tissues of OA patients using DESeq2 [|log2(FoldChange)| ≥ 1, adjusted P ≤ 0.05]. A total of 353 differentially expressed genes (DEGs) were identified, including 142 upregulated genes and 211 downregulated genes (Figure 4A). The volcano plot shows a scattered distribution of DEGs between the two groups, indicating extensive and complex transcriptional changes under OA inflammation. Proteomic analysis using DIA-NN (fold change > 1.5, P ≤ 0.05) revealed 177 differentially expressed proteins (DEPs), 37 of which were upregulated and 140 of which were downregulated (Figure 4B). To assess the consistency between the transcriptomic and proteomic data, we conducted correlation analysis on commonly identified genes/proteins. The results revealed a weak positive correlation between protein expression and gene expression (Pearson = 0.166) (Figure 4C). A nine-quadrant plot was constructed for the joint analysis of the DEPs and DEGs (Figure 4D).
Figure 4
KEGG pathway enrichment analysis of the DEPs (Figure 4E) revealed significant pathways such as extracellular matrix (ECM)-receptor interaction, ascorbate and aldarate metabolism, pyruvate metabolism, and glutathione metabolism, suggesting pronounced ECM degradation, oxidative stress, and metabolic dysfunction in inflamed OA regions. KEGG analysis of the DEGs (Figure 4F) revealed significant pathways related to ECM-receptor interactions, cytokine-cytokine receptor interactions, the PI3K-Akt signaling pathway, and the regulation of lipolysis in adipocytes. Notably, inflammatory pathways were more significantly enriched in the transcriptomic data, whereas metabolic pathways were more strongly enriched in the proteomic data. A comparison of KEGG pathway enrichment between the proteomic and transcriptomic data (Figure 4G) revealed that metabolic pathways (glycosphingolipid biosynthesis and glycosaminoglycan degradation) were substantially enriched at the protein level, whereas the enrichment of ECM-receptor interactions was relatively balanced in both datasets.
2.4 Integrative analysis of gut microbiota-cartilage interactions
We performed an integrated analysis of the gut microbiota, cartilage transcriptome, and proteome in OA patients. By constructing a triomic correlation network, we identified two significantly correlated communities. The first community revealed significant correlations between the gut microbes g_Lachnospiraceae_FCS020_group and g_un_f_Muribaculaceae and the expression levels of genes (MAPK11, ITGB3, CD55, ANGPT2) and proteins (phosphoacetylglucosamine mutase isoform 1, etc.). g_Lachnospiraceae_FCS020_group was negatively correlated with MAPK11 and ITGB3 expression, whereas g_un_f_Muribaculaceaewas positively correlated with both genes. Genes such as MAPK11 are involved in cellular stress, inflammation, differentiation, angiogenesis, and autoimmunity, whereas proteins such as NP_001186846.1 are linked to cell signaling, adhesion, and angiogenesis (Figure 5A). The second community linked gut microbes (g_Helicobacter, g_Pseudomonas, and g_Phocea) with cartilage transcriptome genes (CXCL14, GNGT2) and proteome proteins (acyl-coA synthetase long chain family member 4). All these elements showed significant pairwise correlations. These genes were negatively correlated with GNGT2 and positively correlated with CXCL14 (Figure 5B). These genes play key roles in inflammation and cell communication, suggesting that these microbes may play a role in OA inflammation by regulating their expression. Proteins such as NP_001305438.1 (long-chain-fatty-acid-CoA ligase 4 isoform 2) are involved in inflammation and ECM remodeling, indicating potential microbial impacts on cartilage inflammation and degeneration.
Figure 5
3 Discussion
We identified distinct dysbiosis patterns in the gut and oral microbiota of OA patients. These multi-layered data support the existence of an “oral-gut-joint axis”, tentatively connecting specific gut microbes to cartilage inflammatory genes and matrix remodeling proteins.
Our study revealed that the α-diversity of the gut microbiota in OA patients was significantly greater than that in healthy controls, as reflected by increased richness and evenness indices. The unchanged Simpson diversity might suggest that this elevation derives from the expansion of rare taxa rather than dominant community restructuring. These findings differ from those of previous studies reporting reduced gut microbiota diversity in OA patients (Liu et al., 2022). We suspect this discrepancy stems from various factors, including patient geographical location, dietary habits, disease staging, comorbidities, and sequencing technologies (Yatsunenko et al., 2012; Falony et al., 2016). Geographically, our cohort was recruited from northern China, where dietary patterns may shape distinct gut microbial profiles compared to Western or southern Chinese populations. However, an increase in microbial diversity is not always indicative of health; it may reflect a decline in gut ecological stability and the opening of ecological niches, thereby providing opportunities for the enrichment of potential pathogens (Yao et al., 2024). Significant β-diversity separation indicates reproducible, disease-associated community restructuring between OA patients and healthy controls. β-diversity captures between-group compositional dissimilarity, and its divergence in our cohort suggests that OA is characterized by a consistent microbial community signature rather than random variation. The 737 unique OTUs identified in OA patients in this study might include more opportunistic pathogens with proinflammatory potential. At the taxonomic level, we observed enrichment of Ruminococcaceae in the gut of OA patients. Notably, members of the Ruminococcaceae family exhibit extremely high functional heterogeneity (Wang et al., 2023); some are major butyrate producers, whereas others are associated with gut inflammation and the presence of LPS (Joossens et al., 2011; Davis et al., 2021; Milton-Laskibar et al., 2022; Han et al., 2025).
Our study also revealed significant alterations in the oral microbiome of OA patients, characterized by a comprehensive increase in α-diversity and significant enrichment of the LPS biosynthesis pathway. As the second most microbially dense site in the human body, the oral cavity continuously interacts with the gut microbiota (Yeo et al., 2022). Research has confirmed a significant bidirectional relationship between periodontitis and OA (Ma et al., 2022). Mechanistically, oral pathobionts (e.g., Porphyromonas gingivalis) and their products can migrate to the gut via the “oral-gut-joint” axis, disrupting gut immune homeostasis (Lorenzo et al., 2019), and can also activate systemic inflammatory responses (de Oliveira et al., 2022; Yamazaki and Kamada, 2024). Nanovesicles carrying Porphyromonas gingivalis antigens have been detected in the plasma of periodontitis patients, providing direct evidence for the entry of oral pathogens and their products into systemic circulation (Farrugia et al., 2020; Wang et al., 2022). The predicted enrichment of the LPS biosynthesis pathway in the oral microbiota in this study further supports the possibility that orally derived LPS may be linked to OA inflammation through systemic routes.
In this study, differential expression analysis of transcriptomic and proteomic data from inflamed and noninflamed cartilage tissues from OA patients was performed. Proteomic data highlighted significant pathways such as ECM-receptor interaction, ascorbate and aldarate metabolism, and glutathione metabolism. The enrichment of the ECM-receptor interaction pathway suggests abnormal degradation and remodeling of the ECM in OA cartilage, which is consistent with characteristic cartilage damage in OA (Hao et al., 2021; Liu et al., 2022; Moore et al., 2022). Changes in glutathione metabolism point to the important role of oxidative stress in OA, which is considered a key driver of chondrocyte damage and apoptosis through the production of reactive oxygen species that impair cellular structure and function (Zhu et al., 2020; Zhao et al., 2023). Transcriptomic data, on the other hand, revealed inflammation-related pathways, including cytokine-cytokine receptor interactions, the PI3K-Akt signaling pathway, and the regulation of lipolysis in adipocytes. Upon binding to receptors on the cell membrane, many cytokines (e.g., interleukin-1β, tumor necrosis factor-α, and interleukin-6) activate the NF-κB signaling pathway through the PI3K/Akt and MAPK signaling pathways, thereby affecting OA (Lee et al., 2022). Integrated analysis revealed that ECM-receptor interaction and focal adhesion pathways were highly enriched in both the transcriptomic and proteomic datasets. These findings are highly consistent with reports in the literature (Sun et al., 2024) and provide a multilevel molecular perspective for understanding OA. The weak transcriptome-proteome correlation (Pearson = 0.166) may reflect several biological and technical factors. Post-transcriptional regulatory mechanisms, including microRNA-mediated mRNA degradation, variable translational efficiency, protein turnover dynamics, and post-translational modifications, commonly weaken the correspondence between mRNA and protein levels. Additionally, sample heterogeneity in cartilage tissue composition, such as variable chondrocyte density, extracellular matrix content, and inflammatory cell infiltration, may contribute to this divergence. These considerations suggest that transcriptomics and proteomics capture distinct yet complementary molecular layers, and their integration remains valuable for understanding OA cartilage pathophysiology.
In the constructed “microbiota-gene-protein” coexpression network, two significant correlation modules were identified. The first module linked the gut bacteria Lachnospiraceae and Muribaculaceae with the expression levels of genes (MAPK11, ITGB3, CD55, ANGPT2) and proteins (NP_001186846.1). The activation of MAPK11 (p38β), a member of the p38 MAPK family, is closely related to chondrocyte apoptosis and matrix metalloproteinase 13 expression, potentially serving as an important signaling pathway for OA inflammation and cartilage degradation (Zheng et al., 2024; Meng et al., 2025). ITGB3, a member of the integrin family, plays a crucial role in mechanosignal transduction between cells and the ECM; functional abnormalities in this gene may affect the response of chondrocytes to mechanical stress and ECM maintenance (Li et al., 2023; Lian et al., 2025). Although both Lachnospiraceae and Muribaculaceae are important short-chain fatty acid producers, Lachnospiraceae was negatively correlated with MAPK11 and ITGB3, whereas Muribaculaceae was positively correlated with MAPK11 and ITGB3. These findings raise the hypothesis that Lachnospiraceae and Muribaculaceae may be associated with chondrocyte inflammation through other metabolites (e.g., branched-chain fatty acids and amines), outer membrane vesicles, or other signaling molecules. The second module linked the gut microbes Helicobacter, Pseudomonas, and Phocea with cartilage transcriptome genes (CXCL14 and GNGT2) and proteome proteins (NP_001305438.1, etc.). Helicobacter and Pseudomonas are generally considered potential pathogens (Raman et al., 2018; Duan et al., 2025); their overgrowth in the gut may lead to impaired intestinal barrier function, promoting the entry of bacterial products such as LPS into the systemic circulation and triggering chronic low-grade systemic inflammation (Ning et al., 2024; Sun et al., 2025). This systemic inflammation is considered a significant driving factor in OA pathogenesis, exacerbating cartilage degradation and synovial inflammation (Sun et al., 2023). CXCL14 is a chemokine that is involved primarily in immune cell recruitment and inflammatory responses (Gowhari Shabgah et al., 2021). The positive correlation of Helicobacter and Pseudomonas with CXCL14 may indicate that these bacteria activate immune responses, leading to the upregulation of CXCL14 expression, thereby promoting the infiltration of inflammatory cells into the joint area and exacerbating the inflammatory process of OA.
Our study links oral-gut microbiome dysbiosis to OA cartilage degeneration. However, we must admit that the modest sample size limits the robustness of our multiomics correlation networks. The limited statistical power may increase false positive risks. The networks in Figure 5 should be interpreted as exploratory hypotheses requiring validation in larger cohorts. Additionally, this study did not conduct subgroup analyses for females and males. The cross-sectional design of this study prevents the determination of causality between dysbiosis and OA progression, and the sample size limits the statistical power of subgroup analyses. Furthermore, direct evidence for further mechanistic research is lacking; thus, cellular and animal experiments should be conducted in the future to further clarify the underlying mechanisms involved.
From a clinical perspective, our findings suggest that the oral-gut-joint axis may represent a modifiable therapeutic target in OA. The identification of reproducible microbial signatures and their correlation with cartilage inflammatory markers provides a rationale for future interventional studies, including microbiome-based biomarker development for risk stratification, targeted microbiome modulation as adjunctive therapy, and oral health management as a component of comprehensive OA care.
4 Materials and methods
4.1 Study design and participants
This cross-sectional case-control observational study included participants from Weifang People’s Hospital and the community between July 1, 2025, and October 15, 2025. A total of 45 participants (25 patients with knee OA and 20 matched HCs) were included in the multiomics analysis. The study was approved by the Medical Research Ethics Committee of Weifang People’s Hospital on October 23, 2024 (Approval No: KYLL20241023-1) and registered with the Chinese Clinical Trial Registry on June 23, 2025 (Registration number: ChiCTR2500104749). Written informed consent was obtained from all participants.
OA patients were diagnosed according to the Chinese Guidelines for the Diagnosis and Treatment of Osteoarthritis (2021 Edition), fulfilling clinical and radiographic criteria (Kellgren-Lawrence grade ≥ II) and experiencing active symptoms (mean pain score ≥ 4 on a 0–10 numeric rating scale over the past week) with a disease duration of at least 6 months. HCs had no clinical diagnosis of OA, rheumatoid arthritis, gout, or other systemic inflammatory joint diseases and reported no chronic knee pain, stiffness, or low back pain requiring medical attention during the past year.
The inclusion criteria for all participants included age over 18 years (no sex restrictions), willingness to cooperate, and provision of informed consent. The exclusion criteria included other serious diseases (e.g., rheumatoid arthritis or gout), active infection or severe infection within the past three months, immunosuppression or use of immunosuppressants, joint surgery within the past six months, current use of medications affecting joint inflammation (e.g., long-term high-dose steroids), pregnancy or breastfeeding, and severe mental disorders. To minimize potential biases, HCs were community-recruited and matched with OA patients for age and sex. Standardized protocols were employed for sample collection and laboratory analysis to reduce information bias.
For continuous variables including age and BMI, we employed independent samples t-tests when normality assumptions were met (assessed via Shapiro-Wilk testing), otherwise using Mann-Whitney U tests. Categorical variables were analyzed using Chi-square tests or Fisher’s exact probability tests where expected cell counts were low.
4.2 Sample collection and storage
All laboratory procedures were conducted at the Central Laboratory of Weifang People’s Hospital. Fecal samples were self-collected using sterile containers. Oral samples were collected from the posterior oropharynx using sterile swabs. Blood was collected in EDTA tubes. Discarded cartilage tissues were collected during surgery, rapidly frozen in liquid nitrogen, and stored at -80 °C. All the samples were transported on ice to the laboratory within 2 hours. Fecal and oral samples were stored at -80 °C until DNA extraction. Plasma was obtained from blood by centrifugation (3,000×g for 10 min), aliquoted, and stored at -80 °C. The tissue samples were kept at -80 °C until analysis.
4.3 16S rDNA gene sequencing and analysis
Microbial DNA was extracted using a QIAamp DNA Fecal Mini Kit. The V3-V4 hypervariable region was amplified by PCR using forward (5’-CCTACGGGNBGCASCAG-3’) and reverse (5’-GGACTACNVGGGTWTCTAAT-3’) primers. PCR products were purified and subjected to paired-end sequencing on an Illumina HiSeq 2500 platform.
Raw paired-end reads were assembled using FLASH (Magoč and Salzberg, 2011). Primers and low-quality reads were removed with Cutadapt (Martin, 2011). Chimera checking and OTU clustering were performed using usearch (Edgar, 2010). Representative OTU sequences were aligned against the silva_132_97_16S (DeSantis et al., 2006) database for taxonomic classification via the RDP (Cole et al., 2014) Classifier. Phylogenetic trees were constructed using FastTree (Price et al., 2009).
Functional prediction was performed using PICRUSt (Langille et al., 2013) to infer metagenomic function content from 16S data. α-diversity (Observed richness, Shannon, Simpson, ACE, Chao1, Pielou’s evenness (J’)) and β-diversity (Bray-Curtis distance) were calculated using the R package vegan (Oksanen et al., 2013). Statistical significance for α-diversity differences was determined by a nonparametric Kruskal-Wallis rank sum test with Benjamini-Hochberg correction. β-diversity differences were assessed using PERMANOVA with 999 permutations in the R package vegan (Oksanen et al., 2013). The LEfSe (Paulson et al., 2013) analysis was performed to identify taxa with differentiating abundance in the different group. LEfSe (Paulson et al., 2013) is an algorithm for high-dimensional biomarker discovery and explanation that identifies genomic features characterizing the differences between two of more biological conditions.
4.4 Proteomic analysis of cartilage
Cartilage tissue was ground in liquid nitrogen and lysed with SDT buffer. Lysates underwent centrifugation, heating, alkylation, and acetone precipitation. Protein pellets were dissolved in DB buffer and digested with trypsin (Wiśniewski et al., 2009). Peptides were desalted using C18 StageTips and separated on a PepMap™ Neo C18 column. MS analysis was performed on an Orbitrap Astral mass spectrometer with an Easy-Spray ion source. Mass spectrometry data were acquired in data-independent acquisition (DIA) mode, with a full MS scan range of m/z 380–980 and a resolution of 240,000. DIA settings included 300 windows of 2 Th, NCE 25%, MS2 range m/z 150-2000, and Astral resolution 80,000. Protein identification and quantification were performed using DIA-NN software (Demichev et al., 2020) against a protein database. DEPs were defined as those with a fold change greater or less than a specified value. Functional analysis included GO and KEGG enrichmentanalyses (Huang da et al., 2009; Franceschini et al., 2013). Detailed protocols and quality control data are provided in the Supplementary Materials S2.
4.5 Transcriptomic analysis of cartilage
Total RNA was extracted using TRIzol®. RNA integrity was verified on an Agilent Bioanalyzer. Libraries were prepared using the NEBNext® Ultra™ II Directional RNA Library Prep Kit and sequenced on an Illumina NovaSeq 6000 platform (150 bp paired-end). Clean reads were aligned to the human reference genome (GRCh38.p14) using HISAT2 (v2.2.1). Ribosomal RNA was depleted from total RNA. Transcript assembly and quantification were performed with StringTie (v2.2.3). Gene expression was calculated as FPKM values using featureCounts (v2.0.6). Differential expression analysis was conducted using DESeq2 (v1.42.0), which implements the Benjamini-Hochberg false discovery rate (FDR) method for multiple testing correction via the independent hypothesis weighting approach. Genes with adjusted p-values (padj) ≤ 0.05 and |log2FoldChange| ≥ 1 were considered significantly differentially expressed. GO and KEGG enrichment analyses were performed using clusterProfiler (4.8.1), with corrected P values less than 0.05 considered significant. Detailed protocols and quality control data are provided in the Supplementary Materials S2.
4.6 Integrated multiomics analysis
Pairwise correlation analysis between microbial, gene expression, and protein data was performed using the R package Hmisc. Prior to analysis, each omics data type was preprocessed with specific normalization methods: for microbiome compositional data, centered log-ratio transformation was applied using the compositions package in R to address compositional effects. For transcriptomics data (gene expression), counts were normalized using DESeq2’s variance-stabilizing transformation, followed by Z-score standardization across samples. For proteomics data, abundance values were log2-transformed and then Z-score standardized. All datasets were checked to ensure consistent sample names and order. Correlation coefficients and corresponding P-value matrices were obtained using the rcorr function. To correct for multiple testing, the Benjamini-Hochberg FDR method was applied. A “microbiota-gene-protein” interaction network was constructed by retaining pairwise associations that met the following thresholds: Spearman’s correlation coefficient |ρ| > 0.6 and FDR-adjusted P < 0.05. Only correlations meeting these criteria in all three pairwise comparisons (microbiota-gene, microbiota-protein, gene-protein) were retained for network construction. The resulting significant associations were visualized in Cytoscape software, where network layout (force-directed), node attributes (shape, size), and edge colors (positive: red; negative: blue) were adjusted to illustrate complex relationships and identify key nodes or modules.
5 Conclusion
Our integrated multiomics study constructed a “microbiota-gene-protein” correlation network to elucidate the dysbiotic features of the oral and gut microbiomes in OA patients and their complex functional connections to pathological changes in cartilage. The constructed “microbiota-gene-protein” correlation network revealed two significant cross-omics modules.
Statements
Data availability statement
The raw 16S rDNA sequencing data have been deposited in the NCBI SRA under BioProject accession number PRJNA1467634 and PRJNA1467632. The raw transcriptomic data have been deposited in GEO and the accession number is GSE331536. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (https://proteomecentral.proteomexchange.org) via the iProX partner repository (Ma et al., 2019; Chen et al., 2022) with the dataset identifier PXD078593.
Ethics statement
The studies involving humans were approved by the Medical Research Ethics Committee of Weifang People’s Hospital. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.
Author contributions
YL: Visualization, Formal analysis, Conceptualization, Data curation, Writing – original draft. JG: Formal analysis, Writing – original draft, Data curation, Visualization, Conceptualization. YZ: Writing – original draft, Resources, Funding acquisition, Validation. HW: Writing – original draft, Visualization, Data curation. HF: Writing – review & editing, Conceptualization, Supervision, Project administration.
Funding
The author(s) declared that financial support was received for this work and/or its publication. Funding for this study was provided by the Shandong Provincial Natural Science Foundation, China (grant number ZR2023MH381) (Yuying Zhang).
Acknowledgments
We are deeply thankful to all the participants and collaborators for their engagement in this study and to AJE (www.aje.com) for expert language editing.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb.2026.1833218/full#supplementary-material
Abbreviations
OA, Osteoarthritis; HC, Healthy Control; 16S rDNA, 16S ribosomal DNA; OTU, Operational Taxonomic Unit; DEGs, Differentially Expressed Genes; DEPs, Differentially Expressed Proteins; ECM, Extracellular Matrix; LPS, Lipopolysaccharide; DIA, Data-Independent Acquisition; FDR, False Discovery Rate.
References
1
BinvignatM.Herrero-BeaumontG.RannouF.SellamJ.BerenbaumF.ValdesA. M.et al. (2024). The role of obesity and adipose tissue dysfunction in osteoarthritis pain. Nat. Rev. Rheumatol.20, 565–584. doi: 10.1038/s41584-024-01143-3
2
BoerC. G.RadjabzadehD.Medina-GomezC.GarmaevaS.SchiphofD.ArpP.et al. (2019). Intestinal microbiome composition and its relation to joint pain and inflammation. Nat. Commun.10, 4881. doi: 10.1038/s41467-019-12873-4
3
BorodyT. J.CampbellJ. (2012). Fecal microbiota transplantation: techniques, applications, and issues. Gastroenterol. Clinics North. America41, 781–803. doi: 10.1016/j.gtc.2012.08.008
4
CatalanE. A.Seguel-FuentesE.FuentesB.Aranguiz-VarelaF.Castillo-GodoyD. P.Rivera-AsinE.et al. (2024). Oral pathobiont-derived outer membrane vesicles in the oral-gut axis. Int. J. Mol. Sci.25 (20), 11141. doi: 10.3390/ijms252011141
5
ChenT.MaJ.LiuY.ChenZ.XiaoN.LuY.et al. (2022). iProX in 2021: connecting proteomics data sharing with big data. Nucleic Acids Res.50, D1522–D1527. doi: 10.1093/nar/gkab1081
6
ColeJ. R.WangQ.FishJ. A.ChaiB.McGarrellD. M.SunY.et al. (2014). Ribosomal Database Project: data and tools for high throughput rRNA analysis42, D633–D642. doi: 10.1093/nar/gkt1244
7
DavisJ. A.CollierF.MohebbiM.PascoJ. A.ShivappaN.HébertJ. R.et al. (2021). The associations of butyrate-producing bacteria of the gut microbiome with diet quality and muscle health. Gut Microbiome (Cambridge England)2, e2. doi: 10.1017/gmb.2021.2
8
DemichevV.MessnerC. B.VernardisS. I.LilleyK. S.RalserM.. (2020). DIA-NN: neural networks and interference correction enable deep proteome coverage in high throughput. Nat. Methods17, 41–44. doi: 10.1038/s41592-019-0638-x
9
de OliveiraR. C. G.GardevE.ShaddoxL. M. (2022). Dysbiotic relationship between arthritis and the oral-gut microbiome. A critical review. J. Periodontal Res.57, 711–723. doi: 10.1111/jre.13002
10
DeSantisT. Z.HugenholtzP.LarsenN.RojasM.BrodieE. L.KellerK.et al. (2006). Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl. Environ. Microbiol.72, 5069–5072. doi: 10.1128/aem.03006-05
11
DongX.LinL.LiuL.ChenL.ShangN.WangR.et al. (2025). Oral administration of Clostridium butyricum alleviates high-fat diet-induced obesity in mice by modulating gut Akkermansia muciniphila abundance via direct growth promotion. J. Microbiol. Biotechnol.35, e2509005. doi: 10.4014/jmb.2509.09005
12
DuanY.XuY.DouY.XuD. (2025). Helicobacter pylori and gastric cancer: mechanisms and new perspectives. J. Hematol. Oncol.18, 10. doi: 10.1186/s13045-024-01654-2
13
EdgarR. C. (2010). Search and clustering orders of magnitude faster than BLAST. Bioinformatics26, 2460–2461. doi: 10.1093/bioinformatics/btq461
14
FalonyG.JoossensM.Vieira-SilvaS.WangJ.DarziY.FaustK.et al. (2016). Population-level analysis of gut microbiome variation. Science352, 560–564. doi: 10.1126/science.aad3503
15
FarrugiaC.StaffordG. P.MurdochC. (2020). Porphyromonas gingivalis outer membrane vesicles increase vascular permeability. J. Dent. Res.99, 1494–1501. doi: 10.1177/0022034520943187
16
FranceschiniA.SzklarczykD.FrankildS.KuhnM.SimonovicM.RothA.et al. (2013). STRING v9.1: protein-protein interaction networks, with increased coverage and integration. Nucleic Acids Res.41, D808–D815. doi: 10.1093/nar/gks1094
17
G.O. Collaborators (2023). Global, regional, and national burden of osteoarthritis, 1990-2020 and projections to 2050: a systematic analysis for the Global Burden of Disease Study 2021. Lancet Rheumatol.5, e508–e522. doi: 10.1016/S2665-9913(23)00163-7
18
Gowhari ShabgahA.Haleem Al-QaimZ.MarkovA.Valerievich YumashevA.EzzatifarF.AhmadiM.et al. (2021). Chemokine CXCL14; a double-edged sword in cancer development. Int. Immunopharmacol.97, 107681. doi: 10.1016/j.intimp.2021.107681
19
HanJ.MengX.KongH.LiX.ChenP.ZhangX. A. (2025). Links between short-chain fatty acids and osteoarthritis from pathology to clinic via gut-joint axis. Stem Cell. Res. Ther.16, 251. doi: 10.1186/s13287-025-04386-3
20
HaoX.ShangX.LiuJ.ChiR.ZhangJ.XuT. (2021). The gut microbiota in osteoarthritis: where do we stand and what can we do? Arthritis Res. Ther.23, 42. doi: 10.1186/s13075-021-02427-9
21
HecquetS.TotosonP.MartinH.AlgrosM. P.SaasP.Pais-de-BarrosJ. P.et al. (2023). Increased gut permeability and intestinal inflammation precede arthritis onset in the adjuvant-induced model of arthritis. Arthritis Res. Ther.25, 95. doi: 10.1186/s13075-023-03069-9
22
Herrero-BeaumontG.et al. (2024). Systemic osteoarthritis: the difficulty of categorically naming a continuous condition. Aging Clin. Exp. Res.36, 45. doi: 10.1007/s40520-024-02714-w
23
HongM.LiZ.LiuH.ZhengS.ZhangF.ZhuJ.et al. (2023). Fusobacterium nucleatum aggravates rheumatoid arthritis through FadA-containing outer membrane vesicles. Cell. Host Microbe31, 798–810.e7. doi: 10.1016/j.chom.2023.03.018
24
HuangZ.ChenJ.LiB.ZengB.ChouC. H.ZhengX.et al. (2020). Faecal microbiota transplantation from metabolically compromised human donors accelerates osteoarthritis in mice. Ann. Rheumatic Dis.79, 646–656. doi: 10.1136/annrheumdis-2019-216471
25
Huang daW.ShermanB. T.LempickiR. A. (2009). Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res.37, 1–13. doi: 10.1093/nar/gkn923
26
HunterD. J.Bierma-ZeinstraS. (2019). Osteoarthritis. Lancet393, 1745–1759. doi: 10.1016/j.berh.2011.11.008
27
JoossensM.HuysG.CnockaertM.De PreterV.VerbekeK.RutgeertsP.et al. (2011). Dysbiosis of the faecal microbiota in patients with Crohn's disease and their unaffected relatives. Gut60, 631–637. doi: 10.1136/gut.2010.223263
28
LangilleM. G.ZaneveldJ.CaporasoJ. G.McDonaldD.KnightsD.ReyesJ. A.et al. (2013). Predictive functional profiling of microbial communities using 16S rRNA marker gene sequences. Nat. Biotechnol.31, 814–821. doi: 10.1038/nbt.2676
29
LeeY. T.YunusM. H.M.UgusmanA.YazidM. D. (2022). Natural compounds affecting inflammatory pathways of osteoarthritis. Antioxidants (Basel Switzerland)11 (9), 1722. doi: 10.3390/antiox11091722
30
LiQ.ChenG.JiangH.DaiH.LiD.ZhuK.et al. (2023). ITGB3 promotes cisplatin resistance in osteosarcoma tumors. Cancer Med.12, 8452–8463. doi: 10.1002/cam4.5585
31
LianQ.LiuH.LiJ.LuoC.LiuC.ZhaoH.et al. (2025). Enhancing radiosensitivity of osteosarcoma by ITGB3 knockdown: a mechanism linked to enhanced osteogenic differentiation status through JNK/c-JUN/RUNX2 pathway activation. J. Exp. Clin. Cancer Res. CR44, 159. doi: 10.1186/s13046-025-03417-4
32
LiuY.DingW.WangH. L.DaiL. L.ZongW. H.WangY. Z.et al. (2019). Gut microbiota and obesity-associated osteoarthritis. Osteoarthritis Cartilage27, 1257–1265. doi: 10.1016/j.joca.2019.05.009
33
LiuS.LiG.ZhuY.XuC.YangQ.XiongA.et al. (2022). Analysis of gut microbiome composition, function, and phenotype in patients with osteoarthritis. Front. Microbiol.13, 980591. doi: 10.3389/fmicb.2022.980591
34
LiuL.et al. (2022). The effects and significance of gut microbiota and its metabolites on the regulation of osteoarthritis: close coordination of gut-bone axis. Front. Nutr.9, 1012087. doi: 10.3389/fnut.2022.1012087
35
LiuS.LiG.XuH.WangQ.WeiY.YangQ.et al. (2023). Cross-talk" between gut microbiome dysbiosis and osteoarthritis progression: a systematic review. Front. Immunol.14, 1150572. doi: 10.3389/fimmu.2023.1150572
36
LiuS.XuH.LiuL.MaW.FanH.LiuF.et al. (2025). Gut microbiome dysbiosis accelerates osteoarthritis progression by inducing IFP-SM inflammation in "double-hit" mice. Arthritis Res. Ther.27, 137. doi: 10.1186/s13075-025-03602-y
37
LorenzoD.GianvincenzoZ.Carlo LucaR.KaranG.JorgeV.RobertoM.et al. (2019). Oral-gut microbiota and arthritis: is there an evidence-based axis? J. Clin. Med.8 (10). doi: 10.3390/jcm8101753
38
MaJ.ChenT.WuS.YangC.BaiM.ShuK.et al. (2019). iProX: an integrated proteome resource. Nucleic Acids Res.47, D1211–D1217. doi: 10.1093/nar/gky869
39
MaK. S.LaiJ. N.ThotaE.YipH. T.ChinN. C.WeiJ. C.et al. (2022). Bidirectional relationship between osteoarthritis and periodontitis: a population-based cohort study over a 15-year follow-up. Front. Immunol.13, 909783. doi: 10.3389/fimmu.2022.909783
40
MagočT.SalzbergS. L. (2011). FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics27, 2957–2963. doi: 10.1093/bioinformatics/btr507
41
Martel-PelletierJ.BarrA. J.CicuttiniF. M.ConaghanP. G.CooperC.GoldringM. B.et al. (2016). Osteoarthritis. Nat. Rev. Dis. Primers2, 16072. doi: 10.1016/j.joca.2003.10.002
42
MartinM. (2011). Cutadapt removes adapter sequences from high-throughput sequencing reads. Embnet J.17, 10–12. doi: 10.14806/ej.17.1.200
43
MengC.ChenS.LiuQ.XuH.HeY.QuY.et al. (2025). Therapeutic potential of CHI3L1 in osteoarthritis: inhibition of cartilage matrix degradation and inflammation through TLR4-MAPK-STAT1 pathway. Int. Immunopharmacol.156, 114684. doi: 10.1016/j.intimp.2025.114684
44
Milton-LaskibarI.Cuevas-SierraA.PortilloM. P.MartinezJ. A.. (2022). Effects of resveratrol administration in liver injury prevention as induced by an obesogenic diet: role of Ruminococcaceae. Biomedicines10 (8), 1797. doi: 10.3390/biomedicines10081797
45
MoonJ.LeeA. R.KimH.JhunJ.LeeS. Y.ChoiJ. W.et al. (2023). Faecalibacterium prausnitzii alleviates inflammatory arthritis and regulates IL-17 production, short chain fatty acids, and the intestinal microbial flora in experimental mouse model for rheumatoid arthritis. Arthritis Res. Ther.25, 130. doi: 10.1186/s13075-023-03118-3
46
MooreL. Q.PanZ.BrottoM. (2022). RNAseq of osteoarthritic synovial tissues: systematic literary review. Front. Aging3. doi: 10.3389/fragi.2022.836791
47
NingY.HuM.GongY.HuangR.XuK.ChenS.et al. (2022). Comparative analysis of the gut microbiota composition between knee osteoarthritis and Kashin-Beck disease in Northwest China. Arthritis Res. Ther.24, 129. doi: 10.1186/s13075-022-02819-5
48
NingP.et al. (2024). Potential role of gut-related factors in the pathology of cartilage in osteoarthritis. Front. Nutr.11, 1515806. doi: 10.3389/fnut.2024.1515806
49
OksanenJ.BlanchetF. G.KindtR.LegendreP.MinchinP. R.O'HaraR. B.et al. (2013). Vegan: community ecology package. R package version. 2.0-10. J. Stat. Software48, 1–21. doi: 10.32614/cran.package.vegan
50
PaulsonJ. N.StineO. C.BravoH. C.PopM.. (2013). Differential abundance analysis for microbial marker-gene surveys. Nat. Methods10, 1200–1202. doi: 10.1038/nmeth.2658
51
PriceM. N.DehalP. S.ArkinA. P. (2009). FastTree: computing large minimum evolution trees with profiles instead of a distance matrix. Mol. Biol. Evol.26, 1641–1650. doi: 10.1093/molbev/msp077
52
RamanG.AvendanoE. E.ChanJ.MerchantS.PuzniakL. (2018). Risk factors for hospitalized patients with resistant or multidrug-resistant Pseudomonas aeruginosa infections: a systematic review and meta-analysis. Antimicrobial Resistance Infection Control7, 79. doi: 10.1186/s13756-018-0370-9
53
RobinsonW. H.LepusC. M.WangQ.RaghuH.MaoR.LindstromT. M.et al. (2016). Low-grade inflammation as a key mediator of the pathogenesis of osteoarthritis. Nat. Rev. Rheumatol.12, 580–592. doi: 10.1038/nrrheum.2016.136
54
Sánchez RomeroE. A.Meléndez OlivaE.Alonso PérezJ. L.Martín PérezS.TurroniS.MarcheseL.et al. (2021). Relationship between the gut microbiome and osteoarthritis pain: review of the literature. Nutrients13 (3), 716. doi: 10.3390/nu13030716
55
SunC.ZhouX.GuoT.MengJ. (2023). The immune role of the intestinal microbiome in knee osteoarthritis: a review of the possible mechanisms and therapies. Front. Immunol.14, 1168818. doi: 10.3389/fimmu.2023.1168818
56
SunN.ZhaoY.ZhangA.HeY. (2024). Identification of HTRA1, DPT and MXRA5 as potential biomarkers associated with osteoarthritis progression and immune infiltration. BMC Musculoskeletal Disord.25, 647. doi: 10.1186/s12891-024-07758-7
57
SunN.et al. (2025). Gut microbiota and osteoarthritis: epidemiology, mechanistic analysis, and new horizons for pharmacological interventions. Front. Cell. Infect. Microbiol.15, 1605860. doi: 10.3389/fcimb.2025.1605860
58
WangF.PengL.GuF.HuangP.ChengB.ChenG.et al. (2022). Circulating small extracellular vesicles from patients with periodontitis contribute to development of insulin resistance. J. Periodontology93, 1902–1915. doi: 10.1002/jper.22-0035
59
WangY.LiY.BoL.ZhouE.ChenY.NaranmandakhS.et al. (2023). Progress of linking gut microbiota and musculoskeletal health: casualty, mechanisms, and translational values. Gut Microbes15, 2263207. doi: 10.1080/19490976.2023.2263207
60
WiśniewskiJ. R.ZougmanA.NagarajN.MannM.. (2009). Universal sample preparation method for proteome analysis. Nat. Methods6, 359–362. doi: 10.1038/nmeth.1322
61
YamazakiK.KamadaN. (2024). Exploring the oral-gut linkage: interrelationship between oral and systemic diseases. Mucosal Immunol.17, 147–153. doi: 10.1016/j.mucimm.2023.11.006
62
YaoX.NieW.ChenX.ZhangJ.WeiJ.QiuY.et al. (2024). Two Enterococcus faecium isolates demonstrated modulating effects on the dysbiosis of mice gut microbiota induced by antibiotic treatment. Int. J. Mol. Sci.25 (10), 5405. doi: 10.3390/ijms25105405
63
YatsunenkoT.ReyF. E.ManaryM. J.TrehanI.Dominguez-BelloM. G.ContrerasM.et al. (2012). Human gut microbiome viewed across age and geography. Nature486, 222–227. doi: 10.1038/nature11053
64
YeoL. F.LeeS. C.PalanisamyU. D.KhalidB.AyubQ.LimS. Y.et al. (2022). The oral, gut microbiota and cardiometabolic health of indigenous Orang Asli communities. Orang Asli Health Microbiota12, 812345. doi: 10.3389/fcimb.2022.812345
65
ZhaoB.YuJ.LuoY.XieM.QuC.ShiQ.et al. (2023). Deficiency of S100 calcium binding protein A9 attenuates vascular dysfunction in aged mice. Redox Biol.63, 102721. doi: 10.1016/j.redox.2023.102721
66
ZhengX.WangY.QiuX. (2024). Comprehensive analysis of MAPK genes in the prognosis, immune characteristics, and drug treatment of renal clear cell carcinoma using bioinformatic analysis and Mendelian randomization. Eur. J. Pharmacol.980, 176840. doi: 10.1016/j.ejphar.2024.176840
67
ZhuS.MakosaDMillerBGriffinTM. (2020). Glutathione as a mediator of cartilage oxidative stress resistance and resilience during aging and osteoarthritis. Connective Tissue Res.61, 34–47. doi: 10.1080/03008207.2019.1665035
Summary
Keywords
16S rDNA, gastrointestinal microbiome, multi-omics research, oral-gut-joint axis, osteoarthritis, proteomics, transcriptome
Citation
Liu Y, Gong J, Zhang Y, Wang H and Feng H (2026) The oral-gut-joint axis in osteoarthritis: a multiomics case-control study. Front. Cell. Infect. Microbiol. 16:1833218. doi: 10.3389/fcimb.2026.1833218
Received
18 March 2026
Revised
25 May 2026
Accepted
31 May 2026
Published
22 June 2026
Volume
16 - 2026
Edited by
Siyuan Ma, Vanderbilt University Medical Center, United States
Reviewed by
Feza Korkusuz, Hacettepe University, Türkiye
Monica Gulati, Lovely Professional University, India
Ke You, University of Macau, China
Updates
Copyright
© 2026 Liu, Gong, Zhang, Wang and Feng.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Haotian Feng, sd.feng@outlook.com
†These authors have contributed equally to this work
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.