Abstract

Objective

   With the aim of clarifying the therapeutic mechanisms of the American
   Ginseng-Achyranthes bidentata (AG&A) herbal pair in primary Sjögren’s
   syndrome (pSS), this study employs an integrated approach combining
   network pharmacology, molecular docking, molecular dynamics
   simulations, and animal experiments.

Methods

   Network pharmacology & LC-MS/MS was utilized to identify the active
   components and potential targets of A&A. Molecular docking and dynamics
   simulations were performed to evaluate binding affinity and complex
   stability with key targets. Animal experiments using non-obese diabetic
   (NOD) mice were conducted to validate symptom improvement by critical
   active components.

Results

   Network pharmacology identified baicalin and quercetin as key active
   components. Molecular docking revealed strong binding affinities
   (binding energy ≤ -8.0 kcal/mol) between these compounds and
   apoptosis-related proteins, BAX and CASP3. Molecular dynamics
   simulations confirmed the stability of these complexes. Animal
   experiments demonstrated that baicalin can significantly reduce
   inflammatory cytokines of IL-18, TNF-α, IFN-α, and IFN-β,CXCL-10 (p <
   0.05), decrease mtDNA release, and downregulate cGAS-STING
   pathway-related proteins including cGAS, STING, CASP3, ZBP1, TBK1,
   p-STING, p-TBK1, IRF3, p-IRF3 and BAX.

Conclusion

   The critical components baicalin and quercetin from AG&A, particularly
   in aqueous extracts, exhibit therapeutic efficacy against pSS. This
   study provides experimental evidence for their action mechanism through
   modulating the mtDNA-cGAS-STING pathway. While highlighting their
   therapeutic potential, additional in vivo and clinical studies are
   warranted to validate these findings.

   Keywords: network pharmacology, molecular docking, cGAS-STING pathway,
   mtDNA, Sjögren’s syndrome

1. Introduction

   Sjögren’s syndrome (SS) is a chronic autoimmune disorder characterized
   by the involvement of exocrine glands, particularly the salivary and
   lacrimal glands, leading to common clinical symptoms such as dry mouth
   and eyes ([37]1). With the prevalence ranging from 0.29% to 0.77%, SS
   has exhibited an increasing trend observed in recent years ([38]2).
   This condition predominantly affects women over the age of 40, with a
   male-to-female incidence ratio of approximately 1:9 ([39]3). Patients
   with SS are at a higher risk for complications and multisystem damage,
   with about 30% to 40% experiencing extra-glandular manifestations such
   as interstitial lung disease and hematologic issues. Additionally,
   recent studies indicate that SS patients have an elevated risk of
   developing other tumors compared to the healthy population ([40]4).

   Over the years, individuals with SS have been burdened by the disease,
   leading to a reduced quality of life, yet treatment remains mainly
   symptomatically and empirically managed, with no biologically or
   chemically synthesized antirheumatic agents available to modify the
   disease process ([41]5). Hence, it is urgently needed to identify
   effective preventive and therapeutic strategies for SS, which holds
   substantial clinical and socioeconomic significance.

   The pathogenesis of SS is multifactorial, involving genetic,
   environmental, hormonal, and immune elements ([42]6). Although research
   utilizing genome-wide association studies, epigenomics,
   transcriptomics, and metabolomics has been extensively conducted
   ([43]7), the complete mechanisms underlying the disease remain elusive.
   Current treatments in modern medicine tend to focus on localized
   symptoms, with artificial fluid replacement therapies commonly used to
   alleviate dryness. However, these interventions often yield suboptimal
   results ([44]8), and there are significant challenges in systemic
   treatment and disease progression management, compounded by safety
   concerns. Consequently, many patients suffer from persistent clinical
   symptoms that are not adequately managed.

   Hydroxychloroquine (HCQ) is the most frequently prescribed
   immunomodulatory agent for pSS, primarily acting through interfering
   Toll-like receptor (TLR) signaling and inhibiting type I interferon
   (IFN) pathway. Although HCQ demonstrates clinical efficacy in reducing
   disease activity and is extensively used in clinical treatment, its
   therapeutic effectiveness in pSS remains incompletely characterized. In
   addition, current evidence regarding HCQ’s efficacy in pSS patients
   shows conflicting results. Hence, it is imperative to identify
   alternative therapeutic agents for treatment ([45]9).

   In prior studies, the efficacy of American Ginseng and Achyranthes
   (AG&A) in treating SS has been highlighted ([46]10, [47]11). Therefore,
   the present study aims to explore the mechanisms by which AG&A exerts
   its therapeutic effects in SS treatment through network pharmacology,
   LC-MS/MS and molecular docking methodologies, and validate the
   enrichment results of network pharmacology through in vivo experiments
   in NOD mice.

2. Methods

2.1. Network pharmacology

2.1.1. Composition of PEs and SS target acquisition

   We identified the chemical compositions of the effective
   pharmacological extracts (PEs) using the methods outlined above. The
   key active ingredients in the remaining PEs were screened via the
   Traditional Chinese Medicine Systems Pharmacology (TCMSP) database
   ([48]12), with oral bioavailability (OB) ≥ 30% and drug-likeness (DL)
   ≥0.18 the criteria. The SMILES format for these components was
   collected from the PubChem database ([49]13). Ingredient targets,
   defined as active ingredients with a probability greater than zero,
   were obtained from the Swiss Target Prediction database, and any
   duplicate targets were subsequently removed ([50]14).

2.1.2. Analysis of PE components and SS targets

   Targets associated with SS were sourced from the GeneCards ([51]15) and
   Online Mendelian Inheritance in Man (OMIM)) ([52]16) databases. The
   intersection of SS-related targets with the identified targets was
   considered as potential therapeutic targets. Then, we utilized
   Cytoscape 3.9.1 software to construct the network linking active
   ingredients, potential targets, and the disease associated with
   effective PEs for SS treatment.

2.1.3. Construction of the protein–protein interaction network

   All targets related to PEs and SS were entered into the STRING database
   ([53]17) to construct the PPI network. The minimum interaction score
   was set to above 0.9, with free nodes hidden and all other settings
   left as default.

2.1.4. GO and KEGG enrichment analysis

   To further elucidate the mechanisms of the key targets of the effective
   PEs in treating SS, these targets were analyzed using the DAVID
   database ([54]18).

2.1.5. LC-MS/MS analysis of American ginseng and achyranthes components

   An aliquot (150 µL) of the herbal formula sample was mixed with 150 µL
   of 70% methanol (containing internal standards, 2 µg/mL). The mixture
   was vortexed for 1 min, ultrasonically extracted in an ice-water bath
   for 60 min, and centrifuged at 12,000 rpm (4°C) for 10 min. The
   supernatant (200 µL) was transferred to an LC-MS vial and then
   separated on an ACQUITY UPLC HSS T3 column (100 mm × 2.1 mm, 1.8 µm;
   Waters) at 45°C for subsequent analysis. Mobile phases consisted of (A)
   water with 0.1% formic acid and (B) acetonitrile. The gradient elution
   was: 0–2 min, 5% B; 2–8 min, 5–50% B; 8–10 min, 50–80% B; 10–14 min,
   80–100% B; hold until 15 min; return to initial conditions at 15.1 min.
   The flow rate was set to 0.35 mL/min with a 2-µL injection volume. The
   analysis was conducted using a Thermo Orbitrap QE HF mass spectrometer
   equipped with a heated electrospray ionization (HESI) source. Data were
   acquired in data-dependent acquisition (DDA) mode, alternating between
   full MS scans (resolution: 60,000; m/z range: 100–1500) and MS/MS scans
   of the top 8 ions (resolution: 15,000). For the key parameters, raw
   data were processed using XCMS software (v4.5.1) for peak detection,
   alignment, and integration. Components were identified by matching
   against the proprietary LuMet-TCM database (>5,000 herbal standards)
   with the following criteria: Extracted ion chromatograms (EIC) and
   annotated MS/MS spectra were generated for verification. Compounds with
   a total identification score ≥40 were retained for analysis.

2.2. Molecular docking and molecular dynamic simulation

2.2.1. Molecular docking

   The 3D structure files of the target protein were retrieved from the
   RCSB Protein Data Bank ([55]19). Structural preparation for molecular
   docking was performed using AutoDockTools 1.5.7.

2.2.2. Molecular dynamics simulation

   The molecular dynamics simulations were conducted utilizing Gromacs
   ([56]20), under static conditions at a temperature of 310 K and an
   atmospheric pressure of 1 bar. For preprocessing small molecules, the
   potential data were incorporated into the topology files of the
   molecular dynamics system ([57]21). The Amber99sb-ildn force field was
   employed, with water molecules modeled using the Tip3p water model as
   the solvent.

2.3. Laboratory animal and groups

   In this study, NOD mice [a well-established Sjögren’s syndrome animal
   model documented in scientific literature ([58]22)] were used as the
   disease model and C57BL/6 mice as the healthy control group(6-10weeks,
   males, Beijing Beiyou Biotechnology Co., Ltd). We randomly divided them
   into six groups (n=10 per group), with oral gavage applied for all drug
   administration.

   Specifically, they were Normal control group (C57BL/6 mice), Disease
   model group (untreated NOD mice), Positive control group
   (Hydroxychloroquine Sulfate, 6.5 mg/kg/day), Baicalin treatment group
   (100 mg/kg/day), Quercetin treatment group (100 mg/kg/day), and Herb
   group (0.1625 g/mL herbal preparation, 16 mL/kg/day). The herbal
   medicine for oral gavage was prepared by Guang’anmen Hospital in
   accordance with standard clinical protocols. Animals had free access to
   feed and water. The room was maintained at a temperature of 23 ± 2°C
   and a relative humidity of approximately 45%, with a 12-hour light/dark
   cycle. A one-week acclimatization period preceded the experiment. Water
   intake was measured gravimetrically over 24h; resting saliva flow was
   measured: saliva pooled for 1 min was aspirated and volume recorded
   (mL/min). All therapeutic interventions were administered at equivalent
   volumes and concentrations through appropriate administration routes.
   The experimental protocol-maintained consistency in dosage formulation
   and administration methods across all treatment groups to ensure
   comparability. Baicalin and quercetin dosages were referenced from the
   literature and were identified by the additional WB experiment ([59]
   Supplementary Material 1 ) ([60]23, [61]24); the herbal dosage was
   derived from preliminary clinical data ([62]9).

2.4. Transmission electron microscopy observation

   Three submandibular gland specimens (1mm³) were collected and immersed
   immediately in 2.5% glutaraldehyde/PBS. They were then rinsed with PBS,
   post-fixed in osmium tetroxide, dehydrated through ethanol/acetone
   gradients, and infiltrated with epoxy resin overnight. After that, they
   were embedded, sectioned stained and finally examined using
   Transmission electron microscopy (TEM).

2.5. Elisa

   Tissue lysates or serum samples were analyzed for TNF-α, IFN-β, CXCL10,
   IL-1β, IL-16, IFN-α, ATP and NAD^+/NADH concentrations using a
   commercially available sandwich ELISA kit (RXW202412M, RUXIN BIOTECH,
   QUANZHOU, CHINA; RXW203063M, RUXIN BIOTECH, QUANZHOU, CHINA;
   RXW203064M, RUXIN BIOTECH, QUANZHOU, CHINA; RXW203106M, RUXIN BIOTECH,
   QUANZHOU, CHINA; RXW203124M, RUXIN BIOTECH, QUANZHOU, CHINA;
   RXW202930M, RUXIN BIOTECH, QUANZHOU, CHINA; INSSH-0089, INSELISA,
   HUANGSHI, CHINA; INSSH-0028, INSELISA, HUANGSHI, CHINA) according to
   the manufacturer’s instructions. Absorbance was measured at 450 nm
   using a microplate reader, and concentrations were interpolated from a
   standard curve.

2.6. Western blotting and RT-PCR

   The Western blotting & RT-PCR experimental procedure is consistent with
   our previous study ([63]25). In the present study, we used: sting:
   immniway T5488, 1:1000; cGAS: immniway YM8509, 1:5000; TBK1: immniway
   YM8322, 1:5000; ZBP1: immniway YN2476, 1:1000; BAX: Proteintch
   50599-2-IG, 1:50000; Caspase-3: CST 9662S, 1:1000; GAPDH: Servicebio
   GB11002, 1;1000; IRF3: immniway YM8227, 1:1000;p-IRF3: immniway YP0326,
   1:1000; p-STING: immniway YP1518, 1:1000; p-TBK1: immniway YM8674,
   1:2000.

   Tissues were collected by centrifugation at 600 × g for 5 min, washed
   with PBS, and repelleted by centrifugation (5 min, 600 × g). The pellet
   was resuspended in cytosol extraction buffer and incubated on ice for
   10 min. Cell disruption was performed using a Dounce tissue grinder,
   followed by centrifugation at 700 ×g for 10 min to pellet nuclei and
   intact cells. The supernatant was transferred and subjected to
   centrifugation at 10,000 g for 30 min. The resulting supernatant was
   discarded, and the pellet (enriched mitochondria) was resuspended in
   fresh cytosol extraction buffer, then re-centrifuged at 10,000 g for 30
   min to further purify mitochondrial fractions. The final mitochondrial
   pellet was resuspended in mitochondrial lysis buffer and incubated for
   10 min. An enzyme mix was added to the lysate and incubated for 60 min,
   followed by ethanol addition and a 10-min incubation. Mitochondrial DNA
   was recovered by centrifugation (5 min), with the resultant pellet
   retained as purified mtDNA. Primer sequence: CCCCATATTAAACCCGAATGATA;
   TAGGCTTCGTTGCTTTGAGGTA.

2.7. Immunofluorescence

   Tissue sections were subjected to immunofluorescence staining using
   primary antibodies against BAX, cGAS and STING, followed by appropriate
   Alexa Fluor-conjugated secondary antibodies and DAPI counterstain.
   Slides were mounted and imaged at 200x magnification using a
   SlideViewer2.5 scanner, ensuring that tissue filled the field of view
   and background illumination was consistent across all samples. For each
   section, three representative fields were quantitatively analyzed.
   Using image analysis software, the integrated density (IntDen) and
   positive pixel area (Area) for green, red, pink, and co-localization
   signals were measured in pixels. The mean fluorescence intensity (MFI)
   for each channel was calculated as MFI = IntDen/Area.

3. Results

3.1. Network pharmacology and analysis

   We conducted a comprehensive retrieval of pSS (primary Sjögren’s
   syndrome) -related targets, yielding a total of 564 targets from the
   GeneCards database and the OMIM database, with a relevance score
   greater than 3.41. Following database calibration, the pSS targets
   identified in the PEs target libraries were imported into the Venny 2.1
   tool to generate a Venn diagram, which resulted in an identification of
   106 common targets associated with pSS, visually represented in the
   Venn diagram ([64] Figure 1a ). The active ingredients from PEs and
   their corresponding targets were imported into Cytoscape 3.9.1 to
   construct a comprehensive drug-active ingredient-target network. In
   this network, nodes represent targets, with node size proportional to
   the degree of connectivity: larger nodes indicate targets with more
   biological functions and enhanced significance([65] Figure 1b ).
   Removing irrelevant nodes, the network diagram was generated. Based on
   LC-MS/MS identification ([66] Figure 2 , [67]Supplementary Material 2
   ), baicalin and quercetin were selected as the active compounds.
   Through network pharmacology analysis, several key targets were
   identified (degree>6), including ESR1, PPARG, CASP3, CA2, and BAX.
   Additionally, the analysis highlighted critical active ingredients
   (degree>18), such as quercetin, baicalin, beta-sitosterol, and wogonin,
   which play significant roles in the therapeutic effects of PEs. To
   elucidate the interactions between PEs and pSS targets, the
   intersection targets were imported into the STRING database to
   construct a PPI network, thus highlighting the most critical genes
   within the PPI network that may significantly influence pSS. The
   results revealed a network comprising 83 nodes and 196 edges among the
   106 overlapping proteins, with an average node degree of 4.77. This
   indicated a strong correlation among the targets, with a confidence
   score of ≥ 0.9, suggesting robust interactions within the network ([68]
   Figure 1c ). The 106 intersection targets were then submitted to the
   DAVID database for GO and KEGG enrichment analysis, with the top 10 GO
   entries across the three categories—BP, MF, and CC— identified and
   illustrated in [69]Figure 1d , and the top 10 KEGG pathways associated
   with these targets screened and presented in [70]Figure 1e . These
   analyses provided insights into the biological roles and pathways that
   the common targets may influence in the context of SS.

Figure 1.

   [71]a) Venn diagram showing overlap between disease (554) and target
   (1911) groups; 106 shared. Bar chart displays list sizes. b) Network
   graph illustrating protein interactions. c) Network of interactions
   between compounds and genes, with compounds in blue and genes in
   orange. d) Pathway analysis scatter plot with pathways on the y-axis
   and enrichment scores on the x-axis, showing significant pathways. e)
   Bar graph with GO results in three ontologies (BP, CC, MF) showing
   enrichment scores for cellular processes.
   [72]Open in a new tab

   Network pharmacology predicts potential mechanisms of AG&A. (a) Disease
   Target Venn Diagram. Illustrates the overlap between disease-associated
   targets(bule, contain all databases) and potential therapeutic targets
   (green, contain all databases) of AG&A. The intersection highlights
   common targets critical for AG&A’s therapeutic effects. (b)
   Drug-Molecular-Target Network Diagram A bipartite network showing
   interactions between bioactive compounds and their targets. Orange
   nodes represent potential therapeutic targets; green nodes denote key
   compounds from Achyranthes bidentata; blue nodes indicate key compounds
   from American Ginseng. Node size may reflect compound/target
   significance. (c) Protein-Protein Interaction (PPI) Network Visualizes
   functional associations between AG&A target proteins. Nodes represent
   proteins, with edges indicating interactions. Core targets were
   identified via topology analysis. (d) KEGG Pathway Enrichment Analysis
   Bubble plot displaying enriched GO terms (Biological Process, Molecular
   Function, Cellular Component) in AG&A targets. X-axis: Statistical
   significance (-log[10](p-value); Y-axis: Term name; Bubble size: Gene
   count; Color:p-value. (e) Gene Ontology (GO) Enrichment Analysis Bubble
   plot showing AG&A target-enriched KEGG pathways. X-axis: Rich Factor;
   Y-axis: Pathway name. Top pathways are highlighted. (e). KEGG Pathway
   Enrichment Analysis Bubble plot showing AG&A target-enriched KEGG
   pathways. X-axis: Rich Factor; Y-axis: Pathway name; Bubble size: Gene
   count; Color: Statistical significance [-log[10](p-value)]. Top
   pathways are highlighted.

Figure 2.

   [73]Four panels display analytical data. Panel a shows a BPC graph with
   red lines representing TIC data in negative mode. Panel b presents a
   similar TIC graph. Panel c features two graphs illustrating Baicalin
   and Quercetin identification, with absorbance and relative intensity
   data. Panel d contains a pie chart categorizing compounds, highlighting
   proportions like terpenes at 23.46% and carbohydrates at 12.15%.
   [74]Open in a new tab

   Analysis of baicalin and quercetin by LC-MS/MS. (a) Representative base
   peak chromatogram (BPC) of the AG&A analyzed. The BPC shows the most
   intense ion current at any retention time during the analysis,
   providing a chromatographic profile of the most abundant compounds. (b)
   Representative total ion chromatogram (TIC) of the AG&A analyzed. The
   TIC represents the summed ion current over the entire mass range
   detected at each retention time, providing an overall chromatographic
   signature of all detected compounds. (c) Mass spectrum of baicalin and
   quercetin. This spectrum displays the relative abundance of ions at
   different mass-to-charge ratios (m/z) for the identified compounds,
   enabling structural identification and confirmation. (d) Component
   Classification Quantity Distribution Diagram. This chart illustrates
   the relative proportions of different compounds. Each sector of the pie
   represents a different compound or compound category, with the size of
   the sector corresponding to the relative quantity of that compound in
   the sample.

3.2. Molecular docking and molecular dynamics simulations

   Molecular docking was conducted using the key genes and PEs identified
   in section 3.1 as active ingredients and targets. The docking &
   molecular dynamics simulations results ([75] Figure 3 ) demonstrated
   that the selected targets exhibited strong binding affinities with the
   core active ingredients. This finding further validated the reliability
   of the network pharmacology analysis, confirming the potential
   interactions between the active components and their respective targets
   in the therapeutic context of SS.

Figure 3.

   [76]A grouped image shows two panels. Panel i presents a heatmap of
   docking scores for baicalin and quercetin against five proteins: ESR1,
   PPARG, CASP3, CA2, and BAX, ranging from -8.97 to -6.35. Panel ii
   consists of subfigures a to d, each containing analyses of
   root-mean-square deviation (RMSD), root-mean-square fluctuation (RMSF),
   hydrogen bonds number, radius of gyration, and solvent-accessible
   surface area (SASA) for different molecular dynamics simulations.
   [77]Open in a new tab

   Computational prediction of protein-ligand interactions and molecular
   dynamics simulations. (i) Heatmap depicting predicted binding
   affinities (kcal/mol) between the two active ingredients (Quercetin,
   Baicalin) and the five hub targets. Darker green indicates stronger
   binding (more negative values). (ii) Representative snapshots from
   molecular dynamics simulations of protein-ligand complexes, from top to
   bottom: RMSD, RMSF, Hbonds Number, Gyrate, SASA: (A) BAX in complex
   with Quercetin (B) BAX in complex with Baicalin (C) CASP3 in complex
   with Quercetin (D) CASP3 in complex with Baicalin.

   While BAX and Caspase-3 (CASP3) were identified as critical targets for
   AG&A in pSS treatment (binding energy ≤ -8.5 kcal/mol) through network
   pharmacology and molecular docking, their biological roles extend
   beyond apoptosis regulation. As a key mediator of mitochondrial outer
   membrane permeabilization (MOMP), BAX facilitates mitochondrial DNA
   (mtDNA) release into the cytosol during cellular stress. Concurrently,
   CASP3—traditionally associated with apoptosis execution—has recently
   been implicated in modulating innate immune responses via cleavage of
   cGAS to suppress interferon production ([78]26). Crucially, cytosolic
   mtDNA acts as a potent ligand for cGAS, initiating the cGAS-STING
   signaling cascade that drives type I interferon (IFN-I) inflammation—a
   hallmark of pSS pathogenesis. Given that IFN-I signature correlates
   with pSS disease severity and glandular dysfunction, we hypothesized
   that AG&A’s modulation of BAX/CASP3 may functionally intersect with the
   mtDNA-cGAS-STING axis to resolve immunopathology. Therefore, to
   elucidate the holistic mechanism underlying AG&A’s efficacy, we
   experimentally validated its impact on mtDNA leakage, cGAS-STING
   activation, and downstream inflammatory cascades in submandibular
   glands.

3.3. AG&A ameliorates submandibular gland dysfunction in Sjögren’s syndrome
models

   AG&A and its key components can significantly improve submandibular
   gland function in NOD model mice, manifested through enhanced salivary
   flow rate, reduced water consumption, and histopathological
   restoration. Specifically, NOD mice exhibited reduction in salivary
   flow rate compared to normal controls (p<0.001), while pharmacological
   intervention with AG&A and its active compounds restored salivary
   secretion to 70-80% of physiological levels ([79] Figure 4a ).
   Concomitant with xerostomia, model mice demonstrated increase in daily
   water intake (p<0.01) ([80] Figure 4b ). Histopathological examination
   revealed [81]Figure 4c : The normal control group exhibited intact
   salivary gland architecture with no pathological changes such as acinar
   degeneration, necrosis, or inflammatory cell infiltration in the
   stroma. In contrast, the model group showed multifocal, dense
   inflammatory cell infiltration within the submandibular glands, forming
   lymphoid follicle-like structures, with mild acinar atrophy observed in
   surrounding areas. Relative to the model group, all treatment groups
   demonstrated varying degrees of pathological improvement: The herb
   group exhibited the most significant amelioration, followed by the
   baicalin group which presented only minor vacuolar degeneration in
   acini, and the quercetin group and positive control group showed the
   mildest changes, with scant inflammatory cell infiltration between
   acini and slight vacuolar degeneration.

Figure 4.

   [82]Graphical data and microscopic images are displayed. (a) Bar graph
   showing salivary flow rates for various groups over 8 and 16 weeks,
   with significance levels indicated. (b) Line graph plotting water
   intake over 8, 10, 12, 14, and 16 weeks. Column graph showing water
   intake at 8 and 16 weeks with significance levels. Below are
   histological images at 50X and 200X magnification for six groups:
   Normal, Model, Positive, Baicalin, Quercetin, and Herb, with areas
   highlighted for comparison.
   [83]Open in a new tab

   Effects of AG&A and its active compounds ameliorates submandibular
   gland dysfunction. (a) Salivary flow rate. (n=3 per group, Two-way
   anova was used, Data are presented as mean ± SEM, ns p>0.05, *p < 0.05,
   **p < 0.01, ***p < 0.001, ****p < 0.0001). (b) Daily water intake. (n=3
   per group, Two-way anova was used, Data are presented as mean ± SEM, ns
   p>0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). (c)
   Representative HE stained sections of submandibular. 50X:Scale bar =
   200 µm;200X:Scale bar = 50 µm.

3.4. AG&A ameliorates mitochondrial function in submandibular glands of
Sjögren’s syndrome model mice

   Experimental results demonstrated that compared to the normal control
   group, NOD model mice exhibited significant mitochondrial dysfunction
   in submandibular glands, characterized by a reduction in ATP synthesis
   (p<0.01), a 1.8-fold increase in cytoplasmic mtDNA copies, and an
   abnormal elevation in NAD+/NADH ratio. Trough pharmacological
   interventions, these abnormalities were effectively reversed:
   cytoplasmic mtDNA overamplification was suppressed, ATP levels
   recovered to normal values, and the NAD+/NADH ratio normalized to
   physiological ranges ([84] Figure 5A ), indicating substantial
   restoration of mitochondrial energy metabolism.

Figure 5.

   [85](a) Bar charts comparing mtDNA, NAD+/NADH, and ATP levels across
   six groups: Normal, Model, Positive, Baicalin, Quercetin, and Herb,
   with statistical significance indicated. (b) Electron microscope images
   of samples from the same groups showing cellular structures at 15,000x
   and 30,000x magnifications, highlighting differences in morphology.
   [86]Open in a new tab

   Analysis of AG&A ameliorates mitochondrial function in submandibular
   gland tissue. (A) Concentrations of ATP and NAD^+/NADH ratio measured
   by ELISA, and relative mitochondrial DNA (mtDNA) copy number determined
   by RT-PCR. (ATP and NAD^+/NADH: n=5 per group, mtDNA:: n=10 per group,
   one-way anova was used, Data are presented as mean ± SEM. ns p>0.05, *p
   < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001). (B) Representative
   TEM images depicting mitochondrial ultrastructure in submandibular.
   Scale bars = 500 nm.

   These findings were then corroborated through TEM ultrastructural
   analysis, demonstrated by classical mitochondrial pathologies in acinar
   cells of SS model mice. Specifically, mitochondria exhibited swelling,
   and also showed cristae fragmentation or complete loss of mitochondrial
   structure, as well as vacuolar degeneration. Post treatment, all groups
   except the quercetin group demonstrated marked morphological recovery,
   with the most pronounced restoration observed in the herb group ([87]
   Figure 5B ).

3.5. AG&A corrects mitochondrial dysfunction via cGAS/STING signaling to
ameliorate immune microenvironment in submandibular glands

   Mechanistically, mitochondrial dysfunction in submandibular glands of
   NOD model mice triggers pathological leakage of mtDNA into the cytosol
   through mitochondrial permeability transition pores, thereby activating
   the cytosolic DNA-sensing cGAS-STING signaling axis. This activation
   initiates a proinflammatory cascade characterized by upregulated type I
   interferons (IFN-α/β) and inflammasome-related mediators.

   Western blot analysis confirmed significant dysregulation of key
   pathway components: Model mice exhibited marked upregulation of BAX,
   cGAS, STING, ZBP1, TBK1, p-STING, IRF3, p-IRF3, p-TBK1 and CASP3
   compared to normal controls (P<0.05 for all). This hyperactivation was
   differentially suppressed by pharmacological interventions: Baicalin &
   herb Group showed most pronounced reductions (P<0.05) ([88] Figures 6 ,
   [89]7 ) across all targets, and Quercetin Group demonstrated marginal
   efficacy (P>0.05). Notably, the therapeutic hierarchy (AG&A/baicalin >
   positive control > quercetin) aligned precisely with functional
   recovery patterns observed in salivary flow restoration &
   histopathological improvements (Section 3.3) and mitochondrial
   protection (Section 3.4). These biochemical alterations substantiate
   AG&A’s dual mechanism: protecting mitochondria to prevent mtDNA leakage
   and modulating cGAS/STING axis to resolve interferon-driven
   inflammation, thus synergistically restoring glandular immune
   homeostasis.

Figure 6.

   [90]Western blot and bar graph data illustrate protein and cytokine
   levels across six conditions: Normal, Model, Positive, Baicalin,
   Quercetin, and Herb. Panel (a) shows western blot bands and
   corresponding bar graphs for proteins including BAX, cGAS, ZBP1, and
   others, indicating variability in expression levels. Panel (b) features
   bar graphs for cytokines like IFN-α, IFN-β, IL-1β, IL-18, CXCL10, and
   TNF-α, with significant statistical comparisons highlighted.
   Statistical significance is marked with asterisks, indicating varying
   levels of significance.
   [91]Open in a new tab

   AG&A corrects mitochondrial dysfunction via cGAS-STING signaling to
   ameliorate immune microenvironment. (A) Western blot analysis of BAX,
   cGAS, STING, p-STING, ZBP1, TBK1, p-TBK1, IRF3, p-IRF3 and CASP3
   protein levels. Densitometric quantification of protein bands
   normalized to loading control is presented as mean ± SEM. (n=3 per
   group, one-way anova was used, ns p>0.05, *p < 0.05, **p < 0.01, ***p <
   0.001, ****p < 0.0001). (B) Concentrations of IFN-α, IFN-β, CXCL10,
   IL-1β, IL-16, and TNF-α in serum measured by ELISA. Data are presented
   as mean ± SEM (n = 5 per group, one-way anova was used,ns p>0.05, *p <
   0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Figure 7.

   [92]Fluorescence microscopy images of five groups: Normal, Model,
   Positive, Baicalin, Quercetin, and Herb. Rows show different stains:
   DAPI (blue), cGAS (green), STING (pink), BAX (red), and merged images.
   Charts on the right compare expression levels of BAX (red), cGAS
   (green), and STING (pink) with statistical significance markers.
   [93]Open in a new tab

   Immunofluorescence analysis of BAX, cGAS, and STING expression in
   submandibular gland tissue. (A) Representative immunofluorescence
   images showing localization of (red) BAX, (green) cGAS, and (pink)
   STING. Nuclei were counterstained with DAPI (blue). Scale bar = 50 µm.
   (B) Quantitative analysis of mean fluorescence intensity (MFI) for BAX,
   cGAS, and STING. MFI was calculated as Integrated Density (IntDen)/Area
   within regions of interest. Data are presented as mean ± SEM (n = 3 per
   group; one-way anova was used, Scale bars = 50 µm; ns p>0.05, *p <
   0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

4. Discussion

   It is recognized that the treatment with traditional Chinese medicine
   (TCM) is effective for SS. Clinical treatment has shown that the
   application of TCM in SS management can significantly alleviate
   symptoms, slow disease progression, and maintain safety. Therefore, a
   deeper exploration of its therapeutic mechanisms was conducted to.
   provide a new foundation for the further clinical application of TCM in
   SS treatment.

   Previous studies have indicated that American ginseng can counteract
   mitochondrial dysfunction (MDF) and apoptosis induced by oxidative and
   inflammatory damage through inhibiting calcium (Ca²^+) influx, thereby
   maintaining cellular homeostasis ([94]27). Additionally, it has been
   proved that achyranthes bidentata possesses antioxidant properties,
   thus being helpful to protect mitochondrial ultrastructure ([95]28).
   Recent research shows that the cGAS-STING pathway likely contributes to
   the pathogenesis of autoimmune diseases, including desiccation
   syndrome. Furthermore, the abnormal leakage of mtDNA due to
   mitochondrial damage has been recognized as a critical mechanism for
   activating the cGAS-STING pathway. Thus, the present study explored how
   AG&A regulates the mtDNA-cGAS-STING signaling axis, aiming to correct
   MDF and regulate the metabolic reprogramming of immune cells to prevent
   immunoinflammatory alterations. This regulation may further impact the
   immune microenvironment of the submandibular gland, which supports the
   notion that Bidentate Achyranthes and American Ginseng represent an
   effective therapeutic intervention for SS, highlighting its significant
   research value.

   The superior therapeutic outcomes of the whole herb AG&A formulation
   compared to high-dose baicalin or quercetin monotherapy—particularly in
   mitochondrial ultrastructural restoration and cGAS-STING pathway
   suppression—indicate multifaceted synergism. This likely arises from
   pharmacokinetic enhancement, target diversification, and pathway
   crosstalk amplification. In contrast, quercetin’s marginal efficacy
   stems from extensive first-pass metabolism, and context-dependent
   activity, and its weak CASP3 binding further limits apoptosis
   regulation and mitochondrial rescue.

   As baicalin and quercetin are flavonoids, we included virtual screening
   data for other AG&A flavonoids ([96] Supplementary Material 3 ).
   Notably, our results revealed two constituents within the mixture,
   Trifolirhizin and Rottlerin, which demonstrated superior binding
   affinity compared to Baicalin (previously identified by Network
   Pharmacology). Besides, Wang, D ([97]29). proved that Rottlerin can
   inhibit cGAS-STING signaling pathway in mice. Furthermore, both
   Trifolirhizin and Rottlerin have been independently documented in other
   studies as potent inhibitors of mitophagy ([98]30). This strongly
   suggests that these two compounds may be the key synergistic substances
   responsible for the enhanced efficacy observed in the herb, compared
   with baicalin alone.

   The submandibular gland is the organ most commonly affected in SS, with
   an abnormal immune microenvironment in the gland being the primary
   cause of dry mouth due to decreased secretory function. In this altered
   immune microenvironment, the epithelial cells of the submandibular
   gland can directly activate CD4+ T cells by expressing major
   histocompatibility complex class II (MHC-II), co-stimulatory molecules
   (CD80 and CD86), and intercellular adhesion molecule-1 (ICAM-1) under
   inflammatory stimuli. Furthermore, these activated epithelial cells
   enhance the immune response through elevated expression of chemokines
   such as CXCL12 and CXCL13, as well as pro-inflammatory cytokines
   including IL-1, IL-6, and BAF, thereby driving the inflammatory
   response. In autoimmune diseases, an uncontrolled inflammatory response
   often prompt the immune system to attack and destroy healthy cells and
   tissues, a cascade commonly associated with MDF ([99]31). Research
   indicates that the pathogenesis of SS is linked to changes in the
   immune microenvironment of the submandibular gland, which closely
   correlates with MDF and immune inflammation ([100]32). MDF will result
   in the release of damage-associated molecular patterns (DAMPs),
   including mitochondrial DNA (mtDNA), into the cytoplasm. This release
   then further activates pattern-recognition receptors (PRRs), thus
   stimulating signaling pathways, including the cGAS/STING pathway. DAMPs
   can activate PRRs and initiate signaling cascades such as the NLRP3
   inflammasome, TLR9, and ZBP1, ultimately triggering an inflammatory
   immune response.

   Recent studies demonstrate that abnormal leakage of mtDNA can activate
   several key innate immune response pathways, particularly the
   cGAS-STING signaling pathway, whose activators play crucial roles in
   cellular immunity. Prolonged immune dysregulation can lead to various
   pathological immune responses, disrupting the immune microenvironment
   and the body’s homeostasis under mitochondrial stress conditions.
   Specifically, when mitochondria are stressed, mtDNA may be released
   into the cytoplasm through mechanisms such as BAX/BAK-dependent
   mitochondrial outer membrane permeabilization (MOMP) or the
   mitochondrial permeability transition pore (mPTP). Once in the
   cytoplasm, mtDNA readily binds to cGAS, activating the downstream STING
   signaling pathway, which generates type I interferon (IFN) and triggers
   inflammatory responses. Under normal physiological conditions, caspases
   effectively inhibit mtDNA-induced cGAS-STING activation by cleaving
   cGAS and IRF3, thereby preventing the production of IFN and subsequent
   inflammatory responses. However, during immune inflammation,
   apoptosis-mediated caspases may lose their enzymatic activity, leading
   to an intensified IFN response. This abnormal activation of the
   mtDNA-cGAS-STING signaling axis results in the overexpression of IFN
   and various inflammation-related genes ([101]33). Moreover, the
   activated ZBP1 interacts with TBK1, thus promoting the phosphorylation
   of IRF3 and further enhancing the production of type I IFN. Also, ZBP1
   can activate the NF-kB signaling pathway through TBK1. This intricate
   interplay underscores the significance of the mtDNA-cGAS-STING
   signaling axis in mediating immune responses and highlights potential
   therapeutic targets for managing autoimmune conditions like SS.

5. Conclusion

   Novel technologies and methods have brought unprecedented opportunities
   for the research of traditional medicine ([102]34–[103]36). In this
   study, we identified AG&A as a potential effective herbal combination
   based on the preliminary clinical data. Through LC-MS/MS and network
   pharmacology analysis, we determined the potential active components
   and targets, and thus identified baicalin and quercetin as potential
   active substances, and BAX and CASP3 as potential effective targets.
   The promising binding performance of these small molecules with the
   potential targets was confirmed by molecular docking and molecular
   dynamics simulations. Additionally, we selected the most relevant
   cGAS-STING pathway for in vivo animal experiments, providing
   multi-level evidence of baicalin’s therapeutic efficacy, including
   histopathological examination, mitochondrial function assessment,
   pathway activation, and local immune microenvironment analysis.
   Notably, this study also found that the efficacy of the herbal mixture
   was comparable to high dose baicalin, and even better in certain
   aspects, such as mitochondrial function protection. Although further
   exploration was limited by space constraints, this suggests that other
   substances in the AG&A combination may act as potential mitochondrial
   protectors or exhibit broader drug effects. Besides, this study focuses
   only on small clinical doses, necessitating further research on
   different dosages.

Acknowledgments