1. Introduction
Rheumatoid arthritis (RA) is a chronic systemic autoimmune disease affecting approximately
of the population [
1,
2]. Despite recent medical advancements in RA therapy, the disability rate remains alarmingly high at
for patients with RA duration
years, and successful treatment outcomes remain below
[
3]. The roots of the plant Tripterygium wilfordii Hook.
f. (TwHF) are known for its anti-inflammatory and immune-modulatory properties and are used in traditional Chinese medicine [
4]. Among the various preparations of TwHF, Tripterygium glycosides tablets (TGTs) are widely used in clinical practice and recognized by the World Health Organization as a "China-innovated plant-derived drug" for treating RA [
5]. Numerous clinical randomized controlled trials have demonstrated that TGTs are safer and more effective than the RA anchor drug methotrexate and other disease-modifying antirheumatic drugs [
6⇓-
8]. Evidence has shown that TGTs can ameliorate bone destruction and synovial hyperplasia in RA through various pathways and mechanisms, such as modulation of immunity and metabolism, alleviation of local inflammation, and inhibition of angiogenesis. These diverse effects contribute to the efficacy of TGTs in managing RA and preventing joint deterioration [
9⇓⇓-
12].
Nevertheless, TGTs face several challenges in terms of efficacy and safety. The narrow effective dose range of TGTs may make it difficult to accurately determine the appropriate indications for clinical medication, leading to differential responses among patients [
13,
14]. A growing number of clinical studies have shown a relatively high risk of drug-induced liver injury (DILI) caused by improper use of TGTs in clinical practice [
15⇓-
17]. Of the 472 reported adverse reactions associated with TGTs, 54 (11.44%) were associated with abnormal liver function. Among the 45 patients with severe adverse reactions, 11 (24.44%) experienced liver injury [
18]. In a previous study, we integrated chemical and target profiling of TGTs, systematic network analysis, and in vitro and in vivo experimental validations to elucidate the molecular mechanisms underlying TGT-induced DILI [
19]. Moreover, our research group has disseminated guidance titled "Clinical practice guidelines for Tripterygium glycosides/Tripterygium wilfordii tablets in the treatment of rheumatoid arthritis” (No. T/CACM 1337-2020) to provide comprehensive recommendations for the rational use of TGTs in clinical settings [
20,
21]. However, the relationship between the efficacy and toxicity of TGTs remains unclear, posing challenges in designing and optimizing therapeutic strategies to improve the clinical outcomes of patients with RA with minimal adverse effects.
In the current study, we integrated multi-omics profiling data associated with the clinical efficacy and DILI of TGTs based on clinical samples with the chemical and target profiling of this drug to perform a systematic network analysis. Our findings were substantiated using a collagen-induced arthritis (CIA) mouse model (in vivo) and further validated using MH7A and AML12 cell lines (in vitro), as depicted in
Fig. 1. This integrative approach enhances our understanding of the dual effects of TGTs, combining clinical insights and molecular mechanisms.
2. Materials and methods
2.1. Clinical sample collection
The clinical sample collection protocol followed was as per the guidelines outlined in the Declaration of Helsinki. This study was approved by the Research Ethics Committee of Guang’anmen Hospital (No. BZYSY-2021KYKT8L-02) and the First Affiliated Hospital of Anhui University of Chinese Medicine (No. 2019AH- 12). Prior to sample collection, all the patients provided written informed consent.
The serum samples were obtained from a group of RA patients consisting of 34 individuals. Among these patients, 17 had DILI and 17 did not. The samples were collected between February 2018 and May 2019 at the Department of Rheumatology, the First Affiliated Hospital of Anhui University of Chinese Medicine. Additional details are provided in Tables S1 and S2 in Appendix A. The inclusion criteria, patient treatment, and definition of DILI to targeted therapies (TGTs) were based on previous studies conducted by our team [
22]. Peripheral blood mononuclear cells (PBMCs) were collected from 12 patients with RA. The patients were divided into two subgroups: six well responders and six poor responders. The samples were collected between January 2015 and October 2019 at the Department of Rheumatology, Guang’anmen Hospital. Similar to those in the previous group, the inclusion criteria, patient treatment, and definitions of well and poor responders to TGTs were based on previous studies by our team [
23]. Venous blood samples were collected from all participants in the morning, before breakfast. Subsequently, the sera were separated and stored at
until use.
2.2. Transcriptomics detection and data analysis
The expression of the messenger RNAs (mRNAs) in PBMCs obtained from TGTs well and poor responders was analyzed using Affymetrix miRNA 4.0 and EG1.0 arrays. The detection of microar-ray signals was conducted by Shanghai GMINIX Biotech Limited Company (China). The obtained microarray data were made publicly available on the Gene Expression Omnibus, National Center for Biotechnology Information under accession numbers GSE106893 and GSE106894.
To identify the genes that showed differential expression between the TGTs well and poor responders, a screening process was performed using specific criteria. These criteria included a fold change (FC) or , and false discovery rate (FDR) . Screening was performed using the random variance model -test and FDR analysis. Hierarchical clustering analysis was conducted using the R package (version 4.2.2; R Core Team, Austria).
2.3. Metabonomics detection and data analysis
The thawed serum sample was deproteinized by adding a
aliquot to a mixture of
and acetonitrile (precooled to
) in a 1:1 ratio. Samples were vortexed for
and sonicated for
in an ice bath. To further enhance protein precipitation, the samples were stored overnight at
. The following day, the samples were centrifuged at
for
at
, and only
of the resulting supernatant was used for metabonomics analysis. For the untargeted metabolomics profiling, an ultra-high performance liquid chromatography (UPLC)-quadrupole time-of-flight mass spectrometry (Q-TOF/MS) system (AB SCIEX, USA) was employed. A Waters ACQUITY UPLC BEH C18 column (1.8 μm,
; USA) was used for chromatographic separation. The volume of injection for the samples was
, and there was no streaming. Helium (99.9996%) was used as the carrier gas and the front inlet purge flow rate was set to
. Additionally, the gas flow rate through the column was set to
. The UPLC and mass spectrometer conditions were set according to previous reports [
24].
The raw data of the metabonomics analysis were imported into the Progenesis QI software (Nonlinear Dynamics, UK). This software was used for peak alignment, which involved matching the peaks in the data to obtain a list of peak areas and identified metabolites. To match the molecular mass data with the metabolites, several online databases were used, including Human Meta-bolome Database
†(†
https://hmdb.ca/.), LIPIDMAPS
† (†
https://www.lipidmaps.org/.), Kyoto Encyclopedia of Genes and Genomes (KEGG)
† (†
https://www.kegg.jp/.), and METLIN
## (##
https://metlin.scripps.edu.). These databases provide information on the molecular masses of various metabolites, allowing their identification and alignment with data obtained from metabolomics analysis.
2.4. Tandem mass tag (TMT) quantitative proteomic analysis
TMT quantitative proteomic analysis was performed by Beijing Zhimeiyinuo Biotechnology Co., Ltd. (China). Protein concentrations were determined using the Bradford method [
25]. Total protein was extracted from the PBMCs of patients with and without DILI. PBMC lysates were collected and quantified using a Bicin-choninic Acid Protein Assay Kit (Beyotime Biotechnology, China), following the manufacturer’s instructions. The digested samples were then desalted using a XBridge Peptide BEH 5 µm C18 column (Waters) and subsequently vacuum dried. For TMT labeling, the peptides were reconstituted in
tetrabutylammonium bromide (TMT 10plex
kit; Thermo Fisher Scientific, USA) and processed according to the manufacturer’s protocol. For sample preparation, tryptic peptides were fractionated using high
reverse-phase high-performance liquid chromatography (HPLC) with a C18 column (Art. No. 186003581; Waters), and a gradient of
acetonitrile at
9.0. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) was performed as previously described [
26].
2.5. Establishment of CIA mouse model and drug treatment
All the animals were maintained in a room with a constant temperature of and with a light and dark cycle. The mice had free access to water and food. This study was approved by the Research Ethics Committee of the Institute of Chinese Materia Medica, China Academy of Chinese Medical Sciences (ethics approval numbers: 2022B069 and 2022B070). All protocols involving animal experimentation strictly adhered to the guidelines and regulations governing the ethical treatment and management of animals.
A total of 30 male DBA-1 mice, aged 7-9 weeks with an average weight of
, were procured from Beijing Vital River Laboratory Animal Technology Co., Ltd. (SCXK 2021-0011; China). All 30 mice were divided into five groups, and the treatments were administered orally for five and seven weeks, starting from the day after the second immunization. The groups are as follows: control group (Con,
), group subjected to CIA for five weeks (CIA
), group subjected to CIA for seven weeks (CIA 7
,
), group treated with TGTs for five weeks (TGT
), and group treated with TGTs for seven weeks (TGT
). The CIA mouse model was established as described in our previous studies [
27]. Briefly, mice were subcutaneously immunized in the tail with
of type II collagen (CII; Cat No. #20022; Chondrex, USA), emulsified in complete Freund’s adjuvant (CFA; Cat No. #7001; Chondrex) and boosted with the same emulsion after three weeks using the same route. Mice in the control group received subcutaneous injections of saline solution at the same dosage. The dosage of TGTs in the treatment group was
, which is equivalent to twice the clinical daily dosage given to patients with RA and has shown significant therapeutic efficacy in our previous studies [
19,
28].
2.6. Cell culture and treatment
MH7A cells, derived from the human RA synovium (RIKEN, Japan), were cultured in Roswell Park Memorial Institute (RPMI)- 1640 medium supplemented with fetal bovine serum at in a atmosphere. AML12 cells, representing normal mouse liver cells, were obtained from Guangzhou Hui Yuan Pharmaceutical Technology Co., Ltd. (China) and maintained in Dul-becco’s modified Eagle medium (DMEM)-F12 supplemented with fetal bovine serum, insulin-transferrin-selenium, and 40 dexamethasone at in .
For experimental treatments, MH7A cells underwent a induction with tumor necrosis factor , followed by treatments with Stattic (a signal transducer and activator of transcription 3 (STAT3) phosphorylation inhibitor; Cat No.: HY-13818; MCE, USA), colivelin TFA (a STAT3 phosphorylation activator; Cat No.: HY-P1061A; MCE), and TGTs for . AML12 cells were stimulated with TGTs, and subsequently treated with varying concentrations of Stattic and colivelin TFA for . Cell supernatants and precipitates were collected for molecular biology assays to investigate the key proteins and genes in both cell lines.
2.7. Evaluation of relevant indicators for arthritis and liver injury
The severity of arthritis was evaluated based on the arthritis incidence, arthritis score, and hind paw thickness, as described in our previous studies [
29]. The severity of liver injury was evaluated by measuring the hepatobrain index and serum alanine transaminase (ALT) and aspartate transaminase (AST) levels. Hepatic ALT and AST levels were determined using the respective kits (C009- 2-1 and C010-2-1; Nanjing Jiancheng Bioengineering Institute, China). Furthermore, we assessed the severity of lipid metabolic disorders in liver tissues by measuring total cholesterol (TC), triglyceride (TG), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C) levels. Hepatic TC, TG, LDL-C, and HDL-C levels were quantified using the corresponding kits (BC1985, BC0625, BC5335, and BC5325; Beijing Solarbio Science & Technology Co., Ltd., China). To measure the iron content in both the inflamed joint and liver tissues, we used the respective kits (BC4355 and BC5415; Beijing Solarbio Science & Technology Co., Ltd.). Additionally, the expression levels of malondialdehyde (MDA, ml094962), 4-hydroxynonenal (4-HNE, ml023341), hep-cidin (HAMP, ml025039), lipocalin 2 (LCN2, ml022573), and lysophosphatidylcholine acyltransferase 3 (LPCAT3, ml023384) were estimated using enzyme-linked immunosorbent assay (ELISA) kits (Shanghai Enzyme-linked Biotechnology Co., Ltd., China) following the manufacturer’s instructions. The absorbance was measured using a Multiskan
GO microplate spectrophotometer (Thermo Fisher Scientific).
2.8. Histopathological observations and scoring
Histopathological alterations in the joint and liver tissues of CIA mice were examined using histopathological analysis, focusing on the articular cartilage, synovium, and DILI. To ensure objectivity, the tissue sections were stained with hematoxylin and eosin (H&E) staining. Subsequently, two anatomical pathologists blinded to the experimental groups scored the stained sections according to the established protocol outlined in our previous study [
19]. Furthermore, using the protocol outlined in our earlier studies [
30], Masson staining was used to assess histopathological variations in knee tissues.
2.9. Immunohistochemical (IHC) staining
IHC staining was conducted to investigate the expression levels and subcellular localizations of transferrin receptor protein 1 (TFR1) and acyl-CoA synthetase long-chain family member 4 (ACSL4) proteins in the inflamed joint tissues and liver tissue of mice from different groups. The antibodies used for IHC analysis included TFR1 (ET1702-06, 1:100 dilution) and ACSL4 (ET7111- 43, 1:100 dilution). The staining protocol used is as described previously [
31]. Immunoreactivity was quantified using the ImageJ software (Bethesda, USA), which involves the examination of three distinct representative areas within the field.
2.10. Transmission electron microscopy (TEM)
Changes in mitochondria and lipid droplets in the inflamed liver tissues were observed using TEM according to the protocol described in our previous studies [
32,
33].
2.11. Quantitative polymerase chain reaction (qPCR) analysis
qPCR analysis was conducted in triplicates to assess the expression levels of diverse efficiency- and toxicity-related genes, including HAMP, LPCAT, and tartrate-resistant acid phosphatase type 5 (ACP5) genes, in both inflamed joint and liver tissue samples from the groups. The experimental procedure was performed as described previously [
34].
2.12. Western blot analysis
To assess the regulatory effects of TGTs on STAT3 and phospho-signal transducer and activator of transcription 3 (p-STAT3) in arthritis and liver tissue samples, we conducted Western blot analysis following the protocol described in our previous study [
22]. The antibodies used for this analysis included STAT3 and p-STAT antibodies (1:1000 dilution, rabbit polyclonal antibody; Hangzhou Hua’an Biotechnology Co., Ltd., China), glyceraldehyde 3- phosphate dehydrogenase antibody (1:1000 dilution, rabbit poly-clonal antibody; Hangzhou Hua’an Biotechnology Co., Ltd.), and
-actin antibody (1:10000 dilution, rabbit polyclonal antibody; Cell Signaling Technology, USA). All experiments were conducted in triplicates.
2.13. Statistical analyses
Data are presented using the mean values and standard deviations (SDs). Group comparisons were conducted using one-way analysis of variance, and pairwise comparisons were made using Student’s -test. Statistical significance was determined using a -value < 0.05. Visualizations were performed using GraphPad Prism 8.0 software (USA). The screening criteria for various omics differential marker included a FC or , and . A threshold of 0.60 was determined in this correlation network analysis.
The statTarget package was used for principal component analysis and partial least squares discriminant analysis (PLS-DA). The ggcor R package was used to calculate the Pearson’s rank correlation between the two variables. Statistical significance was achieved when .
2.14. Ethics approval and consent for participant
This study was approved by the Medical Ethics Committee of the First Affiliated Hospital of Anhui University of Chinese Medicine. Written informed consent was obtained from all study subjects. The animal experiments were approved by the Institute of Basic Theory for Chinese Medicine, China Academy of Chinese Medicine Science.
3. Results
3.1. Clinical transcriptome analysis reveals that iron and bone metabolism is closely related to the efficacy of TGTs in RA
In a previous study, we analyzed the transcriptomic profiles of PBMCs from RA patients with well or poor responses to TGTs treatment and identified 212 differentially expressed genes (DEGs) associated with its clinical efficacy, including 102 upregulated and 110 downregulated DEGs. All DEGs were mapped to Gene Ontology terms, KEGG, and Reactome pathways using the corresponding databases (Tables S3 and S4 in Appendix A). Functional enrichment analysis showed significant associations between DEGs related to the efficacy of TGTs and pathways involved in bone resorption, iron regulation, and osteoclast differentiation (
Fig. 2(a)). Following the construction and analysis of an interaction network among these functional DEGs, we found that the TGTs efficacy-related genes were significantly involved in iron metabolism-related pathways, such as iron ion binding, iron ion channel activity, and iron storage; and bone metabolism-related pathways, such as osteoclast differentiation and osteoprotegerin (OPG)/receptor activator of nuclear kappa-B ligand (RANKL)/receptor activator of nuclear kappa-B signaling (RANK) (
Fig. 2(b)).
3.2. Clinical metabolomics and proteomics analyses reveal that the imbalance between iron and lipid metabolisms is implicated in TGTs-induced DILI
To determine the molecular mechanisms underlying TGT-induced DILI, 17 RA patients with DILI (DILI group), 17 RA patients without DILI (non-DILI group), and 17 healthy volunteers (NC group) were enrolled for metabolomics and proteomics analyses of their serum samples. The output of partial least squares-discriminant analysis (PLS-DA) on the metabolome and proteome revealed distinct metabolic profiles among the three clinical cohorts, including the NC, DILI, and non-DILI groups (
Figs. 2(c) and
(d)). Differential metabolic analysis revealed significant differences in 24 metabolites among the three groups (Table S5 in Appendix A) and a strong correlation between glycerol phospholipid metabolism and the occurrence of DILI (
Fig. 2(e)).
Considering that most of the identified differential metabolites were lipids or lipid derivatives, which are not categorized in existing databases, we integrated the proteomic data for further investigation. A total of 49 differentially expressed proteins (13 downregulated and 36 upregulated; Table S6 in Appendix A) were associated with DILI (
Fig. 2(f)). Subsequently, we performed Mantel’s correlation analysis to examine the relationship between the differentially expressed proteins and nine distinct metabolite types (
Fig. 2(g)) and found that these proteins and metabolites were not only involved in pathways related to lipid metabolism such as glycerophospholipid, sphingolipid, and arachidonic acid metabolism but also notably enriched in pathways such as cysteine, glutamine, and glutathione metabolism (
Fig. 2(h)), which have been reported to play a role in ferroptosis via two significant processes: "glutamine-cysteine pathway" and "lipid peroxidation" [
35], implying a potential relationship between TGTs-induced DILI and ferroptosis.
3.3. STAT-HAMP signal axis mediated bone and lipid metabolism is respectively associated with the clinical efficacy and toxicity of TGTs
The above clinical multi-omics profile analysis indicates that the regulation of bone and lipid metabolism, mediated by iron homeostasis, is closely associated with the clinical efficacy and toxicity of TGTs. Based on this, we further screened the relevant signaling pathways according to the following steps: ① We identified the core targets involved in iron and bone metabolism related to the clinical efficacy of TGTs (
Figs. 3(a) and
(b)), as well as the those involved in iron and lipid metabolism related to DILI caused by TGTs (
Figs. 3(c) and
(d)). ② Docking analysis of identified core targets and chemical profiling of TGTs from previous studies (Tables S7 and S8 in Appendix A) [
19]. ③ The interaction network between the components and targets was analyzed, and cluster analysis was performed using STRING online tools. ④ The results are shown in
Figs. 3(e) and
(f) using Cytoscape 3.9.1. Based on the calculation of topological features and functional information, STAT3 and HAMP, which are involved in the regulation of iron homeostasis, were identified as key targets associated with the clinical efficacy and toxicity of TGTs. In addition, LCN2 and ACP5, which are involved in the regulation of bone metabolism, were identified as key targets associated with the effects of TGTs. ACSL4 and LPCAT3 are closely related to lipid metabolism in terms of TGTs toxicity. Notably, STAT3/HAMP may be common upstream regulators of the LCN2-ACP5 and ACSL4-LPCAT3 signal axes. These findings prompted us to hypothesize that the STAT3- HAMP/LCN2-ACP5 and STAT3-HAMP-ACSL4-LPCAT3 signaling axes are potential targets associated with the clinical efficacy and toxicity of TGTs, respectively (
Fig. 3(g)).
3.4. TGTs improve the iron and bone metabolic disorders in the inflamed joint tissues of CIA mice by inhibiting the STAT3-HAMP/ LCN2-ACP5 signal axis
As illustrated in
Fig. 4(a), behavioral measurements were taken every three days after the second immunization, and the animals were sacrificed at five and seven weeks after administration. Compared to the Con group mice, CIA mice experienced a significant decrease in body weight (6-12 days after administration,
), followed by a gradual recovery. The TGTs group also exhibited a decrease in body weight after administration, reaching its lowest value on 12th day after administration; however, overall, the values were significantly higher than those in the CIA model group before 24th day
. Additionally, TGTs induced a significant improvement in the arthritis score, paw swelling, and pain threshold in CIA mice (
Figs. 4(b)-(f)). Notably, the organ/brain index data revealed that TGTs exacerbated the increase in the liver/brain index in CIA mice; however, it reversed the spleen/brain index and the thymus/brain index (
Figs. 4(g)-(j)) in a time-dependent manner, indicating significant effects of TGTs on both the liver and immune system.
Regarding iron homeostasis, the content of both
and
in the inflamed joint tissues of CIA mice were significantly lower than that in those of normal control mice, which were effectively reversed by treatment with TGTs (all
,
Fig. 5(a)), suggesting that CIA induces an overall deficiency of iron levels in the inflamed joint tissues. According to the
ratio in joint tissues (
Fig. 5(b)), TGTs increased iron levels, particularly
levels. In addition, the IHC results, as shown in
Fig. 5(c), demonstrated that TGTs effectively suppress the expression of TFR1 in inflamed cartilage tissues (all
). TGTs regulate iron metabolism disorders in joint tissues to improve cartilage destruction caused by CIA. The expression of TFR1 protein was also measured using Western blot analysis (Fig. S1(a) in Appendix A), and similar results were obtained. Interestingly, in the TGT 5W group, cartilage expression was negative but synovial expression was positive, indicating a time-dependent feature with a heterogeneous distribution of cartilaginous synovium (
Fig. 5(d)).
Regarding bone metabolism, the histopathological scores of the ankle and knee joints in the CIA group showed a significant increase in inflammatory cell infiltration, bone destruction, and synovial hyperplasia compared to those in the Con group. After treatment with TGTs, changes in histopathological scores showed a distinct decrease in drug duration dependence (all
,
Figs. 5(e)-(g)). Additionally, Masson staining (
Figs. 5(h) and
(i)) of the inflamed knee joint tissues obtained from Con mice revealed that the articular cartilage matrix and fibers appeared blue, whereas the cartilage cytoplasm and glia appeared red, exhibiting uniform coloring, clear tide lines, and consistent gradients between the cartilage and subchondral bone boundaries. In contrast, the articular cartilage surface of the mice in the CIA 5W group was rough and damaged, with degenerated and destained superficial cartilage. In the CIA 7W group, fibrosis was evident, and there were numerous flame-like protruding red stains near the tidal line, with an unclear boundary between the subchondral bones. Notably, the TGTs group displayed various degrees of improvement in the morphology of the articular cartilage of CIA rats, with no rough or damaged cartilage surface, superficial cartilage degeneration or discoloration, or collagen fibers at the bone tidal line. These results indicate that TGTs effectively alleviated CIA-induced arthritic infiltration, bone destruction, synovial hyperplasia, and collagen fibrosis.
Mechanistically, there was an increase in the proportion of STAT3 and p-STAT3 protein expression in the CIA model group compared to that in the Con group, which was significantly reduced by the treatment (P<0.001,
Fig.5(j).Accordingly,the pro-) tein content and transcription levels of HAMP, LCN2, and ACP5 were significantly decreased in the TGTs treatment than in the Con group (P<0.01,
Figs.5(k)-(n)).These findings indicate that TGTs) improve iron and bone metabolic disorders in inflamed joint tissues of CIA mice by inhibiting the STAT3-HAMP/LCN2-ACP5 axis.
3.5. TGTs induce iron and lipid metabolic disturbances in the liver tissues of CIA mice by modulating the STAT3-HAMP-ACSL4-LPCAT3 signal axis
As shown in
Figs. 6(a) and
(b), there were no noticeable abnormalities in liver tissue morphology or hepatic sinusoids in normal control mice. The overall morphology of liver tissues in the CIA group was normal, with some areas showing slight inflammatory infiltration in the liver cord. Notably, the liver tissues of the TGTs treatment group exhibited enlarged and irregularly shaped hepatic cells with inflammatory cell infiltration, which was more pronounced in the TGT 7W treatment group, indicating that TGTs can exacerbate liver injury as the intervention time is prolonged. Additionally, we assessed the changes in liver lipid profiles (TC, TG, HDL-C, and LDL-C), liver function (AST and ALT), and lipid peroxide indicators (MDA and 4-HNE) among the different groups. As shown in
Figs. 6(c)-(j), the TGTs group exhibited a time-dependent increase in ALT and AST levels after administration compared to the Con and CIA groups
. Similarly, the levels of TC, TG, LDL-C, MDA, and 4-HNE were significantly elevated in a time-dependent manner after TGTs administration
, whereas HDL-C levels were significantly reduced
. Moreover, the severity of these markers was more pronounced in the TGT 7W treatment group than in the TGT 5W treatment group. These data indicate that TGTs induce abnormal liver function and lipid metabolism, leading to lipid peroxidation in liver tissues.
To investigate the association between disruption of iron metabolism and TGT-induced liver injury, the content of iron ions in the liver tissues detected herein demonstrated a significant increase in the content of
after the treatment (P<0.01,
Figs.7(a) and
(b)),) which subsequently led to liver injury due to the unstable
content. The results of TEM shown in
Figs. 7(c) and
(d) revealed that the nuclei of the Con group, CIA
group, and CIA
group exhibited regular and smooth characteristics, with clear distribution of inner chromatin. Cytoplasmic lipid droplets were moderate in number and displayed regular shapes. Under
magnification, the boundaries of the mitochondria were distinct and had a uniform texture. In contrast, the TGT 5W and TGT 7W groups exhibited shrunken nuclei, increased accumulation of lipid droplets with aggregation, numerous vacuoles in the chromosomes, uneven texture, and disappearance of ridge lines. These observations indicated that TGTs effectively enhanced the key characteristics of ferroptosis, including the accumulation of lipid droplets and serious mitochondrial damage in liver tissues, which was further confirmed by IHC staining of the key upstream proteins TFR1 and ACSL4 involved in "iron-lipid metabolism" (
Figs. 7(e) and
(f)). Similar results regarding the expression of TFR1 and ACSL4 proteins were obtained using Western blot analysis (Figs. S1(b) and (c) in Appendix A).
Mechanistically, the TGTs treatment groups exerted a markedly reduced effect on the expression ratio of p-STAT3 and STAT3 proteins in the liver tissues compared to the CIA group
,
Fig. 7(g)). Accordingly, the content and transcription levels of HMAP and LPCAT3 significantly decreased in the TGTs treatment groups
. These findings suggest that TGTs induce iron and lipid metabolism disorders in the liver tissues of CIA mice by regulating the STAT3-HAMP-ACSL4-LPCAT3 cross-linking signaling axis (
Figs. 7(h)-(k)).
3.6. In vitro mechanistic validation of TGTs for its efficacy and toxicity in MH7A and AML12 cells
To elucidate the mechanisms underlying the efficacy and toxicity of TGTs, we conducted in vitro experiments using the MH7A and AML12 cells. The doses of TGTs and TNF
were determined based on our preliminary experiments and previous studies [
19,
27] (Fig. S2 in Appendix A). In MH7A cells, TNF
-induced conditions were employed, and cells were treated with different concentrations of Stattic, colivelin TFA, and
TGTs for
. The results revealed that Stattic significantly inhibited p-STAT3 protein expression at concentrations of 5.0 and
, with the most pronounced effect observed at
. Conversely, col-ivelin TFA at
exhibited the most significant activation effect (P<0.001,
Fig.8(a)).Following TGTs intervention,a) synergistic effect between p-STAT3 expression and Stattic was observed
, whereas an antagonistic effect was observed with colivelin TFA (P<0.001,
Fig.8(b)).Similar observations were ") made for other components of the STAT3-HAMP/LCN2-ACP5 signaling axis (P<0.05,
Figs. 8(c)-(g)).
In AML12 cells, under conditions of STAT3 phosphorylation inhibition/activation, cells were treated with varying concentrations of Stattic and colivelin TFA, followed by TGTs treatment. Stat-tic at concentrations of2.5,5.0, and
exhibited varying degrees of inhibition of p-STAT3 protein expression, with the most significant effect at
, while
col-ivelin TFA demonstrated the most pronounced activation effect (P<0.01,
Fig.8(h)). Post TGTs intervention,a synergistic effect with) Stattic and an antagonistic effect with colivelin TFA were observed on p-STAT3 expression (
,
Fig. 8(i)). Similar effects were noted for other components of the STAT3-HAMP-ACSL4-LPCAT3 signaling axis (P<0.05,
Figs.8(j)-(n)).Additionally,iron ion ratio) imbalance and lipid droplet aggregation in AML12 cells were found after administration of TGTs (Fig. S3 in Appendix A), indicating a link between TGTs-induced toxicity and disruptions in the "iron-lipid” metabolism in AML12 cells.
4. Discussion
Elucidating the molecular mechanisms underlying the clinical efficacy and DILI of TGTs is helpful in providing the complete knowledge about the therapeutic advantages and ensure the rational application of this Chinese patent drug in clinics. Despite their evident clinical efficacy, the mechanisms underlying the use of TGTs in treating RA remain insufficiently understood and warrant further investigation. The primary active components of TGTs influence inflammatory signaling pathways, such as Toll like receptors (TLR), Janus kinase (JAK)/STAT, NOD-like receptor thermal protein domain associated protein 3 (NLRP3) inflammasome, and cyclic guanine monophosphate-adenosine monophosphate synthase (cGAS)-stimulator of interferon gene (STING) activation; however, the key targets for RA treatment remain unclear [
36⇓⇓-
39]. Molecular-level studies have primarily focused on triptolide and celastrol, raising questions about their effects on the overall effects of herbal medicines. Notably, the dual efficacy and toxicity of TGTs suggest that active targets may contribute to the adverse effects. Clarifying the active targets of TGTs could elucidate both the therapeutic and adverse mechanisms, supporting the theoretical foundation for combination therapies in clinical practice. In contrast to previous studies that focused on exploring the mechanisms of the efficacy or toxicity of the TGTs alone, our comprehensive analysis revealed significant variations in transcriptome-protein-metabolite profiles based on clinical cohorts, distinguishing RA patients with favorable responses to TGTs from those with poor outcomes. Intriguingly, the clinical efficacy and DILI of TGTs against RA were associated with metabolic homeostasis between iron and bone and between iron and lipids, respectively. Among these, the STAT3-HAMP/LCN2-ACP5 and STAT3-HAMP-ACSL4-LPCAT3 axes were identified as the key drivers of the efficacy and toxicity of TGTs. Experimentally, TGTs have been demonstrated to play dual roles in ameliorating cartilage damage and synovial hyperplasia in CIA mice, as well as in inducing hepatic dysfunction, lipid metabolism disruption, and hepatic lipid peroxidation. Notably, TGTs effectively reversed "iron-bone" disruptions in the inflamed joint tissues of CIA mice by inhibiting STAT3-HAMP/LCN2-ACP5 axis, subsequently leading to "iron-lipid" disturbances in the liver tissues via modulation of the STAT3-HAMP-ACSL4-LPCAT3 axis. To reinforce our findings, additional bidirectional validation experiments were conducted using MH7A and AML12 cells to confirm the bidirectional regulatory effects of TGTs on key targets. To the best of our knowledge, this is the first study to comprehensively determine the clinical efficacy and toxicity of TGTs and to clarify the underlying mechanisms, unlike previous studies that solely focused on the exploration of the clinical efficacy or toxicity of TGTs alone.
Iron is a vital trace element that plays a crucial role in normal functioning of the human body. Under normal physiological conditions, the body maintains iron homeostasis through precise regulatory mechanisms. Disturbances in iron homeostasis can lead to the onset and progression of various diseases [
40]. Iron homeostasis has been indicated to be closely related to bone metabolism, and both iron overload and iron deficiency have adverse effects on bone metabolism [
41,
42]. Clinical data have reported that approximately
of RA patients experience iron deficiency, leading to anemia, which not only exacerbates joint destruction, but also serves as an independent factor for assessing the prognosis of RA patients [
43]. Recent studies have revealed that iron deficiency not only increase the expression of osteoclast signature genes ACP5, cathepsin K, and matrix metalloproteinase 9 and promote osteoclast activation, but also disrupt bone formation by reducing the expression of bone formation-related genes like bone morphogenetic proteins (BMPs) [
44,
45]. These findings highlight the significance of abnormal iron homeostasis in osteoclast activation and bone destruction during RA progression. Reversing the imbalance of iron metabolism has been proven to be an effective approach in preventing and treating bone destruction in RA [
46,
47]. Accordingly, our clinical transcriptomic profiling revealed that the TGTs’ efficacy-related genes are significantly involved in a series of biological processes in iron and bone metabolism pathways, such as iron ion binding, iron ion channel activity, iron storage, bone resorption, iron regulation, and osteoclast differentiation. HAMP acts as a master regulator of iron homeostasis and is an upstream regulatory target of STAT3 [
48]. Inflammatory conditions activate the phosphorylation of STAT3 proteins, subsequently increasing the transcription of HAMP and generating iron deficiency in the body circulation [
49]. The active ingredients of TGTs, such as celastrol and triptolide, block the phosphorylation of STAT3 proteins in osteoclast [
50⇓-
52]. Our chemical and target profiling of TGTs revealed that STAT3 is a putative target of the main active ingredients of TGTs such as celastrol and triptohairic acid. Interestingly, STAT3 was one of the downregulated DEGs after treatment with TGTs, according to our clinical transcriptomics. LCN2 is an acute-phase protein that regulates iron transport under inflammatory conditions [
53,
54]. LCN2 and JAK2/STAT3 have been reported to mutually promote each other, leading to increased inflammatory damage [
55]. Notably, our data showed a significant increase in HAMP and LCN2 expression in the inflamed joint tissues of CIA mice, which was distinctly downregulated in the TGTs treatment groups. Moreover, TGTs have been shown to prevent the phosphorylation of STAT3 protein, reduce HAMP transcription, inhibit crosstalk activation with
, regulate iron uptake and absorption by joints, and control the release of iron stored in cells, leading to increased utilization of
and
and attenuation of iron deficiency. Therefore, the restoration of iron content in joint tissues subsequently ameliorates the overexpression of the bone metabolism marker ACP5, ultimately improving "iron-bone" metabolic disorder in inflamed joints.
The liver plays a crucial role in iron uptake, storage, and transportation in the body [
56]. It is also the primary organ affected by drug poisoning [
57]. Maintaining iron homeostasis is essential for liver cell function and the prevention of various liver pathologies [
58]. Recent studies have shown that liver cells may experience excessive
accumulation in the presence of iron overload. This accumulation leads to the production of excessive reactive oxygen species through the Fenton reaction, disrupting the balance between the intracellular oxidation system and the antioxidant system [
59,
60], is consistent with our results indicating a significant increase in
, MDA, and 4-HNE content in CIA mice after treatment with TGTs. Additionally, the Fenton reaction generates hydroxyl radicals
, which serve as synthetic substrates for ACSL4. These radicals, along with LPCAT3, catalyze the peroxidation of polyunsaturated fatty acids (PUFAs), resulting in abnormal lipid metabolism and ultimately inducing DILI [
61]. It has been speculated that TGT-induced DILI is affected by iron and lipid metabolic disorders. Based on clinical metabolomic and proteomic profiling, we identified differentially expressed proteins and metabolites associated with TGT-induced DILI. These proteins and metabolites are involved in metabolic pathways that directly affect iron overload sensitivity and lipid peroxidation. HAMP is mainly synthesized and released by the liver in response to circulating iron levels [
62]. Notably, celastrol decreases hepatocyte HAMP levels during inflammation. This reduction in HAMP levels leads to an increase in iron absorption by the body and an increase in iron ion concentration, ultimately resulting in hepatotoxicity [
63]. Accordingly, in our study, a time-dependent decrease in HAMP expression and content in the liver tissues of CIA mice after the administration of TGTs and the observed time-dependent increases in ALT and AST levels indicated the development and progression of hepatotoxicity. A recent study showed that exogenous phosphatidic acid can reduce acetaminophen-induced injury by inhibiting the IL-6/STAT3 pathway, complementing our research on TGT-induced DILI mechanisms [
64]. In accordance with our findings, TGTs inhibit the STAT3-HAMP signal axis, subsequently resulting in a distinct elevation in the concentration of
in the liver tissues, after liver tissue overload, the Fenton reaction activates the ACSL4-LPCAT3 signaling axis, which induce the liver injury-related lipid peroxidation of PUFAs and "iron-lipid" metabolic disorders.
In conclusion, our data highlighted the association between iron-mediated metabolic homeostasis and the clinical efficacy and toxicity of TGTs in RA therapy, offering guidance for the rational clinical use of TwHF-based therapies with dual therapeutic and toxic potential. Further research based on organoids chips and single cell sequencing, which may be helpful in investigating the molecular mechanisms underlying the heterogeneity of the efficacy and toxicity of TGTs in different target organs and cells, will be carried out in our future study.
Acknowledgments
This study was supported by the Scientific and Technological Innovation Project of the China Academy of Chinese Medical Sciences (CI2021A03807 and CI2021A01501), the National Natural Science Foundation of China (82330124), the Beijing Municipal Natural Science Foundation (7212186), the Innovation Team and Talents Cultivation Program of National Administration of Traditional Chinese Medicine (ZYYCXTD-C-202002), and the Key Laboratory of Beijing for Identification and Safety Evaluation of Chinese Medicine, Institute of Chinese Materia Medica, China Academy of Chinese Medical Sciences.
Compliance with ethics guidelines
Zihe Ding, Xiaoyue Wang, Yi Zhang, Jian Liu, Lei Wan, Tao Li, Lin Chen, Na Lin, and Yanqiong Zhang declare that they have no conflict of interest or financial conflicts to disclose.
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2024.04.003.
PDF
(6526KB)
