Bioactive fractions from Securidaca inappendiculata alleviated collagen‐induced arthritis in rats by regulating metabolism‐related signaling

Abstract

Securidaca inappendiculata is a xanthone rich medicinal plant that has been used in the treatment of inflammation and autoimmune diseases like rheumatoid arthritis (RA) for centuries; however, the material base and mechanism of action responsible for its anti‐arthritis effect still remains elusive. The objective of this study is to evaluate the therapeutic effects of xanthone‐enriched extract of the plant against collagen‐induced arthritis (CIA) in rats and explore the underlying mechanisms. The xanthone‐deprived fraction (XDF) and xanthone‐rich fraction (XRF) were obtained by using a resin adsorption coupled with acid‐base treatment method, and their chemical composition difference was characterized by UPLC‐MS/MS analysis. Effects of the two on CIA were analyzed using radiographic, histological, and immunohistochemical analyses. The results indicated that XRF alleviated joint structures destructions with the higher efficacy than XDF, and decreased levels of TNF‐α, IL‐6, and anti‐cyclic citrullinated peptide antibody in CIA rats significantly. Furthermore, XRF inhibited nicotinamide phosphoribosyl transferase (NAMPT) mediated fat biosynthesis and utilization indicated by clinical evidences and metabonomics analysis, which thereby disrupted energy‐metabolism feedback. In addition, Toll‐like Receptor 4 and High Mobility Group Protein 1 expressions were downregulated in XRF‐treated CIA rats. Collective evidences suggest NAMPT could be an ideal target for RA treatments and reveal a novel antirheumatic mechanism of S. inappendiculata by regulating NAMPT controlled fat metabolism.

INTRODUCTION

Rheumatoid arthritis (RA) is considered as one of the most common and prevalent autoimmune disorders that primarily affects the joints. The negative impact of RA cannot be overemphasized, as it significantly decreases the quality of life and ability to effectively perform daily activities of people living with the disease. RA is characterized with chronic inflammation of the synovial, swelling, and deformities of the joints, destruction of the bone and cartilage as well as autoantibody production. RA patients suffer consistent joint pain and progressive disability. Pathogenesis of RA is very complex and it is thought to be mediated through several pathways which are yet to be fully comprehended. However, several bodies of evidences have suggested that the chronic inflammation is critically implicated in the pathology of RA.

Securidaca inappendiculata Hassk. is a medicinal plant belonging to the family Polygalaceae and it is native to the Southern China provinces of Yunnan, Guangxi, Guangdong, and Hainan. The bark and root of the plant are used to treat inflammations and RA for centuries. Our previous studies have solidly validated its therapeutic effects on experimental arthritis in rodents and xanthone derivatives were found to contribute significantly to the antirheumatic potentials. However, the distribution of every single xanthone in the plant is limited. As such we speculated that all xanthone derivatives should be deemed as a whole when investigating the anti-arthritis properties, because most compounds exert bioactivities in the dose-dependent manners. Furthermore, oral bioavailability of xanthones is poor, while the intestinal absorption will be improved when in the form of mixture. Under such conditions, development of preparations with enriched xanthone is of great clinical interest for the full exploration of S. inappendiculata. Since most of xanthones are highly lipophilic, and substituted with aromatic hydroxyl groups, we divided the ethanol extract into two different parts based on xanthone distributions by using a resin adsorption coupled with acid-base treatment method, and subsequently compared their therapeutic effects on collagen-induced arthritis (CIA) in rats in this study. Analysis of obtained results preliminary indicated that xanthone enrichment is a simple and realistic way to improve clinical performance of S. inappendiculata. Additionally, experimental evidences revealed a novel therapeutic mechanism of S. inappendiculata on CIA by regulating metabolism-related signaling.

METHODS Chemicals and reagents

Immunization grade incomplete Freund's adjuvant and lyophilized bovine type II collagen were products of Chondex Inc. (Redmond, Washington). Primary antibodies used in the immunohistochemical experiments including anti-nicotinamide phosphoribosyl transferase (NAMPT), Sirtuin1 (Sirt1), Toll-like Receptor 4 (TLR4), and High Mobility Group Protein 1 (HMGB1) antibodies were purchased from Affinity Biosciences (Cincinnati, Ohio). Nicotinamide adenine dinucleotide (NAD) qualification kit and ELISA kits were provided by Solarbio Science and Technology (Beijing, China) and Multisciences Biotech (Hangzhou, China), respectively. Other solvents of analytical grade were supplied by Merck Chemicals (Shanghai, China).

Plant material, extraction and fraction separation

The bark of S. inappendiculata Hassk. was obtained from a medicinal herb market in Bozhou, Anhui Province and authenticated at Wannan Medical College by associated professor J.Z. A reference specimen (2017-11-034) was deposited at Herbarium of Anhui College of Traditional Chinese Medicine.

The plant material was thoroughly washed, dried, powdered, and exhaustively extracted with 95% ethanol under reflux and the resulting ethanolic solution was concentrated with a rotary evaporator. The crude ethanol extract was further subjected to a HPD300 macro porous resin column and eluted with gradient mixtures of EtOH/H₂O (9%-95%). The fractions collected from 9% to 15% ethanol was designated as part A, while fractions collected from 16% to 95% ethanol was designated as part B. Part B was dried and dissolved in 0.5% sodium hydroxide solution. The insoluble portion was then combined with part A and assigned as xanthone-deprived fraction (XDF). The solution was then acidified with hydrochloric acid and the insoluble precipitate obtained was designated as -rich fraction (XRF). The yields of XDF and XRF were 9.9% and 2.0%, respectively. Folin-Ciocalteu reagent test indicated that polyphenol contents including xanthones in XDF and XRF were 1.53% and 12.69%, respectively.

UPLC-MS/MS analysis

Main compounds in the fractions were identified by a UPLC-MS/MS-based method. Chromatographic separation was performed on a Ultimate 3000 system (Thermo, San Jose, California) equipped with an ACQUITY UPLC HSS T3 C18 column (150 × 2.1 mm, 1.8 μm, Waters). The temperatures of column and autosampler were set at 40°C and 8°C, respectively. A gradient elution at the flow rate of 0.25 mL/min was adopted, and 0.1% formic acid in water and 0.1% formic acid in acetonitrile were served as phases A and B, respectively. The elution schedule was as follows: 0 to 1 minute, 2% B; 1 to 9 minutes, 2% to 50% B; 9 to 12 minutes, 50% to 98% B; 12 to 13.5 minutes, 98% B; 13.5 to 14 minutes, 98% to 2% B; 14 to 20 minutes, 2% B. ESI-MS detection was achieved on a Thermo Q Exactive Focus mass spectrometer (San Jose, California) under the positive ion mode. Main parameters adopted were as following: spray voltage, 3.8 kV; capillary temperature, 325°C; mass range, 150 to 400 (M/Z); mass resolution, 70 000. The data-dependent acquisition MS/MS analysis was carried out under HCD scan mode with normalized collision energy of 30 eV.

Animals

Male Sprague Dawley rats (7 weeks old) were purchased from Qinglongshan Laboratory Animal Company (Nanjing, China). The animals were housed in a controlled environment (21°C ± 2°C, 50% ± 5% relative humidity) with a 12 hours light/dark cycle. The animals were allowed unlimited access to food and water and acclimatized for 7 days. All animal experiments were performed in accordance with the guidelines of the Institutional Animal Care and Use Committee of Wannan Medical College.

Induction of CIA and evaluation of arthritis

CIA was induced in rats according to the previously described protocol. In short, rats in CIA and treated groups received a multiple-point subcutaneous injection with bovine type II collagen emulsion at day 0 and a booster injection was administered 7 days later. After the first induction, the animals were randomly assigned into four different groups and received treatments for 44 days. Rats in treatment groups were administrated with XRF or XDF by oral gavage at the dose of 80 mg/kg/day, while the health and CIA model controls were treated with 0.5% sodium carboxymethylcellulose instead. The progress of CIA was evaluated by arthritis score, which was defined in our previous report. Briefly, clinical manifestation in each paw was scored on a scale of 0 to 4:0, no obvious changes; 1, limited edema; 2, moderate local swelling; 3, significant edema and erythema; and 4, severe swelling throughout the whole paw. Accordingly, the highest sum was 16. Prior to sacrifice, hind limbs of all the rats were examined via lateral and frontal digital radiography (DR).

Sampling and sacrifice

At the end of the experiment (3 hours after the last treatment), all the rats were anesthetized and blood samples were collected through the abdominal aorta. The blood was allowed to stand at room temperature for 15 minutes and then centrifuged at 3000 rpm for 15 minutes to obtain serum, which was then stored at −80°C until further analyses. One portion of the serum was used for TNF-α, IL-6, anti-cyclic citrullinated peptide (anti-CCP) antibody and extracellular NAMPT (eNAMPT) determination with the aid of commercially available ELISA kits according to the manufacturers' instructions. Another portion of serum was used to perform metabolomics analysis using a ¹H NMR-based method. The rats were then sacrificed and spleen, thymus, fat pad (from left kidneys), and hind paws including ankle joints were obtained, weighted, and preserved in buffered formalin.

Histological and immunohistochemical examinations

The fixed organs/tissues were trimmed and washed extensively with running water. The joint specimens were further decalcified in ethylenediaminetetraacetic acid solution for 2 weeks. Afterwards, they were subjected to dehydration and clearing processes with ethanol and xylene, respectively. Processed specimens were then embedded in paraffin and sectioned at 4 μm thickness. After heating at 60°C for 5 hours, the sections mounted on glass slides were dewaxed and dehydrated. Finally, hematoxylin/eosin (HE) staining was carried out and pathological progression of CIA was evaluated based on histological changes observed under a BH-2 microscope (Olympus, Tokyo, Japan). The fat pads were frozen in liquid nitrogen and sectioned at 8 μm thickness. After pretreatments with isopropanol, the specimens were dyed with filtered Oil Red O working solution for 10 minutes at room temperature and then destained with 60% isopropanol. Thereafter, rinsed sections were mounted on glasses. Major characteristics of fat tissue including adipocyte morphology and lipid droplets distribution were observed and photographed using a light microscope with an in-build camera.

The sections for the immunohistochemical assay were dewaxed in xylol and dehydrated using gradient ethanol. Subsequently, the antigen retrieval was performed using citric acid coupled with microwave heating method and endogenous peroxidase was depleted with hydrogen peroxide. Further immunohistochemical staining was based on the diaminobenzidine peroxidase-antiperoxidase technique. Briefly, slides were incubated with normal goat serum, primary antibodies, biotinylated secondary antibody, peroxidase-conjugated streptavidin biotin complex in turns. The slides were thoroughly rinsed with PBS after every other step. The signals were visualized by using diaminobenzedine as the chromogen and the nuclei were counterstained with hematoxylin. Immunohistochemical analysis was also performed on the frozen fat tissues using the same procedure as described above.

NMR metabolomics analysis

A total of 400 mL of serum was mixed with 200 μL of phosphate buffer solution (90 mM K₂HPO₄/NaH₂PO₄, pH 7.4, 99.9% D₂O). The mixture was centrifuged at 10 000 rpm for 10 minutes at 4°C, and 500 μL of the supernatant obtained was pipetted into 5 mm NMR tube for NMR experiment. The ¹H NMR spectra of the samples were acquired on a 600 MHz Bruker spectrometer at 296 K. Standard ¹H spectra were acquired with a water-suppressed CPMG pulse sequence. For each sample, 64 FIDs were collected into 64 K data points over a spectral width of 12 000 Hz with a relaxation time of 6.5 μs and an acquisition time of 2.66 seconds.

Data preprocessing and metabolomics analysis

All ¹H NMR spectra were phased and baseline corrected manually using MestReNova software (version 9.0.1, Mestrelab Research S.L.). The processed spectra were then calibrated to the internal lactate CH₃ resonance at 1.33 ppm. All signals were also peak aligned manually to reduce random peak-shift. Afterwards, chemical shifts representing resonances from water and urea, as well as the baseline (peak-free) were removed from further analysis. Spectra resonances between 0.5 and 9.0 ppm were binned into buckets using adaptive binning method, and each peak was binned as one bucket. The binned data were saved as Microsoft Excel format file and fed into SIMCA-P software (version 14, Umetrics AB, Umea, Sweden) for multivariate analysis. The principal component analysis (PCA) was performed for data distribution overview. Then, partial least-squares discriminant analysis (PLS-DA) and orthogonal PLS-DA (OPLS-DA) were implemented to identify the metabolites accounting for differences among groups. All of the above analyses were based on Pareto scaling. PLS-DA and OPLS-DA models were evaluated and validated using sevenfold cross-validation and permutation test. By combining with HMDB database (http://www.hmdb.ca/) and an in-house developed NMR database, Chenomx NMR suite (https://www.chenomx.com/) was used to assign metabolite peaks by fitting the standard spectra of metabolites in the database to the obtained ¹H NMR spectra of bio-fluid samples.

Statistical analysis

Results were presented as mean ± SD. Statistical differences were evaluated using one-way ANOVA and Student's t test with Bonferroni correction using SPSS software (version 14.0, SPSS, Inc., Chicago, Illinois). The linear regression models for all index-pairs were developed using SPSS software, and the determination coefficients (R²) of the models were also given.

RESULTS Chemical profile identification of XRF

Since most of polyphenols including xanthones obtained from S. inappendiculata were with small molecular weight, we set the scan range (M/Z⁺) between 150 and 450. The total ion chromatograms indicated that retention times of majority of compounds in XRF were in the range of 11 to 13 minutes, and these signals were very weak or absent in XDF (Figure ). By comparing MS characteristics with our previously isolated compounds, we identified six peaks with the highest abundance: (a) 11.48 minutes, 1,3,7-trihydroxy-2-methoxyxanthone; (b) 11.75 minutes, 1,3,7-trihydroxy-2,8-dimethoxyxanthone; (c) 11.92 minutes, 7-hydroxy-1, 2-dimethoxyxanthone; (d) 12.14 minutes, 1,7-dihydroxy-3,4-di-methoxyxanthone; (e) 12.54 minutes, 1,7-dihydroxy-4-methoxyxanthone; (f) 12.95 minutes, 1,7-dihydroxyxanthone. Tandem mass spectra of all these compounds were shown in S1.

Total ion chromatograms of xanthone-deprived fraction (XDF) and xanthone-rich fraction (XRF) acquired by UPLC-MS/MS analysis under the positive iron mode

Effects of treatments on the severity of CIA in rats

Overall, therapeutic effects of XRF and XDF on CIA severity are shown in Figure . As observed, 10 days after the first injection, extensive signs of CIA were developed in rats, including limp, paw edema, and joint deformation. These pathological changes progressed rapidly after the second booster injection with dramatic increase in arthritis scores (Figure A). During the later stages, the soft tissue swelling was gradually ameliorated, but ankylosis and dysfunction of ankle joints were readily observed in all CIA rats. Furthermore, the body weight gain in CIA rats was significantly reduced compared with healthy controls since day 13 (Figure B). Although treatments with both XRF and XDF exhibited promising effects on arthritis scores, the overall clinical efficiency of XRF was better than XDF. As shown in Figure C, XRF efficiently reduced the bulbous swelling on paws, while XDF brought no benefit in this regard. The therapeutic advantages of XRF were further confirmed by DR and histological analyses. DR examination showed that CIA rats suffered severe bone density loss and joints structure damages. XRF largely restored these pathological changes, while obvious joint space narrowing and bone/cartilage erosion can still be noticed in XDF-treated rats (Figure C). Histological examination found similar results. The joint space in both CIA model and XDF-treated rats was greatly occupied by expanded synovium, which eventually led to fusion of nearby bones and ankylosis of joints. The hyperplastic synovium also caused notable cartilage/bone erosion. Meanwhile, arthritis manifestations including pannus formation, inflammatory infiltration, and cartilage/bone erosion were effectively abrogated by XRF (Figure C). Additionally, histological examination revealed significant hyperplasia occurred in both spleen and thymus in CIA rats. Splenic white pulp in these animals expanded a lot, which should account for the increased volume of spleen. Meanwhile, the lymphoid progenitor population was increased in spleen of CIA rats too. These signs of immune hyperactivation were restored to a certain extent by the both treatments (S2). Despite of weight difference, no obvious histological changes in thymus was found among groups. Surprisingly, both XRF and XDF did not restore the weight loss in CIA rats, but rather aggravated this trend (Figure B). Upon further analysis, we noticed that the negative effect on body weight occurred prior to onset of CIA (since day 5). It suggested that this phenomenon was not resulted from arthritis progression, but served as a direct indication of the intervention of treatments into fat metabolism.

Therapeutic effects of xanthone-rich fraction (XRF) and xanthone-deprived fraction (XDF) on collagen-induced arthritis (CIA) in rats. A, Periodic changes of arthritis score, B, periodic changes of body weight, C, the clinical manifestation of arthritis: a, morphological observation of hind paw (red arrow: Bulbous swelling); b, frontal radiography examination of hind paws; c, lateral radiography examination of hind limb (red arrow: Joint space narrowing and fusion of bones); and d, histological examination of ankle joints from hind paw (black arrow: Synovium caused cartilage erosion), D, levels of serological biomarkers evaluated by ELISA method. *P [less than] .05 and **P [less than] .01 compared with CIA models

Effect of treatments on RA-related biomarkers in CIA rats

To further validate therapeutic effects of the treatments on CIA, we analyzed some RA-related biomarkers, including TNF-α, IL-6, and anti-CCP antibody, and found all these parameters were significantly suppressed by both treatments (Figure D). To our disappointment, there was no significant differences observed between XDF and XRF-treated rats, which was contradictory to DR and histological examinations. We speculated that some mechanisms different from classic cytokines dominated theories could account for this paradox.

XRF intervened into fat metabolism in CIA rats

The volume of fat pad varied a lot among different groups. Generally, fat reserves in healthy rats were much more than that in CIA models. XRF treatment resulted in even further reduction in fat distribution and shrinkage in size of adipocytes in CIA rats (Figure A). Meanwhile, it suppressed inflammatory cells infiltration into white adipose tissues (WAT) with high efficacy (Figure B). The inhibition of XRF on fat accumulation was further supported by decreased weight of fat pad (Figure C). Although XDF exerted even more profound effect on body weight than XRF, it had no effects on either histological structure or volume of fat pad. Subsequently, we qualified two serological biomarkers related to fat metabolism. It was found that both XDF and XRF treatments resulted in remarkable reduction in Malonyl-CoA, but only XRF suppressed the abnormal increase of circulating fatty acids significantly (Figure C).

Effects of xanthone-rich fraction (XRF) and xanthone-deprived fraction (XDF) on fat metabolism in collagen-induced arthritis (CIA) rats. A, Histological examination of fat pad (Oil Red O staining), B, histological examination of fat pad (HE staining), C, effects of treatments on fat metabolism-related indicators, D, effects of treatments on paw index and circulating extracellular nicotinamide phosphoribosyl transferase (NAMPT) (eNAMPT)/nicotinamide adenine dinucleotide (NAD) levels. *P [less than] .05 and **P [less than] .01 compared with CIA models

Previously, we found xanthones can alleviate inflammations by manipulating NAMPT/NAD pathway, a pivotal role in fat metabolism and energy homeostasis. We speculated that similar mechanism was also involved in therapeutic actions of S. inappendiculata on CIA, and roughly tested this claim by analyzing levels of eNAMPT and NAD in serum. Indeed, both treatments downregulated eNAMPT concentration in CIA rats. At the same time, XRF reduced circulating NAD a bit although it was not significant because of huge individual variation, while XDF did not bring down but slightly increased the concentration (Figure D).

Correlation between fat metabolism and CIA severity in rats

Inspired by the conceptualized energy-inflammation feedback, we attempted to explore the clinical implication of fat metabolism intervention in CIA treatments by analyzing the correlation among different parameters, and the overall results were summarized in Figure A. Of great interest, levels of methylmalonyl-coenzyme A (Malonyl-CoA) were fluctuated a lot among individuals proportionate to RA-related serological indicators, including anti-CCP antibody, TNF-α and IL-6. The distribution of eNAMPT, a typical adipokine in rats was also highly correlated to levels of anti-CCP antibody and TNF-α (Figure B). These results provided the preliminary evidences supporting the notion that intervening in fat metabolism could have significant implication on RA treatments.

Correlation among different clinical parameters from individual rats. A, The summary of correlation analysis. B, The statistical correlation between fat metabolism-related parameters and diagnostic biomarkers of RA

XRF inhibited fat utilization in CIA rats

Considering the relevance between fat metabolism and CIA severity revealed above, the effects of XRF on fat metabolism would give it some advantages in CIA treatments over XDF. Thereafter, it is necessary to further characterize the impacts of XRF on metabolism. Typical 600 MHz ¹H NMR spectra of rat serum samples from all groups (XRF, XDF, CIA model, and healthy control) are shown in S3. A total of 40 metabolites were assigned and labeled on the spectra with references. To explore metabolic differences among different groups, PCA, PLS-DA, and OPLS-DA models were developed on the binned data. The PCA score plot shows no outlier in the samples, that is, all samples fell in the 95% confident intervals in PC1-PC2 subplace. Score plots for most individual rats from different groups were well separated in PLS-DA model (Figure A). It confirmed the successful development of CIA in rats, and substantial metabolism alterations induced by treatments.

Metabolomics study based on 1H NMR analysis of serum samples. A, Principal component analysis (PCA) (a) and partial least-squares discriminant analysis (PLS-DA) (b) score plots for rats in all groups (PC1 and PC2 represent first two principal components in PCA modeling; t1 and t2 represent first two latent variable in PLS-DA modeling), B, orthogonal PLS-DA (OPLS-DA) (OPLS-DA) modeling developed on xanthone-rich fraction (XRF) and collagen-induced arthritis (CIA) groups: a, OPLS-DA score plots (t1 and t2 represent parallel component and orthogonal component, respectively); b, results of sevenfold cross-validation tests (X-axes represent correlation coefficients between the response variable and permuted response variable, and Y-axes represent R2 and Q2 values of the permuted models), C, discriminating metabolites differentiated XRF and CIA groups: a, discriminating metabolites visualized in volcano plot (X-axes represent logarithm of fold changes within XRF group compared with CIA controls, and Y-axes represent logarithm of P-value of one-way analysis of variance); b,c, levels of glycerol and acetone in individual rats

To specify metabolic changes occurred in XRF-treated rats, OPLS-DA modeling was carried out based on raw data from XRF and CIA groups. As shown in Figure B, this multivariate statistics method perfectly differentiated the two groups. Cross-validation further tested the reliability of this modeling. Subsequently, we visualized altered metabolites in colored volcano plot, and found that metabolites involved in fat metabolism were significantly affected. Among them, decreased 3-hydroxybutyric acid, glycerol, and acetone in XRF-treated rats were especially notable and meaningful, as they are important intermediates involved in fat mobilization and β-oxidation processes (Figure C). By the comparison, XDF had no influence on all these indexes. The metabolomics clues together with free fatty acids changes collectively demonstrated XRF rather than XDF significantly inhibited fat utilization in vivo.

XRF downregulated NAMPT signaling in CIA rats

Considering its effects on levels of eNAMPT/NAD and fat metabolism, XRF could have profound effects on NAMPT signaling. Firstly, we analyzed NAMPT expression in WAT by the immunohistochemical method. Consistent to anticipation, XRF but not XDF significantly decreased this level. However, as the main downstream target of NAMPT/NAD signaling, Sirt1 was barely affected by the two treatments (Figure A).

Effects of xanthone-rich fraction (XRF) and xanthone-deprived fraction (XDF) on protein expressions in rats. A, Nicotinamide phosphoribosyl transferase (NAMPT) and Sirt1 expressions in fat pad, B, NAMPT, TLR4, and HMGB1 expressions in synovium. *P [less than] .05 and **P [less than] .01 compared with collagen-induced arthritis (CIA) models; #P [less than] .05 compared with XDF-treated rats

Overexpression of TLR4 in synovium is one of the dominating factors leading to chronic inflammation and progressive damages in joints under RA circumstances. As the positive correlation between NAMPT and TLR4 status has been preliminarily established, we investigated the two in joints consequently. It was found that NAMPT in CIA rats was dramatically increased compared with normal controls. Different with XDF, XRF exhibited potent inhibitory effects on the abnormally increased NAMPT. It almost eliminated all accumulated NAMPT in synovium except from some synovial lining region. Similar effect was observed for the expression of TLR4 too (Figure B). Additionally, the anoxia condition related to TLR4 activation was also significantly ameliorated upon both treatment suggested by the significant decrease of HMGB1 (Figure B). Further semi-quantification of these results was performed by using the ImageJ software (version 1.52a, NIH, Bethesda, Maryland). All imaged used in the analysis were included in S4-S8.

DISCUSSION

As a traditional antirheumatic drug, S. inappendiculata has been used for RA treatment for centuries. Although the chemical profile and its antirheumatic effects have been preliminarily investigated, the material base accounting for its clinical application and relevant therapeutic mechanisms are still not fully understood. We previously discovered that xanthone derivatives are important bioactive ingredients in the plant, which are capable of manipulating many stress sensitive pathways, such as NF-κB and MAPKs, and subsequently exert anti-inflammatory effects. However, the natural distribution of xanthones and their oral bioavailability are limited. Current study was designed to test if xanthone-enriched preparation of S. inappendiculata. had more promising application in RA treatments. Obtained results are encouraging, as with the dose equivalent to less than 10 mg/mL of pure xanthones, XRF efficiently reduced the severity of CIA. Previously, we investigated therapeutic effects of some S. inappendiculata derived xanthones on adjuvant-induced arthritis in rats. However, it is impossible for us to compare these experimental outcomes because the two studies were performed on different disease models, while a followed up research demonstrated that XRF could have better antirheumatic potentials than pure xanthones. α-Mangostin is a naturally occurring xanthone isolated from mangosteen, and possesses many pharmacology and pharmacokinetics advantages over simple xanthones because of long chain substituents. We found it can significantly ameliorate clinical manifestation of CIA in rats at the dose of 40 mg/kg/day. Despite of this, certain pathological changes including pannus formation and cartilage/bone erosion were still noticed under such treatment. By comparison, XRF treatment resulted in excellent protection on joints against damages (Figure B). Although mechanism underlying the superior therapeutic effects of xanthones in the form of mixture is still elusive, available clues suggested this phenomenon had something to do with the improved biological availability.

According to Traditional Chinese Medicine Theory, S. inappendiculata is a cold natured drug, and mainly used to expel pathogenic hot and dampness. Hence, it is reasonable to hypothesize that alteration of energy metabolism could be critically involved in its therapeutic actions on hot syndrome-related RA. Both clinical and experimental evidences suggested that fat metabolism disorder is a common complication of RA. Accumulation of fat inevitably fuels the development of chronic inflammation by providing extra energy supply, and some degradation products from β-oxidation will also contribute to the aggravation of inflammation by acting as agonists of pro-inflammatory receptors like TLR4 or second messengers implicated in inflammatory reactions. These clues inspired us to think over the therapeutic mechanisms of S. inappendiculata on RA/CIA from a novel perspective. According to revealed correlation between Malonyl-CoA and RA-related biomarkers, both XDF and XRF seem to be capable of alleviating CIA by inhibiting fat biosynthesis. However, XDF did not reduce fat reserve actually (Figure ). It hinted that the negative effect of XDF on Malonyl-CoA mediated fat biosynthesis could be compensated by other mechanisms, and XDF induced body weight loss probably due to accelerated degradation of protein. This claim was then supported by metabonomics evidence, as XDF treatment augmented urea production in CIA rats significantly (S9). Besides from the effects on fat biosynthesis, XRF also substantially suppressed fat utilization as it reduced free fatty acids, glycerol and acetone simultaneously. Collectively, it can be deduced that XRF and the within xanthones contribute largely to the cold nature of S. inappendiculata and its antirheumatic effects by modifying fat metabolism.

Because energy metabolism and inflammation converge on NAMPT signaling, theoretically inhibition on this pathway will disrupt the energy-inflammation feedback. Based on well-recognized promoting role of NAMPT in adipopexis, reduced fat reserves hinted XRF could inhibit NAMPT signaling efficiently. Reduced eNAMPT under XRF treatment partially supported above hypothesis. However, it cannot serve as a solid evidence for the regulation of XRF on NAMPT/NAD pathway, as eNAMPT is usually taken as an adipokine rather than metabolic enzyme, and is not necessary to be proportionate to intracellular NAMPT (iNAMPT). Decreased circulating NAD under XRF treatment was more meaningful, although this effect was not statistically significant due to the individual variation. The immunohistochemical experiments provided more convincing evidences. XRF did not only reduce NAMPT expression in WAT but also in synovium. As NAD is indispensable for fat utilization, decreased NAMPT expression under XRF treatment was crucial for the inhibition on fat mobilization and oxidation. In sum, these clues suggested XRF could alter fat metabolism profile in CIA rats by inhibiting NAMPT signaling.

Although energy-inflammation feedback has been well-accepted, exact mechanism underling it is still under investigation. As a key component in innate immunity, the role of TLR4 in this feedback cannot be ignored. Available evidences suggested eNAMPT can activate TLR4/NF-κB pathway directly by acting as a pro-inflammatory cytokine. Besides, we also found upregulation of NAMPT/NAD can substantially promote TLR4 expression. In this study, the correlation between NAMPT and TLR4 expressions was noticed once again. Despite the mechanism underlying NAMPT elicited TLR4 expression is not thoroughly understood, inhibition of eNAMPT/iNAMPT was favorable to the alleviation of TLR4 controlled inflammation under XRF treatment. Together with decreased HMGB1 and harmful metabolites from β-oxidation, all these factors collectively led to sustained downregulation of TLR4 pathways. These evidences do not only partially explain the diversified therapeutic effects of XDF and XRF on CIA, but also further prove that NAMPT could be a feasible target for RA treatments.

CONFLICT OF INTEREST

The authors declare no potential conflict of interests.