The benefit of combining curcumin, bromelain and harpagophytum to reduce inflammation in osteoarthritic synovial cells

Background Osteoarthritis (OA) is the most common form of arthritis, affecting millions of people worldwide and characterised by joint pain and inflammation. It is a complex disease involving inflammatory factors and affecting the whole joint, including the synovial membrane. Since drug combination is widely used to treat chronic inflammatory diseases, a similar strategy of designing plant-derived natural products to reduce inflammation in OA joints may be of interest. In this study, we characterised the response of OA synovial cells to lipopolysaccharide (LPS) and investigated the biological action of the combination of curcumin, bromelain and harpagophytum in this original in vitro model of osteoarthritis. Methods Firstly, human synovial cells from OA patients were stimulated with LPS and proteomic analysis was performed. Bioinformatics analyses were performed using Cytoscape App and SkeletalVis databases. Additionally, cells were treated with curcumin, bromelain and harpagophytum alone or with the three vegetal compounds together. The gene expression involved in inflammation, pain or catabolism was determined by RT-PCR. The release of the encoded proteins by these genes and of prostaglandin E2 (PGE2) were also assayed by ELISA. Results Proteomic analysis demonstrated that LPS induces the expression of numerous proteins involved in the OA process in human OA synovial cells. In particular, it stimulates inflammation through the production of pro-inflammatory cytokines (Interleukin-6, IL-6), catabolism through an increase of metalloproteases (MMP-1, MMP-3, MMP-13), and the production of pain-mediating neurotrophins (Nerve Growth Factor, NGF). These increases were observed in terms of mRNA levels and protein release. LPS also increases the amount of PGE2, another inflammation and pain mediator. At the doses tested, vegetal extracts had little effect: only curcumin slightly counteracted the effects of LPS on NGF and MMP-13 mRNA, and PGE2, IL-6 and MMP-13 release. In contrast, the combination of curcumin with bromelain and harpagophytum reversed lots of effects of LPS in human OA synovial cells. It significantly reduced the gene expression and/or the release of proteins involved in catabolism (MMP-3 and -13), inflammation (IL-6) and pain (PGE2 and NGF). Conclusion We have shown that the stimulation of human OA synovial cells with LPS can induce protein changes similar to inflamed OA synovial tissues. In addition, using this model, we demonstrated that the combination of three vegetal compounds, namely curcumin, bromelain and harpagophytum, have anti-inflammatory and anti-catabolic effects in synovial cells and may thus reduce OA progression and related pain. Supplementary Information The online version contains supplementary material available at 10.1186/s12906-021-03435-7.


Introduction
Osteoarthritis (OA) is a debilitating and painful disease characterised by inflammation of the synovial membrane and the progressive destruction of articular cartilage [1,2]. It is one of the top 10 causes of physical disability [3]. However, its aetiology and pathogenesis are still not fully understood. Long considered a simple degenerative cartilage disease, OA is now described as a global joint disease [4]. To date, no treatment has been able to reverse OA progression.
Although OA is not classified as an inflammatory disease, many reports suggest that inflammation could be a major driver of OA development. In fact, elevated joint inflammation has been correlated with progression of the disease [5]. Therefore, although OA pathogenesis remains unclear, inflammation is widely regarded as an extremely important factor for the progression of this disease [2,[6][7][8] and pain severity [9][10][11].
Synovitis, i.e. the inflammation of synovial tissues, is common in OA [12] and is mediated, in part, by fibroblast-like synoviocytes (FLS). These cells play an important role in OA inflammation and joint destruction, primarily by secreting a wide range of proinflammatory mediators, such as IL-6 and prostaglandin E2 (PGE2) [12], which leads the release of neurotrophins such as NGF, leading to pain during OA. They also secrete various type of proteases, including matrix metalloproteinases (MMPs) and the A Disintegrin and Metalloproteinase with Thrombospondin Motifs family (or enzymes) (ADAMTS) [13], thus promoting the degradation of extracellular cell matrix (ECM) and further aggravating the progression of OA. Therefore, alleviating synovial inflammation may prevent the onset or minimise the progression of OA and symptoms [2,[14][15][16]. Conventional anti-inflammatory drugs are nonsteroidal anti-inflammatory drugs (NSAIDs) [17]. However, these entail several side effects and drug interactions, including the risk of gastrointestinal, cardiovascular and kidney problems. The use of natural compounds may be a relevant alternative.
Herbal medicine has been used since ancient times for healing purposes and is still used today. Curcumin (CUR), which is extracted from the rhizome of Curcuma longa L., is one of the most ancient medicinal herbs and is widely used in human health due to its various therapeutic features, such as anti-inflammatory, antioxidant, anticancer and antimicrobial effects [18]. In patients with OA, oral administration of curcumin improves the clinical manifestation of the disease [19][20][21][22], improves quality of life and enables a decrease in the consumption of NSAIDs [23]. This beneficial effect of curcumin is associated with its ability to reduce OA inflammation in cells, animal models and human studies [24,25]. The action of curcumin may be reinforced by combining with other natural compounds [18,26].
The purpose of the study was to investigate the effects of the combination of curcumin (CUR), bromelain (BRO)a food obtained from pineapple which has analgesic properties [27] -and harpagophytum (HAR) -a traditional remedy for articular diseases [28] -on inflammation in an original in vitro model of osteoarthritis, using human synovial cells treated with lipopolysaccharide (LPS).

Culture cells and treatments
Human synoviocytes were recovered from the synovial membrane of six patients undergoing hip replacement surgery (mean age = 75 years). The cells were released by enzymatic digestion of the synovial membrane with collagenase type I (2 mg/ml, 12 h; ThermoFisher, Waltham, USA). The cells were cultured in Dulbecco's Modified Eagle Medium high glucose with glutamine and sodium pyruvate (DMEM, Dutscher), supplemented with 10% Foetal Bovine Serum (FBS, Dutscher) and penicillinstreptomycin (Lonza), then incubated at 37 °C in a humid atmosphere containing 5% CO 2 .
vegetal compounds, namely curcumin, bromelain and harpagophytum, have anti-inflammatory and anti-catabolic effects in synovial cells and may thus reduce OA progression and related pain.
To achieve the desired number of cells, passages were performed. The cells were rinsed with DPBS, then detached with 0.05% trypsin (ThermoFisher). The cells were recovered in a culture medium and seeded at approximately 7500 cells/cm 2 . The absence of mycoplasmas was checked by PCR.
The cells were processed at the confluence stage. Treatments were diluted in a new culture medium to the desired concentration. Each molecule was tested alone or in the presence of LPS. The three molecules were also tested together in order to see the effects of the combination of these three extracts, in the presence and absence of LPS.

Proteomic experiment
Five μg of each protein extract was prepared using a modified Gel-aided Sample Preparation protocol [30]. Samples were digested with trypsin/Lys-C overnight at 37 °C. For nano-LC fragmentation, protein or peptide samples were first desalted and concentrated onto a μC18 Omix (Agilent) before analysis.
The chromatography step was performed on a nanoElute (Bruker Daltonics) ultra-high pressure nano-flow chromatography system. Approximatively 200 ng of each peptide sample was concentrated onto a C18 Pep-Map 100 (5 mm × 300 μm i.d.) precolumn (Thermo Scientific) and separated at 50 °C onto a reversed phase ReproSil column (25 cm × 75 μm i.d.) packed with 1.6 μm C18 coated porous silica beads (IonOpticks). Mobile phases consisted of 0.1% formic acid, 99.9% water (v/v) (A) and 0.1% formic acid in 99.9% ACN (v/v) (B). The nanoflow rate was set at 400 nl/min, and the gradient profile was as follows: from 2 to 15% B within 60 min, followed by an increase to 25% B within 30 min and further to 37% within 10 min, followed by a washing step at 95% B and re-equilibration.
Mass spectrometry (MS) experiments were carried out on an TIMS-TOF pro mass spectrometer (Bruker Daltonics) with a modified nano electrospray ion source (CaptiveSpray, Bruker Daltonics). The system was calibrated each week and mass precision was better than 1 ppm. A 1600 spray voltage with a capillary temperature of 180 °C was typically employed for ionising. MS spectra were acquired in the positive mode in the mass range of 100 to 1700 m/z. In the experiments described here, the mass spectrometer was operated in Parallel Accumulation Serial Fragmentation (PASEF) mode with the exclusion of single charged peptides [31]. A number of 10 PASEF MS/MS scans were performed for 1.25 s from a charge range of 2-5.
Before the post-processing, the samples were analysed using Preview software (Protein Metrics) in order to estimate the quality of the tryptic digestion and predict the post-translational modifications present. The result, below, is used for the 'bank research/identification' part. The fragmentation pattern was used to determine the sequence of the peptide. Database searching was performed using the Peaks X+ software. A UniProt Homo sapiens database (October 2020) was used. The variable modifications allowed were as follows: N-terminal acetylation, methionine oxidation, Deamidation (NQ), Methylation (KR) and Carbamylation. In addition, C-Propionamide was set as the fixed modification. 'Trypsin' was selected as Specific. Mass accuracy was set to 30 ppm and 0.05 Da for the MS and MS/MS modes respectively. Data were filtered according to a false discovery rate (FDR) of 0.5% and protein redundancy was eliminated on the basis of proteins being evidenced by the same set or subset of peptides.

Identification of differentially expressed proteins
To quantify the relative levels of protein abundance between different groups, samples were analysed using the label-free quantification feature of PEAKS X+ software [32]. Feature detection was separately performed on each sample by the expectation-maximisation algorithm. The features of the same peptide from all replicates of each sample were aligned through the retention time alignment algorithms. Mass error tolerance was set at 30 ppm, Ion Mobility Tolerance (1/k0) at 0.07 and retention time tolerance at 10 min. Normalisation factors of the samples were obtained by the total ion current (TIC) of each sample. Quantification of the protein abundance level was calculated using the sum area of the top three unique peptides. A 1.5-fold increase in relative abundance and a significance of ≥5 using ANOVA as the significance method were used to determine those enriched proteins.

Enrichment analysis and comparison with datasets related to skeletal biology
The heatmap technique was performed with a Spearman clustering method using the ComplexHeatmap R package. Enrichments in the molecular processes, cellular processes and pathways (KEGG and Reactome) were performed using the ClueGo App from the Cytoscape software. Network specificity was set to medium; the GO tree interval was set between 2 and 4. Clusters were performed using a selection set to a minimum of three genes in addition to 4% of genes. Enrichments were performed using a Bonferroni step-down method.
Additionally, differentially expressed proteins were compared to existing gene expression datasets related to skeletal biology using the SkeletalVis application (http:// skele talvis. ncl. ac. uk/ skele tal/, [33]). Proteins encoded by genes associated with osteoarthritis joint damage in animals were also identified using OATargets databases [34].

RNA extraction and RT-PCR
RNA was extracted from the cell layer using the RNeasy mini kit (Qiagen, Hilden, Germany) in accordance with the supplier's protocol. Then, DNase treatment and reverse transcription were carried out using the DNase I kit (Sigma Aldrich) and the reverse transcriptase M-MLV (Invitrogen, Carlsbad, USA) as previously described [35]. Next, cDNA was amplified by real-time PCR using a PCR master Mix (Power SYBR Green, Applied biosystems, Courtaboeuf, France) and read on the Step One Plus Real 5′-TGG CCC AGG AGG AAA AGC -3′. The relative mRNA level was calculated using the 2 −ΔΔCT method. RPL13a was used as the invariant housekeeping gene. The decision to opt for this gene was based on our previous experience in the field [29,35,36].

Elisa
PGE2 and MMP released into conditioned media were quantified using a commercially available enzyme immunoassay kit (R&D Biosystem) as previously described [29]. For IL-6, we proceeded in the same way but using the Human beta-NGF ELISA Kit and the Human IL-6 ELISA kit (Sigma Aldrich). The immunoassays were all carried out in accordance with the manufacturer's protocol. Absorbance was determined at 450 nm with a wavelength correction set at 540 nm using the Multiskan GO spectrophotometer (Thermo Scientific).

Statistical analyses
All results are expressed as the mean value of three or four patients (biological replicates) + the standard error of the mean (SEM). Statistical analyses were carried out on the GraphPad prism 8 software. After checking the normal distribution of samples, two-way ANOVA tests were used for multiple comparisons. In significant cases, Tukey's multiple comparisons test for matched samples was performed as a post-hoc analysis. P-values < 0.05 were considered significant.

Stimulation of human OA synovial cells with LPS, an efficient OA model in vitro
Lipopolysaccharide (LPS) has recently been considered a stimulus which is able to trigger inflammation and OA onset, and is used to model the inflammatory component of OA. Therefore, we planned to test the effects of curcumin, harpagophytum and bromelain in human OA synovial cells stimulated with LPS. Before doing so, we wanted to validate the model and its ability to model OA inflammation. So, we conducted a proteomic analysis to define the differentially expressed proteins from among unstimulated OA synovial cells and LPS-stimulated OA synovial cells. Two thousand nine hundred seventeen proteins were identified in the control group, and 3011 in the LPS treated-group. Among them, 106 proteins were differentially expressed between the two groups (Peaks Sign > 5, Fold-change > 1.5, Fig. 1 and Table 1). More precisely, 66 proteins (i.e. 62%) were significantly downregulated by LPS, and 40 (i.e. 38%) were upregulated by LPS. ClueGo analysis revealed that these differentially expressed proteins are mainly involved in the biological processes of oxidative stress-induced cell death (45%, p-value < 0.01) and in the molecular processes of intramolecular oxidoreductase activity (25%, p-value < 0.01) and collagen binding (12.5%, p-value < 0.01) (Fig. 2, Tables 2 and 3). Furthermore, KEGG pathway enrichment (Fig. 3A, Table 4) showed the presence of proteins involved in protein digestion and absorption, fructose and mannose metabolism, and antigen processing and presentation (33% for each, p-value < 0.01). Enrichment using Reactome (Fig. 3B, Table 5) also showed the presence of proteins involved in the assembly of collagen fibrils and other multimeric structures (24%, p-value < 0.05).
The signature comparison of the proteomic profiles of the control group and the LPS-stimulated synovial cells using the SkeletalVis database, which allowed us to explore skeletal biology-related expression datasets [33], suggested that deregulated proteins were encoded by genes which are also differentially expressed in several other OA models (suppl. Data 1), namely 'Synovial cells from inflammatory and normal areas of osteoarthritis synovial membrane' (signed Jaccard index (sig) = 0.015; z score = 5.08) and observed in 'Rat model of surgically induced knee osteoarthritis' (signed         Jaccard (sig) = 0.0118; z score = 3.98). In addition, using OATargets databases [34], we were able to observe that several identified proteins were encoded by genes associated with OA, such as Thrombpondin-1 (THBS1), collagen alpha-1(VI) chain (COL6A1), superoxide dismutase [Mn] mitochondrial (SOD2) and Nicotinamide phosphoribosyltransferase (NAMPT) ( Table 1). In addition, about half of these genes were also found at least once as a human OA DEG, and around 90% are known to interact with OA genes (Table 1). Altogether, this proteomic analysis clearly confirms that LPS-stimulated synovial cells from OA human patients are a good model for studying the osteoarthritis process in vitro.

LPS increases the expression of genes associated with inflammation, catabolism and pain
Next, using a commonly targeted strategy, we investigated the effect of LPS treatment in human OA synovial cells. After 24 h of treatment, LPS stimulated inflammation through the production of pro-inflammatory cytokines (Interleukin-6, IL-6), catabolism through an increase of metalloproteases (MMP-1, MMP-3, MMP-13), and the production of pain-mediating neurotrophin (Nerve Growth Factor, NGF). These increases were observed in terms of mRNA levels and protein release. LPS also increased the amount of PGE2, another pain mediator (Fig. 4).

The combination of curcumin with bromelain and harpagophytum significantly reduced the LPS-induced expression of genes associated with catabolism
Having validated our model, we continued by studying the effect of vegetal extracts (curcumin bromelain and harpagophytum) on OA-associated genes. On the doses tested, vegetal extracts had little effect on the expression of catabolic genes. Only curcumin slightly counteracted the effects of LPS on MMP-13 mRNA and protein release. However, the combination of curcumin with bromelain and harpagophytum reversed the effects of LPS on the mRNA levels of MMP-1, MMP-3 and MMP-13, and on the release of MMP-3 and MMP-13 proteins (Fig. 5). These data suggested that the combination of curcumin, bromelain and harpagophytum may reduce cartilage degradation during the OA process.

The combination of curcumin with bromelain and harpagophytum significantly reduced the LPS-induced expression of genes associated with inflammation and pain
Next, we investigated the effect of these vegetal compounds on the expression of genes involved in inflammation and pain (Fig. 6). We observed that only curcumin was able to slightly reduce the LPS-induced expression of NGF and the release of PGE2 and IL-6. Interestingly, the combination of all three vegetal compounds (curcumin, bromelain and harpagophytum) significantly reduced the gene expression of IL-6 and NGF mRNA expression. It also decreased the IL-6 release and the production of PGE2. This suggests that the combination of these three compounds may reduce inflammation and pain.

Discussion
To date, no efficient treatment exists to reverse osteoarthritis. As a result, it is crucial that we identify strategies that can slow down OA progression and that are usable in the long term. Some natural compounds are known to present anti-oxidative and anti-inflammatory actions, so they may be an alternative to pharmacological drugs. In this study, after proteomic characterisation of the in vitro OA model which was used, and after confirming that it was able to induce changes in gene expression profiles similar to that observed during OA, we demonstrated that the combination of curcumin, and bromelain and harpagophytum is efficient in counteracting numerous LPS-induced effects in human OA synovial cells. Firstly, we evaluated the potential of lipopolysaccharide to induce changes in gene/protein expression by mimicking some features of OA. LPS is an endotoxin and a classical activator of the innate immune system. Because of its pathophysiological properties, LPS has been used to induce arthritis in conjunction with collagen in animal models [37,38]. More recently, researchers have started to connect LPS with the pathogenesis of OA [39]. LPS is released by gut microbiota and is correlated with the pathophysiology of osteoarthritis, in part through the activation of macrophages. In addition, local LPS administration to joints induces synovitis and is used as a model to evaluate potential treatments for acute synovitis [40]. . Since LPS is now considered a trigger for OA pathology, especially by activating synovial cells, we have proposed that stimulated human OA synovial cells may induce inflammation and reproduce in vitro some changes observed during the OA process. Using proteomics, we demonstrated here that treating human OA synovial cells with LPS induces the expression of OA signature genes, and in particular reproduces some gene expression changes observed in synovial cells from inflammatory and normal areas of the osteoarthritis synovial membrane. A more targeted strategy showed, for instance, that LPS induced the expression of MMPs, IL-6, PGE2 and NGF, which are mainly markers of catabolism, inflammation and joint pain. Consequently, the stimulation of human OA synovial cells by LPS appeared to be a good in vitro model for studying inflammation during OA. With the knowledge that alleviating inflammation may prevent the onset or minimise the progression of OA [2,14,15,39], we suggested the use of this in vitro model to test the ability of several natural substances to reduce inflammation.
Firstly, we demonstrated that curcumin has some anticatabolic and anti-inflammatory effects in human OA synovial cells. This correlates with the literature, which demonstrates that curcumin reduces MMP-3 and MMP-13 expression in rabbit chondrocytes and in the articular cartilage of oestrogen-deficient rats, preventing collagen degradation [41,42]. Also, curcumin prevents the activation of nuclear factor kappa B (NF-κB), the major mediator of inflammation [42,43]. Another study shows that curcumin favours cartilage anabolism by increasing type II collagen synthesis [25,44].  6 The combination of curcumin with bromelain and harpagophytum significantly reduced the LPS-induced expression of genes associated with inflammation and pain. Human synovial cells were treated with LPS (1 μg/ml) for 24 h in the presence of curcumin (CUR, 13 μM), bromelain (BRO, 14.7 μg/ml) and harpagophytum (HAR, 36 μg/ml), and all three together. At the end of the experiments, RNA were extracted and the media collected. Relative mRNA expression of NGF and IL-6 were determined by RT-PCR. Culture media were also collected and ELISA performed to assayed IL-6 and PGE2 release in medium. Values were compared to LPS-treated cells and presented as relative release (compared to the LPS group). n = 3. *: p-value< 0.05, **: p-value< 0.01, ***: p-value< 0.001