The following products were included in the study: Arthocare (Bonapharm S.A.C., Lima, Peru), Arthrocen (Pharmin United States, LLC, San Jose, CA, United States), Avovida (Pharma Nature, Saint Hippolyte du Fort, France), Insaponifiable 300 (GIPHAR Group, Paris, France), Persemax (Laboratorios Synthesis SAS, Bogota, Colombia), PIASCLEDINE-ExpASU^® (Laboratoires Expanscience, Courbevoie, France), and Saponic (Laboratorio Gador La Paz, Montevideo, Uruguay). All products were presented as 300 mg capsules.

For each product, the mass of 20 individual capsules was weighed and recorded. The content of the capsules was collected and subjected to Soxhlet extraction using chloroform. Soxhlet extraction is a very standard and well-recognized method that leads to the extraction of water-insoluble and slightly water-soluble organics identified as semi-volatile organic compounds (SVOCs) (Abubakar and Haque, 2020). The mass of the SVOC fraction was determined by gravimetry and recorded as the lipidic part. The contents of the lipidic part were characterized by gas chromatography analysis using a VARIAN 3400 chromatograph equipped with a septum-equipped programmable injector (SPI), a flame ionization detector (FID), and a capillary column with 5%-phenyl-95%-methylpolysiloxane as the stationary phase and helium as the mobile phase. A solution of squalene at 0.2% (m/v) in hexane and a solution of avocado oil unsaponifiables at 0.4% (m/v) and soybean oil unsaponifiables at 0.8% (m/V) in chloroform were used as standards for the identification of unsaponifiables. Quantification was performed by gas chromatography using purified alkyl furans and alkyl triols in addition to aliphatic alkane fractions from avocado, purified sterol fractions from soy, and tocopherol standards.

OA human articular cartilage was obtained from 12 different patients (five women and seven men; mean age of 62 ± 6.2 years) with knee OA at the time of total knee joint replacement surgery. All the subjects provided their informed consent, and ethical approval (ethics committee agreement of Liège University, no. B70720108313) was granted for this study.

To allow testing the product on primary human OA chondrocytes, which are one of the therapeutic targets of ASU, we used freshly isolated chondrocytes expressing inflammatory phenotypes defined by a high rate of IL-6 production (up to 40 ng/µg DNA). To obtain OA chondrocytes, the following process was used: 1. On dissection, the femoral, patellar, and tibial articular surfaces were evaluated for gross morphological modifications of cartilage according to the Collins diagram (Collins D and McElligott T, 1960). The severity of lesions was, thus, recorded for each sample. Different grades were considered: 0, normal white cartilage in all areas examined; I, the presence of a yellow–gray area with some superficial fibrillations on one or more articular surfaces; II, irregular surface with deep fibrillations on one or more articular surfaces; III, subchondral bone denudation on one or more articular surfaces, in less than 50% of the most damage compartment; and IV, subchondral bone denudation on more than 50% of the articular surface at least in one tibial plateau. All donors showed OA cartilage lesions of grade III to IV. 2. We evaluated the IL-6 levels produced by cultured chondrocytes using an immunoassay.

Full-depth articular cartilage was excised and immersed in Dulbecco’s modified Eagle’s medium (DMEM) (with phenol red and 4.5 g/L of glucose) supplemented with N-(2-hydroxyethyl)piperazine-N’-(2-ethanesulfonic acid) (HEPES) 10 mM, penicillin (100 U/ml), and streptomycin (0.1 mg/ml) (all from Biowest, Nuaillé, France). After three washes, the chondrocytes were released from the cartilage by sequential enzymatic digestions with 0.5 mg/ml of hyaluronidase type IV-S (Sigma-Aldrich, Bornem, Belgium) for 30 min at 37°C, 1 mg/ml of pronase E (Merck, Leuven, Belgium) for 1 h at 37°C, and 0.5 mg/ml clostridial collagenase IA (Sigma-Aldrich) for 16–20 h at 37°C. The enzymatically isolated cells were then suspended in alginate beads, as previously described (Sanchez et al., 2002), and maintained in culture for 12 days in DMEM supplemented with 1% ITS+ (Lonza, Verviers, Belgium), 10 mM HEPES, penicillin (100 U/ml) and streptomycin (0.1 mg/ml), 200 µg/ml of glutamine (Lonza), 50 µg/ml of ascorbic acid (Sigma-Aldrich), and 2 mM proline (Sigma-Aldrich). The cells remained in this culture medium (washout medium) for 48 h. After this washout period, the culture medium was changed every 3 days, and the collected supernatants were stored at −20°C until analysis. The chondrocytes were cultured for 12 days either without any ASU (negative control, Ctl) or with 10 µg/ml PIASCLEDINE-ExpASU^® or one of the six other nutraceutical products being evaluated in comparison, namely, Arthocare, Avovida, Arthrocen, Insaponifiable 300, Persemax, or Saponic (3 wells/condition). At the end of the culture period, the culture medium was carefully discarded, and the beads were dissolved in 1 ml of 0.1 M citrate for 10 min. The resulting suspension was centrifuged at 1,200 rpm for 10 min. With this method, two fractions were collected: the supernatant containing macromolecules originating from the further-removed matrix (FRM) and a pellet containing cells with their associated matrix (CM).

Twelve cultures have been performed. Each culture was performed with chondrocytes coming from one single cartilage specimen. Given the limited number of cells per specimen, it was not possible to test all products on the same specimen. Therefore, the products were separated into two groups. Each group was composed of 3 products and the reference product PIASCLEDINE-ExpASU^®. Each product group has been tested on six independent cultures. Each culture condition was performed in triplicates.

The DNA content of the culture was measured according to the fluorometric method using Hoechst (Labarca and Paigen, 1980).

The total AGG production corresponded to the AGG accumulated in the alginate beads. AGG at D12 and IL-6, IL-8, and MMP-3 only in the supernatant (cumulative production D3_D6_D9 and D12) were measured by specific enzyme-amplified sensitivity immunoassays (Invitrogen, Merelbeke, Belgium).

Cell viability was estimated by quantifying the release of LDH in the culture supernatant, as previously described (Mathy-Hartert et al., 2009). A sample of the supernatant or dilutions of the standard solution (LDH from rabbit muscle) was mixed with Tris buffer [10 mM Tris-HCl (pH 8.5) and, 0.1% bovine serum albumin] containing 800 mM lactate. Then, a colorimetric reagent containing 1.6 mg/ml iodonitrotetrazolium chloride (Sigma-Aldrich), 4 mg/ml nicotinamide adenine dinucleotide (Roche Diagnostics, Brussels, Belgium), and 0.4 mg/ml phenazine methosulfate (Sigma-Aldrich) was added, and the absorbance at 492 nm was read after 10 min of incubation at room temperature.

Nitrite and nitrate are stable end products of nitric oxide. Nitrate was reduced to nitrite by its addition to the supernatant of nitrate reductase (0.25 U/ml) for 20 min at 37°C. The cumulative nitrite concentrations in conditioned culture supernatants after 3, 6, 9, and 12 days of culture were determined via a spectrophotometric method based upon the Griess reaction. The absorption was measured at 540 nm. Sodium nitrite (NaNO₂) was used for calibration.

Data from in vitro experiments were analyzed and compared using a paired Student’s t-test and bilateral hypothesis followed by Bonferroni adjustment. Differences were considered significant when p < 0.05. Data are presented as histograms. The exact p values are provided, and asterisk representations were also performed. All data were analyzed by R software.

For all the products except Piascledine^®300, the lipidic part extracted by the Soxhlet device represented less than 300 mg/capsule lipidic part, while the PIASCLEDINE-ExpASU^® capsules contained 305.5 mg of the lipidic part (Table 1).

The composition of the products was compared by superimposing their chromatographic profiles (Figure 1). The purified standard of unsaponifiable fractions from avocados and soybeans was used for characterization and quantification. PIASCLEDINE-ExpASU^® showed a specific profile with the presence of chromatographic peaks corresponding to alkyl furans and alkyl triols. Subsequent quantification showed that the PIASCLEDINE-ExpASU^® product contained 65.2 ± 2.0 mg/capsule of alkyl furans and 3.0–15.0 mg/capsule of alkyl triols and aliphatic alkane fractions. All other products investigated in our study were devoid of these unsaponifiable compounds (Table 1).

Arthocare (0.1 mg/capsule), Arthrocen (1.5 mg/capsule), Insaponifiable 300 (1.8 mg/capsule), Persemax (1.8 mg/capsule), and PIASCLEDINE-ExpASU^® (73.7 mg/capsule) contained quantifiable amounts of tocopherol, while tocopherol was undetectable in Avovida and Saponic. Only PIASCLEDINE-ExpASU^® contained squalene (2 mg/capsule). The sterol content varied depending on the product: 102 mg/capsule in Saponic, 106 mg/capsule in PIASCLEDINE-ExpASU^®, 124 mg/capsule in Persemax, 126.8 mg/capsule in Arthrocen, 138.4 mg/capsule in Avovida, 154.2 mg/capsule in Insaponifiable 300, and 195.4 mg/capsule in Arthocare.

Primary OA human chondrocytes were cultured for 12 days in alginate beads with or without ASU products. Cell viability, evaluated by LDH release and DNA content, were not affected by the tested compounds (data not shown).

Under basal conditions, the NO production was 69.11 ± 20.08 nmoL/µg of DNA. As illustrated in Figure 2, Arthrocen, PIASCLEDINE-ExpASU^®, and Persemax significantly decreased NO production (Ctl vs. PIASCLEDINE-ExpASU^®: p < 0.0001; Ctl vs. Arthrocen: p = 0.01563; Ctl vs. Persemax: p = 0.00044). However, the effects of PIASCLEDINE-ExpASU^® on NO production were significantly higher than those of other compounds (p < 0.0001). In contrast, Arthrocare, Avovida, Insaponifiable 300, and Saponic had no significant effect on NO production.

Except for Persemax and Saponic, all other products either slightly or significantly increased the total AGG production by human OA chondrocytes (Ctl vs. PIASCLEDINE-ExpASU^®: p = 0.10116; Ctl vs. Arthrocen: p = 0.04979; Ctl vs. Arthocare: p = 0.00010; Ctl vs. Persemax: p = 0.00063) (Figure 3).

Except for Avovida and Saponic, all other products significantly decreased IL-6 production by human OA chondrocytes (Ctl vs. PIASCLEDINE-ExpASU^®: p < 0.0001; Ctl vs. Arthrocen: p = 0.00031; Ctl vs. Arthocare: p = 0.00244; Ctl vs. Insaponifiable 300: p = 0.00038; Ctl vs. Persemax: p = 0.00014) (Figure 4). However, the inhibitory effect of PIASCLEDINE-ExpASU^® was significantly higher than that of the other compounds (p < 0.0001).

As illustrated in Figure 5, except for Avovida and Arthocare, all other ASU products significantly decreased IL-8 production by chondrocytes (Ctl vs. PIASCLEDINE-ExpASU^®: p < 0.0001; Ctl vs. Arthrocen: p = 0.00711; Ctl vs. Insaponifiable 300: p = 0.00658; Ctl vs. Persemax: p = 0.00192; Ctl vs. Saponic: p = 0.00603). However, the inhibitory effect of PIASCLEDINE-ExpASU^® was significantly higher than that of the other compounds (p < 0.0001).

PIASCLEDINE-ExpASU^® and Insaponifiable 300 significantly decreased the basal production of MMP-3 (Ctl vs. PIASCLEDINE-ExpASU^®: p = 0.00019; Ctl vs. Insaponifiable 300: p = 0.00204) while Arthocare, Arthrocen, Avovida, and Persemax had no significant effects, and Saponic significantly increased MMP-3 production (Ctl vs. Saponic: p = 0.04103) (Figure 6).