The Dysregulated Galectin Network Activates NF-κB to Induce Disease Markers and Matrix Degeneration in 3D Pellet Cultures of Osteoarthritic Chondrocytes

Human recombinant galectins, fluorescent galectins and rabbit polyclonal antibodies against Gal-1, -3, or -8 were prepared, purified and tested as described previously [21‐23]. For details, please refer to Supplemental File 1.

Specimens of human articular cartilage were obtained from endstage OA patients (24 female, 12 male; age range 49–84 years; Knee Society Scores: Knee Score 25–87, Functional Score 30–70; Mankin score range 6–12) during total knee replacement surgery with written informed consent and in accordance with the terms of the ethics committee of the Medical University of Vienna (EK-No. 1822/2017 and 1555/2019). Comorbidities in included patients comprised cardiovascular diseases (24 cases), obesity (15 cases), neurological pathologies (9 cases), hyperlipidemia (9 cases), nicotine abuse (6 cases), hyperuricemia (6 cases), and cancer (4 cases).

Chondrocytes were isolated from femoral condyles and tibial plateaus including OA lesions and cultured in growth medium (DMEM GlutaMAX (Gibco) supplemented with 10% fetal calf serum (Biochrom), 1% penicillin/streptomycin (Gibco), and 0.1% amphotericin B (Sigma)) in a humidified atmosphere of 5% CO₂/95% air at 37 °C. For all assays, primary chondrocytes were used without subculturing to preserve the chondrocyte phenotype. 2D chondrocyte cultures (90% confluency) were kept overnight without serum addition and exposed to galectins combined in a standard mixture based on previous experience [21]. The NF-κB pathway was inhibited using Bay 11-7082, caffeic acid phenethyl ester (CAPE) and IKK inhibitor VII (all from Merck). Concentrations of reagents and time periods of cell treatment with reagents are listed in respective figures and their legends.

5 × 10⁵ chondrocytes were seeded per 1.5 ml tubes in growth medium and centrifugated at 1000 rpm for 10 min at room temperature (RT). The pelleted cells were cultured for 2 days in growth medium, then brought into starvation medium [DMEM GlutaMAX, 1% penicillin/streptomycin mixture, 0.1% amphotericin B and 1% insulin-transferrin-selenium (Gibco)]. Following culture for 3 weeks, resulting pellets were treated with the mixture of Gal-1/-3/-8 (5 µg/ml, 1 µg/ml and 5 µg/ml, respectively) either for 48 h and for 1 week (mRNA isolation), or for 2 weeks (histological examination). An additional set of pellets was used for RT-qPCR and ELISA analyses after 2 weeks of treatment (i.e., at the time of histological evaluation), to determine the effect of CAPE. Culture medium with galectins was replaced twice a week. The pellet size was analyzed with a Nikon ECLIPSE TE2000-U microscope (×2 magnification) and NIS-Elements software (Version 4.20.03). Macroscopic pictures were taken with the ZEISS Lumar V12 equipment (×24 magnification). GAGs released into the pellet supernatant were analyzed using the DMMB assay (Supplemental File 1).

Cartilage preparation, assessment of degree of degeneration using the Mankin score (MS), and immunohistochemical galectin stainings were performed as previously described [18, 21]. For details, please refer to Supplemental File 1. Pellets were processed accordingly, followed by HE, SO, and DMMB stainings, as well as immunohistochemical stainings using anti-collagen type II (mouse; Acris), anti-collagen type I (mouse; Novus Biologicals), anti-MMP-13 (mouse; R&D Systems), anti-aggrecan (mouse; Santa Cruz Biotechnology) antibodies and the set of non-cross-reactive anti-galectin antibodies. Immunohistochemical stainings were quantified using the Image J software with the plugin “Color Deconvolution” program. Of each pellet, five random pictures were recorded and quantified by splitting RGB images into defined codes (Supplemental File 1). Equal threshold settings of signal intensities were used, and the measured signal intensity was related to the average number of nuclei per image.

Isolation of total RNA, cDNA synthesis and SYBR-green-based qPCR experiments were performed as previously described [18, 21]. Levels were calculated as relative quantities compared to controls considering amplification efficiencies and normalization to succinate dehydrogenase complex, subunit A (SDHA). A detailed checklist according to the MIQE guidelines is provided in Supplemental File 2.

Cell suspensions of 3 × 10⁵ chondrocytes in 50 µl PBS were incubated at 4 °C for 10 min with a mixture of 2 µg/50 µl AlexaFluor647-labeled Gal-1, 4 µg/50 µl AlexaFluor488-labeled Gal-3, and 5 µg/50 µl AlexaFluor555-labeled Gal-8 in the presence or absence of cognate sugar, i.e., 0.1 M lactose. Photomicrographs were immediately taken without fixation by laser scanning microscopy at ×630 magnification (LSM700 microscope; Zeiss).

OA chondrocytes were grown in 96-well plates, exposed to the standard galectin mixture (5 µg/ml Gal-1, 1 µg/ml Gal-3, and 5 µg/ml Gal-8) fixed with methanol (-20 °C). Primary antibodies included NF-κB p65 (mouse; 1:1000; Cell Signaling) and pNF-κB p65 (Ser536; rabbit; 1:800; Cell Signaling) and secondary antibodies were donkey anti-mouse IgG IRDye 800CW (1:1000) and goat anti-rabbit IgG IRDye 680RD (1:1000). Signals were recorded using the Odyssey CLx Infrared Imaging System and Image Studio Version 5.2 (LI-COR Biosciences). For details, please refer to Supplemental File 1.

Cell culture supernatants were obtained from untreated pellets and pellets exposed to the galectin mixture in the presence or absence of CAPE by centrifugation and stored at −80 °C. ELISAs for pro-MMP-1, pro-MMP-13, total-MMP-3 (all from R&D Systems), Gal-1 (R&D Systems), Gal-3 (R&D Systems), and Gal-8 (Cloud-Clone Corp.) were performed with culture supernatants following the manufacturers’ protocols. ELISA standard curve ranges were 0.156–10 ng/ml (pro-MMP-1, total-MMP-3, Gal-3, Gal-8), 0.313–20 ng/ml (Gal-1), and 78–5000 pg/ml (pro-MMP-13).

Cell viability of chondrocytes and pellets was determined with the EZ4U assay (Biomedica) and the CyQUANT LDH Cytotoxicity Assay (Thermo Fisher Scientific), respectively, according to the manufacturers’ instructions. For details, please refer to Supplemental File 1.

All experiments were repeated independently with primary chondrocytes from different OA patients. The number of biological replicates (i.e., different patients) is given in the respective figure legends. Technical replicates of each biological replicate were averaged and the resulting mean values were used for statistics. Statistical analyses were performed using SPSS 25.0. Normal distribution of the data was analyzed using the Shapiro–Wilk test. Statistical significance of normally distributed data was delineated using paired t-test or ANOVA with Dunnett’s post hoc test. For non-normally distributed data, the Wilcoxon test or Friedman test with pairwise comparison was used. p-Values < 0.05 were considered significant.

After 5 weeks of culture, chondrocyte pellets from four OA patients were histologically processed. HE-stained sections revealed cells of round morphology in lacunae, surrounded by considerable amounts of ECM (Fig. 1a). Immunohistochemical analysis revealed a pronounced staining of the ECM for collagen type II, the main collagenous component of native articular cartilage (Fig. 1b). Positivity of the ECM for collagen type I was also found (Fig. 1a). Despite some interindividual variability regarding staining intensities between the donors, collagen type II was far more abundant than collagen type I in each of the four independent cell populations, as it is expected for articular chondrocytes. In addition, the ECM was positive in 2 cases for glycosaminoglycans (SO and DMMB staining) and in 3 cases for aggrecan (Supplementary File 3). Immunohistochemistry further developed strong signals for the presence of MMP-13, a major collagenase with a key role in cartilage degeneration, in the cytoplasm of chondrocytes (Fig. 1a, bottom). This staining demonstrates the intrinsic ability of the OA-derived cells in pellets to produce this typical catabolic enzyme (Fig. 1a).

Additional characterization of the pellet cultures was performed at the level of mRNA expression (Fig. 1b). In line with the immunohistochemical findings, transcripts of COL2A1 were significantly more abundant than those of COL1A1, resulting in COL2A1/COL1A1 ratios of 61 ± 25 (n = 4 OA patients). The mRNA levels of the aggrecan-encoding gene (ACAN) as well as that of the chondrocyte differentiation factor SOX9 were detectable at levels exceeding that of the reference gene SDHA. The extent of expression of the ‘tissue inhibitors of metalloproteinases’ TIMP-1 and TIMP-2 (but not of TIMP-3) was significantly higher than that of MMP-1 and MMP-13, respectively, resulting in a TIMP-1/MMP-1 ratio of 24 ± 36 and a TIMP-2/MMP-13 ratio of 50 ± 58.

Taken together, these data demonstrate (i) the maintenance of the chondrocyte phenotype in 3D pellet cultures over several weeks of culture, (ii) the accumulation of cartilage-like ECM and (iii) the expression of catabolic enzymes, thereby suggesting the suitability of this model for the functional investigation of galectin activity on ECM degradation in vitro.

We next proceeded to examine the expression of Gal-1, -3, and -8 in the OA chondrocyte pellet model at the level of transcription and of secretion. Quantification of their mRNA levels disclosed high signal intensities, exceeding that of the reference gene SDHA (Fig. 2a). Since secretion of effector proteins is a prerequisite for triggering cell signaling via cell surface receptors, we next analyzed culture supernatants for presence of these galectins. As shown in Fig. 2b, presence of galectins in the culture medium of pellets was detected in each case. Thus, cells in the 3D culture system can export galectins after their production on free ribosomes. This initial location without common entry to the endoplasmic reticulum is visualized by the predominant cytoplasmic localization of the galectins throughout cross sections of histological specimens (Fig. 2c).

In order to unravel similarities between the 3D culture model and the physiologic setting in vivo, clinical specimens were processed in parallel under identical conditions, using serial sections for assessing galectin presence and their distribution as network. As illustrated in Supplemental File 4(A), staining profiles for the three galectins monitored simultaneously in OA cartilage (with the typical upregulation in relation to an increase in the MS) closely resembled the pattern in the sections of pellet cultures. The sizes of the Spearman correlation coefficients and of associated significance levels confirmed the expectation for a positive association among the galectins [Supplemental File 4(B) and (C)]. Pellet cultures obviously retained the typical upregulation of galectins, previously seen in clinical specimens and in 2D OA chondrocyte cultures [19‐21]. In order to establish pellet cultures as tools for studying therapeutic means in vitro to interfere with disease onset and progression, we further pursued work with pellets to answer the pertinent question whether galectins will promote typical characteristics of disease manifestation in these models. Thus, we aimed to evaluate the impact of a mixture of Gal-1, -3, and -8 (i.e., Gal-1/-3/-8) on functional ECM breakdown in vitro.

Setting an internal standard for these functional assays in pellets, we first proceeded with experiments in 2D-cultured OA chondrocytes. The obtained results demonstrated the ability of Gal-1, -3, -and -8 to simultaneously bind to OA chondrocytes when applied as mixture to mimic the in vivo condition (Fig. 3a). In addition, the data, here obtained by using 3-color fluorescence microscopy, confirmed previous results on galectin-surface counterreceptor pairing [19‐21]. Presence of the cognate sugar lactose completely abolished the binding of the three galectins, excluding a carbohydrate-independent binding process (not shown). Neither the mixture of galectins nor the presence of CAPE affected chondrocyte viability (Fig. 3b). Whereas IL-1β-specific mRNA levels were upregulated by Gal-1/-3/-8, the additional presence of lactose alleviated this induction (Fig. 3c), highlighting glycan binding of the galectins as a prerequisite for triggering an effect. Lactose (without Gal-1/-3/-8) did not significantly modulate IL1B mRNA levels in this experimental setting (not shown). Intracellularly, the increase of phosphorylated p65 as sensor of galectin-dependent NF-κB activation underscored the activity of the galectin mixture at this point (Fig. 3d and e), again confirming and extending previous observations made in assays with separately tested galectins [19‐21]. The involvement of NF-κB signaling prompted us to select an appropriate inhibitor in the attempt to attenuate the expression of functional disease markers. Among the three compounds tested, CAPE actively reduced the mRNA levels of IL1B (Fig. 3f) and MMP-13 (Fig. 3g), which were selected as representative indicators for the disease-associated inflammatory and degradative processes.

With this experience in hand, we proceeded to expose pellet cultures to the galectin mixture in the absence or presence of CAPE. Shown exemplarily as photomicrographs (Fig. 4a) and as dot plot after quantitative analysis (Fig. 4b), the galectin mixture was effective to reduce the pellet size during the course of treatment, while the presence of CAPE ameliorated this status. HE-stained histological sections of the treated pellets showed no obvious signs of necrosis or apoptosis that might explain the reduction of pellet sizes after treatment with Gal-1/-3/-8 (Fig. 4c). Additional analyses of LDH activity in supernatants of pellets support the absence of cytotoxic effects of a Gal-1/-3/-8 treatment in pellet cultures (not shown). As reference, untreated control pellets were characterized by a small number of chondrocytes per field of view and highly abundant ECM that was strongly immunopositive for collagen type II (Fig. 4c). Treatment with the galectin mixture, however, resulted in a significantly increased cell density per field of view (Fig. 4d), accompanied by a significantly reduced intensity of staining of the ECM for collagen type II (Fig. 4c and d). Of note, the additional treatment of pellets with CAPE alleviated these effects of Gal-1/-3/-8, resulting in cell density and collagen type II-staining that was comparable to untreated pellets (p > 0.05; Fig. 4c and d). Immunohistochemical staining for MMP-13 resulted in an apparently increased staining of chondrocytes in the Gal-1/-3/-8-treated pellets per field of view (Fig. 4c). Since this staining concerned the chondrocytes and not the ECM, normalization to the number of cells was mandatory. When done, the obtained data made clear that neither Gal-1/-3/-8 nor CAPE had an effect on the presence of MMP-13 in the chondrocytes within the pellets (Fig. 4d). In those cases where the pellet ECM was positive for GAGs and aggrecan, treatment with Gal-1/-3/-8 reduced this positivity [Supplementary File 6(A)]. Analysis of pellet supernatants of 13 OA patients indicated levels of GAGs between 8 and 53 µg/ml, which were significantly reduced by Gal-1/-3/-8 [p < 0.05; Supplementary File 6(B)]. Addition of CAPE was not able to recover this GAG loss (p > 0.05; n = 9; not shown).

To identify the molecular triggers that are responsible for the observed ECM breakdown induced by the galectin mixture, RT-qPCR analyses and ELISAs were performed in chondrocyte pellets that were treated with the galectin mixture for 2 or 7 days. Figure 5a shows that the mRNA levels of MMP-1, MMP-3, and MMP-13 were significantly upregulated at both time points, whereas transcripts of COL2A1 and ACAN were markedly downregulated. The COL1A1 mRNA level, in contrast, was not significantly affected. At the end of the culture period (i.e., after 2 weeks of treatment, when histology was performed), protein levels of secreted MMPs were strongly induced by Gal-1/-3/-8, i.e., from 3.4 ± 1.5 to 68.4 ± 33.2 ng/ml pro-MMP-1 (p > 0.05), from 0.3 ± 0.2 to 4.0 ± 0.8 µg/ml total-MMP-3 (p < 0.05), and from 0.3 ± 0.2 to 2.9 ± 3.0 ng/ml pro-MMP-13 (p = 0.81). The presence of CAPE evoked the significant reduction to basal levels in case of pro-MMP-1 (1.0 ± 2.1 ng/ml) and pro-MMP-13 (0.8 ± 1.4 ng/ml). Albeit not statistically significant, Gal-1/-3/-8-induced total-MMP-3 levels were alleviated by CAPE in all samples (0.9 ± 0.4 µg/ml).