A personalized osteoarthritic joint-on-a-chip as a screening platform for biological treatments

Abstract

Osteoarthritis (OA) is a highly disabling pathology, characterized by synovial inflammation and cartilage degeneration. Orthobiologics have shown promising results in OA treatment thanks to their ability to influence articular cells and modulate the inflammatory OA environment. Considering their complex mechanism of action, the development of reliable and relevant joint models appears as crucial to select the best orthobiologics for each patient. The aim of this study was to establish a microfluidic OA model to test therapies in a personalized human setting. The joint-on-a-chip model included cartilage and synovial compartments, containing hydrogel-embedded chondrocytes and synovial fibroblasts, separated by a channel for synovial fluid. For the cartilage compartment, a Hyaluronic Acid-based matrix was selected to preserve chondrocyte phenotype. Adding OA synovial fluid induced the production of inflammatory cytokines and degradative enzymes, generating an OA microenvironment. Personalized models were generated using patient-matched cells and synovial fluid to test the efficacy of mesenchymal stem cells on OA signatures. The patient-specific models allowed monitoring changes induced by cell injection, highlighting different individual responses to the treatment. Altogether, these results support the use of this joint-on-a-chip model as a prognostic tool to screen the patient-specific efficacy of orthobiologics.

Keywords: Cartilage; Chondrogenic matrix; Inflammation; Joint-on-a-chip; Mesenchymal stem cells; Microfluidics; Orthobiologics; Osteoarthritis.

Graphical abstract

Fig. 1

Design and development of the microfluidic device. (A) Schematic of an OA knee joint and application of an injective therapy based on mesenchymal stem cells, illustrating how each component is reproduced in the chip. (B) Micrograph of the device with channels dedicated to synovial and chondral compartments highlighted in yellow and red, respectively. Scale bar: 2 mm. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 2

Characterization of the synovial compartment. (A) Viability of synovial fibroblasts in fibrin hydrogel after 1, 4, and 10 days of culture (n = 3). (B,C) Expression of typical markers of synovial fibroblasts: Collagen-I (Col1) and Lubricin (both in red) in fibrin hydrogels. Nuclei are counterstained with DAPI (blue). Scale bars: 100 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3

Comparison of different hydrogels for the cartilage compartment. (A) Graph showing the hydrogel viscosity as a function of applied shear rate, (B) time sweep curves measured during hydrogel polymerization, and (C) G′ and G″ moduli of the tested hydrogels. (D) Chondrocyte viability in the hydrogels after 1, 4, and 7 days of culture (n = 3; *p < 0.05, **p < 0.01, ***p < 0.001 as compared to the same gel at day 1; §§p < 0.01, §§§: p < 0.001 as compared to fibrin at the same day). (E,F) Expression of Collagen-II (Col2) and Aggrecan (both in green) at day 4 and 10. Nuclei are stained with DAPI (blue). Scale bars: 100 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4

Evaluating the effects of OA-SynFlu on cells within the joint-on-a-chip model. (A) Viability of synovial fibroblasts and chondrocytes after 4 and 10 days of culture in the presence of Culture Medium or OA-SynFlu. Cell viability is expressed as the percentage of live cells over the total number of cells (n = 3; **p < 0.01). (B) Assessment of synovial fibroblasts and chondrocyte senescence through SA-β-Galactosidase activity assay. The result is expressed in terms of percentage of positive cells over total cells (n = 4; *p < 0.05; ***p < 0.001). (C,D) Expression of tissue-specific markers in synovial fibroblasts (Col1 and Lubricin in red, nuclei in blue) and articular chondrocytes (Col2 and Aggrecan in green and Col1 in red, nuclei in blue) cultured in the presence of Culture Medium or OA-SynFlu in the chip. Scale bars: 100 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 5

Verifying the induction of a pro-inflammatory microenvironment with OA-SynFlu in the joint-on-a-chip model.(A) Heatmap showing cytokine content in pooled OA-SynFlu at day 0 and in the supernatants collected at day 10 and pooled from 3 independent chips cultured with Culture Medium or OA-SynFlu. (B) Representative immunofluorescence images of IL8 (red) and TNFα (green) in synovial fibroblasts and articular chondrocytes. Nuclei are stained with DAPI (blue). Scale bars: 100 μm. (C) Quantification of IL8 and TNFα signal in the presence of Culture Medium and OA-SynFlu (n = 3; ***p < 0.001). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 6

Assessment of the pro-degradative microenvironment in presence of OA-SynFlu in the microfluidic model. (A) Quantification of MMP1 and MMP13 signal (n = 3; *p < 0.05, **p < 0.01). (B) Expression of matrix metalloproteinases (MMP1 in red and MMP13 in green) in the presence of Culture Medium or OA-SynFlu in synovial fibroblasts and articular chondrocytes. Scale bars: 100 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Fig. 7

Patient-specific OA joint-on-a-chip models as screening platforms for biological treatments. (A) Cytokine quantification in OA-SynFlu from Patients 1–3. (B) Fold increase for each cytokine measured in patient-specific models in the presence of BMSCs (OA-SynFlu + BMSCs) and ASCs (OA-SynFlu + ASCs) normalized with respect to the control condition (OA-SynFlu) set at 1. For each condition, supernatants were pooled from 3 independent chips. White crosses in the heatmap indicate that one of the values used to calculate the fold increase fell below the sensitivity range of the assay. In this case, to estimate the fold increase, we used a value that was half of the lower standard curve limit. (C) Quantification of MMP1 and MMP13 in patient-specific joint-on-a-chip models, non-treated or treated with BMSCs or ASCs (n = 3; *p < 0.05, **p < 0.01, ***p < 0.001). (D) Representative images of MMP1 (red) and MMP13 (green) for Patient 1. Nuclei are stained with DAPI (blue). Scale bars: 100 μm. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

figs1

CAD design of the microfluidic chip including measures expressed in mm (A) and stereomicroscope picture of the obtained microfluidic chips with numbers (black squares) indicating the measurements of the device features (B). The comparison between the theoretical values set in the design phase and the measurements assessed in the microfluidic chip is reported in the table (C).

figs2

Strain sweep tests carried out for HA-MA (A), MIX 1:2 (B), and MIX 1:6 (C) formulations at 25 °C, at 37 °C, and at 37 °C after UV curing for photo-crosslinkable gels to study the effects of physical (i.e. temperature-mediated) and UV-mediated cross-linking. For HA-MA the strain sweep was run only at 25°C.

figs3

Comparison of chondrocyte morphology in fibrin (A) and HA-PEGDA (B). Scale bars: 100 µm.