We provide detailed methods for generating four types of tissues from human mesenchymal stem cells, which are used to recapitulate the cartilage, bone, fat pad, and synovium in the human knee joint. These four tissues are integrated into a customized bioreactor and connected through microfluidics, thus generating a knee joint-on-a-chip.
The high prevalence of debilitating joint diseases like osteoarthritis (OA) poses a high socioeconomic burden. Currently, the available drugs that target joint disorders are mostly palliative. The unmet need for effective disease-modifying OA drugs (DMOADs) has been primarily caused by the absence of appropriate models for studying the disease mechanisms and testing potential DMOADs. Herein, we describe the establishment of a miniature synovial joint-mimicking microphysiological system (miniJoint) comprising adipose, fibrous, and osteochondral tissue components derived from human mesenchymal stem cells (MSCs). To obtain the three-dimensional (3D) microtissues, MSCs were encapsulated in photocrosslinkable methacrylated gelatin before or following differentiation. The cell-laden tissue constructs were then integrated into a 3D-printed bioreactor, forming the miniJoint. Separate flows of osteogenic, fibrogenic, and adipogenic media were introduced to maintain the respective tissue phenotypes. A commonly shared stream was perfused through the cartilage, synovial, and adipose tissues to enable tissue crosstalk. This flow pattern allows the induction of perturbations in one or more of the tissue components for mechanistic studies. Furthermore, potential DMOADs can be tested via either "systemic administration" through all the medium streams or "intraarticular administration" by adding the drugs to only the shared "synovial fluid"-simulating flow. Thus, the miniJoint can serve as a versatile in vitro platform for efficiently studying disease mechanisms and testing drugs in personalized medicine.
Joint diseases like osteoarthritis (OA) are highly prevalent and debilitating and represent a leading cause of disability worldwide1. It is estimated that in the US alone, OA affects 27 million patients and occurs in 12.1% of adults aged 60 and above2. Unfortunately, most drugs currently used to manage joint diseases are palliative, and no effective disease-modifying OA drugs (DMOADs) are available3. This unmet medical need primarily stems from the absence of an effective model for studying the disease mechanisms and developing potential DMOADs. The conventional two-dimensional (2D) cell culture does not reflect the 3D nature of joint tissues, and the culture of tissue explants is often hindered by significant cell death and usually fails to replicate the dynamic tissue interconnections4. In addition, genetic and anatomical differences significantly reduce the physiological relevance of animal models4.
Organs-on-chips (OoCs), or microphysiological systems, are a promising research field at the interface of engineering, biology, and medicine. These in vitro platforms are minimal functional units that replicate defined healthy or pathological features of their in vivo counterparts5. Furthermore, these miniaturized systems can host diverse cells and matrices and simulate the biophysical and biochemical interactions between different tissues. Therefore, a microphysiological system that can faithfully recapitulate the native synovial joint promises to offer an effective platform for modeling joint diseases and developing potential DMOADs.
Human mesenchymal stem cells (MSCs) can be isolated from many tissues throughout the body and differentiated into osteogenic, chondrogenic, and adipogenic lineages6. MSCs have been successfully used to engineer various tissues, including bone, cartilage, and adipose tissue6, thus meaning they represent a promising cell source for engineering the tissue components of the knee joint. We recently developed a miniature joint-mimicking microphysiological system, named miniJoint, that comprises MSC-derived bone, cartilage, fibrous, and adipose tissues7. In particular, the novel design enables tissue crosstalk by microfluidic flow or permeation (Figure 1). Herein, we present the protocols for the fabrication of the chip components, the engineering of the tissue components, the culture of the engineered tissues in the chip, and the collection of tissues for downstream analyses.
Figure 1: Schematic of the miniJoint chip showing the arrangement of the different tissue components and medium flows. OC = osteochondral tissue. Please click here to view a larger version of this figure.
The following protocol follows the ethical guidelines of the University of Pittsburgh and the human research ethics committee of the University of Pittsburgh. Information on the materials used in this study is listed in the Table of Materials.
1. Manufacturing 3D-printed bioreactors
Figure 2: Fabrication of the different components to make the miniJoint bioreactor. (A,B), 3D models of bioreactors for creating (A) osteochondral and (B) miniJoint chips. (C,D) 3D printed (C) lids and (D) inserts with the O-ring installed. (E,F) 3D printed chambers for (E) osteochondral and (F) miniJoint tissue culture. (G,H) Assembly of (G) osteochondral and (H) miniJoint chips. Please click here to view a larger version of this figure.
2. Engineering the tissue components
NOTE: The processes for the fabrication of lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) and methacrylated gelatin (GelMA) are described in previous studies8,9.
3. Establishing the miniJoint chip
Figure 3: Assembly of the miniJoint. (A,B) Tissue-specific media are introduced from inlets 1-3 (I1-3) and moved out from outlets 1-3 (O1-3). The shared medium is perfused from I4 to O4. (C) The full setup of the miniJoint culture. Drugs (green sun like shapes) can be either introduced into (D) the shared medium only or (E) all the mediums to respectively simulate "intraarticular administration" or "systemic administration". Please click here to view a larger version of this figure.
4. Individual tissue collection
All the tissues of the miniJoint were collected to analyze their phenotypes following 28 days of culture in the miniJoint (Figure 4A). This has been detailed in our previous publication7.
Through the use of RT-qPCR, immunostaining, and histological staining, it was confirmed that the tissue-specific phenotypes were well maintained for the individual microtissues (Figure 4). For example, the osseous component of the OC microtissues (OC-O), but not the other tissue components, expressed high levels of osteocalcin (OCN). By contrast, in the cartilage portion of the osteochondral tissues (OC-C), collagen type II (COL2) and aggrecan (ACAN) were expressed at significantly higher levels than in the other tissues (Figure 4B). The expression levels of adiponectin (ADIPOQ) and leptin (LEP), two representative adipogenic genes, were much higher in the AT than in the other tissues. The deposition of calcium minerals (Figure 4F,G) as well as alkaline phosphatase (ALP) and OCN proteins (Figure 4D) was primarily seen in the OC-O, and the OC-C showed well-retained glycosaminoglycan (GAG) (Figure 4C) and COL2 (Figure 4D). In addition, the expression of two chondrocyte hypertrophy-associated markers, collagen type X (COL10) and Indian hedgehog (IHH), which were detected in the OC-C on day 28, was found to be significantly downregulated after 28 days of culture in the miniJoint (Figure 4E).
The results of RT-qPCR, immunostaining, and histology confirmed that the phenotypes of AT and SFT were successfully maintained after 4 weeks of miniJoint culture (Figure 4B,H,I). Using BODIPY staining and Oil Red O staining, we observed abundant lipid droplet deposition in the AT (Figure 4H). The immunofluorescence staining images in Figure 4I show robust lubricin and cadherin 11 (CDH11) expression by the SFT after 4 weeks of miniJoint culture.
In combination, these results indicate the creation of a functional, multi-tissue synovial joint model in the miniJoint system.
Figure 4: Maintenance of the individual tissue phenotypes of the microtissues after 4 weeks of co-culture in the miniJoint chip. (A) Timeline of generating a normal miniJoint chip with the tissue-specific phenotypes maintained. (B) RT-qPCR results showing key marker genes in all the tissue components. The OCN data were normalized to the values in the OC-O, the COL2 and ACAN data to the values in the OC-C, the ADIPOQ and LEP data to the values in the AT, and the TNC and COL1 data to the values in the SFT. The data were analyzed by one-way ANOVA (N = 3 biological replicates). * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001. (C) Safranin O staining of the biphasic OC unit. Scale bar = 1 mm. (D) Histological staining and immunostaining images confirming the presence of bone-specific and cartilage-specific markers in the corresponding components of the OC unit. Scale bar = 50 µm. (E) Immunostaining showing much lower expression of two hypertrophy markers, COLX and IHH, in the OC-C on day 56 than on day 28. Scale bar = 50 µm. (F) Alizarin Red staining of the OC unit showing the presence of calcium deposits primarily in the OC-O. Scale bar = 500 µm. (G) Magnified views of the Alizarin Red staining image in (F). Scale bar = 200 µm. (H) Oil Red O and BODIPY staining images showing the retention of lipid droplets in the AT. Scale bar = 50 µm. (I) Immunostaining images showing the expression of lubricin and CDH11 by the synovial-like fibrous tissue (SFT). Scale bar = 50 µm. Reproduced with permission from Li et al.7. Please click here to view a larger version of this figure.
To create a disease model, interleukin 1β (IL-1β) was introduced to the FM stream after 28 days of co-culture (Figure 5A,B). IL-1β treatment resulted in cell apoptosis and elevated MMP-13 levels in the SFT (Figure 5C). Interestingly, we observed cartilage degradation in the miniJoint treated by IL-1β (Figure 5D), suggesting the occurrence of crosstalk between the OC-C and SFT.
Figure 5: Modeling joint inflammation and degeneration in the miniJoint. (A) Schematic showing the induction of "synovitis" by challenging the SFT with IL-1β. (B) Timeline of establishing and analyzing the disease model. (C) TUNEL assay and MMP-13 immunostaining showing the pathological changes in the SFT. DNA fragmentation is indicated by the red arrowheads. Scale bar = 50 µm. (D) Safranin O staining and MMP-13 immunostaining images showing the degeneration of the OC-C. Scale bar = 50 µm. Reproduced with permission from Li et al.7. Please click here to view a larger version of this figure.
Lastly, we tested the therapeutic efficacy of naproxen (NPX) through "systemic administration" (Figure 6A), which was shown to reduce cartilage degradation in the IL-1β-treated miniJoint (Figure 6B). We also examined four other potential DMOADs through "intraarticular administration" (Figure 6A). The results from the real-time PCR indicated that these four compounds were able to partially reverse the cartilage loss (Figure 6C).
Figure 6: Testing drugs via "systemic" and "intraarticular" routes in the established disease model. (A) Schematic showing the two different drug administration routes, including the "systemic administration" of naproxen (NPX) and the "intraarticular administration" of fibroblast growth factor 18 (FGF18), SM04690, sclerostin (SOST), and IL-1 receptor antagonist (IL-1RN). (B) Safranin O and MMP-13 staining images showing alleviated OC-C degeneration after NPX treatment. Scale bar = 50 µm. (C) Gene expression in the OC-C after the "intraarticular administration" of four different drugs.The statistical differences between each drug-treated group and the control group are indicated by * p < 0.05; ** p < 0.01; *** p < 0.001; and **** p < 0.0001. The data were analyzed by the Student's t-test (N = 4 biological replicates). Reproduced with permission from Li et al.7. Please click here to view a larger version of this figure.
In this article, we present a protocol for creating a knee joint-on-a-chip system, in which bone, cartilage, adipose tissue, and synovium-like tissues are formed from MSCs and co-cultured within a customized bioreactor. This multi-component, human cell-derived system with plug-and-play features represents a new tool for studying the pathogenesis of joint diseases and developing drugs.
Given that different tissues favor specific culture media, it is critical to provide the respective medium for each tissue and prevent free medium exchange between flows. In particular, during the generation of biphasic osteochondral tissues, the fate of naive MSCs is determined by the medium to which they are exposed. In the current design, we use a gelatin-based hydrogel as the scaffold, which provides a template for cell growth and seals the potential gap between the tissues and the walls of the inserts. Therefore, it is critical to in situ photocrosslink the gel within the inserts. In addition, growth factors such as BMP7 and TGFβ3 are supplemented in the osteogenic medium and chondrogenic medium, respectively. To maintain their bioactivities for several days, the fresh media must be kept outside the incubator before being introduced to the chip culture. Therefore, we need to use syringes to withdraw the media from the reservoirs instead of infusing the media in the incubator.
Another critical point while handling the bioreactor is avoiding unnecessary pressure. The tissues rely on physical binding to the insert wall to stay in position. Since they are exposed to the top and bottom medium flows, if one phase of the flow has a higher pressure, it may push the tissues out of the inserts, thus resulting in leaking. Therefore, during the handling processes, such as when removing bubbles, a gentle push of the syringe is critical. If the hydrogel scaffold is pushed out, it can be put back, additional uncured hydrogel can be applied to fill the gap, and it can be cured, allowing for a secure and firm scaffold fit within the insert.
Given its proven biocompatibility, gelatin-based scaffolds are used to create all four tissues in the current system. It should be noted that gelatin may not represent the best material for supporting tissue formation. Therefore, if necessary, other types of scaffolds can be adapted for use. For example, one can use a porous and stiff scaffold and combine it with gelatin to further enhance MSC osteogenesis11. In this case, one needs to ensure there is no free medium change between the top and bottom flows. As discussed above, one can use biocompatible hydrogels to seal the potential leaking points. In addition, MSCs are used in the current tissue chip. Given their demonstrated differentiation potential into musculoskeletal tissues, induced pluripotent stem cells (iPSCs) can also be used in the future to replace MSCs. As a first step toward this investigation, we have recently used iPSCs to create osteochondral tissues12.
Synovium inflammation, or synovitis, is a key feature of OA and many other joint diseases. Furthermore, Atukorala et al. found that synovitis is a strong predictor of subsequent radiographic OA13. Therefore, we induced SFT inflammation by IL-1β treatment to generate OA-like features in the miniJoint. However, we are aware that this disease induction method cannot capture all the facets of OA. Thus, in our future research, we will explore alternative approaches to modeling OA in the miniJoint by, for example, using hyperphysiological loading to induce mechanical injury of the cartilage component14. To apply mechanical loading, the adaptation of the miniJoint design will be necessary. For example, the bottom of the miniJoint chip can potentially be modified to make the cartilage tissue accessible to the impactor tip of a customized spring-loaded impact device developed in our lab15.
To the best of our knowledge, the knee joint-on-a-chip described here is the first in vitro model that includes multiple tissues within one system to simulate a synovial joint. The novel plug-and-play capacity allows for investigating the role of a single tissue in disease progression and its response to various treatments. This system can also be employed to model joint diseases other than OA. For example, bacteria and other pathogens can be possibly introduced into the SM to model septic arthritis16. In addition, the design enables real-time crosstalk between the tissues, thus overcoming the limitation of using the conditioned medium. Specifically, adipose, synovial, and cartilage tissues can communicate through the shared medium, and bone and cartilage can interact through direct physical binding. However, there are some limitations to the current miniJoint. First, stem cell-derived tissues are used, and whether their phenotype and function resemble their counterparts in the native knee joint needs further investigation. Second, immune cells such as macrophages, which play a critical role in OA pathogenesis, are not included. Our previous study has demonstrated the feasibility of including macrophages in gelatin scaffolds17. Lastly, a physical stress-enabled mechanism is not included to simulate the mechanical loading on the tissues in the native knee joint. Recently, Occhetta et al. developed a cartilage-on-a-chip model, in which strain-controlled compression was applied to stimulate the engineered cartilage tissue14. A similar method can be adopted to enable mechanical loading in the miniJoint.
In summary, the miniJoint can serve as a unique platform to investigate the pathogenesis of OA and related conditions in vitro and provide a mechanism for exploring potential DMOADs and interventions for personalized medicine. The miniJoint system can also be integrated with OoCs mimicking other organs to establish body-on-a-chip systems that can be used to study the interactions between various organ mimics.
The authors have nothing to disclose.
This research was primarily supported by funding from the National Institutes of Health (UG3/UH3TR002136, UG3/UH3TR003090). In addition, we thank Dr. Paul Manner (University of Washington) for providing the human tissue samples and Dr. Jian Tan for their help in isolating the MSCs and creating the cell pool.
3-isobutyl-1-methylxanthine | Sigma -Aldrich | I17018-1G | |
6 well non-tissue culture plate | Corning Falcon® Plates | 351146 | |
24 well non-tissue culture plate | Corning Falcon® Plates | 351147 | |
30 mL syringes | BD Syringe Luer Lock Cascade Health | 302832 | |
Alcian blue stain | EK Industries | 1198 | 1% w/v, pH 1.0 |
Advanced DMEM | Gibco | 12491-015 | |
αMEM | Gibco | 12571-063 | |
Antibiotic-antimycotic | Gibco | 15240-062 | |
Biopsy punch | Integra Miltex | 12-460-407 | |
BODIPY® fluorophore | Molecular Probes | ||
Bone morphogenic protein 7 (BMP7) | Peprotech | ||
Curved forceps | Fisher Brand | 16100110 | |
DMEM | Gibco | 11995-065 | Dulbecco’s Modified Eagle Medium |
Dexmethasome | Sigma -Aldrich | 02-05-2002 | |
E-Shell 450 photopolymer in | EnvisionTec | RES-01-4022 | |
Fetal Bovine Serum | Gemini-Bio Products | 900-208 | |
GlutaMAX | Gibco | 3505-061 | |
gelatin from bovine skin | Hyclone | 1003372809 | |
Hank’s Balanced Salt Solution | Sigma -Aldrich | SH30588.02 | |
indomethacin | Sigma -Aldrich | I7378-56 | |
Insulin-Transferrin-Selenium-Ethanolamine (ITS) | Gibco | 51500-056 | |
interleukin 1β | Peprotech | 200-01B | |
Leur-loc connectors | Cole-Parmer Instruments | 45508-50 | |
L-proline | Sigma -Aldrich | 115388-93-7 | |
β-glycerophosphate | Sigma -Aldrich | 1003129352 | |
Medium bags | KiYATEC | FC045 | |
Methacrylic Anhydride | Sigma -Aldrich | 102378580 | |
Phosphate buffered Saline | Corning | 21-040-CM | |
Pointed forceps | Fisher Brand | 12000122 | |
Silicon mold | McMaster-Carr | RC00114P | |
Silicon o-rings | McMaster-Carr | ZMCCs1X5 | 1mm x 5mm |
SolidWorks | Dassault Systèmes SE, Vélizy-Villacoublay, France | ||
Surgical Blades | Integra Miltex | 4-122 | |
Syringe pump | Lagato210P, KD Scientific | Z569631 | 10 syringe racks |
T-182 tissue culture flasks | Fisher Brand | FB012939 | |
Tissue Culture Dish 150 mm | Fisher Brand | FB012925 | |
Transforming Growth Factor Beta (TGF-β3) | Peprotech | 100-36E | |
Trypsin | Gibco | 25200-056 | |
UV Flashlight | KBS | KB70109 | 395 nm |
Vida Desktop 3D Printer | EnvisionTec | ||
Vitamin D3 | Sigma -Aldrich | 32222-06-3 | 1,25-dihydroxyvitamin D3 |