Tissue Engineering for Musculoskeletal Regeneration and Disease Modeling

Abstract

Musculoskeletal injuries and associated conditions are the leading cause of physical disability worldwide. The concept of tissue engineering has opened up novel approaches to repair musculoskeletal defects in a fast and/or efficient manner. Biomaterials, cells, and signaling molecules constitute the tissue engineering triad. In the past 40 years, significant progress has been made in developing and optimizing all three components, but only a very limited number of technologies have been successfully translated into clinical applications. A major limiting factor of this barrier to translation is the insufficiency of two-dimensional cell cultures and traditional animal models in informing the safety and efficacy of in-human applications. In recent years, microphysiological systems, often referred to as organ or tissue chips, generated according to tissue engineering principles, have been proposed as the next-generation drug testing models. This chapter aims to first review the current tissue engineering-based approaches that are being applied to fabricate and develop the individual critical elements involved in musculoskeletal organ/tissue chips. We next highlight the general strategy of generating musculoskeletal tissue chips and their potential in future regenerative medicine research. Exemplary microphysiological systems mimicking musculoskeletal tissues are described. With sufficient physiological accuracy and relevance, the human cell-derived, three-dimensional, multi-tissue systems have been used to model a number of orthopedic disorders and to test new treatments. We anticipate that the novel emerging tissue chip technology will continually reshape and improve our understanding of human musculoskeletal pathophysiology, ultimately accelerating the development of advanced pharmaceutics and regenerative therapies.

1. Introduction

The human musculoskeletal system, also known as the locomotor system, provides the human body with structural support and load-bearing capacity and offers protection to the delicate internal organs. The musculoskeletal system enables mechanical functions that subject the tissues to wear and tear as well as injuries over a lifetime of use, leading to debilitating pain and weakening or even loss of functions. In fact, musculoskeletal conditions are the leading cause of physical disability worldwide (James et al. 2018). As musculoskeletal disorders are most prevalent in the elderly, the world’s aging population represent the major contributing factor to the increasing medical and socioeconomic burdens of treating musculoskeletal impairments. Conventional treatments include the use of autografts, allograft, and xenografts, but have several drawbacks such as disease transmission, limited availability and reproducibility, donor scarcity, sterilization-induced alteration in natural matrix properties, and immune rejection. Tissue engineering represents a promising strategy of regenerative medicine to restore musculoskeletal structures and functions.

The field of tissue engineering interfaces engineering, life science, and medicine, with the capability of creating living tissues outside the human body (Langer and Vacanti 1993). Traditionally, these in vitro grown tissues are intended to be implanted into the human body to restore, maintain, augment, or replace diseased, injured, or degenerated tissues. Regenerative medicine, as the term indicates, aims to enable and enhance the body’s natural repair mechanisms to restore the function of otherwise irreparable tissues or organs in situ.

The tissue engineering triad consists of three key elements: scaffolds, cells, and signaling molecules (O’Brien 2011). In the past few decades, considerable progress has been made in the tissue engineering and regenerative medicine (TERM) field, which continues to evolve rapidly. In particular, new biomaterials are being designed, and novel approaches are emerging to fabricate scaffolds with existing biomaterials; the identification of new stem cell sources and functions and the establishment of more biologically accurate organotypic models promise to broaden the applications of TERM technologies; and innovative delivery and administration strategies are being proposed for growth factors and other biologically active molecules. These exciting developments underscore the importance of comprehensive evaluation of the three TERM components to gain an in-depth understanding of their safety, efficacy, and optimal use. In this chapter we focus specifically on the essential TERM elements of relevance to orthopedic applications.

Successful translation of TERM techniques and products requires evidence-based safety and efficacy assessment prior to the clinical trial. Currently, animal models are an essential component in preclinical studies and have been used to assess various biomaterials, cells, and signaling molecules for musculoskeletal regeneration. Small rodents, especially mice and rats, are the most commonly used animal models in musculoskeletal research. The past few decades have witnessed a continuous increase in the use of mice in orthopedic research, while the use of rats remains relatively unchanged due to the difficulties in genetic manipulations (Ericsson et al. 2013). However, in recent years, concerns have risen about the benefit of animal research to humans. A number of studies have shown that most animal experiments are unable to predict the observations in human trials (Mak et al. 2014; Pound and Bracken 2014). In fact, among the 76 highly influential animal studies (each with >500 citations) published on 7 prominent scientific journals, only 37% could be replicated in human randomized trials (Hackam and Redelmeier 2006). Therefore, the benefits of animal models still need to be supported by more systematic evaluations.

In addition to the many commendable achievements in tissue repair, TERM applications have also been successfully extended to disease modeling and drug development over the past decade. Exemplary in vitro models have been established using TERM principles for mimicking musculoskeletal tissues, including bone and cartilage (Lin et al. 2014; Occhetta et al. 2019; Arrigoni et al. 2020), and skeletal muscle (Truskey 2018), and other tissue/organ systems, such as lung (Huh et al. 2010), blood-brain barrier (Phan et al. 2017), gut (Kim et al. 2012), kidney (Jang et al. 2013), liver (Ribeiro et al. 2019), pancreas (Shik Mun et al. 2019), and heart (Nunes et al. 2013). These exciting studies have suggested that in vitro organotypic models and microphysiological systems (MPS), generated with tissue engineering principles and technologies, may serve as convenient and versatile platforms for the comprehensive evaluation of multiple TERM elements. Once the clinical relevance is validated, in vitro models like MPS will be a powerful alternative to current animal models.

We begin this chapter by describing the properties and performances – both in vitro and in vivo – of the scaffolds, cells, and signaling molecules in musculoskeletal tissue engineering. The derivation of organoids from stem cell aggregates and their potential applications are introduced. We then describe the optimal use of the three TERM elements in representative MPS established thus far for modeling musculoskeletal tissues. Finally, we summarize the advantages and disadvantages of different models and envision their utility in the development of drugs/treatments in the future.

2. Key Elements of Musculoskeletal Tissue Models

Scaffolds, cells, and signaling molecules are considered the three key, enabling components of tissue engineering. For musculoskeletal regeneration, the structure and function of native bone, cartilage, skeletal muscle, tendon, and ligament tissues have significant implications on the selection of the three elements mentioned above. Best reparative and regenerative outcomes are achieved when the three elements act synergistically, where the appropriate cell types are seeded into a scaffold that possesses mechanical, structural, and biochemical characteristics akin to those of the native tissue and are given the biochemical and physical signals that induce anabolic and regenerative responses.

2.1. Biomaterials and Scaffolds

Enormous progress in biomaterials research has been made in the past decades. Many different types of natural and synthetic biomaterials have been utilized to fabricate tissue engineering scaffolds, which can be cell-laden or cell-free when implanted. The safety and efficacy of a new biomaterial have to be rigorously assessed prior to its clinical application. This evaluation process routinely includes both in vitro and in vivo tests. We recognize the myriad biomaterial types that have been utilized to regenerate musculoskeletal tissues. This section thus does not intend to provide a thorough description of existing biomaterials, but rather focuses on an analysis of the assessment and performances of the representative, novel biomaterials and scaffolds developed in the past decade.

Bone

Biomaterial preparation for bone regeneration has been facing a persistent challenge to recapitulate the mechanical, structural, and functional characteristics of native bone. Natural biomaterials such as acellular bone matrix and extracellular matrix (ECM) generated by in vitro cultured mesenchymal stem cells (MSCs) have led to robust osteogenesis (Ni et al. 2014; Liu et al. 2019; Rothrauff and Tuan 2020). A myriad of synthetic biomaterials, including metals, ceramics, polymers, and composites, have been developed to address the unmet medical need of bone grafts. Metallic biomaterials still dominate the market of load-bearing bone substitute materials (Chen and Thouas 2015). One of the most commonly used biometals is Ti-6Al-4V alloys. Human MSCs were cultured on additive manufactured Ti-6Al-4V scaffolds, and surface anodization was found to increase the alkaline phosphatase (ALP) production, osteocalcin (OCN) and type I collagen expression, and mineral deposition (Li et al. 2020a; Groessner-Schreiber and Tuan 1992). A major drawback of most metallic implants is the associated “stress shielding effect.” Biodegradable metals such as magnesium alloys have thus been proposed to address this issue. Fracture repair with magnesium-containing orthopedic implants has shown considerable efficacy. Magnesium intramedullary rods were implanted in rat femur, leading to abundant new bone formation through an osteogenic mechanism that involves calcitonin gene-related polypeptide-α (CGRP) (Zhang et al. 2016). It has been noted that there exists a large gap in mechanical properties between existing biodegradable materials and traditional metallic implants. Biodegradable zinc alloys have therefore been developed, tested in vitro with MC3T3-E1 mouse preosteoblast cells and human umbilical vein endothelial cells (HUVECs), and implanted in rat femur to evaluate their degradability (Yang et al. 2020). This group of newly designed zinc alloys was found to show excellent promise in load-bearing orthopedic applications.

Bioceramics mostly possess high chemical stability and biocompatibility; bioactive ceramics are usually osteoconductive and/or osteoinductive. Osteoconductivity is the ability of a surface to support the attachment, proliferation, and migration of bone-forming osteoblasts, while osteoinductivity implies the recruitment and induction of undifferentiated stem cells to become osteoprogenitors (Albrektsson and Johansson 2001). Hydroxyapatite and Bioglass^®, as two extensively studies bioactive ceramics, can form direct chemical bonds with surrounding tissues. Because of the close compositional resemblance of calcium phosphates (CaP, minerals containing Ca²⁺ and phosphate anions) to the inorganic phase of natural bone, CaP ceramics are among the most frequently used bone biomaterials (Bose and Tarafder 2012). The suitability of bioceramics for bone repair is typically assessed by culturing osteoblast-like cells and MSCs on bioceramic surfaces or in bioceramic-conditioned medium (Li et al. 2017a, b, 2020b). As a relatively new bioceramic, 2D nanosilicates have become a topic of increasing interest in bone tissue engineering because of their outstanding osteoconductivity and osteoinductivity. Nanosilicate platelets can be conveniently mixed in three-dimensional (3D) hydrogels to form nanocomposites. A 2D in vitro culture model has been used to assess the osteogenic responses of the MC3T3 E1 subclone 4 cells to nanosilicate-containing scaffolds (Xavier et al. 2015). The cells showed higher quantitative ALP activity and calcium deposition induced by nanosilicate addition. In another study, both in vitro and in vivo models were employed to examine silicate/methacrylated glycol chitosan scaffolds. The in vitro osteoinductivity tests showed markedly increased ALP activity and mineralization as well as upregulated osteogenic gene expression by the encapsulated MSCs in the presence of silicate. In a mouse nonunion calvarial defect model, the silicate-containing, cell-free composites were found to overtly promote native cell infiltration as well as cause remarkably higher bone volume density, bone growth surface area, and trabecular number than either constituent material (Cui et al. 2019).

A main limitation of many bioceramics is their low fracture toughness. To deal with this issue, bioceramics have been combined with biopolymers to obtain composites with improved ductility and toughness (Rezwan et al. 2006). A broader application of biopolymers, especially hydrogels, for musculoskeletal repair is seen in cell- or drug-delivery therapies (Nöth et al. 2010), as will be discussed in Sects. 2.2 and 2.3.

Cartilage

Various natural and synthetic biopolymers have been utilized to maintain or enhance the chondro-phenotype of chondrocytes or chondrogenic differentiation of stem cells, a process where new cartilage ECM is synthesized and deposited. Cartilage scaffolds come in various forms such as porous sponges (Wang et al. 2005), fibers/meshes (Li et al. 2005), and hydrogels (Deng et al. 2019). As a natural biomaterial, MSC-derived ECM was shown in our previous study to be a robust substrate for enhancing the chondrogenesis of articular chondrocytes (Yang et al. 2018). A composite hydrogel, consisting of gelatin, fibrinogen, hyaluronan (HA), and glycerol, was used to 3D print scaffolds containing rabbit articular chondrocytes (Kang et al. 2016). In vitro culture and in vivo implantation in mouse dorsal subcutaneous pockets both showed abundant new cartilage matrix deposition in the 3D printed structure. Sharma et al. (2013) designed a poly(ethylene glycol) diacrylate hydrogel and used it in conjunction with a bioadhesive to repair focal cartilage defects in a caprine model and in human patients (Fig. 1a–c). This soft hydrogel could augment the microfracture treatment and promote cartilage regeneration. Electrospun polymer fiber meshes are usually too dense for cell infiltration. To tackle this issue, cryoelectrospinning, with the mandrel kept at −78 °C, was employed to create an ultraporous nanofiber network that permits the infiltration of cell-laden hydrogels (Formica et al. 2016). The scaffolds were assessed with both bovine and human chondrocytes and led to robust production of type II collagen and sulfated glycosaminoglycans (GAGs), characteristic markers of cartilage, in vitro.

Fig. 1.

Representative cartilage regeneration strategies harnessing adhesive-hydrogel composites (a–c), MSC chondrogenesis (d–g), and growth factor-induced endogenous cell homing (h–k). In a pilot clinical study, focal cartilage defects in human patients were applied with an adhesive and microfractured (a), filled with injectable, photo-crosslinkable hydrogel (b), which trapped bleeding from the microfracture holes (c). In MSC-based cartilage tissue engineering, an optimized differentiation protocol with exposure to Wnt/β-catenin inhibitor and shorter TGF-β treatment time was found to significantly inhibit chondrocyte hypertrophy both in vitro (d, e) and in a mouse intramuscular implantation model (f, g). Scale bars: 100 μm. A 3D printed scaffold (h) infused with collagen hydrogel and with/without TGF-β3 loading was used to replace a rabbit proximal humeral joint. After 4 months’ implantation, TGFβ3-infused scaffolds showed full articular surface coverage by newly formed cartilage (i) similar to the native tissue (k), while TGFβ3-free scaffolds had only isolated tissue formation (j). Reproduced with permission from Sharma et al. (2013), Deng et al. (2019), and Lee et al. (2010)

Native cartilage undergoes substantial mechanical shear, wear, and compression during a lifetime. The recapitulation of the mechanical and frictional properties has become a key consideration in the design of cartilage biomaterials and scaffolds (Liao et al. 2013). By virtue of its high biocompatibility and safety (and with FDA approval), poly(ethylene glycol) (PEG) has been widely utilized, despite their low mechanical strength, to prepare various hydrogel scaffolds (Benoit et al. 2008; Brandl et al. 2010). An injectable hydrogel derived from 4-arm star PEG showed a local maximum strength of ~20 MPa. This high-strength PEG-based hydrogel supported the proliferation and phenotype maintenance of encapsulated murine chondrocytes, and the injected chondrocyte-laden hydrogels resulted in formation of new hyaline cartilage integrated with the host tissue in a murine osteochondral defect model (Wang et al. 2017). We recently developed a photo-crosslinkable poly-D,L-lactide acid/PEG hydrogel and assessed its chondrogenic potential with human MSCs (Sun et al. 2017). This hydrogel possesses a compressive modulus in the physiological range of native cartilage and supports the differentiation and maintenance of human MSCs and thus holds promise in point-of-care treatment of cartilage defect.

Skeletal Muscle

Generally there are two approaches to skeletal muscle tissue engineering: (1) transplantation of biomaterial scaffolds seeded with muscle cells and other supporting cells and (2) delivery of biomaterial scaffolds, with or without paracrine signaling cells, growth factors, or cytokines, to induce in situ muscle tissue engineering (Kwee and Mooney 2017). In the design of biomaterials for skeletal muscle tissue engineering, it is important to consider the microstructural, physical, and biochemical properties, such as porosity (Hill et al. 2006), degradability (Hong et al. 2011), 2D and 3D patterns (Ku et al. 2012), injectability (Rossi et al. 2011), and native biochemical cues (Perniconi et al. 2011). Electrospinning is suited to a wide variety of natural and synthetic biomaterials, capable of fabricating fibers with varying diameters and orientations and generating scalable, ECM-mimicking scaffolds with desired degree of anisotropy. An alginate-based bioink containing HUVECs has been electrospun onto uniaxially micropatterned PCL/collagen struts, generating scaffolds that provide both topographical and biochemical cues that facilitated the alignment and differentiation of subsequently seeded myocytes (Yeo and Kim 2020). Hydrogels are also commonly used biomaterials for muscle repair. Interestingly, 3D free-standing skeletal muscle fibers engineered from muscle cell-laden Matrigel were found to be able to support the differentiation of neural stem cells into neurons to form neuromuscular junctions (NMJs) (Morimoto et al. 2013).

Cell-free scaffolds used in regenerative medicine eliminate the stringent cell harvest and administration process and obviate the associated regulatory hurdles and possible immune responses. A biologic-free ferrogel, without the incorporation of any bioreagents or cells, could generate externally actuated mechanical suppressions to reduce fibrosis and inflammation and heal myotoxin-induced severe muscle injuries (Cezar et al. 2016). To regenerate volumetric muscle loss (VML), the self-regeneration capability of native muscle does not suffice. Decellularized bladder ECM was used to repair VML in both mice and human patients and resulted in de novo muscle remodeling linked to the recruitment of perivascular stem cells (Sicari et al. 2014).

Tendon and Ligament

Tendons are tough connective tissues that bind muscles to bone, while ligaments connect bones to other bones. A number of design factors, including native tissue anatomy, physical and chemical properties of the materials, and material interactions with native cells, need to be considered to select an optimal biomaterial in tendon and ligament tissue engineering (Kuo et al. 2010).

Our previous study explored the potential application of decellularized tendon-derived, solubilized extracellular matrix (tECM) in adipose stem cell (ASC)-based tendon tissue engineering (Yang et al. 2017). The tECM-supplemented 3D scaffolds not only enhanced the tenogenic differentiation of ASCs and the scaffolds’ mechanical properties but also downregulated the expression of osteogenic markers and matrix metalloproteinases. Decellularized tendon ECM combined with stem cells has also been researched for tendon repair, and the natural, decellularized scaffolds were found to provide an inductive environment for the tenogenic differentiation of MSCs (Youngstrom et al. 2013; Ning et al. 2015). Using the synthetic, degradable biopolymer poly(ε-caprolactone), Wang et al. (2018) fabricated a 3D scaffold with tendon-like mechanical properties and microstructural and hierarchical anisotropy. This scaffold supported tenogenic matrix production by human tenocytes, and the acellular scaffolds showed robust pro-tenogenic properties in a micropig model.

Ligament and tendon have very similar structure in spite of differences in collagen and water content in their ECM. Many biomaterials have, therefore, been designed to engineer both tissues and tested with similar methods (Barber et al. 2013). While tenocytes are the major cellular component of tendons, the cells residing in ligaments are mostly fibroblasts. In in vitro studies, ligament biomaterials and scaffolds are typically evaluated with fibroblasts and MSCs (Correia Pinto et al. 2017; Chang et al. 2020), and in vivo models for ligament tissue engineering are mostly created in small-scale animals.

Tendons and ligaments attach to bone at junctions known as entheses. Biomaterials for enthesis tissue engineering are expected to facilitate integration and smooth load transfer between tendon/ligament and bone. Such biomaterials have been investigated in a number of previous studies and are not detailed here (Font Tellado et al. 2015; Tang et al. 2020).

2.2. Cells

The common cell sources employed to engineer musculoskeletal tissues are shown in Fig. 2a. Because of their multipotency, high proliferative capacity, and relative ease of isolation and expansion, MSCs remain the most promising cell source in musculoskeletal regeneration and in general TERM applications. Although embryonic stem cells (ESCs) have wide differentiation potential and extensive expansion properties, for TERM applications, the hurdles include not only ethical issues related to sourcing but also the risks of tumorigenicity and immune rejection as well as the absence of a standardized protocol for ESC differentiation into musculoskeletal tissues (Jukes et al. 2010). Induced pluripotent stem cells (iPSCs) have received much attention for their pluripotency and virtually unlimited supplies. Since iPSCs can be reprogrammed from somatic cells, they not only obviate the need for embryos but can be made individual-specific, thus avoiding immune rejection issues (Takahashi and Yamanaka 2006). Primary cells from native tissues have also been employed for musculoskeletal regeneration, but they usually have lower availability, are more difficult to culture, and show larger patient-to-patient variability.

Fig. 2.

Commonly used cell sources (a) and growth factors (b) for musculoskeletal tissue engineering

Bone

Human/animal primary osteoblastic cells and stabilized osteoblastic cell lines have both been used in in vitro investigations. Osteoblastic cell lines, including SaOs-2, MG-63, and MC3T3-E1, have been compared with primary human osteoblasts (HOBs) in terms of their proliferation ability, mineralization behavior, and gene expression profile. It was found previously that HOBs only share some of the characteristics with each of the cell lines (Czekanska et al. 2014). Although the uses of HOBs are more clinically relevant than cell lines, their applications are limited by the complicated isolation procedures and heterogeneous phenotypes (Czekanska et al. 2012). As a multipotent stem cell, MSCs can be readily differentiated into osteoblasts and have been applied in various scaffolds to induce osteogenesis. Tissue-engineered bone scaffolds are designed to mimic bone ECM and recruit surrounding cells, forming a bone tissue to repair the bone defect (Amini et al. 2012). Recent studies show that MSCs possess robust osteogenic ability when seeded in collagen scaffolds, polymer-mineral scaffolds, fibrous nanocomposite scaffolds, silica-coated scaffolds, as well as their own ECM (Kuttappan et al. 2020; Duan et al. 2018; Gandhimathi et al. 2019; Harvestine et al. 2018; Meinel et al. 2003). However, autologous MSCs have limited availability. Due to patient heterogeneity, allogeneic MSCs show donor age-dependent proliferation rates and raise immunogenic concerns; MSC isolation involves invasive harvesting procedures in the case of bone marrow MSCs. iPSCs have emerged as a novel alternative to MSCs. Osteoblasts derived from iPSC were seeded on a gelatin scaffold and were found to secret calcium, OCN, and bone sialoprotein in vitro and in vivo (Bilousova et al. 2011). Retinoic acid was used to induce human iPSCs to differentiate into osteoblast-like and osteocyte-like cells, and these cells could generate human bone tissues in mouse calvarial defects (Kawai et al. 2019). Nevertheless, with current differentiation protocols, the multi-lineage differentiation capability of iPSCs generally does not match that of MSCs (Diederichs and Tuan 2014), and the optimization of differentiation protocols is warranted in future research.

Cartilage

Chondrocytes are the only cell type present in articular cartilage and exist within abundant ECM that is neither vascularized nor innervated. The presence and functional state of chondrocytes are of significant orthopedic clinical importance in the case of degenerative joint diseases, such as osteoarthritis (OA), where the articular cartilage suffers from extensive degeneration, resulting in serious physical debilitation. In order to understand OA progression, it is important to examine the changes in chondrocytes within the cartilage. There has been a relentless pursuit of an optimal chondrocyte culture method to reduce their dedifferentiation during in vitro expansion (Caron et al. 2012), and a number of chondrogenic cell lines from mouse or rat, including C1, C3H10T1/2, ATDC5, CK2, and RCJ3.1, have been generated and widely used (Brown et al. 2014). It is now accepted that abnormal biomechanical and genetic factors target chondrocytes and alter their normal functions (Goldring 2000). Dudek et al. (2016) reported that the chondrocyte clock gene BMAL1 plays a key role in the normal function of articular chondrocytes and hence in cartilage integrity. Besides chondrocytes, MSCs, iPSCs, and ESCs have also been utilized to form cartilage tissues. One of the key challenges in the long-term culture of chondrocytes, either native or derived from chondrogenically differentiated stem or progenitor cells, is the process of hypertrophy, whereby the cells enlarge and enter terminal differentiation, accompanied by apoptosis and calcification. While this differentiation and maturation of chondrocytes is a normal process of long bone development from transitional cartilage as part of the process of endochondral ossification (Saito et al. 2010), the hyaline cartilage of the articular joint surface is a permanent cartilage, and the appearance of hypertrophic chondrocytes is in fact part of the degenerative process in OA (Wang et al. 2004). Researchers have thus explored different methods to generate non-hypertrophic, hyaline cartilage. We found in our previous study that inhibiting the WNT signaling pathway during the differentiation process can promote the MSC chondrogenesis and suppress chondrogenic ossification in vitro and in vivo (Fig. 1d–g) (Deng et al. 2019; Narcisi et al. 2015). Different protocols have been proposed for iPSC chondrogenic differentiation (Dicks et al. 2020; Lach et al. 2019; Hu et al. 2020), and further research is needed for protocol optimization and standardization.

Skeletal Muscle

Regeneration of skeletal muscle is a complex process orchestrated by heterogeneous cell populations. Satellite cells/muscle-derived stem cells (MDSCs) and MSCs are among the most frequently used cell types for repairing defective muscles (Sacco et al. 2008; Montarras et al. 2005; Koponen et al. 2007; Usas and Huard 2007). Human pluripotent stem cells can be successfully differentiated into induced myogenic progenitor cells, which can readily form 3D contractile multinucleated myotubes (Rao et al. 2018). Other less commonly used cell types in muscle regeneration include adipose-derived stem cells and pericytes (Dunn et al. 2019). Satellite cells are positive for PAX7 and PAX3 (Relaix et al. 2005) and generally considered to stay dormant when there is no injury to the muscle. Interestingly, Keefe et al. (2015) demonstrated that satellite cells contribute to myofibers in both injured and healthy adult mouse muscles. Collins et al. (2005) transplanted as few as seven satellite cells in a single intact myofiber into radiation-ablated muscles, where the satellite cells vigorously self-renewed and expanded, generating clusters of new, compact myofibers. Satellite cells have also been delivered in HA hydrogels to a muscle ablation model in mouse, resulting in both structural and functional recovery (Rossi et al. 2011). Multipotent MDSCs have been employed to regenerate muscle and other musculoskeletal tissues (Usas and Huard 2007). A preplate technique was introduced to isolate MDSCs. In this procedure, the minced muscle tissues first undergo enzymatic digestion, and the resultant slurry is plated on collagen-coated flasks. Non-adhering cells in the culture medium are transferred and plated in new collagen-coated flasks, and this process is repeated ~5 times to obtain a slowly adhering cell population that contains MDSCs (Gharaibeh et al. 2008). The maintenance of MDSC potency and quiescence is challenging in in vitro culture. Quarta et al. (2016) created collagen-based artificial muscle fibers that extended the quiescence of mouse and human MDSCs. Bone marrow-derived MSCs, as another promising cell source for muscle repair, have been seeded onto 3D, porous alginate scaffolds loaded with vascular endothelial growth factor (VEGF) and insulin-like growth factor 1 (IGF1), where the local growth factor stimulation could enhance MSC paracrine signaling and support endogenous muscle regeneration in a rat muscle injury (blunt crush) model (Pumberger et al. 2016). The trophic actions of MSCs have been proposed to be a major mechanism underlying the enhanced endogenous muscle repair and regeneration (Sassoli et al. 2012).

Tendon and Ligament

Both terminally differentiated cells and stem cells have been utilized for tendon/ligament regeneration. Tenocytes are characterized by specific markers, such as Scleraxis, Mohawk, and early growth response factor (Asahara et al. 2017). To maintain tenocyte phenotype, many collagen-based scaffolds have been engineered. It was reported that collagen-GAG scaffolds were helpful for the long-term maintenance of tenocyte transcriptomic stability (Caliari et al. 2012). In addition, supplementation of growth/differentiation factor 5 (GDF5) and IGF1 together can rescue the tenocyte phenotype and drive cell proliferation (Caliari and Harley 2013). Tendon stem/progenitor cells (TSPCs) possessing the universal characteristics of stem cells were first identified by Bi et al. (2007). TSPCs reside in a biglycan- and fibromodulin-rich niche and can regenerate tendon-like tissues both in vitro and in vivo. Many studies have explored MSC-based tendon tissue engineering strategies, where MSCs were successfully induced to express tenocyte phenotypes (Kryger et al. 2007; Lee et al. 2011; Li et al. 2015; Tokunaga et al. 2015). It has also been shown that co-culture with stem cells derived from bone marrow or adipose tissue can promote tenocyte proliferation (Chen et al. 2018; Kraus et al. 2013). Recently, Komura et al. (2020) generated a tenocyte induction protocol to differentiate murine iPSCs into tenocyte-like cells, which could significantly reduce scar formation and promote tendon regeneration when implanted in injured mouse tendons. In ligament tissue engineering, ligament-derived fibroblasts (Chang et al. 2020), human dermal fibroblasts (Correia Pinto et al. 2017), and human MSCs have all been employed.

3D Organoids

ESCs, adult stem cells (including MSCs), and iPSCs have been used to engineer almost every musculoskeletal tissue (Fig. 2a), because these cells possess extensive proliferation potential and multi-differentiation potency. Aggregates of stem cells can form 3D structures that recapitulate certain architectural and functional characteristics of some native tissues in vitro, and such 3D cell structures are termed organoids (Takahashi 2019). Various ESC- and iPSC-derived organoids have been derived by employing developmental biology principles, and a number of other organoids were generated by subjecting adult stem cells to conditions that mimic tissue renewal or repair processes in vivo (Clevers 2016). For example, organoids derived from pluripotent stem cells have been studied previously to model the brain (Lancaster et al. 2013), retina (Eiraku et al. 2011), adenohypophysis (Suga et al. 2011), stomach (McCracken et al. 2014), liver (Takebe et al. 2013), lung (Dye et al. 2015), and kidney (Takasato et al. 2014); adult stem cells, which can also undergo extensive proliferation and differentiation to form organoids, have been utilized to mimic stomach (Stange Daniel et al. 2013), prostate (Chua et al. 2014), lung (Desai et al. 2014), salivary gland (Maimets et al. 2016), and esophagus (DeWard et al. 2014), just to name a few. Organoid cultures can often proliferate extensively and thus generate sufficient cells to replace damaged or diseased tissues (Yui et al. 2012). In addition, organoids offer several advantages over animal models and conventional cell culture systems in studying human development, physiology, and pathobiology. Organoids possess multiple cell types as well as 3D structural and morphological characteristics similar to native tissues; shorter duration and lower cost can generally be expected for experimentation with organoid models than with animal models; using patient-derived iPSCs and gene manipulation techniques such as the CRISPR/Cas9 system, patient- and disease-specific organoids with edited genome can greatly advance precision medicine research (Ran et al. 2013; Takebe and Wells 2019). In the history of musculoskeletal research, in 1929, Fell and Robison (1929) cultured skeletal tissue fragments from fowl embryos to model the skeletal development process in vitro. However, a relatively small number of musculoskeletal organoids have been established thus far (Mori et al. 2019). To engineer skeletal muscle tissues, a biomaterial scaffold is usually required to induce myofiber alignment under tension (Maffioletti et al. 2018). Recently, mouse iPSCs were used to fabricate 3D spherical bone/cartilage complex by micro-space culture followed by mechanical shaking (Limraksasin et al. 2020). We recently showed that after brief trypsinization, MSC-impregnated ECM experienced mesenchymal condensation and robust chondrogenesis (Yang et al. 2019).

2.3. Signaling Molecules

Cells residing in native musculoskeletal tissues receive a broad array of signals which can be chemically transmitted via growth factors and other molecules (Fig. 2b) or physically exerted through the immediate ECM. The application of such complex signals in TERM has considerable influences on the outcome. This section focuses on the signaling molecule-conveyed biochemical signals for musculoskeletal regeneration. The effects of some other signals are described in Sect. 3.

Bone

Bone is a highly vascularized tissue that undergoes constant remodeling throughout our lifetime, and signaling molecules play an important role in the complex bone remodeling process. Bone morphogenetic proteins (BMPs), particularly BMP2 and BMP7, show outstanding osteoinductivity and have therefore been widely incorporated in bone scaffolds (Yilgor et al. 2009). BMP2 was previously co-spun with hydroxyapatite-containing silk fibroin/poly(ethylene oxide) solution to fabricate fibrous bone scaffolds (Li et al. 2006; Lee et al. 2013). The presence of BMP2 markedly enhanced mineralization and upregulated osteogenic gene expression. Platelet-derived growth factor-BB (PDGF-BB), one of the five isomeric forms of PDGF, was found to enhance osteogenesis of adipose-derived MSCs and promote bone formation in a distraction osteogenesis rat model (Hung et al. 2015; Moore et al. 2009).

Vascularization plays a critical role in sustaining transplanted cells and/or scaffolds. The use of angiogenic growth factors, especially VEGF, in bone repair has led to encouraging outcomes (Murphy et al. 2000). Poly(lactic-co-glycolic acid) scaffolds containing human bone marrow-derived MSCs and condensed plasmid DNA encoding for BMP-4 and for VEGF were implanted into SCID mice, and the scaffolds containing all three components produced robust bone regeneration (Huang et al. 2005). Besides signaling molecules that act locally, systemic agents such as human growth hormone and parathyroid hormone are also important for bone regeneration (Dimitriou et al. 2011).

Cartilage

Unlike bone, articular cartilage is avascular and recalcitrant to repair and regeneration. Despite the relatively simple tissue composition of cartilage, i.e., chondrocytes embedded in a dense ECM, clinically successful cartilage tissue engineering remains a challenge. The survival and efficacy of transplanted cells in cartilage scaffolds remain controversial. Many approaches utilize biologically active molecules, such as growth factors, to enhance the recruitment of endogenous cells for expedited tissue regeneration. Transforming growth factor beta (TGF-β) plays a critical role in the maintenance of both articular cartilage and subchondral bone homeostasis (Zhen et al. 2013; Zhen and Cao 2014). TGF-β1 promotes MSC chondrogenesis (Miura et al. 1994); TGF-β1 releasing alginate-sulfate scaffold was shown to promote chondrogenesis both in vivo and in vitro (Re’em et al. 2012). Another member in the TGF-β family, TGF-β3, also possesses strong chondrogenic effects (Barry et al. 2001). It was reported that TGF-β3 incorporated within collagen-hyaluronic acid scaffolds could support the chondrogenesis of MSCs and produce cartilage-like ECM (Matsiko et al. 2015). Lee et al. (2010) 3D printed an anatomically correct scaffold and infused it with TGF-β3-loaded collagen hydrogel. It was found that these scaffolds led to regeneration of hyaline cartilage with superior compressive and shear properties and recruited cells significantly more than the TGF-β3-free scaffolds in a rabbit humeral head defect model (Fig. 1h–k). The use of biologically active molecules, therefore, holds promising potential in facilitating cell homing without cell delivery.

BMPs and the growth differentiation factors (GDFs), as members of the TGF-β superfamily, are also able to enhance cartilage regeneration. BMP2, BMP4, and BMP7 have been incorporated in various controlled delivery strategies for cartilage repair (Lam et al. 2015). Sun et al. (2019) found that GDF5 enhanced the migration and chondrogenesis of MSCs in vitro; furthermore, implanting a 3D-bioprinted GDF5-conjugated MSC-laden scaffold led to robust cartilage regeneration in a rabbit knee defect site.

Muscle

Several growth factors, including IGF1, fibroblast growth factor 2 (FGF2), nerve growth factor (NGF), placenta-derived growth factor (PIGF), and VEGF, contribute to muscle repair through different mechanisms (Wei and Huard 2008). IGF is capable of mediating the proliferation and differentiation of muscle stem cells (Adams 2000). Gel delivery of VEGF to ischemic muscle tissue increased the expression of neurotrophic factors and promoted the regrowth and maintenance of damaged axons via NGF/GDNF (nerve growth factor/glial-derived neurotrophic factor) signaling (Shvartsman et al. 2014). The co-delivery of the myogenic factor IGF1 and the angiogenic factor VEGF via an injectable alginate gel has been shown to recover muscle functions from ischemic injuries in a mouse model (Borselli et al. 2010). Wang et al. (2014) engineered a degradable, shape-memory alginate scaffold that delivers myoblasts, IGF, and VEGF through a minimally invasive approach to injured mouse muscle. The implant led to reduced muscle fibrosis, enhanced vascularization, and enhanced functional recovery. A 3D PEG-fibrinogen hydrogel incorporating mouse mesoangioblasts transduced with a PIGF lentivirus was found to attract host vessels and nerves and generate newly formed tissues histologically similar to native muscle in an ablated muscle injury model (Fuoco et al. 2015).

Tendon and Ligament

No consensus has been reached on the optimal tenogenic induction protocol. Previous studies have proposed the use of growth factors such as IGF1, FGF2, PDGF-BB, GDF-5, TGF-β3, connective tissue growth factor (CTGF), and BMP12 to promote tendon cell proliferation and enhance tendon regeneration (Raghavan et al. 2012; Rossi et al. 2011; Wolfman et al. 1997). Many of these growth factors were similarly used for ligament repair (Pauly et al. 2017; Hee et al. 2012). It was previously shown that a 12 h BMP-12 (10 ng/ml) treatment could significantly enhance MSC tenogenic marker expression, and this phenotype could be sustained in vivo in rat tendon defects (Lee et al. 2011). Barsby et al. (2014) found that by loading TGF-β3 into an ESC-seeded 3D collagen scaffold, tendon-associated gene and protein expression by the ESCs could be significantly upregulated.

3. Musculoskeletal Microphysiological Systems

As an in vitro experimental research platform, MPS mimic a tissue or organ by providing living cells with a microenvironment in which the cells can display tissue- or organ-specific phenotypic, structural, and functional characteristics. Successful MPS have been shown to exhibit characteristics that bear high structural and biologic fidelity to tissues that are difficult to achieve in conventional, static 2D, or 3D cultures (Grayson et al. 2010). Namely, MPS-derived musculoskeletal tissues aim to present tissue maturation and complexity as seen in in vivo models, where diverse cellular composition, structurally complex ECM, and interactive signals from biochemical and physical stimuli co-exist. More importantly, the application of MPS not only contributes to achieving musculoskeletal regeneration via the technology of fabricating tissue implants but may also impact the field by establishing disease models that facilitate understanding of musculoskeletal pathogenesis and expediting the development of potential therapies and therapeutic agents. Over the past decade, MPS have been utilized to model various musculoskeletal diseases such as OA (Lin et al. 2019), bone metastasis (Marturano-Kruik et al. 2018), and muscle injury (Agrawal et al. 2017). Examples of pharmacological agents that were tested in musculoskeletal MPS for treating OA include the glucocorticoids dexamethasone and triamcinolone, celecoxib, rapamycin, and HYADD^®4 (hyaluronic acid alkylamide) (Lin et al. 2019; Occhetta et al. 2019; Rosser et al. 2019). The physiological relevance and clinical applicability of the MPS depends critically on optimal combination of biomaterial scaffold, appropriate cells, and biochemical and physical signals. Table 1 summarizes the biomaterials, cells, and signals utilized in representative MPS developed to recapitulate musculoskeletal tissues.

Table 1.

Representative MPS that capitulate musculoskeletal tissues/organs, the target tissues/organs, and the biomaterials, cells, and type of stimuli employed

Target tissues	Biomaterials	Cell types	Signals/stimuli	Physiological/clinical relevance	Ref.
Cartilage	PEG-based hydrogel	Human chondrocytes	Mechanical compression; dexamethasone, IL-1Ra, rapamycin, celecoxib treatment	Mechanical overload induced OA phenotype in the tissue	Occhetta et al. (2019)
Cartilage	Fibrin	Equine chondrocytes	Exposure to TNF-α and IL-1β and steroid (triamcinolone) treatment	The microtissues emulate basic properties of native cartilage and respond to biochemical insults and concurrent steroid treatment	Rosser et al. (2019)
Cartilage and synovium	Fibrin	Human articular chondrocytes and synovial fibroblasts	Synovial fluid from OA patients	Synovitis is characterized by the increased infiltrating monocytes and macrophages in synovium (to be verified in model)	Mondadori et al. (2018)
Bone	Devitalized bovine trabecular bone	Human MSCs	Interstitial flow of medium	Geometrically complex temporomandibular joint condylar bone structures were created with high structural and biologic fidelity	Grayson et al. (2010)
Bone	N.A. (cells directly seeded in PDMS mold)	Murine preosteoblast MC3T3-E1	Breast cancer cells MDA-MB-231-BRMS1GFP and MDA-MB-231GFP	Formation of mineralized collagen was seen in the chip and hallmarks of breast cancer bone colonization observed	Hao et al. (2018)
Bone	Fibrin	Human MSCs, osteodifferentiated MSCs and HUVECs	Breast cancer cell MDA-MB-231	A vascularized bone-mimic was generated to study breast cancer cell extravasation and test drugs	Jeon et al. (2015)
Bone	Decellularized bovine bone	Human MSCs and HUVECs	Breast cancer cell MDA-MB-231	Drug resistance and progression of breast cancer cells colonizing the bone were observed in the bone perivascular niche-on-a-chip	Marturano-Kruik et al. (2018)
Osteochondral tissue	Gelatin	Human MSCs	Interleukin (IL)-1β stimulation	IL-1β treatment of either bone or cartilage induced degenerative responses in the other tissue, suggesting their crosstalk	Lin et al. (2014)
Osteochondral tissue	Gelatin	Human iPSCs	IL-1β stimulation on cartilage; celecoxib treatment	The presence of bone aggravated cartilage degeneration	Lin et al. (2019)
Skeletal muscle	Collagen type I	Murine skeletal muscle cell line C2C12	Electrical stimulation	The engineered muscle microtissues contracted when electrically stimulated	Shimizu et al. (2015)
Skeletal muscle	Collagen type I with Matrigel™	Optogenetically encoded myoblasts C2C12 cell expressing Channelrhodopsin-2	Optical stimulation	The device allows convenient testing of the effects of various factors on muscle maturation, structure, and function	Sakar et al. (2012)
Skeletal muscle	Gelatin	C2C12 cell	Cardiotoxin treatment	Cardiotoxin induced structural destruction and reduced passive tension in engineered muscle tissue	Agrawal et al. (2017)
Neuromuscular junction	N.A. (scaffold-free culture in a PDMS chip)	Human skeletal muscle myocytes and human iPSC-derived motoneuron	Electrical stimulation and treatment with 3 NMJ toxins	Muscle contraction induced by motoneuron activation declined or ceased with toxin treatment	Santhanam et al. (2018)

Selection of Biomaterials

Most MPS employ hydrogels as the cell carrier to create a 3D culture environment. In addition to their biocompatibility and high water content, a major advantage of hydrogels in MPS is their injectability, allowing the cell-laden structure to crosslink in situ and conform to a desired geometry. Among the frequently used hydrogels are collagen (Sakar et al. 2012), fibrin (Rosser et al. 2019), and gelatin (Lin et al. 2014). Devitalized bone, as a natural biomaterial, possesses the microstructural, biochemical, and mechanical properties of native bone, provides an inductive niche for mineralization and angiogenesis, and has been successfully used in bone-on-a-chip systems (Marturano-Kruik et al. 2018; Grayson et al. 2010).

Cell Types

As multiple cell types are found in the majority of human tissues, MPS must also replicate this characteristic. Both stem cells and terminally differentiated cells have been utilized to construct MPS. Previous research has shown the feasibility of generating heterogeneous tissues from a single stem cell source through the perfusion of separate induction medium streams (Lin et al. 2014). Cells isolated from diseased tissues can bring with them “memories” of the disorders, thus displaying different phenotypes in vitro compared to those harvested from healthy tissues. For example, chondrocytes from healthy and OA joints were found to possess altered chondrogenic potential (Yang et al. 2006). Many musculoskeletal diseases are patient-specific, with varying stages, genetics, etiology, and drug sensitivity in different individuals. Using individual-specific cells, MPS offer a personalizable approach to understanding disease mechanisms and testing treatment options. Although the use of primary patient cells best recapitulates patient specificity, it is impractical when large cell numbers are needed. iPSCs, with an almost unlimited proliferative capacity, overcome this limitation. Although there is still a long way to go to realize wide clinical applications of iPSCs due to their possible immunogenicity and tumorigenicity and the presence of genetic and epigenetic aberrations, the use of patient-specific iPSCs has contributed significantly to the advancement of precision medicine (Tabar and Studer 2014; Sayed et al. 2016). For example, MPS engineered from iPSCs can serve as an individualized platform for various preclinical tests, thus facilitating the identification of potentially efficacious therapies tailored to individual patients.

Stem cell-derived organoids have been extensively researched to study human physiology and build disease model-in-a-dish (Lancaster et al. 2013). As tissue analogs generated in vitro, organoids are recognized for their higher biological fidelity and have been used in MPS to generate organoid-on-a-chip systems (Skardal et al. 2016). MPS create a dynamic microenvironment mimicking in vivo conditions such as continuous fluidic flow and varying oxygen concentrations, promoting communications between multiple cell/tissue components, and generating potentially more clinically relevant organoid responses to toxins, drugs, and potential medications. In the organoid-on-a-chip research community, increasing attention has been drawn to the use of custom-designed MPS that allow multi-organoid integration to ultimately generate body-on-a-chip systems with sufficient physiological relevance for accurate prediction of native tissue/organ responses to selected treatments (Skardal et al. 2016).

Biochemical and Physical Stimulations

Cells and tissues grown in musculoskeletal MPS can be subjected to various forms of environmental signals, including biochemical (Lin et al. 2014), mechanical (Occhetta et al. 2019), optical (Sakar et al. 2012), and electrical (Santhanam et al. 2018) stimulations, and the elicited responses constitute the MPS readouts. Generally, the application of MPS cultures involves three distinct stages: (1) tissue maturation, (2) disease modeling, and (3) therapeutic drug testing. In the first stage, biochemical signals provided by signaling molecules such as growth factors are usually required if the differentiation of stem cells or progenitors is involved. Other stimulations, including mechanical loading or stretching, interstitial fluid flow, and electromagnetic field, have been shown to be beneficial for enhancing matrix deposition and/or functions of musculoskeletal tissues (Langelaan et al. 2011; Mauck et al. 2000; Grayson et al. 2010). The most common strategy for generating musculoskeletal disease models in MPS is through the introduction of biochemical insults by proinflammatory cytokines or toxins (Lin et al. 2014; Agrawal et al. 2017). Given the high prevalence of mechanical injury-caused musculoskeletal disorders, hyperphysiological mechanical stress-induced pathogenesis is also believed to be of high clinical relevance (Occhetta et al. 2019). Once tissue abnormalities are observed, potential therapeutics targeting the corresponding mode of tissue injury can be introduced to test efficacy in disease modification as well as to evaluate potential toxicity. In some studies disease modeling and drug testing steps occur simultaneously (Rosser et al. 2019). When therapeutics are introduced prior to and/or along with disease causative agents, MPS may provide valuable information on the drugs’ efficacy in not only disease treatment but also disease prevention.

Generally, in the design of the MPS, there exists a tradeoff between device complexity and throughput in the process of balancing the level of biological accuracy and the required technological elements, and it is highly dependent on the needs of the application (Arrigoni et al. 2020). In basic, mechanistic studies, much higher levels of physiological accuracy are desired than in drug screening applications, where throughput and access to resources are important considerations.

As an example, MPS modeling of human synovial joints can serve different purposes. For investigating the pathogenesis and etiology of joint disorders like OA, it is critical to take into account the crosstalk and communication among different joint tissues because OA has long been considered a whole-joint disease (Loeser et al. 2012). The incorporation of multiple tissue components, including cartilage, bone, synovium, and infrapatellar fat pad, in the MPS would be critical to create sufficient physiological relevance. A particular challenge posed by the highly prevalent and debilitating disease of OA is the absence of an effective, FDA-approved disease-modifying medication. A number of candidate disease-modifying OA drugs have been proposed for OA treatment, and there is thus an urgent need of a convenient, efficient, and reliable model for high-throughput drug screening. For this purpose, MPS with low technical sophistication and high cost-effectiveness would be more appropriate. Nevertheless, we believe that investigations using a physiologically relevant and complex MPS, after extensive evaluation and validation, can offer useful guidance, e.g., identification of key readouts, to inform the design of practical and high-throughput systems.

4. Summary and Future Perspectives

It has become clear that intimate crosstalk between different cell types and cell-matrix interactions significantly influence cell proliferation, migration, and differentiation during the regeneration process (Kuraitis et al. 2012; Marsell and Einhorn 2011). Therefore, the traditional 2D culture using one cell type is less informative in predicting future clinical outcomes of potential regenerative treatments. However, 2D cultures are relatively fast and inexpensive and have lower technical requirements compared to animal models and MPS. Thus, conventional 2D dish culture should serve as the first and high-throughput platform to exclude the drugs that display obvious cytotoxicity or inefficacy for further tests. In addition, a 2D dish model allows the easy manipulation of cells, such as gain- or loss-of-function assessment, which enables mechanistic study. The use of animal models is able to partially mimic the local and global physiological changes in humans, recapturing the cell-cell and tissue-tissue crosstalk in tissue injury and repairing processes. In addition, by targeting different mechanisms, animal models are able to simulate different injuries in humans. Therefore, animal models, in particular those using clinically relevant animals, still represent the most powerful models in predicting treatment outcomes in humans. However, as has been recognized for a long time, due to the inherit difference in physiology and anatomic structure, the translation from animal models to human is challenging and not straightforward (Muschler et al. 2010). For example, autologous chondrocyte implantation, a clinically used regenerative method to treat chondral defects, showed robust cartilage repair in the first animal study in rabbits (Grande et al. 1989). However, this technology is mostly only applicable for the treatment of focal defects, and the clinical benefits over older techniques such as microfracture are sometimes questioned (Mollon et al. 2013). Such potential outcome discrepancies between animal models and human patients must be considered when translating regenerative therapies from animal studies to human clinical applications.

As mentioned above, the MPS and organoids have emerged as the next generation of models to assess the utility of regenerative medicine in treating musculoskeletal diseases. The advantages and disadvantages of these technologies, however, are both obvious. With the use of human cells and connective tissues in the manner that is observed in vivo, we may be able to recapitulate the tissue crosstalk and repair processes in humans. The major limitation of current MPS is the insufficient fidelity in replicating native tissues on the anatomical structure, phenotype, or function. In addition, the simulation of some physiological activities, such as mechanical loading, still presents significant technical challenges. Kaarj and Yoon (2019) recently reviewed the methods of delivering mechanical stimuli to MPS. With the progress in scaffold fabrication technologies, in particular 3D printing, as well as the tissue-specific differentiation of iPSCs, the rapid evolution of MPS is expected in the near future. It is noteworthy that immune cells, such as macrophages, which have been shown to play a major role in tissue regeneration (Wynn and Vannella 2016), are often neglected in current MPS. Therefore, successful development and application of MPS requires collaboration among experts from different disciplines and research fields.

The different etiologies and pathogenesis of musculoskeletal injuries and diseases further amplify the challenges in developing generally accepted regenerative treatments. Additional heterogeneity of musculoskeletal injuries and responses to treatments also arise from the patient’s genetic background; for example, single nucleotide polymorphisms between the major and minor alleles of expressed genes have recently reported to display significantly different regulatory activities (Klein et al. 2019). Therefore, future models must have the capacity to model patient-specific physiology and pathology, in order to allow the development of personalized regenerative treatments. MPS will be the most promising models for such applications. In particular, given that iPSCs have theoretically unlimited expansion capacity as well as potential to generate all human tissues/organs (Shi et al. 2017), MPS derived from patient-specific iPSCs should possess unique advantages over animal models in developing personalized medicine (Fig. 3). However, significant technological advances in stem cell biology to achieve controlled differentiation of iPSCs are clearly needed.

Fig. 3.

Schematic of traditional and next-generation drug development pipelines

The key requirement for a clinically relevant model is the capability to replicate the human physiology, under both normal and disease conditions, as well as to precisely predict the human response to a treatment. Achieving this relies on extensive validation. In particular, the model should be able to replicate the success or failure of known treatments that have been used clinically. Owing to the limited number and relatively short history of regenerative medicine products that have been used in clinical applications, the validation of new regenerative medicine products with known treatment is often not feasible. Thus, the establishment of a “gold standard” translation pathway is urgently needed, which will require the collaboration among practitioners of life science, pathology, engineering, and clinical medicine.