Introduction

Articular cartilage is a thin connective tissue covering the end of bones to transmit loads to the subchondral bone and to provide a smooth surface for low-friction motion. Many joint disorders, such as intra-articular fracture, can lead to cartilage defects. The repair ability of articular cartilage is very limited because of its avascular nature and the paucity of resident stem cells. Partial-thickness defects of mature articular cartilage do not heal spontaneously [1, 2], while full-thickness lesions are always repaired with fibrocartilage that has inferior mechanical properties [3, 4]. Much effort has been devoted to cartilage regeneration; nevertheless, current medical interventions, such as microfracture and mosaicplasty, can relieve clinical symptoms to some extent but fail to restore functional and phenotypically stable hyaline cartilage [5,6,7].

Autologous chondrocyte implantation (ACI) has been considered as a useful repair technique for cartilage defects [8]. In this cell-based technique, chondrocytes are harvested from a non-weight bearing articular cartilage after biopsy, expanded in vitro and then implanted into the cartilage defect [9]. Nevertheless, chondrocytes gradually lose their phenotype during in vitro expansion [10, 11], and dedifferentiated chondrocytes have been reported to be unable to generate stable hyaline cartilage [12,13,14]. Therefore, the phenotype changes of expanded chondrocytes may compromise the long-term clinical outcomes of ACI.

Because stem cells can generate hyaline-like cartilage tissue under specific conditions, stem cell-based therapy has been regarded as a promising approach to address the complexity of cartilage injury. Recently, mesenchymal stem cells (MSCs) have attracted increasing attention for cartilage regeneration owing to their ease of isolation, relatively high expansion rates, low immunogenicity, and multipotency. MSC-specific markers are not yet clearly defined, and the phenotype and comparison of MSCs from different tissues isolated using different protocols are still based on the combination of different parameters according to the minimal criteria proposed by the International Society for Cellular Therapy [15].

MSCs are comprised of heterogeneous cell populations, and their regenerative capacity varies among tissue sources [16,17,18]. MSCs from different tissues have been shown to differ dramatically in chondrogenic potency [19,20,21]. Hence, the innate tissue-specific characteristics should be taken into consideration when developing MSC-based cartilage regeneration approaches. For instance, bone marrow-derived MSCs (BM-MSCs) are apt to undergo endochondral bone development rather than form stable hyaline cartilage after chondrogenic induction. Without any osteogenic induction, rat BM-MSCs can spontaneously calcify during in vitro culture [22]. The subcutaneous implantation of chondrogenically-primed BM-MSCs leads to bone formation, but not stable cartilage formation [23,24,25,26]. Furthermore, the transplantation of chondrogenically-primed BM-MSC pellets into an atrophic, non-union environment resulted in pronounced bone regeneration through endochondral ossification [27]. These findings raise considerable concern about the improper differentiation of BM-MSCs during cartilage regeneration.

Recent studies found that stem/progenitor cells derived from synovial joint tissues showed superior chondrogenic ability when compared with those derived from bone marrow and adipose tissue [20, 28, 29], thus suggesting that the resident stem/progenitor cells in synovial joints could be ideal cell sources for cartilage repair. A further understanding of the biology and regenerative potential of these progenitors is likely to yield new therapeutic approaches that could successfully repair cartilage defects. Therefore, in this review, we discussed the distribution and function, preclinical and clinical applications, and future perspectives of stem/progenitor cells isolated from three types of synovial joint tissues: articular cartilage, synovial membrane and synovial fluid.

In vivo Distribution of Progenitors in the Synovial Joint

The synovial joint comprises different tissues enclosed within the joint capsule, including articular cartilage, bone, synovium, ligaments, and the fibrous capsule. The following techniques have been employed to determine the distribution of progenitors in the synovial joint: 1) DNA labeling methods to detect the slow-cycling cells [30,31,32]; 2) transgenic animal models to track the fate of progenitors [33]; 3) immunohistological analysis to detect the distribution of putative stem cell markers [34,35,36,37,38].

Tissue Distribution of Progenitors in the Synovial Joint

The distribution of stem/progenitor cells in synovial joint tissues has been studied in some rodent species, such as rat, mouse and rabbit. Progenitors are slowly proliferating cells in vivo; therefore, DNA labeling agents, such as [3H]-thymidine, bromodeoxyuridine (BrdU) and 5-ethynyl-2′-deoxyuridine (EdU), are useful to identify their localization due to the long-term labeling capacity of these agents in stem/progenitor cells. Usually, the distribution of cells that retain a long-term label is determined by a pulse-chase method.

In embryonic and young rats, cells with long-term [3H]-thymidine-labeling were observed in the proximal portion of the growth plate, the perichondrial ring and the surface of articular cartilage [30]. Similarly, in rabbit joints, long-term BrdU-positive cells were observed in the germinal zone of the growth plate, the perichondrial groove of Ranvier and all the zones of articular cartilage [31]. In the developing synovial joints of mice, long-term EdU-labeled cells were predominantly present at the surface zone of the articular cartilage, but were also detectable in the other areas of articular cartilage, the perichondrium/periosteum and the synovium [32]. Collectively, these studies showed that stem/progenitor cells reside in the articular cartilage and its adjoining tissues.

Tissue Distribution of Progenitors in the Articular Cartilage

The articular cartilage consists of histologically distinct zones with different cell populations: the superficial zone, the transition zone, the deep zone, and the calcified zone [39]. During development, articular cartilage grows appositionally from the articular surface, indicating that the superficial zone contains progenitor cells that provide transit-amplifying progeny for growth [40]. Using a transgenic mouse model, Kozhemyakina et al. tracked the fate of articular cartilage progenitor cells and found that the Prg4+ cells, a progenitor population for the deeper layers of mature articular cartilage, were observed at the joint surface in the embryo [33].

Several putative stem cell markers have been used to determine the location of progenitors in the cartilage. CD166, a putative marker of articular cartilage stem/progenitor cells, was primarily located in the superficial and middle zones of cartilage [34]. Interestingly, the expression of Notch-1, a cell surface marker of in vitro cultured cartilage progenitors, was observed in all zones of human articular cartilage [35]. Likewise, other putative MSCs markers, such as CD90, Stro-1, Oct-3/4, and CD105, were found in various zones of human cartilage [36]. From the superficial to the deep zone of articular cartilage, a high percentage of cells were positive for Notch-1, Stro-1 and VCAM-1 [37]. In view of the low frequency of stem/progenitor cells in the cartilage, these abundantly expressed cell markers may not be sufficient to determine the precise location of stem/progenitor cells; thus, further investigation is needed.

Tissue Distribution of MSCs in the Synovial Membrane

The synovial membrane has two layers: the synovial lining and the sub-intimal layer. The synovial lining is comprised of fibroblast-like synoviocytes and macrophage-like synoviocytes, while the sub-intimal space contains fibrous tissue, blood vessels and immune cells [41]. In vitro, the ultrastructural and immunocytochemical features of synovial membrane-derived MSCs (SM-MSCs) are similar to the fibroblast-like synoviocytes, indicating that they may originate from the synovial lining [42].

In a mouse model of joint-surface injury, SM-MSCs have been identified in the lining layer and the subsynovial tissue of synovium [43]. In healthy humans, SM-MSCs were found to be localized in the subintimal zone of the synovial membrane [44]. In patients with osteoarthritis, however, the distribution of SM-MSCs was more diffuse; they have been found around veins in the perivascular matrix [44], as well as in synovial surface projections [38].

Response of Stem/Progenitor Cells to Cartilage Injury

To maintain tissue homeostasis, adult stem cells undergo asymmetric cell division to self-renew and generate functional cells to replenish the dead ones [45]; this process can be intensified by injury signals to restore tissue function. After cartilage injury, stem/progenitor cells in the joint tissues, such as synovium, undergo proliferation and serve as a pool of reparative cells [43, 46,47,48].

In vivo Studies

Partial-thickness articular cartilage injuries (PTCIs) provide a useful model to study the intrinsic reparative response of resident stem/progenitor cells in the synovial joint [46, 47]. Notably, in the PTCIs model, cartilage defects do not penetrate the subchondral bone; and consequently, there is no infiltration of bone marrow cells from the subchondral bone, which usually serves as the major source of reparative cells in full-thickness cartilage defects [52, 53].

In immature rats, the repair response occurs immediately after PTCIs, and progenitor cells, identified as CD105+ and CD166+ cells, have been found in the superficial and transitional zones of the reparative cartilage tissue [46]. In mature rats, putative progenitor cells, defined as CD105+ or BrdU-label-retaining cells, have been found around the injury sites of PTCIs and increased during the repair process, indicating that cartilage injury activated the proliferation of these progenitors [47]. Although there is no recruitment of BM-MSCs from the subchondral bone in PTCIs, the precise tissue origin of these reparative stem/progenitor cells remains uncertain.

The synovium is highly responsive to cartilage injury. MSCs residing in the synovium have been shown to respond to full-thickness articular cartilage injury through cell proliferation and chondrogenesis [43]. At PTCIs, the recruitment of repair cells from the synovial membrane was evident, as shown by a continuous layer of mesenchymal cells extending from the synovial membrane across the surface of normal articular cartilage into the defect [48]. Additionally, the recruited synovial cells were able to differentiate into chondrocytes when stimulated by TGF-β1 [49]. Collectively, these data clearly showed that the synovium contributes to articular cartilage repair in both full- and partial-thickness defects.

Synovial fluid (SF) is a clear, viscous, hyaluronic acid-rich liquid in contact with the synovial membrane and articular cartilage [54]. It provides a route for exogenous stem cells to access cartilage defects after intra-articular injection [55]. In normal joints, the number of synovial fluid-derived MSCs (SF-MSCs) is very low, but it increases notably under injury or in osteoarthritic conditions [50, 51].

Ex vivo Studies

Cartilage explants are very helpful for investigating the reparative mechanism of cartilage injury. Following a mechanical injury, two repair mechanisms have been described in porcine cartilage explants: 1) proliferation of cells close to the injured cartilage, and 2) chemotactic migration of cells toward the injured surface followed by binding to the repair tissue [56]. Likewise, in blunt-impact-injured cartilage explants, nonviable areas were repopulated by MSC-like cells migrating from the surrounding matrix [57], and this repair response can be enhanced by stromal cell-derived factor 1 alpha through stimulation of the recruitment of local cartilage progenitors [58]. At the wound edge of explants, cells responsible for neocartilage formation originated from the deep zone of cartilage [59]. Taken together, these studies revealed that tissue-resident progenitors were actively engaged in the repair of injured cartilage.

Biological Characteristics

Articular Cartilage-Derived Mesenchymal Progenitor Cells (AC-MPCs)

Chondrocytes represent a highly differentiated cell type that producing cartilage-specific extracellular matrix, and they have long been regarded as the only cell type in articular cartilage. In 2002, Dowthwaite et al. first isolated AC-MPCs, a cell type with multi-differentiation potential, from the surface of bovine articular cartilage by a method of differential cell adhesion to fibronectin-coated dishes (termed as the fibronectin adhesion assay). They observed that AC-MPCs formed large numbers of colonies from an initially small seeding density and expressed α5β1 integrin and Notch-1 [60]. Subsequently, they found that AC-MPCs exhibited phenotypic plasticity in an embryonic chick tracking system [61]. Similarly, AC-MPCs isolated from human articular cartilage have been shown to be multipotent under specific induction conditions, and to express a panel of cell surface markers typical of MSCs (Table 1). Based on the similarities with MSCs, in literature AC-MPCs are also referred to as articular cartilage-derived mesenchymal stem cells [62, 103, 105].

Table 1 Multipotency and cell surface markers of AC-MPCs, SM-MSCs and SF-MSCs
Full size table

In addition to the fibronectin adhesion assay, other techniques, as shown in Fig. 1, have been applied to isolate AC-MPCs, including the Hoechst 33,342 exclusion assay and the cell sorting techniques. The Hoechst 33,342 exclusion assay is a valuable technique for identifying and sorting AC-MPCs because of the property of adult stem/progenitor cells to exclude this DNA binding dye. The cell sorting techniques are used to isolate AC-MPCs based on the positive and/or negative expression of specific cell surface markers. Interestingly, due to differences in the isolation methods, variations in some of the biological properties of AC-MPCs, such as differentiation potency, were presented. For instance, AC-MPCs isolated by the Hoechst 33,342 exclusion assay have been shown to be osteogenic and chondrogenic, but not adipogenic, after in vitro induction [37]. Additionally, the proliferation and differentiation abilities among AC-MPCs subpopulations have been shown to vary enormously [62].

Fig. 1
figure 1

Schematic representation of the isolation of stem/progenitor cells from articular cartilage, synovial membrane and synovial fluid. AC-MPCs: articular cartilage-derived mesenchymal progenitor cells; SF-MSCs: synovial fluid-derived mesenchymal stem cells; SM-MSCs: synovial membrane-derived mesenchymal stem cells; FACS: fluorescence-activated cell sorting

Full size image

AC-MPCs have greater chondrogenic potency than BM-MSCs [63] and adipose-derived MSCs [28, 64, 65]. Unlike the full-depth chondrocyte populations, AC-MPCs preserved chondrogenicity after extensive expansion [66]. Notably, AC-MPCs did not obtain a hypertrophic cartilage phenotype after chondrogenic induction, and could thus form stable hyaline cartilage without calcification [67].

SM-MSCs

The surface markers of SM-MSCs are similar to those of BM-MSCs (Table 1), but the expression is influenced by various factors such as cell passage number [68, 69]. SM-MSCs are multipotent, and their differentiation ability is not influenced by donor age, cell passage or cryopreservation [70]. SM-MSCs are highly clonogenic, with a clone forming efficiency more than 100-fold higher than that of BM-MSCs [71, 72]. Furthermore, SM-MSCs isolated by different methods (Fig. 1) have robust in vitro expandability; when subcultured at low density, SM-MSCs retained proliferation ability after extensive expansion [71].

SM-MSCs possess greater chondrogenic ability than MSCs derived from extra-joint tissues, such as the adipose tissue [73], bone marrow [71] and umbilical cord [20]. It is reported that gene expression profiles of chondrocytes and SM-MSCs are closer to each other than those of extra-articular tissue-derived MSCs [74]. At the single-cell level, SM-MSCs are heterogeneous in chondrogenic potency [75]. Therefore, as shown in several studies, an enriched subpopulation of SM-MSCs could be more efficient for chondrogenic differentiation than the mixed SM-MSCs populations [69, 76, 77].

SF-MSCs

SF-MSCs are fibroblast-like cells with a phenotype similar to that of BM-MSCs (Table 1). SF-MSCs are usually isolated by the method of direct cell seeding in the culture flask (Fig. 1), and they are highly proliferative in vitro. For instance, most bovine SF-MSCs could expand for at least 1 million-fold [78]. The proliferation ability and the expression of pluripotent transcription factors of SF-MSCs were higher than BM-MSCs [79].

In contrast to the proliferation ability, SF-MSCs were inferior in adipogenic, osteogenic and neurogenic differentiation when compared to BM-MSCs [78, 79]. However, the chondrogenic potential of SF-MSCs was comparable to SM-MSCs [29]. The detailed mechanism of cell-source-dependent variant in the biological properties of MSCs has not been fully understood. Gene profiles indicated that SF-MSCs were more similar to SM-MSCs than to BM-MSCs and adipose tissue-derived MSCs [80].

Interestingly, the health state of the donor greatly influences the chondrogenic potential of SF-MSCs. For example, SF-MSCs derived from normal joints did not require a micro-mass step for efficient chondrogenesis, while those obtained from osteoarthritic joints needed the micro-mass procedure [81].

Animal Studies

AC-MPCs

Very few animal studies have been conducted to determine the cartilage repair potential of AC-MPCs (Table 2). To determine the in vivo plasticity of AC-MPCs, fluorescent PKH26 labeled cells were injected into the thigh muscle of severe-combined immunodeficient (SCID) mice; at two weeks post implantation, no robust cartilage pellet, but only diffuse cartilage nodules, were found, suggesting that AC-MPCs required further signals for chondrogenic differentiation after ectopic implantation [82].

Table 2 Animal studies of AC-MPCs, SM-MSCs and SF-MSCs for cartilage repair
Full size table

Nevertheless, in a full-thickness cartilage defect model, the transplantation of autologous AC-MPCs resulted in improved reparative tissue and significantly reduced central osteophyte formation [83]. Moreover, the transplantation of AC-MPCs showed histological repair scores similar to those of full-depth chondrocytes, and both groups showed evidence of collagen type II-positive repair tissue [66].

SM-MSCs

Compared with AC-MPCs, more animal studies have been performed to explore the cartilage repair potential of SM-MSCs (Table 2). After induction in a traditional chondrogenic medium, primed SM-MSCs were unable to form stable hyaline cartilage after ectopic implantation [84, 85]. Interestingly, when transplanted into cartilage defects, SM-MSCs showed high cartilage repair ability in both small and large animal experiments [86,87,88,89,90,91,92], indicating that the microenvironment of graft sites influenced greatly on the fate and behaviors of grafted cells.

In a rat osteochondral defect model, it was observed that articular cartilage defects could be repaired by grafted SM-MSCs [86]. Likewise, in rabbit full-thickness osteochondral defects, transplantation of SM-MSCs improved cartilage repair [87,88,89,90]. Interestingly, placing a suspension of SM-MSCs on the surface of cartilage defects resulted in rapid adherence of grafted cells and an improved cartilage repair outcome in full-thickness osteochondral defects [91, 92]. Due to the promising outcomes and the simple treatment procedure, this cell delivery strategy is attractive for clinical application.

In partial-thickness chondral defects (PTCDs), however, transplanted cells have difficulty in attaching to the surface of lesions, which may be due to the anti-adhesive properties of proteoglycan, a rich component of the cartilage matrix existing at the surface of PTCDs [93]. Nakamura and his colleagues evaluated the reparative ability of allogenic SM-MSCs in a porcine PTCDs model [94,95,96,97]. They first generated 3-dimensional scaffold-free tissue engineered constructs (TECs) from allogenic SM-MSCs in vitro and then implanted TECs into the PTCDs; the results showed that TECs promoted the repair of chondral lesions, and the reparative tissue exhibited mechanical properties similar to normal cartilage in static compression, friction and unconfined compression tests [94, 95]. However, further analysis revealed some compromised mechanical properties of the reparative tissue: 1) the surface stiffness of the reparative tissue, measured by a micro-indentation analysis, was significantly lower than that of normal cartilage [97], and 2) the tensile strength of the integration boundary between native cartilage and reparative tissue was significantly lower than that of uninjured cartilage [96]. Hence, further investigations are needed to improve the integrity of reparative tissue.

SF-MSCs

Only one report has determined the cartilage repair ability of SF-MSCs (Table 2). In this study, the SF-MSC-laden platelet-rich plasma (PRP) hydrogel possessed better therapeutic potential than PRP hydrogel alone in repairing porcine osteochondral defects, which was evidenced by an increase in cell growth and the maturation of chondrocytes [98].

Clinical Trials

Clinical trials focusing on cartilage regeneration by articular cartilage-, synovial membrane- and synovial fluid-derived stem/progenitor cells are very limited. According to the clinical trials database (www.clinicaltrials.gov), to date (Jun 1, 2017), there is only one completed trial (Identifier: NCT01879046) that explored the cartilage repair potential of SF-MSCs, while there are no trials investigating AC-MPC- or SM-MSC-based cartilage repairs.

In literature, two studies have reported the clinical application of AC-MPCs and SM-MSCs for cartilage repair [99, 100], but there were no clinical results regarding SF-MSC-based cartilage regeneration. For AC-MPCs, Jiang and colleagues obtained a population of AC-MPCs from fully differentiated human articular chondrocytes and evaluated their repair ability for large knee cartilage defects in 15 patients; the clinical outcomes of AC-MPCs implantation were highly encouraging [99]. Concerning SM-MSCs, it was reported that transplantation of SM-MSCs could improve the clinical outcomes of patients with a symptomatic single cartilage lesion of the femoral condyle, in terms of magnetic resonance imaging score, qualitative histology and clinical evaluation scores [100].

Current Challenges and Future Perspectives

The high chondrogenic potential of AC-MPCs, SM- and SF-MSCs makes them promising graft cells for cartilage repair. However, many questions need to be addressed before extensive clinical application. Particular attention should be paid to the following unanswered questions: 1) How can a sufficient amount of the aforementioned stem/progenitor cells with high therapeutic potential be obtained? 2) How can the therapeutic potential of grafted cells be promoted/enhanced? 3) What is the repair mechanism? 4) How safe and efficient are these strategies?

First, obtaining a clinically relevant number of cells is an important premise for cell-based therapy. The small number of progenitors in articular cartilage and the synovial fluid hampers the acquisition of a sufficient number of cells. Furthermore, clinical-grade cell expansion protocols for the aforementioned stem/progenitor cells have yet to be successfully developed. Hence, further studies are needed to establish protocols that comply with good manufacturing practices; and it is necessary to determine the quality and therapeutic potential of cells after extensive expansion. For instance, detailed evaluations of the genetic stability, phenotype, differentiation potential, migration ability, and paracrine effects are suggested.

Second, it is necessary to develop new methods that can induce stable cartilage formation in vivo. The ability of grafted cells to maintain a chondrocyte phenotype and thus to produce hyaline cartilage-specific extracellular matrix is critical for articular cartilage repair. Traditional chondrogenic induction protocols often result in transient cartilage formation. To promote the therapeutic potential of graft cells, smart strategies that mimic the development of articular cartilage, such as a combination of mechanical stimulation and growth factors, are required to produce permanent articular cartilage.

Third, many questions need to be answered at the preclinical level, especially the safety and efficiency of these cell-based cartilage repair approaches. Although previous animal studies have shown improved outcomes after cell transplantation, the fate and repair mechanism of grafted progenitors remains largely unclear. A combination of stable cell labeling techniques and non-invasive cell tracking methods, such as magnetic resonance imaging, is proposed to monitor the fate of implanted cells for a long-term period.

Finally, high quality clinical trials are still missing. Although promising results have been shown in some pilot clinical cases, the clinical evidence is still very limited due to a small patient population and a short-term follow-up [99, 100]. Therefore, the efficiency of these cell-based treatments needs to be further confirmed by reliable clinical data from double-blind, controlled, prospective and multicenter studies with long-term follow-up, especially clinical studies comparing the aforementioned stem cell-based strategies with traditional treatments, such as the arthroscopic procedures.