Simultaneous regeneration of full-thickness cartilage and subchondral bone defects in vivo using a three-dimensional scaffold-free autologous construct derived from high-density bone marrow-derived mesenchymal stem cells

Repairing damaged cartilage, as in traumatic lesions of the joint surface or degenerative damage due to osteoarthritis, remains a clinical challenge because of the poor self-healing ability of cartilage [1]. Several existing treatments for the repair of damaged cartilage, including mosaicplasty [2], microfracture [3], and autologous chondrocyte implantation (ACI) [4], have been used successfully in patients to relieve pain and improve joint function. However, the long-term results of mosaicplasty and microfracture are unsatisfactory because the replaced tissue is not physiological hyaline cartilage, but mostly fibrous cartilage, which lacks suitable mechanical properties [5],[6]. Furthermore, in histological evaluations of the repaired tissue, ACI exhibited no significant difference compared to microfracture in a randomized trial [7], and the repaired tissue was predominantly fibrous cartilage.

Tissue engineering of functional articular cartilage could help address these issues. In recent years, several methods have been developed to repair full-thickness cartilage defects. Some of these methods use synthetic or biological scaffolds to achieve sufficient thickness, mechanical function, support cell attachment, migration, proliferation, and differentiation [8],[9]. However, these materials can influence the surrounding microenvironments [10] or cause immunological problems [11],[12].

Several scaffold-free systems have also been investigated [13]-[16]. Without using a scaffold, however, it is difficult to create enough thickness to fill in the defect. A thickness of 5 mm is necessary to fill a full-thickness articular cartilage defect in the knee. Furthermore, with or without using a scaffold, it is even more challenging for the current methods to regenerate both articular cartilage and subchondral bone in the context of an osteochondral defect, which sometimes occurs in trauma, osteochondritis dissecans, osteonecrosis, or osteoarthritis.

Adherent cells grown in suspension form spheroids (aggregates) in order to avoid cell death. We confirmed this phenomenon using bone marrow-derived mesenchymal stem cells (BM-MSCs). Furthermore, we developed a novel method for fusing spheroids derived from BM-MSCs to make a large construct without using a scaffold. Loading the spheroids into a molding chamber can produce a columnar structure consisting of fused spheroids; we named this structure "high-density mesenchymal stem cell, scaffold-free autologous construct" (HDMAC). We hypothesized that HDMACs could be used to regenerate both articular cartilage and subchondral bone in the context of an osteochondral defect. Furthermore, because many other reports have achieved successful short-term cartilage repair, we focused on achieving a relatively long-term result (up to 1 year) in an animal model.

The purposes of this study were to demonstrate our method for making spheroids and a scaffold-free construct (HDMAC) using only undifferentiated BM-MSCs and to evaluate the 1-year outcome of simultaneous regeneration of cartilage and subchondral bone in an osteochondral defect in rabbit knees.

In this study, we demonstrated successful regeneration of articular cartilage and subchondral bone in an osteochondral defect of rabbit knee joints using three-dimensional HDMACs. The HDMACs were made without using any differentiation agents, such as growth factors, and were implanted to the osteochondral defect while the cells in the construct were undifferentiated. The implanted MSCs spontaneously differentiated into bone or cartilage, depending on the environment in which they were implanted, and maintained the original border between bone and cartilage. Eventually, by the 52-week follow-up, the simultaneous regeneration of bone and cartilage had been completed. The unique features of our method are as follows: 1) A three-dimensional MSC construct with sufficient thickness for bone and cartilage repair can be made without a scaffold. 2) The construct is implanted without induction of differentiation. 3) Simultaneous regeneration of bone and cartilage occurs spontaneously by reproducing the differentiation process involved in bone and cartilage formation. A major strength of our study is that we could confirm that the regenerated bone and cartilage were maintained for up to 52 weeks.

However, one limitation of our study is that we did not make an effort to increase the sample sizes for the short-term results (e.g., within 12 weeks). After confirming in a few samples that MSCs spontaneously differentiate into bone or cartilage in vivo, as we originally hypothesized, we placed greater emphasis on investigating the long-term results. We chose this course because a large number of previous studies using other methods had succeeded over the short term, but it remained that deterioration of the repaired cartilage would occur over long-term observation. Some previous studies using a scaffold reported frequent invasion of a bony structure into the repaired cartilage, causing the cartilage to become thinner than the surrounding normal articular cartilage [9],[17]. Furthermore, a scaffold has the potential to influence surrounding microenvironments, which can affect the regional specification of implanted MSCs [10]. The presence of nonphysiological scaffolds could interfere with the physiological distribution of signal molecules, such as growth factors, that influence the differentiation of cells toward bone and cartilage.

Following a report of the efficacy of ACI by Brittberg et al. [4], chondrocytes have been widely used in the field of articular cartilage regeneration. The form of the grafts used for implantation varies from monolayer culture [4] to tissue-engineered three-dimensional cultures in agarose [18] or atelocollagen gel [19]. A number of clinical trials have been conducted, and some of them reported successful outcomes in regard to patients' symptoms or imaging endpoints [19],[20]. However, it should be noted that ACI was more effective in patients younger than 45 years than in older patients [21]. In fact, the mean age of clinical trials with satisfactory outcomes for relatively long follow-up periods ranges from 26.4 to 33.1 years old [19],[22]. This result may be related to the shortening of telomeres in many adult cells, including chondrocytes [23]. Therefore, it may be necessary to use alternative methods in middle-aged and elderly patients.

Bone and cartilage have the same origin, i.e., MSCs. In a previous work, patient age made no difference in outcomes of methods using MSCs [21]. Therefore, MSCs represent an attractive alternative to adult chondrocytes as a cell source for bone and cartilage regeneration. However, it is challenging to artificially control differentiation into the osteocyte or chondrocyte lineage after implantation. The underlying molecular mechanism of our method, in which a bone and articular cartilage defect was repaired while maintaining the border between cartilage and subchondral bone, is currently unclear. However, cell tracking analysis revealed that both cell types in the repaired tissues were derived dominantly from the implanted MSCs within the construct rather than the migrated cells from the surrounding tissues (Figure 5). Our HDMACs may mimic the cellular integration of embryonic osteogenesis and chondrogenesis that originates from mesenchymal condensation. The HDMACs were implanted without pretreatment to induce differentiation. The micrographs by SEM revealed that the HDMACs contained flattened cells, not ovary-shaped chondrocyte-like cells (Figure 2). Combined with the fact that the HDMACs were not stained with safranin O (Figure 1d), these cells are suggested to be undifferentiated. On the other hand, we also confirmed that the constructs could undergo differentiation into chondrocytes in vitro when cultured in chondrocyte induction medium (data not shown). In vivo, the implanted MSCs spontaneously differentiate into articular cartilage and subchondral bone; eventually, the normal structures of bone and cartilage were completely formed (Figures 6 and 7). In addition, type X collagen was not expressed in the implanted area. This suggests that hypertrophy of articular cartilage, one of the major problems using MSC-based constructs, did not occur during the regeneration period.

Our study is distinguished not only by the use of undifferentiated MSCs but also by the use of a construct generated by a scaffold-free method. Our HDMACs have enough volume to fill a full-thickness cartilage defect. Cell nutrition particularly in the middle of a big construct can be a concern [24]. However, the nucleus was stained with hematoxylin in the middle of the HDMACs and there was no fragmentation or chromatin condensation (Figure 1d). This suggests that the cells in the construct were viable. Previous studies developed various scaffold-free methods using chondrocytes or MSCs; the simplest reported method is injection of synovial MSCs into a cartilage defect in vivo[16]. In that study, however, although the histological score was significantly greater than the control, the depth of the defect was restricted to 2 mm and the surface was concave. Another method using a scaffold-free construct derived from synovial MSCs resulted in successful repair of a cartilage defect [14], but again, the depth was limited to 2 mm. In yet another study, scaffold-free chondrocyte sheets were used successfully to repair of a partial-thickness cartilage defect in vivo[13]. However, this method was insufficient for subchondral bone repair and tissue integration in the context of a full-thickness cartilage defect [25]. In efforts to obtain enough cell mass to fill a full-thickness cartilage defect, pellet cultures of chondrocytes or MSCs have been investigated. In one animal study, chondrocyte pellet implantation into a 3-mm osteochondral defect was attempted [15]. Although the thickness of the repaired cartilage remained relatively constant, overgrowth of the surface was observed, and subchondral bone regeneration was extremely limited.