Research ArticleTissue Engineering

Orchestrated biomechanical, structural, and biochemical stimuli for engineering anisotropic meniscus

See allHide authors and affiliations

Science Translational Medicine  10 Apr 2019:
Vol. 11, Issue 487, eaao0750
DOI: 10.1126/scitranslmed.aao0750

Engineering anisotropy

The meniscus is a fibrocartilage structure within a joint that helps reduce friction during joint movement. The outer and inner regions within the knee meniscus differ in cell types, extracellular matrix components, organization, and corresponding mechanical properties (anisotropy). Here, Zhang et al. used biomechanical stimulation and growth factor treatment during culture of mesenchymal stem cell–seeded polymer scaffolds to generate tissue constructs that mimic the native knee meniscus. The engineered meniscus constructs demonstrated long-term chondroprotection when implanted into the knees of rabbits. This study helps guide tissue engineering efforts to generate anisotropic constructs.

Abstract

Reconstruction of the anisotropic structure and proper function of the knee meniscus remains an important challenge to overcome, because the complexity of the zonal tissue organization in the meniscus has important roles in load bearing and shock absorption. Current tissue engineering solutions for meniscus reconstruction have failed to achieve and maintain the proper function in vivo because they have generated homogeneous tissues, leading to long-term joint degeneration. To address this challenge, we applied biomechanical and biochemical stimuli to mesenchymal stem cells seeded into a biomimetic scaffold to induce spatial regulation of fibrochondrocyte differentiation, resulting in physiological anisotropy in the engineered meniscus. Using a customized dynamic tension-compression loading system in conjunction with two growth factors, we induced zonal, layer-specific expression of type I and type II collagens with similar structure and function to those present in the native meniscus tissue. Engineered meniscus demonstrated long-term chondroprotection of the knee joint in a rabbit model. This study simultaneously applied biomechanical, biochemical, and structural cues to achieve anisotropic reconstruction of the meniscus, demonstrating the utility of anisotropic engineered meniscus for long-term knee chondroprotection in vivo.

INTRODUCTION

Over the past decade, the field of tissue engineering has made remarkable advances in the regeneration of damaged tissues through the manipulation of cells, biomaterials, and various stimuli (1, 2). However, the majority of bioengineering efforts fail to capture the anisotropic nature of physiological systems [heterogeneous populations of connective tissue cells and extracellular matrix (ECM)], most notably in fibrocartilage structures such as the knee meniscus, intervertebral disc, and temporomandibular joint disc.

As a heterogeneous fibrocartilage, the knee meniscus is characterized by a complex structural organization of outer and inner regions. In the outer region, fibroblast-like cells exist within an ECM that mainly comprises type I collagen (COL-1, 80 weight %), enabling resistance against tensile loads (3). Meanwhile, in the inner region, chondrocyte-like cells are embedded within an ECM that largely consists of type II collagen (COL-2) and glycosaminoglycans (GAGs), contributing to resistance against compression (4, 5). Thus, the anisotropic structure of the meniscus is directly related to its spatial biomechanics and chondroprotective functions. However, reconstruction of the native morphology and anisotropic structure of the meniscus remains a great challenge. Previous studies on meniscal engineering have revealed a paucity of viable cells (chondrocytes and stem/progenitor cells) that are capable of regenerating anisotropic tissue phenotypes from a single-cell source (6, 7), even in the presence of various growth factors (8). Furthermore, engineered menisci with homogeneous structures were unable to withstand tensile or compressive stresses, resulting in improper load bearing and shock absorption, and eventually in joint degeneration (9).

Because cellular differentiation requires spatial regulation by complex signals in the microenvironment [such as cell-cell contacts (10), matrix structures (11), and external stimuli (12)], only a few studies to date have achieved anisotropy in engineered meniscus tissues. Although recent studies using coculture seeding strategies (10) and electrospun fiber-aligned scaffolds (11) reported heterogeneous matrix deposition, these approaches led to poor performance in knee chondroprotection compared with the native meniscus in vivo. Therefore, we hypothesized that the reconstructed meniscus tissue should restore not only the structural properties, such as those of the cell and matrix, but also the biomechanical functions (2, 13).

Biomechanical stimulation has recently emerged as a viable strategy for fibrocartilaginous tissue repair, mimicking the complex biophysical environment of the native fibrocartilage and using defined regimens of physical conditioning such as compressive load (14, 15), hydrostatic pressure (1618), and tension (15, 19). Identification of specific mechanotransduction pathways that can spatially regulate stem cell fate may provide new clues toward engineering of meniscus tissue with anisotropic structures resembling those in the native tissue (20). For example, biomechanical load was shown to modulate the differentiation of native meniscus cells and mesenchymal stem cells (MSCs) (1517), and compressive loading led to increased chondrogenesis with elevated GAG and COL-2 synthesis (21, 22), whereas tensile loading resulted in fibrogenesis with elevated COL-1 synthesis and improved tensile strength (19, 23). Despite these encouraging reports, biomechanical stimulation alone failed to yield anisotropic properties in the stem cell–engineered meniscus, suggesting that spatial regulation of fibrochondrogenesis requires a well-orchestrated symphony of biochemical, structural, and biomechanical signals resembling the native microenvironment (24).

Here, we used a combinatorial approach of applying biomechanical and biochemical stimuli to bone marrow–derived MSCs seeded into a polymer scaffold to reconstruct the anisotropic structure and reestablish the proper function of the knee meniscus. Together with connective tissue growth factor (CTGF) and transforming growth factor–β3 (TGF-β3), which induced differentiation of MSCs into fibrochondrocytes, the orchestrated biomechanical and biochemical stimuli led to the generation of an anisotropic meniscus. Transplantation of the engineered meniscus into rabbit knees conferred long-term chondroprotection.

RESULTS

Scaffold fabrication

Using three-dimensional (3D) printing, we created wedge-shaped ring scaffolds of poly(ε-caprolactone) (PCL) (Fig. 1A) with a mean pore size of 215 μm, which our previous studies revealed led to improved proliferation, differentiation, and ECM production ability of endogenous and exogenous stem/progenitor cells seeded within the scaffolds (25). The scaffold was designed to match the size and adapt to the stress load of a rabbit meniscus for implantation (Fig. 1B). A customized dynamic biomechanical loading system was applied to exert compression-tension strains on the MSC-laden PCL scaffold with regional differences (Fig. 1C, a to c).

Fig. 1 Schematic diagrams for reconstruction of functional anisotropic meniscus.

(A) Flowchart of stem cell–based strategies for construction of a tissue-engineered meniscus with anisotropic structures. BMSCs, bone marrow–derived stem cells. (B) 3D-printed PCL scaffolds for implantation. (a) Photograph of rabbit knee during implantation; (b) 3D scaffold design model; (c and d) top and section views of the wedge-shaped circular PCL scaffold; (e and f) top views of the outer and inner regions of PCL scaffolds by scanning electron microscopy (SEM); (g and h) section views of the outer and inner regions of PCL scaffolds by SEM. (e) to (h) are higher magnifications of regions outlined in (c) and (d). (C) Biomechanical simulation. (a) Forces typically transduced on the knee meniscus in the human body; (b and c) applied loading forces across the meniscal construct; (d and e) calculated stress fields across the meniscal construct at 10% displacement of the loading platen: (d) von Mises stress distribution with a gradual decrease in stress from the internal rings to external rings; (e) compressive circumferential stress in the internal rings and tensile circumferential stress in the external rings.

Finite element analysis

To induce anisotropic fibrocartilaginous tissue repair, we designed a dynamic biomechanical loading system that could simultaneously generate regional compressive and tensile stresses. Using finite element (FE) analysis, we calculated that a minimum compressive load of 81.5 N applied to the biomimetic scaffold was required to move the loading platen down by 0.15 mm. The imposed von Mises stress on the meniscal construct was rotationally symmetrical around the scaffold ring structures and gradually decreased from the internal (0.48 to 0.85 MPa) to the external (0.12 to 0.36 MPa) rings (Fig. 1C, d and e). As the total displacement of the loading platen increased to 10%, the stress and strain patterns across the meniscal construct were primarily compressive, with partial stress/strain being translated into circumferential stress/strain and radial strain. Through this loading mode, both compressive circumferential stress in the internal rings (−0.23 to 0.41 MPa) and tensile circumferential stress in the external rings (0.04 to 0.14 MPa) were achieved, demonstrating the feasibility for simultaneous generation of regional compressive and tensile stresses with a simplified compressive loading step. Meanwhile, when the same load was applied to a circular scaffold with a homogeneous structure, the imposed von Mises stress on the construct was only observed in the contact areas and thus failed to generate regional compressive and tensile stresses (fig. S1).

In vitro evaluation of regional anisotropy in the meniscal scaffold

To reconstruct the native morphology and anisotropic organization of the meniscus, appropriate cues including signaling biomolecules and biomechanical stimuli need to be provided in the cellular microenvironment. TGF-β was shown to up-regulate fibrochondrogenic and chondrogenic gene expression (COL2A1, ACAN, and SOX9) upon biomechanical stimulation (22, 25), whereas CTGF up-regulated fibrochondrogenic and fibrogenic gene expression (COL1A1, FN1, and TNC) with appropriate biomechanical stimulation (19, 23). After seeding MSCs on the scaffolds, combined treatments with CTGF and TGF-β3 were applied for 4 weeks (biochemical stimulus). Dynamic compression-tension on the meniscus constructs was added for the last 2 weeks (biomechanical stimulus). The seeded MSCs in the group with both biochemical and biomechanical treatments (double stimuli) showed zonal differentiation into a fibrochondrocyte-like cell phenotype with synthesis of both COL-1 and COL-2, similar to the native meniscus (Fig. 2A). The other three groups did not exhibit zonal fibrochondrocyte differentiation of MSCs in the scaffold (Fig. 2B).

Fig. 2 Biochemical and biomechanical stimuli synergistically induce fibrochondrocyte differentiation of MSCs to build functional anisotropy after 4 weeks of in vitro culture.

(A) Zonal fibrochondrocyte differentiation of MSCs in 3D PCL scaffolds in the double-stimuli versus native meniscus (green, COL-1; red, COL-2). (B) Immunofluorescence observation of COL-1 (green) and COL-2 (red) deposition by MSCs treated in the following study groups: (i) growth medium as a control (no stimulus); (ii) CTGF and TGF-β3 for 4 weeks (biochemical stimulus); (iii) growth medium for 4 weeks with biomechanical stimulation during the last 2 weeks (biomechanical stimulus); and (iv) CTGF and TGF-β3 for 4 weeks with biomechanical stimulation during the last 2 weeks (double stimuli).

To evaluate the anisotropy, the ECM contents, including COL-1, COL-2, and GAGs, in the inner and outer regions of the meniscus construct were measured and compared. In the double-stimuli group, a relatively higher COL-1 content was detected in the outer region, whereas relatively higher COL-2 and GAG contents were detected in the inner region, resembling the ECM organization of the native meniscus (Fig. 3, A to C, and table S1). Only the double-stimuli group exhibited zonal variation (inner versus outer) in all three depositions (COL-1, COL-2, and GAG), indicating that the dual-stimuli treatment may produce a synergistic response. Furthermore, in the double-stimuli group, the expression of fibrogenic genes (COL1A1, FN1, and TNC) was higher in the outer zones and the expression of chondrogenic genes (COL2A1, ACAN, and SOX9) was higher in the inner zones, suggesting zone-specific mRNA phenotypes (Fig. 3D and table S2). Collectively, the double-stimuli treatment was sufficient to induce fibrochondrogenic differentiation of MSCs and the synthesis of zone-specific COL-1 and COL-2 in vitro.

Fig. 3 Biochemical analysis and gene expression of engineered meniscus after 4 weeks of in vitro culture.

(A to C) COL-1, COL-2, and GAG contents in the inner and outer regions of each study group. *P < 0.05 between the inner region and outer region in the same group; #P < 0.05 between the double-stimuli group and other groups in the same region. (D) Expression of fibrogenic genes (COL1A1, FN1, and TNC) and chondrogenic genes (COL2A1, ACAN, and SOX9) in MSCs in the inner and outer regions of each group. *P < 0.05 between the inner region and outer region in the same group; #P < 0.05 between the double-stimuli group and other groups in the same region (outer zone for fibrogenic genes; inner zone for chondrogenic genes). All data are presented as means ± SD (n = 6) and were analyzed by two-way analysis of variance (ANOVA) with Tukey’s test. The complete statistical analysis is shown in tables S1 and S2.

Anisotropic construct for regeneration of the knee meniscus in vivo

The efficacy of meniscus regeneration in vivo was evaluated by transplanting the anisotropic constructs into the knee joints of rabbits. Double-stimuli treatment (double stimuli), no-stimulus treatment (control), and native meniscus treatment (native) groups were chosen for in vivo evaluation. At 6, 12, and 24 weeks after implantation, whole joints along with regenerated tissues were collected for gross observations and histological and SEM analyses (Fig. 4 and fig. S2). To evaluate the presence of an inflammatory response, synovial fluid was collected for analysis of interleukin-6 (IL-6) and tumor necrosis factor–α (TNF-α). No obvious inflammatory reactions were observed at any time point after implantation (fig. S3). After 24 weeks, the regenerated meniscus in the double-stimuli group had a normal gross appearance with a shiny white color and a smooth surface (Fig. 4A). Analysis by the Gross Evaluation of Meniscus Implant Score revealed differences in the total score between the double-stimuli group and the control group from week 6 through week 24 (table S3). On immunofluorescence images, the double-stimuli group showed zone-specific COL-1 and COL-2 at 24 weeks, similar to the native rabbit meniscus (Fig. 4A).

Fig. 4 Regeneration of a knee meniscus resembling the native tissue after 24 weeks in vivo.

(A) Gross view and low-magnification immunofluorescence (IF) images of native or regenerated menisci at 24 weeks after in vivo implantation in rabbit knees. Green, COL-1; red, COL-2. (B and C) Zone-specific matrix phenotype analysis in engineered versus native tissue. Tissue sections were stained by immunohistochemistry for COL-1 and COL-2 (B) or with toluidine blue (TB) for proteoglycans and picrosirius red (PR) for COL-1 and COL-3 (C). (D) Zone-specific cell phenotypes in the regenerated meniscus [hematoxylin and eosin (H&E) staining]. Fibroblast-like and chondrocyte-like cells were both observed in the implants in the double-stimuli group and native group. (E) SEM images of regional variations in the ultrastructure of the implants in the double-stimuli group and native group. A total of six replicates were tested, with representative images selected from the same construct.

Histological evaluation of the regenerated meniscus in the double-stimuli group at 24 weeks revealed zone-specific matrix phenotypes in the regenerated meniscus that resembled those in the native tissue (native group). The outer zone exhibited an aligned fibrous matrix dominated by COL-1, whereas the inner zone presented a cartilaginous matrix containing COL-2 and proteoglycans (Fig. 4, B and C). The control group showed amorphous fibrous tissue throughout the construct and lacked any zone-specific tissue phenotypes. Because the PCL scaffolds were dissolved by the organic solvents used during the staining procedure, holes in the regenerated tissue indicated the residual scaffolds (Fig. 4B, control group). The relative density of the integrated optical density (IOD)/area value in each zone of the regenerated meniscus was used for semiquantitative analysis of the distribution of collagen types by immunohistochemistry. In the double-stimuli group and the native group at week 24, the relative density of the IOD/area value for COL-1 in the outer zone was greater than that in the inner zone, whereas relative density of the IOD/area value for COL-2 was greater in the inner zone, suggesting the formation of a heterogeneous tissue at 24 weeks after implantation (fig. S4 and table S4).

Similar to the anisotropic organization of the ECM, the regenerated meniscus tissue in the double-stimuli group also exhibited zone-specific cell phenotypes as shown by the meniscus histology, whereas the control group did not show differential cellular organization (table S5). At 24 weeks, abundant round-shaped chondrocyte-like cells appeared around cartilage islands in the inner zone, whereas the outer zone was populated by fusiform-shaped fibroblast-like cells, similar to the appearance of the native meniscus (Fig. 4D). In the control group, many spindle-shaped fibroblast cells with elongated nuclei were surrounded by fibrous tissue across the entire implant.

SEM images confirmed regional variations in the ultrastructure (Fig. 4E). In the double-stimuli group, a contiguous fiber network surrounding circumferential fiber bundles was visible in the outer region, whereas the inner region showed small pore-like structures with fibrochondrocytes in lacunas, similar to the native meniscus (native group). In contrast, loose collagen fibers were randomly oriented in all regions with no anisotropy in the control group.

Biomechanical evaluation of meniscus constructs and regenerated knee meniscus

The biomechanical properties of the in vitro meniscus constructs were assessed before in vivo implantation. The double-stimuli group showed higher tensile modulus and greater ultimate tensile strength (UTS) than the other groups (Fig. 5A and table S6). The double-stimuli group also had a greater aggregate modulus in the confined compressive creep test, despite a lack of difference in permeability. For comparison, the biomechanical properties of the meniscus constructs were assessed after in vivo implantation for 24 weeks. Although the values in the native group were higher than those in both the double-stimuli group and control group at 6 weeks, the UTS and tensile modulus in the double-stimuli group had improved by 12 and 24 weeks, respectively, to reach the values for the native meniscus, with no remarkable differences between the two groups (Fig. 5F, fig. S5, and tables S7 to S11). The double-stimuli group consistently outperformed the control group in tensile strength at all time points. The confined compressive creep test showed that the implants in both the double-stimuli group and control group were 20 to 50% as stiff as those in the native group at week 6, although the double-stimuli group outperformed the control group in the aggregate modulus value at all time points. By week 12, there was no difference in aggregate modulus between the double-stimuli group and the native group. The permeability of the scaffolds in the control group was reduced just after implantation, but there were no differences in permeability for each group after 12 weeks (fig. S5).

Fig. 5 Biochemical properties of the regenerated knee meniscus after 4 weeks in vitro or 24 weeks in vivo.

(A) Biomechanical properties of the in vitro construct, including bulk tensile modulus, aggregate modulus, UTS, and permeability, after 4 weeks of culture. *P < 0.05 between the double-stimuli group and other groups. All data are presented as means ± SD (n = 6) and were analyzed by one-way ANOVA. The complete statistical analysis is shown in table S6. (B to E) Biomechanical properties of the in vitro construct in bidirectional tensile testing and compressive testing in different locations on the tissue constructs. (B) Circumferential tensile modulus. +P < 0.05 between the inner region and outer region in the same group; *P < 0.05 between the double-stimuli group and other groups in the inner zone; #P < 0.05 between the double-stimuli group and other groups in the outer zone. All data are means ± SD (n = 6) and were analyzed by two-way ANOVA with Tukey’s test. The complete statistical analysis is shown in table S8. (C) Radial tensile modulus. *P < 0.05 between the double-stimuli group and other groups. All data are means ± SD (n = 6) and were analyzed by one-way ANOVA with Tukey’s test. The complete statistical analysis is shown in table S9. (D and E) Reduced modulus (D) and hardness (E). +P < 0.05 between the inner region and outer region in the same group; *P < 0.05 between the double-stimuli group and other groups in the inner zone; #P < 0.05 between the double-stimuli group and other groups in the outer zone. All data are means ± SD (n = 6) and were analyzed by two-way ANOVA with Tukey’s test. The complete statistical analysis is shown in tables S10 and S11. (F) Biomechanical properties of the regenerated meniscus 24 weeks after in vivo implantation. *P < 0.05 between the control group and other groups. All data are means ± SD (n = 6) and were analyzed by two-way ANOVA with Tukey’s test. The biomechanical properties of the regenerated meniscus at weeks 6 and 12 are shown in fig. S5. The complete statistical analysis is shown in table S7. (G to J) Biomechanical properties for the double-stimuli group in bidirectional tensile testing and compressive testing in different locations on the tissue constructs at 24 weeks after in vivo implantation. (G) Circumferential tensile modulus. +P < 0.05 between the inner region and outer region in the same group; *P < 0.05 between the native or double-stimuli group and control group in the inner zone; #P < 0.05 between the native or double-stimuli group and control group in the outer zone. All data are means ± SD (n = 6) and were analyzed by two-way ANOVA with Tukey’s test. The complete statistical analysis is shown in table S12. (H) Radial tensile modulus. *P < 0.05 between the native or double-stimuli group and control group. All data are means ± SD (n = 6) and were analyzed by one-way ANOVA with Tukey’s test. The complete statistical analysis is shown in table S13. (I and J) Reduced modulus (I) and hardness (J). +P < 0.05 between the inner region and outer region in the same group; *P < 0.05 between the native or double-stimuli group and control group in the inner zone; #P < 0.05 between the native or double-stimuli group and control group in the outer zone. All data are means ± SD (n = 6), and were analyzed by two-way ANOVA with Tukey’s test. The complete statistical analysis is shown in tables S14 and S15.

To verify the heterogeneity of the mechanics, we performed bidirectional tensile testing in the radial and circumferential directions, as well as compressive testing at different anatomic locations for the in vitro 4-week construct and in vivo 24-week tissue-engineered meniscus (Fig. 5, B to E and G to J, and tables S8 to S11 and S12 to S15, respectively). For the in vitro 4-week construct, the double-stimuli group showed anisotropy in circumferential tensile modulus, reduced modulus, and hardness. In addition, the double-stimuli group had the highest values in bidirectional tensile and compressive testing. For the in vivo 24-week tissue-engineered meniscus, both the double-stimuli group and native group presented zonal variations (inner versus outer) in circumferential tensile modulus, reduced modulus, and hardness. Moreover, both groups had higher values in bidirectional tensile testing and compressive testing than the control group. All of these results clearly demonstrate that the MSC-seeded constructs with dual-stimuli treatment showed restored functional anisotropy at 24 weeks after in vivo implantation, as well as biomechanical properties of the meniscus.

Evaluation of joint cartilage degradation

According to the scoring system issued by the International Cartilage Repair Society (ICRS), the worst cartilage damage was grossly visible in the control group (Fig. 6, A and D, and fig. S6, A and D). For the Mankin scoring system based on histological observations (Fig. 6, B and E, and fig. S6, B and E), the control group exhibited diffuse chondrocyte clones with markedly lower intensity of TB staining in the FC and TP at 24 weeks. The cartilage damage worsened over time, with the control group maintaining a higher Mankin score compared with the other groups at all time points examined. In addition, although obvious differences in the ICRS and Mankin scores were observed between the double-stimuli group and native group, the double-stimuli group exhibited only minor cartilage surface irregularities under microscopic observation at 24 weeks.

Fig. 6 Gross and microscopic observations of the recovered joints 24 weeks after implantation.

(A) Macroscopic observations of the femoral condyle (FC) and tibial plateau (TP). (B) H&E and TB staining of articular cartilage surfaces in the FC and TP. (C) SEM images of articular cartilage surfaces in the FC and TP. (D) ICRS and (E) Mankin scores assessing cartilage degeneration. *P < 0.05 between the indicated groups. The gross and microscopic observations of the articular cartilage and the scores at weeks 6 and 12 are shown in fig. S6. All data are means ± SD (n = 6) and were analyzed by two-way ANOVA with Tukey’s test.

The microscopic changes in the articular cartilage surfaces were further assessed by SEM (Fig. 6C and fig. S6C). At 24 weeks, the surfaces in the double-stimuli group exhibited no superficial splits and had a minor hillocky appearance arising from synovial fluid crystals. In contrast, the surfaces in the control group showed superficial splits and small amounts of uniform fibers. In the native group, the FC and TP presented smooth surfaces. Thus, both the microscopic and gross observations revealed a chondroprotective effect of the regenerated meniscus in the double-stimuli group.

DISCUSSION

The purpose of this study was to explore the synergistic effect of dynamic compressive-tensile loading and growth factors in inducing the differentiation of MSCs seeded onto a biomaterial scaffold to reconstruct the anisotropic organization of a tissue-engineered fibrocartilage, the knee meniscus. The biomechanical and biological properties of the constructs were evaluated in vitro for 4 weeks and in vivo for 24 weeks. The dual-stimuli treatment was confirmed to successfully enable meniscal fibrocartilage tissue regeneration with anisotropic properties and functional recovery on par with those of the native meniscus. The results highlight the potency of orchestrated biomechanical, structural, and biochemical stimuli for engineering of regenerated anisotropic tissue.

Although the connective tissues of the knee joint are known to derive their biomechanical properties from their biochemical constituents, the precise structure-function relationships remain elusive (24, 26). In theory, the meniscus tissue is hypocellular and has zone-specific collagens and GAGs that are responsible for the tensile and compressive integrity of the fibrocartilaginous tissue (4). The type and magnitude of the physiological loads often determine the extent of anisotropy (20), such that load transmission is not only crucial for proper functioning of musculoskeletal tissues but also directly involved in their development and maintenance (27). Compressive load is generally important for chondrogenic differentiation, leading to elevated COL-2 and GAG expression and improved compressive modulus (21, 22), whereas tensile load generally increases fibrogenesis, resulting in elevated COL-1 expression and improved tensile strength (19, 23). Neither type of biomechanical stimulation alone can fully restore the fibrocartilage characteristics of meniscal tissue. Therefore, in the present study, a loading pattern simulating the full range of the knee environment was developed to simultaneously apply tensile and compressive stimuli to a scaffold. Because the native fibrocartilage tissue supports a wide range of compressive, tensile, and shear stresses, recapitulation of its anisotropic composition and spatial tissue structures is critical for the maintenance of these diverse biomechanical functions. The present study revealed an effective impact of the anisotropic structure and composition of tissue-engineered fibrocartilage on its biomechanical properties in both the in vitro scaffold and in vivo regenerated meniscus, particularly in the double-stimuli group.

Previous studies have described the effects of dynamic loading alone in the absence of growth factors on MSC differentiation (21). Our in vitro analyses confirmed that the biomechanical stimulus group failed to show substantial collagen and GAG contents, suggesting that a suitable substitute for biomechanically induced differentiation in the absence of biochemical factors may not currently exist. Previous studies showed that compressive or tensile loading in the presence of TGF-β superfamily members can synergistically enhance MSC chondrogenic or fibrochondrogenic differentiation, resulting in remarkable collagen and proteoglycan expression (19, 21, 23, 28). However, none of these methods can replicate the heterogeneous and anisotropic properties of native fibrocartilage tissue. In our study, the double-stimuli group showed fibrochondrocyte differentiation of MSCs and deposition of zone-specific COL-1 and COL-2 in vitro. Compared with the control group, the ECM in the double-stimuli group showed differences in the outer and inner zones that could be explained by a synergistic effect of the biomechanical and biochemical stimuli. Upon statistical analysis, a significant synergistic effect was determined between the biochemical and biomechanical stimuli. A previous study confirmed the regulatory effect of biomechanical stimulation on cell differentiation and the amplification of biomechanical stimuli by chemical stimuli (15). The mechanism may be related to the regulation of nutrient distribution and the upgrading of a specific signal pathway by respective biomechanical stimuli. The rationale for inclusion of CTGF in neo-tissue engineering was based on its regulatory role in the expression of fibrochondrogenic or fibrogenic genes (COL1A1, FN1, and TNC) under biomechanical stimulation (26, 2931). Our biochemical and reverse transcription polymerase chain reaction (RT-PCR) assays revealed that MSCs treated with dual stimuli underwent fibrochondrogenic differentiation with zone-specific phenotypes at both the mRNA and ECM levels. Smad and p38 pathways were previously shown to be involved in collagen and aggrecan synthesis in the presence of biomechanical stimulation (22). Further signaling studies should be conducted to identify the factors that enhanced the fibrocartilage regeneration in the present study.

To evaluate the anisotropic structure of the resulting fibrocartilage and the chondroprotective function for articular cartilage, we followed up our in vitro examination of double stimuli–induced fibrochondrogenesis with long-term in vivo evaluation of scaffold implantation and meniscus regeneration. The double-stimuli group showed a consistent yield of fibrochondrocytes with zone-specific matrix phenotypes in the regenerated tissue at 24 weeks. With their high incidence in knee injury, meniscus lesions are a major predisposing factor for osteoarthritis (32), and total or partial meniscectomy is performed for over 1.5 million people in the United States and Europe every year (33). However, restoration of function and chondroprotection in the tissue-engineered meniscus poses a challenge. Some studies have used exogenous biological factors to investigate partial healing with homogeneous tissue constructs in animal models (34, 35), but there are few reports on restoration of function or regeneration of anisotropic meniscal tissue phenotypes (26). In the present study, the functional anisotropy of the reconstructed meniscus was confirmed by chondroprotection analysis, with the results suggesting profound clinical implications for knee stability and health.

In this study, postimplantation maturation was indicated by an increase in ECM staining. In accordance with the theory for the biomechanical microenvironment of fibrocartilage tissue (3, 4), the rounded shape of cells in the inner region indicated a response to compressive strain and demonstrated that compressive loading induced chondrogenesis in endogenous and exogenous stem/progenitor cells (21). In contrast, the fibroblast-like cells around the scaffold in the outer region appeared elongated, suggesting that tensile strain applied circumferentially and radially may have stimulated stem/progenitor cells to differentiate into fibroblasts and produce COL-1 (23, 36). Although the precise mechanisms and signaling pathways for the differentiation of endogenous or exogenous stem/progenitor cells in vivo were not investigated in the present study, we hypothesize that the anisotropic fibrocartilage construct produced in vitro may have been prompted to undergo further maturation in the in vivo environment after implantation.

The double-stimuli group showed not only reconstitution of the biological structure of the meniscus at the cellular and ECM levels but also restoration of the biomechanical properties and long-term stabilization of the knee joint. The ICRS and Mankin scores both indicated a chondroprotective effect of the double-stimuli treatment, and no notable cartilage degradation was observed. We hypothesize that the heterogeneous and anisotropic organization of the implanted constructs made an important contribution to long-term improvement of the knee joints. The increased biomechanical properties of the dual stimuli–treated tissue-engineered meniscus, such as improved tensile strength, tensile modulus, and compressive strength, likely played a key role in enabling joint load distribution on par with that of the native meniscus. Although the initial implants at 6 or 12 weeks did not yield biomechanical strengths similar to those of the native rabbit meniscus, the double-stimuli group at 24 weeks showed fully restored biomechanical properties. These enhanced biomechanical properties may be primarily attributed to improved tissue ingrowth (25, 37), because the fibers in the polymer scaffold may have degraded during the early stages of implantation (25, 38), resulting in decreased biomechanical properties during the initial stages in all treatment groups, whereas the regenerated tissue at the later stages of implantation may have supported a larger proportion of the circumferential and vertical stresses. The tensile and compressive properties were better in the double-stimuli group compared with the control group, probably through tissue deposition and anisotropic organization with heterogeneous deposition of collagens and proteoglycans. The control group showed poorer biomechanical values and lower ICRS/Mankin scores, suggesting that poor ECM deposition and homogeneity of the regenerated tissue translate to poor biomechanical functions for chondroprotection.

Despite these results, the structural and functional properties of the PCL-based constructs were still slightly inferior to those of the native meniscus. Some parameters such as the scaffold material and cell seeding conditions may require further evaluation to improve the structure and mechanics of the regenerated meniscus. In addition, we expect that the degradation of the PCL scaffold can be extended to 9 to 12 months. In this way, with tissue ingrowth, an adequate ECM can maintain biomechanical support. Finally, the PCL scaffold could degrade and become remodeled in the tissue-engineered meniscus. Although PCL has been used for meniscus regeneration and shows excellent biomechanical intensity and biocompatibility (39), the compliance for tensile and compressive stresses is not adequate. Although hydrogels do not generally have sufficient biomechanical strength for meniscus tissue engineering, certain high-strength formulations may be suitable for regenerating the heterogeneous meniscus (40). In addition, the microenvironmental signals that regulate spatial differentiation could be further optimized with the proper type and dosage of biochemical or biomechanical stimuli.

The major limitation of the present study is that we could not follow the fate of the endogenous or exogenous stem/progenitor cells within the knee joint in vivo, which would serve to interpret the mechanisms underlying the cell proliferation and differentiation. Although tracking of cell migration and differentiation is challenging, future studies may utilize green fluorescent protein or other genetically encoded labeling methods to achieve this goal (35). We also defined the inner and outer regions of the meniscus as a measure of anisotropy, but more stringent models may be needed to consider the gradual spatial transition of the heterogeneous phenotypes and ECM compositions of the native meniscus. Furthermore, owing to the limited size of the rabbit meniscus, it was impossible to obtain two samples from one meniscus for creep compressive testing on the inner and outer regions. Thus, we used nanoindentation to test the compressive properties in the different anatomic locations (inner and outer regions). Large animal models (sheep, goat, or primate) may be more relevant to future studies with respect to the development of clinical treatments.

In this study, we demonstrated that a tissue-engineered meniscus seeded with MSCs under biomechanical and biochemical dual stimuli in vitro could promote functional anisotropy and support a chondroprotective effect. The simplicity and efficiency of this strategy applying spatial, biomechanical, and biochemical cues may have profound implications in the field of fibrocartilaginous tissue engineering in general, leading to breakthroughs in the regeneration of the tendon-bone junction, intervertebral disc of the spine, and temporomandibular joint. Fundamentally, this study furthers our understanding of engineering anisotropy in functional biomaterials through the strategic application of biomechanical and biochemical stimuli.

MATERIALS AND METHODS

Study design

The overall objective of the study was to determine whether combined biomechanical and biochemical stimulations (double stimuli) would offer benefits in heterogeneous meniscus regeneration and protective effects on the articular surface. The efficacy for rebuilding functional anisotropy in vitro was systematically investigated, including the following: (i) ECM synthesis and gene expression of GAGs, COL-1, and COL-2, as the primary macromolecules in the native meniscus; (ii) fibrochondrogenic differentiation of MSCs under biomechanical or biochemical stimulation; and (iii) biomechanical properties of regenerated tissues.

The ability of the heterogeneous tissue to perform articular cartilage protection was further investigated in vivo using a rabbit model. A power analysis was performed before the study as described in the “Statistical analysis” section. At each time point, 12 knees per group were collected for histological and biomechanical analyses. A total of 108 rabbits were operated on, and all 108 reached the endpoint. No outliers were eliminated in the study. The treatment was randomized and blinded. The endpoints were predefined as the presence or absence of significant differences in the parameters listed above. The protocol was approved by the local Institutional Animal Care and Use Committee (IACUC) and complied with the Guide for the Care and Use of Laboratory Animals published by the National Academy Press (National Institutes of Health Publication No. 85-23, revised 1996). Primary data are reported in data file S1.

Fabrication of 3D PCL scaffolds

A wedge-shaped circular disk was fabricated by fused deposition modeling as described in our previous reports (25, 41). Briefly, the medial meniscus of skeletally mature male rabbits was captured by magnetic resonance imaging (3.0 T Signa HDxt; GE Healthcare) at high resolution (512 × 512–pixel bitmap image; 0.31-mm resolution), followed by calculation of its size and reconstruction by computer-aided design. For biomechanical stimulation in vitro, a wedge-shaped circular disk was selected as a typical model (Fig. 1B, b). PCL (number-average molecular weight, 74,600 g mol−1; melting point, 52.9°C; Changchun SinoBiomaterials Co. Ltd.) was softened and dispensed, according to the layer path directed by the 3D design for the internal microstructures. For example, the first layer was built by projecting the first pattern from the outer region to the inner region (Fig. 1B, f). After the first layer was fabricated, a radial alignment was projected on the surface, and the second layer was a circumferential alignment (Fig. 1B, e). For the fabrication process, the width of the upper layer was controlled to be less than the previous layer to ensure that the outer region was thicker than the inner region and the cross section was wedge-shaped (Fig. 1B, d, g, and h). Movement in the z direction was controlled at 200 μm, and the road width was approximately 300 to 500 μm. The heterogeneous structure with the three types of alignments was successfully fabricated by a 3D printing process (Fig. 1B, c). The mean pore size of the scaffold was about 215 μm.

Bone marrow–derived MSC harvesting, culture, and seeding

Bone marrow–derived MSCs were isolated from 3-month-old New Zealand white rabbits. Isolation, culture, trilineage differentiation potential assay, and immunophenotypic identification of MSCs were performed as previously described (25, 42). MSCs in the third passage were seeded onto a scaffold (5 × 106 cells per scaffold) using a centrifugal method (43).

Biochemical stimulation

To investigate the single or synergistic effect of biochemical and biomechanical cues on the differentiation of MSCs, cellular scaffolds were treated according to the following study groups: (i) growth medium [α-minimum essential medium (α-MEM); Gibco BRL Co. Ltd.] as a control (no stimulus) and (ii) CTGF (100 ng/ml; PeproTech) and TGF-β3 (10 ng/ml; PeproTech) for 4 weeks (biochemical stimulus). Fibrogenic induction supplement [ascorbic acid (50 μg/ml)] and chondrogenic induction supplement [0.1 μM dexamethasone, sodium pyruvate (100 μg/ml), l–ascorbic acid 2-phosphate (50 μg/ml), l-proline (40 μg/ml), and 1% 1× insulin transferrin selenium (ITS)] were included in the CTGF and TGF-β3 treatments, respectively (26, 31).

Biomechanical stimulation

Dynamic biomechanical loading was applied during the last 2 weeks of culture to constitute the following study groups: (iii) cellular scaffolds cultured in α-MEM with biomechanical stimulation for the last 2 weeks (biomechanical stimulus); (iv) continuous CTGF and TGF-β3 for 4 weeks with biomechanical stimulation during the last 2 weeks (double stimuli). For biomechanical stimulation, a custom-built bioreactor was constructed with a stepper motor driver and a computer controller to apply cyclic sinusoidal deformations to the constructs during in vitro culture (fig. S7A).

The platens were fabricated on the basis of the previous model (15). Before loading, an FE analysis was performed to determine an appropriate protocol by analyzing local strain fields in the in vitro constructs during loading. A linear elastic FE model with two contact components, loading platen and meniscal construct, was established by ABAQUS (Dassault Systèmes). The loading platen was modeled as a rigid body. The meniscus was modeled as linear elastic PCL with a Young’s modulus of 40 MPa and a Poisson’s ratio of 0.35. The values for the biomechanical properties of the PCL scaffold are shown in fig. S7 (B and C). The stress and strain field distributions in the meniscal construct were analyzed with axial displacement up to 1.5 mm (10% of the outer wall height of the meniscal construct). A previous report indicated that 10% compressive strain was the most successful loading regimen and matched the native meniscus (44).

In the in vitro experiment with the customized designed bioreactor, dynamic loading was simulated by moving the loading platens in the axial direction according to a sinusoidal loading pattern: Amplitude·(cos(2πt) − 1). Biomechanical stimulation was applied after a 14-day period of unloaded static culture in the presence of growth factors. The frequency of the dynamic load was 1 Hz, and the stimulation was applied for 1 hour/day and 5 days/week. The above protocol was selected by reference to previous reports describing positive effects on the synthesis of fibrocartilage/cartilage tissue in vitro (15, 22, 28, 45). After 4 weeks of culture, the constructs were collected and subjected to in vitro analysis or in vivo implantation (Fig. 1B, a).

Biochemical analysis

Samples were digested in a prepared papain solution [papain (1 mg/ml), 0.5 M EDTA, and 0.05 M cysteine-HCl] at 60°C overnight. The COL-1/COL-2 and GAG contents were quantitatively assayed using an enzyme-linked immunosorbent assay (ELISA) kit (Rabbit COL-1/COL-2 ELISA Kit, Chondrex Inc.) and 1,9-dimethylmethylene blue dye binding (ZSGB-BIO Inc.), respectively. All reagents were used according to the manufacturer’s protocol.

Gene expression analysis

Total RNA was extracted using TRIzol reagent (Invitrogen). Isolated RNA was reverse-transcribed and subjected to real-time PCR using a StepOnePlus Real-Time PCR System (Applied Biosystems). Commercial primers for COL1A1, COL2A1, FN1, TNC, ACAN, and SOX9 were used. The sequences of the primers are shown in table S16. The amount of GAPDH expression was evaluated as an internal control. The relative expression of the target genes was calculated by the ΔΔCt method.

Immunofluorescence and immunohistochemistry analyses

For immunofluorescence, the scaffolds were fixed, permeabilized, and blocked with 10% (v/v) fetal bovine serum in phosphate-buffered saline at 37°C for 60 min. The specimens were then incubated with primary antibodies against COL-1 (ab90395; Abcam; 1:10,000 dilution) and COL-2 (CP18; Calbiochem; 1:1000 dilution) for 24 hours, followed by Alexa Fluor 488 or 594 Goat Anti-Mouse IgG Antibody ReadyProbes Reagent (Invitrogen, USA) for 1 hour at room temperature. Cell nuclei were counterstained with Hoechst 33258 (2.0 μg/ml). The samples were viewed under a TCS-SP8 confocal microscope (Leica).

For immunohistochemistry, paraffin sections (6 μm) were treated for antigen retrieval and incubated overnight at 4°C with primary antibodies against COL-1 (ab90395; Abcam; 1:100 dilution) and COL-2 (CP18; Calbiochem; 1:200 dilution), followed by a goat anti-mouse secondary antibody (ZSGB-BIO Inc.) for 2 hours at room temperature. The diaminobenzidine (DAB) substrate system was applied for color development. Cell nuclei were stained with hematoxylin. Initially, serial dilutions of the primary antibodies were evaluated on the native meniscus to delineate the structural anisotropy. The meniscal tissue displayed differences between the inner and outer zones at the appropriate dilutions. After this step was optimized, the same antibody dilutions were applied to the experimental samples. Tendon (COL-1) and articular cartilage (COL-2) samples from rabbits were used as positive and negative controls (fig. S8).

Surgical procedure

The implantation procedure was performed as previously described (25, 46). A total of 108 skeletally mature male New Zealand white rabbits (6 months) weighing 3.0 kg were randomly allocated for animal experiments. All treatments were blinded from the members who performed the operations and postsurgical care. All animals underwent medial total meniscal meniscectomy of the right knee and were randomly divided into three groups. Overall, 36 knees received implantation of heterogeneous tissue (double stimuli), 36 knees received implantation of cellular scaffolds without stimulation (control), and 36 knees received implantation of native menisci (native). Twelve rabbits from each group were randomly selected for euthanasia at 6, 12, and 24 weeks. At each time point, six knees were collected for histological evaluation of implant regeneration and articular cartilage degeneration, and six knees were analyzed for implant biomechanical testing and microscopic observation of the cartilage and implant surface.

Briefly, the operation was performed under general anesthesia combined with local anesthesia by intravenous injection of sodium pentobarbital solution (30 mg/kg) and subcutaneous injection of 1% lidocaine (3 ml). The knee of the rabbit was approached through a medial parapatellar incision. A total meniscectomy was performed by resecting the medial meniscus sharply along the periphery and detaching it from its anterior and posterior junctions without injuring the medial collateral ligament. The anterior horns and periphery of the scaffold were attached to the respective root attachments and appropriate adjacent synovium with absorbable No. 4-0 sutures (Ethicon). Because the posterior root attachment of the medial meniscus was adjacent to the posterior cruciate ligament, we used a self-made threading apparatus and extracapsular knotting technique to fix the posterior horn of the scaffold with ligamentous structures (fig. S9). For the native group, the total medial meniscus was resutured in situ. After the operation, all animals were returned to their cages and allowed unlimited movement. The animals were given antibiotic prophylaxis and pain-killing medication (narcotics and nonsteroidal anti-inflammatory drugs) as necessary. After the rabbits had achieved normal gait movement with no signs of infection (2 to 3 weeks postoperatively), they were transferred to an IACUC-approved farm for unrestricted movement and exercise.

Evaluation of implants

At each time point, synovial fluid was collected by aspiration without dilution for assessment of IL-6 and TNF-α as previously reported (7). The supernatants were assayed for IL-6 and TNF-α using standard ELISA kits (Rabbit IL-6 ELISA Kit and Rabbit TNF-α ELISA Kit, respectively; Hermes Criterion Biotechnology).

The meniscus in each joint was grossly evaluated by the Gross Evaluation of Meniscus Implant Score (6). After dissection and fixation, the samples were cut to produce blocks that exposed the wedge-shaped profile and showed the inner and outer zones of the regenerated meniscus in histological sections. The specimens were then dehydrated and embedded in paraffin. Sections were stained with H&E, TB (positive for proteoglycans), and picrosirius red (to distinguish COL-1 and COL-3). The procedures for labeling COL-1 and COL-2 were described above. The implant sections were evaluated blindly according to a meniscus histology scoring system (6). In particular, lymphocytes and neutrophils were stained with H&E and identified in the histological sections. Descriptors for the relative density of the IOD/area value were analyzed for semiquantification of COL-1 and COL-2 deposition as previously reported (47). Briefly, the wedge-shaped meniscus in histological sections was divided into equal inner and outer regions as areas of interest (AOIs). Next, 10 digital images at 1600 × 1200–pixel resolution and ×400 magnification were captured with a DP21 charge-coupled device camera (Olympus) coupled to a BH2 microscope (Olympus) for each AOI. The measurement parameters included the IOD, area sum, and mean density. The OD was calibrated and the AOI was set using the following parameters: hue, 0 to 30; saturation, 0 to 255; intensity, 0 to 255. The image was converted to grayscale, and the values were counted. The positive and negative controls were described above.

For ultrastructure evaluation, each engineered construct was processed for cryofracturing to create a cut surface in the wedge section and observed with a JSM5600LV scanning electron microscope (JEOL USA Inc.).

Evaluation of joint cartilages

The cartilages of the medial FC and TP were grossly evaluated according to the criteria of the ICRS cartilage lesion classification system (48). For histological evaluation, the osteochondral specimens were sectioned on the coronal plane at the midpoint of the TP and on the sagittal plane at the midpoint of the medial FC as previously described (46). The sections were stained with H&E and TB and graded blindly according to the Mankin grading system. The cartilage surfaces were also observed by SEM.

Biomechanical analysis

The biomechanical properties of the scaffold in vitro or regenerated meniscus were assessed using a materials testing machine (AG-IS; Shimadzu) as previously reported (49, 50). Bidirectional tensile testing was performed with uniaxial tests in the radial and circumferential directions within the samples. In addition, each sample (1 mm thick) was cut into a rectangular shape in the designated regions, leaving a 3-mm gauge length for testing (fig. S10). In bulk tensile testing, the posterior portion of the regenerated meniscus or scaffold in vitro was used. Uniaxial tensile testing was performed on the samples in the circumferential direction as previously described (37). The samples were tested to failure at a rate of 0.06 mm/s. The elastic modulus was analyzed from the linear portion of the stress-strain curve. In confined compressive testing, a cylindrical sample of 2-mm diameter and 1-mm thickness from the anterior portion of the regenerated meniscus or scaffold in vitro was used (fig. S10). The creep response of the specimen under a step force (0.02 N) was monitored until reaching equilibrium (defined as slope < 1 × 10−6 mm/s, for at least 7200 s). The test force was then automatically removed, and the recovery phase was monitored until equilibrium was reached (defined as slope < 1 × 10−6 mm/s, for at least 3600 s). The aggregate modulus and permeability were calculated using the Mow biphasic theory (51). For the empty PCL scaffold, the samples were compressed at a constant loading rate of 0.1 mm/min. The elastic modulus was calculated from the linear portion of the stress-strain curve (25).

Compressive testing in the different anatomic locations (inner or outer regions) in the tissue-engineered meniscus was also performed by nanoindentation. Briefly, samples were isolated from the inner or outer regions of the regenerated tissue. All indentations were analyzed using a TriboIndenter (Hysitron Inc.) with a 400-mm radius curvature conospherical diamond probe tip. A trapezoidal load function was applied to each indent site (inner or outer region) with loading (10 s), hold (2 s), and unloading (10 s). The operation of the indentations was force-controlled to a maximum indentation depth of 500 nm. The values of elastic modulus and hardness were then calculated.

Statistical analysis

Sample sizes for all quantitative data were determined by power analysis with one-way ANOVA or two-way ANOVA using an α level of 0.05, a power of 0.8, and an effect size of 1.50 chosen for performance of biochemical, gene expression, and biomechanical analyses in vitro and in vivo upon verification of a normal data distribution. The necessary sample size was 5 to 7 to achieve a power value of 0.8 for the in vitro and in vivo parameters in this study. Expected mean and SD values were entered on the basis of our previous study on tissue-engineered meniscus (7).

All statistical data were expressed as means ± SD. Comparisons of differences between construct regions for each stimulus group were used to test the anisotropic properties of the engineered tissue. The efficacy of stimuli in building functional properties within a given region was also tested. One-way ANOVA or two-way ANOVA with Tukey’s test was used to analyze the data. When the two-way ANOVA showed significance (P < 0.05), Tukey’s test was applied. Interaction effects were estimated using a general linear model. The independent variables were biochemical stimulus, biomechanical stimulus, and their interaction term. A significant synergistic effect was determined between the biochemical and biomechanical stimuli (table S17). All data analyses were performed using SPSS statistical software (version 15.0; SPSS Inc.). Values of P < 0.05 were considered statistically significant.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/11/487/eaao0750/DC1

Fig. S1. Calculated stress fields across a circular scaffold with homogeneous structure at 10% displacement of the loading platen.

Fig. S2. Representative images of regenerated tissues at 6 and 12 weeks postoperatively.

Fig. S3. Representative H&E-stained images of the interfaces between the implant and regenerated tissues and synovium in the double-stimuli group and control group.

Fig. S4. Semiquantitative analysis of collagen distribution by immunohistochemistry.

Fig. S5. Biomechanical properties of the regenerated meniscus at 6 and 12 weeks postoperatively.

Fig. S6. Gross and microscopic observations of joints at 6 and 12 weeks postoperatively.

Fig. S7. Dynamic loading and biomechanical properties.

Fig. S8. Representative images of positive and negative controls using tendon (COL-1) and articular cartilage (COL-2) samples from rabbits.

Fig. S9. Schematic diagram showing implantation of an engineered meniscus.

Fig. S10. Samples for tensile and compressive testing.

Table S1. Statistical significance for comparisons between the groups in Fig. 3 (A to C).

Table S2. Statistical significance for comparisons between the groups in Fig. 3D.

Table S3. Gross evaluation of meniscus implant scores.

Table S4. Statistical significance for comparisons between the groups in fig. S4.

Table S5. Histological features of implants.

Table S6. Statistical significance for comparisons between the groups in Fig. 5A.

Table S7. Statistical significance for comparisons between the groups in Fig. 5F and fig. S5.

Table S8. Statistical significance for comparisons between the groups in Fig. 5B.

Table S9. Statistical significance for comparisons between the groups in Fig. 5C.

Table S10. Statistical significance for comparisons between the groups in Fig. 5D.

Table S11. Statistical significance for comparisons between the groups in Fig. 5E.

Table S12. Statistical significance for comparisons between the groups in Fig. 5G.

Table S13. Statistical significance for comparisons between the groups in Fig. 5H.

Table S14. Statistical significance for comparisons between the groups in Fig. 5I.

Table S15. Statistical significance for comparisons between the groups in Fig. 5J.

Table S16. Primer sequences used for RT-PCR.

Table S17. Statistical significance for synergism between the biochemical and biomechanical stimuli.

Data file S1. Primary data.

REFERENCES AND NOTES

Acknowledgments: We thank J. Yang and J. P. Kim for technical assistance. Funding: This work was supported by the National Natural Science Foundation of China [grant nos. 51273004 (J.-K.Y.), 31200725 (D.J.), 31670982 (D.J.), 81630056 (J.-K.Y.), 81802142 (Z.-Z.Z.), 81722031 (X.-G.W.), and 81770873 (X.-G.W.)], the National High Technology Research and Development Program of China (863 Program) [grant no. 2012AA020502 (J.-K.Y.)], and the Natural Science Foundation of Guangdong Province [grant no. 2018A030313675 (Z.-Z.Z.)]. Author contributions: Z.-Z.Z., Y.-R.C., S.-J.W., and F.Z. contributed equally to conceiving the study, designing/performing the experiments, analyzing the data, preparing the figures, and writing the manuscript. W.-Y.D., Y.-C.Y., and T.-Q.Z. designed the biomechanical stimulation system and performed the FE analysis. J.-Y.Z. performed the histological analysis. X.-G.W., F.Y., and J.-J.S. helped edit the manuscript, and Z.-G.G. provided oversight. J.-K.Y. and D.J. designed the experiments. Competing interests: The authors declare that they have no competing financial interests. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article