Osteoinductivity of engineered cartilaginous templates devitalized by inducible apoptosis

Significance

It has been previously reported that hypertrophic cartilage tissues engineered from human mesenchymal stromal cells can efficiently remodel in vivo into bone organs, recapitulating developmental steps of endochondral ossification. We have here demonstrated that the extracellular matrix (ECM) of such engineered cartilage, even in the absence of a living cell component, retains frankly osteoinductive properties. The use of an apoptosis-driven devitalization technique revealed the importance of preserving the ECM integrity and, in particular, the embedded factors to trigger the regenerative process. Although exemplified in a skeletal context, our work outlines the general paradigm of cell-based but cell-free off-the-shelf materials capable of activating endogenous cells toward the formation of specific tissues.

Abstract

The role of cell-free extracellular matrix (ECM) in triggering tissue and organ regeneration has gained increased recognition, yet current approaches are predominantly based on the use of ECM from fully developed native tissues at nonhomologous sites. We describe a strategy to generate customized ECM, designed to activate endogenous regenerative programs by recapitulating tissue-specific developmental processes. The paradigm was exemplified in the context of the skeletal system by testing the osteoinductive capacity of engineered and devitalized hypertrophic cartilage, which is the primordial template for the development of most bones. ECM was engineered by inducing chondrogenesis of human mesenchymal stromal cells and devitalized by the implementation of a death-inducible genetic device, leading to cell apoptosis on activation and matrix protein preservation. The resulting hypertrophic cartilage ECM, tested in a stringent ectopic implantation model, efficiently remodeled to form de novo bone tissue of host origin, including mature vasculature and a hematopoietic compartment. Importantly, cartilage ECM could not generate frank bone tissue if devitalized by standard “freeze & thaw” (F&T) cycles, associated with a significant loss of glycosaminoglycans, mineral content, and ECM-bound cytokines critically involved in inflammatory, vascularization, and remodeling processes. These results support the utility of engineered ECM-based devices as off-the-shelf regenerative niches capable of recruiting and instructing resident cells toward the formation of a specific tissue.

The clinical gold standard solution to critical bone defects consists of autologous bone transplantation. However, it is associated with severe donor site morbidity, risks for infection, and limited availability of the material (1, 2). Off-the-shelf synthetic or naturally derived bone substitute materials (e.g., ceramics, collagen) allow bypassing of these issues (3) but have reduced regenerative potency, especially in challenging scenarios (e.g., atrophic nonunions, comminuted fractures, large substance loss, compromised environment). Cell-based approaches could introduce a superior biological functionality, but their clinical use remains rather limited (4), predominantly because of their nonpredictable effectiveness combined with their economic, logistic, and interpatient variability issues (5, 6).

Modern approaches to bone tissue engineering aim at triggering regenerative processes by matching the corresponding developmental program and thus recapitulating the embryonic stages of bone tissue development (7). During embryonic development, long bones typically develop by endochondral ossification, a process involving the formation and subsequent remodeling of a hypertrophic cartilage template (8). Following the principles of “developmental engineering,” the process of endochondral ossification has been successfully reproduced using embryonic stem cells (9) and human mesenchymal stromal cells (hMSCs) (10–12). A further step was achieved with the upscaling of the graft size, leading not only to the successful formation of large bone tissue but also to the development of a mature organ that includes a fully mature hematopoietic compartment (13).

The increased recognition of the potency of the extracellular matrix (ECM)-derived materials in regenerative processes (14) led us to investigate whether a living cell compartment is strictly required or whether the endochondral route could be initiated by a cell-free ECM, represented by a devitalized hypertrophic template. Addressing this question may lead to a better understanding of the elements regulating the endochondral ossification process and to the generation of cell-based but cell-free off-the-shelf materials capable of instructing host osteoprogenitors toward bone formation. A devitalized approach to endochondral ossification has been envisioned from the beginning of the research in this field (9, 10), as it would bypass the complexity of delivering living cells of possibly autologous origin, but it has never been realized to date.

Existing studies converge on the importance of preserving ECM integrity to elicit the desired regenerative effect (15–17). This implies the use of a devitalization strategy reducing alterations in the composition and architecture of the generated template to mimic both the physiologic regenerative milieu and the 3D structure of the fracture callus. Toward this objective, a devitalization approach has been proposed via the induction of apoptosis (18). In particular, an inducible genetic system (19) can be incorporated into primary hMSCs to specifically induce their apoptosis on exposure to a clinical-grade chemical compound. This strategy offers the possibility to generate hypertrophic cartilage templates that can be subsequently devitalized with, theoretically, minimal changes in the ECM.

In this study, we aimed to induce de novo bone organ formation, using cell-free hypertrophic cartilage templates devitalized by apoptotic induction. We hypothesized that the preservation of the ECM integrity, serving as a reservoir of multiple growth factors at physiological levels, is a key prerequisite to recruiting and instructing endogenous progenitors to initiate bone regeneration.

Results

Assessment of the Osteoinductive Properties of Freeze & Thaw Devitalized Constructs.

As a first attempt, we tested whether hypertrophic cartilage templates, engineered according to a previously established protocol (13) and devitalized by standard “freeze & thaw” (F&T) treatment, are capable of inducing bone formation (Fig. 1A). F&T is a well-established physical devitalization method efficiently leading to cell bursting and limited ECM alteration (20, 21). Primary hMSCs were then seeded at high density on cylindrical collagen meshes (8 mm diameter, 4 mm thickness) and cultured for 3 wk in chondrogenic medium, followed by 2 wk of hypertrophic medium. The resulting grafts successfully displayed the typical pattern of a hypertrophic cartilaginous tissue, consisting of a glycosaminoglycan (GAG)-rich core surrounded by a mineralized ring (Vital; Fig. 1B). Some constructs were subsequently devitalized, using a recognized method consisting of three F&T cycles followed by a bath in hypertonic solution (F&T) (22), leading to the absence of human living cells after implantation (Fig. S1A). The F&T constructs were still positively stained for GAG and mineral deposits, but quantitative assessments indicated a marked reduction in the wet weight percentage of GAG (−22%) and mineral content (−35%) (Fig. 1B). The F&T treatment also reduced the amount of growth factors that are known to play a critical role in hypertrophic cartilage remodeling into bone, including VEGFα (−57%), matrix metalloproteinase 13 (MMP-13; −62%), and bone morphogenetic protein 2 (BMP-2; below detection level; Fig. 1C).

Fig. 1.

Assessment of the osteoinductive properties of living (Vital) and F&T (F&T) devitalized engineered hypertrophic cartilage constructs. (A) Experimental scheme for the generation of hypertrophic tissues starting from hMSCs, devitalized and subsequently implanted in nude mice. (B) Representative sections of in vitro (5 wk) generated hypertrophic cartilage samples and the effect of the F&T treatment on the GAGs and calcium content (n = 3). (Scale bars, 100 μm.) (C) Effect of the F&T devitalization on the content of VEGF-α, BMP-2, and MMP-13 protein. (D) 3D reconstruction of microtomographic images (MicroCT) and hematoxylin and eosin (H&E) stained representative sections of samples retrieved 12 wk after ectopic implantation in nude mice (n = 3). (Scale bars, 1 mm.) Error bars represent SEMs of n = 3 measurements.

On ectopic implantation in nude mice for 12 wk, Vital samples were macroscopically colonized by blood cells, in contrast to F&T tissues (Fig. S1B). Vital samples were extensively remodeled, resulting in interconnected mineralized bone trabeculae embedding bone marrow elements (Fig. 1D), with minimal cartilage residuals (Fig. S1B). In contrast, the retrieved F&T samples maintained an outer shell of calcified matrix and abundant remnants of devitalized cartilage, with no evidences of bone or bone marrow formation (Fig. 1D and Fig. S1B). The failure of F&T devitalized hypertrophic constructs to remodel and induce bone formation could be attributed to the absence of living cells within the implanted graft and/or the effect of the devitalization process, leading to a loss of osteoinductive (e.g., BMP-2), angiogenic (e.g., VEGFα), and remodeling (e.g., MMP-13) factors.

Generation of Hypertrophic Constructs Using Death-Inducible hMSCs.

To assess whether a better-preserved devitalized hypertrophic ECM can induce endochondral bone formation, we introduced the strategy of devitalization by apoptosis induction (Fig. S2), which was previously proposed to target the living cell fraction with minimal changes in the ECM components (18). Primary hMSCs were thus retrovirally transduced with a death system based on the inducible dimerization of modified caspase 9 (iDS; Fig. S2A). To ensure the maximum cell death on induction, the population was purified on the basis of the expression of the CD19 reporter surface marker (Fig. 2A). This allowed for the generation of hMSC-iDS capable of being efficiently induced toward apoptosis in 2D culture (>97%; Fig. 2B) by adding the soluble inducer in the culture media. hMSC-iDS could generate hypertrophic cartilage tissues similar to the untransduced cells, as assessed by histological stainings (Fig. 2C) and gene expression analysis showing successful induction of chondrogenic and hypertrophic genes (Fig. 2D). Hence, the gain of the inducible-apoptosis function did not impair the chondrogenic differentiation and subsequent hypertrophy of hMSC-iDS. Importantly, hMSC-iDS hypertrophic templates continued to express the iDS, as revealed by CD19 immunostaining (Fig. 2C), suggesting the possibility of devitalizing the engineered graft by activation of the system.

Fig. 2.

Generation of hypertrophic constructs using death-inducible hMSCs. (A) Flow cytometry measurements of the CD19 (iDS reporter marker) expression by primary hMSC (Untransduced) after iDS retroviral transduction (Transduced) and subsequent enrichment using magnetic beads (Sorted fraction, hMSC-iDS) (n = 5). (B) Assessment of the efficiency of hMSC-iDS apoptosis induction in 2D culture on overnight exposure to the soluble inducer (+chemical inducer of dimerization) (n = 4). (C) Histologic sections of in vitro constructs (5 wk) generated by primary untransduced hMSCs and hMSC-iDS displayed a similar hypertrophic cartilage pattern (Safranin-O and Alizarin-red stainings). (Scale bars = 100 μm.) Only the hMSC-iDS expressed the iDS (CD19 immunostaining). (Scale bar = 50 μm.) (D) Gene expression analysis of hypertrophic templates generated by hMSCs and hMSC-iDS. Error bars represent SEMs of n ≥ 4 measurements.

Devitalization by Apoptosis Induction of Hypertrophic Cartilaginous Templates.

Treatment of hypertrophic constructs by F&T or apoptosis induction (Apoptized) allowed for an effective devitalization, leading to, respectively, 91% and 93% cell killing efficiency, as assessed by flow cytometry measurement of propidium iodide (PI) and annexin V staining (Fig. 3A). Conversely, most of the cells from the Vital group remained viable (16% of annexin V/PI positivity; Fig. 3A). Because the assay measures the apoptosis-driven extracellular translocation of annexin V, the measured cell death is not biased by the reported natural expression of annexin V by chondrocytes (23); in particular, during their mineralization phase (24). Histologic analyses indicated the successful activation of the apoptotic pathway, with clear morphologic evidence of cell and nuclear fragmentation (late stage of apoptosis) throughout Apoptized and F&T constructs, further confirmed by the presence of cleaved caspase 3 in nucleated cells (Fig. 3B). In the Vital group, apoptotic cells were mainly found within the hypertrophic outer ring. Luminex-based analysis showed, in Apoptized samples, the overall maintenance of factors involved in inflammation [monocyte chemoattractant protein-1 (MCP-1), macrophage colony-stimulating factor (M-CSF), IL-8], angiogenesis (VEGFα), and remodeling [MMP-13, osteoprotegerin (OPG)] processes, with levels similar to those of the Vital group. Instead, F&T treatment resulted in a severe impairment of ECM composition, with a significant loss in IL-8 (64.9%; P < 0.0001), MCP-1 (49.4%; P = 0.0388), OPG (37%; P = 0.0015), VEGFα (58.7%; P < 0.0001), and MMP-13 (32.1%; P = 0.0307) compared with the protein content in the Vital constructs. Thus, although the two devitalization methods led to a similar killing of the cells, the induction of apoptosis allowed for a better preservation of representative ECM components.

Fig. 3.

Devitalization of hypertrophic cartilaginous templates. (A) Annexin V/PI flow cytometric analysis of cells retrieved from nondevitalized (Vital) or devitalized hypertrophic constructs, based on F&T cycles (F&T) or apoptosis induction (Apoptized). Both methods led to efficient tissue devitalization, with a killing efficiency superior to 90%. (B) Biochemical (Safranin-O) and immunofluorescence (Cleaved caspase-3) stainings of Vital and devitalized constructs. The devitalization processes efficiently induce cell death within the constructs. (Scale bars, 50 μm.) (C) Quantitative measurement of ECM proteins loss on devitalization of hypertrophic cartilage. Although the apoptotic method led to a minimal protein loss, the F&T treatment dramatically affected the ECM content. Error bars represent SEMs of n = 6 measurements. *P ≤ 0.05; **P ≤ 0.005; ***P ≤ 0.0005.

In Vivo Assessment of Hypertrophic Cartilaginous Templates.

To assess whether a better-preserved acellular ECM is sufficient to induce vascularization and endochondral bone formation, Apoptized, F&T, and Vital templates were implanted ectopically in immunodeficient mice. On retrieval, samples displayed distinct morphologic patterns. Colonization by host blood cells was evident in Vital samples and, to a lower extent, in Apoptized ones, whereas F&T samples did not display macroscopic evidence of vascularization (Fig. 4A). Confocal microscopy confirmed these macroscopic observations, as Vital and Apoptized samples showed the presence of a mature vasculature characterized by CD31+ vessels stabilized by pericytic cells (NG2 staining; Fig. 4B and Fig. S3A). In contrast, F&T samples were marked by the absence of either cells or blood vessels within the constructs. Collectively, these data indicate the successful recruitment of the host vasculature by Apoptized, but not by F&T constructs, a prerequisite for the recapitulation of the endochondral ossification route.

Fig. 4.

Vascularization assessment of implanted hypertrophic cartilage tissues. (A) Macroscopic view of the samples at the time of explantation. Vital and Apoptized constructs displayed signs of blood cell colonization, in contrast to F&T samples. (B) Fluorescence microscopy of representative sections of explanted tissues. Vital and Apoptized constructs contained sinusoid-like vascular structures, positively stained for CD31 and stabilized by NG2+ pericytes. F&T devitalized constructs did not display evidences of vessel formation. (Scale bars, 50 μm.)

Samples were further processed to investigate the presence of bone, cartilage, and bone marrow tissue. Remarkably, Vital and Apoptized samples, in strong contrast to F&T constructs, underwent intense remodeling, giving rise after 12 wk to bone structures, including bone marrow spaces (Fig. 5A). Interestingly, although cortical external structures were observed in Vital and Apoptized groups, only Vital specimens displayed inner bone trabeculae (Fig. S3 B and C). The amount of mineralized tissue quantified after segmentation of micro-computerized tomography images was highest in the Vital specimens, followed by the Apoptized ones (Fig. S3D). However, as this technique does not allow discriminating between calcified cartilage and frank bone tissue, more specific quantification of the tissue types in the different groups was carried out, using histological sections. Histomorphometric analysis indicated in Vital samples the predominant formation of bone and bone marrow tissues (respectively, 24.9% and 32%), whereas cartilaginous regions were negligible (2%; Fig. 5B). Apoptized samples displayed a significantly higher bone (14.8%) and bone marrow formation (5.7%) than F&T samples (1.5% bone, 0.2% marrow). The latter, in turn, contained the highest percentage of cartilage remnants, confirming the limited efficiency of the remodeling process (Fig. 5B).

Fig. 5.

Endochondral bone formation assessment of implanted hypertrophic cartilage tissues. (A) Safranin-O and Masson’s Trichrome stainings of constructs retrieved 12 wk after implantation. Vital constructs underwent a full remodeling into bone, whereas F&T samples resembled an immature collagenous matrix with abundant cartilage remnants. Apoptized samples displayed evidence of perichondral bone formation, embedding a hematopoietic compartment. (Scale bars, 200 μm.) (B) Histologic quantification of bone, cartilage, and marrow tissue areas in sections of explanted living and devitalized constructs. (C) Masson’s Trichrome, tartrate-resistant alkaline phosphatase, osterix, and Alu stainings were performed to, respectively, assess the presence of bone tissue, osteoclasts, osteoblasts, and human versus host cells. The presence of host-derived osteoclasts and osteoblasts was detected only in Vital and Apoptized samples. Human cells were present only in Vital constructs (black arrows). (Scale bars, 100 μm.)

Ossicles of Vital and Apoptized constructs were characterized by the presence of osterix and tartrate-resistant-acid-phosphatase-expressing cells, respectively, representing osteoblastic and osteoclastic lineages (Fig. 5C). Those cells were predominantly lining the edges of the bone marrow regions, suggesting their involvement in tissue remodeling for marrow colonization. In Vital constructs, human cells participated in the bone formation, as assessed by the presence of cells positive for human Alu repeats among nucleated cells (Fig. 5C). In contrast, no human cells could be detected within Apoptized samples, so the formation of perichondral bone could only be attributed to host osteoprogenitors. F&T samples were also marked by the absence of human cells, but with no evidence of frank bone structures or osteoblastic/osteoclastic cells (Fig. 5C).

Discussion

The present study demonstrates the hitherto unreported capacity of devitalized hypertrophic cartilage templates to induce de novo the formation of bone, including a mature vasculature and the presence of a bone marrow compartment, as well as the strict dependency of the regenerative process on the preservation of the ECM matrix and the growth factors and chemokines bound to it. In fact, the formation of heterotopic bone and bone marrow could be achieved only through the implementation of a devitalization strategy minimally affecting ECM integrity.

The deliberate activation of the apoptotic pathway allowed for the preservation of key embedded factors identified as being involved in inflammation (IL-8, MCP-1, M-CSF), vasculature recruitment (VEGFα), and bone remodeling processes (OPG, MMP-13). The initiation of the stage-specific cartilage template remodeling is known to require digestion of the engineered ECM through the identified factors, leading to the attraction of blood vessels via the release of entrapped VEGFα. The in-growing blood vessels could subsequently deliver osteoblastic, osteoclastic, and hematopoietic precursors, completing cartilage resorption and directing formation of bone and associated stromal sinusoids, and in turn providing the microenvironment for hematopoiesis. Conversely, the use of the F&T as a “crude” devitalization technique led to the dramatic loss of those proteins, resulting in negligible vasculature, host cell recruitment, and tissue remodeling on implantation. These observations provide important information on the nature of the signals to be delivered to initiate endogenous formation of bone tissue and also warrant further investigation to identify the complete set of factors necessary and sufficient to instruct an efficient de novo formation of bone and bone marrow. In particular, our work highlights the paramount role played by the growth factors and chemokines embedded within the MSC-deposited ECM in triggering the tissue regeneration process. The potent biological role of the MSC secretome, appropriately bound to the ECM, is also in line with the recent view of MSCs as an “injury drugstore” and emphasizes the trophic effect of MSCs over their direct participation to the tissue formation (25).

In our study, the induced apoptosis of human hypertrophic chondrocytes is compliant with the physiological apoptosis of hypertrophic chondrocytes occurring during endochondral ossification. However, recent studies indicate that endochondral bone formation necessitates the presence of living hypertrophic chondrocytes, part of which directly contributes to the formation of trabecular structures (13) or the stromal niche for hematopoietic cells (26). As a consequence, despite the use of a cartilage intermediate, bone formation induced by devitalized ECM cannot be defined as being of canonical endochondral origin. Indeed, because the deposited bone tissue was predominantly perichondral, it could be attributed to the direct ossification of the mineralized cartilage template. A possible strategy to further improve the bone regeneration capacity of Apoptized constructs could thus be based on the increase of the surface:volume ratio by manufacturing channeled tissues, as recently described (27), to achieve a higher extent of perichondral bone formation.

Current acellular osteoinductive materials are typically enhanced by the delivery of single growth factors (BMPs). Because of the absence of critical accessory cues and ECM ligands that potentiate their effect, these strategies require supraphysiological doses of the morphogen, raising economic and safety issues (28, 29). Conversely, the osteoinductivity of the devitalized hypertrophic template relies on a mixture of factors accumulated at doses within physiological ranges and presented through a backbone of ECM molecules. Thus, Apoptized devitalized hypertrophic cartilage could offer an attractive alternative to currently available off-the-shelf osteoinductive materials, with the potency of cell-based grafts but bypassing both the logistically complex and regulatory costly use of autologous cells and the still-controversial introduction of allogeneic MSCs. As opposed to ECMs derived from native tissues, which are receiving an increased therapeutic interest (30), the present approach offers the opportunity not only to mimic a developmental process to efficiently form bone tissue but also to enrich the engineered templates in targeted proteins. This could be achieved through the overexpression of key identified factors (e.g., BMP-2, VEGFα) during in vitro culture by modified cells, leading to their embedding in the ECM and their preservation by apoptotic induction. The production of “customized” grafts, with an enhanced angiogenic or osteoinductive potency, would be required to target specific classes of patients or compromised environmental conditions at the repair site (e.g., atrophic nonunions requiring extensive vascularization, or simple bone losses requiring only osteoinduction) (31, 32). The ECM-based embedding of different growth factors in a controllable and customizable fashion for specific clinical needs clearly distinguishes the proposed approach from the “smart” ceramic materials, which have been proven to be osteoinductive in large animal models (33).

Obviously, the clinical translation of the proposed system necessitates further development and extensive preclinical studies. In the present work, the apoptosis induction relies on the use of a retroviral vector, leading to the integration of the system in the target cells. Although the approach has been validated for clinical practice (34), the development of alternative nonviral methods capable of efficiently devitalizing hypertrophic cartilaginous templates (e.g., proapoptotic adjuvants) may be preferred. One important issue with clinical implementation of the developed approach is also related to the efficacy in bone formation of Apoptized versus Vital grafts. At the assessed time, Apoptized constructs remained inferior to Vital ones, probably because of the lack of donor cells initially contributing to bone formation. Our study thus needs to be extended to a longer time of observation to assess whether this initial difference will be overcome.

The ectopic model used in the present work allows using human cells and investigating the de novo formation of bone tissue independent from osteoconductive events. Therefore, it represents the most stringent proof of effective osteoinductivity of the devitalized grafts. However, a relevant assessment of the long-term bone-forming capacity of the implanted constructs will require an orthotopic model in immunocompetent animals. This will also allow the study of the regulatory role of osteoprogenitor and inflammatory cells from a bone environment, as well as of mechanical loading parameters.

In conclusion, we demonstrated that engineered hypertrophic cartilaginous matrix, provided a suitable devitalization technique, can deliver the set of factors to induce its remodeling and develop into bone tissue and bone marrow tissue. The findings outline a broader paradigm in regenerative medicine, relying on the engineering of cell-based but cell-free niches capable of recruiting and instructing endogenous cells on the formation of predetermined tissues. The approach can thus be extended to other biological systems to support both innovative translational strategies and fundamental investigations on the role of engineered ECM, decoupled from that of living cells.

Materials and Methods

All human samples were collected with informed consent of the involved individuals, and all mouse experiments were performed in accordance with Swiss law. All studies were approved by the responsible ethics authorities and by the Swiss Federal Veterinary Office.

HMSC-iDS were generated with the use of a retrovirus carrying the inducible death system (iCasp9-ΔCD19) and were purified by magnetic bead sorting based on CD19 expression (35). Cells were expanded for up to four passages (accounting for an average of 15–20 population doubling) to minimize the loss of chondrogenic potential, and were characterized by flow cytometry for putative MSC markers, with results consistent with our previous report (13).

Hypertrophic templates were generated by either the seeding of cells onto type I collagen meshes at a density of 70 × 10⁶ cells/cm³ (upscaled construct) (13) or using transwell culture seeded at 5 × 10⁵ cells per insert (10). Constructs were cultured in chondrogenic conditions for 3 wk in a serum-free chondrogenic medium, followed by 2 wk in a serum-free hypertrophic medium (13). Samples were devitalized following two different protocols: F&T treatment or apoptosis induction. Samples were implanted in s.c. pouches of nude mice (four samples per mouse) and retrieved after 12 wk. The resulting in vitro and in vivo tissues were analyzed histologically, immunohistochemically, biochemically (glycosaminoglycans, DNA, protein content), by real-time RT-PCR, and by microtomography. Tissue development in vivo was evaluated histologically, immunohistochemically, and by microtomography. The survival and contribution to bone formation by HMSCs was evaluated with in-situ hybridization for human Alu sequences. A more complete and detailed description of the methods is included in SI Materials and Methods.