Currently, most in vitro approaches focusing on bone regeneration are based on the BTE strategy. This strategy uses 3D scaffolds seeded with cells or cells in combination with bioactive molecules to develop an osteogenic replacement for bone defects. In general, BTE is built on the following pillars [105]:

(i)

Mimicking the extracellular bone matrix by using biocompatible scaffolds;

(ii)

Using osteogenic progenitor cells or osteogenic cells capable of forming bone tissue matrices;

(iii)

Inducing differentiation of the implemented cell types toward the desired phenotype through morphological signals;

(iv)

Supporting the growing tissue with oxygen and nutrients by fully restoring vascularity.

Therefore, many new scaffolds have been developed that are able to closely mimic the mechanical and structural properties of bone and, in combination with cells, induce tissue formation [106]. There are a variety of options for the colonization of these scaffolds. Mainly stromal cells from bone marrow but also from adipose origin are used. Furthermore, osteoblasts, dental pulp cells, periodontal ligament cells, hESCs and iPSCs are used for colonization [107]. In addition, the differentiation and proliferation of the implemented cells is promoted by the addition of BMPs, fibroblast growth factors (FGFs) or VEGFs [108,109,110,111]. Of note, the osteogenic effects of, for instance, BMPs tend to diverge in animals compared to humans. This is suggested to be a result of the different expression pattern of transcription factors that additionally have different functions, especially when comparing MSCs from rodents and from humans [112]. Although VEGF is known to induce osteogenesis in vitro and in vivo in animal studies, no human clinical investigation on this has been performed for in vivo bone formation in humans [113]. Furthermore, biologics and biological compounds that are highly species-specific are becoming more important but fail to function in animals. These findings highlight the need to consider the biological inter-species difference and related specifications with regard to drug development. Thus, it is not surprising that human in vitro models in the field of bone regeneration are gaining more attention.

To generate scaffold-based artificial bone, scaffolds and bone cells are brought together and optionally combined with growth factors. Subsequently, the seeded cells are cultured in a 3D architecture either in non-perfused or perfused cultivation systems such as bioreactors, microfluidic devices. In this context, the scaffold supports cell adherence/attachment and thus migration and colonization and positively influences cell activation, differentiation and proliferation, thus acting as an osteogenic inducer [78]. To fulfill its orthopedic purpose as an implant, a suitable scaffold should have the following properties:

(i)

Biocompatibility to promote cell adhesion and survival;

(ii)

Appropriate scaffold architecture to initiate cell differentiation and proliferation;

(iii)

Adequate mechanical properties to mimic the mechanical properties of the tissue of interest;

(iv)

A bioactive material that allows interaction with the host tissue and has osteoinductive properties;

(v)

Proper biodegradability [78,114,115,116].

Generally, scaffolds are used in orthopedic surgery as a temporary matrix for bone growth [117]. A wide range of materials can be considered to produce scaffolds. Due to the variety of materials and the improvement of manufacturing processes, the quality of research on the biological application of scaffolds has greatly improved in recent years. Sintered metal implants such as titanium or iron–magnesium scaffolds are strong and durable materials that exhibit superior biocompatibility and mechanical properties [118,119]. However, these implants are often not degradable or resorbable, and therefore remain as foreign bodies in the organism for the rest of their lives [116]. Interestingly, Lee and colleagues conducted a clinical trial with a magnesium alloy that proved to be fully biodegradable within one year, thereby supporting full regeneration of the bone defect [120]. One further development in the field of scaffold materials is the group of bioceramics. Bioceramics are often used in bone regeneration due to their osteoinductive capacity and ability to integrate cells such as MSCs into the scaffold. Commonly used materials, such as hydroxyapatite (HA) and tricalcium phosphate (TCP), show significant similarities to the mineral bone content and provide a suitable 3D scaffold architecture for implants [116,121]. In particular, HA is an almost perfect material due to its high biocompatibility, lack of cytotoxicity and controlled degradation properties [122]. However, brittleness and hardness of these materials are a concern with regard to adequate mechanical properties, which decrease in the course of their use [116]. Nevertheless, various approaches using ceramics have shown promising results in terms of bone regeneration in vitro and in vivo [123,124,125]. Scaffolds made of biodegradable polymers, both natural and synthetic, have recently attracted great interest in the field of BTE research and are considered ideal in terms of biocompatibility, durability, bioactive behavior, interaction with host tissue, low immunogenicity and biodegradability [116,117]. Hydrophilic, hydrogel-generating polymers such as gelatin or collagen demonstrating osteo-inductive properties are often used. Due to its natural occurrence in bone, cells easily attach to collagen and show their typical characteristics such as adherence and structural and functional properties, and collagen-based hydrogels reveal good remodeling and biodegradation properties [126,127,128]. In addition, other naturally occurring polymers such as alginate or silk used for BTE approaches have the advantage of easy processing [129,130]. However, synthetically produced polymers have the advantage over natural polymers in that they can be tailored to the specific requirements of the application through chemical modifications or molecular change [126]. As such, several polymers have been approved by the FDA, for instance, polycaprolactone (PCL) [131,132], poly (l-lactic acid) (PLLA) [133,134] or poly (ethylene glycol) (PEG), and are now in use [135]. Recently, bioglass was added to the extensive repertoire of bone scaffold materials [126]. Although scaffold-based approaches are already part of routine clinical practice, e.g., in the reconstruction of the jawbone after tooth extraction, these approaches have certain limitations in terms of modeling fracture healing processes. Most of these models focus on cell–scaffold interactions, biocompatibility and resorbability, thereby disregarding the fine-tuned mechanism of bone healing processes involving cell–cell and cell–matrix interactions. Thus, it is not surprising that the replication of in vivo processes such as morphological, biochemical and biomechanical features are mostly waived [97,136]. In addition, the generation of scaffolds that can either mimic the bone matrix for modeling fracture healing in vitro or be implanted into critical size defects to accelerate fracture gap bridging is focused primarily on cell colonization, biocompatibility and resorbability. While appropriate for implantation, this approach is limited for modeling fracture healing, because adequate diffusion of both oxygen and nutrients is restricted throughout the scaffold region above a certain size. This restriction based on diffusion limits results in heterogeneous cell colonization and thus distribution of cells on the corresponding scaffold [97,102,137,138]. In particular, non-porous scaffolds, which form a durable, dense and solid matrix, face the challenging conditions of low initial cell seeding numbers. Conversely, hydrogel-based approaches that allow homogeneous distribution of cells in larger cell numbers suffer from low durability and stability [102]. Natural biopolymers, on the other hand, have insufficiently resilient mechanical properties, show high batch-to-batch variability and behave in a potentially immunogenic manner, while synthetic scaffolds face the problem of undesirable acidic degradation [126]. In general, research on scaffold-based models focuses on the development and improvement of implants. The in vitro experiments for this are the precursor for the subsequent experiments in animals to finally find application in clinical practice. When using scaffold-based models as fracture healing models or bone models, vascularization is the main challenge in BTE. Above a certain size, tissue thickness limits nutrient and oxygen diffusion, which is required to support osteogenesis as well as osseointegration during bone healing and regeneration. Angiogenesis influences osteogenesis, with bone progenitor cells and osteoblasts located near vascular endothelial cells during new bone formation. In this regard, VEGF is the most important growth factor for vascular growth and is crucial for the effective coupling of angiogenesis and osteogenesis during bone healing and bone regeneration, respectively. Therefore, various strategies have been explored to develop a suitable vascular network in engineered scaffolds, such as (i) the use of biocompatible materials in scaffold design, (ii) micro-nano-structure, morphology and porosity, and roughness of scaffolds, (iii) ion-doped materials and (iv) the addition of angiogenic growth factors or recombinant proteins [139]. To address the limitation of insufficient vascularization, Ma and colleagues used an approach in which they incorporated magnesium particles into 3D-printed porous tantalum scaffolds via dopamine self-polymerization. Using this approach, they demonstrated improved osteogenic and angiogenic potential in vitro and improved osseointegration in vivo [140]. Xu et al. used an electrospun, fiber-porous PLLA/gelatin composite material doped with ceria nanoparticles that exhibited angiogenic properties in vivo, as demonstrated by the hen’s egg chorioallantoic membrane test (HET-CAM) [141]. Quazi et al. demonstrated that bioactive ions released from bioglass formed a vascular network in an in vitro 2D tube formation assay using HUVEC cells [142]. These endeavors demonstrate the sophisticated strategies that are being used to significantly improve the inadequate vascularization of scaffold-based constructs and make them useful as in vitro models.

Scaffold-free approaches exploit the ability of, for example, MSCs to self-organize and self-assemble. Spheroids are already frequently in use in order to mimic complex tissues such as brain, liver, solid tumors, cartilage and bone [143,144,145,146]. These models consist of self-assembled and self-organized cells and their ECM in a close cell–cell contact [147]. Once the conditions for self-assembly and self-organization are explored and established, handling and production becomes easy, offering the possibility for mid- to high-throughput approaches [78]. To date, a wide range of spheroid production methods have been developed, including folding of cell sheets, assembly of cells in hanging droplets, pelletization or centrifugation techniques to form micro- or mini-mass cultures, or the use of non-adhesive surfaces that lead to cell clumping [97]. In case of 3D bone tissue spheroid cultures, the osteogenic potential of MSC pellets is specifically enhanced [148,149]. The use of spheroid models mimics the in vivo situation better than 2D cultures due to their 3D architecture and has distinct advantages over scaffold-based approaches, e.g., no foreign or even exogenous materials are used, so challenging parameters such as biocompatibility and resorbability do not play a role. In addition, tissue formation is less time-consuming than in scaffold-based approaches due to the initially high cell density, eliminating the pitfalls of cell colonization involving cell proliferation and migration as decisive factors [102]. Since spheroids are very small items often based on micro- and mini-mass cultures and spatial or temporal changes on a superordinate tissue level are very challenging to obtain, we recently developed scaffold-free bone constructs on a macroscale-level using a combination of mesenchymal condensation and intermittent mechanical stimulation. This approach is based on the work of Ponomarev and colleagues, who introduced scaffold-free cartilage transplants only consisting of cells and their ECM, which showed distinct similarities to human cartilage [150]. We also used these models for studies on the cytokine-driven cellular and matrix-related changes during cartilage degradation [151] and as the cartilaginous part of a 3D osteochondral model which we established to mimic cytokine-induced features of arthritis in vitro [152]. The generation of scaffold-free bone-like constructs (SFBCs) was advanced by slightly adjusting the manufacturing protocol of the scaffold-free cartilage grafts and by adding osteogenic components during the maturation process toward endochondral ossification. With this approach, we achieved mineralization of the complete SFBCs without the typical formation of necrotic areas in the center of the constructs. Furthermore, by analyzing the expression of specific bone-related markers at the mRNA and protein levels, we were able to demonstrate the functionality of SFBCs based on the enhancement of the osteogenic potential of immune cells in an in vitro fracture hematoma, mimicking the initial phase of fracture healing [56,153]. The main advantage of such macroscale approaches compared to spheroids is the 3D architecture, low cell number, physiologically relevant size, matrix density and typical mechanical properties [151,153].

However, the lack of vascularization of spheroids above a critical size often leads to the formation of necrotic centers, which is a well-known problem of spheroidal cell cultures [154]. Above a critical spheroid size, cells of the center region face challenging conditions of a reduction in supply of oxygen and nutrients. Although cells, especially MSCs, are capable of adapting to a hypoxic microenvironment and insufficient nutrient supply by metabolic reprogramming, they are not able to cope with an anoxic microenvironment and the complete lack of nutrient supply. Reduction of physiological oxygen availability with sufficient supply of nutrients, e.g., glucose, seems to favor MSC survival and their differentiation towards the osteogenic lineage, while adipogenesis is reduced [56,155,156]. Conversely, hypoxia was also shown to inhibit osteogenesis in MSCs through direct regulation of RUNX2 and/or by activation of the Notch signaling pathway [157,158]. In addition, hypoxia was demonstrated to promote cell proliferation and limit cell differentiation of MG-63 osteoblasts [159,160]. Thus, the adaptation processes of osteogenic cells and their precursors is still a matter of research, and new insights could eventually lead to considerable improvements in terms of spheroidal bone tissue cultures. However, the more complex a tissue is, the less suitable it is for modeling using self-organized spheroids. Bone itself is composed of different structures including a variety of cell types. Therefore, only individual components are suitable for modeling with spheroids, such as the anabolic bone components including osteoblasts and osteocytes, while catabolic osteoclasts are difficult or impossible to include. This may explain why spheroids equivalent to human bone are still difficult to find. However, there are several approaches to overcome the problem of cell homogeneity by co-culturing arrangements using different cell types for spheroid formation.

A very recent approach to using different cell types in an organized manner combines scaffolds with scaffold-free spheroid models, using the latter as building blocks for tissue engineering of bone, thus combining the synergistic effects of both components in terms of osteo-differentiation and osteo-integration [102,161,162]. In this approach, the scaffold represents the ECM to which cell-containing spheroids adhere, into which the cells can migrate and thereby proliferate and differentiate. In addition, scaffolds must be designed to maintain and, at best, preserve spheroids in vitro and if needed in vivo for the purpose to regenerate injured tissue by promoting repair processes and vascularization [163,164]. Thus, morphogenesis is mimicked by the intrinsic capacity of spheroids to fuse to each other [161]. To merge with scaffolds, spheroids are most often combined with cage-like scaffolds of various materials. For instance, Sankar et al. used electrospun fiber mats in combination with MSC spheroids to obtain an enhanced osteogenic differentiation [162]. The essential bottom-up engineering strategy that enables the combination of scaffolds and spheroids in a highly coordinated and reproducible manner is 3D bioprinting. This technique not only allows the production of micro-scaffolds in almost any desired shape, but also the combination of cells and scaffolds [97]. Daly and colleagues used an inkjet-based bioprinting approach to seed MSCs and chondrocytes into 3D-printed micro-chambers, resulting in a cartilage construct similar to native cartilage [165].

Although 3D bioprinting is a promising technology and allows the generation of highly reproducible models in mid-throughput output, several challenges must be addressed. Since the biomaterials and spheroids have the same output requirements, these face the same limitations, as already discussed. Additionally, the implementation of vasculature to supply nutrient and oxygen to inner tissue structures is still elusive.

Healthy bone is characterized by a distinct vascularization to supply and extract nutrients, oxygen, metabolites, growth factors and hormones from the bone. Therefore, vascularization plays an essential role in bone development, bone maturation, bone growth, bone regeneration and bone remodeling [166]. Several approaches have been attempted to implement vascularization within in vitro models of bone tissue. The endeavor is quite complex, which means that even for the approaches of in vitro modeling of angiogenic and osteogenic material, a living organism seems to be still needed. Zhang et al. developed a double-cell sheet complex in vitro, whereby the first cell sheet is composed of an osteogenic cell layer (osteogenic potential) and the second of vascular endothelial cells (blood vessel potential), transplanted to nude mice. After 12 weeks, they observed both osteogenic and blood vessel formation potential in vivo [103]. Thus, replacement of animal experiments for this approach is not yet achievable. As a real alternative approach to the still necessary use of animals at least as hosts for in vitro models, Chiesa et al. developed a vascularized in vitro bone model by employing 3D bioprinting. They combined pre-differentiated osteogenic MSCs and HUVECs on a gelatin-nanohydroxyapatite 3D bioprinted scaffold. After 4 weeks of cultivation, the HUVECs formed a tubular-like structure and a capillary-like network within the bone constructs, alongside ongoing osteogenesis [104].

To mimic the crosstalk between immune cells and bone during the initial phase of fracture healing, we added the immune component to bone cells and established a fracture hematoma model. In brief, we combined scaffold-free macroscale bone constructs (SFBCs) with a coagulant of human peripheral blood and MSCs and demonstrated the osteo-inductive potential of SFBCs on fracture hematoma, mimicking several key features of the initial phase after fracture in vitro [153]. Thus, the combination of cells and tissues in sophisticated in vitro models may be promising tools to unravel distinct fracture-healing processes in order to gain insights into human fracture healing and finally to improve human healthcare. To significantly improve the cultivation of both scaffold-based and scaffold-free constructs, dynamic bioreactor systems offer an attractive way to increase differentiation and proliferation of implemented cells via monitoring of metabolic parameters and controllable perfusion. These include, but are not limited to, perfusion bioreactors, spinner flasks, and systems with electromagnetic or mechanical stimulation properties, and allow adequate monitoring and control of various important properties such as physical, biological, or chemical parameters during the process of bone formation in vitro, as has been extensively reported by Rauh and colleagues [167].