Bone tissue engineering with human stem cells

Treatment of extensive bone defects requires autologous bone grafting or implantation of bone substitute materials. An attractive alternative has been to engineer fully viable, biological bone grafts in vitro by culturing osteogenic cells within three-dimensional scaffolds, under conditions supporting bone formation. Such grafts could be used for implantation, but also as physiologically relevant models in basic and translational studies of bone development, disease and drug discovery. A source of human cells that can be derived in large numbers from a small initial harvest and predictably differentiated into bone forming cells is critically important for engineering human bone grafts. We discuss the characteristics and limitations of various types of human embryonic and adult stem cells, and their utility for bone tissue engineering.

elegantly that the coupling between osteoclastic bone resorption and osteoblastic bone formation is needed to maintain bone health, and that both processes are mediated by tumor growth factor-β [4]. Similarly, coordinated responses of bone tissue, bone marrow, periosteum and surrounding soft tissues are needed for bone repair, which involves elements of both endochondral and intramembranous ossifi cation [5]. A tight control of the balance between bone formation and resorption determines normal turnover of bone tissue throughout the lifetime.
Frequently, clinical situations arise where extensive injury, congenital malformations or diseases cause large bone defects, and reconstruction with tissue grafting is needed. Autologous grafts contain the essential components of bone regeneration -osteogenic cells, osteoinductive growth factors and bone-supporting matrixbut are not readily available in every situation. Other limitations include morbidity at the donor site (which often heals more slowly than the repair site) and diffi culties in preparing anatomically shaped grafts from the harvested bone. Alternatives, including allogeneic bone transplantation, have their own limitations, such as tissue matching, disease transmission, batch variability, and an inability to survive and integrate following implantation [6]. Additionally, the large variability in bone defects, the huge complexity of bone architecture, and the high metabolic activity of bone, necessitating immediate blood supply, motivate the development of new treatment strategies [7].

Engineering viable human bone in vitro
Current models of bone formation in vitro are based on the paradigm that cellular diff erentiation and function can be modulated by the same factors known to play a role during embryonic development [8]. In order to engineer an environment supporting bone formation, combina tions of biochemical and biophysical signals need to be presented to the cells in a three-dimensional setting in a way that allows cellular interactions with the surrounding cells and extracellular matrix (Figure 1, left). Th e complexity of signaling -with temporal and spatial gradients of molecular and physical factors aff ecting bone morphogenesis -presents signifi cant challenges to engineering fully viable, functional bone.
One key component of the in vitro bone model is the scaff old, which provides a structural and logistic template for the developing tissue, and can markedly aff ect cell behavior (Figure 1, middle). Several types of porous scaff olds have been shown to support in vitro bone formation by human cells, including those made of ceramics [9,10], native and synthetic polymers [11,12] and composite materials [13].
Scaff old properties important for bone formation include: the size, distribution and shape of the pores [14]; surface roughness [15]; the presence of cell attachment sites [16]; and the biomechanics of both the material [17] and the scaff old structure. In general, the most suitable scaff olds for bone formation are those with large and interconnected pores (which facilitate cell infi ltration and matrix deposition) and rough inner surfaces (which facilitate cell attachment), made of osteoconductive materials (such as bone protein and hydroxyapatite), and with mechanical properties similar to those of native bone (both to enable load-bearing and stimulate osteo genesis). Additional features of interest include aniso tropic structure, capacity for vascularization, and process a bility into anatomically correct shapes. Scaff olds can also incorporate and modulate delivery of molecular signals controlling cellular functions [18].
Another key component of bone tissue engineering is the culture system or bioreactor. Bioreactor systems can be designed to control transport of nutrients and oxygen to cells in clinically sized constructs and provide lineagespecifi c biological stimuli in various regions of the graft [19]. Additionally, the development of functional, loadbearing characteristics of the graft would be enhanced by the application of biophysical stimulation in order to attain mechanical competence in both the cartilage and bone regions.
Advanced bioreactor designs maintain the physiological milieu in the cell microenvironment (pH, temperature, oxygen and nutrient delivery) by perfusion and conditioning of culture medium (Figure 1, right). Bioreactors can also be designed to recapitulate one or more of the developmentally relevant biophysical signals in a time-controlled manner. For example, increased mass transport and fl uid shear by medium perfusion [20][21][22], and cyclic loading [23] have been shown to improve osteogenesis and enable formation of homogenous bone constructs. Ideally, a bioreactor system should be capable of coordinating biological, physiological and mechanical stimuli, and applying them in a spatially and temporally controlled manner to provide lineage-specifi c stimulation within clinically sized grafts.
Th e clinical and scientifi c utility of tissue engineering largely depends on our ability to predictably direct cells to diff erentiate into the right phenotypes in a spatially and temporally defi ned pattern. Th e control of environmental conditions provided through the design of bioreactors -in conjunction with scaff olds -can help gain more insight into the interplay of molecular and physical factors that guide the development of bone from various types of osteogenic cells. Understanding of the developmental process may then serve as feedback to the optimization of engineering parameters toward better graft designs, and towards the use of engineered grafts as models of development and disease.

Sources of human osteogenic cells
Th ere are several basic considerations when choosing a cell source for bone tissue engineering: the choice between the patient's own (autologous) cells or the use of another person's (allogeneic) cells; the availability and ease of tissue harvesting with minimal donor site morbidity; the effi ciency of cell isolation and cell yield; the use of primary osteogenic cells with limited potential for proliferation, or self-renewing stem cells; the use of cells that have both osteogenic and vasculogenic potential, to support the formation of vascularized bone; the homogeneity of the obtained cell population and controllable induction of the osteogenic phenotype; phenotype stability and cell safety after implantation; and the possibility of automation and the development of quality control measures for the generation of cells and grafts.
In most cases, cells are isolated from a tissue harvest and expanded in vitro prior to bone construct preparation. An expansion step increases the number of osteogenic cells (approximately 70 × 10 6 osteoblasts are needed to form 1 cm 3 of new bone) [24], but could also be used for selection and enrichment of the appropriate cell population(s). Cells with high biosynthetic activity are desirable for enhanced bone formation in vitro and subsequent integration with the host tissue. Stable expres sion of the osteogenic phenotype could aid bone regeneration, and is critical in order to avoid non-specifi c tissue development. Several clinical reports of cell-based tissue engineering approaches for bone regeneration are summarized in Table 1.

Cells from bone tissue and periosteum
Adult bone tissue and periosteum can be used as sources of primary osteogenic cells [25][26][27][28]. Isolation techniques usually involve preparation of explant cultures from the dissected tissues, or enzymatic release of progenitor cells from the endosteal and periosteal layers ( Figure 2a).
Stepwise collagenase digestion has been used for the preparation of osteoblast-like populations with lower propor tions of adherent stromal cells [25,29]. In some studies, fetal bone tissue has been used as a potential alogeneic source due to fast cell proliferation and has demonstrated osteogenic potential [30].
In vitro, between 20 and 40 population doublings have been reported for primary bone and periostal cells [31,32]. Studies have indicated diff erences in proliferation rates of the bone cells isolated by diff erent methods and originating from diff erent donors, as well as age-related declines in cell proliferation [26,27,30]. Diff erences in the proliferation potentials of bone cells isolated from diff erent sites have been observed [29,33], similar to bone marrow stromal cells originating from diff erent sites [34].
In most studies, the expression of osteogenic markers (for example, increased alkaline phosphatase activity, synthesis of osteopontin, bone sialoprotein, osteocalcin and extracellular matix calcifi cation) has been noted in the presence of the osteogenic supplements 1,25-dihydroxy vitamin D3, dexamethasone, β-glycerophosphate and L-ascorbic acid [35,36], but only a limited amount of work has directly compared the functional potentials of osteogenic cells isolated from diff erent sources [37]. In addition to osteogenesis, it has been reported that periosteal and endosteal populations also exhibit chondro genic and adipogenic diff erentiation potential [31,32].
Primary human bone and periosteal cells cultured on porous scaff olds formed bone-like tissue [11,38,39]. In separate studies, bone constructs have been engineered from periosteal cells and used clinically to enhance healing of periodontal defects (Table 1).
Taken together, these studies demonstrate that primary osteogenic cells can be isolated from tissues discarded during surgical procedures and used for in vitro studies [33], and suggest that harvests of small tissue volumes from relatively accessible sites (for example, jaw bones during dental implant placement) could potentially be used for cell isolation and preparation of autologous grafts up to several millimeters in diameter and length. In contrast, due to donor site morbidity [40][41][42] and limited proliferation of primary cells, it would be diffi cult to envision routine preparation of large autologous grafts (several centimeters in diameter and length) from primary bone or periosteum-derived cells (Table 1). Th e applicability of such approaches will strongly depend on developing robust cell preparation procedures from source tissues that are inherently variable due to donor age, gender, health status, systemic conditions and genetic background.

Adult stem cells
Adult mesenchymal stem cells capable of diff erentiation into bone, cartilage, adipose, muscle, tendon, ligament and marrow stroma have been found in a variety of tissues, including bone marrow, adipose tissue, synovium, dental pulp, cord blood, umbilical cord and others [43][44][45][46]. For bone regeneration, the most studied source has been the bone marrow, as it was recognized early that its stroma contains bone marrow mesenchymal stem cells (BMSCs) capable of forming bone and cartilage [47]. Bone marrow transplantation is also being used clinically in combination with osteoconductive materials to augment bone healing [6,48].
BMSCs are commonly isolated based on their adherence and growth on tissue culture plastics (Figure 2b). Alternatively, pure bone marrow aspirates can be used to immunoselect BMSCs using specifi c surface markers. Th e small initial numbers of immunoselected cells are then expanded in culture to obtain suffi cient cell mass for therapeutic purposes. Th e number of stem cells (0.001 to 0.01% of the nucleated marrow cells) [43] varies between diff erent patients and reportedly declines with patient age [46]. Additionally, marrow aspiration volume (up to 150 ml) and technique can infl uence the number of isolated stem cells [48]. BMSCs can, however, be cultureexpanded to large numbers and have been reported to reach up to 50 population doublings in vitro [49]. Importantly, studies suggest that the osteogenic potential of BMSCs is maintained in older individuals [46], and appro priate conditions in vitro (for example, culture on collagen substrate, growth factor supplementation of culture media) [50,51] can help maintain cell diff erentiation potential.
Adipose tissue stem cells (ASCs), discovered more recently [44], represent another attractive source for bone tissue engineering due to their accessibility and potential for diff erentiation into osteogenic, chondrogenic, adipogenic and endothelial lineages [52]. Lipoaspirate volumes can range from 100 ml to several liters and contain a relatively high frequency of ASCs (1 to 5% of isolated nucleated cells) [53]. Cell isolation protocols usually include density gradient centrifugation of the collagenase-digested tissue (lipoaspirate or minced adipose tissue) and culture of the adherent cell population (Figure 2c). Similar to BMSCs, the numbers of  (11 patients) rate in group 2 compared to group 1 in the period Group 2: cultured mandibular periosteal 3 to 9 months after surgery; comparable clinical and cells; seeding and culture on polymer fl eece radiological results (80% fusion in group 1, 90% fusion in osteogenic conditions (13 patients) in group 2) 12 months after surgery isolated stem cells are infl uenced by the tissue harvesting procedure, as well as the site of tissue harvesting (for example, arm, thigh, abdomen, breast) [52]. Several groups have reported the formation of bonelike constructs from BMSCs and ASCs cultured on porous scaff olds [54][55][56], and noted positive eff ects of dynamic bioreactor culture on cell distribution and matrix formation [12,20,21]. Recently, our group has reported engineering of fully viable, clinically sized, and precisely shaped temporomandibular joint grafts by culturing BMSCs on anatomically shaped scaff olds in specially designed 'anatomical' bioreactors [57]. Th is study illustrates the feasibility of using adult stem cells for engineering functional human bone grafts, and underlines the importance of perfusion culture to support physiologic cell densities, as well as formation of dense, homogenously distributed bone matrix.
Survival of large engineered grafts after implantation remains an open question due to the need for immediate connection to the host vasculature, which is an unsolved problem of all tissue engineering. Various strategies for pre-vascularization are currently under investigation [7]. For example, Scherberich and colleagues [58] obtained bone constructs with intrinsic vascularization potential by culturing isolated adipose stromal vascular fractions on porous scaff olds in perfusion for 5 days. Th ese eff orts could enhance the survival of implanted grafts once the methods become available for the connection of the graft to the blood supply of the host.
A few clinical studies have reported on bone constructs prepared from adult stem cells and implanted to enhance bone regeneration. Importantly, adverse side eff ects of the transplanted cells have not been reported, and the authors suggested possible positive eff ects of transplanted cells on bone regeneration (Table 1).

Diff erentiation potential and phenotype stability of adult stem cells
Diff erentiation potentials of adult stem cells obtained from various sources are under investigation, as are the culture conditions required to achieve the functional properties of terminally diff erentiated cells. Another focus is determining correlations between the pheno types of cultured cells and their potential for functional diff erentiation. Stem cells isolated from various tissues are frequently evaluated for the expression of surface antigens by fl ow cytometry [43,45], and share common patterns between various BMSC preparations, and a highly conserved profi le between ASC preparations [52]. In spite of the relative uniformity of marker expression, the cell potential to deposit bone matrix can vary quite signifi cantly between diff erent donors and cell popu la tions [21,59]. Simple models for evaluating the diff erentiation potential of cells are provided by micromass and pellet cultures. Pellets are prepared by centrifugation of several hundred thousand cells, and incubated in diff erentiation media for specifi c diff erentiation paths -in most cases osteogenic, chondrogenic and adipogenic. Micromass cultures are prepared by plating droplets of high cell density suspensions on tissue culture plates, which are also incubated in specifi c cell diff erentiation media. In both systems, high cell density helps mimic cell interactions and cell condensation events present during native development of cartilage and bone.
Osteogenesis and chondrogenesis can be measured quantitatively using molecular, biochemical and histological assays [12,21]. Bone formation capacity can also be evaluated in vivo -for example, in ectopic bone formation models [60]. In a recent study, correlations between bone marker gene expression and functional osteogenesis assays have been made to construct a mathematical model for predicting the bone forming capacity of the synovial and periosteal stem cells [60]. In future, such models could possibly be implemented in culture protocols to help develop robust procedures for manufacturing bone grafts.
Several reports of long-term BMSC and ASC cultures (≥4 months, ≥30 doublings) have indicated changes in cell cycle kinetics and the possibility of abnormal karyotype development, leading to malignant cell transfor mation [61,62]. Th ese studies identifi ed some limitations of ex vivo manipulation, which should be taken into consideration and explored further to ensure the biosafety of adult stem cells before their clinical application.

Embryonic stem cells and induced pluripotent stem cells
Pluripotent human embryonic stem cells (ESCs) can form any tissue of the body and have exhibited an unsurpassed (possibly unlimited) potential for proliferation in vitro [63]. ESCs were fi rst successfully isolated and cultured in 1998 by Th omson and colleagues [64], and have enormous value as a potential source of cells for regenerative medicine, as well as a model of early human development. In bone tissue engineering, ESCs could be used as a single source for the derivation of multiple lineages present in adult bone, including osteogenic cells, vascular cells, osteoclasts, nerve cells and others.
Compared to adult stem cells, ESCs require complex culture conditions: they are commonly derived from blastocyst-stage embryos and cultured on mitotically inactivated murine feeder cells in media supplemented with basic fi broblast growth factor and other factors [63]. ESCs grow in colonies, and are passaged as small aggregates by mechanical or enzymatic dissociation from the feeder cells. In recent years, progress has been made, and completely defi ned feeder-free conditions have been reported [65]. In an alternative approach, human feeder cells (for example, skin fi broblasts) have been used for ESC culture [66].
Similar to adult stem cells, pluripotent ESCs are charac terized by the expression of specifi c surface antigens, including stage-specifi c embryonic antigen-4 (SSEA-4), tumor rejection antigens TRA-1-60 and TRA-1-81, and the absence of SSEA-1 [67]. Other markers associated with undiff erentiated ESCs are high alkaline phosphatase and telomerase activities, and expres sion of transcription factors Oct4, Sox2 and Nanog, which are crucial for the maintenance of pluripotency [68]. A standard test for confi rming human ESC diff erentiation potential in vivo is the formation of teratomas after cell injection in immunodefi cient mice. Pluripotency can also be evaluated in vitro by inducing diff erentiation in embryoid bodies (aggregates of cells cultured in suspension) and observing formation of tissues from all three germ layers [67].
Spontaneous development of abnormal karyotypes and other genetic alterations have been observed during prolonged cultivation of ESCs [69,70]. Th erefore, frequent monitoring of the karyotype is recommended, and further studies are needed to ensure stability and safety of the ESC-derived progenitor populations before their potential use in regenerative medicine.
Recently, the prospect of using ESCs for autologous therapies has gained attention with reports of induced pluripotent stem cells derived from adult diff erentiated cells [71]. Induced pluripotent stem cells share many characteristics with ESCs, including morphology, prolifera tion, surface antigens, gene expression, epigenetic status and pluripotency. Development of safer alternatives for cell reprogramming (for example, excluding genetic manipulation) could potentially provide a cell source for autologous therapy [72]. Currently, however, these cells present a unique opportunity to study the development and progression of genetic diseases in vitro.

Osteogenesis of embryonic human stem cells
Th e pluripotent nature of ESCs presents a challenge to the development of effi cient protocols for directing cells into specifi c lineages. Th e embryoid body step has been an integral part of many diff erentiation protocols, including osteogenic diff erentiation. Cells capable of osteo genesis have been found in mixed populations of progenitor cells present in embryoid bodies after 4 to 5 days of culture [73,74], and populations arising from co-cultures with primary bone and periodontal ligament cells [75,76]. A direct diff erentiation protocol excluding the embroid body step has also been tested, and seemed to enhance ESC osteogenesis in vitro [77].
Alternatively, induction and isolation of mesenchymal stem cells (MSCs) from ESCs has been attempted, and these MSC-like progenitors have been subsequently directed into the osteogenic lineage (mostly in monolayer culture). In one study, MSC-like progenitors have been obtained by mechanical isolation of spontaneously diff erentiated cells from ESC cultures, followed by longer culture in confl uent monolayers [78]. In another study, co-culture with stromal cells was used to induce diff erentiation, followed by immunoselection of a MSC-like population [79]. More recently, exposure of ESCs to serum and growth factor supplemented media [80,81] has been used to induce diff erentiation, and MSC-like progenitors have been expanded in a subsequent monolayer culture. Whereas these studies elucidate some of the factors involved in osteogenesis of ESCs, further work is needed to gain a better understanding of the developmental processes involved in specifi cation to bone-forming cells, as well as to evaluate the stability and functionality of the obtained populations [82].

Genetically engineered osteogenic cells
In vivo implantation of stem cells genetically engineered to carry osteogenic genes has been shown to induce rapid bone formation, indicating the possibility of enhancing regenerative processes by combining cell and gene therapy strategies. It has been hypothesized that genetically modifi ed cells exert both autocrine and paracrine eff ects, recruiting host cells to the site of implantation and resulting in enhanced osteogenesis [83]. Importantly, recruitment of host cells could allow for a reduced number of exogenous cells that need to be implanted. In many studies, adult stem cells have been engineered to express genes of the bone morphogenetic protein family (for example, BMP2, BMP4, and BMP7). Other genes of interest include those encoding transcription factors essential for osteoblast diff erentiation (for example, core bind ing factor α1 (Cbfa1), and Osterix), factors enhancing angiogenesis (for example, vascular endothelial growth factor), and bone formation antagonists (for example, noggin) for an additional level of control over bone formation, and combinations of several factors [83]. Th e challenges lay in effi ciently delivering therapeutic genes into the cells without adenoviral and retroviral vectors -for example, by nucleofection (a form of electropermeabilization) -in order to increase safety and allow for the subsequent development of clinical applications.

Future prospects, clinical translation and regulation
In addition to scientifi c challenges, several manu facturing, safety and regulatory issues need to be addressed before cell-based therapies can become routine clinical practice for the treatment of bone defects. Recently, the conditions allowing clinical scale production of cells for therapeutic purposes have been evaluated, including the possibility of automated culture [84,85]. Functional and structural criteria for bone grafts are not yet fully established, and might vary depending on the therapeutic purpose. In pre-clinical studies, load-bearing large animal models should generally be used to assess graft functionality, as research on small animals does not give relevant results due to major diff erences in graft size and healing properties.
Cell-based products are those requiring cell isolation, proliferation and diff erentiation, as well as application of supporting scaff olding materials. Under European Union regulations, cell-based products need to be manufactured in good manufacturing practice facilities under classic pharmaceutical standards [86,87]. Th e regulation of cellbased products falls under the categories of 'human cell, tissue, and cellular and tissue based products' [88,89] in the USA, and 'advanced cell therapies' in Europe as defi ned in [86]. In addition, the International Society for Stem Cell Research published Guidelines for the Clinical Translation of Stem Cells [90], which highlight the scientifi c, clinical, regulatory, ethical, and social issues to be addressed for cell-based products and services. Th ese documents address the safety and use of therapeutic cells, and regulate the necessary evaluations and permissions for the sourcing of material, especially for cells of allogeneic origin (patient information, genetic and infection screening, written informed consent for donors).
As a general principle, stem cell-based approaches should be clinically competitive or superior to existing therapies. Th e risks of using cell-based products should be carefully evaluated with respect to the benefi ts of enhanced functional outcome, application of one single procedure, reduction of cost, and improved quality of life. Clinical trials should be based on a clear rationale and justifi cation of the procedure (with supporting preclinical data and comparisons to existing treatments), and should include characterization of the product, description of administration (including drugs and surgery) and plans for clinical follow-up and adverse events reporting.

Conclusions
Tissue-engineered bone constructs have the potential to alleviate the demand arising from the shortage of suitable autograft and allograft materials for augmenting bone healing. Th ey also can serve as controllable in vitro models of high biological fi delity for studies of bone development, disease or regeneration. Each of the sources of osteogenic human cells -primary cells, MSCs, ESCs and induced pluripotent stem cells -has distinct advantages when used for bone tissue engineering, and the quest for an 'ideal' cell source is still in progress.
Technologies are critical for unlocking the full biological potential of any cell type. To this end, advanced scaff olds (with molecular, structural and mechanical properties designed to mimic bone) and bioreactors (with environmental control and biophysical signaling for enhanced osteogenesis) are being developed to engineer bone grafts and to test the osteogenic capacity of stem cells. Because bone is a vascularized tissue that is actively remodeled through coordinated sequences of bone growth and resorbtion, the requirements are much more complex than 'just' the formation of mineralized bone matrix. Th e need for a vascular compartment, as well as for coordinated activity of osteoblasts and osteoclasts, pose major challenges to directed diff erentiation of stem cells. Ongoing research is advancing from the ability to engineer centimeter-size bone tissue constructs containing viable cells and mineralized matrix to engineering more complex tissue constructs that more closely resemble native bone. It remains to be seen how much can be done (and needs to be done) in vitro to obtain bone grafts for implantation, and to study disease and screen cells and therapeutic agents.