|Home | About | Journals | Submit | Contact Us | Français|
Tissue engineering describes an initiative whereby a deficit of tissue may be replaced with an engineered construct, typically thought to be some combination of a structural support element and a cellular element. There are several mechanical aspects that come into play during the design of such a construct. First, the way in which the mechanical behavior of a tissue is characterized varies depending on the tissue type. For example, one would not consider the ultimate strength of a non–load-bearing tissue such as adipose. However, in bone, where this property helps to describe a functional role, it is of paramount importance. In addition, the arrangement of material chosen to represent the design space has implications regarding the mechanical performance of the scaffold on several different size scales, from the cellular toward the macroscopic. The loading experienced by the implant must also be within the native tissue's mechanical usage window. Future knowledge gained on this subject will continue to characterize the mechanical requirements of various tissues, so that engineering solutions, such as computer-aided tissue engineering, may be utilized from this knowledge. In this article we describe some of these requirements and solutions using bone tissue as an example.
It is well known that biological activity is regulated to a varying degree by mechanical signals. Evidence of this exists in the mechanical stimulation of axons through stretch,1 regulation of bone shape and Wolff's law,2,3,4,5 induction of biochemical cellular responses,6 the modulation of vascular diseases such as hyperplasia through pressure modification,7,8 and so forth. Nevertheless, very little success has been achieved using basic science information about mechanosensitivity for the goal of designing tissue-engineered constructs (TECs). An underestimation of the importance of mechanical issues (with respect to issues such as cellular differentiation, growth factor delivery, and material development) and an uncertainty about how to apply the known information can be attributed to this lack of published knowledge.
Tissue engineering is a gateway into a future in which few will suffer because of organ availability or maladies caused by dysfunctional organs. Its development affects all branches of medicine and certain areas of plastic surgery. Insofar as it relates to the latter, mechanical knowledge is most critical when the tissue in question has a mechanically functional role. Even so, aesthetic reconstructions with no obvious function still require mechanical support to maintain the shape (e.g., soft tissue reconstruction of the ear). The tissues that stand to benefit the most from the addition of a mechanical component to the tissue engineering paradigm are orthopedic in nature (i.e., bone and muscle), where the structure is intimately related to the function. For example, in partial glossectomy, an ideal tongue reconstruction demands the coordination of soft tissue and muscle function to account for speech, food digestion, and aesthetics.9
Before delving specifically into orthopedics, it is worth summarizing some advances in tissue engineering that utilize biomechanical principles. Even traditional “mechanical tissues” are often not tissue engineered through direct mechanical stimulation. For example, cell therapy and growth factors are frequently used for tendon, ligament, and cartilage.10,11 The rationale is that these tissues have a limited capacity for regeneration of nonfibrous tissue and have repair cascades that are sensitive to growth factors. Thus, a biochemical catalyst may operate in place of native mechanical loading. Biochemical strategies that replace mechanical strategies are an exception to the rule but nonetheless portray the status of a field that has had great success in material development and chemistry.
Mechanically induced tissue formation could be considered the fourth factor of the tissue engineering paradigm. This paradigm is a generalized schema whereby (1) cells, (2) growth factors, and (3) substrate are combined into an implantable replacement (Fig. 1). Mechanical aspects such as the appropriate mechanical cues, and the frequency and magnitude of these cues, remain elusive, yet arguments that tissue is accentuated under biomechanical culture12 and has superior cell distributions13 can scarcely be refuted.
There has been significant groundwork investigating the role of biomechanics for a wide variety of tissues. In cartilage, dynamic loading of chondrocytes as well as other cell types14 is stimulatory as evidenced by markers of mature cartilage such as hydroxyproline and glycosaminoglycan.15 Subsequently, tissue substitutes using chondrocytes seeded in hydrogels16 or other scaffolds have been cultured under physiological deformation, which is believed to be responsible for tissue maintenance and to bring about the appropriate zonal morphology.17 Ion channels in vascular endothelial cells can be regulated by static stretch, which has implications for cardiac tissue as well as blood vessel formation.18 Moreover, on the tissue level, it was found that a human saphenous vein responded differently to static culture, constant pressure, and flow perfusion.19 The latter two raised the levels of a protein kinase involved in the signal transduction pathway of vascular remodeling. Indeed, this is relevant in tissue engineering as vascular tissue appears susceptible to mechanically induced pathologies such as high pressure or shear stresses.8 Skeletal and cardiac muscle is also biomechanically active. With respect to cardiac muscle, it has been shown that human heart cells align in the direction of loading and demonstrate an increased maximum tensile strength under cyclic strain.20,21 Cardiac tissue engineering strategies may be performed either by injecting cardiomyogenic cells (from various cell sources) directly into the myocardium or through substitute tissue equivalents.21 One study proposed a promising tissue equivalent consisting of parallel-oriented polymer microfibers acting as a conduit for skeletal myoblasts.22 These examples represent just a portion of the large volume of research that is currently being done in mechanobiology.
The majority of research thus far linking biomechanics to tissue engineering suffers because it is limited to the cellular level and the genesis of tissue. Therefore, no distinction is made as to which mechanical cues are necessary to coordinate growth past its developmental form. Likewise, there is little hope to bring into light information on the reference frame of measurement unless a concerted effort is made to apply a unifying metric of scale, in which the mechanical conditions of similar experiments may be compared. Coincidentally, there is very little known regarding the mechanical aspects of continuum portions of a tissue or a whole organ, with the exception of one tissue, bone.
Each tissue in the body has a unique function and therefore a different set of mechanical requirements. Bone has an exceptional list that includes the capability to provide structural rigidity, act as a reservoir for ions and calcium regulation, and provide a framework for the transfer of muscle forces.23 The inability of bone defects to be adequately and quickly replaced has led to a conglomeration of research seeking to provide cell-seeded scaffolds for both mechanical support and biological functionality. This need arises in several clinical situations including tissue degeneration related to osteoporosis,24,25 voids caused by tumor resection,26 damage caused by trauma,27 and a variety of other genetic diseases affecting the formation of mature bone.28
Because bone is a living tissue, the manner in which local mechanical signals are transduced is salient. If a tissue's mechanical usage may be understood, tissue engineering solutions may be formulated accordingly. Wolff's law, theorized in 1892, was the first to credit bone mechanics for bone shape.29 His trajectorial theory stated that trabecular bone struts intersected at right angles that were aligned with the principal stress axes, which has since been proved incorrect and updated.30,31 A mechanical usage window was introduced by Harold Frost to explain the metabolic adaptation of bone to mechanical signals (see Fig. Fig.22).32 This terminology as well as his “mechanostat” theory helped to explain the apparent level biological machinery of bone. The “usage” of bone is defined as the voluntary mechanical loads on a skeleton during a typical week33 and is delineated by units of strain. Studies have shown that strain levels and strain rates in humans and animals are predominately constant,34 which makes strain a robust metric of consideration.
Frost's mechanostat was one of the first theories to explain adequately many facets of mechanically induced bone formation through a lumped-parameter model. Without considering all the cellular detail, his model describes a control process of bone mass/strength changes. The term mechanostat comes from the analogy to thermostat, where deviations from a set point trigger the mechanism to turn “on” and with no deviation the mechanism remains “off.”35 Frost's model was able to predict 32 verifiably occurring phenomena related to bone, such as the existence of a safety factor in load-bearing bones.32
The usage is described in four quadrants as seen in Figure Figure2.2. In the disuse window (DW), strains below 50 μ cause an osteopenia-type loss of mass in which material near marrow is evenly resorbed, such as occurs in age-related osteoporosis. In the adapted window (AW) that spans the remodeling and modeling thresholds, strains between 100 and 1000 μ trigger conservation remodeling, in which architecture and strength are maintained yet “old” tissue is replaced by “new” tissue. In the mild overload window (MOW), bone mass increases not because of higher remodeling metabolism but because of lamellar modeling drifts that seek to restore the lower strain levels of the AW. At strains larger than 3000 μ, microdamage is proportionally larger than the reparation by remodeling drifts, resulting in decreased strength in the pathological overload window (POW).33 In addition, the intermediate set points, or thresholds, may be subject dependent or altered by a state of disease to explain bone drifts.36
Some limitations to Wolff's law and the mechanostat theory have become evident in recent years. Dynamic loading is not elegantly addressed, but it is critical in the mechanotransduction of at least bone37 and cartilage tissue.38 In bone, a short bout of dynamic loading increased bone formation by recruitment of surface cells,39 whereas longer loading regimes at similar magnitudes appeared to be detrimental to a loaded implant within a rat tibia when compared with static controls.37 Alternative cell types also have different sensitivities to magnitude of loading. Cartilage cells prefer smaller dynamic loading,15 and it is unclear whether there is a preference in bone. Thus, the effect of the magnitude and type of loading is not addressed by Wolff's law or the mechanostat; neither does each qualify completely why the architecture of bone appears the way it does morphologically.
Mechanical properties are not constant in each structural reference frame. The structural hierarchy from largest to smallest is the whole-bone, architectural, tissue, lamellar, and ultrastructural levels (see Table Table1).1). The whole-bone level (3–750 mm) describes the most macroscopic look at bone, including muscles and tendon attachment sites. The two distinct architectural regions (75–300 μm) are trabecular bone, composed of rods and plates found in the ends of long bones, and cortical bone, comprising the concentric shaft of long bones. The tissue level (20–100 μm) consists of individual trabeculae or osteons that have properties determined by the constituent material, a combination of organic matrix and mineral, as opposed to apparent properties that include void space.40 The lamellar level (1–20 μm) is one step smaller and is composed of sheets of collagen and minerals deposited by osteoblasts.41 The ultrastructural level (0.06–0.4 μm) is the smallest basis, which consists of chemical interactions and quantum level relations and is less often considered because of difficulty in its characterization.40
The ability to determine mechanical properties at each individual level is somewhat limited because of mechanical testing protocols, yet there is much information on the tissue and architectural levels. For an in-depth review, see Liebschner.42 As seen in Table Table1,1, the architectural modulus values for trabecular43 and cortical bone44 are less than the respective tissue properties.44,45 This occurs because the former is estimated from a continuum composite of bone and void, whereas the latter captures the mineral phase and as a result is an order of magnitude larger. Naturally, the whole-bone level42 yields a wider spectrum of modulus values—as the structure of the whole is composed of its individual hierarchies, each of which has an associated range.
The architecture has a great impact on the structural properties of bone and, in turn, on its mechanobiology. Native tissue “sees” mechanical stresses coincident with the shape and stress patterns on the local level, which are not adequately capture by a single “modulus” value. The structural hierarchy of bone is one of the most compelling arguments for a biological machinery and ranges from the radial configuration of growing osteons, to the rods and plates of trabecular architecture, to the symmetric interface that occurs during endochondral ossification.46 One simple example illustrating the importance of architecture may be seen in Figure Figure3.3. Consider a composite bone/cell or scaffold/tissue arrangement where the composite experiences a uniform level of deformation or isostrain. In the case of the bone/tissue arrangement, tissue will deform consistent with the scaffold but will experience much lower internal stresses (assuming tissue has a lower modulus). Isostress, where each constituent receives equal stress, results in proportionately larger deformation of the softer material, in this case the tissue. All bone architectures may be viewed as receiving some combination of isostress and isostrain. Therefore, a given architecture contains a complex milieu of stress and strain profiles that are a function of the geometry.
Bone architecture is not only patient specific but site specific. The ratio of trabecular bone to cortical bone, bone mineral density, and boundary geometry of a femur are different from those of a vertebral body. Figure Figure44 illustrates some of these key architectural differences for a trabecular portion of a human vertebral body and iliac crest. The former has a combination of rods and plates; the latter is composed of thicker plate sections, devoid of a truss-like structure. Aside from morphological differences, there are clear structural anisotropy distinctions. The vertebral sample is more aligned in the superior-inferior direction in accordance with the direction of load due to weight in the spine. The iliac crest is less organized in its fabric directions. In addition, the two samples (although unmeasured) probably have similar yield and ultimate strains on the apparent level, but the stresses they experience are drastically dissimilar. Disuse osteoporosis, which is sensitive to mechanics, affects the vertebral body, causing perforations and reductions in trabeculae, although the iliac crest is not vulnerable. The implication for tissue engineering is that constructs must be designed with a priori information on the site, its boundary geometry, and the loading it experiences.
This previous arguments give rise to further questions regarding the mechanotransductive nature of bone. If, in fact, a certain variety of bone cell is responsive to a stress-derived parameter as opposed to a strain-derived parameter; shouldn't the tissue engineering design strategy be tailored accordingly? Similarly, won't the significance of isostress and isostrain vary with this dependence? On the other hand, if the scaffold material or bone tissue can be considered linearly elastic, shouldn't the independent variable, strain, be considered as the mechanotransductive element? The reality is that it is unclear which signaling events are prominent for the evolution of structure under mechanical loading.
Nevertheless, experiments have addressed this issue and arrived at varied conclusions. One problem involves the difficulty in isolating an experimental loading type. For example, how does one apply strain to a substrate in culture media without inadvertently creating fluid flow conditions (causing stresses)? In addition, because cell seeding always requires adhesion to a substrate, much of the literature inherently provides information about substrate-cell relationships,47 which, while helpful for the whole of tissue engineering, only adds complexity to the investigation of mechanostransduction.
Three types of mechanical indicators have been considered: (1) fluid shear stress, (2) direct deformation, and (3) changes in stored energy. Fluid shear stress is believed to be a major mode of cell transduction. Small-diameter canals surrounding osteocytic cell processes cause large pressure gradients even when small mechanical loads are applied to the exterior. These loads are thought to induce a fluid gradient, which then acts as a shear stress on the osteocytes.48 One way mechanical signals influence tissue formation is through differentiation of cells. Figure Figure55 portrays the relationship between interstitial fluid velocity, strain, and the differentiation pathway.49,50 It is clear that bone tissue prefers a balance of small strain and small fluid velocities, and an overexposure (to either) promotes fibrous tissue. You et al found that human and rat osteoblastic cells are less sensitive to substrate deformation than oscillatory fluid flow, at least in regard to expression of messenger RNA levels of osteopontin and intercellular calcium concentrations, two factors thought to be responsible for bone cell recruitment and proliferation.51 Despite these findings, there is certainly evidence of strain-mediated remodeling at the cellular level. Cowin and Weinbaum described how strain may be amplified in the lacuna-canalicular network. Their argument is that flow generated through the canaliculi passes over the fiber glycocalyx and causes a hydraulic resistance. This resistance then distends the cell, producing a hoop stress52 and a straining of the cell in the direction of flow. You et al further deduced that the effect of the drag force on the pericellular matrix is an order of magnitude higher than the shear stress, which can lead to hoop strains 10 to 100 times larger than the applied strain.53 Direct experimental measures of stored energy have been much sparser because there is no good experimental protocol for its measurement. Kunnel et al suggested that hysteresis energy may be a valid indicator of anabolic growth in bones upon mechanically applying a cyclical load to neonatal mouse tibias.54 Inferences to stored energy, however, are most often accomplished through model techniques in which the mechanical properties of the tested material are known so that the stresses may be ascertained.2 The model of Ruimerman et al on three-dimensional representations of trabecular microarchitecture concludes that strain energy density leads to the closest approximation in trabecular bone morphology.55 This conclusion was based on comparing the results of their simulations with reasonable values of volume fraction, trabecular spacing, and net full turnover with the values observed experimentally.
Designing TECs for bone is a daunting task, in part because of the wealth of mechanical factors discussed. Realistically, only a few of these criteria may be optimized at once. Ideally, the TEC would have an architecture that (1) conforms to the boundary, (2) has mechanical properties that are site specific to the defect region, (3) exists within the mechanical usage window (at least on the architectural level), (4) degrades in a time frame that is synchronous with the sum of the infiltrating and developed tissues, and (5) promotes tissue growth by supplying the appropriate mechanical cues on the tissue and cellular level.
Computed tomography and magnetic resonance imaging have had success in isolating the boundaries of bone.56 Computed tomography produces three-dimensional reconstructions of x-ray attenuation coefficients, which can be correlated with density and apparent level properties. This, in turn, is useful in creating finite element models that can predict the internal stresses and strains within an architecture or be correlated with whole-bone strength.57
Many of the design criteria are particularly amenable to computer-aided design, which has spawned the field of computer-aided tissue engineering (CATE). For example, in vivo applications such as computer-assisted surgery prohibit obtaining mechanical data through experiment testing of bone samples. However, if an in vivo scanner can delineate the necessary bone tissue, it may be possible to model the internal stresses and architecture of the undisturbed environment.
Site-specific and patient-specific mechanical design may be addressed by CATE (see Fig. Fig.6).6). CATE describes a complete process of using imaging techniques to determine the boundaries of an implant region, modeling and/or optimization techniques for the scaffold, and computer-aided design methods of manufacture (with or without cellular components) most often utilizing rapid prototyping and negative molding techniques.58 Results of finite element models involved in the second step may provide approximations to continuum elastic moduli. In one example, Wettergreen et al used a library of unit blocks with predefined mechanical properties59 to develop a scaffold of a human vertebral body based on its anatomic stiffness “map.” In this way a scaffold shape was developed with different space filling characteristics than bone but with similar regional stiffnesses.60
The rate at which a TEC degrades is critical and should be proportional to the rate of tissue infiltration and production. In practice, this matter has met with little success. The degradation depends on the type of biomaterial, its surface chemistry, and the local environment. Because design constraints of scaffold microarchitectures have not been identified, material scientists have been unable to develop biomaterials that have the strength of bone tissue and the rapid degradation rates required for metabolic turnover. In addition, the degradation profile severely affects the tissue, if not the apparent, mechanical properties. As the tissue accumulates in the scaffold domain, the locations of de novo tissue determine the isostress and isostrain relationships (see Fig. Fig.7).7). Hutmacher proposed a degradation profile in which initially, at time t0, the implant accounts for the entire mass and volume of the design space whereas at t∞, after scaffold degradation, the bone is self-supporting.61 An augmented version is presented in Figure Figure8,8, with emphasis on the mechanical properties as opposed to mass. As the volume of tissue within the TEC slowly increases, the apparent mechanical properties of the new tissue increase only after sufficient extracellular matrix has deposited to support loading. It remains unclear what strength characteristics of the scaffold need to be present initially to achieve a final desired strength. Should the construct be stiffer initially or maintain a constant strength? Figure Figure88 (bottom) presents three proposed schemas that, with the combination of degradation profiles, could define tissue viability during the regeneration process.
Based on current beliefs, the TEC will support only loads within the DW to POW (see Fig. Fig.1).1). In the DW, a tissue-engineered solution should support biofactors that stimulate growth despite insufficient loading. In the AW, a combination of loading and biofactors should be issued or incorporated within the scaffold. At strains larger than the modeling threshold, mechanical regulation must be minimized, although it is unknown whether biofactors will be useful (by speeding up metabolism) or detrimental (by adding mass) within this loading range. The mechanical environment of the defect site is not likely to change; therefore, the manner in which the tissue infiltrates and scaffold degrades will be responsible for determining the position along the abscissa in Figure Figure2.2. If the mechanical usage reaches the POW and the right combination of isostress and isostrain exists, irreversible cell and tissue apoptosis will occur through mechanical overload. A scaffold's usage may be dictated quite well with CATE58,62 or topology optimization63,64,65 techniques on the apparent level, but limitations with computing technology prevent similar methodologies on the tissue level hierarchy. One study used 7.6 million partitions to discretize the femur geometry at a resolution suitable to obtaining tissue level stresses.66 Even so, it is not clear what mechanical objectives should be met at this level once the stresses are determined.
There is some empirical evidence witnessed in nature for a uniform surface hypothesis. In one example, it was shown that bone responds through minimizing the stresses at the surface, which is believed to reduce the probability of fatigue fracture or crack propagation.67 Mattheck et al implemented a simple optimization routine that dramatically improves the fatigue life of structures by the reduction of peak stresses,68 which could be characterized as a tissue level optimization goal. Apart from this generalized surface criterion, little is known of the impetus behind bone shape evolution and thus an appropriate scaffold configuration at the tissue level. The best research to date has been focused on the design of directional, anisotropic moduli of bone on the apparent level,69 which adds specificity to the scaffold design problem but does not increase additional hierarchical information.
There are number of mechanical issues to consider when fabricating tissue-engineered constructs. The design cannot be as simple as finding a biomaterial that matches the apparent stiffness of tissue. Also critical are the following issues:
Universal protocols for a generic tissue type are doubtful, and thus solutions will be unique to the particular case considered. One such problem and solution strategy was described for bone using work in the field of computer-aided tissue engineering.
The author's would like to thank Jeremy Lemoine for providing micro-CT images of trabecular bone cores of the iliac crest and vertebral body, presented in Figure Figure33.