A bioreactor system was developed to provide high-amplitude cyclic hydrostatic compressive stress (cHSC) using compressed air mixed commercially as needed to create partial pressures of oxygen and carbon dioxide appropriate for the cells under investigation. Operating pressures as high as 300 psi are achievable in this system at cyclic speeds of up to 0.2 Hz. In this study, ligamentous fibroblasts from human periodontal ligaments (n = 6) were compressed on two consecutive days at 150 psi for 3 h each day, and the mRNA for families of extracellular matrix protein and protease isoforms was evaluated by real-time PCR array. Several integrins were significantly upregulated, most notably alpha-3 (6.4-fold), as was SPG7 (12.1-fold). Among the collagens, Col8a1 was highly upregulated at 53.5-fold, with Col6a1, Col6a2, and Col7a1 also significantly upregulated 4.4- to 8.5-fold. MMP-1 was the most affected at 122.9-fold upregulation. MMP-14 likewise increased 17.8-fold with slight reductions for the gelatinases and a significant increase of TIMP-2 at 5.8-fold. The development of this bioreactor system and its utility in characterizing periodontal ligament fibroblast mechanobiology in intermediate-term testing hold promise for better simulating the conditions of the musculoskeletal system and the large cyclic compressive stresses joints may experience in gait, exertion, and mastication.
- gene expression
- extracellular matrix
- cell adhesion
- real-time PCR
- PCR array
for the various tissues that make up the body, each experiences a characteristic set of mechanical stresses. In turn, the distinct stresses within each tissue of the musculoskeletal system define the tissue's mechanical characteristics. The periodontal ligament (PDL), in particular, has a complex relationship with the local mechanical environment as it experiences both tension and compression at high magnitudes, similar to that seen in the meniscus of the knee and annulus fibrosus of the spine. Bone, by contrast, is optimized to withstand primarily compressive stress, as is articular cartilage, although bone transmits shock load while cartilage absorbs it. Blood vessels experience mostly tension due to dilatory stress, as well as a component of fluid shear stress, which also exists deep within bone in the lacuno-canalicular network. Extensive work has been done since the early 1990s on two of these stress states—tension and fluid shear—especially tension, due to the advent of the Flexcell system (4) and related controlled stretching devices (23).
The stress state that has received less attention is compressive stress, despite its fundamental role in the mechanobiology of cells in cartilaginous, fibrocartilaginous, and concentrically loaded ligamentous tissues. Two types of experiments have been reported for compressive stress. One is axial compression, using a loading piston or platen operating on the basis of a materials testing machine. The other is hydrostatic compression, inducing equal pressure in all directions that results in negligible deformation. In the axial compression method, a tissue explant or cells grown in a regenerative matrix are subjected to confined or unconfined compression. The former relies on the porosity of the platen to allow trapped fluid to exude out of the otherwise constrained tissue and provide for realistic deformation (17, 18, 21, 49, 50). One difficulty of this method is the standardization of the platen porosity. Unconfined compression simply mimics free-standing axial stress with bulging at the sides (9, 22, 56). One challenge for both axial compression methods is providing an apparatus with the function of a materials testing machine within a culturing environment. The compromise typically has been to test the culture outside of an incubator, sometimes employing a buffer to extend cell survivability in a non-physiological, benchtop environment.
The second type of compression study reported is based on hydrostatic compressive stress. Typically, this method relies on hydraulic components (15) to provide rapid cyclic pressurization with a high level of control of the input conditions. Because in some tissues physiological levels of stress may exceed several megapascals (58), there is a substantial mechanical demand placed on such a test apparatus, for which servo-hydraulic technology is particularly useful. However, systems of this type are mechanically closed and thus preclude the exchange of gases, which is critical for balancing pH and maintaining the partial pressures of O2 and CO2 appropriate for most musculoskeletal tissues. One design, reported in several versions, uses the servo-hydraulic cycling capacity of a materials testing machine, converting the force generated by the actuator into a hydraulic stress inside a nearby vessel (16, 51).
Alternatively, hydrostatic stress also can be achieved with a pneumatic apparatus, which provides greater control of dissolved gases and pH. In fact, earlier pioneering studies on the effects of compression commonly relied on pneumatic designs. One study on fetal whole bones in mouse applied 1.8 psi (13 kPa) of 5% CO2 in air at 0.3 Hz for 5 days to study the effect on mineralization and growth (38). A complementary system was designed to study the effects of higher levels of compression (44 psi, or 0.1–0.3 MPa) in a pre-osteoblast mouse cell line (46). For both designs, it was determined theoretically and by measurement that the extent of dissolution of the mixed gases was negligible (55). The primary difference between the two systems was that the low level of compression, but not the high level, could be applied cyclically. Achieving a high level of stress, however, is critical for mimicking gait, mastication, and other essential activities of daily life. Providing simultaneously a physiological rate of cycling and magnitude of stress in a biologically supportive system would represent a fundamental advance in studying the mechanobiology of joint tissues.
We designed and fabricated a system for this purpose using mixed gases. To more efficiently realize the mechanical function of the bioreactor, we also designed a system for culturing cells in monolayer that allows them to be passed undisturbed from a quiescent environment to the bioreactor for testing. The utility of this system serves both the basic science study of mechanobiology, as it shows the effects of hydrostatic compressive stress on cellular metabolism and signal transduction, as well as the translation of these findings to tissue-engineered regenerative constructs through long-term culture of seeded scaffolds and their development into mechanically viable implants.
Our initial focus was on the PDL and the response of genes encoding extracellular matrix proteins and adhesion molecules. The mechanical function of the PDL, in particular its role in distributing the load of mastication, suggests that compressive stress may be as essential to its tissue characteristics as the strain. PDL fibroblasts are critical for remodeling extracellular matrix to both distribute and resist loading. In disease, this response may become pathologic, thus altering the stress-strain relationship that defines PDL function. In this study, we applied a hydrostatic compressive stress of 1 MPa (∼10 atm, 150 psi) to PDL cells in the new bioreactor system over 2 days and assayed the change in gene expression. One megapascal is estimated from published reports to approximate the compressive stress experienced while chewing small nuts (1).
MATERIALS AND METHODS
Coverslip cell culture systems.
A custom culturing system was devised to fit into a conventional 150-mm plate and allow transfer of the monolayer samples between the culture plate and bioreactor. The system consists of two stacked circular plates with 18 matching holes, each 20 mm in diameter (Fig. 1). The top plate is made of 3/8-in.-thick polycarbonate, and the bottom plate, holding the culture slip, is made of 1/8-in.-thick stainless steel. The holes in the top plate form the wells to hold the medium, and those in the bottom plate provide a clear view for microscopic visualization with long-working-distance optics. Coverslips 22 mm in diameter are placed in between the plates with preleached EPDM o-rings above and below to allow seeding of the cells, and the plates are fastened together with 3/8-in. screws. A shoulder recess, 1 mm wide, allows the culture slip to rest flush with the top of the bottom plate. In these experiments, the coverslip used was a Thermanox slide (Nalgene Nunc International) coated to enhance cell attachment and sterilized in situ in the two-plate culture system using ozone gas. Other experiments with this system have used 22-mm glass coverslips. Both surfaces facilitate a high proportion of cell attachment. A second culturing device in this bioreactor system is based on a polycarbonate carousel design that holds the transferred culture slips in the circular bioreactor vessel (Fig. 2). The base of the 24-position carousel is machined with radial slots to hold the culture slips using minimal edge contact. In operation, the carousel rests fully submerged in medium in the bioreactor vessel. All culturing components are autoclavable. The pressure vessel, fabricated by Douglas Fluid and Integration Technology (Prosperity, SC) is 76 mm in diameter, 60 mm deep, and made of 1/16-in. stainless steel weld tube construction typically used for chemical processing. The top of the vessel consists of a ferrule cap with a shallow circumferential recess located near the edge that mates with a similar recess on the shoulder of the vessel cup through a formed quick-clamp gasket. A seal is created by compressing the three parts together with a two-bolt split ring clamp.
The function of this bioreactor is to provide high physiological levels of joint-related compressive stress through cyclic injection of high-pressure gases. In these experiments, a mixture of 5% CO2, 5% O2, and 90% N2 was used. The system design allows the gases to be injected from outside the incubator to inside, allowing the bioreactor to function long-term under normal culture conditions of controlled temperature and humidity (Fig. 3). Servo-hydraulic systems, by contrast, typically exceed the housing capacity of incubators, and thus commonly force the cultures to be tested in benchtop conditions that lack proper gas exchange to control pH without additional buffers or to support aerobic metabolism. By tuning the proportions of the various component gases, specific conditions for targeted musculoskeletal tissues can be approximated, such as the low oxygen tension in the deep regions of thick articular cartilage or radial gradients in the knee meniscus. Previous work has treated the issue of variable solubility of the gases in medium and found, both theoretically and experimentally, only a negligible effect for sufficiently quick cycling (55). Our own qualification testing has indicated a shift in pH of <0.2 points after several hours of 0.1 Hz cyclic application of 300 psi. Fast transient shifts in pH cannot be entirely ruled out, but the bulk of observations and data suggest a functionally stable environment for addressing in vivo conditions with this apparatus, assuming that most of the daily culture period supports an equilibrating gas exchange.
The bioreactor system uses a proportional solenoid valve to control the flow of high-pressure mixed gases from the tank to the pressure vessel. The solenoid valve (model QB1-TFEE-500, ProportionAir, McCordsville, IN) mounts behind the incubator and is controlled by a function generator (model FG2, Beckman Industrial, Brea, CA). The air passes from the tank to the valve through conventional refrigerant hosing, then through 1/16-in. stainless tubing with quick-disconnect fittings from the valve to a sintered stainless steel 0.2 μm filter above the vessel. Flow is not appreciably restricted by the filter. In this configuration, tests have been carried out for 17 days without infection or mechanical incident.
Isolation and culture of PDL cells.
Human PDL cells were isolated from freshly extracted teeth in our dental clinic. PDL tissue was harvested from the middle third of the root to avoid gingival and pulpal fibroblasts, rinsed three times in HBSS medium supplemented with antibiotics, then finely minced into pieces 1 to 2 mm on a side. The pieces were placed in a tissue culture dish, allowed to dry briefly to enhance attachment, then the dish was filled carefully with UltraCulture medium (Lonza 12–725F, Switzerland) supplemented with 20% FBS, nonessential amino acids and antibiotics, and changed three times a week. After 7–10 days of routine culture, fibroblasts were seen growing prolifically around the tissue pieces. At confluence, the cells were cryopreserved for later testing at passages 3–6. In this study, a total of six cell lines were cultured from six individuals.
Preparation of test cultures, bioreactor conditions, and gene profiling with PCR array.
Following brief 0.05% trypsin detachment, the PDL cells were subcultured in triplicate on Thermanox coverslips in the custom 18-well system at 100,000 cells/well, or ∼25,000 cells/cm2, and fed with UltraCulture supplemented with 20% fetal FBS (the lone source of ascorbic acid) for 2 days then serum-free medium 24 h prior to the start of the experiment. Experimental and control groups were plated in separate 18-well constructs with parallel medium changes throughout the study. Coverslips of the experimental groups were then transferred to the 24-position carousel and immersed in the bioreactor vessel, while the control groups remained in the 18-well constructs. The carousel top was secured, and the vessel was filled further with medium until a headspace of 2–3 mm remained to distribute the gas pressure. The vessel was then sealed and connected to the gas transfer and control system. The amplitude of the pressure was 150 psi (∼1 MPa, 10 atm) and was cycled at a frequency of 0.1 Hz for 3 h on 2 consecutive days, with the gas source disconnected between test periods. On the first day, the medium was changed in both control and experimental cultures after the stimulation period to minimize any residual effects of partial pressure changes.
Following the second stimulation period, the culture slips were removed from the bioreactor and transferred back to the custom 18-well system, then the RNA was isolated with a Qiagen kit (Cat no. 74106, Valencia, CA). Quality and concentration of the RNA was assessed from 1-μl samples on a Thermo Scientific NanoDrop1000 Spectrophotometer and further validated on a Bioanalyzer (Agilent 2100), requiring a 260/280 ratio of 2.0–2.1 and an RNA integrity number of 9.8–10.0 for inclusion in analysis. RNA was reverse-transcribed into cDNA using RT2 First Strand Kit (SABiosciences). Pairs of biologically distinct specimens were pooled to reduce variance, which yielded n = 3 separate cDNA pools for gene analysis. The cDNA was then amplified in real-time (ABI 7300) using RT2 Profiler reagents (SABiosciences) in a 96-well, SYBR-green-based PCR array designed to target selected families of extracellular matrix proteins and their proteases, as well as cell-matrix adhesion molecules. The primers assay for 84 target genes and 5 housekeeping genes, which were averaged to provide the reference for threshold cycle (delta-Ct) values, plus 3 PCR process controls. Analysis of delta-delta-Ct values between stimulated and unstimulated samples was performed on a spreadsheet using SABiosciences software, which output P values for fold-regulation change, wherein upregulation is relative to a value +1 for controls and downregulation is relative to a value of −1 for controls.
Cyclic hydrostatic stress at 150 psi significantly increased gene expression of the primary α- and β-integrin isoforms while producing less dramatic effects in the other cell adhesion molecules and the common collagen isoforms. ITGA1 (2.3-fold), ITGA2 (3.4), ITGA3 (6.4), ITGA5 (4.9), ITGB1 (2.5), and ITGB5 (4.9) all were significantly upregulated (Table 1, Fig. 4). Constitutive analysis indicated a robust expression for most of the integrin isoforms, with the majority in the upper half of the genes assayed in this ECM plate. Increases also were notable for the laminin gene LAMB3 (5.4-fold), and particularly SPG7 (12.1; Table 2), which plays several intracellular roles, including motility, membrane trafficking, organelle biogenesis, and protein folding. For the few cell adhesion genes that showed downregulatory trends, none achieved significance. Likewise, in the collagen family, there were few genes downregulated by the cyclic hydrostatic stress. Upregulation was significant, however, for COL6A1 (4.4-fold), COL6A2 (4.8), and COL7A1 (8.5), and was especially strong for COL8A1 (53.5). The upregulation of types 6, 7, and 8 collagen may be of particular consequence, as it was found in these PDL cells that the α1 chain of type 6, for example, was expressed constitutively at essentially the same level as that of the α1 chain of type 1 collagen. This suggests a fundamental role for type 6 in defining the physical characteristics of the ligament matrix. The effect of the stress on COL1A1 also appeared upregulatory (3.1-fold), but did not achieve significance. In the remaining ECM proteins, one gene, KAL1, was marginally but significantly downregulated at −2.3-fold (Table 3).
The hydrostatic compressive stress regimen also produced large, statistically significant increases in the expression of two of the ECM proteases, especially MMP-1 (122.9-fold; Table 4) and, to a lesser degree, membranous MMP-14 (17.8). However, there also was a smaller, but still significant decrease in the expression of MMP-8 (−2.5-fold) and MMP-7 (−3.6). As well, stromelysin (MMP-3, −2.9-fold) and both gelatinases (MMP-2, −4.2; MMP-9, −3.0) were trendwise downregulated. To complement the expression of the matrix metalloproteinases, mRNA of one of the tissue inhibitors of MMP, TIMP-2, was significantly upregulated at 5.8-fold; TIMP-1 and -3 were only marginally affected.
This study investigated the effects of cyclic hydrostatic compression on human PDL fibroblasts in a recently developed pneumatic pressure bioreactor. Compression represents a mechanical counterpoint to the high level of tensile stress also experienced by PDL cells. These stresses arise through both normal mastication, which was simulated in this cyclic study design, and the special circumstance of orthodontic tooth movement. Clinically, periodontal tissues are known to remodel rapidly compared with other connective tissues (53). In tooth movement there is bone resorption on the compressive side and apposition on the tensile side (7). Matrix metalloproteinases and tissue inhibitors of MMPs are considered fundamental to both processes (31).
While both these stress components promote remodeling through modulation of MMP expression, the effect each has on particular isoforms can differ. For instance, MMP-1 is upregulated by static and cyclic stretch (4, 11) and by compression (19), as this study corroborates. However, MMP-2 production, which is stimulated by stretch (4, 11, 25), was found here to be trendwise (4-fold) downregulated by compression, in contrast to an earlier report of compression applied on a silastic substrate (25). A similar pattern applies for MMP-9, which may not be affected by stretch (4, 11), but was found in this study to be trendwise (3-fold) downregulated by compression. In the clinic, active forms of MMP-2 and -9 have been inversely correlated with collagen fiber density (3), and both forms of MMP-2 are found in inflamed gingiva (3).
Under stretch stimulation, MMP-14 also has been reported to be unchanged (4, 11), whereas in this study compression yielded a significant 17-fold upregulation of MMP-14. The upregulation of both MMP-1 and MMP-14 seen in this study of cyclic loading supports the findings of complementary in vivo models of static orthodontic loading, which have shown proteolytic effects due to tooth movement in mice (26), rats (54), dogs (48), and humans (52). Similarly, the increase in TIMP-2 found in the current study indicates a stronger regulatory effect due to compression than that reported for tension. Although MMPs as a family can have broad-spectrum proteolytic effects, the possible suppression of gelatinases MMP-2 and MMP-9 by compression seen here, along with promotion of membrane-bound MMP-14 and TIMP-2, may signal distinct remodeling modes in the response of PDL cells to tension vs. compression during tooth movement. According to the literature, this distinction extends to inflammatory mediators as well. Static compression can upregulate IL-6 (40), PGE2, and COX2 (14), supporting the catabolic side of remodeling, whereas tension can suppress IL-1b-induced upregulation of PGE2, COX2, and MMP-1 and -3 (43).
Among the cell adhesion molecules, the effect of compression in this study was upregulatory, in general, particularly for the integrins, suggesting the cells attempted to increase their attachment to the substrate when pressurized. The sensitivity of the integrin genes in some cases followed the pattern of relative constitutive expression; that is, the more common the isoform, the more sensitive its response. Two exceptions were the highly expressed isoforms, αv and β1, which doubled, on average, but did not experience statistically significant changes.
Among the collagens, type I actually was trendwise (3-fold) upregulated in this study, despite its role of resisting tension in skin, ligaments, and fibrocartilage. This finding contrasts with previous reports of downregulation under compression (25, 62). Rather, upregulation is a common finding in studies of stretch, both at transcript (45) and protein levels (25). In this study, the largest increases in collagen isoforms were observed in the minor collagen types of 6, 7, and 8. Although not included with the PCR array employed in this study, type II collagen, as the dominant collagen in articular cartilage, is a presumptive candidate for mechanosensitivity under compression, which has been verified in chondrocyte studies applying cyclic stimulation (51). Other studies have further revealed the sensitivity of chondrocyte to a range of mechanical conditions (27, 47, 51).
To understand soft tissue mechanobiology, it is essential to distinguish how compression is applied. The hydrostatic mode tested in this study isolates the pure compressive stress component experienced in joint tissues, and the cyclic application simulates repetitive motion related to function. Hydrostatic stress is experimentally efficient for the study of compression as it nearly eliminates extraneous movement and strain of the specimen (5, 8, 38, 42, 61). Unconfined compression, on the other hand, imparts strain and, when coupled with a static rather than cyclic regimen, ultimately can downregulate total collagen and proteoglycan biosynthesis in tissues that otherwise respond anabolically to compression, such as the meniscus of the knee (29). Elucidating the molecular pathways of hydrostatic stress presents a challenge due to the lack of conspicuous deformation normally associated with strain-inducing, deviatoric stresses. Mechanical coupling that facilitates interaction between the matrix and cell is implied but unclear.
Despite the lack of shape change, however, there can be substantial remodeling of the cytoskeleton network under compression. Thus it likely shares similar signaling with tensile stress through the ECM-integrin-cytoskeleton linkages, in turn relying on integrins (20, 28, 30, 32), cytoskeleton proteins (33, 57, 59), G proteins (13, 24), receptor tyrosine kinases (RTKs) (36, 39), mitogen-activated protein kinases (MAPKs) (12, 34, 35), and stretch-activated ion channels (15, 32, 37) to transduce mechanical signals. In this way, the current study appears to connect the responsiveness of PDL cells to loading with that shared by other fibroblasts, ranging from tenocytes to cardiac muscle (19, 40, 41, 60, 61).
The bioreactor system presented in this study combines the versatility of a readily transferable culturing system with the relevance of a stress apparatus that achieves cyclic physiological stress magnitudes in a long-term growth environment. In distinguishing between the cellular effects of tension vs. compression, it is essential to be able to clearly differentiate the simulation of these two mechanical conditions. Regarding the latter, the common approach of unconfined axial compression mimics joint deformation properties and joint loading characteristics together (10, 44, 58). Analytically, however, the effect is to lump tension, fluid shear, and compression into a single stimulation. Hydrostatic stress theoretically resolves these issues but, for monolayer culture, may in practice still produce asymmetric binding and shear strain of the integrins to the substrate surface during pressurization (2). An advantage of 3-D culture is that it can minimize transverse strain at the articular surface (6).
The use of culture slips, either plastic or glass, promotes relative ease of maintaining cells in a manner similar to conventional techniques and provides for unencumbered imaging of the stimulated cultures in individualized formats. This may help to detect, for instance, colocalization of proteins that are synergistically related to the cellular response to compressive stress. Changing to a 3-D culture system, either with ex vivo tissue punches or de novo engineered constructs, provides added relevance to cell-matrix interactions, although imaging may be compromised. For 3-D systems, nutrient exchange and mass transport are not yet clearly characterized in a nondeforming system. For these purposes, the efficacy of perfusion systems remains the standard by which to gauge the utility of the hydrostatic approach.
The conclusions of this study are that this new bioreactor design for applying hydrostatic compressive stress through the use of variable compressed gas mixtures efficiently promotes the study of how connective tissue cells react to physiological stress. In PDL cells, cyclic compressive stress strongly affects gene expression of the integrin family of transmembrane proteins, along with some proteases and collagens. The potential of the bioreactor system to serve as a tissue engineering tool in long-term culture is currently under investigation as the next application of its utility. Future developments will assess to what degree the hydrostatic compression, acting alone and in combination with supplementation components, can drive the cellular processes needed for realizing the regenerative capacity of various cell types to recapitulate native tissue.
This work was funded by departmental sources, with personnel support provided by the Egyptian government.
No conflicts of interest, financial or otherwise, are declared by the authors.
Dr. Mohamed Sharawy, GHSU Oral Biology Dept., has been instrumental in establishing the pathway for training young dental scientists, such as Dr. El-Awady, at our campus. Renderings kindly provided by Diantha La Vine, Medical Illustrations Dept., GHSU.
- Copyright © 2011 the American Physiological Society