Integrin Medicated Mechanotransduction of Cyclic Hydrostatic Pressure in Rabbit Intervertebral Disc Cells Cultured In Vitro
A.J.M. Yee, MD, MSc, FRCS(C)
Background: The significant role of certain integrin cell surface proteins in mechanotransduction has been demonstrated in tissues such as cartilage (1). Literature is scant, however, regarding disc integrin cell surface proteins, their potential role in intervertebral disc cell metabolism and how that may be influenced by mechanical loading. We have previously characterized the expression of α5 and β1 integrin by immunoblotting and immunohistochemistry in porcine disc cells. It was the purpose of this study to evaluate the potential effects of cyclic mechanical load on intervertebral disc cell integrin protein expression.
Materials and Methods: Appropriate institutional animal care use approval was obtained for the current study.
Disc Cell Isolation and Culture: Intervertebral discs were harvested from ten 9wk old, 25-30Kg porcine lumbar spines. The nucleus pulposus (NP) and annulus fibrosus (AF) were grossly dissected (transition zone discarded) and disc cells subsequently isolated by serial enzymatic digestion. AF disc tissue was digested using 0.2% w/v pronase for 90 minutes and 0.03% w/v collagenase type IA overnight. NP disc cell isolation utilized 0.4% and 0.012% pronase and collagenase, respectively. Isolated AF and NP cells were seeded and cultured in T75 flasks supplemented with low-glucose DMEM containing 10%FBS and 1%antibiotic/antimycotic solution at 37oC with 5% CO2.
Mechanical Loading: The first passage AF and NP cells were maintained in monolayer culture for up to ten days until subconfluence. The monolayer cultures were trypsinized and seeded onto 6-well tissue culture plates (2x105cell in 5 ml DMEM per well). The cells were grown for 2 days at 37oC with 5%CO2. The cells were refed with media prior to loading and sterile Canola oil added cautiously to the top of the wells prior to placing the tissue culture plates into a custom hydrostatic chamber. The chamber was filled with Canola oil under sterile conditions and pressure applied using a computer controlled servohydraulic materials testing system (MTS 858 Bionix, MTS Systems Corporation, Eden Prairie, MN). Temperature regulation to 37oC was maintained during loading. The cells were exposed to cyclic mechanical pressure of 1 or 3 MPa at a frequency of 1Hz for three days (1 hr/day). Unloaded control plates were also prepared inclusive of the addition of Canola oil during ‘sham’ loading durations. Oil on top of tissue culture media was carefully aspirated from each well following experimentation, and immunoblotting for α5 integrin, β1 integrin, actin, and fibronectin expression was performed.
Immunoblotting: Following loading, each tissue culture well was washed with PBS. In one well, cell viability was determined by trypan blue exclusion. In the remaining wells, 200 µl of lysis buffer was added. Protein quantification of the total cell lysate was performed using DC Protein Assay (Bio-rad Laboratories, Hercules, CA). A standard curve was made from Bovine Serum Albumin (Sigma, St. Louis, MO) and lysis buffer. Quantification was determined spectrometrically at 750nm. Standardized protein samples were mixed with reduced loading buffer (5X) or non-reduced loading buffer (4X; α5 integrin) and boiled for 5 minutes (2). Proteins were separated by 10% polyacrylamide gel electrophoresis (7% for Fibronectin) containing 0.1% SDS with constant current (120V) for 2.5 hrs at room temperature and transferred to a nitrocellulose membrane at 27V over night in a cold room (4ºC). The nitrocellulose sheets were blocked with 5% non-fat dry milk / blocking buffer for 60 min at room temperature (2). Protein expression was determined using primary antibodies [1:2000 rabbit anti-integrin α5 (Chemicon International Inc., Temecula, CA), 1:2000 mouse antiintegrin β1 (Chemicon International Inc., Temecula, CA), 1:1000 rabbit anti-human fibronectin (Sigma, St. Louis, MO) and 1:3000 mouse monoclonal antibody anti-actin (Sigma, St. Louis, MO)] diluted with 5% non-fat dry milk / blocking buffer, incubated overnight at 4ºC. Secondary antibodies used were HRP-conjugated goat anti-rabbit and anti-mouse IgG antibodies (Bio-Rad Laboratories, CA, U.S.A), 1:5000 diluted in 5% non-fat dry milk/0.1%TBST-Tween 20. An ECL detection kit was used, the bands were scanned and densitometry read with Gel Doc.
Results: The majority of cells mechanically loaded using 3mPa, 1Hz, 3 days (1hr/day) were not viable following the loading duration. In contrast, cell viability of greater than ninety percent was observed in those cells loaded using the lower 1mPa amplitude. A reduction in α5 and β1 integrin expression was observed by immunoblotting following loading at 1mPA for 3 days (1 hr/day) at 1Hz in both NP and AF cells. The effect on reduction in β1 expression appeared to be greater in AF when compared to NP cells. Following loading, AF cells demonstrated a reduction in FN expression in the cell lysate which corresponded to a greater expression of FN in the culture media of loaded AF samples. There was no apparent difference in actin expression in the cell lysate comparing loaded to unloaded samples for either disc cell type.
Discussion: Excessive mechanical loading has been shown to be detrimental to disc cell viability through mechanisms that include apoptosis (3). This is consistent with our findings of low cell viability observed in those cells loaded at high, non-physiologic pressures (3mPa experimental group). In those cells loaded at the lower pressure amplitude of 1mPa, cell viability was largely preserved which facilitated the determination of an effect of cyclic mechanical load on the expression of integrin proteins in the intervertebral disc. In comparing loaded to unloaded samples standardized to the total cell lysate protein concentration with immunoblotting, a reduction in α5 and β1 integrin expression was observed following loading at 1mPa (1hr/d) over a 3 day duration. This effect may reflect the down regulation of protein expression and/or the degradation of the cell surface proteins as lower molecular weight bands were noted in several loaded samples. The greater reduction in β1 integrin expression observed in AF cells following loading when compared to NP cells supports the heterogeneity of disc cell populations. With the known role of the α5β1 integrin as a fibronectin (FN) receptor, we note that the reduction in β1 integrin expression in AF cells also resulted in changes to fibronectin expression. Greater fibronectin expression was observed in culture media of AF cells following loading, concordant with an observed reduction in its expression in the cell lysate of the loaded cells. The mechanisms by which mechanical loads are perceived by intervertebral disc cells and its effect on cell metabolism requires ongoing study.
Deformation Characteristics of the Vertebral Pedicle During Pedicle Screw Insertion: Characteristics of Deformation During Insertion of Tapered and Cylindrical Screws
R.F. McLain, MD; B.K. Bay, PhD; L.A. Ferrara, MS; E.C. Benzel, MD
The initial fixation strength of pedicle screws is commonly tested using a standard pull-out test, with load applied at a constant rate. This method overlooks the cyclic nature of in-situ loading responsible for clinical failure. This study was undertaken to determine the effects of stress relaxation properties at the bone-screw interface on screw fixation strength. Pedicle screws were inserted into calf lumbar vertebrae, using a paired testing array. After embedding and mounting in a custom fixture, axial pull-out tests were performed at rates of 1mm/min, 5mm/min and 25 mm/min. For each vertebra, one screw was pulled at a continuous rate. The other screw was pulled at increments of 0.5mm, at the same rate, with 1,000 seconds pause between increments. Peak load, energy-to-failure, displacement-to-failure and stiffness were calculated for each screw pull-out test. Two-way ANOVA showed that the standard pull-out method yielded significantly higher peak loads (p<0.05) at faster pull-out rates and higher stiffness (p<0.05) at all rates compared to the stress relaxation pull-out model. These results suggest that the stress relaxation properties of bone significantly affect the pull-out behavior of pedicle screws, reducing the peak load and stiffness values observed during testing. This mode of testing may provide a better biomechanical model of screw pull-out failure and a more accurate estimate of initial fixation strength.
Our study of microCT morphometry showed that the trabeculae in the pedicle are plate-like and isotropic. Anisotropy in the trabecular network is mainly derived by a particular direction of loading in the bone. Considering the combined bending moments caused by facet orientations in the spine, it is not surprising to observe an absence of anisotropy in the pedicular trabeculae.
The correlation analysis between BV/TV and Tb.Th and Tb.Sp showed weak interrelations and negligible correlations between BV/TV and Conn.D. and Tb.N were observed. These findings can be interpreted such that thinning but not loss of trabeculae causes the change in the bone volume in pedicle cancellous bone. Mathematical models showed that the loss in strength by trabecular thinning was recoverable by therapeutical treatments.
Some studies in the literature documented the morphometric properties of vertebral trabeculae, ie, BV/TV (0.138 and 0.17), Tb.Th (0.156-0.217mm and 0.17mm), Tb.Sp (1.500-1.500mm and 1.11mm) and Tb.N (0.95). Our results indicated that pedicular trabeculae have somewhat different architectural properties than vertebral trabeculae. Trabeculae in pedicle were larger in thickness and number and had less spacing in the network. Moreover, the cancellous bone in vertebral body was anisotropic and consisted of mainly rod-like trabeculae whereas trabeculae in pedicle were more isotropic and plate-like. This suggests that pedicle screw fixation, because of better fixation strength within the pedicle compares favorably to anterior screw fixation where the screws are inserted transversely through the vertebral body.
Localized Trabecular Damage Adjacent to Interbody Fusion Devices and Its Effect on Construct Stability
D.T. Davy, PhD; K.E. Warden, PhD Candidate
Background: Interbody fusion devices (IFDs) have been in widespread clinical use for a number of years. While the early outcome reports contained high clinical and fusion success rates, more recently complications such as cage migration, subsidence and loosening have been described. We hypothesize that a loss of mechanical integrity results from trabecular damage accumulation, in the form of microdamage or trabecular fracture, occurring adjacent to an IFD subjected to load. This study seeks to quantify trabecular damage and its effect when adjacent to IFDs under load.
Purpose: 1) To determine the quantity and distribution of acute, local in-vitro trabecular damage resulting from the presence of IFDs when subjected to load and 2) to determine the loss in mechanical properties due to the trabecular damage accumulation surrounding an IFD.
Methods: Fifteen cadaveric human thoracolumbar spines were obtained for this evaluation. QCT scans of each spine were obtained prior to specimen preparation and bone mineral density (BMD) for each vertebral body was determined using a calibration phantom.
Twenty-four vertebral pairs were cleaned of all soft tissue and potted in polymer resin following the removal of the posterior elements. Each vertebral pair was mounted into a fixture and attached to a materials test machine. Four treatment groups consisting of six specimens each were evaluated in this testing. Generic IFDs of two cross-sectional shapes, square and round, and mechanical damage at two strain levels, 1% and 2.5%, defined the four treatments. Each vertebral pair was randomly assigned to a treatment group and implanted with two bilateral IFDs. The mechanical testing sequence was applied using a trapezoidal waveform (load ramp, hold, unload ramp) and included the application of a diagnostic cycle to 0.4% strain before and after application of the damage cycle. The mechanical properties obtained from the diagnostic cycles included loading stiffness (LS), secant stiffness (SS) and stress relaxation (SR). Ratios of the second to the first diagnostic cycle (i.e., post-damage/pre-damage) for each property were compared within strain levels using unpaired t-tests and correlations between these parameters and BMD were performed using the Pearson correlation coefficient. Before and after the mechanical testing sequence, vertebrae were labeled with the fluorescents alizarin complexone and calcein, respectively. Following mechanical testing, vertebral halves were embedded in polymethylmethacrylate, sectioned, glued to acrylic slides, ground and polished to 150-200µm thickness. One longitudinal and two transverse sections from each vertebra were examined histologically under ultraviolet light to measure tissue area (T.Ar), trabecular bone area (B.Ar), and diffuse damage area (Dx.Ar). Bone area fraction (B.Ar/T.Ar) and damage area fraction (Dx.Ar/B.Ar) were compared within strain levels using unpaired t-tests and correlations between these parameters and BMD were performed using the Pearson correlation coefficient. Results: Vertebral BMDs ranged from 43.8 to 135.9 mg/cc although there was no significant difference between the mean BMD for each treatment group within each strain level.
Mechanically, damage is evidenced by a degradation in the mechanical properties, i.e., a property ratio < 1. At 1% strain, no mechanical damage was detected in any of the three mechanical parameters for any of the square IFD specimens. Ratio means (sd) for the square IFD group were: LS = 1.48 (0.29), SS = 1.33 (0.26) and SR = 1.26 (0.30) and for the round IFD group were: LS = 0.93 (0.11), SS = 0.91 (0.11) and SR = 0.87 (0.12). There was a significant difference between the type of IFD for each ratio mean (p<0.02). At 2.5% strain, the ratio means for the round IFD group were greater than those for the square group, however these differences were not significant. There was no statistical correlation of any of the mechanical parameters and BMD for either IFD group or strain level.
The types of physical damage identified in cancellous tissue include complete trabecular fracture, linear and cross-hatched microcracks and regions of diffuse damage. Histologic analysis of the specimen sections revealed the existence of each of these, however diffuse damage was, by far, the most commonly occurring and occurred in all specimens. Due to an histology error, all of the sections from some of the specimens were lost to analysis; two of the 1% round, three of the 2.5% square and one of the 2.5% round specimens could not be analyzed. In the remaining specimen sections, all labeled damage occurred within 3 or 4mm of the bone-implant interface. Unfortunately, with the loss of specimen sections, little statistical power remained and group comparisons were not performed. Across IFD type and strain level, BMD was significantly correlated to B.Ar/T.Ar (p<0.001) but not to Dx.Ar/B.Ar.
Conclusions: At the lowest strain level, significantly more mechanical damage occurred in the IFDs with round versus square cross-sections, although this was not repeated at the higher strain. It may be that at the onset of load, a consolidation of bone is facilitated by the flat bone-implant interface. In this series however, this effect was mitigated with additional strain. Even at the higher strain, both devices showed only small mechanical effects of damage. Histologic damage was found in all specimens, including those that did not demonstrate mechanical damage. The predominant form of physical damage to the vertebral bone implanted with IFDs was that of diffuse damage and this was located in close proximity to the boneimplant interface. In this study, we did not find a relationship between BMD and either the mechanical properties or histologic measures associated with damage.