Samples
Osteochondral dowels (8 mm diameter with approximately 1 cm attached subchondral bone) were obtained from the medial tibial plateaus of 18-20 month old bovine specimens (n = 17) using a custom hardened-steel coring bit attached to a standard bench-top drill press. Stifle (knee) joints with an undisturbed joint capsule were procured within six hours post mortem. A sample diameter of 8 mm was chosen because it was relatively large compared with the test indenter (1.048 mm diameter), preventing applied loads from being distributed beyond the boundaries of the loading surface. A rounded, semi-spherical indenter tip minimized stress concentration effects at the interface between the indenter and the sample. Overheating and dehydration were prevented during specimen removal by spraying both the coring bit and the cartilage specimen with isotonic phosphate buffered saline (PBS). Each sample was stored in 5 ml PBS at 4°C until testing (48 hours maximum). Such storage did not alter mechanical properties when compared to freshly extracted samples. All experiments were conducted at room temperature (22°C).
Freeze-Thaw Protocols
Samples were placed in polystyrene culture tubes (VWR) containing 5 ml isotonic PBS solution and stored for 24 hours either at 4°C, -20°C, -80°C, or snap frozen in liquid nitrogen (-196°C) for 2 minutes and subsequently stored at -80°C. The culture tubes (containing samples and PBS) were placed individually by either standing the container in a refrigerator (4°C) freezer (-20°C, -80°C), or individually freezing samples in liquid nitrogen prior to placement at -80°C. Before mechanical testing, each sample (remaining in the culture tube) was rapidly warmed to the testing temperature (22°C) in a 37.5°C saline bath [1, 4, 16, 17]. The total time required for the samples to thermally equilibrate to 22°C were one minute, three minutes, or five minutes, for samples stored at 4°C, -20°C, or -80°C respectively, as determined using a digital thermometer placed in contact with the articular surface of the specimens. While each treatment group consisted of 15 samples, the control group (unfrozen) contained 96 samples, and was used to identify any treatment effects by establishing baseline properties and their variance. Sample positions within the tibial plateau were accounted for by grouping according to limb (right or left) and within plateau (anterior, middle, posterior). Samples from each group (eg. anterior position from the left limb) were then equally divided among the various storage protocols. The different groups were compared using two-way analysis of variance (ANOVA) and no significant differences were found. Mechanical testing occurred twice: immediately after thawing and after four hours of storage at 22°C. The latter mechanical tests were performed to evaluate the permanence of any mechanical property changes resulting from the various temperature treatments.
Mechanical Analysis
Mechanical properties were assessed by applying a dynamically varying, cyclic compression to the articular surface and comparing this with the output of a load cell supporting the sample. The differences between the phase of the applied waveform and that reaching the support was attributed to the intervening specimen. Movement of the indenter was produced by an electrodynamic vibrator (Ling Dynamics, UK). A displacement transducer mounted between the vibrator and the rigid indenter beam monitored displacement of the indenter and indicated the excitation signal (equal amplitude sinusoids). A custom built piezoelectric force beam, with a resolution of 1 × 10-4 N and a frequency response of 2500 Hz, measured force at the base of the specimen, providing the analysis signal. The electrodynamic vibrator was controlled by the function generator component of a spectrum analyzer (Stanford model SR780). The excitation and analysis signals were then compared using Fast Fourier Transform (FFT) analysis within the spectrum analyzer to determine specimen properties. This arrangement is shown in Figure 1 and has been previously used as reliable, sensitive way to measure dynamic mechanical properties of viscoelastic materials [18]. Maximum indenter displacements reached 0.01 mm creating strains less than 5% and producing loading forces of 28.45 N. Output signals from each transducer were amplified prior to input to the spectrum analyzer to increase resolution.
Two separate dynamic tests were performed to adequately span the loading rate ranges of interest. One was limited to a lower loading rate range (0.1-20 Hz), with input and output signals DC (direct current) coupled to allow low frequency analysis. The other included higher loading rate ranges (30-100 Hz) and was AC (alternating current) coupled to allow filtering of DC drift. Recent studies have shown substantial viscoelastic properties at these higher loading frequencies [13], and reveal an importance of assessing properties at high frequencies. Identifying the mechanical response of articular cartilage to these loading rates, covers the physiologic range of loading that occurs during the heel strike [10–12].
While creating the loading protocol used in the current study, the affect of loading frequency (low frequency testing followed by high frequency testing, vs. high frequency testing followed by low frequency testing) was investigated and found not to influence the results of the mechanical testing. Data was collected after 30 cycles to mechanically condition the specimen and ensure that the viscoelastic steady state properties were assessed (as determined by experimentation during loading protocol creation).
The electrodynamic vibrator motion was driven by sinusoidal waveforms spanning a range of frequencies produced in an exponential chirp pattern. The phases of each sine wave were arranged so that they do not add in phase. The chirp function allowed for the calculation of transfer functions quickly without multiple discrete measurements from single frequency sine waves. This decreased testing time, approximately 3 minutes for each sample and avoided potential changes in tissue properties over the testing period.
Each test began with pre-loading of the cartilage sample with the indenter to 0.1 MPa to ensure that contact between indenter and sample was not lost during dynamic oscillation of the indenter head. The complexities involved with the non-destructive assessment of articular cartilage thickness on intact osteochondral dowels creates an inherent difficulty in calculating applied strain. As a result, the tissue was pre-loaded with a consistent indenter to 0.1 MPa rather than pre-strained to a specific level. After pre-loading, the sample was allowed to reach equilibrium (identified by the signal from the force transducer reaching pre-contact levels) for 5 minutes, previously reported as a reasonable equilibrium time for articular cartilage in standard indentation loading [19]. Each sample was tested in 5 ml of PBS contained by a small plastic cup (with stiffness properties two orders of magnitude greater than cartilage) directly attached to the force beam (this is labeled as the hydration chamber in Figure 1). The ability to maintain continuous contact during the course of each test has been reported by other authors [13] and was confirmed in this study by monitoring real-time oscilloscope traces of displacement transducer and force beam signals. Loss of contact would be quite evident as either complete signal loss from the force beam, or a severely diminished force beam signal indicating that the indenter was moving only the solution around the sample.
Complex stiffness was calculated from the product of the calibration constants of the force and displacement transducers and a transfer function (giving complex moduli) provided by the spectrum analyzer. The complex stiffness function (K*) represents force per unit displacement for the specified frequency [20, 21]. The stiffness measurement is complex due to the inclusion of energy loss through hysteresis and consists of both real (storage modulus, E') and imaginary (loss modulus, E") components; E* = E'+ iE", where E* is the complex modulus and i is an imaginary number [20, 21]. Hysteresis levels were observed as a shift of the phase relationship between stress and strain waveforms. Hysteresis was calculated by a simple relationship between the storage and loss moduli; tan δ = E"/E', the tangent of the phase angle. For the purposes of this study, tan δ will be reported in units of degrees.
Detection of Extracellular Matrix Changes
Blyscan assay (B1000 Biocolor, Ireland), a quantitative dye-binding method that uses 1,9-dimethylmethylene blue to label the sulphated polysaccharide component found in both proteoglycans and the protein-free sulphated GAG chains, was used to measure the total sulphated glycan (with 0.5 μg sensitivity) content within the supernatant surrounding the sample, identifying the potential GAG release due to freeze-thaw treatment. Sircol assay (S1000, Biocolor, Ireland), a quantitative dye-binding method using Sirius Red (Direct Red 80), an anionic dye whose molecules become aligned parallel to the long, rigid structure of native collagens that have intact triple helix organization, was used to quantify the potential for freeze-thaw to break the collagen network and release acid-soluble collagens (with 2.5 μg sensitivity) into the supernatant. These assay techniques have previously been used to identify the extraction of extracellular macromolecules from the tissue [22, 23]. The ability of these assays to test the supernatant, avoiding analyses destructive to the sample, allowed samples to be used as their own controls thereby decreasing biological variability and strengthening statistical comparisons.
Statistical Analyses
Two-way repeated measures analysis of variance (ANOVA) with Tukey post hoc analysis were used to compare treatments and controls, taking into account both storage temperatures and the various loading rate frequencies. One-way ANOVA was used to compare frequency bins within each treatment group. A Bonferroni correction was applied to the loading rates for each treatment over both testing times (9 frequency bin comparisons required a corrected significance level of α = 0.006). A post-hoc power analysis verified the ability of this experimental protocol to distinguish between treatments, finding a statistical power > 0.80 for this study. All statistical methods were conducted using SPSS v.16 (Chicago) and α = 0.05 was accepted as significant for all statistical comparisons unless otherwise noted.